Final Report: Own-Source Revenue Potential and Tax Gap Study of Kenya’s County Governments August 2018 Acronyms ASI Adam Smith International CIT Corporate Income Tax COB Office of the Controller of Budget COG Council of Governors CRA Commission for Revenue Allocation DEA Data Envelopment Analysis GCP Gross County Product GDP Gross Domestic Product GoK Government of Kenya GVA Gross Value Added IMF International Monetary Fund ISS Integrated Survey of Services KNBS Kenya National Bureau of Statistics KRA Kenya Revenue Authority KSH Kenya Shillings LATF Local Authorities Transfer Fund MoL Ministry of Lands MSME Micro, Small and Medium Enterprises NT Kenya National Treasury OSR Own-Source Revenue PFM Public Financial Management ToR Terms of Reference VAT Value Added Tax WB World Bank Contents Executive Summary 5 Introduction and Methodology 9 Our understanding of the context 9 Study Objectives 9 Methodology 10 Legal and Policy Review 12 Approach and Principles for the Review 12 Mapping of current legal and policy framework 16 Critique of policy and legal framework and practice 20 Options for new revenue sources 29 Revenue administration key findings 32 Revenue management key findings 35 Revenue Potential Estimation 37 3.1 Motivation - Uses for revenue potential estimates 37 Selection of main county revenue sources 38 County revenue collections by source 40 Data availability and gaps 41 Results from estimation of aggregate revenue potential 46 Top-down analysis: Results for main revenue sources 47 Recommendations: Strategies for Enhancing County Own-Source Revenue 74 Data quality and availability 74 Legal framework – General Recommendations 75 General Recommendations – Policy 75 Property Tax 76 Entertainment Tax 76 Single Business Permits 77 Cess 77 Other Revenue Sources 77 Options for new revenue sources 78 Revenue Administration 78 Revenue Management 79 Conclusions and Next Steps 80 Annex 1: Questionnaire Summary Report 82 Annex 2: Questionnaire 97 Annex 3: Aggregate county revenue potential estimates (frontier analysis) 104 Annex 4: County revenue potential estimates for six main sources (top-down analysis) 106 Annex 5: List of consultation meetings 112 Table 1: Example pricing structures for non-tax fees and charges (cost-recovery) ....................................................... 15 Table 2: Application of selection criteria against counties’ main revenue sources ......................................................... 40 Table 3: Actual revenue collections for selected counties, FY2016/17, Ksh million ....................................................... 40 Table 4: Policy rationale and legal basis for selected main revenue sources. ............................................................... 41 Table 5: Available audited county revenue data for six main revenue sources, FY2014/15-FY2016/17. ...................... 42 Table 6: Examples for variation in business data between censuses, surveys and taxpayer registries. ....................... 44 Table 7: Comparison of actual trading business licence fees by size of business for sample districts (Ksh). ............... 45 Table 8: Method 1 - Frontier analysis (Data Envelopment Analysis, DEA) model specifications and overview of findings. ........................................................................................................................................................................... 46 Table 9: Five counties with the highest share of total revenue potential for different DEA specifications...................... 47 Table 10: Overview of model specifications for top-down analysis. ............................................................................... 48 Table 11: Method 2 - Top-down revenue potential estimates for six main county revenue sources. ............................. 49 Table 12: Potential revenues from property taxes for selected counties (in million KShs) ....................................................... 54 Table 13: Potential revenues from building permit fees for selected counties (in million KShs) a ................................... 58 Table 14: Comparison of actual licence fees with modelled fee structure (Ksh) ............................................................ 59 Table 15: Counties with highest (unrealized) revenue potential from liquor licences, Ksh million. ................................ 62 Table 16: Comparison of actual licence fees with modelled fee structure (Ksh). ........................................................... 63 Table 17: Counties with highest (unrealized) revenue potential from liquor licences, Ksh million. ................................ 66 Table 18: Parking fees for sample counties. ................................................................................................................... 66 Table 19: Counties with highest (unrealized) revenue potential from parking fees, Ksh million .................................... 69 Table 20: Counties with highest (unrealized) revenue potential from advertisement fees, Ksh million. ......................... 72 Table 21: Trends in cess collections relative to total county collections and national VAT. ........................................... 72 Table 22: Counties with high relative cess collections (>15% of total collections), Ksh million, FY2016/17 .................. 72 Figure 1. Illustration of using county revenue potential in intergovernmental fiscal transfer calculation .................................... 36 Figure 2: Land rates for selected counties between 1982 and 2000. ............................................................................. 45 Figure 3: County revenue potential based on frontier analysis, Ksh billion. ................................................................... 47 Figure 4: Estimated total value of residential and commercial properties across counties, Ksh billion, 2015/16. ......... 51 Figure 5. Value of properties in the low-value group (<600,000 KShs) across Kenya counties.................................................. 52 Figure 6. Estimated property tax revenue potential by county under different rate scenarios (in Ksh million) ............................. 53 Figure 7: Estimated property tax revenue gaps by county under different rate scenarios (only counties with data on property tax collections) .................................................................................................................................................. 54 Figure 8. Total value of residential and commercial construction by county - KSh. ....................................................... 56 Figure 9. Estimated building permit revenue potential by county under different rate scenarios (in Ksh million) ........................ 57 Figure 10. Estimated building permit revenue gap by county under different rate scenarios (only counties with data on building permit collections) ............................................................................................................................................................. 58 Figure 11: Estimated total full-time equivalent employees and working owners across Kenya counties ....................... 60 Figure 12: Estimated business licence revenue gaps by county (only counties with data on business licence collections) ...................................................................................................................................................................... 61 Figure 13: Estimated building licence revenue gaps by county (only counties with data on business licence collections) ........................................................................................................................................................................................ 62 Figure 14: Estimated total full-time equivalent employees and working owners in establishments serving alcohol across Kenya counties .................................................................................................................................................... 64 Figure 15: Estimated liquor licence revenue potential by county (in Ksh million) ........................................................... 65 Figure 16: Estimated liquor licence revenue gaps by county ((only counties with data on liquor licence collections .... 65 Figure 17: Estimated county vehicle parking units/spots across Kenya counties .......................................................... 67 Figure 18: Estimated vehicle parking fee revenue potential by county (in Ksh million) .................................................. 68 Figure 19: Estimated vehicle parking fee revenue gaps by county (only conties with data on vehicle parking fee collections) ...................................................................................................................................................................... 69 Figure 20: Estimated annual company spending on public/outdoor advertising across Kenyan counties, Ksh millions 70 Figure 21: Estimated outdoor advertising fee revenue potential by county (in Ksh million) ........................................... 71 Figure 22: Estimated outdoor advertising fee revenue gaps by county (only counties with data on outdoor advertising fee collections) ................................................................................................................................................................ 72 Executive Summary Introduction and Methodology This is the final report produced by Adam Smith International for the Own-Source Revenue (OSR) Potential and Tax Gap Study of Kenya’s county Governments. This study was commissioned by The World Bank on behalf of the Kenya National Treasury. The key objectives of the study are to:  Map out counties’ current local revenue base and potential.  Support more credible projections by counties of future revenue from assigned taxes, fees, levies and charges.  Develop a framework for monitoring improvements in OSR performance by counties. Methodology The methodological approach adopted for this study is divided into two parts: 1. Legal and policy review: Based on desk research from secondary sources on Kenya OSR and international best practice, as well as selected county visits and responses to a questionnaire circulated to all 47 counties. 2. Revenue potential by county and by revenue stream: Estimates of revenue potential have been established that provide answers to the following questions:  How much revenue would each county be able to raise in total and from each OSR stream if it operated in line with the best performing county in the country? This is calculated using frontier analysis.  How much revenue would each county be able to raise from each OSR stream if it fully utilised the fiscal instruments at its disposal, resolved issues relating to administration and eliminated evasion? This is calculated using the ‘top-down’ approach. Revenue Potential Estimates The frontier analysis of aggregate revenue potential, and particularly the top-down analysis for six key county revenue sources show substantial unrealized county revenue potential (see tables below) ranging between Ksh 55 billion and Ksh 173 billion, compared to current collections of Ksh 35 billion. While data gaps hinder revenue gap analysis for most counties, where data was available, estimated potential compared to actual collections show gaps between 35% and 94% for different county revenue sources. Such substantial gaps are likely to be representative of most, if not all, counties. This suggests that counties can gradually fund an increasing share of local service delivery from own source revenue if they are able to realize more of the available potential over time (while intergovernmental fiscal transfers will continue to play an important role for local goods and services, particularly in health, education and infrastructure). Results – Method 1: Frontier analysis (Data Envelopment Analysis, DEA) model specifications and overview of findings 2018 Data Envelopment Analysis 2015 DEA1 Input (data source) County GDP estimates County household County consumption, County income, (World Bank, 2015) consumption (KIHBS urbanization, education urbanization, education 2015/16) (KIHBS 2015/16) (unknown) Output Average of actual county collections over FY2015/16- Same excluding natural Actual county revenue for FY2016/17 (audited county revenue database) resource revenue 2014/15 Frontier county Weighted average for Laikipia Nairobi and Mombasa Unclear (none shows Isiolo, Kericho and relative efficiency) Baringo Frontier counties Narok (natural resource Narok (natural resource None None excluded due to revenue), Nairobi and revenue), Nairobi and unique features Mombasa (main hubs, Mombasa (main hubs, density) density) Revenue potential Ksh 55 billion Ksh 53 billion Ksh 66 billion Ksh 48 billion estimate (0.83% of GDP) (0.81% of GDP) (1.0% of GDP) (0.84% of GDP) Revenue gap 36% 35% 50% 30% KSh 55 1 Office of the Controller of Budget, Republic of Kenya, 2016. County Revenue Baseline Study 2015. Nairobi. 5 Results – Method 2: Top-down revenue potential estimates for six main county revenue sources Revenue source Total potential (Ksh) % of GDP Revenue gap Property tax: Only for 26 counties (i) assuming 90% lowest value properties exempt, 1% rate 66.2 billion 1.01% 91% with data others (ii) assuming 1% rate all 84.3 billion 1.28% 93% (iii) assuming 0.5% low value, 1.5% high value 108.3 billion 1.65% 94% Building permits Only for 8 counties (i) 1% on all construction value 6.0 billion 0.09% 35% with data (ii) 1% on low, 2% on high value 11.8 billion 0.18% 66% Business licences 23.4 billion 0.31% 75% Only for 41 counties with data Liquor licences 10.2 billion 0.14% 89% Only for 5 counties with data Vehicle parking fees 12.6 billion 0.17% 61% Only for 39 counties with data Outdoor advertising 6.3 billion 0.10% 83% Only for 8 counties with data OVERALL TOTAL RESULTS Low Scenario 125 billion 1.8% Medium Scenario 143 billion 2.1% High Scenario 173 billion 2.6% Total potential (high scenario) = 43% of total county budget Actual (all sources, average FY16-FY17) 35 billion 0.50% (FY16/17) Actual cess collections FY17 1.2 billion 0.02% Total County Budgets 399 billion 5.7% Legal and policy review This study has identified a wide range of revenue streams from which Counties are currently collecting revenue. Across the board, there are at least 100 reported streams, and in some counties, several hundred separate fees and charges. However, this reflects an inconsistency in reporting of revenue streams, which are often the same or similar sources reported under a slightly different name. In fact, the majority of revenue is collected from a handful of sources and the most commonly reported sources of OSR are as follows:  Single Business Permits.  Land and property rates.  Trade/building permits.  Parking fees.  Advertisement and billboard fees.  Market fees.  Liquor licenses.  Lease rents.  Cess.  Hospital / health service charges. Not all revenue streams are suitable for revenue enhancement effort e.g. user charges, which are based on payment of a fee for accessing a service. Health services, for example, should not be targeted for revenue enhancement per se, in case they make important medical care inaccessible. The key policy findings are that (a) is that there is a clear disconnect between revenue collection and policy objectives. The justification and design of taxes, fees and charges at county level is not clear and in most cases was ‘inherited’ from Local Authorities without adequate review and refinement; and (b) counties should focus revenue improvement efforts on a few sources that have a clear policy rationale, the greatest revenue potential and are cost-effective to collect. The legal review found overall, that counties cannot rely on transitional provisions in the Constitution 2010, but need their own legislation. However, this creates inconsistencies and a proliferation of rates, bases and approaches to administration of revenues. There is a need to consider how to guide counties to establish some common frameworks, to ensure consistency but allow discretion to local contexts, as appropriate. County Finance Acts are also being used as omnibus laws to impose all fees and charges. This is not sufficient to provide adequate regulatory functions and collection procedures, which require separate, substantive legislation. Recommendations. 6 Key recommendations from the study include: General Legal and Policy Recommendations  Counties should create a County (Taxes, Fees and Charges) Act that states in one Act all the revenue streams (authorised by legislation) and specifies the relevant tax rate, fee or charge.  Focus revenue raising ‘tax’ on few, coherent sources that have grea test potential, have a clear policy rationale and are most cost-effective to administer.  Ensure local taxes have a clear policy rationale in terms of principles of taxation, market failure, equity/social services provision, revenue-generating, development and regulation. Policy objectives and features should be clearly communicated, such as definitions and communication of base, rates, who is liable, when to pay and how to pay.  For user-charges, develop pricing policies for cost-recovery of services, which identifies and justifies which services will be subsidised, those selected for partial cost recovery, any for which market pricing will apply, and on what basis the cost will be applied.  Strengthen monitoring and analysis of costs.  Policy changes to be informed by impact analysis, including costs and benefits of new policy measures and economic and welfare impacts on users/taxpayers. Specific Recommendations by Revenue Source: Property rates:  Counties to focus most effort on this tax, which represents by far the most revenue potential;  Consider establishing a national Act for property rates (to replace the Rating Act), with Counties setting their own rates, bands and discounts in relation to their fiscal objectives.  Simplify valuation methods and provide for regular updates.  Review and apply a more consistent base.  Rates to be reviewed and adjusted with re-valuation.  Amend the legal framework and procedures for implementing a shift in collection of the tax from owner to occupier.  Reinstate contribution in lieu of rates (CILOR) with clear methodology and process for government payment/transfer.  Strengthen compliance and recovery methods, such as: Compliance certificates; interruption of services, especially electricity; recovery from tenants or beneficial occupant; early payment discounts. Entertainment tax  Consider enactment of new Act with wider tax base consistent with the functional responsibilities under Constitution Fourth Schedule Part 2, paragraph 4.  Consider combining gaming and betting licensing of premises with higher SBP license fee for ease of administration and to capture social cost of gambling addiction.  Fees for entertainment venues are also a form of regulation of business and therefore could be considered as part of SBP, with higher rates for larger venues to reflect the public safety risk and cost of regulation. Single Business Permits (Trade Licences):  Consider national guidelines for the relative, simplified structure of fees, based on size (e.g. employees or turnover).  Variation of fees outside simplified framework to have a clearer rationale, such as the regulation of specific sectors.  County legislation should be enacted to establish a clear licensing framework that is consistent with the principle of a Small Business Permit (SBP). Cesses:  Cess fees should not be collected without clear legislative authority enacted by the County Assembly.  Replace cesses due to high economic burden, double taxation and risk of excessive cost crossing borders etc. o Agriculture cess to be replaced with easily with property tax revenue (or a category of property rates, agricultural land tax) 7 o Quarrying ‘Cess’ to be handled under SBP as special category or replaced with environmental levy. Parking Fees and Market Fees:  Parking fees: Counties to review rates in line with benchmarks, location, peak periods and zoning of areas to manage traffic flow etc. and ensure cost of provision and maintenance is at least covered by revenue.  Market fees: Counties to clarify the definition and objectives of market fees e.g. is it a type of cess on produce, an access fee to market space and facilities, or rental charge for use of government property (market stall). Administration, Compliance and Enforcement:  Counties should enact legislation to set out compliance obligations and powers in a County (Revenue Administration) Act, and the legislation can be based on the existing model, reviewed and updated through the Intergovernmental Relations mechanisms.  Counties should engage in information sharing arrangements with agencies whose data can contribute to ensure the integrity of the County OSR tax base.  Strengthen taxpayer/user awareness and support.  Establish risk management approaches in revenue administration.  Investment in staff and skills.  Review and, if necessary, simplify IT systems and databases to improve data accuracy and integrity.  Strengthen performance monitoring of revenue administration. Revenue Management:  Greater automation of revenue management systems and strengthening of IT connectivity.  Establish better integration of revenue management systems with IFMS reporting.  Strengthen forecasting methods.  Improve transparency of forecasts by publishing forecast assumptions.  Undertake regular performance review of forecasting and outturn revenue performance. Conclusions and Next steps The objectives have been met as follows:  Counties’ current local revenue base and potential has been mapped out in Section 3, to the extent possible using available data;  These estimates and the tax bases provide a framework that will support counties to make more credible projections of future revenues from 6 key revenue sources. Counties (or national government) will need to establish comprehensive databases for these (and potentially other) revenue sources in order to improve county forecasting quality; and  The potential estimation exercise (and resulting data/modelling sheets provided) establishes a framework for monitoring improvements – future revenue collections can be assessed against potential (i.e. how much of the gap is reducing over time, as in Section 3.1). It will be critical for counties to improve data availability and quality, including through more systematic and consistent reporting of collections by revenue source as well as information on actual revenue bases, as the basis for any improvement monitoring. Going forward, the draft policy and bill on county revenue is under discussion and will help frame discussions between National Treasury, CRA and the Counties, among others to strengthen and streamline the legal and policy frameworks in each county. Counties will need to focus efforts in on the key sources highlighted, which may involve removing or replacing some existing revenues, while maintaining a lighter touch approach on others. We envisage that further technical assistance will be needed to help counties to address legal gaps and inconsistencies, administrative challenges and policy design and analysis. 8 Introduction and Methodology This is the final report produced by Adam Smith International for the Own-Source Revenue Potential and Tax Gap Study of Kenya’s county Governments. This study has been commissioned by The World Bank on behalf of the Kenya National Treasury and is being undertaken by a team from Adam Smith International between February and July 2018. The team comprises: Nick Spyropoulos, an international tax economist and forecasting expert; Graeme Keay, an international tax legislative expert with extensive experience across the region; Hazel Granger, a tax policy and PFM specialist; Desmond Boi, an experienced Kenyan statistician; Johannes Wolff, an experienced economist, and Iain Nelson, an ASI Project Manager. Our understanding of the context Kenya has a positive economic outlook, achieving an average growth rate of 5-6% in recent years. Nonetheless, there has been an increase in the fiscal deficit (to 8.8 percent in 2016/17) 2, partly due to domestic revenue shortfalls and spending pressures relating to a drought and Presidential elections. As a result, Kenyan Authorities have committed to a number of corrective actions, including ways to broaden the tax base, including strengthening revenue performance at the county level3. Local level own-source revenue (OSR) powers in Kenya had traditionally been assigned to Local Authorities (post- independence) through property rates and business licenses. However, following centralisation reforms in the 1980s and weak enforcement of property rates, OSR fell short and Local Authorities fell into debt and faced difficulties financing expenditure commitments including salaries. Since devolving powers to the 47 counties through the revised Constitution in 2010 (and subsequent county elections in 2013), county governments inherited the range of revenue sources available to the former LAs, and several of the staff and systems formerly employed in revenue collection functions. However, counties also therefore inherited significant debts, inefficiencies and inadequate administrative practices 4. An initial expansion of revenues that continued after devolution, now appears to be slowing down or even reversing. OSR contributes only up to 12-13%5 of the total financing of County Governments, with increasing dependency on transfers from the National Government. This has raised concerns that own-source revenues are not commensurate with the scale, growth and nature of the expanding economic activity at the County level and with the expanding value of the tax/fee base. A new draft policy and County Revenue Bill (2018) therefore aim to broaden the County revenue base, to strengthen administrative capacity of counties to raise own revenues, and to provide a regulating framework for county imposition and variation of rates, in line with national tax policy and economic objectives. The Commission on Revenue Allocation also aims to strengthen incentives for counties to enhance own-source revenues through an OSR performance element in the central allocation formula. In order to inform the draft policy and reform of the CRA formula, a better understanding in needed of the potential revenues possible at county level and of the barriers or opportunities available to counties to achieve more of that potential in order to reap the benefits of devolution, to generate efficiencies from consolidation of local government structures and be able to better fulfil their mandates for decentralised service delivery through more sustainable financing. In some cases, investment in the administration of OSR might be needed in the infrastructure, equipment or staff administering the taxes, fees or charges, and therefore, having a better understanding of the revenue potential may help to identify whether such investments are worthwhile. Study Objectives The Terms of Reference for this study outlined three key objectives, namely, to: a) map out counties’ current local revenue base, and the associated tax potential vis -à-vis tax effort, leading to a clear determination of counties’ tax and non-tax revenue potential, a systematic identification of revenue streams which can enable each county to maximize its revenue potential and a comparative assessment of counties’ fiscal capacities; b) bring about more credible projections by counties of future revenue from assigned taxes, fees, levies and charges, leading to improved alignment between budgets and policy priorities; and, 2 IMF Country Report No. 18/83, March 2018 3 IMF Country Report No. 17/25, January 2017 4 Decentralisation in Kenya: the governance of governors , Cheeseman et al., Journal of Modern African Studies, 54, 1 (2016) pp 1-35 5 Draft National Policy to Support Enhancement of County Governments’ Own -Source Revenue, National Treasury and Ministry of Planning, February 2018 9 c) develop a framework for monitoring improvements by counties in terms of OSR performance including efficiency in collection, leading to more objective assessment -- particularly by the Controller of Budget -- of county budgets, specifically focusing on actual OSR vis-à-vis forecasts. This report addresses each of these three objectives, with conclusions and recommendations relating to each, as well as drawing together the overall findings, particularly from Phase III of the study, relating to ‘ Data analysis, modelling and recommendations’. In particular, Phase III is expected to address the following activities and issues: (i) Simulate revenue potential for each county. Among other relevant considerations, the simulations should illustrate: a. how counties’ existing revenue administration structures and systems might be influencing optimization of OSR, including their efficiency and collection costs; and, b. the extent if at all, of under-exploitation of taxes, fees, charges or levies over which counties have discretion. For instance, the simulations should show: which taxes are deliberately not being rolled-out? What are the attenuating practices e.g. on reliefs and exemptions, particularly on core areas such as property rates? Are there any other decentralized revenues streams from which counties are currently not collecting anything? (ii) Review Kenyan counties’ constitutional and legal OSR -raising powers including the discretion to introduce taxes/fees/charges/levies, set rates and boundaries (e.g. floor and ceilings) and to grant reliefs or full exemptions. This review will be undertaken against the experience of other countries in the region as well as internationally. Outputs from this review will contribute towards strengthening the legislative framework for counties’ OSR enhancement, which the Government is currently developing. (iii) Make specific recommendations on how different counties’ should strategize to e nhance OSR collection, based on their unique macroeconomic, fiscal, geographic and urban profiles. The recommendations will form inputs into the Government’s ongoing formulation of an overarching policy framework for county OSR. Therefore, among policy considerations to be explored are: a. whether or not it is recommendable for counties to focus on 20 percent of the revenue streams which bring in 80 percent of OSR; and, b. how the National Government should incentivize counties to enable them enhance OSR collection and improve efficiency. Methodology Legal and policy review methodology This review was based on desk research from secondary sources on Kenya OSR and international practice, as well as selected county visits and responses to a questionnaire circulated to all 47 counties. Our approach to the policy and legal review for this study followed a number of principles, including that it should:  Be strategic, by identifying common issues and challenges across all 47 counties and consider how best to support them to strengthen policy and legislation in a consistent way, balanced with their local discretionary powers.  Add value over and above what is already known and well-documented in the literature and studies on Kenya OSR and the draft national OSR policy and draft Bill.  Link to the revenue potential estimation, in terms of providing a guiding framework for the potential analysis and also draw from the findings of the potential analysis to focus the policy and legal recommendations on what is relevant to Kenya counties.  Recognise that the policy and legal frameworks are interdependent, and are both essential for successful revenue collection.  Consider not just a technocratic approach, but present practical and realistic advice, appropriate to the Kenya context. Based on these principles, our analysis and recommendations by considering policy objectives in terms of the problem(s) that are to be addressed, relevant considerations to inform options for improvement and how the legal framework may need to be amended to support any policy changes, as well as making reference to the administrative processes and capacities that would support the implementation of these measures to enhance revenues. 10 Revenue potential methodology A key objective of this study is to provide estimates of revenue potential by county and by revenue stream. Revenue potential is a term that is often used inconsistently in the literature. For the purposes of this study, we produced estimates of revenue potential that provide answers to the following questions: 1. How much revenue would each county be able to raise in total and from each OSR stream if it operated in line with the best performing county in the country? This is calculated using frontier analysis. 2. How much revenue would each county be able to raise from each OSR stream if it fully utilised the fiscal instruments at its disposal, resolved issues relating to administration and eliminated evasion? This is calculated using the ‘top-down’ approach. Stakeholder Engagement An important consideration highlighted in the ToR was the need for stakeholder engagement, to allow an opportunity for inputs and feedback prior to, during and at the end of the study. As mentioned above, a survey was circulated to all county finance departments to provide an opportunity to contribute to the study, highlight any challenges or good practice and raise any concerns. In addition to this, the team visited 4 counties to get a more detailed picture of how county revenue departments operate and to better understand the local context, particularly on administrative challenges, which was not readily available from secondary sources. Furthermore, consultation meetings were held with a range of stakeholders to provide a range of perspectives, test and validate findings and proposed recommendations. A list of all consultees is provided in Annex 5. 11 Legal and Policy Review This section frames the policy and legal issues surrounding County Own Source revenue (OSR) with key principles from international practice and theory, as a basis for comparison with experience from the current situation in Kenya and key findings from our analysis of the County policy and legal framework and practice. We outline general or overarching policy and legal issues, followed by analysis of each key revenue source individually, drawing on experience form relevant international examples. Possible solutions and options for enhancement of county OSR are outlined in the final sub- section. Approach and Principles for the Review This review was based on desk research from secondary sources on Kenya OSR and international practice, as well as selected county visits and responses to a questionnaire circulated to all 47 counties. A full report of the questionnaire responses is provided in Annex 1. Our approach to the policy and legal review for this study followed a number of principles, including that it should: Be strategic, by identifying common issues and challenges across all 47 counties and consider how best to support them to strengthen policy and legislation in a consistent way, balanced with their local discretionary powers; Add value over and above what is already known and well-documented in the literature and studies on Kenya OSR and the draft national OSR policy and draft Bill; Link to the revenue potential estimation, in terms of providing a guiding framework for the potential analysis and also draw from the findings of the potential analysis to focus the policy and legal recommendations on what is relevant to Kenya counties; Recognise that the policy and legal frameworks are interdependent, and are both essential for successful revenue collection; and Consider not just a technocratic approach, but present practical and realistic advice, appropriate to the Kenya context. Policy and legislation are closely interlinked and, ideally, form part of a process that underpins and enables the collection and assessment of taxes and fees. Typically, the process would involve the following steps: a) Analysis and understanding of the ‘problem’ which the policy is attempting to address (revenue raising, regulatory controls, behaviour change); b) Through the analysis and consultation, the policy design informs proposals for changes to various policy features, such as the revenue base, who is liable for the payment, the rate or fee/charge, the frequency and date of payment and method of payment; c) Those proposals form the basis for legal drafting, in the form of amendments or new laws and regulations. These legal amendments are submitted to Parliament or the County Assembly for enactment and gazetting before they become the formal legal basis for revenue collection; d) Changes to the law are then implemented through administrative procedures and systems, which may need to be adjusted with new policy changes. New procedures need to be communicated to taxpayers, perhaps providing guidance and sensitisation through appropriate media of communication; and e) Finally, these changes should be assessed, monitored and evaluated in terms of their revenue and wider economic impacts, in order to determine whether they met the intended objectives. This analysis informs any further amendments or new policy design in future. Based on these principles, our analysis and recommendations reflect the above process, by considering policy objectives in terms of the problem(s) that are to be addressed, relevant considerations to inform options for improvement and how the legal framework may need to be amended to support any policy changes, as well as making reference to the administrative processes and capacities that would support the implementation of these measures to enhance revenues. OSR Policy Principles – Role and Assignment of Local Tax and Non-Tax The role of revenue policy in enhancing revenues can either support the achievement of existing potential, by making the tax work better in practice or by setting a wider base of economic activity from which to raise revenue. Making existing taxes work better could include fixing known problems that are limiting administrative efficiency or causing weaknesses that create opportunities for avoidance., Expanding the potential through broadening of the tax base could include removing exemptions or re-defining the base on a broader value or range of taxpayers), by adjusting rates or introducing new sources of revenue from an ‘untapped’ economic activity. It is worth noting that adjusting rates upwards 12 does not always yield additional revenue overall, particularly if it places a heavy burden on taxpayers that they cannot pay, or if there are significant behavioural changes as a result e.g. moving away from an area, consuming different (cheaper) alternative services or selling property in order to avoid the tax. Lower, simpler tax rates can often improve overall compliance, meaning that the burden of tax can be spread across a wider number of taxpayers while improving total collections. In order to reap efficiency gains from decentralisation, it is usual to allow local discretion over revenue source as well as what they are spent on. The rationale for decentralisation of service delivery and assignment of revenue sources between central and local government stems from a traditional theory that decentralised services offer a potential efficiency gain from allowing local discretion over the provision of goods and services to meet local needs. By accessing local goods and services (through consumption, payment of a fee or otherwise e.g. local voting), the local population can reveal their preferences, thus creating a form of local market for those goods and services, which would otherwise be difficult to observe at national level. It provides information on how users value services as well as a mechanism for rationing what might otherwise become over-consumed services if provided freely at the point of consumption. The assumption being that well-functioning markets provide the most efficient allocation of resources by matching supply with demand6. For this to be effective, the costs and benefits of local services need to be internalised to that geographic area. Additionally, decentralisation tends to support greater democratic accountability, with greater participation in decision- making helping to build a stronger ‘social contract’ with citizens and an opportunity to improve the quality of service delivery. The appropriate assignment of powers to levy taxes between central and local government, usually considers factors such as distributional effects, stabilisation policy and the efficient allocation of goods and services. In principle, policy that affects distribution of wealth should be undertaken at national level, since taxpayers bearing the burden of distributional taxes might move out of the locality and therefore make the tax base unviable or inefficient. Stabilisation policy instruments that aim to control aggregate demand and inflation are ineffective at local level due to spillover effects outside the locality and lack of policy instruments, such as interest rates (unless federal states have central banks, for example). Therefore, the main function of decentralised policy would be allocation of goods and services. This principle suggests that local tax policy should therefore be limited to adjusting rates and not the base itself, since the design of the tax base has distributive effects e.g. if certain groups are exempt. It is also argued that variation of tax bases across geographic areas reduced transparency and accountability. Financing for decentralised government aims to provide some revenue sources (tax or user charges) over which the local government has some control, combined with central government grants and/or borrowing options. A key guiding principle for the selection of revenue sources is that the burden of the revenue stream should not be possible to shift outside the locality and/or that the tax base is limited in mobility. In choosing revenue sources and the assignment of revenues at the local level, international practice suggests universally accepted principles of taxation, as well as several ideal ‘qualities’ that a good local tax should possess, including 7: Broad based: revenue raised at rates low enough to be acceptable in terms of burden Buoyant: automatic growth in annual revenue with growth in the base or periodic increase in the tax rate Stable revenue: for accurate prediction and planning Internalised: borne by local population to which benefits of the services provided accrue, difficult to avoid by moving across borders Neutral: low influence on behaviour/decisions of consumers and producers Autonomous: amenable to local administration Developmental: Connected to benefits, incentivises development, synergies with local investment spending Acceptable: understood and accepted by taxpayers (role of benchmarks, historic, industry or international standards) Ease of Administration: feasibility and efficiency of collection Fairness and equity: ability to pay, shared burden/not regressive, applied to all liable Land and property taxes are widely accepted to fulfil these criteria, as it is immobile, easy to observe and difficult to avoid. Nonetheless, some countries advocate the use of local or regional income taxes, although they are based on mobile factors. In principle, user charges are also appropriate where it is possible to charge a fee, since the idea is to provide financing for local service delivery and for the charges to reflect the cost of provision and/or capital investment. 6 Tax Policy Handbook, Shome et al. (1990) 7 Bahl, Wallace, & Cyan, 2008 and Norregaard in Shome et al (1995) Tax policy Handbook. 13 Not all of these potential sources are relevant for revenue enhancement. Only general taxes are expected to raise revenue for general spending purposes. Direct user charges are, in principle, expected to cover the costs of provision, with varying degrees of subsidy, depending on the policy objective and equity considerations (users’ ability to pay). Incentivising counties to collect more revenue from hospital charges, for example, could have negative consequences on the health of the local community if higher charges make some health services inaccessible or more expensive than users are willing to pay. Licensing fees should, ideally, only cover the cost of administration or regulatory service (e.g. cost of printing a certificate or carrying out an inspection), or at no fee, if the objective is to encourage businesses to register formally in order to collect information on local business activities to use for planning purposes. However, they can sometimes take on revenue-raising functions in practice, with fees being set above cost, particularly if there is a perceived ability to pay and the measure could be administratively more efficient than alternative taxes, or if there is a policy intention to control or reduce an activity with negative social costs (e.g. polluting activities). Direct user charges would normally be provided at a subsidised or cost-basis and therefore revenue potential is only for cost recovery purposes and not for enhancing revenues to pay for other expenditures . That is, unless providing a service in direct competition with private sector alternatives and can be charged at the market rate (margin cost plus profit mark-up). Similarly, fines and penalties should ideally be minimised, since they are incurred for violations of rules, which is not to be encouraged. County shares of royalties could enhance general revenues, but other than the initial negotiation of shares with central government, counties have little control over the receipts of these sources. Therefore, only genuine taxes are typically used for revenue-raising, unless case can be made for user charges or regulatory fees to have revenue-raising potential. In principle, setting of tax rates, fees or charges may involve a number of considerations. For taxes, as outlined above, considerations would include: ability to pay (equity); the ‘benefit’ principle (fee to be commensurate with benefit received); simplicity (e.g. uniform rates for similar tax base); efficiency/neutrality (that individuals’ or business behaviour or decisions are, on the whole, unaffected by the tax); acceptable and administration is efficient. For the overall efficiency of a tax system, voluntary compliance is encouraged as far as possible to focus limited administrative resources on the most significant revenue risks. Therefore, where possible, voluntary compliance is likely to be higher when using low, uniform rates that are easily understood, transparent and have payment or filing procedures that are as easy as possible for taxpayers to follow. For cost recovery through user charges for providing local goods and services, pricing of fees and charges and setting a pricing policy may involve a number of considerations such as user ability to pay. This is particularly important if the fee is mandatory compared to a voluntary charge for accessing a service. Other considerations include county service delivery obligations, cost recovery principles, benchmarking of similar services or charges, expected use of the service and how it might be affected by changes to the fee and compliance or alignment with other County and National Government policy. Key factors to be considered in setting cost recovery principles and some example pricing structures are outlined below 8. 1. Subsidy The degree of ‘excludability’ matters in this case. That is, if it is not possible to exclude someone from accessing the service, it may not be feasible to charge e.g. a sports event with no fences or gate or a public fireworks display. Pure ‘public’ goods are typically under-provided in private market situations due to the problem of free-riding or collective action problem. If it is not possible to exclude someone from the service, then some are likely to allow others to pay and then benefit from the service themselves. This problem is usually solved through general taxation or regulation to coordinate the provision fairly and ensure adequate levels are provided. Some services are provided directly for equity reasons, based on income levels, if there are wider social benefits in consumption of the service (e.g. public health campaign, medical check-ups, education, installing a smoke alarm), but an access fee or price may deter some from consuming them (particularly if they are not affordable), then they can be provided at a subsidised or free rate through general taxation (see also positive externalities, below). So, there is need to consider the income profile of users or burdens it places on business. 2. Partial or full cost-recovery (marginal or average cost) Estimating the costs of local service provision is based on the financial cost, that is, the initial fixed costs of construction and then any pursuant operating and maintenance costs. Estimating the marginal (unit) cost of provision might include a range of overheads, such as building rent, equipment, utilities, staff and payroll, training, transport, advertising, printing, management or other associated costs. With capital projects, the presence of high fixed costs (economies of scale or ‘sunk’ costs) suggests that prices need to take into consideration the overall average cost, even if marginal costs decline 8 R.M. Bird (2003), User Charges in Local Government Finance 14 as output (number of users) increase. In this case, average cost of provision exceeds marginal cost and includes the total fixed costs divided by the number of users. How costs are then attributed to the service provision or to any one additional user may vary according to the factors outlined above, e.g. if the marginal cost or average cost represents the cost more appropriately, for example. A further factor to consider is the economic or “opportunity” cost, which represents the highest value of the resource in its next best alternative use e.g. land used for a park instead of for residential housing. Time and distance can also affect marginal costs e.g. road use at peak times causes congestion, which incurs higher private, economic and social costs. 3. Market price (marginal cost & profit mark-up) If a County is providing a service that essentially competes with the private sector, then the good or service can be provided at market price. This could be based on the price of other similar services. 4. ‘Internalising’ externalities The presence of marginal social costs or benefits need to be factored into the price to either deter negative externalities or encourage positive externalities, which markets would normally fail to capture. In this case the marginal private benefit or cost is actually less than the marginal social benefit or cost. In a County Government context, this could include open access problems in which resources could be rapidly depleted, or polluted, which would not be in the greater interest of the community, e.g. Quarrying, over-grazing public land, littering or dumping, traffic congestion or damaging the natural environment of a park or natural habitat. Typical solutions include restricting access or usage through regulation, physical barriers, quotas, licenses, fees, or issuing fines and penalties. The degree of ‘public-ness’ (or rivalry) of the good matters; the more private in nature is the good or service, the more desirable it is to charge to access it, as consumption by one person has a cost to others, by using up the resource. In these cases, cost recovery pricing is less appropriate because the rationale for levying a fee or charge is to do with rationing a valuable resource or minimising negative impacts or, alternatively, encouraging investment in a social ‘good’ when it would otherwise be under-provided e.g. wildlife conservation or universal education services. In this case, some estimation of the benefit derived from the ‘service’ or action is needed, using willingness to pay (e.g. using survey questions), revealed preference (observing usage patterns), or contingent valuation techniques (asking about willingness to accept compensation to give it up). Table 1: Example pricing structures for non-tax fees and charges (cost-recovery) Charge/Fee type Pricing method Considerations/applicability of method example Prescription drug fee Marginal cost: marginal Below market price. Ideally, should capture opportunity cost financial cost (e.g. wholesale and adjust for positive/negative externalities (see below). price) This method is not practical if there is no market alternative or goods are non-excludable Toll road fee Single average cost: Practical for financing capital project, acceptable, but Construction cost and inefficient and inequitable – does not take into consideration maintenance divided by relative wider costs of usage e.g. at busy times or by large number of expected users trucks over recovery period Road license fee + Average incremental cost or Useful for financing capital project and/or infrastructure congestion charge multi-part tariff: basic access maintenance, public transport etc. Charge can be varied charge to cover fixed costs + according to scale of damage (e.g. size of truck, or time of Or metered usage fee day used) electricity/water Park/tourism fees Variable block: unit cost Consider incentives or disincentives this creates e.g. (one-off or annual recovery that varies with encourages long-term usage which can improve stability of membership) usage (e.g. declines with revenue, but may want to restrict over-consumption at busy higher frequency usage) periods to prevent damage or congestion e.g. charge higher at peak times Source: adapted from Bird (2003) 15 Mapping of current legal and policy framework This study maps the current and proposed framework for county OSR around 3 pillars: (i) Revenue streams – powers to tax (including waivers and adjustment of rates); (ii) Revenue administration and compliance – powers to collect, administer and enforce; and (iii) Revenue management –duties to report and account (linked to public financial management procedures and the central government revenue allocation formula for county inter-governmental fiscal transfers) This includes discussion of: institutions (such as assemblies, executives, agencies and committees); laws (Acts, regulations, notices); and policy and legal processes (and their intended interaction). Kenya County OSR Revenue Streams Typical revenue sources and Kenya’s current practice International experience of local revenue sources is mixed. Anglo-Saxon countries tend to focus on property taxes almost exclusively, while Scandinavian countries use a mixture of revenue sources, including income taxes (Norregaard, 1995). Typical local tax instruments across Anglophone Africa include, for example 9: Property taxes (usually urban or commercial); Rental income from Local Authority property; Per capita tax (usually individuals over 18); Tax on activities (e.g. trading licenses); Charge on utilities e.g. water/electricity service; Fees for public facilities e.g. markets; Livestock or grazing fees; and Royalties on natural resource use. From a review of County Finance Acts, published notices and available reports, our study identified a wide range of revenue streams from which Counties are currently collecting revenue. Across the board, there are at least 100 reported streams, and in some counties, several hundred separate fees and charges. However, this reflects an inconsistency in reporting of revenue streams, which are often the same or similar sources reported under a slightly different name. In fact, the majority of revenue is collected from a handful of sources. The most commonly reported sources of OSR are as follows: Land and property rates Parking fees Market fees Lease rents Hospital / health service charges Single Business Permits Trade/building permits Advertisement and billboard fees Liquor licenses Cess Due to inconsistencies in the labelling and reporting of revenue streams, it is difficult to report accurately the number of counties making use of each of these revenue streams or the value of revenues by source. Nonetheless the above list (in no particular order) has been selected based on the general frequency observed of counties reporting these types of revenues and those documented in county Finance Acts. While there are many documented revenue sources, most or all can be categorised in terms of policy objectives, as follows: 9 Fjeldstad and Heggstad (2012) “Local Government Revenue Mobilisation in Anglophone Africa.” 16 a) Taxes: a compulsory contribution levied by a Government body on income or the value of goods, service or transactions for general revenue purposes and not connected to a particular benefit e.g. property rates and entertainment tax; b) User charges: governmental charges that are incurred in exchange for a benefit, which should reasonably approximate the payer's fair share of the costs incurred by the government in providing the benefit 10 e.g. hospital fees, market fees/rent, parking fees, lease rents and wildlife park entry fees; c) Regulatory licensing or user charges with regulatory ‘elements’: fees or charges connected with licensing and regulatory activities of government e.g. Single Business Permits, advertising, building permits, liquor licensing; d) Fines and penalties: incurred as a penalty for violation of rules and procedures set out in the legal and regulatory framework; e) Share of royalties: government’s share of economic ‘rent’ collected from activities e.g. from Mining Act 2016; f) Other income (non-tax): interest on loans, tender document sales, asset disposal; Ad-hoc In some cases, the type of revenue is defined by its objective and use of revenues, rather than the label or name it is given (e.g. when a fee is actually a tax). For example, licensing or user fees may receive some sort of benefit, but are often not in any way related to cost of provision of a service, and may or may not be voluntary. In this way, business license fees or road user charges can appear a lot like taxes, particularly if used mainly for general revenue raising and not earmarked for any specific purpose. Alternatively, some types of cess could appear as ‘regulatory’ if the objective of charging a fee for the right to undertake an activity (e.g. quarrying or sand harvesting) is to control or even reduce the activity, or to raise revenue to pay for mitigating measures, such as to reduce environmental damage. In some cases, however, cess takes on the form of a tax, where it is mandatory and contributes to general spending and is not earmarked for a specific purpose, or to general a specific benefit for the taxpayer/user. Some counties receive other (non-tax) revenue from the rent of county-owned assets, such as residential property or equipment. In this case, there is limited scope to enhance revenue potential without investing in new assets, which is not generally the role of a county government. In fact, in international practice the investment and holding of residential housing stock is usually to provide housing at sub-market rents as social safety nets for low income households who cannot afford private rented accommodation. Nonetheless, we did not find this type of policy rationale from our (albeit limited) fieldwork. Where there is potential to enhance revenue from the rental value of assets, it would be driven by strengthening the management and maintenance of assets and ensuring that the pricing policy of assets is appropriate to the market for that asset e.g. if it is intended to be equivalent to market rates, then the county would need a flexible policy and legislative framework that would enable regular price adjustments in line with market prices. Legal Framework: Kenya’s County powers to raise taxes and fees The framework for powers to tax (and charge fees and levies) is based on the Constitution 2010. The Constitution provides, in Chapter 11, for devolved government. In particular, it establishes in Part 1, Art. 175(b), the fundamental principle that county governments are to have reliable sources of revenue to enable them to govern and deliver services effectively. 175. County governments established under this Constitution shall reflect the following principles: (a) County governments shall be based on democratic principles and the separation of powers. (b) County governments shall have reliable sources of revenue to enable them to govern and deliver services effectively. Chapter 12 elaborates the public finance framework for counties and the arrangements between counties and the national government. The principles are set out in Part 1. In terms of the context of county own source revenue, Art. 201(2)(b) provides several principles, notably that the burden of taxation shall be shared fairly [Art. 201(2)(b)(i)], and that revenue raised nationally shall be shared equitably among national and county governments [Art. 201(2)(b)(ii)]. Counties are empowered to raise taxes under Art. 209(3). Part 3 of the Constitution sets out revenue raising powers of national and county governments (and, pursuant to Art. 175(b), the latter requires to be reliable sources of revenue. Part 3 reserves the power to raise income tax, values tax, customs duties (and other duties on imports) and excise duty to the national government [Art. 209(1)]. The national government may be empowered to raise any other tax by Act of Parliament, except the two express taxes that are reserved to county governments [Art. 209(2)]. County powers under 10 Tax Foundation Background Paper No. 63, “How Is the Money Used? Federal and State Cases Distinguishing Taxes and Fees” by Joseph Henchman 17 Art. 209(3) are specified as (a) property taxes, (b) entertainment taxes and (c) any other tax that is authorised by an Act of Parliament. The terms “property taxes” and “entertainment taxes” are not defined. The term “tax” is not defined, and would therefore be implied to have its ordinary meaning, namely an imposition by a state (at any level) for the purpose of generating revenue on a defined class of income, transaction, activity or property. The term “property” is defined in Art. 260 (by expressly including) various forms of rights in land and buildings, goods and personal property, intellectual property, and money, choses in action and negotiable instruments. Whilst the term may commonly be read narrowly so as to mean land and buildings, the use of “property”, and its constitutional definition, widens the scope of the tax raising power to more than just land and buildings. The power to tax property in the form of land and buildings is circumscribed, however, by Art. 67 which establishes the National Land Commission. One of the express functions of the NLC is set out in Art. 67(2)(g) as the power to assess tax on land and premiums on immovable property in any area designated by law. The reserved power of county governments to impose property taxes (in respect of land and buildings) must therefore be read in conjunction with the NLC’s power to assess tax. The NLC’s power to assess is limited to those areas designated by law (i.e. any other constitutional provision of an Act of Parliament). In addition to the power to impose tax, county governments (as well as the national government) may impose charges for services provided [Art. 209(4)]. Any tax or revenue raising power needs to be imposed or exercised in a way that is consistent with single market principles. Any tax or revenue raising power of county government is subject to the principles in Art. 209(5) that ensure there is a national market in Kenya. Art 209(5) provides that such county powers shall not be exercised in a way that prejudices national economic policies, economic activities across county boundaries or the national mobility of goods, services, capital or labour. No tax or licensing fee may be imposed (or waived or varied) except as provided by legislation [Art. 210(1)]. This Article has narrower application than the preceding articles as it refers expressly to “licensing fee” as opposed to the preceding articles in which charges for services are provided. Whilst a licensing fee may be classified as a service charge, a service charge cannot be described as being the equivalent of a licensing fee. For example, a license to operate a business compared to a charge for medical services. Schedule 4 of the Constitution also sets out the division of functional responsibilities of the national and county governments, pursuant to Arts. 185(2), 186(1) and 187(2). Under Part 2 of this Schedule, there are 14 categories of functions and powers that are distributed to the county governments. Category 14 is a general function of co-ordination with the national government. Categories 1 to 13 are therefore those specific areas where county governments may exercise their powers under Arts. 209(3) and (4). These functional responsibilities and powers. The 13 categories are:  Agriculture.  County Health.  Air pollution, noise pollution, other public nuisances and outdoor advertising.  Cultural activities, public entertainment and public amenities.  County transport.  Animal control and welfare.  Trade development and regulation.  County planning and development.  Pre-primary education, village polytechnics, homecraft centres and childcare facilities.  Natural resources and environmental conservation (implementation of national government policies).  County public works and services.  Firefighting services and disaster management.  Control of drugs and pornography. These categories need to be read in conjunction with the Part 1, which sets out the national government’s powers and functions. Where there is uncertainty as to the distribution of the functional responsibilities and powers (including the power to impose a fee or charge), the Intergovernmental Relations Act 2012 provides a mechanism to deal with intergovernmental relations for the purpose of avoiding disputes between different levels of government, in particular the facilitation of devolution under Art. 175 (revenue sources). For example, the Intergovernmental Relations Technical Committee (IGTRC) is a “body formed by an act of parliament to establish a framework for consultation and cooperation between the National and County Governments and amongst county governments”. 18 Each tax and fee or charge requires to be established by legislation. This may be an Act of Parliament or an Act of the County Assembly. (i) Act of Parliament (pre-2010): If an Act of Parliament, that relates to the functional responsibilities of the counties, was in force at the time of the coming effect of the Constitution 2010, the Act requires to be read subject to the transitional provisions in the Sixth Schedule, Part 2, and in particular paragraph 7(2). This paragraph applies to an existing law that assigns a responsibility to a State organ, and that assignment of responsibility is changed by the Constitution to another State organ. In this case, the Act is to be read on the basis that the transfer provided by the Constitution prevails. The effect of this transitional provision means that a pre-existing Act that provided for a tax to be imposed or a fee to be charged by an organ of the national government, and where the Constitution has assigned such functional responsibility to the county governments, is to be read as the county organ having the same power. In addition, the Public Fees Act 1961 (Cap.256) provides that a Minister may make regulations prescribing the fees to be levied for licences, permits and other acts, matters and things issued or performed by or in any public office. This empowers the Minister with responsibility for the Act to determine and prescribe the fees. For the purposes of the Sixth Schedule of the Constitution, the equivalent of a Minister needs to be determined. (ii) Act of Parliament (2010 onwards): If an Act of Parliament passed after the coming into effect of the Constitution 2010 grants a power to county governments to impose a tax or impose a fee, that Act is to be read as the authority for the county government to impose and collect the revenue. (iii) Act of County Assembly (2010 onwards): Under Art. 185(2) of the Constitution, County Assemblies have legislative competence to enact legislation in relation to the functional responsibilities assigned under the Fourth Schedule. Thus a County Assembly may legislate for a fee or charge in relation to a functional responsibility. A number of Counties have done so [examples include the Machakos County Trade Licence Act 2014 (Act 9 of 2014, Machakos County Gazette Supplement No. 25, 29th December 214, page 339). These Acts are published as part of the County Gazette, by the Government Printer, Nairobi. There is also a constitutional requirement to propose and enact an Appropriation Act for each financial year [Art. 224]. An Appropriation Act is usually accompanied by a Finance Act. The purpose of the Finance Act is to enact the new and amending fiscal measures arising from the budget in respect of which the appropriation is authorised. The amendments in a Finance Act are of substantive laws that create tax obligations and, in the case of counties, impose fees and charges. Legal Framework for Revenue administration and Compliance The Constitution makes no express provision for powers to collect, administer and enforce taxes or other forms of revenue. National taxes (reserved to the national government under the Constitution) are collected, administered and enforced in accordance with the Tax Procedures Act 2015. There is no equivalent law (setting out all the administrative powers for enforcement and collection) for the taxes, fees and charges reserved for Counties. Historically, these powers are located within the Act that imposes the relevant tax or fee/charge. This results in differing administration frameworks, dependent of the revenue stream and its enabling legislation. The Public Finance Management Act 2012 establishes receivers of county revenue and empowers appointment of collectors of county revenue, including the KRA in accordance with specified criteria. The actual powers of collection are not (nor should they be) set out in the PFM Act 2012. County Revenue from Central Government (Revenue Allocation Formula) Equalisation measures can range from simple measures such as per capita grants to more complex formulas for grant allocation, based on statistical determination of expenditure needs and the grant revenue interaction with taxation. Expenditure is also controlled indirectly e.g. by limiting borrowing and limiting taxation e.g. setting rate ceilings. Central government control over local expenditure can be required for a number of reasons, such as the need to control domestic demand in the economy, given the scale of consumption at local level. Some argue that the more discretion a local government has over expenditure, the more participatory decision-making is within the local community and expenditure will be self-regulating, by supplying services only in accordance with demand. Nonetheless, others argue that the political nature of local government and the relatively short political cycle means that self-regulation is not effective. As a result, there is a wide range of policies and financing mechanisms across countries for local government finance, typically including a mixture of central conditional grants, discretionary grants and own-source revenue streams. Kenya’s revenue allocation formula for county distribution of central revenues includes an equal share across all 47 counties and specific weightings based on various factors, such as population, poverty, land area and development needs. There is also provision for a ‘fiscal effort’ factor, which could capture incentives to counties to enhance OSR performance. This is discussed further in section 2.6. 19 Critique of policy and legal framework and practice General Policy Findings The most significant policy finding from the review was a clear disconnect between revenue streams and policy objectives. In many cases, the basis for the current set of revenue bases, rates and charges is that ‘inherited’ by counties from the previous local authority / municipal laws. Where adjustments have been made under the county system, these have been mostly incremental inflation adjustments, rather than amending any policy features based on new policy objectives or analysis. For example, SBPs were based previously on a policy developed by Ministry of Local Government and the Ministry of Planning under a World Bank study in 2000, which provided a framework of rates across the whole of Kenya for local authority areas. Kiambu County has maintained the differential between rates under this framework, but applied a fixed percentage increment across all license fees to account for inflation and revenue-raising needs. More substantial changes e.g. to adjust revenue bases, such as to introduce a waiver, appear not to be based on any clear rationale and may be mostly politically motivated. For example, one county introduced a waiver from SBPs for disabled traders, when it was unclear why a disabled person operating a business would require any different treatment from a non-disabled business person. There could be a case for waiving the fee in order to encourage disabled businesses to register formally in order for counties to better plan for facilities that might aid people with disabilities, but this rationale was not put forward by the county finance department. Recommendations  Ensure County taxes and charges have a clear policy rationale e.g. in terms of principles of taxation, market failure (public goods, externalities), equity/social services provision, revenue-generating, development, regulation;  Counties should focus revenue enhancement effort on fewer, coherent sources that have a clear policy rationale, have greatest revenue-raising potential and are most cost-effective to administer e.g. Property tax, building permits, SBP, parking, advertising, liquor licensing and relevant user charges; and  For user-charges, develop pricing policies for cost-recovery of services, which identified and justifies which services will be subsidised, those selected for partial cost recovery, any for which market pricing will apply, and on what basis the cost will be applied (e.g. average or marginal cost, or alternative structure). There may be a lack of capacity in the area of policy design and analysis. While it is not necessarily essential to have a revenue policy unit, the fact that most counties responding to the questionnaire did not have one, suggests that capacity is low. Most counties (1 in 10 respondents) reported not to have a clear county revenue generation policy and 60% of counties admitted to having less than adequate policy guides for at least one existing revenue stream. In some cases the tax bases were not well understood by revenue managers and officers, which was particularly challenging for the assessment and collection of property rates, such as the conditions under which occupants as opposed to owners could be held liable for the tax. Some support for policy development has apparently been offered by CRA, although in the long-term this is likely to be outside CRA’s mandate and in any case this support did not appear to have reached all counties at the time of the study. Recommendation  County finance/revenue departments to develop capacity in tax (revenue) policy design and analysis e.g. with technical assistance, if possible. In particular, to build capacity in the following areas: (a) Monitoring and analysis of costs: to assess cost of service provision (average and marginal) and consider links to fees; and (b) impact analysis of policy changes, including costs and benefits of new policy measures and economic and welfare impacts on users/taxpayers. Ideally, revenue impacts of changes in fees and charges will take into account the relative elasticities (responsiveness) of taxpayers/users to changes in the price or tax payable. Policy consultation exercises are not being used to effectively inform policy. A number of counties reported undertaking consultation with stakeholders, which is essential for effective policy making and for establishing acceptance and improving compliance. Nonetheless, due to the lack of clear understanding and link to policy objectives, the policy justification for adjustments to rates or revenue bases is often not communicated and can weaken the purpose and outcome of consultation exercises. In this situation, consultation can simply become a political negotiation with taxpayers without any basis in evidence or principles. This also increases the risk (and opportunity) of political interference in revenue policy and administration. For example, one county introduced an exemption for small traders from market fees in response to complaints during consultation. However, there was no ‘hard’ analysis to inform the decision, based on, for example, a comparison of rates against average earnings and ability to pay, and it was not clear how the county finance department expected to continue to pay for facilities provided at the market, such as water, washrooms and shade. Recommendation 20  Policy objectives and features should be clearly communicated to taxpayers, users and stakeholders, including key policy features e.g. definition of tax base, rates, who is liable, when to pay and how to pay. General Legal Framework Findings Some Counties have used the mechanism of the Finance Act to create an omnibus law that imposes all fees and charges, but without any of the regulatory functions and processes for the licence or service in respect of which fee or charge is imposed. For example: Kiambu County Finance Act, 2016 (Act 5 of 2016), Kiambu County Gazette Supplement No. 19, 21st December 2016, page 1. County Finance Acts are intended to introduce amendments (e.g. of fees and rates, or tax bases) to substantive laws and are not supposed to confer taxes in their own right as a stand-alone law, without the necessary accompanying provisions relating to how a tax or charge will be imposed and collected, for example. Recommendations: (i) County legislation that creates a regulatory duty or obligation, and imposes a licensing fee, should not be set out in a Finance Act, but instead in dedicated, separate County legislation; (ii) Counties should create a County (Taxes, Fees and Charges) Act that states in one Act all the revenue streams (authorised by legislation) and specifies the relevant tax rate, fee or charge (iii) Finance Acts should be reserved for annual amendments to fiscal provisions, arising from the County annual budget submitted and passed under Art. 224, including any necessary amendments to the County (Taxes, Fees and Charges) Act (see above). The intention of the Intergovernmental Relations Technical Committee may be open to challenge as being neither the empowered body (Summit) or exercising a valid power (s.12). For example, item 34 in Part 1 relates to national betting, casinos and other forms of gambling, which is shared with the counties under item 4(a) of Part 2 which covers betting, casinos and other forms of gambling. The functions of the national government and the county governments in relation to betting, casinos and other forms of gambling are purported to be delineated by approval of the Intergovernmental Relations Technical Committee, as published in the Gazette Notice No. 8753. However, it is not clear that the Technical Committee has such a power. The functions of the Committee set out in s.12 of the Intergovernmental relations Act 2012 do expressly state that the Committee may decide on such delineation. It would appear that the Committee could report to the Summit and for the Summit to exercise powers in accordance with s.8 of that Act. The summit has functions under s.8(j) to co-ordinate and harmonise the development of county and national government policies, and to facilitate and co-ordinate the transfer of functions, powers or competencies from and to either level of government. The notice purports to “delineate” powers and not “transfer” powers. The Technical Committee may perform other functions conferred on it. Recommendation: To take stock of the various intergovernmental frameworks under the constitution and recommend any changes to ensure conformity with the Constitution e.g. Provide clarity in the Intergovernmental Relations Act 2012 on mandate of Intergovernmental Relations Technical Committee. Key Findings on Land and Property Rates Weaknesses in the legal framework Property rates is one of the two taxes expressly provided as a revenue stream dedicated to the Counties. The existing law used by the former local councils (Rating Act Cap.267) was first passed in 1963. This was supported by the Rating for Valuation Act Cap.266, first enacted in 1956. Under the Rating Act, Counties (as the local authority) are under a legal duty to impose and collect rates, pay them into a general rate fund and to all discharge their liabilities from that fund (subject to a general reserve fund). Thus, this primary revenue source is required to be used to discharge liabilities. The liability for rates is imposed on the rateable owner, as defined in the Valuation for Rating Act. This is defined narrowly by reference to formal tenure and registration arrangements. This constrains the administration of the Act and the collection of revenue, particularly where there are varying tenure types and traditional tenures for which there is no formal registration of land. The rate may be chosen from one of six rating methods, as prescribed in the Act. Multiple rating methods may be adopted and different rating methods, by area, may be adopted, within prescribed limitations. The rating methods must be approved by the Minister responsible for local government. Site value rates must be applied uniformly across the local authority (Council) area. However, variation is possible, in the form of exemption. This must be approved by the Minister responsible for local government, and may be subject to conditions imposed by the Minister. 21 The amount of rate for both unimproved and improved land is generally limited to 4%. Variation is possible, with the approval of the Minister. For improved site value land, the rate is also subject to a requirement that the product of the rate not exceed 25% of the total amount forecast to be collected for the year. Both site rate and improved rate are calculated and imposed on the valuation set out in the roll. A valuation roll (and supplementary rolls) are required to be prepared and published in accordance with Cap. 266. Recovery of unpaid rates may be made in accordance with the enforcement powers set out in the Rating Act ss.17-20. These are primarily based on seeking a court judgement for payment of unpaid rates. This is slow, uncertain and costly. The enforcement powers do not reflect the range of timely and effective recovery powers that would appropriate for the evolving nature of the property occupation in Kenya, and the limitations of formal judicial recovery. The Counties may continue to use these Acts, on the basis of the interpretation provisions set out in the Sixth Schedule. As referred to above, this is legally possible but undesirable, on the basis of its complexity, and the continuation of out- dated law. Nonetheless, the creation of separate property rating legislation by each County, even if based on a model (e.g. one that sets outs appropriate “best fit” principles and practice to improve administration, and to address registration and valuation capacity challenges), runs a severe risk of divergence, with multiple property rating systems and tax bases developing. Recommendation: In order to provide for a harmonised, updated, national property tax base, consider establishing a national Act for property rates (to replace the Rating Act), with Counties setting their own rates, bands and discounts in relation to their fiscal objectives in light of the economic context of each County. A modern rating legislation would be more appropriate to the current and anticipated land tenure and land use structure in Kenya. This national Act could be created through agreement reached by County Summit and other intergovernmental mechanisms, so as to address political and constitutional issues (noting that this is new constitutional territory, the principles of devolved administration are evolving and not entirely certain, and that legal advice from the Attorney-General would be helpful to guide the process). This Act would expressly reserve to the counties powers in relation to the rates, bands, waivers and discounts for the rates (thus limiting the new Act to establishing a consistent tax base and valuation system(s)). Policy challenges Most, if not all, counties collect some form of land and property taxes, which is highly appropriate, since these are widely accepted to fulfil the desirable criteria for a good local tax (as described above), in particular they provide: Potentially large and stable revenue source; High level of buoyancy: the value and scope of the property base expands with economic growth; A strong, positive connection between local service provision and property revenue, since successful local governments will attract more businesses and households into the area; Ability to pay: it is a progressive tax, if levied on the value of property placing a higher burden on those who are property owners, who are likely to benefit disproportionately Government protection of allocated property rights and local services or infrastructure e.g. roads from which to access the property; Promotion of efficient use of land and economic development: whether levied on vacant or occupied land, property taxes encourage more productive use of land and therefore promote economic development; Supports property-related institutions, such as property registers and deeds of sale, use of courts for resolution of title disputes, town planning (building permits, demolitions, surveys); Compliance: property is easy to observe and therefore relatively difficult to avoid paying tax; Nonetheless, there are a number of potential drawbacks, which often mean that property taxes are under-used or not achieving their potential, including: Administrative challenges in taxing owners, particularly if the owner is overseas or not easily traceable; Administration can be complex and expensive, particularly if based on market value, which is based on a multitude of factors and needs regular updating to remain accurate; Low political acceptability, as opposition from land and property owners can be strong; and Low capacity at local level for revenue administration. Evidence from our review indicated that counties face challenges in all these typical areas of collection of property tax. In particular, the problem of outdated valuation rolls, which affects the majority of counties, especially Nairobi, which still 22 uses valuation rolls from 1982. In addition, there is a complexity of rates, across and within counties, the policy rationale for which is unclear. For example, Machakos County applies different rates in different localities across the county based on old municipal council lines, which was partly a result of the existing valuation roll, which had been carried out under local authorities. The revenue bases also differ between localities, due to differences in the definitions and/or inclusion of freehold and leasehold property. Some counties reported challenges in applying the liability for the tax on the ‘rateable’ owner, which was either not possible or difficult to ascertain when land or property is not formally registered with a title. Some effective recovery methods are already used, such as requiring compliance (or tax clearance) certificates to access services. Some countries use interruption of services to improve recovery from non-compliant taxpayers. Any new power to support collection of property rates would need to be consistent with the existing legal framework. For example, if recovery could be through interruption of electricity services, it would need to be consistent with the legal framework covering the duties and obligations of Kenya Power to supply and maintain electricity to consumers, plus their contractual arrangements. Thus, it could be possible for the County Government to be empowered to serve a statutory notice on Kenya Power that a Kenya Power customer is in default to the County Government for the payment of property rates, and Kenya Power would be accordingly obliged to suspend the service to that customer, in accordance with prescribed criteria. This follows the principles of tax collection through garnishee notices, which is a statutory or court-enforced mechanism for collecting debts directly from a debtor’s creditor. It is a standard d ebt recovery mechanism for tax. Nonetheless, there could be equity concerns and may, in some cases, be an excessive measure. If there is interest to take such a mechanism forward, it would be subject to legal due diligence of the legal framework, as mentioned. Recommendation: Simplify valuation methods and provide for regular updates: e.g. banding and indexing between valuation exercises to add new property and allow for inflation; or Computer Assistance Mass Appraisal (CAMA) e.g. South Africa; Consider testing approaches to strengthen compliance and recovery methods, such as:  Compliance certificates: these may already be in force and are effective.  Interruption of services e.g. electricity.  Encumbrance of estates is already implied, and is effective if the legal probate system is in place.  Early payment discounts, monthly payment to spread cost. On the issue of the tax bases, property taxes are based on either (a) land value (Unimproved Site Value); (b) immovable property on the land (and/or structural improvements); or (c) both (a) and (b) combined (Improved Site Value). Valuation of the structural improvements (buildings/immovable property) is likely to consist mainly of the construction cost (materials, labour etc.) and depreciation, whereas the land value (USV) will be affected by a range of factors, including location, proximity to public investments or infrastructure, population growth and demand. ISV is more common and can be revealed in transaction values, which tend to capture both the factors affecting land values as well as the value derived from buildings, particularly the rental value, for example. International practice is mixed. The tax base for property tax in Tanzania, for example, is the value of immovable properties, compared to Rwanda, Malawi, Zambia and Botswana, in which the value of land and property are isolated and taxed separately. By contrast, the property tax in Uganda is based on both land and property, using a measure of their composite value. Similarly, in Lagos, three different rates on land and property are consolidated into one ‘land use charge’11. From a fairness and equity point of view, taxing only immovable property is less closely linked to ability to pay since it is not the value paid by the owner to purchase the property. Taxing land alone is preferable, since the fixed supply of land means that this kind of tax does not affect investment in improving land and, in fact, it discourages inefficient land use, because landlords are less likely to hold on to under-developed land if they are liable to pay tax on it 12. However, this does not capture the variation in value or ability to pay due to the structures on the land (ISV), which can vary significantly in a central business district between an undeveloped plot and one with a large office building receiving rental income. It is therefore more equitable to tax the ISV. In practice, depending on the tax base, valuation can be assessed on, for example, the rental value, capital value, land value or land area (fee per square metre etc.). Methods of assessing the value (and therefore also of tax liability) range from individual property valuation by registered valuation professionals (labour intensive), to a simpler, cheaper (but less fair) method of using a flat rate per registered property. It may also be administratively easier to tax both land and property together (ISV), if it is difficult to isolate the land value (e.g. if property costs are not available) and transactions data reflects the value of both land and property. It also might be more progressive, in that plots of land with similar values might have an entirely different scale of buildings on it, indicating variation in the owners’ ability to pay. A land -based tax on a poorer 11 Collier at al. (2017), “Land and property taxes for municipal finance” 12 Collier, P., Glaeser, E., Venables, A., Manwaring, P., and Blake, M. (2017) Land and property taxes for municipal finance –version 1. IGC Cities that Work Policy Brief. 23 owner might force them to sell their land, which might require either a composite tax base to be used, or to provide a time- bound exemption in order to allow them time to find alternative accommodation (Collier et al., 2017). Recommendation: Review and apply a more consistent base: consider the shift from ‘unimproved site value’ to ‘improved site value’, which may depend on an assessment of data availability and consistency with the chosen method of valuation. The liability for the tax can fall on either the owner or occupier (in some cases these are one and the same). As discussed above, it is fairer to tax the owner, as the occupier already pays for the benefit of the property through rent. Nonetheless, it can be easier to administer through occupiers, as they are easier to locate and in some places the registration of land ownership may be incomplete, inaccurate or otherwise not sufficiently formalised in order to identify the owner. This is also problematic where there are a number of non-resident owners based overseas. Having an ability to collect from occupiers when it is not possible to identify owners can be instrumental in improving revenue collections, as in Hargeisa, Somalia in 200513. Recommendation: Consider implementing a shift in collection of the tax from owner to occupier: this should aim to address the problem of taxing informally held or traditional land as well as absentee owners. That is, counties would need to identify tenants or beneficiaries, make them liable as taxpayers, and allow them to withhold from the landlord, all with appropriate documentation. This is a practical way to improve administration of property tax, but could be subject to legal challenges and would likely require legislative amendment, which could be considered in developing a modern national framework, as described above. Exemptions and reliefs from property taxes typically fall into the following categories: uses for social care e.g. schools and hospitals; equity-based exemptions for lower value assets or for low income households; politically-driven exemptions; or for publicly-owned land and property or non-profit organisations. To reduce inequality, improve political acceptability and encourage development for key social sectors, these can be valid policy rationales for providing exemptions. However, they come with a risk of revenue loss, potentially more complex administration and opportunity for fraud and evasion, as well as placing a greater burden on those who are liable to the tax. The UK and US, for example, provide reliefs from property tax based on occupation or income levels. This can be difficult to implement if data is not available or detailed enough to link property to incomes. Examples in Africa include Tanzania, in which local governments provide exemptions for individuals based on age or income levels. Where income or age data is not available, rental values are used as a proxy for income levels (Collier et al., 2017). Property tax rates across UK, Europe, US and East Asia are in the range of 0.15 to 2% of the market value. Rates in Rwanda are also low, at 0.1% of asset values (Collier et al. 2017). Other African countries have higher rates, often reflecting out of date valuation registers, as in Kenya. Rates of property tax are typically applied based on category of land use e.g. residential, commercial or industrial, or by geographic area, in some cases. Kenya County property rates are based on a few categories of land use, such as agricultural, industrial and residential. However, some do not have a ‘commercial’ category and significant numbers of ‘Business cum residential’ (BCR) properties, both of which need a clear definition for rates purposes. Administration of property tax is much simpler if there is a single rate applied across all properties. However, Collier et al. (2017) identify a number of reasons why variation of rates may be justified, providing there are sufficient administrative capabilities to put these into effect efficiently, such as: Reducing land ‘speculation’: Relatively higher tax rates on underdeveloped land compared to developed land discourage the holding back of land when it is needed for development; Rates commensurate with use of public services e.g. residential properties generally make greater use of local services than non-residential properties; To support achievement of urban land use master plans by using rates to incentivise certain land uses or investment in certain locations; and Wealth distribution: by applying higher rates to higher value land and property. The burden and affordability of the tax should be seen in context of the range of the incomes of taxpayers and other taxes placed on them as individuals and on land and property or associated transactions, such as stamp duty on transfers of 13 Walters (2011) “Land and Property Tax: A Policy Guide.” 24 property, taxes on rental income, VAT on property transactions (although land itself is typically exempt from VAT), taxes on capital gains, inheritance, and so on. Uganda property tax has an exemption for all residential property. Due to this and administrative difficulties, property taxes only raise slightly more than revenues from taxi and parking fees 14. Nonetheless, even modest improvements in local investment in Land and Property Tax collections can have a major impact. For example, reforms to property taxes in Lagos from 1999 helped the state to achieve a five-fold increase in public revenues to over $1 billion in 2011 15, which helped finance some major and popular infrastructure investments. In another example, estimates indicated that a 1% tax on land and property in Kigali could generate over $60 million annually, assuming full compliance16. Recommendation: Rates to be reviewed and adjusted with re-valuation: Kenya county rates should be adjusted to a low, uniform rate on a broader, up to date base. Setting rates may require an assessment of impact on ability to pay and consideration of a more progressive regime (e.g. a higher rate for high value properties above a threshold). Relative use of services could also be a consideration in rate setting, as well as property use (agricultural, residential and industrial). Counties should consider introducing a ‘commercial’ category and a methodology for establishing a ratio for ‘mixed’ properties ; Regarding contribution in lieu of rates (CILOR), the relevant policy principle is that Government property should be taxed in the same way as the private sector. Special purpose infrastructure should be assessed separately, usually by an international expert. However, in practice, as the usual enforcement techniques are not available for government entities, payment of tax is dependent on the good will of government institutions and can be difficult to collect through standard channels. The concept of CILOR, therefore, is to do with how Government’s tax liability to the County for owning property in that County can be paid. That is, it is a payment of rates owed on buildings occupied by government, which require servicing. The logistics of making the payment to counties could be done through the central revenue allocation mechanism, but then the allocation formula should capture liabilities for property rates, if not already. CILOR was in place during the time of Local Authorities, but has lapsed since the introduction of Counties. The issue, therefore, is whether to exempt government from the liability for property tax. There is no real reason to do so other than the practical (lack of) capacity to enforce. It could therefore be more practical to maintain the principle of payment in lieu, but using an agreed methodology for calculating and processing (government transfer) of the payment. Recommendation: Reinstate CILOR with a clear methodology and effective process for government payment/transfer to Counties that captures the tax liability for government-owned property, with the principle that government property should be taxed in the same way as the private sector. Entertainment Tax Weaknesses in Legal framework Entertainment tax is the second of the two taxes prescribed in the Constitution. It is imposed on operators of venues for “entertainment” at a rate of 18% of the amount charge d by the operator for admission, by ticket, to the venue. The term “entertainment” is defined to include exhibition or performance for which paid entry is required). However, there is a very significant exemption from the scope of the Act. An operator that is registered for VAT is expressly excluded from the definition. Thus the tax base is limited to those operators who have turnover below the VAT registration threshold or are otherwise exempt from VAT. Taxes on turnover for the provision of defined services, such as the entertainment Tax, would normally be replaced following introduction of VAT. In Kenya’s case, following the introduction of VAT, this amendment to exempt VAT registered businesses was introduced. It is consistent with the principle that VAT should replace turnover taxes. However, this treatment significantly narrows the tax base under this revenue stream. The tax base is limited to admission income, based on amounts paid for tickets to enter venues. The Act provides additionally for tax to be recovered from operators of venues run as clubs, associations or societies, where the tax is calculated on the lump sum paid for the right of admission over a period of time, for example, an annual subscription. Although the Constitution, Schedule 4, Part 2, includes “betting, casinos and other forms of gambling” as one of the group of functions in paragraph 4 within the general definition of “cultural activities, public entertainment and public amenities” , there is nothing in this allocation or the Entertainments Act that would create a power of the Counties to impose a form of 14 Taylor (2016) “How One African City Is Flipping the Script on Urban Development. 15 Paice (2015) “Lagos Proves Africa’s Property Tax Potential.” 16 Murray, Kopanyi, and McSharry (2016) “A Land Value Tax for Kigali: Analysis and Policy Considerations. 25 tax on betting or gambling. Further, the substance delineation notice in the Gazette (see earlier), whilst subject to scrutiny in terms of the proper decision making body, makes clear that the revenue streams of Counties within the category of betting and gambling are licensing fees imposed as part of the discharge of regulatory functions. Recommendation: Given the very limited base, consider repeal of the Entertainment Tax Act (Cap.479) and enactment of a new Act with a wider tax base consistent with the functional responsibilities under Constitution Fourth Schedule Part 2, paragraph 4 but without being double taxation of those services subject to VAT. Policy Challenges In terms of policy rationale, a case can be made to impose an additional tax (over and above the standard income tax and VAT) on gaming and betting activities in order to capture a greater share of what is typically a more highly profitable sector, as well as to address concerns about wider costs to society from the impacts of gambling addiction. Nonetheless, if in practice the taxing rights assigned to counties are limited to the licensing of premises, then this would appear to be akin to a regulatory activity only as opposed to a tax and might ‘fit’ more efficiently and logically under the existing SBP regime. Other forms of entertainment tax based on the number of admissions to entertainment venues and the value of admission charges might be difficult for counties to administer in practice, due to difficulties in validating admissions and receipts. As the underlying objective of this tax may be similar in nature to regulatory licensing of premises, a simpler or more efficient method might be to capture the additional tax revenue under the SBP regime (as with county gaming and betting tax). The additional ‘tax’ could be captured, for example, as a higher rate for larger venues to reflect the likely number of admissions. This might also adequately capture the likely use of local services at events, such as public safety management, roads and waste disposal. Recommendations: Consider combining gaming and betting licensing of premises with higher SBP license fee for ease of administration and to capture social cost of gambling addiction, in context of other taxes on gaming and betting industry and industry norms and standards; Fees for entertainment venues are also a form of regulation of business and therefore could be considered as part of SBP, with higher rates for larger venues to reflect the public safety risk and cost of regulation. Single Business Permits Weaknesses in the Legal framework Counties have a functional responsibility for trade development and regulation. This covers markets, trade licences, fair trading practices, local tourism and co-operative societies. Counties are entitled, therefore, to raise revenue in the form of licencing trading activities. However, counties may not seek to license professional firms (such as lawyers, accountants and doctors). These are regulated by the National Government or entities established under legislation of the Parliament (for example, the Medical Practitioners and Dentists Board established under the Medical Practitioners and Dentists Act Cap.253). An additional requirement of a county government for a professional services firm to apply for a single business permit (trade licence) would effectively be additional regulation of the firm, and consequently is excluded by application of paragraph 7. Under the pre-2010 Constitution, the Local Governments Act provided for a Single Business Permit (see Cap.265 s.163A). Counties may therefore impose licence fees by applying the transitional provisions of the Constitution to the pre-existing Acts relating to the functional areas covered above. As discussed above, this is complex and requires careful application of legislative interpretation principles and constitutional provisions to ensure that a practical outcome is arrived at. This is not desirable. Since counties may not rely on the transitional provisions of the Local Government Act Cap. 265 in relation to Single Business Permit as that Act has been repealed, they may not collect fees for SBPs or trade licence without clear legislative authority of an Act of the County Assembly. Some counties have taken the step of legislating for the trade licences. For example, the Machakos County Trade Licence Act 2014 (Act No.9 of 2014), MCGS, 29th December 2014, page 339, which follows, mainly, the basic model recommended by the Kenya Law Reform Commission in its suite of model County laws. Recommendations: 26 County legislation should be enacted to establish a clear licensing framework that is consistent with the principle of a SBP. The model legislation on trade licences should be reviewed to ensure that it is consistent with the principles of SBP. County legislation should not seek to impose a SBP requirement on a professional business that is regulated by statute Policy Challenges As discussed above, Business licensing is, in principle, a means of regulation of business activities, by granting the right to conduct business in a particular locality and sector. The setting of permit or license fees should, ideally, be limited to the cost of the regulatory activity i.e. administration and enforcement of registering, issuing licenses and monitoring or inspection activities. Where there are negative externalities involved and the rationale for licensing is to control a particular activity that has a wider social cost, then the pricing needs to capture the social cost as well as the marginal financial or economic cost of regulation. Nonetheless, it is also often used as a means of raising revenues and is a common source of local government revenue, particularly in countries such as Kenya in which the standard income tax is not yet sufficiently formalised and cannot capture a number of businesses operating in what is effectively a ‘shadow’ economy. The recommended approach internationally to taxation of informal traders, for example, is presumptive taxes. In the case of hawkers, or street traders, there are a number of methods, many of which are already in place: demarcation of permitted street markets, sale of a limited number of trading permits, daily rental of demarcated stalls, daily rental of certified hawking barrows, among others. It is good practice to use the tax regime to encourage formalisation. Local level business taxation is therefore common globally, with varied types across different countries. Bird (2003) 17 describes various trends, including North America (corporate income, VAT, property), Europe (business capital/profits in Germany, business capital/payroll in France), Latin America and Asia (turnover w/ flat fee for small businesses in Brazil, Venezuela and the Philippines, Octroi in South Asia), Francophone Africa (business capital/payroll/turnover) and Anglophone Africa (typically licensing). Any revenue-raising objective applied to this type of fee needs to be balanced against the cost of implementing counties’ regulatory responsibility as well as the impact that imposing additional fees might have on the cost of doing business and competitiveness. A further rationale is provided by the principle that businesses benefit from local services provided and should therefore make a contribution to those services not covered by direct user charges. Traditionally, there is a need for regulation of businesses in a number of areas 18, such as (i) public health risks associated with food and alcohol sale and processing; (ii) public safety, such as at public events, gaming and betting activities or casinos, risk of pollution or industrial waste etc.; or (iii) public welfare, including zoning of locations to manage congestion, quality control, or regulating competitive practices. A business licensing regime therefore not only provides a source of revenue, but also supports information gathering that provides a basis for local planning and budgeting of service delivery. Local business licenses typically consist of a range of categories of license with varying fee rates. While with general taxation, it is typically more efficient to maintain a single, low rate across all activities so as not to distort economic activities and investment decisions, in the case of regulatory activities, the fee may vary according to the relative cost on society. Ability to pay may also factor into pricing of fees and therefore can be progressive i.e. with rates increasing with the level of turnover. Depending on the type of business and market for their products and service, the burden or incidence of business licensing may vary. Businesses engaging in activities for which there is a low price elasticity of demand, such as gaming and betting, businesses are likely to pass on the cost of licensing fees and taxes to the consumer, who will bear most or all of the incidence. In Kenya, Single Business Permits were introduced in the late 1990s at Local Authority level, after central government trade licensing was repealed. It also provided a means to consolidate multiple licensing activities and agencies, which had created a complex environment for doing business. Under previous LA structure, MoLG determined a relative fee structure or schedule, from which Authorities could choose and set the absolute fees. A study of Kenya business licenses in 2006 identified around 300 licensing requirements, and a subsequent comprehensive inventory found well over 1,300 business licenses and associated fees imposed by over 60 government agencies and 175 local governments19. This appears to be a common problem across a range of countries, according to similar findings. While this was under the former Local Government structure and not under Counties, in practice, as discussed above, the County Governments appear to be adopting the old fee and rate structures without significant review and reform, meaning that the plethora of categories of licenses and fee rates still persists. 17 Bird, R. (2003), “Local Business Taxes” 18 Kelly, R. (2003), “Mobilizing Local Revenue from the Business Sector” 19 W orld Bank (2010), “Policy Framework paper on Business Licensing Reform and Simplification” 27 Recommendations: Consider national guidelines for the relative, simplified structure of fees, based on size (e.g. employees or turnover), discretion for counties to set fees within the framework. Variation of fees outside simplified framework to have a clearer rationale, such as the regulation of specific sectors (gambling, liquor licensing) and/or addressing negative externalities (e.g. environmental damage). Cess Cess was imposed by the previous local authorities under the Agriculture Act Cap.318, s.192A. Cess was imposed on agricultural produce, if approved by the Minister. The Act provided no criteria on the method, timing or scope of the imposition. The Agriculture Act Cap.318 was repealed and replaced by Agriculture, Fisheries and Food Authority Act, 2013 (No. 13 of 2013). The new Act does not provide for Cess in any form. Thus Counties may not rely on the transitional provisions in the Sixth Schedule of the Constitution to continue to impose Cess pursuant to Cap.265. Any Cess that is to be collected requires to be legislated for by each County through new legislation passed by the County Assembly. Any cess, as a fee, legislated by a County, would require to be consistent with the “single national market” tests set out in Article 209 of the Constitution. In addition, as summing this test can be satisfied, the need to legislate for Cess is dependent upon an analysis of the potential revenue and satisfaction that the revenue stream would cost-efficient, in relation to the relatively high costs involved in administering a low yield revenue stream. Recommendation: Cess fees should not be collected without clear legislative authority enacted by the County Assembly. County Cess legislation should therefore be repealed unless: • A clear revenue potential case can be made. • The Art. 209 tests can be satisfied Cess is essentially another term for a tax. In some countries it is used to denote a local tax (e.g. UK, Ireland and India) and refers to property tax in Scotland and India 20. However, in the Kenya context it mostly refers to a levy on agricultural produce. Some counties also apply what they refer to as cess to other activities, such as quarrying or sand harvesting (sometimes also referred to as ‘transport cess’, since it is levied on trucks carrying rock or sand). Another county example, levied an ‘infrastructure maintenance fee’ on the produce from a tea fa ctory (also similar to agriculture cess, but for a value-added agricultural processing activity). In some counties, ‘market fees’ refer to a type of cess. That is, a kind of income (turnover) tax on the sale of agricultural produce in local markets. Whereas, as discussed earlier, market fees could also relate to a form of rental charge or access fee to enter and sell from a market, or for use of a market stall or other market facilities provided by the County. The cess- type market fees can be highly punitive on very informal, micro traders on low incomes. This is in conflict with wider income tax policy, which excludes very low incomes for equity reasons. They are also typically excluded for administrative efficiency reasons, since it is highly resource intensive to position revenue officers all day across local markets to collect a very small amount of tax. Recommendation: Counties to clarify the definition and objectives of market fees e.g. is it a type of cess on produce, an access fee to market space and facilities, or rental charge for use of government property (market stall). Since there is a case to remove cess, enhancement of this type of revenue source should therefore focus on management of assets (market stalls, facilities) and ensure that fees are commensurate with the cost of provision and ability to pay. A key problem caused by cess, which is already well-documented in the draft county revenue policy and complaints reported to National Treasury and others by private business, is that of double taxation and opportunities for informal cess collection or bribery at county border crossings. This is a particular problem for agriculture businesses transporting goods from across several counties, in which each county expects to be able to collect a similar tax from the same truckload of goods. A 2016 study of cess in Kenya 21 found evidence of the existence of multiple cess levies along trading routes, as revealed by traders in urban counties located away from the major production areas that face higher cess charges and that produce cess adds a significant cost to the distribution of produce around the country. In particular “a one percent increase in cess raises the average distribution cost by 0.8%” and a “one percent increase in cess increases the average 20 http://www.businessdictionary.com/definition/cess.html 21 The Bayesian Consulting Group (2016), “The Burden of Produce Cess and Other Market Charges in Kenya” 28 cost of production by 0.2%”. The study therefore argues that cess and market fees add to the overall cost of doing business and that counties should aim to reduce cess collections. Recommendation: Replace cesses due to their high economic burden, double taxation risk and barrier to trade across county borders. In particular: a. Agriculture cess could be replaced with a flat land tax (e.g. as a category of property rates); and b. Quarrying ‘Cess’ (or other labels) to be potentially covered (replaced) through SBP as a special category or replaced with an environmental levy (potentially with a revenue sharing arrangement if part of the existing mining levy/royalty). Advertising, Parking and Liquor Licensing Advertising fees are considered a form of regulation, since the objective is to manage and control advertising that causes outdoor ‘pollution’ (e.g. fly posting) and also to make the most effective use of county property for advertising. It can also be a source of revenue, by providing a service to private businesses in a more managed way that also protects the target audience. Fees could be based on the average cost of providing advertising platforms (e.g. billboards) and/or ability to pay, commensurate with the private benefit to businesses from the expected size of the ‘audience’ reached. Vehicle parking fees also provide a means of recovering the cost of providing and managing county property (parking spaces and car park facilities). It can also provide a form of regulation and/or mitigation of wider social and economic costs, such as traffic congestion. Higher parking charges at peak times or in the most congested areas of a city Central Business District provides a disincentive to drive at those times and can help control the flow of traffic. If congestion management is a primary objective, revenues from parking fees and fines could contribute to more efficient public transport provision to reduce individual car use. Recommendation: Counties to review rates of parking fees in line with benchmarks, location, peak periods and zoning of areas to manage traffic flow etc. and ensure cost of provision and maintenance is at least covered by revenue; and Liquor licensing is part of a regulatory responsibility of the Counties, which can correct for negative impacts on society from the effects of alcohol over-consumption. Insofar as the burden of a ‘tax’ or license on liquor manufacturers, distributors and/or retailers is passed on to consumers, it can have a disincentive effect on the quantity consumed, while raising revenue for provision of valuable services. Mining Royalties County Revenue Sharing Extraction of minerals, oil and gas are subject to standard national taxes, as with any other business activity, but also typically subject to an additional fiscal regime. Since natural resources are considered as public or Government-owned resources, fiscal regimes for mining usually include some form of ‘access fee’, including license fees, signing bonuses and surface rental charges for the right to extract minerals from the ground or under the sea and to use the surrounding land. In addition, there is a further tax, representing Government’s ‘return’ on the resource asset itself (or economic ‘rent’). This can take the form of a royalty (percentage of the value of the mineral produced), a resource rent tax (a form of income tax on the economic rent or ‘super-normal’ profit from natural resources) or additional windfall tax, to capture any further profit achieved when commodity prices rise above an agreed level. While these taxes are usually levied, collected and monitored at the national level, there is often some form of sharing of revenues with the local communities from which the resource was extracted. The exact shares assigned between central and local government and communities is usually a political negotiation, and represents partly a form of compensation for any negative environmental, social or economic impacts of the mining activity itself in the local area, and also to recognise the local ownership of the natural resource and improve acceptability of the project and of the need to give up part of the control to central government in order to manage the resource more efficiently. Options for new revenue sources Local Hotel / Tourism Tax: A form of ‘City tax’ is fairly widespread across Europe, such as in Germany, Netherlands, Italy, Spain and Belgium, as well as US and Dubai, among others around the World. In these examples, the rate is either a fixed amount per room per night or a percentage of the final hotel bill. The purpose for revenues varies, including the development of tourism (Cologne), to support tourism boards, town halls and the Tourism Agency (Catalonia) or purely revenue-raising (Italy). 29 While Kenya, as across East Africa, already has a form of tourism tax, primarily in the form of a national Tourism Development Levy. Across East Africa, the scope of VAT on services has also broadened to include tourism services. Tanzania’s Tourism Development Levy is charged at 2% on hotels and lodge s by the Tanzania Revenue Authority. An additional local tax on hotel occupancy may therefore encounter double taxation issues. Nonetheless, there could be a policy case based on the rationale that local hotel occupants benefit from local services and tourism infrastructure. In Uganda, the 2008 Local Government Amendment Act, also provides for Local Hotel and Local Services Tax, in addition to the national Tourism Development Levy, which is imposed on leisure businesses in general as well as hotels. The Local Hotel Tax is aimed at urban tourism regulation and is a specific amount levied on hotels (of between US$0.15 and US$2 per room per night) determined at the national level in the Local Government Act and is intended to contribute to local provision of street lighting, road maintenance, refuse collection etc. The introduction of a Local Hotel/Tourism Tax would need to satisfy the Constitutional requirements (1) to enable County Taxes and (2) for their impact to satisfy the various criteria in the constitution. Definitions would be part of agreeing the scope of the tax. Recommendation: Consider whether there could be a policy case based on the rationale that local hotel occupants benefit from County services and tourism infrastructure. Careful review in line with the Constitution, VAT and East Africa Tourism Levy is needed, as well as enactment of appropriate legislation to empower Counties to collect/receive revenue. Environmental taxes Economic activity that causes localised environmental damage may provide a case for local taxation. For example, mining, quarrying or an industrial activity that causes air, water or other types of pollution. An environmental tax ‘i nternalises’ negative externalities from environmental degradation, with a rate based on the cost of mitigating environmental damage and/or capturing the environmental cost. Typically, environmental taxes are levied as part of a regulatory regime. For example, the base for a tax could be an access fee (a license or right to carry out the activity) and/or a specific tax per unit of production, or gross revenue/turnover of the business. Across the Kenya counties, taxes on quarrying and sand harvesting, for example, are viewed less as environmental management taxes and more as a revenue-raising cess. Some counties refer to this as ‘transport tax’, as it is levied on trucks on the road and was intended to pay for road maintenance investment. However, transport has little to do with the environmental effects of aggregates extraction and in practice, revenues were generally not used for transport investment, but for general revenue raising. Pricing is usually on the basis of price per ‘trip’ (truck -load) crossing the quarry gate or road block. While simple to administer, this is a fairly blunt measure and is not commensurate with the environmental damage or extraction rate, as the same fee applies to a large, heavily loaded truck as a small, light load. While the cess suffers from a risk of double taxation due to levies provided in the Mining Act 2016, it may be possible to make a separate case for a dedicated environmental tax, if there are specific challenges that are not currently addressed through the central revenue sharing arrangement from mining levies. Kenya’s construction industry represent s about 7% of GDP, a part of which consists of primary materials from extraction and production of rock, gravel, sand, clay, cement etc. Since demand is relatively inelastic, as there are few viable alternatives and demand for aggregates as construction materials is high, this could potentially be a good revenue raiser, as well as mitigating environmental damage from natural resource extraction. Alternatively there could be other polluting activities that could warrant environmental taxation, such as road use or fuel consumption or congestion charging to tackle air pollution. Littering is a localised issue and some countries impose a tax on plastic bags. Since Kenya has an active ban on plastic bags, there may be other pollutants that could be better managed through a tax. Aggregates taxes: European Examples An evaluation by the European Environment Agency in 2008 reported that aggregates taxes were in effect in the UK, 22 Denmark, Sweden, Belgium and Italy, with other countries levying mining or extraction charges. The aggregates taxes were introduced for a range of reasons, including revenue, compensation for environmental costs, efficient management of natural resources, reducing demand for primary aggregates and therefore reducing the rate of extraction and encouraging recycling. In the UK, the combined effect of the aggregates levy and landfill tax, which placed a specific tax per tonne of industrial waste (landfill tax) and aggregates production (aggregates levy), with relief provided for any landfill waste taken for recycling, helped reduce demand for primary aggregates and expanded the market for recycled aggregates. This was simple to administer and had little impact on industry competitiveness or cross-border effects. Furthermore, for a limited period a share of revenues were provided to a Aggregates Levy Sustainability Fund, which was devolved to the regions to provide mitigating measures for environmental damage. However, due to budget pressures this was later removed. 22 Effectiveness of environmental taxes and charges for managing sand, gravel and rock extraction in selected EU countries, EEA Report No. 2/2008 30 Italy introduced a tax as a local revenue measure, which was earmarked for ‘compensatory investments’ to mitigate environmental costs in quarrying areas. Charges were equivalent to 5% of turnover in the industry, which had little effect on demand. In practice, there was also little evidence to suggest that the revenues were used for environmental damage mitigation, although there were some qualitative examples, such as restoring old used quarry sites. It did provide an additional revenue source to local governments and there were also indirect benefits, in the improved monitoring and regulation of quarrying, resulting from better information gathered on sites and production, which helped to detect and deter illegal quarrying. Recommendation: Consider the case for environmental taxes in Counties that particularly suffer from environmental degradation from harmful activities e.g. quarrying, polluting industrial activity or waste disposal. Careful consideration in context of alignment with existing mining law in case of quarrying or other existing national regulation. County income tax/VAT levy: While property taxes are regarded as the ‘best fit’ in terms of meeting the taxation principles, some countries advocate the use of local or regional income taxes, although they are based on mobile factors. Payroll taxes are typically earmarked for specific purposes e.g. national social security or health insurance. These are normally levied as a single percentage rate, with few or no exemptions, on salaries and withheld from employment income. The burden of payroll taxes tends to fall on the employee, since the cost of labour reflects the overall productivity of labour, regardless of payroll taxes. In this way, they tend to be regressive taxes. An alternative mechanism for local own-source revenue, is local revenue sharing arrangements with central government. For example, most mineral, oil and gas revenues have provisions in national minerals acts defining revenue sharing arrangements, providing for a certain (negotiated) percentage to be allocated to central, regional, local and/or communities in which the extraction takes place. This serves as both a compensating mechanism to make any disruption from extractive activity more acceptable to local communities or mitigating any negative impacts, and also to ensure that future generations can benefit from the natural resource discovered in that locality. Sharing of national tax revenues through central government grants is also similar, in terms of principles. An advantage of this approach is that it allows for redistribution of revenue to correct imbalances in the economic structures and needs of each locality and therefore ensures a minimum level of service is financed. Earmarking of revenue for a specific spending purposes places inflexibility on budgets to spend without being responsive to need. Nonetheless, it can be a motivating force to encourage compliance with the tax, if taxpayers know it is for a worthy cause and can see a noticeable difference. Earmarking also provides a level of certainty on spending and planning for delivery, which can foster better quality investment management and ensures a minimum level of provision for that purpose is protected. Examples of this in other countries has been applied to sector-specific taxes, such as Ghana’s health insurance levy. It can also be more efficiently collected through existing national revenue collections (e.g. by KRA), which can compensate for low administrative capacities at county level. This approach also encourages counties to spend in areas that central government wants to encourage that would not otherwise be provided i.e. where there are positive spillovers e.g. public health campaigns or investment in research and technology. Some drawbacks, however, include the limited control that counties would have over the revenue source and that it could be relatively inefficient in terms of providing a financing structure that allows counties to be responsive to meet local demand. Ideally, tax rate variation should reflect the variation in service delivery provided at each county, and allowing counties to adjust the rate of additional tax levied on the national income tax or VAT would provide for some discretion for local variation. However, this also adds complexity to the system that could discourage voluntary compliance, and by basing the tax on ‘mobile’ factors, this would not meet the ‘internalised’ criterion as discussed above. That is, that taxpayers could choose to minimise tax by moving their activities to another county, which could undermine county revenue sources. A uniform, low rate across counties may therefore be more effective if this route is chosen. Recommendation: Consider feasibility and efficiency gains from such a tax against the costs/disincentive effects. Any income tax or VAT levy would require to be authorised by the National Parliament, and stated to be a tax that County Governments are authorised to collect (consistent with Article 209(3)(c) of the Constitution). This would require political agreement between the different levels of government. 31 Revenue administration key findings Almost all the counties responding to our questionnaire reported to have challenges with revenue non-compliance, including: Ineffective policies and laws or the lack of adequate enforcement legal framework Lack of property rating and valuation legislation Political interference Weak enforcement unit to enforce compliance Corruption among revenue collection officers Out-dated databases inherited from the defunct local authorities Absentee landlords making it hard to collect land rates Scanty information on existing and potential taxpayers Huge overdue collections inherited from defunct local authorities which continue to accumulate making it harder for the affected to comply High poverty levels in some counties Lack of effective and efficient revenue collection and management systems Tax evasion and resistance to pay Failure by some to register their property Prior to 2010, local authorities used administration and enforcement powers that were prescribed in each relevant Act enabling rates, entertainment tax and various forms of fees for licensing and regulatory functions. The responsible administration entity was the local authority. As discussed previously, there is no law equivalent to the administration procedures law for national tax collection, leaving the collection, assessment and enforcement procedures to Counties, which leads to a diversity of administration frameworks and approaches, which can be confusion, unfair or unnecessarily burdensome to taxpayers. Methods reported in our survey include: eviction, closure of business, confiscation, clamping and inspection accompanied by police officers. While these measures were considered appropriate, sufficient, and clear to both officials and the public by a majority of responding county officers, in some cases these methods may require review to ensure they are in line with good international practice approaches adopted by KRA. For example, in the counties we visited and from stakeholder feedback, the revenue administration focus was often on enforcement rather than encouraging voluntary compliance. Furthermore, a number of tax amnesties have been used to undertake periodic drives to widen the tax net and improve compliance. However, repeated amnesties in Nairobi, for example, are considered to have undermined enforcement. In some cases, (especially Nairobi) the extent of arrears is so great as to make amnesties more appealing. Nonetheless, there is a high risk that Taxpayers will simply delay and evade, and wait for the next amnesty. Therefore, while a one-off amnesty might seem an effective way to improve compliance, amnesties can undermine compliance over the longer- term, particularly if carried out periodically. A key principle of voluntary compliance is that revenue can be collected more efficiently form a wider base than if collection requires regular assessment and enforcement by a revenue authority. Ways in which voluntary compliance can be encouraged include simplification of tax and fee regimes (to ensure that assessment is simple), clear communication to ensure taxpayers understand their liabilities and ease (and convenience) of payment procedures e.g. mobile money payments or on-spot payments. Recommendation: Counties should enact legislation to set out compliance obligations and powers in a County (Revenue Administration) Act, and the legislation can be based on the existing model, reviewed and updated through the Intergovernmental Relations mechanisms. Alternatively, such legislation could be enacted by Parliament for exercise at a County level, to ensure consistent county tax/fee/charge administration, compliance and enforcement treatment throughout Kenya. Counties should introduce risk management approaches in revenue administration i.e. to identify, assess and monitor revenue risks using data on registration, filing, assessment and payment, in order to develop appropriate strategies to improve compliance according to risk. Counties to consider ways to improve voluntary compliance e.g. through ease of self-assessment and payment and to strengthen taxpayer/user awareness and support, including access to information, guides and quick reference on 32 liabilities and procedures. This will improve communications with taxpayers/users on policy objectives and the benefits of tax to service delivery. In the absence of a prescribed revenue administrative framework, steps are required to develop the most appropriate and efficient method of administering the various County laws and collecting revenue across the various revenue streams. This can be done in a variety of ways – By the County Finance Department; By a county revenue agency, modelled on national semi-autonomous revenue agencies; By engagement of the Kenya Revenue Authority; or By engagement of a commercial entity. The features and potentials risk of these approaches are summarised below – Arrangement Benefits Risks or disadvantages County Finance  Uses existing staff and infrastructure  Existing staff may be Department  No cost to establish inadequately trained or skilled for focussed revenue collection  Hiring additional staff constrained by County HR policies and procedures County Revenue Agency  Allows selection of staff  Costs and time to establish  Allows easier allocation of staff to new  Additional governance and functions administration challenges Kenya Revenue Authority  Benefits from existing KRA knowledge,  Limited to those areas where skills, expertise and infrastructure (in KRA can add value (tax and tax- respect of formal tax or tax-like fees) type fees)  KRA has a national presence and  Challenge for Counties where reputation, and maintains relevant KRA has no infrastructure infrastructure and networks presence  MoU sets out roles and responsibilities and fair remuneration  Capable of adjustment by agreement to changing circumstances (such as new county revenue sources)  Supports information sharing to mutual benefit of Counties and KRA External agencies  Provides rapid implementation of  Requires public procurement collection processes based on existing processes to be in place and to commercial collection systems be followed  Insufficient commercial negotiation expertise may result in disadvantageous terms  Transparency on performance management and performance payment is challenging  Susceptible to long terms without review/ termination/ amendment/ All four of these methods have been adopted in various Counties. The Counties that have engaged external collection agencies, on a commercial basis, have done so on terms that are favourable to the collection agency, and not to the County, and are weak in terms of the oversight, audit and contract review provisions. Data on the average cost of revenue collection in OECD countries ranges between 0.3% and 1.7% of revenues collected23. There are few estimates of the cost of collection in developing countries, but the cost of collection for property tax in Kampala, for example, is estimated at around 10-12%24. In some cases the costs can be much higher, particularly using private contractors. In Tanzania, a private agent contracted to collect revenue from a bus terminal in Dar es Salaam in 2006 retained almost 60% of revenues collected25. Other studies of developing countries have found that outsourcing to private collectors can undermine government capacity that is required to monitor and enforce private collection contracts. In Kampala, for example, the 23 Administrative costs for tax administration/net revenue collected %, OECD (2013) 24 Haas and Manwaring (2017), “Private vs. public collection in enhancing local tax revenues” 25 OH Fjeldstad et al. (2008), “Outsourcing Revenue Collection: Experiences from Local Government Authorities in Tanzania” 33 Kampala Capital City Authority (KCCA) brought revenue collection back in-house due to corruption and high costs of private outsourced collections and achieved double the collections of “road user fees” from minibus taxis, for example, in one year, and revenues at the KCCA expanded by 89 percent after inflation, from US$9 million to $24 million between 2010 and 2015. Indicative information from the Kenya counties suggested that the cost of collection is at least 4-7% and could be above 10% if all relevant costs are included (e.g. receipting and payment platform cost, revenue/finance wage bill, equipment, transport and other office overheads etc.). Counties do not appear to measure the cost of collection, making it difficult to monitor performance improvements in this area and to assess the relative cost-efficiency of private versus public collections. From the interviews and survey results, a small number of Counties have entered into arrangements with the KRA (as empowered by the PFM Act 2012, although there are no current regulations as foreseen by s.160). The agreements are set out in writing in the form of a Memorandum of Understanding, and define the revenue streams to be collected, the fees for collection, and the processes for transferring and accounting for revenue. The agreed rates were not uniform, but were significantly lower than the cost-of-collection estimated from the indicative information. There would be a strong argument, and in some cases compelling argument, for a county, faced with immediate needs to put in place an effective revenue collection system, to engage KRA. Any engagement of KRA should, of course, be done to ensure mutual benefit on fair terms. For some Counties, the engagement of KRA may be seen as a transition step in the short-term, if their medium term objective is to build county level capacity to undertake revenue (which is estimated to be at a better cost-to- collection ration than the KRA services) Recommendation: For the short-term, Counties should consider the engagement of the KRA in respect of the collection of property rates and trade licences, in terms of a standard Memorandum of Understanding that sets out service level agreements for KRA and exchange of information to assist both County and KRA to maintain accurate taxpayer registration. Counties should not engage external collectors of county OSR unless the engagement is pursuant to transparent public procurement processes and a cost benefit analysis justifies the level of fees to be paid for the services. Counties (perhaps guided by central government) should strengthen performance monitoring of revenue administration, including indicators of efficiency e.g. cost of collection and compliance ratios to be able to track and measure improvements in efficiency and effectiveness. Virtually all Counties face challenges in the form of insufficient staff with appropriate training or skills to implement a modern, cost effective revenue administration, collection and enforcement operation. Counties reported on average about 215 revenue officers. Compared to the average county population, each staff member ‘covers’ at least 4,000 citizens 26. This compares to an international study of national revenue authorities that found on average that revenue staff cover about 2,200 citizens27. Many of those revenue officers were retained from the local authorities and have insufficient training (reported in 40% of counties). On average, about 40% of revenue staff were reported as ‘qualified’ and only 50% of counties reported having OSR procedures and/or a training manual. Recommendation: Counties should consider further investment in staff and skills. This may require reviewing recruitment policies, providing induction training on ethics and integrity as well as technical training on procedures and introducing guidance or handbooks. On the whole, counties collect data on taxpayers (or users) and their transactions and revenues collected and have some form of system for detecting late/non-payment. However, 30% of counties reported not having taxpayer records/databases and relied on other registers such as the debtors register, business register and plot registers to keep track of their revenue streams. There is room for improvement in the comprehensiveness and accuracy of taxpayer data and is not analysed sufficiently to inform policy design. Recommendations: Counties should invest in improving the accuracy and comprehensiveness of taxpayer data, as an essential basic platform for monitoring revenue risks and designing appropriate compliance strategies. This may require reviewing and, if necessary, simplifying IT systems and databases to focus effort on good quality entry and maintenance of information. Counties should engage in information sharing arrangements (through written agreements) with agencies whose data can contribute to ensure the integrity of the County OSR tax base. This requires establishing protocols for sharing of 26 Based on 2015 population projections from 2009 Census, Statistical Abstract 2016 27 A. Lemgruber et al., (2015), “Understanding revenue administrations” 34 data e.g. which specific data is required by whom, when, for what purpose and in what format. Understanding and agreement on treatment of confidential data will form part of these protocols. Revenue management key findings Revenue forecasts appear to consistently over-estimate revenue collections on average, compared to actual outturns. On average, county revenue collections were almost 40% below projections between 2013/14 and 2016/17. Reported methods of forecasting included use of historic trends, tax base projection, consideration of economic conditions, development trends and surveys. Nonetheless, in practice, county revenue projections do not appear to reflect actual revenue bases or historic outturn trends. Furthermore, revenue forecast assumptions are not transparent and therefore are not open to scrutiny or challenge. County revenue management IT systems are generally not integrated with the Financial Management System, some of which are based on those of the defunct local authorities (LAIFMS), which still used different geographical classifications that are inconsistent with the current classification under county governments. Revenue data are therefore entered manually into the IFMS. IT systems are also regularly affected by power failure and weak network connections. Smaller sub-counties therefore use manual receipting, which is transferred and consolidated into excel sheets and combined into Excel and entered manually into IFMIS for financial reporting purposes. Most counties reported to be using electronic payment systems, such as Point of Sale (POS) devices, used to capture revenue at the point of collection and for generating receipts. These systems have the advantage of capturing real time data and can incorporate mobile payment methods. Some counties have become entirely cashless using this technology, which has helped them to mitigate fiduciary risks. They can generate some reports from such transactions. Recommendations: Counties (supported, where possible by central government for efficiency and consistency) should consider investing in greater automation of revenue management systems and strengthen IT connectivity, speeds and infrastructure to support the replacement of manual receipting and reporting. This should also involve consideration of establishing better integration of revenue management systems with IFMS reporting. Make use of existing and new available data on tax bases (e.g. from this study) to strengthen forecasting methods, such as use of monitoring, modelling and projection of tax bases, analysis of historic trends and impact analysis of new measures. The transparency of forecasts should be improved by publishing forecast assumptions. Counties should undertake regular performance reviews (e.g. annually) of forecasting and outturn revenue performance, providing reasons for deviations and propose improvements, incrementally. Incentives for revenue enhancement through the county allocation formula As discussed in section 3.1 (3), revenue potential estimates can be considered in the allocation formula for intergovernmental fiscal transfers. If such a system were to be considered in Kenya, review of good practices and country examples would be important. Nonetheless, as an illustration, the following steps indicate how such a system could be implemented (basic illustration in Figure 1): 1. How much does each county need (relative to others if the aggregate transfer amount is limited)? This requires calculating county financing needs to provide an adequate level of county goods and services within their constitutional functional assignment. This can consider (a combination of) counties’ population, demographic, poverty, geographic, local price, infrastructure or other characteristics. 2. To what extent are individual counties able to cover their financing needs through their own source revenue potential? a. Which county revenue sources should be factored in when establishing counties’ abilities to cover their own service delivery needs? Total own source revenue potential could be applied; however, this may create incentives with unintended and undesirable consequences, resulting in risks that counties ramp up collection of socially or economically harmful revenue sources to cover the financing needs. Such risks can be limited by focusing on selected key revenue sources, e.g., property tax and revenue from natural resources. b. How can the revenue potential objectively/fairly be established across counties? It is critical to design an appropriate methodology building upon high-quality dataset(s) that cover all 47 counties and is regularly updated, supporting broad acceptance of potential estimation results. A detailed study comparing different options for any selected revenue source(s) would likely be useful to generate more exact and reliable estimates than the scope of this study allows. c. What is a realistic share of the potential that counties can be expected to collect? Revenue potential is never fully realized due to resource and capacity limitations, non-compliance, etc. This is particularly the case in evolving 35 revenue administrations such as in Kenya and especially at the county level. Based on capacity and systems, only a share of the revenue potential can be assumed to be collected, which can increase over time as capacities improve and to incentivize increasing revenue mobilization efforts. d. What share of financing needs have to be covered through intergovernmental fiscal transfers? Deducting (a realistic share of) local revenue potential of the identified source(s) from established financing needs will result in the balance to be covered through equitable, needs-based transfer allocations for each county. This can incentivize counties to collect revenue, while leaving policy decisions on rate levels to counties (lower rates equals less local goods and services). e. How should the ‘saved’ transfer amounts offset through local revenue be used? Examples for how to use the deducted amounts are to redistribute amounts based on relative county needs (focus on equity) or based on county revenue performance relative to potential (focus on incentivizing collections) but also specific earmarking for priority purposes at the local level or other uses could be designed. Figure 1. Illustration of using county revenue potential in intergovernmental fiscal transfer calculation Transfer funds for redistribution (e.g., based on need or fiscal effort) Balance (to be filled through transfers) (Share of) local revenue potential Fiscal need to fund basic level of local goods and County A County B County C County D services Recommendation: A detailed study comparing different options for incentivising counties, based on the potential estimates from selected revenue source(s) from this study (and perhaps other sources), to generate more exact and reliable estimates than the scope of this study allows. 36 Revenue Potential Estimation 3.1 Motivation - Uses for revenue potential estimates Revenue potential estimates at sub-national level can be useful for a number of purposes, including: 1. Benchmarking counties: Revenue potential estimates (correcting for county characteristics that affect revenue bases) allow comparison of counties’ revenue collection efforts, tracking county performance over time, and supporting identification of good practices across counties. 2. Focusing revenue collection efforts: By showing different levels of revenue potential, revenue potential estimates can inform where revenue collection units should focus their efforts. Often, historically-evolved collection priorities (e.g., on the transport of goods such as cess in Kenya) have low potential and are relatively more harmful in a modern economy in comparison to other sources, such as taxes on immovable property for which revenue bases have been growing steeply through urbanization and substantial construction growth over the past decades. Revenue potential estimates can show those imbalances and redirect focus where collection efforts have the highest return. 3. Informing intergovernmental fiscal transfers: Local revenue potential can be considered in the calculation of intergovernmental fiscal transfers per county with the aim of increasing horizontal equity among counties. For example, Papua New Guinea applies such a system to the distribution of operational transfers for basic services to provinces and districts, factoring in unequally distributed natural resource revenues and local revenue generation.28 Revenue potential estimation approach A key objective of this study is to provide estimates of revenue potential by county and by revenue stream. Revenue potential is a term that is often used inconsistently in the literature. For the purposes of this study, we produced estimates of revenue potential that provide answers to the following questions: a) How much revenue would each county be able to raise in total and from each OSR stream if it operated in line with the best performing county in the country? This is calculated using frontier analysis. b) How much revenue would each county be able to raise from each OSR stream if it fully utilised the fiscal instruments at its disposal, resolved issues relating to administration and eliminated evasion? This is calculated using the ‘top-down’ approach. These two key methodological approaches used by the study are discussed in more detail below. A bottom-up approach to establishing revenue gaps is not carried out. Such an approach is generally used to estimating the tax gap (usually utilised in the case of direct taxes, such as income tax) that can also be used to develop econometric tax compliance models – i.e. models to assess the risk of not declaring or paying taxes for taxpayers with particular characteristics. The approach utilises detailed individual-level data of audited taxpayers to ‘gross up’ detected tax evasion to the population of taxpayers. The underlying data can also be used in principle to develop a model that can help ‘predict’ the amount of tax evasion any given taxpayer is likely to be engaging in given their characteristics. Given the nature of most OSR streams and the absence of detailed data on audits of individual taxpayers by each county, applying the bottom up approach in this context is not feasible. Furthermore, preliminary work seems to suggest that for many counties the main reason behind relatively low OSR collections is inconsistent application of the available instruments combined with weak administration rather than predominantly taxpayer non-compliance, limiting the usefulness of the bottom-up approach. Overall, it should be noted that project resource limits as well as limitations in available data and its quality result in high- level, ballpark estimates that give an indication of revenue potential and collection efficiency across counties. Despite this, broad findings are judged to be reliable. Subject to uses of revenue potential estimates in practice, building upon this study revenue potential estimation models can be further refined in the future. While the study reviewed the policies/fiscal instruments utilised by counties as well as carried out some qualitative work to better understand possible administrative deficiencies that may be contributing to reduced revenues, it is not possible to produce a detailed breakdown of the tax gap by whether it arises due to ‘policy’ or ‘administration’. This is largely due to practical reasons (e.g. lack of consistent data/documentation at the county level and project resource limits), but mainly because it is important the revenue potential of each county is assessed on a consistent basis based on objective indicators rather than each county’s policy choices (e.g. decisions by some counties to not collect a particular tax/fee, or to charge low rates). In other words, in the context of this study and the need to define fiscal effort on the same basis for 28 Papua New Guinea National Economic and Fiscal Commission, 2009. Reform of Intergovernmental Financing Arrangements, Plain English Guide to the New System of Intergovernmental Financing. Port Moresby. 37 all counties, the distinction between underperformance due to poor administration/compliance and underperformance due to policy choices is less relevant, and the revenue potential of each county should be assessed on the same basis regardless of county-level choices on how to apply the various fiscal instruments at their disposal. While potential estimates for aggregate county revenues were produced, the study team’s efforts were focussed on modelling potential for revenue streams that currently raise, or have the potential to raise, the most revenue while minimizing negative impact on social and economic policy objectives (see details on selection of revenue sources under 3.3 below). Frontier Analysis Frontier analysis provides the answer to the question ‘how much revenue would each county be able to raise from OSR (aggregat e or a specific stream) if it operated in line with the best performing county in the country?’. In general , frontier methods are used to establish the ‘best performing’ unit – in this case the county that, given its characteristics, is most effective at raising OSR revenue under each stream – and measure the deviation of other units from that ‘frontier’. Data Envelopment Analysis (DEA) is a standard frontier method. The DEA methodology will be adopted to identify OSR potential taking into consideration county-specific characteristics related to different OSR streams. We will attempt to identify the most appropriate set of data to arrive at the best possible estimate of county revenue bases at the aggregate and for each OSR stream. Following this, each county will be assigned a coefficient indicating how close they are with respect to optimal collection of revenues, or the optimal frontier, allowing estimation of the revenue potential of this revenue stream should all counties be able to reach their respective optimal frontier. Therefore, the implementation of the DEA will result in a technical analysis of gaps between actual and potential revenues across the counties, having taken into consideration both best-practices already adopted in Kenyan counties and between-county differences which potentially influence tax collection. Top-down analysis The “Top Down” approach provides the answer to the question ‘how much revenue would each county be able to raise from each OSR stream if it fully utilised the fiscal instruments at its disposal, resolved issues relating to administration and elimin ated evasion?’. It involves estimating the size of the tax base for each OSR stream by county, and then applying the prevailing tax rate to calculate potential revenues. Estimated differences between the potential and the actual revenues collected by the counties reflect the possibilities for enhancement of revenues if existing tax policies were implemented effectively. Top-down analysis also allows for exploring the effects of different designs of tax policies on enhancing revenues. Alternative rates can be selected and applied to the revenue bases in order to arrive at theoretical liabilities. This way, we can estimate a range of potential revenues from fees and taxes levied under various policy scenarios, e.g. applying best international practices or designing different taxation systems (e.g. graduated rates, progressive taxation etc.). In addition to the selection of applicable tax rates, top-down estimations of potential revenues depend on the selection of the revenue base. A major benefit of top-down analysis is that it allows for using consistently collected data to come up with the best possible proxies for revenue bases, ensuring that there is no disadvantaged county as a result of varying data collection protocols. Selection of main county revenue sources The selection of main county revenue sources for the modelling of revenue potential estimates was based on a two-step process using five criteria. The table below provides an overview applying these criteria against counties’ main revenue sources to arrive at a selection for which revenue potential will be modelled. Subsequently, details on actual county revenue collections and a recap of the policy rationale and legal basis for the selected main revenue sources is provided to substantiate the selection. Criteria 1. County actual collections were ranked by importance based on the available data (see discussion of data issues below, which substantially affected the exercise). This resulted in a list of 11 main sources that constitute the vast majority of county revenue collections, i.e., property taxes/land rates, building permits, business licences/permits, liquor licences, vehicle parking fees, advertisement fees, cess, charges on sand harvesting and quarrying, market fees, health-related fees, and rents from county assets. This list is broadly consistent with the findings of the County Revenue Baseline Study for 2015, which identified about 150 different sources of revenues with the highest share of revenues (approx. 90% of total own source revenue) is collected from around 10 key types of taxes/charges in the majority of counties. More specifically, the leading sources of revenue the 2015 study identified were land rates, vehicle parking fees, business licences, approval of building plans, market fees, public health and medical levies, sign boards and advertisement fees, rents and cess. 38 These sources were then filtered based on four criteria, which are interconnected: Policy rationale a. Adequate policy rationale: An adequate policy basis/justification should be available for each of the revenue sources to avoid charges that are harmful for social, economic or other policy goals. This is discussed in more detail in the legal and policy review High actual section. Legal county Revenue raising framework revenue objective b. Legal basis: Revenue sources should be legal within the existing collections legislative framework and, if at all, only require limited adjustments to subordinate legislation under the overarching constitutional revenue assignments. This is discussed in more detail in the legal and policy review section. Applicability across counties c. Revenue raising objective: Revenue sources that are primarily linked to social or economic objectives and should be based on (partial) cost-recovery calculations were eliminated from the list. Health-related fees fall under this category. There are a number of reasons for this: Linking such fees with revenue potential estimates suggests that such fees should be set based on a revenue maximization calculation, which poses the risk that access to services for these fees is reduced. This is a particular issue in areas such as health or education. Further, fees aim to recover (partially) the cost of such services. This means that for any fees collected an equivalent or higher expenditure is incurred to deliver the service. Together these points suggest that a different approach to fees with primary social objectives should be adopted. Based on this study’s findings, a separate review is recommended that looks in detail into (i) which services should be provided by counties and which can be better provided by the private sector; (ii) the costs of delivering such services; and (iii) decisions on cross-subsidy levels for different services. Substantial scope to simplify and harmonize services at the county level appears to be existing. d. Applicability across counties: Revenue sources that are only relevant for a few selected counties were eliminated. Such items include revenue sources dependent on county assets that exist only in a few counties (e.g., revenue from natural resources) or that cannot be assumed to be available across the majority of counties and where asset distribution was unknown (e.g., rents from county land and property assets). Analogue to the finding on the county fee review and reform above, support to maximize use of/return on county assets could be useful. This could comprise a review of which asset types should be owned by counties (policy rationale), their registration and management, including appropriate rental charges and contractual arrangements. Assessment overview Based on these criteria, six out of the 11 main county revenue sources were selected for revenue potential estimate modelling. These include (1) property taxes/land rates, (2) building permits, (3) business licences, (4) liquor licences, (5) vehicle parking fees, and (6) advertisement fees. 39 Table 2: Application of selection criteria against counties’ main revenue sources Adequate policy Revenue raising Applicability across Actual collectionsa Legal basis rationale objective counties Property taxes/land High (most counties) Yes Yes Yes Yes rates Yes (within Building permits High (some counties) Some Yes (within limits) Yes limits) Business Yes (within High (most counties) Some Yes (within limits) Yes licences/permits limits) Low to moderate Liquor licences Yes Yes Yes Yes (some counties) Low to high (most Cess No Unlikely Yes Yes counties) Low to high (most Market fees Some Yes No Varies counties) Moderate to high Health-related fees Yes Yes No Yes (most counties) Low to high (most Vehicle parking fees Yes Yes Yes Yes counties) Low to high (some Advertisement fees Yes Yes Yes Yes counties) Charges on natural resources (incl sand Low to high (some Yes Unlikely Yes Varies harvesting and counties) quarrying) Rents from other Low to high (some Subject to asset type Yes Varies by asset type Varies county assets counties) a Based on available data. See discussion of data limitations below. County revenue collections by source The table below gives an overview of actual revenue collections for selected counties that provided detailed breakdowns of their collections for the review (see discussion on available data and limitations further below). Each of the six selected main revenue sources constitute a significant revenue source for at least two out of the eight sample counties (two highest relative shares highlighted in bold in the table). Table 3: Actual revenue collections for selected counties, FY2016/17, Ksh million County Land Building Business Liquor Parking Advertisement Sub- Significant other Total rates permits licences licences fees fees total sources Embu 13.8 4.2 63.1 0.0 27.7 0.9 109.7 217.1 Cess (45m), market fees (6%) (2%) (29%) (0%) (13%) (0%) (51%) (100%) (20m) Kericho 13.3 0.3 37.6 1.3 32.5 2.1 87.1 183.9 Health fees (42m), market (7%) (0%) (20%) (1%) (18%) (1%) (47%) (100% fees (28m) Kirinyaga 42.0 4.1 99.9 44.5 19.8 3.6 213.8 390.4 Health fees (84m), market (11%) (1%) (26%) (11%) (5%) (1%) (55%) (100%) fees (42m), Kisumu Health fees (246m), 144.5 42.7 96.9 10.9 206.9 60.8 562.7 1,004 market fees (75m), rents (14%) (4%) (10%) (1%) (21%) (6%) (56%) (100%) (44m) Kwale Royalties/cess (27m), 53.7 2.1 55.0 0.0 11.1 12.9 134.7 221.0 health fees (26m), (24%) (1%) (25%) (0%) (5%) (6%) (61%) (100%) auction/ market fees (12m) Machakos Quarrying/sand (319m), 159.8 224.7 180.4 49.3 79.2 0.0 693.4 1,259 health fees (91m), rent (13%) (18%) (14%) (4%) (6%) (0%) (55%) (100%) (50m), market fees (45m) Makueni Market fees (34m), cess 6.7 4.7 65.5 32.7 25.8 4.1 139.5 219.1 (15m), sand harvesting (3%) (2%) (30%) (15%) (12%) (2%) (64%) (100%) (5m) Nairobi 2,253 1,361 1,776 0.0 1,975 720.0 8,084 11,006 General miscellaneous (20%) (12%) (16%) (0%) (18%) (7%) (73%) (100%) (1.6bn), rents (284m) Note: Revenue sources showing zero collections may be reported by counties under summary categories. Source: Unaudited county revenue collection reports. Recap of policy and legal basis for main revenue sources Table 3 summarizes policy rationales and legal basis for the six selected main revenue sources that are discussed in more detail in the legal and policy review section. 40 Table 4: Policy rationale and legal basis for selected main revenue sources. Revenue source Policy rationale Legal basis Property tax/land Closely linked to benefits of local goods and service provision; linked to Constitutional assignment rates household wealth; main revenue source for subnational governments (Article 209 (3)) worldwide Building permits Appropriating share of ‘windfall’ profits from building construction; County function (8. County planning compliance with building codes for public health and safety, construction and development) quality and easier property valuation Business licences Broadening the tax base and appropriating share of profits given that County function (7. Trade development national income tax system is still evolving and business formalization is and regulation, incl. trade licences) still low (temporary justification) Liquor licences Correcting for negative externalities from alcohol consumption, including County function (4. (…) public on health, public safety and waste pollution entertainment (…) incl. liquor licensing) Parking fees Effective use of county property; basic congestion charge to correct for County function (5. County transport, negative externalities from traffic incl. traffic and parking) Advertisement fees Effective use of county property; limiting of outdoor ‘visual pollution’ County function (3. Control of (…) outdoor advertising) Note: See legal and policy review section for details. County functions are listed in the Fourth Schedule of the Constitution of Kenya, 2010. Data availability and gaps County revenue data A database with audited county revenue data covering fiscal years FY2013/14 to FY2016/17 (interim) was compiled under phase 1 of this study. However, several issues were identified with the dataset. Reporting is inconsistent across counties and counties frequently use summary categories (e.g., miscellaneous or other charges) to report data in aggregates. Some of the issues appear to be aggravated by the reporting template, with the revenue classification not reflecting all of the common main revenue sources across counties. For example, no specific revenue item is included for property taxes/land rates. Further, no guidelines appear to be available that guide counties of how to fill the template. In other cases, counties (or national entities responsible for database compilation if counties are not involved) do not seem to record their revenue according to the categories. Several counties report revenue only under one item. For example, Kajiado County reported total collections in FY2013/14 and FY2016/17 under ‘other miscellaneous revenues’ and FY2014/15 and FY2015/16 under ‘other receipts not classified elsewhere’. Bomet county reported between 80 -90% of total revenue under ‘receipts from administrative fees and charges - collected as AIA’ and the remainder under ‘other property income’ in FY2013/14 and FY2014/15 while using ‘other receipts not classified elsewhere’ for total collections in FY2015/16 and ‘various fees’ in FY2016/17. Wajir county consistently reported total collections under ‘various fees’. Table 5 provides an overview of county reporting for aggregate collections and the selected six main revenue sources. 41 Table 5: Available audited county revenue data for six main revenue sources, FY2014/15-FY2016/17. Revenue source Assessment Data on total county revenue appears to be reasonably reliable. Aggregate collections across all counties were relatively stable in the range of Ksh30.5-36.5 billion over FY2013/14 to FY2016/17. At the county level, total Total revenue collections also show limited volatility over the years, except for a few outliers such as Samburu where collections of Ksh175m in FY2014/15 reduced to Ksh16m in FY2015/16 before increasing again to Ksh200m in FY2016/17, which could indicate revenue accounting and/or reporting issues. No specific revenue item entitled property taxes/rates or land rates exists in the collection template. Matching collection data from selected available unaudited county revenue reports with the database points to counties using the categories ‘poll rates’, ‘housing’ and ‘plot rents’ to report property tax revenues. However, other counties appear to Property taxes/rates use these categories differently (e.g., ‘housing’ for income from rental of county housing assets and ‘plot rents’ for rental of county plots). Anecdotal evidence suggests that poll rates are only used to report property taxes; however, only 15 counties in 2014/15, 8 counties in 2015/16, and 10 counties in 2016/17 reported poll rate collections. Overall, it is therefore not possible to establish a dataset with property tax/land rate collections across all counties. A specific revenue item for ‘plan approval fees’ exists. However, only two counties in 2014/15 and no counties in the Building permits other fiscal years reported collections under this category. A specific revenue item for ‘business permits’ exists in the database. 27 counties in 2014/15, 32 counties in 2015/16, Business licences and 40 counties in 2016/17 reported collections under this category. A specific revenue item for ‘liquor licence fees’ exists. However, only one counties in 2014/15 and no counties in the Liquor licences other fiscal years reported collections under this category. A specific revenue item for ‘vehicle parking fees’ exists in the database. 24 counties in 2014/15, 30 counties in Vehicle parking fees 2015/16, and 34 counties in 2016/17 reported collections under this category. A specific revenue item ‘advertisement’ exists. However, no counties reported collections under this category in any Advertisement fees fiscal year. Source: Database of audited county revenue collections. Recommendations: Counties should address audit recommendations on total collections. While they appear reasonably reliable, there are some data issues (e.g., incomplete reporting, reconciliation issues, spending at source) outlined in county audit reports. The reporting template for individual county revenue sources needs to be redesigned (e.g., to be GFS-based), with clear supporting guidelines for counties. Individual revenue items are reported inconsistently across counties and there is substantial aggregation in summary categories (e.g., miscellaneous/other charges), partially due to unsuitable reporting templates. Following review of the database, as part of a county questionnaire the team requested counties to submit unaudited revenue reports based on the counties’ individual revenue classifications used for county internal reporting. As of 8 June 2018, eight counties had submitted such revenue reports (Embu, Kericho, Kirinyaga, Kisumu, Kwale, Machakos, Makueni and Nairobi City). While these internal revenue reporting formats are inconsistent across counties, the reports provided useful breakdowns of county collections and were used to identify counties’ main sources and as a basis for calculating revenue gaps compared to the estimated potential. Recommendation: Use redesigned standard reporting template also for county-internal reporting. Overall, the lack of detailed actual revenue data disaggregated by main sources for the majority of counties has implications on the analysis. Data on total collections is available across all 47 counties, which allows revenue collection efficiency calculations through the frontier analysis. However, frontier analysis for individual revenue sources and calculation of respective revenue gaps is not possible at this stage. Following discussions of the initial study findings, the National Treasury embarked on another attempt to collect county revenue data based on a simplified data collection template focusing on the six identified main sources. This data can still be added later by counties or National Treasury when it becomes available. Recommendation: 42 Once available, data from counties on actual collections of the 6 key revenue sources can be entered into the designed models to carry out frontier analysis for individual revenue sources and systematically calculate revenue gaps compared to the estimated potentials. County revenue base (proxy) data Apart from detailed information on base definitions in county finance acts, only limited data on actual county revenue bases was available. Among the available data were land valuation rolls for 20 counties and building permits issued for Nairobi City County. However, such data appeared to be neither comprehensive for individual counties (e.g., land valuation rolls outdated with limited coverage and substantial gaps) nor available and comparable across counties. Recommendation (see also recommended actions under Section 2.5): Counties to maintain readily available databases, ideally harmonized across counties using standardized IT system(s). Counties should aim to institutionalize regular data sharing across relevant stakeholders, including counties, KRA, KNBS and relevant ministries (e.g., lands). We therefore mainly used micro-data drawn from KNBS household and enterprise surveys in order to estimate a consistent tax bases across counties. County finance acts have detailed descriptions of applicable rates and charges for each type of revenue stream but they only report limited substantial information regarding county-specific revenue bases. Additionally, existing information on revenue bases are not consistent across counties. For example, property valuation rolls are only documented across 20 counties while they have major drawbacks with respect to coverage (e.g. private and/or public land, some parts of the county are not covered at all) and completeness (e.g. high shares of missing values). Therefore, data from sources such as finance acts, MoL valuation rolls and city development plans are only used partly for the purpose of extrapolating key economic indicators across the counties. The proxies for revenue bases were constructed using three sources of micro-data: i) the National Housing Survey 2012/13, ii) the Kenya Integrated Household Budget Survey (KIHBS) 2015/16 and iii) the Small and Medium Enterprises Survey (MSME) 2016. Both the KIHBS and MSME cover all 47 counties in Kenya while the National Housing Survey covers 44 counties. Garissa, Mandera and Wajir are not covered at all because the sampling frame has not been developed yet at the time of the survey due to lack of security. All surveys adopt probability sampling designs which ensure that the samples are representative at the national and county level. The household surveys (KIHBS and the National Housing Survey) sampled households using stratified random sampling based on enumeration areas developed in the context of the 2009 Kenya Population and Housing Census. Sampling units (households) were selected randomly from two levels of clusters; across counties and in urban and rural areas within counties. The MSME survey sampled micro (<10 employees), small (10-49 employees) and medium (50-99 employees) sized enterprises based on different sampling designs depending on the licensing status of the businesses. For licensed businesses, a stratified random sampling design was adopted with units being drawn from registration lists based on ISIC codes. For unlicensed business, the sampling design resembles that of the household surveys. Conducting analysis at the micro-economic level, we applied the weighting factors to our estimates in order to arrive at representative proxies of revenue bases for each county. The weighting factors, which are included in each dataset reflect the representativeness of the sampled units (for example, households or establishments). Weighting is based on the probability of selection of each unit of observation both across and within (urban vs. rural regions) the regions. Each unit of observation is assigned a weighting factor which represents the actual population of units accounted for by this specific sampled unit. While census and survey data has been useful in estimating revenue bases for the various revenue sources, it should be noted that substantial variation between different datasets exists. While some of the variation can be explained by different scope, definitions and methodologies, the surveys do not appear to paint a consistent picture. Table 6 below shows variation for numbers of licensed businesses and employment in licensed businesses from different data sources. In the potential estimation models, we have selected the dataset that appeared to most comprehensively capture the revenue base (e.g., the MSME survey for employment numbers, which builds upon county business registries). Conservative assumptions have been used to capture revenue bases excluded from used survey datasets. 43 Table 6: Examples for variation in business data between censuses, surveys and taxpayer registries. Census of Enterprises, KRA non-individual MSME Survey, 2016 KIHBS 2015/16 2017 taxpayer database Licensed businesses Total 138,190 1,489,200 218,775 361,232 Micro (0-9) 98,669 1,370,913 Small (10-49) 35,501 107,345 Employment in licensed businesses Total 2,094,715 8,167,800 Further, county GDP estimates were available from a World Bank Policy Research Working Paper (2015)29. The paper uses satellite imagery from outer space for a night lights approach to estimate growth and levels of subnational 2013 GDP for 47 counties in Kenya. The night lights approach is independent of official data and can provide more accurate estimates of informal activity. However, the approach may underestimate economic activity in sectors that are largely unlit such as agriculture. For Kenya, Nairobi County is the largest contributor to national GDP (13%); however, its contribution is lower than generally expected. As the paper notes, to get a composite picture of subnational economic activity, it is important to estimate subnational gross domestic product using standard approaches (production, expenditure, income). While this was beyond this study KNBS is in the process of developing county GDP estimates. Lastly, a master plan study for urban transport in the Nairobi metropolitan area contained a car parking study for Nairobi’s central business district, providing insights into available parking units and average parking durations 30. Looking forward, an ongoing KNBS survey exercise gathering detailed information on economic activities at the county level and KNBS plans to develop detailed county GDP estimates will be useful additions to the available data and fill gaps by providing specifically county level information. While the county economic activity survey is a stand-alone exercise at this point, a possible future integration of such variables into the standard set of KNBS censuses and surveys could provide regularly updated information on revenue bases across counties. Recommendations: KNBS (in collaboration with other stakeholders and counties) to ensure that different surveys are complementary and paint together a realistic picture. Future censuses and surveys should capture relevant information that supports county revenue collections, ideally based on harmonized and simplified revenue sources and bases. County rate schedules The team reviewed selected county annual finance acts in order to map out the specific tax and fee rates promulgated by counties for the key revenue sources. Several county finance acts are available online and provided useful information on tax and fee rates for the revenue potential estimates, even if in many cases acts are outdated or only finance bills were available. For most counties, the finance acts provide extensive lists of taxes, fees and other charges (often more than 100 pages). While this is can be appropriate given the extensive list of services counties provide (e.g., in the area of primary health care), there appears to be still scope to group and/or simplify rate structures. Analysis of rate structures for property tax/land rates showed that rates are used in many counties to compensate for outdated valuation rolls. For example, land rates in Nairobi had climbed gradually to 14% of the unimproved site value in 1998 and further increased up to 34% in the 2015 county finance act. In comparison, Nairobi charges 0.12% on unimproved site value for properties with updated valuations. Overall, using rates to compensate for outdated valuations appears to be less than ideal. Even if such a system would adequately capture property value increases over time across counties compared to outdated valuations, it will likely make communication to taxpayers more difficult, both through the need to apply high rates and the significant variation across counties. 29 Bundervoet et al., 2015. Bright Lights, Big Cities—Measuring National and Subnational Economic Growth in Africa from Outer Space, with an Application to Kenya and Rwanda. Policy Research Working Paper 7461. World Bank: Washington, DC. 30 Japan International Cooperation Agency (JICA), Ministry of Roads and Public Works, Ministry of Local Government, 2006. The Study on Master Plan for Urban Transport in the Nairobi Metropolitan Area in the Republic Of Kenya. Final Report. Katahira & Engineers Internationalr Recs International Inc. 44 Figure 2: Land rates for selected counties between 1982 and 2000. Source: Ministry of Lands. A review of the extensive business licence rate schedules pointed to potential scope for simplification across business categories as well as generally higher rates for informal and small businesses compared to larger businesses (Table 6). Two arguments could theoretically support such a rate structure: (1) larger businesses are generally formalized and have therefore already a higher tax burden through national taxes on income, profits, and goods and services through VAT, and should therefore be taxed less at the local level; and (2) taxing informal businesses relatively higher may incentivize formalization. However, both points do not appear convincing in view of benefits from local services and perceived equity of the tax system. Table 7: Comparison of actual trading business licence fees by size of business for sample districts (Ksh). County Samburu Kilifi Nairobi Total fee Per employed Total fee Per employed Total fee Per employed Hawkers (w/o vehicle) 1,500 1,500 800 800 5,000 5,000 Temp/informal kiosk 1,500 1,500 2,400 2,400 4,000 4,000 Small trader (1-4) 2,000 500-2,000 3,000 750-3,000 5,000 1,250-5,000 Medium trader (6-20) 4,000 200-667 6,000 300-1,000 15,000 750-2,500 Large trader (21-50) 7,000 140-333 12,000 240-570 30,000 600-1,425 Mega (50-100) 7,000 70-333 12,000 120-240 80,000 800-1,600 Hyper (over 100) 21,000 Up to 210 36,000 Up to 360 120,000 Up to 1,200 Sources: Kilifi County Trade Licence Bill, 2016; Nairobi City County Business Permits Fee Schedule (http://eregulations.invest.go.ke/media/Single_Business_Permit.2.pdf, accessed on 12 June 2018); Samburu County Finance Act, 2013. Recommendations: Consider reducing the length of rate schedules by grouping and/or eliminating some charges As discussed in Section 2, consider harmonising land rates, updating valuations and review business licence schedules. Conclusions on data validity for revenue potential estimation Overall, substantial data gaps do undermine the revenue potential and gap analysis, warranting caution when interpreting specific results at the county level. However, broad findings on the scale of the potential of key revenue sources, including the broad distribution across counties, appear to be sufficiently reliable to allow conclusions and recommend policy and administrative reforms. Improving data availability and quality is one of the key recommendations to allow for more accurate estimates for the main revenue sources in the future. While the use of several data sources can result in some 45 inconsistency, only data that appeared reasonably reliable has been used in the analysis. Where this was not available, the report highlights data gaps and does not estimate revenue gaps. For example, the lack of detailed actual revenue data disaggregated by main sources for the majority of counties only allows frontier analysis to be carried out at the aggregate level and hinders the calculation of revenue gaps for individual revenue sources. The models built as part of this study allow for inserting county data once available to fill this gap. Ongoing data collection exercises on revenue bases at the county level are noted, and will provide useful information to refine revenue potential estimates and establish revenue gaps based on counties’ actual collection by main revenue sources. Going forward, efforts to improve data quality are important for evidence-base policy making in general and specific exercises such as this one. The study notes that significant efforts are being undertaken on a regular basis, including collation of county revenue data in databases and reports of national scope, as well as a multitude of censuses and surveys. Further, more information is available within national institutions and counties, which was not readily available for this exercise. This suggests that efforts should be undertaken more strategically to satisfy as many data needs as possible, i.e., producing not more but more useful data as well as improving data sharing/exchange of information should be a priority. Results from estimation of aggregate revenue potential As outlined above, the audited county revenue data appears to provide reliable information on total collections by county, allowing for calculation of revenue collection efficiencies and revenue gaps using frontier analysis. However, the results should be interpreted with care given the wide range of different county services that relate to different revenue bases as well as the fact that most county services do not have a primary revenue raising objective, with results potentially incentivizing socially and economically harmful collections. More detailed modelling, while beyond the scope of this study, would theoretically be possible but the purpose for this should be clear in view of the highlighted issues. Overall, separate modelling of counties’ main revenue sources, both top -down and through frontier analysis (as data becomes available) appear to be more useful. Frontier analysis: summary of model specifications and results As outlined above, Data Envelopment Analysis (DEA) is a standard frontier method and used here to estimate efficiency levels and aggregate potential at the frontier. 31 An output-oriented DEA model specification is used in this setting to answer the question ‘by how much can output quantities (i.e., county revenue collections) be expanded without altering the input quantities (i.e., the county revenue base) used?’. Table 8 provides an overview of model specifications applied as part of this study and resulting potential and revenue gap estimates. Overall, aggregate estimates of different DEA specifications arrive at comparable levels, also in comparison with results from the 2015 revenue baseline study 32. Table 8: Method 1 - Frontier analysis (Data Envelopment Analysis, DEA) model specifications and overview of findings. 2018 Data Envelopment Analysis 2015 DEA Input (data source) County GDP estimates County household County consumption, County income, (World Bank, 2015) consumption (KIHBS urbanization, education urbanization, education 2015/16) (KIHBS 2015/16) (unknown) Output Average of actual county collections over FY2015/16- Same excluding natural Actual county revenue for FY2016/17 (audited county revenue database) resource revenue 2014/15 Frontier county Weighted average for Laikipia Nairobi and Mombasa Unclear (none shows Isiolo, Kericho and relative efficiency) Baringo Frontier counties Narok (natural resource Narok (natural resource None None excluded due to revenue), Nairobi and revenue), Nairobi and unique features Mombasa (main hubs, Mombasa (main hubs, density) density) Revenue potential Ksh 55 billion Ksh 53 billion Ksh 66 billion Ksh 48 billion estimate (0.83% of GDP) (0.81% of GDP) (1.0% of GDP) (0.84% of GDP) Revenue gap 36% 35% 50% 30% Note: In cases where counties were excluded from being at the national frontier due to dominant unique features, such counties were considered to be at their (own) frontier, with actual revenue collections equalling potential. The collections are reflected in the total revenue potential estimate. 31 DEA models attribute all deviations from the frontier to inefficiency since they neither account for measurement errors nor ‘noise’. Thus, special caution should be taken when interpreting the estimates. 32 Office of the Controller of Budget, Republic of Kenya, 2016. County Revenue Baseline Study 2015. Nairobi. 46 Figure 3: County revenue potential based on frontier analysis, Ksh billion. 70 66 60 55 53 48 50 40 35 35 34 Actual collections 33 30 20 Revenue potential estimate 10 - 2018 DEA (GDP) 2018 DEA 2018 DEA (three 2015 DEA (three (consumption) inputs) inputs) Note: 2018 DEA with three input factors excludes collections from natural resources (significant for Narok county). While aggregate estimates of different DEA specifications, both from the 2015 baseline study and this study, arrive at comparable levels, distribution of revenue potential across counties varies across both studies (Table 9). This appears to be driven by the use of county income in the 2015 study compared to consumption or GDP in the 2018 model specifications. Since none of the counties’ main revenue sources are income based but largely based on economic activity and consumption (including rent), and partially also due to limited information on county income levels 33, consumption and county GDP were selected in the 2018 specifications. The counties with highest shares of total revenue potential are Nairobi, Kiambu Mombasa and Nakuru (all consistently among the highest four across the three 2018 specifications) as well as Nyeri and Meru counties. Table 9: Five counties with the highest share of total revenue potential for different DEA specifications. 2018 DEA (consumption, 2015 DEA (income, urbanization, 2018 DEA (GDP) 2018 DEA (consumption) # urbanization, education) education) 1 Nairobi 20% Nairobi 21% Nairobi 17% Nairobi 34% 2 Kiambu 10% Mombasa 6% Kiambu 6% Nakuru 7% 3 Nakuru 7% Kiambu 6% Nakuru 5% Mombasa 6% 4 Mombasa 6% Nakuru 5% Mombasa 5% Kiambu 6% 5 Nyeri 3% Meru 3% Meru 4% Narok 5% Top-down analysis: Results for main revenue sources This sub-section comprises top-down analysis for the six identified main revenue sources; however, revenue gap calculation is hindered by lack of disaggregated county revenue data. As outlined in the data section above, due to the incomplete and inconsistent actual revenue collection data by revenue source, comparing counties’ performance relative to their revenue bases, as is done in the frontier analysis, is not feasible. The basic Excel models prepared as part of this study allow the plugging in of county revenue data for the six selected sources include tabs for carrying out the frontier analysis and calculate efficiencies of different county revenue administrations in view of the proxy revenue bases estimated as described below. Summary of model specifications As outlined under 3.2 above, the top-down approach involves estimating the size of the revenue base for each OSR stream by county, and then applying the prevailing rate, or an alternative rate based on international practice or comparator countries, to calculate potential revenues. Estimated differences between the potential and the actual revenues collected by the counties reflect the possibilities for enhancement of revenues if existing revenue policies were implemented effectively. Both require simplifications and assumptions, which are by definition not reflective of the actual circumstances at the county level. The accuracy and verifiability of assumptions is subject to data availability and quality, and varies by revenue source modelled. The results should therefore be interpreted in view of the explicitly stated simplifications and assumptions. Table 10 below provides an overview of the model specifications. Further, potential estimates should be viewed as indicative only, with models for each source allowing adjustment of bases and rate structures based on policy preferences and if additional data becomes available. 33 KNBS reports and micro-surveys do not include comprehensive information regarding income by county. For example, while data on household incomes from rent, pensions and financial investment are included in KIHBS 2015/16, income from employment is not reported. 47 To calculate revenue gaps by county, the available county revenue collection datasets have been merged, selecting the maximum of annual collections by county over FY2014/15 to FY2017/18 (July-March only) from the database of audited county revenue collections and available unaudited county revenue reports, to arrive at a more comprehensive dataset. This data should be replaced by actual collections for FY2016/17 once received. Table 10: Overview of model specifications for top-down analysis. Revenue Actual Model base(s) Model rate structure(s) source collections Property tax Estimated value of residential and commercial properties Three scenarios: (i) 1% rate on all property values; (ii) excluding agricultural land, differentiated between low and 0.5% on property with values below and 1.5% above cut- high value properties (cut-off Ksh600,000) off; and (iii) exemption (0%) on values below and 1% above cut-off Building Estimated value of residential and non-residential Two scenarios: (i) 1% rate on all construction values, and permits construction based on national private sector construction (ii) 1% on value below and 2% above cut-off Maximum of output and relative shares of county construction, collections by differentiated between low and high value construction (cut- county over off Ksh200,000) FY2014/15 to Business FY2017/18 (July- Estimated full-time equivalent employees/self-employed by Base rate of Ksh2,000 per employee/ self-employed, licences March only) from county adjusted by a county multiplier based on average county database of consumption per adult relative to average national audited county consumption Liquor revenue Estimated full-time equivalent employees/self-employed Base rate of Ksh10,000 per employee/self-employed, licences collections and working in businesses serving alcohol by county adjusted by a county multiplier based on average county available consumption per adult relative to average national unaudited county consumption revenue reports Vehicle Estimated utilization of per capita car parking units in Base rate of Ksh150 per parking vehicle, adjusted by a parking fees Nairobi City extrapolated to core urban populations by county multiplier based on average county consumption county, with multiplier for bus parking per adult relative to average national consumption Outdoor Business expenditure on advertisement/marketing, adjusted Not applicable advertising by share of businesses spending primarily on fees outdoors/public advertising 3.7.2 Aggregate results Overall, results from the top-down revenue potential estimates show significant unrealized potential across the identified six main revenue sources, particularly from property rates (Table 11). This suggests that counties can gradually fund an increasing share of local service delivery from own source revenue (while intergovernmental fiscal transfers will continue to play an important role for local goods and services, particularly in health, education and infrastructure). Property rates have by far the largest potential, even if 90% of the lowest value properties are exempt. However, this is only the case if the revenue base and rate structures are updated in line with global practices that were applied in the model. Should such reforms be politically feasible in Kenya, prioritizing property tax collections and investing in strengthening the property tax system, in terms of regulatory frameworks, systems, human resource capacity and taxpayer education, could have substantial returns. These estimates also suggests that current cess collections could be replaced easily by collections from property rates in terms of potential, which would require refocusing of capacities with the benefit of phasing out any potentially harmful practices. It should be noted that revenue potential can only gradually be realized. It will be critical to (i) prioritize revenue sources to focus reform and administration efforts where these have highest returns, (ii) simplify and ideally harmonize revenue sources, bases and rate structures across counties, to allow easier administration, development of consistent regulatory frameworks and training of county administrations, (iii) explore efficient implementation arrangements, including through potential collaboration with the KRA (e.g., exchange of taxpayer information, potentially delegated collection on formal sector businesses), and (iv) strengthen administration capacities, both human resources and systems (e.g., ICT, business processes). 48 Table 11: Method 2 - Top-down revenue potential estimates for six main county revenue sources. Revenue source Total potential (Ksh) % of GDP Revenue gap Property tax: Only for 26 counties (i) assuming 90% lowest value properties exempt, 1% 66.2 billion 1.01% 91% with data rate others (ii) assuming 1% rate all 84.3 billion 1.28% 93% (iii) assuming 0.5% low value, 1.5% high value 108.3 billion 1.65% 94% Building permits Only for 8 counties (i) 1% on all construction value 6.0 billion 0.09% 35% with data (ii) 1% on low, 2% on high value 11.8 billion 0.18% 66% Business licences 23.4 billion 0.31% 75% Only for 41 counties with data Liquor licences 10.2 billion 0.14% 89% Only for 5 counties with data Vehicle parking fees 12.6 billion 0.17% 61% Only for 39 counties with data Outdoor advertising 6.3 billion 0.10% 83% Only for 8 counties with data OVERALL TOTAL RESULTS Low Scenario 125 billion 1.8% Medium Scenario 143 billion 2.1% High Scenario 173 billion 2.6% Total potential (high scenario) = 43% of total county budget Actual (all sources, average FY16-FY17) 35 billion 0.50% (FY16/17) Actual cess collections FY17 1.2 billion 0.02% Total County Budgets 399 billion 5.7% Disaggregated results by source and methodology Property tax/rates Taxation of property is a major source of own-source revenues for the counties. Thus far, only the land portion of the property is taxed. Under prevailing legislation, taxes on property are calculated on the basis of unimproved site value (USV) of the land rather than the capital value of property. Counties charge either a flat rate depending on the size of the land or define its own rates based on the land value. The future direction for property taxation is currently under discussion, with tax rates on capital improved value instead of USV being under consideration34. Using the land value as the tax base for property rates faces several challenges. First, existing valuation polls are inconsistent across countries. Second, it is conceptually and practically difficult to estimate land values when a building is already on site, as the value of the building is likely to substantially influence the value the land itself. Moreover, taxing land values (rather than values of the full property) might be seen as unfair and not commensurate with ability to pay compared to taxing the value of the property as a whole. Therefore, rates applied to combined value of land and immovable properties may be advantageous and we explore potential revenues from taxing total property values. In practice, basic proxies for total property value can be used and gradually improved, for example physical size of built (e.g., using square meter) and unbuilt land adjusted by ‘neighbourhood’ multipliers/zoning and banding. A detailed design study could be useful to determine most feasible approach in practice if reform of the current system is politically feasible. Finally, we are only looking at potential from taxes on residential and commercial properties, not considering potential revenues from taxes on agricultural land. A separate framework could be designed for taxes on agricultural properties (for example, to define whether market values of land or its current use should be taxed). International practice 34 See here for the policy document on enhancement of county governments’ own -source revenues issued by KRA: http://treasury.go.ke/publications/bills-acts-agreements/category/161-county-governments.html?download=738:draft-national-policy- to-support-enhancement-of-county-own-source-revenue. 49 In Kenya collection of revenues is modest compared to other non-OECD middle-income countries. According to IMF’s technical assistance report on property rates in Kenya, average revenues from property taxes during the period between 2004 and 2010 amounted to 0.15% of GDP which is much lower compared to average importance of property tax revenues in middle income countries accounting for 0.76% of GDP during the same period 35. Moreover, collection of property taxes is particularly important for the generation of domestic revenues at the regional (or county) level. In Kenya, the tax base for property rates is assessed based on either the value of the land itself without considering the value of its use (e.g. buildings) or the size of the land. Internationally, alternative policy frameworks are implemented which may result in higher revenues at comparable tax rate structures. In the majority of developed countries, such as Canada, Germany and the United States, property is taxed with respect to its total value, including land and building values. This type of property taxation is also implemented in emerging economies, such as Brazil, Argentina and South Africa. Particularly, in South Africa, generation of revenues from property rates is quite high (1.8% of GDP on average during the 2004-2010 period)36 suggesting that taxing property value instead of land value is likely to result in higher revenue potentials for Kenya. Taxes on capital improved values of land (land values and building values) are also implemented in Liberia, Gambia, Ghana and Tanzania. For example, in Ghana, local authorities implement tax rates on estimated value of property which vary with respect to the classification of the area where the property is located. The applicable rates ranged from 0.5 to 3%,37 being roughly higher compared to the international standards in property rates, ranging from 0.5% to 2% of market values 38,39. Methodology Given that valuation polls are dated and not consistent across the counties, we used self-reported data drawn from household and business surveys to infer the present value of properties. Based on dwelling owners’ self -assessments of the present value of their dwellings and the rental income they would earn if they rented their property reported in KIHBS 2015/16, we calculated national average returns to residential property. Average rental yields for residential properties amount to 6.7% of the total value of the property. The estimated national average level of rental yields was then used to calculate the present value of properties in the KIHBS subsample of renters who reported their monthly expenditure on rent. Missing property values (5%) due to lack of information both on present value of dwelling or rental expenditure were imputed using a linear model to predict property values based on the following characteristics: i) number of habitable rooms, ii) area of residency (urban/rural), iii) type of dwelling (e.g. bungalow, flat, maisonette, shanty, manyatta) and iv) building materials for roof, wall and floor. Assuming that rental yields do not differ across residential and commercial properties, we applied national average rental returns to rental expenditure data drawn from MSME 2016 to arrive to estimates of commercial property values. 25% of the observations were zero implying that the owners of the establishments did not pay any rent to use it. Zero values were imputed using gross-up factors to provide high-level estimates of total commercial property value for each county. Key findings: Estimated value of tax base The selected tax base for property rates is total value of property, including the value of buildings and land. County- specific tax bases for property taxation vary with respect to the urbanisation levels of the counties. Figure 4 presents the distribution of total value of residential and commercial properties across Kenya counties. Property values in counties which are predominantly rural, such as Isiolo, West Pokot and Samburu, are much lower compared to the value of residential and commercial properties in urban counties with big cities, such as Nairobi, Kiambu and Mombasa. Properties in the top ten counties (Nairobi, Kiambu, Mombasa, Kajiado, Nakuru, Machakos, Kisumu and Meru) account for 70% of the total property values in Kenya. The estimated value of residential and non-residential properties in Nairobi, which is not included in the chart, accounted for 40% of total property value at the country level (app. 3.4 billion in 2015/16). On the contrary, the sum of property values in the bottom ten counties (Bomet, Mandera, Elgeyo Marakwet, Tharaka- Nithi, Marsabit, Tana River, Wajir, Isiolo, West Pokot and Samburu) is much lower, accounting to just 2% of the total value of property at the national level. Notably, despite being a rural county, Makueni is ranked among the top 10 counties with highest property values, based on data from KIHBS 2015/16 and MSME 2016. In Makueni, self-reported values of residential properties before any 35 IMF (2015), Kenya: Improving Property rates within the unfolding fiscal decentralization framework. 36 IMF (2015), Kenya: Improving Property rates within the unfolding fiscal decentralization framework. 37 http://www.ipti.org/wp-content/uploads/2015/09/IPTI-Xtracts-Ghana-April-20161.pdf 38 ICG, 2017. Land and property taxes: exploiting untapped municipal properties. Available at: https://www.theigc.org/wp- content/uploads/2017/08/Land-and-property-taxes-Exploiting-untapped-municipal-revenues.pdf 39 IMF (2015), Kenya: Improving Property rates within the unfolding fiscal decentralization framework. 50 imputation of missing data amounted to 122 billion and were 4th in the ranking of counties with respect to residential property values. Moreover, total value of commercial properties amounted to approximately 4 billion. Figure 4: Estimated total value of residential and commercial properties across counties, Ksh billion, 2015/16. Kiambu 706 Mombasa 534 Kajiado 392 Nakuru 335 Machakos 198 Kisumu 194 Meru 167 Kilifi 164 Makueni 144 Uasin Gishu 142 Narok 133 Kakamega 131 Trans Nzoia 119 Nyeri 115 Turkana 106 Murang'a 101 Kisii 95 Kwale 87 Nyandarua 77 Bungoma 77 Kirinyaga 76 Kitui 74 Kericho 69 Siaya 61 Embu 59 Laikipia 56 Busia 53 Taita/Taveta 52 Homa Bay 51 Nyamira 50 Nandi 46 Migori 43 Vihiga 37 Garissa 28 Lamu 27 Baringo 27 Bomet 27 Mandera 26 Elgeyo/Marakwet 23 Tharaka-Nithi 21 Marsabit 19 Tana River 15 Wajir 13 Isiolo 13 West Pokot 12 Samburu 11 Note: Nairobi is not included in the chart and accounted for 40% of total property value at the country level (approximately Ksh3.4 billion in 2015/16). Values of residential and commercial properties also vary within counties with rural counties having fewer high-value properties compared to urban counties. Overall, 90% of households in KIHBS 2015/16 and 90% of establishment owners in MSME 2016 reported that their residential property or business site was worth less than 600,000 KShs. We used this cut-off point to explore potential revenues from applying different rates to high- and low-value residential and non- residential properties. Figure 5 shows the distribution of high and low value properties across the counties. In Nairobi, total estimated value of properties in the low-value group (around 326 billion) amounts to 10% of total property, which is the lowest share among all Kenya counties. Baringo is the county with largest share of low-value properties, with residential and commercial properties in the low-value group (being worth less than 600,000 KShs) represent 60% of county-total property value. 51 Figure 5. Value of properties in the low-value group (<600,000 KShs) across Kenya counties Low-value property (% of total property value) Nairobi Kajiado Kiambu Mombasa Lamu Trans Nzoia Turkana Narok Kakamega Kisumu Kwale Makueni Marsabit Bungoma Uasin Gishu Machakos Kilifi Busia Garissa Nakuru Kisii Kericho Laikipia Taita/Taveta Embu Mandera Isiolo Nandi Kitui Vihiga Kirinyaga Nyeri Nyamira Murang'a Tharaka-Nithi Wajir Nyandarua Elgeyo/Marakwet Migori Tana River West Pokot Siaya Homa Bay Samburu Bomet Meru Baringo 0% 10% 20% 30% 40% 50% 60% 70% Key Findings: Revenue potential According to Nairobi City’s financial act, a rate of 0.12% is currently applied on newly assessed unimproved land values. We used self-assessments of the value of construction of dwellings reported in the National Housing Survey 2012/13 to estimate the value of land as a share of the present value of residential property. In the weighted sample, costs of dwelling construction, reflecting the capital improved value of the property, were found to account for 40% of total property value. Assuming that total value of property comprises land and buildings value, the remaining share of total property value is attributed to unimproved value of land (60%). If the existing rate of 0.12% was applied on estimated land values across counties, total revenues would be quite low, accounting for only 0.09% of national GDP in 2015/16. This is only slightly higher compared to the current contribution of property taxes to domestic revenues (0.05% of national GDP in 2015/16). Globally, revenues from recurrent taxes on immovable property contribute substantially to domestic resource mobilization, particularly at the subnational level, with the average for OECD countries at around 0.9% (recurrent property taxes only, excluding other property taxes, e.g., on wealth or transactions) and the range for OECD countries being up to 3% of GDP 40. Therefore, we estimate revenues from implementing alternative tax systems (e.g. progressive taxation) and rates in order to explore an extended range of potentials for enhancing domestic revenues for the counties. If the internationally common rate of 1% (scenario 1) is applied uniformly on total value of property (including land and building values), potential revenues amount to 1.28% of national 2015/16 GDP. Progressive taxation of property is likely to result in higher potential revenues. For example, a progressive tax system, with a rate of 0.5% being applied to low- value properties and 1.5% to high-value properties (scenario 2) would result in potential revenues amounting to 1.65% of national GDP. Moreover, if low-value properties were exempt from taxes and a 1% rate was applied only to properties in the high-value group, total potential revenues would amount to 1% of 2015/16 GDP. Applying taxes only to the 10% most 40 See here for detailed data on property rates across OECD countries: https://data.oecd.org/tax/tax-on-property.htm 52 valuable properties would result in estimated revenue potential reductions ranging between 35% and 55% for counties with low shares of high-value properties.41 Figure 6. Estimated property tax revenue potential by county under different rate scenarios (in Ksh million) - 10,000 20,000 30,000 40,000 50,000 60,000 Nairobi Kiambu Mombasa Kajiado Nakuru Machakos Kisumu Meru Kilifi Makueni Uasin Gishu Narok Kakamega Trans Nzoia Nyeri Turkana Murang'a Kisii Kwale Nyandarua Bungoma Kirinyaga Kitui Kericho Siaya Embu Laikipia Busia Taita/Taveta Homa Bay Nyamira Nandi Migori Vihiga Garissa Lamu Baringo Bomet Mandera Elgeyo/Marakwet Tharaka-Nithi Marsabit Tana River Wajir Isiolo West Pokot Samburu Potential estimate (0.5% low value, 2% high value) Potential estimate (1% all properties) Potential estimate (0% low value, 1% high value) In all scenarios, all counties have the potential to significantly increase their revenues through property taxes. Table 12 presents highest unrealized potentials for a list of the top 9 counties with highest revenue potential in scenarios 1 (1% rate applied uniformly) and 2 (0.5% for low-value properties and 1.5% for high value properties). As shown in figure 6, urban counties, such as Nairobi, Kiambu and Kajiado, where properties have higher values compared to rural counties, will benefit more by implementing a progressive system of property taxation as the one outlined in scenario 2. Table 12 41Changes in estimated potential revenues under the tax exemption for low-value properties scenario are calculated using scenario 2 (0.5% for low-value properties and 1.5% for high-value properties) as the reference point. 53 presents further details regarding unrealised potentials and revenue gaps for the counties with highest estimated potential revenues from property taxes. Table 12: Potential revenues from property taxes for selected counties (in million KShs) Scenario 1 (1%) Scenario 2 (0.5%, 1.5%) County Actual Collections Revenue Revenue Unrealised Revenue Unrealised (max Potential Potential Revenue Gap potential Gap potential FY2015- (estimate) (estimate) FY2018) Nairobi 3,160 34,244 31,084 91% 48,104 44,944 93% Kiambu -a 7,061 7,061 100% 9,357 9,357 100% Mombasa -a 5,347 5,347 100% 7,080 7,080 100% Kajiado -a 3,920 3,920 100% 5,459 5,459 100% Nakuru -a 3,350 3,350 100% 3,804 3,804 100% Machakos 184 1,974 1,791 91% 2,301 2,117 92% Kisumu 145 1,941 1,797 93% 2,393 2,248 94% Meru 37 1,669 1,632 98% 1,573 1,536 98% Kilifi 213 1,643 1,430 87% 1,897 1,685 89% a Actual collections unknown. The unrealized potential and revenue gap may therefore be smaller than shown here. For the counties for which reasonably reliable data for actual revenue collections from property taxes is available, either from detailed county revenue reports or from the audited county revenue database (category: land/poll rates), tax gaps range between slightly above 80% up to almost 100% of estimated revenue potential (Error! Reference source not found.). Kilifi county appears to be best performing based on the potential estimates; however, actual collections only equal about between 20% in the lowest rate scenario (gap of about 80%) and 12% in the highest yield scenario (slightly over 88% gap). The data shows that across counties substantial revenue potential remains unrealized, which is highly likely also the case for most, if not all other counties for which actual collection data is missing. Figure 7: Estimated property tax revenue gaps by county under different rate scenarios (only counties with data on property tax collections) 80% 82% 84% 86% 88% 90% 92% 94% 96% 98% 100% Kilifi Machakos Nairobi Kericho Kisumu Uasin Gishu Kwale Samburu Nyamira Kirinyaga Bungoma Murang'a Lamu Taita/Taveta Nandi Tana River Mandera Embu Meru Kakamega Trans Nzoia Makueni West Pokot Busia Homa Bay Gap estimate (0.5% low value, 1.5% high value) Gap estimate (1% all properties) Gap estimate (0% low value, 1% high value) 54 Building permits Charges for approvals of building permits and plans collected by counties are important sources of OSR. In Nairobi, fees for building permits are defined at 1% of the estimated cost of construction based on various pre-defined rates for different types of buildings (e.g. residential properties, retail outlets, industrial sites, hotels, sports facilities etc.). Similarly, in Mombasa, a 1.01% fee is applied to the total value of construction for single-storey buildings with slightly lower rates being applied to buildings with additional floors. The applicable rates for non-residential buildings are 1.5% for commercial buildings and 2% for industrial sites. In other counties, fees are applied with respect to the size and the use of the buildings. For example, in Uasin Gishu, pre-defined charges vary with respect to number of floors and use of the building with higher fees being applied to construction of commercial buildings. Charges for building permits and approval of building plans contribute substantially to domestic revenues for the counties. 25 of the counties collect fees for either building permits or plan approvals while there are some counties which implement both types of fees. For example, only fees from building permits are collected in Nairobi city while in Mombasa, fees for both permits and plan approvals are levied. However, there appear to be 22 counties which do not collect these type of charges at all, though it is difficult to say with certainty this may be down to inconsistent reporting – i.e. revenues from these charges recorded under a different category. 42 Methodology Given that we could not distinguish between different types of properties based on available county data, we used the 2012/13 National Housing Survey to estimate the value of construction of residential dwellings between 2008/09 and 2012/2013. Grossing-up factors based on the ratios between the values of residential and commercial properties calculated using KIHBS 2015/16 and MSME 2016 data were applied to provide high-level projections of total construction value of residential and non-residential properties for each county. Estimated construction value relied on valid observations of construction costs reported by owners of dwellings who had constructed their houses as reported in the National Housing Survey. The three counties which were not covered in the survey (Garissa, Mandera and Wajir) were among the bottom group of counties with lowest values of residential and non- residential properties. In order to impute the missing values, we calculated average construction value as a share of total residential and non-residential building stock for the bottom ten group of counties. Then, we applied this adjustment factor to total property value for each of the three counties to arrive to high-level estimates of annual construction value of residential and non-residential properties for each of the counties. Data limitations did not allow for a precise estimation of the construction value of buildings used for residential purposes. For example, half of the sampled households reported that their dwelling was built long before 2008. Moreover, values of construction were often misreported as respondents who had built their own houses tend to underestimate the value of formal and informal work. Therefore, we did not depend on estimated construction values using self-reported data from the National Housing Survey 2012/13 to arrive to reliable proxies for building permits tax bases. Instead, we used these estimates to calculate county- specific shares of total value of annual construction at the national level. Then, we applied these adjustment factors to total output of construction in 2016 after adjusting for government construction expenditure drawn from the 2017 Statistical Abstract issued by KNBS.43 Key findings Tax base The selected tax base for building permits is annual costs of residential and non-residential construction. In counties with big urban centres, such as Nairobi, Mombasa and Kisumu, the revenue base for charges on building permits is quite broad compared to counties with high shares of rural population. As shown in Figure 7, our estimated sizes of residential and non-residential construction are in line with information reported at the 2017 Statistical Abstract pointing to counties in the Nairobi Metropolitan Area (Kajiado, Kiambu, Machakos and Nairobi) as being top-counties for urban development. Kisumu, Mombasa, Meru, Siaya and Kakamega are also among the top ten counties with highest total value of new buildings. According to our estimates, total construction value in the top ten counties accounts for more than 60% of total construction size in Kenya, with 20% of total size of construction taking place in Nairobi alone. 42 These counties are: Kwale, Tana River, Garissa, Marsabit, Isiolo, Tharaka- Nithi, Kitui, Nyandura, Nyeri, Kiambu, Turkana, Samburu, Trans Nzoia, Elgeyo Marakwet, Nandi, Baringo, Narok, Bomet, Bungoma, Busia and Migori. 43 https://www.knbs.or.ke/statistical-abstract-2017/ 55 Figure 8. Total value of residential and commercial construction by county - KSh. Total annual value of residential and non-residential construction Nairobi Machakos Kisumu Mombasa Kiambu Meru Siaya Kakamega Nyeri Kajiado Muranga Uasin Gishu Bungoma Bomet Makueni Kericho Homa Bay Kitui Trans-Nzoia Elgeyo/Marak Nyamira Migori Nakuru Kirinyaga Kisii Busia Taita Taveta Kilifi Embu Baringo Laikipia Kwale Garissa Nandi Mandera Narok Tharaka Nyandarua Isiolo Turkana Marsabit West Pokot Vihiga Wajir Tana River Samburu Lamu 0 20,000,000,000 40,000,000,000 60,000,000,000 80,000,000,000 100,000,000,000 While in counties with big cities and towns rapid urbanisation results in approvals of expensive construction plans, in rural counties the size of construction is rather limited. According to our estimates, the bottom ten group of counties with lowest values of annual construction mainly comprises rural counties. Total value of residential and non-residential construction in the bottom ten counties (Lamu, Samburu, Tana Rive, Wajir, Marsabit, Turkana, Isiolo and Nyandarua) accounts for only 2% of total construction at the national level. The distribution of annual construction value within counties is also influenced by the level of urbanisation in each county. In the National Housing Survey subsample of dwelling owners who had paid for the construction of their own property from 2008 and onwards, 90% of households reported that their properties had costed less than 200,000 KShs. Despite the possibility that dwelling owners underestimate construction costs, we use this threshold to provide rough estimates of potential revenues from applying different charges to new building approvals depending on construction costs. The value of new residential and non-residential buildings belonging to the low value group is around 6% of the total value of annual construction across the counties. In counties with large urban centres, such as Machakos and Mombasa, the value of new buildings as a share of total county construction is less than 6% while in Nairobi there were no buildings costing less than 200,000 KShs in the National Housing Survey subsample. On the contrary, in rural counties, such as West Pokot, Vihiga, Turkana and Nyandarua, the values of new buildings in the low value group represent more than 40% of county total construction value. More specifically, in Tana River, Samburu and Lamu, there were no households responding that their dwellings costed more than 200,000 KShs in the National Household Survey subsample. Potential revenues If the existing 1% rate, which is the charge for building plans currently collected in Nairobi based on pre-defined rates, was uniformly applied on annual construction value of residential and non-residential properties, total revenues from building permits and plans would amount to 0.09% of 2015/16 GDP (app. 6 billion KShs). This is approximately three times higher compared to actual collections reported by 9 counties (Embu, Kericho, Kirinyaga, Kisumu, Kwale, Machakos, Makueni and Nairobi) amounting to 1.8 billion. A progressive system of fees could also be applied. Counties could benefit from such a system by increasing their revenues from approving expensive building plans without placing financial burdens to the construction of new residential or commercial buildings of lower value. Increasing the rates for new buildings with construction costs which exceed the 200,000 KShs threshold to 2% while retaining the 1% rate for construction of buildings in the low-value group would double total revenues, which would then amount to 0.18% of 2015/16 GDP. 56 Figure 9. Estimated building permit revenue potential by county under different rate scenarios (in Ksh million) - 500 1,000 1,500 2,000 2,500 Nairobi Machakos Kisumu Mombasa Kiambu Meru Siaya Kakamega Nyeri Kajiado Murang'a Uasin Gishu Bungoma Bomet Makueni Kericho Homa Bay Kitui Trans Nzoia Elgeyo/Marakwet Nyamira Migori Nakuru Kirinyaga Kisii Busia Taita/Taveta Kilifi Embu Baringo Laikipia Kwale Garissa Nandi Mandera Narok Tharaka-Nithi Nyandarua Isiolo Turkana Marsabit West Pokot Vihiga Wajir Tana River Samburu Lamu Potential estimate (1% low value, 2% high value) Potential estimate (1% all values) Figure 8 presents estimated potential revenues from building plans and permits for two alternative scenarios: uniform rate of 1% (scenario 1) and graduated rates with increased rates for construction plans of higher value (scenario 2: 1% for the plans in the low-value group and 2% for the plans in the high-value group). Evidently, Nairobi is an outlier with much higher estimated potential revenues compared to the rest of the counties. Implementing scenario 2 would result in revenues from building permits amounting to approximately 2.3 billion, which is far more than double than the estimated potential revenues for almost all urban and rural counties. 44 44 In Nairobi, actual collection of revenues from building permits is already very high (1.7 billion), accounting for more than 90% of total reported counties revenues (including 9 counties). While uniformly implementing the rate of 1% would not result in substantial 57 Urban counties, where development plans of high value are approved on an annual basis, would benefit more by implementing graduated rates (scenario 2). In Nairobi, Machakos, Kisumu, Mombasa and Kiambu, estimated potential under scenario 2 are much higher compared to scenario 1. As shown in Table 13, a transition from scenario 1 to scenario 2 would result in a 138% increase in unrealised potential revenues in the case of Machakos county while for Kisumu, implementation of scenario 2 would result in a 95% increase in unrealised potential compared to scenario 1. Table 13: Potential revenues from building permit fees for selected counties (in million KShs)a Scenario 1 (1%) Scenario 2 (1%, 2%) County Actual Revenue Unrealised Revenue Revenue Unrealised Revenue Collections Potential potential Gap Potential potential Gap (max over (estimate) (estimate) FY2015- FY2018) Embu 6 68 62 92% 121 115 95% Kericho 8 98 89 91% 174 165 95% Kirinyaga 5 81 76 94% 135 130 97% Kisumu 43 525 482 92% 987 944 96% Kwale 5 41 36 89% 67 63 93% Machakos 225 724 500 69% 1,414 1,190 84% Makueni 6 100 94 94% 177 171 96% Nairobi 1,654 1,163 -490 -42% 2,327 673 29% a Covers 8 counties that submitted detailed revenue reports. Negative values imply actual collections exceed potential estimate. Figure 10 visualizes the revenue gap for the eight counties for which data on actual collections is reasonably reliable. In the lower rate scenario, a negative revenue gap for Nairobi implies that collections are, in practice, higher than the potential estimate, which changes in the higher rate scenario and results in a revenue gap of almost 30%. With this, Nairobi is leading in realizing the revenue potential from building permits, while for other counties revenue gaps are substantial and range between about 70% and 97%. Figure 10. Estimated building permit revenue gap by county under different rate scenarios (only counties with data on building permit collections) -60% -40% -20% 0% 20% 40% 60% 80% 100% Nairobi Machakos Kwale Kericho Embu Kisumu Makueni Kirinyaga Revenue gap (1% all values) Revenue gap (1% low value, 2% high value) Business licences To create a revenue base proxy for business licences, business physical size (e.g., in square metres), rental value, number of businesses, and employment numbers were considered. Data on physical size of businesses was not available but counties could use this in the future, for example based on their registry for property taxes/rates, for presumptive business licence taxation assessments. While the MSME Survey 2016 includes businesses self-reported rental expenditure (excluding large businesses with 100 and more employees), anecdotal evidence from KRA rental income tax assessments suggests a relatively high complexity, which could pose difficulties for most counties. (Inconsistent) business numbers are available from the KRA non-individual taxpayer database, the Census of Establishment 2017 and the MSME Survey 2016, with the latter two providing some business size information by employment ranges. Employment information is also available from the Census of Establishment 2017 and the MSME Survey 2016, with the latter appearing increases in revenues, a progressive system of fees would contribute in enhancement of own-sources revenues from new building approvals. 58 more comprehensive for micro and small businesses that make up the majority of the revenue base for the vast majority of counties. In practice, counties apply charges to different types of businesses and for different business sizes, with employment numbers used as a proxy for size of trading businesses. While business numbers by employment ranges and employment could both be used to model proxy revenue bases, exact employment numbers appear to be more closely reflective of business activity than employment ranges, which charges, e.g., businesses with 6 and 20 employees the same. Businesses’ self -reported employment data from the MSME Survey, 2016 was therefore selected to model revenue bases by county. Since the survey does not cover data on large businesses with 100 and more employees the resulting proxy revenue base is assumed to reflect a conservative estimate. To arrive at full-time equivalent positions, available employment data was weighted, with working owners and full time employees counted fully (100%), part-time employees half (50%) and casual workers one-quarter (25%). In the future, business licences (for businesses in the formal sector) could also be linked to corporate income or turnover which would more accurately reflect ability to pay. Such a revenue base would also likely require agreement by the national government (given that these would be taxes on income or sales), and may require changes to the legal framework but could be efficiently administered by KRA together (e.g., as a surcharge) with their ongoing corporate income and/or VAT collections. Such an approach could also facilitate gradual formalization of businesses, e.g. counties could collect business licence fees from micro and small businesses, and notify such businesses and KRA that appear to have surpassed KRA thresholds for filing returns and registering for VAT. KRA could then take over collections in behalf of counties (e.g., as a small surcharge on corporate income or VAT) and remit revenue to counties while counties focus on (presumptive) business licence taxation of micro and small businesses, which builds on counties’ comparative advantage having more local knowledge and staff based at the county level compared to KRA (complementing roles). An annual base rate of Ksh2,000 per full-time position has been assumed while in practice counties differentiate charges by business categories and sizes, for some of which using employment number ranges as a proxy for size. The base rate is adjusted by a county multiplier reflecting relative household consumption levels across counties derived from KIHBS 2015/16 (average consumption in county divided by average national consumption per adult). The underlying assumption is that higher consumption levels reflect higher turnover and thus ability to pay of businesses (plus overall higher price levels). In practice, business licence rates vary by county, e.g., Nairobi rates are substantially higher than rates in poorer counties, which is broadly reflected in the modelled rate structure (Table 14). Further, as discussed under 3.3 above, in practice counties appear to charge hawkers and smaller, informal businesses higher, and medium and large businesses lower than in the modelled fee structure. Table 14: Comparison of actual licence fees with modelled fee structure (Ksh) Samburu Kilifi Nairobi Total fee Per employed Total fee Per employed Total fee Per employed Hawkers (w/o vehicle) 1,500 1,500 800 800 5,000 5,000 Temp/informal kiosk 1,500 1,500 2,400 2,400 4,000 4,000 Small trader (1-4) 2,000 500-2,000 3,000 750-3,000 5,000 1,250-5,000 Medium trader (6-20) 4,000 200-667 6,000 300-1,000 15,000 750-2,500 Large trader (21-50) 7,000 140-333 12,000 240-570 30,000 600-1,425 Mega (50-100) 7,000 70-333 12,000 120-240 80,000 800-1,600 Hyper (over 100) 21,000 Up to 210 36,000 Up to 360 120,000 Up to 1,200 Consumption relative to 52% 98% 199% national average Assumed charge 1,030 per employed 1,965 per employed 3,980 per employed Sources: Kilifi County Trade Licence Bill, 2016; Nairobi City County Business Permits Fee Schedule (http://eregulations.invest.go.ke/media/Single_Business_Permit.2.pdf, accessed on 12 June 2018); Samburu County Finance Act, 2013. Key findings Revenue base estimates show high full-time equivalent employee and working owner in urban centres, with Nairobi accounting for about one-third of the total estimated revenue base (Figure 11). In contrast, predominantly rural counties have low revenue bases. Based on the modelled rate structure, this pattern is intensified in the revenue potential estimates, with half of total revenue potential in Nairobi County (Ksh11.8 billion of Ksh23.4 billion aggregate potential). Other urban centres follow with Nakuru (Ksh1.9 billion), Machakos (Ksh1.4 billion), Kiambu (Ksh1.0 billion), Kisumu (Ksh904 million) and Kakamega (Ksh686 million) also showing substantial revenue potential. The revenue gap among the 41 counties that separately reported business licence collections is relatively high at 75%, highlighting that under the modelled assumptions, substantial revenue potential remains unrealized at the moment, both 59 due to the modelled higher rates for larger businesses and the larger revenue base likely capturing a substantially higher share of the informal employment and self-employed working owners. A range of smaller counties including Muranga and Kitui appear to make good use of the business licences for revenue raising compared to their estimated potential (Figure 11). On absolute terms, Kiambu appears to efficiently administer business licences, collecting about 60% of their estimated potential. Figure 11: Estimated total full-time equivalent employees and working owners across Kenya counties Nairobi Nakuru Machakos Kakamega Kisumu Kiambu Kajiado Mombasa Meru Uasin Gishu Narok Nyeri Homa Bay Kericho Nyamira Siaya Migori Bungoma Kilifi Kisii Baringo Busia Embu Kirinyaga Wajir Taita Taveta Makueni Laikipia Mandera Kwale Trans Nzoia Nyandarua Murang'a Bomet Kitui Nandi Turkana Vihiga Samburu Lamu Tharaka Nithi Elgeyo Marakwet Garissa Isiolo Marsabit West Pokot Tana River - 500,000 1,000,000 1,500,000 2,000,000 2,500,000 3,000,000 3,500,000 Error! Reference source not found. shows revenue gaps for business licences for counties with reasonably reliable data on actual collections. Collections of several counties are in practice higher than potential estimates, which can at least in parts be explained by the higher rate structure some counties apply in comparison to the conservative assumptions of the model configuration applied that are discussed in Error! Reference source not found. above. All of these counties have also generally low total potential, as shown in Error! Reference source not found. above. For most other counties, however, revenue gaps are positive and span between 8% and 93%. Several counties with substantial potential have large unrealized potential, such as Kisumu (gap of 88%), Nairobi (gap of 85%) and Nakuru (78%). 60 Figure 12: Estimated business licence revenue gaps by county (only counties with data on business licence collections) Nairobi 11,837 Nakuru 1,923 Machakos 1,447 Kiambu 1,006 Kisumu 904 Mombasa 704 Kakamega 686 Kajiado 515 Meru 418 Nyeri 373 Narok 328 Uasin Gishu 323 Homa Bay 227 Kericho 205 Kilifi 204 Nyamira 186 Siaya 183 Bungoma 163 Migori 157 Kisii 154 Baringo 134 Embu 126 Kirinyaga 125 Taita/Taveta 98 Laikipia 82 Makueni 81 Busia 70 Trans Nzoia 67 Kwale 67 Murang'a 66 Nyandarua 66 Wajir 59 Mandera 48 Bomet 45 Nandi 45 Kitui 45 Lamu 41 Vihiga 34 Tharaka-Nithi 34 Turkana 32 Samburu 23 Elgeyo/Marakwet 17 Isiolo 14 Garissa 14 Marsabit 6 Tana River 6 West Pokot 6 61 Figure 13: Estimated building licence revenue gaps by county (only counties with data on business licence collections) Marsabit -170% Murang'a -128% Kitui -126% Tana River -43% Turkana -37% West Pokot -32% Nyandarua -28% Trans Nzoia 8% Kwale 12% Makueni 18% Kirinyaga 20% Elgeyo/Marakwet 24% Embu 29% Garissa 30% Nandi 32% Uasin Gishu 32% Kiambu 39% Busia 39% Vihiga 40% Mombasa 43% Bungoma 44% Taita/Taveta 52% Kisii 57% Migori 64% Baringo 65% Kilifi 68% Meru 69% Nyamira 72% Nyeri 75% Kericho 76% Nakuru 78% Mandera 79% Samburu 80% Lamu 80% Siaya 82% Nairobi 85% Machakos 86% Homa Bay 87% Kakamega 87% Kisumu 88% Narok 93% Table 15: Counties with highest (unrealized) revenue potential from liquor licences, Ksh million. Actual collections Revenue potential Unrealized potential County Revenue gap (max FY15-FY18) (estimate) (estimate) Nairobi City 1,814 11,837 10,023 85% Nakuru 430 1,923 1,492 78% Machakos 199 1,447 1,248 86% Kisumu 109 904 794 88% Kakamega 91 686 594 87% Kiambu 617 1,006 388 39% Kajiado Unknown 515 Unknown Unknown Total (41 counties with 22,642 5,598 17,044 75% data only) (All: 23,392) 3.6.2.4 Liquor licences Counties are constitutionally mandated to regulate local entertainment including licensing venues for selling liquor, and offering gaming and betting services. Liquor licences provide a means to limit and oversee the number of venues selling alcoholic beverages, facilitating better control over public health and security issues including limiting the sale of alcohol to minors and detecting quality issues. Liquor licences can also be seen to correct for a share of negative externalities from alcohol consumption on health and public order. Methodology 62 To create a revenue base proxy for liquor licences, the number of businesses engaged in the production, trade and sale of alcoholic beverages, and employment numbers in businesses engaging in food and beverage serving activities were considered. Only limited information on business numbers by ISIC classification and county were available from a KRA non-individual taxpayer database; however, numbers were low (likely only larger businesses covered) and no information on business size was available. Self-reported employment numbers for businesses engaged in food and beverage serving activities from the MSME Survey, 2016 were therefore selected to more closely reflect business sizes in absence of other data. No data on manufacturing, wholesale and retail of alcohol were available, and large businesses with 100 and more employees are also not covered. The proxy revenue base therefore reflects a conservative estimate. In the future, liquor licences could also be linked to corporate income or turnover from liquor sales (for businesses in the formal sector) which would more accurately reflect ability to pay and be more closely linked to the correction of negative externalities. Such a revenue base would also likely require agreement by the national government (given that these would be taxes on income or sales), and may require changes to the legal framework but could be efficiently administered by KRA together with their ongoing corporate income and/or VAT collections. To arrive at full-time equivalent positions, available employment data was weighted, with working owners and full time employees counted fully (100%), part-time employees half (50%) and casual workers one-quarter (25%). It was further assumed that half of all (50%) establishments, and therefore employee positions, engaging in food and beverage serving activities serve alcohol. An annual base rate of Ksh10,000 per full-time position has been assumed. In practice, counties differentiate charges by business categories and sizes, also requiring liquor licences for manufacturing, wholesale and retail activities. The base rate is adjusted by a county multiplier reflecting relative household consumption levels across counties derived from KIHBS 2015/16 (average consumption in county divided by average national consumption per adult). The underlying assumption is that higher consumption levels reflect higher turnover and thus ability to pay of businesses. In practice, county finance acts do not appear to include comparable liquor licence fee categories systematically across counties (some charge only health-type liquor certificates while others also require full liquor licences with higher rates). Overall, modelled rates are substantially higher than currently applied in the sample counties for most business sizes. The rationale is to use liquor licensing more actively for revenue raising purposes given their potential to regulate liquor- related businesses and correct for negative externalities. The modelled rates also treat different liquor-selling business types of the same (employment) size equally, which does not seem to be the case in practice. Table 16: Comparison of actual licence fees with modelled fee structure (Ksh). Kajiado Nairobi Tana River Restaurants 10,000-20,000 30,000 30,000 Current 10,961 per employed 19,900 per employed 6,760 per employed Modelled More expensive for >1-2 full-time More expensive for >1 full-time More expensive for >4 full-time Difference employed employed employed Hotels 20,000-30,000 50,000 50,000 Current 10,961 per employed 19,900 per employed 6,760 per employed Modelled More expensive for >2-3 full-time More expensive for >2 full-time More expensive for >7 full-time Difference employed employed employed Key findings Revenue base estimates show high full-time equivalent employee and working owner numbers in establishments serving alcohol in urban centres, with Nairobi accounting for almost half of the total estimated revenue base. Instead, predominantly rural counties have low to negligible revenue bases. Based on the modelled rate structure, this pattern is intensified in the revenue potential estimates, with over 60% of total revenue potential in Nairobi county (Ksh6 billions of Ksh10 billion aggregate potential). Other urban centres follow with Machakos (Ksh660 million), Kiambu (Ksh411 million), Meru (Ksh370 million) and Kisumu (Ksh370 million) also showing substantial revenue potential (see Figure 15). The revenue gap among the only five counties that reported liquor licence fee collections separately is high at 89%, highlighting that with the modelled higher rate structures, substantial revenue potential remains unrealized at the moment. Kirinyaga and Makueni appear to make good use of liquor licence fees already relative to their estimated potential (around 90% and 60% respectively); Machakos and Kisumu to a lesser extent. If politically feasible, counties could take a more active stance in using liquor licences for revenue raising to realize (a larger share of) the estimated revenue potential. 63 Figure 14: Estimated total full-time equivalent employees and working owners in establishments serving alcohol across Kenya counties Nairobi Machakos Meru Kisumu Kiambu Nakuru Kajiado Kericho Kakamega Mombasa Homa Bay Kisii Migori Uasin Gishu Narok Bungoma Nyeri Kilifi Makueni Nyamira Kwale Taita Taveta Embu Kirinyaga Laikipia Bomet Siaya Turkana Murang'a Lamu Samburu Baringo Nandi Trans Nzoia Busia Kitui Tharaka Nithi Nyandarua Isiolo Mandera Vihiga Elgeyo Marakwet Tana River West Pokot Marsabit Garissa Wajir - 50,000 100,000 150,000 200,000 250,000 300,000 350,000 64 Figure 15: Estimated liquor licence revenue potential by county (in Ksh million) Nairobi 6,249 Machakos 660 Kiambu 411 Meru 370 Kisumu 357 Nakuru 260 Kajiado 194 Mombasa 185 Kericho 117 Kakamega 99 Nyeri 99 Homa Bay 98 Uasin Gishu 93 Narok 91 Kisii 89 Migori 78 Kilifi 70 Bungoma 61 Makueni 56 Nyamira 55 Kwale 51 Taita/Taveta 51 Kirinyaga 50 Embu 45 Laikipia 38 Siaya 30 Lamu 29 Murang'a 27 Bomet 25 Baringo 21 Trans Nzoia 20 Turkana 18 Nandi 17 Tharaka-Nithi 16 Samburu 15 Kitui 12 Nyandarua 12 Isiolo 11 Busia 10 Vihiga 5 Mandera 4 Elgeyo/Marakwet 4 Tana River 4 Marsabit 3 West Pokot 2 Garissa 1 Wajir 0 Error! Reference source not found.6 shows revenue gaps for five counties that submitted detailed revenue reports that included separate accounting for liquor licence revenues. Kirinyaga appears to be well performing in comparison to the estimated potential, realizing 90% of it. Instead, Machakos, Kisumu and Kericho counties collect less than 10% of estimated potential, with gaps ranging from 93% to 99%. Figure 16: Estimated liquor licence revenue gaps by county ((only counties with data on liquor licence collections Kirinyaga 10% Makueni 41% Machakos 93% Kisumu 97% Kericho 99% 65 Table 17: Counties with highest (unrealized) revenue potential from liquor licences, Ksh million. Actual collections Revenue potential Unrealized potential County Revenue gap [max FY16-FY17) (estimate) (estimate) Machakos 49 660 611 93% Kisumu 11 357 346 97% Kajiado 1 117 116 99% Nairobi City Unknown 6,249 Unknown Unknown Kiambu Unknown 411 Unknown Unknown Meru Unknown 370 Unknown Unknown Nakuru Unknown 260 Unknown Unknown Mombasa Unknown 185 Unknown Unknown Total (5 counties with 1,240 139 1,101 89% data only) (All: 10,210) 3.6.2.5 Parking fees Counties collect parking fees for various types of vehicles, predominantly from cars and bus parking units but also from taxis, trucks and motorcycles. Various rate schedules exist charging vehicles by hour, day, month or year. Methodology To estimate counties revenue potential from parking fees, county vehicle registrations, household and business transport expenditure, and parking units/spots were considered as data inputs to generate proxy revenue bases. Data on vehicle registrations at the county level was not available, with vehicles registered nationally (unclear if owner residency by county is recorded). The KIHBS 2015/16 questionnaire includes a set of questions on households’ transport spending but the specific dataset is not available and available business surveys do not include specific information on transport spending. This left parking units/spots as the only possible approach to estimate revenue bases, which mirrors to current county revenue collection approach. The only available information on parking units in Kenya is from the master plan study for urban transport in the Nairobi metropolitan area.45 Assuming conservatively that all Nairobi county parking spots are located in the central business district, average parking spots per 1,000 core urban residents (based on the 2009 Population Census) amount to about 4.7. This information is used to extrapolate parking units based on counties’ core urban populations, only considering towns with a total population of more than 25,000 (based on 2009 Population Census). It is assumed that parking fees are charged on 6 workdays per week (excluding public holidays), which results in about 300 days per year. Based on the average parking time of 3 hours derived from the Nairobi transport masterplan study, it is conservatively assumed that each parking spot is used 2 times daily. Lastly, a bus parking multiplier based on the weighted ratio from four counties is applied (40:60 car to bus) for which disaggregate data was available (Embu, Kericho, Kisumu and Machakos). A base rate of Ksh150 per parking vehicle has been assumed. This is a simplified assumptions with counties differentiating different parking zones and durations in practice. The base rate is adjusted by a county multiplier reflecting relative household consumption levels across counties derived from KIHBS 2015/16 (average consumption in county divided by average national consumption per adult). The underlying assumption is that higher consumption levels reflect higher ability to pay of citizens. In practice, parking fees indeed vary by county, e.g., Nairobi rates are substantially higher than rates in poorer counties, which the modelled fee structure broadly reflects (Table 18). Table 18: Parking fees for sample counties. Bomet Machakos Nairobi Saloon car Ksh 50 Ksh 50 Ksh 200-400 Lorry or bus Ksh 100-200 Ksh 500 Ksh 1,000 Model parking fee based base rate and Ksh92 Ksh171 Ksh298 consumption level Key findings Almost 70% of estimated county parking units are concentrated in the five main urban centres (Nairobi, Mombasa, Kiambu, Nakuru and Kisumu), with Nairobi leading by a wide margin (Figure 17). The majority of predominantly rural counties instead have very low estimated revenue bases and thus low revenue potential. Based on the methodology, Tana River and Lamu have no revenue potential from parking fees. This is based on the assumption (driven by available 45 The study provides information on the total number of parking spots in the central business district of Nairobi ( almost 15,000) as well as information on average parking times (180 minutes). 66 data) parking is only available in urban centres above 25,000 inhabitants, resulting in conservative revenue base and potential estimates. In practice, rural counties may have limited potential from car parking but could have some potential from bus park and truck parking fees. When better data becomes available (e.g., of actual parking units by county, vehicle registrations and/or county transport spending), more accurate estimates can be prepared. That said, the estimates point to that professional management of county parking assets could lead to substantial revenue. Overall, the estimated revenue potential is above Ksh12 billion, with a revenue gap of 61% (only based on available data for 39 counties, with remaining counties excluded). Nairobi, Mombasa and Kiambu show highest unrealized potential at Ksh4.6 billion, Ksh995 million and Ksh 916 million respectively. Based on the estimates, Kisumu and Nakuru collect more than half of their substantial estimated potential revenue while in some counties with low potential estimates (Busia, Nyandarua, Muranga, Nyamira) collections exceed estimated potential two to four-fold. This can be explained by their low level of estimated potential (which is ‘easier’ to exceed in absolute terms) and, as discussed above, the high -level assumptions that were necessary to arrive at the revenue base proxy (which reduce accuracy of estimates). Figure 17: Estimated county vehicle parking units/spots across Kenya counties Nairobi Mombasa Kiambu Nakuru Kisumu Uasin Gishu Kajiado Kilifi Machakos Bungoma Kakamega Garissa Migori Kisii Mandera Turkana Trans Nzoia Laikipia Kwale Nyeri Busia Meru Kericho Siaya Isiolo Vihiga Homa Bay Taita Taveta Narok Embu Kitui Baringo Murang'a Nandi Marsabit Kirinyaga West Pokot Nyamira Wajir Samburu Elgeyo Marakwet Tharaka Nithi Bomet Nyandarua Makueni Tana River Lamu - 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 67 Figure 18: Estimated vehicle parking fee revenue potential by county (in Ksh million) Nairobi 6,655 Mombasa 1,509 Kiambu 1,196 Nakuru 528 Kisumu 355 Uasin Gishu 243 Kajiado 237 Kilifi 211 Machakos 202 Bungoma 101 Kakamega 88 Nyeri 85 Migori 79 Garissa 74 Kisii 74 Trans Nzoia 70 Meru 62 Laikipia 62 Kwale 60 Kericho 47 Turkana 46 Narok 44 Siaya 44 Mandera 42 Isiolo 40 Embu 38 Taita/Taveta 38 Homa Bay 35 Busia 33 Baringo 33 Murang'a 32 Vihiga 31 Kirinyaga 29 Kitui 27 Nandi 23 Marsabit 18 Nyamira 16 West Pokot 12 Wajir 10 Samburu 9 Tharaka-Nithi 8 Elgeyo/Marakwet 7 Nyandarua 6 Bomet 5 Makueni 4 Lamu - Tana River - Error! Reference source not found.9 below shows revenue gaps for counties that separately accounted for vehicle parking fees in their revenue reports. Gaps show a substantive range, with Makueni collecting over five times more than its estimated potential while Mandera and Narok counties collect less than 5% of their estimated potential. While substantial ‘over-collection’ (compared to estimated potential) is in parts possible due to very low estimated potential in some counties (e.g., Makueni), the data issues to establish a reliable revenue base by county that are described above (extrapolated parking spaces per urban capita in Nairobi CBD) are highly likely playing a leading role here, requiring caution when interpreting results for this revenue source. Improved estimates should be prepared once data on vehicle (public and private) ownership and usage are available by county. 68 Figure 19: Estimated vehicle parking fee revenue gaps by county (only counties with data on vehicle parking fee collections) Makueni-561% Busia -327% Nyandarua -201% Murang'a -107% Nyamira -101% Nandi -50% Vihiga -26% Nyeri -24% Meru -17% Kisii -15% Migori -5% Kericho -2% Embu 14% Kitui 18% Homa Bay 20% Uasin Gishu 28% Kakamega 32% Kirinyaga 33% Bungoma 34% Machakos 36% Kisumu 42% West Pokot 45% Nakuru 45% Elgeyo/Marakwet 45% Trans Nzoia 46% Taita/Taveta 59% Siaya 64% Mombasa 66% Nairobi 69% Kiambu 77% Kwale 81% Garissa 85% Marsabit 89% Kilifi 89% Turkana 92% Narok 96% Mandera 99% Table 19: Counties with highest (unrealized) revenue potential from parking fees, Ksh million Actual collections Revenue potential Unrealized potential County Revenue gap [max FY16-FY17) (estimate) (estimate) 1. Nairobi City 2,038 6,655 4,617 69% 2. Mombasa 514 1,509 995 66% 3. Kiambu 279 1,196 916 77% 4. Nakuru 292 528 235 45% 5. Kilifi 22 211 188 89% 6. Kisumu 207 355 148 42% 7. Kajiado Unknown 237 Unknown Unknown Total (39 counties with 12,173 4,750 7,424 61% data only) (All: 12,567) 3.6.2.6 Advertisement fees Counties charge advertisement fees for companies to market their products and services on public billboards, street lighting posts and other county infrastructure. Methodology The revenue potential estimates build on businesses’ self -reported data on advertisement expenditure included in the MSME Survey, 2016. The MSME Survey further includes information on the share of establishments which use 69 outdoors/public advertising as primary and secondary method for advertisement by county. The study assumes that businesses using outdoor/public advertising as primary method allocate 50% of their total advertisement spending to outdoor/public advertisement while businesses using it as their secondary method only allocate 30% of total marketing spending. For all other businesses, it is assumed that spending amounts to 10% of their total advertisement budget. Based on these assumptions, a multiplier is calculated to be applied to the total business advertisement spending by county. Since the MSME Survey does not cover large companies with 100 and more employees, a multiplier of 100% that doubles total estimated public advertisement spending by micro, small and medium enterprises is assumed given the significant role large companies, such as mobile network operators and banks, play in using public advertising (roadside billboards, lighting posts, etc.). The applied methodology using company spending on public advertisement does not require assumptions regarding counties’ rate schedules since estimated business spending equals county revenue from renting advertisement ‘infrastructure’ assets. Key findings Nairobi has, as expected, the highest revenue potential, followed by Kisumu and other major urban settlements (Nakuru, Kiambu, etc.). The majority of predominantly rural counties instead have very low revenue potentials. It should be note that the limited data and assumptions required to arrive at the revenue estimate implies that the results are indicative only and should be interpreted with caution. For example, businesses’ self -reported spending on outdoor/public advertisement in Kisumu appears high and in Mombasa low in comparison to Nairobi. Analogue to parking fees discussed above, when better data becomes available, more accurate estimates can be prepared. That said, estimates highlight that professional management of county advertisement assets could lead to substantial revenue. Figure 20: Estimated annual company spending on public/outdoor advertising across Kenyan counties, Ksh millions Nairobi 2,735 Kisumu 2,111 Nakuru 310 Kiambu 245 Kakamega 194 Machakos 162 Kajiado 152 Mombasa 103 Nyeri 39 Migori 31 Uasin Gishu 24 Homa Bay 21 Kwale 18 Kericho 14 Garissa 13 Embu 10 Narok 10 Kilifi 9 Taita Taveta 8 Laikipia 7 Nyamira 6 Lamu 4 Kisii 4 Meru 4 Turkana 3 Nandi 3 Kitui 3 Kirinyaga 3 Baringo 3 Mandera 3 Tharaka Nithi 3 Busia 2 Isiolo 2 Bomet 2 Bungoma 2 Nyandarua 2 Samburu 1 Murang'a 1 Makueni 1 Trans Nzoia 1 Siaya 0 Elgeyo Marakwet 0 Vihiga 0 Tana River 0 Wajir 0 Marsabit 0 West Pokot 0 - 500 1,000 1,500 2,000 2,500 3,000 Millions Only limited data is available on counties’ actual collections from advertisement fees. Nairobi collects a substantial share of total revenue from such fees, as does Machakos. Unrealized revenue potential is highest in Kisumu (as discussed above, Kisumu businesses reported very high advertisement spending, including for outdoor/public advertisement), followed by Nairobi. For the other counties with substantial estimated revenue potential (Nakuru, Kiambu, Kakamega and Kajiado) no data on actual collections is available, not allowing estimation of unrealized potential and revenue gaps. Counties with highest actual collections as a share of potential revenue (based on limited available information on actual collections) are Makueni, Kirinyaga, and Kwale, all of which have low estimated potential. 70 Figure 21: Estimated outdoor advertising fee revenue potential by county (in Ksh million) Nairobi 2,735 Kisumu 2,111 Nakuru 310 Kiambu 245 Kakamega 194 Machakos 162 Kajiado 152 Mombasa 103 Nyeri 39 Migori 31 Uasin Gishu 24 Homa Bay 21 Kwale 18 Kericho 14 Garissa 13 Embu 10 Narok 10 Kilifi 9 Taita/Taveta 8 Laikipia 7 Nyamira 6 Lamu 4 Kisii 4 Meru 4 Turkana 3 Nandi 3 Kitui 3 Kirinyaga 3 Baringo 3 Mandera 3 Tharaka-Nithi 3 Busia 2 Isiolo 2 Bomet 2 Bungoma 2 Nyandarua 2 Samburu 1 Murang'a 1 Makueni 1 Trans Nzoia 1 Siaya 0 Elgeyo/Marakwet 0 Vihiga 0 Tana River 0 Wajir 0 Marsabit 0 West Pokot 0 Error! Reference source not found.2 below shows advertising fee revenue gaps by county for eight counties that separately accounted for this revenue source in their revenue reports. Again, Makueni collects substantially more than its estimated potential, partially explainable by its very low potential estimate (Ksh 1 million) based on the best available proxy data. Kirinyaga county also collects more than estimated while revenue gaps for other counties range between 30% and 97%. Several counties with high estimated potential appear to have significant unrealized revenues, including Kisumu (97% gap), Machakos (78% gap) and Nairobi (74% gap). 71 Figure 22: Estimated outdoor advertising fee revenue gaps by county (only counties with data on outdoor advertising fee collections) Makueni -237% Kirinyaga -30% Kwale 30% Nairobi 74% Machakos 78% Kericho 85% Embu 91% Kisumu 97% Table 20: Counties with highest (unrealized) revenue potential from advertisement fees, Ksh million. Actual collections Revenue potential Unrealized potential County Revenue gap [max FY16-FY17) (estimate) (estimate) 1. Kisumu 61 2,111 2,050 97% 2. Nairobi 720 2,735 2,015 74% 3. Machakos 35 162 127 78% 4. Nakuru Unknown 310 Unknown Unknown 5. Kiambu Unknown 245 Unknown Unknown 6. Kakamega Unknown 194 Unknown Unknown 7. Kajiado Unknown 152 Unknown Unknown Total (8 counties 5,055 839 4,215 83% with data only) (All: 6,272) Additional results: potential alternatives to cess revenues As discussed under 3.3 and in more detail in the legal and policy review section, cess collections are potentially harmful, impacting on the free flow of goods, are prone to illegal practices, and not consistent with the existing legal framework. A detailed assessment has been prepared in 2016, which highlights issues with cess collections. 46 Based on audited county revenue data, total cess collections have seen a slight upward trend since 2014/15 but still only account for 3.5% of total collections (Table 21). Relative to national VAT collections, also a tax on goods, cess collections are negligible, amounting only to less than half a percent of total VAT collections. It is unclear to what extent there are other cess-like collections not specifically recorded under the cess category; however, it is unlikely that such collections would lift total collections to substantially higher levels. Table 21: Trends in cess collections relative to total county collections and national VAT. 2014/15 2015/16 2016/17 (interim) Cess actual collections Ksh 976m Ksh 1,279m Ksh 1,213m Cess share of total county collections 2.7% 3.6% 3.5% Cess collections relative to national VAT 0.38% 0.44% 0.36% While on aggregate cess is not a major revenue source for counties, is an important revenue source for several. In Tana River County it accounts for about half of total collections and for about a third of total collections in a number of counties. However, considering estimated revenue potential from less harmful revenue sources, primarily property rates, even in those counties cess collections could be replaced and total county collections increased through a change in focus and a redeployment of capacities from cess to, for example, property tax/rates. The phasing out of cess could also be financially incentivized, e.g., compliant counties could receive a (temporary) top-up of intergovernmental fiscal transfers. This could help fast-tracking counties abandoning cess and other potentially harmful and possibly illegal collections (and may be a less confrontational approach than attempting to enforce a stop of such collections countrywide). Table 22: Counties with high relative cess collections (>15% of total collections), Ksh million, FY2016/17 Cess collections Revenue potential of alternative sources Total County Business Comparison with cess collections Amount Share Property rates Total licences collections 46 Kenya Markets Trust, 2016. Burden of Produce Cess and Other Market Charges in Kenya. 72 Tana River 25.9 13.7 53% 150.6 5.9 156.5 11 times higher Nyamira 126.5 42.1 33% 499.6 185.7 685.3 16 times higher Marsabit 163.9 49.4 30% 192.9 6.4 199.3 4 times higher Nandi 244.7 70.5 29% 456.8 45.2 502.0 7 times higher Elgeyo Marakwet 97.3 25.7 26% 228.2 16.6 244.8 10 times higher Migori 360.8 92.5 26% 427.4 157.3 584.7 6 times higher Nyandarua 300.1 51.3 17% 773.3 66.1 839.4 16 times higher Note: Estimated revenue potential from property rates are based on a 1% rate on all properties (low and high value). If lower rates are applied, e.g., low value properties exempt, potential from alternative sources would be less than shown here, in which case (temporary) compensatory transfers could incentivize counties to phase out cess collections. 73 Recommendations: Strategies for Enhancing County Own-Source Revenue Following the analysis presented in Section 3.3 and international good practice examples, we have identified a number of specific recommendations for Counties to enhance their revenue collections, including recommendations relating to: a) Data quality and availability b) Legal framework – general recommendations; c) Policy – general recommendations; d) Specific recommendations for key revenue streams; e) Options for potential new revenue sources; f) Options for incentivising Counties to enhance OSR through the CRA formula. Data quality and availability Data source Issue Recommendation Database of county Coverage: All 47 counties revenue data from Quality: audit reports  Total collections: Appear reasonably reliable, with some data issues (e.g., Counties to address audit incomplete reporting, reconciliation issues, spending at source) outlined in recommendations county audit reports  Individual revenue items: Inconsistent reporting across counties and Redesign of reporting template substantial aggregation in summary categories (e.g., miscellaneous/other (e.g., GFS-based), with clear charges), partially due to unsuitable reporting template (e.g., no specific item supporting guidelines for for land rates) counties Unaudited county Coverage: Only 8 counties submitted (up to 8 June 2018) Use redesigned standard revenue reports Quality: Some inconsistencies in report coverage (revenue items, item names, template also for county-internal (collected as part of fiscal vs. calendar year) but better information on main revenue items reporting project questionnaire) Actual bases for Detailed information on base definition in county finance acts but only very Counties to maintain readily county revenue items limited information on actual revenue bases (e.g., info on charge per off-street available databases, ideally parking slot but no info on number of available slots), except for: harmonized across counties using standardized IT system(s)  Property tax valuation rolls for 20 counties; however, coverage (e.g., private and/or public land, part of county only) and completeness (e.g., values Institutionalize regular data missing) major issues sharing across relevant  Building permit approvals for Nairobi City County only stakeholders, including counties,  Other studies and assessments (e.g., Nairobi transport master plan with actual KRA, KNBS and relevant parking slots in CBD) ministries (e.g., lands)  Info not useful to establish consistent revenue bases across 47 counties and only partly used to extrapolate Census and survey Multiple censuses and surveys available, several of which very useful including Ensure different surveys are data the KIHBS 2015/16, National Housing Survey 2012/13, and MSME Survey 2016, complementary and paint but some significant inconsistencies across surveys and data gaps at county level together a realistic picture exists, and some data has not (yet) been available. For example: Ensure future censuses and  Substantial differences in business and employment numbers between various surveys capture relevant info surveys (MSME, COE, KIHBS, KRA registry), even when comparing similar that supports county collections, business categories ideally based on harmonized and  Value of new construction only available for selected counties and very low simplified revenue sources and compared to actual county collections from building/construction permits bases  Planned KNBS county survey delayed; Census of Establishments (2017) (KNBS in collaboration with detailed data not yet available other stakeholders including counties) County rate schedules County finance acts provide detailed information on rate schedules, with the Consider reducing length of rate following noticeable findings: schedules by grouping and/or eliminating some charges  Rate schedules are very extensive (often 100+ pages)  For land rates, rates are used to compensate for outdated valuations, which Harmonize land rates and update leads to high rates and substantial variation across counties valuations  For business licences, larger businesses are often relatively lower taxed than Review business licence smaller and informal businesses schedules 74 Legal framework – General Recommendations Issue Recommendation Finance Acts should be reserved for annual amendments to fiscal provisions, arising from the County annual budget submitted and Some Counties have used the mechanism of the Finance Act to create passed under Art. 224, including any necessary amendments to the an omnibus law that imposes all fees and charges, but without any of County (Taxes, Fees and Charges) Act (see above). the regulatory functions and processes for the licence or service in respect of which fee or charge is imposed Counties should create a County (Taxes, Fees and Charges) Act that states in one Act all the revenue streams (authorised by legislation) and specifies the relevant tax rate, fee or charge County Finance Acts are intended to introduce amendments (e.g. of fees and rates, or tax bases) to substantive laws and are not supposed County legislation that creates a regulatory duty or obligation, and to confer taxes in their own right as a stand-alone law, without the imposes a licensing fee, should not be set out only in a Finance necessary accompanying provisions relating to how a tax or charge Act, but instead in dedicated, separate County legislation will be imposed and collected The functions of the national government and the county governments To take stock of the various intergovernmental frameworks in relation to betting, casinos and other forms of gambling are under the constitution and recommend any changes to ensure purported to be delineated by approval of the Intergovernmental conformity with the Constitution e.g. Provide clarity in the Relations Technical Committee, as published in the Gazette Notice Intergovernmental Relations Act 2012 on mandate of No. 8753. However, it is not clear that the Technical Committee has Intergovernmental Relations Technical Committee. such a power. General Recommendations – Policy Issue Recommendation Ensure local taxes have a clear policy rationale: in terms of principles There is a clear disconnect between revenue streams and policy of taxation, market failure (public goods, externalities), equity/social objectives. In many cases, the basis for the current set of revenue services provision, revenue-generating, development, regulation. bases, rates and charges is that ‘inherited’ by counties from the Policy objectives and features should be clearly communicated, such previous local authority / municipal laws. as definitions and communication of base, rates, who is liable, when to pay and how to pay; For user-charges, develop pricing policies for cost-recovery of Not all revenue streams are suitable for revenue enhancement effort services, which identified and justifies which services will be e.g. user fees, which represent payment for accessing a service. subsidised, those selected for partial cost recovery, any for which Counties do not have a clear policy on how these fees are set. market pricing will apply, and on what basis the cost will be applied (e.g. average or marginal cost, or alternative structure); Counties should focus revenue enhancement effort on fewer, coherent Counties collect a wide range of revenues, some of which place a sources that have greatest potential, have a clear policy rationale and relatively high burden on taxpayers/users and are costly to collect, are most cost-effective to administer e.g. Property tax, building while others have potential, but are not well designed. permits, SBP, parking, advertising, liquor licensing and relevant user charges. County finance/revenue departments to develop capacity in tax (revenue) policy design and analysis e.g. with technical assistance, if possible. In particular, to build capacity in the following areas: (a) Monitoring and analysis of costs: to assess cost of service County capacity in policy design and analysis is low - most counties provision (average and marginal) and consider links to fees; and do not have a clear county revenue generation policy, guidelines or a policy unit. (b) impact analysis of policy changes, including costs and benefits of new policy measures and economic and welfare impacts on users/taxpayers. Ideally, revenue impacts of changes in fees and charges will take into account the relative elasticities (responsiveness) of taxpayers/users to changes in the price or tax payable. Policy consultation exercises are not being used to effectively inform Policy objectives and features should be clearly communicated to policy. Due to the lack of clear understanding and link to policy taxpayers, users and stakeholders, including key policy features e.g. objectives, the policy justification for adjustments to rates or revenue definition of tax base, rates, who is liable, when to pay and how to bases is often not communicated and can weaken the purpose and pay. outcome of consultation exercises. 75 Property Tax Issue Recommendation Counties should focus most revenue enhancement effort on Property tax represents by far the largest revenue potential, according improving efficiency and effectiveness of the property tax. The to the potential estimation exercise. potential estimates only hold if counties can update the revenue base (including valuation) and rates in line with international practice In order to provide for a harmonised national property tax base, consider establishing a modernised national rating legislation appropriate to the current and anticipated land tenue and land use The creation of property rating legislation by each County runs a high system in Kenya. This Act would expressly reserve to the counties risk of divergence with multiple property rating systems and tax bases powers in relation to the rates, bands, waivers and discounts for the developing. rates in relation to their fiscal objectives in light of the economic context of each County. Legal advice from the Attorney-General would be helpful to guide the process Simplify valuation methods and provide for regular updates: e.g. Most counties are operating property tax from outdated valuation rolls, banding and indexing between valuation exercises to add new which are reported to be expensive to update property and allow for inflation; or Computer Assistance Mass Appraisal (CAMA) e.g. South Africa; Review and apply a more consistent base: consider the shift from Kenya counties levy property rates on the unimproved site value. ‘unimproved site value’ to ‘improved site value’, which may depend International practice typically uses improved site value (i.e. land on an assessment of data availability and consistency with the chosen including buildings). method of valuation. There are a multiplicity of rates applied to land and property across Rates to be reviewed and adjusted with re-valuation, to maintain a and within Kenya counties. Globally, rates are in the range of 0.15 – low, more uniform rate across counties. Setting rates may require an 2%. Some rates are exceptionally high compared to international assessment of impact on ability to pay and consideration of whether a practice, partly due to outdated valuation rolls and administrative more progressive regime is preferred. challenges. Kenya County property rates are based on a few categories of land Relative use of services could also be a consideration for rate setting, use, such as agricultural, industrial and residential. However, some do as well as property use (agricultural, residential and industrial); not have a ‘commercial’ category and significant numbers of Counties should consider introducing a ‘commercial’ category and a ‘Business cum residential’ (BCR) properties, both of which need a methodology for establishing a ratio for ‘mixed’ properties. clear definition for rates purposes. Consider implementing a shift in collection of the tax from owner to Owner definition does not adequately cover the range of land tenures occupier: this should aim to address the problem of taxing informally in Kenya and limits the tax base. Counties face administrative held or traditional land as well as absentee owners. This would challenge collecting from unregistered properties or from absentee require amendment of the legal framework and procedures, which owners. Having an ability to collect from occupiers can be could be considered in developing a modern national framework, as instrumental in improving revenue collection described above. In principle, government property would be taxed in the same way as private property, according to its value (or rental value) of the Reinstate CILOR with clear methodology and process for government buildings. But practical collection can be difficult and depends on payment/transfer goodwill. Strengthen compliance and recovery methods, such as: Compliance Counties face multiple compliance and administration challenges in certificates; Interruption of services, (e.g. electricity); Recovery from collecting property rates. tenants or beneficial occupant: encumbrance of estates (already implied); early payment discounts; monthly payments to spread cost. Entertainment Tax Issue Recommendation The introduction of VAT would normally replace turnover taxes, such Consider repeal of Entertainment Tax Act (Cap.479) and enactment as the entertainment tax. For counties, there is a significant exemption of new Act with wider tax base consistent with the functional from the scope of the Act (operators registered for VAT are excluded), responsibilities under Constitution Fourth Schedule Part 2, paragraph thus the revenue base is limited to turnover below the VAT 4 but without being double taxation of those services subject to VAT. registration threshold or VAT-exempt. In practice, the taxing rights assigned to counties are limited to the Consider combining gaming and betting licensing of premises with licensing of premises. This would appear to be akin to a regulatory higher SBP license fee for ease of administration and to capture social activity only, as opposed to a tax. cost of gambling addiction, in context of other taxes on gaming and betting industry and industry norms and standards. 76 Issue Recommendation Other forms of entertainment tax based on the number of admissions Fees for entertainment venues are also a form of regulation of to entertainment venues and the value of admission charges might be business and therefore could be considered as part of SBP, with difficult for counties to administer in practice, due to difficulties in higher rates for larger venues to reflect the public safety risk and cost validating admissions and receipts. of regulation Single Business Permits Issue Recommendation Counties may not rely on the transitional provisions of the Local County legislation should be enacted to establish a clear licensing Government Act Cap. 265 in relation to Single Business Permit as that framework that is consistent with the principle of a SBP. Act has been repealed. Counties may not collect fees for SBPs or trade The model legislation on trade licences should be reviewed to ensure licence without clear legislative authority of an Act of the County that it is consistent with the principles of SBP. Assembly. Professional firms (such as lawyers, accountants and doctors) are County legislation should not seek to impose a SBP requirement on a regulated by the National Government or entities established under professional business that is regulated by statute legislation of the Parliament. An additional requirement of a county government for a professional services firm to apply for a single business permit (trade licence) would effectively be additional regulation of the firm, and consequently is excluded. There has been a proliferation of business licenses and fee rates across Consider national guidelines for the relative, simplified structure of Kenya’s counties. fees, based on size (e.g. employees or turnover), discretion for counties to set fees within the framework. Counties adopted Local Authority rate structures for SBP and do not Variation of fees outside simplified framework to have a clearer appear to have clear policy on SBP fees. rationale, such as the regulation of specific sectors (gambling, liquor licensing) and/or addressing negative externalities (e.g. environmental damage). Cess Issue Recommendation The Agriculture Act Cap.318 was repealed and replaced by Cess fees should not be collected without clear legislative authority Agriculture, Fisheries and Food Authority Act, 2013 (No. 13 of 2013). enacted by the County Assembly. County CESS legislation should be The new Act does not provide for Cess in any form. Any cess, as a repealed unless: fee, legislated by a County, would require to be consistent with the  A clear revenue potential case can be made. “single national market” tests set out in Article 209 of the  The Art. 209 tests can be satisfied Constitution. A 2016 study of cess in Kenya found evidence of the existence of Consider replacing cesses due to high economic burden, double multiple cess levies along trading routes, as revealed by traders in taxation and risk of excessive cost crossing borders etc. (see also urban counties located away from the major production areas that face Legal review analysis): higher cess charges and that produce cess adds a significant cost to the  Agriculture cess to be replaced with property tax (or, for distribution of produce around the country. example, a flat land tax (category of property rates), which relates more to the agriculture sector) Our potential estimates indicate that cess revenue collections can be  Quarrying ‘Cess’ (or other labels) to be handled under SBP as relatively easily replaced by collections from enhanced property tax. special category or replaced with environmental levy (revenue sharing) on existing mining levy/royalty Other Revenue Sources Issue Recommendation Parking Fees can raise revenue to provide for maintenance of assets Counties to review rates in line with benchmarks, location, peak and roads, as well as regulate congestion at peak periods. Most periods and zoning of areas to manage traffic flow etc. and ensure counties do not have a clear policy on pricing of parking charges. cost of provision and maintenance is at least covered by revenue Rationale and definition of ‘Market fees’ varies across counties e.g. in Counties to clarify the definition and objectives of market fees e.g. is some cases it is a form of agriculture cess, in others it is an access fee it a type of cess on produce, an access fee to market space and or rent for use of market facilities. facilities, or rental charge for use of government property (market stall); Since we recommend removing cess, enhancement of this type of revenue source should focus on management of assets (market stalls, facilities) and ensure fees are commensurate with the cost of provision and ability to pay 77 Options for new revenue sources Issue Recommendation A form of ‘City tax’ is fairly widespread internationally. The rate is Consider whether there could be a policy case based on the rationale either a fixed amount per room per night or a percentage of the final that local hotel occupants benefit from County services and tourism hotel bill. The purpose for revenues varies, including the development infrastructure. Careful review in line with VAT and East Africa of tourism or general revenue-raising. Tourism Levy is needed. Economic activity that causes localised environmental damage may Consider the case for environmental taxes in Counties that provide a case for local taxation. An environmental tax ‘internalises’ particularly suffer from environmental degradation from harmful negative externalities from environmental degradation, with a rate activities e.g. quarrying, polluting industrial activity or waste based on the cost of mitigating environmental damage and/or disposal. Careful consideration in context of alignment with existing capturing the environmental cost. Typically, environmental taxes are mining law in case of quarrying or other existing national regulation. levied as part of a regulatory regime. An alternative mechanism for local own-source revenue, is local Any income tax or VAT levy would require to be authorised by the revenue sharing arrangements with central government e.g. through National Parliament, and stated to be a tax that County Governments sector levy or income tax. There could be collection efficiency are authorised to collect (consistent with Article 209(3)(c) of the benefits, but could incentivise ‘mobile’ factors to avoid tax in some Constitution). This would require political agreement between the counties. different levels of government. Revenue Administration Issue Recommendation Counties should enact legislation to set out compliance obligations and powers in a County (Revenue Administration) Act, and the There is no law equivalent to the administration procedures law for legislation can be based on the existing model, reviewed and updated national tax collection, leaving the collection, assessment and through the Intergovernmental Relations mechanisms. enforcement procedures to Counties, which leads to a diversity of administration frameworks and approaches, which can be confusion, Alternatively, such legislation could be enacted by Parliament for unfair or unnecessarily burdensome to taxpayers. exercise at a County level, to ensure consistent county tax/fee/charge administration, compliance and enforcement treatment throughout Kenya. For the short-term, Counties should consider the engagement of the KRA in respect of the collection of property rates and trade licences, The PFM Act 2012 empowers counties to engage KRA for revenue in terms of a standard Memorandum of Understanding that sets out collection. Some counties have entered MoUs with KRA to do this. service level agreements for KRA and exchange of information to The agreed rates appear to be significantly lower than the prevailing assist both County and KRA to maintain accurate taxpayer cost-of-collection estimated from indicative information. There would registration. be a strong argument for a county, faced with immediate needs to put Counties should not engage external collectors of county OSR unless in place an effective revenue collection system, to engage KRA. the engagement is pursuant to transparent public procurement processes and a cost benefit analysis justifies the level of fees to be paid for the services. There is potential to improve compliance rates and revenue collections Counties should engage in information sharing arrangements (through through validation and cross-matching of county taxpayer/user data written agreements) with agencies whose data can contribute to with third parties. ensure the integrity of the County OSR tax base. Strengthen taxpayer/user awareness and support, including access to There is a lack of clear guidance and support to taxpayers and users on information, guides and quick reference on liabilities and procedures; tax obligations, procedures and payment methods. These features can improve communications with taxpayers/users on policy objectives improve voluntary compliance. and the benefits of tax to service delivery; Establish risk management approaches in revenue administration: Most counties do not apply a risk-management approach to identify, assess and monitor revenue risks using data on registration, compliance and enforcement activities (i.e. using limited to resources filing, assessment and payment; develop appropriate strategies to to focus follow up on highest revenue risks). improve compliance according to risk; Counties to review and, if necessary, simplify IT systems and databases to improve data accuracy and comprehensiveness; establish Risk assessment is limited by data quality issues. protocols for sharing of data with third parties and government departments and agencies; and Counties to consider investment in staff capacity building and Most counties have under-qualified revenue staff and insufficient recruitment procedures, guidance and training to include technical training. knowledge and also professionalism and integrity/attitude. 78 Issue Recommendation Strengthen performance monitoring of revenue administration, Revenue performance is mostly tracked by collection outturns, but not including indicators of efficiency e.g. cost of collection and administrative efficiency measures. compliance ratios. Revenue Management Issue Recommendation Counties reported challenges in reporting and revenue administration Greater automation of revenue management systems and strengthen due to connectivity issues and out of date IT systems/configuration IT connectivity, speeds and infrastructure to support the replacement (e.g. using old Local Authority system). of manual receipting and reporting. Most revenue management systems used by counties are not integrated Establish better integration of revenue management systems with with IFMS, which increases the risk of errors and missing data IFMS reporting. entering IFMS reports. Forecasting of revenues is reported to have wide deviations with Strengthen forecasting methods, such as use of monitoring, modelling outturns. Methods vary and not all counties are using estimates of and projection of tax bases, analysis of historic trends and impact actual tax bases to monitor and forecast revenue streams. Assumptions analysis of new measures. are not transparent. Improve transparency of forecasts by publishing forecast assumptions. Undertake regular performance review of forecasting and outturn revenue performance, providing reasons for deviations and propose improvements, incrementally. Revenue potential estimates could be used to inform a factor within A detailed study comparing different options for incentivising the county revenue allocation formula to incentivise counties to counties, based on the potential estimates from selected revenue enhance OSR. source(s) from this study (and perhaps other sources), to generate more exact and reliable estimates than the scope of this study allows. 79 Conclusions and Next Steps The key objectives, as stated in the Terms of Reference (see Introduction Section 1.2) have been met as follows: a) map out counties’ current local revenue base, and the associated tax potential vis-à-vis tax effort, leading to a clear determination of counties’ tax and non-tax revenue potential, a systematic identification of revenue streams which can enable each county to maximize its revenue potential and a comparative assessment of counties’ fiscal capacities; Counties’ current local revenue base and potential has been discussed in Section 2, in terms of policy and legal framework, and estimated in Section 3, to the extent possible using available data. The study found significant potential, both in terms of existing frameworks and capabilities (using the frontier analysis) and in terms of longer- term potential that could be gained from expanding the tax bases and improving compliance, ranging between Ksh 55 billion and Ksh 173 billion, compared to current collections of Ksh 35 billion. While data gaps hindered revenue gap analysis for most counties, where there was data available, estimated potential compared to actual collections show gaps between 35% and 94% for different county revenue sources. Such substantial gaps are likely to be representative of most, if not all, counties. This suggests that counties can gradually fund an increasing share of local service delivery from own source revenue if they are able to realize more of the available potential over time (while intergovernmental fiscal transfers will continue to play an important role for local goods and services, particularly in health, education and infrastructure). b) bring about more credible projections by counties of future revenue from assigned taxes, fees, levies and charges, leading to improved alignment between budgets and policy priorities; and, The revenue potential estimates and estimated tax bases provide a framework that will support counties to make more credible projections of future revenues from 6 key revenue sources. Counties (or national government) will need to establish comprehensive databases for these (and potentially other) revenue sources in order to improve county forecasting quality. c) develop a framework for monitoring improvements by counties in terms of OSR performance including efficiency in collection, leading to more objective assessment -- particularly by the Controller of Budget -- of county budgets, specifically focusing on actual OSR vis-à-vis forecasts. The potential estimation exercise (and resulting data/modelling sheets provided) establishes a framework for monitoring improvements, by monitoring revenue collections against potential to assess on a regular basis how much of the gap is reducing over time. In order to track this consistently and accurately, it will be critical for counties to improve data availability and quality, including through more systematic and consistent reporting of collections by revenue source as well as information on actual revenue bases, as the basis for any improvement monitoring. In addition, during Phase III, ‘Data analysis, modelling and recommendations ’, the study undertook the following specific tasks: (i) Simulation of revenue potential for each county. Section 3 presents simulations of 6 key revenue sources, representing those with the greatest revenue potential. These simulations, supported by policy and legal analysis, also shed light on the following issues: a. Optimisation of county OSR requires that counties address a number of legislative, policy and administration challenges and capacity gaps, including a relatively high collection cost across a range of smaller revenue streams, which could be better focused on a few key sources, notably the property tax. While a number of lower value revenue streams will remain necessary (e.g. user charges for services), there are some levies that could be replaced e.g. Cesses and entertainment tax. b. There is no specific evidence that counties are deliberately not rolling out any available revenue sources, but the analysis indicates some significant constraints. For example, out of date valuation rolls, which limit the value of the base for property tax. The property tax base is also limited to unimproved land and placing liability only on the registered owner. In the case of entertainment tax, the legal base for the tax is significantly restricted by exempting VAT-registered businesses. As a result, counties are collecting almost nothing from this tax. (ii) Reviewed Kenyan counties’ constitutional and legal OSR-raising powers, including the discretion to introduce taxes/fees/charges/levies, set rates and boundaries (e.g. floor and ceilings) and to grant reliefs or full exemptions. Drawing on international practice and examples from other countries, this review identified a 80 number of proposals, outlined in Section 4, which could strengthen the legislative framework for counties’ OSR enhancement. Most of these suggestions are in line with the approach being developed already by Government and may further inform the process, where required; and (iii) Proposed specific recommendations on how different counties’ should strategize to enhance OSR collection , based on their unique macroeconomic, fiscal, geographic and urban profiles. The recommendations set out in Section 4 cover data quality issues, policy and legal framework strengthening, how to enhance specific revenue streams and revenue administration and management. In particular: a. That the majority of revenue potential was found in a few sources of revenue, particularly property tax, which represents up to Ksh 108 billion out of a possible Ksh 173 billion; and, b. Possible incentives for counties to enhance OSR collections and improve efficiency could be introduced in the county revenue allocation formula (see Section 2.6). Going forward, it is important to note that revenue potential is never fully realised, due to non-compliance, resource and capacity constraints, among other things. This is likely to be the case particularly in Kenya, in which the county administration are still relatively new and evolving. Counties are therefore likely to require further guidance and assistance in implementing the proposed recommendations to achieve an additional share of the potential and to track and improve performance over time. We envisage that further technical assistance will be needed particularly to help counties to address legal gaps and inconsistencies, as well as administrative challenges and policy design and analysis. The draft policy and bill on county revenue is under discussion and will help frame discussions between National Treasury, CRA and the Counties, among others to strengthen and streamline the legal and policy frameworks in each county. Counties will need to focus efforts in on the key sources highlighted, which may involve removing or replacing some existing revenues, while maintaining a lighter touch approach on others. 81 Annexes Annex 1: Questionnaire Summary Report Introduction This report captures the responses to a standard questionnaire sent to all counties which was filled by the relevant heads in charge of each of the counties’ revenue collection and management. 1 1 questionnaires were returned filled and those who responded held the following titles depending on the specific counties: i. Ag. Director Revenue Management ii. CEC Finance iii. Chief Finance Officer iv. County Chief Officer- Finance and Economic Planning v. Director-Revenue vi. Head County Revenue vii. Ag. Head of Revenue The report follows the format of the questionnaire. It is divided into eight sections with titles matching those on the questionnaires. 1. REVENUE COLLECTION This part reports on the revenue collected by the counties; the specific streams and their performance for a period of 3 to 5 years. Half of the returned questionnaires omitted this section of the questionnaire. Of the remaining half, most indicated to attach data but failed to do so while only one respondent forwarded the needed data. Table 1 below breaks this up: Table 1: Availed Data on Revenue Streams No data Data Provided Number 10 1 Percentages 90.91% 9.09% Source: Computed from County OSR questionnaire responses County Own Source Revenue Streams Based on data provided by Respondent 5, the main sources of revenue collection from the counties are: land rates, single business permits, house and stall rents, market fees, royalties and cess, transit goods tax, auction fees, advertisement charges, parking fees, revenue from county hospitals, building plan approval fees, land revenue, garbage fees, charges on slaughter houses, revenue from agriculture livestock and fisheries activities, charges connected to public health and sanitation, tourist charges, and a range of miscellaneous charges. The aggregated contribution of each for the period 2012/13-2016/17 is illustrated by figure 1 below. 82 Figure 1: OSR Collections per Stream for Period 2012/13-2016/17 Miscellaneous Revenue from Public 4% Trade and Revenue from Health and Garbage Fees Cooperative Agriculture Livestock Sanitation 1% Slaughter 0% and Fisheries 1% 0% 1% Land Revenue 0% Tourist Charges Building Plan Approval 0% 1% OSR Streams Collections Revenue from (2013-2017) Hospital 11% Land Rates 28% Parking Fees 3% Advertisement 5% Auction Fees 3% Single Business Transit Goods Parmits 22% Royalties 4% and Cess 11% Market Fees House and Stalls Rent 2% 1% Source: Computation from data from Respondent 5 Notably, land rates collected generated more revenue than any other source at 28%. This was followed by single business permits with 22% of the collected revenue. County hospital collections and collections from royalties and Cess tied at 11% of total revenue collected as the third largest revenue source. 2. LAWS AND LEGAL FRAMEWORK Laws Defining Revenue Collection and Administration The Annual County Finance Act was reported as the main legal framework that defines the administration and collection of revenue in the counties. The other laws reported to define revenue collection were those that were specific in regard to the revenue streams and included the following:  County Liquor Control Act/Liquor  County Parking Act Licensing Act  County Non-Motorised Vehicles Act  County Trade License Act/ County  County Public Participation Act Business Licensing Act  County Health Services Act  County Outdoor Advertisement Act  County Omnibus and Parking Act  County Quarrying Act  County Alcoholic Drinks Control Act  County Sand Harvesting Act  County Sand Conservation and  County Revenue Administration Act Utilisation Act  County Ratings Act 83 This list of laws and regulations is a combination from all the counties indicating a variation from county to county on the laws adopted in regard to revenue streams. Laws Regulating Revenue Collection and Administration In responses from most respondents, the above laws were reported also as the basis of regulations on county revenue collection and administration. The aggregated list is as follows:  Yearly County Finance Act  Liquor Control/Licensing Act  County Revenue Act  County Revenue Authority Act  County Quarrying Act  County Quarry Regulations  County Outdoor Advertising Act  County Trade & Markets Act  County Livestock Sales Yard Act  County Business Licensing Act  County Transport Act  County Sand Harvesting Act  County Revenue Administration Act  County Rating Act  County Parking Acts  County Non-Motorised Vehicles Act  County Public Participation Act  County Public Finance Management Act 84 Laws Defining and Regulating Revenue Management (Accounting and Reporting) The following list indicates the laws that defines and regulates the revenue management in the counties as reported:  County Finance Act  Public Finance Management Act  County Government Financial Accounting and Reporting Manual  County Revenue Authority Act  Public Financial Management Regulations  County Revenue Administration Act Notably, apart from the County Finance Act ubiquitous in all the counties in regard to revenue collection, the list leaves out a number of laws noted in the previous two lists. This is so because those that were left were noted as being specific to revenues stream. Among the 11 respondents collected, two of them reported not to have a framework for the county’s revenue stream management. Matching of Revenue Streams to Appropriate Law County revenue streams are broadly matched to their respective laws. When asked whether they have done a mapping exercise to establish the appropriateness of the revenue streams to the existing legislative source, more than half of the respondents affirmed they had. The other group declined and asserted that their counties are yet to do such a mapping. Revenue Sources with Little Legal Authority Responses as to whether the county has identified a revenue source with no, insufficient or appropriate legal authorities were as follows: Table 2: The County has Revenue Stream with Inadequate Legal Authority Response Frequency Percentage (%) Yes 8 72.73% No 3 27.27% Blank 0 0.00% Total 11 100.00% Source: Computed from county OSR questionnaire responses From table 2, it is clear that most counties have a revenue stream that is not sufficiently supported by existing laws and regulations (72.73% of responding counties). Only 27.27% of the respondents had proper laws and regulations that backed all their county’s revenue sources. Shortcoming of Existing Laws The following are the gaps that the respondents identified in the laws that existed to support revenue collection sources.  Total lack of a supporting policy and primary legislation for certain revenue sources  Limited enforcement authority in executing certain revenue collection sources for example, the collection of cess affected by counties lacked authority to place cess barrier points on the highways by KENHA  Lack of both a valuation roll and a law governing valuation and rating  Interference by the national government officials in the collection processes of certain revenues such as liquor license interfered with by County Commissioners.  The Ratings Act is too sketchy  There is lack of clarity on the government level with the capacity to impose tax on entertainment  There is still a heavy reliance on by-laws passed by the defunct local authorities Publishing the Laws 85 As a constitutional requirement, all counties publish their laws through the Government printer’s National Gazette. Besides the gazette, counties avail their legislations through:  Their websites  County assembly buildings  County revenue offices Without any county government’s input, some of these laws are made available to the public by the Kenya Law.Org (a non-governmental organisation) through its website. 3. REVENUE POLICY Determining Charges to Impose  It was noted that every county had its own ways of determining charges/fees to impose on the various taxable activities. There were varied responses given when asked to respond to how they determine the levies to impose. Some of these responses included: New rates being a result of consultations with the public and other stakeholders through public participation forums  New rates guided by existing laws especially the count y Finance Act  The rates a result of benchmarking between counties to compare and learn about the levies imposed by other counties  Rates determined by prevailing economic condition  Basing new rates on existing ones and the rates previously charged by the defunct local authorities County Revenue Policy Most counties reported they did not have a clear county revenue generation policy. Out of the 11 returned questionnaires, only one (Respondent 3) reported to have a tax policy. In their view the county’s existing policy is adequate. It is reported to have clear reasons and assumptions supporting the existing revenue streams, their rates and tax base for every stream. The rest of the counties had none with Respondent 8 reporting to be in the process of formulating a policy. Table 3 gives a summary of this. Table 3: Presence of Tax Policy per County Respondent Has tax policy Adequacy of policy 1 No N/A 2 No N/A 3 Yes Adequate 4 No N/A 5 No N/A 6 No N/A 7 No N/A 8 No Developing 9 No N/A 10 No N/A 11 No N/A Source: Computed from county OSR questionnaire responses The above implies that most counties are not adequately facilitated to formulate their own-source revenue policies. Apart from respondent 3, the other 8 counties did not mention having a plan to formulate a revenue policy. Revenue Streams with Insufficient or No Policy Rationale The following represents responses from the respondents asked if their counties had any revenue stream lacking in a policy rationale or whose guiding policy is insufficient. 86 Table 4: Responses on Existence of Revenue Stream with No/Insufficient Policy Rationale Response Frequency Percentage Yes 6 54.55% No 5 45.45% Total 11 100.00% Source: Computed from county OSR questionnaire responses Majority of the counties (54.55%) admitted having less than adequate policy guide for at least one existing revenue stream. The gap was not big compared to the counties whose policies were considered adequate, that is, 45.45% of the respondents. Among the identified fees which were found wanting by at least one county among the 54.55% of the counties that admitted such deficiencies include:  Parking fees  Agriculture produce cess  Property rates  Market fees and charges  Single business permits  Property rates,  Building plans approvals fees  Advertisement charges Consistency of County Revenue Policy with National Policy Majority of the respondents felt that their county’s revenue policies were consistent with that of the national government. Figure 2 illustrates how the responses were distributed among the respondents. Figure 2: Is Your County Revenue Policy Consistent with that of the National Government? No answer 20% No 20% Yes 60% Source: Computed from county OSR questionnaire responses Double Taxation within Counties This is how counties responded when asked whether there is any case of double taxation noticed in their revenue collection process: Table 5: Cases of Double Taxation 87 Response Frequency Percentage Yes 2 18.18% No 8 72.73% No answer 1 9.09% Total 11 100.00% Source: Computed from county questionnaire responses Most respondents (72.73%) believed that their systems lacked cases of double taxation with only two counties reporting a different condition. One questionnaire had no response. Addressing Double Taxation Among the two counties with reported double taxation cases, one described how they handle the scenarios while the other did not elaborate. The county which identified double taxation noted that it emanated from cases of businesses operating in different locations and so taxed in every location they operate. The county is currently handling this by issuing single business permits and basing the tax on permits other than taxing at location. Stakeholders Involvement in Changing Revenue Streams and Rates The questionnaire gauged how different stakeholders are involved in the process of changing revenue stream or rates. These were: the members of county assembly (MCAs), the county Governors, members of county business community, and the general public. This is how the respondents reacted: Table 6: Stakeholders Involvement Stakeholder Level of Involvement MCAs Debates and motions which approve such changes once tabled in county assembly County Governor Debates on changes tabled before cabinet (county executive committee) as member of the cabinet/committee before submitting to assembly for approval. Approves/Assents to bills/regulations Business Community Their views are sort in a number of public participation forums Other Stakeholders Their views sort in a number of public participation forums Source: Computed from county questionnaire responses Forecasting County OSR Revenue Notably, the revenue collection forecast is mainly based on these pre-eminent factors:  Previous years’ revenue collection performance,  Individual revenue stream tax base/ licenses issued,  The prevailing economic conditions and expected development trend,  Unrealised/unexploited revenue,  Conducting baseline surveys 4. REVENUE PAYMENTS Revenue Collectors The forms used to collect revenues are similar for the responding counties. Table 7 lists these forms, showing the number of respondents which affirmed their use in their counties, and the percentage such numbers constitute vis-a-vis other counties which responded. Table 7: Means of County Revenue Collection 88 Revenue Collector Observations % Counties Using County service providers 2 18.18% Country treasury office 7 63.64% County finance officials at taxpayer’s premises, 7 ^3.64% market etc. KRA 0 0.00% A collecting agent appointed by the County 1 9.09% Others 0 0.00% Source: Drawn from county OSR questionnaire answers Prescribed Payment Methods Most payments were reported to be done in the forms listed in Table 8. Table 8: Common Mean of Revenue Payments Mean of Payment Observations % Counties Using Cash 9 81.82% Bank cheque 11 100.00% Cash/cheque deposit at a bank 9 81.82% Electronic transfer 8 72.73% Other 5 45.45% Sources: Computer from county questionnaire Payment by bank cheque was the most preferred means; it was affirmed by all the respondents. The common alternative payment beside the 4 popular means was mobile phone transfer payment (mainly by M-pesa paybill services); this is used by 45.45% of the surveyed counties. 5. COMPLIANCE AND DEBT MANAGEMENT Existence of a Register or Database of Current or Potential Tax Payers In gauging the measures counties use to determine the number of people to tax, they were asked to verify if their counties have a register or database that lists current and potential taxpayers. Responses on whether the county had a database to identify existing and potential taxpayers are presented in table 8. Table 8: Counties with Database Listing Current and Potential Taxpayers Response Frequency Percentage (%) Yes 8 72.73% No 3 27.27% Blank 0 0.00% Total 11 100.00% Source: Computed from county questionnaire The counties which did not have records/database relied on other registers such as debtors ’ register, business register and plot registers to keep track of their revenue streams. This constituted 27.27% of the study’s respondents. Even though-some counties reported having no database of existing and potential taxpayers, all respondents reported to have a means of identifying potential taxpayers. Inter Government Information Sharing 89 Information sharing was reported mainly between the counties and national government institutions. Such feedback happens only when need arise and through official communications such as circulars and letters. To identify taxpayers, some respondents reported to have their counties linked to critical national government agencies like KRA, land ministry, ministry of public health among others. The counties use land registers from ministry of land to identify land owners and use approvals given by regulating bodies (such as ministry of public health) to businesses to identify taxpayers. One respondent reported the sharing of such information to be the base of taxpayers’ dat abase maintained by their county. Challenges with Revenue Non-Compliance All the responding counties reported to have some challenges with revenue non-compliance except one (respondent 8). The challenges they faced included:  Ineffective policies and laws or the lack of adequate enforcement legal framework  Lack of property rating and valuation legislation  Political interference  Weak enforcement unit to enforce compliance  Corruption among revenue collection officers  Out-dated databases inherited from the defunct local authorities  Absentee landlords making it hard to collect land rates  Scanty information on existing and potential taxpayers  Huge overdue collections inherited from defunct local authorities which continue to accumulate making it harder for the affected to comply  High poverty levels in some counties  Lack of effective and efficient revenue collection and management systems  Tax evasion and resistance to pay  Failure by some to register their property County Systems' Capability to Detect and Act on Arrears When asked whether they have a system to detect overdue payment and action on such, most respondents’ answer was in the affirmative. The ratios are displayed on table 9. Table 9: Systems Capable of detecting and Acting on Overdue Payments Response Frequency Percentage (%) Yes 9 81.82% No 2 18.18% Blank 0 0.00% Total 11 100.00% Source: Computed from county questionnaire Most counties have systems capable of detecting accounts late in payment and have a way of acting on it. Only 18.18% of the responding counties felt their systems were inadequate. Recovering Outstanding Amounts In order to recover outstanding revenue amounts, the counties used a number of methods. Below are the unique responses collected from all the respondents to the question asking how their counties recover outstanding debts:  Issuing demand notices and legal suit notices to overdue rate payers  Conducting follow-ups with the affected taxpayers  Confiscating property 90  Closing businesses with unpaid levies  Initiating legal action  Negotiating with defaulters  Issuing penalty waivers to encourage debt payment  Employing independent debt collectors to follow some debtors  Clamping vehicles  Continually updating taxpayers records  Eviction from county houses Many of the above answers were repeated when the respondents were asked about legal and administrative powers their county exercise to recover outstanding amounts. The additional information given was that those powers are governed by the county Finance Act, County Revenue Administration Act and other relevant Acts. Such powers are considered appropriate, sufficient, and clear to both officials and the public by a majority of responding county officers (54.55%). Their response distribution is captured on table 10. Table 10: Adequacy and Clarity of County Debt Collection Legal and Administrative Powers Response Frequency Percentage (%) Yes 5 45.45% No 6 54.55% Blank 0 0.00% Total 11 100.00% Source: Computed from county questionnaire 6. REVENUE ADMINISTRATION POLICY Administration of OSR It is clear the Finance Department of every county is the body in charge of collecting and managing all county revenue. All respondents (100%) from the counties selected the department as the sole role executor of OSR. The execution of OSR administration mandate is vested with the Chief Finance Officer or anybody who hold san equivalent position in a county. They perform this function through County Finance Management Unit which operates under the Department of Finance. Number of Officers in Revenue Collection The number of officers involved in revenue collection and administration varied from county to county. Respondent 6 reported the lowest number at 40 staff while the largest number of 500 staff was reported by respondent 4. Table 11 below summarizes this. Table 11: Summary Statistics for Recorded Revenue Officers Standard Variable Observations Mean Deviation Min Max Officers 11 217.18 122.99 40 500 Source: Computed from county questionnaire On average, the counties involved about 217 staff in the collection and administration of own source revenue with a deviation of about 122. Existence of OSR Procedures Manual 91 In the determination of the capacity of each county to train finance officers, they were asked if they had an own source revenue (OSR) procedures manual. Figure 3 outlines the results. Figure 3: Proportion of Counties with OSR Manual Blank Answer 9% No OSR Procedure Manual 36% Has OSR Procedure Manual 55% Source: Drawn from county OSR questionnaire responses More than half of the counties have a procedures manual that guides their revenue collection and administration. 36% of the counties reported to have no procedures manual. Procedures Used in Absence of a Manual Three counties which reported to have no manual stated that they used the following as a guide to their procedures;  National Treasury procedures  Financial regulations  County Financial Act Existence of Standard Training Manual Asked if their counties have a standard training manual, the respondents replied as follows, tabulated on table 11. Table 11: Having Standard Training Manual Response Frequency Percentage (%) Yes 5 45.45% No 6 54.55% Blank 0 0.00% Total 11 100.00% Source: Computed from county OSR questionnaire responses As shown in table 11, a majority (54.55%) of the counties lacked a standard manual to use in training their revenue administration staff. The responding officers for these counties suggested the following given the absence of a training manual in their counties:-  Offering of KRA’s tax administration course to all county revenue staff  Development of a standard training course on revenue administration and collection for all counties by the national government or through the council of governors Counties Offering Other Training 92 Counties were asked if they offer any training in relation to county revenue engagement processes. Their results are given on table 12 below. Table 12: Alternative Training by Counties Response Frequency Percentage (%) Yes 4 36.36% No 1 9.09% Blank 6 54.55% Total 11 100,00% Source: Computed from county OSR questionnaire responses County OSR Training Needs Respondents identified the following as among the areas that county revenue staff mostly need to be trained on:  Financial reporting, accountability, and fraud detection training  Revenue collection, enhancement, and administration capacity building,  Management, supervisory and, leadership skills,  Conflict resolutions and public relations  Customer service  Report writing, excel and computer training,  Debt management and recovery,  Basic legal training Minimum Qualification as OSR Administration/Collection Staff Most counties recruited their staff for the collection and administration based on set minimum qualifications. This represented 90.91%% of the responding counties. Only one county admitted having no minimum qualification requirement to be employed as a revenue officer. Table 13 illustrates: Table 13: Counties with Minimum Qualification for New Staff Response Frequency Percentage (%) Yes 10 90.91% No 1 9.09% Blank 0 0.00% Total 11 100.00% Source: Computed from county OSR questionnaire responses Number of Staff Qualified for their Roles Table 14 below is based on responses given when the responding officials were asked to give the number of officers qualified for their roles (column 3). In order to make it more meaningful, the responses were reinforced by asking for total revenue officers per county (column 2) to calculate the percentage of qualified staff per county (column 4). Table 14: County Revenue Staff: Total, Qualified and Percentage Qualified County Total Staff Qualified Staff % Qualified 1 210 60 28.57% 2 157 157 100.00% 93 3 250 200 80.00% 4 500 230 46.00% 5 140 70 50.00% 6 40 3 7.50% 7 345 100 28.99% 8 237 22 9.28% 9 140 0 0.00% 10 138 94 68.12% 11 210 60 12.93% Mean 217.18 87.82 39.22% Source: Computed from county OSR questionnaire responses The following can be deduced from table 14. First, the number of qualified staff ranges from 0 to 230 with a mean of 87 in absolute terms. Second, in percentages, the proportion of qualified staff varies from 0% to 100%. Among these, only 3 out of the 11 counties operate with above 50% qualified staff. Lastly; on average, 39.22% of revenue staff are qualified for their roles. This means the majority of county revenue officers are unqualified. Presence of a County Policy Unit Table 15 summarises answers given when responding officials were asked to verify if they have a revenue policy unit in their counties. Table 15: County Policy Unit Response Frequency Percentage (%) Yes 1 9.09% No 8 72.73% Blank 2 18.18% Total 11 100.00% Source: Computed from county OSR questionnaire responses Most counties lacked a revenue policy unit. From the above, 72.73% admitted to having no county unit with less than 10% of them answering in the affirmative. Therefore, there was little that was collected with regards to the role of such units. The only county which did attributed the role of the unit to emanate from the Public Service Management. 7. REVENUE IT INFRASTRUCTURE Capability of Existing IT Systems In order to capture the adequacy of existing county IT systems; the respondents were asked the questions to gauge: (i) if their county had a system for revenue administration and collection, and (ii) a system for revenue management and reporting. Their responses are presented on table 16. Table 16: Presence of Revenue Administration and Management IT System IT System Present Yes No % of Counties With For Revenue admin & collection 9 2 81.82% Revenue management & 10 1 90.91% reporting Source: Computed from county questionnaire 94 Description of Existing OSR IT System Given that every county had their own system in place, varied system descriptions were given. Below are some of the descriptions as captured in the respondents’ own words. “Revenue management and reporting through IFMIS and reporting through excel worksheets” (Respondent 1) “It is an electronic online revenue collection, generates of receipts and reports” (Respondent 2) “It comprises of Point of Sale (POS) gadgets with backups. The system is real time and incorporates mobile payment methods. This makes the system completely cashless.” (Respondent 3) “Capabilities are as per the National Treasury and Commission of revenue Allocation guidelines. It’s a GIS based” (Respondent 5) “It has front end users using point of sale devices to collect revenue and back end users for reporting” (respondent 10 ) In summary, the systems are mainly payment systems to capture collected revenue at the point of collection. In addition to this, they can generate some reports from such transactions. However, little was provided on their ability to help manage the collected revenue. Effectiveness and Challenges of the System Apart from one respondent, most respondents reported their systems to have some level of ineffectiveness. For one respondent, their report was that the existing system is “very effective, user friendly, reduces revenue leakage loop holes and provides real time reporting...is steady, fast and reliable” (Respondent 3). Some reported their system to be effective in some levels but not in all. Among the challenges faced were:  Incomplete data occasioned by loss of data and the systems inability to capture all revenue streams  System incapable of real-time auto reconciliation  System insufficient for revenue management and reporting, problem generating accounting reports  System based on defunct local authorities which used different geographical classification inconsistent with the current classification under county government  Some reported the system to be affected by frequent power failure, weak network connection, and entails tedious process  Some of the county officials are challenged with the new IT infrastructure The Ability of the IT System to Generate Accurate and Timely Reports In determining whether the counties have systems capable of reporting on critical revenue management functions, they were asked to confirm if their system could report accurately and timely on: (i) the taxpayer base, (ii) the OSR received, and (iii) the arrears of OSR. Table 17 reports on the answers given. Table 17: Reponses on Capability of County Revenue IT System to Accurately Report Accurately Reports on: Yes No % of Confirming Counties The taxpayer base 6 5 54.55% The OSR received 8 3 72.73% The arrears of OSR 6 5 54.55% Source: Computed from county OSR questionnaire responses It general, county IT systems can fairly generate accurate reports on the taxpayer base, OSR received, and be able to capture the arrears. All options had a confirmation rate of above 50% as seen on table 17 above. 95 8. INFORMATION AND COMMUNICATION Information Available to the Public At least all county governments share some information on OSR with the public. Mostly, the information they share are those availed through the county annual Finance Acts which, by law, must be published by the Kenyan Gazette. County Finance Acts contain information on: the existing county revenue streams, the applicable rates, applicable procedures, and payment forms. Some counties also make available their revenue reports with information on the revenue collected, means of payment, arrears, etc. Forms of Sharing OSR Information The above information is shared on different platforms. The extent at which each mean of communication was employed by the counties is shown in figure 4. Figure 4: Common Means Counties Use to Avail OSR Relevant Information to the Public 100.00% % of Counties Using a Platform 90.00% 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% Information Sharing Platform Source: Drawn from county OSR questionnaire responses From figure 4, it can be deduced that the most popular means of sharing information is by posting them on a county’s website. This was confirmed by 90.91% of the responding counties. This was followed closely by use of newspaper advertisement (81.82%) town hall meetings (81.82%), and posters in public places (72.73%). No county reported using podcast while television programmes was the second least used means. 96 Annex 2 - Questionnaire To be completed and returned to CountyOSR@adamsmithinternational.com, by Monday 14th May 2018 Background This study has been commissioned by The World Bank on behalf of the Kenya National Treasury and is being undertaken by a team from Adam Smith International between February and July 2018. A new draft policy and County Revenue Bill (2018) aim to broaden the County revenue base, to strengthen administrative capacity of counties to raise own revenues, and to provide a regulating framework for county imposition and variation of rates, in line with national tax policy and economic objectives. In order to inform the implementation of the policy, a better understanding is needed of the potential revenues possible at county level and of the barriers or opportunities available to counties to achieve more of that potential. The key objectives of the study are to:  Map out counties’ current local revenue base and estimated potential;  Estimate the scale of revenue ‘loss’ due to non-compliance and/or low capacity;  Support more credible projections by counties of future revenue from assigned taxes, fees, levies and charges;  Develop a framework for monitoring improvements in OSR performance by counties;  Review the legal framework for county revenue sources; and  Make policy recommendations for county revenue enhancement strategies. This questionnaire has therefore been designed to complement the use of secondary data, particularly to enable Counties to contribute directly to the study and ensure that any local issues, challenges or data are taken into consideration and help make the outputs of the study more realistic and of practical use. Please complete all Sections below: Name of County Name of Contact Person Designation Contact (Email / phone) 97 Note: “taxpayer” means a person liable to pay a County tax or a fee/charge in a County 1. Revenue Collections Kindly share with us data on revenue collections: this should be detailed by: (a) individual revenue stream (not aggregated, as in financial reports); and (b) for 3-5 years, depending on what is available. This is for purposes of this study only and not for official reporting to NT, so we can accept unaudited or incomplete records. We prefer Excel files, but if not convenient, please provide in any other format available e.g. Word, hard copy/Scanned. 2. Laws and Legal Framework a. Please describe the legal framework (laws, regulations, legal notices) currently in place in your County to define the revenue streams of your County. b. Please describe the legal framework (laws, regulations, legal notices) currently in place in your County to define and regulate the revenue collection and administration by your County. c. Please describe the legal framework (laws, regulations, legal notices) currently in place in your County to define the revenue management (accounting and reporting) by your County. d. Has your County undertaken a mapping exercise to match revenue streams to the appropriate legislative source/authority (YES/NO) ? 98 e. Have you identified revenue streams where there is no, insufficient or inappropriate legal authority for the revenue streams (YES/NO)? f. If yes, please describe any gaps or weaknesses in the law that you have identified. g. How are these laws published and made officially available? Please provide links to any laws etc. that are available online 3. Revenue Policy a. How does the County decide which rates and fees to charge? b. Is there a County Tax/Tariff/Fees & Charges Policy (YES/NO)? c. If yes, to what extent does the Policy explain the basis, assumptions and justification for the choice of revenue streams, the base for each stream and justification for each rate or amount? d. Are there any revenue sources for which there is no or insufficient policy ‘rationale’ (YES/NO)? If yes, which ones? e. Is the County Revenue Policy consistent with National policies in respect of Tax, Trade and the Economy (YES/NO)? If No, why or in what ways is it inconsistent? f. Are there examples of double taxation in your county (YES/NO)? If Yes, how do you address them? g. When the County considers making changes to revenue sources or rates, what engagement is there with: (a) the County Assembly members; 99 h. (b) The County Governor: i. (c) The business community, business representative organisations and j. (d) Other stakeholders? k. How do you forecast for revenue collection? For example: do you make forecasts based on previous years’ performance, or based on each tax base (e.g. number of licenses businesses) or another method? 4. Revenue payments a. Who collects the payment of taxes and fees/charges payable to the County? Indicate all that apply i. County service providers? ii. Country treasury office? iii. County finance officials at the taxpayer’s premises, market etc.? iv. KRA? v. A collecting agent appointed by the County? vi. Other? __(Describe)__ b. What are the prescribed payment methods for the above? Indicate all that apply i. Cash? ii. Bank cheque? iii. Cash/cheque deposit at a bank? iv. Electronic transfer? v. Other: _(Describe)_ 5. Compliance and debt management a. Is there a register or database of current/potential taxpayers (YES/NO)? If No, what is used to keep track of revenues? 100 b. Is there any process to identify potential taxpayers (YES/NO)? If NO, how do you currently identify them? c. How do you share information with other County or National institutions (e.g. Land Registry; Motor Vehicle Registry; KRA) to identify taxpayers? d. Does your County have any challenges with revenue non- compliance (YES/NO)? If Yes, what are the most significant challenges? e. If a tax/fee is not paid by the due payment date, does the County have a system to – i. Detect this, ii. Take action to follow up? f. How does the County recover outstanding amounts? g. What legal and administrative powers are exercised? How often? h. Are these powers: Appropriate? Sufficient? Clear to County officials and to the Public? 6. Revenue administration capacity a. Is the administration of OSR undertaken by – i. Finance Department ii. County Corporation iii. KRA (under MoU) iv. Agent (under contract with the County) b. Under what arrangements is the above done? c. Please provide a copy of any agency agreement or MOU (if necessary this can be redacted/edited before sharing) 101 d. How many officers are involved in administration and collection of OSR? e. Is there a procedures manual (YES/NO)? If No, what is used? f. Is there a standard finance training course covering the administration of the County revenue (YES/NO)? If No, what do you propose? g. Is any other training provided? What are your training needs? h. Are there any minimum qualifications needed to be employed in the administration /collection of OSR (YES/NO)? If No, what other selection criteria area considered? i. How many finance officers have the required qualifications for their role? j. Does the County have a Policy Unit? Describe its role, functions, expertise/skill sets, current activities: 7. Revenue IT Infrastructure a. Is there any IT system in place for – i. Revenue administration and Revenue collection ii. Revenue management and reporting? b. Describe the above system: 102 c. How effective is this system? Or what challenges have you encountered with the system? d. Can the system generate accurately and timely reports on: – i. The taxpayer base? ii. The OSR received? iii. The arrears of OSR? 8. Information and communication a. What sort of information of revenue / taxes and fees is available to the public? E.g. sources, rates, procedures, payment forms, policy, collections etc. b. Is the information available or shared in any of these forms? i. Leaflets/booklets ii. Website iii. Mobile phone applications iv. Podcasts v. Television programmes vi. Radio programmes vii. Newspaper articles/interviews viii. Newspaper advertisements ix. Posters in public places (e.g. markets) x. Letters xi. Town hall meetings (general public) xii. Business group forums xiii. Other: ___ 9. Any other comments Please provide any further comments, feedback or suggestions that you feel would be useful for this study to take into consideration: Thank you! Your contributions are greatly appreciated and we look forward to further engagement with you on this study. 103 Annex 3―Aggregate county revenue potential estimates (frontier analysis) Revenue potential and gaps by county (Ksh million) County Actual 2018 DEA (county GDP) 2018 DEA (county 2018 DEA (consumption, urbanization, 2015 DEA (income, urbanization, education)c collections consumption) education) (average Estimated Estimated Actual Estimated Actual collections Estimated Revenue gap Revenue gap Revenue gap Revenue gap FY16-FY17)a potential potential collectionsb potential FY15 potential Baringo 297.6 311.9 5% 678.1 56% 297.6 888.0 66% 249.7 332.4 25% Bomet 200.0 438.9 54% 618.3 68% 200.0 799.8 75% 206.4 288.9 29% Bungoma 496.5 849.2 42% 1,298.7 62% 496.5 1,803.6 72% 504.6 610.6 17% Busia 354.3 413.2 14% 478.7 26% 354.3 593.7 40% 315.2 405.1 22% Elgeyo Marakwet 112.7 188.8 40% 361.2 69% 112.7 420.4 73% 128.9 194.2 34% Embu 414.5 554.7 25% 629.4 34% 414.5 816.1 49% 401.1 524.6 24% Garissa 92.5 500.0 81% 290.3 68% 92.5 413.4 78% 130.7 168.8 23% Homa Bay 168.4 544.4 69% 955.2 82% 168.4 1,296.8 87% 157.9 235.9 33% Isiolo 102.6 99.1 -3% 145.8 30% 102.6 207.6 51% 133.7 212.4 37% Kajiado 604.0 1,757.1 66% 1,092.6 45% 604.0 1,499.5 60% 785.8 876.3 10% Kakamega 627.2 1,287.8 51% 1,529.2 59% 627.2 2,143.7 71% 516.9 631.9 18% Kericho 478.0 467.2 -2% 837.7 43% 478.0 1,123.5 57% 413.6 532.4 22% Kiambu 2,255.3 5,209.9 57% 2,969.7 24% 2,255.3 3,857.9 42% 2,110.9 2,506.6 16% Kilifi 1,017.3 1,760.0 42% 1,526.9 33% 952.9 2,140.4 55% 545.5 716.3 24% Kirinyaga 355.4 828.4 57% 766.8 54% 355.4 1,018.8 65% 311.6 439.4 29% Kisii 544.6 846.7 36% 1,174.6 54% 541.5 1,620.5 67% 296.8 398.4 26% Kisumu 994.4 1,055.9 6% 1,425.0 30% 994.4 1,990.0 50% 970.9 1,143.7 15% Kitui 362.5 658.7 45% 868.7 58% 362.5 1,169.3 69% 320.5 415.6 23% Kwale 234.8 1,593.7 85% 729.1 68% 214.9 963.2 78% 254.0 384.0 34% Laikipia 476.5 853.3 44% 476.5 0% 476.5 590.5 19% 400.5 562.7 29% Lamu 59.2 102.5 42% 142.5 58% 59.2 202.9 71% 61.7 118.5 48% Machakos 1,192.2 1,749.1 32% 1,606.7 26% 919.0 2,258.2 59% 1,356.6 1,597.4 15% Makueni 244.6 642.6 62% 897.5 73% 244.6 1,211.8 80% 215.3 301.6 29% Mandera 67.9 477.7 86% 361.2 81% 67.9 514.3 87% 87.7 106.8 18% Marsabit 137.9 203.2 32% 202.8 32% 137.4 186.6 26% 99.1 160.7 38% Meru 584.5 1,478.8 60% 1,653.3 65% 584.5 2,326.8 75% 539.2 661.0 18% Migori 371.9 502.2 26% 864.3 57% 367.3 1,162.7 68% 355.1 464.1 23% Mombasa 3,101.8 3,101.8 0% 3,101.8 0% 3,101.8 3,101.8 0% 2,492.6 3,314.3 24% Murang'a 591.9 1,695.5 65% 1,136.9 48% 584.8 1,564.9 63% 562.2 734.2 23% Nairobi 11,085.6 11,085.6 0% 11,085.6 0% 11,058.9 11,058.9 0% 11,500.0 15,291.2 24% Nakuru 2,170.3 3,951.1 45% 2,548.6 15% 2,170.3 3,455.5 37% 2,200.3 2,409.2 9% Nandi 219.6 481.7 54% 802.3 73% 219.6 1,071.2 79% 298.0 390.6 24% Narok 1,638.6 1,638.6 0% 1,638.6 0% 241.5 1,702.3 86% 1,639.2 6,830.0 24% Nyamira 116.7 288.0 59% 642.7 82% 116.7 835.7 86% 104.3 165.9 37% Nyandarua 291.0 431.9 33% 684.8 58% 291.0 897.8 68% 240.6 335.0 28% Nyeri 726.2 1,818.4 60% 1,194.5 39% 726.2 1,649.9 56% 680.7 892.2 24% Samburu 107.8 116.4 7% 152.5 29% 107.8 217.1 50% 195.7 262.2 25% 104 County Actual 2018 DEA (county GDP) 2018 DEA (county 2018 DEA (consumption, urbanization, 2015 DEA (income, urbanization, education)c collections consumption) education) (average Estimated Estimated Actual Estimated Actual collections Estimated Revenue gap Revenue gap Revenue gap Revenue gap FY16-FY17)a potential potential collectionsb potential FY15 potential Siaya 154.1 499.9 69% 886.5 83% 154.1 1,195.4 87% 143.3 213.3 33% Taita Taveta 209.4 381.9 45% 383.4 45% 205.3 453.1 55% 216.6 324.9 33% Tana River 26.0 310.7 92% 223.8 88% 26.0 318.7 92% 33.0 59.1 44% Tharaka Nithi 119.8 190.1 37% 438.8 73% 119.8 534.9 78% 115.7 147.3 21% Trans Nzoia 274.4 498.6 45% 1,021.7 73% 274.4 1,394.9 80% 301.3 395.7 24% Turkana 160.1 472.3 66% 638.3 75% 139.9 829.2 83% 126.5 194.5 35% Uasin Gishu 730.1 1,002.6 27% 1,145.8 36% 730.1 1,578.0 54% 800.8 985.2 19% Vihiga 182.3 543.2 66% 480.9 62% 182.3 597.0 69% 115.9 193.5 40% Wajir 78.5 424.7 82% 252.6 69% 78.5 359.7 78% 107.7 156.0 31% West Pokot 129.9 274.0 53% 376.7 66% 129.9 443.3 71% 103.9 163.8 37% Total 34,961.6 54,582.2 36% 53,417.6 35% 33,140.7 66,479.6 50% 33,848.5 48,448.3 24% Notes: a Used for 2018 DEA (county GDP) and 2018 DEA (county consumption). b Average collections over FY2015/16 and FY16/17 excluding ‘Council’s natural resource extraction’ revenue, which is uniquely h igh for Narok county. c Source: Table 5-5 of the Office of the Controller of Budget, Republic of Kenya, 2016. County Revenue Baseline Study 2015. Nairobi (pages 70-71). 105 Annex 4―County revenue potential estimates for six main sources (top-down analysis) Revenue potential and gaps for property taxes/rates County Actual collections Lower yield scenario (1% flat rate on all properties) Higher yield scenario (0.5% rate on low value and 1.5% on high value [maximum FY2015- properties) FY2018 (Jul-Mar)] Estimated potential Unrealized potential Revenue gap Estimated potential Unrealized potential Revenue gap Baringo unknown 267,737,838 unknown unknown 236,336,486 unknown unknown Bomet unknown 267,428,571 unknown unknown 254,701,681 unknown unknown Bungoma 41,782,861 771,833,333 730,050,472 95% 909,190,678 867,407,817 95% Busia 2,333,077 533,080,460 530,747,383 100% 613,701,149 611,368,072 100% Elgeyo/Marakwet unknown 228,166,667 unknown unknown 237,050,000 unknown unknown Embu 13,753,276 587,013,699 573,260,423 98% 647,379,703 633,626,427 98% Garissa unknown 283,709,756 unknown unknown 326,372,529 unknown unknown Homa Bay 1,312,424 506,837,838 505,525,414 100% 493,051,277 491,738,853 100% Isiolo unknown 128,246,667 unknown unknown 141,082,676 unknown unknown Kajiado unknown 3,920,310,345 unknown unknown 5,458,513,136 unknown unknown Kakamega 19,078,434 1,311,187,500 1,292,109,066 99% 1,639,043,545 1,619,965,111 99% Kericho 52,396,796 690,295,775 637,898,978 92% 772,980,975 720,584,179 93% Kiambu unknown 7,061,250,000 unknown unknown 9,357,459,416 unknown unknown Kilifi 212,623,214 1,643,112,676 1,430,489,462 87% 1,897,161,768 1,684,538,554 89% Kirinyaga 41,992,781 760,750,000 718,757,219 94% 825,875,000 783,882,219 95% Kisii unknown 953,666,667 unknown unknown 1,074,635,135 unknown unknown Kisumu 144,518,414 1,941,173,913 1,796,655,499 93% 2,392,793,656 2,248,275,242 94% Kitui 1,236,050 735,205,882 733,969,832 100% 802,808,824 801,572,774 100% Kwale 53,682,884 867,649,123 813,966,239 94% 1,067,746,411 1,014,063,527 95% Laikipia unknown 558,607,595 unknown unknown 623,729,574 unknown unknown Lamu 10,268,749 271,914,493 261,645,744 96% 351,354,092 341,085,343 97% Machakos 183,794,860 1,974,476,190 1,790,681,330 91% 2,300,571,429 2,116,776,569 92% Makueni 11,683,488 1,444,793,103 1,433,109,615 99% 1,751,534,483 1,739,850,995 99% Mandera 6,471,717 260,218,987 253,747,270 98% 285,940,169 279,468,452 98% Marsabit unknown 192,853,659 unknown unknown 232,878,049 unknown unknown Meru 36,981,926 1,669,234,568 1,632,252,642 98% 1,572,601,852 1,535,619,926 98% Migori unknown 427,444,444 unknown unknown 441,870,892 unknown unknown Mombasa unknown 5,347,352,941 unknown unknown 7,080,103,486 unknown unknown Murang'a 44,569,292 1,011,000,000 966,430,708 96% 1,056,366,667 1,011,797,375 96% Nairobi 3,160,240,000 34,243,809,524 31,083,569,524 91% 48,103,822,394 44,943,582,394 93% Nakuru unknown 3,350,273,973 unknown unknown 3,804,284,198 unknown unknown Nandi 12,429,790 456,835,443 444,405,653 97% 501,516,322 489,086,532 98% Narok unknown 1,331,289,157 unknown unknown 1,673,810,278 unknown unknown Nyamira 27,995,599 499,603,175 471,607,576 94% 531,468,254 503,472,655 95% Nyandarua unknown 773,333,333 unknown unknown 803,222,222 unknown unknown Nyeri unknown 1,147,000,000 unknown unknown 1,229,787,671 unknown unknown Samburu 6,435,854 112,778,947 106,343,093 94% 108,268,421 101,832,567 94% Siaya unknown 609,297,297 unknown unknown 600,945,946 unknown unknown Taita/Taveta 18,350,359 517,500,000 499,149,641 96% 576,916,667 558,566,308 97% Tana River 3,936,579 150,622,727 146,686,148 97% 153,797,727 149,861,148 97% Tharaka-Nithi unknown 210,427,027 unknown unknown 219,720,541 unknown unknown 106 Trans Nzoia 16,821,107 1,189,176,471 1,172,355,364 99% 1,528,549,020 1,511,727,913 99% Turkana 80,430 1,056,454,545 1,056,374,115 100% 1,349,015,152 1,348,934,722 100% Uasin Gishu 88,574,604 1,421,869,565 1,333,294,961 94% 1,660,727,425 1,572,152,821 95% Vihiga unknown 368,333,333 unknown unknown 401,612,676 unknown unknown Wajir unknown 128,718,182 unknown unknown 134,158,225 unknown unknown West Pokot 655,280 122,092,537 121,437,257 99% 121,702,442 121,047,162 99% Total 4,213,999,845 (known) 84,305,967,926 unknown unknown 108,348,160,319 unknown unknown Note: Actual collections are reflecting the maximum a county collected of the particular revenue source over the period FY2014/15 to FY2017/18 (first 9 months). Once a comprehensive county dataset is available, this can be replaced with collections of the most recent year, and unrealized revenue potential and gaps calculated for counties for which actual collections are currently unknown. Revenue potential and gaps for building permits County Actual collections Lower yield scenario (1% flat rate on all construction) Higher yield scenario (1% rate on low value and 2% rate on high value [maximum FY2015- construction) FY2018 (Jul-Mar)] Estimated potential Unrealized potential Revenue gap Estimated potential Unrealized potential Revenue gap Baringo Unknown 59,370,385 unknown unknown 89,922,298 unknown unknown Bomet Unknown 103,297,548 unknown unknown 173,812,700 unknown unknown Bungoma Unknown 116,251,509 unknown unknown 191,392,257 unknown unknown Busia Unknown 79,792,775 unknown unknown 124,362,739 unknown unknown Elgeyo/Marakwet Unknown 86,151,058 unknown unknown 161,342,970 unknown unknown Embu 5,707,504 68,080,055 62,372,551 92% 121,043,293 115,335,789 95% Garissa Unknown 38,545,867 unknown unknown 63,610,373 unknown unknown Homa Bay Unknown 95,740,508 unknown unknown 150,144,584 unknown unknown Isiolo Unknown 24,413,748 unknown unknown 38,012,633 unknown unknown Kajiado Unknown 131,706,568 unknown unknown 242,193,745 unknown unknown Kakamega Unknown 136,616,445 unknown unknown 200,458,361 unknown unknown Kericho 8,365,000 97,606,020 89,241,020 91% 173,521,813 165,156,813 95% Kiambu Unknown 340,111,550 unknown unknown 626,354,501 unknown unknown Kilifi Unknown 70,206,330 unknown unknown 113,808,156 unknown unknown Kirinyaga 4,547,705 80,745,028 76,197,323 94% 134,540,761 129,993,056 97% Kisii Unknown 80,505,839 unknown unknown 118,147,758 unknown unknown Kisumu 42,727,457 524,943,907 482,216,450 92% 987,220,092 944,492,635 96% Kitui Unknown 92,847,030 unknown unknown 154,710,966 unknown unknown Kwale 4,632,580 40,982,884 36,350,304 89% 67,446,706 62,814,126 93% Laikipia Unknown 57,049,415 unknown unknown 103,947,071 unknown unknown Lamu Unknown 1,209,887 unknown unknown 1,209,887 unknown unknown Machakos 224,736,091 724,315,771 499,579,680 69% 1,414,410,948 1,189,674,857 84% Makueni 6,438,310 100,487,529 94,049,219 94% 177,460,975 171,022,665 96% Mandera Unknown 35,407,563 unknown unknown 56,627,539 unknown unknown Marsabit Unknown 19,322,895 unknown unknown 29,144,533 unknown unknown Meru Unknown 150,400,466 unknown unknown 219,560,995 unknown unknown Migori Unknown 84,342,862 unknown unknown 126,821,217 unknown unknown Mombasa Unknown 379,978,971 unknown unknown 747,150,852 unknown unknown Murang'a Unknown 121,050,050 unknown unknown 215,569,964 unknown unknown Nairobi 1,653,800,000 1,163,499,911 (490,300,089) -42% 2,326,999,821 673,199,821 29% Nakuru Unknown 81,592,287 unknown unknown 123,232,488 unknown unknown Nandi Unknown 37,169,077 unknown unknown 45,194,219 unknown unknown 107 Narok Unknown 33,895,552 unknown unknown 52,404,307 unknown unknown Nyamira Unknown 84,601,391 unknown unknown 146,798,461 unknown unknown Nyandarua Unknown 31,380,560 unknown unknown 43,287,010 unknown unknown Nyeri Unknown 132,066,898 unknown unknown 240,912,033 unknown unknown Samburu Unknown 7,740,285 unknown unknown 7,740,285 unknown unknown Siaya Unknown 138,883,599 unknown unknown 239,898,937 unknown unknown Taita/Taveta Unknown 70,788,264 unknown unknown 125,081,115 unknown unknown Tana River Unknown 10,173,111 unknown unknown 10,173,111 unknown unknown Tharaka-Nithi Unknown 33,403,861 unknown unknown 53,793,231 unknown unknown Trans Nzoia Unknown 92,313,611 unknown unknown 161,697,074 unknown unknown Turkana Unknown 20,489,781 unknown unknown 25,087,488 unknown unknown Uasin Gishu Unknown 116,652,446 unknown unknown 203,576,949 unknown unknown Vihiga Unknown 17,933,977 unknown unknown 22,702,549 unknown unknown Wajir Unknown 17,507,890 unknown unknown 27,007,996 unknown unknown West Pokot Unknown 18,302,637 unknown unknown 22,749,041 unknown unknown Total 1,950,954,647 (known) 6,049,875,600 unknown unknown 10,902,286,802 unknown unknown Note: Actual collections are reflecting the maximum a county collected of the particular revenue source over the period FY2014/15 to FY2017/18 (first 9 months). Once a comprehensive county dataset is available, this can be replaced with collections of the most recent year, and unrealized revenue potential and gaps calculated for counties for which actual collections are currently unknown. Revenue potential and gaps for business licences and liquor licences County Business licences Liquor licences Actual collections Estimated potential Unrealized potential Revenue Actual collections Estimated potential Unrealized potential Revenue gap gap Baringo 46,431,538 133,882,876 87,451,338 65% unknown 21,415,278 unknown unknown Bomet unknown 45,468,251 unknown unknown unknown 24,875,981 unknown unknown Bungoma 91,283,142 162,648,864 71,365,722 44% unknown 61,389,582 unknown unknown Busia 43,034,366 70,145,322 27,110,956 39% unknown 9,897,137 unknown unknown Elgeyo/Marakwet 12,642,375 16,630,698 3,988,323 24% unknown 4,010,849 unknown unknown Embu 89,090,299 125,546,593 36,456,294 29% unknown 44,629,651 unknown unknown Garissa 9,520,010 13,539,151 4,019,141 30% unknown 756,906 unknown unknown Homa Bay 30,376,180 227,158,407 196,782,227 87% unknown 97,774,761 unknown unknown Isiolo unknown 13,823,764 unknown unknown unknown 10,621,315 unknown unknown Kajiado unknown 515,495,302 unknown unknown unknown 193,986,650 unknown unknown Kakamega 91,237,493 685,510,660 594,273,167 87% unknown 99,057,941 unknown unknown Kericho 48,602,000 204,834,194 156,232,194 76% 1,290,500 116,890,645 115,600,145 99% Kiambu 617,409,070 1,005,632,207 388,223,137 39% unknown 410,957,806 unknown unknown Kilifi 66,067,805 203,527,141 137,459,336 68% unknown 70,174,366 unknown unknown Kirinyaga 99,862,816 125,467,788 25,604,972 20% 44,453,876 49,539,222 5,085,346 10% Kisii 66,688,755 153,828,293 87,139,538 57% unknown 88,593,958 unknown unknown Kisumu 109,192,352 903,584,036 794,391,684 88% 10,912,848 357,289,132 346,376,285 97% Kitui 101,194,240 44,692,899 (56,501,341) -126% unknown 12,219,786 unknown unknown Kwale 58,603,940 66,664,275 8,060,335 12% unknown 50,934,396 unknown unknown Laikipia unknown 82,491,730 unknown unknown unknown 38,420,967 unknown unknown Lamu 8,190,200 40,882,310 32,692,110 80% unknown 28,673,035 unknown unknown Machakos 198,978,358 1,447,366,296 1,248,387,938 86% 49,323,970 660,424,957 611,100,987 93% Makueni 66,533,409 80,976,567 14,443,158 18% 32,685,505 55,637,494 22,951,989 41% 108 Mandera 10,266,950 48,102,235 37,835,285 79% unknown 4,335,318 unknown unknown Marsabit 17,354,370 6,431,400 (10,922,970) -170% unknown 2,516,375 unknown unknown Meru 130,306,318 417,524,327 287,218,009 69% unknown 369,985,237 unknown unknown Migori 57,092,030 157,263,686 100,171,656 64% unknown 78,106,252 unknown unknown Mombasa 403,819,234 703,947,152 300,127,918 43% unknown 184,825,857 unknown unknown Murang'a 151,599,620 66,468,578 (85,131,042) -128% unknown 26,945,459 unknown unknown Nairobi 1,814,200,000 11,837,356,732 10,023,156,732 85% unknown 6,249,384,365 unknown unknown Nakuru 430,281,392 1,922,671,688 1,492,390,296 78% unknown 259,713,489 unknown unknown Nandi 30,589,401 45,188,231 14,598,830 32% unknown 17,102,516 unknown unknown Narok 21,472,633 327,571,924 306,099,291 93% unknown 90,597,217 unknown unknown Nyamira 51,595,600 185,693,107 134,097,507 72% unknown 54,611,373 unknown unknown Nyandarua 84,861,880 66,127,164 (18,734,716) -28% unknown 11,902,520 unknown unknown Nyeri 94,132,281 372,837,073 278,704,792 75% unknown 98,660,732 unknown unknown Samburu 4,644,960 23,128,067 18,483,107 80% unknown 14,755,692 unknown unknown Siaya 32,899,734 182,536,098 149,636,364 82% unknown 30,161,979 unknown unknown Taita/Taveta 47,613,598 98,445,669 50,832,071 52% unknown 50,822,746 unknown unknown Tana River 8,432,789 5,887,021 (2,545,768) -43% unknown 3,522,041 unknown unknown Tharaka-Nithi unknown 33,667,621 unknown unknown unknown 15,682,545 unknown unknown Trans Nzoia 61,386,301 67,026,049 5,639,748 8% unknown 19,963,992 unknown unknown Turkana 43,638,750 31,962,117 (11,676,633) -37% unknown 18,414,638 unknown unknown Uasin Gishu 218,481,639 323,328,460 104,846,821 32% unknown 92,659,411 unknown unknown Vihiga 20,315,889 33,734,484 13,418,595 40% unknown 4,707,830 unknown unknown Wajir unknown 59,129,923 unknown unknown unknown 311,737 unknown unknown West Pokot 7,634,980 5,798,487 (1,836,493) -32% unknown 2,457,032 unknown unknown Total 5,597,558,697 (known) 23,391,624,914 unknown unknown 138,666,699 (known) 10,210,318,169 unknown unknown Note: Actual collections are reflecting the maximum a county collected of the particular revenue source over the period FY2014/15 to FY2017/18 (first 9 months). Once a comprehensive county dataset is available, this can be replaced with collections of the most recent year, and unrealized revenue potential and gaps calculated for counties for which actual collections are currently unknown. 109 Revenue potential and gaps for parking fees and advertisement fees County Parking fees Advertisement fees Actual collections Estimated potential Unrealized potential Revenue gap Actual collections Estimated potential Unrealized potential Revenue gap Baringo Unknown 32,677,642 unknown unknown unknown 2,671,052 unknown unknown Bomet Unknown 4,600,776 unknown unknown unknown 2,136,192 unknown unknown Bungoma 67,299,456 101,270,700 33,971,244 34% unknown 2,109,305 unknown unknown Busia 142,221,660 33,331,985 (108,889,675) -327% unknown 2,203,918 unknown unknown Elgeyo/Marakwet 3,610,700 6,623,551 3,012,851 45% unknown 280,752 unknown unknown Embu 32,754,170 38,164,078 5,409,908 14% 896,280 10,284,137 9,387,857 91% Garissa 10,838,785 74,310,257 63,471,472 85% unknown 13,022,482 unknown unknown Homa Bay 27,649,190 34,757,598 7,108,408 20% unknown 21,318,870 unknown unknown Isiolo Unknown 39,819,802 unknown unknown unknown 2,175,402 unknown unknown Kajiado Unknown 236,812,311 unknown unknown unknown 151,948,623 unknown unknown Kakamega 59,800,975 87,887,867 28,086,892 32% unknown 193,751,331 unknown unknown Kericho 48,383,000 47,227,765 (1,155,235) -2% 2,131,950 14,130,398 11,998,448 85% Kiambu 279,489,985 1,195,668,067 916,178,082 77% unknown 245,274,847 unknown unknown Kilifi 22,455,474 210,755,008 188,299,534 89% unknown 8,512,092 unknown unknown Kirinyaga 19,782,900 29,495,288 9,712,388 33% 3,559,360 2,731,805 (827,555) -30% Kisii 85,064,970 74,128,609 (10,936,361) -15% unknown 4,054,099 unknown unknown Kisumu 206,888,650 354,904,807 148,016,157 42% 60,772,871 2,111,158,483 2,050,385,612 97% Kitui 22,219,585 27,142,767 4,923,182 18% unknown 2,746,771 unknown unknown Kwale 11,054,159 59,556,216 48,502,057 81% 12,905,110 18,411,898 5,506,788 30% Laikipia Unknown 62,103,329 unknown unknown unknown 6,825,583 unknown unknown Lamu Unknown - unknown unknown unknown 4,411,860 unknown unknown Machakos 129,153,300 201,737,913 72,584,613 36% 35,013,041 161,965,934 126,952,893 78% Makueni 25,841,269 3,907,851 (21,933,418) -561% 4,111,260 1,221,189 (2,890,071) -237% Mandera 312,720 42,112,092 41,799,372 99% unknown 2,620,817 unknown unknown Marsabit 1,975,740 17,825,934 15,850,194 89% unknown 117,854 unknown unknown Meru 72,585,487 62,220,649 (10,364,838) -17% unknown 3,896,588 unknown unknown Migori 82,482,110 78,680,558 (3,801,552) -5% unknown 30,756,605 unknown unknown Mombasa 513,941,197 1,508,770,647 994,829,450 66% unknown 102,705,337 unknown unknown Murang'a 66,578,580 32,209,919 (34,368,661) -107% unknown 1,248,910 unknown unknown Nairobi 2,037,870,304 6,655,282,681 4,617,412,377 69% 720,000,000 2,734,970,194 2,014,970,194 74% Nakuru 292,414,437 527,696,439 235,282,002 45% unknown 309,901,885 unknown unknown Nandi 35,222,291 23,465,992 (11,756,299) -50% unknown 3,007,438 unknown unknown Narok 1,655,695 44,322,856 42,667,161 96% unknown 9,858,684 unknown unknown Nyamira 31,553,300 15,716,633 (15,836,667) -101% unknown 6,453,163 unknown unknown Nyandarua 19,394,794 6,450,481 (12,944,313) -201% unknown 1,766,288 unknown unknown Nyeri 104,568,670 84,539,341 (20,029,329) -24% unknown 39,089,952 unknown unknown Samburu 6,000 8,718,147 8,712,147 100% unknown 1,415,840 unknown unknown Siaya 15,713,913 43,802,986 28,089,073 64% unknown 442,780 unknown unknown Taita/Taveta 15,454,744 38,137,014 22,682,270 59% unknown 8,495,025 unknown unknown Tana River 1,854,300 - (1,854,300) unknown 242,739 unknown unknown Tharaka-Nithi Unknown 8,087,795 unknown unknown unknown 2,529,892 unknown unknown Trans Nzoia 37,774,873 70,458,311 32,683,438 46% unknown 708,442 unknown unknown Turkana 3,647,077 45,545,033 41,897,956 92% unknown 3,062,572 unknown unknown Uasin Gishu 174,030,936 243,185,624 69,154,688 28% unknown 24,306,714 unknown unknown Vihiga 39,260,860 31,205,375 (8,055,485) -26% unknown 279,617 unknown unknown 110 Wajir Unknown 9,576,561 unknown unknown unknown 207,513 unknown unknown West Pokot 6,733,539 12,147,632 5,414,093 45% unknown 68,591 unknown unknown Total 4,749,539,795 (known) 12,567,042,884 unknown unknown 839,389,872 (known) 6,271,500,467 unknown Note: Actual collections are reflecting the maximum a county collected of the particular revenue source over the period FY2014/15 to FY2017/18 (first 9 months). Once a comprehensive county dataset is available, this can be replaced with collections of the most recent year, and unrealized revenue potential and gaps calculated for counties for which actual collections are currently unknown. 111 Annex 5: List of consultation meetings Participant Organisations Purpose Date Location ASI, NT, World Bank, CRA, To introduce project team to NT and discuss 15th March National Treasury COB, KNBS, KRA initial work. ASI, NT, World Bank, CRA, Presentation on project methodology, 20th March National Treasury KNBS, KRA, Ministry of introduction of project to NT Permanent Lands Secretary, discussion on project priorities ASI, NT, KRA To discuss project data requirements and 22nd March KRA introduce the project in more detail ASI, NT, KNBS To discuss project data requirements and 23rd March KNBS introduce the project in more detail ASI, NT, Ministry of Lands To discuss project data requirements and 23rd March Ministry of Lands introduce the project in more detail ASI, NT, COB To discuss project data requirements and 27th March National Treasury introduce the project in more detail Annex NT, COG To discuss project data requirements and 4th April Council of Governors introduce the project in more detail HQ ASI, NT, Institute of To discuss issues relating to administration 24th April ICPA, Upper Hill, Certified Public of county level tax and fees Nairobi Accountants ASI, NT, Nairobi County To discuss issues and challenges pertinent 25th April NCC, City Hall to OSR collection at the county level ASI, NT, National To discuss issues and challenges pertinent 25th April NTA Offices, Nairobi Taxpayers Association to OSR collection at the county level ASI, NT, Machakos County To discuss issues and challenges pertinent 26th April Machakos County, to OSR collection at the county level Athi River ASI, NT, Kiambu County To discuss issues and challenges pertinent 27th April Kiambu County to OSR collection at the county level ASI, NT, Commission on To discuss project data requirements and 3rd May CRA Offices, Revenue Allocation introduce the project in more detail. Also to Riverside Drive, gather ideas on what would be useful for Nairobi CRA and feedback on methodology. ASI, NT, Kenya Law Issues relating to legislative drafting and 9TH May KLRC Offices, CBD Reform Commission legal basis for county OSR ASI, NT, Kakamega To discuss issues and challenges pertinent 22nd May Kakamega County HQ County to OSR collection at the county level 112 www.adamsmithinternational.com Headquarters Africa South Asia Asia Pacific 240 Blackfriars Road 2nd Floor Cavendish Bharat Yuvak Bhawan 301/88 Foveaux Street London 14 Riverside, Riverside 1 Jai Singh Road Surry Hills SE1 8NW Drive New Delhi 110 001 Sydney NSW 2010 United Kingdom PO Box 26721-00100 India Australia T: +44 20 7735 6660 Nairobi T: +91 11 4150 T: +61 2 8265 0000 Kenya 2291/93/94/95 T: +254 20 444 4388 Copyright © 2014 Adam Smith International. All rights reserved.