69094 Sustainable Development — East Asia and Pacific Region D I S C U S S I O N P A P E R S P h i l ippines The KALAHI-CIDSS Impact Evaluation A Revised Synthesis Report January 2013 © 2013 The World Bank Group January 2013 Printed in the United States All rights reserved. The World Bank Group World Bank Office Manila 1818 H. Street, N.W. 25th Floor, One Global Place Washington DC 20433, USA 5th Avenue, Bontacio Global City Tel: (202) 473 1000 Taguig City, Philippines Fax: (202) 477 6391 Tel: (632) 465 2500 www.worldbank.org Fax: (632) 465 2505 www.worldbank.org.ph This note has been prepared by Julien Labonne. Thanks to Fermin Adriano, Mae Arevalo Radu Ban, Amanda Beatty, Sean Bradley, Cliff Burkley, Robert Chase, Patricia Fernandes, Alex Glova, Malu Padua, Anne Pizer, Matt Stephens, Lawrence Tang, Susan Wong, and Mark Woodward for comments on previous drafts. This publication is available online at http://www.worldbank.org/. The pictures on the cover page were taken from: http://kalahi.dswd.gov.ph/index.php?option=com_phocagal- lery. This volume is a product of the staff and consultants of the Sustainable Development Department of the East Asia and Pacific Region of the World Bank / The World Bank. The findings, interpretations and conclusions expressed herein are those of the author(s) and do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denomi- nations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Pub- lisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax 202-522-2422; e-mail pubrights@world- bank.org. Note: This report presents revised estimates of the KALAHI-CIDSS impacts on household welfare, access to services, social capital and local governance. Earlier estimates were reported in a report prepared by Asia Pacic Policy Center (2011) and summarized in World Bank (2011a). The original estimates of APPC were revised as an error was made in generating the results. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report* PHILIPPINES January 2013  This report was prepared by Julien Labonne. I am grateful to Fermin Adriano, Mae Arevalo, Radu Ban, Amanda Beatty, Sean * Bradley, Cliff Burkley, Robert Chase, Patricia Fernandes, Alex Glova, Malu Padua, Anne Pizer, Simon Quinn, Matt Stephens, Lawrence Tang, Susan Wong, and Mark Woodward for comments on previous versions of this note. Table of Contents Executive Summary...................................................................................................................5 1. KALAHI-CIDSS.........................................................................................................................9 2. Background on the Evaluation Strategy......................................................................... 12 2.1 Design.......................................................................................................................................................................... 12 .............................................................................................................. 13 2.2 Testing the Parallel Trend Hypothesis. 2.3 The Regressions........................................................................................................................................................ 13 2.4 Implementation and Challenges....................................................................................................................... 14 3. Results of KALAHI-CIDSS in terms of Welfare................................................................. 16 3.1 Who did the project reach?.................................................................................................................................. 16 3.2 Key welfare impacts................................................................................................................................................ 16 3.3 Access to basic services......................................................................................................................................... 18 4. The Results of KALAHI-CIDSS in terms of Governance and Social Capital............... 19 5. Areas for Improvement...................................................................................................... 21 ................................................................................................................... 21 5.1 Implications for project expansion. 5.2 Suggestions for additional analytical work.................................................................................................... 22 References................................................................................................................................ 23 Annexes..................................................................................................................................... 63 List of Tables Table 1. KALAHI-CIDSS Impact Evaluation Report Card.......................................................................................7 Table 2. Distribution of subproject types (December, 2010).......................................................................... 10 Table 3: Impacts on log per capita expenditures ............................................................................................... 25 Table 4: Impacts on log per capita expenditures (poor households) ......................................................... 26 Table 5: Impacts on log per capita expenditures (non-poor households) . .............................................. 27 Table 6: Impacts on poverty levels........................................................................................................................... 28 Table 7: Impacts on non-food share of total consumption............................................................................. 29 Table 8: Impacts on log per capita food expenditures...................................................................................... 30 Table 9: Impacts on log per capita non-food expenditures............................................................................ 31 Table 10: Impacts on self-rated poverty levels..................................................................................................... 32 Table 11: Impacts on employment........................................................................................................................... 33 Table 12: Impacts on male employment................................................................................................................ 34 Table 13: Impacts on female employment............................................................................................................ 35 Table 14: Impacts on house accessibility............................................................................................................... 36 Table 15: Impacts on number of trips to municipal center............................................................................. 37 The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  3 ............................................................... 38 Table 16: Impacts on log per capita transportation expenditures. Table 17: Impacts on access to level II and III water supply............................................................................. 39 Table 18: Impacts on access to safe water............................................................................................................. 40 Table 19: Impacts on access to water-sealed toilets.......................................................................................... 41 Table 20: Impacts on access to health services.................................................................................................... 42 Table 21: Impacts on access to health services (Male sample)....................................................................... 43 Table 22: Impacts on access to health services (Female sample).................................................................. 44 Table 23: Impacts on school enrollment................................................................................................................ 45 Table 24: Impacts on school enrollment for boys............................................................................................... 46 Table 25: Impacts on school enrollment for girls................................................................................................ 47 Table 26: Impacts on attendance in village assemblies.................................................................................... 48 Table 27: Impacts on willingness to contribute money to community projects..................................... 49 Table 28: Impacts on trust that others are willing to help if needed........................................................... 50 Table 29: Impacts on willingness to contribute time to community projects.......................................... 51 Table 30: Impacts on participation in bayanihan................................................................................................ 52 Table 31: Impacts on group membership.............................................................................................................. 53 Table 32: Impacts on trust of community members.......................................................................................... 54 Table 33: Impacts on need to be alert..................................................................................................................... 55 Table 34: Impacts on trust of other with money................................................................................................. 56 Table 35: Impacts on trust of local officials............................................................................................................ 57 .................................................................................................... 58 Table 36: Impacts on trust of national officials. Table 37: Impacts on trust of strangers................................................................................................................... 59 Table 38: Impacts on perceptions of peace........................................................................................................... 60 Table 39: Impacts on knowledge of village budget........................................................................................... 61 Table 40: Impacts on participation in planning of development activities............................................... 62 Table A-1: Comparing treatment and control municipalities at baseline (Welfare indicators).......... 64 Table A-2: Comparing treatment and control municipalities at baseline (Access indicators)............ 65 Table A-3: Comparing treatment and control municipalities at baseline (Social capital and local governance indicators)................................................................................................................................................. 66 Table A-4: Parallel Trend Hypothesis: Consumption.......................................................................................... 67 Table A-5: Parallel Trend Hypothesis: Asset........................................................................................................... 68 Table A-6: Are the determinants of attrition different in treatment and control municipalities? (Welfare indicators)........................................................................................................................................................ 69 Table A-7: Are the determinants of attrition different in treatment and control municipalities? (Access indicators).......................................................................................................................................................... 70 Table A-8: Are the determinants of attrition different in treatment and control municipalities? (Social capital and local governance indicators)................................................................................................. 71 List of Boxes and Figures Box 1. Political Engagement: Virgie Niebres, Barangay Rawis, Pio Duran.................................................. 20 Figure 1. Distribution of subproject types in sample municipalities........................................................... 24 Figure A-1: Project Coverage...................................................................................................................................... 63 Executive Summary T he KALAHI-CIDSS program was set up households in treated municipalities indicated in 2002 to alleviate rural poverty. The being aware of the project and three in every program, following a Community-Driven five of them expressed their satisfaction with the Development (CDD) approach, aims to achieve project. Local, elected officials also viewed the this by pro­viding resources to poor rural munici- project in a positive light, with 75 percent of Lo- palities to invest in public goods and by reviving cal Government Units (LGUs) officials indicat- local institutions to enhance peoples’ participa- ing being satisfied with the project. Respondents tion in governance. KALAHI-CIDSS originally identify infrastructure improvement, better ac- targeted the poorest 25 percent of municipalities cess to services and community empowerment in 42 of the poorest provinces. As of December as key project benefits. Feedback from baran- 2010, the project had covered 4,583 barangays gays that were not prioritized by the Municipal (communities) in 200 municipali­ ties and sup- Inter-Barangay Forum (MIBF), and therefore ported 5,645 sub-projects, worth PHP 5.7 billion did not receive sub-project financing, was more (about USD 140 million)1 and bene�ting about negative. 1.26 million households.2 Communities follow very detailed participatory processes to secure The KALAHI-CIDSS was designed to minimize resources for planning and implementation of the risk of elite capture and it appears to have public investments. been successful in doing so. At the national lev- el, the program directed resources to some of the A rigorous impact evaluation was designed in poorest municipalities in the country, identi�ed 2003 to evaluate general impacts on poverty re- through a ranking process undertaken by Dr. Balisacan at the University of the Philippines duction, social capital, empowerment, and gov- School of Economics. At the local level, avail- ernance. Quantitative and qualitative data were able evidence indicates that project processes collected in 2003, 2006 and 2010 on a broad were not subject to elite capture, at least in its range of indicators from a sample of KALAHI- most malign form. First, barangay captains do CIDSS municipalities and of comparable muni- not appear to be a driving force behind proposals cipalities that did not receive project support. put forward in the MIBF. Their preferences and The report presents the main results from the those of community members are equally repre- �nal quantitative and qualitative impact evalua- sented in community proposals. Second, the im- tions as well as from other studies that were car- pact evaluation reveals that, within municipali- ried out throughout project implementation. ties, KALAHI-CIDSS targeted the poorest and best-organized villages, suggesting that better- Available data indicate that participation rates off and connected individuals and villages did in project activities were relatively high, sug- not receive a disproportionate share of project gesting that households and local elected offi- bene�ts. cials in targeted municipalities see value in the KALAHI-CIDSS approach. About 80 percent of The project had a positive impact on household consumption. Speci�cally, per capita consump- 1  The exchange rate is USD 1 = PHP 40.6 on tion increased by about 12 percent as a result 01/26/2013. of the project, which is consistent with �ndings 2  A barangay is the lowest administrative unit in the from the evaluation of the Kecamatan Develop- Philippines; corresponding to a village. ment Program (KDP), a similar CDD project The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  5 in Indonesia. Those impacts are stronger for that bene�t the community and on the propor- households that were classi�ed as poor in 2003; tion of respondents who thought that most peo- which experienced a 19 percent increase in per ple in the village are willing to help if need be. capita consumption. The impacts on per capita No impacts were detected on other measures of consumption are associated with a 6 percentage- barangay governance and social capital, such as point decline in the probability that households group membership or trust levels. are classi�ed as poor. There is some evidence that individuals, especially women, are more Finally, �ndings from the evaluation suggest ar- likely to be employed as a result of the project eas for improvement. First, despite signif­icant which could explain how per capita consump- investment in water systems in treatment muni- tion increased. cipalities, no impacts on access to water systems were detected. Further qualitative �eld work The project also had a positive impact on ac- suggests that it is due to the fact that some of the cessibility. Speci�cally, a 9 percentage-point sub-projects were unable to reach all community increase in the proportion of households whose members and some barangays did not manage house is accessible year-long can be attributed to to adequately maintain the investment. This, in the project. This is associated with greater mo- turn, may be the result of project resources be- bility. Households in treatment areas were going ing allocated on a per barangay, rather than on to the municipal center more regularly as a result a per capita, basis, which in some cases led to of the project. However, no effects were detected limited per capita allocations. The evaluation on other measures of access to basic services, in- also showed that the key impact on increased cluding access to improved water sources, sani- consumption levels is stronger on poorer house- tation and use of health facilities. In addition, the holds, suggesting that it might make sense to program led to a decline in school enrollment. vary municipal allocation by poverty levels (adjustments incorporated into the new national Results from the qualitative evaluation indicate CDD program). that the project led to changes in how village assemblies (a feature of the lowest level of lo- Second, there are challenges in sustaining em- cal government) are perceived. Prior to project powerment and barangay-level governance implementation, they were, at best, considered impacts, and in affecting improvements in mu- avenues for reporting, while now they tend to be nicipal-level governance. This could require seen as mechanisms for participation, transpar- greater LGU involvement and better integra- ency and accountability. This change seems to tion of project processes with the local planning be partly driven by a new breed of village lead- cycle, along the lines currently being followed ers. Indeed, especially in villages that received by the Makamasang Tugong initiative. Fur­ ther, financing for a sub-project, some of the village �ndings from the qualitative study suggest that volunteers have been empowered. This new pool the project was relatively successful at empow- of leaders can effectively engage elected village ering project volunteers but that the broader citi- offcials. They are considered to be more service- zenry was not as positively affected. oriented and committed than previous village leaders and, in some cases, they have been elect- Third, while a large proportion of barangays in ed to village offce. Ensuring the sustainability of targeted municipalities receive at least one sub- those impacts once project implementation has project during the 3 cycles, some do not. Project ended appears more challenging, however. volunteers who engaged in the relatively time- consuming KALAHI-CIDSS processes and did The quantitative evaluation was able to detect not manage to get a project for their barangay, positive impacts on the proportion of house- might be reluctant to engage in similar processes holds willing to contribute money to projects in the future. 6  Philippines Table 1. KALAHI-CIDSS Impact Evaluation Report Card Key Indicators Sign Size* Comments/Explanation Household Welfare Per capita consumption (log)— Positive Medium overall Per capita consumption (log)— Positive Medium poor households Per capita consumption (log)— None non-poor households Poverty levels are lower as Poverty levels Negative Small a result of the project Non-food share to total Positive Small consumption Labor force participation Positive Small Stronger for women Access to Services Stronger in prioritized Year-long road access Positive Medium barangays Visits to health stations None Issues with subproject Access to water None maintenance Low level of investments School enrollment Negative Small in sample barangays Social Capital and Local Governance Contribution to community projects Positive Medium Others are willing to help Positive Small * Size refers to the difference in the changes between baseline and endline in the treatment and control groups, taking into account the baseline value of the relevant indicator. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  7 8  Philippines 1. KALAHI-CIDSS T his report reviews available evidence local institutions mandated by the 1991 Local on the KALAHI-CIDSS program with Government Code. Speci�cally, the project had the aim of identifying both its strengths the objectives of strengthening local communi- and weaknesses. It will serve as an input into ties’ participation in barangay governance, and the planned revisions to project operating pro- developing their capacity to design, implement cedures and for the on-going scaling up of the and manage development activities that reduce program. The report starts by indicating the pro- poverty (World Bank 2002). grams main achievements in terms of outputs. It then briefly presents the systems put in place to The government of the Philippines commit- measure project impacts and to learn from the ted USD 82 milllion to the project, which was various studies that were implemented during complemented by a USD 100 million loan from the course of the program. The report presents the World Bank. Given the project emphasis on the main results from the �nal quantitative and alleviating rural poverty, it targeted the poor- qualitative impact evaluations with a special fo- est 25 percent of municipalities in each of the cus on the project impacts on poverty, access to poorest 42 provinces.3 At �rst, the project was basic services, local governance and social capi- implemented in 184 municipalities and then tal. The last section of this report indicates areas expanded to an additional 16 municipalities in for potential improvement. 2010. The project is currently being expanded through a USD 120 million grant from the Mil- At the turn of the new millennium, poverty in the lennium Challenge Corporation and a USD 59 Philippines, on the increase due to the aftermath million loan from the World Bank. Selection of of the 1997 Asian Crisis, was mostly a rural municipalities for project expansion took place phenomenon. In 2000, about 44 percent of the in the first half of 2011. rural population was poor and about three-fourth of the poor lived in rural areas (World Bank 2002). The passage of the 1991 Local Government As of December 2010, the project had sup- Code (RA 7160) provided opportunities for local ported 5,645 subprojects, worth PHP 5.7 billion poverty reduction efforts but implementation (about USD 140 million) and benefiting about fell short of original expectations. While 1.26 million households. The five most common signi�cant responsibilities were devolved to subproject types were roads, water systems, Local Government Units (LGUs), transfers were school buildings, health stations and pre/post not deemed sufficient to pay for these services. agricultural production facilities. The distribu- Further, poor rural communities often lacked tion of subprojects financed under the program opportunities to effectively engage in local is shown in Table 2. development processes. 3  Concerns about the capacity of regional DSWD of- The KALAHI-CIDSS program sought to respond fices to cover a large number of municipalities prevented to some of these short-comings. Set up in 2002, the program from targeting the poorest municipalities regardless of their province of origin. In addition, a deci- the program aimed at alleviating rural poverty sion was made not to implement the project in ARMM. A by providing resources to poor rural municipali- similar project, the ARMM Social Fund, was implemented ties for public goods investment and reviving instead. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  9 Table 2. Distribution of subproject types (December, 2010) % of % of HH % Subprojects Beneficiaries Total Cost Basic social services (e.g., health, edu- 50.1 49.1 44.5 cation, water) Basic access infrastructure (e.g., roads, 27.5 26.1 36.5 bridges) Community production, economic support, and common service facili- 11.4 12.7 8.9 ties Environmental protection and conser- 10.2 11.7 9.6 vation Other 0.8 0.5 0.5 Source: KALAHI-CIDSS National Project Management Office. Data on bene�ciaries are taken from sub- project proposals and correspond to the number of households in each barangay that are expected to bene�t directly from the sub-project. KALAHI-CIDSS applied a detailed participa- of criteria they themselves developed. tory process to the identification, prioritization, implementation and evaluation of community- 4. Subproject Implementation, Monitoring, level subproject investments. The process fol- and Evaluation (M&E), and Operations and lows what is known as the community empow- Maintenance stage for approved subproject erment activity cycle (CEAC), which consists of proposals. five main stages:4 5. Transition stage to enter into the sec- 1. Social Preparation Stage—during which ond implementation of the CEAC communities participate in a series of activi- after subprojects are completed. ties to identify and prioritize their problems and needs. The program has a number of noteworthy de- sign features that are consistent with Commu- 2. Subproject Identification Stage—during nity Driven Development programs worldwide. which community members are technically First, once a barangay has been prioritized for trained to design and package subproject subproject investment, a community bank ac- proposals that hope to address their needs. count is opened and funds from the project flow directly from the Philippine Government’s 3. Subproject Preparation, Selection, and Ap- implementing agency (the Department of Social proval stage—during which community Welfare and Development; DSWD) accounts representatives through the Municipal Inter- into the community account. Second, commu- Barangay Forum select which proposals will nity volunteers are fully responsible for pro- be funded by KALAHI-CIDSS using a set curement of subproject inputs and reporting to community at large and municipal authorities on the usage of funds. Third, municipal mayors 4  http://kalahi.dswd.gov.ph/index.php?option=com content&view=article&id=3&Itemid=3 visited on role in approving subprojects is limited by their 12/16/2010. non-voting status in the Municipal Inter-Baran- 10  Philippines gay Forum. Fourth, communities are required cial preparation and project identi�cation and to provide local counterpart contributions either design training). in cash or in-kind that are pooled from various sources (province, municipality, barangay and Third, common to all CDD operations, the KA- community), and which develops community LAHI-CIDSS �nances a number of different capacity for resource leveraging/mobilization. subprojects, which are likely to affect different dimensions of household welfare. Indeed, one In reviewing the impact of the KALAHI-CIDSS would not expect similar impacts for a farm- in participating municipalities, it is important to to-market road and for a school building. As a consider a few key aspects of the program. First, result, project impacts are diluted over a broad participating municipalities receive an annual range of outcome indicators and one should ex- grant equivalent to PHP 300,000 (about USD pect relatively smaller impacts on a number of 7,400) for each barangay; the total municipal indicators. Due to sample size restrictions, no grant is then allocated competitively between attempts were made to assess impacts by types barangays in the municipality. This corresponds of subprojects. to about 19 percent of Internal Revenue Allot- ment (IRA), i.e. regular �scal transfers from the Fourth, to better understand the impacts of KA- central government, in KALAHI-CIDSS munic- LAHI-CIDSS, ideally these results should be ipalities and to an average annual per capita al- compared to those of similar efforts to support location of approximately PHP 300 (about USD basic community infrastructure and services in 7.40). Given the small size of the per capita allo- the Philippines. Unfortunately, a limited number cation, expectations of the likely poverty reduc- of such programs in the Philippines have been tion impact of the program should be similarly subjected to this kind of robust analysis. As a modest. result, it is difficult to judge whether the KALA- HI-CIDSS is a cost-effective way of achieving Second, given the competitive nature of the the observed impacts. However, the large-scale prioritization process to allocate funding to vil- impact evaluation of the Pantawid Pamilyang lages within municipalities, one is unable to Pilipino Program (4Ps), also implemented by know ex-ante which villages will receive a sub- the DSWD, will generate useful comparative project and which villages will not. As a result, information. among the treatment municipalities surveyed, the sample covers both villages that were pri- oritized and villages that did not receive any subproject �nancing (but which did receive so- The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  11 2. Background on the Evaluation Strategy5 2.1 Design5 analysis to select two pairs of comparison and treatment municipalities in each of four provinc- As part of the project’s overall M&E efforts, a es. The pairs with the best match were selected. rigorous impact evaluation was designed in 2003 Chase and Holmemo (2006) report results indi- to evaluate project impacts on poverty reduction, cating that, unsurprisingly, given the strict pov- social capital, empowerment, and governance erty targeting procedures used by the project, and, to examine processes by which poverty has control municipalities are slightly richer than the been reduced and communities empowered. The treatment municipalities but appear similar along evaluation followed best practices in that it col- other dimensions. We provide further baseline lected quantitative and qualitative data before, descriptive statistics on the main household-lev- during and after project implementation in a el outcomes of interest and test for differences sample of KALAHI-CIDSS municipalities that between the treatment and control at baseline. received support (treatment municipalities) and Results, available in Tables A-1-A-3, are simi- of comparable municipalities that did not receive lar to the ones obtained by Chase and Holmemo support (control municipalities). Data were col- (2006). As discussed below, the analysis will be lected on a broad range of indicators: service carried out using either municipal or household delivery (access to health, education), poverty �xed-effects which will be picking up any pre- (employment, per capita consumption, self-rated existing differences between the treatment and poverty) and empowerment/governance (group control municipalities. More importantly, we membership, participation in barangay assem- also provide evidence that the two sets of mu- blies, collective action). The quantitative sample nicipalities were on similar paths before project includes 2,400 households in 135 barangays in implementation (Section 2.2). 16 municipalities in 4 provinces. The qualitative assessment, using focus group discussions, key The evaluation was designed to capture medi- informant interviews and direct observations, um-term impacts. Therefore, while baseline data took place in a subset of 20 barangays in 4 mu- collection took place in 2003, endline data were nicipalities in 2 provinces. not collected until early 2010. More than a year went by between the end of project activities in The control group was selected through cluster the sample municipalities and endline data col- analysis and, as discussed in more detail below, lection. As such, the design was able to pick up provides a credible estimate of what would have lasting impacts that materialize more slowly.7 happened in the treatment municipalities in the absence of the project.6 The team used cluster Chase and Holmemo (2006). 7  A large number of evaluations are designed to cap- ture impacts within a relatively short time-frame (e.g., one 5  This sections builds on Chase and Holmemo (2005) or two years). As King and Behrman (2009) and Woolcock and Labonne and Chase (2011). (2009) judiciously pointed out, this can lead to unreliable 6  Cluster analysis is a statistical method that allows re- results if either project impacts take time to materialize, searchers to pair together similar municipalities along with short-term evaluations underestimating project im- a set of chosen indicators. More details can be found in pacts, or if they fade away quickly, with short-term evalu- 12  Philippines 2.2 Testing the Parallel Trend similarly in the absence of the project and that the observed differences can be attributed to the Hypothesis project. Combined with previous tests compar- ing treatment and control municipalities at base- The key identifying assumption in the impact line, this suggests that there is a need to control evaluation is that, without the program, the two groups of municipalities would have evolved for either municipal or household fixed-effects. similarly. While it is impossible to test this hy- They will pick up any differences between treat- pothesis directly, it is possible to test if prior to ment and control municipalities at baseline. the project the two groups evolved similarly, the so-called parallel trend hypothesis (Bertrand, 2.3 The Regressions Duflo and Mullainathan 2004). Rejection of the parallel trend hypothesis would cast doubts on Taking advantage of the panel structure of the the validity of our estimation strategy. data, we estimate a series of regressions of the form: For this purpose, we use data from the Family Income and Expenditure Survey (FIES). The Yijt = αKCjt + βXijt + γt + uij + vijt (2) FIES is a large-scale nationally representative survey carried out every three years by the Na- Where Yijt is the outcome of interest for house- tional Statistics Office (NSO). We have access to the 2000 and 2003 data. Out of the 16 mu- hold i in municipality j at time t, KCjt is a dummy nicipalities included in the KALAHI-CIDSS equal to one if the KALAHI-CIDSS program impact evaluation sample, 13 were included in was implemented in municipality j at time t, uij the 2000 and in the 2003 FIES. This leaves a captures household fixed-effects and vijt is the repeated cross-section of households in 13 of idiosyncratic error term. For each outcome of our sample municipalities. We run the following interest, we start with a simple OLS regression placebo test: (Column 1 of each table), we then include mu- nicipal fixed effects (Column 2) and household Yijt = αTijt + βXijt + γt + uij + vijt (1) fixed effects (Column 3 of each table). We then add controls for overall time trends (Column 4 where, Yijt is the parameter of interest for house- of each table), basic household controls (Col- hold i in municipality j at time t, Tijt is a dummy umn 5 of each table) and regional time trends equal to one in 2003 for our sample treatment (Column 6 of each table).8 municipalities and zero otherwise, Yijt is a set of household characteristics. For each outcome of interest, we report results on various samples and with different estima- For each outcome indicator, we run four dif- tors. Specifically, we estimate equation (2) on ferent regressions (with and without municipal both the balanced sample (Panels A and B of dummies; with and without household controls). each table) and on the full sample (Panels C and Results are available in Table A-4 and A-5. We D of each table). In each case, we report results are unable to reject the parallel trend hypothesis. both with (Panels B and D) and without survey For none of the 24 regressions are the estimated weights (Panels A and C). Given that the pro- coefficients on the placebo treatment dummy statistically different from zero at the usual lev- 8  In the household-level regression, the set of house- els of confidence. This gives credence to the hold controls includes the number of female in the view that the two groups would have evolved household, the number of household members age 0-5, age 6-14, age 15-24, age 25-34, age 35-59 and age 60+. In the individuals-level regression, the set of controls in- ations overestimating project impacts. cludes a full set of age dummies and a gender dummy. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  13 gram was implemented at the municipal-level, that treatment municipalities in our sample were standard errors are clustered at that level. For similar to other KALAHI-CIDSS municipalities prior to project implementation.11 completeness, we also provide standard errors clustered at the village-level.9 Second, to reap the benefits from having a household panel dataset, efforts were devoted In each table, the preferred specification is the to keeping attrition to a minimum. Nonetheless, one with the most controls, clustering of stan- the sample size was reduced from 2,400 house- dard errors at the municipal-level, on the full holds during the baseline survey to a little less sample with survey weights. That is, the pre- than 1,900 households during the endline sur- ferred estimates of project impacts are the ones vey, mostly due to migration and deaths. Lev- presented in Column 6 of Panel D in each of the els of attrition are similar in the treatment and regression tables. control group (21 percent vs. 22 percent). In ad- dition, we test whether the determinants of at- 2.4 Implementation and trition are similar in the treatment and control groups. Specifically, for each household-level Challenges outcome of interest, we run a probit regression of a dummy indicating whether the household The impact evaluation, carried out in three drops out of the sample between 2003 and 2010 phases between 2003 and 201110, faced challen- on the interaction of the outcome of interest with ges in implementation. Implementation was the treatment dummy, its interaction with the not without challenges, however. First, due to control dummy, the treatment dummy and a full budgetary and logistical constraints, data were set of province dummies. The interaction terms only collected in 16 municipalities for the quan- are reported in Columns 1 and 2 of Tables A-6 titative surveys and in 4 municipalities for the -A-8. Results from a chi-square test of equality qualitative survey. As a result, one could ques- of the coefficients are available in Column 3. tion whether results from the evaluation are externally valid, that is, whether results from Results suggest that such attrition is unlikely to the evaluation would carry over in other project significantly bias the results as the determinants areas. While it is not possible to adequately of attrition do not appear to differ between the answer this question, available data indicate control and treatment groups. Third, one of the original control municipality 9  There is an additional challenge associated with the in the Province of Albay (Malinao) ended up be- limited number of municipalities. Indeed, with less than ing included in the PODER project, a KALAHI- 40 clusters, standard methods to account for clustering CIDSS-type program supported by the Spanish will provide downward biased standard errors, and as a result will tend to over-reject the null hypothesis of no aid agency. As a result, baseline data had to effect. While some bootstrap methods have been devel- be collected in a replacement control municipal- oped, an alternative is to use a t-distribution with G − c ity (Oas). degree of freedoms; with G= # of clusters(16) and c = # of variables that are fixed within clusters (1: the constant). The relevant critical values for a t-distribution with 14 de- In the impact evaluation sample, about two- grees of freedom are 1.75 (10 percent), 2.13 (5 percent) and 2.95 (1 percent). The main results, discussed below, are robust to using those critical values to determine sig- 11  For example, the small area estimates released by nificance. National Statistical Coordination Board indicate that in 10  The actual timing of data collection was as follows: 2000 poverty incidence was 64.8 percent in the 8 treat- Quantitative baseline in Sept/Oct 2003; Qualitative base- ment municipalities in the sample and 62.8 percent in line in April/June 2005; Quantitative midterm: Oct/Nov the other KALAHI-CIDSS municipalities, a difference that 2006; Qualitative and quantitative endlines: Feb/March is not different from zero at usual levels of statistical sig- 2010. nificance. 14  Philippines thirds of treatment barangays were prioritized three in every five of them expressed their sat- for subproject investment at least once. Put dif- isfaction in the project. Participation rates were ferently, about a third of the sample barangays around 65 percent in the preparatory and plan- in treatment municipalities did not receive a ning phases and 31 percent in the sub-project single subproject throughout the three subproj- implementation phase. Of particular interest, ect cycles thereby reducing likely measurable women are more likely to participate in proposal impacts in these areas. selection and preparation. Conversely, men are more likely to participate in subproject imple- mentation. This might reflect traditional gender The actual distribution of subprojects in the roles in those communities. Interestingly, wom- sample barangays is shown in Figure 1. The en volunteers belong to the same socio-econom- relative importance, and level of investment by ic status as most of the constituents but are more subproject type, should be borne in mind when available for and interested in barangay projects. interpreting results. Specifically, in our sample, This is consistent with the view that project pro- project impacts should only be expected on out- cesses are not dominated by local elites. comes that can be affected by subprojects that were chosen by the community. The impacts Local elected officials also view the project in a might differ in areas where communities select- positive light. About 75 percent of barangay and ed a different mix of subprojects. municipal officials indicated being satisfied with the project. When asked about the benefits of the Available data indicate that participation rates in KALAHI-CIDSS, the most common responses project activities are relatively high, suggesting are infrastructure improvement and better access that households in targeted municipalities see to services; community empowerment also fig- value in the KALAHI-CIDSS approach. About ures among the top responses. Not surprisingly, 80 percent of households in treated municipali- feedback from barangays not prioritized to re- ties indicated being aware of the project and ceive subproject financing was less positive. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  15 3. Results of KALAHI-CIDSS in terms of Welfare 3.1 Who did the project reach? village officials) did not appear to be an over- whelming force behind proposals put forward to This section of the report reviews the program the MIBF (subproject prioritizing committee), targeting procedures both at the provincial, as their preferences and those of community municipal and barangay level. A map of proj- members were equally represented in commu- ect areas for the period 2002-2009 is shown in nity proposals. Not surprisingly, however, indi- Annex 1. viduals who were already active in community affairs prior to the project are more likely to As previously noted, the project targeted the have their preferences represented in the submit- poorest 25 percent of municipalities in 42 of ted community proposal. Moreover, and consis- the poorest provinces identified through a cus- tent with the challenges of engaging marginal- tomized index developed in collaboration with ized groups, the survey found that women and Dr. Balisacan at the University of the Philip- individuals who had not attended school were pines, School of Economics. The rankings used less likely to have their preferences represented for targeting municipalities proved to be con- in the subproject proposal. However, this result sistent with official rankings released by the was obtained after only one subproject cycle and National Statistical Coordination Board (NSCB) DSWD revised its operating procedures shortly in 2005 (World Bank n.d.) afterwards to promote greater inclusiveness. There is no evidence available on the effects of While the program was successful in directing those revisions, however. resources to the poorest municipalities, the pos- sibility of elite capture -i.e., better-off and con- The impact evaluation also reveals that KALA- nected individuals dominating project processes HI-CIDSS was successful in targeting the poor- and receiving a disproportionate share of proj- est, best-organized villages. Surprisingly, how- ect benefits (Mansuri and Rao 2004) remained ever, more unequal villages were more likely a concern. To avoid this risk, KALAHI-CIDSS to have their proposals funded. This appears to was specifically designed to help ensure that be due to the fact that the barangay captain was poor households and communities within eli- more likely to take control of a disorganized gible municipalities could benefit from the proj- community preference, and to influence inter- ect. For example, the specific poverty concerns village competition at the MIBF. This is akin to of the program were emphasized in the social benevolent forms of elite capture as the com- preparation and subproject design phases of the munity, as well as the barangay captain, benefits project, and in the meetings in which communi- from receiving a subproject, which might not ties developed criteria to rank project proposals. otherwise happen. Facilitators were also instructed to encourage participation of marginalized households. 3.2 Key welfare impacts Available evidence indicates that KALAHI-CI- This section of the report reviews program im- DSS subprojects were not subject to elite cap- pacts on per capita consumption. Results are ture, at least in its most malign form (Labonne shown in Tables 3-13. As indicated above, the and Chase 2009). Barangay captains (elected preferred estimates of project impacts are the 16  Philippines ones presented in Column 6 of Panel D in each the Remedios Farmers Cooperative. Accord- of the regression tables. ing to community members, the project has cut corn and rice production costs by 30 percent by Per capita consumption increased by about 12 bringing the mill (and the grain to be milled) percent as a result of the project (Table 3). This closer to the people. Previously, there were few is strikingly similar to findings from an impact milling facilities on the western part of the Agu- evaluation of a similar CDD project in Indone- san river and transport costs to the producers sia. Voss (2008) found that the project led to an were therefore much higher. The community 11 percent increase in per capita consumption. also noted that because the mills’ services are Once we distinguish between households that better, other barangays - i.e., Bakingking, New were classified as poor in 2003 and those that Gingoog, Tagabase, and Hawilian - are now were not, an interesting pattern emerges.12There using it. is evidence that the project led to a 19 percent increase of per capita consumption for poor The project also led to a 5 percentage-point in- households (Table 4) but that it had no impact crease in the non-food share of consumption, on non-poor households (Table 5). This further which some researchers have argued is a better reduces concerns over elite capture of project measure of household welfare.13 (Table 7). benefits. Indeed, if project benefits had been On the other hand, the self-perception of pov- captured by local elites, one should expect to erty (i.e., the share of households rating them- observe larger impacts on non-poor households selves as poor) does not seem to be affected by than on poor households. the project (Table 10). A potential explanation The impacts on per capita consumption are as- for this finding is that the increases in per capita sociated with a 6 percentage-point decline in the consumption are not large enough for house- probability that the household is poor (Table 6). holds to switch from feeling poor to feeling Again, this is of a similar order of magnitude non-poor. Alternatively, self-reported poverty found on the KDP in Indonesia (Voss 2008). measures might not be very good measures of household welfare. Findings from the qualitative evaluation high- light how the program could have generated One possible source for the increase in per such impacts. In San Ramon, Libon (Albay), capita consumption is that individuals in treat- community members indicated that, among ment areas are more likely to be employed as a the development projects in their barangay, the result of the project. Indeed, there is some evi- KALAHI-CIDSS-funded road-improvement dence that a 4 percentage-point increase in the project created the most impact as more trans- likelihood of employment can be attributed to port and utility vehicles are now plying to and the project (Table 11). The effect mainly comes from the area. This increased traffic is creating from the female sample, who experience an 8 business opportunities in the community, and percentage-point increase in their likelihood of has also made transportation available at much employment (Table 13). It is important to note lower cost than before. that these improvements in employment more likely reflect greater economic activity gener- Similary, the barangay of Remedios, Esperanza ated by the project rather than direct, project (Agusan del Sur) built a rice and corn mill, with related employment opportunities as the survey was conducted at least one year after subproject related employment ended. 12  In the Philippines, households are classified as poor if their per capita income falls below a certain threshold. As a second-best strategy, since no data on income were collected in the survey, households were classified as 13  Measures of per capita consumption do not account poor in 2003 if their baseline per capita consumption was for (i) potential economies of scale within the household lower than their regional poverty line. and (ii) relative needs of children and adults. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  17 3.3 Access to basic services test that hypothesis directly due to small sample sizes. This section of the report reviews program im- pacts on access to basic services. Results are The quantitative evaluation did not identify im- shown in Tables 14-25. pacts on other measures of access to basic ser- vices, such as access to improved water sources (Tables 17-18) , sanitation (Table 19) and use of Consistent with the large number of roads that health services when sick (Tables 20-22). were financed by the project in sample areas, the project had a positive impact on accessibility. Surprisingly, individuals in project areas are Specifically, the project resulted in a 9 percent- less likely to be enrolled in school as a result age-point increase in the proportion of house- of the project, with the effect concentrated on holds whose house is accessible year-round the male sample (Tables 23-25). This could be (Table 14). This increase translates into greater due to improved employment opportunities in mobility, with households making more trips to KALAHI-CIDSS municipalities that increased the municipal centers and with higher expendi- the opportunity cost of going to school. How- tures on transportation (Tables 15-16). While ever, it is important to note that a small number roads financed under the project are most likely of prioritized barangays in our sample decided driving this impact, no attempts were made to to invest in school buildings (cf. Figure 1). 18  Philippines 4. The Results of KALAHI-CIDSS in terms of Governance and Social Capital T his section of the report reviews pro- to have been empowered (See box 1). This new gram impacts on village governance group of leaders can effectively engage elect- and social capital. Results are shown in ed barangay officials. They are considered to Tables 26-40. be more service-oriented and committed than previous barangay leaders and, in some cases, While the quantitative evaluation was only able they have been elected to barangay office. In- to detect marginally significant effects on house- terestingly, most of those volunteers are women. hold’s participation in barangay assemblies However, as discussed in more detail below, (Table 26), the qualitative evaluation detected these empowerment benefits have yet to reach changes in how assemblies are perceived. Prior the broader community outside of the project to project implementation, they were, at best, volunteers. considered avenues for reporting. Now they tend to be seen as mechanisms for participa- Interestingly, according to the qualitative evalu- tion, transparency and accountability, as reflect- ation, there is also a shift in how community ed in the following quotes from the qualitative members perceive their barangay captains. Tra- evaluation: ditionally, leaders are rated highly if they are available, understanding and able, within limits, “More often, barangay assemblies (…) are to bring resources to the community. Households reduced to occasions for reporting accom- in treatment barangays in Agusan del Sur now plishments and expenditures, and for presen- tation of plans, programs, and projects, that also care about whether leaders are consultative, are in most cases already approved by the transparent and able to plan for the future. barangay council.� (focus group discussion (FGD) participant in Balangibang, Polan- Apart from the effects on participation in ba- gui, a control municipality) rangay assemblies discussed above, the quanti- tative impact evaluation did not detect signifi- “Barangay Assemblies are good and effec- cant impacts on measures of local governance tive venues for the people to be heard.� (FGD and social capital. There are two exceptions, participant in Bacolod, Libon, a KALAHI- however. First, more households indicate being CIDSS treatment municipality) willing to contribute money to projects that will benefit the community as a result of the project This change seems to be partly driven by a new (Table 27). Second, the project led to an increase breed of barangay leaders. Indeed, especially in the proportion of households indicating that in barangays that received subproject financ- others community members are willing to help ing, some of the community volunteers appear if needed (Table 28). The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  19 Box 1. Political Engagement: Virgie Niebres, Barangay Rawis, Pio Duran Virgie Niebres is a 36-year-old resident of Rawis. She began studying nursing at Bicol University, but due to poverty was forced to stop schooling after only her first semester. Her husband is 37 and an elementary graduate. Together they have five children. Before the KALAHI-CIDSS project, their only source of income was from harvesting copra. The KALAHI-CIDSS road project in Rawis has created the opportunity for Virgie to improve and diversify her family’s livelihoods options. With a more efficient way to transport copra to market, they were also able to purchase a motorcycle operated by her husband for “habal-habal� (motorcycle rental). Vergie also has benefited directly by working closely with KALAHI-CIDSS as project prep- aration team chairman and as a bookkeeper. She was then elected as the chair of the Ba- rangay Subproject Management Committee (BSPMC). During project preparation, Virgie learned how to develop project proposals, and assisted with mapping impoverished re- gions. Her experience as a BSPMC chair also taught her various aspects of project imple- mentation. She was able to overcome her shyness and enhance her public relations skills because she had to convince people in the barangay to attend barangay assemblies. She also gained the confidence to talk in front of a large crowd. Because of KALAHI, she learned to participate in barangay affairs. Being a volunteer also allowed her to attend numerous training events and seminars. She has traveled not only within the municipality, but even in other provinces. These experiences resulted in a new career for her as center chief of Simbag sa Pag-asenso, a Catholic social action lending microfinance program. As center chief, Vergie handles 52 members from four barangays. She is also the Secre- tary of the Barangay Power Association, a local electrification association in charge of the maintenance of the barangay’s electrification. The association’s activities include the col- lection of payments from each household. She also became the manager of the distribu- tion of fertilizers and seedlings provided by the Department of Agriculture in the munici- pality. She also takes part in the decision making in the barangay. Source: World Bank (2011b). 20  Philippines 5. Areas for Improvement T his section of the report seeks both to While the competitive allocation of resources highlight implications from the evalua- through the MIBF is a key feature of the proj- tion for project expansion and to identify ect, some of its downsides need to be acknowl- areas where additional analytical work could edged. In practice a large proportion of baran- prove fruitful. gays in targeted municipalities receive at least one sub-project during the 3 funding cycles, but 5.1 Implications for project some do not. Project volunteers who engaged in the relatively time-consuming KALAHI-CIDSS expansion processes but did not manage to get a project for their barangays might be reluctant to engage in Results from the qualitative evaluation indicate similar processes in the future. There is a need to that the project did not have any measurable im- better manage expectations. Further, the project pact on governance at the municipal-level. There could systematically consider offering support are two possible explanations for this. First, very to non-prioritized communities in seeking fund- little direct capacity building was initially tar- ing for their KALAHI-CIDSS proposals through geted at municipalities, therefore great changes in behavior should not have been expected. Sec- other sources. ond, the relatively small and short-term nature of the project funding (as compared to other Findings from two recent studies suggest that the available forms of support) may be insufficient program might have led to a temporary increase to influence great change in the dynamic be- in conflict levels, especially in areas where the tween municipalities and barangays. In relation, New Peoples Army (NPA) is present (Arcand, KALAHI-CIDSS has been experimenting with Bah and Labonne 2010; Crost and Johnston the so-called Makamasang Tugong initiative 2010). The first study uses newspaper reports that shifts responsibility for management of the of conflict incidence between the Armed Forces program to the municipal LGUs. While it is too of the Philippines (AFP) and either the NPA or early to know whether this has made a differ- the Moro Islamic Liberation Front (MILF) and ence, the project team might want to review the finds that the project led to a decline in MILF- conditions of this initiative to ensure that they related events, but to an increase in NPA-related promote the needed transparency and participa- events. The second study uses AFP data and tion in LGU activities. finds that the program led to an increase in both MILF and NPA-related events. Both studies rely Findings from the qualitative study suggest that on nationwide conflict data and estimate proj- the project was relatively successful at empow- ect impacts using difference-in-differences and ering project volunteers but that the broader ba- regression discontinuity techniques. The differ- rangay citizenry was not as positively affected. ences between the two sets of results could come While this might reflect lack of interest by some from the variation in data sources but also from of the community members, and the unwilling- the different definitions used to classify conflict ness to challenge local leaders that are seen as events. Indeed, the first study looks at events bridges to resources, this could also indicate that with a 50km radius of eligible municipalities further efforts from the facilitators are necessary while the second study is only concerned with throughout social preparation. events in KALAHI-CIDSS municipalities. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  21 While more research is necessary to understand Second, results from the evaluation also suggest which project component is driving this shift in that maintenance arrangements for water proj- conflict occurrence, available results call for a ects might not be adequate. The project team more cautious approach in conflict-affected ar- should carry out a thorough maintenance review eas. Nonetheless, it is important to keep in mind of various project types. The study should also that there are no similar analyses available for propose ways to improve maintenance arrange- other government programs in the Philippines ments in the future. and, as such, it is impossible to compare the KALAHI-CIDDS approach with other develop- Third the project team should compile and fa- ment interventions in the Philippines along those cilitate access to data on the efficiency and ef- dimensions. fectiveness of different types of sectoral invest- ments. Further, while there is evidence that KALAHI-CIDSS sub-projects are less expen- 5.2 Suggestions for additional ana- sive than comparable sectoral investment (Arar- lytical work al and Holmemo, 2007), computations should be updated and compiled in a user-friendly format. Findings from the evaluation are consistent with This could serve as the basis for a long-term en- the view that subprojects are what drive the proj- gagement with sectoral colleagues. ect impacts on poverty reduction. Long-term impacts will require sustained efforts and both Fourth, the project team should carefully review social and physical investments. Moreover, the the questionnaires used in the evaluation to bet- greatest impacts are found where poverty among ter capture most relevant data on outcomes and households and communities is the highest. As a impacts. result, to increase the poverty reduction impacts of the project, the project team should explore A final note of caution is also in order. While ways to (i) support local communities access to a number of studies were carried out through- alternative additional sources of funding and, out the project cycle, some of them were not (ii) differentiate barangay grants by poverty lev- adequately linked to operations and, as a result, els. An option would be to vary municipal grants their recommendations were not acted upon. If with municipal poverty levels. Alternatively, in one wants to build an empirical basis to inform richer municipalities, communities should be policy and operational decisions, systems need required to provide larger Local Counterpart to be put in place to (i) allow DSWD to identify Contribution (LCC). This is especially impor- areas where more research is needed, (ii) carry tant as the project expands in relatively richer out the studies in close collaboration between areas. Such options could be carefully piloted the project team and researchers and, most im- and evaluated in a subset of municipalities. portantly, (iii) to act upon the findings. 22  Philippines References Arcand, Jean-Louis, Adama Bah, and Julien nal of Development Economics, 2011, Labonne, “Conflict, Ideology and For- 96 (2), 348–358. eign Aid,� CERDI Working Paper 2010- 21, 2010. Mansuri, Ghazala and Vijayendra Rao, “Community-Based and -Driven De- Asia Pacific Policy Center, Final Survey for the velopment,� World Bank Research Ob- KALAHI-CIDSS Impact Evaluation Re- server, 2004, 19 (1), 1–40. vised Final Report 2011. Voss, John, “Impact Evaluation of the Second Bertrand, Marianne, Esther Duflo, and Send- Phase of the Kecamatan Development hil Mullainathan, “How Much Should Program in Indonesia,� The World Bank We Trust Differences-in-Differences Indonesia, 2008. Estimates?,� Quarterly Journal of Eco- nomics, 2004, 119 (1), 249–275. Woolcock, Michael, “Toward a plurality of methods in project evaluation: a con- Chase, Robert and Camilla Holmemo, “Com- textualised approach to understanding munity Driven Development and Social impact trajectories and efficacy,� The Capital: Designing a Baseline Survey Journal of Development Effectiveness, in the Philippines,� World Bank -Social 2009, 1 (1), 1–14. Development Department, 2005. World Bank, Project Appraisal Document for Crost, Benjamin and Patrick Johnston, “Aid the KALAHI-CIDSS, Report No: 24642- Under Fire: Development Projects and PH, 2002. Civil Conflict,� University of California – Berkeley, mimeo, 2010. ___, The KALAHI-CIDSS Impact Evaluation: A Synthesis Report, Sustainable Devel- King, Elizabeth and Jere Behrman, “Timing opment -Asia and the Pacific Region, and Duration of Exposure in Evaluation 2011. of Social Programs,� World Bank Re- search Observer, 2009, 24 (1), 55–82. ___, Making everyone Count: Gender-sensitive Monitoring and Evaluation in a Com- Labonne, Julien and Robert Chase, “Who’s munity-Driven Development Program: at the Wheel when Communities Drive The Case of the Philippines’ KALAHI- Development? The case of the KALA- CIDSS., East Asia and Pacific Social, HI-CIDSS in the Philippines,� World Rural and Environmentally Sustainable Development, 2009, 37 (1). Development Unit, mimeo, 2011. ___ and ___, “Do Community-Driven Develop- ___, Comparison of Rankings of Municipalities ment Projects Enhance Social Capital? in the Philippines from two poverty- Evidence from the Philippines,� Jour- mapping methodologies n.d. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  23 Figure 1. Distribution of subproject types in sample municipalities School Buildings 4% Other 10% Day Care 7% Roads 33% Health Stations 9% Pre- and Post- productions facilities Water Systems 13% 24% Source: KALAHI-CIDSS National Project Management Office. 24  Philippines Table 3: Impacts on log per capita expenditures (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.066 0.289 0.289 0.119 0.133 0.133 (0.076) (0.028)*** (0.028)*** (0.031)*** (0.029)*** (0.022)*** [0.038]* [0.022]*** [0.022]*** [0.026]*** [0.024]*** [0.022]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.124 0.072 0.121 0.279 0.287 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.028 0.279 0.279 0.122 0.135 0.129 (0.075) (0.025)*** (0.025)*** (0.028)*** (0.026)*** (0.019)*** [0.040] [0.023]*** [0.023]*** [0.028]*** [0.025]*** [0.024]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.106 0.061 0.105 0.271 0.279 Panel C: Full sample (no weights) KALAHI-CIDSS 0.077 0.283 0.279 0.110 0.126 0.125 (0.075) (0.027)*** (0.025)*** (0.028)*** (0.028)*** (0.022)*** [0.037]** [0.020]*** [0.021]*** [0.024]*** [0.023]*** [0.022]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.003 0.133 0.069 0.119 0.274 0.282 Panel D: Full sample (with weights) KALAHI-CIDSS 0.034 0.272 0.272 0.117 0.130 0.124 (0.074) (0.025)*** (0.023)*** (0.026)*** (0.025)*** (0.019)*** [0.040] [0.021]*** [0.021]*** [0.026]*** [0.024]*** [0.023]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.111 0.060 0.103 0.268 0.275 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is the log per capita expenditures. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  25 Table 4: Impacts on log per capita expenditures (poor households) (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.185 0.462 0.462 0.204 0.203 0.203 (0.059)*** (0.036)*** (0.036)*** (0.037)*** (0.037)*** (0.031)*** [0.030]*** [0.025]*** [0.025]*** [0.030]*** [0.027]*** [0.026]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 3,126 3,126 3,126 3,126 3,126 3,126 R-squared 0.031 0.159 0.188 0.293 0.418 0.430 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.160 0.446 0.446 0.200 0.199 0.196 (0.065)** (0.033)*** (0.033)*** (0.037)*** (0.035)*** (0.028)*** [0.031]*** [0.026]*** [0.026]*** [0.032]*** [0.028]*** [0.026]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 3,126 3,126 3,126 3,126 3,126 3,126 R-squared 0.023 0.151 0.171 0.276 0.406 0.417 Panel C: Full sample (no weights) KALAHI-CIDSS 0.209 0.463 0.458 0.202 0.202 0.203 (0.058)*** (0.035)*** (0.032)*** (0.034)*** (0.035)*** (0.030)*** [0.028]*** [0.024]*** [0.024]*** [0.028]*** [0.027]*** [0.025]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 3,536 3,536 3,536 3,536 3,536 3,536 R-squared 0.038 0.180 0.188 0.294 0.416 0.428 Panel D: Full sample (with weights) KALAHI-CIDSS 0.180 0.447 0.445 0.202 0.201 0.199 (0.065)** (0.034)*** (0.030)*** (0.034)*** (0.033)*** (0.027)*** [0.030]*** [0.024]*** [0.024]*** [0.030]*** [0.027]*** [0.025]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 3,536 3,536 3,536 3,536 3,536 3,536 R-squared 0.028 0.167 0.173 0.278 0.406 0.418 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is the log per capita expenditures (only for households that were classified as poor in 2003). In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 26  Philippines Table 5: Impacts on log per capita expenditures (non-poor households) (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.028 0.048 0.048 -0.019 0.020 0.009 (0.069) (0.052) (0.052) (0.050) (0.047) (0.024) [0.046] [0.035] [0.035] [0.037] [0.035] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 2,526 2,526 2,526 2,526 2,526 2,526 R-squared 0.001 0.065 0.002 0.010 0.164 0.184 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.064 0.032 0.032 -0.021 0.022 0.009 (0.068) (0.052) (0.052) (0.050) (0.049) (0.025) [0.053] [0.036] [0.035] [0.041] [0.038] [0.034] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 2,526 2,526 2,526 2,526 2,526 2,526 R-squared 0.002 0.060 0.001 0.006 0.179 0.198 Panel C: Full sample (no weights) KALAHI-CIDSS -0.039 0.031 0.036 -0.033 0.007 -0.004 (0.067) (0.044) (0.050) (0.048) (0.046) (0.023) [0.044] [0.032] [0.032] [0.034] [0.033] [0.029] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 2,866 2,866 2,866 2,866 2,866 2,866 R-squared 0.001 0.069 0.001 0.010 0.159 0.180 Panel D: Full sample (with weights) KALAHI-CIDSS -0.073 0.023 0.022 -0.031 0.012 0.000 (0.066) (0.045) (0.050) (0.048) (0.048) (0.024) [0.052] [0.032] [0.033] [0.038] [0.036] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 2,866 2,866 2,866 2,866 2,866 2,866 R-squared 0.003 0.062 0.000 0.006 0.175 0.194 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is the log per capita expenditures (only for households that were classified as non-poor in 2003). In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  27 Table 6: Impacts on poverty levels (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.043 -0.190 -0.190 -0.069 -0.079 -0.079 (0.057) (0.023)*** (0.023)*** (0.031)** (0.029)** (0.023)*** [0.026] [0.018]*** [0.018]*** [0.023]*** [0.023]*** [0.022]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.002 0.106 0.036 0.065 0.151 0.155 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.025 -0.184 -0.184 -0.075 -0.085 -0.079 (0.058) (0.020)*** (0.019)*** (0.026)** (0.024)*** (0.020)*** [0.028] [0.017]*** [0.017]*** [0.024]*** [0.023]*** [0.022]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.097 0.031 0.056 0.157 0.159 Panel C: Full sample (no weights) KALAHI-CIDSS -0.049 -0.183 -0.177 -0.056 -0.067 -0.067 (0.055) (0.020)*** (0.018)*** (0.026)** (0.025)** (0.020)*** [0.026]* [0.016]*** [0.016]*** [0.021]*** [0.021]*** [0.020]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.002 0.104 0.033 0.063 0.150 0.154 Panel D: Full sample (with weights) KALAHI-CIDSS -0.027 -0.175 -0.173 -0.064 -0.074 -0.069 (0.057) (0.017)*** (0.016)*** (0.022)** (0.021)*** (0.017)*** [0.028] [0.017]*** [0.017]*** [0.023]*** [0.022]*** [0.021]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.095 0.029 0.054 0.155 0.157 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the household is classified as poor. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 28  Philippines Table 7: Impacts on non-food share of total consumption (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 3.089 9.187 9.187 5.027 5.078 5.030 (2.076) (1.239)*** (1.238)*** (1.704)*** (1.717)*** (1.391)*** [0.967]*** [0.715]*** [0.714]*** [0.889]*** [0.907]*** [0.799]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.008 0.100 0.065 0.091 0.098 0.109 Panel B: Balanced panel (with weights) KALAHI-CIDSS 2.091 9.305 9.305 5.845 5.915 5.752 (2.079) (1.230)*** (1.228)*** (1.613)*** (1.636)*** (1.481)*** [1.021]** [0.742]*** [0.741]*** [0.918]*** [0.943]*** [0.880]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.094 0.063 0.082 0.088 0.097 Panel C: Full sample (no weights) KALAHI-CIDSS 3.525 9.051 9.025 4.861 4.929 4.901 (2.090) (1.134)*** (1.237)*** (1.721)** (1.727)** (1.374)*** [0.947]*** [0.641]*** [0.686]*** [0.867]*** [0.881]*** [0.763]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.009 0.108 0.065 0.091 0.099 0.110 Panel D: Full sample (with weights) KALAHI-CIDSS 2.340 9.192 9.163 5.777 5.847 5.680 (2.117) (1.103)*** (1.178)*** (1.574)*** (1.602)*** (1.413)*** [1.004]** [0.643]*** [0.688]*** [0.865]*** [0.891]*** [0.816]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.100 0.062 0.081 0.089 0.097 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is the share of non-food to total expenditures. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  29 Table 8: Impacts on log per capita food expenditures (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.020 0.127 0.127 0.030 0.043 0.043 (0.040) (0.030)*** (0.030)*** (0.037) (0.034) (0.026) [0.024] [0.021]*** [0.021]*** [0.026] [0.023]* [0.021]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.075 0.020 0.043 0.221 0.227 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.003 0.117 0.117 0.021 0.033 0.030 (0.037) (0.030)*** (0.030)*** (0.036) (0.033) (0.025) [0.025] [0.022]*** [0.022]*** [0.027] [0.023] [0.022] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.062 0.016 0.039 0.222 0.225 Panel C: Full sample (no weights) KALAHI-CIDSS 0.024 0.126 0.122 0.025 0.040 0.040 (0.039) (0.031)*** (0.029)*** (0.036) (0.034) (0.026) [0.023] [0.020]*** [0.020]*** [0.025] [0.022]* [0.020]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.082 0.019 0.042 0.212 0.217 Panel D: Full sample (with weights) KALAHI-CIDSS 0.006 0.114 0.113 0.018 0.030 0.027 (0.036) (0.031)*** (0.028)*** (0.035) (0.032) (0.025) [0.024] [0.021]*** [0.021]*** [0.026] [0.022] [0.021] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.067 0.015 0.038 0.216 0.219 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is the log per capita food expenditures. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 30  Philippines Table 9: Impacts on log per capita non-food expenditures (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.192 0.588 0.588 0.276 0.292 0.290 (0.134) (0.062)*** (0.062)*** (0.079)*** (0.081)*** (0.070)*** [0.062]*** [0.038]*** [0.038]*** [0.045]*** [0.046]*** [0.042]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.008 0.138 0.093 0.145 0.213 0.224 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.127 0.585 0.585 0.312 0.326 0.317 (0.133) (0.056)*** (0.056)*** (0.071)*** (0.073)*** (0.068)*** [0.065]* [0.039]*** [0.038]*** [0.048]*** [0.049]*** [0.047]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.123 0.085 0.127 0.199 0.208 Panel C: Full sample (no weights) KALAHI-CIDSS 0.219 0.583 0.574 0.261 0.279 0.278 (0.133) (0.057)*** (0.059)*** (0.078)*** (0.081)*** (0.070)*** [0.061]*** [0.035]*** [0.037]*** [0.045]*** [0.045]*** [0.042]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.010 0.148 0.090 0.142 0.210 0.221 Panel D: Full sample (with weights) KALAHI-CIDSS 0.142 0.579 0.575 0.305 0.320 0.311 (0.133) (0.050)*** (0.052)*** (0.068)*** (0.071)*** (0.065)*** [0.065]** [0.034]*** [0.036]*** [0.046]*** [0.047]*** [0.045]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.131 0.084 0.126 0.197 0.207 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is the log per capita non-food expenditures. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within mu- nicipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  31 Table 10: Impacts on self-rated poverty levels (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.012 -0.037 -0.037 0.000 -0.002 -0.002 (0.048) (0.036) (0.036) (0.041) (0.042) (0.032) [0.024] [0.021]* [0.021]* [0.026] [0.026] [0.024] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.047 0.001 0.004 0.006 0.013 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.029 -0.023 -0.023 0.001 -0.001 0.004 (0.050) (0.038) (0.038) (0.041) (0.041) (0.031) [0.024] [0.021] [0.021] [0.025] [0.025] [0.023] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.045 0.000 0.002 0.004 0.010 Panel C: Full sample (no weights) KALAHI-CIDSS 0.003 -0.039 -0.040 -0.004 -0.006 -0.006 (0.047) (0.037) (0.035) (0.039) (0.040) (0.031) [0.023] [0.021]* [0.020]* [0.025] [0.025] [0.023] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.046 0.002 0.004 0.006 0.013 Panel D: Full sample (with weights) KALAHI-CIDSS 0.023 -0.024 -0.027 -0.004 -0.007 -0.001 (0.049) (0.039) (0.038) (0.040) (0.041) (0.030) [0.023] [0.021] [0.021] [0.024] [0.024] [0.022] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.043 0.001 0.002 0.004 0.010 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the household classified itself as poor. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 32  Philippines Table 11: Impacts on employment (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.006 -0.026 -0.021 0.035 0.040 0.042 (0.021) (0.031) (0.030) (0.032) (0.031) (0.016)** [0.012] [0.016] [0.016] [0.018]* [0.017]** [0.013]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 17,035 17,035 17,035 17,035 17,035 17,035 R-squared 0.000 0.017 0.000 0.003 0.314 0.319 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.013 -0.022 -0.016 0.046 0.050 0.050 (0.029) (0.027) (0.026) (0.029) (0.028)* (0.015)*** [0.013] [0.015] [0.016] [0.017]*** [0.017]*** [0.014]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 17,035 17,035 17,035 17,035 17,035 17,035 R-squared 0.000 0.019 0.000 0.004 0.314 0.318 Panel C: Full sample (no weights) KALAHI-CIDSS -0.007 -0.025 -0.022 0.031 0.038 0.039 (0.021) (0.029) (0.029) (0.031) (0.030) (0.016)** [0.011] [0.016] [0.016] [0.017]* [0.017]** [0.013]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 18,928 18,928 18,928 18,928 18,928 18,928 R-squared 0.000 0.016 0.000 0.003 0.317 0.323 Panel D: Full sample (with weights) KALAHI-CIDSS -0.014 -0.020 -0.014 0.046 0.051 0.050 (0.029) (0.025) (0.026) (0.028) (0.027)* (0.015)*** [0.013] [0.015] [0.015] [0.017]*** [0.016]*** [0.013]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 18,928 18,928 18,928 18,928 18,928 18,928 R-squared 0.000 0.019 0.000 0.004 0.316 0.320 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual is employed. In Panels A and B, the sample is restricted to house- holds that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  33 Table 12: Impacts on male employment (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.001 -0.046 -0.044 0.003 0.004 0.004 (0.022) (0.018)** (0.016)** (0.016) (0.015) (0.014) [0.013] [0.012]*** [0.012]*** [0.013] [0.013] [0.012] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,970 8,970 8,970 8,970 8,970 8,970 R-squared 0.000 0.017 0.002 0.005 0.304 0.305 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.005 -0.044 -0.042 0.010 0.010 0.012 (0.022) (0.016)** (0.014)*** (0.018) (0.017) (0.015) [0.014] [0.012]*** [0.012]*** [0.015] [0.014] [0.014] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,970 8,970 8,970 8,970 8,970 8,970 R-squared 0.000 0.016 0.001 0.006 0.311 0.312 Panel C: Full sample (no weights) KALAHI-CIDSS -0.003 -0.046 -0.042 0.005 0.004 0.004 (0.022) (0.016)** (0.016)** (0.015) (0.014) (0.013) [0.013] [0.011]*** [0.011]*** [0.013] [0.012] [0.012] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 9,910 9,910 9,910 9,910 9,910 9,910 R-squared 0.000 0.018 0.002 0.005 0.300 0.301 Panel D: Full sample (with weights) KALAHI-CIDSS -0.007 -0.043 -0.040 0.013 0.012 0.013 (0.022) (0.015)*** (0.014)** (0.017) (0.016) (0.014) [0.014] [0.012]*** [0.012]*** [0.015] [0.014] [0.014] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 9,910 9,910 9,910 9,910 9,910 9,910 R-squared 0.000 0.017 0.001 0.006 0.307 0.307 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual is employed. In Panels A and B, the sample is restricted to house- holds that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 34  Philippines Table 13: Impacts on female employment (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.013 0.007 0.016 0.075 0.080 0.080 (0.034) (0.048) (0.048) (0.052) (0.049) (0.020)*** [0.018] [0.026] [0.027] [0.029]** [0.028]*** [0.022]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,065 8,065 8,065 8,065 8,065 8,065 R-squared 0.000 0.046 0.000 0.005 0.169 0.188 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.023 0.011 0.021 0.088 0.092 0.088 (0.041) (0.042) (0.044) (0.049)* (0.044)* (0.021)*** [0.021] [0.025] [0.026] [0.029]*** [0.027]*** [0.023]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,065 8,065 8,065 8,065 8,065 8,065 R-squared 0.000 0.045 0.000 0.006 0.162 0.176 Panel C: Full sample (no weights) KALAHI-CIDSS -0.013 0.008 0.013 0.070 0.075 0.074 (0.034) (0.047) (0.047) (0.051) (0.049) (0.021)*** [0.018] [0.025] [0.026] [0.029]** [0.028]*** [0.021]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 9,018 9,018 9,018 9,018 9,018 9,018 R-squared 0.000 0.048 0.000 0.004 0.168 0.186 Panel D: Full sample (with weights) KALAHI-CIDSS -0.023 0.014 0.023 0.089 0.093 0.088 (0.041) (0.042) (0.042) (0.047)* (0.043)** (0.019)*** [0.020] [0.024] [0.025] [0.028]*** [0.027]*** [0.022]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 9,018 9,018 9,018 9,018 9,018 9,018 R-squared 0.000 0.046 0.000 0.006 0.160 0.174 Notes: Results from OLS (Column 1) and fixed-effects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual is employed. In Panels A and B, the sample is restricted to house- holds that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes significance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  35 Table 14: Impacts on house accessibility (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.016 0.124 0.124 0.094 0.094 0.094 (0.073) (0.040)*** (0.040)*** (0.040)** (0.040)** (0.036)** [0.037] [0.028]*** [0.028]*** [0.030]*** [0.030]*** [0.029]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.144 0.016 0.018 0.019 0.030 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.043 0.134 0.134 0.112 0.111 0.108 (0.082) (0.042)*** (0.042)*** (0.050)** (0.050)** (0.034)*** [0.041] [0.030]*** [0.030]*** [0.033]*** [0.033]*** [0.031]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.002 0.133 0.018 0.020 0.021 0.037 Panel C: Full sample (no weights) KALAHI-CIDSS -0.004 0.121 0.126 0.093 0.094 0.095 (0.070) (0.041)*** (0.039)*** (0.039)** (0.039)** (0.036)** [0.035] [0.027]*** [0.027]*** [0.029]*** [0.029]*** [0.029]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.141 0.017 0.020 0.021 0.030 Panel D: Full sample (with weights) KALAHI-CIDSS -0.032 0.132 0.136 0.111 0.110 0.107 (0.080) (0.040)*** (0.040)*** (0.048)** (0.048)** (0.033)*** [0.040] [0.028]*** [0.029]*** [0.033]*** [0.033]*** [0.031]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.131 0.019 0.021 0.022 0.037 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent’s house is accessible all year. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for po- tential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 36  Philippines Table 15: Impacts on number of trips to municipal center (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -1.352 0.430 0.513 0.625 0.639 0.639 (0.777) (0.239)* (0.222)** (0.289)** (0.284)** (0.289)** [0.558]** [0.215]** [0.230]** [0.281]** [0.274]** [0.276]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,629 5,629 5,629 5,629 5,629 5,629 R-squared 0.007 0.085 0.001 0.002 0.008 0.008 Panel B: Balanced panel (with weights) KALAHI-CIDSS -1.455 0.415 0.496 0.656 0.677 0.675 (0.667)** (0.235)* (0.215)** (0.241)** (0.233)** (0.233)** [0.581]** [0.218]* [0.232]** [0.289]** [0.280]** [0.277]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,629 5,629 5,629 5,629 5,629 5,629 R-squared 0.008 0.069 0.001 0.001 0.009 0.009 Panel C: Full sample (no weights) KALAHI-CIDSS -1.271 0.405 0.458 0.571 0.587 0.587 (0.775) (0.176)** (0.210)** (0.268)* (0.265)** (0.268)** [0.544]** [0.187]** [0.213]** [0.264]** [0.257]** [0.259]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,376 6,376 6,376 6,376 6,376 6,376 R-squared 0.006 0.087 0.001 0.001 0.008 0.008 Panel D: Full sample (with weights) KALAHI-CIDSS -1.435 0.363 0.445 0.590 0.614 0.619 (0.676)* (0.169)** (0.202)** (0.217)** (0.214)** (0.213)** [0.578]** [0.187]* [0.215]** [0.273]** [0.264]** [0.261]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,376 6,376 6,376 6,376 6,376 6,376 R-squared 0.007 0.070 0.001 0.001 0.009 0.009 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is the number of trips to the municipal center taken by the respondent in the month before the survey took place. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in paren- theses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  37 Table 16: Impacts on log per capita transportation expenditures (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.146 0.579 0.579 0.353 0.364 0.366 (0.169) (0.095)*** (0.094)*** (0.141)** (0.145)** (0.138)** [0.097] [0.081]*** [0.080]*** [0.105]*** [0.105]*** [0.103]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.050 0.017 0.023 0.034 0.036 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.033 0.597 0.597 0.405 0.417 0.399 (0.168) (0.068)*** (0.068)*** (0.108)*** (0.111)*** (0.108)*** [0.091] [0.081]*** [0.081]*** [0.100]*** [0.100]*** [0.099]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.045 0.018 0.022 0.033 0.034 Panel C: Full sample (no weights) KALAHI-CIDSS 0.180 0.576 0.559 0.325 0.338 0.340 (0.172) (0.090)*** (0.102)*** (0.147)** (0.149)** (0.145)** [0.096]* [0.074]*** [0.079]*** [0.103]*** [0.102]*** [0.101]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.002 0.051 0.016 0.022 0.032 0.034 Panel D: Full sample (with weights) KALAHI-CIDSS 0.055 0.581 0.568 0.372 0.383 0.365 (0.164) (0.068)*** (0.072)*** (0.109)*** (0.114)*** (0.114)*** [0.089] [0.070]*** [0.075]*** [0.095]*** [0.094]*** [0.094]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.045 0.016 0.020 0.031 0.032 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is the log per capita transport expenditures. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 38  Philippines Table 17: Impacts on access to level II and III water supply (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.004 -0.013 -0.013 -0.015 -0.017 -0.017 (0.057) (0.045) (0.045) (0.053) (0.053) (0.038) [0.036] [0.032] [0.032] [0.038] [0.038] [0.036] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.087 0.000 0.000 0.003 0.027 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.043 -0.024 -0.024 0.009 0.007 -0.006 (0.052) (0.057) (0.057) (0.062) (0.062) (0.037) [0.038] [0.036] [0.036] [0.041] [0.041] [0.037] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.002 0.077 0.001 0.003 0.005 0.034 Panel C: Full sample (no weights) KALAHI-CIDSS 0.007 0.017 -0.005 -0.007 -0.008 -0.009 (0.057) (0.046) (0.043) (0.052) (0.052) (0.039) [0.036] [0.030] [0.031] [0.036] [0.036] [0.034] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.088 0.000 0.000 0.002 0.026 Panel D: Full sample (with weights) KALAHI-CIDSS -0.034 -0.001 -0.015 0.017 0.016 0.003 (0.054) (0.059) (0.056) (0.061) (0.060) (0.037) [0.038] [0.034] [0.034] [0.039] [0.039] [0.035] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.081 0.000 0.002 0.005 0.034 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the household has access to either level II or level II water supply. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  39 Table 18: Impacts on access to safe water (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.052 0.004 0.004 -0.001 0.000 0.000 (0.031) (0.015) (0.015) (0.024) (0.024) (0.024) [0.020]** [0.014] [0.014] [0.021] [0.021] [0.021] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.005 0.048 0.000 0.000 0.002 0.005 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.051 0.005 0.005 -0.004 -0.003 -0.001 (0.039) (0.013) (0.013) (0.021) (0.021) (0.023) [0.023]** [0.014] [0.014] [0.020] [0.020] [0.021] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.004 0.050 0.000 0.000 0.003 0.005 Panel C: Full sample (no weights) KALAHI-CIDSS 0.048 0.001 0.000 -0.003 -0.002 -0.002 (0.029) (0.014) (0.014) (0.022) (0.022) (0.022) [0.019]** [0.013] [0.014] [0.020] [0.020] [0.020] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.047 0.000 0.000 0.002 0.005 Panel D: Full sample (with weights) KALAHI-CIDSS 0.047 0.003 0.001 -0.006 -0.004 -0.003 (0.037) (0.013) (0.013) (0.021) (0.021) (0.023) [0.022]** [0.013] [0.013] [0.019] [0.019] [0.020] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.003 0.047 0.000 0.000 0.003 0.005 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the water is safe for drinking. In Panels A and B, the sample is restricted to house- holds that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 40  Philippines Table 19: Impacts on access to water-sealed toilets (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.007 0.107 0.107 0.023 0.022 0.022 (0.041) (0.029)*** (0.029)*** (0.032) (0.032) (0.029) [0.029] [0.023]*** [0.023]*** [0.026] [0.026] [0.026] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.068 0.015 0.034 0.036 0.039 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.002 0.107 0.107 0.034 0.033 0.027 (0.051) (0.025)*** (0.025)*** (0.029) (0.030) (0.026) [0.032] [0.021]*** [0.020]*** [0.025] [0.025] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.072 0.014 0.029 0.033 0.036 Panel C: Full sample (no weights) KALAHI-CIDSS 0.016 0.118 0.095 0.011 0.011 0.011 (0.039) (0.025)*** (0.030)*** (0.032) (0.032) (0.029) [0.029] [0.023]*** [0.024]*** [0.027] [0.027] [0.027] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.064 0.013 0.032 0.035 0.037 Panel D: Full sample (with weights) KALAHI-CIDSS 0.008 0.112 0.096 0.023 0.022 0.016 (0.051) (0.023)*** (0.026)*** (0.030) (0.030) (0.026) [0.032] [0.020]*** [0.020]*** [0.025] [0.025] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.070 0.012 0.027 0.031 0.034 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the household has water sealed toilets. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  41 Table 20: Impacts on access to health services (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.020 -0.005 -0.008 0.001 -0.003 -0.003 (0.052) (0.048) (0.045) (0.059) (0.058) (0.058) [0.024] [0.028] [0.026] [0.032] [0.031] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 16,136 16,136 16,136 16,136 16,136 16,136 R-squared 0.000 0.044 0.000 0.000 0.041 0.041 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.013 0.021 0.023 0.044 0.038 0.038 (0.069) (0.062) (0.060) (0.067) (0.068) (0.064) [0.030] [0.031] [0.032] [0.036] [0.036] [0.035] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 16,136 16,136 16,136 16,136 16,136 16,136 R-squared 0.000 0.054 0.000 0.001 0.043 0.043 Panel C: Full sample (no weights) KALAHI-CIDSS -0.020 -0.011 -0.012 -0.003 -0.007 -0.007 (0.052) (0.046) (0.041) (0.053) (0.051) (0.051) [0.024] [0.028] [0.025] [0.030] [0.029] [0.029] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 18,040 18,040 18,040 18,040 18,040 18,040 R-squared 0.000 0.045 0.000 0.000 0.042 0.042 Panel D: Full sample (with weights) KALAHI-CIDSS -0.012 0.012 0.017 0.036 0.031 0.032 (0.069) (0.057) (0.056) (0.062) (0.062) (0.059) [0.030] [0.030] [0.030] [0.034] [0.034] [0.033] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 18,040 18,040 18,040 18,040 18,040 18,040 R-squared 0.000 0.055 0.000 0.001 0.042 0.043 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual visited a health clinic when sick. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for po- tential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 42  Philippines Table 21: Impacts on access to health services (Male sample) (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.017 0.006 0.001 0.026 0.017 0.017 (0.048) (0.045) (0.046) (0.060) (0.058) (0.058) [0.024] [0.028] [0.029] [0.034] [0.033] [0.034] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,296 8,296 8,296 8,296 8,296 8,296 R-squared 0.000 0.039 0.000 0.001 0.050 0.050 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.011 0.029 0.034 0.073 0.062 0.063 (0.064) (0.055) (0.060) (0.069) (0.068) (0.065) [0.030] [0.033] [0.036] [0.041]* [0.040] [0.039] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,296 8,296 8,296 8,296 8,296 8,296 R-squared 0.000 0.048 0.001 0.003 0.051 0.052 Panel C: Full sample (no weights) KALAHI-CIDSS -0.019 -0.002 -0.008 0.016 0.009 0.009 (0.047) (0.043) (0.042) (0.054) (0.051) (0.051) [0.024] [0.028] [0.027] [0.032] [0.031] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 9,231 9,231 9,231 9,231 9,231 9,231 R-squared 0.000 0.039 0.000 0.001 0.051 0.051 Panel D: Full sample (with weights) KALAHI-CIDSS -0.012 0.020 0.025 0.062 0.052 0.052 (0.063) (0.052) (0.056) (0.063) (0.061) (0.059) [0.030] [0.031] [0.034] [0.038] [0.037] [0.036] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 9,231 9,231 9,231 9,231 9,231 9,231 R-squared 0.000 0.048 0.000 0.002 0.052 0.052 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual visited a health clinic when sick. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for po- tential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  43 Table 22: Impacts on access to health services (Female sample) (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.024 -0.017 -0.015 -0.022 -0.017 -0.018 (0.057) (0.053) (0.050) (0.062) (0.061) (0.059) [0.027] [0.030] [0.030] [0.036] [0.035] [0.035] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 7,840 7,840 7,840 7,840 7,840 7,840 R-squared 0.001 0.051 0.000 0.000 0.051 0.051 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.016 0.014 0.018 0.021 0.025 0.026 (0.075) (0.069) (0.068) (0.076) (0.072) (0.068) [0.032] [0.035] [0.034] [0.040] [0.040] [0.038] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 7,840 7,840 7,840 7,840 7,840 7,840 R-squared 0.000 0.062 0.000 0.000 0.052 0.053 Panel C: Full sample (no weights) KALAHI-CIDSS -0.021 -0.021 -0.014 -0.020 -0.014 -0.015 (0.057) (0.049) (0.045) (0.056) (0.054) (0.052) [0.027] [0.030] [0.028] [0.034] [0.033] [0.032] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,809 8,809 8,809 8,809 8,809 8,809 R-squared 0.000 0.053 0.000 0.000 0.050 0.050 Panel D: Full sample (with weights) KALAHI-CIDSS -0.013 0.004 0.016 0.017 0.022 0.023 (0.075) (0.063) (0.063) (0.069) (0.066) (0.062) [0.032] [0.032] [0.032] [0.038] [0.037] [0.036] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 8,809 8,809 8,809 8,809 8,809 8,809 R-squared 0.000 0.065 0.000 0.000 0.051 0.052 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual visited a health clinic when sick. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for po- tential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 44  Philippines Table 23: Impacts on school enrollment (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.002 0.001 -0.067 -0.021 -0.031 -0.031 (0.012) (0.014) (0.012)*** (0.014) (0.014)** (0.011)** [0.012] [0.014] [0.013]*** [0.015] [0.013]** [0.013]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 12,573 12,573 12,573 12,573 12,573 12,573 R-squared 0.000 0.004 0.003 0.006 0.453 0.454 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.007 0.005 -0.061 -0.021 -0.031 -0.030 (0.012) (0.017) (0.017)*** (0.015) (0.016)* (0.013)** [0.013] [0.014] [0.014]*** [0.016] [0.015]** [0.015]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 12,573 12,573 12,573 12,573 12,573 12,573 R-squared 0.000 0.003 0.002 0.005 0.451 0.452 Panel C: Full sample (no weights) KALAHI-CIDSS 0.000 -0.002 -0.068 -0.025 -0.038 -0.037 (0.013) (0.011) (0.012)*** (0.013)* (0.014)** (0.011)*** [0.011] [0.011] [0.012]*** [0.014]* [0.012]*** [0.012]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 13,961 13,961 13,961 13,961 13,961 13,961 R-squared 0.000 0.004 0.003 0.006 0.449 0.449 Panel D: Full sample (with weights) KALAHI-CIDSS 0.005 0.001 -0.061 -0.024 -0.035 -0.033 (0.012) (0.014) (0.016)*** (0.014) (0.016)** (0.013)** [0.013] [0.013] [0.014]*** [0.016] [0.015]** [0.015]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 13,961 13,961 13,961 13,961 13,961 13,961 R-squared 0.000 0.003 0.002 0.005 0.448 0.449 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual is enrolled in school. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  45 Table 24: Impacts on school enrollment for boys (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.025 -0.015 -0.124 -0.047 -0.052 -0.052 (0.015) (0.016) (0.013)*** (0.018)** (0.017)*** (0.013)*** [0.016] [0.019] [0.018]*** [0.020]** [0.017]*** [0.017]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,847 6,847 6,847 6,847 6,847 6,847 R-squared 0.001 0.007 0.011 0.020 0.406 0.407 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.024 -0.015 -0.123 -0.054 -0.053 -0.052 (0.014) (0.019) (0.016)*** (0.017)*** (0.017)*** (0.012)*** [0.019] [0.020] [0.018]*** [0.021]** [0.018]*** [0.018]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,847 6,847 6,847 6,847 6,847 6,847 R-squared 0.001 0.006 0.010 0.017 0.409 0.411 Panel C: Full sample (no weights) KALAHI-CIDSS -0.032 -0.023 -0.126 -0.051 -0.060 -0.059 (0.016)* (0.014) (0.012)*** (0.017)*** (0.016)*** (0.012)*** [0.016]** [0.017] [0.017]*** [0.019]*** [0.017]*** [0.017]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 7,525 7,525 7,525 7,525 7,525 7,525 R-squared 0.001 0.007 0.011 0.020 0.403 0.404 Panel D: Full sample (with weights) KALAHI-CIDSS -0.028 -0.021 -0.122 -0.055 -0.056 -0.055 (0.015)* (0.017) (0.016)*** (0.017)*** (0.017)*** (0.012)*** [0.018] [0.019] [0.018]*** [0.021]** [0.019]*** [0.019]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 7,525 7,525 7,525 7,525 7,525 7,525 R-squared 0.001 0.006 0.010 0.017 0.406 0.407 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual is enrolled in school. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 46  Philippines Table 25: Impacts on school enrollment for girls (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.027 0.014 -0.056 0.000 -0.020 -0.020 (0.013)* (0.014) (0.008)*** (0.013) (0.012) (0.012) [0.015]* [0.016] [0.014]*** [0.017] [0.014] [0.014] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,726 5,726 5,726 5,726 5,726 5,726 R-squared 0.001 0.007 0.003 0.009 0.466 0.466 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.035 0.023 -0.047 0.007 -0.021 -0.019 (0.013)** (0.015) (0.011)*** (0.016) (0.013) (0.012) [0.016]** [0.017] [0.016]*** [0.020] [0.017] [0.016] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,726 5,726 5,726 5,726 5,726 5,726 R-squared 0.002 0.007 0.002 0.008 0.456 0.457 Panel C: Full sample (no weights) KALAHI-CIDSS 0.031 0.018 -0.053 -0.001 -0.024 -0.024 (0.013)** (0.011) (0.007)*** (0.012) (0.011)* (0.011)** [0.014]** [0.014] [0.013]*** [0.016] [0.013]* [0.013]* Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,436 6,436 6,436 6,436 6,436 6,436 R-squared 0.001 0.007 0.003 0.008 0.460 0.460 Panel D: Full sample (with weights) KALAHI-CIDSS 0.039 0.023 -0.046 0.005 -0.025 -0.023 (0.013)*** (0.012)* (0.010)*** (0.015) (0.012)* (0.011)* [0.015]** [0.016] [0.015]*** [0.019] [0.016] [0.015] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,436 6,436 6,436 6,436 6,436 6,436 R-squared 0.002 0.007 0.002 0.007 0.453 0.453 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the individual is enrolled in school. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  47 Table 26: Impacts on attendance in village assemblies (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.115 0.124 0.124 0.109 0.107 0.107 (0.060)* (0.048)** (0.048)** (0.056)* (0.055)* (0.052)* [0.033]*** [0.031]*** [0.031]*** [0.036]*** [0.035]*** [0.034]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.013 0.150 0.015 0.015 0.024 0.027 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.110 0.101 0.101 0.079 0.078 0.082 (0.060)* (0.051)* (0.051)* (0.056) (0.054) (0.051) [0.031]*** [0.028]*** [0.028]*** [0.033]** [0.032]** [0.032]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.011 0.125 0.009 0.010 0.020 0.023 Panel C: Full sample (no weights) KALAHI-CIDSS 0.112 0.121 0.106 0.090 0.087 0.087 (0.057)* (0.049)** (0.049)** (0.057) (0.055) (0.053) [0.031]*** [0.029]*** [0.030]*** [0.034]** [0.034]** [0.033]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.012 0.147 0.011 0.012 0.020 0.024 Panel D: Full sample (with weights) KALAHI-CIDSS 0.108 0.099 0.088 0.064 0.063 0.068 (0.058)* (0.052)* (0.052) (0.057) (0.055) (0.052) [0.030]*** [0.027]*** [0.027]*** [0.032]** [0.032]** [0.031]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.011 0.123 0.007 0.008 0.018 0.021 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if an household member joined a village assembly in the 6 months preceding the survey. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in paren- theses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 48  Philippines Table 27: Impacts on willingness to contribute money to community projects (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.090 0.140 0.140 0.104 0.102 0.104 (0.075) (0.072)* (0.072)* (0.079) (0.079) (0.036)** [0.033]*** [0.036]*** [0.036]*** [0.040]** [0.040]** [0.029]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.007 0.099 0.015 0.017 0.020 0.082 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.062 0.129 0.129 0.116 0.114 0.102 (0.082) (0.073) (0.073)* (0.080) (0.079) (0.037)** [0.035]* [0.036]*** [0.036]*** [0.042]*** [0.042]*** [0.031]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.100 0.012 0.012 0.015 0.068 Panel C: Full sample (no weights) KALAHI-CIDSS 0.099 0.155 0.146 0.109 0.107 0.107 (0.075) (0.077)* (0.074)* (0.081) (0.081) (0.038)** [0.032]*** [0.036]*** [0.035]*** [0.040]*** [0.040]*** [0.028]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.009 0.091 0.017 0.019 0.021 0.084 Panel D: Full sample (with weights) KALAHI-CIDSS 0.065 0.136 0.131 0.116 0.114 0.101 (0.083) (0.078) (0.075) (0.081) (0.081) (0.037)** [0.035]* [0.036]*** [0.036]*** [0.042]*** [0.041]*** [0.030]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.096 0.012 0.013 0.015 0.069 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable s a dummy equal to one if the respondent indicated being willing to contribute money to community projects. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signi- cance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  49 Table 28: Impacts on trust that others are willing to help if needed (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.028 0.021 0.021 0.055 0.053 0.054 (0.030) (0.037) (0.037) (0.045) (0.044) (0.023)** [0.017] [0.024] [0.024] [0.030]* [0.030]* [0.025]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.012 0.000 0.003 0.007 0.029 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.024 0.029 0.029 0.068 0.066 0.066 (0.026) (0.032) (0.032) (0.038)* (0.037)* (0.025)** [0.016] [0.023] [0.023] [0.029]** [0.029]** [0.026]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.010 0.001 0.004 0.008 0.024 Panel C: Full sample (no weights) KALAHI-CIDSS 0.029 0.027 0.024 0.057 0.055 0.055 (0.031) (0.042) (0.041) (0.048) (0.047) (0.024)** [0.017]* [0.024] [0.024] [0.029]* [0.029]* [0.024]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.012 0.001 0.003 0.007 0.030 Panel D: Full sample (with weights) KALAHI-CIDSS 0.025 0.032 0.030 0.068 0.066 0.065 (0.027) (0.037) (0.035) (0.039)* (0.038) (0.024)** [0.017] [0.023] [0.023] [0.029]** [0.028]** [0.025]*** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.010 0.001 0.004 0.008 0.024 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent agrees with the statement “Most people in this barangay/neighbor- hood are willing to help if you need it�. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 50  Philippines Table 29: Impacts on willingness to contribute time to community projects (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.023 0.016 0.016 0.006 0.007 0.007 (0.045) (0.038) (0.038) (0.043) (0.043) (0.032) [0.020] [0.021] [0.021] [0.027] [0.027] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.040 0.000 0.000 0.002 0.019 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.017 0.009 0.009 0.032 0.033 0.027 (0.046) (0.036) (0.036) (0.044) (0.044) (0.038) [0.022] [0.020] [0.020] [0.027] [0.027] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.039 0.000 0.001 0.002 0.020 Panel C: Full sample (no weights) KALAHI-CIDSS 0.030 0.024 0.019 0.008 0.010 0.009 (0.043) (0.036) (0.036) (0.041) (0.041) (0.032) [0.020] [0.022] [0.021] [0.027] [0.027] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.036 0.000 0.001 0.003 0.018 Panel D: Full sample (with weights) KALAHI-CIDSS 0.021 0.015 0.011 0.033 0.034 0.028 (0.045) (0.035) (0.034) (0.042) (0.042) (0.038) [0.021] [0.020] [0.020] [0.027] [0.027] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.037 0.000 0.001 0.002 0.019 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable s a dummy equal to one if the respondent indicated being willing to contribute time to community projects. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  51 Table 30: Impacts on participation in bayanihan (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.037 -0.012 -0.012 -0.002 -0.003 -0.002 (0.095) (0.029) (0.029) (0.050) (0.050) (0.032) [0.039] [0.021] [0.021] [0.030] [0.030] [0.027] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.213 0.000 0.000 0.003 0.016 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.016 -0.031 -0.031 -0.002 -0.003 -0.012 (0.110) (0.041) (0.041) (0.055) (0.056) (0.039) [0.042] [0.025] [0.025] [0.033] [0.032] [0.030] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.216 0.001 0.003 0.007 0.019 Panel C: Full sample (no weights) KALAHI-CIDSS 0.040 0.004 -0.015 -0.009 -0.009 -0.009 (0.092) (0.033) (0.030) (0.051) (0.051) (0.033) [0.038] [0.021] [0.021] [0.030] [0.030] [0.026] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.208 0.000 0.000 0.004 0.018 Panel D: Full sample (with weights) KALAHI-CIDSS 0.017 -0.017 -0.032 -0.004 -0.005 -0.016 (0.109) (0.046) (0.042) (0.055) (0.056) (0.039) [0.042] [0.024] [0.024] [0.032] [0.032] [0.029] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.211 0.001 0.003 0.007 0.020 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if an household member joined bayanihan activities in the 6 months preceding the survey. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in paren- theses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 52  Philippines Table 31: Impacts on group membership (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.056 0.113 0.113 0.010 0.008 0.008 (0.060) (0.020)*** (0.020)*** (0.033) (0.032) (0.032) [0.030]* [0.019]*** [0.019]*** [0.025] [0.025] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.112 0.012 0.032 0.037 0.041 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.061 0.116 0.116 0.027 0.024 0.022 (0.070) (0.016)*** (0.016)*** (0.031) (0.030) (0.029) [0.032]* [0.019]*** [0.019]*** [0.026] [0.026] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.119 0.012 0.028 0.034 0.039 Panel C: Full sample (no weights) KALAHI-CIDSS 0.062 0.121 0.103 0.005 0.003 0.003 (0.057) (0.019)*** (0.019)*** (0.032) (0.031) (0.031) [0.029]** [0.019]*** [0.019]*** [0.025] [0.025] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.003 0.104 0.011 0.030 0.035 0.038 Panel D: Full sample (with weights) KALAHI-CIDSS 0.065 0.121 0.109 0.024 0.022 0.019 (0.068) (0.016)*** (0.017)*** (0.030) (0.029) (0.029) [0.031]** [0.018]*** [0.019]*** [0.025] [0.025] [0.024] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.114 0.011 0.026 0.032 0.037 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if an household member is a member of a formal group. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  53 Table 32: Impacts on trust of community members (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.011 0.022 0.022 0.042 0.043 0.044 (0.045) (0.032) (0.032) (0.038) (0.037) (0.029) [0.023] [0.023] [0.023] [0.028] [0.028] [0.027] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.034 0.000 0.001 0.003 0.010 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.017 0.023 0.023 0.058 0.059 0.057 (0.040) (0.030) (0.030) (0.035) (0.034) (0.025)** [0.022] [0.023] [0.023] [0.027]** [0.027]** [0.025]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.028 0.000 0.002 0.005 0.011 Panel C: Full sample (no weights) KALAHI-CIDSS -0.012 0.014 0.015 0.033 0.034 0.034 (0.043) (0.029) (0.030) (0.037) (0.036) (0.029) [0.022] [0.023] [0.023] [0.029] [0.029] [0.028] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.034 0.000 0.001 0.003 0.009 Panel D: Full sample (with weights) KALAHI-CIDSS -0.016 0.017 0.018 0.048 0.049 0.047 (0.038) (0.027) (0.029) (0.033) (0.032) (0.024)* [0.022] [0.023] [0.023] [0.026]* [0.026]* [0.025]* Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.028 0.000 0.002 0.004 0.011 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent agrees with the statement “Most people who live in this barangay/ neighborhood can be trusted�. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 54  Philippines Table 33: Impacts on need to be alert (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.008 -0.054 -0.054 0.028 0.026 0.026 (0.032) (0.052) (0.052) (0.058) (0.058) (0.053) [0.018] [0.027]** [0.027]** [0.032] [0.032] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.034 0.002 0.011 0.015 0.018 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.008 -0.080 -0.080 -0.012 -0.016 -0.010 (0.037) (0.047) (0.047) (0.059) (0.059) (0.046) [0.019] [0.026]*** [0.026]*** [0.035] [0.035] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.033 0.004 0.011 0.015 0.020 Panel C: Full sample (no weights) KALAHI-CIDSS -0.013 -0.054 -0.059 0.015 0.013 0.013 (0.036) (0.056) (0.055) (0.059) (0.060) (0.056) [0.019] [0.027]** [0.027]** [0.032] [0.032] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.038 0.003 0.011 0.014 0.016 Panel D: Full sample (with weights) KALAHI-CIDSS -0.013 -0.079 -0.084 -0.022 -0.026 -0.022 (0.039) (0.050) (0.050) (0.061) (0.061) (0.048) [0.020] [0.026]*** [0.026]*** [0.035] [0.034] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.036 0.005 0.011 0.015 0.019 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent agrees with the statement “In this barangay/neighborhood, one has to be alert or someone is likely to take advantage of you�. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  55 Table 34: Impacts on trust of other with money (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.017 0.014 0.014 0.017 0.015 0.016 (0.028) (0.041) (0.041) (0.054) (0.053) (0.046) [0.018] [0.023] [0.023] [0.029] [0.029] [0.027] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.008 0.000 0.000 0.002 0.004 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.009 0.011 0.011 0.002 -0.001 -0.004 (0.026) (0.038) (0.038) (0.050) (0.049) (0.046) [0.019] [0.024] [0.024] [0.029] [0.029] [0.028] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.000 0.007 0.000 0.000 0.003 0.004 Panel C: Full sample (no weights) KALAHI-CIDSS 0.018 0.018 0.013 0.014 0.012 0.013 (0.029) (0.045) (0.043) (0.056) (0.056) (0.048) [0.017] [0.024] [0.023] [0.029] [0.030] [0.027] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.008 0.000 0.000 0.002 0.005 Panel D: Full sample (with weights) KALAHI-CIDSS 0.010 0.015 0.009 0.000 -0.004 -0.008 (0.027) (0.041) (0.041) (0.052) (0.052) (0.047) [0.018] [0.024] [0.024] [0.029] [0.029] [0.028] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.000 0.006 0.000 0.000 0.003 0.004 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions.The dependent variable is a dummy equal to one if the respondent agrees with the statement “In this barangay/neighborhood, people generally do not trust each other in matters of lending and borrowing money�. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 56  Philippines Table 35: Impacts on trust of local officials (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.081 -0.138 -0.138 -0.055 -0.055 -0.055 (0.051) (0.043)*** (0.043)*** (0.054) (0.054) (0.050) [0.024]*** [0.025]*** [0.025]*** [0.031]* [0.031]* [0.030]* Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.006 0.064 0.015 0.025 0.025 0.032 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.059 -0.160 -0.160 -0.079 -0.080 -0.083 (0.061) (0.049)*** (0.049)*** (0.060) (0.060) (0.052) [0.029]** [0.030]*** [0.030]*** [0.035]** [0.035]** [0.033]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.069 0.018 0.029 0.030 0.039 Panel C: Full sample (no weights) KALAHI-CIDSS -0.096 -0.143 -0.143 -0.060 -0.061 -0.062 (0.050)* (0.040)*** (0.042)*** (0.052) (0.052) (0.050) [0.023]*** [0.024]*** [0.025]*** [0.031]* [0.031]** [0.030]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.008 0.066 0.016 0.027 0.027 0.033 Panel D: Full sample (with weights) KALAHI-CIDSS -0.069 -0.156 -0.159 -0.078 -0.079 -0.082 (0.057) (0.042)*** (0.045)*** (0.056) (0.057) (0.050) [0.027]** [0.027]*** [0.028]*** [0.033]** [0.033]** [0.032]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.068 0.019 0.029 0.031 0.039 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent trusts local ocials. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  57 Table 36: Impacts on trust of national officials (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.066 -0.110 -0.110 -0.014 -0.014 -0.014 (0.042) (0.038)** (0.038)** (0.051) (0.051) (0.051) [0.020]*** [0.023]*** [0.023]*** [0.030] [0.030] [0.030] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.004 0.060 0.010 0.026 0.027 0.031 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.039 -0.116 -0.116 -0.037 -0.038 -0.030 (0.047) (0.041)** (0.041)** (0.055) (0.055) (0.052) [0.023]* [0.030]*** [0.030]*** [0.035] [0.035] [0.035] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.002 0.064 0.011 0.022 0.023 0.030 Panel C: Full sample (no weights) KALAHI-CIDSS -0.078 -0.125 -0.116 -0.024 -0.024 -0.024 (0.040)* (0.037)*** (0.039)*** (0.052) (0.052) (0.053) [0.019]*** [0.023]*** [0.024]*** [0.030] [0.030] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.006 0.064 0.012 0.026 0.027 0.031 Panel D: Full sample (with weights) KALAHI-CIDSS -0.047 -0.124 -0.117 -0.041 -0.041 -0.033 (0.045) (0.038)*** (0.040)** (0.055) (0.055) (0.053) [0.022]** [0.028]*** [0.029]*** [0.034] [0.034] [0.034] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.002 0.065 0.011 0.022 0.024 0.030 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable s a dummy equal to one if the respondent trusts national ocials. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 58  Philippines Table 37: Impacts on trust of strangers (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.029 -0.024 -0.024 0.023 0.023 0.022 (0.011)** (0.014) (0.014) (0.023) (0.023) (0.017) [0.007]*** [0.011]** [0.011]** [0.014] [0.014] [0.012]* Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.004 0.020 0.002 0.017 0.018 0.029 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.023 -0.027 -0.027 0.008 0.008 0.010 (0.010)** (0.014)* (0.014)* (0.019) (0.019) (0.019) [0.007]*** [0.011]** [0.011]** [0.014] [0.014] [0.013] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.002 0.019 0.003 0.012 0.015 0.021 Panel C: Full sample (no weights) KALAHI-CIDSS -0.031 -0.019 -0.021 0.026 0.025 0.025 (0.012)** (0.012) (0.013) (0.023) (0.022) (0.016) [0.008]*** [0.010]* [0.011]** [0.014]* [0.014]* [0.012]** Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.004 0.024 0.002 0.016 0.018 0.029 Panel D: Full sample (with weights) KALAHI-CIDSS -0.024 -0.022 -0.024 0.011 0.011 0.014 (0.009)** (0.012)* (0.013)* (0.018) (0.018) (0.017) [0.007]*** [0.011]** [0.011]** [0.014] [0.013] [0.013] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.003 0.020 0.002 0.012 0.014 0.019 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent trusts strangers. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  59 Table 38: Impacts on perceptions of peace (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS -0.051 -0.077 -0.077 -0.028 -0.026 -0.026 (0.035) (0.028)** (0.028)** (0.033) (0.033) (0.028) [0.023]** [0.025]*** [0.025]*** [0.031] [0.031] [0.030] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.003 0.035 0.006 0.011 0.013 0.020 Panel B: Balanced panel (with weights) KALAHI-CIDSS -0.055 -0.076 -0.076 -0.010 -0.008 -0.016 (0.037) (0.026)** (0.026)*** (0.031) (0.031) (0.025) [0.023]** [0.023]*** [0.023]*** [0.028] [0.028] [0.026] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.004 0.032 0.006 0.016 0.017 0.025 Panel C: Full sample (no weights) KALAHI-CIDSS -0.061 -0.082 -0.081 -0.037 -0.035 -0.035 (0.035) (0.026)*** (0.028)** (0.034) (0.034) (0.029) [0.023]*** [0.026]*** [0.025]*** [0.031] [0.031] [0.030] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.005 0.039 0.008 0.012 0.013 0.020 Panel D: Full sample (with weights) KALAHI-CIDSS -0.061 -0.078 -0.078 -0.016 -0.014 -0.021 (0.036) (0.025)*** (0.025)*** (0.031) (0.031) (0.026) [0.022]*** [0.023]*** [0.022]*** [0.027] [0.027] [0.025] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.005 0.034 0.006 0.016 0.017 0.025 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is the XXXXXXXXXXXXX. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weightare used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlationwithin municipality (resp. within vil- lage). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 60  Philippines Table 39: Impacts on knowledge of village budget (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.019 0.036 0.036 0.007 0.005 0.005 (0.020) (0.020)* (0.020)* (0.033) (0.033) (0.027) [0.013] [0.013]*** [0.013]*** [0.019] [0.019] [0.018] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.011 0.002 0.005 0.007 0.022 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.018 0.028 0.028 0.011 0.009 0.011 (0.022) (0.022) (0.022) (0.037) (0.037) (0.025) [0.013] [0.013]** [0.013]** [0.020] [0.020] [0.018] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.001 0.012 0.001 0.002 0.005 0.023 Panel C: Full sample (no weights) KALAHI-CIDSS 0.022 0.043 0.036 0.006 0.005 0.004 (0.019) (0.020)* (0.019)* (0.032) (0.032) (0.026) [0.013]* [0.013]*** [0.012]*** [0.018] [0.018] [0.017] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.013 0.002 0.005 0.007 0.022 Panel D: Full sample (with weights) KALAHI-CIDSS 0.019 0.034 0.028 0.011 0.009 0.010 (0.021) (0.021) (0.021) (0.035) (0.035) (0.024) [0.012] [0.014]** [0.013]** [0.019] [0.019] [0.017] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.001 0.013 0.001 0.002 0.005 0.022 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if the respondent knows the village budget. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential cor- relation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  61 Table 40: Impacts on participation in planning of development activities (1) (2) (3) (4) (5) (6) Panel A: Balanced panel (no weights) KALAHI-CIDSS 0.058 0.039 0.039 0.055 0.054 0.054 (0.043) (0.040) (0.040) (0.054) (0.054) (0.055) [0.024]** [0.025] [0.025] [0.032]* [0.032]* [0.032]* Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.004 0.041 0.001 0.002 0.010 0.011 Panel B: Balanced panel (with weights) KALAHI-CIDSS 0.060 0.020 0.020 0.025 0.023 0.024 (0.043) (0.037) (0.037) (0.052) (0.052) (0.052) [0.024]** [0.026] [0.026] [0.033] [0.032] [0.032] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 5,652 5,652 5,652 5,652 5,652 5,652 R-squared 0.004 0.040 0.000 0.000 0.010 0.012 Panel C: Full sample (no weights) KALAHI-CIDSS -0.061 -0.082 -0.081 -0.037 -0.035 -0.035 (0.035) (0.026)*** (0.028)** (0.034) (0.034) (0.029) [0.023]*** [0.026]*** [0.025]*** [0.031] [0.031] [0.030] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.005 0.039 0.008 0.012 0.013 0.020 Panel D: Full sample (with weights) KALAHI-CIDSS 0.057 0.024 0.018 0.022 0.020 0.020 (0.042) (0.038) (0.039) (0.053) (0.053) (0.053) [0.023]** [0.023] [0.024] [0.031] [0.031] [0.031] Fixed effects No Municipal Household Household Household Household Time trends No No No Overall Overall Regional Additional controls No No No No Yes Yes Observations 6,402 6,402 6,402 6,402 6,402 6,402 R-squared 0.003 0.038 0.000 0.000 0.010 0.012 Notes: Results from OLS (Column 1) and xed-eects (Columns 2-6) regressions. The dependent variable is a dummy equal to one if an household member joined planning activities for barangay development programs in the 6 months preceding the survey. In Panels A and B, the sample is restricted to households that are observed three times. In Panels C and D the full sample is used. In Panels B and D sample weights are used. The standard errors (in parentheses) (resp. [brackets]) account for potential correlation within municipality (resp. within village). * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. 62  Philippines Annexes Figure A-1: Project Coverage The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  63 Table A-1: Comparing treatment and control municipalities at baseline (Welfare indicators) Variable Treatment Control OLS OLS log per capita expenditures 9.499 9.623 -0.124 -0.122 (0.583) (0.586) [0.033]** [0.041]** log per capita expenditures 9.121 9.209 -0.094 -0.089 (poor households) (0.356) (0.314) [0.001]*** [0.000]*** log per capita expenditures 10.028 10.080 -0.043 -0.039 (non-poor households) (0.400) (0.464) [0.401] [0.485] poverty 0.584 0.525 0.059 0.063 (0.493) (0.500) [0.121] [0.109] non-food share of total consumption 33.025 34.675 -1.652 -1.683 (15.460) (17.565) [0.530] [0.499] log per capita food expenditures 9.063 9.149 -0.086 -0.082 (0.508) (0.466) [0.024]** [0.031]** log per capita non-food expenditures 8.258 8.395 -0.138 -0.135 (1.004) (1.111) [0.331] [0.330] self-rated poverty 0.708 0.697 0.011 0.018 (0.455) (0.460) [0.689] [0.507] Notes: Each row presents the 2003 average of the listed variable for the treatment (Column 1) and control (Column 2) groups. Each cell in Columns 3 and 4 is either the coecient on the dummy variable indicating whether the project was implemented in the municipality or the associated p-value in [brackets] from a dif- ferent OLS regression with a full set of province dummies. In Column 3 the full 2003 sample is used while in Column 4, the sample is restricted to households which are still in the sample in 2010. 64  Philippines Table A-2: Comparing treatment and control municipalities at baseline (Access indicators) Variable Treatment Control OLS OLS (1) (2) (3) (4) house accessibility 0.439 0.542 -0.102 -0.120 (0.496) (0.498) [0.111] [0.073]* number of trips to municipal center 3.376 5.794 -2.422 -2.561 (6.112) (9.124) [0.009]*** [0.010]*** log per capita transportation expenditures 5.201 5.410 -0.209 -0.258 (2.070) (2.139) [0.447] [0.330] access to level II and III water supply 0.440 0.447 -0.008 0.015 (0.497) (0.497) [0.905] [0.823] access to safe water 0.876 0.842 0.034 0.030 (0.330) (0.365) [0.195] [0.224] access to water-sealed toilets 0.530 0.592 -0.062 -0.064 (0.499) (0.492) [0.307] [0.304] Notes: Each row presents the 2003 average of the listed variable for the treatment (Column 1) and control (Column 2) groups. Each cell in Columns 3 and 4 is either the coecient on the dummy variable indicating whether the project was implemented in the municipality or the associated p-value in [brackets] from a dif- ferent OLS regression with a full set of province dummies. In Column 3 the full 2003 sample is used while in Column 4, the sample is restricted to households which are still in the sample in 2010. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  65 Table A-3: Comparing treatment and control municipalities at baseline (Social capital and local governance indicators) Variable Treatment Control OLS OLS attendance in village assemblies (1) (2) (3) (4) 0.611 0.612 0.000 -0.012 willingness to contribute money to community (0.488) (0.488) [1.000] [0.777] projects 0.361 0.376 -0.015 -0.018 trust that others are willing to help if needed (0.480) (0.485) [0.723] [0.638] 0.759 0.760 -0.001 -0.002 willingness to contribute time to community (0.428) (0.427) [0.981] [0.946] projects 0.749 0.706 0.044 0.033 participation in bayanihan (0.434) (0.456) [0.159] [0.228] 0.597 0.484 0.113 0.110 group membership (0.491) (0.500) [0.107] [0.095]* 0.324 0.312 0.012 0.013 trust community members (0.468) (0.464) [0.632] [0.604] 0.545 0.614 -0.070 -0.080 need to be alert (0.498) (0.487) [0.063]* [0.034]** 0.620 0.585 0.035 0.037 trust other with money (0.486) (0.493) [0.541] [0.493] 0.490 0.457 0.033 0.035 trust local officials (0.500) (0.498) [0.314] [0.304] 0.528 0.566 -0.038 -0.028 trust national officials (0.499) (0.496) [0.614] [0.711] 0.418 0.452 -0.034 -0.052 trust strangers (0.493) (0.498) [0.606] [0.426] 0.052 0.125 -0.073 -0.061 perceptions of peace (0.223) (0.331) [0.027]** [0.050]* 0.835 0.857 -0.021 -0.017 knowledge of village budget 0.105 0.111 -0.006 -0.007 (0.307) (0.314) [0.840] [0.821] participation in development activities 0.284 0.263 0.021 0.018 (0.451) (0.441) [0.604] [0.659] confidence to participate in development 0.416 0.400 0.016 0.010 activities (0.493) (0.490) [0.649] [0.770] Notes: Each row presents the 2003 average of the listed variable for the treatment (Column 1) and control (Column 2) groups. Each cell in Columns 3 and 4 is either the coecient on the dummy variable indicating whether the project was implemented in the municipality or the associated p-value in [brackets] from a dif- ferent OLS regression with a full set of province dummies. In Column 3 the full 2003 sample is used while in Column 4, the sample is restricted to households which are still in the sample in 2010. 66  Philippines Table A-4: Parallel Trend Hypothesis: Consumption Panel A: Per capita food consumption Placebo -1.268 -0.634 -0.643 -0.177 (1.052) (1.079) (0.835) (1.079) Observations 724 724 724 724 HH controls No No Yes Yes Mun. Dummies No Yes No Yes R-squared 0.01 0.15 0.38 0.47 Panel B: Log per capita food consumption Placebo -0.127 -0.043 -0.060 0.004 (0.130) (0.133) (0.107) (0.126) Observations 724 724 724 724 Mun. Dummies No Yes No Yes R-squared 0.01 0.15 0.48 0.56 Panel C: Per capita non-food consumption Placebo 2.464 4.202 5.969 5.744 (5.633) (5.478) (7.232) (6.971) Observations 724 724 724 724 HH controls No No Yes Yes Mun. Dummies No Yes No Yes R-squared 0.00 0.02 0.07 0.09 Panel D: Log per capita non-food consumption Placebo -0.188 0.116 -0.052 0.121 (0.199) (0.303) (0.156) (0.287) Observations 724 724 724 724 HH controls No No Yes Yes Mun. Dummies No Yes No Yes R-squared 0.02 0.12 0.46 0.50 Notes: Results from OLS regressions using 2000 and 2003 FIES data. The dependent variable is the per capita food consumption (PHP 1,000) in Panel A, the log per capita food consumption in Panel B, the per capita non-food consumption (PHP 1,000) in Panel C and, the log per capita non-food consumption in Panel D. The standard errors (in parentheses) account for potential correlation within municipality. All re- gressions include time-trends. * denotes signicance at the 10 percent, ** at the 5 percent and, *** at the 1 percent level. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  67 Table A-5: Parallel Trend Hypothesis: Asset Panel A: Electricity Placebo -0.159 0.185 -0.132 0.118 (0.099) ().222) (0.083) (0.190) Observations 724 724 724 724 HH controls No No Yes Yes Mun. Dummies No Yes No Yes R-squared 0.03 0.13 0.20 0.26 Panel B: Index of durable goods Placebo -0.389 1.124 -0.149 0.789 (0.506) (0.760) (0.356) (0.596) Observations 724 724 724 724 HH controls No No Yes Yes Mun. Dummies No Yes No Yes R-squared 0.01 0.13 0.37 0.42 Notes: Results from OLS regressions using 2000 and 2003 FIES data. The dependent variable is an as- set index in Panel A and a dummy equal to one if the household has access to electricity in Panel B. The standard errors (in parentheses) account for potential correlation within municipality. All regressions include time-trends. * denotes signicance at the 10 percent, ** at the 5 percent and, ***at the 1 percent level. 68  Philippines Table A-6: Are the determinants of attrition different in treatment and control municipalities? (Welfare indicators) Variable Treatment Control Chi-sq (1) (2) (3) log per capita expenditures 0.007 -0.035 0.185 (0.072) (0.157) [0.667] log per capita expenditures (poor households) -0.115 -0.111 0.000 (0.160) (0.357) [0.988] log per capita expenditures (non-poor households) 0.131 0.119 0.003 (0.158) (0.336) [0.955] poverty -0.005 0.018 0.039 (0.086) (0.187) [0.844] non-food share of total consumption 0.000 0.000 0.027 (0.003) (0.006) [0.870] log per capita food expenditures -0.017 -0.081 0.292 (0.083) (0.184) [0.589] log per capita non-food expenditures -0.008 -0.030 0.164 (0.041) (0.089) [0.685] self-rated poverty -0.004 -0.050 0.130 (0.092) (0.203) [0.718] Notes: Each row presents coeffcients from a different probit regression of the probability of dropping out of the sample between 203 and 2010. Each regression includes controls for the interaction of the variable listed with the treatment dummy (Column 2), its interaction with the control dummy (Column 3), the treat- ment dummy and a full set of province dummies. Column 3 reports tests of equality of the two coecients reported in Columns 1 and 2. Standard errors are in (parenthesis) and p-values are in [brackets]. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  69 Table A-7: Are the determinants of attrition different in treatment and control municipalities? (Access indicators) Variable Treatment Control Chi-sq (1) (2) (3) house accessibility 0.078 0.263 2.490 (0.086) (0.187) [0.115] number of trips to municipal center -0.001 0.000 0.027 (0.007) (0.015) [0.869] log per capita transportation expenditures -0.014 0.003 0.341 (0.020) (0.044) [0.559] access to level II and III water supply -0.297 -0.576 5.516 (0.086) (0.189) [0.019]** access to safe water 0.058 0.148 0.273 (0.130) (0.283) [0.601] access to water-sealed toilets -0.274 -0.327 0.203 (0.084) (0.187) [0.652] Notes: Each row presents coecients from a different probit regression of the probability of dropping out of the sample between 203 and 2010. Each regression includes controls for the interaction of the variable listed with the treatment dummy (Column 2), its interaction with the control dummy (Column 3), the treat- ment dummy and a full set of province dummies. Column 3 reports tests of equality of the two coecients reported in Columns 1 and 2. Standard errors are in (parenthesis) and p-values are in [brackets]. 70  Philippines Table A-8: Are the determinants of attrition different in treatment and control municipalities? (Social capital and local governance indicators) Variable Treatment Control Chi-sq (1) (2) (3) attendance in village assemblies -0.087 0.092 2.229 (0.090) (0.194) [0.135] willingness to contribute money to community projects -0.121 -0.110 0.007 (0.088) (0.193) [0.932] trust that others are willing to help if needed -0.064 -0.083 0.022 (0.095) (0.212) [0.882] willingness to contribute time to community projects -0.072 0.035 0.687 (0.095) (0.209) [0.407] participation in bayanihan -0.269 -0.316 0.152 (0.088) (0.190) [0.697] group membership -0.238 -0.277 0.098 (0.091) (0.200) [0.755] trust community members 0.114 0.262 1.583 (0.083) (0.186) [0.208] need to be alert 0.005 0.002 0.001 (0.086) (0.189) [0.980] trust other with money -0.024 -0.069 0.147 (0.083) (0.184) [0.701] trust local officials 0.003 -0.135 1.400 (0.083) (0.185) [0.237] trust national officials 0.171 0.362 2.685 (0.083) (0.185) [0.101] trust strangers -0.295 -0.628 2.010 (0.202) (0.421) [0.156] perceptions of peace 0.008 -0.065 0.199 (0.112) (0.254) [0.655] knowledge of village budget -0.316 -0.318 0.000 (0.147) (0.327) [0.993] participation in development activities -0.098 -0.037 0.212 (0.092) (0.206) [0.645] confidence to participate in development activities -0.018 0.021 0.108 (0.084) (0.187) [0.742] Notes: Each row presents coecients from a different probit regression of the probability of dropping out of the sample between 203 and 2010. Each regression includes controls for the interaction of the variable listed with the treatment dummy (Column 2), its interaction with the control dummy (Column 3), the treat- ment dummy and a full set of province dummies. Column 3 reports tests of equality of the two coecients reported in Columns 1 and 2. Standard errors are in (parenthesis) andp-values are in [brackets]. The KALAHI-CIDSS Impact Evaluation: A Revised Synthesis Report  71 Sustainable Development Department East Asia and the Pacific Region THE WORLD BANK 1818 H, Street N.W. Washington, D.C. 20433 USA Tel: 202 473 1000 Fax 202 477 6392 Internet URL: www.worldbank.org World Bank Office Manila 25th Floor, One Global Place 5th Avenue, Bontacio Global City Taguig City, Philippines Tel: (632) 465 2500 Fax: (632) 465 2505 Internet URL: www.worldbank.org.ph