99426 CASE STUDY Determinants of participation and transaction costs in Rwanda’s land markets DANIEL ALI, KLAUS DEININGER, MARKUS GOLDSTEIN, ELIANA LA FERRARA AND MARGUERITE DUPONCHEL INTRODUCTION Land markets play a limited role in subsistence economies with low skill-intensity of agricultural cultivation, equally distributed land endowments and little movement out of agriculture to join the non-farm economy. But, as the economy starts to diversify, the scope for efficiency-enhancing land transfers beyond immediate kin and for longer than just one season assumes significantly greater importance (Badiane et al. 2012). Lease markets can contribute to the diversification of the economic base in rural areas by allowing those with limited farming skills to take up non-agricultural employment or to migrate temporarily without losing their links to rural areas, and those with comparative advantage in agriculture to increase the size of the land they farm. Land sales, on the other hand, may allow households who want to move into the non-agricultural economy to mobilize the capital that will help them to exploit profitable economic opportunities. Fig. 1. Parcels on a Rwandan hill Programs aiming to provide higher level of tenure security and better land information systems have limited local economic opportunities for tenants Government, soon after peace was established, often been justified by noting their impact on lowering who were kept in semi-feudal relationships. embarked on far-reaching legal and institutional transaction costs in land markets. Registered land changes. Following these changes, a program of THE RWANDAN CONTEXT rights make it easier to identify rightful land owners, land tenure regularization (LTR) was launched in negotiate and enforce contracts, and reduce the risk of 2010. Almost 11 million parcels were demarcated With 416 inhabitants per km2 (NISR, 2012), land owners not being able to recover land that they in a participatory, low cost (5 USD per parcel) Rwanda is one of the most densely populated had rented out. This could in turn increase the number process, making the Rwandan LTR program an countries in the world and pressure on land has of efficiency-enhancing land market transactions; exemplary case of land tenure regularization long been identified by the Government as a facilitate credit access via the use of land as collateral; across Africa. The exercise was complemented by challenge to development (EDPRS1, EDPRS2). In and foster structural change and transformation. At a rigorous impact evaluation that used baseline particular, unfettered operation of land markets in the same time, in settings where market imperfections and follow-up surveys aimed at exploring early the 1980s was viewed as one cause of conflict and are prevalent, reducing transaction costs in one market effects of land tenure regularization on, among even considered as a possible contributing factor to will not necessarily improve outcomes across the others, transaction costs of land market the 1994 Genocide by some scholars (Andre and board. For example, historically, distress sales due to participation and the potential for these to foster Plateau, 1998). credit market imperfections have allowed diversification and structural transformation. moneylenders to amass vast land holdings that often To secure rights to Rwandans’ most important asset and deal with other land related issues, the DATA We use data from a two-round panel survey of 3,600 rural households in 300 randomly selected villages collected for the impact evaluation of the LTR program. Before the LTR roll-out, 100 sectors nationwide (4 in each of the country’s 25 districts) were randomly chosen and then half of them were randomly assigned to be among the first program areas and the other half to be among the last.1 The baseline was collected in early 2011 before the project started in the selected villages. The second round of interviews was conducted one year later, after the implementation of the LTR in the ‘early’ program areas and before the start of the implementation of the LTR in the ‘late’ program areas, thus enabling us to observe the short term impacts of the program (in this case, Fig. 2. Parcel demarcation demarcation and adjudication). In particular, the panel data allow us to describe how rural land period, the figures show that 31 percent of the more educated than landlords and those households that sales and rental markets function, identify sampled households leased- in, 12 percent are remaining in autarky; and are more likely to engage similarities and differences between them, and leased-out, 9.5 percent bought land and 4 percent in off-farm wage employment and non-farm household assess to what extent they have contributed to sold land. based enterprises. Interestingly, both tenants and changes in agricultural production and land use. landlords are less credit constrained in the semi-formal Rental markets seem to help equalize land sector2 than households that did not participate in land endowments, increasing the average tenant’s BOTH LAND RENTAL AND SALES rental market transactions. Once participation in land cultivated area by 0.16 ha (from 0.44 ha of MARKETS ARE VERY ACTIVE markets is accounted for, output per ha is not found to owned land). Participation in rental seems to be statistically significantly different across landlord- allow younger household heads with relatively Descriptive statistics reported in Table 1 cum-cultivator, tenant and autarkic households. more dependent children to access cultivable illustrate that participation in both land rental land. Tenants also appear to be significantly and sales market is high. Using a 12 month recall Table 1: Selected descriptive statistics by status of current season land rental market participation Rental Market Sales Market Only market transactions Total Tenant Autarky Landlord Buyer Autarky Seller Land endowment and participation in sales and rental markets Owned land in hectares 0.64 0.44 *** 0.64 1.18 *** 0.8 *** 0.6 0.73 Operated land in hectares 0.71 0.65 * 0.71 0.85 *** 0.86 *** 0.69 0.71 Net purchase of land over the past year in ha. 0.01 0.14 *** 0 -0.1 *** Net land leased-in only through market transactions 0.02 0.16 *** 0 -0.28 *** 0.01 0.02 -0.04 *** Basic household characteristics Female headed household 0.3 0.21 *** 0.33 0.37 ** 0.11 *** 0.32 0.31 Age of household head in years 47 41 *** 49 50 ** 39 *** 48 47 Head has been displaced 0.47 0.47 0.46 0.52 *** Head has primary school education 0.57 0.65 *** 0.54 0.54 0.66 *** 0.57 0.57 Head has secondary school education 0.07 0.09 0.06 0.08 *** 0.09 *** 0.06 0.08 Number of dependent 2.35 2.44 *** 2.3 2.32 2.62 *** 2.31 2.43 Assets and participation in off-farm activities Sale value of household assets in USD 86 82 80 126 *** 122 *** 82 88 Sale value of livestock in USD 158 146 155 201 *** 226 *** 150 163 Constrained in the semi-formal credit market 0.72 0.69 *** 0.75 0.68 *** 0.58 *** 0.73 0.74 Involved in off-farm wage employment 0.42 0.48 *** 0.39 0.39 0.36 *** 0.42 0.4 Has non-farm enterprises 0.24 0.28 *** 0.21 0.25 ** 0.34 *** 0.22 0.25 Output per ha cultivated (USD/ha) 559 560 542 641 573 564 428 Farming ability -0.00 0.14 *** -0.02 0.00 Community characteristics LTR village dummy 0.5 0.54 *** 0.48 0.48 0.53 0.49 0.5 Number of observations 6980 2132 4016 832 663 6035 282 Note: Asterisks indicate the difference in significance between means for tenant and landlord groups as compared to those in autarky with *, **, and *** denoting significance at 10%; 5%; and 1%. DANIEL ALI, KLAUS DEININGER, MARKUS GOLDSTEIN, ELIANA LA FERRARA AND MARGUERITE DUPONCHEL household members per ha of operated land is significantly lower for tenants (23 household members or 12 adults) as compared to landlords (30 and 16) or those in autarky (30 and 14). In Table 3 we compute factor ratios for households depending of their status of participation in the land sales market. We find that buyers have more total or adult household members than sellers (31 and 15 per ha of owned land, as compared to 18 and 9 for sellers) but these ratios are much lower than those in autarky. For those able to participate, sales markets equalize labor-land ratios, but this amplifies the contrast with non-participants. Land rental markets do seem to attenuate some of this effect, and further reduces the labor-land ratio for buyers below that for sellers. To identify determinants of participation on either side of land rental or sales markets as well as the extent to which the LTR program affected transaction costs in these markets in the short-term, we use a standard friction model Fig. 3. Couple having just collected their land titles (Skoufias 1995). Results for both land rental and land sales market participation are given in table 4. LAND RENTAL MARKET DETERMINANTS Areas transferred in the sales markets were relatively smaller than those in the rental markets (0.14 ha bought and 0.10 ha sold vs. 0.16 ha leased-in and 0.28 ha leased-out), Columns 1 and 2 of table 4 show that lease markets are and parties involved complemented transactions by leasing-in or –out land. With an characterized by an endowment equalization effect whereby average of 0.8 ha of owned and 0.86 ha of operated land, buyers are more land abundant the coefficient on the amount of land owned is significant than those in autarky, suggesting that the existing pattern of land sales in the market may and positive for leasing out and negative for leased in, lead to concentration of land ownership. On average, buyers are less likely to be female implying that larger land owners are more likely to lease out headed households than those remaining in autarky. The household heads among land and smaller ones more likely to lease in. However, the extent buyers are also younger, more educated and have larger families (in particular they have of adjustment is far from what would be required to achieve more female adults) than those in autarky. The sale values of their owned household assets full adjustment to the desired level of cultivated area (i.e., and livestock are higher and these buyers are, consistent with expectation, less constrained we reject that the absolute value of the coefficient of land in the semi-formal credit markets. In addition, buyers are also more involved in operating owned equals one). Land rental provides access to younger non-farm enterprises than both sellers and households who did not participate in the land producers as suggested by the fact that the head’s age is sales markets. While no difference in agricultural productivity is observed among buyers, positive for net land leased out and negative for net land sellers and households that do not participate in the land sales market, buyers tend to have leased in. Interestingly, the same pattern is observed for the significantly higher levels of farm abilities.3 value of household assets, implying that Rwanda’s land rental markets effectively transfer land from relatively rich LAND/LABOR RATIOS to relatively poor households. Positive signs on the Comparing land-labor ratios for households before and after participation in the land coefficients for number of male adults and dependents (both rental markets (table 2) and in the land sales markets (table 3) allows us to test the input at 10% significance level) and dummies for having a non- ratio equalization hypothesis. We note that, with 47 household members (or 24 adults) per farm enterprise and being credit unconstrained suggest that ha of owned land, those leasing in land have higher labor-land ratios than those who act rental markets transfer land to those with higher as landlords (with 15 and 8) or remain in autarky (with 36 and 17). Market participation endowments of family labor, do not impede diversification allows for the reversal of this; once relevant transfers are accounted for, the number of Table 2: Labor-Land Ratio before and after Land Rental Market Participation Market transactions only Total Tenants Autarky Landlord Before participation in land rental market Household size per hectare of owned land 36.78 46.51 *** 36.31 15.09 *** Number of adults per hectare of owned land 18.20 23.69 *** 17.48 8.18 *** Number of observations 6815 2045 3938 832 After participation in land rental market Household size per hectare of operated land 27.63 23.04 *** 29.64 29.73 Number of adults per hectare of operated land 13.82 11.70 *** 14.43 16.38 Number of observations 6950 2132 4016 802 Note: Asterisks indicate the difference in significance between means for tenant and landlord groups as compared to those in autarky with *, **, and *** denoting significance at 10%; 5%; and 1%. 4 4 Table 3: Labor-Land Ratio before and after Land Rental and Sales Markets Participation participation allowed households to expand Market transactions only income sources and join the non-agricultural Total Buyer Autarky Seller economy. Second, although female-headed If not participated in land sales market Household size per hectare of owned land 36.92 30.87 ** 38.51 17.53 *** households are less likely to participate in Number of adults per hectare of owned land 18.34 15.30 * 19.13 8.93 *** rental markets than male headed ones, welfare Before participation in land rental market effects need not be negative as leasing out Household size per hectare of owned land 36.78 26.21 *** 38.51 25.45 ** provides them with more resources than they Number of adults per hectare of owned land 18.20 12.20 *** 19.13 13.10 ** After participation in land rental market could get from own cultivation. Investigation Household size per hectare of operated land 27.63 15.75 *** 29.23 21.29 ** of this as well as the way in which plots to be Number of adults per hectare of operated land 13.82 8.04 *** 14.58 10.98 * leased out are chosen and resources received Number of households 6980 663 6035 282 Asterisks indicate the difference in significance between means for buyer and seller groups as from doing so are distributed within the compared to those in autarky with *, **, and *** denoting significance at 10%; 5%; and 1% household will be useful. Regarding land sales, a particular area of interest will be to explore the extent to which other measures that of the rural non-farm economy, but are more affected the transaction cost of land sales, in LAND MARKETS PARTICIPATION difficult to use for credit constrained households. particular changes in the consent clause, Female headship is estimated to increase net land affected land market participation. This would From the perspective of the Rwanda’s land leased out and decrease net land leased in. We allow us to assess if such a restriction had regularization program, the negative and discuss below the extent to which this may be a affected the incidence of certain types of land significant coefficient on the interaction between source of concern. Results suggest that there is no transfers as a first step to appreciate potential dummies for the LTR program and the follow up strong impact of farming ability (proxied by impacts on other outcome variables such as the survey (round 2) in the estimates for leasing -in household fixed effects from a panel production protection of women’s rights or the number of suggests that LTR significantly reduced the function regression) on net leasing of land while conflicts. threshold for land rental market participation having a non-farm enterprise is estimated to (Table 4 – column 2). This is in line with the 1 Areas where the regularization program had already increase rather than reduce the likelihood of notion that LTR-induced clarification of land started by the time of data collection were dropped leasing in land. ownership reduced transaction cost-induced from the frame, implying that Kirehe district in friction in land rental markets. Eastern Province and Rubavu district in Western LAND SALES MARKET DETERMINANTS Province, as well as Kigali city were excluded. In the case of land sales, threshold equations point 2 Details of the construction of the "credit towards a significant widening of the autarky In contrast, in column 3 of table 4 the magnitude constrained" variable in the semi-formal sector can be band on the purchasing side with time that is only of the estimated coefficient on land owned in the found in Ali et al. (2014). partly compensated by a LTR-induced 3 We use the standard method of measuring farming net seller equation is much smaller than in the transaction cost reduction. At the village level, abilities using household fixed effects from a Cobb- land rental market equation and the coefficient in having more land with a record that Douglas production function regression. Details are the equation for net buying (column 4) is unambiguously identifies the owner will reduce available upon request from the authors. insignificant. This suggests that the contribution the transaction cost of land purchases by of land sales markets to equalization of eliminating the need for a costly investigation endowments is marginal. Similarly to what was into whether or not the seller is the true owner, observed for land rental, being credit constrained thus making it easier to buy land. At the same reduces land sales market participation. While time, the regression suggests that this variable coefficients on relevant variables suggest that net will be of little relevance for the costs borne by land buyers are younger, less likely to be female land sellers (column 3 of table 4), a finding that headed, and have more male adults and may be partly related to the exclusion from our dependents, all the estimated coefficients are sample of households who sold land and no rather small in magnitude compared to the longer engage in agriculture. estimated coefficients on assets, both household and livestock, suggesting that such markets transfer land to relatively affluent households and CONCLUSION contribute less to poverty reduction (Deininger 2003). Finally, being engaged in off-farm wage While our results suggest that clarifying land employment is negatively correlated with net rights had a positive impact on functioning of land purchased. Interestingly, there is a positive land rental as well as sales markets even in the correlation with farming ability that suggests that very short term, they leave open a number of land is transferred to the households with more questions that would be worth exploring. First, it agricultural skills. would be useful to better uncover the underlying mechanisms, in particular the extent to which IMPACTS OF SECURING LAND RIGHTS ON clarification of land tenure and land market DANIEL ALI, KLAUS DEININGER, MARKUS GOLDSTEIN, ELIANA LA FERRARA AND MARGUERITE DUPONCHEL 5 Table 4: Determinants of Net Land Leased- and of Net Land Purchases Leased-out Leased-in Net-Seller Net-Buyer Owned land in hectares 0.154*** -0.109** 0.018*** -0.002 (2.729) (-2.404) (2.748) (-0.191) Head has been displaced 0.028 -0.043 0.033 0.009 (0.531) (-1.624) (1.259) (0.488) Female headed household 0.090** -0.097*** 0.011 -0.151*** (2.475) (-2.955) (0.461) (-5.061) Age of household head in years 0.004*** -0.008*** -0.000 -0.006*** (2.946) (-5.859) (-0.098) (-6.373) Head has primary school education 0.016 0.062** 0.007 -0.001 (0.410) (2.200) (0.317) (-0.049) Head has secondary school education 0.101 0.053 0.048 -0.003 (1.502) (0.723) (1.221) (-0.106) Number of dependents -0.013 0.013* 0.004 0.015** (-1.057) (1.675) (0.535) (2.198) Number of male adults -0.004 0.030* 0.003 0.023** (-0.211) (1.929) (0.275) (2.566) Number of female adults -0.015 0.023 0.015 0.006 (-0.643) (1.320) (1.256) (0.471) Sale value of household assets in USD1000 0.286** -0.128* -0.095 0.128** (2.344) (-1.847) (-1.562) (2.362) Sale value of livestock in USD1000 0.008 0.045 0.002 0.100*** (0.070) (0.769) (0.046) (2.805) Constrained in the semi-formal credit market -0.040 -0.061*** 0.003 -0.060*** (-1.006) (-2.743) (0.124) (-3.738) Involved in off-farm wage employment 0.044 0.026 -0.022 -0.095*** (1.047) (1.233) (-1.126) (-4.327) Has non-farm enterprises -0.005 0.113*** 0.019 0.023 (-0.124) (3.335) (0.710) (1.159) Farming ability 0.036 0.003 -0.000 0.031** (1.205) (0.110) (-0.023) (2.269) Threshold effect LTR village X time dummy 0.037 -0.067** -0.028 -0.057** (0.891) (-2.156) (-1.062) (-2.055) Time dummy 0.001 -0.066*** 0.013 0.098*** (0.027) (-2.606) (0.577) (4.033) Sigma 0.630*** 0.347*** (6.346) (8.339) Number of observations 6,657 6,657 Log-Likelihood -5,522.96 -2,137.97 Wald chi2 test 44.666 16.493 Note:. Robust z-statistics adjusted for clustering at the village level are in parenthesis: *** significant at 1%; ** significant at 5%; * significant at 1%. Coefficients for leasing out (except threshold effects and constant parameters) are multiplied by -1 for ease of interpretation. Results include a range of constant parameters at the village level that are not reported for presentation purposes. REFERENCES Ali, D. A., K. Deininger and M. Duponchel. 2014. Deininger, K. W. (2003). “Land policies for "Credit Constraints and Agricultural Productivity: growth and poverty reduction”. World Bank Evidence from Rural Rwanda." The Journal of Publications. Development Studies, 50(5), 649-665. Government of Rwanda (2007). “Economic André C &Plateau JP, 1998. “Land Relations Development and Poverty Reduction Strategy, Under Unbearable Stress: Rwanda Caught in the 2008-2012”. Malthusian Trap.” Journal of Economic Behavior and Organization, 34, 1-47 Government of Rwanda (2012). “Economic Development and Poverty Reduction Strategy 2, Badiane, O., J. Ulimwengu and T. Badibanga. 2013-2018”. 2012. "Structural Transformation among African Economies: Patterns and Performance." Skoufias, E. (1995). “Household resources, Development, 55(4), 463-476. transaction costs, and adjustment through land tenancy.” Land Economics, 42-56. This case study was prepared by a team from the Development Economics Research Group and Africa Gender Innovation Lab of the World Bank in collaboration with the Rwanda Natural Resources Authority. The UK Department for International Development, the Bank Netherlands Partnership Program and the International Growth Centre provided generous financial support. Photo credit: Marguerite Duponchel The findings, interpretations, and conclusions expressed are entirely those of the authors. They do not necessarily represent the views of the Government of Rwanda, and of the World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.