74812 CASE STUDY WORLD BANK | AGRICULTURE AND DEVELOPMENT DECEMBER 19, 2012 Going Digital: Credit Effects of Land Registry Computerization in India KLAUS DEININGER AND APARAJITA GOYAL ABSTRACT —————–—————————————————————————————— Digitization of land records represents a unique way to test for credit supply effects of improved land administration information over time. We exploit variation in the timing of the shift from manual to digital operation of Andhra Pradesh’s 387 sub registry offices during the state-wide rollout of this intervention from 1999 to 2005. Data on credit disbursement and registered mortgages for the 1995-2007 period point to significant, though quantitatively modest, increases in credit access in urban areas. Institutional factors allow us to explain these results. BACKGROUND digitized all encumbrance certificates (ECs)1 rural ones where land records are used as an from 1983. This occurred in three rounds, alternative. Development economists have long starting from offices with the highest highlighted the central role of institutions transaction volumes, and finally covering all of DESIGN and how they impact growth and distribution the state’s 387 sub registry offices (SROs). As of gains among populations (Greif 1993, a result, ECs, as well as market valuations, In order to determine whether the North 1971). Since land is an important deed extracts, and other key documents computerization of land registry affected credit household asset in virtually all countries, became available online. It became possible access over time in Andhra Pradesh, we property institutions can have a significant for banks to ascertain property ownership conduct rigorous empirical analysis using data impact on economic outcomes. Secure status or pre-existing liens that could be from three sources. First, quarterly data from property rights can reduce expropriation offered as collateral. We hypothesized that 1995 to 2007 on credit disbursed by all risk, thereby increasing investment this would reduce the bank cost of extending scheduled commercial banks to retail incentives and reducing the need for credit and facilitate an expansion of credit customers in all of the state’s 1,064 taluks is individuals to spend resources on protecting supply, either by lending to those who had collected from the Reserve Bank of India their rights. They can also facilitate market previously been ineligible, or by lending more (RBI)2. Second, we use the state’s transactions by allowing land assets to be to existing customers. Department of Stamps and Registration’s traded and used as collateral in financial annual data on different types of registered markets. Institutional factors that are characteristic to land transactions from 1995 to 2007 for all the setting, however, must be taken into 387 SROs. Third, in light of the differences Although many empirical studies show that account. India’s land administration system between rural and urban land administration, secure property rights can have significant differentiates between land records and land we use the 2001 Census for the initial share of investment effects, there is little evidence registration. In rural areas, land records urban population in any taluk or SRO. on their credit effects. To partly fill this gap, maintained by village level officials are more we examine the credit effects of land accessible and up to date than land registries, Since CARD was implemented in three registry digitization in Andhra Pradesh, a which only operate at the taluk (block) level. In rounds, we exploit the variation in timing southern Indian state. From 1999 to 2005, contrast, land registries are the only available across SROs to single out the impact of Andhra Pradesh implemented a program for form of documentary land ownership evidence computerization on credit supply and different Computer-Assisted Registration of Deeds in urban areas. Thus we expect the types of registered land transactions. If (CARD), which streamlined procedures, computerization of land registries to have a c om p ut e r i z at i o n d o es i nd ee d h a v e a provided automatic property valuation, and more pronounced effect in urban areas than in significant credit effect, we expect to see its WWW.WORLDBANK.ORG/ARD 2 CREDIT EFFECTS OF LAND REGISTRY COMPUTERIZATION IN INDIA clear impact on mortgages but not on other Table 1: Evolution of retail credit and transaction volume in Andhra Pradesh, 1995-2007 types of land transactions. Panel A Panel B Non-Monetary Before After Year Credit Sales Mortgages All DESCRIPTIVE EVIDENCE Transactions Comp. Comp. 1995 525 0.48 0.19 0.11 Total Credit 0.24 0.10 0.37 Panel A in Table 1 shows that credit markets 1996 579 0.51 0.21 0.12 Sales 1704 1115 2302 in Andhra Pradesh evolved rapidly during the 1997 627 0.52 0.23 0.12 Mortgages 558 552 563 period in question. The volume of credit almost Non-Mon. 1998 667 0.54 0.31 0.13 488 258 722 quintupled in real terms, from Rs. 525 billion in Transactions 1999 736 0.55 0.29 0.13 Rural Credit 0.06 0.04 0.09 1995 to Rs. 2,503 billion in 2007. Prima facie, 2000 817 0.60 0.27 0.15 Sales 1491 1014 2003 the rate of growth seems to have 2001 874 0.61 0.26 0.16 Mortgages 548 545 551 accelerated after 2003, at a time when most Non-Mon. SROs had already been computerized. 2002 973 0.65 0.26 0.18 393 227 570 Transactions Registered sales, mortgages, and non- 2003 1094 0.72 0.27 0.20 Urban Credit 2.00 0.97 2.55 monetary transactions also increased notably. 2004 1432 0.84 0.28 0.26 Sales 4455 3171 5077 2005 1986 0.88 0.28 0.30 Mortgages 712 673 798 Panel B summarizes total credit amounts as Non-Mon. 2006 2050 0.93 0.29 0.34 1720 891 2122 well as the mean number of transactions for Transactions each land transaction type by SRO (overall, 2007 2503 1.09 0.33 0.36 rural, and urban). The total volume of credit Notes: Nominal values of credit are deflated using the RBI whole sale price index with 1993-94 disbursed by commercial banks differed as the base year; credit is reported in billions of 1993 Rupees. In Panel A, volume of land transactions (land sales, mortgages, non-monetary land transactions) are reported in millions. greatly between urban and rural areas, with Panel B reports average number of land transactions registered per SRO. the amount of urban credit (Rs. 2 billion) significantly higher than that of rural credit RESULTS computerization. Post Q 0-4 is an (Rs. 0.06 billion) on average. After indicator variable for the quarter in which computerization, credit is estimated to have Although our empirical analysis points computerization was introduced and the doubled in both rural and urban areas; from towards an insignificant impact of first four quarters thereafter. Similarly, approximately Rs. 1 billion to Rs. 2.55 billion in computerization on aggregate credit Post Q 5-8 is an indicator variable for the urban settings and from Rs. 0.04 billion to supply, we find that it does have a fifth to eighth quarters (representing the Rs. 0.09 billion in rural settings. It is also significant impact in areas that are at least second year) after computerization, and evident that the number of registered land partly urbanized. Computerization in Post 9+ is an indicator variable for the transactions (for all types) is much higher in urban areas considerably increased the ninth quarter and beyond. In column 3, urban areas than rural areas. Focusing overall number of registered mortgages, there appear to be no evidence for lagged specifically on mortgages, there is a significant but had no effect on the number of other effects of computerization in the difference in their numbers for rural areas and types of registered land transactions aggregate. In contrast, column 4 suggests urban areas (548 vs. 712). It must also be (i.e. land sales and non-monetary that computerization’s impact on credit noted that the increase in registered transactions). This is consistent with the disbursement increases with the degree mortgages after computerization is confined to idea that reduced bank transaction costs of urbanization; some delay in effect is the urban setting (673 to 798). are the underlying driver to increasing also evident, with increases of credit supply. approximately 3 percent to 5 percent in the first eight quarters, followed by a Results in Table 2 point towards 15 percent increase thereafter. This may computerization’s robust effect on credit be due to the time it takes for banks to supply that increases over time in fully or learn about the information provided by partly urbanized areas. While a naïve computerized registries and to adjust their estimate (column 1) suggests a large processes in a way that can take effect, inclusion of proper controls in advantage of this information. column 2 implies that computerization had no effect on overall credit supply. Yet, Examining computerization’s effect on the according to the point estimate, in volume of registered mortgages provides completely urban SROs computerization an additional robustness check to support did increase credit supply by 10.5%. the plausibility of credit supply effects postulated above. Table 3 provides the To examine whether and how the effect evidence. Although computerization does varies over time, we include variables that not appear to have a significant effect on indicate time periods immediately the volume of mortgages overall, in urban preceding or following computerization in areas the number of mortgages is columns 3 and 4. Pre Q 1-5 is an indicator estimated to have increased by roughly Figure 1: Accessing records via a computer variable for five quarters leading up to the 31 percent. Interaction with post-variables 3 CREDIT EFFECTS OF LAND REGISTRY COMPUTERIZATION IN INDIA Table 2: Effect on credit access Table 3: Effect on the volume of registered land transactions Dependent variable: log of credit disbursed by banks Mortgages (1) (2) (3) (4) (1) (2) Computerization 0.854*** 0.0163 Computerization 0.13 [0.0109] [0.0147] [0.0794] Computerization * Urban Share (US) 0.105** Comp. * Urban Share (US) 0.311** [0.0495] [0.169] Comp. * Pre Q 1-4 0.0149 0.0202 Comp. * Pre Y 1-4 0.0375 [0.0132] [0.0135] Comp. * Post Q 0-4 0.0155 0.0193 [0.0878] [0.0123] [0.0238] Comp. * Post Y 0-2 0.1801 Comp. * Post Q 5-8 0.0232 0.0344 [0.1069] [0.0199] [0.0215] Comp. * Post Y 3+ 0.2123 Comp. * Post Q 9+ 0.0289 0.0260 [0.0298] [0.0362] [0.1647] Comp. * US * Pre Y 1-4 0.134 Comp. * US * Pre Q 1-4 0.0112 [0.0313] [0.137] Comp. * US * Post Q 0-4 0.0313** Comp. * US * Post Y 0-2 0.177** [0.0165] [0.0751] Comp. * US * Post Q 5-8 0.0513** Comp. * US * Post Y 3+ 0.325** [0.022] [0.182] Comp. * US * Post Q 9+ 0.147** Observations 5,031 5,031 [0.0599] Quarter*Year Fixed Effects No Yes Yes Yes R-squared 0.859 0.859 Taluk Fixed Effects No Yes Yes Yes Notes: All regressions include year and SRO fixed effects. Observations 55,328 55,328 55,328 55,328 Robust standard errors clustered at the SRO level. R-squared 0.18 0.96 0.96 0.96 All dependent variables are in logs. Notes: Robust standard errors clustered at taluk level. *significant at 10%, ** significant at 5%, *** significant at 1% * significant at 10%; ** significant at 5%; *** significant at 1% 1 suggests that computerization’s impact on i n t e rp ret at i on i s c orrec t, c om bi ni ng Encumbrance certificates (ECs) are the volume of mortgages increases with computerization with efforts to improve the abstracts that list all registered transactions time, from approximately 18 percent in the quality and relevance of the underlying by person or for a specific parcel of land. first two years to 33 percent thereafter. information could potentially result in 2 additional benefits. For example, records Scheduled commercial banks include all CONCLUSION and registries could be synchronized, or a public and private sector banks with the survey could be added to unambiguously exception of cooperative banks which Our empirical analysis shows that locate the property in question and contributed less than 10 percent to total interventions which reduce the cost of determine its boundaries. This may improve lending from 1997 to 2006 (RBI 2008). updating registries and make information not only credit access, but also land more available to lenders (i.e. governance and conflict resolution. REFERENCES computerization) can increase credit access. Exploring other dimensions of land However, factors such as limited record administration and their impact is an Greif, A. 1993. "Contract Enforceability and coverage, problems in the underlying important area for future research. Economic Institutions in Early Trade: the information, and structural characteristics of Maghribi Traders' Coalition." American the land administration system can limit the Economic Review 83 (3): 525-48. scope of this effect. Thus, when assessing * * * the quality of a country’s land information North, D. C. 1971. Structure and Change in system, other meaningful metrics will likely Economic History. New York: W.W. Norton. need to be considered in addition to the cost This case study was prepared by Klaus Deininger from the Development Economics of registering transactions. Our results also Research Group and Aparajita Goyal from the Economics and Policy Group of the Agriculture imply that it is unrealistic to expect an and Environmental Services Department of the World Bank, in collaboration with the Reserve immediate, universal increase in credit Bank of India, the Andhra Pradesh State Government’s Department of Stamps and Duties, access when land administration is and the Commissioner Survey and Settlement. Global Land Tools Network provided improved. generous financial support. The impact of computerization is estimated The findings, interpretations, and conclusions expressed are entirely those of the authors. They do to be of rather modest magnitude, even in not necessarily represent the views of the Government of India, and of the World Bank and its fully urbanized areas. This may be due to the affiliated organizations, or those of the Executive Directors of the World Bank or the governments limitations of registry information. If this they represent.