| ||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © The Author 2007. Published by Oxford University Press.
This article appears in the following Oxford Review of Economic Policy issue: INDIA [View the issue table of contents]
Land reform and agricultural productivity in India: a review of the evidence
* London School of Economics, e-mail: m.ghatak{at}lse.ac.uk
** London School of Economics, e-mail: s.roy2{at}lse.ac.uk
 |
Abstract |
|---|
In this paper we review as well as contribute to the empirical literature on the impact of land reform on agricultural productivity in India. We find that, overall for all states, land-reform legislation had a negative and significant effect on agricultural productivity. However, this hides considerable variation across types of land reform, as well as variation across states. Decomposing by type of land reform, the main driver for this negative effect seems to be land-ceiling legislation. In contrast, the effect of tenancy reform, averaged across all states, is insignificant. There seems to be a wide range of state-specific effects, which suggests that focusing on average treatment effects can hide a considerable amount of heterogeneity. In particular, allowing a separate slope for West Bengal, one of the few states that implemented tenancy laws rigorously, we find that land reform had a marginal positive effect relative to the rest of India.
Key Words: agricultural productivity • land reform • West Bengal
We would like to thank Robin Burgess for access to the data set relating to the study of land reforms in India in Besley and Burgess (2000). We would also like to thank Oriana Bandiera, Erlend Berg, Tim Besley, Robin Burgess, Dave Donaldson, an anonymous referee, and the editors for very helpful comments.
1 See Joshi (1975) for a discussion of land-reform legislation in India and its implementation. See also Besley and Burgess (2000), who provide a systematic description of these laws and their amendments that were passed in individual states over time.
2 Often, such eviction was euphemistically referred to as ‘voluntary surrender’, although in most cases they were anything but voluntary.
3 Here, land resources are calculated as the difference between geographical land and forest land.
4 This ignores the issue of whether the empirical observations mentioned above control for unobserved heterogeneity of tenants and soil quality. If these observations are purely driven by unobserved variations in farmer quality or land quality, then the efficiency case for land reform is weakened. For example, if higher-ability farmers prefer to cultivate smaller plots of land (as opposed to being constrained to do so), or landowners prefer to lease out lower-quality plots to sharecroppers, then, in the absence of any other frictions such as those mentioned above, land reform will not raise average productivity. However, there is some evidence (e.g. Rosenzweig and Binswanger (1993) on farm size and productivity, and Shaban (1987) on sharecropping) to suggest that these empirical facts are not purely driven by heterogeneity in farmer quality or land quality. For example, Shaban (1987) finds, after controlling for land quality, that the same farmer puts in less effort in plots of land that he cultivates as a sharecropper compared to plots of land that he cultivates as an owner-cultivator.
5 It measures the number of legislative reforms concerning land in any particular state, calculated cumulatively over time. There are four key land-related laws: redistributive land reform, tenancy reform, abolition of intermediaries, and land-ceiling legislation. The 4-year lag is based on the reasonable assumption that the effect of legislative reform on productivity or poverty will not be instantaneous, owing to implementation and adjustment lags.
6 The original dataset (available at http://chd.ucla.edu/dev_data/index.html#hhsurveys) covered the years 1971–87, and was later updated to 1999 by Duflo and Pande (2007) for their paper. We use the updated version. This dataset has also been used by Banerjee and Iyer (2005).
7 The states that are dropped include Assam, Jammu and Kashmir, and Kerala.
8 We also report in the last two columns of Table 1(a) the results obtained when we used data on rice yields from the MoA, Government of India, for the purpose of comparison with Banerjee et al. (2002): see section III(ii).
9 The reason the coefficients in column (3) are very close but not exactly identical to those reported in Besley and Burgess (2002, Table 4) seems due to some missing values in our version of the Besley–Burgess estimation.
10 We also ran the robustness checks for the West Bengal experience, using the same policy controls as in Table 1(b), and the results are similar.
11 We also ran the robustness checks in this case, using the same policy controls as in Table 1(b), and the results are similar.
12 See Bardhan et al. (2007) for a formal analysis of the latter effect.
13 For the relevant regression results, see Conning and Robinson (2007, p. 440, Table 4, column 3).
14 Haryana is dropped owing to missing observations.
15 Owing to the very small sample, we do not use the more standard method of interacting the reform dummies with the state dummies to capture the state-wise effects.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  V. Bhaskar and B. Gupta India's development in the era of growth Oxf. Rev. Econ. Policy, June 1, 2007; 23(2): 135 - 142. [Abstract] [Full Text] [PDF]
|
|