Adoption and valuation of soil conservation measures: an analysis of the Teesta watershed programme

Soil erosion is a problem in Indian agriculture and is particularly acute in mountain areas, because the slopes are too unstable to allow the soil cover to evolve. Unchecked, continued soil erosion has adverse consequences for agricultural productivity. There are several soil conservation measures, some of which are large-scale, and others that can be implemented by farmers on their own farms, that can limit soil erosion, or even restore soil cover.

This provides the context for this dissertation, which has as its focus the Teesta River Watershed, in Darjeeling District in the Eastern Himalayas. In this watershed, there have been soil conservation interventions both by the government at the sub-watershed level (that we term treatment), and by individual farmers (that we term adoption). But there has been thus far little economic analysis of how successful these interventions have been in improving agricultural outcomes.

The dissertation has three major objectives: 1) to assess if farm profits and demand for farm labour are significantly different in treated sub-watersheds than in untreated sub- watersheds (Chapter 3), 2) to identify the determinants of adoption of on-farm soil conservation practices, with a special focus on whether there is a spatial dimension to adoption decisions (Chapter 4), and 3) to examine the causal impact of on-farm soil conservation measures on profit, value of agricultural output and cost of cultivation (Chapter 5). The data for this study was collected through a primary survey conducted during 2013 for the monsoon and winter crops.

To study the first objective, we estimate a quadratic flexible form profit function jointly with labour demand, using seemingly unrelated regression, and find that profits are higher in treated sub-watersheds in the winter season but not in the monsoon season. Also, labour demand appears to be no different between treated and untreated sub- watersheds.

The second objective is the subject of Chapter 4, which attempts to assess the key determinants of the decision to adopt soil conservation. The distinguishing feature of our analysis is that it explicitly accounts for possible neighbourhood effects in

influencing adoption. This is captured both by identifying adoption practices among farmers who are immediately upstream, and using spatial econometric techniques that incorporate the spatial distance between neighbouring farms. We use Bayesian formulation of a standard probit model in conjunction with Markov Chain Monte Carlo to estimate the model. The findings suggest strong and positive evidence of neighbourhood impact on farmers in making soil conservation decisions. We also examine if adoption decisions differ between farmers residing in treated and untreated sub-watershed and conclude that they do not.

The third objective is addressed in Chapter 5, which uses Propensity Score Matching to measure the impact of the adoption of soil conservation on farmer profits, revenues and variable costs. The comparison includes not merely adoption status, but also adoption intensity, to see if the adoption of multiple measures results in higher estimates of impact than the adoption of single measures of conservation. Also, unlike the conventional literature, we use the spatial probit model (estimated in Chapter 4) to determine propensity scores. Our results from binary adoption case suggest that there is no difference in the profits in both the winter and monsoon seasons. Although revenues from adoption are higher, these appear to be associated with higher variable costs, thus resulting in no difference in profits. Furthermore, joint adoption of contour, afforestation, bamboo plantation or at least two from these measures can provide a significant gain in revenues but also increase costs.

Overall, these results suggest that sub-watershed treatment has been successful in achieving only a limited set of objectives of Teesta River Valley Programme. While it is important to incorporate spatial features in modeling adoption decisions, it is likely that farms that have adopted these measures are able to maintain their soil quality at the same level as non-adopters; which perhaps explains why there is no difference in net profits, although revenues are higher. And although simultaneous adoption of some soil conservation measures may have some impact on reducing top soil loss, the increase in revenues appears to be offset by increased costs.