Three Essays on Evaluating Forest Conservation Programs in Developing Countries

Deforestation and forest degradation in developing countries are leading causes of environmental problems such as soil erosion, biodiversity loss, and climate change. As a result, policies aimed at slowing down or reversing the trend of deforestation and forest degradation have attracted considerable attention. This dissertation consists of three essays on evaluating forest conservation programs in developing countries. Although the focus of each essay differs, they all use rigorous econometric methods to provide insights on impacts of historical forest conservation programs and assist stakeholders in modifying existing policies and making future ones more efficient and effective.

I begin by assessing the nutritional impact of payments for ecosystem services (PES) in the context of rural China (i.e., Chapter 2). PES is a special type of conditional cash transfer (CCT) in which the conditionality is explicitly attached with conservation practices. In this chapter I develop a stylized household-farm model to show that when households participate in a land-diversion PES program, they would settle for lower levels of food consumption if they lack market access. Exploiting panel data from the China Health and Nutrition Survey (CHNS), I use a triple difference (TD) model to examine the impact of China’s Sloping Land Conversion Program (SLCP), one of the largest PES programs in the world, on the nutrient intake of farming households. My findings suggest that the SLCP had a significant negative impact, though small, on calorie intake and this effect was likely driven by missing market in areas that implemented the SLCP. This essay demonstrates that land-diversion PES, which is a dual conservation and development tool, could affect food consumption and nutrition in ways very different from other conservation programs such as protected areas (PAs) as well as regular CCT programs that only aim for poverty reduction.

I then shift the focus from PES to PAs by implementing innovative evaluation methods to assess the effectiveness of mangrove protection in Southeast and South Asia (i.e., Chapter 3 and Chapter 4). Economists typically estimate the average treatment effect on the treated (ATT) when evaluating government programs. The welfare interpretation of the ATT can be ambiguous when program outcomes are measured in purely physical terms, as they often are in evaluations of environmental programs (e.g., avoided deforestation). In Chapter 3, my co-authors and I present an approach for inferring welfare impacts from physical outcomes when the ATT is estimated using propensity-score matching. We employ the discrete-choice Roy model of selection into treatment to show that the ex post net social value of a forest conservation program can be proxied as a weighted ATT, with the weights being utility measures derived from the propensity of being treated. We apply this new metric to mangrove forest conservation in Thailand during 1987–2000. Wefind that the Thai government’s conservation program protected approximately 30% of the social welfare that would have been lost if all the protected mangrove area had been deforested. This magnitude is very similar to the magnitude of a conventional ATT that measures avoided deforestation, but we show that the potential range of the welfare-based ATT extends from barely a quarter of the conventional physical measure to nearly twice as large as it.

While Chapter 3 adopts an indirect approach to infer the welfare impact of PAs, Chapter 4 exploits the same idea in a direct approach. In Chapter 4, my co-authors and I exploit rich data on carbon stock and land values in India toestimate and predict spatial heterogeneity in the benefit (i.e., carbon sequestration) and cost (i.e., forgone land value) of mangrove conservation. We combine this information with satellite-based data on India’s mangrove coverage in 1990– 2010 to construct a net land value, and then estimate the causal impact of PAs on the net land value. This new approach allows us to account for spatial heterogeneity in the net economic benefit of conservation. Our results show that incorporating the economics of conservation into evaluation could detect impact of PAs that would not be detected under the conventional approach that focuses only on avoided deforestation. Estimates from our heterogeneity treatment effect model suggest that the level and direction of PA’s impact is associated with the road proximity of mangrove sites and differs between the short run and the long run.

The three essays in my dissertation examine the heterogeneity in effects of forest conservation programs in one way or another. They highlight that the efficiency and effectiveness of conservation programs depend on local contexts. When designing and implementing future conservation programs, policymakers should assess local contexts and adjust program features accordingly.