Panel Paper: Predicting Adoption of Clean Household-Level Cooking Technologies: Do Stated Preference Surveys Reflect Reality

Saturday, November 9, 2013 : 2:25 PM
Boardroom (Ritz Carlton)

*Names in bold indicate Presenter

Marc Jeuland, Jessica Lewis, Subhrendu Pattanayak and Jie-Sheng Tan Soo, Duke University
In addition to reducing damaging emissions that contribute to global climate change, the purported benefits of improved cook stoves (ICS) to poor households in developing countries include reduced damages to health, fuel cost savings, and the more diffuse benefits from the preservation of forested areas (Jeuland & Pattanayak, 2012). However, use of improved cooking technologies requires a complex set of adjustments by users, along several margins that affect net benefits: the time budget, given changes in cooking efficiency, as well as time required for fuel preparation; the pecuniary budget, given changes in stove and fuel cost; and the generation of household benefits from cooking (as affected by adjustments in both the intensive types, and extensive amounts, of food cooked). In fact, little is known about how user cooking preferences and local context actually relate to benefits and long-term behaviors related to clean stoves. This makes it difficult for stove promoters to know how to most successfully target ICS interventions to achieve the types of benefits described above.

This study explores preferences for ICS among rural households in India, and provides initial comparisons of these preferences with short-term adoption behavior. During a pre-intervention baseline study conducted June-August 2012, we interviewed about 2060 households living in 66 villages in two very different regions of India – in the plains of central Uttar Pradesh and in the mountainous state of Uttarakhand. During the baseline survey, we collected detailed information on stove ownership, cooking practices, and perceptions of traditional stoves, and we implemented a conjoint experiment to assess uses tradeoffs in four stove attributes (fuel requirement, smoke emissions, number of cooking surfaces, and price). We analyze these data using a series of mixed logit methods that vary in their ability to accommodate preference heterogeneity across different respondents. Collectively, these initial analyses reveal considerable heterogeneity in preferences and among many households, and a relatively strong default preference for traditional stoves, which together suggest that there may be significant demand barriers to achieving greater adoption of ICS.

Using latent class methods, we next generate a typology of different households, and assess the degree to which baseline (pre-intervention) characteristics of households and respondents are related to class membership. In addition, during the spring and summer of 2013, sample households will have the opportunity to purchase two different types of ICS during a marketing campaign targeting survey communities. Once collected, data on initial adoption and interest in the ICS will be compared with the typology derived from the baseline data, to determine whether: a) conjoint methods deliver useful predictive information on which households are likely to adopt; and b) whether the collection and analysis of such data provides information that is valuable for targeting of ICS interventions above and beyond that provided by more readily-observed household characteristics. Given the undeveloped nature of the market for ICS, such evidence could be useful to policy makers and program implementers.