Zachary S. Brown

Abstract    This paper evaluates the potential for a positive externality associated with the prevention of an infectious disease.  Previous economic research has provided a theoretical foundation for the provision of Pigovian subsidies to incentivize private effort to reduce disease transmission.  I calculate optimal subsidies for participation in a malaria prevention program in northern Uganda, assuming a policymaker who seeks to maximize household welfare and who faces a static budget constraint dictated by available program funds.  Using results from a discrete choice experiment (DCE) I designed for this purpose, I first calculate the value that households place on malaria risk reduction via insecticide spraying programs in northern Uganda.  I then calibrate a malaria transmission model to published epidemiological surveys of malaria transmission for the region.  By coupling the disease model with the welfare and behavioral parameters estimated from the DCE, as well as estimates of IRS program costs, I am able to calculate optimal Pigovian subsidies for spray program participation for each of the 15 villages surveyed for this exercise.  Estimates from the DCE suggest that reducing monthly malaria risk by 1% is worth roughly 2% of (stated) household monthly income in northern Uganda.  After linking the DCE estimates with the malaria transmission model, I find that some form of subsidy worth between $3 to $10 per household per month, or around a quarter of the median household income in any given village, is an economically efficient mechanism for correcting the positive externality associated with malaria prevention via insecticide spraying.

Abstract    Biological theory and evidence have pointed to the existence of fitness tradeoffs associated with organisms’ evolution of resistance to many biocides (e.g. insecticides and antibiotics).  We use a simple model of malaria transmission, insecticide spraying impacts, and evolution of insecticide resistance coupled with an optimality criterion to develop intuition about how different types of fitness costs in malaria vectors impact economic insecticide policies for disease reduction.   We show that large economic losses from short- and long-term malaria burden due to qualitative differences in management strategies can be incurred through misinformed assumptions about sources of evolutionary fitness costs in vectors.

Abstract    

BACKGROUND The spraying of insecticides on the interior of homes, commonly referred as indoor residual spraying (IRS), is an important but costly strategy for reducing malaria in northern Uganda. This region has some of the highest malaria transmission rates in the world, implying large potential benefits from IRS. In the global and environmental health communities, IRS has been a controversial strategy because of increased insecticide exposure in homes, and diffusion in the environment. Little work, however, has investigated how households perceive how effective IRS is at reducing malaria.

METHODS In June through November of 2009, we organized and observed 6 focus group discussions and administered a structured questionnaire to 612 household representatives in Gulu and Oyam districts of Uganda to assess knowledge of malaria transmission and local perceptions of the spray programs.

RESULTS Overall, 90% of surveyed households correctly identified mosquitoes as the cause of malaria, however 30% incorrectly identified "cold weather" as a cause. Sixty-one percent of respondents identified "standing/still water," and 55% identified "uncleared bush/grass," 26% identified rainfall, and 12% identified temperature as major factors contributing to the presence of mosquitoes in their community. Less than 1.5% of respondents identified "farming methods," "insecticide spraying," or "sleeping under a mosquito net" as a factor affecting mosquito abundance. Overall, 80% (95% CI 74.3% - 82.9% in Gulu and 80% - 89.3%) of respondents thought that spraying their homes with insecticides reduces their risk of malaria. Those who previously participated in a government spray program thought that spraying was more effective than those who had not participated (P <0.012).

CONCLUSIONS Among Gulu and Oyam residents, IRS programs are highly desirable because of their perceived effectiveness at reducing malaria. Participation in the first round of IRS in these districts was quite high, concomitant with expectations that IRS will reduce malaria. Meeting these expectations with future rounds of IRS, while simultaneously scaling up ITN distribution, are the primary challenges for malaria control programs in northern Uganda. In addition, more education about the causes of malaria and contributors to mosquito abundance would be useful in reducing malaria burden in northern Uganda.

Abstract    This paper presents findings from a study which revisits a game formulated by Loomes (1991) to test Expected Utility Theory (EUT).  As with Loomes, we find evidence against EUT, namely the Independence Axiom (IA) which is a core part of that theory.  However, the significance of the test of EUT was around 10%, allowing us to only marginally reject EUT in this study.
We also illustrate in this paper how the observations from Loomes’ game can be used, in conjunction with other data from the survey, to analyze the determinants of risk aversion levels.  We find that gender, income, and the amount won in the first of two rounds of the game are significant determinants of how risk-averse a respondent’s allocation is in the game.  Some measures of education and wealth were found to be statistically significant determinants of respondents’ allocations, but these findings were not robust to changes in model specification. Future work with these data will explore the relationship between risk preferences as revealed through the game and risk preferences elicited through questions regarding malaria and agriculture.

Abstract    This paper reexamines the time consistency of discount functions as analyzed by Strotz (1956) and Koopmans (1960). Following on an observation by Heal (1998) regarding the time consistency of logarithmic discounting, I fully characterize the class of time consistent—but not necessarily autonomous—discount functions. I use my findings to explain an apparent “paradox,” pointed out by Gollier (2004), in the use of the gamma discounting method formulated by Weitzman (2001).