Summary/Abstract The poor suffer disproportionally from poor mental and physical health. Many causes for these disparities have been considered, including low income. But, poor families' incomes are not only low, but also often unstable and unpredictable. This instability creates uncertainty about whether they will be able to safeguard their future wellbeing. According to the allostatic load framework, prolonged activation of physiological stress responses will cause “wear and tear” on the body, heightening risks of cardiovascular disease and of age-related metabolic diseases, promoting cognitive decline and dementia, and accelerating cellular aging. This R01 will study the causal effects of income instability on the psychological and physical health of the poor. Our specific aims are to: 1) Identify the causal effect of income instability on psychological health (e.g. depression, anxiety), biomarkers of stress (e.g. cortisol), and physical health (e.g. blood pressure), 2) Decompose the effects identified in aim 1 into the effects of predictable and unpredictable instability and compare to the impact of increasing the average level of income, and 3) Investigate the channels through which effects on health occur, including both economic and behavioral channels and estimate the impact of key moderating factors (e.g. age, gender, baseline mental health). The trial will be conducted in southwestern Bangladesh. We will manipulate income instability by varying the number of work hours (and hence earnings) of participants in a cash-for-work program. Participants in the first treatment arm will have a fixed work schedule, with the same hours and earning each period. The hours and earnings of a second treatment arm will vary over time, but the fluctuations will be known in advance. Finally, the number of work hours and earnings of a third treatment arm will fluctuate unpredictably. Each of these arms will be compared to a control group that is surveyed, but not offered additional work. Importantly, we will vary income instability while holding the average level of income constant in order to disentangle the impact of instability from the level-effect. The study will create 1,867 new jobs that would not otherwise be available during the lean season when jobs are scarce. The intervention has been designed so that the job opportunity cannot make them worse off than they would otherwise have been in the absence of research.