A critical part of adapting to the higher temperatures that climate change brings will be the deployment of existing technologies to new sectors and regions. This paper examines the evolution of the temperature-mortality relationship over the course of the entire 20 th century in the United States both for its own interest but also to identify potentially useful adaptations that may be useful in the coming decades. There are three primary findings. First, the mortality impact of days with a mean temperature exceeding 80° F has declined by about 70%. Almost the entire decline occurred after 1960. There are about 14,000 fewer fatalities annually than if the pre-1960 impacts of high temperature on mortality still prevailed. Second, the diffusion of residential air conditioning can explain essentially the entire decline in hot day related fatalities. Third, using Dubin-McFadden's discrete-continuous model, we estimate that the present value of US consumer surplus from the introduction of residential air conditioning (AC) in 1960 ranges from $83 to $186 billion ($2012) with a 5% discount rate. The monetized value of the mortality reductions on high temperature days due to AC accounts for a substantial fraction of these welfare gains.

This paper estimates the effects of humidity and temperature on mortality rates in the United States (c. 1973–2002) in order to provide an insight into the potential health impacts of climate change. I find that humidity, like temperature, is an important determinant of mortality. Coupled with Hadley CM3 climate-change predictions, I project that mortality rates are likely to change little on the aggregate for the United States. However, distributional impacts matter: mortality rates are likely to decline in cold and dry areas, but increase in hot and humid areas. Further, accounting for humidity has important implications for evaluating these distributional effects.

Recent research exploring associations between environmental factors and influenza outcomes has devoted substantial attention to the role of absolute humidity. However, the existing literature provides very little quantitative epidemiologic evidence on the relations between absolute humidity and other weather variables and influenza outcomes in human populations. In the present study, the authors helped fill this gap by analyzing longitudinal weather and influenza mortality data, observed every month between January 1973 and December 2002, for each of 359 urban US counties. A flexible regression model was used to simultaneously explore fully nonlinear relations between absolute humidity and influenza outcomes and temperature and influenza outcomes. Results indicated that absolute humidity was an especially critical determinant of observed human influenza mortality, even after controlling for temperature. There were important nonlinear relations; humidity levels below approximately 6 g of water vapor per kilogram of air were associated with increases in influenza mortality. Model predictions suggested that approximately half of the average seasonal differences in US influenza mortality can be explained by seasonal differences in absolute humidity alone. Temperature modestly influenced influenza mortality as well, although results were less robust.

We reconsider the effect of very low birth weight classification on infant mortality. We demonstrate that the estimates are highly sensitive to the exclusion of observations in the immediate vicinity of the 1,500-g threshold, weakening the confidence in the results originally reported in Almond, Doyle, Kowalski, and Williams (2010).

This study uses Monte Carlo simulations to demonstrate that regression-discontinuity designs arrive at biased estimates when attributes related to outcomes predict heaping in the running variable. After showing that our usual diagnostics are poorly suited to identifying this type of problem, we provide alternatives. We also demonstrate how the magnitude and direction of the bias varies with bandwidth choice and the location of the data heaps relative to the treatment threshold. Finally, we discuss approaches to correcting for this type of problem before considering these issues in several non-simulated environments.