This paper examines the causal association between family background characteristics--parental education and family size-and returns to schooling using data from the Occupational Change in a Generation Survey. I first develop a formal model of schooling and earnings, with heterogeneous returns to education. Family environment is shown to influence the marginal return to schooling through its effects on the marginal benefit and the marginal cost of an additional year of education. Using two types of exclusion restrictions, I find that men raised in larger families have substantially lower returns to education, while the combined effects of parental education on the returns to education are more modest. I also examine the difference between OLS and TSLS estimates of the return to schooling. Like other “supply-side” IV studies of the causal effect of education, this paper documents TSLS estimates that are larger than the corresponding OLS estimates. The results of this paper provide an alternative explanation for this phenomenon: constant marginal returns to schooling, combined with a negative ability bias and a positive self-selection bias (i.e. non-hierarchical sorting).

This paper quantifies the change in the causal effect of education on labor market earnings in the United States between 1979 and 2000. In absence of valid instrumental variables for schooling, a causal model for earnings and schooling that incorporates heterogeneity in absolute and comparative advantage across individuals is used to impose some structure on the observed relationship between schooling and earnings. A simple intuition arises from the model: if individuals with higher returns to schooling acquire more schooling, the relationship between log earnings and schooling will be convex in the population. Likewise, for a fixed cohort of individuals, the degree of convexity will rise over time if the causal return to education rises. Differences across cohorts in the mapping between schooling and ability will lead to permanent differences in the profiles of the earnings-schooling relationship. Changes in the observed relationship between schooling and earnings can therefore be decomposed into year-specific and cohort-specific factors corresponding to causal and confounding components. Using CPS data for cohorts of men born between 1930 and 1970, I find that the causal return to education increased by 30% between 1979 and 2000, after controlling for the confounding effects of time-varying ability and comparative advantage biases across cohorts.

This paper measures the economic impact of climate change on US agricultural land. We replicate the previous literature's implementation of the hedonic approach and find that it produces estimates of the effect of climate change that are very sensitive to decisions about the appropriate control variables, sample and weighting. We find estimates of the benchmark doubling of greenhouse gases on agricultural land values that range from a decline of $420 billion (1997$) to an increase of $265 billion, or -30% to 19%. Despite its theoretical appeal, the wide variability of these estimates suggests that the hedonic method may be unreliable in this setting. In light of the potential importance of climate change, this paper proposes a new strategy to determine its economic impact. We estimate the effect of weather on farm profits, conditional on county and state by year fixed effects, so the weather parameters are identified from the presumably random variation in weather across counties within states. The results suggest that the benchmark change in climate would reduce the value of agricultural land by $40 to $80 billion, or -3% to -6%, but the null of zero effect cannot be rejected. In contrast to the hedonic approach, these results are robust to changes in specification. Since farmers can engage in a more extensive set of adaptations in response to permanent climate changes, this estimate is likely downwards biased, relative to the preferred long run effect. Together the point estimates and sign of the likely bias contradict the popular view that climate change will have substantial negative welfare consequences for the US agricultural sector.

This paper measures the economic impact of climate change on US agricultural land by estimating the effect of the presumably random year-to-year variation in temperature and precipitation on agricultural profits. Using long-run climate change predictions from the Hadley 2 Model, the preferred estimates indicate that climate change will lead to a $1.3 billion (2002$) or 4.0% increase in annual profits. The 95% confidence interval ranges from -$0.5 billion to $3.1 billion and the impact is robust to a wide variety of specification checks, so large negative or positive effects are unlikely. There is considerable heterogeneity in the effect across the country with California’s predicted impact equal to -$0.75 billion (or nearly 15% of state agricultural profits). Further, the analysis indicates that the predicted increases in temperature and precipitation will have virtually no effect on yields among the most important crops, which suggest that the small effect on profits are not due to short-run price increases. The paper also implements the hedonic approach that is predominant in the previous literature and finds that it may be unreliable, because it produces estimates of the effect of climate change that are extremely sensitive to seemingly minor decisions about the appropriate control variables, sample and weighting. Overall, the findings contradict the popular view that climate change will have substantial negative welfare consequences for the US agricultural sector.

This paper produces the first large-scale estimates of the US health related welfare costs due to climate change. Using the presumably random year-to-year variation in temperature and two state of the art climate models, the analysis suggests that under a ‘business as usual’ scenario climate change will lead to an increase in the overall US annual mortality rate ranging from 0.5% to 1.7% by the end of the 21st century. These overall estimates are statistically indistinguishable from zero, although there is evidence of statistically significant increases in mortality rates for some subpopulations, particularly infants. As the canonical Becker-Grossman health production function model highlights, the full welfare impact will be reflected in health outcomes and increased consumption of goods that preserve individuals’ health. Individuals’ likely first compensatory response is increased use of air conditioning; the analysis indicates that climate change would increase US annual residential energy consumption by a statistically significant 15% to 30% ($15 to $35 billion in 2006 dollars) at the end of the century. It seems reasonable to assume that the mortality impacts would be larger without the increased energy consumption. Further, the estimated mortality and energy impacts likely overstate the long-run impacts on these outcomes, since individuals can engage in a wider set of adaptations in the longer run to mitigate costs. Overall, the analysis suggests that the health related welfare costs of higher temperatures due to climate change are likely to be quite modest in the US.

China has experienced drastic climate change and severe environmental pollution since the 1980s. This dissertation provides economic analyses of these two types of environmental changes in China by focusing on the temporal evolution of agricultural sensitivity to extreme heat and the political economy explanations for severe air pollution in China. Chapter 1 examines the time-varying impacts of extreme temperatures on Chinese agriculture over 1981 to 2010. By estimating a period-specific panel regression model using nationwide county-level agriculture production data combined with fine-scale meteorological data, I primarily find the impact of a daily exposure to extreme temperatures on corn and soybean yields in the post-1996 period is 40% to 50% less than that in the pre-1996 period and the decline in the extreme temperature impacts on crop yields mainly occurs in counties with expanding irrigation coverage. Chapter 2 explores reasons for severe air pollution from the perspective of political economy by examining the environmental consequences of the county-to-city upgrading policy which delegates the autonomy of building cities to upgraded counties. In a centralized system like China, economic decentralization without changing the promotion metrics centered around economic performance for local government officials would likely lead to worse environmental quality because local officials compete for promotion on economic performance. Using a comprehensive county-level dataset on economic performance indicators and air pollutant concentrations, I primarily find significantly positive policy effects on economic growth and air pollutant concentrations, suggesting that the upgraded counties promoted economic performance at the cost of local air quality. I also calculate the total loss due to the increasing air pollution as valued in terms of statistical life to indicate the magnitude of the social cost of the upgrading policy.

Following the finding of decline in agricultural sensitivity to extreme heat in Chapter 1, Chapter 3 quantifies the contribution of the temporal evolution of extreme temperature impacts to the growth of agricultural revenue during 1981-2010 using an Oaxaca-Blinder decomposition which attributes the growth of agricultural revenue to the change in the levels of predictors and to the change in the coefficients for the predictors. I find extreme temperature impacts on agricultural revenues per hectare in the post-1996 period is more than 60% lower than that in the pre-1996 period, contributing 6.1 percentage points of revenue growth over the two periods, which is 5.4% of the overall growth of agricultural revenue. The significant increase in marginal benefit of irrigation in terms of moderating extreme temperature impacts may have contributed about 40% of the decline in the extreme temperature impacts on agricultural revenue per hectare.

As the world transitions toward a low-carbon energy economy, two features of natural gas set it apart from other fossil fuels. First, the combustion of natural gas generates about half as much carbon dioxide as coal. Second, gas-fired electricity generation can be ramped up and down quickly and efficiently, making it well-suited for balancing out intermittently available wind and solar energy. These two factors lead many to believe natural gas will play an extended role in the electricity generation mix. However, there are many ways in which current markets for natural gas not well-adapted for its role in the energy transition. This dissertation explores two areas where natural gas markets can be improved for its efficient utilization in the transition to low-carbon energy. The first two chapters develop policy tools to help reduce methane emissions from the natural gas supply chain, and the third chapter investigates a previously unknown market failure that can arise in interconnected natural gas and electricity markets.

The first chapter of this dissertation empirically estimates the cost of reducing methane emissions from the extraction segment of the natural gas industry. Although natural gas has important climate benefits, it is composed of about 90 percent methane, which is itself a greenhouse gas that is far more potent than carbon dioxide. A small fraction of emitted gas (in the form of equipment leaks and intentional venting) can therefore have severe warming effects. This chapter estimates the cost of reducing these emissions by examining how production facilities' emission rates respond to changes in natural gas prices. Because firms mitigate emissions up to the point at which their marginal cost of mitigation equals their marginal private benefit of being able to sell captured gas, an estimated relationship between emission rates and prices can be used to determine mitigation costs. Results indicate that methane emissions from natural gas production can be reduced at very low cost relative to other sources of greenhouse gas emissions. For example, an emissions price equivalent to the social cost of methane is predicted to decrease emissions by about 76 percent while increasing the net cost of natural gas extraction by less than one percent.

Building on this result, the second chapter explores how emissions pricing can be used to regulate methane emissions in practice. Previously, emissions pricing programs have been implemented based on the carbon content of fossil fuels or by using continuous emissions monitoring sensors placed in smokestacks. However, because methane emissions from the natural gas industry are released from many different sources in a variety of different ways, comprehensively monitoring them is prohibitively costly at this time. This chapter outlines a novel estimate-based approach for implementing emissions pricing in this setting. Rather than monitoring emissions at all facilities continuously, the regulator randomly selects a subset of each firm's facilities to perform measurements at. The regulator then uses these measurements to develop a firm-level estimate of emissions, which can then be used to apply an emissions tax or account for the use of permits. A theoretical model demonstrates that this approach preserves the efficiency benefits of emissions pricing with comprehensive measurement. Furthermore, a simulation calibrated to be representative of the U.S. natural gas industry predicts that this approach can achieve climate mitigation benefits roughly two orders of magnitude greater than the cost of measurement.

The third chapter, which is coauthored with Charles F. Mason, Kristina Mohlin, and Matthew Zaragoza-Watkins, explores a market failure that can arise from the increasing interdependence of natural gas and electricity markets. It develops a theoretical model that illustrates conditions under which a firm that owns both electricity generation plants and contracts for natural gas pipeline capacity may find it optimal to withhold those contracts from secondary markets. By artificially limiting the available supply of pipeline capacity on constrained days, this behavior increases electricity prices in the downstream electricity generation market, which benefits non-gas generators owned by the withholding firms. We document pipeline scheduling patterns exhibited by two firms in New England that are consistent with this behavior. We then estimate the impacts of this behavior, finding that it increased wholesale natural gas and electricity prices by 35 percent and 18 percent, respectively. We estimate that substitution from natural gas generation to coal and oil generation due to these artificial supply constraints resulted in economic losses of $1.5 billion over a three-year period. While this behavior may have been within the firms' contractual rights, these findings underscore a need to improve regulation and coordination of these increasingly linked energy markets.

The first chapter concerns the eruption of Mount St. Helens in Washington State on May 18th, 1980 which resulted in a massive and unanticipated particulate air pollution shock. I use the incidence of the ash cloud fallout across Washington as a natural experiment to estimate the effect of a particulate shock on birth outcomes and infant mortality. I find that while there is no statistical effect on infant mortality, there were statistically fewer low birth weight babies born. The measured effect indicates about 230 fewer babies were born as a result of exposure while in the womb. These results suggest that about 1 in 10 pregnancies were terminated from the 20th percentile group of the weight distribution, increasing to almost 1 in 5 in the lowest 5th percentile. The effect is found to be the strongest in the early stages of pregnancy. Using these findings I estimate that the cost of a single particulate shock of typical magnitude on a metropolitan area of median size is $3.7 million. This quasi-experimental treatment is unique since the ash particulate is bioreactively inert and uncorrelated with other pollution. It clearly identifies the pernicious nature of all particulates, not just particulate categorized by source. In addition, the discrete timing of the event identifies the most vulnerable window for suspended particulate shocks on expectant mothers.

In the second chapter, my colleague Daniel Moncayo and I investigate the informational effect that social comparisons play in household electricity use. We employ a randomized field experiment to present individuals with a social norm that encourages them to conserve electricity by comparing their consumption to that of an ``energy efficient neighbor,'' consisting of the average of the 10th percentile of participants' electricity consumption. Utilizing smart-meter data, we find households that received our informational treatment reduced their electricity consumption by an average of 7%, even though the treatment was private and there was no financial incentive to conserve. We also discover that conservation gains are largest during the peak morning and evening hours.

The third chapter examines the effect of country-level political stability and investment security on forestland use. Using cross-section data I find that these measures are associated with benign outcomes for overall rates of forest area change and roundwood production. These associations are robust when instrumented for endogeneity, and reveal stronger impacts than OLS estimates would imply. Two-stage least squares results indicate that a one standard deviation increase in political stability, as currently measured by the World Bank, increases forest area by 12% over 10 years, and increases roundwood production by a factor of 10 to 19. Targeting political stability and investment security may be one of the most effective tools in mitigating carbon emissions through forest expansion and increasing forest productivity.

The rise in the divorce rate over the past 40 years is one of the fundamental changes in American society. A seemingly ever-increasing number of women and children spend some fraction of their life in single female-headed households, leading many to be concerned about the economic circumstances of these women their and children. Estimating the cause-to-effect relationship between marital dissolution and female economic status is complicated because the same factors that increase marital instability may also affect the economic status and labor market behavior of women. We propose an instrumental variables solution to this problem based on the sex of the firstborn child. This strategy exploits the fact that the sex of the firstborn child is random and the fact that marriages are less likely to survive following the birth of girls as opposed to boys. Our IV estimates cast doubt on the contention that marital instability causes large declines in woman’s economic status. Once the negative selection into divorce is accounted for, we find that women who have experienced marital dissolution have considerably higher levels of personal income and annual wages than women who remain married. At the same time we find little evidence of differential poverty rates and equivalized household incomes among ever-divorced women and never-divorced women. We further show that the higher wages of ever-divorced women mostly reflect increased labor supply intensity (hours and weeks of work) of woman who experienced marital dissolution.