This article addresses the problem of forecasting time series that are subject to level shifts. Processes with level shifts possess a nonlinear dependence structure. Using the stochastic permanent breaks (STOPBREAK) model, I model this nonlinearity in a direct and flexible way that avoids imposing a discrete regime structure. I apply this model to the rate of price inflation in the United States, which I show is subject to level shifts. These shifts significantly affect the accuracy of out-of-sample forecasts, causing models that assume covariance stationarity to be substantially biased. Models that do not assume covariance stationarity, such as the random walk, are unbiased but lack precision in periods without shifts. I show that the STOPBREAK model outperforms several alternative models in an out-of-sample inflation forecasting experiment.

When applied to time series processes containing occasional level shifts, the log-periodogram (GPH) estimator often erroneously finds long memory. For a stationary short-memory process with a slowly varying level, I show that the GPH estimator is substantially biased, and I derive an approximation to this bias. The asymptotic bias lies on the (0,1) interval, and its exact value depends on the ratio of the expected number of level shifts to a user-defined bandwidth parameter. Using this result, I formulate the Modified GPH estimator, which has a markedly lower bias. I illustrate this new estimator via applications to soybean prices and stock market volatility.

In commodity futures markets, contracts with various delivery dates trade simultaneously. Applied researchers typically discard the majority of the data and form a single time series by choosing only one price observation per day. This strategy precludes a full understanding of these markets and can induce complicated nonlinear dynamics in the data. In this paper, I introduce the partially overlapping time series (POTS) model to model jointly all traded contracts. The POTS model incorporates time-to-delivery, storability, seasonality, and GARCH effects. I apply the POTS model to corn futures at the Chicago Board of Trade and the results uncover substantial inefficiency associated with delivery on corn futures. The results also support two theories of commodity pricing: the theory of storage and the Samuelson effect.

As the second-largest economy globally, China's economy has developed rapidly in recent years, and China's stock market has developed particularly fast as one of the most representative and important markets. Its total market capitalization had grown from 4.03 trillion US Dollars in 2010 to 8.52 trillion US Dollars in 2019. It is an important investment channel for individual investors on the supply side. It also provides financing opportunities for enterprises on the demand side. The fast growth has brought many systemic problems that need to be solved urgently, like imperfect regulation and informed trading. Considering these characteristics, I choose China's stock market as the theme for deep research and study. The first chapter studies the informed trading problem in China's A-Share market. It investigates the abnormal drop of stock returns on announcement day for the listed firms on A board in China's stock market. It finds that the firms who reported a huge irrational goodwill impairment experienced a more significant decline before the announcement day, and their rebounds in abnormal return are, on average, more powerful after the announcement day. There is a significant negative jump in stock return on announcement day for these abnormal firms, while the other firms' jump is not significant. The difference in difference model over various cut-off days suggests that external reasons like informed trading are not the dominating factor determining the decline in abnormal returns of the abnormal firms before announcement day. Rather, the internal difference between firms plays a more important role in the stock disaster.

The second chapter focuses on the profit potential of China's stock indexes by applying neural network models. The chapter uses a neural network approach to predict China's stock indexes' returns from the Shanghai Stock Exchange(SSE) and the Shenzhen Stock Exchange(SZSE). It compares the ARIMA model, Multiple Layer Perception(MLP) model, and Recurrent Neural Network(RNN) model in terms of mean squared error and forecast accuracy to study the added value of improvement from model architecture. It turns out that MLP and RNN models failed to provide a significantly better prediction than ARIMA models if using historical information of stock prices alone. The paper also uses a standard quantitative trading strategy to backtest the value of predictions from these three models. The paper finds that the ARIMA model predicts the SSE Composite Index well, while the RNN model is the best in predicting Industrial Index and Composite Index. Most of the strategy's annualized returns surpassed the return of the benchmark stock index in the bear market from 2017 to 2020, which offers investors a better choice in their stock trading.

While my first two chapters focus on China's stock market, my third part of the dissertation studies the US's options market. China has not yet established a developed options trading system like the US, and the options market is an indispensable supplement of a complete stock market. Actually, China's options trading for stock indexes started in 2019. As the first stock index options product in China's domestic market, CSI 300 stock index options were listed and traded on the China Financial Futures Exchange on Dec 23rd, 2019. Considering the fact, Chapter 3, joint work with my colleague Yitian Xiao, study the US's options market. It is well known that the options market enables traders to hedge positions in asset markets, thereby reducing risk exposure. Traders can choose from many different strike prices -- analogous to the coverage level in an insurance contract -- when hedging. In Chapter 3, we argue that the hedging motive leads a typical hedger to take positions slightly out of the money, i.e., just below the asset's current value. We develop a simple theoretical model and validate its predictions using data on S

Decarbonizing the transportation sector, which is the largest contributor to U.S. greenhouse gas (GHG) emissions, requires a transition away from fossil fuels to renewables like biofuels and electricity. An increasing number of policies have been enacted to incentivize the use of renewable fuels. The U.S. Renewable Fuel Standard, California Low Carbon Fuel Standard, and Oregon Clean Fuels Program are in the vanguard of such policies in the United States. California and Oregon’s policies are examples of a carbon intensity standard, a tool that has become increasingly popular among U.S. policymakers. This dissertation explores the past, present, and future of these policies. The first chapter begins with the present, addressing an important challenge of achieving efficient outcomes currently faced by policy stakeholders: pass-through of policy costs and incentives to fuel prices. Findings suggest that compliance costs are fully passed through, and biofuel incentives are fully passed through in some regions and less than fully passed through in others. The second chapter looks ahead to the end of the decade, forecasting a range of compliance outcomes under California’s LCFS through 2030. Annual compliance requires that by 2030, the state’s transportation sector must achieve a 20 percent carbon intensity reduction below 2010 levels. Achieving the target will require the majority of diesel demand to be supplied with biomass-based diesel unless electric vehicle adoption grows substantially. Finally, lessons from the last decade are drawn in the third and final chapter, exploring trends in the three standing carbon intensity standards in California, Oregon, and British Columbia utilizing publicly available historical data.

Agricultural production around the world has undergone tremendous changes over the past century. However, two features remain salient in modern agricultural production. One is that global crop production depends heavily on a few major crop-producing regions, and the other is that climatic conditions still play a vital role in determining crop production in these regions. Not surprisingly, anthropogenic global warming has been considered one of the greatest threats to feeding the world’s growing populations. This dissertation studies the impacts of climatic events on agriculture in three aspects: farmers’ decisions in the first chapter, farm profits in the second chapter, and agricultural production in the third chapter.

In the first chapter, I ask the following question: can we find statistical evidence that farmers in the US Corn Belt show recency effects associated with local yield shocks? As climate change is projected to increase extreme heat events in frequency and severity, farmers are expected to see an increase in crop yield variability. Using field-level crop choice data in the major US Corn Belt states (Iowa, Illinois, and Indiana), this chapter uncovers statistical evidence that crop choice decisions in the region feature recency effects associated with local yield shocks largely driven by plausibly random weather. In the region, short-run acreage adjustment occurs mainly along the intensive margin (transition between corn and soybeans). I show that farmers are less likely to plant corn---which is more susceptible to heat/water stress than soybeans---after a hotter or drier than average year, irrespective of the within-season timing of the heat. This means that low yields of corn or soybeans in one year predict less corn being planted in the subsequent year. Interestingly, relative yield does not predict the probability of growing corn next year. Based on the insights from my conceptual model, these empirical results suggest that farmers respond sensitively to total exposure to extreme heat during the previous growing season but do not respond to when it was concentrated (e.g., 1st half or 2nd half of the growing season) or which crop it affected most.

In the second chapter, I ask the following question: what are the economic costs of large-scale droughts to crop producers? Extreme weather events, such as heatwaves and droughts, in a major crop-producing area decrease crop yields but tend to increase crop prices. Such a negative correlation makes it difficult to quantify net crop revenue impacts. This paper proposes a panel approach to estimating the impacts of extreme weather events in major crop-producing regions on crop revenues accounting for the correlation between crop price and yield in the context of US crop production. I first show that, under some conditions, weather-induced changes in crop revenues are identical to those of crop profits, for which data are scant. To estimate weather-induced crop revenue effects, I use a national-level yield shock as an explanatory variable in addition to local weather variables. This variable bears two appealing features. First, it can account for heterogeneous crop yield responses to weather across regions. Second, it permits coefficients that can be interpreted as the conditional price flexibilities of demand. I show that, when estimating crop revenue impacts of weather, it is important to additionally account for spatially varying degrees of the correlation between local and aggregate yield shocks, temporally varying price flexibilities of demand, and spatially heterogeneous yield responses to weather. I apply this approach to the 1988 and 2012 US Midwest droughts to quantify the impacts of the droughts on crop revenues across US counties. I estimate that crop revenue was impacted by (-)11% for corn and (+)1% for soybeans in 1988, and (+)11% for corn and 0% in 2012. I also document that, in the two years, regional inequality of crop revenues substantially deteriorated.

In the third chapter, I work with John Abatzoglou to answer the following question: how does planting-season weather affect agricultural production? Record-high prevented planting of staple crops such as corn and soybeans in the US Corn Belt due to heavy rainfall in recent years has spurred the concern over food security, as growing evidence suggests winter and spring precipitation extremes will occur more frequently in the upper US Corn Belt in the coming decades. We examine within-season time-varying effects of planting-season water balance---precipitation minus reference evapotranspiration---on prevented planting of corn and soybeans in the US Corn Belt. Our results show significant impacts of excess moisture on preventing planting and suggest a 58-176% increase in prevented planting during the months of April-June per standard deviation increase in water balance. This framework is additionally used alongside downscaled climate change projections to estimate future changes in county-level prevented planting during the mid-century (2036--2065) under the moderate emission scenario (RCP 4.5). We find that prevented planting will increase in parts of Iowa, Minnesota, and Wisconsin by 0-30% and generally decrease in the other parts of the US Corn Belt. This work highlights the value of incorporating water balance data in assessing prevented-planting impacts and is the first known study to examine changing risk of prevented planting under future climate scenarios that may help inform adaptation efforts to avoid losses.

This thesis consists of three chapters. In the first two chapters, I study rebound effects in solar adoption and an energy efficiency program (air-conditioning units (AC) upgrading program). Both solar adoption and AC upgrades reduce households’ energy bills and lower their average electricity prices. Households might adjust their energy consumption behavior which results in the actual reduction in energy use lower than the anticipated reduction. In chapter 1, I use novel data which contain the detailed household-level hourly purchase, sale, and solar generation in the Sacramento area. Contrary to existing literature, results do not show significant rebound effects. Using fixed effects models, I find a statistically insignificant rebound effect of 0.96%, which translates to a 0.0096 kWh increase in solar homes’ total electricity consumption when the solar generation increases by 1 kWh. This effect is also economically negligible. Results from chapter 1 enrich the current literature on solar adoption rebound effects.In contrast to chapter 1, significant rebound effects from the AC Energy Efficiency rebate program in the Sacramento Municipal Utility District (SMUD) serving area are identified in chapter 2. Household-level daily electricity consumption data and daily temperature data in the Sacramento area are utilized in this analysis. Regression mixture models are estimated by an expectation-maximization algorithm to recover premise-level temperature response functions and AC usage behavior functions. These functions are then used to calculate direct savings, total savings, and rebound effects from AC upgrades through a difference-in-difference design. On average, the AC energy efficiency rebate program reduces energy uses in cooling by 347.10 kWh per household in one summer. The rebound effects are estimated at 20.61%. When a household saves 1 kWh in cooling from AC upgrading, its total daily electricity consumption will increase by around 0.21 kWh. The increases in consumption are mainly caused by turning on AC units more often after AC upgrades. In chapter 3, I and coauthors study R&D lag structure in agriculture. Quite diverse models of R&D lag structures have been used by economists studying economic growth com- pared with those estimating returns to investments in industrial R&D or agricultural R&D. In this paper, we and coauthors empirically compare and contrast these alternative models and their implications for R&D knowledge stocks using data on multifactor productivity (MFP) in U.S. agriculture and U.S. public agricultural R&D investments. We employ a model selection procedure based on a combination of time-series properties of data, econo- metric estimation performance, and consistency of estimates with strongly held economic priors. We reject the models used in studies of economic growth and industrial R&D both on prior grounds and using statistical tests. The preferred model is a 50-year gamma lag distribution model similar in shape to Huffman and Evenson’s (1993) trapezoidal lag model. In this gamma lag model the effects of an investment in agricultural R&D on the R&D knowledge stock rise to a peak after 13 years and are mostly dissipated after 35 years. The estimated elasticity of MFP with respect to the knowledge stock is 0.28 and the implied marginal benefit-cost ratio is 23:1.

Genetic modification of crops has revolutionized food production, but it remains controversial due to food safety concerns. A recent food safety scare provides a natural experiment on the market's willingness to accept an increase in perceived risk from genetically modified (GM) food. We analyze the market impact of contamination of the U.S. food-corn supply by a GM variety called StarLink. We find that the contamination led to a 6.8 percent discount in corn prices and that the suppression of prices lasted for at least a year.

U.S. agricultural policy strives to boost farm incomes and secure the nation's food supply, yet new policies are also tasking agriculture with reducing or offsetting the country's greenhouse gas emissions. California's Low Carbon Fuel Standard uses a system of tradable compliance credits to incentivize the consumption of alternative fuels instead of fossil fuels. These credits provide additional value for the production of agricultural biofuels including ethanol. My dissertation examines the effects of the LCFS on commodity prices by first building a spatial corn market model of ethanol plants selling to California and then empirically estimating the pass-through of LCFS credit values changes in local corn prices. I find that the LCFS could be providing a competitive advantage in the corn market to ethanol plants with lower emissions, and these plants only pass-through around 40\% of LCFS credit value changes to local corn farmers as a result. Moreover, policymakers across the globe see working farms as a possible carbon sink through increasing the carbon sequestered in cropland soils. The Inflation Reduction Act of 2021 provides billions of dollars to USDA programs that tangentially influence carbon sequestration. However, soil carbon sequestration is a complex process that cannot be easily influenced by policy. I review the scientific literature on soil carbon sequestration and relate it to economic research on environmental externalities. While practices like no-till or biochar can increase carbon sequestration on agricultural lands, effective policy design faces substantial challenges due to heterogeneity, costly measurement, and uncertainty.