Blake Hoarty, Jeffrey H. Dorfman
Abstract. Farmers have historically exhibited a tremendous amount of heterogeneity in their hedging patterns. Previous research has mostly modeled farmer behavior by employing the mean-variance hedge ratio in a 2-period investment setting. In this paper, we use expected utility theory to rationalize hedging decisions made by farmers in a dynamic setting, allowing for subjective beliefs about market conditions. We enable our representative farmer to make hedging decisions on a weekly basis to account for incremental changes in market beliefs. We find that subjective prior beliefs and crop insurance coverage play a significant role in hedging decisions. Other factors such as initial wealth play a small but non-negligible role in guiding hedging decisions. Our results indicate that only pessimistic farmers hedge in concordance with traditional financial wisdom. We also find that unbiased farmers, regardless of the degree of risk aversion, forgo hedging until late in the calendar year.
The Evolving Nature of Seasonality in Meat Demand
Blake Hoarty, Barry Goodwin, Jeffrey H. Dorfman
Meat demand has been a topic of interest for agricultural economists for the past 100 years and the methodologies used to model demand features such as seasonality and changes in tastes have been subject to substantial debate. In this paper, we estimate seasonality in the Rotterdam demand system from Barten (1964) and Theil (1965) for beef, pork, chicken and turkey from the years 1980-2020 across three different methodologies and assess their results and implications. Patterns of seasonality in meat demand appear to have shifted significantly in recent years, and we evaluate the evolving nature of seasonality in the demand for individual meat products. We first estimate a standard Rotterdam demand system with seasonal dummies. Next, we estimate a Rotterdam demand system with second order Fourier expansion to assess smoother patterns of seasonality. Lastly, we estimate a gradually switching regime model that enables us to assess seasonality while imposing only minimal restrictions on the representative agent's utility function.
Moment Based Prior Sensitivity Analysis in High Dimensional Bayesian Econometric Models
Blake Hoarty, Sujit Ghosh, Jeffrey H. Dorfman
Variable selection is an imperative feature of high-dimensional statistics. The problem of variable selection can be approached through Bayesian methodologies. Given the fact that the posterior draws inference from the prior and the likelihood function, the choice of the prior is often non-trivial, especially under large sample sizes. Mitchell and Beauchamp (1988) coined the Bayesian spike and slab regression that employs a two- part prior on regression coefficients; namely, a spike at 0 for noise variables, and a slab distribution for regressors that differ significantly from 0. The flexible nature of the spike and slab priors give the model a rich environment for prior sensitivity analysis. The goal of this paper is to conduct a prior sensitivity analysis on spike and slab regressions to assess how local perturbations in the prior affect posterior diagnostics under different data generating processes. We employ a skewed generalized t-distribution (SGT) from Theodossiou (1998) in lieu of a normal distribution. The 5-parameter distribution is amenable to moment-preserving local changes, which make it ripe for sensitivity analysis. We evaluate how these moments influence variable selection accuracy, estimation bias, and related posterior diagnostics across high-dimensional data generating processes. Our findings indicate that posterior inference is highly sensitive to even small deviations in prior specification. In particular, the SGT-based prior often improves variable selection performance relative to the conventional normal prior, albeit with notable tradeoffs in false positive rates. These results highlight the importance of flexible prior design and underscore the non-negligible role of prior choice in high- dimensional Bayesian regression settings.
Optimally Combining Information Sources to Estimate the Federal Poverty Guidelines: Robustness and Red Flags
Jeffrey H. Dorfman, Blake Hoarty, Peibin Hou, Wenying Li, and Falin Sun
Equivalence scales measure the income difference needed to compare income across households of different sizes and types, important for policy actions such as setting welfare program eligibility thresholds in an equitable manner. Empirical estimates of equivalence scales have been subject to criticism over features such as significant sensitivity to identifying assumptions. This paper presents a Bayesian method of combining estimates of equivalence scales from two disparate sources of empirical estimates: surveys and scanner data. We treat estimates from a survey in which respondents were asked how much extra (or less) income it would take to equalize their financial situation if their household size went up (or down) as our prior. Then we use revealed expenditure patterns from households with two adults and either with zero or one child as our data and numerical Bayesian techniques to estimate a posterior distribution of the equivalence scale for this change in household type. This takes advantage of the fact that while Bayesian methods are typically thought of as a method for combining subjective prior beliefs with data-based information, they are actually a flexible, optimal method for combining any two sets of information on the same measures of interest. Our results provide insight on the equitable setting of poverty lines and welfare program eligibility standards as well as provide a future framework for producing more robust estimates by utilizing multiple different data sources. Further, the Bayesian methodology raises red flags when empirical estimates of equivalence scales from the two sources disagree to such an extent that the combined estimates (and at least one of the component estimates) should be considered unreliable.
Is There A Future for Tobacco Farming? (Link)
Jeffrey H. Dorfman, Blake Hoarty, Younghyeon Jeon
The number of tobacco farms has been in sharp decline in the United States. Over half of the remaining tobacco farms and an even larger share of tobacco acres are in either Kentucky or North Carolina. A new survey of tobacco growers in North Carolina suggests some uncertainty over the survival of the crop in the U.S. We surveyed tobacco farmers in North Carolina by emailing all of them a link to an online survey as well as using county agents to remind them to complete the survey. We ended up with a sample of 114 usable responses for most sets of variables. The results mgive us the most detailed current picture we have of tobacco growers.
PPI and CPI seasonal adjustment during the COVID-19 pandemic (Link)
Blake Hoarty, Steven M. Muri, Daniel J. Pallotta, Marie Rogers, Jonathan C. Weinhagen, Jeffrey S. Wilson
Published in: Monthly Labor Review U.S. Bureau of Labor Statistics, May 2022.
The U.S. Bureau of Labor Statistics publishes seasonally adjusted Consumer Price Index (CPI) and Producer Price Index (PPI) data monthly. Seasonal adjustment removes within-year seasonal patterns from data. To seasonally adjust data and estimate seasonal patterns of time series, the CPI and PPI use a filter-based approach that employs moving averages of historical data. In 2020, many PPIs and CPIs experienced extreme movements because of the coronavirus disease 2019 (COVID-19) pandemic. For example, the PPI and CPI for gasoline decreased 53.0 percent and 16.5 percent in April 2020, respectively. Because the CPI and PPI use historical data to estimate seasonal patterns, the extreme price movements in 2020 could have adversely affected the capability of the two price programs to accurately estimate seasonally adjusted data. This article explains how the CPI and PPI mitigated the effects of the COVID-19 pandemic on their seasonally adjusted price indexes. Mitigation steps included identifying price indexes whose movements were affected by the pandemic, estimating time series models to quantify these effects, and removing pandemic-related price movements from the data before estimating seasonal patterns.
On regulation and excess reserves: The case of Basel III (Link)
Stephen Matteo Miller, Blake Hoarty
Published in: Journal of Financial Research 44(2): 215–247, March 2021.
Common explanations for the observed rise in excess bank reserves include payment of interest on reserves and liquidity regulations, but capital regulations may also matter. We use Kuhn-Tucker conditions to show that a profit‐maximizing bank substitutes from higher risk‐weighted loans to lower risk‐weighted reserves and Treasuries if the risk‐based capital ratio, but not the leverage ratio, increases. Using a fully-flexible difference-in-differences model, we show estimated treatment effects for “advanced approaches” bank holding companies, the focus of Basel III capital regulations, initially increased for reserves after the regulatory changes but later decreased when treatment effects for Treasuries increased as Treasury yields rose; treatment effects for loans became increasingly negative.
Licensing massage therapists in the name of crime: the case of Harper v Lindsay (Link)
Darwynn Deyo, Blake Hoarty, Conor Norris, Edward Timmons
Published in: Journal of Entrepreneurship and Public Policy 10(1): 1-14, February 2021.
This study aims to analyze the trends for crime and STDs after the passage of massage therapist licensing. In 1977, Texas passed a law permitting county-level licensing laws for massage therapists, which was soon followed by a statewide licensing requirement in 1985. This early massage therapy law was upheld by the Fifth Circuit Court of Appeals. Massage therapy licensing is commonly associated with preventing crime, specifically prostitution. However, massage parlors also represent an opportunity for entrepreneurs starting businesses, who face significant barriers to entry across the USA.
Who Participates? An Analysis of School Participation Decisions in Two Voucher Programs in the United States (Link)
Corey DeAngelis, Blake Hoarty
Published in: Cato Institute Policy Analysis no. 848, September 2018
Most empirical studies of school choice examine academic outcomes of students who attend private schools compared to students who do not. We eschew this traditional approach and examine the quality of schools that decide to participate in school voucher programs in Milwaukee, WI and the state of Ohio. We use probit regression analysis to examine the quality of schools that elected to participate in voucher programs in Ohio and Milwaukee. Using tuition and enrollment levels—proxies for price and quantity demanded — we find evidence suggesting that lower‐quality schools are more likely to participate in voucher programs. Specifically, a $1,000 increase in tuition is associated with 3 percent lower likelihood of participation in the Milwaukee voucher program and a 3.8 percent lower likelihood of participation in the Ohio Educational Choice voucher program. We also find that a one‐point increase in a GreatSchools review score is associated with a 14.8 percent reduction in the likelihood of voucher program participation in Milwaukee. (GreatSchools is an online nonprofit that provides educational information and reviews for private, public, and charter schools.)