Click here to view my research statement.
Job Market Paper
This paper examines animal replacement behavior for over 1,000 Wisconsin dairy farms during the period 2011-2014 and analyzes the rationale for high replacement rates. I model the replacement decision using a dynamic discrete choice model and incorporate unplanned mortality as a source of uncertainty that drives farmers to replace dairy cows before they maximize production. The empirical model incorporates cow and herd heterogeneity in mortality rates to back out the implied cost of cow mortality. Using the conditional choice probability method, I estimate the cost of mortality at 1,800 USD per death, 800 dollars more than estimates based on simulation studies. Utilizing farm size heterogeneity, I also find that mortality costs are three times higher on small dairies than on larger ones. In a counterfactual estimation, dairy farmers were willing to pay 1,300 USD to eliminate mortality risk completely for first year dairy cows. These results suggest that genetic selection in U.S. dairy favors relatively large farms and may be accelerating the exit of small farms.
with Brent Hueth and Guilherme Rosa
Estimates of productivity growth in the dairy sector attribute as much as half of observed growth to genetic improvement. Unobserved match quality is an important determinate of genetic selection by dairy farmers that confounds attribution to genetic improvement alone. Using data from a large sample of Wisconsin dairy farms, and national-level data on sire rankings, we develop and estimate a model that accounts for selection behavior, and decompose total productivity change into separate effects for genetic improvement and endogenous selection. We find that selection accounts for as much as 75 percent of the total productivity improvement in our sample. Our results provide evidence for positive assortative matching, whereby farmers who adopt above-average yield genetics also perform better than average for their chosen genetics. Further, we find that management behavior accounts for a significant portion of within-herd cow-level heterogeneity, suggesting that dairy farmers manage their herds at the level of individual cows. Overall, our results indicate that a large portion of productivity growth in dairy farming can be explained by farmers’ ability to identify and select genetics well suited to their production environment.
with Brent Hueth
We study the impact of Production Credit Associations (PCAs) during the decade-long period shortly after their introduction as one component of the 1916 Federal Farm Loan Act. Using county distances to PCAs as a proxy for cost of access to credit, we examine the effects of credit expansion on county-level crop yield, crop revenue, and input use. Despite serving only about 7% of U.S. farmers during the period we study, we estimate that counties 100 kilometers closer to a PCA had roughly 10% higher crop revenue per acre. We also find that counties closer to PCA locations experienced significantly higher growth rates in tractor and fertilizer utilization, relative to more distant counties. In years prior to the arrival of PCAs, farms in relatively close-by counties earn on average less revenue and use fewer purchased inputs than farms in counties further away. This relationship reverses in subsequent years, suggesting that the mechanism for identifying PCA locations (one per state, initially) targeted less well-off counties. Our estimates therefore represent lower bounds for the true causal effect of access to credit on farm revenue and input use during the period we study.
with Brent Hueth
Milk production is allegedly inelastic to changes in milk price and feed cost in the short-run. In the literature on dairy farm supply response, studies almost always find short-run response to be small or insignificant. Such studies, however, are usually done at the herd and quarterly level where the mechanisms of supply response cannot be distinguished. Using a monthly, animal level data set, we analyze supply response at the animal level which isolates the intensive margin response, that is use of more inputs, subject to the production process. In our empirical analysis of over ten million animal records, we reject the null hypothesis of no response, finding that dairy cow milk production is impacted by changes in milk price and slaughter price. Specifically, milk production increases in response to milk price at the point in the lactation curve where the marginal returns to feeding are highest. We also find that current month milk price does not explain milk production but the milk price lagged two months does. Further, movements in the slaughter price have a much larger effect on production than milk or ration prices, suggesting a future area of research for dairy farm supply response.
with Karen Marshall and Ayao Missohou
In Senegal, several government campaigns have been undertaken to expand access to AI to increase the productivity of the dairy sector with at times mixed success. In this survey, dairy farmers in Senegal were asked about the advantages and disadvantages of AI and natural breeding as well as their willingness to pay (WTP) for natural breeding and AI of various exotic cattle using a double-bound contingent valuation experiment. Farmers surveyed listed the main disadvantage of AI as a low success rate and lack of breed choice and generally did not value AI differently than natural breeding. In contrast to previous results, farmer characteristics had relatively no bearing on WTP whereas previous AI use and using a private provider had a significant effect on WTP; in general, those using a private provider valued the technology more. Given the low success rates of public campaigns, farmers using the public service may value the technology less because of these experiences.