Nonlinear Temperature Effects of Dust Bowl Region and Short-Run Adaptation A. John Woodill University of Hawaii at Manoa Seminar in Energy and Environmental Policy March 28, 2016 A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 1 / 33
Outline Motivation Identification Strategy Data Regression Results Conclusion A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 2 / 33
Motivation Source: Center for Air Pollution Impact and Trend Analysis A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 3 / 33
Motivation Source: National Archives and Records Administration, Records of the Bureau of Agricultural Economics A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 4 / 33
Motivation Source: Historic Adobe Museum A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 5 / 33
Motivation A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 6 / 33
Motivation Hornbeck 2012 Limited adaptation through land use Outward migration never fully returned Introduction of crops and over farming practices What about potential impacts of changing weather and adaptation of crops through time? A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 7 / 33
Motivation Dust Bowl Literature Problems with weather adjustments 1890 1925 (Libecap and Hansen 2002) Small farms were less likely to control for erosion (Hansen and Libecap 2004) Most severe drought and moisture deficit since 1855 (Burnette and Stahle 2012) A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 8 / 33
Motivation Adaptation Literature High temp for corn are harmful above 29 C (Schlenker and Roberts 2006, 2009) Indiana: High sensitivity at 2000, but less sensitive from 1940 1960 (Roberts and Schlenker 2011) No increase in heat tolerance east of 100 meridian (Roberts and Schlenker 2012) Major drought in 1936 increased adoption of hybrid corn (Sutch 2009) A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 9 / 33
Goal of Study Focus on weather impacts for great plains and subsets Investigate short and medium-run impacts using degree days What were these marginal effects over time? Is there evidence of adaptation? Linear and Nonlinear Estimation of corn yield and temperature A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 10 / 33
Outline Motivation Identification Strategy Data Regression Results Conclusion A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 11 / 33
Identification Strategy What are degree days? Source: Hand Calculating Degree Days Snyder 1984 A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 12 / 33
Identification Strategy Degree Days How long during a day was the temperature above a threshold? Sum of truncated degrees on each day during growing season Provides heat exposure to identify impact on crop growth (yield) Example (Degree Days above 29 C) 34 C all day = 5 Degree Days 34 C 1/2 day (5 C x 0.5 day) = 2.5 Degree Days A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 13 / 33
Identification Strategy Why is this important? Degree Days between 10 C and 29 C important to crop yield Degree Days above 29 C shown to have negative effect Cross-section and time series link between annual yields and weather outcomes Historical agriculture and weather data to conduct study 1910 1960 in The Great Plains Region A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 14 / 33
Outline Motivation Identification Strategy Data Regression Results Conclusion A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 15 / 33
Data Agriculture Data Michael Haines Historical, Demographic, Economic, and Social Data: The United States, 1790-2000 (Hornbeck 2012) 1910, 1920, 1925, 1930, 1935, 1940, 1945, 1950, 1954, 1959 National Agricultural Statistics Service (NASS) 1910 1960 Fine-scale Interpolated Weather Data PRISM Monthly: Min, Max, Precipitation (Grids) NCDC Daily: Min, Max, Precipitation (Stations) Relative Anomaly Spline Interpolation Daily Fine Scale Grid Data Integrated Sine approach to degree day calculation A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 16 / 33
Data A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 17 / 33
Data Average Corn Yield for Great Plains A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 18 / 33
Data Average Irrigated Yield for Great Plains A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 19 / 33
Data County Level Corn Yield Changes 1910 1960 A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 20 / 33
Data U.S. Corn Yields, 1866 2002 Source: The Impact of the 1936 Corn Belt Drought on American Farmers Adoption of Hybrid Corn Richard Sutch 2011 A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 21 / 33
Data 10 C 29 C Degree Day Changes 1910 1960 A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 22 / 33
Data >29 C Degree Day Changes 1910 1960 A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 23 / 33
Outline Motivation Identification Strategy Data Regression Results Conclusion A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 24 / 33
Linear Regression Model y it = β 1 DD H it + β 2 DD M it + β 3 p it + β 4 p 2 it + t + t 2 + c i + ɛ it y it : log Corn Yield : Degree Days >29 C : Degree Days 10 C 29 C p it : Precipitation t it : Time trend c i : County-level Fixed Effect DDit H DDit M Weighted by county corn acreage Huber-White standard errors for heteroskedasticity Clustered by state for spatial correlation A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 25 / 33
Linear Regression Results A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 26 / 33
Linear Regression Results A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 27 / 33
Nonlinear Regression Model y it = β 1 DD H it + β 2 DD M it + f p (p it ) + f t (t) +f M (t) DD M it + f H (t) DD H it + f t2 (t) f p2 (p it ) + c i + ɛ it y it : log Corn Yield : Degree Days >29 C : Degree Days 10 C 29 C p it : Precipitation f x ( ) : Restricted cubic splines c i : County-level Fixed Effect DDit H DDit M Weighted by county corn acreage Huber-White standard errors for heteroskedasticity Clustered by state for spatial correlation A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 28 / 33
Nonlinear Regression Results 3-knot Spline A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 29 / 33
Nonlinear Regression Results A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 30 / 33
Nonlinear Regression Results A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 31 / 33
Conclusion Results: No signification adaptation and decreasing in Dust Bowl region Precipitation more sensitive in Dust Bowl region Shift north for corn production during time period Possible medium and long-run effects? Was there crop switching occurring to adapt? A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 32 / 33
Conclusion Thank you! Questions? A. John Woodill Nonlinear Temperature Effects of Dust Bowl March 28, 2016 33 / 33