Table 2

Regression Kink Design Estimates of Subsidy Incidence

  Linear with Controls: Individual Kink
Linear without Controls
(1)		Linear with Controls
(2)		Quadratic with Controls
(3)		Cubic with Controls
(4)		Left
(5)		Right
(6)
Subsidy incidence0.537***
(0.080)		0.378***
(0.078)		0.396***
(0.085)		0.377***
(0.086)		0.372***
(0.079)		0.479***
(0.128)
AIC−21,401−22,419−22,417−22,418
Observations13,68013,68013,68013,68013,1304,383

  • Note: The table presents the subsidy incidence after scaling δ by −1/.85 from equation [8]. Robust standard errors, additionally scaled, are clustered by county. The Akaike information criteria (AIC) measures model fit. Specifications 2-6 include the following control variables that were divided by BR: the proportion of land planted to corn, soybeans, wheat, and sorghum; farm-level per-acre revenue; year dummy variables. The sample only includes Midwest farms. Kinks were estimated at 0.845 and 1.010. Specifications 5 and 6 were only estimated across a single kink. The left kink is where payments become capped. The right kink is the payment threshold. AIC is not presented for specifications 5 and 6 because they contain different samples.

  • ↵*** p < 0.01;

  • ** p < 0.05

  • * p < 0.1