The impact of US sugar prices on the financial performance of US sugar-using firms

Agricultural and Food Economics

Table 4 Regression with ROA_t + 1 as the dependent variable

Variable	Model 1	Model 2	Model 3
ln(SGA)	− 0.003	− 0.005	0.006
ln(SGA)	(0.146)	(0.147)	(0.147)
t	− 0.02	− 0.03	0.04
ln(ATO)	0.600***	0.586***	0.581***
ln(ATO)	(0.179)	(0.177)	(0.178)
t	3.35	3.31	3.26
ln(Size)	0.153***	0.154***	0.140***
ln(Size)	(0.041)	(0.039)	(0.039)
t	3.76	3.98	3.62
ln(Growth)	0.010	0.007	0.008
ln(Growth)	(0.011)	(0.010)	(0.011)
t	0.87	0.71	0.76
ln(FRisk)	− 0.056***	− 0.056***	− 0.058***
ln(FRisk)	(0.016)	(0.015)	(0.016)
t	− 3.59	− 3.64	− 3.75
ln(UWpra)	0.179**
ln(UWpra)	(0.068)
t	2.62
ln(UWpre)		0.259**
ln(UWpre)		(0.099)
t		2.61
ln(UWprb)			0.199***
ln(UWprb)			(0.068)
t			2.92
Constant	0.861	0.793	0.926
Constant	(0.609)	(0.574)	(0.586)
t	1.41	1.38	1.58
Observations	953	953	953
# of firms	26	26	26
R-squared	0.297	0.295	0.284
F value	54.42	71.30	68.58

Notes: Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. Reported R² are “within” R² statistics. Variable definitions are in Table 1. The data were winsorized at the 1% and 99% percentiles (Adams et al. 2019; Tukey 1962). The Akaike Information Criterion (AIC) and Bayesian Information (BIC) tests confirmed that winsorized data explain better the models