Fixed-effects panel Poisson model of wildland arson fire occurrences, 18 California national forests, 1993-2002.

In contrast with our uncertainty about the role of law enforcement and socioeconomic factors in wildland arson, the annual models provide results consistent with expectations. Biophysical variables also explain significant variation in annual levels of wildland arson ignitions on national forests.

Variable	Coefficient	Standard Error
Population, t	-7.89E-07	3.11E-07	**
Unemployment Percent, t	4.81E-02	3.05E-02	*
Per Capita Sworn Law Enforcement Officers, t-1	-655	195	***
Pacific Decadal Oscillation, t	0.50	0.09	***
Palmer Drought Severity Index, January-March, t	-0.21	0.05	***
Palmer Drought Severity Index, April-June, t	0.13	0.04	***
Palmer Drought Severity Index, July-September, t	9.1E-02	2.8E-02	***
Palmer Drought Severity Index, October-December, t	-9.8E-02	2.5E-02	***
Nino-3 SST Anomaly, t	-0.43	0.07	***
Wildfire Area, t-1	-2.14E-06	1.35E-06	*
Wildfire Area, t-2	4.78E-06	1.10E-06	***
Wildfire Area, t-3	3.47E-06	1.27E-06	***
Wildfire Area, t-4	-7.21E-06	2.37E-06	***
Wildfire Area, t-5	-1.03E-06	2.44E-06
Wildfire Area, t-6	-2.07E-06	1.49E-06
Wildfire Area, t-7	6.16E-07	1.58E-06
Wildfire Area, t-8	-3.57E-06	1.66E-06	**
Wildfire Area, t-9	-9.96E-07	1.46E-06
Wildfire Area, t-10	-2.44E-06	1.66E-06	*
Wildfire Area, t-11	2.58E-06	1.23E-06	**
Wildfire Area, t-12	4.72E-06	1.42E-06	***
Observations	180
Number of Cross-Sections (18 national forests)	18
Number of Periods (Ten years, 1993-2002)	10
Log-Likelihood Model	-521.96
Likelihood Ratio Statistic, Model Versus Intercept Only (21 d.f.)	1,559.56	***

Note: “***” indicates significance at 1 percent, “**” at 5 percent, and “*” at 15 percent.

Copyright: Public Domain