Table 4. Sensitivity of ATT to matching parameters and thresholds.

	Control with matching on physical characteristics in treated ejidos (A)					Control with matching on physical characteristics in non-treated ejidos (B)					Control with matching on physical and socioeconomic characteristics in non-treated ejidos (C)
	Minimum forested area (ha)	0.1		1		Minimum forested area (ha)	0.1		1		Minimum forested area (ha)	0.1		1
	Minimum PES overlap (ha)	0.5	2	2	8	Minimum PES overlap (ha)	0.5	2	2	8	Minimum PES overlap (ha)	0.5	2	2	8
	n treated	1,646	1413	1413	707	n treated	1,646	1413	1413	707	n treated	1,646	1413	1413	707
Match 1:1 (1)	n control	939	826	805	473	n control	1309	1120	1111	601	n control	888	793	761	435
	ATT	1.00	1.11	1.11	1.00	ATT	0.96	1.09	1.10	1.20	ATT	0.78	0.91	0.90	0.89
	Variables with bias <20%	9	9	9	9	Variables with bias <20%	7	8	10	10	Variables with bias <20%	10	10	12	10
Best matches(2)	n control	894	796	776	473	n control	1259	1086	1070	594	n control	859	773	742	431
	ATT	1.06	1.14	1.10	0.97	ATT	0.94	1.02	1.03	1.15	ATT	0.84	0.94	0.91	0.79
	Variables with bias <20%	12	12	11	9	Variables with bias <20%	7	7	7	10	Variables with bias <20%	9	10	11	12
Match 2:1(3)	n control	1486	1315	1273	783	n control	2291	1997	1953	1088	n control	1374	1237	1208	738
	ATT	1.09	1.17	1.21	1.04	ATT	0.98	1.11	1.13	1.19	ATT	0.76	0.89	0.87	0.82
	Variables with bias <20%	11	11	11	11	Variables with bias <20%	7	7	7	10	Variables with bias <20%	9	10	11	10

We provide an estimation of the Average Treatment effect on the Treated (ATT). We provide statistics on the performance of matching using the two specifications. We provide variations of results according to different matching parameters. (1) is a 1:1 matching based on Mahalanobis distance with replacement. (2) only considers the best 95% matches of the 1:1 matching. (3) corresponds to a 2:1 matching with replacement. We provide different results by varying the minimum forest area and the minimum overlap with PES polygons.