Table 5. Robustness of specifications using difference-in-difference estimation.

Matching DID
Matching on physical characteristics (1)			Matching on physical characteristics (2)			Matching on physical and social characteristics (3)
	DID	SE					DID	SE
PES overlap>2 ha	1.28***	0.17	PES overlap>2 ha	1.24***	0.15	PES overlap>2 ha	1.1***	0.1693211

*** p-value< 0.01.

We provide results of various Difference-in-Difference (DID) estimations. (1),(2) and (3) correspond to OLS estimation of DID for each of our matching specification. We estimate the equation by adding control variables likely to explain conservation outcomes (elevation, slope, years since ejido foundation, ejido size, proportion of commons area, total population, proportion of active population, proportion of people with secondary education, number of right holders, distance to nearest city and marginality index. (4), (5) and (6) estimate using a DID in panel version using a fixed-effect. We add variables in panel version (marginality index, population, proportion of active population and proportion of population with secondary education).