Table 8.

Specialization Pattern Across Non-Contiguous Partitions of the Same Language Group

	(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES	Pasture		Maize		Sorghum
Pasture Across Buffer Regions	0.749*** (0.094)	0.420*** (0.137)
Pasture in the non-Adjacent Partition of the Same Group		0.307*** (0.161)
Maize Across Buffer Regions			0.945*** (0.018)	0.881*** (0.050)
Maize in the non-Adjacent Partition of the Same Group				0.540*** (0.144)
Sorghum Across Buffer Regions					0.926*** (0.023)	0.853*** (0.059)
Sorghum in the non-Adjacent Partition of the Same Group						0.367*** (0.135)
Mean Land Quality	−0.102** (0.039)	−0.059** (0.028)	0.064 (0.206)	0.055 (0.182)	0.154 (0.155)	0.120 (0.149)
Observations	418	418	372	372	372	372
Log pseudolikelihood	−40.581	−30.356	−54.667	−43.563	−65.098	−57.339

The unit of analysis is a partition of a language group. Standard errors in parenthesis clustered at the language group level. All columns are estimated by maximum-likelihood Probit models.

^***p < 0.01;

^**p < 0.05;

^*p < 0.1

The coefficients report the change in the probability of the dependent variable induced by a change in the respective regressor. In (1) and (2) the dep. var. is an indicator that takes the value 1 if pasture dominates land use in the partition. In (3) and (4) the dep. var. is an indicator that takes the value 1 if maize is the dominant crop in the partition. In (5) and (6) the dep. var is an indicator that takes the value 1 if sorghum is the dominant crop in the partition. See web Appendix for variable’s definitions.