TABLE 7

Effect of Renting Land on Household Per Capita Total Income and Net Crop Income (Dynamic Panel Regression)

	Per Capita Total Income			Per Capita Net Crop Income			Per Capita Off-Farm Income
	OLS	IVGMM	IVGMM	OLS	IVGMM	IVGMM	OLS	IVGMM	IVGMM
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Log of dependent variable	0.176^***	0.065^**	0.063^**	0.186^***	0.094^***	0.068^**	0.192^***	0.092^***	0.098^***
	(7.90)	(2.02)	(2.03)	(7.89)	(2.91)	(2.03)	(8.03)	(2.87)	(2.96)
Rented-in dummy ( = 1)	0.129^***	0.159^**	0.178^**	0.246^***	0.318^***	0.331^***	- 0.024	- 0.072	- 0.087
	(3.69)	(2.02)	(2.14)	(4.97)	(3.92)	(4.11)	(0.14)	(0.21)	(0.26)
Log of total agricultural assets	0.193^***	0.081^***	0.079^***	0.123^***	0.062^***	0.071^*	0.348^***	0.002	0.003
	(14.39)	(3.96)	(2.69)	(7.16)	(2.78)	(1.88)	(6.16)	(0.03)	(0.02)
Log of adult equivalents	-0.496^***	-0.560^***	-0.703^***	-0.421^***	— 0.587^***	-0.584^***	-0.400^***	-0.119	- 0.488
	(14.33)	(10.12)	(5.82)	(9.40)	(10.39)	(3.80)	(2.84)	(0.55)	(0.84)
Land endowment per capita	0.246^***	0.207^***	0.353^***	0.390^***	0.319^***	0.494^***	- 0.053	- 0.068	0.104
	(8.31)	(5.68)	(5.83)	(8.87)	(10.28)	(7.61)	(0.56)	(0.61)	(0.47)
Log of Iwestock value	0.058^***	0.056^***	0.055^***	0.035^***	0.028^***	0.021^***	- 0.000	0.020	0.037
	(12.77)	(10.22)	(8.57)	(8.83)	(5.49)	(2.82)	(0.02)	(1.20)	(1.47)
Female-headed household ( = 1)	— 0.174^***	- 0.070	-0.313	-0.100^**	-0.152	-0.815	-0.445^***	-0.678^**	- 2.024
	(4.22)	(0.63)	(0.63)	(2.09)	(1.29)	(1.39)	(3.20)	(1.98)	(0.90)
Death of household head since prior survey ( = 1)	0.037 (0.58)	0.068 (0.59)	0.260 (0.59)	0.057 (0.65)	0.026 (0.18)	0.609 (1.20)	0.071 (0.27)	0.574 (1.17)	1.676 (0.85)
Log of household head’s age	3.1O3^**	2.425	5.133^*	1.801	8.920^***	12.790^***	6.534	-3.701	5.303
	(2.59)	(1.03)	(1.78)	(1.00)	(2.80)	(3.19)	(1.11)	(0.36)	(0.38)
Log of household head’s age squared	-0.412^*** (2.71)	- 0.268 (0.88)	-0.625^* (1.66)	-0.208 (0.91)	— 1.129^*** (2.78)	- 1.648^*** (3.15)	-0.931 (1.26)	0.473 (0.36)	- 0.728 (0.39)
Household head completed primary school ( = 1)	0.139^*** (3.45)	0.159^*** (2.69)	0.134^** (2.08)	0.077^* (1.91)	0.015 (0.24)	- 0.029 (0.42)	0.498^*** (3.27)	0.191 (0.82)	0.129 (0.50)
Log of rainfall	0.042	-0.155	0.145	0.933	1.346	1.603	1.990	1.378	2.401
	(0.06)	(0.16)	(0.14)	(1.55)	(1.08)	(1.24)	(0.75)	(0.26)	(0.45)
Log of rainfall (lagged)	0.601	0.339	0.703	-0.486	- 0.029	0.234	1.340	1.081	1.973
	(0.84)	(0.43)	(0.83)	(0.75)	(0.02)	(0.18)	(0.39)	(0.19)	(0.34)
Distance to nearest piped water source	- 0.006^* (1.89)	- 0.005 (1.37)	- 0.005 (1.49)	- 0.005 (1.24)	- 0.002 (0.64)	- 0.003 (0.94)	-0.013 (0.96)	0.004 (0.34)	0.004 (0.32)
Distance to the nearest	0.003	0.007	0.009	0.001	0.001	0.002	0.022	0.037	0.038
electricity supply	(0.43)	(1.02)	(1.41)	(0.35)	(0.14)	(0.37)	(0.84)	(1.39)	(1.36)
Distance to extension services	- 0.002	0.001	0.001	0.003	0.005	0.005	-0.035^*	-0.033^*	-0.033^*
	(0.37)	(0.20)	(0.16)	(0.61)	(1.18)	(1.07)	(1.80)	(1.73)	(1.71)
Distance to main road (km)	0.005	- 0.003	- 0.002	0.012	-0.023^*	-0.019	0.004	0.045	0.045
	(0.40)	(0.20)	(0.13)	(0.87)	(1.79)	(1.43)	(0.06)	(0.81)	(0.80)
Credit constrained ( = 1)	- 0.023	0.046	0.036	-0.127	- 0.063	- 0.098	0.051	- 0.002	- 0.026
	(0.32)	(0.51)	(0.36)	(1.41)	(0.61)	(0.84)	(0.17)	(0.00)	(0.06)
Hanson J overidentification test (p-value)	—	0.14	0.21	—	0.83	0.73	—
Test for autocorrelation (p-value)^{^a}	—	0.13	0.13	—	0.53	0.53	—
R-squared	3,607	2,381	2,381	3,513	2,290	2,290	3,620	2,394	2,394
Observations	0.62			0.64			0.25

Note: Absolute value of i-statistics in parentheses. All standard errors adjusted for clustering effect at the village level. Village dummies were added to control for local fixed factors that affect the change in income, time dummies were included to control for time trend of income growth, and interactions of village and time dummies were included to control for any policy change over time that may affect the income. The inverse probability weighting approach was used to account for potential attrition bias in all the regressions. Columns 1 and 4: Pooled OLS regression with village dummies and year dummies included. Columns 2 and 5: First differencing equation, and and are treated as endogenous variable with the former being instrumented by Y_ijt-2 and the latter being instrumented by R_ijt- 1 and R_ijt-2 Columns 3 and 6: First differencing equation, and and are again treated as endogenous and instrumented the same way as for Columns 2 and 5. Δ(Log of value of agricultural assets)y Δ(Land endowment)_z, Δ(Log of livestock value)_i, and Δ(Female head)_i are all treated as predetermined, and they are instrumented by the level of themselves lagged by one period and lagged by two periods (i.e.,¾~ ¡ and x_t~ 2)·
↵a The autocorrelation of the original error term in the level equation is tested using Wooldridge’s (2002) Xtserial command in the Stata program. An alternative and more commonly used option is the test for the second-order autocorrelation of the error term in the first differenced GMMfV estimation equation. This is not possible in a four time period panel.
↵* Significant at 10%;
↵** Significant at 5%;
↵*** Significant at 1%.