Table A1

Probit Estimation of Marginal Probability of Remaining in the Next Round of Panel Survey

Panel Period
Variable (dependent variable: Dummy for remaining in the panel)1997-20002000-20042004-2007
Adult equivalents in the initial period-0.003 (1.32)0.002 (0.77)-0.014 (1.13)
Rented-in land in the initial period-0.010 (0.87)0.011 (0.70)0.003 (1.16)
Members < 14 in the initial period0.016*** (4.42)- 0.002 (0.65)0.009 (0.23)
Members > 60 in the initial period0.018** (2.32)0.008 (0.75)0.003 (0.49)
Female-headed household in the initial period-0.024** (1.69)- 0.020 (0.95)0.006 (0.57)
Household head with primary education-0.016 (1.61)- 0.001 (0.05)-0.002 (0.17)
Household head with secondary education0.017 (1.55)-0.014 (0.80)- 0.007 (0.62)
Secondary education as the highest education of0.002 (0.17)0.025 (0.79)0.012 (0.58)
household members
Productive assets in the initial period0.004*** (2.85)0.002 (0.95)- 0.001 (0.97)
Value of durable consumer goods in the prior period- 0.001 (0.78)0.006** (2.44)0.001 (0.33)
Share of income from livestock production0.001 (0.05)- 0.001 (0.77)0.012 (1.38)
Share of income from off-farm activities- 0.009 (0.63)- 0.007 (0.50)-0.010 (0.73)
Land endowment in the initial period-0.001 (1.61)0.005** (2.03)0.003** (2.02)
Distance to tarmac road (km)0.001 (1.28)- 0.001 (0.88)-0.001 (1.55)
Distance from extension advice (km)- 0.001 (0.72)0.002** (1.85)0.000 (0.27)
Survey team 2-0.056 (1.49)0.034 (1.57)0.020 (0.83)
Survey team 3-0.032 (1.44)0.048*** (3.29)-0.050 (1.47)
Survey team 4- 0.057** (2.49)0.026 (1.19)0.006 (0.41)
Predicted probability of remaining in the panel0.950.930.96
Observations1,5231,4391,333
  • Note: Robust z-statistics in parentheses. Provincial and agroecological zone dummies are included in all the regressions. Clustering effects at the village level are controlled for in all the regressions.

  • * Significant at 10%;

  • ** significant at 5%;

  • *** significant at 1%.