Remittances and Household Investment Decisions: Evidence from sub-Saharan Africa

[i] Notes: This table presents the mean and standard deviation of the key variables.

[ii] Standard deviations are presented in parentheses.

Notes: This table presents the first stage estimates of our instrumental variable estimation. Robust standard errors are presented in parentheses. The SW F-statistics is a test of weak identification with a null hypothesis that the endogenous regressor is weakly identified. The Hansen's J-statistics are the test statistics from the Sargan–Hansen test of overidentifying restrictions. The joint null hypothesis is that the instruments are the valid instruments. The outcome variable in all columns is an indicator that equals one if a household received remittances.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

SW, Sanderson–Windmeijer.

Notes: This table reports the average marginal effects from the probit and recursive bivariate probit models. Robust standard errors are presented in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is above 60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

Notes: This table reports the estimates of OLS, 2SLS, and IIV estimation. Robust standard errors are presented in parentheses. IIV estimation bounds are reported in square brackets and corresponding confidence intervals are reported in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

IIV, imperfect instrumental variable; IV, instrumental variables; OLS, ordinary linear regression.

Notes: This table reports the average marginal effects for probit and IV-probit models. Robust standard errors are presented in parentheses. The variable of interest is the amount of cash remittances received (in log scale) in the last 12 months. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

IV, instrumental variables.

Notes:

This table reports OLS and 2SLS estimates. Robust standard errors are presented in parentheses. Outcome variables are capital investment expenditure in US dollars (in log scale) in the past 12 months. The variable of interest is the amount of cash remittances received in US dollars (in log scale) in last 12 months. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

OLS, ordinary linear regression.

Notes: This table reports the average marginal effects for recursive bivariate probit models. Columns 1–4 show estimates for households with a migrant household member and columns 5–8 show estimates for households with no migrant household member. Robust standard errors are presented in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

Notes:

This table reports the average marginal effects for recursive bivariate probit models. Robust standard errors are presented in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

Notes: This table reports the average marginal effects for recursive bivariate probit models. Robust standard errors are presented in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban. Estimates of columns 1 and 2 are conditional on households investing on physical or social capital. Similarly, estimates of column 3 are conditional on households investing on human or social capital. Finally, column 4 is conditional human or physical capital investment.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

Notes: Columns 1–3 report the average marginal effects for recursive bivariate probit models and columns 4–6 report the estimates of 2SLS regression. Robust standard errors are presented in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

[i] Notes: Number of observations in Panel-A, 1,603; in Panel-B, 1,821; in Panel-C, 2,030; in Panel-D, 1,895; and in Panel-E, 1,705.

[i] Notes: This table presents the mean and standard deviation of the key variables. Standard deviations are presented in parentheses.

Notes: This table reports the coefficients of our probit and recursive bivariate probit models. Robust standard errors are presented in parentheses. The variable of interest, received remittances, is an indicator that takes one if a household received remittances, and zero otherwise. Outcome variables are also indicator variables that equal one if a household made capital investment, and zero otherwise. Control variables are female household head, head is a paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.

Notes: This table reports 2SLS estimates. Robust standard errors are presented in parentheses. Outcome variables are capital investment expenditure in US dollar (in log scale) in past 12 months. The variable of interest is amount of cash remittances received in US dollar (in log scale) in last 12 months. Columns 1–4 includes amount of cash remittances received in US dollar (in log scale) squared. Columns 5–8 includes an indicator of high remittance (above district-level average remittances) and an interaction between cash remittances received (in log scale) and high remittances. Control variables are female household head, head is paid employee, head is self-employed, head has secondary education, head has above secondary education, head's age is 45–60 years, head's age is >60 years, log household income, number of children in the household, number of elderly in the household, and location is urban.

*** p < 0.01,

** p < 0.05,

* p < 0.10.