Skip to main content

Table 3 Estimated W-L model for food consumption: first upper stage

From: Income and food Engel curves in Rwanda: a household microdata analysis

Explanatory variablesEstimated coefficients
Rural householdsUrban households
Log(total household expenditure)0.946***− 0.408
(0.215)(0.533)
[Log(total household expenditure)]2− 0.039***0.010
(0.008)(0.018)
Household size0.012***0.017***
(0.001)(0.002)
Number of adults− 0.008***− 0.006*
(0.003)(0.004)
HH age0.207***− 0.001
(0.062)(0.185)
HH age squared− 0.024***− 0.000
(0.008)(0.025)
Household with non-farm business− 0.014***− 0.021***
(0.002)(0.008)
Monogamy household0.013***0.026***
(0.003)(0.007)
Spouse with no education− 0.022− 0.015
(0.025)(0.026)
Spouse with primary education− 0.0230.003
(0.024)(0.029)
Spouse with secondary education− 0.026− 0.016
(0.023)(0.027)
HH has no education− 0.0010.055**
(0.021)(0.021)
HH has primary education− 0.0110.037***
(0.020)(0.013)
HH has secondary education− 0.0180.013
(0.017)(0.012)
Southern Province− 0.025***− 0.018
(0.005)(0.011)
Western Province− 0.019***− 0.026**
(0.006)(0.012)
Northern Province0.011**0.002
(0.005)(0.017)
Kigali City 0.227***
 (0.013)
Constant− 5.749***4.006
(1.437)(3.969)
Turning point200,1381.30e+ 09
[95% confidence interval][66753–333522][− 2.84e+ 10 to 3.10e+ 10]
Anderson-Rubin test21.54***12.33***
Anderson correlation LR test225.29***32.84***
Observations11,6452502
R-squared0.2540.539
Expenditure elasticities0.76***0.63***
  1. Robust standard errors reported in parentheses are obtained using a clustering approach according to household levels of total expenditures. The Anderson-Rubin test is used to check the joint significance of the endogenous regressors: (Log(total expenditures) and [Log(total expenditures)]2). The Anderson correlation likelihood ratio (LR) test aims at checking whether the two estimated equations are identified, i.e., that the excluded instruments are relevant. The null hypothesis of the test that the equation is under-identified is highly rejected, i.e., implying that the model is identified. The dependent variable for each equation is household expenditure share of food among rural and urban HH. Asterisks “***”, “**”, and “*” denote the significance levels at 1, 5, and 10%, respectively. HH household head