Table 4.

Regression analysis for cooking fuel expenditures (in RWF): past 30 days1,2

Dependent variables:	Cooking fuel purchased	Charcoal purchased for cooking
DiD coefficient (treatment effect)	0.28 (0.39)	−1.02* (0.60)
FE coefficient	0.08 (0.38)	−1.23*** (0.44)
Observations	288	288

Robust standard errors in parentheses.

^***p<0.01,

^**p<0.05,

^*p<0.1

^1.Household characteristics controlled for: household size, baseline stove used in past 30 days, number of durable goods, ownership of land, log of per capita baseline total expenditure in last 4 weeks (in RWF), log of per capita baseline cooking fuel expenditure in last 4 weeks (in RWF), log of per capita baseline hygiene expenditure in last 4 weeks (in RWF). Traditional stoves is the referent category for 'stove used in the past 30 days: baseline'.

^2.Primary cook characteristics controlled for: hired, female. Household head characteristics controlled for: female, age (in years), married, secondary and above education level. TVET and levels below (primary, pre-primary and no education) is the referent category for 'secondary and above education level of household head'.