Table 1.

Descriptive statistics of the primary data

	Wave 1		Wave 2
	Non-PW	PW	Non-PW	PW
Voluntary labour contributions to community works
Number of workdays in last 12 months on […]
All sectors combined	5.60	6.62	10.69	13.45
	(10.4)	(12.4)	(15.5)	(19.1)
Afforestation	0.05	0.90	0.68	2.03
	(0.5)	(4.5)	(4.5)	(7.3)
Land conservation	0.01	0.30	0.25	0.28
	(0.2)	(2.8)	(2.9)	(2.8)
Nursery/seedling production	0.00	0.12	0.31	1.62
	(0.0)	(1.8)	(3.2)	(6.8)
Road work	0.62	0.58	2.14	2.22
	(3.6)	(3.5)	(7.1)	(6.4)
Non-road construction	3.25	2.62	7.57	7.81
	(7.3)	(6.3)	(13.4)	(13.9)
School-related activities	2.25	2.49	7.11	7.46
	(6.4)	(7.0)	(13.4)	(13.7)
Control variables
Household size	5.32	5.70	5.23	5.77
	(2.3)	(2.2)	(2.2)	(2.1)
Age of household head (in years)	42.20	43.07	43.50	45.29
	(15.8)	(14.7)	(15.2)	(14.6)
Married household head	0.80	0.88	0.77	0.87
Maximum education attained by head or spouse
Primary completed	0.32	0.41	0.32	0.39
Secondary completed or more	0.05	0.10	0.06	0.09
Head or spouse with disability or chronic illness	0.11	0.08	0.28	0.26
Business or wage employment	0.29	0.31	0.26	0.23
Number of seven productive assets owned	0.92	1.12	1.14	1.42
	(1.0)	(1.1)	(1.2)	(1.2)
Number of 14 domestic assets owned	3.46	3.96	3.61	4.41
	(2.0)	(2.2)	(2.0)	(2.2)
Number of months with not enough food, last 12 months	3.63	3.83	2.90	2.30
	(2.7)	(3.0)	(2.6)	(2.4)
Observations	234	266	218	282

The reported values are the means with the standard deviation (SD) in parentheses below for non-binary variables. PWP participants are not necessarily the same across waves because some respondents dropped out of the programme and others newly entered between Waves 1 and 2. The sample size of each column group is reported in the last row. The dependent variables are winsorised at fraction 0.98. This corresponds to how the dependent variables are used in the preferred empirical specifications