matrices based on recalled self-assessment of poverty by the beneficiary respondents
are used for the purpose of examining dynamic aspect and testing hypotheses. This
qualitative definition of poverty may produce incidence different from quantitative
poverty. The current incidence of self-assessed poverty is compared with income
poverty. It is often argued that income/consumption-based definition of poverty has
the advantage of clearly dividing a population into mutually exclusive categories
however consumption-based definition is usually considered more stable (Lipton and
Ravallion 1995). Our data set contains only a cross-section of income data. In order to
address whether the model facilitates a particular livelihood strategy we rely on asset-
base approach (Siegel 2005, Alwang et al. 2005). This framework assumes that
household welfare results from its livelihood strategies determined by its access to
assets in the given institutional, policy and vulnerability environment. Some of the
assets affect welfare indirectly through livelihood strategies. The relation may be
expressed as follows:
(1) L = f(X, Y)
(2) W = f(X, L)
Where, L represents the vector of livelihood strategy pursued by households, X is the
matrix of assets that affect welfare directly and indirectly, Y is the matrix of assets
that affect welfare only directly and W is a vector of welfare measure. We use
multinomial logistic regression to explain livestock based livelihood strategies in
equation 1. Household welfare is measured by income per person and the equation 2
is estimated using two-stage regression.
3. Poverty transitions
The respondents who were the beneficiary of the BPM were asked to assess their
poverty situation in two points in time- prior to their entry into the programme and at
the time of interview in August, 2006. Their answers in four categories are depicted in
Figure 2 which shows that poverty reduced considerably. Poor and very poor
constituted of 42% before they entered into the programme and the proportion
dropped down to 26% in 2006. Years of entry varied widely; a quarter of the sample
entered the programme during the eighties, more than 60% were beneficiaries in the
nineties. So this is not a contrast between two particular years but before-after
situation of programme participation. The transition is not due entirely to programme
because effects of other sources of development are not separated.
Figure 2.
Before beneficiary
Poor
39%
Very poor
3%
Rich
3%
Middle class
55%
After beneficiary (now)
Poor
24%
Middle class
69%
Very poor
2%
Rich
5%
The transition matrix in Table1 indicates that 67 participants (more than 40% of the
poor) escaped poverty partly due to programme and only 3 out of 234 non-poor