Most studies examining the dynamics of welfare have found large fluctuation
s in consumption over relatively short periods, suggesting substantial shor
t-run movements in and out of poverty. The consequence is that cross-sectio
n poverty research may not be able to identify the pool: In this study, we
explore this short-run variability further: We use a data set on a panel of
1450 households in different communities in rural Ethiopia, surveyed thric
e, over 18 months. On average year-to-year poverty is very similar However
we find high variability in consumption and poverty, over the seasons and y
ear-by-year Econometric analysis suggests that consumption is affected by i
diosyncratic and common shocks, including rainfall and household-specific c
rop failure, while households respond to seasonal incentives related to cha
nging labour demand and prices. The results imply that a larger number of h
ouseholds are vulnerable to shocks than implied by the standard poverty sta
tistics, while some of the non-poor in these statistics are in fact otherwi
se poor households temporally boosting their consumption as an optimal resp
onse to changing seasonal incentives.