Most of studies in poverty analysis focused on analyzing poverty at a point in time or the trend over time based on cross sectional data. The drawback of such approach is that it measures poverty on a given date without distinguishing between those that are chronically poor due to low asset base and those that are transiently poor due to shocks, i.e., a household may be poor because it is not able to generate sufficient income to meet the basic needs due to lack of assets (physical as well as human) under normal circumstances or a household may be poor due to negative shocks like temporary layoff of the major income earner of the household due to illness from the labour market, bad agricultural weather condition, which pulls down the mean income or consumption of the household below poverty line for that particular period or vice versa. This might lead to the wrong policy direction, since the policy required for addressing the needs of transient poor is different from the chronic ones.
Indeed there is little research based on panel data that tries to investigate the poverty dynamics in Ethiopia in the 1990s. Exceptions are series of papers by Dercon and Krishnan (1998, 2000, and 2001), Hagos and Holden (2003), Bigsten and Shimeles (2003), Swanepoel (2005) used panel data in poverty analysis.
The analysis of poverty dynamics distinguish between the transient and chronic poverty or the exit, entry and re-entry into poverty. This means that there is a chance that a household that is not poor becomes poor, one that is poor remains poor. This leads to vulnerability assessment in terms of ‘vulnerability to poverty’. ‘Vulnerability to poverty’ is the probability that a household will be poor next period. So it is an important concept to deepen the understanding of poverty, since it reveals information on what measures should be taken to prevent poverty while poverty dynamics largely imply the past and contemporary poverty situation and is helpful on how to alleviate the existing situation.
Using three round rural household Tigray panel data for 1997, 2000 and 2003, this thesis examined poverty dynamics, vulnerability to poverty of households. It also analyzed the determinants of poverty dynamics and correlates of vulnerability to poverty.
Methodologically, the thesis used the Foster-Greer-Thorbecke (FGT) poverty measures in order to measure poverty and both component and spells approach to decompose poverty into chronic and transient. Three econometric models were used in this thesis. One, a fixed effects regression used to analyze the determinants of the poverty dynamics. In the fixed effects regression, the poverty dynamics captured using the entity demeaned value of log per adult equivalent consumption expenditure and set of household characteristics used as explanatory variables. Two, to identify the factors that affect the likelihood of being chronic and transient to non-poor, I estimated a multinomial logit model. Three, to estimate the vulnerability to poverty of a household I adopted the vulnerability measure in Chaudhuri (2003). In order to estimate this measure I followed a three step Feasible Generalized Least Squares (3FGLS) to estimate the expected log per adult equivalent consumption expenditure and variance log per adult equivalent consumption expenditure. Using these estimates and assuming that per adult equivalent consumption expenditure is log normally distributed I estimated the vulnerability measure as the probability that the standard normal variate will fall below standardized poverty line.
The main findings of this thesis are as follows: Firstly, poverty in rural Tigray is chronic, however, there is some evidence of dynamics in the rural poverty as one can infer from the transient component of poverty. Secondly, the fixed effects regression identifies that household’s farm size improve significantly welfare of the household while the number of children, juniors and adults in the household found to adversely affect household’s welfare. Thirdly, the explanatory variables for chronic and transient poverty are the same, where the household’s head age and the numbers of children in the household increase the probability of being chronic and transient poor compared to non-poor. Farm size and off farm participation of the head of the household reduce the likelihood of becoming chronic and transient poor. Lastly, there is high vulnerability to poverty, i.e. high probability of becoming poor in a period ahead, in the region and it is significantly correlated with household head age and education, household size, asset ownership and number of seniors (or elderly) in the household.
The thesis suggests two different policy interventions. One, since poverty is chronic in rural Tigray the policy interventions in the region should focus on assisting poor households to accumulate assets through increased investment and employment generation that enhances their mean consumption level. Two, the evidence that there are more vulnerable households’ calls for policy intervention that reduce consumption variability through reducing exposure to risk or improving the ex post coping mechanisms of the vulnerable