Measuring poverty requires long time periods. Different from other macroeconomic variables like the GDP or the inflation rate of a country, which can be determined immediately and quite precisely at every point in time, it is not that straightforward to measure the level or degree of poverty. If we were to count all people below a certain poverty line at a particular time, we would know only half of the story behind those poor. Some one can fall below the poverty line in one period but climb above it in the next; on the other hand, some one can be persistently below the poverty line. Therefore it is not enough to take only one snapshot of the scenario but one has to take into account that people can be either chronically or transiently poor and that there is a lot of movement in and out of poverty.
illustration not visible in this excerpt
Commonly poverty is measured by looking at consumption of households rather than their incomes. The reason is that income in many cases is only difficult to capture precisely. A self-employed farmer may not have a monetary income but only his harvest, which can be only inaccurately translated into monthly incomes. However, his consumption of food is easy to determine and can also be properly reported. This aspect allows for tracking the households’ poverty level at their different states such that a farmer’s consumption before the harvesting season is most likely to be lower than after and thus his poverty level might change from below the poverty line to above it. This kind of household moves in and out of poverty depending on the season and therefore it is not enough to interview him only once. Figure 1 shows how income can develop over a time period of 5 units. Whereas individual 1’s income is persistently below the poverty line and it experiences permanent deprivation, individual 3 manages to escape poverty after the third period. On the other hand, individual 2’s income rises above the poverty line in period 2 but declines again after the third, which is the typical pattern of transient poverty. However, we do not know for sure what happened after the fifth and before the first period and therefore cannot draw unambiguous conclusions.
Considering the fact that poverty has two faces, one should analyse the shares of people that are chronically and transiently poor, respectively. Not only is this a correct measure of poverty but it also provides crucial information for the policymakers. One can easily draw a comparison to different states of unemployment and its implication for policy. If most people in an economy are short time unemployed then it will need a different policy compared to the case when the majority is long time unemployed. Short time unemployment means a lot of moving between jobs, whereas long-term unemployment represents stagnation in the job market which is a more serious issue and therefore requires other policy actions than short time unemployment (McKay, Lawson 2002). For the same reason it is important to distinguish between chronic and transient poverty in order to give appropriate policy advice.
Chronic versus Transient Poverty
In panel data households were observed several times during several years, whereby particular interest was in their monetary conditions of living standards, as well as weight-for-height as an aspect of malnutrition, both of which are significantly fluctuating in the short term. Additionally, it is also possible to look retrospectively at historical family data like illiteracy and stunting, which are supposed to be less volatile in short term.
Different approaches have been made to figure out the share of chronic and transient poor; one of these is the “spell” approach which analyses the number of spells of poverty a person experiences. According to this approach, Baulch and McCulloch (1998) found out for rural Pakistan that only 3% of the surveyed households were chronically poor in a 5 round period, and 58% were experiencing movement in and out of poverty. For rural South India, Gaiha and Deolalikar (1993) observe 21.8% of households suffering chronic poverty and 87.6% are transient poor. Furthermore, Baulch and Hoddinott (2000) classify the observed households in three poverty levels, namely as “always poor”, “sometimes poor”, or “never poor”. Among the poor, only a small amount of households can be classified as “always poor”, although the probability of being always poor decreases with higher observation periods. Moreover, the result also depends on the study population and varies from 25% for rural South India to only 5% in rural Pakistan. Apart from this, taking income rather than consumption as a measure for welfare may increase transitory poverty because income is more volatile than consumption. As well as this, transitional poor may include households that are poor for only one period but also those that are always poor except one period, therefore the classification of the sometimes poor is not unambiguous. According to the quintile-based measure of poverty movement, a substantially high proportion of the households in the lowest quintile remained there over several periods of time. In the evidence from India, Peru, South Africa and Vietnam, this number of households accounts for over 40%.
Jalan and Ravallion (1998) find in rural China in a five-year period between 1985-1990 that only 7-13% of all households in the sample are chronically poor and also that in provinces with higher average consumption poverty is more likely to be transient. For South Africa this number is substantially higher. Hulme et al. (2002) report Aliber’s estimation for chronically poor households for 2000 of about 18-24% and due to disastrous spreading of AIDS this numbers reached 24-30% in 2010. Moreover, he divided the chronically poor in different categories according to which the majority (1 million) was situated in the rural area, followed by the female-headed households with over 700,000, elderly households with 378,000, and former farm workers (300,000 - 600,000). Also amongst the chronically poor were with a number of 60,000 AIDS families and with 38,000 disabled people.
McCulloch and Baulch (1999) use panel data from 686 households in 52 villages in rural Pakistan in the years 1986 through 1991 from the International Food Policy Reseach Institute (IFRPI). From those households, 400 were experiencing poverty, which can be divided in 295 transitorily poor and 105 chronically poor, and the rest was never poor. All in all, over the whole period the majority (about 75%) of the poor households were transitorily poor, but when looking at separate years, almost half of the poor households were classified as chronically poor.
McCulloch and Baulch (1999) suggest two different policy means to reduce transitory and chronic poverty, namely income growth and smoothing incomes. The effect of income growth on overall poverty, as well as chronic and transient poverty can be seen in Figure 2. By increasing income means by 10 to 50% (shift in the mean), both chronic and transient poverty are reduced significantly.