Many countries measure poverty using annualized survey data generated on the basis of a one-time “snapshot” of household consumption. Such designs gather information for a single period of reference of 30, 15 or 7 days, collected using either a consumption diary or an interview recall approach. More frequent data collection is certainly possible, but because welfare questionnaires are very long, and it is expensive conduct multiple field visits, reporting poverty dynamics at greater frequency is rare in practice.

Often, this limitation is ignored on the theoretical grounds that consumption, our preferred measure of welfare, can be “smoothed” over time—in other words, that households can accumulate money or goods during times of plenty, and draw down this reserve in times of scarcity. Credit markets can also allow smoothing by enabling households to borrow during times of need, and pay down loans in periods when they enjoy higher incomes. However, this assumption hinges on the existence of well-functioning formal and informal household smoothing mechanisms; often, we do not have the data to allow us to verify that these exist.

Indeed, given market imperfections, the fact that markets often do not even exist in some developing countries, and the inability of poor households to smooth labour and non-labour income fluctuations, there are many reasons that we would expect poverty to fluctuate in the short-term.

Weather patterns and their natural impacts on agricultural output is an example. When harvests are gathered, there is more work available, and more income for those who produce and sell agricultural goods. During seasons in which little agricultural activity takes place however, incomes tend to be lower, reducing consumption. Similar fluctuations are likely present in other economic activities as well, such as construction. In some cases, effects can be compounded when there is significant seasonal labor migration, such as in some countries in Central Asia and Central America. National festivities and religious holidays can also play a significant role, as in the case of Ramadan in Muslim countries.

There are a few instances in which higher frequency data are available however, and these cases can help to get a feel for the story behind annualized poverty rates. Take the national poverty measurement method in Tajikistan for instance – which is based on the country’s Household Budget Survey (HBS) administered by TajStat. Taking advantage of the survey’s continuous quarterly panel design, which generates nationally representative data in each quarter of the year, Tajikistan’s approach allows policy-makers to monitor both seasonally adjusted (annualized) and non-seasonally adjusted (quarterly) poverty trends.

The graph above shows substantial within-year fluctuations around the annual average rate in Tajikistan. Especially in contexts with such strong seasonality, understanding how poverty rates change over short periods of time can be an important element in designing policies that target poor households.

The seasonality trends in Tajikistan also draw attention to the importance of consumption smoothing mechanisms, such as borrowing money from friends or family or from a financial institution. In the absence of functional savings and credit options, households can be subject to large swings in consumption or income, putting them at greater risk of falling into poverty when times are lean.

(João Pedro Azevedo is a Lead Economist at the World Bank in Washington. William Seitz is an Extended Term Consultant at the World Bank in Washington)