Development Methodology

Development Methodology

DEVELOPMENT METHODOLOGY

________________________

LEARN HOW THE A2IE WAS DEVELOPED

T

o create the Asset to Income Estimator (A2IE), we combine information from the Demographic and Health Survey (DHS) or the Multiple Indicator Cluster Survey (MICS) with data from the World Bank to provide an estimate of individual income. Estimations of individual income, using per capita consumption or income data from household surveys, are calculated using the full DHS wealth index and applied to each individual according to their wealth quintile. Our approach follows methodology developed by Kenneth Harttgen and Sebastian Vollmer (Harttgen and Vollmer, 2013).

Additional detail on estimation methods

The methodology for the A2IE was developed by Kenneth Harttgen and Sebastian Vollmer (Harttgen and Vollmer, 2013). They describe how the distribution of incomes in most countries can be approximated by a log-normal distribution, which is statistically re-creatable. To create the A2IE, we used mean per capita income or consumption data and the Gini coefficient from the World Bank poverty surveys to construct the income distribution for an individual. We then overlaid the income distribution onto the asset based wealth rankings from the Demographic and Health Survey (DHS) or the Multiple Indicator Cluster Survey (MICS). We calculated estimated household income by multiplying the estimated individual income at each percentile of the income distribution by the average number of household members in that percentile. Income is expressed in constant 2011 international dollars, which allows for purchasing power parity. This gives the value of a dollar the same purchasing power across countries, not only in the United States.

Accuracy of the Estimations

We compared the World Bank poverty headcount ratio (the proportion of people making less than $1.90/day) with our estimates. Over 50% of our estimates are within 2 percentage points of the World Bank poverty headcount ratio, and over 90% are within 5 percentage points (see below graph). This demonstrates that our methodology is accurate and provides a good estimate of income for each wealth quintile.

Assumptions

There are two assumptions associated with this analysis which should be kept in mind when interpreting the estimated income associated with each wealth quintile. First, there is an assumption that national income distributions are, indeed, log-normally distributed. Lopez and Serven (2006) suggest that this is a reasonable assumption for income data. Second, there is an assumption that the order of individuals in the income distribution is the same as the order of the individual in the asset-based wealth index distribution. There is evidence which supports this assumption (Filmer and Scott 2012), although they found some differences in order in the lowest quintile, where there may be short-term fluctuations in income while asset ownership stays relatively stable.

LEARN MORE

Contact us to learn how Metrics for Management can help you use targeted, accurate, and measurable data to expand and improve your services.