Engel Coefficient

THE ENGEL COEFFICIENT is the proportion of family income that is spent on food. It received its name in honor of German statistician Ernst Engel (1821–96).

In a famous study using the budget data of 153 Belgian families, Engel found that the lower a family's income, the greater is the proportion of it spent on food. Later studies have confirmed this empirical regularity, which has been called Engel's Law. According to H.S. Houthakker (1957), “of all the empirical regularities observed in economic data, Engel's Law is probably the best established.”

Engel's Law is a ceteris paribus relationship. That is, it holds when prices and preferences are kept constant. An alternative way to enunciate Engel's Law is by saying that the income elasticity of the demand of food is smaller than one.

The Engel coefficient plays an important role in several methods for the estimation of the poverty line. Since the seminal study by B. Seebohm Rowntree, the Cost of Basic Needs (CBN) approach has been one of the most commonly used methods to estimate poverty lines. According to this methodology, poverty is the lack of some basic consumption needs, and the poverty line is the cost of these needs. To implement this method, we should stipulate a basic consumption bundle and then calculate its cost using information on prices. The implementation of this method poses some practical problems. In particular, the choice of the basic needs bundle is quite arbitrary, especially for nonfood products.

However, objective nutritional requirements can provide a defensible anchor to define a bundle of basic food products. For instance, the food poverty line could be the cost of buying 2,000 calories per person per day. However, we are typically interested in a poverty line for total consumption or income, and not just in a food poverty line.

An approach that has been used in many poverty studies, for example by G.M. Fisher, is to estimate the poverty line as is the average Engel coefficient for the group of households with income below percentile p, where p is typically between 25 percent and 50 percent. As argued by M. Ravallion and B. Bidani, this approach can provide biased estimates of the poverty line. Ravallion and Bidani propose to use household-level data to estimate an Engel curve for food consumption and use the estimated curve to calculate the poverty line.