M4. If median income is better than GDP and the HDR, why doesn’t anyone use it?

2014/05/23 by Jonathan Andreas

Median income is such a simple and feasible improvement over GDP/GNP (as a measure of welfare) that I am mystified why it had not become a standard measure long ago and I have been researching why nobody else has been promoting the idea.  I think the main reason is the dominance of mmutilitarian ontology.  The earliest paper that I have found that briefly suggested both components of MELI was written by Norman Hicks & Paul Streeten (1979, p. 569), but they did not combine the two components into a composite measure.  Instead, they rejected both ideas and mischaracterizing median income as a mere “adjustment to GNP” which they said lacks a “logical basis” and tends instead to “result in a confusion of concepts” (1979, p. 577).  Instead, they favored creating a complex new composite of ‘social and human indicators’ that would combine life expectancy and basic needs.

Hicks & Streeten characterized median income as just an adjustment of mean GNP, but median income is not simply an adjustment to mean GNP.  GNP is not really a measure of income at all.  GNP is a measure of production.  It should be quite similar to income, but it is not exactly the same and to measure median income requires gathering completely different data. After they rejected MELI, Hicks & Streeten advocated a composite measure of human indicators beginning with life expectancy.  MELI is also a composite measure of human indicators that begins with life expectancy, like Hicks & Streeten advocate, and it has two enormous advantages over their approach. First, MELI is already feasible to calculate using existing data for many countries, and the data is relatively easy to collect compared with most of the kinds of data Hicks & Streeten desired.  Secondly, MELI produces something that Hicks & Streeten say is important (1979, p. 577).  MELI is “a single number which, like GNP per head, can be quickly grasped and gives a rough indication of ‘social’ development.”  Hicks & Streeten’s work on an alternative indicator of welfare was too utopian to go anywhere.  In the best light, their work can be seen as part of the ongoing effort that ultimately produced the HDI, but the HDI is also inferior to MELI.

Hicks & Streeten almost invented MELI, but they didn’t because they were too utopian. They hoped to invent a much more perfect measure of wellbeing which wasn’t feasible.  The HDI also came out of a utopian effort to create a measure of human capabilities, and this led to an excessively complex measure that is no closer to the original utopian inspiration to measure capabilities than MELI is.  Most efforts to replace GDP have been too utopian which has led to such complexity that the efforts have either produced nothing or at best they have produced something like the HDI which is less useful and accurate than a simpler measure like MELI.

People haven’t used MELI because of its location between the utopian imaginations that lead to excessive complexity and the failure of imagination due to mutilitarian ontology which has produced a long history of investing in infrastructure for GDP that gave it momentum down this path dependence.  The failure of imagination should have been clear in the 1930s when historians of economics say there was an “ordinal revolution” in economics.  They claim that this was when economists came to accept that money can only give an ordinal measurement of utility (welfare).  One problem with this historiography is that economists adopted mutilitarianism as a result of the so-called “ordinal revolution” and mutilitarianism adds up cardinal money-metric mutility to measure welfare.  That is how economists came to revere mean GDP ( average income).  If there had really been an ordinal revolution economists would not have accepted cardinal mutility.  If they had believed that welfare is ordinal, they would have sought an ordinal measure and a median is the appropriate statistic for ordinal data, not a mean.  A mean of ordinal data is meaningless.  The “ordinal revolution” is a misnomer that mutilitarians began promoting as Whiggish historical propaganda to hide the philosophical shortcomings of mutilitarianism.  It should be called the mutilitarian revolution because it supported a huge investment in measuring mean GDP rather than the development of an ordinal statistic like MELI.

There have been hundreds of groups that have tried to create a better measure of economic welfare than GDP.  Many of these groups point out that GDP is misleading because GDP can rise even when median income falls.  For example, the Sarkozy Commission’s report makes this argument (Stiglitz, Sen, & Fitoussi, 2010, p. xi) and it uses median income to show that mean income is an incorrect measure (2010, pp. xxi, 47).  Although the authors use median income as an example of the right way to measure income, they dismiss median income as being too difficult to measure (2010, p. 45)!  They do not not even consider the possibility of using it to replace GDP because they are too utopian.

There is some truth to the claim that median income is difficult to measure, but again it reveals a failure of imagination.  In many ways, MELI is easier to measure than GDP, but very few resources have been invested in measuring median income.  For example, the United States Bureau of Economic Analysis produces American GDP statistics and it had a $96m annual budget in 2013 and over 500 employees partly because GDP is also very difficult to measure.  But we surmounted the tremendous difficulties of measuring GDP and the BEA releases detailed, up-to-date GDP estimates four times a year.

In contrast, the Census Bureau publishes median income data only once a year as a tiny, almost accidental part of its overall mission and the data is always more than a year out of date.  For example, as I write this, halfway through 2014, the government has only published information about 2012 median income.  Median income data is so neglected that a private firm, Sentier Research, has been helping fill the void in the official statistics to produce better and more up-to-date median income statistics.  The government is neglecting median income if Sentier Research, a tiny private organization with only a handful of employees, can produce better median income statistics in their spare time than the government is doing.

The biggest problem with measuring median income is a failure of imagination to invent better methodologies for measuring it.  MELI is a strong framework for improvement.  To understand how to improve median income, we must first understand why (almost) nobody currently uses median income measures.


Hicks, N., & Streeten, P. (1979). Indicators of development: The search for a basic needs yardstick. World Development, 7(6), 567–580. doi:10.1016/0305-750X(79)90093-7

Stiglitz, J. E., Sen, A., & Fitoussi, J. P. (2010). Mismeasuring Our Lives: Why GDP Doesn’t Add Up. The New Press.