A common refrain about economic policy and performance in the United States over the past thirty years is that inequality is rising and the middle class is disappearing.  The implication of the latter claim is that people are falling out of the middle class into the lower income class.  Of course, if you define “middle class” to be the middle part of the income distribution, centered at the median, then there always will be a middle class.  So, in order to be meaningful, “decline of the middle class” is usually taken to mean flat or falling income for the median household.  Combining this with strongly rising income for the top 10% or 1% of the income distribution leads to the conclusion that inequality is rising.

Economists Thomas Picketty and Emmanuel Saez (PS) are among the leading scholars in this field and have documented large increases in the share of income at the top of the income distribution.  Using IRS data extending back to 1914, PS show that inequality as measured by the income share at the top of the distribution fell through the early 20^th century, but then flattened out in the 1980s and has been rising sharply ever since.  One reason for this result is rapid growth in incomes at the top of the income distribution.  But this does not necessarily mean that the middle class is declining.  To investigate this we need to look at how the middle of the income distribution is faring.  On this point, PS estimate very modest income gains for the median household; in particular, their methodology shows a real income gain of 3% for the median household over the past thirty years (1979-2008).  This compares to an income gain of 50% for the top decile household for the same period and strongly supports the notion of rising inequality.

As noted in a prior note (May 22, 2012), it is interesting and important to examine reasons for these income dynamics.  Partly, these dynamics are due to a heightened return to education and a growing college wage premium.  Another part of the explanation is poor education opportunity for young people in many parts of the country.  Surely it is a worrisome trend if the financial position of the average American is flat or declining.

Yet, it may be that the PS conclusions about the median household are not correct.  There is no doubt that PS are highly respected researchers who have been meticulous in their scrutiny of the data.  However, assessing trends in the income distribution is very complex.  Economist Richard Burkhauser (RB) has spent the last 15 years studying similar data.  His source is Census data (based on the annual Current Population Survey) and he is able to replicate certain features of the PS data, in particular the 3% real income gain for the median “tax unit” (this is the observation unit utilized by PS) over the past 30 years. 

However, RB argues that the PS analysis overlooks or ignores several critical issues including changes in the number of the people in the average household (this has been falling for several decades) and the effects of progressive taxes and transfer payments (the PS analysis looks at pre-tax income before transfer payments).  After making suitable adjustments (which he is able to do with the Census data), RB finds that the median worker net income (including transfer payments) has increased by 32% in real terms between 1979 and 2008.  This is close to the long-term increase in the economy-wide per capita standard of living, and therefore does not support the notion of a declining middle class.  This conclusion is supported by studies that examine consumption patterns.  The consumption “basket” of the poverty line family in 2010 appears to be greater than the consumption basket of the median family in 1970. 

It is disconcerting when equally competent researchers come up with dramatically different answers to what appears to be a straightforward question – what is happening over time to income for the median household?  Actually, the question is not all that simple.  In fact, I don’t think either PS or RB really address the issue.  That is because they don’t look at the same households over time.  If you really want to see what is happening to people over time, you have to follow the same individuals!  This requires the use of so-called panel data, where the same “panel” of individuals is followed over time.  The longest running and most comprehensive such data set is the Panel Study of Income Dynamics (PSID).  Using this data, a recent study by the Brookings Institute shows that the median income household in 1970 has enjoyed a real income increase of 29% in one generation (that is, the children of the median household in 1970 earned 29% more in real terms than their parents).

While not the identical question addressed by PS or RB, the Brookings finding does suggest that the middle class has been doing reasonably well over the past thirty years or so.