Unions and Economic Mobility: Is There a Link?

Governor Scott Walker's victory in Wisconsin on Tuesday is bad news for public sector unions, the only growing sector of organized labor.

Wisconsin voters supported Governor Walker because his policies of ending public sector union collective bargaining and increasing public employees' pension and health insurance contributions have resulted in declining budget deficits, unemployment rates, and property taxes in the state.

Nevertheless, an April 2012 Pew study entitled "Economic Mobility of the States" (found here) suggests that Americans have more upward mobility, measured by changes in earnings, in states with a substantial union presence. Conducted by the Pew Center on Economic Mobility, the study has received substantial media attention in past weeks.

For instance, Eric Hoyt and John Schmitt of the union-friendly Center for Economic Policy Research wrote on Tuesday that "there is a strong, positive relationship between the share of a state's workforce that is unionized and the Pew measure of upward mobility."

And the study is listed in the American Federation of State, County, and Municipal Employees Information Highway Web site, a library of resources for union members. It is cited because it bolsters the union case that people are worse off in right-to-work states and better off in states with a substantial union presence.

So, who is right, the Wisconsin voter or Pew? Will Walker's victory reduce economic mobility in Wisconsin, which now ranks "not statistically different from average" in the Pew study?

The Pew study concludes that Maryland, New Jersey, New York, Pennsylvania, Connecticut, Massachusetts, Pennsylvania, Michigan, and Utah, have experienced relatively higher upward and lower downward mobility than other states. California and Oregon have better upward mobility but their downward mobility is about average.

In contrast, nine states in the South, known as right-to-work states because by law workers cannot be compelled to join a union as a condition of employment, have lower upward mobility and higher downward mobility, Pew found. This is why union-friendly groups are interpreting the results as supportive of their agenda of higher taxes and more government spending.

Unfortunately, the study has several important methodological flaws that call into question its results.

The study looks at the mobility of 64,686 Americans in 42 states by calculating their earnings between ages 35 and 39, and later between 45 and 49. It then examines whether these individuals moved up or down the income distribution, both in absolute terms, adjusted for inflation, and relative to others in the state. Individual state sample sizes were not published and are not available from the researchers.

One problem is that the study uses different years for comparisons between different individuals. The authors write, "For the sample as a whole, earnings for 35 to 39-year-olds are measured anytime between 1978 and 1997, while earnings for these same individuals at the ages of 45-49 cover the years 1988 to 2007."

This would be fine if economic growth were steady during the period 1978 through 2007, but growth fluctuated substantially over that period. For a true comparison among the states, the same years have to be used.

Second, the study's results are influenced by migration to parts of the country where people with high levels of education find jobs, such as Washington D.C., New York, and Boston. Therefore, it appears to me that the study is measuring returns to education, with upward mobility greatest in cities where elite college graduates prefer to settle.

University of California (Berkeley) economist Enrico Moretti has written extensively on the higher demand for college graduates in certain geographic areas. For example, finance and high technology industries expanded in New York, Boston, and San Francisco during the 1980s and 1990s. In addition, the movement of educated people to cities results in combinations of skills which increase productivity.

Pew center researchers told me that they account for geographical mobility by comparing the place of birth of a subsample of 48,000 people with their residence at age 40, information derived from the Survey of Income and Program Participation. If the places are the same, they assume no mobility from birth through age 49.

But they do not have data on residences of individuals at ages 35 to 39, or 45 to 49, the years that form the basis of the earnings and economic mobility calculations from the Social Security database. This is because the Social Security data do not include residence. So the study assumes no geographic mobility in those years, in my view an oversimplification.

A third problem with the study is that it excludes those with no earnings in one of ten years, leaving out people who suffer a year's unemployment or women who leave the workforce and then return after taking care of children. This creates a sample that is not representative of the workforce as a whole. In particular, it contains fewer workers with low skills, who have higher rates of unemployment, and fewer women, who are more likely to take time off.

Fourth, the study reports only Social Security wage earnings, which leaves out the self-employed and entrepreneurs, an increasingly important component of the workforce.

Badger State residents voted overwhelmingly to support Governor Walker in reining in the power of public sector unions. Perhaps they know that their economic mobility will be improved by lower levels of government spending, unemployment, and taxes.