It is widely recognized that income inequality increased in the 1990’s, but nobody knows quite why. Despite the lack of hard evidence, there are plenty of theories.
One says the culprit was declining unionization. Another ties it to immigration and outsourcing. A third theory is that the demand for high-level cognitive skills has increased, while other explanations range from changes in executive compensation to the lack of policy initiatives directed toward the working poor.
All these factors may have contributed to the increase in income inequality, but there is little solid evidence about their relative importance.
Two University of Texas researchers, James K. Galbraith and Travis Hale, added an interesting twist to this debate in a paper, “Income Distribution and the Information Technology Bubble” (utip.gov.utexas.edu/abstract.html#UTIP27).
According to Mr. Galbraith and Mr. Hale, much of the increase in income inequality in the late 1990’s resulted from large income changes in just a handful of locations around the country — precisely those areas that were heavily involved in the information technology boom.
Their study used data on average income and population by county available from the Bureau of Economic Analysis, available at bea.gov/bea/regional/reis. Unlike most other studies, their work does not examine inequality among individuals, but rather differences in average income across counties.
The authors compute a measure of inequality known as the Theil index, named for the econometrician Henri Theil. They present their results in two ways. The most visually arresting is an animated map at utip.gov.utexas.edu, which shows how per capita income by county has changed over time. The other depiction is a simple plot of the Theil index from 1990 to 2000.
The Theil index exhibits the standard shape found in other income inequality studies: income inequality was flat in the first half of the 1990’s, then rose sharply in the second half. After 2000, the inequality index declined again.
A big advantage of looking at county data is that it is possible to identify counties that contributed the most to the increase in income inequality from 1994 to 2000. It turns out that the five biggest winners in this period were New York; King County, Wash. (with both Seattle and Redmond); and Santa Clara, San Mateo and San Francisco, Calif., the counties that make up Silicon Valley. The five biggest losers were Los Angeles; Queens; Honolulu; Broward, Fla.; and Cuyahoga, Ohio.
What do the counties in the first list have in common? Their economies were all heavily driven by information technology in the late 90’s. This is true for the rest of the list of winners as well. Harris, Tex. (home to Houston and Enron); Middlesex, Mass. (home to Harvard and M.I.T.); Fairfield, Conn.; Alameda, Calif.; and Westchester, N.Y., were also among the top 10 income gainers in this period.
The authors point out that half the 80 American companies in the CNET Tech Index are in those top 10 counties. Furthermore, when income inequality decreased after 2000, the income drop in the high-tech counties contributed most to the decline.
New York, interestingly enough, showed large increases in per capita income both during the Internet boom and the Internet winter that followed.
The implication is that the income gains of the 1990’s associated with the technology bubble not only accrued to a relatively small number of people but also occurred in a relatively small number of geographic areas.
To drive this point home, the authors asked what would have happened to the index if just 4 of the 3,100 counties in the United States exhibited average income growth in the technology boom years. The four are Santa Clara, San Mateo, San Francisco (all associated with Silicon Valley) and King County, Wash. (home of Microsoft).
Note the remarkable difference in the Theil index computed with the adjusted growth rates for these four counties: If the per capita income in just these four counties had grown at the same rate as the average in the United States, income inequality across counties would have changed little in the late 1990’s. In other words, only four counties drove most of the change across the 3,100 counties.
The findings by Mr. Galbraith and Mr. Hale offer something to all sides in the debate. Perhaps options grants and initial public stock offerings had a lot to do with income inequality. Changing compensation patterns for technology-related skills could also be significant.
But the biggest point that I take away is a simple one: there’s no substitute for being in the right place at the right time.
