The analysis in this report is based on the hourly wage. For workers in the sample who reported only a weekly wage, the hourly wage is estimated as the weekly wage divided by the usual hours worked in a week. Using the hourly wage factors out differences in earnings across workers that arise from differences in hours worked, such as weekly earnings of full-time versus part-time workers.
Workers earning less than $2 per hour or more than $100 per hour are dropped from the sample. That amounts to a loss of less than 1% of the sample.
Why the Numbers of Workers in the Five Wage Groups Are Not Equal
The wage distribution used in this report is based on the principle of separating workers into wage quintiles. Each quintile would, in theory, consist of one-fifth of the workforce. However, as shown in Table A1, this ideal is not achieved in either 1995 or 2005. The reason is that workers at the boundaries of wage classes often have the same hourly wage. For example, in 2005, workers who in principle would be at the top of the low-middle group (the 40th percentile) had the same wage, $12 per hour, as workers who would be at the bottom of the middle-wage group (the 41st and 42nd percentiles). Since it would be arbitrary to separate workers with the same hourly wage, it was decided to classify them all into the middle-wage group. As a result, the low-middle group lost a percentile of workers and the middle-wage group gained a percentile. The consequence is that the low-middle group in 2005 had 23.6 million workers and the middle-wage group had 25.2 million.
Labor Market Conditions in 1995 and 2005
A concern that arises when comparing economic outcomes across time is whether the periods in question represent similar phases of the business cycle. For example, if wage outcomes in an economic expansion are compared with outcomes in a past recession, that would exaggerate the improvement in earnings relative to the underlying trend. The periods in question for this report are 1995 and 2005. A key similarity between the periods is that they are about four years removed from the ends of the last two recessions, the first lasting from July 1990 to March 1991 and the second from March 2001 to November 2001. In that sense, 1995 and 2005 can be said to represent similar points in the business cycle.
Data from the Bureau of Labor Statistics show that labor market conditions in the two periods were about the same. One indicator suggests the labor market in 2005 may have been tighter than in 1995, but other indicators suggest the opposite. In particular, the 5.1% unemployment rate in 2005 was less than the 1995 unemployment rate of 5.6%. However, the percentage of the working-age population that was employed, or the employment-population ratio, was slightly higher in 1995—62.9%, versus 62.7% in 2005. Also, the percentage of the working-age population that is employed or actively seeking work, or the labor force participation rate, was higher in 1995—66.6%, compared with 66% in 2005. On balance, labor markets conditions in 1995 and 2005 appear roughly the same.
Related Findings in the Literature
The findings in this report regarding the progress of Mexican-born workers and newly arrived foreign-born workers are echoed in some recent research papers. In a report issued in December 2006, the Congressional Budget Office (CBO) examined the changing characteristics of the low-wage labor market in the U.S. between 1979 and 2005. One of its findings is the recent decline in the percentage of workers born in Mexico or Central American who are paid low wages (less than the 20th percentile hourly wage). The CBO analysis shows that percentage falling from 49% in 1994 to 44% in 2005.
Borjas and Katz (2007) document an improvement in the education profile of Mexican-born workers, albeit not as rapid an improvement as achieved by native-born workers. They also report that declines in the wages of Mexican-born workers relative to native-born workers in the 1980s came to a halt in the 1990s and that the relative wage of newly arrived Mexican-born workers increased in the 1990s. A more detailed examination of the earnings of new arrivals is found in Borjas and Friedberg (2006). They point to a shift away from agricultural labor as a factor in raising the entry wage for newly arrived workers from Mexico.