Changes in the Composition of Labor for BLS Multifactor Productivity Measures

Labor is an important input into the production process, but calculating MFP using only the total number of labor hours implies that every hour worked is equally productive and ignores the fact that labor input is composed of different types of labor. Demographic trends, such as increasing educational attainment and the aging of the population, have changed the composition of the workforce in ways that impact the productivity of an hour of labor. Other influences, such as business cycles, bring about more transient changes. For example, the workforce tends to become more experienced in recessions because older and more experienced workers are less likely to lose their jobs than younger and less experienced workers.[1] The effect of these underlying trends on labor composition and ultimately multifactor productivity are described below.

Ideally MFP would be calculated using numerous labor inputs, one for each type of worker. However, the BLS’s primary source of hours data for productivity measurement, the Current Employment Statistics (CES) survey, does not include any demographic information and cannot distinguish hours by type of worker. Therefore, the BLS uses data from the Current Population Survey (CPS)—Outgoing Rotation Group (ORG), which includes demographic details in addition to information on hours worked and earnings, to adjust private business labor hours for changes in the composition of the workforce. On March 26, 2015, the BLS changed its source data from the March CPS Income Supplement to the CPS ORG, going back to 2004. CPS ORG data have more than twice the number of observations as compared to the March CPS and the ORG data are more current. Labor composition for the previous year can be computed as early as February using the CPS ORG, while measures using the March CPS couldn’t be computed until October. On March 24, 2016, the BLS expanded its estimates of labor composition using CPS ORG to reflect the entire published time series, which begins in 1987.

Changes in labor input in the private business and private nonfarm business sectors are equal to the changes in labor hours plus the changes in labor composition. Labor hours for all employees are estimated by combining CES production and nonsupervisory worker employment and hours paid data, with an hours-worked adjustment from the BLS National Compensation Survey, and information on hours worked by nonproduction and supervisory workers from the CPS.[2]The changes in labor composition are estimated using CPS-ORG data as the difference between the weighted sum of changes in the hours of each type of worker and the un-weighted total change in the hours of all workers combined. The weights are the relative cost shares that more heavily weight the hours worked by more productive workers, i.e. those whose marginal product is estimated by their higher wage. To compute the change in labor composition, two consecutive years of CPS-ORG data are sorted into types of workers, defined by combinations of age, education and gender. For each of these worker types, “cells”, total hours worked and median hourly wage are calculated in each year. The hours and wage data are used to calculate each type of worker’s share of total wages, averaged over the two years; the weight for each type of worker. Next, the year-to-year percentage change in hours worked is computed for each type of worker, as well as for all workers combined. The labor composition adjustment that is used to adjust total hours is calculated as the difference between the percent change in total hours worked and the weighted sum of the percent changes of hours worked by each age/education/gender worker type. The labor composition index reported in the following section is computed by selecting a base year and computing the compound growth of labor composition between the base year and the current year.

Recent Changes in Labor Composition

Between 2013 and 2014, labor composition grew by 0.1 percent in both the private business and the private nonfarm business sectors.

Because the two sectors are similar—the only difference in coverage is the farm sector, which made up 2.2 percent of private business sector hours worked in 2014—subsequent estimates will be presented only for the private business sector. [3]

The effect of labor composition on labor input increased over the 1987-2014 period, with positive growth rates in all but two years. Changes in the labor composition index have a cyclical component. The labor composition index tends to increase faster during a recession and the early stages of the recovery, because younger and less-educated workers usually lose their jobs first, leaving an older and more-educated workforce. The labor composition index grew faster during the 1990-1992, 2000-2002, and 2007-2009 periods.

Recent Changes in the Distribution of Hours Worked

Over the 2010-2014 period, the educational attainment of the work force continued to increase. For both men and women, the share of total hours worked by those with college and advanced degrees increased. These increases were fairly evenly distributed over the 5-year period. Looking at all workers, workers with 16 or more years of schooling saw the largest gain in their share of hours worked from 2010 to 2014.

The hours-weighted average level of educational attainment for men was 13.7 years in 2014, which was lower than women’s level of 13.8 years. The difference between the hours-weighted average level of educational attainment for men and women has remained 0.1 percent between 2010 and 2014. The hours-weighted average level of educational attainment for men and women combined was 13.8 years in 2014. The hours weights are the workers’ shares of labor compensation, according to the relative size and hourly wages of the group of workers, with groups defined by age and education levels.

The fraction of hours worked by those 31 to 55 has decreased from 59.5 percent in 2010 to 57.2 percent in 2014, while the fraction of hours worked by those over 55 has increased from 15.6 percent in 2010 to 17.7 percent in 2014. Movement towards an older age distribution of hours worked will tend to increase the labor composition index.

Wage Structure

Wage structure can affect the labor composition index when the hours worked by the different demographic groups change at different rates. To illustrate, suppose that the hours of college-educated workers increase at the same rate as the hours of high school educated workers decrease. Labor input can either increase or decrease depending on the relative weights of these two groups. The higher wages of college-educated workers tends to increase their relative weight in the labor composition index, while their smaller total number of hours worked tends to reduce their weight. Wage growth matters if growth rates differ across demographic groups. In the above example, faster wage growth among college graduates would increase their relative weight and cause labor input to increase at an even faster rate. It is important to note, however, that if the hours of all groups grow at the same rate then differences in wage levels and growth rates across demographic groups do not affect the growth of the labor input.

Earnings for men with a college degree and women with an advanced degree decreased in 2014; the only two groups to see earnings decrease in 2014. However, all other groups for both men and women saw an increase in earnings in 2014.

Published labor input measures

Labor composition growth generally provides a small but steady positive contribution to labor input. Within a growth accounting framework, an increase in the labor composition index, or workers' skill levels, has the same effect on output and productivity growth as an increase in hours worked. That is, a 1.0 percent increase in labor composition is equivalent to a 1.0 percent increase in hours worked.

For the 1987-2014 period, labor input increased by 1.2 percent annually; labor composition increased 0.4 percent annually and hours worked increased 0.8 percent. For the 2007-2014 period, labor input increased 0.3 percent; labor composition increased 0.4 percent and hours worked declined at 0.1 percent. In 2014, labor composition increased 0.1 percent for the second year in a row. These two years showed the smallest increase since 2005, yet hours worked grew at 2.5 percent.

Notes

Date last modified: March 24, 2016