A closer look at the Employment Cost Index

Business Economics, July, 1999 by Alan Garner

Many economic analysts believe the employment cost index is the best available measure of U.S. labor costs. For example, Abate referred to the ECI as "the best measure of compensation costs," and Griggs and Santow, Inc., described the ECI as "the best measure of wage behavior and benefits being paid, and of the pace at which such employment costs are rising." In many respects, these sentiments are probably correct, but other potentially useful measures of labor costs exist, including the average hourly earnings index and unit labor costs.

An Introduction to the ECI

The ECI is a quarterly measure of labor compensation per hour worked, including all wages, salaries, and benefit costs paid by employers. Wages and salaries are based on straight-time average hourly earnings, whether or not the employee is normally paid by the hour. Wages and salaries have historically accounted for a little over 70 percent of total employment costs. Nonwage benefits include paid leave, other supplemental payments, and employer contributions for insurance, retirement and savings plans, and legally required benefits.

The inflation rate of the private-sector ECI has roughly paralleled the overall inflation rate, measured by the GDP price index, since 1980.(1) The GDP price index is a broader measure of the general price level than the CPI, reflecting purchases by businesses and governmental units as well as consumers.

Growth rates of the two major ECI components have sometimes differed substantially in the 1980s and 1990s. The growth rate of benefit costs exceeded the growth rate of wages and salaries during most of this period. For example, benefit costs grew at an average annual rate of 5.6 percent from the second quarter of 1981 to the fourth quarter of 1994, well above the average growth rate for wages and salaries of 4.0 percent over the same period. However, benefit cost inflation has slowed sharply in the past few years, while wage and salary gains have increased moderately. As a result, benefit costs have risen at only 2.2 percent annually from the fourth quarter of 1994 to the first quarter of 1999, well below the 3.3 percent rate for wages and salaries.

As with all economic statistics, the ECI is subject to measurement problems. For example, the ECI does not capture certain forms of labor compensation, such as stock options and signing bonuses. Moreover, the Bureau of Labor Statistics collects information for only a small percentage of all relevant employees because of the costs involved. As a result, the ECI will have "sampling errors" when wages and benefits for this small group do not behave exactly the same as wages and benefits for all relevant employees. In addition, the ECI is computed for a fixed basket of industries and occupations, similar to the fixed market basket of goods and services in the consumer price index (CPI). As a result, the ECI is potentially subject to statistical biases caused by the fixed industry-occupation weights. Economists have criticized the CPI for this and other biases, suggesting that the index has overstated the inflation rate. However, research conducted at the BLS found the use of fixed industry-occupation weights does not cause large statistical biases in the ECI.(2)

Comparison with Other Labor Cost Measures

The ECI differs in various ways from two other measures of labor costs: the average hourly earnings index and unit labor costs. The average hourly earnings index reflects changes in basic hourly and incentive wage rates as well as such variable factors as premium pay for overtime and late-shift work. Being a monthly series, the average hourly earnings index is more timely than the ECI. But the ECI is a more complete measure of labor costs, because it includes many important elements of labor compensation, such as nonproduction bonuses, health insurance, and payroll taxes paid by employers, that are not in average hourly earnings. Another drawback of the average hourly earnings index is that its coverage is restricted to production workers and nonsupervisory employees, thereby excluding the compensation of business managers. The average hourly earnings index also does not apply fixed weights across industry-occupation categories, as does the ECI. As a result, a shift in the mix of jobs away from low-paying occupations toward high-paying occupations would be recorded as wage inflation even if the wages paid by the particular jobs had not changed.(3)

In spite of these differences, the ECI and average hourly earnings have displayed broadly similar movements since the early 1980s. Except for the past four years, the ECI has risen faster than average hourly earnings, partly reflecting the rapid increase in health insurance costs, which is not captured by the average hourly earnings index. Recently, average hourly earnings have accelerated somewhat more than the ECI.

The ECI differs from unit labor costs, another widely quoted measure of labor costs, in several ways. Unit labor costs equal labor compensation per hour divided by output per hour, where output per hour measures labor productivity. Like the ECI, unit labor costs include nonwage benefit costs, such as Social Security taxes and health insurance costs paid by employers. But unit labor costs also include some compensation, such as proprietor's income, that is not in the ECI. A potential disadvantage is that the compensation per hour measure used to calculate unit labor costs does not apply fixed industry-occupation weights as does the ECI. As a result, an increase in compensation caused by a shift in the employment mix toward higher paying jobs could be misinterpreted as labor cost inflation.