A Natural Approach to Estimating the ‘Natural Rate’ of Unemployment
The unemployment rate is a widely viewed gauge of U.S. labor market slack or tightness. Because of structural changes to the labor market over time, assessments about slack/tightness require a reference point called the “natural rate of unemployment.”
The natural rate of unemployment—the unemployment rate that would prevail in a “neutral” labor market after removing all movement due to the business cycle—is not directly observable and must be estimated.
Changing demographics, technological disruption and exposure to foreign competition, among myriad other factors, can cause the natural rate of unemployment to vary. Consequently, the unemployment rate today could signal a slack labor market, whereas the same unemployment rate in the past might have signaled a tight labor market.
This is the first of three articles in which we derive and present a new natural rate of unemployment series. This article shows how we estimate the natural rate of unemployment. Our next article describes the trend in our natural rate series and contrasts it with a frequently used series produced by the Congressional Budget Office (CBO). The final installment investigates and explains wage growth based on the new series.
Examining Worker Cohorts
To capture the various factors that can change the natural rate of unemployment, we examine data on 35 million individuals, age 16 and older, from the Current Population Survey (CPS) for the years 1976 to 2018. Specifically, we follow 360 cohorts of workers over their careers, where we define each cohort based on the individual’s decade of birth (pre-1930, 1930–39, … , 1990–99, post-2000), sex, race/ethnicity and level of education completed. An example of a cohort is white females with a high school education who were born in the 1960s.
An important aspect of our approach is that we can see individuals in each cohort reach a specific age over a 10-year calendar span. For example, during 1985–94, we have many observations of individuals in our 1950–59 cohorts who turned 35 in a different month and year, and therefore, at a different point in the business cycle.
The ability to look at age-unemployment profiles across a full decade helps to smooth out cyclical effects in the data. By controlling for cyclicality, our estimated life-cycle unemployment rate profile for each cohort results in a cohort-specific natural rate series, whereby the estimated unemployment rate series depicts the expected unemployment rate for individuals in a specific cohort at each age in a neutral labor market.
Life-Cycle Unemployment Profiles
When we estimate a cohort life-cycle unemployment rate profile, we need to initially select years where we assume the labor market is neutral for all cohorts—that is, years for which there are no cyclical unemployment effects. For each cohort, we can then identify and remove the cyclical effects associated with the remaining years.
We use the CBO’s published non-accelerating inflation rate of unemployment (NAIRU) to provide an initial value for neutral labor market years—years when the CBO’s NAIRU equaled the actual unemployment rate and the unemployment rate was declining. (NAIRU is the rate of unemployment below which inflation should rise or, conversely, above which inflation should fall.)
We find some common features in the estimated cohort life-cycle unemployment rate profiles. These features, summarized below, are important to understanding the drivers behind changes to the natural rate of unemployment:
- Unemployment rates tend to decline with age until individuals reach their mid-40s.
- Increasing education tends to reduce unemployment rates.
- Unemployment varies by race/ethnic group, but this difference is mitigated by the time individuals reach their 50s, except for black workers.
- Women have lower rates of unemployment than men if both have less than a college degree, but this relationship reverses if both have a college degree.
Life-cycle unemployment rates tend to be the highest early in a worker’s career (Chart 1).
As workers gain labor market experience and stronger job attachments, their expected unemployment rates tend to decline until they reach their late 40s. Unemployment rates tend to rise during their early 50s and then decline as workers move toward retirement.
Individuals with less than a high school education have significantly higher expected unemployment rates than those with a high school diploma. These differences persist but narrow over a person’s career. However, there is little difference in expected unemployment rates between individuals with a college degree and those with a post-college (graduate) degree.
Race/Ethnicity, Gender Influence Unemployment Rate
Life-cycle unemployment rate profiles also differ significantly by race/ethnicity (Chart 2).
Among males born in the 1950s who have a high school education, black workers early in their careers have an expected unemployment rate roughly 10 percentage points higher than that for whites. The gap in the black and white expected unemployment rates for these two cohorts does not disappear even as these workers reach retirement.
In contrast, this cohort of Asians has a lower expected unemployment rate than whites until the Asians reach their mid-30s, when they closely track each other until retirement. For Hispanics, the expected unemployment rate exceeds that for whites and Asians until they reach their early 50s.
Comparing high school-educated white men and women born in the 1950s, the expected unemployment rate tends to be several percentage points lower for women than for men, with the gap widening in their 50s (Chart 3).
In contrast, the expected unemployment rate tends to be slightly higher for college-educated women in this cohort than for men, with the difference largely disappearing in their 50s.
Impacts on the Natural Rate
These charts illustrate several factors that will cause the natural rate of unemployment to change.
The rapid decline in expected unemployment rates in the early years of a worker’s career indicates that a workforce that skews older, as it does today, will act to reduce the natural rate of unemployment. In addition, increases in education across cohorts will also act to reduce the natural rate as the relatively more educated cohorts enter the labor force.
Given lower unemployment rates for many women, the rapid rise of female labor force participation in the 1960s to 1980s is also a factor in reducing the natural rate of unemployment.
About the Authors
The views expressed are those of the authors and should not be attributed to the Federal Reserve Bank of Dallas or the Federal Reserve System.