Are Initial Jobless Claims a Useful Gauge of Labor Market Conditions?

Along with inflation measures, Federal Reserve officials pay close attention to labor market indicators in maintaining the dual mandate of price stability and maximum sustainable employment. Since early 2023, our blog posts have concerned the first side of the balance: inflation persistence, the predictive content of different inflation components and the content horizon of core PCE. In this On the Economy blog post, we shift our focus to the second side: labor market conditions.

Our readers may be more familiar with monthly indicators commonly cited by economists and policymakers, such as the unemployment rate or total nonfarm payroll from the Bureau of Labor Statistics. However, weekly data, such as initial claims for state unemployment insurance benefits, can capture important movements in the labor market that occur between monthly reports. Revisiting a 2011 Economic Synopses essay (PDF) by Kevin Kliesen, Michael McCracken and Linpeng Zheng, this blog post examines whether there exists a threshold of initial claims that indicates a turning point in labor market conditions: Claims rising above this level signal weakening conditions while claims falling below this level signal improving conditions. The “rule-of-thumb” threshold commonly used by business economists is 400,000.

We put this rule to the test by running a threshold regression to determine the relationship between the one-month-ahead growth rate in total nonfarm payroll and the current month’s average initial claims. Rather than deciding on a threshold beforehand, we try an array of thresholds and report the one that yields the highest accuracy in predicting future change in nonfarm employment, henceforth the “optimal” threshold.

To account for the possibility that the optimal threshold has evolved with structural changes in the economy, we do this exercise for 10-year rolling samples between 1948 and 2024. We first estimate our model with (1) employment change between February 1948 and January 1958, and (2) claims data between January 1948 and December 1957, i.e., on a one-month lag relative to employment. We then estimate the model with (1) employment change between March 1948 and February 1958, and (2) claims data between February 1948 and January 1958. We repeat this process, maintaining a window size of 10 years (120 months), until we reach the end of the data in November 2024.

The figure below plots the 12-month moving average of the optimal claims threshold, in solid red, and the 12-month average of the actual number of claims, in dashed black. For context, we also provide the 12-month average of actual change in total nonfarm employment, in gray columns, and the rule-of-thumb threshold of 400,000, in dotted gray.

Realized Initial Jobless Claims Relative to an Optimal Threshold

SOURCES: Bureau of Labor Statistics and Labor Department (retrieved via Haver Analytics) and authors’ calculations.

NOTE: Realized data are 12-month moving averages.

The threshold varies significantly over time, though on average it is close to 400,000. Two patterns emerge that align with two regimes in U.S. economic history: the period of economic volatility leading up to the mid-1980s and the period of relative stability that followed, commonly known as the Great Moderation.

Before 1984, the optimal threshold was increasing over time but generally below 400,000. Its average between January 1958 and December 1983 was about 306,090 (the threshold line, in red, begins in December 1958 since it was the first observation with enough lag to take a 12-month moving average). Recall the informal rule that when initial claims fall below a certain threshold and remains so for an extended period, labor market conditions are improving, and vice versa. In line with the long economic expansion between 1961 and 1969, the 12-month moving average of claims steadily declined from July 1961 onward. It was below the optimal threshold by late 1964 and would remain so until early 1972.

During significant downturns, such as the 1973-75 recession, the actual number of claims clearly exceeds the optimal threshold as employment levels plummet. In fact, initial claims began to consistently exceed the optimal threshold in June 1974, providing an early signal of weakening labor market conditions; in contrast, claims didn’t exceed the rule-of-thumb threshold until March 1975. At the peak of employment loss in June 1975, the 12-month moving average of claims was about 462,167—well above the corresponding optimal threshold of 275,000. This pattern persists for the 1981-82 recession, when actual claims surpassed the optimal threshold of 325,000 by about 246,167 claims at peak employment loss in October 1982. Before 1984, it seems that the optimal claims threshold provided a meaningful benchmark for gauging labor market conditions, especially during bad economic times.

After 1984, the optimal threshold was about 434,165 on average, significantly higher than it was before the Great Moderation. During the recovery from the 1981-82 recession, the number of realized claims was consistently below the optimal threshold. Even as a brief recession in the early 1990s punctuated this relatively peaceful expansion, claims rose above the rule-of-thumb threshold of 400,000 but not the optimal threshold.

In the thick of the Great Recession (2007-09), claims did exceed the optimal threshold but by much smaller magnitudes than those of serious downturns before 1984. During the long recovery that followed, claims were steadily declining and far below the optimal threshold as well as the rule-of-thumb threshold.

For the COVID-19 recession, claims did not exceed the optimal threshold until February 2021, 10 months after the end of the recession in April 2020. Since 1984, the optimal threshold seems more informative for gauging labor market conditions during expansionary rather than recessionary periods.