U.S. weekly jobless claims are a key early indicator for the U.S. economy and global financial markets. A new Kansas City Fed paper suggests that to use these data efficiently one should first estimate a time varying benchmark for the “neutral level” of claims. Claims above (below) the benchmark would indicate deterioration (improvement) of the U.S. labor market.

Revisiting the Use of Initial Jobless Claims as a Labor Market Indicator
John Carter Braxton, Federal Reserve Bank of Kansas City, May 3, 2013
 
http://www.kansascityfed.org/publicat/reswkpap/pdf/rwp13-03.pdf

“Initial jobless claims measures the number of individuals who have been laid off either temporarily or permanently and have filed for unemployment insurance benefits in the previous week. By its construction, initial claims capture one of the primary margins of the labor market: layoffs. Since firms generally conduct layoffs when their business prospects worsen, initial claims are considered a leading indicator of labor market conditions.”

“[The paper introduces] a threshold of initial jobless claims that serves as a benchmark of comparison for the weekly reporting of initial jobless claims… this benchmark represents a threshold of initial jobless claims that is associated with a constant unemployment rate. Observed initial claims below the threshold correspond with an improving labor market and an expected decline in the unemployment rate, while observed claims above the threshold correspond with a weakening labor market and increasing unemployment.”

“Frequently 400 thousand initial claims in a week is cited as a labor market threshold. A downside to using a fixed threshold number is that it cannot incorporate changes in labor market dynamics. To account for these changing dynamics the presented threshold incorporates multiple margins of the labor market, such as layoffs, hires, quits, and labor force participation. Considering these margins allows the presented threshold to explain more than twice the amount of volatility in the unemployment rate compared to simply using 400 thousand as a threshold.”

“Since layoffs are captured through initial jobless claims and are a component of separations, initial claims and the change in employment are negatively correlated.”

“As layoffs rise (fall), the unemployed’s share of the labor force tends to increase (decrease). This relationship induces initial jobless claims and the unemployment rate to largely move together. However…the unemployment rate typically remains elevated longer than initial claims. This occurs because the unemployment rate includes multiple margins of the labor market such as hires, quits, layoffs and labor force participation while initial claims is associated with layoffs. Thus, any approach aiming to use initial claims as an indicator for the overall labor market must account for these additional margins.”
 
“Further…the fraction of laid-off individuals who apply for unemployment benefits varies over time… They show that the take-up rate is counter-cyclical and explain that laid-off individuals expect to be unemployed longer when the labor market is weaker, making them more likely to apply for unemployment benefits…The threshold of initial jobless claims moves counter cyclically… The countercyclical nature of the threshold indicates that in a weaker labor market an unchanging unemployment rate can occur despite an elevated level of initial claims.”
An elevated level of observed initial claims can indicate improvements in the labor market so long as observed initial claims are below the threshold. For example, between May 2010 and October 2011 observed initial claims were above 400 thousand but below the threshold in each month. Thus, over this period, the common benchmark of 400 thousand suggests labor market conditions were worsening, while the threshold indicates improvement. In fact, during this period the three month moving average of the unemployment rate decreased by nearly 0.9 percentage points.”

“Analyzing the difference between observed initial jobless claims and the threshold provides an estimate of future changes in the unemployment rate. [The figure below]shows that there is a strong linear relationship between the size of the deviation in month t and the change in that month’s unemployment rate.”