Do Markets Respond More To More Reliable Labor Market Data? A Test of Market Rationality

Abstract

Since 1979, the Bureau of Labor Statistics (BLS) has nearly quadrupled the size of the sample used to estimate monthly employment changes. Although first-reported employment estimates are still noisy, the magnitude of sampling variability has declined in proportion to the increase in the sample size. A model of rational Bayesian updating predicts that investors would assign more weight to the BLS employment survey as it became more precise. However, a regression analysis of changes in interest rates on the day the employment data are released finds no evidence that the bond market’s reaction to employment news intensified in the late 1980s or 1990s; indeed, in the late 1990s and early 2000s the bond markets hardly reacted to unexpected employment news. For the time period as a whole, an unexpected increase of 200,000 jobs is associated with about a 6 basis point increase in the interest rate on 30 year Treasury bonds, and an 8 basis point increase in the interest rate on 3 month bills, all else equal. Additionally, unexpected changes in the unemployment rate and revisions to past months’ employment estimates have statistically insignificant effects on long-term interest rates.