An empirical study of United States employment fluctuations

This paper studies U.S. nonfarm aggregate and sectoral employment fluctuations by employing the stochastic variance modeling technique to capture the changing and serially correlated variances in employment series. To estimate the SV model, I use the quasi-maximum likelihood (QML) approach developed in Harvey and Shephard (1993), Harvey, Ruiz, and Shephard (1994), and Breidt and Carriquiry (1996). The QML method applies the Kalman filter in the estimation, thereby rendering inference about the unobserved volatility component of the series. The main findings of the paper are as follows. First, aggregate employment volatility is higher in recessions than in expansions. Second, aggregate employment volatility has moderated over the past fifty years, especially in expansions but also in recessions. Third, the decline in overall employment volatility can be partly attributed to the growth of the less volatile services industry along with the decline of the highly volatile manufacturing sector. Fourth, volatility in manufacturing employment and output has been reduced over the past 15 years, and this pattern is also seen in durable goods manufacturing. Fifth, the patterns found in the volatility of overall employment are closely matched by those in durable goods manufacturing employment and output