Understanding the Decline in U.S. Output Volatility: An Analysis of the Automobile Industry

University Of California-San Diego, La Jolla CA

Investigators

Abstract

Why has output volatility in the post-1984 period been so much lower than in earlier periods? Is it just good luck, better monetary policy, or structural change in the economy? Understanding the source the decline in output volatility not only helps forecast future volatility patterns, but also sheds light on potential improvements in monetary policy and firm behavior. The proposed research seeks to shed light on the source of the decline in volatility by conducting a case study of changes in output and sales volatility in the U.S. automobile industry. The U.S. automobile industry represents an ideal case study of the decline in volatility because it exhibits patterns similar to, but more dramatic than, the aggregate data. Previous work by Kahn, McConnell and Perez-Quiros presents evidence that changes in production and inventory behavior are a key source of the decline in volatility. In the post-1984 period, the variance of production fell much more than the variance of sales. Furthermore, the covariance of inventory investment and sales has changed from being positive (which adds to volatility) to being negative. These facts represent circumstantial evidence that the key to the decline in volatility is structural change in production and inventory management. More specificially, this research explores the alternative hypothesis that these observed changes stem mostly from changes in the sales process rather than structural changes in firms' production and inventory scheduling. The U.S. automobile industry, as well as many other industries, face nonconvex costs and lumpy production margins that can lead to a very nonlinear relationship between the variance of production and the variance of sales. Thus, a change in the dynamics of the sales process can lead to radical changes in the way production is scheduled. The first part of the project demonstrates the complex relationship between sales dynamics and output dynamics using simulations. Employing a cost function similar to the one faced by automobile assembly plants, optimal production schedules in response to various sales processes will be derived. The second part of the project gathers and analyzes detailed weekly automobile assembly plant data to investigate how much of the change in production behavior is due to structural changes at the plant level versus changes in the nature of the sales process. Since the sales process can be affected by changes in monetary policy rules or by firms' pricing behavior, identifying how much of the change in production behavior is simply a response to changes in the dynamics of sales is key to tracking the source of the decline in volatility.