Academic journal article Federal Reserve Bank of Atlanta, Working Paper Series

Productivity, Energy Prices, and the Great Moderation: A New Link

Academic journal article Federal Reserve Bank of Atlanta, Working Paper Series

Productivity, Energy Prices, and the Great Moderation: A New Link

Article excerpt

Working Paper 2008-11

March 2008

Abstract: We study how total factor productivity (TFP), energy prices, and the Great Moderation are linked. First we estimate a joint stochastic process for the energy price and TFP and establish that until the second quarter of 1982, energy prices negatively affected productivity. This spillover has since disappeared. Second, we show that within the framework of a dynamic stochastic general equilibrium model, the disappearance of this energy-productivity spillover generates the significantly lower volatility of output and its components. Specifically, the change in the joint stochastic process accounts for close to 70 percent of the moderation in output volatility.

JEL classification: C22, E32, Q43

Key words: Bayesian estimation, productivity, energy price, Great Moderation, business cycles

1 Introduction

Our research is motivated by the fact that output volatility in the United States has declined significantly since the mid 1980s, as first documented by Kim and Nelson (1999) and McConnell and Perez-Quiros (2000). Stock and Watson (2002) also document the fall in volatility in other important macro variables such as consumption, investment, and hours worked. This drop in volatilities of output and most other macro variables is an artifact economists also call the "Great Moderation." (1) In this paper we study the link between energy prices and total factor productivity (TFP) and the role it plays in accounting for the great moderation. (2)

Why would the fluctuations in energy prices be important for the great moderation? Looking at the energy price and its relationship with business cycles reveals that the deep recessions in 1973-74 and 1980-82 were preceded by large energy price spikes. (3) However, the energy price drop in 1986 did not spark a significant acceleration in GDP growth and, likewise, the sustained rise in energy prices since 2002 has not yet led to a recession. (4) Our hypothesis from these observations is that a link between energy prices and business cycles existed in the early period, say, before 1982, but has since disappeared, potentially accounting for the lower volatility of macro variables. This motivates the empirical analysis where we estimate a joint stochastic process for quarterly energy prices and TFP using Bayesian estimation methods. We explicitly model a spill-over effect from the energy price innovations to TFP and the magnitude of this spill-over varies over time. Specifically, we allow for a breakpoint from one regime into another, and the timing of this break itself is a parameter to be estimated. We find the second quarter of 1982 (1982:II) to be the estimated breakpoint. Before 1982:II, innovations in the process for the energy price had a significant and negative spill-over into TFP. This spill-over disappeared afterwards.

Next we use a dynamic stochastic general equilibrium (DSGE) modeling framework to evaluate the impact of the changing nature of the joint stochastic process for energy prices and TFP on key macro volatilities. Specifically, we take the Kim and Loungani (1992) model, which incorporates energy use as a complement to fixed capital on the production side, and simulate it with the pre and post 1982:II specification for the joint stochastic process for the price of energy and TFP. We show that the absence of the spill-over effect after 1982:II reduces output volatility by about 34 percent. (5) Given that the actual drop in output volatility after 1982 was about 55 percent, the changing nature of the stochastic process accounts for about 61 percent of the great moderation in output volatility.

One can object that the significant drop in the share of energy use in GDP since the early 80's can directly account for the reduced volatility, without the added link of energy prices on productivity. Thus, we simulate the model without an energy-productivity spill-over but with different energy shares calibrated to the observed energy to output ratios in the pre and post 1982:II time-periods. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.