The Roles of Comovement and Inventory Investment in the Reduction of Output Volatility
Social Science Research Network
More than 80 percent of the decline in the variance of aggregate output since 1984 is accounted for by a decline in the covariance (and correlation) of output among industries that hold inventories. Using a HAVAR macro model (Fratantoni and Schuh 2003) with only two sectors, manufacturing and trade, we show that this decline in comovement -and thus much of the Great Moderation in aggregate and industry-level output -is explained largely by changes in the structural relationships between
... ips between sectors' sales and inventory investment, rather than by "good luck." A small part of the Moderation is explained by structural changes among interest rate parameters, but the case for better monetary policy is complicated by structural changes in the real side of the economy. We also show that the decline in comovement is concentrated in the automobile industry and related industries that are linked by supply and distribution chains. Immediately prior to the Great Moderation, these industries adopted new production and inventory management techniques, which may explain the structural changes. JEL Classifications: E22, E32, E52, L16 Although the reduction in covariance (and correlation) was generally widespread, it was concentrated disproportionately in the automobile industry and closely related industries. 6 These industries are linked by extensive and complex supply chains, which provide materials, supplies, and intermediate goods to manufacturers of finished goods, and distribution chains, which transport finished goods from manufacturers to wholesale and retail traders. It is widely acknowledged that these industries adopted new production and inventory management techniques, such as just-in-time production, in the late 1970s and early 1980s -immediately prior to the onset of the Great Moderation. Except for the declining ratio of materials inventories to production in manufacturing industries, the macroeconomic impact of these techniques is not well understood. Our results suggest that this technological change may be manifest in reduced comovement; if so, analyses based on representative-agent models and aggregate data cannot identify this change. 7 To identify and quantify the role of comovement between industries in a tractable macroeconometric model, we use the heterogeneous-agent VAR (or HAVAR) framework developed by Fratantoni and Schuh (2003) . 8 HAVAR models are well-suited for this endeavor because they can incorporate disaggregated of industries within an otherwise standard macroeconomic VAR while imposing all of the necessary aggregation conditions to make the model internally consistent. In fact, standard macroeconomic VARs used in the Great Moderation literature are nested in the HAVAR model, so the homogeneity restrictions imposed by macroeconomic VARs can be tested econometrically.