Split Personalities: The Changing Nature of Technology Shocks

This paper analyzes the nature of technology shocks and documents important changes in their propagation over time. We employ a vector-autoregression and identify a shock that explains the maximum variation in total factor productivity (TFP) at a long finite horizon. This agnostic identification suggests that the dominant shock driving TFP is not necessarily a surprise shock, but exhibits features consistent with a shock that is anticipated or diffuses over time: GDP and consumption rise prior to any significant increase in TFP. We further find that shock transmission has changed over time. In a sample that ends in the mid 1980s, the shock triggers a decline in hours-worked and inventories, and a rise in credit spreads. In a post-Great Inflation sample the response of these variables is reversed and the shock generates an outright expansion in hours, inventories, GDP and consumption that is accompanied by a decline in credit spreads. We find that the importance of technology shocks as a major driver of aggregate fluctuations has increased over time - they play a dominant role in the second subsample, but much less so in the first.