Working Paper: NBER ID: w22294
Authors: Hashmat Khan; Christopher R. Knittel; Konstantinos Metaxoglou; Maya Papineau
Abstract: U.S. carbon dioxide emissions are highly procyclical—they increase during expansions and fall during recessions. Given this empirical fact, we estimate the response of emissions to four prominent technology shocks from the business-cycle literature using structural vector autoregressive methodologies and data for 1973–2012. By studying the response of emissions to these shocks, we provide a novel approach to assess the shocks’ relevance as sources of aggregate output fluctuations. We find that emissions rise on impact only after an anticipated investment-specific technology shock; the response is statistically significant after the first quarter. The same shock explains most— roughly a third—of the total variation in emissions at a horizon of 5 years. Notably, emissions decrease on impact after an unanticipated neutral technology shock in a statistically significant way. This negative empirical response has the opposite sign from its theoretical counterpart in recent environmental DSGE (E-DSGE) models. Since the positive response of emissions drives the E-DSGE models’ recommendation for an optimal procyclical policy, our findings suggest that such a policy recommendation should be treated cautiously.
Keywords: carbon emissions; business cycles; technology shocks; procyclicality
JEL Codes: E32; Q43; Q48; Q54
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| anticipated investment-specific technology shock (E22) | emissions (Q52) |
| unanticipated neutral technology shock (O49) | emissions (Q52) |
| anticipated neutral technology shock (O49) | emissions (Q52) |
| unanticipated investment-specific technology shock (O39) | emissions (Q52) |