Will using newer input–output data for general equilibrium modeling provide a better estimate for the CO₂ mitigation cost?

ABSTRACT

We provide a critical evaluation about how updating the input–output data of a computable generation equilibrium model can affect policy results, an assessment that is rarely done in existing literature. Specifically, we explore how datasets with different fossil energy cost shares alter results of policy simulations that aim at reducing CO₂ emissions. We prove analytically that a sudden fossil fuel price surge, which provides little time for adjustment through input substitution, can lead to a higher CO₂ mitigation cost. The finding is demonstrated empirically in a full-scale economy-wide model for a base year with lower fossil fuel prices, contrasted with results from a base year when fossil fuel prices spiked. We then propose an adjustment to resolve the issues of using input–output data that embed abrupt fossil fuel price hikes.

KEYWORDS:

• Input–output data
• energy-economic modeling
• emissions mitigation costs

Acknowledgments

We are thankful for the financial support provided by National Energy Program-Phase II and Atomic Energy Council, Executive Yuan, Taiwan. We appreciate valuable comments from Dr. Cheng-Da Yuan, two anonymous reviewers, Professor Manfred Lenzen, and participants of the following conferences: Taiwan Economic Association 2018 annual conference held in Taipei, Taiwan, and the 22nd GTAP conference held in Warsaw, Poland. We are grateful for the excellent edits provided by Jamie Bartholomay. All remaining errors are our own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 As noted in Roberts (1994), in CGE modeling, using single-observation data for calibration purposes prevails, as the econometric approach to CGE analysis is quite demanding in terms of data requirements.

2 Since in the long run more factors of input can be adjusted, the long-run price elasticity of demand for a commodity will be higher than the short run one, as found empirically. To compare the short- and long-run elasticities in a consistent fashion, in general, one needs to have (1) a framework that can identify both short- and long-run elasticities; and (2) data with enough time periods to extract changes in demand in response to prices and other explanatory variables over time (Tsai & Mulley, 2014). Regarding the data needed, some studies are based on the sectoral data prepared by the government agencies (e.g., Lim et al., 2012; Maddala et al., 1997), while others use the individual/firm level data from various sources (e.g., Joutz, 2009; Barnes et al., 1982).

3 We are thankful for an anonymous reviewer in referring us to Jorgenson et al. (2012).

4 ${\alpha }_h>{\alpha }_l$ suggests that the energy cost share for technology $h$ is higher than that for technology $l$. ${\lambda }_h|Y⟩{\lambda }_l|Y$ means that after imposing the constraint $E=E_1$, $P_E^h+{\lambda }_h|Y^{\ast }>P_E^l+{\lambda }_l|Y^{\ast }$. The combination of a higher energy cost share and a higher consumer price for energy leads to a higher level of price increase for the commodity produced by sector $A$, i.e., $P_1^h-P_0^h>P_1^l-P_0^l$. Since $P_0^h>P_0^l$, therefore $P_1^h>P_1^l$.

5 In this exercise, our goal is not to reproduce the dataset of a particular year using that of another year.

6 The model used to produce the adjusted database is a typical static CGE with a long run closure, i.e., it allows mobility of capital and labor across sectors. See Chai et al. (2019) for details.

Additional information

Funding

This work was supported by The National Energy Program Phase II: 107-3113-F-042A-001 and Atomic Energy Council, Executive Yuan, Taiwan.