We show that under typical conditions the simple “excess‐burden triangle” formula substantially underestimates the excess burden of commodity taxes, in some cases by a factor of 10 or more. This formula performs poorly because it ignores general equilibrium interactions—most important, interactions between the taxed commodity and the labor market. Many prior studies have shown that general equilibrium interactions affect excess burden but have not appreciated the bias associated with ignoring these interactions or the quantitative importance of this bias. We derive an implementable alternative to the simple formula. This alternative formula captures interactions that the simple formula omits; as a result it is both unbiased and usually more accurate.

High nominal gas prices, new awareness of threats to national security, and growing concern about global warming have reignited discussion of ways to reduce gasoline consumption in the United States. Debate centers on changing two policies already in place: the federal gas tax and Corporate Average Fuel Economy (CAFE) standards. Two influential recent reports find that increasing the gas tax would attain a given reduction in gas consumption at lower cost than would tightening CAFE standards (National Research Council, 2002; Congressional Budget Office, 2003). The gas tax has this advantage because it encourages not just increases in fuel efficiency, but also reductions in miles driven. In contrast, CAFE standards actually encourage more driving, because increases in fuel efficiency reduce the cost of gas per mile driven. We also compare the costs of the gas tax and CAFE standard but take into account interactions with preexisting tax distortions. Many papers examine the effects of these tax interactions in other contexts, but to our knowledge none of them considers the CAFE standard (see e.g., Lars Bovenberg and Ruud de Mooij, 1994; Ian Parry, 1995; Lawrence Goulder, 1995; Bovenberg and Goulder, 1996; Parry et al., 1999; Don Fullerton and Gilbert Metcalf, 2001). These interactions reduce the cost of the gas tax but increase the cost of CAFE, thus expanding the cost advantage enjoyed by the gas tax. This difference does not arise because the gas tax raises revenue, while the CAFE standard does not. Rather, this result is similar to that in West and Williams (2004a), which showed that, since gasoline and leisure are relative complements, raising the gas tax will increase labor supply, generating additional efficiency gains. In this paper, we estimate a consumer demand system using data from the Consumer Expenditure Survey and the California Air Resources Board and find that miles driven and leisure are relative complements. Thus, the gas tax encourages labor supply by raising the cost per mile driven, producing an additional efficiency gain. Conversely, because CAFE reduces the cost per mile, it discourages labor supply and yields an additional efficiency loss. While the induced changes in labor supply are tiny relative to the labor market, they are still substantial relative to the gas market and thus have a dramatic effect on the relative costs of the two policies. Our point estimates imply that they reduce the social marginal cost of the gas tax (starting from the status quo gas tax rate and ignoring the benefits of reduced gas consumption) by almost 30 percent, while increasing the marginal cost of CAFE by nearly 60 percent. This result implies that the case for raising the gas tax rather than tightening the CAFE standard is far stronger than previous studies suggest. Indeed, it strongly suggests that any tightening at all of the CAFE standard would lower welfare unless the benefits of reduced gas consumption have been seriously underestimated.