اثرات توزیعی قیمت گذاری کربن: یک رویکرد تعادل عمومی با میکرو داده ها برای خانواده ها

Many policies to limit greenhouse gas emissions have at their core efforts to put a price on carbon emissions. Carbon pricing impacts households both by raising the cost of carbon intensive products and by changing factor prices. A complete analysis requires taking both effects into account. The impact of carbon pricing is determined by heterogeneity in household spending patterns across income groups as well as heterogeneity in factor income patterns across income groups. It is also affected by precise formulation of the policy (how is the revenue from carbon pricing distributed) as well as the treatment of other government policies (e.g. the treatment of transfer payments). What is often neglected in analyses of policy is the heterogeneity of impacts across households even within income or regional groups. In this paper, we incorporate 15,588 households from the U.S. Consumer and Expenditure Survey data as individual agents in a comparative-static general equilibrium framework. These households are represented within the MIT USREP model, a detailed general equilibrium model of the U.S. economy. In particular, we categorize households by full household income (factor income as well as transfer income) and apply various measures of lifetime income to distinguish households that are temporarily low-income (e.g., retired households drawing down their financial assets) from permanently low-income households. We also provide detailed within-group distributional measures of burden impacts from various policy scenarios. Highlights ► We develop a simulation model with 15,588 households to study the distributional impacts of carbon pricing in the US. ► Sources side impacts have typically been ignored in the literature biasing studies towards finding carbon pricing to be regressive. ► Our general equilibrium framework allows us to capture uses and sources side impacts from carbon pricing. ► We find that variation in impacts within broad socioeconomic groups may swamp average variation across groups. ► We find that progressivity on the sources side is sufficiently strong to offset regressivity on the uses side.

Carbon pricing, whether through a cap and trade system or a tax, can have widely varying distributional impacts. Variation in impacts arises for three reasons. First, households differ in how they spend their income. Carbon pricing will raise the price of carbon intensive commodities and disproportionately impact those households who spend larger than average shares of their income on these commodities. In a general equilibrium setting, carbon pricing also impacts factor prices. Households which rely heavily on income from factors whose factor prices fall relative to other factor prices will be adversely impacted. In the public finance literature on tax incidence, the first impact is referred to as a uses of income impact while the latter a sources of income impact (see, for example, Atkinson and Stiglitz, 1980, for a discussion of incidence impacts). Third, regional differences in the composition of energy sources affect the carbon content of various commodities, most notably electricity. In previous work, we have used a new general equilibrium simulation model of the U.S. economy (the MIT USREP model) to explore distributional implications of various ways of distributing allowances from a cap and trade system (Rausch et al., 2010b) and alternative schemes for returning revenues from an auctioned cap and trade system or equivalently a carbon tax (Rausch et al., 2010a). This paper is similar in spirit to Rausch et al. (2010a) but employs a variant of the USREP model that endogenously incorporates 15,588 households as individual agents within a general equilibrium framework. This allows us to explore distributional impacts of carbon policy over a number of new dimensions that previously have not been explored. We find the following. First, the use of revenues that can be raised through carbon pricing affects both the efficiency and equity of the policy. Analyses that focus solely on the impacts of carbon pricing without considering the use of revenues can lead to seriously misleading results. Second, the use of a model with a large number of households allows us to consider distributional impacts over different sub-populations. It also drives home the point that variation in impacts from a carbon pricing policy within sub-groups may swamp the variation across groups. Third, we provide two measures that proxy for lifetime income and find little evidence that the use of annual income biases carbon pricing towards greater regressivity. Finally we find interesting variation across racial and ethnic groups that have not been addressed in the literature to date. We turn next to some background on the measurement of the burden of carbon pricing. Section three describes the model and the following section presents results. We conclude in Section 5.

This paper makes five key points. First, how proceeds of a carbon pricing policy are used affects both the efficiency and equity of the policy. Using revenues to cut tax rates has beneficial efficiency consequences but comes at the cost of higher regressivity. On the other hand, certain distributions have adverse consequences on both efficiency and equity. On these grounds, one cannot find an easy justification for the free distribution of allowances in a cap and trade system to industry. Of course, one can justify free distribution on political economy grounds. Our analysis helps illustrate the cost of using a free distribution to effect political goals. Second, previous policy analyses have been carried out using models with a single representative agent or a small number of households. This analysis uses a model with a large number of households and therefore provides finer level detail on distributional impacts of various policies. Doing so makes the point that variation in impacts within broad socioeconomic groups may swamp average variation across groups. Third, we provide two measures to proxy for lifetime income to address the criticism that studies using annual income bias carbon pricing towards greater regressivity. We do not find evidence of such bias in this analysis but stress that our proxy measures are by no means perfect. Fourth, interesting variation in burden impacts exists across racial and ethnic groups. Such variation has not been pointed out in the literature before. Further work will be needed to understand the causes and implications of this variation. But our initial analysis suggests that much of the impact goes away once income and other household characteristics are controlled for. Thus, we emphasize that while differential burden impacts across racial and ethnic groups reflect differences in income and spending patterns across the groups, they are not inherent to racial characteristics. Five, we note that sources side impacts of carbon pricing have typically been ignored in the literature. Doing so biases distributional studies towards finding carbon pricing to be regressive. We find that progressivity on the sources side is sufficiently strong to offset regressivity on the uses side so that carbon pricing is proportional. Finally we note that advances in computing power and numerical techniques make solving CGE models with large numbers of households quite tractable. This paper provides a brief look at the possibilities for understanding differential impacts of policies across different socioeconomic dimensions. We expect that this will provide a new area for research that should improve our understanding of the distributional impacts of environmental and energy policies.