A Distributional Analysis of Green Tax Reforms - ntanet.org

1 A Distributional Analysis of Green Tax Reforms655 National Tax JournalVol. LII, No. 4 Abstract - I measure the Distributional impact of a shift towardgreater reliance on environmental taxes (a Green tax reform) usingboth annual and lifetime income measures to rank households. Anenvironmental tax reform can be designed that has a negligibleimpact on the income distribution when the funds are rebated tohouseholds through reductions in the payroll tax and personal in-come tax. I also analyze trade-offs among competing goals of effi-ciency, equity, and ease of administration in the design of a greentax we raise environmental taxes? This question hasbeen asked increasingly in the face of widespread envi-ronmental problems, including global warming, air and wa-ter pollution, and a host of other environmental problemsthat we face today.

2 If one is concerned exclusively with effi-ciency, the answer to this question would be yes, we certainlyshould raise taxes to the point where the tax equals the mar-ginal social damage from pollution. Unfortunately, the realworld is more complicated than a textbook world. How dowe measure marginal social damages? Concerns about eco-nomic efficiency intrude given the widespread prevalence ofother taxes. Finally, Distributional concerns come into taxes tend to be regressive: poor people paya disproportionate share of their income in these taxes rela-tive to rich paper addresses how one could design an environ-mental tax reform such that it reduces or even eliminates theregressive nature of environmental taxes. It considers reformsthat combine environmental taxes with reductions in othertaxes such that the increased regressivity of the environmen-tal taxes is offset by increased progressivity resulting fromreductions in other taxes.

3 I also analyze trade-offs amongcompeting goals of efficiency, equity, and ease of administra-tion in the design of a Green tax next section provides some background on the issueof environmental (or Green ) tax Reforms . Next, I describe howI measure the Distributional impact of tax Reforms and de-scribe the data. The fourth section provides results from theanalysis and a concluding section Distributional Analysis ofGreen Tax ReformsGilbert E. MetcalfDepartment ofEconomics,Tufts University,Medford, MA 02155 National Bureau ofEconomic Research,Cambridge, MA 02138 NATIONAL TAX JOURNAL656 BACKGROUND ON Green TAXREFORMST here has been a great deal of interestin recent years in the use of environmen-tal tax revenues to substitute for someportion of existing tax collections.

4 The is-sue of a substitution of environmental forother taxes can be traced back to Tullock(1967) and more recently Terkla (1984).This early literature focused on the effi-ciency implications of an environmentalreform and led to a debate over what hasbeen dubbed the Double Dividend Hy-pothesis. 1 One strand of this literature (astypified by Terkla) considers a reform inwhich environmental regulations are re-placed with tax instruments in such a fash-ion that pollution activities are the switch from a regulatory to a taxa-tion mechanism for limiting pollutionraises revenue that can be used to lowerother distorting taxes. This shift has un-ambiguous welfare gains. Another strandof the literature (typified by Pearce (1991)and Repetto et al.)

5 (1992)) focuses on theuse of environmental taxes both to reducepollution and to raise revenue to lowerother taxes. While it is clear (seeBovenberg and deMooij (1994) as well asParry (1995)) that there are also efficiencycosts with environmental taxes (separatefrom the environmental benefits), it is alsoclear that the efficiency costs depend im-portantly on which taxes are , the benefits of the new revenuesare greatest when used to lower the mostdistorting debate over a Green tax shift andthe Double Dividend Hypothesis has fo-cused on efficiency considerations. In ad-dition, Distributional considerations areclearly important and little work has beendone in this area. These concerns are rel-evant given the sense that most energyand environmental taxes are some authors have challenged thisperception by taking into account lifetimeconsiderations ( , Poterba, 1991a; Bull,Hassett, and Metcalf, 1994), it is clear thatdistributional concerns limit political sup-port for the greater use of discussion of the distribu-tional problem suffers by looking at theenvironmental taxes in isolation that is,with lump-sum recycling of the tax rev-enues.

6 While it might be the case that theimposition of an environmental tax by it-self is regressive, it is quite possible that arevenue neutral tax reform, where an en-vironmental tax replaces some other tax,could be A recent study byHamond et al. (1997) emphasizes thispoint. Below, I will consider Reforms (based on suggestions in Hamond et al.)that are designed to maintain or perhapsincrease the progressivity of the tax sys-tem. I will also consider other Reforms toemphasize the trade-offs among equity,efficiency, and ease of AssumptionsIncidence analyses rest on a number ofimportant assumptions. First, researchersmust make assumptions about the direc-tion of shifting of the various taxes underanalysis. Second, researchers must deter-mine how to rank people by some mea-sure of well-being.

7 Typically, annualincome has been used to sort people orhouseholds ( , Pechman, 1985; Gale,Houser, and Scholz, 1996). Recognizingthat annual income may not always be agood measure of resources available tohouseholds given life-cycle and perma-nent income considerations, a number ofresearchers have either tried to measurelifetime income explicitly or constructed1 See Fullerton and Metcalf (1998) for a survey of this economic terms, the former is an example of an absolute tax incidence Analysis , while the latter is a differ-ential tax incidence Distributional Analysis of Green Tax Reforms657proxies for lifetime income (Poterba, 1989,1991a; Feenberg, Mitrusi, and Poterba,1997; Metcalf, 1993, 1994; Caspersen andMetcalf, 1994) or have constructed com-putable general equilibrium models (Ful-lerton and Rogers, 1993; Altig et al.)

8 , 1997).This paper uses a measure of lifetime in-come constructed by Caspersen andMetcalf from the Panel Study on IncomeDynamics (PSID) and the Consumer Ex-penditure Survey (CES). In brief, the pro-cedure constructs wage profiles from thePSID based on information available inboth the PSID and the CES and then com-putes a measure of lifetime income forhouseholds in the In this paper, Iwill provide Distributional results usingboth an annual income measure and a life-time income its attraction, lifetime income isdifficult to measure and whatever mea-sure is employed rests on strong assump-tions. An alternative approach is to em-ploy a cohort Analysis . Gale, Houser, andScholz (1996), for example, consider theimpact of tax changes on married couplesin the age range of 40 50.

9 By restrictingthe Analysis to households who are likelyto be at the same stage of their earningsprofile, they avoid mixing people fromdifferent stages of the life cycle. The ap-proach is conceptually appealing and doesreduce the measurement problem de-scribed above. It does not, however, ad-dress the problem of transitory incomeshocks. Households with a one time nega-tive income shock may maintain previousconsumption levels under the assumptionthat the poor income realization is a tem-porary setback that is likely to be offsetby positive income shocks in the , consumption to income ratios willbe high for this group and any tax thatapproximates a consumption tax in its ef-fect will look more regressive than itwould if transitory income shocks weretaken into account.

10 Despite this drawback,I will report a cohort distribution of taxesas an alternative measure to my lifetimeincome the methodology in this paper issimilar to that in Caspersen and Metcalf(1994), there are two important differ-ences. First, the Caspersen and Metcalfanalysis is an example of an absolute taxincidence Analysis , while the Analysis inthis paper is a differential tax incidenceanalysis. Therefore, in addition to distrib-uting the new taxes, I also have to con-sider the distribution of existing taxes, inparticular the corporate and personal in-come taxes. Second, the CES data onhealth expenditures only reflects out ofpocket spending on health care. In thispaper, I replace the CES spendingamounts with amounts based on informa-tion drawn from the National MedicalExpenditure Survey (NMES).