Consumers Share and Producers Share
Abstract - We estimate, for each state with a general sales tax, thepercentage of that tax that is levied directly on final consumptionspending of the states residents. State by state, the consumers shareranges from 28 to 89 percent. For all states together, it is 59 percent,findings that generally agree with expectations and with otherestimates for single states. Most of the remaining sales tax is onproducer inputs. Between 1979 and 1989, the consumers share in-creased for 28 states, fell for 17 states, and was unchanged in 1state.
While the state general sales tax is a tax on final consumption, its base also includes substantial sales to busi-nesses, raising the specter of tax pyramiding/cascading. Sup-pose, for example, that a state taxes electricity used in pro-duction. Depending on market conditions, a manufacturersproduction costs, and ultimately its selling price, might in-clude tax already paid. Taxing the output at retail imposes atax more than once on the same piece of value added(Graeser and Maury, 1992).
Besides current production costs, state and local sales taxesextend to the acquisition of capital assets. They are far fromuniform across investments (Joulfaian and Mackie, 1992).Joulfaian and Mackie found that national average effectivesales tax rates for different types of business equipment andstructures varied from 5.9 percent to zero. State-to-statedifferences in tax rates and in taxation or exemption of capi-tal purchases add further distortions, especially in locationdecisions. These and other examples (such as tax incidenceand tax exporting) illustrate the importance of recognizingthat sales taxes affect business purchases, as well as con-sumption, and of knowing the amount of sales tax on eachpart.
This paper reports estimates, for each state with a generalsales tax, of how much of that tax was levied on personalconsumption spending of in-state residents in 1989. (Hereand henceforth, reference to states with a general sales taxincludes the District of Columbia.) These update my earlierestimates for 1979 (Ring, 1989), incorporating several im-
Consumers Share and Producers Shareof the General Sales Tax
Raymond J. Ring, Jr.School of Business,University of SouthDakota, Vermillion, SD57069
NATIONAL TAX JOURNAL
provements in data and methodology. Thenext section briefly explains the improve-ments and shortcomings. The section af-ter that contains the estimates and ananalysis of them. The paper concludeswith a summary and conclusions.
The consumers share of the generalsales tax (SHARE) is given by
 SHARE = CP/(CP + BP)
where CP represents sales tax on pur-chases by resident consumers and BP rep-resents other sales tax revenue. BP in-cludes primarily business purchases, butalso any other items a state taxes, such aspurchases by governments and nonprofitorganizations. For a thorough discussionof SHARE and the factors that affect it,please see Ring (1989).
To estimate each states SHARE in 1989,we used 1990 U.S. Census data on thenumber of each states households in 1989income classes (U.S. Department of Com-merce, 1993), Consumer Expenditure Sur-vey (CES) tabulations of average 1989spending by consumer units in eight in-come classes (U.S. Department of Labor,1991, Table 1), and information on howeach state defines its tax base. (AppendixB lists sources of state-specific informa-tion.) See Ring (1989) for a detailed expla-nation of the methodology and elabora-tion of its advantages and shortcomings.The major advantage is the ability to ac-count for state-to-state differences, provid-ing consistent estimates of SHARE for allstates.
This study incorporates several im-provements over my earlier work. Theearlier study used 19723 CES dataupdated to 1979 (Ring, 1989); the esti-mates presented here incorporate actual
1989 CES data. This is unlikely to greatlyaffect the results, but does increase confi-dence in the estimates.
This update also includes several refine-ments in how each state defines its salestax base. Ring (1989) fit all states into gen-eral tax base categories according towhether they taxed or exempted food athome, clothing, utilities, and gasoline.This update retains those general catego-ries, but distinguishes state-by-state treat-ment of four utilities (electric; natural gas,fuel oil, etc.; water and sewer; and tele-phone), instead of treating each state astaxing or exempting all utilities. Ring as-sumed that all states taxed the sameconsumer services: a few services taxedby nearly all states in 1979. Greater detailin the 1989 CES (70 spending categoriescompared to 41 in the 19723 CES) allowsus to account for each states treatment ofnine categories of consumer services. Wealso now include more precise treatmentof alcohol, tobacco, prescription drugs,and nonprescription drugs. Table 1 indi-cates the broad categories into which eachstate falls; the finer distinctions used forutilities, services, alcohol, etc. are availableupon request.
We also treated unrelated individualsdifferently for this study than for the ear-lier one. Earlier, we used the number ofunrelated individuals (from the Census)directly, treating each as a consumerunit as defined in the CES. For this study,we distinguish financially independentconsumer units from those that pool theirincome. Appendix A explains this adjust-ment. Without it, the number of consumerunits (and thus estimated consumerspending) would be understated.
The CES spending figures include salesand excise taxes; without adjustment, ourmethodology would overestimate theconsumers part of the sales tax base. Ad-justing Joulfaian and Mackies 1987 esti-mates (1992), we find the weighted aver-age state and local rate (weighted by grossstate product (GSP)) for states with a gen-
Consumers Share and Producers Share
TABLE 1CONSUMERS SHARE CALCULATIONS FOR 1989
AlabamaArizonaArkansasCaliforniaColoradoConnecticutWashington, D.C.FloridaGeorgiaHawaiiIdahoIllinoisIndianaIowaKansasKentuckyLouisianaMaineMarylandMassachusettsMichiganMinnesotaMississippiMissouriNebraskaNevadaNew JerseyNew MexicoNew YorkNorth CarolinaNorth DakotaOhioOklahomaPennsylvaniaRhode IslandSouth CarolinaSouth DakotaTennesseeTexasUtahVermontVirginiaWashingtonWest VirginiaWisconsinWyoming
TotalaTax base key:
A: exempt food, clothing, utilities, gasolineB: exempt food, utilities, gasolineC: exempt food, utilitiesD: exempt food, gasolineE: exempt utilities, gasolineF: exempt foodG: exempt utilitiesH: exempt gasolineJ: exempt none
NATIONAL TAX JOURNAL
eral sales tax to be 6.06 percent in 1989.1
To account for the over-estimate, we di-vided each states estimated consumersales tax base by 1.0606.
The CES figures also include travelspending, so our CP estimates (as describedso far) include what each states residentsspent out of state. State sales tax revenuesinclude taxes paid by nonresidents, whichare thus included in BP. To adjust for bothof these, we used Mutti and Morgans(1983) estimates of General Sales TaxesPaid by Residents Traveling Out-of-Stateand General Sales Taxes Paid by Out-of-State Travelers, adjusting them to accountfor state taxes only and for growth between1980 and 1989. This adjustment removesone of the sources of uncertainty in howclosely BP approximates business pur-chases, leaving government and nonprofitpurchases as the only potentially signifi-cant elements besides business purchases.
This study suffers from several limita-tions. Due to the absence of state-specificCES data, we must attribute nationalspending patterns to each state. For morediscussion of this problem, please see Ring(1989).2 The CES under-reports consumerspending, which in itself might be ex-pected to result in underestimates ofSHARE. However, both the Census andthe CES suffer from under-reporting ofincome. Absent any good way to adjustfor these problems, we must hope thatthey are roughly offsetting.
The adjustment to reconcile Censushouseholds and CES consumer units(Appendix A) should provide better esti-
mates (than were in Ring (1989)) of thenumber of consumer units in each incomeclass. Unfortunately, there is no clear wayto adjust for the fact that CES consumerunits will have lower incomes (and thusspend less) than will Census nonfamilyhouseholds. Consequently, this methodstill somewhat overstates spending. Themagnitude of thi