WELFARE AND MACROECONOMIC EFFECTS OF LESSONS FROM CGE MODELS ALAN D. VIARD** AMERICAN ENTERPRISE INSTITUTE

Transcription

1 JAMES A. BAKER III INSTITUTE FOR PUBLIC POLICY RICE UNIVERSITY WELFARE AND MACROECONOMIC EFFECTS OF DEFICIT-FINANCED TA CUTS: LESSONS FROM CGE MODELS By JOHN W. DIAMOND* JAMES A. BAKER III INSTITUTE FOR PUBLIC POLICY RICE UNIVERSITY And ALAN D. VIARD** AMERICAN ENTERPRISE INSTITUTE AUGUST 2008

2 Deficit-Financed Tax Cut THESE PAPERS WERE WRITTEN BY A RESEARCHER (OR RESEARCHERS) WHO PARTICIPATED IN A BAKER INSTITUTE RESEARCH PROJECT. WHEREVER FEASIBLE, THESE PAPERS ARE REVIEWED BY OUTSIDE EPERTS BEFORE THEY ARE RELEASED. HOWEVER, THE RESEARCH AND VIEWS EPRESSED IN THESE PAPERS ARE THOSE OF THE INDIVIDUAL RESEARCHER(S), AND DO NOT NECESSARILY REPRESENT THE VIEWS OF THE JAMES A. BAKER III INSTITUTE FOR PUBLIC POLICY. PREPARED FOR THE AMERICAN ENTERPRISE INSTITUTE CONFERENCE, TA POLICY LESSONS FROM THE 2000S, MAY 30, * Baker Intitute, Rice Univerity, jdiamond@rice.edu. ** American Enterprie Intitute, aviard@aei.org BY THE AMERICAN ENTERPRISE INSTITUTE THIS MATERIAL MAY BE QUOTED OR REPRODUCED WITHOUT PRIOR PERMISSION, PROVIDED APPROPRIATE CREDIT IS GIVEN TO THE AUTHOR AND THE AMERICAN ENTERPRISE INSTITUTE. 2

3 Deficit-Financed Tax Cut I. Introduction Tax cut are often adopted with no explicit proviion for offetting the reulting revenue lo. Example of thi include the tax-cut package enacted on June 7, 2001 and May 28, The decreae in revenue from uch tax cut initially increae the deficit, leading to a larger tock of government debt. However, the government infinite-horizon budget contraint require that offet ultimately be adopted to ervice or retire that debt. Deficitfinanced tax cut have been controverial, with upporter often pointing to beneficial economic effect from reduction in ditortionary taxe and opponent pointing to the advere economic effect of deficit. In view of thee conflicting economic effect, an ultimate aement of deficitfinanced tax cut require an etimate of the magnitude of the variou effect. Computable general equilibrium (CGE) model offer a natural way to obtain uch etimate. In thi paper, we urvey prior work that ha tudied deficit-financed tax cut in CGE model and preent new reult for a range of tax-cut experiment. Prior work ha often emphaized the impact on macroeconomic variable, uch a output, conumption, labor upply, and the capital tock. We report thee impact, but alo emphaize the impact of deficit-financed tax cut on the welfare of variou generation in our overlapping generation (OLG) framework. We find that even when deficit-financed tax cut increae long-run output, they often reduce the welfare of future generation while increaing the welfare of earlier generation. The remainder of the paper i organized a follow. In ection II, we provide the economic background for thi topic, including the early reult obtained by Auerbach and Kotlikoff (1987) in their CGE model of an OLG economy. In ection III, we urvey the leon learned from imilar tudie in the In ection IV, we decribe the tructure and calibration of the model that we ue. We preent our reult in ection V. In ection VI, we decribe recent empirical work that invetigate how actual deficit-financed tax cut are ultimately paid for. In ection VII, we briefly review the work that ha been done in model of infinitely-lived agent. We conclude in ection VIII. II. Background In thi ection, we dicu the baic of modeling OLG economie and the early work by Auerbach and Kotlikoff. A. Baic Theory in OLG Model In an OLG model, the economic impact of a deficit-financed tax cut generally reflect both intergenerational-reditribution and incentive effect. In mot cae, deficit-financed tax cut lighten the fical burden on early generation and increae thoe on later generation. The extent and ditance of the hifting depend upon how rapidly the reulting debt i retired or erviced. It alo depend, however, upon the age of the affected taxpayer. A tax cut for the young financed by fical meaure on the old may actually reditribute reource toward earlier generation, even if the financing meaure are adopted later than the tax cut. Intergenerational reditribution from a deficit-financed tax cut i a zero-um game; any teady-tate lo reflect a gain to tranitional generation (or vice vera). The incentive effect depend upon how the initial tax cut and the financing meaure affect ditortion in the economy. For example, if a ditortionary tax i reduced and the reulting debt i ultimately financed through a le ditortionary fical intrument a net improvement in the efficiency of the fical tructure occur. 3

4 Deficit-Financed Tax Cut Thee effect have an impact on the welfare of the affected generation and alo determine (through income and ubtitution effect) their labor upply, conumption, and other deciion. Thoe deciion determine the impact of the deficit-financed tax cut on the variou macroeconomic variable. B. Rationale for CGE Modeling Some analye have tudied deficit-financed tax cut in an ad hoc manner, in which the variou effect of deficit are dicued heuritically. Thi approach can eaily lead to problem. For example, labor upply may be aumed to depend only upon the explicit marginal tax rate on labor, ignoring the labor-upply impact of taxe on conumption and taxe on capital income. Dynamic may be handled in a very curory manner, with little attention to the impact of incentive on the timing of work and other deciion. Thi approach cannot readily incorporate the imultaneity that i at the heart of general equilibrium analyi. CGE model provide a more diciplined manner of proceeding. The model aume utility and profit maximization, with precie aumption about the economic environment. The interaction of the effect i alo taken into account. The direct income and ubtitution effect of the policy change, for example, caue change in factor price, which in turn have income and ubtitution effect. CGE modeling therefore yield a coherent and elf-conitent et of reult. Of coure, the ultimate validity of the reult depend upon the extent to which the model accurately decribe the actual economy. C. Auerbach-Kolikoff Reult The pioneering CGE modeling of deficit-financed tax cut (among other policie) in an OLG framework wa done by Auerbach and Kotlikoff (1987). We draw on their finding a background for the work that ha been done in the Auerbach and Kotlikoff reported (pp ) the effect of broad-baed income tax cut that lat one, five, and twenty year, followed by income-tax increae that keep the debt-to-output ratio tabilized at it new, higher level. Their model featured a (rather low) initial income tax rate of 15 percent. One experiment reduce the tax rate to 10 percent for 20 year, after which it i increaed to maintain the debt-to-gdp ratio at it year-20 value. The required income-tax rate rie teadily over time, reaching a teady-tate value of 30.4 percent. In the teady tate, the capital tock i reduced 49 percent and labor upply i reduced 5 percent, cauing output to fall by 19 percent and the wage rate to fall by 14 percent. The dynamic are alo intereting. A Auerbach and Kotlikoff emphaized, the crowding out of capital proceed in a very gradual manner, with much of it occurring after the tax cut ha expired. After 10 year, the capital tock ha fallen only 2 percent. Even at the 30-year mark, a decade after the tax cut ha expired, the capital tock ha fallen 20 percent, only two-fifth of the teady-tate decline. Auerbach and Kotlikoff identified two reaon for the low pace. Firt, the income effect inducing additional conumption by the early generation i pread over their entire lifetime, not jut the period that the tax cut i in effect. Second, the ubtitution effect of the tax cut induce additional aving while it i in effect, becaue after-tax return are higher during that period. Thee policy experiment reduce teady-tate welfare, but improve the welfare of earlier cohort. Steady-tate utility i reduced by more than 14 percent (meaured by the fraction of lifetime reource that would have to be taken to compenate for the required 4

5 Deficit-Financed Tax Cut utility decline). A large part of the utility effect are reditributive, however, with utility gain for all generation that are already in the economy at the time the tax cut begin. The duration of the deficit-financed tax cut ha a harply non-linear effect. Compared to the 5-year tax cut, the 20-year tax cut caue 10 time the increae in the teady-tate income tax rate, 7 time the teady-tate crowding out of capital, 9 time the teady-tate labor upply reduction, and 11 time the teady-tate welfare lo. Auerbach and Kotlikoff noted that, in ome repect, a longer tax cut i qualitatively different from a horter one. During the period that the tax cut i in place, the ubtitution effect are equally large regardle of duration, but the income effect are larger for the longer tax cut. For example, they found that crowding out of capital begin immediately with the 20-year tax cut becaue the income effect alway outweigh the ubtitution effect. With the 5-year tax cut, in contrat, there i initial crowding in of capital becaue the ubtitution effect outweigh the income effect within the 5-year period. Their reult paint a dimal picture of the teady-tate effect of deficit-financed tax cut, a the hort-run improvement in incentive from the initial tax cut i followed by a long-term deterioration due to the ubequent tax increae. In their word (p. 93): Although temporary tax cut may initially crowd in capital formation, there i no way to ecape the long-run cot of hort-run deficit finance Although one might think that, having crowded in capital through hort-term tax cut, one could adopt a painle policy for eliminating the accumulated debt (or imply meeting repayment commitment), uch i not the cae when income taxe mut be relied on. One cannot potpone indefinitely raiing tax rate, and once thee rate are raied, the timulu to aving through ubtitution effect i revered; in addition, the cro-generational income effect that are at the heart of the crowding-out proce ultimately play a deciive role in reducing national aving. (emphai added) Negative reult are inevitable when the income-tax cut mut ultimately be financed by income-tax increae. Auerbach and Kotlikoff did not, however, conider deficit-financed tax cut in which a reduction in one type of tax i ultimately financed by increaing a different type of tax or fical intrument that i potentially le ditortionary. In that cae, ome, but not all, of the concluion are modified. Such work ha been developed in the 2000 and will be the focu of our dicuion below. III. Leon Learned in the 2000 A CGE modeling ha become more widepread, the number of tudie examining deficit-financed tax cut ha increaed. Although today OLG model are broadly imilar to Auerbach and Kotlikoff early model, they are more likely to feature multiple production ector and poibly a rudimentary international ector. In the remainder of thi ection, we decribe ome of the reult that have been obtained. Uing the model decribed in ection IV, below, Diamond (2005) conider a permanent extenion of the 2001 and 2003 tax cut. Financing begin after 10 year and tabilize the debt-output ratio at it new higher level. If the extenion i financed by reduction in tranfer payment, GDP i 0.8 percent higher in If the extenion i financed by a reduction in government conumption, GDP i 0.1 percent higher; if the extenion i financed by an increae in tax rate, GDP i 0.2 percent lower. The differing 5

6 Deficit-Financed Tax Cut impact on GDP are largely due to differing impact on capital accumulation; the correponding change in the capital tock are 0.7 percent, 0.3 percent, and negative 1.2 percent. A 2006 Treaury Department tudy 1 employ a imilar model to examine the permanent extenion of the 2001 and 2003 tax cut. Thi tudy find that extending the 2001 and 2003 tax cut would raie long-run GNP and conumption by 0.7 percent if the extenion wa financed by a reduction in government conumption, but would reduce long-run real GNP by 0.9 percent if financed by an acro-the-board increae in income tax rate. (In each cae, the financing meaure tart after 2017 and tabilize the debt-tooutput ratio at it new, higher level). The difference again largely reflect a differing impact on capital accumulation, with the long-run capital tock riing 2.3 percent in the firt cae and falling 1.8 percent in the econd cae. The output impact in 2011 through 2016 i poitive for both cae. The tudy decompoe the effect of the tax-cut extenion into three component. Lowering the dividend and capital gain tax rate ha long-run benefit under either financing mechanim, a do the reduction in the top four ordinary income tax rate. The gain are, however, larger under the government-conumption financing mechanim. Extending the remainder of the tax cut generate long-run loe under either mechanim, although the loe are maller when government conumption i reduced. The Congreional Budget Office produce annual dynamic analye of the Preident budget propoal. In recent year, thee propoal have featured a net tax reduction, primarily reflecting a permanent extenion of the 2001 and 2003 tax cut. The mot recent analyi conider two financing aumption, a gradual reduction in government pending (both government purchae and tranfer payment) and a gradual increae in marginal tax rate. 2 In each cae, the financing meaure are phaed in from 2019 through Uing a cloed-economy OLG model, CBO find (pp ) that the Preident tax and pending propoal would increae GNP by 0.4 percent and GNP by 0.6 percent with the government-purchae adjutment. With the tax-rate adjutment, CBO find the correponding increae to be 0.4 and 0.8 percent. Uing an open-economy OLG model, CBO find that the propoal would increae GNP by even larger amount. With the government-pending adjutment, the increae for the two time period are 0.8 percent and 1.2 percent; with the tax-rate adjutment, they are 0.7 and 1.0 percent. CBO doe not, however, examine the long-run effect. Denni, et al. (2004), another tudy by CBO economit, examine a 10 percent reduction in marginal tax rate, financed either by cut in government pending or by increae in tax rate. The financing mechanim tart 10 year after the tax cut i adopted and become fully effective after 20 year. In the OLG cloed-economy model, teadytate GDP fall by 0.1 percent with the pending cut and by 1.5 percent with the tax-rate increae. In the open-economy model, long-run GDP rie by 0.5 and 0.2 percent, repectively, but the more welfare-relevant GNP fall by 0.4 and 2.1 percent, repectively. Output alway rie in the hort run (Table 3, p. 13). A tudy by the Joint Tax Committee conider a rate reduction in individual income taxe accompanied by bae-broadening meaure. 3 In an OLG model, two 1 Office of Tax Analyi, U.S. Department of the Treaury (2006). 2 U.S. Congreional Budget Office (2008). 3 U.S. Congre, Joint Committee on Taxation (2006). 6

7 Deficit-Financed Tax Cut different offet to change in government debt are ued, one involving change in tranfer payment and one involving change in individual income tax rate. Under either offet, the policy increae real GDP by 1.2 percent in The GDP increae i 1.9 percent with the tranfer offet and 1.1 percent with the tax offet. The long-run increae in GDP i 2.6 percent with the tranfer offet and 1.2 percent with the tax offet (Table 2, p. 10). The higher output growth i largely due to higher growth of the capital tock with the tax offet (Table 3, p. 11); long-run capital growth i 4.5 percent with the tranfer offet and 1.8 percent with the tax offet. The long-run growth in conumption i 1.6 percent with the tranfer offet and 1.0 percent with the tax offet (Table 8, p. 15). Auerbach (2002) examine extenion of the 2001 and 2003 tax cut. He examine a omewhat different policy experiment than mot of the other tudie. The other recent tudie generally make thoe tax cut permanent and then layer a financing mechanim on top of the tax cut. (Of coure, when the financing mechanim i a tax-rate increae, layering on the tax increae effectively undoe the tax cut in whole or in part.) Following the approach of Auerbach and Kotlikoff (1987), Auerbach intead conider experiment in which the 2001 and 2003 tax cut lat 10, 15 or 20 year before expiring, after which the debt-to-output ratio i tabilized uing either increae in wage taxe or capital income taxe. Output increae while the tax cut are in effect, but fall after the tax cut expire. The long-run output decline i greater when the tax cut lat longer, cauing more debt to be accumulated, and when the financing i done through higher taxe on capital income. The negative effect are diminihed if part of the revenue lo i offet by reducing government purchae during the time that the tax cut are in effect. In ummary, the tudie generally find that both the form of the initial tax cut and the financing method matter. An increae in teady-tate output i mot likely if the initial tax cut i targeted toward marginal-rate reduction, if the financing method doe not raie marginal rate and ha income effect that encourage work (tranfer-payment reduction are ideal for thoe purpoe), and if the financing i implemented quickly. The impact on teady-tate output i generally driven by the impact on the teady-tate capital tock rather than teady-tate labor upply. Depite the valuable inight that have been gleaned from previou work, the reult are generally limited in both the range of policy experiment and the extent of the welfare analyi. The paper have drawn from a omewhat limited et of initial tax cut, generally acro-the-board income tax cut or extenion of the 2001 and 2003 tax law. Alo, depite the early example et by Auerbach and Kotlikoff (1987), the more recent tudie generally do not report the welfare impact on the variou generation. To explore thee iue, we report ome new reult for a range of deficit-financed tax cut. IV. Model Structure and Calibration We ue a dynamic overlapping-generation life-cycle computable general equilibrium model that explicitly calculate reform-induced change in all aet value that would accompany a debt-financed tax cut. A noted previouly, the model ha three production ector owner-occupied houing, rental houing, and a compoite good ector that include all non-houing good and ervice. The time path of invetment demand in all three ector i modeled explicitly, taking into account capital tock adjutment cot. On 7

8 Deficit-Financed Tax Cut the conumption ide, the current tax advantage of owner-occupied houing relative to other aet i taken into account in modeling the demand for the three good. The model allow for a fairly detailed decription of the tranitional and long run macroeconomic effect of debt-financed tax cut and an examination of the intergenerational welfare effect. The model doe not allow for multiple income group within each generation. Thi ection outline the baic tructure of the model, which combine variou feature from Auerbach and Kotlikoff (1987) and other imilar and well-known model contructed by Goulder and Summer (1989), Goulder (1989), Keuchnigg (1990), Fullerton and Roger (1993) and Hayahi (1982). A detailed decription of the model i provided in the appendix, and an even more complete decription i provided in Diamond and Zodrow (2005). A. Production Firm in the compoite good production ector produce output uing a CES (contant elaticity of ubtitution) production function with capital and labor a input. Firm chooe the time path of invetment to maximize the preent value of firm profit or, equivalently, to maximize firm value, net of all taxe and ubject to quadratic cot of adjuting the capital tock. Total taxe in the compoite good production ector include the corporate income tax and tate and local property taxe. Each firm maintain a fixed debt/aet ratio and pay out a dividend a contant fraction of earning after taxe and depreciation in each period, which i conitent with the old view of dividend taxation. The model aume individual-level arbitrage, which implie that the after-tax return to bond mut equal the after-tax return received by the hareholder of the firm. The value of the firm in the compoite good ector equal the preent value of all future net ditribution to the owner of the firm. Houing i produced in the owner-occupied and rental houing production ector. Following Goulder and Summer (1989) and Goulder (1989), rental houing i produced by non-corporate landlord and owner-occupied houing i produced by the owner. The technology ued in the production of rental houing and owner-occupied houing i aumed to be identical, with capital and labor combined in the ame CES production function. Landlord and owner-occupier are alo aumed to chooe time path of invetment to maximize the equivalent of firm value, net of total taxe. In the rental houing ector, the firm i modeled a a non-corporate firm, which implie that landlord are imply taxed at the individual level. In the owner-occupied houing ector, the meaurement of the tax burden take into account the fact that imputed rent are untaxed and that maintenance expenditure are not deductible, while mortgage interet and property taxe are deductible. The optimal invetment path i calculated a above. B. Individual Behavior On the individual ide, the model ha a dynamic overlapping generation framework with fifty-five generation alive at each point in time. There i a repreentative individual for each generation, who ha an economic life pan (which begin upon entry into the work force) of fifty-five year, with the firt forty-five of thoe year pent working and the lat ten year pent in retirement. Individual tate are identical o that difference in behavior acro generation are due olely to difference in lifetime budget contraint. An individual accumulate aet from the time of economic birth that are ued to finance 8

9 Deficit-Financed Tax Cut both conumption over the life cycle, epecially during the retirement period, and the making of bequet. The model include a joy-of-giving bequet motive o that the real value of bequet change with change in income and other economic variable. The conumer i aumed to chooe the time path of conumption and leiure to maximize ret-of-life utility, which i a dicounted um of annual utilitie, ubject to a lifetime budget contraint that require the preent value of lifetime wealth including inheritance to equal the preent value of lifetime conumption including bequet. Annual utility i aumed to be a CES function of conumption of an aggregate conumption good and leiure. The aggregate conumption good i modeled a a CES function of the compoite good and aggregate houing ervice (including a minimum purchae requirement for both good), with aggregate houing ervice in turn modeled a a CES function of owner-occupied and rental houing ervice. In addition, the model include a imple ocial ecurity ytem, government purchae of the compoite good, tranfer payment, a hump-backed wage profile over the life cycle, a progreive tax on wage income, and contant average marginal tax rate applied to interet income, dividend, and capital gain. Tranfer payment excluding thoe funded by the payroll tax are modeled a going diproportionately to younger generation with the younget 25 percent of the population receiving 31 percent of benefit. The progreive wage tax i modeled uing a quadratic wage tax function imilar to Auerbach and Kotlikoff (1987). The model aume a cloed economy, no uncertainty, and perfect competition in every ector of the economy. C. Calibration The model i calibrated by chooing a number of parameter value and economic variable o that the initial income tax teady tate in the bae year, which i the year of reform, cloely reemble the prevailing feature of the U.S. economy in Parameter value are choen to be conitent with empirical etimate and parameter value ued in other CGE tudie, epecially Altig, et al. (2001), Auerbach and Kotlikoff (1987), Auerbach (1996), and Fullerton and Roger (1993). The value for economic variable are generally choen to be conitent with etimate from the National Income and Product Account. Table 1 how the value of the model parameter that are the mot important in term of determining individual and firm behavioral repone. The rate of time preference, ρ, i et equal to In CGE model, the rate of time preference (or dicount rate) i typically choen in tandem with the intertemporal and intratemporal elaticitie of ubtitution to generate reaonable level of aving and invetment and labor upply in the initial teady-tate. Uing the Euler equation approach, Ziliak and Kneiner (1999) etimate that the rate of time preference under two pecification that yield value of and Jorgenen and Yun (2001) etimate a higher value of The value we chooe i at the low end of thee etimate but i conitent with other CGE tudie uch a Altig, et al. (2001). The elaticity of intertemporal ubtitution σ determine the willingne of conumer to ubtitute conumption acro period in repone to change in the relative price of conumption and therefore play a critical role in etablihing the reponivene of aving to tax change. Empirical tudie uing aggregate conumption 9

10 Deficit-Financed Tax Cut data typically find that the EIS i between zero and one. 4 The range of aumed value for the EIS ued in CGE model i quite mall, primarily becaue the choen value mut generate a teady-tate capital tock that i conitent with the data and the aumed value of the pure rate of time preference. Auerbach and Kotlifoff (1987), Fullerton and Roger (1993), Jorgenon and Yun (2001), Altig, et al. (2001), and Diamond and Zodrow (2007) all aume a value of the EIS between 0.25 and 0.50, depending partly on the interaction of the EIS with the choice of the pure-rate-of-time-preference parameter. We aume that the EIS i equal to The intratemporal elaticity of ubtitution! and the percentage of the endowment devoted to leiure are key parameter that determine the compenated and uncompenated wage elaticitie. For a given intratemporal elaticity of ubtitution, there i a larger percentage increae in labor upply aociated with an increae in the wage rate if the hare of the initial time endowment devoted to leiure i greater. The intratemporal elaticity of ubtitution determine conumer willingne to ubtitute between labor upply and leiure in repone to change in their relative price and therefore i critical in determining the labor upply repone to a change in the after-tax wage. We aume that the intratemporal elaticity of ubtitution i equal to 0.8 and that the hare of the time endowment devoted to leiure i The elaticitie of ubtitution between the compoite good and aggregate houing conumption! CH and between rental and owner houing! RO are choen o that the value of the compenated own-price elaticitie of owner and rental houing are both roughly -0.8 a reported in Roen (1985). 6 The variou weighting parameter in the production function and utility function are et to replicate a cloely a poible the actual pattern of aggregate production and conumption for the three good in the model. Table 2 how the value for technological parameter value. The rate of population growth i equal to 0.01 and the rate of technological growth i equal to 0.01, o that the economy grow at a 2 percent annual rate in teady tate. The ize of adjutment cot i alo important in determining the effect of debtfinanced tax cut. We aume that the adjutment cot parameter (! ) in the nonhouing production ector i equal to 5, meaning that an increae of 1 percentage point in the ratio of invetment to the capital tock i aociated with an increae of 5 percentage point in q. Thi value i a compromie between the etimate preented in Cummin, Haett and Hubbard (1994), Shapiro (1986) and the earlier and coniderably larger etimate preented in Summer (1981). In the abence of data on the value of the adjutment cot parameter in the owner-occupied and rental houing ector, thee value are aumed to equal the value of the adjutment cot parameter in the compoite 4 See Gunning, Diamond and Zodrow (2007). 5 Thi implie that the aggregate uncompenated labor upply elaticity i approximately 0.24, which i within the range of empirical etimate. Thi value i ignificantly lower than the value aumed in Altig, et al. (2001) and Auerbach and Kotlikoff (1987), but yield an aggregate labor upply elaticity that i conitent with mot of the empirical literature. It i, however, inconitent with the relatively large labor upply elaticitie found in the work of Precott (2005) and Davi and Henreckon (2005). 6 Etimate of houing demand elaticitie pan a wide range. DiPaquale and Wheaton (1994) report an etimated houing demand elaticity equal to while Riddel (2004) report an etimated elaticity equal to

11 Deficit-Financed Tax Cut good ector, although there i no economic reaon why thee value would need to be the ame. Table 3 how the initial teady tate value for output, the capital tock, firm value, invetment and earning in each ector, which are calibrated to data from the U.S. Bureau of Economic Analyi (2008). Table 4 how the initial teady tate value for federal taxe in the bae year. The federal tax ytem raie $2609 billion in total tax revenue in the bae year; federal income taxe raie $1,660 billion and ocial ecurity payroll taxe raie $949 billion, which i aumed to equal the amount of ocial ecurity benefit. Total federal income taxe are 18.9 percent of GDP. Federal government expenditure are 19.5 percent of GDP. Government debt i aumed to be 30 percent of annual GDP. Thi ratio i contant in the teady tate. Table 5 how the federal tax rate in the initial teady tate. Under the progreive wage tax, the income-weighted average marginal wage tax rate i equal to 26 percent and the average wage tax rate i 21.4 percent. 7 The tax rate on individual interet income i 15.2 percent and the tax rate on dividend i 16.3 percent. Capital gain in the compoite good and rental houing ector are taxed at an effective annual accrual rate of 5 percent and capital gain in the owner-occupied houing ector are untaxed. 8 The payroll tax i 10.7 percent; thi i lower than the actual 15.3 percent rate becaue all wage income i ubject to the payroll tax in the model. Social Security benefit are taxed at 5.2 percent. The effective tax rate on invetment in the compoite good ector i 28.8 percent 9 and the effective tax rate in the rental houing ector i 21 percent. 10 The model alo include deduction and credit in the calculation of taxable wage income. Deduction and credit are only allowed in the working period (the firt 45 year) of an individual life; retired individual do not receive deduction or credit. V. Simulation Reult In thi ection we examine the macroeconomic and welfare effect of everal alternative tax cut under a number of different financing option. The magnitude of the tax reduction i determined o that the decreae in revenue over the ten-year period following enactment i $500 billion with no behavioral repone. The decreae in revenue i unanticipated and enacted immediately. For concretene, we refer to the year of enactment a 2007, the year to which, a noted above, the initial teady tate i calibrated. The tax cut are permanent. The three main financing method are: (1) government tranfer (other than ocial ecurity benefit) are reduced immediately to finance the tax cut; (2) government debt i ued to finance the tax cut for ten year and then government 7 The value of marginal and average tax rate i baed on data from the Office of Tax Analyi Department of the Treaury. 8 The effective annual accrual tax rate on capital gain in the owner-occupied houing ector i aumed to equal zero, ince the Taxpayer Relief Act of 1997 exempted gain up to $250,000 on the ale of a houe for ingle taxpayer and up to $500,000 for married taxpayer filing a joint return. 9 The corporate tax rate i aumed to be 35 percent, with accelerated depreciation allowance for allowed for the purpoe of calculating taxable income. The amount of remaining bai for tax purpoe i explicitly calculated in each period and ued in the calculation of depreciation allowance. The corporate hare of the compoite-good ector i 62 percent. Thi i ue to calculate a weighted-average tax rate in each production ector. Note that the non-houing production ector i treated a a corporate firm in the model. 10 For the purpoe of calculating the weighted average tax rate in the rental houing ector the corporate hare of the rental-houing ector i aumed to be 10 percent. Note the rental houing firm i treated a a non-corporate firm in the model. 11

12 Deficit-Financed Tax Cut tranfer (other than ocial ecurity benefit) are reduced o that government debt grow at the teady tate rate of growth; and (3) government debt i ued to finance the tax cut for ten year and then all peronal income tax rate (wage, interet, dividend, and capital gain) are increaed proportionately o that government debt grow at the teady tate growth rate. The financing meaure become known to agent when the tax cut i introduced. Numerou other tax cut and financing arrangement could be evaluated, but are omitted due to pace and other conideration. For two reaon, we do not conider a reduction in government purchae a a financing mechanim. Firt, we cannot model the intergenerational welfare effect, a key part of the analyi, without making arbitrary aumption about the utility derived from the government purchae. Second, a an intitutional matter, it i unclear whether government purchae, which are et in annual appropriation bill and which often repond to military development and other volatile factor, could be reduced permanently a part of a fical reform. The following dicuion begin with a ummary of the macroeconomic effect of enacting each of the tax cut on price, conumption, labor upply, invetment, and output a well a the aociated intergenerational welfare reditribution. We examine the macroeconomic effect of cutting the tax rate on wage, interet, dividend and corporate income and the macroeconomic effect of increaing tax credit. We alo dicu the intergenerational welfare effect and macroeconomic revenue feedback effect of each of the tax cut. A. Immediate Spending Offet Table 6 how the macroeconomic effect of a 3.9 percent reduction in the average and marginal wage tax rate, a 22.1 percent reduction in the effective tax rate on interet income, a 50.6 percent reduction in the effective tax rate on dividend income, a 12.3 percent reduction in the effective tax rate on corporate income, and a 41 percent increae in peronal tax credit. In thi et of imulation, tranfer payment are immediately reduced to hold the amount of government debt contant. Under the wage tax cut, the before-tax interet rate increae in the year of enactment by 23 bai point and then immediately return to it approximate level in the initial teady tate. Labor upply increae by 0.4 percent in every year after reform. The before-tax wage rate decline initially by 0.1 percent a labor upply increae immediately by 0.4 percent (note that the capital tock i initially fixed). Labor upply increae becaue the income weighted after-tax wage rate increae by 1.6 percent in the year of reform (1.8 percent in the long-run). Invetment increae in all three production ector, with invetment in the houing ector increaing by twice a much a in the nonhouing ector in the year immediately following the reform. In the long run, invetment in the non-houing and houing ector increae by 0.5 percent. Conumption increae by 0.2 percent in the year of enactment and by 0.4 percent in the long run. GDP increae by 0.3 percent in the year of reform and by 0.4 percent in the long run. Under the interet income tax cut, the before-tax interet rate decreae in the year of enactment by 5 bai point and by 30 bai point in the long run. Labor upply increae by 0.1 percent in the year of enactment but then return to it initial teady tate level. The before-tax wage rate i initially unchanged and increae by 0.1 percent in the long run a the capital tock increae by 0.4 percent. Invetment increae in all three production ector, with invetment in the owner (rental) houing ector increaing by

13 Deficit-Financed Tax Cut (0.6) percent in the year of reform. Non-houing invetment initially increae by 0.3 percent in the year immediately following the tax cut. In the long run, invetment in the non-houing and rental houing ector increae by 0.4 percent, and invetment in the owner houing ector increae by 0.6 percent. Conumption decreae by 0.1 percent in the year of enactment and then return to it initial teady-tate level after 10 year. GDP increae by 0.1 percent in the long run. The macroeconomic effect of reducing the dividend (recall that we are auming the old view of dividend taxe) and corporate tax rate are imilar to each other. The before-tax interet rate increae by 24 (48) bai point in the year the dividend (corporate) tax rate i cut and then gradually decline to a level that i 4 (5) bai point higher than in the initial teady-tate. Under the dividend and corporate rate cut, labor upply increae by 0.1 percent in every year after enactment. Under the dividend (corporate) rate cut, the before-tax wage rate i initially unchanged and increae by 0.7 (0.5) percent in the long run a the capital tock increae by 2.7 (2.1) percent. Invetment in the owner and rental houing ector decreae 2.3 to 3.2 percent in the year of enactment. Non-houing invetment increae by 2.6 (1.9) percent in the year of enactment under the dividend (corporate) tax cut. In the long run, invetment in the nonhouing ector increae by 2.7 (2.1) percent under the dividend (corporate) rate cut. Conumption decreae by 0.1 to 0.2 percent in the year of enactment and increae by 0.3 to 0.4 percent in the long run. Under the dividend (corporate) tax cut, GDP increae by 0.7 (0.5) percent in the long run. One might think that corporate and dividend tax cut of the ame ize hould have identical quantitative effect, a well a imilar qualitative effect, ince they both reduce the effective tax rate on corporate invetment. The two taxe differ, though, in a ubtle, but crucial, repect. Due to accelerated depreciation, a corporate tax cut include a windfall gain for exiting capital, a ome deferred tax liabilitie are forgiven. In thi oldview world, there i no imilar windfall from the dividend tax cut. The dividend tax cut i therefore more powerful in promoting long-run growth becaue none of it i wated on a lump-um tranfer to holder of exiting aet. Thi i analytically imilar, but quantitatively maller, than the difference between adopting a wage tax and a conumption tax. We return to thi key difference in our dicuion of welfare effect. An increae in peronal tax credit offet by a reduction in tranfer doe not ignificantly affect the before-tax wage rate, labor upply, or GDP. The before-tax interet rate increae by 4 bai point in the year of enactment and then decline to a level 2 bai point lower than in the initial teady-tate. In the long run, invetment in the owner and rental houing ector increae by 0.1 percent and non-houing invetment increae by 0.2 percent. Conumption decreae by 0.1 percent in the year of enactment and then return to it initial teady tate level after 10 year. Both the tax credit and the tranfer payment are lump-um, but the ubtitution of the former for the latter provide a light boot to invetment becaue the tax credit are ditributed to younger cohort. B. Government Tranfer Offet after 10 Year Table 7 how the macroeconomic effect of the variou tax cut if government debt i ued to finance the tax cut for 10 year and tranfer payment are then decreaed to hold the growth rate of government debt contant at the teady-tate rate of growth in the economy. 13

14 Deficit-Financed Tax Cut Under the wage tax cut with a reduction in government tranfer payment after 10 year, the before-tax interet rate increae by 16 bai point in the year of enactment and by 8 bai point in the long run. The before-tax wage rate decline by 0.1 percent a labor upply increae immediately by 0.4 percent (note that the capital tock i initially fixed). Labor upply increae becaue the income weighted after-tax wage rate increae by 1.6 percent in the year of reform (1.8 percent in the long-run). In the year immediately following the tax cut invetment increae in all three production ector, with invetment in the houing ector increaing by twice a much a in the non-houing ector in the year immediately following the reform. In the long run, invetment in the non-houing and houing ector decreae by 0.1 to 0.2 percent. Conumption increae by 0.3 percent in the year of enactment and by 0.4 percent in the long run. GDP increae by 0.3 percent in the long run. The debt-to-gdp ratio increae from 29.7 percent in the initial teady tate to 33.3 percent in the long run. In thi cae, the increae in government debt relative to GDP increae interet rate and crowd out private invetment. A reduction in the tax rate on interet income accompanied by a reduction in government tranfer after 10 year ha very modet effect in the long run on the macroeconomic aggregate. The before-tax interet rate decreae in the year of enactment by 11 bai point and by 22 bai point in the long run. Labor upply, conumption and GDP are eentially unchanged from the initial teady-tate. In the year immediately after the tax cut, invetment increae in all three production ector, with invetment in the owner (rental) houing ector increaing by 0.8 (0.4) percent in the year of reform. Non-houing invetment increae by 0.1 percent in the year of the tax cut. In the long run, invetment in the non-houing and rental houing ector decreae by 0.1 to 0.2 percent a an increaed debt-to-gdp ratio crowd out private invetment. The debt-to-gdp ratio increae from 29.8 percent in the initial teady tate to 32.7 percent in the long run. The debt-to-gdp ratio in thi cenario increae by le than that caued by the wage tax cut with the ame financing aumption a a reult of the lower interet rate which reduce government pending. In thi cae, a under the immediate financing offet, the macroeconomic effect of reducing the dividend and corporate tax rate are imilar to each other. Cutting the dividend tax rate caue the before-tax interet rate to increae by 17 bai point in the year of enactment and in the long run. Cutting the corporate tax rate caue the before-tax interet rate to increae by 41 bai point in the year of enactment and then gradually decline to a level that i 17 bai point higher than in the initial teady-tate. Under the dividend and corporate rate cut, labor upply increae by 0.1 percent in every year after enactment. Under the dividend (corporate) rate cut, the before-tax wage rate i initially unchanged and increae by 0.4 (0.3) percent in the long run a the capital tock increae by 2.7 (2.1) percent. Invetment in the owner and rental houing ector decreae 2.7 to 3.5 percent in the year of enactment. Non-houing invetment increae by 2.3 (1.7) percent in the year of enactment under the dividend (corporate) tax cut. In the long run, invetment in the non-houing ector increae by 1.8 (1.2) percent under the dividend (corporate) rate cut. Conumption decreae by 0.1 to 0.2 percent in the year of enactment and increae by 0.3 to 0.4 percent in the long run. Under the dividend (corporate) tax cut, GDP increae by 0.5 (0.4) percent in the long run. An increae in peronal tax credit offet by a reduction in tranfer after 10 year doe not ignificantly affect the before-tax wage rate, labor upply, or conumption. The 14

15 Deficit-Financed Tax Cut before-tax interet rate decreae by 3 bai point in the year of enactment and then increae to a level 10 bai point higher than in the initial teady tate. In the long run, invetment in all three production ector decreae by 0.7 percent. GDP decreae by 0.1 percent in the long run. C. Acro-the-Board Tax Increae after 10 Year Table 8 how the macroeconomic effect of the variou tax cut if government debt i ued to finance the tax cut for 10 year and then an acro-the-board tax increae i implemented to hold the growth rate of government debt contant at the teady tate rate of growth in the economy. In the cae of a wage tax cut with an acro-the-board increae in tax rate after 10 year, the before-tax interet rate increae by 13 bai point in the year of enactment and by 15 bai point in the long run. The before-tax wage rate decline by 0.1 percent a labor upply increae immediately by 0.4 percent (note that the capital tock i initially fixed). Labor upply increae becaue the income-weighted after-tax wage rate increae by 1.6 percent in the 10 year following enactment. After 10 year, tax rate are increaed to hold the growth of government debt contant. A a reult, the increae in the after-tax wage rate decline from 1.6 percent to 0.6 percent in year In the long run, the after-tax wage rate increae by 0.04 percent a the net wage tax rate i reduced by one percent. Labor upply increae by 0.1 percent in the long run. In the year immediately following the tax cut, invetment in the houing ector increae by 0.3 to 0.6 percent while invetment in the non-houing ector i unchanged. In the long run, invetment in the non-houing ector decreae by 1.6 percent and invetment in houing decreae by 0.7 to 1.3 percent. Conumption increae by 0.3 percent in the year of enactment and decreae by 0.1 percent in the long run. GDP increae by 0.3 percent in the year of enactment and decreae by 0.3 percent in the long run. The debt-to-gdp ratio increae from 29.7 percent in the initial teady-tate to 33.6 percent in the long run. In thi cae, debt financed tax cut increae government debt relative to GDP and thu increae interet rate and crowd out private invetment. Tax burden on invetment are alo higher in the teady tate, due to the tax increae. Decreaing the tax rate on interet income with an acro-the-board tax increae after 10 year ha very modet effect on the fundamental macroeconomic aggregate in the long run. The before-tax interet rate decreae in the year of enactment by 13 bai point and by 18 bai point in the long run. In the 10 year after enactment, the after-tax wage rate and labor upply are unchanged. After 10 year, the increae in the marginal wage tax rate and a decline in the before-tax wage rate, which i related to lower growth in invetment, lead to a 0.2 percent decreae in labor upply in the long run. In the year immediately after the tax cut, invetment in owner houing increae by 0.9 percent and invetment in rental houing increae by 0.2 percent. Non-houing invetment i unchanged in the firt few year following the tax cut. The acro-the-board tax increae reduce invetment beginning in In the long run, invetment in the non-houing ector decreae 1.3 percent, and invetment in the owner and rental houing ector decreae by 0.5 and 1.1 percent. In thi cae, the debt-to-gdp ratio increae from 29.8 percent in the initial teady tate to 32.9 percent in the long run. The debt-to-gdp ratio increae by le than under the wage tax cut, with the ame financing aumption, becaue the decreae in the interet rate reduce government pending on interet payment. 15

16 Deficit-Financed Tax Cut The macroeconomic effect of reducing the dividend and corporate tax rate are imilar to each other in thi cae, an acro-the-board tax increae, a they were under the other financing aumption. Cutting the dividend tax rate caue the before-tax interet rate to increae by 14 bai point in the year of enactment. After 2017, the before-tax interet rate increae teadily until it i 25 bai point higher in the long run. Cutting the corporate tax rate caue the before-tax interet rate to increae by 37 bai point in the year of enactment and then decline to a level that i 15 bai point higher than in the initial teady tate in the following year. After 2017, the before-tax interet rate gradually increae until it i 25 bai point higher than in the initial equilibrium. Under the dividend and corporate rate cut, labor upply decreae by 0.3 percent in the long run. Under the dividend (corporate) rate cut, the before-tax wage rate i initially unchanged and increae by 0.2 (0.0) percent. Under the dividend rate cut, invetment in the owner and rental houing ector decreae by 3.0 to 3.8 percent in the year of enactment. Nonhouing invetment increae by 2.2 percent in the year of enactment. Under the corporate rate cut, invetment in the owner and rental houing ector decreae by 2.6 to 3.3 percent in the year of enactment. Non-houing invetment increae by 1.5 percent in the year of enactment. In the long run, invetment in the non-houing ector increae by 0.4 percent under the dividend rate cut and decreae by 0.2 percent under the corporate rate cut. Conumption change by 0.0 to -0.1 percent in the year of enactment and by 0.0 to -0.2 percent in the long run. In the long run, a dividend tax cut doe not change the level of GDP, while the corporate tax cut caue GDP to decreae by 0.2 percent. An increae in peronal tax credit with an acro-the-board tax increae after 10 year increae the before-tax interet rate by 20 bai point in the long run. The beforetax wage rate gradually decreae a government debt crowd out private capital and the tax increae depre invetment. In the long run, the before-tax wage rate decreae by 0.5 percent and labor upply decreae by 0.4 percent. Invetment decreae in all three production ector. In the long run, non-houing invetment decreae by 2.5 percent, and owner and rental houing invetment decreae by 1.5 and 2.2 percent. Conumption decreae by 0.6 percent and GDP decreae by 0.8 percent in the long run. The debt-to- GDP ratio increae to 34.5 percent. D. Summary of Macroeconomic Effect A one would expect, all four of the reduction in ditortionary taxe increae long-run GDP when they are offet by an immediate reduction in tranfer payment. The GDP boot i larget for reduction in dividend and corporate taxe, becaue thoe taxe are more ditortionary. There i no GDP boot from the increae in peronal tax credit. 16

17 Deficit-Financed Tax Cut When the reduction in tranfer payment i delayed by 10 year, however, the long-run effect are le beneficial. In each cae, the GDP boot i diminihed by 0.1 or 0.2 percentage point. Thi dimunition bring the boot in the interet-tax-cut cae to eentially zero, but a noticeable GDP gain till appear in the other three reduction of ditortionary taxe. The long-run gain are maller becaue the increae in debt crowd out private invetment (there i virtually no change in labor upply). Notably, long-run non-houing invetment i increaed by cut in wage taxe and interet taxe when the tranfer payment cut i immediate, but uch invetment i reduced in the preence of the 10-year lag. The dividend and corporate cut, which are more directly targeted to invetment, till boot long-run non-houing invetment even with the 10-year lag, but by ignificantly le than with an immediate tranfer cut. The macroeconomic effect are much wore when tax rate are increaed to finance the deficit. Long-run GDP fall in every cae, except for the dividend-tax cut under which GDP i virtually unchanged. Non-houing invetment fall in every cae, except the dividend-tax cut. Long-run labor upply fall in every cae, except for the wage-tax reduction. With tax-increae financing, crowding out i reinforced by the ditortionary effect of the tax increae. E. Intergenerational Welfare Effect A dicued in ection III, examining the macroeconomic effect of tax policy change ha become an important tool for tax policy analyt in the None of the analye to date, however, have provided policy maker with anything more than imple macroeconomic aggregate a preented in the dicuion above. Thi omiion i unfortunate, ince macroeconomic aggregate are not alway reliable indicator of whether certain policie make individual better off, are not ufficient to compare alternative policie, and do not allow policy maker to examine the effect of policie acro group. In thi ection, we preent etimate of the intergenerational welfare effect of the variou policie. Figure 1 how the intergenerational welfare effect of the five policie auming that government tranfer are cut immediately o that government debt grow at the teady tate rate of the economy. For generation alive at the time of reform (generation with a year of birth between negative 54 and 0) the larget welfare gain occur under the corporate rate cut, the dividend rate cut, and the interet income tax cut. The difference between the dividend and the corporate tax cut again reflect the windfall gain to exiting aet built into the latter, but not the former. The corporate tax cut i more beneficial to the early generation who hold thoe aet and i therefore le beneficial to later generation. In the cae of an increae in peronal tax credit all exiting generation are wore off while future generation are better off. Thi reflect the fact that the tax credit are ditributed to younger cohort, on average, than the tranfer payment. A the generational accounting literature ha long taught, a balanced-budget lump-um reditribution from old to young harm current generation and help future generation. The deficit-financed tax cut, hown in Figure 2 and 3, uniformly reduce the well-being of future generation. Thi welfare lo occur even when thoe generation are bequeathed a more efficient tax ytem (a i true in all of the cae in which reduction in ditortionary taxe are financed by delayed cut in tranfer payment). The reaon for their welfare lo i the hift in net fical burden; deficit financing move uch 17