Office of Tax Analysis U.S. Department of the Treasury. A Dynamic Analysis of Permanent Extension of the President s Tax Relief

Transcription

1 Office of Tax Analyi U.S. Department of the Treaury A Dynamic Analyi of Permanent Extenion of the Preident Tax Relief July 25, 2006

2 Executive Summary Thi Report preent a detailed decription of Treaury dynamic analyi of the Preident propoal to permanently extend the tax relief proviion enacted in 2001 and 2003 that are currently et to expire at the end of Thee enacted proviion include: Lower tax rate on ordinary income; Lower tax rate on dividend and capital gain; A ten-percent individual income tax rate bracket; Doubling of the child tax credit; and Reducing marriage tax penaltie. The purpoe of the report i to provide a more in-depth, tranparent undertanding of dynamic analyi, while alo illutrating the poitive contribution the tax relief, together with pending reduction, can be expected to continue to make to the U.S. economy. In addition, the analyi how the importance of making the tax proviion permanent for the U.S. economy long-term economic growth. Dynamic Analyi Dynamic analyi goe beyond traditional analyi of tax policy by focuing on the broad economic effect in both the hort and long term. Simply, dynamic analyi provide a more comprehenive and complete approach to analyzing tax policy by including it effect on the overall ize of the economy and other major macroeconomic variable. The Preident FY 2007 Budget propoe to create a diviion of dynamic analyi within the Department of Treaury Office of Tax Analyi. The Economic Benefit of Tax Relief A evidenced by key economic indicator uch a increaed capital invetment and Gro Dometic Product (GDP), and trong job growth, the Preident tax relief played an important role in trengthening the U.S. economy a it wa coming out of the recent receion, and in the longer-term by increaing the after-tax reward to work and aving. Lower tax rate enable worker to keep more of their earning, which increae work effort and labor force participation. The lower tax rate alo enable innovative and rik-taking entrepreneur to keep more of what they earn, which further encourage their entrepreneurial activity. The lower tax rate on dividend and capital gain lower the cot of equity capital and reduce the tax biae againt dividend payment, equity finance, and invetment in the corporate ector. All of thee policie increae incentive to work, ave, and invet by reducing the ditorting effect of taxe. Capital invetment and labor productivity will thu be higher, which mean higher output and living tandard in the long run. Treaury ha conducted it dynamic analyi uing a model that account for the effect of thi greater work effort, increae in aving and invetment, and improved allocation of reource on the ize of the economy. While thi model capture many apect of a modern economy and economic behavior, other are not reflected in the model. For example, the model aume that reource are fully employed in the economy and that capital i only omewhat mobile internationally. Thee are area for future development. i

3 Different Component of Tax Relief Have Different Effect on the Economy Treaury dynamic analyi of the Preident tax relief indicate that making the tax relief permanent can be expected to increae the level of annual output (i.e., national income) ultimately by about 0.7 percent. The analyi alo how eparately the effect of the Preident tax relief in three part reflecting: 1) the lower tax rate on dividend and capital gain; 2) the lower tax rate on ordinary income (i.e., the top four rate bracket); and 3) the 10-percent tax rate bracket, higher child tax credit, and marriage penalty relief. Thi decompoition reveal that thee tax relief component are likely to have very different effect on future economic activity. For example, extending jut the lower tax rate on dividend and capital gain increae output in the long run by 0.4 percent, but when the lower tax rate for the four top income tax bracket are extended a well, output increae by a total of 1.1 percent in the long run. Financing Tax Relief Government Spending Reduction over Increaed Tax Rate The analyi reveal that the long-run effect of thee policie depend crucially on whether they are financed by lower pending or higher taxe in the future and are enitive to aumption on underlying parameter. The iue of how, or even if, thee policie need to be financed remain a ource of dicuion among economit. The analyi preented here ugget thee policie will reult in ubtantially more economic activity if they are financed by a future reduction in government pending than if they are financed by future tax increae. If the tax relief i financed by future tax increae that i, if the aggregate amount of tax relief i temporary then it may reult in lower output in the long run. For that reaon, the Adminitration ha emphaized permanence for the tax relief and pending retraint in it Budget. ii

4 1. Introduction A Dynamic Analyi of Permanent Extenion of the Preident Tax Relief Thi Report preent a detailed decription of Treaury dynamic analyi of the Preident propoal to permanently extend the tax relief proviion enacted in 2001 and 2003 that are currently et to expire at the end of Thee proviion include the lower tax rate on ordinary income, the lower tax rate on dividend and capital gain, the 10-percent individual income tax rate bracket, a doubling of the child tax credit, and a reduction in marriage tax penaltie. Tax relief can be important when the economy i performing below it full potential, and can increae it potential in the longer term. In 2003, real GDP wa below it potential level and the unemployment rate wa elevated. The tax relief enacted in 2001 and 2003, together with reduction in hort-term interet rate by the Federal Reerve, helped timulate economic growth and move the economy out of the 2001 receion more quickly. Previou Treaury analyi uing the Macroeconomic Advier macro-econometric model etimated that without the tax relief paed in 2001, 2002, and 2003, a many a 3 million fewer job would have been created by the end of 2004 and real GDP would have been a much a 3.5 to 4.0 percent lower. Beyond thi hort-term economic timulu, the Preident tax relief alo help encourage economic growth in the longer term by increaing the after-tax reward from work, aving, and invetment. The lower tax rate enable worker to keep more of their earning, which increae work effort and labor force participation. The lower tax rate alo enable innovative and riktaking entrepreneur to keep more of what they earn, which further encourage their entrepreneurial activity. The lower tax rate on dividend and capital gain lower the cot of equity capital and reduce the tax biae againt dividend payment, equity finance, and invetment in the corporate ector. All of thee policie improve incentive for work, aving, and invetment by reducing the ditorting effect of taxe. Capital invetment and labor productivity will thu be higher, which mean higher output and living tandard in the long run. The Treaury Department dynamic analyi relie on a model that take into account the effect of work effort, increae in aving and invetment, and improved allocation of reource on the ize of the economy. The overlapping generation (OLG) general equilibrium model ued for thi analyi (decribed in detail in the appendix to thi report) i tructured to account for the effect of change in the effective tax rate on capital and labor income and the conequent effect on economic growth. Repreentative conumer and firm incorporate future price into their current period deciion of how much to ave, work, and produce. Output i generated by four production ector, and individual level deciion of repreentative conumer determine the aggregate level of labor upply and aving in each year. While thi model capture many apect of the economy and economic behavior, other apect are not reflected in the model. For example, the model ignore cyclical diruption in the employment of capital and labor, auming intead that all reource in the economy are alway fully employed. The model include a relatively imple repreentation of international capital flow in which capital i only omewhat mobile internationally. There i no uncertainty in the 1

5 model and houehold and firm exhibit perfect foreight regarding future price and tax rate. Thee are area for future development. Thi analyi how the likely economic effect of making the tax relief permanent. The reult indicate that the level of annual output (i.e., national income) may ultimately be higher by 0.7 percent becaue of the combined effect of the Preident tax relief. The analyi alo how eparately the effect of the Preident tax relief in three part reflecting: 1) the lower tax rate on dividend and capital gain; 2) the lower tax rate on ordinary income (i.e., the top four rate bracket); and 3) the 10-percent tax rate bracket, higher child tax credit, and marriage penalty relief. Thi decompoition reveal that the tax relief component are likely to have very different effect on future economic activity. For example, extending jut the lower tax rate on dividend and capital gain increae output in the long run by 0.4 percent, but when the lower tax rate for the four top income tax bracket are extended a well, output increae by a total of 1.1 percent in the long run. Extending the remainder of the tax relief the 10 percent rate, the expanion of the child tax credit, and the reduction in marriage penaltie timulated economic activity during and immediately after the receion and erved other purpoe, uch a making the tax code more progreive. However, thee element of the tax relief do not have poitive growth effect in the longer term in way that thi type of model can meaure. The analyi reveal that the long-run effect of thee policie depend crucially on how they are eventually financed and are enitive to aumption on underlying parameter. The iue of how, or even if, thee policie need to be financed remain a ource of dicuion among economit. The analyi preented here ugget thee policie will reult in ubtantially more economic activity if they are financed by a future reduction in government pending than if they are financed by future tax increae. If the tax relief i financed by future tax increae that i, if the tax relief i temporary it may well reult in lower output in the long run. In effect, the temporary tax relief mut be paid back with interet through future tax increae, which implie that future tax rate increae compared to current law. For that reaon, the Adminitration ha emphaized permanence for the tax relief and pending retraint in it Budget. The enitivity of the reult to financing and parameter aumption i decribed in detail below. The remainder of thi report i organized a follow. The next ection decribe previou work done by Treaury that etimated the hort-run economic effect of the Preident tax relief. Section 3 decribe the model ued in the permanence analyi in greater detail. Section 4 outline the methodology employed in imulating the economic effect of extending the 2001 and 2003 tax relief and dicue ome of the limitation of the model. Section 5 decribe and explain the reult and the lat ection conclude. 2. Effect of the Preident Tax Relief in the Near Term The focu of thi Report i on the future economic effect of permanently extending the Preident tax relief. A decribed in the introduction, the model ued for thi analyi aume that the economy i alway performing at it potential. Thi aumption implifie the model and allow for a more detailed repreentation of houehold labor upply and aving behavior in both 2

6 the near term and the long run. Yet thi implification implie the model ued for thi report i not able to capture the hort-run timulu that tax relief may provide when the economy i operating below potential. Such a ituation exited when the Preident tax relief wa paed in 2001 and 2003; real GDP wa below it potential level and the unemployment rate wa elevated. The Treaury Department previouly compared how the economy would have performed if there had been no tax relief uing a different type of model that i deigned to capture the interaction of economic ector a the economy fluctuate around it potential growth path. Thee model attempt to account for change in the level and growth of GDP, employment, inflation, and interet rate. Short-run change in monetary and fical policie are important determinant of acceleration and deceleration of employment and output in thee model. In thi earlier analyi, the Treaury Department ued the Macroeconomic Advier macroeconometric model to etimate how the economy would have performed had there been no legilated fical timulu from 2001 through Thi analyi found that the tax relief increaed employment and output ubtantially above what would have occurred otherwie. Specifically, Treaury found that, without enactment of the Economic Growth and Tax Relief Reconciliation Act of 2001, the Job Creation and Worker Aitance Act of 2002, and the Job and Growth Tax Relief Reconciliation Act of 2003: (1) by the econd quarter of 2003, the economy would have created a many a 1.5 million fewer job and GDP would have been a much a 2 percent lower, and (2) by the end of 2004, the economy would have created a many a 3 million fewer job and real GDP would be a much a 3.5 to 4.0 percent lower. Note that the analyi decribed in thi ection etimate the economic effect that the Preident tax relief ha already had on the economy, auming that interet rate followed the ame path a they did hitorically from 2001 forward. The remainder of the paper dicue the likely future economic effect of making the Preident tax relief permanent. 3. Model decription For the remainder of the analyi in thi Report, the Treaury Department ued a conventional neoclaical growth model with overlapping generation of taxpayer developed by Tax Policy Advier, LLC. 1 In thi life-cycle model, tax policy affect the incentive to work, to ave and invet, and to allocate capital among competing ue. Repreentative conumer and firm incorporate future price into their current period deciion of how much to ave, work, and produce. Output i generated by four production ector, and individual level deciion of repreentative conumer determine the aggregate level of labor upply and aving in each year. An overview of the model follow, with important equation and further explanation provided in the appendix. Firm Behavior Firm behavior i modeled for each of the four production ector corporate, noncorporate, owner-occupied houing, and rental houing. In the owner-occupied houing ector, home owner are treated a "firm" who produce houing and rent it to themelve, taking into account 1 See 3

7 the tax advantage of home ownerhip. Each production function take the tandard Cobb- Dougla form. Firm manager chooe the optimal level of labor demand and invetment to maximize the value of the firm, or profit, in each period. Invetment in each ector i determined according to the q theory of invetment modified to include adjutment cot. Thi implie that firm will continue to invet a long a the increae in the value of the firm i greater than the after-tax cot of invetment. Firm manager explicitly calculate the time path of invetment in repone to a change in the tax tructure a a function of the tax-induced change in q, which denote the ratio of the market value of capital aet to their replacement cot, taking into account convex cot of adjuting the level of invetment from it teady tate level. Difference in the level of depreciation allowance for tax purpoe and economic depreciation are modeled explicitly, a i the value of the exiting tax bai at any point in time. The debt-to-capital ratio i aumed to be fixed in each indutry, and dividend in the corporate ector are aumed to be a fixed fraction of after-tax corporate earning. The model aume the traditional view of dividend taxe, which implie dividend taxe increae the cot of capital to firm. Individual Behavior The model ha a conventional overlapping generation tructure. All individual in a given cohort are identical, with each living for 55 year, the lat 10 of which are pent in retirement. 2 Each individual ha perfect foreight and chooe conumption (and thu aving) to maximize lifetime utility an aggregation of utility in each of the 55 period of the lifecycle, dicounted at a fixed rate of time preference that i common to all individual ubject to a lifetime budget contraint that take into account a hump-haped age-wage profile, inheritance and a target bequet. Utility in each period i a CES function of leiure and an aggregate conumption good which i in turn an aggregation of four good a compoite good produced by the corporate ector, a compoite good produced by the non-corporate ector, owner-occupied houing, and rental houing. The intertemporal elaticity of ubtitution i aumed to be 0.35 and the intratemporal ubtitution elaticity between good and leiure i aumed to be Senitivity to thee and other parameter i conidered in more detail below. Government Behavior The model include a imple characterization of the Social Security program. Government ervice are eparable in the individual utility function and government debt i a contant fraction of Gro National Product (GNP) in the initial teady tate. 2 Thi formulation, in effect, exclude an individual life prior to joining the labor force, but doe include both individual working year plu their retirement. An alternative approach to modeling the retirement deciion would be to allow the retirement age to be endogenou o that individual could come out of retirement and rejoin the work force in repone to reform-induced change in the after-tax wage. Thi potential labor upply repone i precluded by auming a fixed retirement age. 3 See Elmendorf (1996), Engen, Gravelle, and Smetter (1997) and Altig et al. (2001) for dicuion on the plauible range of value for thee parameter. Thee parameter value yield a Frich elaticity of labor upply, which meaure the labor upply elaticity holding the marginal utility contant, equal to 0.4. Thi value i conitent with the range of etimate reported in Browning, Hanen, and Heckman (1999). 4

8 The government finance government expenditure by collecting taxe, and iuing government debt. The tax intrument available to the government in the initial equilibrium include the following: (1) a corporate income tax; (2) an individual income tax with a progreive wage income tax tructure and a tax bae that i adjuted for variou excluion, exemption, deduction and credit; and (3) contant rate capital income taxe applied at different average rate to non-corporate buine income, interet income, dividend and capital gain. An important feature of thi type of model i that a permanent reduction in taxe, a compared to the baeline, would lead to an unutainable accumulation of government debt relative to GNP and the model will not converge without an offetting change to tabilize the debt-to-gnp ratio. In thi type of model, the tax relief i typically financed by an offetting change in taxe or pending, that can occur in the future or contemporaneouly with the initial policy change and can take a multitude of form. In thi analyi it i aumed that the government financing requirement i atified by either cutting future government pending or raiing future taxe, in part to illutrate the enitivity of the reult to the financing aumption. International Capital Flow Although the focu of the model i on the U.S. dometic economy, it include a imple repreentation of international capital flow, which are aumed to repond to difference in after-tax rate of return in the U.S. and the ret of the world through a contant elaticity expreion. 4 Thi approach repreent a compromie between the tandard cloed economy approach and the alternative of a completely open economy in which capital i perfectly mobile and the international return to capital i fixed. A more ophiticated modeling of the international flow of good and capital would be a marked improvement over the current verion of the model. 4. Methodology In the teady tate, per-capita growth in the model i equal to a contant rate of technological change. In the initial teady tate, the model tax parameter are calibrated to match current law average marginal effective tax rate by income ource over the budget window. Simplifying aumption were made in order to meld the data into the tylized model. Firt, given the requirement of contant tax rate in the teady tate, the initial tax rate were et equal to the average of current law rate over the period , when tatutory rate are unchanging. Second, the initial teady tate aume that current law police are fically utainable. That i, tax revenue in each period are jut large enough to pay for government pending and tranfer payment, including interet on the government debt, o that the government debt-to-gnp ratio i contant. 5 The initial hare of tax revenue a a percentage of GNP i et to match current law averaged over the year Thi elaticity i et equal to 0.2 in the bae cae, which implie that international capital flow are not very enitive to difference in the after-tax rate of return in the U.S. compared to the ret of the world. 5 Note that thi approach ignore the tructural fical imbalance of the Social Security and Medicare ytem, but thi aumption eem appropriate in generating the likely independent effect of a tax change and i commonly ued in thi type of analyi. For example, ee Auerbach (2002). 5

9 The tax relief i decompoed into three part to how the economic effect of each: 1. Extend the lower rate on dividend and capital gain. (Dividend and capital gain are taxed at a top rate of 15 percent, a compared to a top rate on dividend of 39.6 percent and for long-term gain of 20 percent in the abence of tax relief); 2. Extend the reduction in the top four ordinary individual rate. (Thee rate are maintained at 25, 28, 33 and 35 percent, a compared to the rate of 28, 31, 36 and 39.6 percent that would apply beginning in 2011 under current law. The repeal of the phaeout of peronal exemption and itemized deduction [PEP and Peae proviion] i alo extended); and, 3. Extend the higher child tax credit ($1,000 per child), reduction in marriage tax penaltie (by increaing the tandard deduction and increaing the ize of the 15 percent bracket for joint filer), and the 10-percent tax rate bracket. 6 The percentage decline in average marginal tax rate by income ource compared to current law for the year i hown in Table 1. 7 Extending the lower tax rate on dividend lead to more than a 50 percent decline in the average marginal dividend tax rate compared to current law for the year 2011 through Extending the relief on capital gain lead to more than a 20 percent decline in the average marginal tax rate on capital gain for the ame period. Lowering ordinary rate lead to a decline in the average marginal tax rate on labor income of 5.6 percent, while the average marginal rate on mall buine income (income from ole proprietorhip, partnerhip, and S-corporation) fall by 11.4 percent. Extending the reminder of the tax relief ha only mall effect on the change in marginal tax rate. 8 Financing the Tax Relief A dicued above, an important feature of thi type of model i that tax relief mut be financed by an offetting change in government revenue or pending to tabilize the ratio of government debt to GNP. There are numerou poibilitie for atifying the government intertemporal budget contraint and two are examined in thi analyi: (1) the tax relief i permanent and future government pending i reduced, and (2) future taxe are increaed. Specifically, in thi analyi, the tax relief i aumed to remain in place through the end of the 10-year budget window (i.e., 2016), holding government pending equal to the baeline amount during thi period, and iuing additional government debt relative to the baeline to account for the decline in tax revenue over the budget window. The tax relief i then financed by either: (1) adjuting government conumption pending in each year after the 10-year budget window to hold the ratio of government debt-to-gnp at the ratio that exit in the firt year after the budget window (2017), or (2) adjuting all income tax rate proportionally in each period after the budget window to hold the government debt-to-gnp ratio equal to the value it take in the firt year 6 The economic effect of the repeal of the etate tax are not included in thi analyi. There i coniderable uncertainty regarding the likely behavioral repone to repealing the etate tax, and the target bequet motive ued in thi model i not very flexible in capturing the range of likely repone. 7 The average marginal rate are weighted by income from that ource. 8 The decline in the marginal tax rate on wage actually become maller (5.1 percent decreae) when the full tax relief i extended. Thi appear to be the reult of more taxpayer being affected by the AMT and the longer phaeout of the child tax credit. 6

10 after the 10-year budget window (2017). 9 The firt financing option i conitent with the Adminitration policy of pending retraint. The econd financing option, in effect, model the tax relief a temporary, which require that future taxe increae enough to pay for the temporary decline in taxe with interet. Senitivity to Underlying Parameter Aumption The reult alo depend on how reponive houehold and firm are to change in after-tax price, uch a the wage rate and the interet rate. The behavioral parameter ued for the bae cae imulation are hown in Table 2. There are three primary parameter that affect the reponivene of houehold labor upply and aving to tax change: the intertemporal elaticity of ubtitution, the intratemporal elaticity of ubtitution between the compoite conumption good and leiure, and the initial hare of leiure in the time endowment. The bae cae imulation ue value for thee parameter that approximate central tendency etimate. However, there i uncertainty about the exact value of thee parameter and reult are alo preented that conider low and high value for thee parameter. 10 In addition, the degree to which houing and non-houing good are ubtitutable i adjuted. The approach taken for thi Report i to adjut the parameter a a group, rather than individually, motly to facilitate eae in preentation. 11 Thi approach provide only limited information on the importance that any given parameter would have on the reult, but it provide an overall ene of the robutne of the reult and highlight the uncertainty that remain in the economic literature on the likely reponivene of taxpayer to change in tax rate. Limitation of the Model The model ued for thi analyi capture many of the likely economic effect that would reult from extending the Preident tax relief, including incentive effect on houehold labor upply and aving, the interectoral reallocation of capital that would reult from reducing the double taxation of corporate profit, and the crowding out of private invetment that would occur by financing the tax relief through iuing government debt. Like all economic model, thi model employ important implifying aumption, and other economic model, which employ different implifying aumption, could yield different economic reult from extending the Preident tax relief. The model ued in thi analyi depart from economic reality in the following way. 9 Auerbach (2002) and Auerbach and Kotlikoff (1987) employ a imilar approach when examining the effect of a temporary tax decreae financed with future tax increae. The Congreional Budget Office (2004, 2006) alo make imilar financing aumption, but the reduction in government pending (or the increae in income taxe) i phaed in over a 10-year period after the end of the 10-year budget window which allow the debt-to-gnp ratio to rie omewhat more in the long-run. 10 The Frich labor upply elaticity equal 0.18 under the low repone parameter and 0.75 under the high repone parameter. Thi i conitent with the reult urveyed by Browning, Hanen and Heckman (1999), and recent paper by Ziliak and Kniener (1999, 2005) and Lee (2001), which etimate the Frich labor upply elaticity for men range between 0.0 and 0.5. The econometric literature ha generally found larger labor upply repone for women compared to men, but there are few tudie that meaure the Frich labor upply elaticity for women. Aaronon and French (2002) ugget thi value i believed to be around 1. Thi Report aume the Frich labor upply elaticity for women range between 0.55 and 1.25, and that women account for one-third of labor earning. 11 A imilar approach i taken by Roger (1997). For each imulation, thi Report alo adjut the rate of time preference in order to maintain the initial capital-output ratio. 7

11 Firt, thi model doe not account for hort-term deviation in output from potential GNP. Thi implie that the model doe not capture ome of the hort-run benefit of tax relief when it occur at a time when the economy i below it potential, which occurred with the 2001 and 2003 tax relief. Section 2 of thi Report decribe a eparate Treaury analyi of the effect of the Preident tax relief that include thee cyclical effect. Second, the treatment of international capital flow i quite imple. A broader model would employ a more ophiticated repreentation of thee flow and would alo include international trade in good. The limited role of international capital flow allowed in thi model ha only a minor impact on the economic reult in thi analyi. It i unclear whether a broader model of international trade and capital flow would lead to reult that are maller or larger in magnitude than the reult preented in thi paper. Third, a mentioned above, the model aume that the traditional view of dividend taxation hold, which implie that taxe on dividend increae the cot of inveting in the corporate ector. An alternative approach that i termed the new view of dividend taxation ugget that dividend taxe are capitalized into the value of the firm, but do not affect marginal invetment deciion. 12 The degree to which each view repreent an accurate portrayal of the economy remain an unettled iue. Recent reearch ugget a egmented market, with ome firm behaving in a manner conitent with the traditional view and other firm behaving in a manner conitent with the new view. 13 To the extent that the new view hold, the output gain reulting from extending the lower tax rate on dividend found in thi analyi are likely to be overtated. However, thi model alo doe not include a meaure of other efficiency gain that would likely reult from lowering the tax rate on dividend due to reducing the ditortion between debt and equity financing. Moreover, the model aume that the level of dividend and corporate payout deciion are held fixed. 14 It i not clear whether a fuller model that account for both new view firm and thee other financial ditortion would how larger or maller effect overall. Fourth, thi model aume perfect certainty and perfect competition. Of coure, a uggeted by empirical reearch, ome individual ave a a precaution againt unforeeen event and at leat a portion of aving i not very enitive to change in the interet rate. Thu, the implied elaticity of aving with repect to the after-tax interet rate in the certainty model i likely to be higher than in a model that incorporate rik. To partially offet the lack of rik in the model, houehold have a imple target bequet motive that tend to mitigate the aving repone. 15 On the other hand, ome model with imperfect competition find that the ditortionary effect of capital income taxation are larger than model that aume perfect competition For an excellent overview of thee iue, ee Zodrow (1991). 13 For example, ee Auerbach and Haet (2003). 14 The importance of thee ditortion i alo diminihed to the extent the new view hold. 15 It i worth noting, however, that Hurt, et al., (2005) recently etimated that precautionary aving account for le than 10 percent of total wealth. 16 For example, ee Judd (2002). 8

12 Fifth, thi model likely overtate the economic cot of deficit finance of temporary tax relief a the rate of return to government bond in the model i greater than the rate of growth in GNP. 17 Hitorically, the average return on government debt i below the average growth rate of the economy, which implie it might not be neceary to increae taxe in the future in order to tabilize the government debt ratio. Some reearch ugget that the need to raie future tax rate to pay for temporary tax relief only occur with a mall probability. 18 Finally, the financing aumption ued in thi model are conventional for thi type of analyi and are not meant to be prediction of what policie actually would be et by thi Adminitration or by a future Adminitration or Congre. Numerou other policy precription alo could be employed. If the revenue cot of the tax relief i financed through reduction in government pending, then the ooner thoe reduction begin, the maller would be the crowding out effect on private invetment and the larger would be the increae in long-run output. If the revenue cot of the tax relief i financed through future tax increae, then the way future taxe are increaed would greatly affect the long-run reult; the more the future tax increae affect marginal rate, the more future economic output will uffer a a conequence. 5. Decription of reult A decribed above, reult are preented auming that the tax relief i financed either through a future decreae in government pending or a future increae in taxe. The firt year in the model i et to be Houehold and firm in the tax relief imulation anticipate the future continuation of lower tax rate after 2010 and the offetting fical policy of reducing government conumption or increaing tax rate beyond the budget window. However, the macroeconomic effect for the firt four year of the budget window ( ) are generally mall a tax rate do not change between the different imulation for thoe year. Reult are preented in Table 3 and 4, and dicued below for only the lat ix year of the budget window ( ) and for the long run. Tax Relief Financed with Future Decreae in Government Spending For thi et of reult, the model aume that government conumption purchae (i.e., government pending) adjut after 10 year to tabilize the government debt-to-gnp ratio. In the model, government conumption purchae do not enter houehold utility function and only indirectly affect houehold deciion through market price. 19 Prior to 2017, the debt-to-gnp ratio i allowed to rie, but, beginning in 2017, government purchae decline in each year to hold the government debt ratio fixed. Mot of the economic effect of the tax relief can be explained by examining houehold budget and price. The tax relief lead to offetting ubtitution and income effect for both 17 The model aume that the after-tax rate of return to government bond, private bond and corporate equity are equal. 18 See Ball, Elmendorf, and Mankiw (1998). 19 That i, government pending i not valued by houehold. An alternative aumption i that valued government pending decreae, uch a government tranfer payment to individual. Thi would motly eliminate the income effect of the tax relief and lead to larger output effect. A more detailed modeling of the government ector would alo be an improvement in the model. 9

13 houehold leiure and conumption choice. The reduction in the marginal tax rate on labor and capital income increae the price of current leiure and conumption, and houehold repond by upplying more labor and aving through the ubtitution effect. 20 The reduction in tax liabilitie increae houehold after-tax wealth, and houehold deire to conume more leiure and conumption through an income effect. There i an additional income effect (termed the human wealth effect) which upport an initial increae in labor upply and aving. Thi effect arie from the increae in the after-tax interet rate that reult from the lower tax on capital, but thi effect become le important over time a the interet rate decline a capital accumulate. However, other effect alo are at work. The lower tax rate on dividend lower the effective tax rate on invetment in the corporate ector relative to the other ector. The reduction in the double tax on corporate profit reult in a more even taxation of invetment acro production ector, a more efficient allocation of capital, and an increae in output. In addition, when lower taxe on capital income are financed initially by iuing government debt, private invetment i crowded out by an increae in government borrowing. Private aving may increae a a reult of the tax relief (and may be augmented by capital inflow from abroad), but private invetment will generally not increae by the ame amount becaue a portion of the increae in private aving fund the increae in government debt. When the majority of the tax relief i on labor income, the crowding-out effect i even larger and private invetment could even decline in the hort run. When government purchae decline after the budget window to tabilize the government debt ratio, more private aving i releaed to fund private invetment, although ome crowding out of private invetment doe perit in the long run. The hort-run and long-run effect for all three tep of the permanence propoal under the bae cae parameter are hown in Table 3 for the two financing aumption. The reult from lowering the dividend and capital gain rate are hown in column (1). Subtitution effect dominate when capital gain and dividend rate fall in 2011, and private aving and invetment increae in both the hort run and long run. The capital tock increae by an average of 0.2 percent from compared to the baeline and GNP increae by 0.1 percent. 21 The increae in output i helped by a more efficient allocation of capital that come from reducing the double taxation of corporate invetment. In the long run, the capital tock increae by 1.2 percent, and output increae by 0.4 percent, with a mall decline in labor upply of 0.1 percent. When reduction to the top four ordinary income tax rate are extended a well, crowding out during the budget window i more pronounced and the average increae in the capital tock i jut 0.1 percent for the year 2011 through Dometically funded invetment actually decline during thi period, but capital inflow from abroad lead to an overall increae in the 20 The reduction in the wage tax rate lead to a hifting from leiure toward labor within a period, while the reduction in the effective capital tax rate lead houehold to hift leiure and conumption into the future. 21 Real gro national product (GNP) i ued a the meaure of national output. Invetment in the dometic economy financed by foreigner would lead to an increae in gro dometic product (GDP), but ome of thi increae mut be returned to the foreign owner of the capital. GNP, which net out the return to foreign-owned capital more accurately reflect the reource available to U.S. citizen. 10

14 capital tock. 22 Column (2) of Table 3 alo indicate that labor upply increae by 0.7 percent on average from leading to an increae in output of 0.7 percent during the ame time period. Dometically funded private invetment increae after the budget window a the government debt ratio i tabilized over time by reducing government pending, and in the long run the capital tock increae by 2.3 percent, labor upply increae by 0.2 percent, and output increae by 1.1 percent. 23 In contrat, column (3) how that when adding the remaining tax relief that increae the deficit with only a mall variation in marginal tax rate, then financing government debt more than offet for the increae in private aving and capital inflow from abroad o that the capital tock decline on average by 0.3 percent from compared to the initial teady-tate, and GNP i only 0.5 percent larger (due to the increae in labor upply of 0.5 percent). Once government pending i reduced to tabilize the government debt ratio, private invetment increae o that in the long-run, the capital tock increae by 2.3 percent and output increae by 0.7 percent. Extending the increae in the child tax credit, the 10-percent marginal tax bracket, and the reduction in marriage tax penaltie primarily increae individual after-tax income, but reult in very little change in marginal tax rate. Houehold repond to thi rie in income by increaing their conumption of good and ervice. Houehold alo conume more leiure, which lead to a long-run decline in labor upply of 0.3 percent through the income effect. 24 Thi decline in labor upply i the primary reaon why the long-run increae in GNP i maller than in the previou imulation. To a leer extent, the greater crowding out of private invetment that reult from a higher government debt burden alo contribute to thi relative decline, a indicated by the decline in dometically financed invetment from 2.6 to 2.3 percent. The overall capital tock how no change between the two imulation due to an offetting increae in foreign capital flow. Tax Relief Financed with Future Increae in Taxe Under thi financing aumption, all average and marginal tax rate on labor and capital income are changed by the ame proportion in each year after the 10-year budget window in order to maintain the baeline amount of government ervice and to maintain the government debt-to- GNP ratio at the value it take at the end of the budget window. In effect, the tax relief i modeled a temporary, a it i more than revered in the future by acro-the-board, proportional tax increae. 22 The row labeled Invetment in Table 3 and Table 4 reflect dometically funded gro invetment, while the row labeled Capital Stock in Table 3 and Table 4 repreent change in the dometic capital tock regardle of whether the new invetment i funded by dometic aving or foreigner. 23 The long-run change in labor upply for the imulation in thi ection i mall a the ubtitution effect reulting from higher after-tax wage i offet by an income effect from the houehold increae in lifetime wealth due to the decline in tax payment. 24 A indicated in Table 1, the decline in the average marginal tax rate on labor income i alo lightly maller under the full extenion of the tax cut, compared to when jut the top four ordinary rate are decreaed. Thi alo contribute to the relative decline in labor upply through the ubtitution effect. 11

15 Again, much of the reult can be explained in term of income and ubtitution effect. But in thi cae, a tax rate increae after the budget window, the ubtitution and income effect dicued above work in oppoite direction. Houehold are forward looking and many of the tranitional generation make choice that are influenced by both the tax decreae and the following tax increae. Thi implie that income effect are le important determinant of behavior during the budget window, and houehold repond by upplying more labor and aving during thi period relative to the imulation dicued above in which future government conumption decreae. The econd et of reult reported under column (1) in Table 3 how the effect of financing the lower dividend and capital gain tax rate with future increae in all income taxe. On net, in the long run thi combination of tax relief and tax increae reduce the burden of taxation on corporate invetment in favor of greater taxation of labor income and, to a leer extent, capital income in other ector. Thi implie ome increae in output reulting from a more efficient allocation of capital acro production ector. Thee gain are offet to a certain extent in the long run by the crowding out of invetment due to a higher government debt ratio and the higher tax rate needed to pay for higher interet payment on the government debt. In the long run, the capital tock increae by 0.7 percent and GNP increae by 0.3 percent (compared to increae of 1.2 percent and 0.4 percent, repectively, when government conumption decline). The reult in column (2) how a imilar pattern. Over the budget window, houehold work more and ave more compared to the ame tax relief with a government pending offet. Labor upply increae on average by 0.9 percent during , rather than 0.7 percent; the capital tock increae by 0.6 percent, rather than 0.1 percent; and GNP increae by 0.9 percent, rather than 0.7 percent. In the long run, the increae in tax rate needed to tabilize the government debt ratio lead to no change in labor upply compared to an increae of 0.2 percent when government pending decline, and the capital tock increae by 0.3 percent compared to 2.3 percent when government pending decline. GNP increae in the long-run by only 0.3 percent, compared to the 1.1 percent increae decribed above. Extending all of the tax relief and then financing with an increae in taxe after the budget window lead to hort-run effect that are again lightly larger than if a reduction in government conumption i ued to finance the revenue cot of the tax decreae. The capital tock increae by 0.6 percent during and GNP increae by 0.8 percent. However, in the long run, the combined effect of increaing marginal tax rate and crowding out lead to a decline in labor upply (0.8 percent), capital (1.8 percent) and GNP (0.9 percent). Senitivity Analyi Thi ection report the macroeconomic reult of the ame tax change and financing aumption decribed above with different value for certain parameter that repreent low and high level of reponivene. Lowering the intertemporal elaticity of ubtitution reduce the degree to which houehold are willing to ubtitute conumption and leiure acro time, which lead to a lower aving upply repone to a decreae in capital taxe. If only dividend and capital gain tax rate are lowered 12

16 and financed by future reduction in government pending, the capital tock increae by 0.9 percent in the long-run uing the lower parameter value, compared to a 1.2 percent increae uing the bae cae parameter. In the long run, if lower tax rate on dividend and capital gain are extended and financed by reduction in future government pending, GNP increae by 0.3 percent uing low parameter value, and by 0.5 percent uing the high parameter value. The choice of parameter value ha little influence on the long-run increae in GNP when uing future income tax increae to finance extenion of the lower rate on dividend and capital gain. For the low parameter value, GNP increae by 0.2 percent in the long run and for the high parameter value, GNP increae by 0.3 percent, a een in column (1) of Table 4. The mall difference i primarily the reult of the parameter having offetting effect. Lowering the intertemporal elaticity of ubtitution lower the aving repone, but lowering the intratemporal elaticity of ubtitution and the initial leiure hare of the time endowment dampen the labor upply repone and labor upply doe not decline a much when future tax rate are increaed. The choice of parameter value ha a greater influence when extenion of the lower ordinary tax rate for the top four individual bracket are added to the lower tax rate on dividend and capital gain. The long-run increae in GNP when the tax relief i financed by reducing future government pending range from 0.4 percent uing the low reponivene parameter to 1.6 percent uing the high reponivene parameter. When the tax relief i financed by increae in future tax rate, then GNP in the long-run fall by 0.1 percent under the low repone parameter and increae by 0.6 percent under the high repone parameter. Similarly, when all of the tax relief i extended and financed by the reduction in future government pending, then long-run GNP increae by only 0.1 percent in the low repone cae and by 1.2 percent in the high repone cae. However, when uing future income tax increae to pay for the tax relief, then, in both the low repone and the high repone cae, GNP fall by 0.9 percent in the long run. Again, thi i the reult of lower behavioral repone parameter working in offetting way when income taxe are raied. Lower intertemporal and intratemporal elaticity of ubtitution imply a maller labor upply repone during the budget window and a higher debt-to-gnp ratio at the end of the window which reult in larger crowding out effect. Thi lead to a large decreae in the capital tock (3.6 percent) in the long run. However, the lower labor upply repone reult in a reduction in labor upply of only 0.4 percent in the long run a future income taxe are increaed. In contrat, under the higher repone parameter value, the capital tock fall by only 1.3 percent, but labor upply fall by 1.1 percent a labor upply i more enitive to the long-run increae in marginal tax rate on labor. 6. Concluion The analyi preented in the paper ugget that permanently extending the Preident tax relief enacted in 2001 and 2003 likely would lead to a long-run increae in the capital tock and an increae in national output in both the hort run and the long run. If the revenue cot of that tax relief i offet by reducing future government pending, the increae in output i likely be about 0.7 percent under plauible aumption. If, intead, the tax relief i extended only through the 13

17 end of the budget window (i.e., it i temporary), the tax relief would increae national output in the hort run, but long-run output would decline a future tax rate increae. The analyi alo ugget that if only the portion of the Preident tax relief that primarily reduce marginal tax rate are extended (i.e., the lower rate on dividend, capital gain and the top four ordinary income bracket), it i likely that output would increae regardle of whether the revenue cot of the relief i financed through a future reduction in government pending or a future increae in tax rate, although the increae would be coniderably larger if government conumption i reduced. 14

18 Reference Aaronon, Daniel and Eric French The Effect of Progreive Taxation on Labor Supply when Hour and Wage are Jointly Determined. Federal Reerve Bank of Chicago WP Altig, David, Alan Auerbach, Laurence Kotlikoff, Kent Smetter and Jan Wallier Simulating Fundamental Tax Reform in the United State. American Economic Review 91: Auerbach, Alan The Buh Tax Cut and National Saving. National Tax Journal 55: Auerbach, Alan Tax Reform, Capital Allocation, Efficiency, and Growth, in Economic Effect of Fundamental Tax Reform edited by Henry J. Aaron and William Gale (Brooking Intitution Pre: Wahington, DC): Auerbach, Alan and Kevin A. Haet On the Marginal Source of Invetment Fund. Journal of Public Economic 87: Auerbach, Alan J. and Laurence J. Kotlikoff Dynamic Fical Policy (Cambridge Univerity Pre: Cambridge). Ball, Laurence, Dougla W. Elmendorf and N. Gregory Mankiw The Deficit Gamble. Journal of Money, Credit and Banking 30(4): Bernheim, B. Dougla Taxation and Saving. in Handbook of Public Economic, Volume 3 edited by Alan Auerbach and Martin Feldtein (Elevier Science Publiher: Amterdam). Browning, Martin, Lar Peter Hanen, and Jame J. Heckman Micro Data and General Equilibrium Model. in Handbook of Macroeconomic, Volume 1A edited by John Taylor and Michael Woodford (Elevier: New York). Congreional Budget Office Macroeconomic Analyi of a 10 Percent Cut in Income Tax Rate, Congreional Budget Office Technical Paper Serie Congreional Budget Office An Analyi of the Preident Budgetary Propoal for Fical Year Cummin, Jaon G., Kevin A. Haet and R. Glenn Hubbard A Reconideration of Invetment Behavior Uing Tax Reform a Natural Experiment. Brooking Paper on Economic Activity 2:1-74. Diamond, John and George R. Zodrow, Decription of the Tax Policy Advier General Equilibrium Model, manucript, Rice Univerity. 15

19 Elmendorf, Dougla W The Effect of Interet-Rate Change on Houehold Saving and Conumption: A Survey. Finance and Economic Dicuion Serie No Board of Governor of the Federal Reerve Sytem. Engen, Eric, Jane Gravelle and Kent Smetter Dynamic Tax Model: Why they do the thing they do. National Tax Journal 50: Fullerton, Don and Diane Lim Roger Who Bear the Lifetime Tax Burden? (Wahington, D.C.: The Brooking Intitution). Goulder, Lawrence H Tax Policy, Houing Price and Houing Invetment. Regional Science and Urban Economic 19: Goulder, Lawrence H. and Lawrence H. Summer Tax Policy, Aet Price and Growth: A General Equilibrium Approach. Journal of Public Economic 38: Goulder, Lawrence H., John Shoven and John Whalley Dometic Tax Policy and the Foreign Sector: The Importance of Alternative Foreign Sector Formulation to Reult from a General Equilibrium Tax Analyi Model. in Simulation Method in Tax Policy Analyi, edited by Martin Feldtein (Chicago: Univerity of Chicago Pre). Hayahi, Fumio Tobin Marginal q and Average q: A neoclaical Interpretation. Econometrica 50: Hurt, Erik, Annamaria Luardi, Arthur Kennickell, and Francico Torralba Precautionary Saving and the Importance of Buine Owner. National Bureau of Economic Reearch Working Paper Judd, Kenneth L Capital Income Taxation with Imperfect Competition. American Economic Review 92: Keuchnigg, Chritian Corporate Taxation and Growth: A Dynamic General Equilibrium Simulation Study, in Johann K. Brunner and Han-Georg Peteren, ed. Simulation Model in Tax and Tranfer Policy (Campu Verlag) Lee, Chul-In Finite Sample Bia in IV Etimation of Intertemporal Supply Model: I the Intertemporal Subtitution Elaticity Really Small? Review of Economic and Statitic 83(4): Muthitacharoen, Athiphat and George R. Zodrow, forthcoming. State and Local Taxation of Buine Property: A Small Open Economy Perpective, Proceeding of the Ninety- Eighth Annual Conference on Taxation, National Tax Aociation. Roger, Diane Lim Aeing the Effect of Fundamental Tax Reform With the Fullerton-Roger General Equilibrium Model, in Joint Committee on Taxation Tax Modeling Project and 1997 Tax Sympoium Paper (JCS-21-97), pp

20 Ziliak, Jame P. and Thoma J. Kniener Etimating Life-Cycle Labor-Supply Tax Effect. Journal of Political Economy 107(2): Ziliak, Jame P. and Thoma J. Kniener The Effect of Income Taxation on Conumption and Labor Supply. Journal of Labor Economic 23(4): Zodrow, George R On the Traditional and New View of Dividend Taxation. National Tax Journal 44(4):