The Evolution of Taxes and Hours Worked in Austria,

Transcription

1 The Evoluion of Taxes and Hours Worked in Ausria, John T. Dalon Wake Fores Universiy Augus 2013 Absrac Aggregae hours worked per working-age person decreased in Ausria by 25% from 1970 o During he same ime period, axes increased, paricularly he effecive marginal ax rae on labor income. Using a sandard general equilibrium growh model wih axes, I quaniaively assess he role played by he evoluion of axes on he evoluion of hours worked in Ausria. The model accouns for 76% of he observed decrease in hours worked per working-age person. My resuls are in line wih oher sudies, such as Presco (2002), which find axes play an imporan role in explaining aggregae hours worked. JEL Classificaion: E24, J22, E13 Keywords: axes, labor supply, growh accouning, dynamic general equilibrium I hank Tim Kehoe for his advice and consan encouragemen. I also hank wo referees for commens leading o subsanial improvemens in he paper. Conac: Deparmen of Economics, Carswell Hall, Wake Fores Universiy, Box 7505, Winson-Salem, NC dalonj@wfu.edu

2 1 Inroducion In examining he causes of he differences in aggregae hours worked boh across counries and wihin counries over ime, macroeconomiss find axes play an imporan role. Presco (2002) and Presco (2004) argue ax raes accoun for much of he difference observed in hours worked beween he Unied Saes and Europe. Ohanian, Raffo, and Rogerson(2008) expands Presco s work o a larger se of counries over a longer ime span and finds much of he variaion in hours worked over ime and across counries can be explained by axes. Conesa and Kehoe (2008) ake a more deailed look a he cases of Spain and France and also show axes play an imporan role in explaining he fall in hours worked. I build on he exising lieraure by analyzing he specific case of aggregae hours worked in Ausria over he years Ausria is represenaive of he experience of many European counries. In 1970 s Ausria, hours worked per working-age person were higher han in he Unied Saes. By he year 2005, hours worked per working-age person in Ausria had decreased by 25% and sood a a level lower han ha in he Unied Saes. I sudy he quesion, How well can he evoluion of axes accoun for he evoluion of aggregae hours worked in Ausria? My work differs from he previously menioned lieraure in he following ways: Presco (2002) only examines he effecs of axes on he differenial in hours worked beween France and he Unied Saes in a paricular period of ime. Presco (2004) expands he analysis of he role played by axes by comparing hours worked across a small group of counries, no including Ausria, beween wo ime periods. Ohanian, Raffo, and Rogerson (2008) considers a larger sample of counries, including Ausria, across a longer ime span bu focuses on he effecs on hours worked of a single ax wedge consruced from consumpion ax raes and average labor income ax raes. I will focus on he changes in hree differen ax raes separaely: consumpion ax raes and marginal ax raes on labor and capial income. In addiion, Ohanian, Raffo, and Rogerson (2008) lumps counries ogeher and presens resuls averaged across groups, whereas I presen more deails relaed specifically o he Ausrian case. Alhough my main focus is also hours worked, I show he effecs of axes on Ausrian real GDP per working-age person and he capial-oupu raio, which are no considered by Ohanian, Raffo, and Rogerson (2008). Lasly, I include sensiiviy analysis no included in Ohanian, Raffo, and Rogerson (2008). 1

3 On he oher hand, his paper closely relaes o he work in Conesa and Kehoe (2008), essenially applying he mehodology employed in ha paper o he case of Ausria over he period The mehodology used is he one developed by Kehoe and Presco (2002) o sudy grea depressions and is based on growh accouning and he dynamic general equilibrium growh model. Kehoe and Presco (2007) conains a collecion of papers employing a similar framework o sudy sixeen depressions hroughou hisory and he world, including he cases of France, he Unied Saes, Japan, and Mexico. Cicek and Elgin (2011) represens a more recen applicaion of his mehodology for he case of Turkey. There are hree seps o he mehodology. Firs, growh accouning quanifies he conribuions of oal facor produciviy (TFP), capial, and aggregae hours worked for he growh of oupu. Second, he neoclassical growh model serves as a heoreical framework for undersanding he dynamics of he economy. The cenral feaure of he model is a represenaive household which akes he evoluion of axes and TFP as given and chooses sequences of consumpion, hours worked, and capial o maximize uiliy. Third, he growh model is calibraed and used o conduc numerical experimens. The numerical experimens generae model daa which can hen be compared o he acual daa observed in he economy. As Kehoe and Presco (2002) poin ou, he mehodology funcions as a diagnosic ool, relying on macro daa and a macro model o deermine he facors of he economy requiring more deailed sudy. The growh accouning for Ausria reveals a large divergence beween TFP and oupu per working-age person. The divergence resuls from he seady decline in hours worked in Ausria from 1970 o Ausria conrass wih he experience of he Unied Saes. In he Unied Saes, hours worked per working-age person remain fairly consan and have even increased since he early 1980 s. The growh accouning for Ausria is, however, in line wih oher European counries experiencing large declines in hours worked, such as Spain, France, and Finland. 1 I find he neoclassical growh model augmened wih axes does a good job of replicaing he daa from my growh accouning exercise. In order o perform his experimen, I exogenously se he consumpion ax rae and he effecive marginal ax raes on labor and capial income o he raes found in he daa. The model wih hese acual ax raes accouns for 76%of he fall in 1 See Conesa and Kehoe (2008) for a similar growh accouning decomposiion of Spain and France. See Conesa, Kehoe, and Ruhl (2007) for he case of Finland. 2

4 hours worked observed in Ausria over he period I show he necessiy of augmening he model wih he sequences of acual ax raes by conducing an addiional experimen which fails o replicae he experience of he Ausrian economy. I es he performance of a model wih consan ax raes agains he daa. This experimen fails o mach he daa as well as he model wih he sequences of he acual ax raes found in he daa. My resuls suppor he evidence found in he lieraure menioned earlier. I do no wish o claim oher labor marke fricions or insiuions play no role in explaining he evoluion of hours worked in Ausria. However, as Conesa and Kehoe (2008) poin ou, o he exen ha he developmen of such insiuions coincides wih he increase in axes, hese explanaions would be correlaed wih he evoluion of axes in Ausria. My analysis also says nohing abou he disribuion of hours worked wihin he working-age populaion. For example, labor force paricipaion among he elderly remains low in Ausria. The pension sysem in Ausria is one of he more generous and complee in Europe, widely recognized as unsusainable, and currenly in a sae of on-going reform. 2 The remainder of he paper is organized as follows: Secion 2 conains he growh accouning exercise. In Secion 3, I describe he neoclassical growh model wih axes. Secion 4 presens he calibraion, resuls of he numerical experimens, and sensiiviy analysis. Secion 5 concludes. 2 Growh Accouning The growh accouning for Ausria is based on he sandard heoreical framework of he neoclassical growh model, as in Kehoe and Presco (2002), and is inended o deec deviaions from balanced growh behavior. The model conains an aggregae producion funcion aking he Cobb-Douglas form, Y = A K α L1 α, (1) where Y is oupu, A is TFP, K is capial inpu, L is labor inpu, and 1 α is labor s share of income. If boh he growh in TFP and he growh in working-age populaion,, are assumed o be consan, 2 See Hofer and Koman (2006) for an overview of he Ausrian pension sysem. 3

5 A +1 = g 1 α A, (2) +1 = η, (3) hen here is a balanced growh pah where oupu per working-age person, Y, grows a he rae g 1; he capial-oupu raio, K Y, is consan; and hours worked per working-age person, L, are consan. Kehoe and Presco (2002) hen rewrie he aggregae producion funcion (1) as he following: Y 1 α = A 1 ( ) α K Y 1 α ( L ), (4) which decomposes oupu per working-age person, Y, ino a produciviy facor, A 1 1 α ; a capial ( ) α K 1 α facor, L Y ; and a labor facor,. On a balanced growh pah, growh in oupu per working-age person arises from changes in he produciviy facor, as boh he capial and labor facors remain consan. In order o show he usefulness of his decomposiion, consider he case of he Unied Saes. Figure (1) repors he daa for he Unied Saes over he period Growh in he Unied Saes appears close o balanced, paricularly over he period ( ) During hese years, growh in Y is close o growh in A 1 α 1 α K 1 α, while Y and L remain fairly consan. However, afer 1983, he Unied Saes growh pah becomes less Y balanced. grows faser han A 1 1 α, which is driven by he gradual increase in L. In order o perform he growh accouning decomposiion for Ausria, daa needs o be colleced for he series of oupu, capial sock, working-age populaion, and hours worked. A value for labor s share of income also needs o be assigned. The series of TFP can hen be calculaed using hese series and he labor share of income. The appendix conains addiional informaion on he daa used hroughou his paper and heir sources. The naional accouns for Ausria do no repor a series for he capial sock, so I consruc he series using he perpeual invenory mehod, K +1 = (1 δ)k +X, (5) 4

6 Figure 1: Unied Saes Growh Accouning Y N 1960= A L N K 1 Y where δ denoes a consan depreciaion rae of capial and X is invesmen. The capial sock series can hen be accumulaed from daa on invesmen and values for δ and an iniial capial sock. The value of δ is chosen o mach he average raio of depreciaion o gross domesic produc (GDP) in he daa over he calibraion period In Ausria, he average raio of depreciaion o GDP over he years is =1970 δk Y = (6) The value of he iniial capial sock is chosen so ha he capial-oupu raio in he iniial period, 1960, maches he average capial-oupu raio over a reference period, : K 1960 = K. (7) Y Y 1961 The equaions (5), (6), and (7) make up a sysem ha can be solved o find he capial sock 5

7 Figure 2: Ausria Growh Accouning A 1960= K 1 Y Y N L N series and he value of δ. The calibraed value for δ in Ausria is The labor income share can be measured direcly from he Ausrian daa over he years My calculaions for he labor income share yield an average value of , which ranslaes ino a capial income share, α, of The Ausrian value of he capial income share is in line wih he resuls in Gollin (2002), which sugges a common value of α = 0.3 across counries. Only he TFP series remains o be calculaed in order o repor he growh accouning for Ausria. This is done by simply rearranging he aggregae producion funcion (1) and using he measures of oupu, capial socks, hours worked, and he labor income share o solve for he following: A = Y K α L 1 α. (8) Figure(2) displays he growh accouning decomposiion(4) for Ausria over he period

8 2005. Three observaions are of noe. Firs, he overall effecs of he Ausrian Wirschafswunder are clearly presen. The Wirschafswunder, or economic miracle, refers o he period of economic recovery and expansion in Germany and Ausria afer World War II. Ausrian oupu per working-ageperson, Y, growsaanaverageannualraeof2.7%overheenireperiod During he years , a period coinciding more closely wih he acual Wirschafswunder, growh in Y is even faser, averaging an annual rae of 3.9%. Second, in conras o he Unied Saes experience, he Ausrian growh pah displays large deviaions from balanced growh afer 1965, as evidenced by he divergence of oupu per working-age person, Y, and he produciviy facor, A 1 1 α. Third, his deviaion from balanced growh occurs due o he seady fall in aggregae hours worked per working-age person, L. Hours worked in Ausria fall by 35% from 1960 o From , he period of focus in his paper, hours worked in Ausria fall by 25%. The decrease in hours worked per working-age person in Ausria is in line wih he experiences of oher European counries. This paper s main purpose is o undersand he role played by axes in explaining he decrease in hours worked in Ausria by esing a model wih axes agains he daa presened in his growh accouning exercise. I now urn o describing such a model. 3 Model The economic environmen is ha of he simple dynamic general equilibrium model augmened wih axes. A represenaive household akes he evoluion of axes and TFP as given and chooses sequences of consumpion, hours worked, and capial o maximize uiliy. A represenaive firm produces oupu wih an aggregae echnology, aking prices as given. Governmen collecs proporional axes on consumpion, labor income, and capial income and rebaes he proceeds o he household in a lump-sum fashion, making sure o balance is budge. Specifically, he represenaive household chooses sequences of aggregae consumpion, C ; aggregaecapialsocks, K ; andaggregaehoursworked, L,osolvehefollowingmaximizaion problem: max β [ γlogc +(1 γ)log( h L ) ] (9) =T o 7

9 s.. (1+τ c )C +K +1 = (1 τ l )w L +[1+(1 τ k )(r δ)]k +T, (10) C,K,L 0, (11) L h, (12) K To given, (13) where β, 0 < β < 1, is he discoun facor; γ, 0 < γ < 1, is he consumpion share; and h is an individual s ime endowmen of hours available for marke work. Equaion (10) represens he household s budge consrain. τ c, τl, and τk are he ax raes on consumpion, labor income, and capial income. w and r are he wage rae and renal rae. δ, 0 < δ < 1, is he depreciaion rae. T is he lump-sum ransfer from he governmen. The inequaliies represened in (11) are he nonnegaiviy consrains on consumpion, capial socks, and hours worked. Inequaliy (12) consrains he household s choice of aggregae hours worked, since he oal number of hours available for work is h. Finally, (13) is he consrain on he iniial sock of capial. The represenaive firm produces oupu according o he producion echnology (1). A compeiive environmen, in which he firm earns zero profis and minimizes coss, gives rise o he pricing rules for he wage rae and renal rae: w = (1 α)a K α L α, (14) r = αa K α 1 L 1 α. (15) The feasibiliy consrain in he economy requires curren oupu be divided beween consumpion and invesmen: C +K +1 (1 δ)k = A K α L1 α. (16) The governmen s budge consrain ensures he oal ax receips exacly equal he lumpsum ransfers o he household: τ c C +τ l w L +τ k (r δ)k = T. (17) 8

10 Rebaing all he ax receips in a lump-sum fashion o he household is equivalen o viewing governmen expendiure as a subsiue for privae consumpion. For insance, he ax revenue migh be used o finance healh care, unemploymen insurance, or public schools. I reurn o his assumpion in he sensiiviy analysis in Secion 4.3 by considering an alernaive specificaion of waseful governmen consumpion. Now, an equilibrium for his environmen can be defined as follows: Given sequences of TFP, A ; working-age populaion, ; consumpion ax raes, τ c; labor income ax raes, τ; l and capial income ax raes, τ k, for = T o,t o + 1,... and an iniial capial sock, K To, an equilibrium wih axes is sequences of aggregae consumpion, C ; aggregae capial socks, K ; aggregae hours worked, L ; wages, w ; ineres raes, r ; and ransfers, T, such ha he following condiions hold: 1. Given wages, w, and ineres raes, r, he represenaive household chooses consumpion, C ; capial, K ; and hours worked, L, o maximize uiliy (9) subjec o he budge consrain (10), he nonnegaiviy consrains (11), he upper bound on he oal number of hours worked (12), and he consrain on he iniial capial sock (13). 2. The wages, w, and ineres raes, r, and he represenaive firm s choices of labor, L, and capial, K, saisfy he cos minimizaion and zero profi condiions (14) and (15). 3. Consumpion, C ; labor, L ; and capial, K, saisfy he feasibiliy consrain (16). 4. Governmen ransfers, T, saisfy he governmen s budge consrain (17). These equilibrium requiremens reduce o a sysem of equaions which characerizes he equilibrium. Taking he firs-order condiions of he household s maximizaion problem, I solve for he household s ineremporal and inraemporal condiions: C +1 = β 1+τc [1+(1 τ C 1+τ +1)(r k +1 c +1 δ)], (18) 9

11 (1 τ l )w ( h L ) = 1 γ γ (1+τc )C. (19) Plugging he firm s opimaliy condiions(14) and(15) ino he household s opimaliy condiions (18) and (19), yields C +1 = β 1+τc [1+(1 τ k C 1+τ+1 c +1 )(αa +1K+1 α 1 L1 α +1 δ)], (20) (1 τ l )(1 α)a K α L α ( h L ) = 1 γ γ (1+τc )C, (21) which, combined wih he feasibiliy consrain (16) and governmen budge consrain (17), is he sysem of equaions characerizing he equilibrium of he model. I use his sysem when compuing he equilibrium of he model in my numerical experimens. 3 4 Numerical Experimens The numerical experimens I perform compare he daa o wo heoreical economies wih differen ax scenarios. The firs is a model wih consan axes in which τ c, τl, and τk are se o he raes observed in he daa in The second is a model wih axes in which he evoluion of τ c, τl, and τk follows he acual evoluion of he raes as measured in he daa. The heoreical economies will deermine he equilibrium evoluion of he endogenous variables given a se of calibraed parameers and he evoluion of he exogenous variables. The exogenous variables are he sequences of TFP, working-age populaion, and he axes raes. The numerical experimens will hen allowme o compare heevoluion ofhe aggregaevariables implied by he model wih hose acually observed in he daa. The aggregaes I compare are real GDP per working-age person, he capial-oupu raio, and, of course, hours worked per working-age person. 3 See Conesa, Kehoe, and Ruhl (2007) for a deailed discussion on solving models of his ype. Accompanying documenaion can be accessed online a 10

12 4.1 Calibraion Following he mehodology in Mendoza, Razin, and Tesar (1994), I use daa on aggregae ax collecions o calculae he sequences of effecive ax raes τ c, τl, and τk. However, I follow oher recen macroeconomic sudies in deviaing from he procedure in Mendoza, Razin, and Tesar (1994) in wo imporan respecs. 4 Firs, I aribue a fracion of household s non-wage income o labor income. Second, I measure effecive marginal ax raes insead of effecive average ax raes. The heoreical framework developed in Secion 3 moivaes he choice of focusing on effecive marginal ax raes. The represenaive household s decisions ake place a he margin, as shown in equaions (20) and (21). Given he progressiviy of income axes, he esimaes of he income axes need o be adjused. In principle, I should use micro daa, such as a represenaive sample of ax records, o esimae effecive income ax funcions o perform my adjusmens. Conesa and Kehoe (2008) does exacly his for he case of Spain. Conesa and Kehoe (2008) muliplies average income axes by a facor of 1.83 o obain marginal ax raes for Spain. In he case of Ausria, I simply follow Presco (2002) and Presco (2004) for he Unied Saes case and muliply average income axes by a facor of 1.6 o obain marginal ax raes. Conesa, Kehoe, and Ruhl (2007) adop he same procedure for he case of Finland. Figure (3) graphs τ c, τl, and τ k for Ausria over he period A key observaion in Figure (3) is ha he effecive marginal ax rae on labor income rends upward over he 35 year period, from a low of 0.36 in 1970 o a high of 0.54 in Deailed informaion on he consrucion of he ax raes appears in he appendix. In he experimens wih axes, I adjus he series of TFP by modifying equaion (8) o A = C +X, (22) K 1 α L α where C +X is real GDP a facor prices in he daa. However, when I repor he conribuion of TFP o growh in he resuls secion, I repor he convenional measure of TFP, 4 See, for example, Conesa, Kehoe, and Ruhl (2007), Conesa and Kehoe (2008), Presco (2002), and Presco (2004). 11

13 Figure 3: Effecive Marginal Tax Raes in Ausria Labor Income 0.4 Capial Income Rae 0.3 Consumpion where  = Ŷ K 1 α L α, (23) Ŷ = (1+τ c T)C +X (24) is real GDP a marke prices of he base year T = The exogenous sequence of he working-age populaion is ha measured from he daa in he growh accouning exercise. I assign a value of h = 100 for an individual s ime endowmen of hours available for marke work per week. The remaining parameers are he iniial capial sock, K To ; capial share, α; depreciaion rae, δ; discoun facor, β; and consumpion share, γ. The iniial capial sock is he 1970 value from he series of capial socks calculaed in he growh accouning exercise. The capial share 12

14 and depreciaion rae are also he same as in he growh accouning exercise, which means α = and δ = Rearranging equaions (18) and (19) allows me o calibrae β and γ as follows: β = (1+τc +1)C +1 (1+τ c )C 1 1+(1 τ k +1 )(r +1 δ), (25) γ = (1+τ c )C (1+τ c )C +(1 τ l )w ( h L ). (26) I calculae a vecor of β s and γ s for he same period used o calculae δ and α, , and hen ake he average of hese vecors o assign values o β and γ. The calibraed values of β and γ vary depending on he ax scenario of he numerical experimen. In he experimen wih consan axes, β = and γ = In he experimen wih he acual ax raes, β = and γ = The calibraed values for β in he experimens wih axes are boh greaer han 1, which means he uiliy funcion (9) is poenially infinie. I avoid his problem by seing β = in he wo ax experimens Resuls Figures (4) - (6) and Table (1) compare daa from he Ausrian economy wih he corresponding resuls from he numerical experimens. Figure(4) compares he growh of he Ausrian economy from 1970 o 2005 wih he growh of he wo heoreical economies over he same period. The model wih consan axes predics a noiceably larger increase in real GDP per working-age person han acually occurred in Ausria. The model wih axes, however, predics a pah for real GDP per working-age person which is much more in line wih he acual experience of he Ausrian economy. This resul suggess models based on he evoluion of TFP alone are inadequae for undersanding recen growh in Ausria. Indeed, he graph highlighs he imporance of recognizing he role played by he evoluion of axes in he Ausrian economy since Figure (5) compares he daa on he evoluion of hours worked per working-age person in Ausria wih he resuls implied by he wo heoreical economies. The model wih consan 5 See Conesa, Kehoe, and Ruhl (2007) for furher discussion on his issue. 13

15 Figure 4: Real GDP per Working-Age Person in Ausria Model wih Consan Taxes 1970= Model wih Taxes Daa axes fails o accoun for he fall in hours worked seen in he daa. The series of hours worked generaed by he model wih consan axes remains fairly consan. The key poin here is ha he mere presence of disorions is no enough o generae he evoluion of hours worked seen in he daa. The acual evoluion of axes is imporan for generaing he fall in hours worked, as seen by he series implied by he model wih axes. Figure (5) shows he model wih axes does a good job accouning for he magniude of he decrease in hours worked in Ausria from 1970 o In he daa, hours worked per working-age person in Ausria fall by 25% from 1970 o 2005, whereas in he model wih axes hey fall by 19%. Hours worked in he model wih axes also seems o qualiaively mach he evoluion of hours worked in he daa, hough he hours worked in he model wih axes flucuae more han hose in he daa. The qualiaive similariies are eviden if he series in Figure(5) are divided ino four period: coincides wih a seady fall in hours worked, hours worked remain consan or increase during he years , he years see anoher fall in hours worked, and is anoher period 14

16 Figure 5: Hours Worked per Working-Age Person in Ausria Daa Model wih Taxes Hours per Week Model wih Consan Taxes of roughly consan hours worked. These four periods are also idenifiable in he series of labor income ax raes presened in Figure (3). Boh models predic similar resuls wih respec o capial deepening. Figure (6) graphs he evoluion of he capial-oupu raio in Ausria and he capial-oupu raios implied by he wo numerical experimens. The wo models generae smaller capial-oupu raios han hose found in he daa. Finally, Table (1) presens he quaniaive implicaions of he numerical experimens by comparing he growh accouning in he daa wih he growh accouning in each of he wo heoreical economies. For ease of exposiion, I ake he naural logarihm of equaion (4): log Y = 1 1 α loga + α 1 α logk Y +log L. (27) Oupu per working-age person now decomposes ino hree addiive facors. The numbers in Table (1) can be viewed as growh raes, as hey are average annual changes muliplied by

17 Figure 6: Capial-Oupu Raio in Ausria Daa 3.7 Model wih Taxes Raio Model wih Consan Taxes Table 1: Decomposiion of Average Annual Changes in Real GDP per Working-Age Person in Ausria (Percen) Daa Consan Taxes Acual Taxes Change in Y/N Due o TFP Due o K/Y Due o L/N Sensiiviy Analysis Table 2 presens simulaion resuls for differen specificaions of he model. The firs wo columns repor he new calibraed values of β and γ under each specificaion. Since he calibraed values for β are ofen greaer han 1, I se β = in mos of he specificaions. The remaining 16

18 Table 2: Sensiiviy Analysis Y/N K/Y L/N per week β γ Daa Consan Taxes Acual Taxes Only τ c Only τ l Only τ k δ = γ = G No Adjusmen Facor McDaniel (2007) Taxes columns coincide wih he aggregae variables from Figures 4-6 bu only repor he values of each variable in 1970 and 2005, no he enire ime series. The firs se of rows reproduces he values from he daa, model wih consan axes, and model wih acual axes in Figures 4-6 as reference. The second se of rows shows he conribuion of each individual ax change, keeping he oher axes consan a he rae in he daa in 1970, e.g. he row labeled Only τ c considers he acual evoluion of he ax rae on consumpion and ses he ax raes on labor income and capial income equal o heir values in These resuls sugges he evoluion of τ l drives he decline in hours worked. The overall fi of he model is also bes in he case when τ l evolves. The hird se of specificaions considers he model wih acual axes subjeced o differen parameers and an alernaive way of modeling governmen. In he row δ = 0.05, I impose a higher depreciaion rae. The only significan difference caused by his change is a decrease in he capial-oupu raio. The row labeled γ = 0.8 considers a differen labor supply elasiciy by seing he consumpion share γ = 0.8. Hours worked increase subsanially in his case, as he household values consumpion more han leisure. The role γ plays in he labor supply elasiciy 17

19 can be seen by he fac ha hours worked decrease by a smaller percenage over he period han in he benchmark model wih acual axes, a 14% decline versus 19%. The row labeled G considers he case when all governmen consumpion is wased or eners he represenaive household s uiliy funcion as a public good. This is he opposie exreme of lump-sum ransfers in he benchmark model. Modeling governmen in his way gives rise o new versions of he feasibiliy consrain (16) and governmen budge consrain (17): and C +K +1 (1 δ)k +G = A K α L 1 α (28) τ c C +τ l w L +τ k (r δ)k = G. (29) Increases in axes cause negaive income effecs when ax revenues are no ransferred back o households, so households respond by increasing he number of hours worked. The resuls in row G show his movemen and are consisen wih he resuls presened in Conesa, Kehoe, and Ruhl (2007) for he case of Finland under he assumpion of waseful governmen consumpion. The las wo rows in Table 2 presen simulaions of he model using differen sequences of axes. The row labeled No Adjusmen Facor considers he same underlying sequences of axes as he case wih acual axes bu does no conver average income axes o marginal ax raes by he adjusmen facor of 1.6. The row labeled McDaniel (2007) Taxes uses he sequences of axes consruced by McDaniel (2007) and used for he analysis in Ohanian, Raffo, and Rogerson (2008). 6 Again, he income ax raes are average ax raes. These las wo simulaions fi he daa worse han he model wih acual axes. Real GDP per working-age person increases more han in he model wih acual axes, overshooing he daa even more. The decline in hours worked is also no as seep. These resuls show he choice of he adjusmen facor maers. 6 See McDaniel (2007) for complee deails, including a comparison wih he procedure used in Mendoza, Razin, and Tesar (1994) for consrucing average ax raes. Boh he procedure I use and he one oulined in McDaniel (2007) improve upon he procedure in Mendoza, Razin, and Tesar (1994) by aribuing a fracion of household s non-wage income o labor income. 18

20 5 Conclusion The workhorse of modern macroeconomics is he general equilibrium growh model. I has been used o sudy business cycles, moneary policy, grea depressions, and a hos of oher economic issues. I apply he model o he sudy of he Ausrian economy. Calibraed o he Ausrian experience, a simple dynamic general equilibrium growh model wih axes can accoun for 76% of he decrease in hours worked per working-age person observed in Ausria over he years My resuls suppor he conclusions of recen sudies sressing he imporance of axes in explaining he evoluion of hours worked. The analysis presened here is silen abou he disribuion of hours worked wihin he working-age populaion. Presco, Rogerson, and Wallenius (2009) find employmen differences beween four European counries (Belgium, France, Germany, and Ialy) and he Unied Saes are concenraed among young and old workers. In he Ausrian case, however, employmen differences relaive o he Unied Saes are concenraed solely among older workers. This feaure of he Ausrian daa suggess an avenue for fuure research. 19

21 Appendix The growh accouning exercises require he use of naional accouns daa and daa on he labor force. The Unied Saes naional accouns daa are aken from he Bureau of Economic Analysis s Naional Income and Produc Accouns, available a and he Inernaional Moneary Fund s (IMF) Inernaional Financial Saisics CD-ROM. The Unied Saes daa on he labor force can be found in he Organisaion for Economic Co-operaion and Developmen s (OECD) Populaion and Labour Force Saisics online a Naional accouns daa for Ausria are from hree sources: he IMF s Inernaional Financial Saisics, he Deailed Tables of Main Aggregaes in he OECD s Annual Naional Accouns, and Angus Maddison s Hisorical Saisics for he World Economy: AD online a Ausrian daa on he labor force are from he OECD s Populaion and Labour Force Saisics and he Conference Board and Groningen Growh and Developmen Cenre s Toal Economy Daabase. The Toal Economy Daabase resides a The Ausrian daa on he composiion of governmen expendiure appearing in Figure (??) are from he OECD s General Governmen Accouns. The General Governmen Accouns are also available online a Toal governmen expendiure is broken down ino differen caegories based on he Unied Naion s Classificaion of he Funcions of Governmen. Given he focus of his paper, a more deailed discussion regarding he consrucion of he Ausrian ax raes seems appropriae. As menioned in he ex, I follow he procedure of previous sudies, such as Conesa, Kehoe, and Ruhl (2007) and Conesa and Kehoe (2008). Calculaing he effecive marginal ax raes requires daa on boh ax revenue and he ax base. For he daa on ax revenue, I use he OECD s Deails of Tax Revenues of Member Counries. The ax base daa are from he Deailed Tables of Main Aggregaes in he OECD s Annual Naional Accouns. Boh daa sources are available online a The following key defines he variables used o calculae he ax raes: Tax Revenue Saisics 1100 = Taxes on income, profis, and capial gains of individuals, 1200 = Taxes on income, profis, and capial gains of corporaions, 20

22 2000 = Toal social securiy conribuions, 2200 = Employer s conribuion o social securiy, 3000 = Taxes on payroll and workforce, 4100 = Recurren axes on immovable propery, 4400 = Taxes on financial and capial ransacions, 5110 = General axes on goods and services, 5121 = Excise axes, Naional Accouns C HH C NPISH = Household final consumpion expendiure, = Final consumpion expendiure of nonprofi insiuions serving households, CE = Compensaion of employees, OSMI = Household gross operaing surplus and mixed income, δk HH Y = GDP, = Household consumpion of fixed capial, T = Taxes less subsidies, δk = Consumpion of fixed capial. The consumpion ax raes are compued as τ c = C HH +C NPISH (30) In order o consruc he ax raes on labor and capial income, I firs calculae he marginal ax rae on household income. Second, I calculae he ax raes on labor and capial income by assigning he ambiguous income caegories in he daa o eiher labor or capial income. I se capial s share of income o be α, which is he same as ha in he aggregae producion funcion (1). The marginal ax rae on household income is τ h 1100 = adj CE 2200+(OSMI δk HH ), (31) where adj is he adjusmen facor aking he progressiviy of he income axes ino accoun and convers he average ax rae o a marginal ax rae. In he case of Ausria, I se adj = 1.6. The labor and capial axes are hen compued as follows: 21

23 τ l = τh [CE 2200+(1 α)(osmi δk HH )] , (32) (1 α)(y T ) τ k = τh α(osmi δk HH ) α(y T ) δk. (33) 22

24 References Cicek, D., and C. Elgin (2011): No-Quie-Grea Depressions of Turkey: A Quaniaive Analysis of Economic Growh over , Economic Modelling, 28(6), Conesa, J. C., and T. J. Kehoe (2008): Produciviy, Taxes and Hours Worked in Spain: , Working Paper. Conesa, J. C., T. J. Kehoe, and K. J. Ruhl (2007): Modeling Grea Depressions: The Depression in Finland in he 1990s, in T. J. Kehoe and E. C. Presco, ediors, Grea Depressions of he Twenieh Cenury, Minneapolis, MN: Federal Reserve Bank of Minneapolis, Gollin, D. (2002): Geing Income Shares Righ, Journal of Poliical Economy, 110(2), Hofer, H., and R. Koman (2006): Social Securiy and Reiremen Incenives in Ausria, Empirica, 33(5), Kehoe, T. J., and E. C. Presco (2002): Grea Depressions of he 20h Cenury, Review of Economic Dynamics, 5(1), (eds.) (2007): Grea Depressions of he Twenieh Cenury. Minneapolis, MN: Federal Reserve Bank of Minneapolis. McDaniel, C. (2007): Average Tax Raes on Consumpion, Invesmen, Labor and Capial in he OECD , working paper. Mendoza, E. G., A. Razin, and L. L. Tesar (1994): Effecive Tax Raes in Macroeconomics: Cross-Counry Esimaes of Tax Raes on Facor Incomes and Consumpion, Journal of Moneary Economics, 34(3), Ohanian, L., A. Raffo, and R. Rogerson (2008): Long-Term Changes in Labor Supply and Taxes: Evidence from OECD Counries, , Journal of Moneary Economics, 55(8),

25 Presco, E. C. (2002): Richard T. Ely Lecure: Prosperiy and Depression, American Economic Review Papers and Proceedings, 92(2), (2004): Why Do Americans Work So Much More Than Europeans?, Federal Reserve Bank of Minneapolis Quarerly Review, 28(1), Presco, E. C., R. Rogerson, and J. Wallenius (2009): Lifeime Aggregae Labor Supply wih Endogenous Workweek Lengh, Review of Economic Dynamics, 12(1),