The Transitory VAT Cut in the UK: A Dynamic General Equilibrium Analysis. Gonzalo Fernández-de-Córdoba and José L Torres 1

Transcription

1 Economic Issues, Vol. 16, Par 1, 2011 The Transiory VAT Cu in he UK: A Dynamic General Equilibrium Analysis Gonzalo Fernández-de-Córdoba and José L Torres 1 ABSTRACT This paper sudies he macroeconomic effecs of he ransiory Value Added Tax (VAT) cu in he UK using a dynamic general equilibrium approach. The emporary VAT cu policy was announced on 24 November 2008, and was due o come ino effec on 1 December 2008, wih VAT revering o is previous level as from 1 January We quanify he effecs of ha emporary fiscal simulus policy on he key aggregae variables of he UK economy and on ax revenues. Overall, we obain ha his policy is oo emporary o have imporan quaniaive effecs on he economy bu qualiaive effecs are of grea ineres. We show ha he emporary VAT cu will generae an overshooing effec on key macroeconomic variables, and will provoke a significan reducion in invesmen. Consumpion and oupu will increase during he VAT cu, bu hey will decrease below heir seady sae values afer VAT revers o he previous level. Our model economy also predics ha fiscal revenues will decrease abou hree per cen during he VAT cu. Finally, we find ha he VAT cu policy would have provided beer resuls if i had been announced earlier. There is no such hing as a good ax Winson Churchill 1. INTRODUCTION THIS PAPER USES A DYNAMIC GENERAL EQUILIBRIUM MODEL along he lines of Baxer and King (1993) o sudy he effecs of he emporary VAT-cu simulus policy in he UK. In paricular, we seek o quanify he consequences of ha policy for key macroeconomics variables and for ax revenues, aking ino accoun all general equilibrium effecs derived from he ax change. In his conex, dynamic general equilibrium models can be a very useful ool for evaluaing he macroeconomic consequences of economic policies such as he one we are ineresed in here

2 G Fernández-de-Córdoba and J L Torres The VAT cu policy in he UK was announced on 24 November 2008 and came ino effec on 1 December, jus six days afer he announcemen. Thus, we firs consider i as an unanicipaed ax shock. The VAT rae was reduced from 17.5 per cen o 15 per cen. Second, he reducion was announced for a period of 13 monhs implying ha VAT was o rever o 17.5 per cen on 1 January Thus, we simulae his ax policy as an unanicipaed ransiory ax shock, wih a known end. Wih hese wo ingrediens in mind we conduc an exercise ha simulaes a ransiory, and unanicipaed VAT reducion using a general equilibrium model calibraed for he UK economy. The recen emporary VAT-cu simulus policy in he UK has been sudied by a number of auhors, such as Crossley e al. (2009), Barrell and Weale (2009) and Blundell (2009). However, hey all use a parial equilibrium approach. Given a change in axes, we can disinguish beween wo effecs: he income effec and he (ineremporal) subsiuion effec. However Crossley e al. argue ha in he case of a emporary change he former is likely o be small because he change in lifeime income would be minimal. 3 The second effec is likely o be of much more imporance because he reducion in he price of goods (given a pass-hrough effec) provides an incenive o increase consumer spending. These auhors poin ou ha he relevan behavioural parameer is he elasiciy of ineremporal subsiuion and ha a value of one seems o be an appropriae elasiciy value for he UK. Crossley e al. (2009) esimae ha, assuming ha prices will fall by exacly he amoun of he VAT cu (full pass-hrough), he proporional change in he relaive prices of curren consumpion will be -1.2 per cen so ha he level of consumpion increases by 1.2 per cen. Blundell (2009) suggess ha around 75 per cen (less han full pass-hrough) of he VAT reducion will be passed on o consumers, who will reac o mainain roughly consan expendiure levels and herefore increase heir demand for consumpion goods. As shown by Barrel and Weale (2009), evidence from he Naional Insiue s Global Economic Model suggess ha he emporary VAT cu will resul in consumpion increasing by less han one per cen by he fourh quarer of 2009 and ha GDP is likely o rise by less han 0.5 per cen relaive o wha would have happened wihou he VAT cu. Moreover, afer he emporary reducion is over, boh consumpion and GDP will be depressed as a resul of he policy. On he oher hand, Crossley e al. (2009) esimae ha VAT revenues will fall by 13.4 per cen. Similarly, he Treasury esimaes he cos of he emporary VAT cu a 12.4 billion, or equivalenly 13.5 per cen of fiscal revenues. However, since all previous sudies use a saic parial analysis approach, hey fail o ake ino accoun ineremporal and general equilibrium effecs. As poined ou by Barrell and Weale (2009), he key policy issue in his conex is o quanify he impac of he emporary VAT cu on oupu, and o ha end a model for he whole economy is needed. In his paper we develop such a model using he general equilibrium approach o simulae a emporary - 2 -

3 Economic Issues, Vol. 16, Par 1, 2011 reducion in he VAT rae from 17.5 per cen o 15 per cen, which a his rae means a cu of 14.3 per cen for his paricular VAT rae. The approach used in his paper has several advanages as i allows us o ake ino accoun all general equilibrium effecs and o esimae he effecs of he fiscal change on he key macroeconomic variables sudied here. Moreover, using a Dynamic General Equilibrium (DGE) model makes i possible o quanify he effecs of he VAT cu no only on VAT revenues bu also on oal fiscal revenues. This is paricularly imporan, as he slowdown in VAT revenues associaed wih he fiscal simulus policy may be offse by a rise in revenues from oher axes. The resuls obained from he model confirm his possibiliy. We find ha his policy is oo emporary o have imporan quaniaive effecs on he economy, bu is qualiaive effecs are of grea imporance. The emporary VAT cu will generae an overshooing effec on key macroeconomic variables wih respec o heir seady sae values. Consumpion and oupu will increase during he VAT cu, bu will decrease below heir seady sae values once VAT revers o is previous level. Invesmen will decrease during he VAT cu, bu slighly increase above is seady sae value when VAT revers o is previous level. Finally, oal fiscal revenues will decrease during he emporary VAT cu and reurn o heir seady sae values afer he policy ends. The res of he paper is srucured as follows. Secion 2 describes he heoreical framework. Secion 3 presens he daa used and he calibraion procedure. Secion 4 shows he resul of simulaing a emporary VAT cu. Secion 5 sudies a counerfacual exercise o sudy he effecs of an anicipaed VAT cu. Finally, Secion 6 presens some conclusions. 2. THE MODEL We use a version of he models of Baxer and King (1993) and Conesa and Kehoe (2008) in which hree axes are considered: a consumpion ax, a labour income ax and a capial income ax. I is assumed ha here exiss habi persisence in consumpion. We consider a producion funcion ha relaes oupu o wo inpus: labour and privae capial socks. 3 The governmen axes privae consumpion goods, capial income and labour income o finance an exogenous sequence of lump-sum ransfers,. {} 0 T = 2.1. Households Consider a model economy where he decisions made by consumers are represened by a represenaive consumer, whose preferences are represened by he following uiliy funcion: - 3 -

4 G Fernández-de-Córdoba and J L Torres UC (, NH L) = γ log( C φc ) + (1 γ)log( NH L) (1) 1 C. 1 Privae consumpion in period is denoed by is consumpion in he previous period and φ > 0 is a consan reflecing habi persisence. The uiliy funcion depends on a quasi-difference of consumpion inroducing nonseparabiliy of preferences over ime. Given ha we wan o sudy he effecs of change in he consumpion ax, habi persisence may be a key elemen of he model. Leisure is NH L, and is calculaed as he number of effecive hours in he week imes he number of weeks in a year H, imes he populaion a he age of aking labour-leisure decisions, N, minus he aggregaed number of hours worked in a year L. The parameer γ (0<γ<1)is he proporion of oal income represened by expendiure on consumpion goods. The budge consrain faced by he represenaive consumer is: (1 + τ ) C + I = (1 τ ) WL + (1 τ ) RK + T (2) where T is lump-sum ransfers received by consumers from he governmen, I is gross invesmen, K is capial sock, W are compensaions o employees, R is renal rae, δ is he capial depreciaion rae, modelled as ax-deducible, c l k and τ, τ, τ, are a privae consumpion ax, a labour income ax, and a capial income ax, respecively. Capial holdings evolve according o: where δ is he capial depreciaion rae. Subsiuing he capial accumulaion equaion in expression (2) resuls: The budge consrain indicaes ha consumpion and invesmen canno exceed he sum of labour and capial renal incomes, ne of axes and lump sum ransfers. The problem faced by he represenaive consumer is o maximise he value of his/her lifeime uiliy given by: c l k subjec o he budge consrain (2), given τ, τ, τ, and given he endowmen of privae capial K 0. The consumer s discoun facor is β (0, 1). C c l k 1 ( 1 δ ) 1 K = K + I (3) [ ] (1 + τ ) C + K K = (1 τ ) WL + (1 τ ) ( R δ) K + T c l k 1 1 β γ log( {, } φ 1) (1 γ)log( ) C L + (5) = 0 Max C C N H L (4) - 4 -

5 Economic Issues, Vol. 16, Par 1, Firms The problem for he firm is o find opimal values for he uilisaion of labour and capial, given prices. The represenaive firm is assumed o have a sandard Cobb-Douglas producion funcion. The producion of final oupu, Y, requires labour services, L, and privae capial, K. Goods and facor markes are assumed o be perfecly compeiive. The firm rens capial and hires labour o maximise period profis, aking facor prices as given. The echnology exhibis consan reurns o scale on all facors and hus firms earn zero economic profi. The echnology is given by: Y = AK L α 1 α 1 (6) where A is a measure of oal facor produciviy, and α denoes he capial share of oupu. The firm s decision problem can be defined as: max { K, L } A K α 1 α 1L R K 1 W L (7) where R is he renal rae on privae capial and W is he real wage Governmen Finally, we consider he dual role of he governmen as a ax-levying eniy and as supplier of privae goods hrough ransfers, T. We assume ha he governmen balances is budge period-by-period by reurning revenues from disorionary axes o agens via lump-sum ransfers. The governmen obains resources from he economy by axing consumpion and income from labour and capial, he effecive average axes on c l k which are τ, τ, τ, respecively. The governmen budge in each period is given by: c [( R δ ) K ] T l k + τ W L + τ 1 τ C = Transfers o consumers are he counerpar of fiscal revenues. The assumpion of a period balanced budge is made o highligh he disorionary effecs of axes, mainly on capial accumulaion. 5 (8) 2.4. Equilibrium Our model has wo producion facors. Based on he firm's profi maximisaion problem, he firs-order condiions are: R = αa K α 1 1 α 1 L α α W = (1 α) A K 1L (9) (10)

6 G Fernández-de-Córdoba and J L Torres From he above equaions we can obain he following relaionships ha will be useful for our calibraion: R K Thus, he economy saisfies he following feasibiliy consrain: 1 = αy W L = ( 1 α ) Y C + W L = Y + I = R K 1 (11) (12) (13) To compue he soluion of he model, we assign he Lagrange muliplier λ o he budge consrain a dae. The firs-order condiions for he consumer are: β γ C 1 φc ) β c + 1 λ (1 + τ γφ 1 C+ 1 1 φc = 0 (14) 1 N H L l ( 1 γ ) + λ (1 τ ) W = 0 (15) E k 1 [(1 τ ) R + 1 δ] λ β 0 β λ = (16) where λ τ, he Lagrange muliplier of he budge consrain a ime, is defined as: λ = γ βγφ (17) c (1 + τ ) C φc 1 C + 1 φc Combining (14) and (15) we obain he condiion ha equaes he marginal disuiliy of addiional hours of work wih he marginal reurn on addiional hours: l 1 (1 τ ) 1 1 (1 γ) = c γ βγφ NH (1 ) C L + τ φc 1 C + 1 φc W (18) Equaion (16) is an ineremporal Euler equaion ha equaes he marginal cos of addiional capial wih he reurn on invesmen. Combining equaion (14) wih expression (16) yields: c 1 1 (1 + τ 1) γ C φc βγφ + 1 C+ 1 φc k = β (1 τ 1) R 1 1 c δ (1 + τ ) 1 1 γ C 1 φc βγφ + C+ 2 φc (19)

7 Economic Issues, Vol. 16, Par 1, 2011 Noe ha he ineremporal equaion for invesmen is no affeced by he consumpion ax if consan. Togeher wih he firs-order condiions of he firm, he budge consrain of he governmen (8), and he feasibiliy consrain of he economy (13), i characerises a compeiive equilibrium for he economy. Definiion: A compeiive equilibrium for his economy is a sequence of consumpion, leisure, and privae invesmen { C, NH L, I} = 0 for he con- sumers, a sequence of capial and labour uilisaion for he firm { K, L} = 0, and a sequence of governmen ransfers { T } = 0, such ha, given a sequence of prices, { W, R}, and axes, { c, k, l τ τ τ } = 0 : 0 = i. The opimisaion problem of he consumer is saisfied. ii. Given prices for capial and labour, he firs-order condiions of he firm are saisfied wih respec o capial and labour. iii. Given a sequence of axes, he sequence of ransfers is such ha he governmen consrain is saisfied. iv. The feasibiliy consrain of he economy is saisfied. Noice ha according o he definiion of equilibrium of our model economy, he governmen eners compleely parameerised, and fiscal policy is drawn up consisenly wih he model and he daa. In oher words, in our model he privae secor reacs opimally o policy changes, and hose policy changes ake place exogenously. This is of paricular imporance as we wan o simulae an unanicipaed emporary ax cu. 3. DATA AND CALIBRATION Before simulaing he emporary VAT cu in he model, values mus be assigned o he parameers. The parameers of he model are: ( α, β, γ, δ, φ, τc, τl, τk) In calibraing he model presened in he previous secion we need hree differen ses of informaion: ax raes ( τc, τl,, τk) echnological parameers, ( α, δ ) and preference parameers, ( β, γ). Following Kydland and Presco (1982), we se as many parameers as possible in advance based upon a priori informaion. Firs, o deermine he value of he oal disposable effecive ime endowmen of individuals, NH, ha is, he non-sleeping hours of he working-age populaion, we assume ha each adul has a ime endowmen of 96 hours per week ( H = 96). Daa on he populaion aged from 15 o 64 and he average hours worked per year were obained from he OECD Corporae Daa Environmen Daabase

8 G Fernández-de-Córdoba and J L Torres Tax raes ( τc, τl, τk) : Compuaional macroeconomic models of fiscal policy depend crucially on realisic measures of ax raes. Agens' decisions depend on marginal ax and herefore effecive marginal axes should be used in he calibraion. However, marginal ax raes are hard o esimae and, as poined ou by Mendoza e al. (1994), i is ofen impracical o do so given he limiaions of daa availabiliy and difficulies in dealing wih he complexiy of ax sysems. Mendoza e al. proposed a mehod for esimaing effecive average ax raes and showed ha hey are wihin he range of marginal ax raes esimaed in previous works and display very similar rends. On he oher hand, hese auhors argue ha heir definiion of effecive average ax raes can be inerpreed as an esimaion of specific ax raes ha a represenaive agen, in a general equilibrium conex, akes ino accoun. Carey and Rabesona (2002) esimae OECD ax raes using a revision of he mehod proposed by Mendoza e al. (1994), hough relaxing some assumpions. They find ha for he period , for he UK, he consumpion ax rae is 0.157, he labour income ax rae is and he capial income ax rae is For he same sample period, he esimaes of Boscá e al. (2008) are fairly similar (0.136 for he consumpion ax, for he labour ax and for he capial income ax). We use effecive average ax raes, 6 borrowed from Boscá e al. (2008), who used he mehodology proposed by Mendoza e al. (1994) o esimae hem. For he year 2004, he esimaed values are a consumpion ax rae of 0.124, a labour ax rae of and a capial ax rae of As poined ou by Adam and Browne (2006) abou 55 per cen of (gross) consumpion is axed a 17.5 per cen. This implies ha abou 51 per cen of (ne) consumpion is axed and ha he average ax rae is abou 8.9 per cen. The VAT cu implies a reducion of per cen in he 17.5 per cen VAT rae. However, oher axes on consumpion remain unchanged. Therefore, for he simulaion we consider a consumpion ax rae of and ha he consumpion ax rae cu is of per cen. Preference parameers ( β, γ, φ) : Preference parameers are calibraed as in Conesa and Kehoe (2008). Firs, he ineremporal discoun facor is 0.99, as we simulae he model using quarerly daa. This is he sandard value used in he lieraure. Second, he parameer γ is obained using firsorder condiions from he model and daa observaions for 2008, aken from he OECD Naional Accouns Daabase. From he firs-order condiions we can obain he following value for γ as a funcion of daa observaions: C γ = l 1 τ ( ) W( N H L ) + C c 1+ τ Finally, he habi persisence parameer is se a 0.7, he esimae repored in Boldrin e al. (2001) and similar o he value frequenly used in he lieraure. Simulaions have also been done for φ = 0-8 -

9 Economic Issues, Vol. 16, Par 1, 2011 Technological parameers (α, δ): Third, we use daa from naional income and produc accouns o calibrae echnological parameers. The capial income share, α, and he depreciaion rae of privae capial, δ, are aken from he EU-Klems daabase. 7 The seleced values for he UK are and for a quarerly basis. The seleced values for he UK are α = and δ = for a quarerly basis. Table 1 summarises he calibraed parameer values for he UK economy o be used in he compuaions. Table 1. Calibraed parameer values Parameer Source Value α EU-Klems daabase β Arbirary γ OECD daabase δ EU-Klems daabase φ Boldrin e al. (2001) τ Boscá e al. (2008) c τ Boscá e al. (2008) l τ Boscá e al. (2008) k 4. THE IMPACT OF THE TEMPORARY VAT CUT Given he calibraion of he model for he UK economy, we can quanify he effecs of he emporary VAT cu on all he variables defined in he model: oupu, consumpion, gross invesmen, labour, oal ax revenues and ax revenues from each ax. We proceed as follows. Firs, he model is calibraed for he year 2008 o replicae he main macroeconomic aggregaes of he UK economy. We consider his calibraion as he seady sae of he economy. Second, we calculae he implici change in he consumpion ax rae derived from he VAT cu policy. Finally, we use he calibraed model economy o simulae a ransiory VAT cu for 2009, wih he ax revering o is previous level in he firs quarer of Our simulaion exercise considers ha he VAT cu was no anicipaed bu ha he ime when he ax will reurn o is previous value is known by he agens and, accordingly used as informaion in heir decision process

10 G Fernández-de-Córdoba and J L Torres A VAT reducion policy resuls in wo effecs: an income effec and a subsiuion effec. Income effecs resul from he fac ha a VAT reducion is equivalen o a rise in disposable income. This effec is imporan in he case of a permanen VAT cu, and indeed is he only effec operaing in he longrun. However, in he case of a emporary VAT cu, income effecs are likely o be small given ha changes in lifeime income will be minimal. However, as shown by Blundell (2009) and Barrell and Weale (2009), only in he case of myopic or credi-consrained consumers would, he income effecs of a emporary VAT cu be imporan. Second, ineremporal subsiuion effecs will operae only in he case of a ransiory VAT change or an anicipaed VAT change. In he case of a permanen, unanicipaed VAT cu, ineremporal subsiuion effecs do no apply as consumers re-opimise heir consumpion pah a he new ax level. Barrell and Weale (2009) repor ha simulaions from he Naional Insiue's Global Economic Model sugges ha he VAT cu will resul in consumpion increasing by less han one per cen by he fourh quarer of 2009 and GDP by less han 0.5 per cen, relaive o wha would have happened wihou he VAT cu. Since he simulaion is carried ou using quarerly daa, we simulae a VAT cu during four quarers as an approximaion o he 13 monhs VAT-cu simulus policy (December 1, 2008-December 31, 2009). We also consider a change in he consumpion ax, which includes oher indirec axes as well as VAT. The VAT reducion o 15 per cen implies an effecive reducion in he consumpion ax of 11.6 per cen. Therefore, his implies ha he consumpion ax of calibraed for he UK will be reduced for four quarers o Firs, we conduc an exercise on he effec of a ransiory VAT cu, replicaing he policy applied in he UK. Table 2 summarises he main resuls. We use his exercise as a benchmark for he permanen VAT cu simulaion. Figure 1 shows he effecs of a ransiory VAT cu where he subsiuion and arbirage effecs are he only relevan effecs, as he ransiory VAT cu does no affec lifeime income. Seady sae values for each variable are normalised o. Impulse-response funcions reflec percenage changes wih respec o seady sae values. We plo he impulse-response of he main variables wih habi persisence and he sandard model wihou habi persisence. The emporary VAT cu increases consumpion, oupu and labour and decreases invesmen. In all cases excep hours worked, we observe an overshooing effec afer he end of he emporary VAT cu. Neverheless, he effecs on seleced key macroeconomic variables are quaniaively small, excep on consumpion (a 1 per cen increase) and invesmen (a 4 per cen decrease) during he emporary VAT cu period. The model predics a rise in oupu of abou 0.2 per cen in he firs quarer, 0.13 per cen in he second quarer and 0.07 per cen in he hird quarer. Afer VAT revers o is previous level, he model predics a persisen

11 Economic Issues, Vol. 16, Par 1, 2011 reducion in oupu. This implies ha he emporary VAT cu gives rise o an undershoo in oupu. Figure 1: Response o he emporary VAT cu: Main variables. Sandard model (solid line), Habi persisence (dash line).5 Oupu Consumpion Quarers Invesmen Quarers Hours Worked Quarers 99 Quarers The ax cu policy was announced only six days before i came ino effec (i.e., i was a surprise), so he possibiliy of agens changing heir consumpion pah prior o he change was very limied. Afer he change, agens vary heir consumpion pah, given ha consumpion during 2009 will be cheaper han in This explains he impulse-response of consumpion, which increased by around one per cen in 2009, and he reducion in consumpion afer he ax revers o is previous level. Gross invesmen shows a similar bu invered pah. The reducion in invesmen is abou 4 per cen. I is imporan o noe ha his large reducion in invesmen resuls in a reducion in he capial sock accumulaion process. This is responsible for he prediced undershooing effec in oupu. When VAT revers o is previous level he model predics an insananeous recovery of invesmen o is seady sae value

12 G Fernández-de-Córdoba and J L Torres Table 2 compares he effec of he emporary VAT cu wih a permanen VAT cu. We show he esimaion under our model of a cu in he consumpion ax on oupu (GDP), consumpion (C), labour (L), Invesmen (I), and fiscal revenues (FR), as percenage changes from he level a he ime of impac. We compue he percenage change of each variable in all four quarers of 2009, he firs and fourh quarer of 2010, he fourh quarer of 2011 and he long-run. Obviously, he long-run effec of he ransiory consumpion ax cu is zero. The differences beween he simulaions of he emporary VAT cu and a permanen VAT cu are due o he income effec. In he case of a emporary ax change he income effec is likely o be very small as changes in permanen income will be minimal, whereas in he case of a permanen ax change all consequences are due o he income effec. Only in he case of credi-consrained agens will income effecs be significan under a ransiory ax change. Table 2. VAT cu policy Period 2009Q1 2009Q2 2009Q3 2009Q4 2010Q1 2010Q4 2011Q4 Long-run GDP Percenage change Temporary VAT cu Permanen VAT cu C L I FR GDP C L I FR As menioned in he Inroducion, Crossley e al. (2009) calculae an increase of 1.2 per cen in consumpion. We obain slighly lower measures of abou 1.05 in 2009, and a much lower figure when he VAT ax cu is permanen, wih an increase of 0.85 per cen in he long run, suggesing ha, as in Blundell (2009), here is an incomplee pass-hrough of he ax. Our measure is herefore closer o ha of Barrell and Weale (2009), who esimae an increase in consumpion of less han one per cen by he fourh quarer of They also esimae an increase in oupu of less han a half of one per cen and we indeed find ha oupu peaks on impac, reaching a mere 0.19 per cen. As prediced by hese auhors, oupu and consumpion reurn o heir previous level afer overshooing o negaive per cen changes. Barrell, Fic and Liadze (2009) repor oupu mulipliers for indirec ax cus in a number of counries and demonsrae ha hey are generally below one

13 Economic Issues, Vol. 16, Par 1, 2011 Anoher relevan quesion is ha relaed o he impac of he fiscal simulus policy on ax revenues. An addiional advanage of a general equilibrium model is ha we can quanify he effecs on all variables considered. This means ha he model allows us o quanify no only he change in consumpion ax revenues bu also he change in he oher wo axes revenues and in oal axes revenues. Obviously, a reducion in VAT will reduce VAT revenues. HM Treasury (2008) provides an esimaion of a reducion of 12.4 billion (13.5 per cen of revenue) from he emporary VAT cu. Crossley e al. (2009) esimae ha he revenue cos of he simulus will be lower by 0.1 per cen, as hey expec a slighly sronger expendiure response. Figure 2: Response o he emporary VAT cu: Fiscal Revenues. Sandard model (solid line), Habi persisence (dash line) 102 Toal ax revenues 105 Consumpion ax revenues Quarers Labour ax revenues Quarers Quarers Capial ax revenues Quarers Figure 2 shows he oal effec on fiscal income of a emporary VAT ax cu, and he breakdown of he oal effec ino he hree ypes of axes ha we consider in our model economy. The measure provided for oal fiscal revenues is a conracion of 3.19 per cen for he fourh quarer of The decomposiion of his can be seen in Figure 2. The 10 per cen fall in VAT revenues prediced by our model, in line wih 13.5 per cen esimaes from HM Treasury and Crossley e al. (2009), is empered by he simulus in labour supply and he increase in he renal rae of capial. From his general equilibrium approach we can race he sources of hese changes: oal labour fiscal income peaks on impac due o increased labour use, alhough i is lowered by he decrease in wages. The opposie occurs wih invesmen: he real reurn on capial increases and his nominal increase

14 G Fernández-de-Córdoba and J L Torres dives he increase in fiscal capial income despie he reducion in oal invesmen. Overall, he emporary consumpion ax cu will decrease oal fiscal revenues by around 3 per cen in he four quarers in which he cu is in place. However, given he overshooing effec of he economy, in he firs quarer of 2010, fiscal revenues will be slighly below he seady sae value. As can be seen from Table 2, and unlike oher esimaes, we regard our resuls beer in erms of fiscal revenue coss as a consequence of he behaviour of consumpion in he shor erm dynamics, and he behaviour of oal wages paid ogeher wih an increase in capial rens. The resuls indicae ha under a permanen consumpion ax cu of he same magniude, oal fiscal revenues would decrease by per cen on impac and by per cen in he long-run, while oher variables such as GDP, consumpion, invesmen and labour would respond proporionally. 5. THE EFFECTS OF AN ANTICIPATED FISCAL POLICY The shor run dynamics of a policy consising of a emporary change of he ax code vary depending on wheher he policy comes as a surprise or anicipaed, even if i is of he same magniude. The model ha we have presened can assess hese differences. We nex presen he equivalen of Table 2, bu assuming ha he governmen had announced he policy one year in advance. In his case agens can change heir consumpion and invesmen pahs in accordance wih heir informaion abou fuure VAT raes. Table 3. Anicipaed emporary VAT cu Period 2009Q1 2009Q2 2009Q3 2009Q4 2010Q1 2010Q4 2011Q4 Long-run Percenage change Anicipaed emporary VAT cu GDP C L I FR Table 3 shows ha modelled raional agens make use of available informaion, improving he resuls of he policy when compared o he same unanicipaed change in he fiscal ax code. Since in his exercise he decrease in he VAT rae is anicipaed, consumers end o reshape heir consumpion pah, waiing for he fiscal advanages of posponing consumpion over a year. Afer he shock, aggregae consumpion increases more han i does if he

15 Economic Issues, Vol. 16, Par 1, 2011 shock is unanicipaed. The negaive effec on fiscal revenues is almos idenical, bu posponed over a year, allowing fiscal auhoriies o reac o he change in he nex fiscal year. The effecs on oupu are similar, wih he difference ha he posiive simulus lass longer and he overshoo is smoohed ou. The impac of he announcemen is posiive on invesmen and also on consumpion in he shor run, while he conracion originaed by he reurn o he previous fiscal ax code is slighly smoohed. Figure 3: Anicipaed response o he emporary VAT cu: Main variables. Sandard model (solid line), Habi persisence (dash line) 101 Oupu Consumpion Quarers Invesmen Quarers 99.5 Quarers Hours Worked Quarers These differences are shown clearly in Figure 3. The overshooing effecs are smoohed over he firs periods, leaving he posiive aspecs of he policy almos unchanged. The minimum level reached by GDP wih he anicipaed change is per cen, whereas he minimum level wih he unanicipaed policy is per cen, which implies ha around hree-quarers of he negaive impac is smoohed by he announcemen. The negaive effec on consumpion peaks a per cen wih he unanicipaed change, compared wih per cen wih he anicipaed policy, meaning ha 13.3 per cen of he negaive effec is smoohed by he announcemen. Invesmen peaks a 1.65 per cen wih he announced policy, compared wih 0.27 per cen under he non-anicipaed policy. The negaive effec is minimal when

16 G Fernández-de-Córdoba and J L Torres compared wih he posiive difference. I is no surprising ha he main difference is found in invesmen, since in his ype of forward-looking models invesmen is he smoohing variable. When he shock is anicipaed, invesors can reac opimally o he decision made by he governmen, and as a consequence he announced policy provides beer resuls. Figure 4: Anicipaed response o he emporary VAT cu: Tax revenues. Sandard model (solid line), Habi persisence (dash line) 102 Toalax revenues 105 Consum pion ax revenues Quarers Labourax revenues Quarers 85 Quarers Capialax revenues Quarers The same quaniaive resuls can be derived from he impac on fiscal revenues of he announcemen of he policy. Figure 4 shows his. Fiscal revenues peak a 0.1 per cen by he end of 2009, while hey are negaive in all periods wih he unanicipaed policy. The minimum change wih respec o he base year is per cen by he end of 2010 wih he anicipaed policy, while he minimum level aained wih he unanicipaed policy is per cen, jus 0.31 per cen more. Overall, i is clear ha anicipaing his kind of policy is a beer opion for he economy. 6. SUMMARY This paper provides a quaniaive measure of he effecs on he emporary VAT cu policy in he UK using a canonical Dynamic General Equilibrium model. We provide a comparison wih oher resuls and wih a raionale for he differences found. We find ha in he long run he general equilibrium

17 Economic Issues, Vol. 16, Par 1, 2011 approach provides similar resuls o hose obained from a linear model where all variables reac proporionally o exogenous change in policy, bu he resuls are differen when he shor run dynamics are analysed. We find ha he emporary VAT cu policy will generae an overshooing effec on consumpion, oupu and invesmen. Consumpion and oupu will increase during he VAT cu, bu will decrease below heir seady sae values afer he VAT rae revers o is previous level. In conras, invesmen will decrease during he VAT cu, bu increase slighly above is seady sae value afer he end of he emporary fiscal simulus policy. The dynamic approach also allows us o perform a differen kind of analysis once he model has been validaed by esimaes made by oher auhors following differen approaches, o measure he policy implicaions of he fiscal change, namely o compare anicipaed and unanicipaed policies. The resuls show ha providing economic agens wih enough informaion o enable hem o smooh heir behaviour, yields beer resuls for he economy. Acceped for publicaion: 7h Ocober 2010 ENDNOTES 1. Deparmen of Economics, Universiy of Málaga, Campus El Ejido, s/n, Málaga (Spain). gfdc@uma.es (Fernández-de-Córdoba); jorres@uma.es (Torres). We would like o hank J Pérez and J Rodríguez and one anonymous referee for very useful commens and suggesions. We graefully acknowledge financial suppor from SEJ-122 and P07-SEJ Crossley e al (2009) argue ha consumers expec he emporary VAT cu oday o be offse by higher axes in he fuure (i.e. Ricardian equivalence holds), and herefore he presen value of wealh is unchanged. 3. An alernaive specificaion, following Guo and Lansing (1997) and Cassou and Lansing (1998), would include public capial. We have disregarded such a producion funcion because he consideraion of public capial socks does no subsanially change our resuls, given he level of aggregaion of he model. 4. Governmen consumpion is no considered in our model in order o keep hings as simple as possible. Neverheless, he same analysis has been repeaed using a model in which governmen consumpion is included, and he resuls do no change significanly (hese resuls are available from he auhors upon reques). 5. This assumpion has been used by Barro (1990), Glomm and Ravikumar (1994), and Cassou and Lansing (1998), among ohers. They argue ha his seup may represen a closer approximaion o acual consrains han one which allows he governmen o borrow or lend large amouns. 6. The use of average effecive ax raes involves he use of conservaive values (smaller implied behavioural responses)

18 G Fernández-de-Córdoba and J L Torres REFERENCES Adam S and Browne J (2006) A survey of he UK ax sysem, IFS Briefing Noe no. 9, Insiue for Fiscal Sudies. Barro R (1990) Governmen spending in a simple model of endogenous growh, Journal of Poliical Economy, 98, Barrell R, Fic T and Liadze I (2009) Fiscal policy effeciveness in he banking crisis, Naional Insiue Economic Review, 207, Barrell R and Weale M (2009) The economics of a reducion in VAT, Fiscal Sudies, 30, Baxer M and King R (1993) Fiscal policy in general equilibrium, American Economic Review, 83, Boldrin M, Chrisiano L and Fisher J (2001) Habi persisence, asse reurns and he business cycle, American Economic Review, 91, Boscá J, García J and Taguas D (2008) Taxaion in he OECD: , Documeno de Trabajo Miniserio de Economía y Hacienda, Gobierno de España. Blundell R (2009) Assessing he emporary VAT cu policy in he UK, Fiscal Sudies, 30, Carey D and Rabesona J (2002) Tax Raios on Labour and Capial Income and on Consumpion, OECD Economic Sudies, no. 35. Cassou S and Lansing K (1998) Opimal fiscal policy, public capial, and he produciviy slowdown, Journal of Economic Dynamics and Conrol, 22, Conesa J C and Kehoe T (2008) Produciviy, Taxes and Hours Worked in Spain: , Working Paper. Crossley T F, Low H and Wakefield M (2009) The economics of a emporary VAT cu, Fiscal Sudies, 30, Glomm G and Ravikumar B (1994) Public invesmen in infrasrucure in a simple growh model, Journal of Economic Dynamics and Conrol, 18, Guo J and Lansing K (1997) 'Tax srucure and welfare in a model of opimal fiscal policy', Federal Reserve Bank of Cleveland, Economic Review, Q1, HM Treasury (2008) Pre-Budge Repor November 2008: Facing Global Challenges: Supporing People Through Difficul Times. Kydland F and Presco E (1982) Time o build and aggregae flucuaions, Economerica, 50, Mendoza E, Razin A and Tesar L (1994) Effecive ax raes in macroeconomics. Crosscounry esimaed of ax raes on facor incomes and consumpion, Journal of Moneary Economics, 34,