The Consequences of Fiscal Illusion on Economic Growth

Transcription

1 (2008) vol. 6, no. 2, pp The Consequences of Fscal Illuson on Economc Growth Paulo Res Mourão Abstract Ths work dscusses the mpact of fscal lluson on economc growth. Its man contrbuton hghlghts the need for reducng the expected return from partcpatng n fscal lluson practces n order to prevent adverse effects on economc growth. Addtonally, ths model renforces the advantages of productve publc goods (not devated for poltcal unproductve rents) n order to mtgate the negatve effects of fscal lluson.. INTRODUCTION Ths orgnal short artcle ams at dscussng the mplcatons of fscal lluson on economc growth rates. For ths purpose, the followng secton wll contextualze the dscusson, ntroduce a model derved upon the orgnal sense of Puvan s (903) Fscal Illuson, and conclude that hgher levels of fscal lluson decrease growth rates. However, ths negatve effect s reduced by hgher values of productve publc consumpton. 2. TOWARD A MODEL FOR DISCUSSING THE FISCAL ILLUSION CONSEQUENCES ON ECONOMIC GROWTH RATES When Amlcare Puvan (903) publshed The Theory of Fscal Illuson he was foundng the economcs of lluson the study of publc choces made by some agents characterzed by mperfect knowledge. After more than a half of a century, James Buchanan (960) gave new lfe to that obscure work and to the fscal lluson theory. James Buchanan, nfluenced by the work of Downs (957), extended Puvan s approach to analyze the substantal lag between the true ntentons of governments and the belefs of the electorate. Ths lag s usually manpulated to ncrease the sze of the government through less vsble (and less reactve) taxaton. Núcleo de Investgação em Polítcas Económcas; Department of Economcs; Unversty of Mnho, Portugal. Emal: paulom@eeg.umnho.pt. The author would lke to thank the anonymous referees of the ejournal of Tax Research, Domenco da Empol, Lnda Vega, Mchael Pckardt, Renato Flores, Francsco Dnz and semnar partcpants at the INFER Workshop "Integraton and Globalzaton: Challenges for Developed and Developng Countres" (Combra, June 2007) at the ISEG Summer School Semnar (Lsbon, 6 July 2007) and at UTAD SPID Semnar (Vla Real, 2 September 2007) for ther useful comments and suggestons. All remanng errors are the author s responsblty. The author acknowledges the support from MCTES/FCT/BD.262/05. 82

2 The Consequences of Fscal Illuson on Economc Growth The orgnal sense of Puvan s deas suggested fscal lluson as a soluton to a pror queston: how can resstance to governmental actons be dmnshed from the perspectve of taxpayers? Accordng to Buchanan (967), the soluton manly studes fscal lluson n the revenue sde of a budget. Illuson can be nserted nto revenues n many ways: obscuraton of the ndvdual shares n the opportunty cost of publc outlays; utlzaton of nsttutons of payments that are planned to bnd the requrement to a tme perod or an occurrence whch the taxpayer seems lkely to consder cheerng; chargng explct fees for nomnal servces provded upon the occurrence of mpressve or pleasant events; levyng taxes that wll captalze on the sentments of socal fear, makng the burden appear less than mght otherwse be the case; use of scare tactcs that have a propensty to make the alternatves to partcular tax proposals seem worse than they are; fragmentaton of the total tax weght on an entty nto numerous small leves; and opacty of the fnal ncdence of the tax. The fnal result of ths lluson s always gatherng hgher amounts of publc revenues wth a mnmum of electorate resstance. Due to the stmulaton from Buchanan s redscovery, ths knd of fscal lluson can properly be labelled the Puvan Buchanan (P B) fscal lluson. However, as far as we are aware, there s a very sgnfcant absence of studes reportng the consequences of P B fscal lluson on economc growth rates. We can pont out some studes relatng fscal lluson and Publc Fnances (Oates, 988; Rogers and Rogers, 995; Easterly, 999), but we have no framework dscussng how economc growth wll react to dfferent levels of fscal lluson. Ths work, more precsely the followng secton, ntends to contrbute to ths purpose, developng the standard AK model (Barro and Sala--Martn, 995, pp ). 3. FISCAL ILLUSION AND A RENT-SEEKING GOVERNMENT The producton functon for a gven frm takes an AK Cobb-Douglas form Y = AL K G, (3.) where 0 < <, A s the level of technology, L s labor nput, K s captal nput and G s the total of government purchases. Therefore, t s assumed that producton for each frm s characterzed by constant returns to scale n the prvate nputs, labor and captal. Addtonally, t s also assumed that the aggregate labor force, L, s constant. For a fxed G, the economy would be characterzed by dmnshng returns to the accumulaton of aggregate captal, K. By statng that G rses along wth K, we assume that (3.) wll not be characterzed by dmnshng returns and that an ncrease n G rases the margnal products of L and K. Thus, the economy s capable of endogenous growth 2, followng the tradtonal AK pattern. Mourao (2007) s an exhaustve survey on the vast lterature that followed the orgnal Puvan (903) Buchanan (960) sense of fscal lluson. 2 The equvalence of the exponent on G to mples that the constant returns to K and G generate endogenous growth,.e., the economy should only ncrease G n a way that t accompanes a rse n K. 83

3 The Consequences of Fscal Illuson on Economc Growth Now, assume that the government has a balanced budget. Ths balanced budget s fnanced by a proportonal tax at rate t charged on the aggregate of gross output G = ty. (3.2) We also suppose that t and, hence, the expendture rato, G/Y, are constant over tme. In our frst case, t s assumed that there s only fscal lluson perceved by frms, that s, frms know there s an announced proportonal tax rate t, however due to the level of fscal lluson f 3, frms actually pay an effectve tax rate (+f)t. In ths frst stuaton, we assume that the government acheves poltcal rents (ft) used for prvate and unproductve ends, and although frms pay the effectve tax rate, the balanced budget only ncorporates t. The frm s after-tax proft s gven by [( ( + f ) t) * Ak G w ( r + δ k ] L ) where k K /L, r s the rate of return on captal, w s the wage rate and δ s the deprecaton rate of captal. The wage rate equals the after-tax margnal product of labor because we assume that frms follow the assumptons of proft maxmzaton and the zero-proft condton. Addtonally, we also obtan that the gross rental rate r + δ equals the after-tax margnal product of captal. Therefore, f we assume k = k the rental prce s gven by r + δ = Y [ ( + f ) t] = [ ( + f ) t] Ak G K (3.3) Usng (3.) and (3.2), we fnd an expresson for G: / ( tal) k G =. (3.4) Substtutng (3.4) nto (3.3) we obtan / [ ( + f ) t] A ( Lt) r + δ = (3.5) On the rght-hand sde of (3.5), the after-tax margnal product of captal plays the same role n the growth process that the constant A played n the standard AK model. 3 Followng Puvan (903), the fnal consequence of nsertng fscal lluson n the process of tax collecton s an ncrease of the amount actually pad by tax-payers and a mnmzaton of ther resstance. For convenence, we assume that f [ 0,] where a hgher f characterzes a hgher degree of fscal lluson. 84

4 The Consequences of Fscal Illuson on Economc Growth As there are no transtonal dynamcs, the growth rates of c, k 4, and y all equal the same constant, γ 5. de, rs γ = ( ) de, rs A Lt [ ( + f ) t] δ ρ. (3.6) θ The effects of government on growth are obtaned through two channels: the term ( + f )t represents the negatve effect of effectve taxaton on the after-tax margnal product of captal, and the term t represents the postve effect of G, the publc servces, on the margnal product. γ Computng we get t γ de t, rs A L = 2 ( Lt) ( + ft + t ) θ. (3.7) Therefore, the golden rule for the sze of the government fnds a maxmum 6 at t =. (3.8) + f The condton (3.8) corresponds to the natural effcency condton for the sze of the Y government = + f,.e., as the socal cost of a unt of G s +f and the beneft s G the margnal product of publc servces, the effcency condton equates the margnal cost to the margnal beneft. Followng (3.8), we can observe that the golden-rule growth rate s * 2 L( ) γ = δ ρ de, rs A. (3.8 ) θ + f 4 In our case, we assume that nfnte-lved households maxmze utlty, as gven by ( ) c θ nt U 0 e ρ = dt, subject to the constrant a = ( r n) a+ w c, where c s consumpton per person, a s θ assets per person, and n s the growth rate of populaton. Assumng t s a closed economy, a=k may hold (Barro and Sala--Martn, 995: 40-4). 5 Some nequalty condtons are requred for the growth rate to be postve and for utlty to be bounded: Y ( f ) t Y + δ>ρ and the transversalty condton ( / ) ( ) ρ f t 0 K θ θ + δ+ >. K θ 2 γ de, rs A ( Lt) (2+ ft+ t ) = t 2 t 2 θ 6 ( ) 2 s negatve as t>, partcularly at t* =. + f + f 85

5 The Consequences of Fscal Illuson on Economc Growth If we want to check the effects of fscal lluson on the optmal decentralzed growth rate, we calculate ts partal dervatve: * γ de f A, rs = ( ) L + f θl (<0). Therefore, we conclude that hgher levels of fscal lluson decrease the growth rate n a decentralzed economy under the prevous assumptons. For the moment, we have shown that (3.8) s the government s best polcy, gven that the growth rate s the result of the decentralzed choces of households and frms n accordance wth (3.6). Now, t s tme to observe whether the outcomes are Pareto optmal by solvng the socal planner s problem. The planner determnes the tme paths G(t) and c(t) n order to maxmze the θ ( ρ n) t c consumer s utlty U = e dt. The planner s constraned by the 0 θ producton functon (3.) and the budget constrant Y = C + G + K + δk. (3.9) It s not dffcult to set up a Hamltonan expresson to reach the condtons for dynamc optmzaton n the socal planner s problem. Ths case wll result n a dfferent growth rate chosen by the socal planner: γ sp, rs = A. (3.0) θ [( ) L] δ ρ Y The socal planner satsfes the condton =. The key dstorton n the G decentralzed model s that nvestors consder the prvate margnal product of captal Y [ ( + f ) t] because of the effectve tax rate ( + f )t, whch s slghtly K Y dfferent from. Ths dfference between socal and prvate returns produces a K shortfall of the growth rate of (3.6). The dfference s also explaned because n (3.0), the negatve effect of the effectve taxaton s replaced by and the sze of government s gven by. Consequently, we conclude that hgher levels of fscal lluson may magnfy the dstorton promoted by taxaton n ths economy. 86

6 The Consequences of Fscal Illuson on Economc Growth 3.2 FISCAL ILLUSION AND A BENEVOLENT GOVERNMENT In ths case, (3.) retans the same producton functon, but (3.2) s now modfed nto (3.): G = ( + f ) ty. (3.) Therefore, we are assumng that the total of government purchases react postvely to the level of fscal lluson, whch can be vewed by ncumbents as a way of wastng more publc resources (Buchanan, 960; Buchanan and Wagner, 977). In ths case, we follow the assumpton of a benevolent government: ths government uses all the collected effectve taxaton n order to stmulate the economy, not hdng values for opportunstc drectons. The frm s after-tax proft s agan: [( ( + f ) t) * Ak G w ( r + δ k ] L ) The rental prce s now: / [ ( + f ) t] A [ Lt( + f )] r + δ = (3.2) And the growth rates of c, k, and y are gven by γ de, b : γ = [ + ] de b A ( f ) Lt, [ ( + f ) t] δ ρ. (3.3) θ Maxmzng the growth rates to t leads agan to growth rate s t =. Therefore, the maxmum + f γ 2 = A θ * de, b. [ L( )] δ ρ * We can check that γ de,b could be acheved n (3.8 ) f there was no fscal lluson, f=0. * It s straghtforward to conclude that γ de,b > γ * de,rs. At ths pont, we can state that a benevolent government can mnmze the harm of fscal lluson on the growth rates. In ths case, we can no longer pont out that fscal lluson s a negatve determnant of economc growth because ts capablty of attrton was reduced by the benevolence of the government, whch released all mones obtaned by the effectve taxaton nto the economy. In ths second case, the planner s constraned by the producton functon and a new budget constrant: Y = C + ( + f ) G + K + δk. (3.4) 87

7 The Consequences of Fscal Illuson on Economc Growth These changes wll lead to a dfferent growth rate: γ sp,b ( ) = L A δ ρ. (3.5) θ + f Checkng what happens to the socal planner s problem of a benevolent government, we fnd that the dfferences between the socal planner s solutons and the decentralzed solutons are smaller n ths second case, ndcatng a proxmty (smaller wedge) between the Pareto soluton and the ratonal choces of households and frms. 7 Wth few assumptons 8, t s straghtforward to conclude that γ < <. de, rs γ de, b < γ sp, b γ sp, rs These nequaltes show that a hgher level of P B fscal lluson orgnatng n poltcal rents used for prvate and unproductve drectons generates low growth rates. When fscal lluson s characterzed by smaller values or when the poltcal rents are beng nvested n the economy (becomng productve), we face ncreasng rates. Therefore, we have shown that the P B fscal lluson can be a sgnfcant determnant n the process of economc growth, functonng as a source of attrton: hgher levels of fscal lluson prejudce the economc growth rates. Therefore, fghtng fscal lluson, makng publc fnances more transparent, s mportant for a healthy budget composton and for the overall economc growth. 4. CONCLUSION Ths work demonstrated that the controversal queston nvolvng the role of fscal lluson practces on publc fnances s not recent, but can be thought of as dervng from the dscusson nvoked by Puvan (903) and substantally enrched by Buchanan (960). In spte of the fact that the Fscal Illuson School of Buchanan and Wagner (977) dentfes hgher levels of fscal lluson promotng ncreasng ncrements n the sze of the publc sector, ths work developed a model that predcts hgher levels of fscal lluson also decrease natonal economc growth rates. Ths model has had further and mportant mplcatons. Manly, t hghlghted the need for reducng the expected return of ncurrng n fscal lluson practces n order to prevent adverse effects on economc growth. Addtonally, ths model renforced the advantages of productve publc goods (not devated for poltcal unproductve rents) n order to mtgate the negatve effects of fscal lluson. 7 If fscal lluson s too hgh,.e, f ln( + f ) ln holds, γ sp, b s smaller than γ de, b. Ths can be thought as a dsadvantage of a too heavy publc sector because the socal planner satsfes the pro-levathan Y condton =+ f n ths case. G 8 We follow the assumptons that all the parameters of the equatons have sgnfcant values and that ln( + f ) < ln. 88

8 The Consequences of Fscal Illuson on Economc Growth REFERENCES Barro, R. and X. Sala--Martn 995, Economc Growth, McGraw-Hll, New York. Buchanan, J. 960, Fscal Theory and Poltcal Economy, Unversty of North Carolna Press, Chapel Hll, pp Buchanan, J. 967, Publc Fnance n Democratc Process: Fscal Insttutons and Indvdual Choce, Unversty of North Carolna Press, Chapel Hll. Buchanan, J. and R. Wagner 977, Democracy n Defct: The Poltcal Legacy of Lord Keynes, Academc Press, New York. Downs, A. 957, An Economc Theory of Democracy, Harper Collns, New York. Easterly, W. 999, When s fscal adjustment an lluson?, Polcy Research Workng Paper Seres No. 209, World Bank, Washngton DC. Mourao, P. 2007, The economcs of lluson A dscusson based n fscal lluson, Journal of Publc Fnance and Publc Choce, 25(), forthcomng. Oates, W. 988, On the nature and measurement of fscal lluson: A survey, Taxaton and Fscal Federalsm: Essays n Honour of Russell Mathews, Brennan, G., B. Grewal, and P. Groenwegen (eds), Australan Natonal Unversty Press, Sydney, pp Puvan, A. 903, Teora della Illusone Fnancaría, Sanaron, Palermo. Rogers, D. and J. Rogers 995, Poltcal competton, causal relatonshp between taxes and spendng, and ther nfluence on government sze: Evdence from state-level data, Internatonal Fnance Dscusson Paper No. 500, Board of Governors of the Federal Reserve System, Washngton DC. 89