What explains the redistribution achieved by the Italian Personal Income Tax? Evidence from Administrative Data

Transcription

1 What eplans the redstrbuton acheved by the Italan Personal Income Ta? Evdence from Admnstratve Data Gan Paolo Barbetta Department of Economcs and Fnance, Catholc Unversty, Mlano (IT) Smone Pellegrno Department of Economcs and Statstcs (ESOMAS), Unversty of Torno, Torno (IT) Glberto Turat Department of Economcs and Statstcs (ESOMAS), Unversty of Torno, Torno (IT) June 22 nd, 204 Abstract We analyze the Italan Personal Income Ta (IRPEF) n the lght of the dfferent methodologes avalable to the government to acheve ncome redstrbuton. We focus n partcular on three tools: margnal ta rates, deductons and ta credts. We then apply the standard Pfähler (990) decomposton, estmatng the contrbuton of these three tools to the Reynolds-Smolensky nde, to a sample of more than,3 mllon ndvdual ta returns. Our estmates suggest that more than a half of the total redstrbutve effect of the PIT s due to the ta credt for employment and the ta credt for retrement ncome, whle the margnal rate schedule contrbute wth 38.5 per cent. On the contrary, most of the temzed ependtures do not show any szable mpact on redstrbuton. JEL Codes: H23; H24 Keywords: Personal Income Ta, Redstrbuton, Pfähler Decomposton, Ta Credts

2 . Introducton In most of the countres around the world, the Personal Income Ta (PIT) s a comprehensve ta on ndvdual ncomes that s thought as a fundamental tool to realze redstrbuton thanks to ts progressve nature. In fact, ths role of PIT s confrmed by the lterature analyzng the dfferences n the degree of redstrbuton acheved by dfferent countres, whch shows however a great varaton around the globe and an mportant role for poltcal nsttutons (e.g., Padovano and Turat, 202). Italy s an nterestng case study n ths respect: t s a country where total publc revenues represent a large share of GDP, the effcency mpact of ths hgh level of taes s pretty much clear from the worryng economc prospect n terms of potental growth and productvty, but the degree of redstrbuton s among the lowest n the contet of Western countres. What about the PIT? What s the degree of redstrbuton acheved thanks to the PIT? And how s t acheved? These are the basc questons we ask n the present paper. We analyze the Italan PIT n the lght of the dfferent methodologes through whch the government can nfluence the degree of progressvty of the ta, hence ts redstrbutve role. We focus n partcular on three tools: margnal ta rates, deductons and ta credts. We then apply the standard Pfähler (990) decomposton, estmatng the contrbuton of these three tools to the Reynolds-Smolensky nde, to a sample of more than,3 mllon ndvdual ta returns. Our estmates suggest that more than half of the total redstrbutve effect of the PIT s due to the ta credt for employment and the ta credt for retrement ncome, whle the margnal rate schedule contrbute wth 38.5 per cent. On the contrary, most of the temzed ependtures do not show any szable mpact. In a polcy perspectve ths calls for an agenda that methodcally revse all categores of ta credts (or at most cut out almost all the ta credts for tems of ependture), n order to both smplfy the structure of the PIT and to obtan resources to ncrease the ta credts ehbtng a hgh contrbuton to the overall redstrbutve effect. The paper s structured as follows. Secton 2 provdes essental background nformaton on the Italan PIT, whle secton 3 presents the Pfähler decomposton. A descrpton of

3 our dataset s n Secton 4, whle the man results are commented n Secton 5. Secton 6 brefly concludes the paper. 2. The Personal Income Ta n Italy Lke n many other countres, the Italan Personal Income Ta (Imposta su Reddt delle Persone Fsche, IRPEF) s a progressve ta on ndvdual ncomes, whch makes use of ncreasng margnal ta rates, deductons and ta credts to defne ts degree of progressvty. In ths secton we descrbe the nsttutonal and techncal detals that characterze the 20 PIT, the baselne of the dataset employed n the emprcal analyss. Let be the personal tapayer s gross ncome, 2,..., n. The Italan ta law consders two dfferent knds of deductons: d s the deducton for the man resdence cadastral ncome; 2 d s the sum of certan tems for whch deductons are allowed (lke socal securty contrbutons, almones and donatons). The taable ncome y s evaluated as: 2 2 d d f d d y 2 0 f d d Margnal ta rates defnng the rate schedule S have been vared last tme n 2007, and have remaned unchanged snce then (Table ). y Table : Rate schedule Taable ncome (euro) Ta rate (%) up to above up to above up to above up to above Once defned the gross ta lablty ta credts - ta credts for earned ncome c wthn the household c 2 GT, the law admts three dstnct knds of effectve ; ta credts for dependent ndvduals ; ta credts for ependture tems c 3 - some of whch 2

4 depends on the tapayer gross ncome (net of the deductons for the cadastral rent d ). More specfcally, the ta credt for employees and pensoners s defned as follows: c ( tm k ( tm a) a k m ) w ( tm a) b w k 0 f f f f m k m w k w where t s the lowest margnal ta rate (23 percent); m s equal to 8,000 euro for employees, 7,500 for pensoners younger than 75, 7,750 for pensoners older than 75, 4,800 for the self-employed, and zero for non workng tapayers; k s equal to 5,000 euro for employees and pensoners, whlst t s equal to w for the self-employed; w s equal to 55,000 euro for all tapayers; a s equal to 502 euro for employees, 470 for pensoners younger than 75, 486 for pensoners older than 75, zero for self-employed and non workng tapayers; b, that ranges from 0 to 40 euro n the bandwdth thousands euro, s appled only to employees. As for dependent ndvduals wthn the household, three dfferent ta credts are H allowed: ta credt for dependent chldren c 2 S, for dependent spouse O and for other dependent household components c 2 c 2 2 2H 2S 2O s then c c c c. In partcular, c 2H c 0 2Hp q ( n ) e q ( n ) e f f 0 c 2,. The overall value for q ( n ) e q ( n ) e where n s the number of dependent chldren, e s equal to 5,000 euro and q s equal to 95,000; Hp c 2 s the correspondng potental ta credt: t s 800 and 900 euro f the dependent chld s younger and older than 3 years, respectvely, and the dependent chldren wthn the households are 3 or less; these correspondng potental ta credts are 200 euro hgher whenever the dependent chldren wthn the households are more than 3. Moreover, 3

5 2H whenever 0 c and the dependent chldren wthn the households are more than 3 the ta law admts a further ta credt equals to,200 euro. The ta credts for dependent chldren have to be splt between parents whenever both of them have a postve gross ncome. Fnally, c 2S c c c 0 2Sp 2Sp 2Sp u k q e w k f f f f k k w k q e w k q e and c 2O c 0 2Op q e q e f f q e q e where u s equal to 0 euro, Op c 2 s equal to 750 euro and Sp c 2 s equal to 800 euro f k and 690 f k. Ta law also admts a ta credt for tenants; t s 300 euros f 5, 494 ; 50 euros f 5,494 30,987 ; 992 euros f 5, 494 and the tapayers are younger than 30 years old. Ependture tems for whch ta credts 3 c are allowed can be classfed n three groups accordng to the percentage of the epense the ta law admts as a ta credt. There are epenses that allow a ta credt of 9 percent, 36 percent and 55 percent. The 9 percent ta credts are abundant (9 dfferent tems, such as epenses for health care and mortgage nterests, but also for veternary epenses); 36 percent ta credt are allowed for home restructurng, whlst 55 percent ta credt are allowed for nterventons for energy savng. All together ta law admts 30 dfferent ta credt for ependture tems, whch makes the PIT structure qute complcated. The net ta lablty T s then evaluated as: 4

6 2 3 2 c c f GT c c 2 f GT c c GT c c T 0 c In what fallows we do not consder regonal and muncpal surcharges, and then we evaluate the net ncome as z T Assessng the PIT redstrbutve effect: The Pfähler methodology As the PIT makes use of ncreasng margnal ta rates, deductons and ta credts to defne ts degree of progressvty, t s mportant to understand n a polcy perspectve whch of these three dfferent tools contrbutes more to defne the PIT redstrbutve effect. To ths end, we use the methodology developed by Pfähler (990). Let G X be the Gn coeffcent of pre-ta ncome X, and C T and C Z the concentraton coeffcents of net taes T and net ncome Z X T, respectvely. As t s well known, the Reynolds-Smolensky nde can be evaluated as: G X C Z RE R APK where RE G X GZ and RAPK GZ CZ. The Kakwan K nde s evaluated as: K C T G X and K are lnked by the average ta rate t n n T as follows: t K t Accordng to Pfähler (990) and Lambert (200), t s possble to decompose the nde n the contrbutons due to deductons, the rate schedule and the ta credts. Here we follow the methodology summarsed by Urban (2006), and we etend t n order to consder the mpact of each of the dozens of ta credts characterzng the Italan ta code. 5

7 As dscussed n the prevous secton, the Italan ta law admts two dfferent knds of deductons. The share of and K due to the deducton for the man resdence cadastral ncome can be evaluated as follows: where n d D n and K C D GX. G C C G D X X D D The share of and K due to deductons for socal securty contrbutons, almones and donatons can be evaluated as follows: where n 2 d D2 n and K C D2 GX. G C C G D2 X X D2 D2 The share of and K due to the margnal ta rate schedule can be evaluated as follows: n RATE C C C C Y Y GT GT RATE where n and K CGT CY. Fnally, the share of and K due to each ta credt j j, 2,..., J can be evaluated as follows: where n Cj GT j C G C G X Cj j c Cj j n and K C Cj GX. X j Cj X X X Y 6

8 Hence, the overall can be decomposed as: D D2 J j Cj j RATE n y J where n and j. j As s well known, n order for the Pfähler decomposton to work, tapayer s total deductons cannot be greater than her gross ncome; or, smlarly, tapayer s total ta credts cannot be greater that her gross ta lablty. These consderatons mpose a choce about the order wth whch to apply deductons or ta credts, for the tapayers wth null net ta lablty (very few n our dataset, about 0 percent of ndvduals, aganst more than one fourth consderng the whole Italan tapayers, see net secton). Let us take an eample. Consder a tapayer who gets GT, 000 euro and 2 c 800 euro, 300 c euro and c euro. In ths case the tapayer s net ta lablty s zero. Total ta credts equal,300 euro, so that she cannot takes advantage of 300 euro. Implctly, researchers have to choose an orderng wth whch ta credts are subtracted from the gross ta lablty. In our eample, f 2 3 c c c the tapayer can fully take advantage from c partally take advantage from c 2 and cannot take advantage from 2 3 f c c c the tapayer can fully take advantage from 3 and she can only partally take advantage from c 2 3 apply c c c, and among the, she can 3 c. Conversely, c and c 2,. In what follows we decded to 3 c famly, we order them accordng to the 2 3 orderng shown n the ta return. Results applyng c c c are n any 7

9 case qute smlar, due to the low share of tapayers wth null net ta lablty wthn our dataset. 4. The data We use data provded by the Centr d Assstenza Fscale (lterally, Ta Assstance Centres, CAF) of the Assocazone Crstana Lavorator Italan (lterally Italan Assocaton of Chrstan Workers, ACLI). The CAF-ACLI asssts more than one mllon tapayers and collects all nformaton requred n fllng n the ta return. We have full nformaton on,370,982 ta returns on the 20 fscal year. We decded to eclude from the analyss few tapayers (56,05) who are not employees or pensoners. Our fnal sample ncludes,34,877 ndvduals. To characterze our sample we frst consder the most mportant redstrbutve ndces (Table 2, Column CAF ). The Gn coeffcent of the gross ncome dstrbuton s 0.39, whlst that of the net ncome dstrbuton s As a consequence, RE Concentraton coeffcent for net ncome s , whlst that of net ta lablty s As a consequence, and K The average ta rate s 8.50 percent. The correspondng fgures for the whole Italan PIT, estmated by means of a mcrosmulaton model (see Pellegrno et al., 20) are shown n Table 2, Column Mcrosmulaton model. The Gn coeffcent of the gross ncome dstrbuton s , whlst that of the net ncome dstrbuton s As a consequence, RE Concentraton coeffcent for net ncome s , whlst that of net ta lablty s As a consequence, and K The average ta rate s 8.70 percent. Despte beng very large (ncludng about 3.2 percent of all Italan ta returns), our dataset s then not a representatve sample of the Italan gross ncome dstrbuton for employees and pensoners: t overestmates employees and tapayers belongng to the mddle class and underestmates poorest tapayers wth respect the correspondng dstrbuton on the same fscal year made avalable by the Department of Fnance of the Mnstry of Economy and Fnance (203) (Fgure ). Ths does not affect our results, A new methodology has been recently proposed by Onruba et al. (204) to overcome the choce of the orderng wth whch each part of the Pfähler-Lambert decomposton can be evaluated. 8

10 snce the same methodology appled by means of mcrosmulaton models on the whole ncome dstrbuton reaches almost the same conclusons. Inde Table 2: Redstrbutve ndees CAF Dataset Mcrosmulaton model Gn coeffcent for the gross ncome Gn coeffcent for the net ncome Concentraton coeffcent for the net ncome Gn coeffcent for the net ta lablty Concentraton coeffcent for the net ta lablty Redstrbutve effect Reynolds-Smolensky nde Kakwan nde Atknson-Plotnk-Kakwan nde Average ta rate (%) Source: Own elaboratons based on CAF (202) and Bank of Italy (202). Fgure : Frequency densty functon 22 Frequency densty functon (%) Gross ncome (thousand euro) Mnstry CAF 5. Results We apply the Pfähler decomposton focusng on the mpact on of each deducton and ta credt. Let us frst concentrate on the dstrbuton across tapayers of the allowances. For each specfc tem, Table 3 shows the Gn coeffcent, the 9

11 concentraton coeffcent, the share of tapayers wth postve deductons and ta credts and ther average values. Few consderatons emerge. Frst, despte the large number of ependture tems for whch ta credts are allowed, very few are used by a large number of tapayers; these nclude for nstance health epenses, used as ta credt by more than 60 per cent of ndvduals. However, despte the large number of tapayers eplotng these ta credts, ther concentraton s pretty much hgh, and both average epenses and the number of ndvduals are ncreasng wth ncome. Second, some ta credts are used by very few ndvduals, so that they are hghly concentrated. A clear eample are the epenses for gude dogs for blnd ndvduals. Average epenses are however pretty much hgh, suggestng that these can represent a large share of tapayers ncome. Thrd, ta credts for employment status and dependent ndvduals wthn the household lnearly decrease wth ncome, showng the most far sde of the whole system of ta credts and deductons. Ths s clearly confrmed by the Pfähler decomposton showed n Table 4. More than half of the total redstrbutve effect of the PIT s due to the ta credt for employment and the ta credt for retrement ncome, whle the margnal rate schedule contrbute wth 38.5 per cent. On the contrary, most of the temzed ependtures do not show any szable mpact, and among them there are also ta credts showng a negatve contrbuton to the overall redstrbutve effect. Of course, one can argue that these ta credts are not thought for redstrbutve purposes. However, even from the pont of vew of effcency they do not perform very well, snce they do not seem to provde enough ncentves to demand, both n the case of the home mortgages, and n the case of lfe nsurances (Jappell e Pstaferr, 2003; 2007). However, they certanly contrbute to make the ta structure more complcated, and more subject to pressures from varous lobby groups. 6. Concludng remarks In ths paper we analyze the Italan PIT n the lght of the dfferent methodologes through whch the government can nfluence the degree of progressvty of the ta, hence ts redstrbutve role. 0

12 Deductons and ta credts Table 3: Indees for deductons and ta credts Gn coeffcent Concentraton coeffcent Share of tapayers wth postve tem (%) Average value for tapayers wth postve tem (euro) Deducton for the man resdence cadastral ncome Deductons for socal securty contrbutons, almones and donatons Total deductons Ta credt for employment ,04 Ta credt for work assmlable to employment Ta credt for retrement ncome ,43 Ta credt for dependent spouse Ta credt for dependent chldren Further ta credt for dependent chldren Ta credt for other dependent ndvduals Ta credt for health epenses Ta credt for non dependent ndvduals Ta credt for health epenses for dsable ndvduals Ta credt for vehcles of dsable ndvduals Ta credts for gude dogs Ta credts for health ependtures of prevous years ,230 Ta credts for mortage nterests on the man resdence Ta credts for mortage nterests on other dwellngs Ta credt for other mortage nterests Ta credt for the constructon of the man resdence Ta credt for agrcultural loans Ta credt for lfe nsurance Ta credt for educaton epenses Ta credt for funeral epenses Ta credt for caregvers Ta credt for sport actvtes of chldren Ta credt for real estate brokerage Ta credt for rents of unversty students Ta credt for other epenses Ta credt for home restructurng Ta credt for nterventons for energy savngs Ta credt for home rent Other ta credts Total ta credts ,682 Source: Own elaboratons based on CAF dataset.

13 Table 4: The Pfähler decomposton The PIT structure % of Deducton for the man resdence cadastral ncome.38 Deductons for socal securty contrbutons, almones and donatons Total deductons. Rate schedule Ta credt for employment Ta credt for work assmlable to employment 0. Ta credt for retrement ncome Ta credt for dependent spouse.45 Ta credt for dependent chldren 2.87 Further ta credt for dependent chldren Ta credt for other dependent ndvduals 0.4 Ta credt for health epenses 0.65 Ta credt for non dependent ndvduals 0.00 Ta credt for health epenses for dsable ndvduals 0.0 Ta credt for vehcles of dsable ndvduals 0.04 Ta credts for gude dogs 0.00 Ta credts for health ependtures of prevous years 0.00 Ta credts for mortage nterests on the man resdence 0.9 Ta credts for mortage nterests on other dwellngs 0.00 Ta credt for other mortage nterests 0.00 Ta credt for the constructon of the man resdence 0.00 Ta credt for agrcultural loans 0.00 Ta credt for lfe nsurance Ta credt for educaton epenses -0.3 Ta credt for funeral epenses 0.07 Ta credt for caregvers 0.09 Ta credt for sport actvtes of chldren Ta credt for real estate brokerage 0.00 Ta credt for rents of unversty students Ta credt for other epenses -0.0 Ta credt for home restructurng -.09 Ta credt for nterventons for energy savngs Ta credt for home rent 0.49 Other ta credts -0.0 Total ta credts Source: Own elaboratons on CAF dataset. 2

14 We focus n partcular on three tools: margnal ta rates, deductons and ta credts. We then apply the standard Pfähler decomposton, estmatng the contrbuton of these three tools to the Reynolds-Smolensky nde, to a sample of more than,3 mllon ndvdual ta returns provded by the CAF-ACLI. Our estmates suggest that more than half of the total redstrbutve effect of the PIT s due to the ta credt for employment and the ta credt for retrement ncome, whle the margnal rate schedule contrbute wth 38.5 per cent. On the contrary, most of the temzed ependtures do not show any szable mpact on redstrbuton, and some of them negatvely contrbute to the overall redstrbutve effect. In a polcy perspectve, ths calls for an agenda that methodcally revse all categores of ta credts, n order to both smplfy the structure of the PIT and to obtan resources to reduce the ta wedge wthout affectng the degree of redstrbuton. References Jappell T. and Pstaferr L. (2003), Ta ncentves and the demand for lfe nsurance: evdence from Italy, Journal of Publc Economcs, 87(7-8), Jappell T. and Pstaferr L. (2007), Do people respond to ta ncentves? An analyss of the Italan reform of the deductblty of home mortgage nterests, European Economc Revew, 5(2), Lambert P. J., (200), The dstrbuton and redstrbuton of ncome, Thrd edton, Manchester Unversty Press, Manchester and New York. Mnstry of Economy and Fnance, Department of Fnance (203), Statstcal Reports. Padovano F. and G. Turat (202), Redstrbuton through a Leaky Bucket. What eplans the Leakages?, Workng Paper , Condorcet Center for Poltcal Economy. Pellegrno, S., Pacenza, M. and Turat G. (20), Developng a Statc Mcrosmulaton Model for the Analyss of Housng Taaton n Italy, The Internatonal Journal of Mcrosmulaton, Vol. 4(2), pp

15 Onruba J., Pcos-Sánchez F., Del Carmen Rodado M. (204), Rethnkng the Pfähler Lambert decomposton to analyse real-world personal ncome taes, Internatonal Ta and Publc Fnance, onlne frst. Pfähler W. (990), Redstrbutve Effect of Income Taaton: Decomposng Ta Base and Ta Rates Effects, Bulletn of Economc Research, 42 (2), Urban I. (2006), Progressvty of personal ncome ta n Croata: decomposton of ta base and rate effects, Fnancal Theory and Practce, Insttute of Publc Fnance, Vol. 30(3), Verbst G. and Fgar F. (203), The redstrbutve effect and progressvty of taes revsted: An Internatonal Comparson across the European Unon, AIAS, Gn Dscusson Paper 88. Wagstaff, A., E. van Doorslaer, H. van der Burg, S. Calonge, T. Chrstansen, G. Cton, U.-G. Gerdtham, M. Gerfn, L. Gross, U. Häknnen, J. John, P. Johnson, J. Klavus, C. Lachaud, J. Laurdsen, R. E. Leu, B. Nolan, E. Peran, C. Propper, F. Puffer, L. Rocha, M. Rodrı guez, M. Schellhorn, G. Sundberg and O. Wnkelhake, Redstrbutve Effect, Progressvty and Dfferental Ta Treatment: Personal Income Taes n Twelve OECD Countres, Journal of Publc Economcs, 72(), 73 98,