Growth and Volatility of Tax Revenues in Latin America

Transcription

1 Growh and Volaly of Tax Revenues n Lan Amerca Hans Frcke Bernd Süssmuh CESIFO WORKING PAPER NO CATEGORY 1: PUBLIC FINANCE JANUARY 2011 An elecronc verson of he paper may be downloaded from he SSRN webse: from he RePEc webse: from he CESfo webse: Twww.CESfo-group.org/wpT

2 CESfo Workng Paper No Growh and Volaly of Tax Revenues n Lan Amerca Absrac Agans he background of a noorously hgh macroeconomc nsably and he need o rase ax revenues o mee he demands of publc spendng, hs paper analyzes he radeoff beween growh and volaly of ax revenues n Lan Amerca. We use a wo-sep Engle- Granger-ype model o esmae shor-run and long-run elasces, accounng for saedependen asymmerc reacons of shor-run elasces over he busness cycle. Due o s dependence on commodes exploaon Lan Amerca s n general suscepble o he boombus cycles of s naural rches. Conrollng for he composon of revenue sources and oher dosyncrases of Lan Amercan economes, we fnd revenues above (below) s long-run equlbrum o reac sronger (weaker) o busness cycle dynamcs. Ths ax revenue channel represens an ndrec argumen for couner-cyclcal dscreonary fscal polcy n he regon. Our dealed elascy esmaes can gve some orenaon on how o reach necessary hgher ax levels whou creang dsncenves and neques hrough busness cycle nsables on he way o develop an adequae nernal ax sysem. JEL-Code: H200, O100, E600. Keywords: ax polcy, developng counres, regme-dependen elascy. Hans Frcke Economc Commsson for Lan Amerca and he Carbbean (ECLAC) / Unversy of Erlangen-Nuremberg hans.frcke@googlemal.com Bernd Süssmuh Insue for Emprcal Research n Economcs Unversy of Lepzg Grmmasche Srasse 12 Germany Lepzg suessmuh@wfa.un-lepzg.de Hans Frcke acknowledges he hospaly of he Economc Developmen Deparmen a ECLAC (Economc Commsson for Lan Amerca and he Carbbean), Sanago de Chle, where hs projec was naed. In parcular, hank s due o Juan Pablo Jménez, Andrea Podesá, and Leandro Cabello for her encouragemen and asssance n he projec. We also hank Marco Sunder for hs help n vsualzng some of our resuls. The usual dsclamer apples.

3 1. Inroducon In general, collecng axes s jusfed by he generaon of revenue o fnance publc goods and servces lke educaon, healh and oher socal programs. Provdng hese prerequses for economc performance s crucal, especally n developng economes, o foser growh and o reduce nequaly and povery. Gven a noorously hgh macroeconomc nsably of Lan Amercan economes (Caão 2007) and resulan capal marke consrans ax revenues need o be boh sable and growng n order o mee hese prerequses. Ths need s all he more obvous gven he fac ha Lan Amercan economes generally are dependen on he shackles of commodes exploaon, whch provdes he lvelhoods of her czens bu leaves her economes perennally suscepble o boom-bus cycles and currency flucuaons. As sae governmens n he U.S. are also consraned n her exernal fnancng and habually end o suffer from cyclcal budge conracons, he vas majory of he exsng leraure on ax revenue growh and volaly s concerned wh U.S. federal saes. I daes back o he semnal sudy by Groves and Kahn (1952). Early sudes ha followed (e.g., Wlford 1965, Legler and Shapro 1968) analyzed sae and local ax revenue, condonng revenues on ncome usng sandard OLS and no dsngushng beween he long and shor run. By he early 1970s, Wllams e al. (1973) demonsraed ha wo axes can follow he same growh rend whle experencng a dsnc varably around. Ther fndngs sugges ha a sngle sasc for revenue elascy can no be used o analyze growh and varably a he same me and ha a possble rade-off beween growh and sably exss. The succeedng sudes by Whe (1983) and Fox and Campbell (1984), herefore, consdered dfferen axes and ax srucures, confrmng hs rade-off and fndng personal ncome ax (PIT) and corporae ncome ax (CIT) o be he fases growng bu also he mos unsable axes. Whle, for example, Whe (1983) resrced hs analyss o one sae, Dye and McGure (1991) appled Whe s mehodology o all federal saes. Sobel and Holcombe (1996) furher mproved hs mehodology by accounng for problems of resdual varably, seral correlaon, and non-saonary of revenue seres. The laes developmen n hs row s Bruce e al. (2006) who combne he srucured approach of Fox and Campbell (1984) wh he refned mehodology proposed by Sobel and Holcombe (1996). For our esmaes, we wll wdely adhere o her approach. To he bes of our knowledge, hese laes echnques by now have besdes for U.S. federal saes only been used o sudy a few oher counres; see Wolswjk (2009) for he Neherlands and Acquaah and Gelard (2008) for Brsh Colomban revenues. Alhough some ax revenue elascy esmaes for Lan Amercan economes can be found n he leraure (usually nended o calculae cyclcally adjused bal- 1

4 ances), he evdence remans scaered across he dfferen naons and mosly sems from researchers locaed n governmenal organzaons n he regon (e.g., Basso 2006, Cárdenas e al. 2008, De Mello and Moccero 2006, Rncón e al. 2003, Salazar and Prada 2003, Schenone and De la Torre 2005, Tapa 2003). Wh few excepons (Anelo 2003, Fuenes and Tobar 2003) hs leraure s focused on a long-run relaonshp,.e., he growh aspec of ax revenues. Neher s he ssue of growh and sably of revenues analyzed jonly nor s a poenal rade-off examned. The presen sudy conrbues o he leraure by applyng he laes echnques o esmae shor-run and long-run elasces of ax revenues n Lan Amerca, accounng for asymmerc reacons of shor-run elasces over he busness cycle. Consderng he composon of revenues of personal and corporae ncome ax (PIT, CIT), value added ax (VAT), socal secury conrbuons, and revenues from commodes exploaon, we fnd revenues above (below) s long-run equlbrum o reac sronger (weaker) o busness cycle dynamcs. Our dealed elascy esmaes can gve some orenaon on how o reach necessary hgher ax levels whou creang dsncenves and neques hrough busness cycle nsables on he way o develop an adequae nernal ax sysem. The remander of he paper s organzed as follows. Secon 2 descrbes he recen developmen of ax collecon n Lan Amercan economes. Secon 3 gves an oulne of he daa and mehodology we use. In Secon 4 we presen and dscuss our fndngs. Fnally, Secon 5 concludes. 2. Some recen developmen n ax collecon across Lan Amerca Throughou Lan Amerca he ax burden has been relavely low. In 2008, cenral governmens on average colleced only 17.9% of gross domesc produc (GDP). 1 Even hough hs amoun s a consderable ncrease over he 12.5% colleced n 1990, remans well under revenues colleced n developed counres: The OECD repors a correspondng 35.2% on average n Smlarly, Tanz and Zee (2000) and Bahl and Brd (2006) documen a level of axaon n ndusralzed counres by he end of he 20 h cenury ha was abou wce ha n developng counres. Hsorcally, hs raher small amoun of ax revenues proved nsuffcen o mee he demands of publc spendng n he regon. Only n 2006 and 2007 dd revenues exceed spendng, makng 1 As can be seen from he fgures repored n Tanz and Zee (2000, p. 303), hs average level of ax revenue for all developng counres n he Wesern Hemsphere aken ogeher has been farly sable, lyng beween %, for approxmaely he second half of he 1980s and 1990s, respecvely. 2 Whn he regon he ax burden s raher heerogeneous. Whle Brazl and Argenna collec more han 30% of GDP, Mexco and Ha do no reach 10% n For he vas majory, however, he respecve share les beween 10 and 20%. 2

5 look lke governmens are now more fscally conservave and suggesng srongly ha on he whole hs s a good hng for her people; see Bahl and Brd (2008, p. 295). Ths perod, however, came o a sudden end n 2008 when he neronal fnancal crss began o h he regon. Besdes comparably low levels of axaon, macroeconomc volaly n Lan Amerca has been hgher han n developed counres and n emergng economes n Asa and Eas Europe (Caão 2007). Flucuaons n macroeconomc acvy have caused major losses n ax revenues. The recen crss has demonsraed how vulnerable hese revenues are o conracons n economc acvy (Fgure 1). As shown n Fgure 1, ax revenues sharply fell a he end of 2008 and durng For example, n he second quarer of 2009 Chlean revenues decreased by as much as 34.6% n comparson o he prevous year. Whle hs drop was no as pronounced as n oher counres, was sll severe. Fgure 1: Evoluon of Tax Collecon durng he Recen Crss a) (consan values, varaon /-4) : ECLAC (2008) a) Whou socal secury conrbuons The nerplay beween srucurally low ax burden and emporary buss n ax revenues has forced several governmens n he regon o cu down on publc servces and fall back on exernal fnancng n unfavorable condons. The margn for counercyclcal polces narrowed for some counres (ECLAC 2008, Fanell 2009). A ax srucure wh posve revenue growh s no suffcen o ensure solvency each year. Transory flucuaons can lead o resource shorages even hough ax revenues grow n he long run, renderng accurae year-o-year budge plannng a mos dffcul ask. Among oher facors such as poor ax admnsraon (cf. Bahl and Brd 2008) hs 3

6 makes ax polcy n developng counres n general he ar of he possble raher han he pursu of he opmal (Tanz and Zee 2000, p. 300). Under he presumpon ha he Lan Amercan economes seek o become fully negraed wh he world economy lke counres such as Canada and Ausrala ha were seen as regons of recen selemen a cenury ago and succeeded n nsallng an adequae nernal ax sysem (Bahl and Brd 2008, p. 279) hey wll as Tanz and Zee (2000, p. 320) pu n he long run probably need a hgher ax level, because of he need o pursue a governmen role closer o ha of ndusral counres. Long-run elascy esmaes capure ax revenue growh, as hey measure he relaonshp beween he cumulave developmen of ax revenues and aggregae ncome or he respecve commody prce for he oal perod of observaon. Shor-run elascy esmaes gve an answer o he queson of how revenues respond o he ups and downs of he busness cycle or of commody prces. In hs sense, hey capure he volaly of revenues. However, here mgh be a rade-off, nasmuch as faser growng ax revenue sources mgh reac more srongly o macroeconomc flucuaons and, hus, prove o be less sable. If hs s he case, polcy makers face he problem of fndng a balance beween polcy goals of revenue expanson and mananng revenue sably. Fgure 2: Major Tax Revenue s n Lan Amerca (% of oal ax revenue, 19 counres) 100% 80% 60% 40% 20% 0% Value Aadded Tax Income Tax Socal Secury a) Ohers : own calculaons based on ECLAC (2008) a) Average for socal secury conrbuons does no nclude Ha. 4

7 Our analyss wll focus on major sources of revenues n Lan Amerca: VAT, ncome ax (IT), socal secury conrbuons, and revenue from commodes. In 2008, IT, VAT, and socal secury conrbuons make up 76.8% of oal ax revenues. As shown n Fgure 2, VAT s he fases growng ax. IT revenue grows more moderaely over he las wo decades. Socal secury conrbuons have decreased slghly from 16.6% o 14.8% wh a mnor ncrease n he frs half of he 1990s. I s noeworhy, however, ha here are consderable dfferences n he composon of hese revenue sources across counres n he regon. For example, n Bolva VAT accouns for 46.2% of oal ax revenues n 2008, whle n Panama amouns o only 13.8%. Gong furher no deal, we dfferenae () personal ncome ax (PIT) from corporae ncome ax (CIT) and () exernal VAT from domesc VAT. Fgure 3 and Fgure 4 show he sample varaon of PIT and CIT as well as he one of he wo VAT componens for he Peruvan economy durng he fnancal crss. Obvously, PIT revenue growh slowed gradually unl revenue fell slghly n he second quarer of 2009 and only recovered slowly aferwards. Ups and downs of CIT revenues are more pronounced and somewha lag he cycle. In fac, CIT revenue growh acually acceleraed a he end of 2008 before fallng n Possble explanaons for he dfferences nclude he fac ha wages are usually more sable n he shor-run due o labor marke frcons, whle companes face a profound negave mpac on profs durng a rough. In he case of domesc and exernal VAT he dfference s even more pronounced. Whle domesc VAT revenue s only slghly affeced by he drop n economc acvy, exernal VAT revenue grew much faser durng 2008 bu fell by as much as 34.9% n he hrd quarer of Impors n he regon manly conss of durable (and luxury) goods, whle domesc producon sasfes basc consumer needs, whch are generally less elasc. Therefore, a conracon of naonal ncome wll lkely be refleced n a decreased demand for mpored goods and, consequenly, a drop n revenue from axes leved on mpors. In he Peruvan case, domesc VAT (PIT) seems o reac less o changes n economc acvy han exernal VAT (CIT). As argued above, a peculary of mos Lan Amercan economes les n her dependence on commodes exploaon. In Venezuela, Bolva, Chle, Colomba, Ecuador, and Mexco 3 non-renewable commodes and naural resources accoun for over 20% of expors. Consderng expors of renewable commodes, he ls also ncludes Argenna as well as several oher counres n Cenral Amerca. In hese economes, ax revenues and non-ax revenues from hese secors generae a szable 3 For smplcy reasons, we wll refer o he Plurnaonal Sae of Bolva as Bolva and o he Bolvaran Republc of Venezuela as Venezuela. 5

8 share of oal fscal revenue. Durng 2008, shares of revenue from commodes n oal revenue ranged from 11.3% n Peru o 49.6% n Venezuela. Fgure 3: Varaon of PIT and CIT n Peru (/-4, consan values) 40% 20% 0% 20% 40% I II III IV I II III IV Personal Income Tax Corporae Income Tax : own calculaons Fgure 4: Varaon of Domesc and Exernal VAT n Peru (/-4, consan values) 40% 20% 0% 20% 40% I II III IV I II III IV Domesc VAT Exernal VAT : own calculaons As commody secors manly produce for expors, revenues are crucally dependen on he prce of he commodes n queson. Moreover, ax and non-ax revenue are usually lnked o he performance of he secor and as a resul depend on he commody s prce. 4 Jménez and Tromben (2006) fnd ha revenues from commodes, n general, show a hgher sandard devaon han non-commody revenue seres. For hs reason, we wll analyze revenue from commodes and non-commody revenue separaely: he frs wh respec o he parcular commody prce and he laer 4 Jménez and Tromben (2006) gve an overvew of ax regmes for non-renewables. 6

9 wh respec o GDP as a measure for aggregae ncome. The exsng leraure s focused on measurng he long-run responses of ax revenues, whereas he shor-run has been wdely negleced. Mos of he esmaes sem from cyclcal adjused balances (CAB) and hus are no esmaed o explcly analyze growh or volaly. Exsng sudes so far also have concenraed eher solely on one counry or, f hey consdered several counres, on oal ax revenues only. The presen sudy wll be he frs o esmae long-run and shor-run elasces of he mos mporan ax revenue sources for Argenna, Bolva, Brazl, Chle, Colomba, he Domncan Republc, Ecuador, Mexco, Peru, Uruguay, and Venezuela. 3. Daa and Mehodology 3.1 Daa In prncple here are wo opons when quanfyng he responsveness of axaon o changes n naonal ncome: eher o use ax base (cf. Dye and McGure 1991, Sobel and Holcombe 1996, Nchols and Tosun 2008) or ax revenue (cf. Bruce e al. 2006, Acquaah and Gelard 2008, Felx 2008) daa. If he relaonshp beween he wo s proporonal hen ax base and ax revenue elasces would be equvalen. However, due o progressvy of he ax schedule, ax exempons, or ax evason, here usually s no such equvalence. In pracce, boh approaches have advanages and dsadvanages. Esmaes of ax revenue elasces can be based f ax code changes alerng he defnon of ax base or ax raes are no conrolled for n he emprcal model s specfcaon. 5 Tax base based esmaes do no suffer from hs bas. Tax base daa, however, s n general no readly avalable and has o be hypohecally consruced wh he help of proxes (Dye and McGure 1991, Sobel and Holcombe 1996), whch do no necessarly concde wh he legal ax base defnon (Dye 2004). In fac, we know only abou one sudy, Nchols and Tosun (2008), where he auhors use exac ax base daa by analyzng gamblng axes, for whch gross casno revenues represen he acual ax base. Due o daa lmaons and for he sake of comparably, we follow he recen sudes of Bruce e al. (2006), Acquaah and Gelard (2008), and Felx (2008) by relyng on ax revenue daa and by conrollng for polcy changes lke ax rae or ax base changes (Appendx B) n our emprcal model. As no for all Lan Amercan economes dsaggregae daa on revenues s avalable, we have o lm our analyss n hese 5 In hs conex, he leraure dsngushes beween ax buoyancy and ax elascy. Tax buoyancy measures he oal response of ax revenues ncludng dscreonary polcy, ha s, he response f ax code changes are no conrolled for n he emprcal model. Tax elascy, n conras, solaes bul-n responsveness o changes n naonal ncome. 7

10 cases o aggregae daa on IT and VAT. Our seres are of quarerly frequency and range for mos of he seres from he frs quarer of 1990 o he frs quarer of To deflae he revenue seres we employ a correspondng GDP deflaor. If he laer s no avalable, we resor o he respecve CPI. To express commody prces n real erms, we use he U.S. Producer Prce Index (PPI). 6 All seres were deseasonalzed applyng he sandard ARIMA X-12 mehod. Deal along wh daa sources s gven n Appendx A. Wh regard o revenues from commodes we consder ax as well as non-ax revenues. Non-ax revenues from commodes, such as ransfers, are usually lnked o secor performance. The laer raher depends on commody prces han on economc acvy. As a resul, we expec non-ax revenues from commodes o reac o changes n he commody prce. In he case of Peru and Argenna, revenue from commodes s, n conras o he oher economes n he regon, no prmarly rased from a sngle good. We, herefore, consruc Peruvan and Argennean prce ndces as weghed averages of prces for commodes wh a subsanal share n expors. 7 Argennean axes on basc goods expors are consdered commody revenues raher han axes. As each produc s axed wh a dfferen rae, an approxmaon of he share n oal expor ax revenue s used as wegh nsead of he share n expor volume. 3.2 Mehodology Followng he mehod proposed n Sobel and Holcombe (1996), we esmae long-run elasces relyng on dynamc OLS (DOLS) echnques (Sock and Wason 1993). A sandard error correcon model (ECM) s used o esmae shor-run elasces (Engle and Granger 1987). Addonally, we allow shor-run elasces o vary for dfferen saes of economc condons. Followng Bruce e al. (2006), sae-dependen asymmery s aken no accoun accordng o he poson of acual revenue o respecve long-run value. Tax revenue measures and cyclcal varables are analyzed n naural log expresson. To conrol for changes n legslaon, ax raes are ncluded n he emprcal models as ndependens. They capure varaons n he schedule and have been consdered for IT (PIT, CIT) and VAT (domesc VAT, exernal VAT) seres. Oher changes n he ax code lke changes n he defnon of legal ax bases are conrolled 6 Noe, he sandard measure o deflae commody prces s he Manufacures Un Value Index (MUV) provded by he IMF or he Worldbank (e.g., Cashn e al. 2000). However, for he perod under consderaon has no been consruced n quarerly or monhly form. Labys (2006) suggess he PPI provded by he U.S. Deparmen of Commerce as an alernave due o he fac ha shows a hgh correlaon as well as srucural and behavoral smlares wh he Worldbank s MUV. 7 For Peru, he prce ndex s compued on he base of prces for copper, gold, znc, and crude peroleum. Each prce s weghed accordng o s share n he expor volume of hese four goods. For Argenna, he prce ndex we used s based on he prces of beef, soybeans, soybean ol, and crude peroleum. 8

11 for by use of dummy varables. As we presume he exsence of a long-run equlbrum n order o quanfy he long-run relaonshp beween revenue and macroeconomc condons, he wo varables mus be conegraed and rendng ogeher, n he sense of followng a common sochasc rend. Ths mples ha for he wo non-saonary seres a lnear combnaon exss ha s I(0). To assess hese echncal condons, we conduc sandard ADF and PP ess for he seres n levels, frs dfferences, and for he resduals from regressng revenue on cyclcal varable, respecvely. Several of he revenue seres are found o be rend saonary, whle he vas majory of cyclcal varables s found o be saonary n frs dfferences (dealed ADF and PP es resuls are avalable on reques). 8 Thus, we decded o follow a wo-rack sraegy. Frs, we nerpre hese es resuls as beng he produc of small sample bas and noorously low power of un roo ess and rea he rend saonary seres as sharng a sochasc rend wh he busness cycle. Ths par of our sraegy corresponds for example o he approach followed by Wolswjk (2009). Addonally, we also consdered anoher sraegy for seres esed o be rend saonary by esmang a deermnsc rend nsead of a long-run mulpler n he frs sep of our analyss (Whe 1983). I measures, how much ax revenue grows each perod whou consderng a relaon o a parcular macroeconomc base. In a second sep, symmerc shor-run elasces are esmaed usng sandard OLS n dfferences whou ncludng an error correcon erm. In he fnal sep, asymmerc reacons are aken no accoun by allowng shor-run elasces o vary accordng o he poson of curren revenue relave o deermnsc rend. Usng Whe s approach, we esmae 35 long-run and 19 shor-run elasces across counres and axes as beng sascally dfferen from zero. Do hese esmaes subsanally dffer from he resuls we would have obaned by reang he rend saonary esed seres as sharng a sochasc rend wh he busness cycle (cf. Wolswjk 2009) and applyng he mehods proposed by Bruce e al. (2006) o esmae elasces? Fgure 5 gves he answer by plong hese esmaes on he ordnae agans he ones ha we oban by adherng o he mehod of Bruce e al. (2006) on he abscssa. If we cener he elascy esmaes obaned from he respecve mehod and regress hem on each oher, we fnd he correlaon coeffcen for he long-run elascy esmaes o amoun o 0.53, he one for shor-run elasces o 0.94, and he one for all elasces o Throughou, hese correlaons are sgnfcan a all convenonal levels of sgnfcance. Ths fndng leads us o absrac n he followng from dscussng esmaes resulan from he mehod proposed by Whe (1983). Dealed 8 Due o he fac ha we are afer peculares and naonal dosyncrases n he relaonshp beween revenues and cyclcal varables and, hence, do no rea he seres n an unbalanced panel framework, hardly makes sense o resor o more powerful panel un roo ess such as Maddala and Wu (1999). 9

12 esmaes based on Whe s approach for all rend saonary esed seres are avalable on reques from he auhors. Therefore, n wha follows, we repor resuls obaned from he mehod of Bruce e al. (2006) hroughou. Fgure 5: Elascy esmaes: Whe s mehod (ordnae) vs. Bruce e al. s mehod (abscssa) (lef scaer: long-run elasces, mddle scaer: shor-run elasces, rgh scaer: all elasces) : Shown esmaes are sascally sgnfcan esmaes for whch p 0.1; pooled over all consdered economes and axes. In fac, Whe-resuls represen quas-elasces measurng a ceran percenage ha respecve revenues grow each perod. In our baselne regressons, he DOLS-model s used o esmae he long-erm elascy β 1 from sngle equaon conegraon relaonshps of he followng form for every economy ln T ( X ) α η j 0 + β1 lny + γ g lny + g + ' (1) g = j = β + j s ' = 0 + β1 ln P + β 2 ln E + ( γ g ln P + g + φh ln E + h ) + X α v, ln R β + (2) g = - j h= -s where T denoes revenues from ax and Y real GDP, respecvely. Covaraes conaned n X are ax raes 9 as well as dummy varables ndcang changes n he ax code for respecve ax (Appendx B). The lag- and lead-operaor,.e., he summaon of frs order dfferences for dfferen forward and backward shfs of Y (as well as of P and E), s employed o adjus for problems of endogeney and auocorrelaon. Lengh j (s) of hs operaor s chosen by means of he Schwarz-Bayesan nformaon creron (BIC), where we allow for a maxmum lengh j = 3 (s = 3) excep for seres of less han 50 observaons, for whch he maxmum lengh s se o one. In equaon (2), R denoes revenues from commodes, P he commody prce n US Dollars, and E he respecve exchange rae. Long-run elasces of revenues from ax are gven by esmaes of β 1 long-run elasces of commody revenues by esmaes of β 1. Errors η and v are assumed o represen..d. normal random shocks. 10

13 Two shor-erm effecs can occur n each perod: Revenues may reac o changes n real GDP (or, n case of commody revenues, o commody prces) and/or may adjus owards her long-erm equlbrum level, based on he assumpon ha a dsequlbrum (ε ) exss a he begnnng of a perod, where ε ( X )' α = ln T β β lny (3) 0 1 ε ' = ln R β 0 β1 ln P β 2 ln E X α. (4) These effecs can be consdered n erms of an error correcon model (ECM) 0 1 ( y y ) + θ ε + ln T = θ + θ (5) ( p p 1 ) + θ2 ( e e 1 ) + θ3ε 1 + ln R = θ + θ (6) 0 1 where (dummy-ype) covarae expressons have been dropped for reasons of noaonal convenence; alhough no shown, hey are ncluded above and n he followng; mnor leers denoe varables n naural log; and represen..d. random varables. Coeffcens θ 1 and θ 1 ndcae nra-perod effecs,.e., shor-erm adjusmen effecs o changes n real GDP and commody prces, respecvely. Thus, hey can be nerpreed as measures of shor-run elasces. A major concern of our sudy s o unravel dfferences beween shor-erm and long-erm effecs of GDP (and commody prces) on revenues. The seleced economerc specfcaon allows a drec comparson of boh effecs. The shor-erm reacon of revenues o ncome (or, n case of commody revenues, o commody prces) s smaller or larger han he long-erm reacon, dependng on wheher he respecve θ 1 s smaller or larger han he respecve β 1. A furher neresng queson s how fas revenues move o her (new) long-run equlbrum, whch may resul due o he changes n real GDP (or, n case of commod- y revenues, n commody prces). Coeffcens θ 2, θ 3 assess he speed of adjusmen of revenues owards her long-erm level,.e., he proporon of dsequlbrum, whch s reduced n each perod. Thus, he larger he absolue value of θ 2 (n case of commody revenues, θ 3 ) s, he faser revenues equlbrae o he new condons and move o her long-erm equlbrum level, respecvely. In equaons (5) and (6) he shor-run elascy of revenues wh respec o changes n real GDP (n commody prces) s he same regardless of wheher revenues are above ( ε > 0 ) or below ( ε < 0 ) her long-erm equlbrum level. A sym- merc reacon s mplcly assumed. To allow he reacon o depend on he parcu- 9 In case of PIT and CIT showng progressve srucure, only maxmum raes are consdered. 11

14 lar sae of he busness cycle, he ECM can be modfed o accoun for possble asymmeres 0 1 ( D y ) + θ ε + λ ( D ε ) υ ln T = θ + θ y + (7) λ ( D p ) + θ e + θ ε + λ ( D ε ) ln R = θ + θ p + λ + υ, (8) where he υ vecors represen..d. random varables, and D denoe dummy varables, whch ndcae he respecve poson of revenues relave o her long-run equlbrum. These dummes wll ake on a zero value f revenues are below her seady sae level and a value of one else. 4. Fndngs and Dscusson Fgure 5 summarzes our long-run and shor-run elascy esmaes of aggregae IT and VAT seres n he regon. The frs row of maps vsualzes our IT elascy esmaes for he long run and shor run, respecvely. The second row of maps does so for VAT elascy esmaes. The sze of bars n he frs column of maps vsualzes he sze of esmaed long-run elasces, whch ypcally s larger han he correspondng elascy esmaes n he shor run when revenues are below her long-run equlbrum level and smaller when above (second column of maps). The laer are represened by a par of bars, where he sze of bars s dencal for symmerc elascy esmaes (eq. 5 and 6) and dffers for sascally sgnfcan asymmerc elascy esmaes (eq. 7 and 8). The smaller he bars are, he less volaly s mpled. The shadng of he bars dsplays he adjusmen speed o he new long-run equlbrum n case of a shor-run devaon from long-run equlbrum as measured by λ 2 and λ 3 n he above specfcaons. The darker he shadng, he faser hs re-adjusmen akes place. For our IT shor-run elasces sx ou of en esmaes are esmaed as clearly asymmerc, whle four ou of en show symmery over he busness cycle. In he case of VAT elascy esmaes, four ou of en are symmerc, wo ou of en show some weak asymmery, and anoher four are characerzed by a clear-cu asymmery. To hghlgh he srengh of our sngle equaon approach, le us consder he resuls counry by counry. The frs hng o noe s ha we canno esmae all four elasces for Uruguay and Ecuador. Thus, we have o absrac from hese economes n he dealed nerpreaon of our resuls, mplyng concree polcy recommendaons. A second remarkable fndng s ha only Mexcan IT revenues reac sronger when below her long-run equlbrum. For all oher sascally sgnfcan asymmerc elascy esmaes he oppose apples, ha s, revenues are found o be more elasc n he shor 12

15 run when above her long-run equlbrum level. Correspondng exac fgures of our elascy esmaes can be found n Table 1. Fgure 5: Long-Run and Shor-Run Elascy Esmaes: IT and VAT Mexco Domncan Republc Mexco Domncan Republc Ecuador Colomba Venezuela Ecuador Colomba Venezuela Peru Bolva Brazl Peru Bolva Brazl Income Tax Long-Run Elascy Chle Paraguay Income Tax Shor-Run Elascy 5 4 Chle Paraguay Uruguay Argenna below above long-run equlbrum Uruguay Argenna 0 Mexco Domncan Republc Mexco Domncan Republc Ecuador Colomba Venezuela Ecuador Colomba Venezuela Peru Bolva Brazl Peru Bolva Brazl Value Added Tax Long-Run Elascy Chle Paraguay Value Added Tax Shor-Run Elascy 5 4 Chle Paraguay Uruguay Argenna below above long-run equlbrum Uruguay Argenna 0 : Own esmaes n absolue values; for dealed fgures see Table 1; for daa sources and deal on me seres see Appendx A. Sze of bars corresponds o sze of elascy esmaes (accordng o scale shown n lower lef corner); shadng of bars for shorrun elascy esmaes reflecs adjusmen speed o long-run equlbrum: he darker he shadng, he faser he adjusmen. Overall, he Mexcan economy has clearly more growh poenal n VAT han n IT revenues. Obvously, revenues from VAT also adjus faser n he shor run. However, 13

16 here s some danger of overheang as Mexcan VAT revenues are more suscepble o conracons when above her long-run equlbrum level. The oppose holds for Mexcan IT revenues n he shor run. Argenna and he Domncan Republc show a hgher growh poenal n IT compared o VAT revenues. Bu hs advanage comes, n he case of he Domncan Republc, a he cos of IT revenues beng relavely more prone o overheang. In he case of Venezuela, neher VAT s srcly preferable o IT, nor he oher way around as he growh poenal s farly hgh and he suscepbly o shor-run flucuaons farly low for boh revenue ypes. Smlarly on he bubble are Colomba and Peru, where he laer s characerzed by an asymmerc elascy over he shor run for boh VAT and IT revenues. The economes of Brazl, Bolva, and Chle face a clear-cu rade-off, n he sense ha ax revenues wh he hgher growh poenal are found o be more volale n he shor and medum run and vce versus. 10 In order o denfy he mos promsng revenue sources more exacly, we also consder he dsaggregae IT componens, PIT and CIT, as well as domesc VAT, exernal VAT, socal secury conrbuons, and revenues from commodes. Ad hoc, dsregardng a poenal rade-off beween growh and (asymmerc) volaly, we fnd ha long-run elasces for CIT revenues ouwegh he ones for PIT n any case (see second column of sub-ables on PIT and CIT revenue elascy esmaes n Table 1). The same apples o VAT componens. Esmaed long-run elasces of exernal VAT revenues hroughou ouwegh he correspondng ones of nernal VAT revenues. For Ecuador he long-run elascy esmae of socal secury conrbuons sands ou and ops he one of any oher consdered revenue seres (see he socal secury par of Table 1). No surprsngly, we clearly fnd he hghes growh poenal for he Chlean economy n s revenues from commodes. Wh he excepon of Argenna, he above assessmen remans unouched f we consder a possble rade-off beween growh poenal and suscepbly of revenues o cyclcal flucuaons. For our esmaes based on dsaggregae revenue componens, we fnd for Argenne exernal VAT revenues o show boh he hghes growh poenal (correspondng long-run elascy esmae s 3.64) bu also he hghes cyclcal volaly (correspondng symmerc long-run elascy esmae s 4.28). Agans he background of he recen Argenne crss of , when severe ros and socal unres were rggered by growng recesson and perssen unemploymen n lae 2001, 11 he relavely hgh suscepbly of exernal VAT revenues o macroeconomc 10 Noe, negave values for shor-run coeffcens esmaes sugges a counercyclcal reacon of revenues (cf. some of he enres n Table 1). Alhough no n accordance wh our nuon, counercyclcal responses are, for example, also esmaed by Bruce e al. (2006) and Nchols and Tosun (2008). 11 In conras, he smlar epsode of polcal nsably n he lae 1980s was rggered by hypernflaon hng he Argenne prvae secor. 14

17 downurns and buss may sugges o exend and raher coun on CIT and/or PIT as prmary and n he shor run more sable source of revenues. Table 1. Long-Run and Shor-Run Revenue Elascy: Dealed Esmaes IT LR Elascy SR Elascy Adjusmen Speed below LR equlbrum above LR equlbrum below LR equlbrum above LR equlbrum Argenna *** ** ** *** *** Bolva *** ** *** ** Brazl *** ** *** *** *** Chle *** *** *** *** Colomba *** *** *** Domncan Rep *** *** *** *** Ecuador *** *** *** Mexco *** *** *** *** *** Peru *** ** *** *** Uruguay *** *** *** Venezuela *** *** *** *** *** PIT Argenna *** * * *** *** Bolva *** *** * *** *** Brazl *** *** *** *** Chle *** * *** *** Peru *** ** ** *** *** CIT Argenna *** * * *** *** Bolva *** * *** *** Brazl *** *** *** *** Chle *** * *** *** *** Domncan Rep *** *** *** *** Peru *** ** *** Uruguay *** *** *** VAT Argenna *** *** *** *** *** Bolva *** * * *** *** Brazl *** *** *** ** ** Chle *** ** ** *** *** Colomba *** *** *** *** Domncan Rep *** *** *** *** Mexco *** *** *** *** Peru *** *** *** *** Uruguay *** *** *** *** *** Venezuela *** ** ** *** ** Domesc VAT Argenna *** *** *** *** *** Bolva *** * *** *** Chle *** ** * *** *** Colomba *** *** *** *** Peru *** *** *** *** Uruguay *** *** *** ***

18 Table 1 (con'ed) Exernal VAT LR Elascy SR Elascy Adjusmen Speed below LR equlbrum above LR equlbrum below LR equlbrum above LR equlbrum Argenna *** *** *** *** *** Bolva *** *** *** *** Chle *** ** *** *** Colomba *** *** *** *** *** Peru *** *** *** *** Uruguay *** *** *** *** *** Socal Secury Argenna *** Brazl *** *** *** *** Chle *** * *** *** *** Ecuador *** * * *** *** Mexco *** *** *** *** *** Peru *** *** *** *** Commodes Rev Argenna *** *** *** *** *** Bolva * *** *** Chle *** *** *** Ecuador *** *** *** *** Mexco *** *** *** *** *** Peru ** *** *** Venezuela *** *** *** *** *** : SR, LR denoe shor-run and long-run, respecvely; *, **, *** denoe sgnfcance a he 10, 5, 1% level. LR elascy esmaes: IT componens: due o mssng nformaon no legslave changes conrolled for Ecuador and Venezuela; VAT componens: due o mssng nformaon no legslave conrols for Venezuela, for Brazl no rae changes ncluded (sae ax) as conrol, no esmaon for Ecuador due o lack of conegraon; Socal secury conrbuons: due o mssng nformaon no legslave conrols for Ecuador, no esmaon for Uruguay due o lack of conegraon; Revenues from commodes: due o mssng nformaon (or oo many changes o conrol for) no legslave conrols for Ecuador, Venezuela (and Argenna); SR elascy esmaes: For above-equlbrum coeffcens he esed hypohess s he jon hypohess of he sum of belowequlbrum coeffcen and dfference equalng zero; shorfall of conrols apples analogously o LR elascy esmaes. Argenne PIT revenues, for example, clearly show a lower growh poenal bu are a he same me less prone o economc flucuaons han are revenues from exernal VAT. The laer, however, adjus abou wce as fas back o he long-run equlbrum pah as do he former. Hence, seen from an ncumben governmen he possbly exsenal busness cycle sably of revenues comes a a cos. For smlar reasons, consderng he Brazlan crss of 1992 ha ended wh he demse of he presden, a focus on revenues from VAT raher ncome axes seems o be he more reasonable as he more cauous sraegy. As noed earler (see foonoe 2), boh, Argenna and Brazl, have n conras o oher Lan Amercan economes nearly caugh up wh he OECD average ax collecon. Today boh governmens col- 16

19 lec axes amounng o abou one hrd of GDP, suggesng a lower wegh on growh compared o volaly aspecs of revenues. 5. Concluson The years before he economc and fnancal crss of 2008 he perennally boom-bus cycle suscepble Lan Amercan economes wnessed for he frs me a shor perod of excess (ax) revenues. Ths perod came o a sudden end n Decreasng revenues led o cus n publc servces and socal programs. They mpled a fallng back on exernal fnancng, a narrowed scope for dscreonary polces, and problems of solvency n some of he counres of he regon. Our sudy by usng laes economerc echnques red o gve some orenaon for wha Tanz and Zee (2000) called he ar of he possble raher han he pursu of he opmal, ha s, for he creaon of growng and a he same me sable ax revenues n he Lan Amercan economes. We fnd ha abou half of he analyzed economes face a clear-cu rade-off beween growh and volaly of revenues. In more han half of he cases, we fnd revenues o reac asymmercally o macroeconomc condons n he shor and medum run: Above her long-run equlbrum level hey reac sronger o economc flucuaons. Below, hey reac weaker. Gven enough scope wh regard o solvency, hs fndng s suggesve for couner-cyclcal dscreonary fscal polcy as a mnor smulus n a phase of above average macro-condons mgh be already que effcen n sablzng revenues, whle a major smulus would be needed n a phase of conracon. References Acquaah, M. and A. M. G. Gelard (2008), The Growh and Sably of Revenues n Brsh Columba, Canada, Journal of Busness and Economc Sudes 14, Anelo, E. (2003), El Balance Fscal Esrucural Cclcamene Ajusado, mmeo, Assenca Técnca Corporacón Andna de Fomeno (CAF), Caracas, Venezuela Bahl, R. W. and R. M. Brd (2008), Tax Polcy n Developng Counres: Lookng Back -- and Forward, Naonal Tax Journal 61, Basso, M. A. (2006), El Balance Esrucural: Meodología y Esmacón para Argenna, mmeo, Asocacón Argenna de Economía Políca, Buenos Ares Bruce, D., Fox, W. F., and M. H. Tule (2006), Tax Base Elasces: A Mul-Sae Analyss of Long-Run and Shor-Run Dynamcs, Souhern Economc Journal 73, Cárdenas, Ó., Venosa-Sanaulàra, D., and M. Gómez (2008), Elascdad Ingreso de los Impuesos Federales en Méxco: Efecos en la Recaudacón Federal Parcpable, El Trmesre Económco 75,

20 Cashn, P., Lang, H., and C. J. McDermo (2000), How Perssen are Shocks o World Commody Prces?, IMF Saff Papers 47, Caão, L. A. V. (2007), Backcasng Lan Amerca, Fnance & Developmen 44, De Mello, L. and D. Moccero (2006), Brazl s Fscal Sance durng : The Effec of Indebedness on Fscal Polcy over he Busness Cycle, OECD Economcs Deparmen Workng Paper No. 485 Dye, R. F. (2004), Sae Revenue Cyclcaly, Naonal Tax Journal 57, Dye, R. F. and T. J. McGure (1991), Growh and Varably of Sae Indvdual Income and General Sales Taxes, Naonal Tax Journal 44, ECLAC (2008), Economc Survey of Lan Amerca and he Carbbean , Sanago de Chle: Uned Naons Publcaon Engle, R. F. and C. W. J. Granger (1987), Co-Inegraon and Error Correcon: Represenaon, Esmaon, and Tesng, Economerca 55, Fanell, J. M. (2009), Volaldad, Cclo y Políca Fscal en Amérca Lana, EUROsocAL Fscaldad, Madrd Felx, R. A. (2008), The Growh and Volaly of Sae Tax Revenue s n he Tenh Dsrc, Federal Reserve Bank of Kansas Cy Economc Revew, 3rd Quarer, Fox, W. F. and C. Campbell (1984), Sably of he Sae Sales Tax Income Elasces, Naonal Tax Journal 37, Fuenes, J. and M. Tobar Slva (2004), La Políca Fscal como Herramena de Ajuse Cíclco en El Salvador, Documeno de Trabajo Banco Cenral de Reserva de El Salvador, No Groves, H. M. and C. H. Kahn (1952), The Sably of Sae and Local Tax Yelds, Amercan Economc Revew 42, Jménez, J. P. and V. Tromben (2006), Fscal Polcy and he Commodes Boom: The Impac of Hgher Prces for Non-Renewables n Lan Amerca and he Carbbean, CEPAL Revew 90, Labys, W. C. (2006), Modelng and Forecasng Prmary Commody Prces, Ashgae Publshng, Hampshre Legler, J. B. and P. Shapro (1968), The Responsveness of Sae Tax Revenue o Economc Growh, Naonal Tax Journal 21, Maddala, G. S. and S. Wu (1999), A Comparave Sudy of Un Roo Tess wh Panel Daa and a Smple New Tes, Oxford Bullen of Economcs and Sascs 61, Nchols, M. W. and M. S. Tosun (2008), The Income Elascy of Gross Casno Revenues: Shor-Run and Long-Run Esmaes, Naonal Tax Journal 61, Rncón, H., Berhel, J., and M. Gómez (2003), Balance Fscal Esrucural y Cíclco del Goberno Naconal Cenral de Colomba, , Banco de la Republca de 18

21 Colomba Borradores de Economía No. 246 Salazar, N. and D. Prada (2003), El Balance Esrucural del Goberno Cenral en Colomba, Repúblca de Colomba Deparameno Naconal de Planeacón Archvos de Economía No Schenone, O. H. and C. De la Torre (2005), Guaemala: Foralecmeno de la Esrucura Trbuara, n M. R. Agosn, Barrex, A., and R. Machado (eds.), Recaudar para crecer: Bases para la reforma rbuara en Cenroamérca, Ineramercan Developmen Bank, Washngon, D.C., Sobel, R. S. and R. G. Holcombe (1996), Measurng he Growh and Varably of Tax Bases over he Busness Cycle, Naonal Tax Journal 49, Sock, J. H. and M. W. Wason (1993), A Smple Esmaor of Conegrang Vecors n Hgher Order Inegraed Sysems, Economerca 61, Tanz, V. and H. H. Zee (2000), Tax Polcy for Emergng Markes: Developng Counres, Naonal Tax Journal 53, Tapa, H. (2003), Balance Esrucural del Goberno Cenral de Chle: Análss y Propuesas, ECLAC Macroeconomía del Desarrollo No. 25 Whe, F. C. (1983), Trade-Off n Growh and Sably n Sae Taxes, Naonal Tax Journal 36, Wlford, W. T. (1965), Sae Tax Sably Crera and he Revenue-Income Elascy Coeffcen Reconsdered, Naonal Tax Journal 18, Wllams, V., Anderson, R. M., Froehle, D. O., and K. L. Lamb (1973), The Sably, Growh and Sablzng Influence of Sae Taxes, Naonal Tax Journal 26, Wolswjk, G. (2009), The Shor- and Long-Run Tax Revenue Response o Changes n Tax Bases, Economcs Bullen 29,

22 Appendx A: Daa Deal and s [Table s connued on he followng 6 pages.] Argenna Income Tax Mnsero de Economía y de Fnanzas Públcas, Admnsracón Federal de Ingresos Públcos Includes whholdng axes Personal Income Tax Perod Q Q Admnsracón Federal de Ingresos Públcos Does no nclude whholdng axes Corporae Income Tax Perod Q Q Admnsracón Federal de Ingresos Públcos Does no nclude whholdng axes Value Added Tax Domesc Value Added Tax Mnsero de Economía y de Fnanzas Públcas, Admnsracón Federal de Ingresos Públcos Exernal Value Added Tax Mnsero de Economía y de Fnanzas Públcas, Admnsracón Federal de Ingresos Públcos Socal Secury Mnsero de Economía y de Fnanzas Públcas Revenue from Commodes Perod Q Q Admnsracón Federal de Ingresos Públcos Expor dues Real GDP Banco Cenral de la Repúblca Argenna In Pesos of 1993 Curren GDP Banco Cenral de la Repúblca Argenna Commody Prces Beef, Ausrala & New Zealand, frozen boneless, U.S. mpor prce FOB por of enry ( /lb.) Soybeans, Uned Saes, n 2 yellow, CIF Roerdam Soybean ol, The Neherlands, FOB ex-mll Crude peroleum, average of Duba/Bren/Texas equally weghed ($/barrel) IMF: Inernaonal Fnancal Sascs Prce ndex weghed wh share n oal expor dues Exchange Rae ECLAC Deflaor GDP Deflaor 20

23 Bolva Income Tax Perod Q Q Mnsero de Economía y de Fnanzas Públcas Personal Income Tax Perod Q Q Mnsero de Economía y de Fnanzas Públcas Corporae Income Tax Perod Q Q Mnsero de Economía y de Fnanzas Públcas Value Added Tax Perod Q Q Mnsero de Economía y de Fnanzas Públcas Domesc Value Added Tax Perod Q Q Mnsero de Economía y de Fnanzas Públcas Exernal Value Added Tax Perod Q Q Mnsero de Economía y de Fnanzas Públcas Revenue from Commodes Perod Q Q Mnsero de Economía y de Fnanzas Públcas Includes specal ax on hydrocarbons, drec ax on hydrocarbons and uly ax on mnng ndusry Real GDP Banco Cenral de Bolva In Bolvanos of 1990 Curren GDP Banco Cenral de Bolva Commody Prce Russan Naural Gas, n Germany U.S. Dollars per Thousand Cubc Meers (FMI) IMF: Inernaonal Fnancal Sascs Exchange Rae ECLAC Deflaor GDP Deflaor Brazl Income Tax Perod Q Q Tesouro Naconal Wholdng axes are ncluded Personal Income Tax Perod Q Q Tesouro Naconal Wholdng axes are excluded Corporae Income Tax Perod Q Q Tesouro Naconal Wholdng axes are excluded Value Added Tax Perod Q Q Recea Federal 21

24 Sae ax Socal Secury Perod Q Q Tesouro Naconal Real GDP Insuo Braslero de Geografa e Esaísca Index 1995=100 Deflaor Consumer Prce Index ECLAC Dec1993=100 Chle Income Tax Servco de Impuesos Inernos Includes CIT from he prvae mnng secor, ne values Personal Income Tax Servco de Impuesos Inernos Gross values Corporae Income Tax Servco de Impuesos Inernos Includes CIT from he prvae mnng secor, gross values Value Added Tax Servco de Impuesos Inernos Ne values Domesc Value Added Tax Servco de Impuesos Inernos Gross values Exernal Value Added Tax Servco de Impuesos Inernos Gross values Socal Secury Perod Q Q Servco de Impuesos Inernos Revenue from Commodes Servco de Impuesos Inernos Specfc ax and provsonal paymens of ncome ax and ne ransfers from CODELCO, does no nclude CIT from prvae mnng secor Real GDP Banco Cenral de Chle In Pesos of 2003 Curren GDP Banco Cenral de Chle Commody Prce Copper, wre bars, U.S. producer, FOB refnery ( /lb.) IMF: Inernaonal Fnancal Sascs Exchange Rae ECLAC 22

25 Deflaor Colomba GDP Deflaor Income Tax Perod Q Q Consejo Superor de Políca Fscal Includes CIT from he mnng secor Value Added Tax Perod Q Q Consejo Superor de Políca Fscal Domesc Value Added Tax Perod Q Q Consejo Superor de Políca Fscal Exernal Value Added Tax Perod Q Q Consejo Superor de Políca Fscal Real GDP Banco de la Repúblca Colomba In Pesos of 2000, already deseasonalzed, lnked seres Deflaor Consumer Prce Index ECLAC Dec2008=100 Domncan Republc Income Tax Perod Q Q Banco Cenral de la Repúblca Domncana Personal Income Tax Perod Q Q Drreccón General de Impuesos Inernos Shorfall n me seres does no allow applcaon of economerc model Corporae Income Tax Perod Q Q Drreccón General de Impuesos Inernos Value Added Tax Perod Q Q Banco Cenral de la Repúblca Domncana Real GDP Banco Cenral de la Repúblca Domncana In Domncan Republc Dollars of 1991 Curren GDP Banco Cenral de la Repúblca Domncana Deflaor GDP Deflaor Ecuador Income Tax Perod Q Q Dreccón General Adjuna de Esadísca de la Hacenda Públca Value Added Tax Perod Q Q Dreccón General Adjuna de Esadísca de la Hacenda Públca Socal Secury Perod Q Q Dreccón General Adjuna de Esadísca de la Hacenda Públca 23

26 Revenue from Commodes Perod Q Q Dreccón General Adjuna de Esadísca de la Hacenda Públca Real GDP Banco Cenral del Ecuador In U.S. Dollars of 2000 Commody Prce Crude peroleum, average of Duba/Bren/Texas equally weghed ($/barrel) IMF: Inernaonal Fnancal Sascs Exchange Rae ECLAC Deflaor Consumer Prce Index ECLAC 2004=100 Mexco Income Tax Secrearía de Hacenda y de Crédo Públco Value Added Tax Secrearía de Hacenda y de Crédo Públco Socal Secury Secrearía de Hacenda y de Crédo Públco Revenue from Commodes Secrearía de Hacenda y de Crédo Públco Real GDP Banco de Méxco In Pesos of 2003, lnked seres Commody Prce Crude peroleum, average of Duba/Bren/Texas equally weghed ($/barrel) IMF: Inernaonal Fnancal Sascs Exchange Rae ECLAC Deflaor Consumer Prce Index ECLAC 2nd forngh Jun2002 = 100 Peru Income Tax Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Gross values Personal Income Tax Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Gross values Corporae Income Tax Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Gross values 24

27 Value Added Tax Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Gross values Domesc Value Added Tax Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Gross values Exernal Value Added Tax Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Gross values Socal Secury Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Supporve conrbuons for penson plans are no ncluded Revenue from Commodes Perod Q Q Supernedenca Naconal de Admnsracón Trbuara Includes CIT on he mnng and hydrocarbon secors and ransfer from he mnng secor, gross values Real GDP Banco Cenral de Reserva del Perú In Nuevos Soles of 1994 Curren GDP Banco Cenral de Reserva del Perú Commody Prces Copper, wre bars, U.S. producer, FOB refnery ( /lb.) Znc, specal hgh grade, LME, cash selemen Gold, 99.5% fne, afernoon fxng London ($/roy ounce) Crude peroleum, average of Duba/Bren/Texas equally weghed ($/barrel) IMF: Inernaonal Fnancal Sascs Prce ndex weghed wh share n oal expors Exchange Rae ECLAC Deflaor GDP Deflaor Uruguay Income Tax Perod Q Q Mnsero de Economía y de Fnanzas Personal Income Tax Mnsero de Economía y de Fnanzas No used (as mplemened no before 2007, replacng varous axes) Corporae Income Tax Perod Q Q Mnsero de Economía y de Fnanzas Value Added Tax Perod Q Q Mnsero de Economía y de Fnanzas Does no nclude ax on socal secury conrbuon Domesc Value Added Tax Perod Q Q Mnsero de Economía y de Fnanzas 25