OIL-DEPENDENT REVENUES AND MACROECONOMIC STABILITY UNDER FISCAL AND MONETARY RULES: AN ANALYSIS FOR MEXICO

Transcription

1 OIL-DEPENDENT REVENUES AND MACROECONOMIC STABILITY UNDER FISCAL AND MONETARY RULES: AN ANALYSIS FOR MEXICO Ana María Aguilar * and Claudia Ramírez Bulos * In his paper we develop a small open economy macroeconomic model, and esimae i for Mexico. We incorporae he effec of oil prices on public finances. Assuming moneary policy follows an opimal policy rule, we evaluae he impac of wo differen fiscal policy rules: a balanced budge rule or a srucural-balance rule on macroeconomic sabiliy. We find ha when he economy faces inflaion or consumpion shocks, boh rules generae almos he same effec. However, when oil price shocks occur, higher macroeconomic sabiliy is achieved and he moneary auhoriy reacs less aggressively under he srucural balance-rule. These resuls are even more relevan wih he recen proposals o flexibilize he energy pricing policy in Mexico. Inroducion Fiscal policy and public deb are imporan for moneary policy since hey can influence he level of aggregae demand and ineres raes and hey may also affec moneary auhoriies abiliy o conrol inflaion. Sandard economic heory suggess ha fiscal policy should be counercyclical. A procyclical fiscal policy ranslaes ino higher public spending and lower ax raes in good imes, and viceversa. However, in pracice governmens seem o follow pro-cyclical fiscal policies, as some auhors have already documened. This feaure is even more imporan in developing economies, paricularly for commodiy-rich counries, where revenues linked o commodiies can be a large porion of governmen revenues. Hence, when commodiy prices are high, someimes linked o posiive global growh perspecives, governmens migh have more revenues and spend more. Addiionally, he fac ha commodiy prices are generally very volaile, overall revenues become volaile as well. 3 Mexico, as oher commodiy exporer counries has faced a number of challenges derived from he managemen of commodiies, paricularly oil. In his counry, public finances are significanly influenced by he movemens on oil prices. In he firs place, more han 3 per cen of he public secor s revenues are from oil, in he second place, energy prices are se by a rule deermined by he governmen. This rule is supposed o ac as a ax when energy prices abroad are low and as a subsidy when hose prices are high. Indeed, before mid-s, his rule funcioned as a ax, however, during he las decade energy prices have been showing an upward rend and, alhough hey declined emporary during he 8-9 financial crisis, recenly hey reached hisorically high levels and elevaed volailiy. In his sense, i is imporan o reconsider he way public expendiures are deermined by such a volaile source of revenues, and o re-hink he appropriaeness of he energy pricing rule in his new environmen. Thus, in his paper we develop a small open economy macroeconomic * 3 Banco de México. amaguilar@banxico.org.mx, and claudia.ramirez@banxico.org.mx The auhors hank Daniel Sámano and Carlos Lever for heir useful commens. MauricioTorres, José Manuel Rodríguez and Miriam González provided us wih excellen research assisance. The opinions expressed in his aricle are solely hose of he auhors and do no necessarily reflec he views of Banco de México. See Barro (979). See Talvi and Vegh (5), Gavin and Peroi (997), Sein e al. (999), and Lane (3). See Céspedes and Velasco ().

2 Ana María Aguilar and Claudia Ramírez Bulos model, and esimae i for Mexico, incorporaing he effec of oil prices on public finances. Assuming moneary policy follows an opimal policy rule, we evaluae he impac of wo differen fiscal policy rules: a balanced budge rule or a srucural-balance rule. We find ha when he economy faces inflaion or consumpion shocks, boh rules generae almos he same effec. However, when oil price shocks occur, he moneary auhoriy reacs less aggressively and higher macroeconomic sabiliy is achieved under he laer rule. These resuls are even more relevan wih he recen proposals o flexibilize he energy pricing policy. 4 I is imporan o menion ha Mexico, like many oher developing economies, has showed some convergence owards greaer ransparency and accounabiliy, recognizing he benefis of he auhoriy aking decisions under clear and ransparen procedures, under a regulaory framework versus a discreionary one. In recen years, Mexico has made some imporan reforms. Indeed, one of he major acions aken was recognizing in he Consiuion, in 994, he auonomy of he cenral bank, hus generaing greaer moneary policy credibiliy. In urn, some reforms regarding fiscal policy were pu in place as well. In 6 he Fiscal Responsibiliy Ac was approved in which, among oher hings, i requires ha he economic package is formulaed based on medium-erm projecions, in paricular esablishing a formula for seing he price of oil o fix he budge for he nex fiscal year, and does no allow he ax auhoriies running deficis. These reforms allowed achieving lower levels of defici and deb, limiing and evenually eliminaing he fiscal dominance siuaion, which allowed he moneary policy o fulfill is mission of mainaining price sabiliy and generaing greaer macroeconomic sabiliy. However here are sill some vulnerabiliies ha prevail in he Mexican public finances and ha during he recen financial crisis were presen, he sill high dependence of public revenues on oil revenues, and he fac ha energy prices are se by a rule deermined by he governmen, as we have menioned. Thus, we consider his a relevan issue for Mexican macroeconomic policies, since all sources of volailiy for he macroeconomy ha are no managed by he fiscal auhoriies migh be addressed by he moneary ones. Even hough he energy pricing rule may be convenien from he inflaion sabiliy perspecive in he shor run, since i absorbs he major par of energy prices volailiy, i could imply such high coss in erms of public finances ha in he medium erm i migh become a hrea for price sabiliy. In addiion o ha, he dependence of public finances on oil prices makes i very common for he governmen o close he governmen budge wih adjusmens in public prices and raes fixed by he governmen. This is a very imporan source of uncerainy for inflaion (especially non-core componen) and herefore inflaion expecaions. The cenral bank has been very successful in accommodaing supply shocks o inflaion, specifically exchange rae flucuaions and volailiy in agriculural prices. However, he Mexican economy has no ye learned o absorb, in an orderly way, energy price volailiy. Therefore, his is an open issue and sill a challenge from he moneary policy perspecive. Given he curren environmen and inernaional condiions in he oil marke, he acions underaken in he recen pas can lead o beer resuls if he auhoriy considers addiional elemens o isolae he effecs of cyclical elemens in he macroeconomic decisions. Thus, hese maers are aken ino accoun in his paper in which we develop a small open economy macroeconomic model for policy analysis, incorporaing he effec of oil prices on public finances, in an environmen in which moneary policy follows an opimal policy rule, and fiscal policy follows one of wo rules: a balanced budge rule or a srucural-balance rule. The purpose is o idenify he effec of hese differen fiscal rules on he Mexican economy o various shocks and heir effecs on macroeconomic sabiliy. 4 The energy pricing policy is described laer in he documen.

3 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 3 The paper is organized as follows. In Secion we describe some common feaures of he fiscal rules ha have been pu in place in a number of economies. In Secion 3 we presen he model. In Secion 4 we describe he daa we used o esimae and calibrae he model. In Secion 6 we analyze he wo policy rules showing some exercises and Secion 5 concludes. Fiscal rules During he pas wo decades moneary policy in many emerging economies was consrained by he high levels of public deb. Neverheless, mainly during he 99s many emerging counries implemened some policies ha led o reduced deb levels generaing a beer fiscal posiion. Indeed many economies implemened formal fiscal rules. Three waves can be observed as (Schaecher e al., ) menions. The firs one in he early and mid-99s in par responding o bank and deb crises as well as consolidaion needs o qualify for he euro area. The second one was driven largely by emerging economies in he early s when many adoped rules and reformed fiscal frameworks in responses o experience wih fiscal excesses, and more recenly he hird wave in response o he recen crisis. In general, he main argumens supporing he use of fiscal rules are o achieve macroeconomic sabiliy, o suppor oher poliical rules, o obain susainable fiscal policies, among ohers. Nowihsanding, in Lain America, some of he rules implemened prior o he recen crises were more effecive in erms of addressing fiscal susainabiliy as compared o he capaciy of responding o shocks, as Berganza () menions. Thus srong medium-erm growh prospecs, in many emerging markes fiscal posiions are sill exposed o financial and exernal demand shocks, paricularly hose commodiy exporers. There have been several empirical sudies regarding fiscal rules. One of he main research areas in relaion o fiscal rules focuses on deermining wheher he rules acually resric he governmen s abiliy o use fiscal policy o smooh business cycle flucuaions. Bayoumi and Eichengreen (995) use daa from 97-9 for he Unied Saes o find ha he reducion in fiscal sabilizers can lead o a significan increase in aggregae oupu variaions. In anoher sudy, Alesina and Bayoumi (996) show ha even hough he ax rules are associaed wih less variabiliy cyclical fiscal balance, his does no lead o increased oupu variabiliy. They used daa for U.S. saes from 965 o 99 and found ha here is saisically significan relaionship beween he acual oupu variabiliy and severiy saes of fiscal conrols. They speculae ha his could occur simply because sae-level sabilizaion migh no be very imporan or because sronger conrols limi flucuaions in he surplus ha could be poliically moivaed and are poenially desabilizing and limiing counercyclical policies, which could lead o uncerainy in he impac on oupu variabiliy. Meanwhile, Levinson (998) noes ha Alesina and Bayoumi (996) do no conrol for unobservable characerisics ha may be correlaed wih he business cycle flucuaions and he exisence of sae fiscal conrols. He suggess ha he size of he sae is correlaed wih he abiliy o affec cycle flucuaions hrough counercyclical fiscal policies and remarks ha sae ax policy maers more in large saes han in small ones, hus he difference beween he business cycle flucuaions in saes wih fiscal conrols and sric conrols is acually higher among large saes han beween small saes, hence concluded ha here is evidence ha a sric fiscal rule as he budge balance exacerbaes he business cycle flucuaions. In urn, Galí and Peroi (3) invesigaed wheher he Maasrich Treay and he Sabiliy and Growh Pac in Europe weakened he abiliy of European governmens o conduc a sabilizing fiscal policy. This is done by esimaing fiscal rules for he discreionary budge defici in he period 98-, using daa from he European counries in he moneary union and conrol groups of counries ha are no par of he moneary union. They find ha discreionary fiscal

4 4 Ana María Aguilar and Claudia Ramírez Bulos policy in EMU counries have become more couner-cyclical over ime, following a rend ha seems o be affecing oher indusrialized economies as well. In urn, here is some evidence on he pro-cyclicaliy of fiscal policy in Lain America (see Gavin and Peroi (997); Kaminsky e al., 5; Daude e al., ) which conribues o macroeconomic volailiy. In sum, afer he economic crisis, some governmens ha already had some sor of fiscal rules in place migh need o adjus hem based on he lessons resulan from experiences afer he crisis and some oher counries will sar implemening hem. I is also noeworhy ha in many counries fiscal rules were violaed during he crisis as many governmens have adoped simulaive policies. However, combining auomaic responses and discreionary simulaive acions has produced some imbalances in many counries. One of he mos common ways o evaluae he possible effec of cerain policies in a paricular economy is hrough he analysis of economic models ha describe he dynamics of he main macroeconomic variables. In he following secion, a macroeconomic model is developed in which fiscal and moneary policies inerac in he conex of a small open economy. 3 Model In his secion we presen he model used in his paper o analyze he effec of implemening fiscal rules on macroeconomic sabiliy. We develop a sandard semisrucural small open economy neokeynesian model. Even hough he coefficiens of he model have a heoreical background, hey are presened in a reduced form. 5 In paricular, he aggregae supply is modeled by a New Phillips Curve; he aggregae demand is deermined by is main componens: privae consumpion, governmen spending, invesmen, expors and impors, hese equaions are derived from each componens opimizaion process and are expressed in erms of reduced form parameers. Addiionally, he model incorporaes an equaion o describe he behavior for he real exchange rae and wo policy rules: an opimal moneary policy rule and a fiscal rule. The design of he fiscal policy akes ino accoun wo differen alernaives: a balanced budge and a srucural balance. This model considers he exisence of oil, whose price is deermined in inernaional markes (i.e., he price is exogenous o he model) and producion of his naural good is owned by he governmen, as i is he case in Mexico Model specificaion 3.. Aggregae demand The aggregae demand is defined as he sum of is differen componens, privae consumpion, privae invesmen, governmen spending and he rade balance (expors and impors). 7 The differen componens of aggregae demand are modeled individually wih reduced form coefficiens, which ake ino accoun a number of feaures. Firs, he aggregae demand in his model is an exension of he opimizing IS-LM model specificaion for an small open economy which incorporaes an exernal secor, for his purpose, he model developed by McCallum and 5 See Mc Callum and Nelson (999). 6 The esimaion of he model for he Mexican economy is presened in he Appendix. 7 Invenories are considered a residual.

5 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 5 Nelson (999) is he main reference for his paper. Second, he equaions used in his work were developed for hree main purposes: i. o capure he persisence of he variables, ii. o incorporae he agens expecaions and allow hem o smooh heir behavior, and iii. o generae a srucural relaionship beween variables. These main feaures allow for a simple model o explain he behavior of an small open economy, such as Mexico. 8 The governmen spending dynamics is described in more deail a he end of he secion. Privae consumpion In order o capure an imporan feaure of Mexican consumers, we assume ha here is a fracion Φ of consumers who have access o financial markes, so ha hey can smooh consumpion over ime by accumulaing financial asses. 9 While a fracion ( Φ) do no have access o financial markes and consume only heir disposable income each period. To model he behavior of consumers wih access o financial markes, we use he specificaion of McCallum and Nelson (999), where consumers opimizaion process, afer some manipulaion, leads o he following represenaion: ( [ ]) FM FM FM c = α + αc + αe c + + α3 R E Δ p+ + υc, () FM c In equaion (), α 3 <, is he logarihmic deviaion of consumpion (Dixi Sigliz ype) of he agens wih access o financial markes of he goods ha he represenaive agen consumes in period, aking ino accoun consumpion habis. R is he nominal ineres rae paid for bonds in he domesic counry, E [Δp ] is he expeced inflaion rae over he nex quarer, so R E [Δp ] + + is he real ex ane ineres rae, and vc is a sochasic shock relaed o he agens preference over presen and fuure consumpion, wih zero mean and consan variance, σ,vc FM. The consumpion for agens wihou access o financial markes, is simply modeled by: ax ( ), c = x + υ () NFM c where x ax, is he log linearized disposable personal income, and ax, is he income ax. Privae invesmen In he spiri of he consumpion equaion, i is assumed ha privae invesmen also mainains a hybrid dynamic so he behavior for presen invesmen depends on he projecs sared in he pas, on he expecaions for fuure invesmen and on he real ex-ane ineres rae, resuling in he following equaion: [ ] ( [ ]) in = ϑ + ϑ in + ϑ E in + ϑ R E Δ p + υ (3) in, 8 These componens are expressed in erms of deviaions from heir endency in an effor o idenify in a simple fashion he ransmission mechanism of he moneary and fiscal policies over he oupu level. 9 Following Woodruff (6) in Mexico around 74 per cen of he populaion has access o financial markes, whereas 6 per cen do no have access. The model equaions can be derived from an infinie horizon scheme where agens choose opimal consumpion and deb pahs and each one of hese agens produce a good wih a cerain level of marke power.

6 6 Ana María Aguilar and Claudia Ramírez Bulos In his equaion, ϑ 3 <, in is he logarihmic deviaion from privae invesmen wih respec o is endency and R E [Δp + ] is he real ex ane ineres rae. υin is a sochasic shock relaed o he invesors preferences over ineremporal invesmen, wih mean zero and consan variance σ. υin Expors Given he conex of a small open economy, he domesic economy expors goods o a small fracion of foreign consumers and he level of expors do no affec he price level of he foreign economy. Aggregae expors in his model are posiively relaed o he real exchange rae, * A Q S P / P, so he equaion ha describes expors is given by he following expression: [ ] ex = δ + δ ex + δ E ex + δ q + δ x + υ (4) US ex, In equaion (4), ex represens he logarihmic deviaion of expors wih respec o is US endency, q represens he real exchange rae, x represens he foreign counry oupu gap and υ ex represens a sochasic shock relaed o he preferences from foreign consumers owards domesic goods, wih mean zero and consan variance, σ υex. Impors The domesic economy does no only consume domesic producion goods, bu also impored goods from abroad ha can be used as producion inpus. In his case, impors, im depend on consumers habis wih respec o consumpion of foreign goods and he expecaions for fuure impors. Impors are also affeced by he real exchange rae, q, and by he domesic oupu gap as defined bellow: [ ] im = γ + γ im + γ E im + γ q + γ x + υ (5) im, υ im represens a sochasic shock relaed o domesic consumers In his case, he erm preferences owards foreign goods, wih mean zero and consan variance, σ υim. Invenories I is assumed ha invenories follow an AR() process given by: inven = ρ inven + υ (6) υe inven, where ρ υe < and υinven is a sochasic shock wih mean zero and consan variance σ υex. This variable is commonly a residual ha allows he oupu level o be adjused; in his case i is assumed exogenous. This componen is also specified as a residual in he esimaions presened in he Appendix.

7 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico New Phillips curve In Neo-Keynesian models, a naural way o describe aggregae supply is hrough a Phillips Curve, which relaes he price level in an economy wih is oupu level. Following McCallum and Nelson (999), in order o incorporae nominal rigidiies, we assume ha he economy has gradual price adjusmens when firms ake producion decisions, so he descripion of price dynamics is fied properly by a Phillips Curve equaion. Many sudies have ried o describe inflaion dynamics for several counries using backward and forward-looking componens, in paricular, for Mexico, Ramos-Francia and Torres (5) find ha he New Neo-Keynesian Phillips Curve properly describes inflaion dynamics in Mexico. Following ha specificaion, we use a hybrid Phillips curve o describe he inflaion dynamics. Addiionally, we use core inflaion since i bes reflecs inflaionary pressures as i migh be considered he srucural inflaion of he economy. The New Phillips curve equaion is he following: ( ) C C C US π = απ + αe π + + α3x + α4 Δ ner + π + υπ (7) C C where core inflaion π is affeced by pas inflaion, π by he expecaions for fuure inflaion, C E π + by he oupu gap, x, and since i is a small open economy, i is also affeced by changes US in he nominal exchange rae and foreign inflaion, Δ ner + π. Addiionally, a sochasic shock is included relaed o changes in inflaion dynamics wih he radiional feaures of zero mean and consan variance Real exchange rae The real exchange rae equaion used in his model considers he real ineres raes pariy: US [ + ] ( ) rer = E rer + r r (8) However, one of he main disadvanages for his equaion is ha he real exchange rae adjuss insanly given any imbalance producing an overshooing of is esimaed value. Several proposals have been made in he lieraure o avoid his abrup adjusmen of he exchange rae (delayed overshooing) as described in Eichenbaum and Evans (995). One of he soluions proposed o his problem consiss in incorporaing an exchange rae lag in he ineres rae pariy condiion. This procedure generally resuls saisically significan in order o explain he real exchange rae since i replicaes he observed persisence in he real exchange rae dynamics generaing a gradual adjusmen. Under his assumpion, i is implicily supposed ha he exchange rae depends on he real exchange rae in he previous period and on he expecaions of he exchange rae on he following period, as presened in he following equaion. US ( [ ] ( )) rer = c rer + c E rer + r r (9) Moneary policy rule In his model, he moneary policy insrumen, he shor-run nominal ineres rae, is obained hrough an opimal rule following Söderlind (999). See Galí and Gerler (999) among ohers.

8 8 Ana María Aguilar and Claudia Ramírez Bulos We calculae opimal discreionary moneary policy rules for each fiscal policy rule esed. This mehodology does no require us o specify he form of he opimal insrumen, since i is a funcion of all he values of he sae-vecor of he model. In order o calculae his opimal policy rule we assume ha he moneary auhoriy minimizes a sandard loss funcion. Hence, we define he loss funcion L as: ( ( ) ( ) ) * π j L= E ϕ α π j π αxx j αi i j i + + j () j= s.. he srucure of he economy (inflaion, aggregae demand, real exchange rae) In his loss funcions, α π, α x and α i represen he relaive imporance of oupu and inflaion gaps as well as ineres rae smoohing, and we assume ha he auhoriy assigns he same weigh o he hree of hem Fiscal policy rule In his model, governmen spending no only is considered par of he aggregae demand, bu i is also subjec o fiscal rules ha deermine how fiscal adjusmens should be done by he auhoriy given exogenous shocks. We assume ha he budge consrain for he governmen is he following: ( ( exp( ) ) ) B = G T + R +Ψ B b B () where B is he real level of governmen indebedness, b represens he deb level in he saionary sae, G represens real governmen spending, T represens public revenues, R is one plus real ineres rae (+r ) and (R + Ψ exp(b b) ) is he cos of deb. 4 In his expression, Ψ>, is he ineres rae elasiciy given changes in deb levels, he cos is defined his way because he price of deb is sensiive o he level of ousanding deb. Addiionally i is supposed he exisence of a commodiy, oil, sold abroad wih producion righs owned by he governmen, so he revenues from his sale are enirely for he governmen. Therefore, public revenue has wo main componens ax revenues on he one hand, and oil revenues on he oher. Toal revenues are described as: ( ϕ) T = ϕτy + P Y () oil oil where ϕ is he share of ax revenues, τ is he ax rae, ( ϕ ) is he share of oil revenues, he price of oil in period and urn. oil Y is oil producion in period. Governmen spending may follow wo differen behaviors depending on he fiscal rule in 3 See Appendix for more deails abou he opimal policy rule. 4 This expression incorporaes an adjusmen cos following Schmi-Grohe and Uribe (), so he deb is no undeermined, given he fac ha R is greaer han one so deb is no saionary. Assuming such adjusmen cos, we can ensure saionariy by supposing ha (R +Ψexp(B b) ) is less han one. oil P is

9 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 9 Balanced Budge Rule: In his case, he governmen mees he balanced budge resricion bu here exiss he possibiliy of issuing deb, so governmen spending adjuss in he following way: 5 ( ( exp( ) ) ) G = T R +Ψ B b B (3) Wih he balanced-budge rule (BB), he governmen can spend only wha i receives from ax and oil revenues minus he deb service, consan over ime. However, afer an increase in oupu, ax revenues increase allowing for a higher spending level while meeing he rule. In he same way, afer a decrease in oupu, governmen spending mus be reduced in order o mee he rule. The same behavior for governmen spending is observed afer changes in oil revenues. Therefore his rule is pro-cyclical. Srucural Balance Rule: In he same spiri as in Galí and Peroi (3) he srucural fiscal defici or srucural balance (SB), aemps o facor ou cyclical componens from he acual budge balance. In general, he acual balance reflecs boh cyclical and srucural facors ha migh over- or underadjus budge developmens by he governmen bu separaing cyclical from srucural facors migh help diminish hese overadjusmens. 6 Assume ha he balance is represened as follows: ( ( exp( ) ) ) BA = T G R +Ψ B b B (4) ( ( exp( ) ) ) B = BA T = T T G R +Ψ B b B (5) s being ( ) T = ϕτ Y Y + ϕ P P Y oupu level and P as a price reference for oil. 7 oil oil oil ref ref oil ref. In his case Y ref is defined as poenial The srucural balance rule used in his paper is based on he rule currenly followed by Chile and akes ino accoun he model described by Medina and Soo (7) for his Lain American economy when defining he srucural balance rule. 8 While following his ype of rule, a shif in oil price causes an increase in he level of cyclical revenues since he gap for oil price wih respec o is reference level increases, and so he srucural defici level diminishes. In order o mainain he srucural defici level unchanged, governmen spending should decrease o compensae for his effec. The same is rue when oupu level deviaes from is poenial level, afer an increase in oupu, cyclical revenues increase wih he same effec on he srucural defici. In his way, fiscal policy does no overadjus over he cycle. 5 Under he balanced budge rule i is possible o isolae he effec of moneary policy on fiscal behavior; while for a posiive level of deb, moneary policy acions hrough ineres rae affec he cos of deb. 6 See Fiess (4). 7 8 The oil reference price for Mexico is deermined as a weighed average of observed prices and fuures, as described in he Fiscal Responsibiliy Ac. Given he log-linearized srucure of he model, he srucural balance level as a fiscal policy rule becomes irrelevan. Consan levels in his ype of models do no affec he resul since variables are expressed in erms of deviaions from he seady sae.

10 Ana María Aguilar and Claudia Ramírez Bulos Equaions () o (5) are expressed in levels, however, in order o be incorporaed in he model and mainain he same srucure as he res of he equaions in he model i is necessary o log-linearize hem. In his case, he equaions are expressed in he following way: Budge consrain: b = s g + s b + s 3 r s 4 (6) oil oil Public revenues: = qx q( p + x ) + (7) oil oil Balanced Budge Rule: g = nx + np + n3x n4b n5r (8) ref oil oil Srucural Balance Rule: g j x j p j x j b j r j ( bas x ) Variables b, g,, = (9) ref oil p, oil p ref, oil x and x, ref x are logarihmic deviaions of real deb, governmen spending, oil price, oil producion and oupu level, respecively; r is he shor-run real ineres rae. 9 We assumed ha non-core inflaion, oil price gap, oil producion gap and invenories follow an AR() process. The oupu gap is compued as he weighed sum of is componens gaps, and he foreign variables follows an auoregressive vecor. Foreign Variables: π = φ + φπ + φ π + φ x + φ x + φ i + φ i + ν () US US US US US US US πus x = k + kπ + k π + k x + k x + k i + k i + ν () US US US US US US US πus i = χ + χπ + χ π + χ x + χ x + χ i + χ i + ν () US US US US US US US πus Daa In his secion we describe he daa used for he analysis. The relevan variables regarding he Mexican economy are: privae consumpion, governmen spending, invesmen, expors, impors, real deb, nominal and real exchange raes. For Mexico and he U.S., he variables of ineres are inflaion, nominal and real ineres raes, oupu gap and he crude oil price. Our sample ranges from he firs quarer of o he hird quarer of. We decided o use his sample since variables such as inflaion sared o be saionary in Mexico since hen (Capisran e al., 8). For Mexico, quarerly daa of GDP componens: privae consumpion, governmen spending, invesmen, rade balance, was used o esimae gaps for each of hese variables. Each gap is defined as he percenage deviaion (logarihmic difference) beween he level of each variable and is rend, calculaed using he Hodrick-Presco filer. Monhly core and non-core inflaion were ransformed ino quarerly daa. The source of all daa is INEGI. Inflaion arge, prior o 4, was calculaed by a quarerly lineal inerpolaion of inflaion arges esablished by Banco de México in is Annual Inflaion Repors. Since 4 i corresponds o he curren 3 per cen arge esablished by Banco de México. Nominal exchange rae refers o FIX exchange rae in pesos per dollar and real exchange rae corresponds o an index (base 99=). The periodiciy of hese series is daily so we calculaed he quarerly average and expressed hem in 9 The parameers s,q,n and j are define afer log-linearizing he model, and are presened in Appendix.

11 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico logarihms. Regarding nominal ineres rae, i corresponds o 9-days CETES yield. Real ineres rae was esimaed from he nominal ineres rae and monhs inflaion expecaion. The source of his daa is Banco de México. Wih respec o U.S variables, CPI was exraced from he Bureau of Labor Saisics. Nominal ineres rae corresponds o 3-monh Treasury bills secondary marke rae, which was provided by he Federal Reserve. Real ineres rae was calculaed using he nominal ineres rae and monhs inflaion expecaion, published by Michigan Universiy (Survey Research Cener). As in he case of Mexican variables, we ransformed hem ino quarerly daa. The source of oupu gap daa is Haver Analyics and he crude oil price was provided by Bloomberg. In order o analyze he dynamics of he Mexican business cycle, we esimaed he volailiy of GDP and is componens, measured as he sandard deviaion, and he cross correlaion beween GDP and he res of he variables. Firs we analyze he performance of hese variables across he enire sample and hen we divide i ino wo subsamples: he firs one ranges from he firs quarer of o he las quarer of 7, jus before he inernaional financial crisis sared; and he second one goes from he firs quarer of 8 o he hird quarer of. In Table we repor he resuls for he enire sample. As we can observe from column seven, all he componens of GDP, excep public consumpion are highly pro-cyclical, being privae consumpion and impors he componens wih higher correlaion wih he GDP gap. Regarding privae consumpion, we can also see from he second column ha in conras wih wha he consumpion heory would sugges, consumpion is more volaile han oupu. Neverheless, his resul is consisen wih previous sudies of business cycles in emerging economies, such as Mexico, Torres () and Cuadra (8). The volailiy in privae consumpion suggess ha consumpion smoohing across he cycle is very limied, which can be due o he exisence of credi consrains or o he cyclical componen of he cos of exernal credi, as menioned in Cuadra (8). In Table we repored he resuls for he enire sample. Cyclical componens of invesmen, expors and impors exhibi grea volailiy across he cycle. Looking o he relaive volailiy, which is defined as he raio of he sandard deviaion of each variable o he sandard deviaion of GDP, we can see ha hese variables are wice as volaile as he oupu. If we compare our resuls wih similar analysis using differen samples, we can see ha he volailiy repored for invesmen and impors in our daa is lower han he one repored using a bigger sample. Cuadra (8) found ha from 98 o 6 invesmen was four imes more volaile han he oupu and impors were six imes more volaile. Our resuls are consisen wih empirical daa showing imporan decreases in volailiy of invesmen and impors hrough ime. Regarding he cyclical componen of public consumpion, we can see in Table and Figure ha i shows low absolue and relaive volailiy. One possible explanaion o his fac is he prevailing fiscal rule in Mexico. As we menioned before, in order o increase he ransparency and reduce he discreionary componen in public spending, in 6 was approved he Federal Budge and Fiscal Responsibiliy Law which forces he governmen o have an equilibrium in public finances a all ime. This is also consisen wih he low correlaion of public spending wih GDP. Wih respec o he real ineres rae, we observed ha during he period of our sudy i exhibied low volailiy. Moreover, as we can see from column four o welve, cross correlaion of his variable wih GDP sugges ha real ineres rae reacs wih some lag o flucuaions in oupu, alhough when we looked ino he subsamples i exhibied differen dynamics before and afer he crisis.

12 Ana María Aguilar and Claudia Ramírez Bulos Figure GDP and Governmen Consumpion (cycle componens) 5 Oupu gap Very pro-cyclical Less pro-cyclical Gov. Consumpion gap Q 98Q 98Q 983Q 984Q 985Q 986Q 987Q 988Q 989Q 99Q 99Q 99Q 993Q 994Q 995Q 996Q 997Q 998Q 999Q Q Q Q 3Q 4Q 5Q 6Q 7Q 8Q 9Q Q Q Q Balanced Budge Rule Comparing our wo subsamples (Table and Table 3), we can observe ha afer he financial crisis volailiy in all he variables increased. Neverheless he dynamics of he relaive volailiy were differen across variables. In he case of consumpion, even hough i showed a considerable increase in volailiy, relaive volailiy remains almos unchanged, and since he crisis consumpion became more correlaed wih GDP. For invesmen, we observed ha in he second period absoluevariance increased bu is relaive volailiy decreased. As we menioned before, his performance is consisen wih an imporan reducion of he volailiy of he cyclical componen of invesmen. The same consideraion applies o he componen of expors. Regarding he relaion beween real ineres rae and oupu cycles we can observe ha i changes significanly beween our wo subsamples. This difference is mainly due o he fac ha he cenral bank mainained ineres raes unchanged for several years afer he crisis and inflaion was sable as well. 4. Oil For several decades, he Mexican economy has been characerized for he high dependence of public revenues o oil revenues. This siuaion has been considered one of he major issues of public finances in Mexico. Evenmore, in recen years he producion of oil in Mexico has been beneah he average of 99-, as we can see in Figure, he barrels of crude oil exraced per Some credi raing agencies have explicily menioned his facor as one of he mos imporan obsacles o upgrade he Mexican deb.

13 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 3 3,5 Mexican Crude Oil Producion (barrels per day) Figure 3,,893,5,,5, day reached is maximum level in 4 and since hen oal producion has been decreasing. A he same ime, we have observed an increasing volailiy of inernaional oil prices (Figure 3). In paricular, he price of he Mexican crude oil has grown more han 6 per cen since, wih higher volailiy hrough 7 and 8 due o he inernaional financial crisis, where we observed a sudden decrease. The combined effec of changes in hese wo variables has been refleced in paricular in public finances. In Mexico, more han a hird of oal public revenues come from oil (Figure ), making revenues quie vulnerable o he dynamics of he siuaion in inernaional markes. Since he early sevenies, he developmen of oil prices has been driven mainly by he ineracion beween supply and demand. Changes in oil supply by OPEP were very relevan facors behind he significan changes in oil prices in he sevenies. However, he sharp increase in oil demand, especially from emerging counries in recen years has been a crucial facor behind he dynamics of inernaional oil prices, of course ogeher wih supply policies applied by some producing counries. Looking ahead, i is likely ha imbalances beween oil supply and demand will coninue o be relevan as oil supply could be limied by physical facors and by reducing invesmens, while demand, especially from emerging economies could coninue o grow, as menioned by ECB ().

14 Variable x Sigma (x) Sigma (x)/ Sigma (GDP) Summary Saisics (period: Q-Q3) Correlaion of GDP wih Table x( 4) x( 3) x( ) x( ) x() x(+) x(+) x(+3) x(+4) Oupu Privae Consumpion Public Consumpion Invesmen Ana María Aguilar and Claudia Ramírez Bulos Expors Impors Real Ineres Rae * I corresponds o he gap of each variable, excep for real ineres rae.

15 Variable x Sigma (x) Summary Saisics* (period: Q-Q4) Table Sigma (x)/ Correlaion of GDP wih Sigma (GDP) x( 4) x( 3) x( ) x( ) x() x(+) x(+) x(+3) x(+4) Oupu Privae consumpion Public consumpion Invesmen Expors Impors Real ineres rae Variable x Sigma (x) Summary Saisics* (period: 8Q-Q3) Table 3 Sigma (x)/ Correlaion of GDP wih Sigma (GDP) x( 4) x( 3) x( ) x( ) x() x(+) x(+) x(+3) x(+4) Oupu Privae consumpion Public consumpion Invesmen Expors Impors Real ineres rae * I corresponds o he gap of each variable, excep for real ineres rae. Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 5

16 6 Ana María Aguilar and Claudia Ramírez Bulos 3.5 Inernaional Marke Price (US $) Figure 3 4 Oil Price (dls/barrel) Gasoline Price (dls/galon, lef axis) Correlaion coefficien = Jan-997 Nov-997 Sep-998 Jul-999 May- Mar- Jan- Nov- Sep-3 Jul-4 May-5 Mar-6 Jan-7 Nov-7 Sep-8 Jul-9 May- Mar- Jan- Nov- Oil Revenues (percen of oal revenues) Figure 4 4 Oil Revenues Average

17 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 7 Under hese circumsances i is harder for he governmen o plan fuure public spending. Moreover, even hough in Mexico public finances have been sound and macroeconomic sabiliy has improved, here is sill room for reducing he dependence of public revenues o oil revenues. The prevailing fiscal rule has allowed he governmen expendiure o reduce is volailiy, however he balanced budge rule currenly in place is sill a procyclical fiscal policy rule. 4.. Oil prices and inflaion One imporan feaure of he Mexican price srucure is ha fuel prices are se by he governmen, hence volailiy in oil prices has no been ransmied o CPI inflaion in any direcion. Unil he second half of s, gasoline prices in Mexico were higher han inernaional gasoline prices, however, given he abrup increase in oil prices in he las years his siuaion has changed. As a response, in he las years he governmen has been announcing gradual increases in gasoline prices in order o close he gap beween domesic and inernaional prices. In his conex, i has been discussed he possibiliy of flexibilizing gasoline prices, or a leas adjus hem o be consisen wih inernaional gasoline prices. We have discussed he poenial vulnerabiliies of public finances o oil prices volailiy, however, if his adjusmen in inflaion occurs he effecs of volailiy in oil prices will no only affec public finances bu inflaion as well. Under hese circumsances, he moneary policy migh need o respond more aggressively. 5 Exercises In his secion we perform a macroeconomic evaluaion of boh fiscal rules allowing he moneary policy o opimally adjus is policy under boh fiscal scenarios. Nex we presen some impulse-response funcions in order o illusrae he funcioning of he model. 5. Evaluaion of policy rules In his subsecion we use he model presened in he previous secion and he loss funcion described above o evaluae he wo fiscal alernaives presened before. We se he loss funcion in order o compare he wo differen policy rules. In paricular, we are ineresed in evaluaing wheher he opimal moneary policy insrumen in combinaion wih a SB rule achieve a beer oucome han an opimal moneary policy rule in combinaion wih a radiional BB rule. To assess he performance of he differen rules we simulaed sochasic shocks for periods allowing he opimal moneary policy rule for each case o adjus endogenously. We repea his 3 imes obaining he average sandard deviaions across repeiions. The resuls from he simulaions are repored in Table 4. As we can observe, under he SB rule, sandard deviaions of he main macroeconomic variables in he model are lower han under he BB rule. In paricular, he oupu gap and headline inflaion show lower volailiy under he former one. Under he BB rule headline inflaion is around. per cen more volaile han under he SB rule. In urn, he oupu gap is 6.4 per cen more volaile. This is in line wih a more sable governmen spending, which is 75 per cen less volaile under he SB rule as compared o he BB rule. Regarding he moneary policy insrumen, i is.4 per cen less volaile under he SB rule, generaing less aggressive responses from he moneary auhoriy under he SB rule. Gasoline prices weigh on inflaion is around 4..

18 8 Ana María Aguilar and Claudia Ramírez Bulos Evaluaion of Sandard Deviaions and Loss Funcions Under Differen Fiscal Rules Table 4 Srucural Balance Rule Balanced Budge Loss funcion Headline inflaion Oupu gap Consumpion Governmen spending Invesmen Real exchange rae.7.75 Nominal ineres rae Real ineres rae Impulse-response funcions In his secion we presen he funcioning of he model under he wo differen fiscal policy alernaives presened in he previous secions, using impulse-response funcions. In order o see he main channels hrough which he model funcions we presen 3 shocks. The firs one is a radiional cos-push shock, he second one is an aggregae demand shock. Finally we analyze an oil price shock. Firs of all we show how he economy responds o a ypical cos-push shock (inflaion shock). A one sandard deviaion increase in he non-core inflaion leads o an increase in headline inflaion (Figure 5), which forces he cenral bank o raise ineres raes o counerac he effec. This increase in he ineres rae leads o a fall in privae consumpion of agens who have access o he financial markes, invesmen, and indirecly, in expors due o he appreciaion of he exchange rae. By reducing economic aciviy when he ineres rae increases, consumers who do no have access o financial markes are affeced since heir disposable income reduces and herefore heir level of consumpion. A he same ime ax revenues fall and spending under he various rules also decreases, alhough in differen magniudes. As expeced, governmen consumpion flucuaes more wih he balanced budge rule han wih he srucural balance rule. For he former case, since he shock generaes lower ax revenues, he governmen reduces is spending as he oupu gap decreases. Addiionally, spending has o fall more because he governmen mus mee is deb paymen duies sill complying wih a zero defici rule. Finally, under a srucural balance rule, since governmen spending reacs o changes in he level of he reference oupu and no o curren changes in he oupu level, he fall in spending is much lower. Nex, we presen a scenario where consumpion is subjec o a posiive shock. When he economy faces his ype of shock, as seen in Figure 6, i generaes demand pressures ha drive he gap o a level above is equilibrium, which in urn generaes inflaionary pressures ha are ransmied hrough he Phillips curve, making he moneary auhoriy o increase he nominal ineres rae so ha he real ineres rae rises enough o counerac he effec. By increasing he

19 IRF.s Non-core Inflaion Shock Headline Inflaion Oupu Gap Consumpion (Financial Markes)..4 Figure periods periods periods Governmen Spending Governmen Deb Toal Governmen Revenue periods periods periods Consumpion (Non-financial Markes) Ineres Rae Real Ineres Rae periods periods periods Real Exchange Rae Expors Impors periods periods periods Balanced Budge Srucural Balance Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 9

20 IRF.s Consumpion Shock Figure Headline Inflaion Oupu Gap Consumpion (Financial Markes) periods periods periods.5 Consumpion (Non-financial Markes). Ineres Rae.6 Real Ineres Rae periods periods periods Governmen Spending Governmen Deb Toal Governmen Revenue periods periods periods x 3 Real Exchange Rae Expors Impors Ana María Aguilar and Claudia Ramírez Bulos periods periods periods Balanced Budge Srucural Balance

21 8 6 4 IRF.s Oil Price Shock x 3 Headline Inflaion Oupu Gap x 3 Consumpion (Financial Markes). Figure periods periods periods Consumpion (Non-financial Markes) Ineres Rae x 3 Real Ineres Rae periods periods periods Governmen Spending Governmen Deb Toal Governmen Revenue periods periods periods x 3 Real Exchange Rae Expors Impors periods periods periods Balanced Budge Srucural Balance Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico

22 Increase of Gasoline Prices and Inflaion Headline Inflaion Oupu Gap Consumpion (Financial Markes) Figure periods periods periods Consumpion (Non-financial Markes) Ineres Rae Real Ineres Rae periods periods periods Governmen Spending Governmen Deb Toal Governmen Revenue periods periods periods Real Exchange Rae Expors Impors Ana María Aguilar and Claudia Ramírez Bulos periods periods periods Balanced Budge Srucural Balance

23 Oil-dependen Revenues and Macroeconomic Sabiliy Under Fiscal and Moneary Rules: An Analysis for Mexico 3 oupu gap, ax revenues increase as well, so he governmen can spend more, bu spending behavior will also depend on he ype of rule followed by he ax auhoriy. In he case of a zero defici, as revenues increase, he governmen can spend more and sill mee he rule. In addiion, he increase in ineres raes by he cenral bank o reduce inflaion, generaes an increase in he financial cos of he governmen and a fall in invesmen, consumpion and, herefore, he oupu gap ha feeds back he effec on spending hrough ax revenues. Addiionally i generaes an appreciaion of he exchange rae, causing an increase in impors and a decline in expors. Finally, he effec on spending under a srucural balance rule mainains he same spiri as he former rule bu generaing much smooher movemens in governmen spending because i only reacs o changes in he reference oupu, and a change in consumpion oday affecs i minimally. Finally, Figure 7 shows he response of he differen variables of he model o an increase in oil prices, bu no in inflaion since movemens in oil prices are no ransmied o movemens in fuel prices under he rule deermined by he governmen. This shock s immediae effec is refleced in he increased governmen revenue due o higher oil revenues, however he reacion of governmen spending depends on he ype of rule ha is being followed. 3 In general, due o he increase in governmen revenue, demand pressures are generaed. The cenral bank has o ac o conain hese pressures and increases he ineres rae on he shor-erm, which affecs boh inflaion and he exchange rae so ha inflaion begins o decline and he increase in he real ineres rae generaes an appreciaion of he real exchange rae, increasing impors and decreasing expors. Consumers and invesors, meanwhile, know ha given he increase in oil prices he governmen will have more money o spend, hence he oupu gap will increase, bu also know ha he moneary auhoriy will reac o hese pressures by ighening he ineres rae o offse his effec on inflaion. For his reason, consumpion, is affeced a he ime of he shock and gradually adjuss during he following periods afer he shock. Considering he balanced budge rule, given an increase of oil revenues, he governmen can increase is spending and sill comply he fiscal rule. This in urn generaes a fiscal policy ha exacerbaes he effec of he increase in he price of oil wih high volailiy. In he case of he srucural balance rule, governmen spending only reacs o changes in he reference price of oil, so an increase in oil prices generae minor effecs on he oupu gap and herefore on inflaion, which leads moneary policy o increase he ineres rae in a lower magniude compared o he balanced budge rule and herefore he effecs on he exchange rae, consumpion and invesmen, are lower as well. 5.3 Oil prices shocks In Mexico, as already menioned, movemens in oil prices are no ransmied o movemens in fuel prices since he governmen fix hese prices every year resuling in a very limied impac on inflaion. However, one of he main purposes of he governmen in recen years has been o close he gap beween domesic and inernaional fuel prices. This modificaion will imply ha oil price volailiy will no only affec governmen revenues volailiy bu also inflaion volailiy. I is imporan o menion ha he curren rule followed by he governmen o fix fuel prices is a very regressive subsidy ha disors he relaive prices of inpus and herefore generaes inefficien allocaion of resources. Even hough from he inflaion sabiliy perspecive, he energy prices rule may be convenien in he shor run since i absorbs he major par of energy prices volailiy, i could imply such high coss in erms of public finances ha in he medium erm i Noe ha his assumpion will be modified in he following excercise. 3 In Mexico, he increase in oil prices affecs governmen revenues by exracing more surplus o he res of world, since he sale of crude oil o oher counries does no direcly affec domesic consumers.