VIII. CHAPTER EIGHT: WHAT IS DEBT SUSTAINABILITY ANALYSIS?

Transcription

1 219 VIII. CHAPTER EIGHT: WHAT IS DEBT SUSTAINABILITY ANALYSIS? Deb susainabiliy analysis (DSA) asks if, under curren policies, a counry or a governmen will be able o service is debs in he medium and long run wihou renegoiaing or defauling, and wihou having o underake policy adjusmens ha are implausibly large economically and poliically. DSA frameworks provide an ineremporal consisency check by esing wheher macroeconomic plans are viable no only from a flow balance perspecive bu also from a sock balance poin of view. They may also help dissuade policymakers from pursuing policies ha deliver shor-erm benefis a he cos of creaing unsusainable debs in he fuure. In recen years, he IMF developed an approach o deb susainabiliy ha is now used in surveillance and lending decisions. 142 These DSAs help he IMF and policymakers assess he risks associaed wih shor-run macroeconomic forecass and he policies on which such forecass are based. A firs risk, as discussed in deail in Chapers 9 and 10, is ha projecions of exernal or public deb may no be always grounded on sufficienly conservaive assumpions. For insance, some IMF-suppored programs have been based on assumpions abou growh in expor volumes and prices ha proved o be opimisic, conribuing o excessive borrowing. A second key risk o he realism of forecass is he assumed pah of he real exchange rae. Counries may be able o susain relaively large socks of foreigncurrency denominaed deb hrough real exchange rae appreciaion over he medium erm. As discussed in Chaper 4, moreover, i may be reasonable o assume ha some counries will experience secular real appreciaion as an equilibrium phenomenon due o cach-up growh. While he assumpion of real appreciaion may be defended in some circumsances, experience in several counries ha underwen subsanial real depreciaions following crises suggess ha i is risky o base policies on he assumpion ha real appreciaion will coninue indefiniely. DSAs also allow policymakers o idenify he economic secors responsible for excessive deb accumulaion, be hey he naional governmen (as in a number of African counries in he 1990s), subnaional governmens and sae enerprises (as in some ransiion economies), or he privae secor (as in he Asian crisis counries). In many emerging marke counries, deb raios may be moderae and he main risk o susainabiliy may arise from liquidiy problems. In some cases, counries do no have sufficien liquidiy o cover mauring obligaions even when hey can be considered solven, i.e., have relaively low and declining exernal deb-o-gdp raios. Concerns abou liquidiy may arise, for insance, if he sovereign or privae secor needs o make large amorizaion paymens o crediors in he near fuure and foreign exchange or governmen revenues are insufficien. In such cases of emporary illiquidiy, much depends on he willingness of 142 See, Assessing Susainabiliy, IMF (2002, 2003).

2 220 crediors o mainain or increase heir exposure in he shor run. Marke confidence is a crucial ingredien, and he vulnerabiliy o confidence crisis needs o be evaluaed and addressed alongside long-erm susainabiliy. For low-income counries ha do no borrow from privae capial markes, he susainabiliy of he public deb is largely de-linked from he senimens of he marke. I depends, insead, on he willingness of official crediors and donors o coninue providing posiive ne ransfers hrough concessional loans and grans. 143 For low-income counries ha have high deb raios, solvency is more of a concern han liquidiy. DSAs allow a sudy of he exposure of he IMF and oher mulilaeral crediors o individual borrowers. Finally, DSAs are also useful o assess he impac of and response o powerful echnological and demographic changes ha consrain governmen policies in he long run. Fiscal DSAs help quanify he fiscal impac of populaion aging, immigraion, and oher long-run populaion changes. The IMF s DSA framework presened in Chapers 9 and 10 is a simple quaniaive model of he evoluion of (exernal or public) deb. I is based on he ineremporal accouning ideniy linking exernal or fiscal defici flows o he accumulaion of he corresponding deb socks over ime. In building any DSA framework, analyss mus make baseline assumpions abou he ime pahs of a number of macroeconomic variables real GDP growh, inflaion, ineres and exchange raes, budge and exernal debs and deficis. In he IMF s DSA framework, he choice of baseline is a judgmen made by he counry eam on he basis of consulaions wih counry auhoriies and oher IMF saff hrough he inernal review process. The baseline projecion is hen sress-esed by subjecing i o plausible macroeconomic shocks. To be useful, sress ess mus choose shocks of reasonable ype, size, and cross-correlaion. Wha plausible shocks are is a maer of judgmen and depends on he specifics of he counry s siuaion and oulook. As Chapers 9 and10 will explain, o make DSAs more sysemaic and disciplined, he sress ess in he IMF s framework are derived from he counry s pas hisory of shocks. Someimes he counry eams complemen he mechanical, hisory-based scenarios wih alernaive scenarios ha assume more adverse exernal condiions and/or subopimal policies. While hey are an exremely valuable ool, DSAs have cerain limiaions. For one hing, DSAs do no assign explici probabiliies o he likelihood of crises. While desirable in principle, probabilisic approaches are more difficul o implemen, especially for counries in which limied daa or rapid srucural change make i difficul o esimae hese probabiliies. A second limiaion of exising DSA approaches, including he IMF s, is ha hey absrac from second-round behavioral responses of economic agens o shocks. For example, shocks o GDP do no affec he governmen s ax collecions or spending plans. 143 For reamen of deb susainabiliy in low-income counries, see IMF,(2004, 2004a).

3 221 The lieraure has begun o incorporae such effecs by esimaing fiscal reacion funcions ha endogenize he economy s response o shocks (Celasun and ohers, 2006). A hird limiaion of DSA analyses is ha hey focus mainly on deb dynamics raher han hreshold levels of deb. DSAs regard deb pahs as susainable so long as he deb-o-gdp raio declines. While his is, in principle, correc in he sense ha i mees he ineremporal budge consrain, i may be problemaic o assume susainabiliy if he deb raio is sabilized a a high level. Clearly, sabilizing he deb-o-gdp raio a 30 percen is differen from sabilizing i a 90 percen. Some indusrial counries including Greece, Ialy, Belgium, and Japan have been able o susain deb-o-gdp raios ha exceed 100 percen for decades wihou having o pay high ineres raes. Developing or emerging marke economies, on he oher hand, ofen do no have such luxury. For example, in he case of Argenina during he pre-2001 crisis period, he deb-o-gdp raio was approximaely 50 percen, no high by inernaional sandards. 144 A fourh limiaion of DSAs is ha hey focus more on deb dynamics and less on liquidiy risk. As noed earlier, he risk ha mauring deb canno be refinanced may arise even in solven counries. The IMF s DSA emplaes provide gross financing needs and oher informaion on rollover risks. On he oher hand, a full evaluaion of hese risks requires more disaggregaed and higher frequency daa on he deb sock. Finally, a word of cauion is in order. Wih he benefi of 20/20 hindsigh, he pah o many of he recen crises and episodes of unsusainable deb dynamics now seems obvious. And ye mos observers failed o spo he iniial policy misakes ha evenually led o hese crises. Why is i so difficul o diagnose susainabiliy problems? No simple or sophisicaed model will be able o predic crises well in advance while avoiding false alarms. The main problem wih all approaches, including DSAs, is wo-fold: firs, changes in he exernal environmen are difficul o predic beyond a shor-erm horizon, and a single se of policies can resul in very differen oucomes depending on exernal evens; second, he reacion of domesic and foreign invesors and he public is difficul o gauge, especially when informaion is scarce, perhaps owing o lack of policy ransparency, and economic agens ac in herd-like manner. Economiss can a bes prepare DSAs or oher model scenarios and explore he circumsances under which crises or deb problems are more or less likely o develop. 144 IMF (2003) akes a closer look a how hese deb raios can be inerpreed, and concludes ha for emerging markes a oal deb above percen of GDP leads o sharply higher crisis probabiliies (he more so he more closed he counry is o foreign rade).

4 222 IX. CHAPTER NINE: FISCAL SUSTAINABILITY This secion examines public deb dynamics. Saring from he governmen s cash-flow consrain, i examines he facors affecing fiscal susainabiliy and shows how a sream of budge deficis can, over ime, lead o unsusainable public deb levels and heir macroeconomic consequences. Boh he closed and open economy cases are considered. 9.1 Deb Dynamics in a Closed Economy Consider firs an economy ha does no rade wih he res of he world. Denoe by Y he economy s real GDP in year and P he GDP deflaor. Nominal GDP is he produc P Y. P Le denoe he rae of increase in prices beween years -1 and, expressed as 1. P Y Similarly, le g denoe he real growh rae of oupu, expressed as g 1. Y Le M -1 denoe he sock of money a he end of year -1 and assume, for simpliciy, ha all ineres-bearing governmen deb has one-year mauriy. Denoe by D -1 he sock of oneperiod governmen bonds ousanding a he end of year -1.The average nominal ineres rae on governmen deb issued a -1 is i. The governmen s expendiure in year consiss of wo componens, non-ineres spending, denoed G, and ineres paymens on he deb, i D -1. Nex consider he governmen s cash-flow consrain in year. As a maer of accouning, governmen expendiure mus be financed by raising ax and nonax revenues ne of ransfers o he privae secor, denoed R, hrough money issuance, M M -1 (=M ), and by issuing ineres-bearing securiies, D D -1. G + i D -1 = R + (D -D -1 ) + (M -M -1 ). (9.1) The governmen s overall budge balance is he difference beween revenue and expendiure, R (G + id -1 ). The primary budge balance, PB, is he difference beween revenue and nonineres expendiure, R - G. As we are ineresed in he evoluion of he sock of ineresbearing public deb, we solve (9.1) for D, yielding D 1 i D PB M. (9.2) 1 To derive an expression for he sock of public deb in relaion o GDP, we divide equaion (9.2) by nominal GDP: 1 1

5 223 D 1 i D 1 PB M PY PY PY PY i D PB M. 1 g P 1Y1 PY PY (9.3) Denoe by lower-case leers he sock of deb, primary balance, and seignorage expressed as shares of GDP: d D / PY, d 1 D 1/ P 1Y1, pb PB / PY, and M / PY. The parameer muliplying d 1, denoed, is key in deb susainabiliy analysis. Use he Fisher equaion linking he nominal and real ineres rae, 1 r (1 i) /(1 ), o wrie as he raio of one plus he real rae of ineres on governmen deb over one plus he real rae of GDP growh: 1 i /[ 1g 1 ] 1 r / 1 g. (9.4) Wih his noaion, he governmen budge consrain can now be rewrien as: 1. d d pb (9.5) We can draw equaion (9.5) in a phase diagram as shown in Figure 9.1 o examine how he deb-o-gdp raio evolves over ime. The horizonal axis plos he deb-o-gdp raio in year -1, d 1, while he verical axis shows he resuling value of d in year. The 45 line shows deb-o-gdp raios ha do no change over ime. Suppose, for simpliciy, ha he parameers, pb, and are consan over ime a, pb, and, respecively, so ha d and d 1 have a linear relaionship. Wheher he public deb-o-gdp raio is explosive or no depends on he value of he parameer. The non-explosive case 1 is shown on he lef-hand side panel of Figure 9.1. In his case, he iniial level of deb-o-gdp raio d 0 evenually falls o * d and says a ha level forever. The explosive deb case 1 is shown on he righ-hand side panel of Figure 9.1. Here, he real ineres rae r which he governmen pays on is deb exceeds he real GDP growh rae g. Saring from any posiive iniial level of deb-o-gdp raio d 0 > d* in year 0, he deb o GDP raio grows wihou bound, which is obviously unsusainable.

6 224 Figure 9.1 Deb Dynamics a. Sable deb dynamics b. Explosive deb dynamics d 45 o line d 45 o line d = d -1 (pb +) where < 1 or r < g d = d -1 (pb +) where < 1 or r < g (pb+) d * d 0 d (pb+) d * d 0 d The speed a which deb can explode in realisic cases is surprisingly fas. Suppose he public deb-o-gdp raio is iniially d 0 = 50 percen. Assume a nominal ineres rae, i =14 percen, real GDP growh rae g = 4.0 percen, annual inflaion = 4.3 percen, primary defici pb = -2.7 percen of GDP, and seignorage, = 1.1 percen of GDP. Applying he Fisher equaion, he real ineres rae is 9.3 percen (= (1.14/ ) 100 percen), which exceeds real GDP growh, implying 1. The deb-o-gdp raio is explosive (see Figure 9.2) and reaches 80 percen of GDP someimes considered he hreshold for severe indebedness in abou five years. Figure 9.2 The Deb-o-GDP Raio in a Closed Economy 200% 180% 160% 140% 120% 100% 80% 60% 40% 20% 0%

7 225 The explosive naure of he governmen s deb dynamics can also been seen by differencing equaion (9.5) o calculae he change in he deb-o-gdp raio, d d d 1. Subracing d -1 from boh sides of equaion (9.5) yields he following 1 d d pb. (9.6) 1 Equaion (9.6) underscores he facors ha affec he change in he deb-o-gdp raio: he size of he primary budge balance pb, seignorage, and he buil-in momenum of deb, 1d 1. If he real ineres rae on governmen deb exceeds real GDP growh, deb becomes explosive. Primary surpluses are hen needed o offse he auomaic deb dynamics. The size of he primary surplus in relaion o GDP, pb, is a good indicaor of he governmen s fiscal adjusmen effor. 145 Equaion (9.6) is useful in calculaing he primary surpluses needed o achieve specific objecives, such as sabilizing he deb a is exising level or even reducing i o a lower level, as needed, for example, o mee he crieria of he Maasrich Treay for European Union member counries. As a firs sep o fiscal susainabiliy, he auhoriies may pick fiscal arges wih a view o hal furher increases in he public deb o GDP raio. This requires raising he primary balance o GDP raio sufficienly o sabilize he deb-o-gdp raio. To obain he debsabilizing primary balance, se d = 0 in equaion (9.6) o obain: 1 d 1. pb (9.7) Coninuing wih our earlier example, if he counry is o avoid he ever-rising deb pah shown in Figure 9.2, he primary balance surplus needs o be a leas 1.45 percen of GDP {[(( )/1.04) 0.5] 0.011} 100 percen insead of 2.7 percen of GDP in defici. The deb-sabilizing primary balance depends on several facors. Firs, if he exising level of deb is large, large primary surpluses are needed o preven i from growing furher. Second, if he difference beween he real ineres rae and real GDP growh is large, hen he primary surplus also needs o be large. Third, if seignorage or oher sources of governmen finance are available (such as privaizaion receips), hese can be used o pay off he deb and will resul in lower deb-sabilizing values for he primary surplus. Of course, many counries likely would like o reduce heir sock of deb relaive o GDP, raher han jus sabilize i. Those counries mus hen achieve a primary surplus in excess of he deb-sabilizing level. 145 The governmen can manipulae he money growh rae o increase revenue from money creaion, or seignorage. Bu raising money growh and inflaion leads o currency subsiuion, which places limis on he amoun of real resources he governmen can obain from seignorage.

8 Deb Dynamics in an Open Economy The analysis of public deb susainabiliy is similar when he governmen can borrow from inernaional financial markes o cover par of is budge defici. Under hese condiions, public deb susainabiliy depends on he pah of he nominal and real exchange rae and foreign ineres raes. When he governmen borrows abroad, a disincion needs o be made beween domesic h f currency-denominaed deb D and foreign-currency denominaed deb D. Leing e be he nominal exchange rae (local currency per uni of foreign currency), he deb sock is h f D D e D and he governmen budge consrain can be wrien * 1 D 1 i D PB M. (9.8) In equaion (9.8), i *, he effecive nominal ineres rae, is a weighed sum of he domesic and foreign ineres raes i h and f where ed / f i, and also depends on he exchange rae h f f i 1 i i 1 i, (9.9) * D is he porion of foreign currency denominaed deb, and is he rae of depreciaion of he currency. I can be shown ha he public deb o GDP raio evolves according o he following equaion, which is analogous o (9.5): * 1 * * * In equaion (9.10), i g d d pb, (9.10) 1 /[ 1 1 ] is analogous o, and h depends on domesic inflaion, foreign inflaion f h f f * *, he GDP deflaor,, and exchange rae movemens: 1 1, (9.11) f f where ep Y / PY is he oupu share of radables in GDP. The inuiion discussed in he closed economy case sill holds: Deb dynamics are explosive * * * if he real ineres rae r 1 i / 1 1 is greaer han real GDP growh g. In he open economy he ineres rae relevan for he DSA calculaion depends on domesic and foreign ineres raes and inflaion, on exchange rae movemens, and on he size of foreign borrowing and foreign rade. In erms of our earlier example, suppose i d =14 percen, f i =8 percen, =0.5, =0, and =0. Then he effecive nominal ineres rae i * is 11 percen = ( percen + 0.5

9 227 * 8 percen) ), and he effecive real ineres rae r is 6.4 percen (= (1.11/ ) 100 percen), which is greaer han he real GDP growh rae of 4.0 percen. As in he closed economy case, he deb-o-gdp raio is explosive (see Figure 9.3). Moreover, if he exchange rae depreciaes by 30 percen, he effecive nominal ineres rae and he effecive real ineres rae become as high as 27.2 percen and 22.0 percen. The debo-gdp raio rises much more rapidly and exceeds he 80 percen hreshold in less han 5 years, assuming a crisis does no force an adjusmen firs (see Figure 9.3). The deb sabilizing primary balance in his case rises o 7.6 percen of GDP (= (22.0 percen 4.0 percen)/ percen). 240% 220% 200% 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% Figure 9.3 The Deb-o-GDP Raio in an Open Economy d d* wih 30% depreciaion d* wih no depreciaion The IMF s Approach o Public Deb Susainabiliy Basic macroeconomic assumpions In his secion, we focus on public DSAs relevan for counries wih access o inernaional capial markes. The fiscal DSA framework consiss of a baseline scenario and sensiiviy ess of deb dynamics o a number of assumpions There are many difficulies in consrucing realisic projecions of public deb and deb service. In paricular, hree imporan risks need o be assessed. A firs risk comes from coningen liabiliies (Box 9.1). Many coningen liabiliies, by naure, go unnoiced in normal imes bu are more likely o emerge in crises. Coningen liabiliies are exceedingly difficul o measure in pracice, boh because he amouns involved are ofen unknown and because he precise circumsances under which hey would urn ino acual liabiliies are ofen unknowable.

10 228 Box 9.1 Coningen Liabiliies The governmen s coningen liabiliies are poenial claims on he governmen ha may or may no be incurred depending on macroeconomic condiions and oher evens. Unlike direc liabiliies, such as pension obligaions, which are predicable and will arise in he fuure wih cerainy, coningen liabiliies are obligaions riggered by discree bu uncerain evens. By naure, coningen liabiliies are difficul o measure. While informaion is usually available on deb formally guaraneed by he cenral governmen, deb no explicily guaraneed has ofen been an imporan conribuor o public deb buildup. Coningen liabiliies, especially hose arising from he need o rescue banks, were responsible for large jumps in he public deb o GDP raio in several counries affeced by pas financial crises. Capuring he hidden fiscal risks arising from coningen liabiliies is herefore an imporan ask for public DSAs. One of he sress ess in he public deb susainabiliy emplae examines he effec on he public deb dynamics of he realizaion of coningen liabiliies, specified as an exogenous increase in he deb raio of 10 percen of GDP. This shock is exogenous and is no linked o he counry s financial secor vulnerabiliies and oher shocks examined in he emplae (e.g., o growh, ineres raes, or he exchange rae). Explici liabiliies are hose recognized by a law or conrac, such as governmen guaranees for nonsovereign borrowing and obligaions issued by subnaional governmens and public or privae secor eniies or rade and exchange rae guaranees. Implici liabiliies are obligaions ha may be assumed by governmen due o public and ineres-group pressures, such as financial secor bailous, or bailous of non-guaraneed social insurance funds. A second risk is an abrup change in financing condiions in inernaional markes affecing boh he availabiliy and he cos of funds. Such changes may reflec developmens in he financial markes, such as conagion, or funding difficulies specific o he counry. These changes may give rise o a liquidiy crisis if he counry is unable o rollover is mauring obligaions or resul in sharply higher ineres raes, calling ino quesion he long-erm solvency of he borrower. A hird risk is a depreciaion of he exchange rae, possibly in he afermah of he collapse of an exchange rae peg, which increases he domesic currency value of he sock of exernal public deb. A key facor in deermining he pos-crisis evoluion of he exchange rae is he exen of iniial overvaluaion and he exen of possible exchange rae overshooing. As some cases have shown, once a crisis erups, he capial ouflows can resul in exchange rae adjusmens far in excess of any iniial esimaes of overvaluaion. To sress es he baseline projecions agains hese and oher risks, he IMF DSA: calibraes he size of shocks (reasonable bu no exreme; use hisorical sandard deviaions or absolue deviaions for global shocks), assesses inerdependencies (perurb correlaed parameers a he same ime), ses duraions of shocks (use shock sequences for serially correlaed parameers), and

11 229 assesses he effec of oher deb-creaing flows (e.g., coningen liabiliies). 146 The public secor DSA emplae The public secor deb emplae racks he behavior of he gross deb-o-gdp raio shown in equaion (9.9). The definiion of deb used in he IMF s DSA is based on gross liabiliies ha is, public secor liquid or oher asses are no need ou. The coverage of public deb is as broad as possible and i includes public enerprises as well as local governmens. Based on equaion (9.9), he emplae idenifies he differen channels ha conribue o he evoluion of he deb o GDP raio, including he primary defici and endogenous/auomaic facors relaed o ineres raes, growh raes and exchange rae changes. The emplae also includes oher deb-creaing operaions, such as would resul from he recogniion by he governmen of coningen liabiliies, as well as deb-reducing operaions, such as privaizaions whose proceeds are used o pay down public deb. The gross financing needs of he public secor are defined as he sum of he public secor defici and all deb mauring over he following 12 monhs. The emplae also calculaes he deb-sabilizing primary balance which would be needed o keep he deb-o-gdp raio consan if all he variables in he deb dynamics equaion remained a he level repored in he las year of he projecion. As discussed in Chaper 8, in he IMF s public deb susainabiliy framework, he baseline pahs of he public deb-o-gdp raio and he variables on which i depends are projeced by IMF saff in consulaion wih counry auhoriies. The baseline projecions are condiional in he sense ha hey assume ha he auhoriies will fully implemen he announced fiscal, moneary, exchange rae, and srucural policies. In addiion, he public deb susainabiliy emplae presens projecions under a hisorical scenario. This is an alernaive pah of he deb raio, consruced under he assumpion ha all key variables say a heir hisorical averages hroughou he projecion period. This scenario is a es of he realism of baseline projecions: if he deviaions of assumed policies and macroeconomic developmens in he baseline are very differen from hose in he hisorical scenario, hese will need o be jusified by referring o credible changes in policies. The emplae also conains a no-policy-change scenario. This is derived under he assumpion ha he primary balance is consan in he fuure and equal o he projecion for he curren year. The no-policy-change scenario can be modified o assume an unchanged cyclically 146 In 2005, he IMF reviewed is DSA framework and revised he size and duraion of he shocks used in he sress ess. The new approach considers he effecs of smaller bu more persisen shocks.

12 230 adjused primary posiion, or o make adjusmens for he expiraion of one-off measures, as necessary. The baseline scenario is also sress-esed using differen assumpions on key parameers. Permanen shocks equal o one-half sandard deviaion are applied o he baseline projecions of each of he parameers, and pahs of deb raios are hen derived. One-quarer sandard deviaion shocks are applied in he combined shock es. These shocks are applied o he ineres rae, growh rae, and primary balance. In addiion, he emplae examines he deb rajecory in he case of a 30 percen depreciaion of he local currency and a coningen liabiliies shock of 10 percen of GDP. The laer is presened as a rough measuremen of an increase in deb-creaing flows, given he difficulies in discussing coningen liabiliies risk. If beer measures are available, he saff is encouraged o use hem in sress ess. Table 9.1 liss he daa inpus needed o calculae he deb-o-gdp raio in he IMF s DSA. Table 9.1 Daa Inpu Requiremens for DSA Fiscal Variables Public secor deb, Public secor balance, Public secor expendiure, Public secor ineres expendiure, Public secor revenue (and grans), Foreign-currency denominaed deb (expressed in local currency), Amorizaion on medium- and longerm public secor deb, Shor-erm public secor deb, Ineres paymens on foreign deb Macroeconomic Variables Nominal GDP, Real GDP, Exchange rae, naional currency per U.S. dollar, end of period, Exchange rae, naional currency per U.S. dollar, period average, GDP deflaor I is also desirable o have daa on privaizaion receips, recogniion of implici or coningen liabiliies, and oher liabiliies (e.g., bank recapializaion). While his daa is much harder o collec, i grealy improves he qualiy of he baseline projecion and he sress ess. Once inpu daa are filled in, he baseline and sress es resuls are auomaically calibraed and presened in a summary able and in chars represening he oucomes of he sress ess, also known as bound ess. See Table 9.2 and Figure 9.4 for an example. Table 9.2 summarizes he baseline scenario. Lines 1 and 2 show how he deb-o-gdp raio evolves over ime. The key macroeconomic assumpions underlying he baseline are repored a he boom of he able. The differen channels ha conribue o he evoluion of he debo-gdp raio are: he primary defici (line 4), he auomaic deb dynamics (line 7), and oher

13 231 idenified deb-creaing flows (line 12), which include privaizaion receips, recogniion of implici or coningen liabiliies, and oher obligaions such as bank recapializaion. These flows are assumed zero in his paricular example. The auomaic deb dynamics, in urn, is broken down ino conribuions from he real ineres rae, real GDP growh, and exchange rae. This decomposiion allows an assessmen of he imporance of differen facors in he buildup of public deb and also serves as he basis for sress ess, he resuls of which are summarized ogeher wih he baseline projecions in Figure 9.3. Changes in gross deb arising from oher below-he-line operaions, such as repaymen of deb financed by a reducion in financial asses, and cross-currency movemens are included in a residual (line 16). I is criical o monior he behavior of his residual, as i may highligh errors in implemening he approach. A large residual may, in paricular, signal a breach of he flow-sock ideniy linking he defici o changes in deb. The residual should be small unless i can be explained by specific facors. The gross financing needs of he public secor, in percen of GDP and in billions of dollars, are also calculaed. Table 9.2 also repors he pahs of deb o GDP raio under he hisorical scenario and under he no-policy-change scenario. These scenarios es he realism of he baseline scenario. Finally, he emplae also calculaes he deb-sabilizing primary balance (las column of Table 9.2).

14 232 Table 9.2 Counry: Public Secor Deb Susainabiliy Framework, Acual Projecions Deb-sabilizing primary balance 9/ 1 Baseline: Public secor deb 1/ o/w foreign-currency denominaed Change in public secor deb Idenified deb-creaing flows (4+7+12) Primary defici Revenue and grans Primary (nonineres) expendiure Auomaic deb dynamics 2/ Conribuion from ineres rae/growh differenial 3/ Of which conribuion from real ineres rae Of which conribuion from real GDP growh Conribuion from exchange rae depreciaion 4/ Oher idenified deb-creaing flows Privaizaion receips (negaive) Recogniion of implici or coningen liabiliies Oher (specify, e.g. bank recapializaion) Residual, including asse changes (2-3) 5/ Public secor deb-o-revenue raio 1/ Gross financing need 6/ in billions of U.S. dollars Scenario wih key variables a heir hisorical averages 7/ Scenario wih no policy change (consan primary balance) in Key Macroeconomic and Fiscal Assumpions Underlying Baseline Table 9.2. Counry: Public Secor Deb Susainabiliy Framework, (In percen of GDP, unless oherwise indicaed) Real GDP growh (in percen) Average nominal ineres rae on public deb (in percen) 8/ Average real ineres rae (nominal rae minus change in GDP deflaor, in percen) Nominal appreciaion (increase in US dollar value of local currency, in percen) Inflaion rae (GDP deflaor, in percen) Growh of real primary spending (deflaed by GDP deflaor, in percen) Primary defici / Indicae coverage of public secor, e.g., general governmen or nonfinancial public secor. Also wheher ne or gross deb is used. 2/ Derived as [(r - g - g + r]/(1+g++g)) imes previous period deb raio, wih r = ineres rae; = growh rae of GDP deflaor; g = real GDP growh rae; = share of foreign-currency denominaed deb; and = nominal exchange rae depreciaion (measured by increase in local currency value of U.S. dollar). 3/ The real ineres rae conribuion is derived from he denominaor in foonoe 2/ as r - π (1+g) and he real growh conribuion as -g. 4/ The exchange rae conribuion is derived from he numeraor in foonoe 2/ as (1+r). 5/ For projecions, his line includes exchange rae changes. 6/ Defined as public secor defici, plus amorizaion of medium and long-erm public secor deb, plus shor-erm deb a end of previous period. 7/ The key variables include real GDP growh; real ineres rae; and primary balance in percen of GDP. 8/ Derived as nominal ineres expendiure divided by previous period deb sock. 9/ Assumes ha key variables (real GDP growh, real ineres rae, and oher idenified deb-creaing flows) remain a he level of he las projecion year.

15 233 Figure 9.4 Counry: Public Deb Susainabiliy: Bound Tes 1 (Public deb in percen of GDP)

16 234 X. CHAPTER TEN: EXTERNAL SUSTAINABILITY Foreign financial resources can be imporan o growing economies, as hey supplemen domesic savings o finance invesmen, help smooh income flucuaions and, in he case of direc foreign invesmen, faciliae echnology ransfer. Access o foreign finance, however, has in he pas also led a number of counries o accumulae unsusainable foreign debs, which hey were unable o honor in full. Alhough excessive deb obligaions can be renegoiaed wih crediors in principle, he process is neiher smooh nor cosless in pracice, no leas because crediors are muliple and fragmened. Someimes an iniial round of deb reducion needs o be followed by furher deb forgiveness. In he meanime, he counry ypically loses access o foreign financing for a susained period, is currency depreciaes srongly in nominal and real erms, and impors and oher foreign spending are compressed. In some insances, as discussed in Chaper 6, balance shee effecs lead o insolvency of domesic firms and deposior runs agains he banking sysem. The effecs of financial convulsions on he real economy may be severe, wih inflaion, ineres raes, and unemploymen spiking up and oupu conracing. Unsusainable foreign debs are hus cosly o a counry and disrup he smooh funcioning of inernaional capial markes. However, while here is a large payoff o prevening hese siuaions, idenifying dangerous imbalances and correcing hem as hey are building up has proven difficul. The purpose of exernal deb susainabiliy analysis (DSA) is o help policymakers in hese endeavors. This chaper presens he analyical and operaional consideraions relevan o he analysis of exernal susainabiliy The IMF s Exernal Deb Susainabiliy Framework Exernal susainabiliy requires a counry o be able o fully service is ousanding deb o foreign residens boh in he shor run and in he long run. To be viewed as susainable, deb should be serviceable wihou assuming unrealisically large policy correcions in he fuure. In conras wih fiscal susainabiliy, which focuses on deb owed by he naional governmen o eiher domesic or foreign residens, exernal deb susainabiliy considers he oal indebedness of he economy (including deb of he governmen, he financial secor, he nonfinancial corporae secor, and households) vis-à-vis foreign residens. To assess exernal susainabiliy, DSAs evaluae he pah of a counry s exernal deb sock over ime in relaion o GDP, expors, or some oher indicaor of capaciy o repay exernal deb. Susainabiliy requires ha hese raios sabilize a reasonable levels evenually by he end of he projecion period a he laes and ha hey do no become explosive hereafer. The ime pah of he exernal deb-o-gdp raio depends on domesic macroeconomic

17 235 condiions, he counry s macroeconomic and srucural policies, and on global rade and capial marke condiions. If, under reasonable macroeconomic assumpions, he raio of exernal deb o GDP or expors does no sabilize a a pruden level, his will raise alarm bells abou he abiliy of he counry o service is foreign deb in he fuure. The IMF s assessmens of exernal susainabiliy proceed in wo seps. Firs, he IMF eam working on a counry, in cooperaion wih counry auhoriies, makes projecions of he pah of policy and endogenous variables. The projecions use he ineremporal budge consrain ha links he exernal deb flows needed o finance a counry s curren accoun deficis o he sock of exernal deb a he end of he projecion period. A key objecive of he DSA is o deermine he pah of he exernal deb-o-gdp raio during he projecion period. These projecions involve many judgmens abou macroeconomic developmens, such as economic growh, inflaion, nominal and real ineres raes, and exchange raes; he porion of he curren accoun defici ha can be financed hrough non-deb flows; and he exen of official inflows. These macroeconomic assumpions are no sandardized bu are adaped o he circumsances of he counry. They mus be scruinized o ensure ha hey are inernally consisen and conform o developmens in he inernaional economy, including likely growh raes and inflaion of rading parners and inernaional ineres raes. As in he case of fiscal DSA, sensiiviy ess are performed on he baseline o examine he effecs on he exernal deb profile of alernaive assumpions abou he ime pahs of key variables The Exernal Susainabiliy Templae The emplae summarizes he DSA s baseline assumpions and is implicaions for exernal deb dynamics. 147 The saring poin is he accouning ideniy linking he economy s ransacions wih he res of he world (see Chaper 4). Le TB be he sum of he non-ineres curren accoun balance, NITB, and non-deb generaing capial inflows. Also le D denoe he sock of exernal deb a he end of year and i w he nominal effecive foreign-currency ineres rae he counry pays on is exernal deb. The increase in he sock of exernal deb over ime is D D i D TB, (10.1) w 1 1 which is equivalen o 147 Exernal deb obligaions should include public secor exernal deb, non-financial privae exernal deb, and financial secor exernal deb.

18 236 w 1 1 D i D TB. (10.2) I is useful o express he sock of a counry s exernal liabiliies in relaion o GDP. This requires ha we obain he foreign currency value of he counry s GDP where for purposes of illusraion, we assume ha he foreign currency in which deb is denominaed is he U.S. dollar. Le P denoe he GDP deflaor, Y denoe GDP, and le e define he exchange rae in unis of domesic currency per U.S. dollar. Then GDP in U.S. dollar erms can be wrien as PY/e. Dividing boh sides of (10.1) by PY/e yields he exernal deb-o-gdp raio. The counry s exernal deb dynamics becomes D PY / e w D P 1Y 1 / e 1 1 i P 1 Y P X P x m 1, (10.3) 1 / e 1 PY / e PY / e PY / e X M where P is he price of expors, X is expors, P is he price of impors in foreign currency, and M is impors. The above equaion can also be expressed as w 1i 1 1 1g d d b, (10.4) 1 M OI where d is he deb-o-gdp raio, π is he growh rae in he GDP deflaor, g is he real GDP growh rae, ρ is he rae of nominal exchange rae appreciaion, and b is he deb-creaing componen of he balance on goods and non-ineres services in percen of GDP. The baseline medium-erm projecion of exernal susainabiliy is obained by exending his equaion o projec he growh raes and balance of paymens several years ino he fuure. To compue he evoluion of he deb o GDP raio, we need saring values for he iniial sock of public and privae exernal deb, is mauriy profile and schedule of ineres paymens. To compue fuure ineres paymens, an esimae of fuure exernal ineres raes mus be made. The sandard pracice is o assume one ineres rae ha applies o boh public and privae exernal deb. An alernaive approach would be o use a separae ineres rae for he public and privae secor and inerpre he exernal ineres rae as he weighed average exernal ineres rae. Forecass of growh raes of real expors and impors, along wih forecass of heir relevan nominal price growh in foreign currency are needed o compue he relaive conribuion of he rade balance o exernal resource needs. Finally, o compue exernal deb o GDP raios, we need forecass of he pah of real GDP growh, he GDP deflaor, and he nominal exchange rae. Table 10.1 provides an example of he sandard emplae for exernal susainabiliy analysis.

19 237 The baseline scenario for exernal susainabiliy should be consruced wih a reasonable se of forecas variables. In oher words, he baseline scenario should no achieve susainabiliy by assuming abnormally high growh raes, abnormally low ineres raes, or unreasonable raes of appreciaion. The economic performance of he counry during he las five o en years is imporan in making realisic assumpions. The underlying assumpions should be ransparen, and opimism or pessimism can be incorporaed by subjecing he baseline projecion o a se of alernaive assumpions. A separae program scenario ha includes an acive policy response can hen be consruced in a separae sep, allowing he auhoriies o evaluae susainabiliy under acive and passive policy sances. Deerminans of exernal deb accumulaion Equaion (10.4) is composed of wo pars. The firs componen is he effec of changes in economic variables on he exising exernal deb-o-gdp raio. This componen is he auomaic deb dynamics since he changes in he economic variables are auomaically applied o he pre-exising sock of exernal deb. The exernal deb-o-gdp raio rises if he nominal exernal ineres rae rises or if he domesic currency depreciaes vis-à-vis foreign currencies. An increase in ineres raes causes deb service coss o rise, some of which may be rolled over ino addiional new deb. A depreciaion reduces he foreign-currency value of domesic GDP. However, increases in he growh rae of he GDP deflaor and/or real GDP iself cause he exernal deb-o-gdp raio o decline. The second componen of exernal deb dynamics is he deb-creaing componen of he balance on goods and non-ineres services. In he exernal susainabiliy emplae, he saring poin is he curren accoun defici, excluding ineres paymens. The curren accoun defici, excluding ineres paymens, is hen adjused by he level of ne non-deb creaing capial inflows from he balance of paymens. The non-deb-creaing capial flows are ne foreign direc invesmen and oher ne equiy invesmen by firms and households. Ne posiive inflows on non-deb creaing capial flows reduce he need for exernal resources and are, herefore, recorded as a negaive value in he emplae. The evoluion of exernal deb over ime is influenced by many facors, including decisions of he governmen and he privae secor. Governmen can clearly conrol is own rae of exernal deb accumulaion. Bu is policies mus also arge acions and expecaions of he privae secor and aim a overall economic sabiliy. The exernal balance, for insance, is affeced by he privae secor s demand for impors and he res of he world s demand for expors, he real exchange rae, compeiiveness consideraions, and domesic and foreign income and demand. The evoluion of exernal deb is also influenced by he volume of nondeb inflows, principally FDI and equiy invesmen. The volume of hese inflows is deermined by inernaional invesors who look a he marginal produciviy of domesic projecs relaive o he reurn available in oher markes, he counry s business climae, and oher consideraions. Thus, several secors joinly help deermine he size of he curren

20 238 accoun balance and he level and composiion of capial inflows ha finance i. A loss of confidence ha resuls in rising ineres raes and exchange rae depreciaion could negaively influence he exernal deb dynamics of he economy. Exernal deb sabilizaion and hreshold levels of deb As in he analysis of fiscal susainabiliy (Chaper 9), we may calculae he non-ineres primary balance needed o sabilize he exernal deb o GDP raio. Seing d =d -1 in (10.4) yields b w 1i 1 1 1g 1d 1, (10.5) which is he level of non-ineres CAB and non-deb generaing inflows needed o keep he exernal deb o GDP raio from rising. Improvemens in he public finances can help improve he exernal curren accoun and arres he accumulaion of exernal deb. These mus normally be complemened by srucural measures and financial secor reforms, wih he aim of improving he supply side of he economy, raising he efficiency of inermediaion, deepening local capial markes, and aracing more non-deb inflows. Deeper local capial markes increase he amoun of domesic currency financing available and help reduce he dependence on foreign currency finance. Improving he currency composiion of public and privae secor balance shees and increasing he ypes of securiies available, including derivaives and opions, allows privae secor borrowers o beer hedge currency flucuaions. Policy may need o do more han sabilize exernal deb raios. The auhoriies may arge a safe level of exernal deb and increase inernaional reserves and fiscal cushions o deal wih liquidiy shocks. They may also improve he srucure of he public secor s exernal deb profile, eiher by subsiuing domesic currency borrowing for exernal borrowing or by lenghening he mauriy of exernal deb. If policy is no successful in sabilizing he exernal deb dynamics, hen he public secor may need o resrucure he deb profile o resore susainabiliy. Sress ess In addiion o he baseline projecion, he sandard framework includes a se of sandard sensiiviy ess. These sress ess examine he implicaions of alernaive assumpions abou policy variables, macroeconomic developmens, and coss of financing. The firs sensiiviy es deails he ambiiousness of he baseline projecion relaive o hisorical experience. As in he fiscal DSA, key variables he rae of growh, ineres rae,

21 239 ec. are se o heir hisorical averages o es wheher he baseline is consisen wih he counry s hisorical norm. The oher sensiiviy ess include: a wo-sandard deviaion adverse shock lasing wo years o each of he key parameers; and a one-sandard deviaion combined shock. These sandard deviaions are compued using hisorical daa for he counry. The combined shock assumes ha each of he variables moves agains exernal deb susainabiliy (e.g., ineres rae increases and exchange rae depreciaions simulaneously). The combined shock is also repeaed using sandard deviaions obained from cross-counry sudies as a robusness check. Finally, counry specific shocks can be ailored o specific feaures of he economy. For example, in a counry ha has a fixed exchange rae regime, he volailiy of he exchange rae may have been low hisorically. This would jusify a scenario in which here is a large exchange rae depreciaion. An imporan consideraion in he implemenaion of DSAs concerns he lengh of he hisorical series used o compue he hisorical averages and sandard deviaions. Long horizons help guard agains excessive euphoria or excessive pessimism abou, say, growh prospecs. Conversely, srucural change may limi he relevance of he disan pas. The IMF DSA framework normally uses he previous en years o calculae averages and sandard deviaions. A five-year period is used in counries ha experienced srucural changes, such as ransiion from cenral planning, or large shocks, such as hyperinflaion or currency crisis. Finally, if he daa available for a counry is oo shor, cross-counry parameers can be subsiued insead. Anoher key judgmen in DSAs is he lengh and serial correlaion properies of he shocks. Low growh in one year may signal a recession, suggesing a posiive serial correlaion in he level of oupu, alhough i is unlikely ha here would be repeaed negaive shocks o oupu growh. The sandard emplae for exernal susainabiliy conducs sensiiviy ess based on a wo-year sequence of shocks followed by a reurn o he mean growh raes for he res of he projecion period. The counry-specific shocks can be modified o consider shorer or longer inervals An Example: Mexico The exernal DSA emplae was applied recenly o Mexico (see Table 10.1). In he baseline, he exernal deb-o-gdp raio was projeced o decline from 26.1 percen in 2003 o 23 percen in Despie a sligh moderaion in GDP growh assumed in he baseline, he sound public finances and healhy non-deb creaing capial inflows were expeced o cover Mexico s exernal resource needs. Hence, Mexico s exernal deb posiion appeared sound under he baseline.

22 240 Table 10.1 also describes he resuls of sress ess o he baseline. The firs alernaive scenario (Table 10.1, A1) uses he en-year hisorical average for he key macroeconomic variables. The second alernaive scenario (Table 10.1, A2) uses a counry-specific shock in his case a negaive shock o oil prices. In addiion, he sress ess consider wo-sandard deviaion shocks o ineres raes, GDP growh, inflaion, and he curren accoun ha las for wo periods (Table 10.1, B1-B4). Finally, a one-sandard deviaion shock o he combined variables combined wih a one-ime, one-year, 30 percen nominal depreciaion were considered (Table 10.1, B5 and B6). Mexico s exernal vulnerabiliy remains low in he higher ineres rae-lower oil price scenario. The exernal deb-o-gdp raio remains in he percen of GDP range hroughou he forecas period. The oucome of he sress ess is differen under an exreme combinaion of shocks, namely a peso depreciaion by wo sandard deviaions, or 24 percen in 2005 and 2006, and a mix of higher nominal ineres raes, lower GDP growh raes, and exchange rae depreciaion by one sandard deviaion. In his scenario, Mexico s exernal siuaion would worsen dramaically. Exernal deb would approach 40 percen of GDP.

23 241 Table Mexico: Exernal Deb Susainabiliy Framework, (in percen of GDP, unless oherwise indicaed)

24 242 REFERENCES Chaper 8 Celasun, Oya, Xavier Debrun, and Jonahan Osry, 2006, Primary Surplus Behavior and Risks o Fiscal Susainabiliy in Emerging Marke Counries: A 'Fan-Char' Approach, IMF Working Paper 06/67 (Washingon: Inernaional Moneary Fund). Inernaional Moneary Fund, 2002, Assessing Susainabiliy, SM/02/166. Available via he inerne: hp:// 2003, Susainabiliy Assessmens Review of Applicaion and Mehodological Refinemens, SM/03/206. Available via he inerne: hp:// 2004, Deb Susainabiliy in Low-Income Counries Proposal for an Operaional Framework and Policy Implicaions. Available via he inerne: hp:// 2004, Deb Susainabiliy in Low-Income Counries: Furher Consideraions on an Operaional Framework and Policy Implicaions. Available via he inerne: hp:// Reinhar, Carmen M., Kenneh S. Rogoff, and Miguel A. Savasano, 2003, Deb Inolerance, Brookings Papers on Economic Aciviy, Vol. 1, pp Chapers 9 and 10 Inernaional Moneary Fund, 2005, Informaion Noe on Modificaions o he Fund's Deb Susainabiliy Assessmen Framework for Marke Access Counries. Available via he hp:// McNeilly, Caryl, 2003, Deb Susainabiliy and he IMF, IMF Insiue Courier, Vol. 5 (Washingon: Inernaional Moneary Fund), pp Available via he inerne: hp:// Morris Goldsein, 2003, Deb Susainabiliy, Brazil, and he IMF, IIE Working Paper 03/1 (Washingon: Insiue for Inernaional Economics). Available via he inerne: hp://