Stress testing in the Debt Sustainability Framework (DSF) for Low-Income Countries

Transcription

1 KNOWLEDGE BRIEF ECONOMIC POLICY AND DEBT DEPARTMENT (PRMED) WORLD BANK Sress esing in he Deb Susainabiliy Framework (DSF) for Low-Income Counries François Painchaud 1 and Tihomir (Tish) Sučka 2 May, 2011 Summary: This echnical noe describes in deail he propagaion of sandardized sress ess in he LIC DSA emplae using an analyical and graphical approach. I is inended o aid echnical saff in low income counries as well as counry eams a he IMF and WB by providing he inuiion as well as mechanics of sress esing. This should enhance he undersanding of he shocks in he DSF and help improve he conduc of deb susainabiliy exercises in low income counries. 1 The views expressed herein are hose of he auhors and should no be aribued o he IMF, is Execuive Board, or is managemen. 2 The views expressed herein are hose of he auhors and should no be aribued o he World Bank, is Execuive Board, or is managemen.

2 2

3 Table of Conens 1 INTRODUCTION DEBT BURDEN INDICATORS DEBT SUSTAINABILITY BASELINE SCENARIO STANDARDIZED STRESS TESTS SHOCK TO REAL GDP GROWTH (PDSA AND EDSA) EXCHANGE RATE DEPRECIATION (PDSA AND EDSA) SHOCK TO THE PRIMARY BALANCE (PDSA) SHOCK TO THE PRIMARY BALANCE AND REAL GDP GROWTH (PDSA) SHOCK TO OTHER DEBT CREATING FLOWS (PDSA) HISTORICAL SCENARIO: SHOCK TO NON-INTEREST EXTERNAL CURRENT ACCOUNT, NET FDI, REAL GDP GROWTH, AND GDP DEFLATOR IN US DOLLAR TERMS (EDSA) SHOCK TO TERMS OF FOREIGN FINANCING (EDSA) SHOCK TO EXPORTS (EDSA) SHOCK TO DOMESTIC GDP DEFLATOR IN DOLLAR TERMS (EDSA) SHOCK TO NON-DEBT CREATING FLOWS (EDSA) SHOCK TO REAL GDP, EXPORTS, GDP DEFLATOR, AND NON-DEBT CREATING FLOWS (ESDA) CONCLUSION REFERENCES

4 4

5 1 Inroducion 1. The objecive of he join Fund-Bank deb susainabiliy framework (DSF) for low-income counries (LICs) is o suppor LICs effors o achieve heir developmen goals wihou creaing fuure deb problems. 3 The DSF is buil on hree pillars: (i) a sandardized forward-looking analysis of deb and deb-service dynamics under a baseline scenario, alernaive scenarios, and sandardized sress es scenarios; (ii) a deb susainabiliy assessmen based on indicaive counry-specific deb-burden hresholds ha depend on he qualiy of policies and insiuions in he counry; and (iii) recommendaions on a borrowing (and lending) sraegy o limi he risk of deb disress, while maximizing he resource envelope o achieve he Millennium Developmen Goals (MDGs). The DSF is operaionalized hrough he join Bank-Fund deb susainabiliy analysis (DSA) which is epiomized by he LIC DSA emplae. 2. The LIC DSA emplae combines he assessmen of exernal and public deb susainabiliy in one file. Public DSA covers exernal and domesic public and publicly guaraneed (PPG) deb 4, whereas he exernal DSA covers he counry s overall financing flows wih he res of he world exernal PPG and privae deb. 3. To adequaely inform borrowing and lending decisions, DSAs need o be based on realisic macroeconomic baseline scenarios. The principal mechanism for promoing realism in DSAs is o scruinize baseline projecions by (i) subjecing hem o realiy checks and (ii) making use of precauionary feaures of he DSF. The realiy checks and precauionary feaures are inended o provide safeguards agains excessive borrowing and a reurn o deb disress, wihou consraining jusified opimism abou he effecive use of exernal resources o promoe growh, reduce povery, and achieve he counry s developmen goals. 4. The LIC DSA emplae conains 16 sandardized shocks. The sandardized shocks are deerminisic wih simplified feedback effecs o ensure cross-counry comparabiliy. To undersand he underlying dynamics of he DSF, i is imperaive o comprehend he propagaion of individual shocks hrough he economy. 3 The LIC DSF was formally inroduced in See Operaional Framework for Deb Susainabiliy Assessmens in Low-Income Counries Furher Consideraions and Deb Susainabiliy in Low-Income Counries Furher Consideraions on an Operaional Framework and Policy Implicaions. 4 PPG deb comprises: (i) deb of he public secor, defined as cenral, regional and local governmens, cenral bank, and public enerprises he laer subsumes all enerprises ha he governmen conrols, such as by owning more han half of he voing shares and (ii) privae secor deb guaraneed by he public secor. Excluding an SOE s exernal deb from he exernal DSAs can be considered, if he company can borrow exernally wihou a public guaranee and is operaions pose a limied fiscal risk. For more informaion, please see Saff Guidance Noe on he Applicaion of he Join Bank-Fund Deb Susainabiliy Framework for Low-Income Counries. 5

6 5. The purpose of he paper is o provide a descripion of he sress ess. I combines an analyical and graphical approach o describing he ransmission mechanisms of he shock in he LIC DSA emplae. 6. The paper proceeds as follows. Secion 2 describes he concep of deb burden indicaors, which are cenral o he deb susainabiliy assessmen. Secion 3 discusses he inerpreaion of deb burden oucomes in erms of risks of deb disress. In Secion 4, he baseline scenario is discussed, which is followed in Secion 5 by a deailed descripion of he sress ess. Secion 6 concludes. 2 Deb burden indicaors 8. Deb susainabiliy is assessed by underaking a forward-looking analysis of he evoluion of deb burden indicaors under a baseline scenario, alernaive scenarios, and sandardized sress es scenarios. Deb burden indicaors compare a measure of indebedness o a measure of capaciy o repay: Measureof Indebedness Deb burden indicaor (1) Measureof capaciy o repay 9. Differen measures of indebedness are used o idenify solvency and liquidiy risks. Liquidiy problems arise when a counry has shor erm difficulies meeing is financial obligaions as hey come due alhough is abiliy o pay is no affeced under normal circumsances. Solvency problems, on he oher hand, arise when a counry s repaymen difficulies are permanen or proraced. Pu differenly, insolvency is associaed wih policies ha lead o ever-increasing deb levels in relaive erms, which necessiae a shif in policies such as an increase in axes, cus in spending, recourse o moneizaion, or even repudiaion. 10. Indicaors based on deb socks are used o idenify possible solvency problems. For LICs, a deb sock measure based on he presen value (PV) of deb is more informaive han he nominal sock of deb 5, reflecing he ypical concessionaliy of loans conraced by LICs (equaion 2). Concessional loans are characerized by belowmarke ineres raes, a grace period, and a long mauriy period. Clearly, he deb burden 5 The presen value of deb is he sum of all fuure deb service paymens (principal and ineres) discouned using an appropriae discoun rae. The PV represens he amoun of money o be invesed oday (earning he discoun rae) required o repay all he financial obligaions semming from he exising sock of deb. Given ha he calculaions assume no capial losses, he discoun rae o be used in PV calculaions should be he risk-free ineres rae (r in equaion 2). The discoun rae in he LIC emplae is relaed o he sixmonh average of he U.S. dollar commercial ineres rae (CIRR). The discoun rae was iniially se a 5 percen and is adjused by 100 basis poins, whenever he 6-monh average of he U.S. dollar CIRR moves by a leas 100 basis poins for a leas 6 consecuive monhs. This approach is inended o srike a balance beween he desire o insulae PV calculaions from cyclical movemens, wihou de-linking i enirely from long-erm marke rends. 6

7 associaed wih an obligaion o repay US$100 million in 5 years a an ineres rae of 7 percen is more onerous compared o he obligaion o repay he same US$100 million in 40 years, wih a 10-year grace period and an ineres rae of 0.75 percen. By discouning he deb service (principal and ineres) using an appropriae discoun rae, he PV capures he effecive deb burden. PV DebService (1 r) 1 DebService 2 (1 r) 2 DebService... n (1 r) n (2) 11. Indicaors based on deb service (ineres paymens and amorizaion) are used o assess liquidiy problems. They represen he share of a counry s resources used o repay is deb (and herefore resources no used for oher public purposes). However, wih he repaymen of concessional loans usually increasing as a loan maures, deb service indicaors are likely o be limied for predicing fuure deb servicing problems. Long projecion periods can miigae his problem, bu he reliabiliy of he projecion ends o diminish wih is lengh. 12. Measures of capaciy o repay include GDP, expors, and governmen revenues. Nominal GDP capures he amoun of overall resources of he economy, while expors provide informaion on he capaciy o produce foreign exchange. Finally, governmen revenues measure he governmen s abiliy o generae fiscal resources. In some specific cases, remiances may be added o GDP and expors o assess exernal deb susainabiliy By defaul, DSAs are done on gross deb. For counries wih significan asses, a ne concep may be applied in he public deb framework; liquid asse accumulaion is no aken ino accoun in he exernal emplae. 14. To appropriaely idenify solvency and liquidiy problems, he LIC DSA focuses on differen deb burden indicaors depending on he coverage of PPG deb. For public and publicly guaraneed (PPG) exernal deb, he deb burden indicaors include raios o expors and are as follows: i. PV of deb-o-gdp ii. PV of deb-o-expors iii. PV of deb-o-revenues iv. Deb service-o-expors v. Deb-service-o-revenues For public and publicly guaraneed exernal and domesic deb (i.e. oal public deb), he deb burden indicaors are as follows: 6 See Saff Guidance Noe on he Applicaion of he Join Bank-Fund Deb Susainabiliy Framework for Low-Income Counries. 7

8 i. PV of deb-o-gdp ii. PV of deb-o-revenues iii. Deb-service-o-revenues 3 Deb susainabiliy 15. The DSF uses policy-dependen deb-burden hresholds o assess PPG exernal deb susainabiliy. These indicaive hresholds do no apply o oher definiions of deb burden indicaors. 7 Based on empirical findings, he DSF assumes ha exernal PPG deb levels ha LICs can susain are deermined by he qualiy of heir policies and insiuions. 8 A counry wih relaively good (weak) policies and insiuions is more (less) likely o allocae resources effecively and is, herefore, beer placed o manage a higher (lower) level of exernal PPG deb. 16. Policy performance and insiuional qualiy is measured by he hree-year moving average Counry Policy and Insiuional Assessmen (CPIA) index, compiled annually by he World Bank. The hree-year average is used o preven volailiy in he hreshold level, and, hus, excess volailiy in he risk of deb disress, which in urn deermines he counry s financing erms from IDA (and possibly oher donors) The DSF divides counries ino hree policy performance caegories: srong, medium, and weak. Table 1 depics he associaed exernal deb-burden hresholds. The risk classificaion depends, among oher facors, on he indicaive hresholds and herefore on he CPIA score. 7 Hisorically exernal borrowing has been he main source of financing for LICs. In addiion, risk raings are used o provide a signal o exernal crediors o possibly change heir erms and condiions of financing in response o changes in exernal risk of deb disress. 8 See Deb-Susainabiliy in Low-Income Counries-Proposal for an Operaional Framework and Policy Implicaions and When Is Exernal Deb Susainable?, Aar Kraay and Vikram Nehru, World Bank, Policy Research Paper, February In addiion, for counries where, following he release of he new annual CPIA score, he updaed hreeyear moving average CPIA raing breaches he applicable CPIA boundary, he counry s performance caegory would change immediaely only if he size of he breach exceeds If he size of he breach is a or below 0.05, he counry s performance caegory would change only if he breach is susained for wo consecuive years. 8

9 Table 1: Thresholds for exernal PPG deb Qualiy of policies and insiuions Weak Medium Srong CPIA < CPIA < 3.75 CPIA 3.75 NPV of deb-o-gdp NPV of deb-o-expors NPV of deb-o-revenue Deb service-o-expors Deb service-o-revenue Depending on he evoluion of exernal PPG deb burden indicaors compared wih heir respecive hresholds under he baseline, alernaive scenarios, and sress ess, a counry is classified as: Low risk. All deb indicaors are well below relevan counry-specific debburden hresholds. Sress esing and counry-specific alernaive scenarios do no resul in indicaors significanly breaching hresholds. In cases where only one indicaor is above is benchmark, judgmen is needed o deermine wheher here is a deb susainabiliy problem or some oher issue, for example, a daa problem. Moderae risk. While he baseline scenario does no indicae a breach of hresholds, alernaive scenarios or sress ess resul in a significan rise in debservice indicaors over he projecion period (nearing hresholds) or a breach of deb or deb-service hresholds. 10 High risk. The baseline scenario indicaes a proraced breach of deb or debservice hresholds bu he counry does currenly no face any paymen difficulies. This is exacerbaed by he alernaive scenarios or sress ess. In deb disress. Curren deb and deb-service raios are in significan or susained breach of hresholds. The exisence of arrears would generally sugges ha a counry is in deb disress, unless here are oher reasons han deb-service burden for no servicing is deb. 4 Baseline Scenario 19. The baseline scenario, or mos likely scenario, is ypically based on a macroeconomic framework designed ouside he LIC DSA emplae. The DSA is based on he evoluion of deb burden indicaors over he projecion period (forwardlooking exercise). This implies ha he DSA needs a projecion of: (i) he measures of indebedness, he numeraor; and (ii) he measures of capaciy o repay, he denominaor. The projecion of he measures of indebedness reflecs eiher oal public deb, including domesic deb, or oal exernal PPG deb. 10 Given he long mauriy profiles of deb in many LICs, deb service is ofen backloaded. Thus, a seady increase in such raios o near heir hresholds could indicae he possibiliy of a breach of hresholds beyond he projecion period. 9

10 20. The evoluion of public deb (Deb public ) can be characerized in erms of he primary defici (PD) rajecory. Public deb is expressed in local currency unis (LCU) and is evoluion akes ino accoun: governmen revenues (ax and non-ax revenues (T), and grans (G)) and expendiures (primary expendiures, ha is, oal expendiures excluding ineres paymens (S), and ineres paymens (INT)) as well as oher nonrecurren facors (OTHER) affecing he sock of deb no aken ino accoun in revenues or expendiures (equaion 3). The laer would include: (i) privaizaion receips; (ii) deb relief; (iii) recogniion of coningen liabiliies such as bank recapializaion coss. Finally, a residual componen is added o capure any changes o he sock of public deb no explained by he variables menioned above. Deb Deb Deb Deb Deb Deb public public public public public public public 1 Deb exp endiures revenues OTHER public 1 Deb S ( T G ) INT OTHER public 1 Deb PD INT OTHER public 1 Deb PD INT OTHER PD INT OTHER Deb exernal Deb residual domesic residual residual residual residual (3) 21. In he LIC DSA emplae, equaion (3) is expressed in percenage of GDP. Once equaion (3) is expressed in percenage of GDP o normalize he nominal amouns, he expression for he change in public deb will include erms describing he endogenous or auomaic deb dynamics, such as he conribuion from changes in he ineres rae, real GDP growh, and prices and exchange rae changes. These conribuions are calculaed auomaically in he emplae The usefulness of fiscal indicaors depends on he coverage of he public secor and he qualiy of he deb daa. If he public secor is defined oo narrowly (cenral governmen, raher han general governmen including public companies), hen public secor deb may be undersaed and is capaciy o repay may be inadequaely measured. The accuracy of he susainabiliy assessmen may also be impeded by daa deficiencies such as incomplee domesic deb daa or inappropriaely measured cos of financing. 23. Similar o he public deb emplae, he evoluion of exernal deb can be characerized by he pah of he non-ineres curren accoun defici (NICA) ogeher wih non-deb creaing flows. The basic equaion for he evoluion of exernal deb (Deb exernal ) akes ino accoun a counry s sources of foreign exchange income/inflows and expendiures/ouflows (equaion 4). A counry s source of foreign exchange includes expors of goods and services (X), ne ransfers (NT), and ne 11 For a complee analyical exposiion see Burnside (2005), Ley (2007) or he Appendix of he Saff Guidance Noe on he Applicaion of he Join Bank-Fund Deb Susainabiliy Framework for Low-Income Counries. 10

11 income (NI). 12 A counry s use of foreign exchange, meanwhile, includes impors of goods and services (M). These componens (X, M, NT, and NI) form he curren accoun (CA) in he balance of paymens and can be rearranged ino he curren accoun defici excluding ineres paymens (NICA) and ineres paymens (INT). The evoluion of he sock of exernal deb akes also ino accoun non-deb creaing sources of financing from he balance of paymens. In paricular, he LIC DSA emplae accouns for he non-deb creaing componen of foreign direc invesmen (ne FDI). 13 Oher facors (residual) conribuing o he evoluion of he exernal sock of deb include deb relief (excepional financing), drawdown of foreign exchange reserves, and errors and omissions. Anoher source of non-deb creaing flows is capial grans, which are no explicily aken ino accoun in he LIC DSA emplae. As such, capial grans are capured by he residual. Deb Deb Deb Deb exernal exernal 1 M X NT NI nefdi OTHER Deb residual (4) exernal exernal Deb 1 NICA INT nefdi OTHER exernal exernal Deb 1 NICA INT nefdi OTHER exernal NICA INT nefdi OTHER residual residual residual 24. If a counry spends more han i earns (curren accoun defici), hen foreigners accumulae ne claims on residens - he counry borrows exernally. This assumes ha FDI and oher facors do no compensae for he foreign income shorfall over spending. Noe ha he exernal borrowing can eiher be PPG or privae. Noe also ha i can also be in domesic or foreign currencies, alhough exernal borrowing is almos exclusively done in foreign currency in LICs. 25. The DSA mus be based on a consisen projecion of he fiscal and exernal accouns. By consrucion, exernal borrowing is he sum of privae and public exernal borrowing, providing a direc link beween he fiscal and exernal accouns. Alernaively, public borrowing is he sum of public exernal and domesic borrowing. 5 Sandardized sress ess 26. Informed and pruden borrowing decisions require an assessmen of he evoluion of deb burden indicaors under differen assumpions compared o he baseline. These differen assumpions are called sress ess and hey assess he robusness of he baseline. Sress esing herefore scruinizes he realism of he baseline and reveals he counry s vulnerabiliies. Wha is he sensiiviy of he evoluion of deb burden indicaors o differen assumpions? For example, if real GDP growh is lower han anicipaed, will i jeopardize fiscal susainabiliy? 12 Noe ha for simpliciy, ne ransfers and ne income are assumed o be sources of foreign exchanges bu his may no be he case in realiy. 13 In addiion o is equiy and porfolio par, FDI also has a deb creaing componen, inercompany loans, which should be added o privae exernal deb. 11

12 27. The LIC DSA emplae sress ess he baseline scenario auomaically. The impac of sress esing is channeled in wo ways: by changing he evoluion of he measures of indebedness and by changing he capaciy o repay compared o he baseline scenario. In he DSF, sress ess are deerminisic raher han sochasic. In oher words, shocks of a cerain magniude are assumed o ake place wih cerainy, based on a paricular algorihm ha will be discussed below. 28. In he fiscal DSA, addiional financing needs riggered by a shock o he baseline are always saisfied by new ne (marginal) borrowing. Shocks o governmen revenues, he primary defici, or he exchange rae, ceeris paribus, impac he governmen financing needs. These addiional financing needs, in urn, resul in an idenical change in borrowing. Accordingly, he emplae assumes ha changes in financing needs arising from shocks (compared o he baseline) are me by changes in borrowing raher han adjusmens in governmen policies In he exernal DSA, exernal PPG borrowing closes he balance of paymens afer a shock o he baseline scenario. Changes in he following variables, ceeris paribus, will affec he evoluion of he measures of exernal indebedness: expors, ne ransfers, ne FDI, ne curren accoun defici excluding ineres paymens, and less concessional new exernal borrowing. As in he public DSA, changes o exernal financing needs resul in a comparable change in exernal borrowing needs. In conras o he public DSA, however, he composiion of marginal borrowing is a priori unclear boh he public and he privae secor could in heory change heir borrowing needs. Given ha he public secor is mos likely o have greaer access o exernal borrowing, he emplae assumes ha all of he addiional exernal financing needs ensuing from a shock are me wih PPG exernal borrowing Shocks can be disinguished according o heir duraion. The LIC DSA emplae disinguishes wo ypes of sress ess: (i) alernaive scenarios; and (ii) bound ess. Alernaive scenarios represen permanen shocks, while bound ess are emporary shocks (one or wo years) afer which he alered variables reurn o heir baseline values. 14 In he public DSA, he users are asked o provide informaion on how he erms and condiions (ineres rae, grace period, and mauriy) of he new borrowing will ake place (exernal borrowing, domesic longerm or shor-erm borrowing). Marginal borrowing assumpions can be enered in he Inp_Oup_deb shee of he LIC DSA emplae under cells J10-Q In he exernal DSA, he new borrowing requiremen semming from he sandardized sress ess are financed according o he exogenously specified average erms (ineres rae, grace period, and mauriy) of new financing in he baseline. Marginal exernal borrowing assumpions can be found in he Inp_Oup_deb shee of he LIC DSA emplae under cells C86-C88. 12

13 31. The LIC DSA emplae calibraes he magniude of he shocks using 10 years of hisorical daa. 16 This defaul period is used o calculae he hisorical averages and associaed sandard deviaion of key macro variables. However, under cerain circumsances, he sandardized sress ess may no capure adequaely he vulnerabiliies of he counry. For insance, srucural breaks in he ime series (i.e. discovery of oil or a civil war) or daa deficiencies may require a change in he period used for calibraion. Bound ess are calibraed so ha he implied oucome for he long-erm deb raio (a a year horizon) has a roughly 25 percen likelihood of occurring. 32. Sress esing should follow an asymmeric approach and be iled oward adverse shocks. To check is robusness, he user should sress he baseline wih meaningful adverse shocks i.e. he DSF is ineresed in downside raher han upside risks. Thus, while i may occur ha sress ess calibraed using 10-year averages and sandard deviaions deliver posiive shocks, implying an improvemen in deb burden indicaors, such shocks could be alered o model downside risks. 33. The LIC DSA emplae conains 16 differen sandardized sress ess in oal. The public DSA (PDSA) comprises 3 alernaive scenarios and 5 bound ess; he exernal DSA (EDSA) encompasses 2 alernaive scenarios and 6 bound ess. To simplify he descripion of he sress ess, comparable shocks will be described in he same secion. 16 For he fiscal DSA average and sandard deviaions are calculaed in cells AK16-AL55 in he fiscalbaseline shee. For he exernal DSA average and sandard deviaions are calculaed in cells B68-P73 in he baseline shee. 17 See Deb Susainabiliy in Low-Income Counries Proposal for an Operaional Framework and Policy Implicaions, IMF and IDA, February Noe ha, bound ess do no exhibi unsusainable deb dynamics wihin his framework, unless he baseline does. Sress ess are consruced in such a way ha he ineres rae, growh rae, and primary balance (or curren accoun excluding ineres paymens) reurn o heir baseline value afer he shock dissipaes. 13

14 Sress ess Public DSA (PDSA) Exernal DSA (EDSA) Alernaive scenarios (A1-A3) Permanen shock over he enire projecion period Bound ess (B1-B6) Temporary shocks (second and hird year of projecion) A1. Hisorical The primary balance-o-gdp raio and real GDP growh are se o heir hisorical average. A2. Primary balance The primary balance-o-gdp raio is se o las year of hisory. A3. Lower real GDP growh Real GDP growh is lowered by a fracion of is sandard deviaion. B1. Real GDP growh Real GDP growh is se o is hisorical average minus one sandard deviaion. B2. Primary Balance The primary balance-o-gdp raio is se o is hisorical average minus one sandard deviaion. B3. Combinaion of B1 and B2 Each reduced only by 0.5 sandard deviaions. B4. Depreciaion One-ime nominal depreciaion of 30 percen (firs year of he projecion). B5. Oher Flows One-ime increase of public deb by 10 percen of GDP (firs year of he projecion). A1. Hisorical Several key macroeconomic variables are se o heir hisorical average. A2. Exernal financing Less concessional exernal borrowing (200 basis poins) B1. Real GDP growh Real GDP growh is se o is hisorical average minus one sandard deviaion. B2. Expors Nominal expors (USD) growh se o is hisorical average minus one sandard deviaion. B3. Deflaor Domesic GDP deflaor (USD) is se o is hisorical average minus one sandard deviaion. B4. Oher flows Curren ransfers-o-gdp and FDI o-gdp se o heir hisorical average minus one sandard deviaion. B5. Combinaion (B1-B5) Each reduced only by 0.5 sandard deviaions. B6. Depreciaion One-ime nominal depreciaion of 30 percen (firs year of he projecion). 14

15 5.1 Shock o real GDP growh (PDSA and EDSA) There are wo emporary shocks (bound es B1 in PDSA and EDSA) and one permanen shock (alernaive scenario A3 in PDSA) o real GDP growh. Descripion: Under he emporary shocks (B1), real GDP growh is se a is hisorical average minus one sandard deviaion over he second and hird year of he projecion period (+1 and +2). Real GDP growh reurns o he baseline projecion hereafer. Under he permanen shock (A3), he baseline real GDP growh projecion is reduced over he enire projecion period by roughly one-fifh of one sandard deviaion For boh he PDSA and he EDSA, he shock o real GDP growh represens a shock o measures of capaciy o repay. The adverse shock o real GDP growh has a permanen impac on he level of real GDP (Y r ) and nominal GDP (Y n ). This is he consequence of wo assumpions: (i) real GDP growh reurns o he baseline projecion afer he shock; and (ii) inflaion remains unchanged, as measured by he GDP deflaor. The decline in nominal GDP (compared o he baseline) has in urn a proporional negaive impac on he nominal level of governmen revenues because he revenues-o- GDP raio is assumed o remain unchanged compared o he baseline. 35. In he PDSA (A3 and B1), he shock o real GDP growh also represens a shock o he measures of indebedness (Figure 1). While he real GDP shock affecs adversely nominal revenues, i is no assumed o have an impac on he level of governmen spending. Smaller ax revenues wih respec o he baseline and unchanged spending resuls in a wider non-ineres (primary) fiscal defici and increased financing needs, and hus addiional borrowing. Noe ha grans are also assumed consan in nominal erms compared o he baseline, and, hus, increase in GDP erms. In he end, fuure financing needs are compounded by new marginal borrowing as i raises fuure deb servicing coss (principal and ineres paymens). 36. In he EDSA, he shock o real GDP growh has no impac on he measures of indebedness. Variables affecing he evoluion of he measures of indebedness such as he level of ne FDI inflows or he curren accoun (expors, impors, ne curren ransfers, oher non-ineres curren accoun flows and ineres paymens) remain unchanged compared o he baseline scenario. Wih he curren accoun and ne FDI unaffeced by he shock, he level of exernal deb, is NPV, and he level of deb service coincides wih he baseline scenario. In oher words, in he EDSA, he shock o real GDP growh affecs deb burden indicaors only hrough he impac on he denominaors, ha is, nominal GDP and governmen revenues. 19 Real GDP growh (scenario STD A3) real GDP growh (baseline) 20 15

16 Figure 1: Negaive shock o real GDP growh (PDSA) - impac on he level of gross financing needs Real GDP growh is emporarily reduced, inflaion in unchanged Tax-o-GDP raio is unchanged, he level of grans remains unchanged Governmen spending is unchanged Real GDP growh shock Nominal GDP Gov. ax revenues Primary defici Gross financing requiremens Amorizaion and ineres paymens New borrowing 5.2 Exchange rae depreciaion (PDSA and EDSA) Depreciaion shocks are presen in boh he fiscal and exernal DSA (bound ess B4 in PDSA and B6 in EDSA). This reflecs he currency mismach risks beween he measures of capaciy o repay (GDP and revenues are in local currency) and he measures of indebedness (exernal deb is denominaed in foreign currency). Descripion: The 30 percen depreciaion shock akes place in he second projecion year. During he hird projecion year, he change in he nominal exchange rae resumes is pah in line wih he baseline scenario. Given ha domesic and foreign prices are no affeced, he nominal depreciaion can be inerpreed as real exchange rae shock. 37. In he EDSA, he nominal depreciaion resuls in a concomian decline in he GDP deflaor denominaed in USD. Changes in he deflaor lead o a proporional and permanen reducion in he value of domesic GDP in US dollars. However, real GDP growh (LCU) remains unchanged compared o he baseline. In addiion, he level of all he componens of he BoP are assumed o be unaffeced by he shock. In oher words, he exchange rae shock is a pure valuaion shock. As discussed in he previous secion (secion 5.1), changes o nominal GDP will impac governmen revenues (expressed in foreign exchange) in he EDSA, which represens he denominaor, bu would no affec measures of indebedness. Hence, while he level of exernal deb is unchanged (measure of indebedness), he deb burden indicaor is affeced by smaller nominal GDP and governmen revenues (he capaciy o repay). Meanwhile, deb burden indicaors based on expors will remain unchanged compared o he baseline as he exchange rae shock is assumed no o affec rade. 38. In he PDSA, he measures of repaymen capaciy (nominal GDP and governmen revenues in LCU) are no affeced by he exchange rae depreciaion. In addiion, governmen spending (excluding ineres paymens) or grans do no change 16

17 afer he depreciaion shock. Therefore, he primary defici remains unchanged afer he depreciaion. 39. In he PDSA, he nominal exchange rae depreciaion affecs only he measures of indebedness and composiion of new borrowing. The depreciaion has wo direc effecs on he level and composiion of new borrowing (Figure 2): i) in local currency erms, i increases he cos of servicing he exising sock of foreign currencydenominaed deb (level effec); and ii) he value of planned PPG foreign borrowing is assumed o remain consan is USD afer he depreciaion, which implies a larger value of foreign borrowing in local currency. Given ha he primary balance is unchanged in local currency and foreign borrowing plans are unchanged is USD, he depreciaion effecively leaves he counry wih more local currency from exernal borrowing o finance he same primary balance,, which may reduce ne domesic deb (composiion effec). Figure 2: Exchange rae depreciaion (PDSA) - impac on he level of gross financing needs Exchange rae depreciaion Planned exernal borrowing remains unchanged in FX Amorizaion and ineres paymens on exising sock of exernal deb Value of projeced FX borrowing in local currency Primary balance remains unchanged in local currency Gross financing requiremens New borrowing: financing mix (exernal and domesic borrowing) Amorizaion and ineres paymens 5.3 Shock o he primary balance (PDSA) There are wo primary balance shocks in he PDSA: one permanen (alernaive scenario A2) and one emporary (bound es B2). Descripion: Under he permanen shock (scenario A2), he primary defici-o-gdp raio is se a a level equivalen o he primary defici in he firs year of he projecion over he enire projecion period. This shock is also called he fiscal policy saus quo. Under he emporary shock (bound es B2), he primary defici-o-gdp raio is se o equal is hisorical average minus one sandard deviaion during he second and hird projecion year. 17

18 40. Under boh scenarios, i is assumed ha he shock o he primary defici occurs hrough changes in primary expendiures, raher han governmen revenues. The idea behind he permanen shock (A2) is o presen he oucome of no policy changes o he curren fiscal sance. The bound es (B2) demonsraes he impac of a emporary spending shock. The wider primary defici increases he gross financing need (Figure 3) and, hus, public borrowing (exernal or domesic), which will furher increase fuure financing needs hrough addiional fuure ineres and amorizaion paymens. 41. Under boh scenarios, he deerioraion in he deb burden indicaors reflecs an increase in he measure of indebedness, raher han a deerioraing repaymen capaciy. Despie an increase in governmen spending, nominal GDP and governmen revenues remain unchanged under boh scenarios. Figure 3: Shocks o he Primary Balance (PDSA) - impac on he level of gross financing needs Tax-o-GDP raio and grans are consan, governmen spending increase Primary Defici Gross financing requiremens New borrowing Amorizaion and ineres paymens 5.4 Shock o he primary balance and real GDP growh (PDSA) There are wo sandardized sress ess combining a shock o he primary balance and a shock o real GDP growh: one permanen (alernaive scenario A1) and one emporary (bound es B3) shock. Descripion: Under he permanen shock (alernaive scenario A1), he primary balance-o-gdp raio and real GDP growh are se o equal heir hisorical average saring in he second year of he projecion and lasing over he enire projecion horizon. Under he emporary shock (bound es B3), real GDP growh and he primary balance-o-gdp raio are se in second and hird year of he projecion o equal heir hisorical averages minus half sandard deviaion. 42. Boh sress ess (alernaive scenario A1 and bound es B3) are a combinaion of individual shocks discussed above (secion 5.1 and 5.2). The deerioraion in he primary defici in he second and hird year of he projecion reflecs he shock o he primary balance. Thereafer, he adverse shock o real GDP growh (deerioraion in repaymen capaciy), which in urn affecs adversely nominal governmen revenues 18

19 (hrough he decline in nominal GDP), leads o a deerioraion of he primary defici and an increase in gross borrowing requiremen (Figure 4). 20 The LIC DSA emplae assumes an unchanged level of grans implying a larger gran-o-gdp raio. Noe also ha under he combo sress es, he magniude of he shocks is smaller (½ sandard deviaion) compared o he shocks o individual macro variables (1 sandard deviaion). 21 Figure 4: Negaive shock o primary balance and real GDP growh - impac on he level of gross financing needs Tax-o-GDP raio is consan, he level of grans is unchanged, governmen spending adjus Primary defici Gross financing requiremen New borrowing Real GDP growh Real GDP growh is reduced, inflaion in unchanged Nominal GDP Tax-o-GDP raio is unchanged Gov. ax revenues Amorizaion and ineres paymens Gross financing requiremen +1 New borrowing +1 Amorizaion and ineres paymens 5.5 Shock o oher deb creaing flows (PDSA) One-off increase in oher deb creaing flows amouning o 10 percen of GDP reflecs a coningen liabiliies shock (bound es B5) 20 Once new borrowing is assumed, he financing requiremen will also reflec addiional amorizaion and ineres paymens. 21 The shock magniude under he combo sress es is smaller as he probabiliy ha very large shocks hi he economy simulaneously is likely o be small and he DSF does no aemp o model exreme risks or scenarios. From a pracical perspecive, if he combo sress es were of equal size compared o oher individual shocks o he same macro variables, hen such individual shocks would always represen subses of he large shock from a mos exreme scenario perspecive.. 19

20 Descripion: During he second projecion year, his generic shock o he financing requiremen embodies, for insance, bank recapializaion coss or oher privae secor bail-ous. This realized coningen liabiliy increases direcly new borrowing during he year of he shock. 43. The coningen liabiliy shock augmens he borrowing needs by 10 percen of GDP wihou affecing oher variables (Figure 5). The defaul value is se a 10 percen of GDP, bu his can be changed if he recogniion of implici coningen liabiliies is expeced o exceed he defaul amoun of 10 percen of GDP. 22 Wih nominal GDP and governmen revenues no affeced by his shock, he now larger deb burden indicaors are affeced hrough he measures of indebedness channel only, and no hrough he direc repaymen capaciy measure. Figure 5: Coningen liabiliy shock impac on he level of gross financing needs Coningen liabiliy Gross financing requiremens New borrowing Amorizaion and ineres paymens 5.6 Hisorical scenario: shock o non-ineres exernal curren accoun, ne FDI, real GDP growh, and GDP deflaor in US dollar erms (EDSA) A permanen shock (alernaive scenario A1) o he non-ineres exernal curren accoun defici, ne FDI, real GDP growh, and GDP deflaor in US dollar erms. Descripion: Saring in he second projecion year and lasing over he enire projecion period, he aforemenioned four key variables are se o heir 10-year hisorical average. This shock represens a benchmark agains which he realism of he baseline assumpions is esed as i assumes a coninuaion of he average hisorical economic performance, which ries o capure he srucural characerisics of he economy. In he following descripion, we assume ha he hisorical averages are below he baseline rajecories. 44. This sress es is a combinaion of individual permanen shocks, some of which were described above (secion 5.1). The shock o real GDP growh and o he GDP deflaor impac nominal GDP expressed in foreign currency and, hence, affec he deb burden indicaors hrough he measure of he repaymen capaciy (Figure 6). A he same ime, he shock o he non-ineres curren accoun defici (NICA) and ne FDI is ransmied o deb burden indicaors hrough he indebedness channel as hey affec direcly he borrowing requiremen. 22 An alernaive scenario could also be designed o capure counry-specific circumsances in which he coningen liabiliy shock spreads over several periods and affecs oher variables. 20

21 45. The hisorical scenario affecs he repaymen capaciy hrough is impac on domesic nominal GDP in US dollar. The permanen reducion in real GDP growh and he change in he domesic GDP deflaor in US dollar erms (compared o he baseline scenario) resuls in a reduced growh rae of nominal GDP and, herefore, a smaller nominal GDP. In addiion, he hisorical scenario assumes ha all curren accoun componens and governmen revenues remain unchanged in percen of GDP, wih respec o he baseline. Accordingly he reducion in nominal GDP implies a proporional reducion in expors and governmen revenues. 46. Shocks o boh he exernal curren accoun and he ne FDI impac he financing need and, herefore, exernal borrowing (measure of indebedness). The increase in deb leads o an increase in deb service paymens as well as he PV of deb. 47. All deb burden indicaors are expeced o deeriorae, reflecing a decline in he measure of he capaciy o repay (nominal GDP, expors and governmen revenues) in conjuncion wih larger indebedness, which in urn leads o an increase in PV of deb and deb service paymens. Figure 6: Key variables se o hisorical averages - impac on he level of gross financing needs : Real GDP growh Nominal GDP Gov. revenues GDP deflaor (in US$) Expor level NICA Ne FDI Gross financing requiremen Deb +1: Real GDP growh Nominal GDP Gov. revenues GDP deflaor (in US$) Expor level Amorizaion, ineres NICA +1 Ne FDI +1 Gross financing requiremen Deb 21

22 5.7 Shock o erms of foreign financing (EDSA) This is a permanen shock (alernaive scenario A2) o erms of foreign financing for public secor borrowing Descripion: Saring in he second projecion year, his permanen shock worsens he erms of foreign financing on public exernal borrowing hrough an increase in he ineres rae of 2 percenage poins 23 (compared o he baseline). 48. The deerioraion in he concessionaliy of new borrowing resuls in larger ineres paymens and, hus, ransmis hrough he measure of indebedness (Figure 7). The rise in ineres paymens resuls in larger gross financing needs, and ulimaely a rise in deb as deb service paymens increase compared o he baseline. Measures of repaymen capaciy remain unchanged. Figure 7: Shock o financing erms - impac on he level of gross financing needs : Nominal ineres rae Ineres paymens +1 Gross financing requiremen +1 Deb +1 +1: Nominal ineres rae +1 Ineres paymens +2 Gross financing requiremen +2 n Deb Shock o expors (EDSA) This is a emporary shock (bound es B2) o he growh rae of expors ha lass for wo years Descripion: Nominal expor growh (is USD) is se emporarily a is hisorical average minus one sandard deviaion. Commencing in he second projecion year, his shock lass for wo periods afer which expor growh reurns o he baseline rajecory. Thus, he emporary shock o expor growh has a permanen effec on he level of expors. To achieve a emporary impac on 23 This shock is locaed in he emplae in shee PV Sress_A2, cell: B5 22

23 he level of he non-ineres curren accoun, he level of impors is reduced from he hird year onward of he projecion period. 49. The shock o nominal expors growh in dollar erms resuls in a deerioraion of he curren accoun defici, and hence affecs he measure of indebedness (Figure 8). Under his bound es, impors are assumed consan during he shock and adjus hereafer o assure an exernal non-ineres curren accoun defici equivalen o he baseline. The wider exernal defici causes a rise in gross financing needs, which elevaes he exernal deb level. During he second year of he shock, in addiion o he decline in expors, larger deb service paymens (amorizaion and ineres paymens) semming from he higher indebedness a year earlier conribue o a furher rise in gross financing needs. 50. A downward adjusmen in he level of impors ensures ha he non-ineres curren accoun defici revers o he baseline rajecory afer he wo-period shock. This adjusmen in impors is permanen and in line wih permanenly lower expor levels following he shock (compared o he baseline). Exernal deb is, neverheless, larger hroughou he projecion period due o larger curren accoun deficis. The increase in deb service paymens leads o an increase in he NPV of exernal deb. 51. The decline in expors does no have an impac on he level of nominal GDP, nor governmen revenues. However, all deb burden indicaors will be adversely impaced by a negaive shock o expors hrough an increase in indebedness. Deb burden indicaors using expors as a measure of capaciy o repay will show a greaer deerioraion given he reducion in expors. Figure 8: Two-year shock o expors growh - impac on he level of gross financing needs : Expor growh NICA n Gross financing requiremen Deb Amorizaion +1, ineres paymens +1 +1: Expor growh NICA +1 n Gross financing requiremen +1 Deb +1 +2: Expor unchanged NICA +2 n unchanged Amorizaion +2, ineres +2 Impors Gross financing requiremen +2 Deb +2 23

24 5.9 Shock o domesic GDP Deflaor in dollar erms (EDSA) This is a emporary shock (bound es B3) ha reduces he growh rae of he domesic GDP deflaor expressed in US dollar erms Descripion: Saring in he second projecion year, he US dollar domesic GDP deflaor is se o is hisorical average minus one sandard deviaion for wo periods. 52. The GDP deflaor shock affecs he deb burden indicaors hrough he measure of repaymen capaciy (GDP and governmen revenues). The decline in he deflaor s growh rae reduces direcly nominal GDP in US dollars erms (Figure 9). As he shock o he deflaor is no compensaed for in subsequen periods, i resuls in permanenly smaller levels of nominal GDP. The reducion in nominal GDP, in urn, implies smaller nominal governmen revenues (secion 5.1.), whereas he repaymen capaciy as expressed by expors is no affeced. 53. This shock o he GDP deflaor has no impac on he non-ineres curren accoun or ne FDI. In line wih he bound es B1, he borrowing needs remain unchanged (compared o he baseline) and, hence, he measures of indebedness are no affeced by he shock o he deflaor. Figure 9: Negaive shock o he GDP deflaor in US dollar erms impac on he level of gross financing needs : Domesic GDP deflaor in US dollars Nominal GDP Gov. revenues +1: Domesic GDP deflaor in US dollars Nominal GDP +1 Gov. revenues 5.10 Shock o Non-deb creaing flows (EDSA) This emporary shock (bound es B4) reduces he inflows of ne FDI and ne privae ransfers Descripion: The shock o ne ransfers and ne FDI akes place in he second projecion year and las for wo periods; i revers o he baseline rajecory hereafer. The size of he sress reflecs he hisorical average of he wo variables minus one sandard deviaion. 54. The shock o non-deb creaing flows impacs he deb indicaors hrough he indebedness channel, raher han hrough a deerioraion of he repaymen capaciy (Figure 10). I does so by widening he non-ineres curren accoun defici. Given ha he measure of indebedness changes, all deb burden indicaors will be affeced by his shock. 24

25 Figure 10: Negaive shock o non-deb creaing flows - impac on he level of gross financing needs : Non-deb creaing flow (Curren Transfers and ne FDI) Gross Financing Requiremen Deb Amorizaion +1, ineres paymens +1 +1: Non-deb creaing flow (Curren Transfers and ne FDI) Gross Financing Requiremen +1 Deb Shock o real GDP, expors, GDP deflaor, and non-deb creaing flows (ESDA) This emporary shock (bound es B5) is he mos comprehensive sress es in he emplae as i combines shocks o five variables: real GDP (B1) and expors growh (B2), he GDP deflaor (B3), and ne privae ransfers and FDIs (B4). Descripion: This shock embodies he simulaneous impac of four bound ess (B1-B4) presened earlier (secions 5.1, 5.8, 5.9, 5.10). As opposed o oher bound ess, however, he hisorical averages under his shock are reduced by one-half raher han a full sandard deviaions. Similar o oher sress ess, his shock sars in he second projecion period and lass over wo periods. 55. This shock deerioraes all measure of he capaciy o repay hrough he decline in real GDP, he deflaor, and expors (Figure 11). The emporary decline in real GDP growh and he growh of he GDP deflaor in US dollar erms ranslaes ino a ransiory reducion in nominal GDP growh in U.S. dollars. However, because he emporary deceleraion in real growh and inflaion (compared o he baseline) are no compensaed by increases in subsequen years, he emporary shock o nominal GDP growh has a permanen impac on he level of nominal GDP. Consequenly, nominal GDP in his sress es will be smaller han in he baseline scenario over he enire projecion period. As in previous bound ess, he share of governmen revenues in GDP remains unchanged compared o he baseline. As a resul, governmen revenues will also be permanenly smaller compared o he baseline scenario. Finally, by reducing emporarily he rae of growh in expors (compared o he baseline), he shock permanenly reduces he level of expors. 25

26 56. The decline in expors growh, ne privae ransfers and FDIs worsens he measures of indebedness. The deceleraion in he rae of growh of expors and he reducion in ne ransfers and FDI widen he non-ineres curren accoun defici in absolue erms compared o he baseline. Afer he shock dissipaes, ne ransfers rever o heir baseline values in absolue erms, and impors decline o help he rade balance reurn o he baseline value in absolue erms. 57. All deb burden indicaors are expeced o worsen compared o he baseline scenario. This reflecs he adverse impac of his shock on all measures of repaymen capaciy (GDP, revenues and expors) as well as on he measures of indebedness (PV of deb and deb service). Figure 11: Negaive combinaion shock - impac on he level of gross financing needs Real GDP growh : Domesic GDP deflaor in US dollars Nominal GDP Gov. revenues Expors Ne Transfers Gross financing requiremen Deb +1: Real GDP growh Domesic GDP deflaor in US dollars Nominal GDP Gov. revenues AMT +1, INT +1 Expors +1 Ne FDI +1 Gross financing requiremen +1 Deb +1 +2: Real GDP growh Domesic GDP deflaor in US dollars Nominal GDP Gov. revenues AMT +2, INT +2 Gross financing requiremen +2 Deb +2 Expors +2 Impors +2 Trade Balance +2 unchanged 26

27 6 Conclusion 58. This echnical noe describes in deail he inuiion and propagaion of deerminisic sandardized sress ess in he LIC DSA emplae using an analyical and graphical approach. I is inended o aid echnical saff in low income counries as well as counry eams a he IMF and WB by providing he inuiion as well as mechanics of sress esing. This should enhance he undersanding of he shocks in he DSF and help improve he elaboraion of deb susainabiliy exercises in low income counries. 27

28 7 References Burnside, Craig ed. (2005). Fiscal Susainabiliy in Theory and Pracice A Handbook, The World Bank, Washingon D.C. Kraay, Aar and Vikram Nehru (2004). When is Exernal Deb Susainable?. World Bank Policy Research Paper No Ley, Eduardo (2007). Fiscal Policy for Growh, PRMED Knowledge Brief, June World Bank and IMF (2010). Saff Guidance Noe on he Applicaion of he Join Bank- Fund Deb Susainabiliy Framework for Low-Income Counries. World Bank and IMF (2008). Saff Guidance Noe on he Applicaion o f he Join Fund-Bank Deb Susainabiliy Framework for Low-Income Counries. World Bank and IMF (2005). Operaional Framework for Deb Susainabiliy Assessmens in Low-Income Counries-Furher Consideraions. World Bank and IMF (2004). Deb-Susainabiliy in Low-Income Counries-Proposal for an Operaional Framework and Policy Implicaions. World Bank and IMF (2004a). Deb Susainabiliy in Low-Income Counries-Furher Consideraions on an Operaional Framework and Policy Implicaions. 28