ADB Economics Working Paper SeriesFiscal Sustainability in Developing AsiaCharles Adams, Benno Ferrarini, and Donghyun ParkNo. 205 | June 2010
ADB Economics Working Paper Series No. 205Fiscal Sustainability in Developing AsiaCharles Adams, Benno Ferrarini, and Donghyun ParkJune 2010Charles Adams is Visiting Professor, National University of Singapore; Benno Ferrarini is Economist,Economics and Research Department, Asian Development Bank; Donghyun Park is Principal Economist,Economics and Research Department, Asian Development Bank. The authors gratefully acknowledge theexcellent assistance of Cynthia Petalcorin and Christian Mina in constructing our database, and the equallyexcellent editorial support provided by Lagrimas E. Cuevas.
I. Introduction,Q UHVSRQVH WR WKH SURQRXQFHG LPSDFW RI WKH JOREDO ¿QDQFLDO DQG HFRQRPLF FULVLV RQ WUDGHDQG JURZWK ZKLFK UHDFKHG LWV FOLPD[ LQ WKH VHFRQG KDOI RI DQG WKH ¿UVW KDOI RI GHYHORSLQJ $VLD KHQFHIRUWK $VLD
KDV EROGO GHFLVLYHO DQG TXLFNO XQOHDVKHG VL]DEOH¿VFDO VWLPXOXV SDFNDJHV1 For the most part, those packages have been skewed towardadditional spending, in particular on infrastructure investments, rather than tax cuts. Whilethe People’s Republic of China’s (PRC) US$586 or Yuan 4 trillion stimulus is the mostKLJKO SXEOLFL]HG H[DPSOH RI $VLD¶V QHZIRXQG SHQFKDQW IRU ¿VFDO DFWLYLVP JRYHUQPHQWVacross the region have aggressively boosted expenditures and slashed taxes to supportJURZWK LQ WKH ZDNH RI WKH JOREDO FULVLV $VLD¶V ¿VFDO DFWLYLVP UHSUHVHQWV DQ XQSUHFHGHQWHGresponse to an unprecedented crisis. Unprecedented because never before has thereEHHQ VXFK D VVWHPDWLF VQFKURQL]HG DQG DFURVVWKHERDUG ¿VFDO H[SDQVLRQ HYHQthough the relative magnitude and composition of the stimulus packages varies widelyacross the region. The regionwide activist countercyclical response was borne out ofsheer necessity in a situation in which plunging exports combined with limp privatedomestic demand to create a gaping vacuum in aggregate demand.There is fairly widespread but unsubstantiated belief that the region’s anticrisisFRXQWHUFFOLFDO ¿VFDO SROLFLHV KDYH ³ZRUNHG´ DQG KHOSHG WKH UHJLRQ ZHDWKHU WKH FULVLV ,Wis too early to tell whether the region’s postcrisis recovery can be sustained. However,what is certain is that the speed and strength of the region’s recovery has exceededall expectations. What makes the region’s spectacular V-shaped recovery all the moreremarkable is that it is taking place against the backdrop of a modest and fragile recoveryLQ WKH * 7KH FRPELQDWLRQ RI WZR VWULNLQJ VWOL]HG IDFWV RI $VLD¶V SRVWFULVLV HFRQRPLFODQGVFDSH LH XQH[SHFWHGO UREXVW UHFRYHU DQG UHJLRQZLGH ¿VFDO H[SDQVLRQ KDYH OHGPDQ WR DWWULEXWH WKH UHFRYHU WR WKH ¿VFDO H[SDQVLRQ DOPRVW E GHIDXOW ,W LV WUXH WKDWgovernments across the region have also pursued monetary expansion, but the impactRI PRQHWDU SROLF LV OHVV LPPHGLDWH DQG GLUHFW WKDQ WKDW RI ¿VFDO SROLF HVSHFLDOOJRYHUQPHQW VSHQGLQJ )XUWKHUPRUH LW LV XQFHUWDLQ KRZ ¿UPV DQG FRQVXPHUV UHVSRQG WRORZHU LQWHUHVW UDWHV LQ D VWDWH RI GHSUHVVHG FRQ¿GHQFH ZKLFK FKDUDFWHUL]HV FULVLV SHULRGV,W LV SUHPDWXUH WR JDXJH WKH H[DFW FRQWULEXWLRQ RI WKH ¿VFDO VWLPXOXV WR $VLD¶V UHFRYHUand there were clearly a number of other contributory factors, including the relativelyKHDOWK VWDWH RI WKH UHJLRQ¶V ¿QDQFLDO VVWHPV 1HYHUWKHOHVV WKHUH LV D SRSXODU SHUFHSWLRQWKDW WKH UHJLRQ¶V ¿VFDO VWLPXOXV PDGH D YLWDO FRQWULEXWLRQ WR WKH UHJLRQ¶V UHFRYHU1 Developing Asia refers to the developing member countries of the Asian Development Bank. In line with ADB practice, the paper disaggregates the developing member countries into various geographical subgroupings, namely Central Asia, East Asia, the Pacific, Southeast Asia, and South Asia.
2 | ADB Economics Working Paper Series No. 2055HJDUGOHVV RI WKH DFWXDO FRQWULEXWLRQ RI WKH ¿VFDO VWLPXOXV WR $VLD¶V UHFRYHU WKHUHJLRQZLGH XVH RI FRXQWHUFFOLFDO ¿VFDO SROLFLHV EULQJV WR WKH IRUH WKH LVVXH RI $VLD¶V ¿VFDOsustainability.2 There are both backward and forward looking reasons for why taking stockRI WKH UHJLRQ¶V ¿VFDO VXVWDLQDELOLW LV D WLPHO DQG RSSRUWXQH H[HUFLVH DW WKLV SRLQW LQ WLPH$FFRUGLQJ WR FRQYHQWLRQDO ZLVGRP ZKDW HQDEOHG $VLD WR UROO RXW ODUJH ¿VFDO VWLPXOXVSURJUDPV VR TXLFNO DQG GHFLVLYHO ZDV WKDW LWV SXEOLF ¿QDQFHV ZHUH JHQHUDOO LQ JRRGshape when the global crisis erupted. In particular, its strong initial public debt positionsUHODWLYH WR LQGXVWULDOL]HG FRXQWULHV DQG PDQ RWKHU SDUWV RI WKH ZRUOG PHDQW WKDW WKHUHJLRQ ZDV DEOH WR DIIRUG FRVWO ¿VFDO H[SDQVLRQV VHH (XURSHDQ RPPLVVLRQ D DQG,0)
7KHUH DUH DOVR D QXPEHU RI DUJXPHQWV IRU ZK FRXQWHUFFOLFDO ¿VFDO SROLFZRUNV EHVW IURP DQ LQLWLDO SRVLWLRQ RI ¿VFDO VWUHQJWK )RU H[DPSOH ¿VFDO H[SDQVLRQ LQ DFRXQWU ZLWK D KLJK GHEW OHYHO FRXOG WULJJHU DGYHUVH UHDFWLRQV LQ WKH ¿QDQFLDO PDUNHWVZKLFK ZRXOG VHULRXVO KDUP EXVLQHVV FRQ¿GHQFH 7KHUHIRUH LW LV ZRUWKZKLOH WR WDNH DEDFNZDUG ORRN DW WKH FXUUHQW DQG SDVW VWDWH RI ¿VFDO KHDOWK DQG VXVWDLQDELOLW DFURVV $VLDDoing so would give us a better idea of the extent to which the conventional wisdom ofD ¿VFDOO UHVSRQVLEOH UHJLRQ LV MXVWL¿HG 6XFK DQ DQDOVLV ZRXOG DOVR DOORZ XV WR LGHQWLIdifferences across subregions and countries. An analysis of how countries have adjustedWKHLU ¿VFDO EDODQFHV LQ UHVSRQVH WR ULVLQJ GHEW OHYHOV LQ WKH SDVW ZLOO SURYLGH IXUWKHUHYLGHQFH RQ WKH ³¿VFDOO UHVSRQVLEOH $VLD´ KSRWKHVLV 7R WKH H[WHQW WKDW ZH ¿QG HPSLULFDOVXSSRUW IRU WKH KSRWKHVLV ZH ZLOO KDYH JUHDWHU FRQ¿GHQFH WKDW $VLD ZLOO WDNH HIIHFWLYHPHDVXUHV WR QRUPDOL]H SXEOLF ¿QDQFHV DV WKH ZRUOG HFRQRP UHWXUQV WR QRUPDO PRGH*RLQJ IRUZDUG $VLD¶V ¿VFDO VXVWDLQDELOLW PDWWHUV IRU D QXPEHU RI UHDVRQV )RU RQHUHJDUGOHVV RI WKH YDOLGLW RI WKH SRSXODU EHOLHI WKDW WKH UHJLRQ¶V ¿VFDO VWLPXOXV PDGH D ELJcontribution to its recovery, the stimulus will raise public debt levels and debt servicingcosts in the short term, relative to the counterfactual of no stimulus. For another, andPRUH ZRUULQJO WKHUH LV D ULVN WKDW WKH VL]DEOH GHWHULRUDWLRQ RI WKH ¿VFDO EDODQFH DV DUHVXOW RI WKH DQWLFULVLV ¿VFDO VWLPXOXV SDFNDJHV ZLOO QRW EH IXOO UHYHUVHG HYHQ DIWHU WKHglobal crisis recedes and normalcy returns. From a broader perspective, the biggest riskLV WKDW WKH UHJLRQ¶V DQWLFULVLV ¿VFDO VWLPXOXV PLJKW WXUQ RXW WR EH D ³JDPH FKDQJHU´ WKDWIXQGDPHQWDOO DOWHUV WKH UHJLRQ¶V FRQVHUYDWLYH ¿VFDO SKLORVRSK DQG RXWORRN )RU WKHPRVW SDUW WKH UHJLRQ¶V ¿VFDO SROLF KDV EHHQ JHDUHG WRZDUG SURYLGLQJ JURZWKFRQGXFLYHpublic goods such as infrastructure and education within the government’s budgetFRQVWUDLQW 7KH JHQHUDO DYHUVLRQ WR ODUJH DQG SHUVLVWHQW EXGJHW GH¿FLWV ZDV URRWHG LQthe region’s high premium on macroeconomic stability. The only persistent exception to$VLD¶V JHQHUDOO EHQLJQ ¿VFDO VLWXDWLRQ ZDV 6RXWK $VLD $ IXUWKHU IDFWRU WKDW OLPLWHG EXGJHWGH¿FLWV KDV EHHQ WKH UHJLRQ¶V JHQHUDO UHOXFWDQFH WR SXUVXH FRXQWHUFFOLFDO ¿VFDO SROLF WRVWDELOL]H RXWSXW )RU H[DPSOH DXWRPDWLF ¿VFDO VWDELOL]HUV VXFK DV XQHPSORPHQW EHQH¿WVremain underdeveloped in the region.2 As discussed below, fiscal sustainability refers to whether government budgets can be smoothly financed without generating explosive increases in public debt. A sustainable fiscal policy is not necessarily an optimal fiscal policy but optimal fiscal policies will need to be sustainable (Horne 1991).
Fiscal Sustainability in Developing Asia | 3,QWHUQDWLRQDO H[SHULHQFH LQIRUPV XV WKDW SROLWLFDO HFRQRP IDFWRUV PDNH LW GLI¿FXOW WRUHYHUVH ¿VFDO H[SDQVLRQ XQGHUWDNHQ GXULQJ EXVLQHVV FFOH GRZQWXUQV ZKHQ WKH FFOHmoves up again. This can lead to a ratcheting up of public spending and debt over time.For example, political opposition may stymie the reversal of tax cuts implemented duringa recession even though the economy may have come out of the recession. Similarly, theH[SDQVLRQ RI VRFLDO ZHOIDUH EHQH¿WV GXULQJ GRZQVZLQJV PD EH GLI¿FXOW WR UROO EDFN ZKHQthe economy recovers. At a broader level, intuitively, it is politically easier for governmentsto increase spending and cut taxes during recessions than to cut spending and increaseWD[HV GXULQJ ERRPV 7KLV JLYHV FRXQWHUFFOLFDO ¿VFDO SROLF DQ LQKHUHQW VWUXFWXUDO ELDVtoward higher public spending and debt over time. From Asia’s viewpoint, the popularSHUFHSWLRQ WKDW ¿VFDO VWLPXOXV SDFNDJHV DUH HIIHFWLYH LQ SURPRWLQJ UHFRYHU PD OHDGWR ¿VFDO DFWLYLVP LQ JHQHUDO DIWHU QRUPDOF UHWXUQV 7KDW LV HYHQ WKRXJK WKH ¿VFDOstimulus packages were an exceptional response to an exceptional crisis, their perceivedHIIHFWLYHQHVV PD SURYRNH SRSXODU FDOOV IRU FRXQWHUFFOLFDO ¿VFDO DFWLYLVP GXULQJ QRQFULVLVperiods.,Q DGGLWLRQ WR WKH GLUHFW HIIHFW RI WKH DQWLFULVLV ¿VFDO VWLPXOXV DQG D ³JDPHFKDQJLQJ´ VKLIWLQ WKH UHJLRQ¶V ¿VFDO SKLORVRSK WKHUH DUH D QXPEHU RI RWKHU VWUXFWXUDO IDFWRUV WKDW LPSOD SRWHQWLDOO ODUJH LQFUHDVH LQ WKH UHJLRQ¶V GHPDQG IRU ¿VFDO UHVRXUFHV LQ WKH PHGLXPterm. One is population aging, a seismic demographic transition fundamentally reshaping$VLD¶V GHPRJUDSKLF SUR¿OH :KLOH WKHUH DUH GLIIHUHQFHV DFURVV VXEUHJLRQV²WKH WUDQVLWLRQtoward older populations is much more advanced in East and Southeast Asia than in6RXWK $VLD²DOO GHPRJUDSKLF LQGLFDWRUV SRLQW WR D IDVWDJLQJ $VLD %XLOGLQJ XS PDWXUHZHOOIXQFWLRQLQJ SHQVLRQ VVWHPV ZLOO UHTXLUH D VXEVWDQWLDO DPRXQW RI ¿VFDO UHVRXUFHVAnother structural force is rebalancing the region’s growth toward domestic demand,DQ LVVXH H[SORUHG LQ JUHDW GHWDLO LQ $% E
7KH LPSDFW RI ¿VFDO SROLF RQ WKHUHEDODQFLQJ SURFHVV ZLOO EH GHWHUPLQHG E WKH HIIHFWLYHQHVV RI VSHFL¿F PLFURHFRQRPLF¿VFDO PHDVXUHV VXFK DV VWUHQJWKHQLQJ KHDOWK HGXFDWLRQ SHQVLRQ DQG VRFLDO SURWHFWLRQSuch measures can promote domestic consumption and hence domestic demand byGLOXWLQJ WKH SUHFDXWLRQDU PRWLYH IRU VDYLQJ $ VXSSOVLGH H[DPSOH LV UHPRYLQJ ¿VFDOincentives that favor export production (e.g., manufacturing), over production for theGRPHVWLF PDUNHW HJ VHUYLFHV
,Q SULQFLSOH VXFK SUREDODQFLQJ ¿VFDO PHDVXUHV QHHGnot involve a quantitative expansion of the government and can be achieved by shiftingthe composition of government expenditures. In practice, given the low-tax, small-JRYHUQPHQW SROLF HQYLURQPHQW WKDW FKDUDFWHUL]HV WKH UHJLRQ LW LV OLNHO WKDW LPSOHPHQWLQJsuch measures will require moderate expansion.7KHUH DUH DOVR WZR SROLFUHODWHG UHDVRQV ZK $VLD¶V ¿VFDO VXVWDLQDELOLW PDWWHUV DQGmatters a whole lot, at this point in time. One is immediate and concerns an exit strategy,RU WKH LVVXH RI ZKHQ DQG KRZ WR XQZLQG WKH UHJLRQ¶V DQWLFULVLV ¿VFDO VWLPXOXV SDFNDJHV7KHUH LV D GHOLFDWH EDODQFH EHWZHHQ H[LWLQJ ³WRR VRRQ WRR TXLFNO´ DQG ³WRR ODWH WRRVORZO´ ZLWK ERWK HQWDLOLQJ VXEVWDQWLDO FRVWV ,Q WKH FDVH RI $VLD WKHUH LV D ULVN WKDW WKHUHJLRQ¶V UHODWLYHO KHDOWK SXEOLF ¿QDQFHV PD OXOO SROLFPDNHUV LQWR GRZQSODLQJ WKH YHU
4 | ADB Economics Working Paper Series No. 205UHDO FRVWV RI H[LWLQJ ³WRR ODWH WRR VORZO´ $W D PLQLPXP RXU H[DPLQDWLRQ RI WKH UHJLRQ¶V¿VFDO VXVWDLQDELOLW ZLOO UHPLQG SROLFPDNHUV DERXW WKRVH FRVWV DQG HQFRXUDJH WKHP WRfactor them in more explicitly into their policymaking calculus. The second policy-relatedPRWLYH IRU RXU DQDOVLV LV UHODWHG WR WKH ¿UVW DQG LW LV WR DOHUW WKH UHJLRQ¶V SROLFPDNHUV WRWKH QHHG IRU VWURQJ PHGLXPWHUP ¿VFDO SROLF IUDPHZRUNV WKDW SURPRWH ¿VFDO VXVWDLQDELOLWLQ WKH PHGLXP DQG ORQJ UXQ 7KH PHGLXPWHUP LPSDFW RI WKH DQWLFULVLV ¿VFDO VWLPXOXVSDFNDJHV DORQJ ZLWK QHZ ¿VFDO GHPDQGV WKDW DUH H[SHFWHG WR DULVH LQ WKH PHGLXP WHUPmake it imperative for regional policymakers to actively explore effective and crediblemedium-term policy frameworks. Regardless of the exact timing and nature of exit fromWKH FXUUHQW ¿VFDO VWLPXOXV VXFK IUDPHZRUNV ZLOO DOVR DVVXDJH WKH SXEOLF¶V FRQFHUQV DERXWWKH WKUHDW WR ORQJWHUP ¿VFDO VXVWDLQDELOLW SRVHG E WKH VWLPXOXV7KH UHVW RI RXU SDSHU LV RUJDQL]HG DV IROORZV 6HFWLRQ ,, UHYLHZV WKH FRQFHSWV RI VWDWLFDQG GQDPLF ¿VFDO VXVWDLQDELOLW DQG VRPH DOWHUQDWLYH DSSURDFKHV WR WKHLU DVVHVVPHQW6HFWLRQ ,,, FRQVLGHUV WKH NH GLPHQVLRQV RI WKH SXEOLF ¿QDQFHV DFURVV GHYHORSLQJ $VLDIRFXVLQJ LQ SDUWLFXODU RQ ¿VFDO EDODQFHV JRYHUQPHQW VSHQGLQJ DQG UHYHQXHV DV ZHOO DVlevels of public debt. Section IV discusses the results from a number of econometric testsRI ¿VFDO VXVWDLQDELOLW EDVHG RQ SDQHO UHJUHVVLRQ WHFKQLTXHV DQG SUHVHQWV HVWLPDWHV RI¿VFDO SROLF UHVSRQVH IXQFWLRQV WR LQYHVWLJDWH WKH UHVSRQVHV RI SULPDU ¿VFDO EDODQFHV WRchanges in debt ratios. Section V presents three alternative scenarios or stress tests ofWKH SRWHQWLDOO DGYHUVH LPSOLFDWLRQV IRU ¿VFDO VXVWDLQDELOLW LI FRXQWULHV ZHUH WR SRVWSRQHfor too long the reversal of recent stimulus packages or adjust in too timid a manner.6HFWLRQ 9, GLVFXVVHV WKH UROH UREXVW PHGLXPWHUP ¿VFDO SROLF IUDPHZRUNV PLJKW SODLQ KHOSLQJ $VLD DWWDLQ DQG PDLQWDLQ ¿VFDO VXVWDLQDELOLW LQ WKH SRVWFULVLV SHULRG )LQDOO6HFWLRQ 9,, FRQFOXGHV ZLWK DQ RYHUYLHZ RI WKH PDLQ ¿QGLQJV RI RXU SDSHU DORQJ ZLWK WKHSROLF LPSOLFDWLRQV RI WKRVH ¿QGLQJVII. Conceptual Review of Fiscal Sustainability)LVFDO VXVWDLQDELOLW LV WKH VWDWH ZKHUHLQ WKH JRYHUQPHQW EXGJHW FDQ EH VPRRWKO ¿QDQFHGwithout generating explosive increases in public debt (or money supply3) over time.When this condition is met, the budget is said to be sustainable and, conversely, whenthe condition is not met.4 In some contexts, it is useful to draw an explicit distinctionEHWZHHQ VWDWLF ¿VFDO VXVWDLQDELOLW ZKHQ WKH EXGJHW FDQ EH ¿QDQFHG VPRRWKO SHULRGE SHULRG
DQG GQDPLF ¿VFDO VXVWDLQDELOLW ZKHQ WKH EXGJHW GRHV QRW OHDG RYHU WLPH3 In what follows, we abstract from monetary financing, as this has generally not been an issue in the region. Moreover, many central banks in the region, under their constitutions, are not permitted to directly finance fiscal deficits and have adopted inflation targeting frameworks, implying that low and stable inflation are the key objective of monetary policy. See BIS (2006).4 Note that fiscal sustainability is different from external sustainability. Whereas fiscal sustainability covers the government budget, external sustainability deals with whether the balance of payments can be smoothly financed without explosive increases in external debt. See Chalk and Hemming (2000) and IMF (2003) for further discussion.
Fiscal Sustainability in Developing Asia | 5to explosive increases in public debt). Loosely, static sustainability refers to the ability ofthe government to fund its budget on a period-by-period basis (funding liquidity) whileGQDPLF VXVWDLQDELOLW LV FRQFHUQHG ZLWK YHU ORQJWHUP ¿VFDO VROYHQF5 Both staticDQG GQDPLF ¿VFDO VXVWDLQDELOLW DUH LPSRUWDQW DQG WKUHDWV WR HLWKHU RU ERWK FDQ KDYHLPSOLFDWLRQV IRU PDFURHFRQRPLF DQG ¿QDQFLDO VWDELOLW6(YHQ WKRXJK ¿VFDO VXVWDLQDELOLW LV JHQHUDOO UHJDUGHG DV YHU LPSRUWDQW WKHUH LV QRuniversal agreement about how it should best be assessed. Various approaches toDVVHVVLQJ ¿VFDO VXVWDLQDELOLW KDYH EHHQ XVHG DQG LW LV XVHIXO WR FRQVLGHU VRPH RIthe key approaches. For illustrative purposes, equation (1) describes the (ex ante)government7 budget constraint under the assumption that all government spending(including gross interest payments) is included in government expenditure (G); andthat all government receipts8 are included in revenues (Z). Abstracting from centralEDQN ¿QDQFLQJ equation (1) implies that differences between government revenuesDQG H[SHQGLWXUHV LQ DQ SHULRG WKH EXGJHW VXUSOXV
ZLOO EH UHÀHFWHG LQ FKDQJHV LQ WKHoutstanding stock of (one period) public debt (B)10 held by the nongovernment sector. Z(t) – G(t) = - (B(t+1) – B(t)) t= 1,2, 3,………..N (1)To better focus on public debt dynamics, it is useful to separate out the interest paymentsRQ SXEOLF GHEW IURP DOO RWKHU JRYHUQPHQW H[SHQGLWXUH DQG GH¿QH 6 DV WKH SULPDU RUnondebt-interest-related) budget surplus.11 With R as the one-period nominal interest ratefactor,12 HTXDWLRQ
GHQRWLQJ SULPDU JRYHUQPHQW VSHQGLQJ B(t+1) = R B(t) – S(t) t= 1,2, 3,………..N (2)5 At a fundamental level, the distinction has to do with the difference between illiquidity and insolvency (see Horne 1991, Blanchard et al. 1990, Bohn 1995).6 Financial stability can be threatened directly in circumstances, such as in much of Asia, where local banking systems hold substantial amounts of public debt. Asian bond markets and bank exposure to public debt are regularly assessed in the Asia Bond Monitor of the ADB.7 As argued in the next section, fiscal policy can be assessed for various levels of government or for the public sector as a whole. In the case of countries with significant state-owned enterprises or government linked companies, the lines between the “private” and “public” sectors can be difficult to draw and may be changing over time under reform programs.8 These would include tax and nontax revenues and, in the case of governments with asset holdings, the income on those assets.9 In any period, some part of government spending may be financed by the central bank acquiring public debt on the primary or secondary market. Such implicit monetization is netted out of equation (1) and is generally small in the region.10 Public debt refers, in general, to outstanding government borrowing, which is assumed for illustrative purposes to take the form of bonds rather than bank loans.11 Here, and in what follows, we abstract from any assets held by the government, and we do not distinguish between gross and net debt.12 That is to say, R is 1 plus the nominal interest rate.
6 | ADB Economics Working Paper Series No. 205According to equation (2), the dynamics of public debt are linked to the interest rate onSXEOLF GHEW DQG WKH VL]H RI WKH SULPDU VXUSOXV 6
interest rate. Conversely, the growth of the public debt will be less than the nominalinterest rate when the primary balance is in surplus. In all cases, the behavior of publicGHEW LQ UHODWLRQ WR JURVV GRPHVWLF SURGXFW *3
ZLOO DOVR EH LQÀXHQFHG E ZKHWKHU WKHnominal interest rate is above or below the growth rate of nominal GDP (as discussedbelow).By solving equation (2) forward over time, the public debt in period t can be written asfollows:13B(t ) = Σ ∞=0 R(t , t + j )−1S(t + j ) + lim j R(t , t + T )−1B(t ,T + 1) (3) T →∞Here, B (t, T+1) is the terminal or very long-term debt stock and R(t, t+j) is the discount j ΠIDFWRU EHWZHHQ SHULRG W DQG SHULRG W M ZKLFK LV GH¿QHG DV k =0Rt + k . Generally, dynamic¿VFDO VXVWDLQDELOLW LV WDNHQ WR UHTXLUH WKDW WKH terminal debt stock (discounted at aSRVLWLYH LQWHUHVW UDWH
DSSURDFK ]HUR DV 7 DSSURDFKHV LQ¿QLW ZKLFK FDQ EH UHJDUGHGDV UXOLQJ RXW ³3RQ]L0DGRII´ VFKHPHV LQ ZKLFK QHZ GHEW H[SORGHV RU LV UROOHG RYHULQGH¿QLWHO14 With this condition imposed, the interpretation of equation (3) is that the¿VFDO VLWXDWLRQ LV H[ DQWH GQDPLFDOO
VXVWDLQDEOH LI WKH SUHVHQW YDOXH RI DOO SULPDUsurpluses matches the value of the current debt stock. Alternatively, if the current debtVWRFN LV ]HUR WKH UHTXLUHPHQW IRU ¿VFDO VXVWDLQDELOLW LV WKDW WKH SUHVHQW YDOXH RI DOO(noninterest or primary) government expenditures should match the present value of allgovernment revenues as given by equation (4). ∞ ∞∑ R(t, t + j )−1G(t + j ) = ∑ R(t, t + j )−1Z (t + j ) (4)j =0 j =0% YLUWXH RI LWV GHSHQGHQFH RQ WKH EHKDYLRU RI SXEOLF GHEW RYHU DQ LQ¿QLWH WLPH SHULRG¿VFDO VXVWDLQDELOLW LV D VRPHZKDW DEVWUDFW FRQGLWLRQ $V QRWHG EHORZ SROLFEDVHGDVVHVVPHQWV RI ¿VFDO VXVWDLQDELOLW LQYDULDEO GHDO ZLWK WKH EHKDYLRU RI GHEW RYHU VKRUWto medium-term time periods and are guided, in varying degrees, by the very long-termconsiderations underpinning the abovementioned terminal condition. Notwithstandingthese considerations, however, budget constraint equations such as given by equations
PDNH FOHDU WKDW WKH ¿VFDO VWLPXOXV LQWURGXFHG GXULQJ WKH JOREDO VORZGRZQ ZLOOneed eventually to be paid back through either higher government revenues or lower13 An alternative approach is to express the items in the budget as proportions of GDP. Under such an approach, the long-run terminal condition discussed below involves not only the nominal interest rate but also the nominal growth rate of GDP. Satisfaction of the terminal condition is related to whether the nominal interest rate is above or below the growth rate of the economy (see below for further discussion).14 See Horne (1991), Chalk and Hemming (2000), and Mendoza and Ostry (2008) for further discussion. In practice, many sustainability tests are based on the weaker condition that the long run debt ratio stabilizes over a finite horizon of 3–5 years (see below).
Fiscal Sustainability in Developing Asia | 7H[SHQGLWXUHV 7KLV LV D FRQVHTXHQFH RI WKH ³QR IUHH OXQFK´ FRQGLWLRQ LPSOLHG E GQDPLF¿VFDO VXVWDLQDELOLW DQG WKH SRVLWLYH GLVFRXQW IDFWRU 5
WLPH KRUL]RQ EXW WKHUH DUH DOVRGLI¿FXOWLHV UHODWHG WR WKH IDFW WKDW WKHUH DUH IHZ WKHRUHWLFDO UHVWULFWLRQV RQ WKH YDOXHVof government spending and revenues. With the possible exception of some extremeboundary conditions, such as revenue should not exceed GDP, or that there be a certainminimum level of noninterest-related government spending, there are, in principle, anLQ¿QLWH QXPEHU RI ZDV LQ ZKLFK WKH VXVWDLQDELOLW FRQGLWLRQ PLJKW EH VDWLV¿HG16 Hence,IRU H[DPSOH LI D JRYHUQPHQW ZDV FXUUHQWO UXQQLQJ D ODUJH ¿VFDO GH¿FLW ZLWK SXEOLF GHEWULVLQJ VKDUSO LW ZRXOG EH GLI¿FXOW WR FRQFOXGH WKDW VXVWDLQDELOLW ZDV QHFHVVDULO YLRODWHG7KLV LV EHFDXVH WKH JRYHUQPHQW PD EH SODQQLQJ WR FXW VSHQGLQJ DQG UXQ ODUJH ¿VFDOsurpluses in the very distant future so as to satisfy the sustainability condition over time.$QG DUJXDEO HYHQ LQ WKH PRVW H[WUHPH FDVHV RI ¿VFDO H[WUDYDJDQFH WKHUH FRXOG DOZDVbe some future level of government spending and taxes to ensure that conditions such asHTXDWLRQ
DUH VDWLV¿HGBased, inter alia, on these types of considerations, it is therefore frequently argued that¿VFDO VXVWDLQDELOLW DQDOVLV LV PRUH RI DQ ³DUW´ WKDQ D ³VFLHQFH´ 6XFK DQDOVLV GHDOVQRW VR PXFK ZLWK ZKDW FRXOG KDSSHQ LQFOXGLQJ ZKHWKHU WKHUH PD EH D ¿VFDO FULVLV LQthe event policy is not sustainable),17 as with what will happen based on factors suchDV SDVW H[SHULHQFHV ZLWK ¿VFDO DGMXVWPHQW ,Q WKHVH FLUFXPVWDQFHV HVWLPDWHV RI ¿VFDOSROLF UHDFWLRQV²ZKLFK PHDVXUH KRZ JRYHUQPHQWV KDYH WUDGLWLRQDOO UHVSRQGHG WR ULVLQJGHEW OHYHOV²FDQ EH LPSRUWDQW WRROV LQ DVVHVVLQJ WKH SURVSHFWV IRU ¿VFDO VXVWDLQDELOLW DVdiscussed below.$VVHVVPHQWV RI ¿VFDO VXVWDLQDELOLW KDYH WHQGHG WR DGRSW RQH RU PRUH RI D QXPEHU RIinterrelated approaches. For the most part, the assessments have focused on dynamicrather than static sustainability, although there was some discussion of the latter in lastyear’s Asia Economic Monitor $% D
7KH Asia Economic Monitor noted thatWKHUH KDG EHHQ D QXPEHU RI ³IDLOHG´ SXEOLF GHEW DXFWLRQV LQ WKH FRQWH[W RI VHHNLQJ WR¿QDQFH VRPH RI WKH ¿VFDO VWLPXOXV SDFNDJHV LQ WKH UHJLRQ LQ DQG LW UHYLHZHG WKH15 A little reflection also makes clear why fiscal sustainability needs to be discussed with reference to the ex ante (rather than ex post) government budget constraint. Necessarily, the ex post government budget constraint must be satisfied since spending must always match receipts when allowance is made for arrears. And, in the case of longer-term public debt, its market value would be expected to take into account whether there was fiscal solvency, implying that the intertemporal budget constraint might always hold in mark to market terms. In ex ante terms, however, the government budget need not be balanced over time, and issues of dynamic fiscal sustainability arise when public debt grows at potentially explosive rates (Horne 1991, IMF 2003).16 See discussion in Chalk and Hemming (2000) and Bohn (1991, 1995, 1998).17 Threats to fiscal sustainability need not necessarily lead to fiscal crises as conventionally defined. Instead, they might lead to debt monetization and inflation, or to (formal or informal) debt restructuring. Moreover, the implications of unsustainable fiscal policy would be expected to depend as well on whether there is an accompanying problem of external debt sustainability.
8 | ADB Economics Working Paper Series No. 205circumstances and implications of these failures. If the government is not able to sell (orUROO RYHU
LWV SXEOLF GHEW DV D UHVXOW RI ³IDLOHG´ DXFWLRQV VWDWLF ¿VFDO VXVWDLQDELOLW ZLOO EHviolated. The Asia Economic Monitor, however, did not judge this to be a major issue inthe region during the global crisis.18%URDGO WKUHH GLIIHUHQW VHWV RI DSSURDFKHV WR DVVHVVLQJ GQDPLF ¿VFDO VXVWDLQDELOLW FDQEH LGHQWL¿HG VHH %ODQFKDUG HW DO HOXVXQ HW DO
2QH DSSURDFK LQYROYHVtime series tests in which the (formal) time series properties of variables such as primaryand nonprimary government spending and revenues, interest payments, and public debtVWRFNV DUH DVVHVVHG VHH IRU H[DPSOH 7UHKDQ DQG :DOVK DQG +DXJ Hostland and Karam 2006). In some instances, the focus is on whether the debt stock(or the debt stock in relation to GDP) is a stationary variable based on the applicationof standard unit root tests. In line with some of the oldest approaches to assessing¿VFDO VXVWDLQDELOLW RPDU
WKHVH WHVWV HIIHFWLYHO FRQVLGHU ZKHWKHU SXEOLF GHEWor the ratio of public debt to GDP is stationary (in a statistical sense) and exhibits meanreverting tendencies. Effectively, these types of tests involve determining whether thetime series for a variable such as the debt stock (or the debt ratio) exhibits the kind ofH[SORVLYH EHKDYLRU WKDW PLJKW EH H[SHFWHG LQ WKH HYHQW WKDW ¿VFDO SROLF LV XQVXVWDLQDEOHIn another time series approach, the literature on asset price bubbles is applied to testwhether the time series for debt stocks include a bubble term (or unstable root) such asPLJKW EH IRXQG LQ WKH FDVH RI ¿VFDO SROLF EHLQJ XQVXVWDLQDEOH +DPLOWRQ DQG )ODYLQ +RUQH KDON DQG +HPPLQJ
,Q DGGLWLRQ WKHUH DUH WHVWV RI WKH FRLQWHJUDWLQJrelationships that would be expected to hold among various variables in the event that¿VFDO SROLF LV VXVWDLQDEOH DV JLYHQ LQ HTXDWLRQ
7KHVH WHVWV LQYROYH FRQVLGHULQJZKHWKHU FHUWDLQ ¿VFDO YDULDEOHV VXFK DV JRYHUQPHQW VSHQGLQJ DQG WD[HV VKDUH FRPPRQVWRFKDVWLF WUHQGV FRQVLVWHQW ZLWK WKH GQDPLF JRYHUQPHQW EXGJHW FRQVWUDLQW DQG ¿VFDOsustainability.By their nature, time series tests are backward looking (in so far as they use historicalGDWD
DQG RQO PHDVXUH LQGLUHFWO WKH H[WHQW WR ZKLFK WKH FRQGLWLRQV IRU ¿VFDOVXVWDLQDELOLW DUH VDWLV¿HG ([SOLFLWO RU LPSOLFLWO WKH DSSURDFKHV DOVR DVVXPH WKDW WKHGLVWDQW IXWXUH ZLOO ³ORRN OLNH´ WKH SDVW DQG WKDW SROLFLHV WKDW ZHUH VXVWDLQDEOH LQ WKH SDVWwill continue to be sustainable in the future. In practice, the time series approaches to¿VFDO VXVWDLQDELOLW KDYH QRW EHHQ YHU FRPPRQ IRU DW OHDVW WKUHH UHDVRQV )LUVW WKHVHapproaches typically require relatively long time series of data that are frequently notDYDLODEOH LQ WKH ¿VFDO DUHD HVSHFLDOO DV UHJDUGV SXEOLF GHEW20 Secondly, the power ofunit root tests tends to be relatively low (especially in small samples) in distinguishing,LQ SDUWLFXODU EHWZHHQ VLWXDWLRQV LQ ZKLFK ¿VFDO SROLF PD EH FORVH WR EHLQJ VXVWDLQDEOHDQG ZKHQ LW LV XQVXVWDLQDEOH $QG ¿QDOO DV DUJXHG E %RKQ DQG
WKH OLQNV18 Viet Nam, in particular, experienced some failed debt auctions but these were judged to have reflected a temporary mispricing of initial issuance.19 See, for example, Adedeji and Thornton (2008).20 This problem is mitigated to some degree through the use of panel cointegration techniques that make use of cross sectional as well as time series data (Adedeji and Thornton 2008).
Fiscal Sustainability in Developing Asia | 9EHWZHHQ WKHVH WHVWV DQG ¿VFDO VXVWDLQDELOLW LV QRW QHFHVVDULO YHU FORVH DQG VRPH EDVLFtime series tests do not fully exploit the implications of uncertainty in deriving appropriateWHVWV RI ¿VFDO VXVWDLQDELOLW21More fundamentally, a shortcoming of the time series approaches is that they do notH[SOLFLWO LGHQWLI WKH ¿VFDO SROLF UHDFWLRQ IXQFWLRQV WKDW XQGHUOLH WKH GDWD $V D UHVXOWWKH GR QRW VKHG PXFK OLJKW RQ WKH NLQGV RI ¿VFDO SROLFLHV WKDW PLJKW GHOLYHU VXVWDLQDELOLWor identify why sustainability may not have held in the past. The next two approachesIRFXV GLUHFWO RQ ¿VFDO SROLF EHKDYLRU DV UHÀHFWHG LQ WKH SULPDU ¿VFDO EDODQFHAnother set of tests can be referred to primary balance tests as they involve estimating³GHFLVLRQ UXOHV´ IRU SULPDU ¿VFDO EDODQFHV DQG LQ SDUWLFXODU IRU KRZ SULPDU EDODQFHVUHVSRQG WR FKDQJHV LQ SXEOLF GHEW DQG RWKHU YDULDEOHV %RKQ KDON DQG +HPPLQJ
7KHVH WHVWV DUH DOVR EDFNZDUG ORRNLQJ LQ VR IDU DV WKH PRGHO ¿VFDO GHFLVLRQ UXOHVbased on past behavior. They have the desirable feature, however, in that they focus on¿VFDO UHVSRQVHV DQG LQ SDUWLFXODU RQ KRZ WKH SULPDU ¿VFDO EDODQFH DV D PHDVXUH RIWKH ¿VFDO DGMXVWPHQW HIIRUW
LV DGMXVWHG DV SXEOLF GHEW LQFUHDVHV ,QWXLWLYHO WKH EDVLV IRUthese tests is that a tendency for primary surpluses to increase as public debt increaseover time will tend to support (dynamic) sustainability.The approach can be illustrated using the explicit uncertainty framework adopted by%RKQ
DERYH LV HIIHFWLYHOreplaced by equation (5): ∞B(t ) = E ∑ E j u ’(c (t + j )) / u ’(c (t ))S(t + 1) + lim Et E T +1u ’(c (t + j + 1)) / U ’(c (t ))B(t + t + 1) (5) j =0 T →∞Here E is the (mathematical) expectations operator22 and u’ (c (t + j))/u’ (c(t)) denotesthe marginal rate of substitution between consumption (c) in two adjacent time periods.Equation (5) is analogous to equation (3) but applies when uncertainty is explicitlymodeled and the (one period) discount factor is replaced by the potentially time-varyingmarginal rate of substitution in consumption. Using equation (5) the condition forsustainability is that the terminal debt stock discounted by the expected marginal rate ofVXEVWLWXWLRQ LQ FRQVXPSWLRQ DSSURDFKHV ]HUR DV T → ∞ , which implies that equation (6)holds. ∞B(t ) = E ∑ E j u ’(c (t + j )) / u ’(c (t ))S(t + 1) (6) j =0,W LV SRVVLEOH IROORZLQJ %RKQ
WR ZULWH WKH UHODWLRQVKLS EHWZHHQ WKH SULPDUsurplus and public debt in a simple linear based decision rule as in equation (7) below.In equation (7), lower case letters are used to denote the public debt (b) and primarysurplus (s) as proportions of GDP, P GHQRWHV WHPSRUDU VWDWLRQDU
LQÀXHQFHV RQ WKH21 Bohn (1998), in particular, has argued that some of the time series tests may reject sustainability in circumstances where fiscal policy is sustainable.22 All expectations are conditional on information through the current period.
10 | ADB Economics Working Paper Series No. 205primary balance to GDP ratio,23 and H LV D ZKLWH QRLVH HUURU WHUP 7KH FRHI¿FLHQW Umeasures the response of the primary balance to changes in the debt ratio while Emeasures the response of the primary balance to the temporary factors included in P.24s(t ) = ρ b(t − 1) + βµ (t ) + ε (t ) ε ⋅ (0,σ 2 ) (7)%RKQ
LV WKH NH HTXDWLRQ IRU DVVHVVLQJ ¿VFDO VXVWDLQDELOLW LQ WKLV IUDPHZRUN VLQFHLW GHWHUPLQHV ZKHWKHU WKH SULPDU EDODQFH UHVSRQGV LQ D VWDELOL]LQJ PDQQHU WR FKDQJHVin the debt ratio. A value for the response parameter U EHWZHHQ ]HUR DQG XQLW LPSOLHVWKDW WKH SULPDU VXUSOXV LQFUHDVHV DV WKH GHEW UDWLR ULVHV ZKLFK LV VXI¿FLHQW IRU GQDPLFsustainability. And the closer U LV WR XQLW WKH ODUJHU RU PRUH IRUFHIXO LV WKH ¿VFDOpolicy response to increases in the debt ratio. Conversely, if U LV ]HUR RU QHJDWLYH WKHimplication is that higher debt ratios lead either to no response of the primary surplus; orWR D VPDOOHU SULPDU VXUSOXV ,Q VXFK FLUFXPVWDQFHV ¿VFDO SROLF PD EH XQVXVWDLQDEOHif it leads to explosive public debt ratios.26 Based on this reasoning, (estimated) valuesof the U FRHI¿FLHQW DORQJ ZLWK WKH UHVSRQVH RI SULPDU EDODQFHV WR WHPSRUDU VKRFNV DVcaptured through the P WHUP FDQ EH XVHG WR ³VFRUH´ ¿VFDO SROLF UHVSRQVHV )RU H[DPSOHa value of U WKDW LV FORVH WR XQLW FDQ EH VFRUHG DV D ³IRUFHIXO´ SROLF UHVSRQVH DQG DYDOXH FORVH WR ]HUR DV D ³GDPSHG´ UHVSRQVH 7KH HVWLPDWHG YDOXHV RI E, on the otherKDQG PHDVXUH WKH UHVSRQVH RI ¿VFDO SROLF WR WHPSRUDU IDFWRUV VXFK DV WKH EXVLQHVVcycle (as measured, say, by gaps between actual and potential output); or to temporaryVZLQJV LQ SULPDU
Notwithstanding the role the U FRHI¿FLHQW SODV LQ GHWHUPLQLQJ WKH SURVSHFWV IRU GQDPLFVXVWDLQDELOLW LW QHHGV WR EH UHFRJQL]HG WKDW WKH EHKDYLRU RI GHEW UDWLRV RYHU WKH VKRUWWR PHGLXP WHUP KRUL]RQV DOVR GHSHQGV LPSRUWDQWO RQ WKH EHKDYLRU RI WKH JDS EHWZHHQthe interest rate on public debt and the growth rate of the economy. In circumstancesZKHUH WKH JURZWK UDWH LV DERYH WKH LQWHUHVW UDWH VKRUWWHUP GHEW WR *3 VWDELOL]DWLRQFDQ RFFXU HYHQ ZKHQ WKH SULPDU EDODQFH LV LQ GH¿FLW ,Q VXFK FLUFXPVWDQFHV D IDLOXUHRI WKH SULPDU EDODQFH WR LQFUHDVH DV GHEW LQFUHDVHV²ZKLFK LV VXI¿FLHQW IRU YHU ORQJ23 As discussed below, these temporary influences include temporary government spending shocks and the deviation of output from trend.24 Note that E and P can be appropriately dimensioned vectors.25 The reason why this is a sufficient (and not necessary) condition is that there may be nonlinear decision rules that are consistent with sustainability. In addition, in circumstances where nominal interest rates are below the growth rate of the economy, debt-to-GDP ratios can decline even in the presence of a primary deficit.26 Sustainability will also be influenced by whether the nominal interest rate is above or below the growth rate of nominal GDP.
Fiscal Sustainability in Developing Asia | 11VXVWDLQDELOLW ZKHQ WKH LQWHUHVW UDWH LV DERYH WKH JURZWK UDWH²ZLOO QRW QHFHVVDU SUHFOXGHa stable debt-to-GDP ratio in the near term. As we note below,27 in some subregions andcountries, the debt ratio has been stable even in circumstances when the primary balanceKDV EHHQ LQ GH¿FLW 2YHU ORQJ WLPH SHULRGV KRZHYHU ¿VFDO VXVWDLQDELOLW ZLOO EH VXSSRUWHGby the primary balance increasing as the debt ratio rises.,I WKH ¿VFDO VLWXDWLRQ LV VXVWDLQDEOH ρε (0,1) ), the long-run expected value of the debt ratiocan be written as in equation (8) below. _E (b(t )) = ( − µ + (1 − ρ )cov(1 + φ, b(t − 1)) / ρ (1 + φ ) − φ (8)Here I is the difference between the nominal interest rate and the nominal growth rate ofthe economy (assumed to be positive in the very long run), the term cov (..) represents acovariance term, and a bar is used to denote a long-run value of a variable. This equationdetermines the expected value of the long-run debt ratio, which depends critically not onlyon the value of U but also on the gap between the nominal interest rate and the nominalgrowth rate of the economy.0RUH JHQHUDOO WKH SULPDU EDODQFH IUDPHZRUN IRU ¿VFDO VXVWDLQDELOLW FDQ EH H[WHQGHGin a number of directions including allowing for nonlinear or lagged relationships betweenthe primary surplus and public debt, or by allowing for the response parameter (U) tovary over time. As regards nonlinearity issues, a particular interesting possibility is thatthere are threshold effects whereby the response of the primary balance to the debt stockPLJKW LQFUHDVH RU GLPLQLVK DV GHEW OHYHOV ULVH DV SDUW SHUKDSV HLWKHU RI ³ZDNH XS FDOOV´RU RI ³DGMXVWPHQW IDWLJXH´28 The presence of such effects would raise the possibility thatWKH LPSOLFDWLRQV RI ¿VFDO VOLSSDJHV GHSHQG RQ WKH GHEW OHYHOV DW ZKLFK WKH RFFXU DQGthat the resulting adjustment path could be different depending on the initial level of debt.A third of tests can be described as scenario or stress tests and have been widely usedby the International Monetary Fund, World Bank, and Asian Development Bank. TheseWHVWV DUH EDVHG RQ ³ZKDW LI´ RU ³VWUHVV WHVW´ WSH PRGHOLQJ HLWKHU ZLWK UHVSHFW WR WKHLPSOLFDWLRQV IRU ¿VFDO VXVWDLQDELOLW RI FRQWLQXLQJ ZLWK FXUUHQW ¿VFDO SROLFLHV DV GH¿QHG Ethe primary balance) or of identifying the gap between the primary balance required toSURGXFH ¿VFDO VXVWDLQDELOLW DQG WKH FXUUHQW ¿VFDO SROLF VHWWLQJVThe application of these scenario approaches can be illustrated by noting that theproportional change in the public debt ratio will (approximately) depend on the differencegap between (one plus) the nominal interest rate and (one plus) the growth rate ofnominal GDP (I
WRJHWKHU ZLWK WKH SULPDU ¿VFDO VXUSOXV DQG WKH GHEW UDWLR 7KLV LV JLYHQ27 See discussion in Section II.28 An extreme example of this would be the Debt Laffer curve effect in which there is a tipping point beyond which higher debt produces a reduction in the primary surplus ( U0). This case would be a fiscal policy analogy to the original Bulow-Rogoff model of sovereign debt problems and the Debt Laffer curve (Blanchard and Fischer 1989).
12 | ADB Economics Working Paper Series No. 205by equation (8) where denotes a proportional change and I denotes the differencebetween the nominal interest rate and the growth rate of nominal GDP. ∆b(t ) = I (t ) − s(t ) / b(t ) (8)8QGHU WKH ¿UVW WSH RI H[HUFLVH WKH WLPH SDWK IRU WKH GHEW UDWLR LV SURMHFWHG RQ WKH EDVLVof a given set of assumptions about the primary surplus (for example, which is constantin relation to GDP) and the gap between the nominal interest rate and the nominalgrowth rate (I
,Q WKH HYHQW WKDW WKLV SURGXFHV DQ ³H[SORVLYH´ LQFUHDVH LQ WKH GHEW UDWLRRYHU D JLYHQ WLPH KRUL]RQ WKH DVVXPHG RU FXUUHQW ¿VFDO SROLF VWDQFH ZRXOG EH MXGJHGunsustainable. This is because it might violate the terminal condition on debt and leadto the debt-to-GDP ratio increasing beyond bounds. Conversely, in the second type ofexercise, the approach would be to compute the primary surplus required to deliver agiven sustainable path for the debt ratio over the medium term (for example, that the debtUDWLR EH FRQVWDQW
LV GH¿QHG DV WKHdifference between the sustainable primary balance (ss) and the actual balance (sa),ZLWK WKH IRUPHU GH¿QHG DV WKH EDODQFH UHTXLUHG WR KROG WKH GHEW UDWLR FRQVWDQW DW LWVcurrent level.30ω (t ) = [(R(t ) − η (t ))b − s a (t )]
%RWK WKHVH WSHV RI VFHQDULR RU VWUHVV WHVWV DUH XVHIXO IRU VKHGGLQJ OLJKW RQ ¿VFDOsustainability. As in the case of the primary-balance approach to sustainability discussedabove, the gap between the nominal interest rate and the nominal growth rate of theeconomy (I) plays a key role. A common assumption is that this gap term will be positivebut, as noted in subsequent sections, there are several Asian economies that have grownfaster than the nominal interest rate over several years and in which I has been negative.$V GLVFXVVHG LQ 6HFWLRQ ,,, D FRPPRQ DVVXPSWLRQ LQ PDQ ¿VFDO VFHQDULR H[HUFLVHV LVthat the nominal interest rate exceeds the nominal growth rate of the economy. In theDEVHQFH RI VXFK D FRQGLWLRQ²ZKLFK FRXOG LPSO D GHSDUWXUH IURP GQDPLF HI¿FLHQF²very long-run terminal conditions on debt may no longer be binding and public debt canbecome unbounded.31 Accordingly, because a positive interest rate gap is necessaryIRU PHDQLQJIXO LQ¿QLWH KRUL]RQ VFHQDULRV LW LV VRPHWLPHV DVVXPHG WR KROG HYHQ RYHU29 If the economy is dynamically efficient, the modified golden rule implies that the nominal interest rate (or, at least the nominal marginal product of capital) will exceed the growth rate of nominal GDP and hence that I will be strictly positive. See Blanchard and Fischer (1989) for further discussion.30 Constancy of the debt ratio is a simplifying assumption. Alternative approaches could allow for any sustainable debt path.31 See Blanchard and Fischer (1989) for further discussion of dynamic efficiency and the modified golden rule. When an economy satisfies the modified golden rule, the nominal interest rate will exceed the nominal growth rate of the economy.
Fiscal Sustainability in Developing Asia | 13the short to medium run (IMF 2003). Over a few years, however, the possibility thateconomies may grow more rapidly than the interest rate on public debt must beconsidered, even if this raises issues about very long-run terminal conditions.Even though the model-based and scenario approaches are different, there are a numberof connections between them that can be usefully exploited. Both approaches addressWKH OLQN EHWZHHQ WKH ¿VFDO DGMXVWPHQW HIIRUW PHDVXUHG WKURXJK WKH SULPDU EDODQFH
and the debt ratio. In the primary balance approach, the policy response parameter forthe primary balance (U) is estimated from the data, whereas in the scenario approaches,WKH SDUDPHWHU PD EH LPSRVHG IURP ³RXWVLGH´ DFFRUGLQJ WR ZKDW LV MXGJHG QHFHVVDUWR SURGXFH ¿VFDO VXVWDLQDELOLW ,Q SULQFLSOH WKH HVWLPDWHG DGMXVWPHQW SDUDPHWHUV IURPthe model-based approaches to sustainability can be used to enrich the scenario-basedDSSURDFKHV WR ¿VFDO VXVWDLQDELOLW DQG DFFRXQW FDQ DOVR EH WDNHQ RI GHEW WLSSLQJ SRLQWVin which the parameter may be different at various debt levels. These are the approachesWDNHQ LQ WKH ¿QDO VHFWLRQ RI WKH SDSHUIII. Fiscal Diagnostics: Evolution of Key Fiscal Indicatorsacross Asia over Time7KLV VHFWLRQ UHYLHZV ¿VFDO EDODQFHV SXEOLF VSHQGLQJ DQG UHYHQXHV DQG SXEOLF GHEW(in relation to GDP) across the region and over time, with particular attention to theEUHDNGRZQ RI ¿VFDO SRVLWLRQV LQWR SULPDU DQG QRQSULPDU FRPSRQHQWV WKH UHODWLRQVKLSEHWZHHQ SULPDU ¿VFDO EDODQFHV DQG SXEOLF GHEW UDWLRV DQG WKH OLQNV EHWZHHQ DFWXDODQG GHEW VWDELOL]LQJ SULPDU ¿VFDO EDODQFHV ,Q DGGLWLRQ WKH VHFWLRQ UHYLHZV NH IHDWXUHVRI FRXQWULHV¶ ¿VFDO SRVLWLRQV LQFOXGLQJ WKH UHODWLRQVKLS EHWZHHQ SXEOLF GHEW UDWLRV DQGYDULRXV PHDVXUHV RI HFRQRPLF VL]H YRODWLOLW DQG UHYHQXH UDWLRV7KH DVVHVVPHQW RI SXEOLF ¿QDQFHV LQ WKH UHJLRQ LV D GDXQWLQJ FKDOOHQJH DQG WKHUH DUHKXJH GLI¿FXOWLHV DVVHPEOLQJ D FRPSUHKHQVLYH VHW RI ¿VFDO DQG SXEOLF GHEW GDWD IRU DOOthe developing country members of the ADB. Not only are there familiar problems ofoccasional missing observations over time and across countries, there are also a numberRI VHULRXV VKRUWFRPLQJV DV UHJDUGV WKH DYDLODELOLW RI FRQVLVWHQW DQG FRPSUHKHQVLYH ¿VFDOdata in some countries and, in particular, as regards public debt obligations. In addition,only a very limited number of countries publish public sector asset positions (as well asSXEOLF GHEW GDWD
PHDQLQJ WKDW ¿VFDO VXVWDLQDELOLW DQDOVLV QHHGV IRU WKH PRVW SDUW WREH EDVHG RQO RQ RQH SDUW RI WKH ¿VFDO SLFWXUH OLDELOLWLHV UDWKHU WKDQ DVVHWV32). In addition,even in the case of countries that report interest payments on debt, consistent data arenot generally available for the actual interest rates on public debt and the extent to whichdebt may be serviced at concessional or market interest rates. Moreover, to the extent32 Income on (any) public asset is implicitly recorded under government revenue.
14 | ADB Economics Working Paper Series No. 205WR ZKLFK VRPH FRXQWULHV LVVXH SXEOLF GHEW WR ³FDSWLYH´ ORFDO EXHUV VXFK DV ORFDO EDQNVpension funds etc.), the true economic costs of servicing that debt may be different fromthe recorded costs.There is also the problem that countries do not generally report contingent (off-balanceVKHHW
WKHVH DUH RIWHQ WKH NH IDFWRUV LQ LQÀXHQFLQJ ¿VFDO VXVWDLQDELOLWRYHU WLPH DQG DUH ZKHUH WKH ³¿VFDO VXUSULVHV´ RIWHQ RFFXU33 The omission of contingent¿VFDO OLDELOLWLHV LPSOLHV WKDW WKH DSSURDFK DGRSWHG LQ WKH SDSHU DUJXDEO UHSUHVHQWV WKH³EHVW FDVH´ LQ VR IDU DV WKH DFFRXQWLQJ IRU FRQWLQJHQW OLDELOLWLHV ZRXOG OLNHO OHDG WR WKHpossibility of higher rather than lower future debt ratios, and potentially larger threats to¿VFDO VXVWDLQDELOLWAt the most fundamental level, there is also the issue of how broadly or narrowly theSXEOLF VHFWRU VKRXOG EH GH¿QHG ,GHDOO ¿VFDO VXVWDLQDELOLW DQDOVLV VKRXOG FRYHU DOODVSHFWV RI SXEOLF VHFWRU RSHUDWLRQV DQG DFWLYLWLHV WKDW KDYH ¿VFDO LPSOLFDWLRQV VXJJHVWLQJa very broad and comprehensive approach. Such comprehensiveness is probably bestDFKLHYHG E IRFXVLQJ RQ WKH RYHUDOO SXEOLF VHFWRU²LQFOXGLQJ DOO DFWXDO DQG TXDVL VWDWHRZQHG HQWHUSULVHV DQG JRYHUQPHQWOLQNHG FRPSDQLHV²EXW LW LV IUHTXHQWO GLI¿FXOW WRGUDZ WKH OLQH LQ GH¿QLQJ WKH SXEOLF VHFWRU34 and comprehensive data for the entire publicVHFWRU LV IUHTXHQWO GLI¿FXOW WR REWDLQ $QG HYHQ LQ WKRVH FDVHV ZKHUH LW LV RQO IHDVLEOH WRcover the (formal) government sector, there are often issues related to the availability ofdata for different levels of government (central, state, and local) and its consistency overWLPH 7KHUH DUH QR VLPSOH VROXWLRQV WR WKHVH DQG UHODWHG ³ERXQGDU´ SUREOHPV LQ GH¿QLQJWKH SXEOLF VHFWRU $QG IUHTXHQWO LW LV QHFHVVDU WR EDVH WKH DQDOVLV RQ D ³QDUURZ´GH¿QLWLRQ RI JRYHUQPHQW GXH WR ODFN RI GDWD IRU EURDGHU GH¿QLWLRQV RI WKH SXEOLF VHFWRU)RU SXUSRVHV RI WKH SDSHU DQQXDO GDWD ZDV DVVHPEOHG RQ JRYHUQPHQW ¿VFDO SRVLWLRQVand debt (and other variables) for over 30 developing member countries of ADB fromWKH HDUO V WKURXJK :KHUH SRVVLEOH GDWD IRU WKH JHQHUDO JRYHUQPHQW ZDVXVHG EXW LQ VHYHUDO LQVWDQFHV FHQWUDO JRYHUQPHQW GDWD KDG WR VXI¿FH ,Q VRPH FDVHVVHULHV KDG EUHDNV RU JDSV DQG YDULRXV DSSURDFKHV ZHUH XVHG WR ³JDS ¿OO´ WKHVH VHULHVEven with these approaches, however, the panel data that was collected is somewhatunbalanced, with the most comprehensive time series coverage generally occurring in themore developed countries in the region.3533 Much of the deterioration in fiscal positions during the Asian financial crisis was associated with bailouts of distressed banks and other financial institutions.34 This is especially the case in economies transitioning from central planning and state ownership to a market economy.35 Implicitly, this can lead to a large country bias in which the large countries for which data are available have a disproportionate effect on the results.
Fiscal Sustainability in Developing Asia | 15One key question that arises in presenting (and interpreting) large amounts of countrydata concerns the way in which the data are aggregated to produce subregionalaggregates. Generally, the aggregates in the Asian Development Outlook $% E
are based on gross national income weighting in which large countries implicitly receivea larger weight than small countries. Even though such an approach is appropriate forDJJUHJDWLQJ YDULDEOHV VXFK DV *3 LW LV QRW FOHDU ZKHWKHU LW VKRXOG EH XVHG IRU ¿VFDOYDULDEOHV ZKHUH WKHUH LV QR PHDQLQJIXO FRQFHSW RI D UHJLRQZLGH ¿VFDO SRVLWLRQ %DVHGRQ WKHVH WSHV RI FRQVLGHUDWLRQV²DQG DOVR EHFDXVH WKH HPSLULFDO DQDOVLV LQ WKH QH[WVHFWLRQ LV EDVHG RQ LQGLYLGXDO FRXQWU H[SHULHQFH²WKH ¿VFDO DJJUHJDWHV IRU WKH UHJLRQpresented in this section are all unweighted (rather than weighted) averages of data forindividual countries and subregions.36,Q DGGLWLRQ WR FRQVLGHULQJ ¿VFDO EHKDYLRU LQ GLIIHUHQW FRXQWULHV WKH FURVV VHFWLRQDOGLPHQVLRQ
GLVSOD NH ¿VFDO LQGLFDWRUV IRU WKH UHJLRQ RYHU WKH WKUHHVXESHULRGV UHIHUUHG WR DERYH DQG IRU ± DV D ZKROH ,Q OLQH ZLWK $%¶VFODVVL¿FDWLRQ WKH UHJLRQ LV EURNHQ GRZQ LQWR ¿YH VXEUHJLRQV HQWUDO $VLD (DVW $VLD WKH3DFL¿F 6RXWKHDVW $VLD DQG 6RXWK $VLD38 %RWK RYHUDOO DQG SULPDU ¿VFDO EDODQFHV LQTable I are measured as the difference between receipts and spending so that a positiveVLJQ LPSOLHV D ¿VFDO VXUSOXV ZKLOH D QHJDWLYH VLJQ LPSOLHV D GH¿FLW6HYHUDO REVHUYDWLRQV FDQ EH PDGH RQ WKH EDVLV RI WKH ¿VFDO LQGLFDWRUV LQ 7DEOH ,L
)RU WKH SHULRG DV ZKROH WKHUH LV D UHODWLYHO QDUURZ GLVSHUVLRQ RI ¿VFDO EDODQFHV across the region. $FURVV WKH UHJLRQ ¿VFDO EDODQFHV DUH LQ VPDOO GH¿FLW RYHU WKH HQWLUH ± SHULRG EXW WKH GH¿FLWV DUH VRPHZKDW KLJKHU LQ WKH 3DFL¿F and especially in South Asia (at over 5% of GDP). With the exception of Central36 Given this approach, there is the possibility that the aggregated data may be distorted by large outlier observations in very small countries. Both mean and median values were considered for data presentation, however, and made little difference.37 Preliminary estimates and forecasts of fiscal positions (including recent fiscal stimulus packages) for 2009 and 2010 are presented and discussed in the final section.38 The countries included in each group are listed in the notes accompanying the tables and charts.39 All balances are measured as the difference between revenues and expenditures with the implication that a positive sign denotes a surplus while a negative sign denotes a deficit.
16 | ADB Economics Working Paper Series No. 205 and South Asia, which were not very badly affected by the Asian crisis,40 other VXEUHJLRQV VDZ WKHLU ¿VFDO GH¿FLWV LQFUHDVH LQ WKH ODWH V EXW WKH GH¿FLWV ZHUH then generally pared back.(ii) Across the entire period, there is a wide range of experience as regards primary ¿VFDO EDODQFHV ZLWK VRPH VXEUHJLRQV DYHUDJLQJ VXUSOXVHV DQG RWKHUV DYHUDJLQJ GH¿FLWV 6RXWKHDVW $VLD KRZHYHU LV WKH RQO VXEUHJLRQ UXQQLQJ D ODUJH SULPDU surplus, while subregions such as South Asia have on average been running UHODWLYHO ODUJH DQG SHUVLVWHQW SULPDU GH¿FLWVLLL
:LWK WKH H[FHSWLRQ RI WKH 3DFL¿F ZKHUH JRYHUQPHQW VSHQGLQJ DQG UHYHQXHV KDYH been close to or above 40% of GDP, the average levels of government spending and revenues across the region are relatively low and clustered mainly at the low end of the 20–25% of GDP range. These ratios are well below the averages in many other part of the world and, in particular, in developed Europe (IMF 2003, (XURSHDQ RPPLVVLRQ D
Table 1: Fiscal Indicators, Ratio to GDP, 1990–2008 (percent ) Subregion Period Public Primary Fiscal Government Goverment Interest Debt Surplus Surplus Expenditure Revenue Payments Central Asia 1 35.10 −4.70 −7.00 21.00 14.00 1.30 2 68.70 −3.50 −5.50 20.20 14.70 1.40 3 34.50 1.00 0.20 25.00 25.20 0.80 All periods 38.00 0.30 −0.70 24.30 23.60 0.90 East Asia 1 17.30 −0.80 −1.40 18.20 16.70 1.00 2 35.20 −3.50 −3.70 23.80 19.00 1.70 3 32.20 0.90 0.20 21.80 22.00 1.00 All periods 27.00 −0.20 −0.80 20.70 19.70 1.10 The Pacific 1 39.00 1.90 −0.20 38.70 38.50 2.40 2 36.70 −0.50 −2.10 36.40 34.30 2.20 3 40.20 −1.70 −2.40 42.70 40.10 1.50 All periods 39.50 −0.40 −1.70 40.70 38.90 1.80 Southeast Asia 1 48.30 4.80 2.40 19.40 21.80 2.80 2 57.60 −1.00 −2.40 21.40 18.70 2.10 3 56.70 0.30 −1.70 20.70 19.10 2.20 All periods 53.90 1.70 −0.40 20.30 20.00 2.40 South Asia 1 63.60 −2.60 −6.00 27.50 21.60 3.30 2 62.00 −1.90 −5.00 24.00 19.10 3.10 3 65.10 −2.10 −5.10 25.50 20.50 2.90 All periods 64.20 −2.30 −5.40 26.10 20.80 3.1040 Fiscal developments in Central Asia during the 1990s were influenced importantly by the international treatment of the debts of the former Soviet Union.
Fiscal Sustainability in Developing Asia | 17Box 1: Various Fiscal Indicators Debt/GDP Ratio by Subregions and Periods Debt/GDP Ratio by Subregions and Periods (mean values, percent) (median values, percent) Central Asia Central Asia East Asia East Asia 1990−1997 The Pacific 1990−1997 The Pacific Southeast Asia Southeast Asia South Asia South Asia Central Asia Central Asia East Asia East Asia 1998−1999 The Pacific 1998−1999 The Pacific Southeast Asia Southeast Asia South Asia South Asia Central Asia Central Asia East Asia East Asia 2000−2008 The Pacific 2000−2008 The Pacific Southeast Asia Southeast Asia South Asia South Asia 0 20 40 60 80 0 20 40 60 80 Debt/GDP Ratio by Subregions and Years Debt/GDP Ratio by Subregions and Years (mean values, percent) (median values, percent)80 8060 6040 4020 20 0 0 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 Year Year Central Asia East Asia The Pacific Central Asia East Asia The Pacific Southeast Asia South Asia Southeast Asia South Asia Debt/Revenue Ratio by Subregions and Periods Debt/Revenue Ratio by Subregions and Periods (mean values, percent) (median values, percent) Central Asia East Asia Central Asia East Asia 1990−1997 The Pacific 1990−1997 The Pacific Southeast Asia Southeast Asia South Asia South Asia Central Asia Central Asia East Asia East Asia 1998−1999 The Pacific 1998−1999 The Pacific Southeast Asia Southeast Asia South Asia South Asia Central Asia Central Asia East Asia East Asia 2000−2008 The Pacific 2000−2008 The Pacific Southeast Asia Southeast Asia South Asia South Asia 0 100 200 300 400 500 0 100 200 300 400 500 Debt/Revenue Ratio by Subregions and Years Debt/Revenue Ratio by Subregions and Years (mean values, percent) (median values, percent)600 600400 400200 200 0 0 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 Year Year Central Asia East Asia The Pacific Central Asia East Asia The Pacific Southeast Asia South Asia Southeast Asia South Asia
18 | ADB Economics Working Paper Series No. 205(iv) Public debt ratios in the region display considerable heterogeneity and variation over time. Across much of the region, public debt ratios have on average been relatively low (below 40–50% of GDP) with the notable exceptions of Southeast and South Asia. In the case of Southeast Asia, the ratio has been in the range of 50–60% of GDP in the period since the Asian crisis. In South Asia, on the other hand, the ratio has been persistently above 60%. Public debt rose sharply in HQWUDO $VLD LQ WKH ODWH V EXW WKHQ IHOO VKDUSO LQ WKH FRQWH[W RI D QXPEHU RI ¿VFDO DGMXVWPHQW DQG GHEW UHVWUXFWXULQJ SURJUDPV(v) Across the region, interest payments on the public debt have been around 1-3% of GDP and show no clear tendency to increase over time. Generally, subregions with the highest debt to GDP ratios also have the highest levels of interest SDPHQWV WR *3 5HÀHFWLQJ GLIIHUHQW IXQGLQJ FRVWV KRZHYHU UHODWHG LQWHU DOLD WR the access to concessional funding, the relationship between interest payment and debt is not always very close.)LJXUH SURYLGHV DQRWKHU SHUVSHFWLYH RQ ¿VFDO SRVLWLRQV LQ WKH UHJLRQ 7KH FKDUWGLVSODV DYHUDJH RYHUDOO
DFURVVHDFK RI WKH ¿YH VXEUHJLRQV RYHU WKH HQWLUH WLPH SHULRG ± ,Q SULQFLSOH WKHREVHUYHG UHODWLRQVKLS EHWZHHQ WKH RYHUDOO ¿VFDO EDODQFH DQG WKH GHEW UDWLR DFURVV WKHsubregions could be either positive or negative. The relationship would be negative ifWKH VXEUHJLRQV ZLWK KLJK SXEOLF GHEW UDWLRV ZHUH UXQQLQJ VPDOOHU ¿VFDO VXUSOXVHV RYHUthe period. Conversely, if subregions with higher public debt ratios were running larger¿VFDO VXUSOXVHV WKHQ WKH UHODWLRQVKLS ZRXOG EH D SRVLWLYH RQH $FFRUGLQJO WKH QDWXUH RIWKH UHODWLRQVKLS EHWZHHQ ¿VFDO EDODQFHV DQG GHEW SURYLGHV LQIRUPDWLRQ DERXW ZKHWKHU¿VFDO DGMXVWPHQW DFURVV WKH VXEUHJLRQV LV FRQVLVWHQW ZLWK VWDELOL]LQJ UHVSRQVHV RI ¿VFDObalances to debt.41Figure 1: Fiscal Balances and Debt (1990–2008 averages) 0 sea eaa ceaFiscal Surplus (% GDP) paa −2 Linear regression line (excluding South Asia) −4 soa −6 30 40 50 60 70 Debt (% GDP)41 Note that the causation between the fiscal balance and the debt ratio runs in both directions. Larger fiscal deficits can lead to larger increases in debt ratios and, conversely, countries can seek to reduce fiscal deficits in order to stabilize or reduce public debt ratios.
Fiscal Sustainability in Developing Asia | 19Inspection of Figure 1 reveals that there is a slight positive relationship between overall¿VFDO VXUSOXVHV DQG SXEOLF GHEW DFURVV WKH UHJLRQ ZKHQ 6RXWK $VLD LV H[FOXGHG 7KLVimplies that, over the entire period, most subregions with higher public debt ratiosWHQGHG WR UXQ ODUJHU ¿VFDO VXUSOXVHV RU VPDOOHU ¿VFDO GH¿FLWV RQYHUVHO LQ 6RXWK $VLDin particular, a high debt ratio over the entire period is accompanied by relatively large¿VFDO GH¿FLWV DQG WKHUH LV OLWWOH HYLGHQFH LQ WKLV VXEUHJLRQ RI ORQJUXQ ¿VFDO DGMXVWPHQWbeing under way. The time snapshots for the subregions included in the following charts)LJXUH
EURDGO FRQ¿UP WKHVH SDWWHUQV WKURXJK WKH QDWXUH RI WKH UHODWLRQVKLS EHWZHHQ¿VFDO EDODQFHV DQG GHEW UDWLRV YDULHV DFURVV WKH ¿YH VXEUHJLRQV IRU WKH WKUHH VXESHULRGV± ± DQG ± (VVHQWLDOO WKH GLIIHUHQW UHODWLRQVKLSV DFURVV WKHWKUHH VXESHULRGV UHÀHFW WKH IDFW WKDW SHULRGV VXFK DV ± ZHUH SHULRGV ZKHQ GHEWincreased sharply during the Asian crisis. On the other hand, periods such as after 2000KDYH EHHQ SHULRGV RI ¿VFDO DGMXVWPHQW HVSHFLDOO LQ (DVW DQG 6RXWKHDVW $VLDFigure 2: Fiscal Balances and Debt(averages by periods and subregions) sea 2 0 eaacea paaFiscal Surplus (% GDP) eaa sea −2 paa paa sea eaa −4 soa soa cea −6 soa cea −8 20 30 40 50 60 70 Debt (% GDP) 1990-1997 1998-1999 2000-2008$QRWKHU SHUVSHFWLYH RQ WKH QH[XV EHWZHHQ ¿VFDO SROLF DQG SXEOLF GHEW LV SURYLGHG Econsidering the behavior of primary balances across subregions and over time. Figures DQG GLVSOD WKH UHODWLRQVKLS EHWZHHQ WKH SULPDU ¿VFDO VXUSOXV DQG SXEOLF GHEW ERWKDV D SHUFHQWDJH RI *3
IRU WKH ¿YH VXEUHJLRQV RI GHYHORSLQJ $VLD RYHU WKH HQWLUH ±2008 time period. The behavior over the three subperiods is displayed in Figure 4.
20 | ADB Economics Working Paper Series No. 205Figure 3: Primary Fiscal Balances and Debt (1990–2008 averages) 1Primary Fiscal Surplus (% GDP) sea cea 0 paa eaa Linear regression line (excluding South Asia) −1 −2 soa 30 40 50 60 70 Debt (% GDP)Figure 4: Primary Fiscal Balances and Debt (averages by periods and subregions) 4 seaPrimary Fiscal Surplus (% GDP) 2 paa cea eaa 0 eaa paa sea paa −2 sea soa soa soa eaa cea −4 cea 20 30 40 50 60 70 Debt (% GDP) 1990-1997 1998-1999 2000-2008In interpreting the two charts, it is useful to note that a positive relationship betweenSULPDU VXUSOXVHV DQG GHEW LPSOLHV WKDW ¿VFDO SROLF KDV EHHQ VXSSRUWLQJ VXVWDLQDELOLWThis is because a positive relationship can be interpreted as implying that the primaryVXUSOXV LQFUHDVHV RU WKH SULPDU GH¿FLW LV UHGXFHG
DV WKH GHEW UDWLR LQFUHDVHVConversely, a negative (or no) relationship between these variables may not be consistentZLWK ¿VFDO VXVWDLQDELOLW RYHU YHU ORQJ WLPH SHULRGV42 As shown in Figures 3 and 4, thereis a relatively strong positive relationship between primary surpluses and debt acrossthe region when South Asia is excluded. This conforms to the pattern for the regionsGLVFXVVHG DERYH ZLWK UHJDUG WR WKH UHODWLRQVKLS EHWZHHQ WKH RYHUDOO
¿VFDO EDODQFHVand debt. As evident from Figure 4, this pattern holds for the most part across the three42 Over short periods during which the interest rate is below the growth rate of the economy, however, the fiscal positions may “appear” sustainable and the debt ratio may be falling.
Fiscal Sustainability in Developing Asia | 21VXESHULRGV VLQFH RQFH DOORZDQFH LV PDGH IRU WKH $VLDQ FULVLVUHODWHG EORZRXW RI¿VFDO SRVLWLRQV LQ WKH ODWH V7KH UHODWLRQVKLS EHWZHHQ DFWXDO DQG GHEWVWDELOL]LQJ SULPDU EDODQFHV DV GLVSODHG LQFigure 5,43 SURYLGHV DQRWKHU SHUVSHFWLYH RQ ¿VFDO VXVWDLQDELOLW 6LQFH ¿VFDO VXVWDLQDELOLWneeds to be evaluated at the country (rather than subregional) level, the chart refers toSDUWLFXODU FRXQWULHV IURP WKH ¿YH VXEUHJLRQV DQG LGHQWL¿HV WKH ORFDWLRQ RI WKH DYHUDJHIRU HDFK VXEUHJLRQ 7KH YHUWLFDO D[LV RI WKH FKDUW PHDVXUHV WKH SULPDU ¿VFDO EDODQFHVUHTXLUHG WR VWDELOL]H GHEW UDWLRV EDVHG RQ DYHUDJH QRPLQDO *3 JURZWK DQG LQWHUHVW UDWHVover the period 2000 through 2008 for each of the displayed countries.44 7KH KRUL]RQWDOaxis displays the average primary balance over the same period with a positive numberLPSOLQJ D SULPDU VXUSOXV DQG D QHJDWLYH LPSOLQJ D SULPDU GH¿FLW $FFRUGLQJO WKHREVHUYDWLRQ IRU HDFK FRXQWU LV D SDLU RI DFWXDO DQG GHEWVWDELOL]LQJ SULPDU EDODQFHVDQG WKH FKDUW LV GLYLGHG LQWR WZR ]RQHV ,Q WKH WRS FKDUW SRLQWV DERYH WKH GHJUHHOLQH GHQRWH SRLQWV ZKHUH WKH GHEW VWDELOL]LQJ SULPDU VXUSOXV LV DERYH WKH DFWXDO SULPDUsurplus and the debt ratio is increasing. Points below the 45-degree line denote situationsZKHQ WKH DFWXDO SULPDU VXUSOXV LV ODUJHU WKDQ WKH GHEWVWDELOL]LQJ VXUSOXV DQG WKH GHEWratio is on average decreasing.45 Conversely, points along the 45-degree line denoteHTXDOLW EHWZHHQ WKH DFWXDO DQG GHEWVWDELOL]LQJ SULPDU EDODQFHV DQG VWDELOLW LQ WKH GHEWto-GDP ratio.43 The calculation of debt-stabilizing primary balances was discussed in the preceding section. Note that debt- stabilizing primary balances can be negative (i.e., primary deficits) if nominal interest rates are below the nominal growth rate of the economy. Such a condition is unlikely to hold over very long time periods and very long-run assessments of fiscal sustainability are typically based on the assumption that the interest rate exceeds the growth rate.44 The calculation is based on equations (8) through (9) in the preceding section.45 As the scale of the axes is different in the lower chart, a comparable 45 degree line cannot be drawn. The indicated line plays a comparable role to the 45 degree line in the top chart.
22 | ADB Economics Working Paper Series No. 205Figure 5: Actual and Debt-Stabilizing Primary Balances 15 10 5 P 0 E P PA A PP P P A C C E E S S S S A AC S C C −5 A S A −10 S Bold letters represent subregional averages CStabilizing Balance (% GDP) C −15 −15 −10 −5 0 5 10 15 Note: GDP growth rates and interest rates: 2000-2008 average. Debt and primary balance: 2005-2008 average. 1.5 S S S 1.0 A C P S S S A P A A PA A 0.5 C E A C E EPC S C P P C C 0 −15 −10 −5 0 5 10 15 Actual Primary Balance (% GDP, 2005-2008 average) C: Central Asia E: East Asia P: The Pacific S: South Asia A: Southeast Asia Note: Interest rate assumed 150bp above 2000-2008 average GDP growth rate. Debt and primary balance: 2005-2008 averages.Based on the top part of Figure 5, it is apparent that over the period 2000–2008 mostFRXQWULHV LQ WKH UHJLRQ IDOO ZHOO ZLWKLQ WKH ]RQH RI VXVWDLQDELOLW HYHQ WKRXJK D QXPEHUare close to the 45-degree line. Most notably, several countries in the region have onaverage been running primary surpluses between 0–5% of GDP, with a small numberrunning even larger surpluses. Findings of this kind are consistent with those of severalRWKHU VWXGLHV RI ¿VFDO VXVWDLQDELOLW ZLWKLQ WKH UHJLRQ DQG WKRVH RI EURDGHU VWXGLHV WKDWKDYH DVVHVVHG ¿VFDO VXVWDLQDELOLW DFURVV D QXPEHU RI UHJLRQV +RUQH ,0) 0HQGR]D DQG 2VWU
7KLV UHÀHFWV WKH IDFW WKDW HFRQRPLFconditions in several countries have been such that the interest rate on debt has beenbelow the growth rate of the economy. Under such circumstances, as illustrated in)LJXUH GHEW UDWLRV FRXOG EH VWDELOL]HG HYHQ ZKHQ UXQQLQJ SULPDU GH¿FLWV WKDW KDYHin some cases been larger than 5% of GDP.46 Either because this condition is unlikelyWR KROG LQGH¿QLWHO RU RQ DFFRXQW RI WKH SRVVLELOLW WKDW WKH DFWXDO LQWHUHVW UDWH PD QRWUHÀHFW WKH WUXH VKDGRZ SULFH RI GHEW WKH UHVXOWV QHHG WR EH WUHDWHG ZLWK FDUH 8OWLPDWHOUREXVW GHEWVWDELOL]LQJ ¿VFDO SROLFLHV QHHG WR EH VXFK DV WR VWDELOL]H WKH GHEW UDWLR LQcircumstances when the interest rate is above the growth rate.$JDLQVW WKLV EDFNJURXQG LW LV DOVR XVHIXO WR FRQVLGHU WKH SRVVLEOH LPSOLFDWLRQV IRU ¿VFDOsustainability of interest rates exceeding the growth rate of nominal GDP. Typically,interest rates on the external public debt of many countries exceed the nominal growthrate so that these rates can, in principle, be used in place of (relatively) low domesticinterest rates. Unfortunately, however, it is not possible to obtain such data for many ofthe countries included in Figure 5. Purely for illustrative purposes, the approach takenLQ WKH ORZHU SDUW RI )LJXUH ZDV WR UHGR WKH GHEW VWDELOL]LQJ FDOFXODWLRQV XQGHU WKHDVVXPSWLRQ WKDW WKH ³XQGHUOLQJ´ JDS EHWZHHQ WKH QRPLQDO LQWHUHVW UDWH DQG WKH JURZWKrate of nominal GDP for all regions was equal to 150 basis points.47 As can be seenfrom the chart, this positive interest rate gap has the effect of reducing the number ofFRXQWULHV ZLWK ¿VFDO SRVLWLRQV WKDW DUH XQDPELJXRXVO VXVWDLQDEOH DQG SXVKLQJ VRPH LQWRthe unsustainable quadrant. The reason for this is straightforward: A positive interest rateJDS LPSOLHV WKDW DOO FRXQWULHV DUH UHTXLUHG WR UXQ SULPDU VXUSOXVHV WR VWDELOL]H RU EULQJGRZQ WKH GHEW UDWLR 7R WKH H[WHQW WKDW ¿VFDO VXVWDLQDELOLW LQ VRPH FRXQWULHV GHSHQGVRQ D IDYRUDEOH²EXW XQOLNHO WR SHUVLVW²DELOLW WR ¿QDQFH GHEW DW LQWHUHVW UDWHV EHORZthe growth rate, the results in this lower chart may give a more realistic long-term senseRI ¿VFDO VXVWDLQDELOLW $W WKH OHDVW WKH FKDUW LQGLFDWHV WKDW ¿VFDO VXVWDLQDELOLW LQ VRPHcountries in the past has been facilitated by very favorable growth and interest rateconditions. Were these conditions to be less favorable in the future, the maintenance of¿VFDO VXVWDLQDELOLW ZRXOG EH PRUH FKDOOHQJLQJ7KH ODVW IHZ FKDUWV DQG WDEOHV SURYLGH NH VWUXFWXUDO LQIRUPDWLRQ IRU WKH ¿VFDO VLWXDWLRQLQ WKH UHJLRQ IRU ± 7KH ¿UVW WZR FKDUWV )LJXUHV DQG
VXPPDUL]H WKHrelationship between public debt and GDP in both absolute and per capita terms acrossthe region. As shown, there is a strong positive relationship between GDP and debt,and a modest negative relationship between per capita GDP and the debt ratio, albeit46 Equations (8) and (9) above describe the relationship between the debt stabilizing primary balance and the debt ratio. Effectively, the primary surplus required to stabilize the debt ratio is equal to the gap between the nominal interest rate and the nominal growth rate of GDP multiplied by the debt ratio. When the gap term is negative, the debt ratio can be stabilized with a primary deficit.47 This is based on the average interest rate growth differential in the 2003 IMF study on fiscal sustainability (IMF 2002). Obviously, in practice, the actual differential will differ across countries and may be higher in countries where there are concerns about fiscal sustainability. To the extent that this is the case, some high-debt countries may be even closer to unsustainable fiscal positions.
24 | ADB Economics Working Paper Series No. 205with both Singapore and Hong Kong, China being outliers.48 Effectively, rich countriesin the region do not have higher public debt ratios than poorer countries. Within theregion, there is some indication that the highest public debt ratios are among the poorercountries, but the relationship is relatively weak.Figure 6: Public Debt and GDP 8 CHN KOR IND 6 HKG IDN THA MYS SGP PHL PAKGDP (log) 4 BGD VNM KAZUZB LKA AZE 2 NPL PNG ARM KHM GEO LAO FJI TJKKGZ MNG 0 MDV SLB BTN VUT WSM −2 PLW MHL −4 −2 0 2 4 6 Debt (log) Note: 1990-2008 averages in log scale. Specific country averages depend on data availability.Figure 7: Public Debt and Per Capita GDP 4 HKG SGP KORGDP per Capita (log) 2 PLW MYS MDV MHL THA FJI KAZ AZE VUT WSM CHN ARM SLB PHL 0 IDN GEO LKA PNG UZB BTN VNM PAK MNG IND BGD KGZ LAO KHM NPL TJK −2 0 1 2 3 4 5 Debt/GDP ratio (log) Note: 1990-2008 averages in log scale. Specific country averages depend on data availability.7UDGLWLRQDOO SXEOLF GHEW UDWLRV KDYH EHHQ DVVHVVHG LQ WHUPV RI WKHLU VL]H LQ UHODWLRQ WRGDP, with the assumption that the level of GDP provides a useful measure of capacityto service the debt. In practice, however, GDP may not be a very good measure of³FDSDFLW´ LQ VR IDU DV LW GRHV QRW PHDVXUH WKH UHVRXUFHV DYDLODEOH WR WKH JRYHUQPHQW IRUdebt service. Under quite general conditions, the debt-to-revenue ratio might be a morerelevant measure of capacity in so far as it measures how much of GDP the government48 In the case of Singapore, the high public debt ratio is matched by an even larger stock of government assets.