加息時代的債務政治.pdf

《加息時代的債務政治.pdf》由會員分享，可在線閱讀，更多相關《加息時代的債務政治.pdf（51頁珍藏版）》請在三個皮匠報告上搜索。

1、The Politics of Debt in the Era of Rising RatesMarina Azzimonti(Federal Reserve Bank of Richmond)aNirvana Mitra(CAFRAL-Reserve Bank of India)November,2024aViews expressed are those of the authors and do not reflect views of the FRB Richmond or CAFRAL.This paperUnderstand how time-varying risk-free r

2、ates r impact near-term sovereign debt manage-ment and default decisions in EMs.Political landscape affects the response to rising rates and associated default risk.We address two questions:1.How do rising rates affect EM when there is scope for corruption in countries with weakinstitutions?2.How do

3、 International Financial Institutions(IFIs)short-term lending programs affect thesedynamics?The Politics of DebtArgentinaBrazilBulgariaColombiaCosta RicaEcuadorEgyptEl SalvadorGreeceIndiaIndonesiaKazakhstanLesothoMexicoPanamaPeruPhilippinesRomaniaRussiaSouth AfricaThailandTunisiaUruguayVenezuelaChil

4、eChinaCzech RepublicHungaryIcelandIsraelItalyKoreaLatviaLithuaniaMaltaPolandPortugalSlovakiaSloveniaTaiwanAustraliaAustriaBelgiumCanadaDenmarkFinlandFranceGermanyIrelandJapanLuxembourgNetherlandsNorwaySingaporeSpainSwedenSwitzerlandUnited KingdomUnited States-22Representation&Accountability-1 0 1-10

5、123Control of CorruptionLow ratedMedium ratedHigh ratedCountries with stronger institutions default less(better Fitch ratings).Risk free ratesBackEMs borrowing rates depend on risk free rate r.01234562020-Q12020-Q32021-Q12021-Q32022-Q12022-Q32023-Q12023-Q32024-Q12024-Q3Risk Free RateDate3 month T-Bi

6、llIMF Rate of ChargeLow Rates+Pandemic:the BuildupEra of Rising Rates:now what?Unusually low during Pandemic=cheap access to credit.Borrowing-20246810122018201920202021202220232024Borrowing/GDPYearAdvancedEM-AsiaEM-EuropeEM-LATAMPandemic+Low RatesPandemic+Low rates=large borrowing/GDP in EM.Debt pre

7、 and post PandemicCountry2022 Debt/GDPLong Run(90-22)Argentina8657Brazil7361Chile3216Colombia5737Costa Rica7050Ecuador5144El Salvador7558Greece172151India8775Korea5536Mexico5032Portugal119100Thailand5532Uruguay7168By 2022,when r,EMs were highly indebted=sustainable?Environment IQuantitative,infinite

8、 horizon model(Azzimonti-Mitra,2023).Het.Agents in n symmetric groups(regions,industries,castes,ethnic/religious/interestgroups)u(c,l,g)+iPublic good g,private consumption c,leisure 1 l.Piece of the piei:political favors,targeted public goods(bridges to nowhere),pork,exemptions,nepotism,bribes corru

9、ption.No access to international capital mkts ct=(1 t)wtlt.Environment IIFirms:competitive f(zt,lt)Government:can default,dt=1Rev(t)+(1 dt)hqtbt+1 bti gt+Xii,t|z the pie,International lenders:as in Arellano(2008).DetailsRisk-free rates Rt=rt,rt+1,.are time-varying and deterministic=affect bond price

10、s.Model-Big pictureBargaining protocolOptimization,1,2,123Interest Rate PathMost of the sovereign default literature assumes a constant risk free rate.We consider this trajectory instead:Risk-free rate(3-month T-Bill,CBO forecast).Long runPolicy FunctionsLong run r constant.Calibrated to Argentina.C

11、alibration DetailsMomentBenchmarkPlannerm nm=n?Debty?53%18.3%(TBy,y)-0.570.55(Favors/y)4.5%=0Spreads6.7%0Rate of Default3.6%0Periods in Default24%0Politics(m over-borrowing,pro-cyclical debt,frequent defaults,and high spreads.Medium term simulation(40 quarters)Assumptions:Start in 2022:Q1.High initi

12、al debt/GDP=64%.rtfollows CBO,converges to r.Caution:ztpersistent,one simulation over medium term can be misleading.Monte Carlo Simulation:draw 1,000 different paths for zt,keep last 40 quarters.Report per-period avgs for those 1000 plausible cases.Medium term debt management(a)Risk free interest ra

13、te(b)Frequency of Default(c)Debt to output ratio(d)SpreadsMedium term debt management,contPandemic+low-rate period built up debt/GDP to unsustainable levels.Facing rising rates60%of EMs default in 2022:Q1(bad and medium TFP shocks).40%of EMs implement steep austerity measures(really good TFP shocks)

14、.Spreads go down because of survivor bias.The party goes onPoliticians adjust by either:(i)defaulting or(ii)imposing stiff austerity measures(hightaxes).(a)The Pie(b)GDP Deviations from SSBut the pie doesnt shrink.non-defaulters cut down everything but corruption.PlotsPainful GDP losses.Internationa

15、l Financial Institutions(IFIs)IMF programsIMF lending5 year IMF program(20 quarters).GBC becomesRev(t)+(1 dt)hqtbt+1 bti+qI,tbI,t+1 bI,t gt+Xii,t.Cost of borrowing from IMF:DetailsA base lending rate,similar to the risk-free rate.Plus a surcharge that depends on the size of the loan.PlotFollowing Bo

16、z(2011),qI,t(bI,t+1)=11+rt+(bI,t+1)Note:No commitment in repaying qI,tb/c IMFs preferag-red cag-reditor status+major EMs almost always repaid(Boz,2011).Debt Management w/IFI program(no conditionality)Plots(a)IFI debt(b)The Pie(c)Private Debt(d)Frequency of DefaultMoral Hazard under No Conditionality

17、IFI debt cheaper if country is near-default(high spreads)=should prevent default.No conditionality:politicians borrow up to 6%of GDP and throw a party.Corruption Moral hazard:IFI debt available in default.Default in 90%scenarios.Lucky EMs dont implement austerity programs.ConditionalityIFI program w

18、ith conditionalitySeveral papers studied IMF lending with conditionality rules.debt ceilings,upper bounds on government spending,borrowing limits,etc.This paper:the root of the problem is corruption.Focus on the effect of governance reforms or anti-corruption efforts.Could be implemented via monitor

19、ing.If country wants to borrow from IMF,its politicians cannot eat the piePerfect monitoring:bI,t+1 0=0.Imperfect monitoring:bI,t+1 0=(1 ),with prob of being caught.IFI program with full monitoringTwo Shocks(a)IMF Debt(b)DefaultsVERY effective in reducing defaults:w/small IFI loans(higher borrowing

20、capacity.Seeding the seeds for recurrent users of the program!IMF program with imperfect monitoring(a)IFI debt(b)Frequency of Default(c)The Pie(d)Private DebtQuantitative Performance Criteria:Debt Ceiling RuleIFI Conditionality:bt+1yt max?0.95btyt,b y?ifbI,t+1 0,Austerity condition:debt/GDP by 5%unt

21、il reaching long-run debt/GDP.Politician cant increase debt if borrowing from IFI=indirectly limits corruption spending.Debt Ceiling Rule(a)IFI debt(b)Frequency of Default(c)The Pie(d)Private DebtWhich conditionality rule is better?Full monitoring!(a)International Investors:Recovery(b)EM Citizens:We

22、lfare GainsInternational investors indifferent between partial monitoring and debt ceiling.EM citizens prefer curbing corruption!ConclusionThe politics of debt:EMs w/weak institutions engage in corruption,over-borrow,defaulttoo often.The era of rising rates,doomed to default or face stiff austerity

23、programs,but nofundamental decrease in corruption spending.IFI lending programs provide short-term relief,but can exacerbate this problem by offeringan attractive outside option.Conditionality:full monitoring is best.If unavailable,International investors indifferent(default).Citizens prefer imperfe

24、ct monitoring.Debt ceilings impacts average citizen,monitoring theaverage political elites(mwc).Moving forwardCalibrate to other EMs.Consider other rules:public spending limits,deficit ceilings,cyclical bounds.Uncertainty about rtpath.Long run effects of IFI lending with conditionality rules.Long-te

25、rm debt.Proposers Problem in Round kContinuation ValuesBackProposers problemmaxtVPt(st,t)U?ct,lt,gt?+p,t+Est+1Jt+1(st+1,t+1|t)s.t.VIt(st,t)Jk+1t(st,k+1t)p,t=t(mt 1)t 0t,gt,t 0.In the mwcVIt(st,t)=U?ct,lt,gt?+t+Est+1Jt+1(st+1,t+1|t)Out of mwcVOt(s,)=U?ct,lt,gt?+Est+1Jt+1(st+1,t+1|t).States st=zt,t,Rt

26、Determination of Jt+1(st+1)Back=11nmt1nnmtnVPt+1(st+1)VIt+1(st+1)VOt+1(st+1)=0t+1Jt+1=1nVPt+1+mt 1nVIt+1+mt nnVOt+1.Functional FormsBackThe utility function specification is GHHU(c,l,g)=11?c l1+1+?1+1 g1.Default involves a productivity cost of the following formh(z)=(zif d=0z max0,0z+1z2if d=1with 1

27、 0.z and m follows an AR(1)processjt+1=(1 j)j+jt+jt+1where Ejt+1=0 and Ejt+12=2jwhere j=z,m.Model FitBackMomentData:ArgentinaBenchmark?Debty?53%53%Matched by Construction?gy?14%13%(r r)7%6.7%IMF Rates ConstructionBackIMF Rates Over TimeBackDebt Management w/IFI program(no conditionality)Back(a)IFI d

28、ebt(b)Price of Private and IFI Debt(c)Spreads(d)GDPDebt Management w/IFI program with conditionalityBack(a)IFI debt(b)Price of Private and IFI Debt(c)Spreads(d)GDPDefault RegionsBackSome factsBackBusiness Cycle MomentsMomentsDeveloped(n=30)Emerging(n=28)(PS,y)0.30.0(NX,y)0.0-0.1(y)5.07.3(c)/(y)1.11.

29、2(g)/(y)0.91.3Fitch Ratings3.82.9Fiscal policy responses to shocks=.Emerging economies:dont smooth.Pro-cyclical debt=amplifies the cycle!Example:Union Leader with Political Influence?backPolicy Functions:institutionsBackb*(mH)bbb*(mL)defaultdefaulttaxesbdefaultdefaultgbfavorsbdefaultdefaultdefaultde

30、fault?=0?0?Policy Functions:shocksPricesBackPolicy Function-Bond PricesBackBorrowing Policy Function and TFPBackThe Pie and TFPBackWhy does rtmatter?BackRisk-free interest rates Rt=rt,rt+1,.are time-varying and deterministic(assume perfectforecast).International lenders:as in Arellano(2008).qt=Zzt+1

31、t+1?1 dt+11+rt+1?zt+1|zt,Sequence Rtaffects bond prices.=This matters for government policy,in particular debt management and default decisions.Proposers ProblemLots of DetailsbackDiverts tfrom budget,offers i,t=to mwc.Keeps p,t=t(mt 1)t.Can write the problem as max welfare of average mwc member(lin

32、earity)maxt,gt,dt,bt+1U(t,gt)+tmt+Est+1J(st+1)s.t.t=Rev(t)gt+(1 dt)hqtbt+1 bti 0.The politicsbackEach group has a leader with seat in bargaining table(governor,legislator,lobbyist,union leader,religious leader,oligarch,etc.).One leader chosen at random to make a policy proposalt=t,gt,bt+1,dt,1,t,2,t

33、,.,n,t|zcorruptionProposals need support of mt n leaders to be implemented(mwc).mtis stochastic(political shocks).Bargaining process opens door to corruption/wasteful spending.Solve symmetric Markov Perfect Equilibrium(MPE).Calibration Targets:ArgentinabackFunctional FormsModel FitPolicy FunctionsPa

34、rameterValueTargetDescription2CRRA2Frisch Elasticity0.993211+rFOC r(annualized)2.8%3 mont T-Bill0.03856.5 Years of Exclusionz0.925Persistence Real GDP?AR(1)z0.018Volatility of Real GDPm0.954Persistence of R&A?AR(1)m0.234Volatility of R&A0-0.36E(Spreads)=7%Jointly Calibrated10.40DebtGDP=53%1.2SpendY=0.14 m5(Quarterly model)IMF program with full monitoring:two TFP sequencesback(a)Non-defaulter(b)Defaulter