Topics on external debt

Transcription

1 Topics on external debt Econ PhD, EUI Lectures 4 and 5: Sovereign default Hernán D. Seoane UC3M Fall 2017

2 Today s lecture Endogenous spread models Strategic default Non-strategic default: a few variants

3 Sovereign debt: some questions Why do countries default? (or better, why do countries repay?) Economic sanctions: if default, seize part of your income Reputation: exclusion from international financial markets (generates costs in terms of consumption smoothing) Financial Costs sector (less explored, getting famous now) Default on domestic bondholders How often countries default? Stylized facts? Sovereign debt levels are sustainable with our models? Define default: credit rating agencies (Standard & Poor s definition)

4 Sovereign debt: some questions Usually default is defined as a situation in which a country fails to pay capital or interests of the outstanding debt. It can be when stop paying or when forces swap in less favorable terms than originally A default is over when the country resumes payments (judgment involved here as a country usually does not get to an agreement with all the lenders at the same time)

5 How long are countries in default? Figure: Source Uribe Schmitt-Grohe (2015) in period the average default probability per year is 0.04 In this period years in state of default per episode is 8

6 Haircuts Defaults are not often solved by resuming payments only There is a haircut involved: i.e. a renegotiation process starts where the lender tries to recover as much as possible and the borrower tries to exchange existing assets by new assets with new conditions Sturzenegger and Zettelmeyer (2008) estimate the losses in sovereign debt restructurings Computing haircuts is not easy, you have to price the present value of outstanding debt in default and compare it with the present value of the new debt. For this you need the discount rate SZ(2008) uses the yield prevailing immediately after the exchange of debt instruments

7 Haircuts investor losses are the % difference between market value of new instruments and the net present value of the remaining contractual payments of the old instrument (inclusive of principal or interest arrears) H NPV = 1 NPV(new, r new) NPV(old, r new ) The key issue here is the discount rate Another option is to use the mark-to-market which would imply a much smaller haircut due to a higher rate for the old debt Haircut Face Value: percentage difference between the market value of new debt and the sum of outstanding face value of the old debt

8 Haircuts conditions Figure: Sturzenegger and Zettelemeyer (2008)

9 Figure: Sturzenegger and Zettelemeyer (2008) Haircuts

10 Haircuts After a default, usually countries renegotiate a haircut on the level of debt, Sturzenegger and Zettelmeyer (2008) The authors measure haircuts to be about 40%, but this has a high dispersion (Uruguay got a 13% haircut while Argentina experienced a 73%) Not clear differences between haircuts to domestic/foreign lenders

11 Haircuts: what then? Cruces and Trebesch (2013) Review the haircut finding, with a larger dataset using bank and bondholder data from (180 defaults 68 countries) Find similar results Higher haircuts are associated with subsequent bond yield spreads and longer periods of capital market exclusion.

12 Haircuts size and volume of debt restructured Figure: Cruces and Trebesch (2013)

13 Haircuts & post restructuring spreads Figure: Cruces and Trebesch (2013)

14 Sovereign debt: when do coutnries default? Too much debt? Expansion or recession? Spreads?

15 Sovereign debt: when do countries default? Figure: Source Uribe Schmitt-Grohe (2015)

16 Sovereign debt: some questions Spreads Akitoby and Stratman (2008) estimates the semielasticity of country spread to debt log(country Spread) D/Y = 1 Spreads tend to rise more for increases in government spending financed by debt than those financed by taxes

17 Sovereign debt: some questions Spreads have a 1 to 1 mapping with default probability... only when investors are risk neutral and all assets have the same liquidity properties These are the assumptions of baseline models but they are off, default probability in our models is over-estimated Additionally, sample mismatch

18 Figure: Source Uribe Schmitt-Grohe (2015) Sample mismatch

19 Correcting sample mismatch Figure: Source Uribe Schmitt-Grohe (2015)

20 Do countries default in bad times? why is this important? Different models produce opposite predictions Suppose you assume a country and an investor sign a contingent contract: the country receives a transfer from ROW if domestic output is lower than average and make payments if it is higher than average Under this contract you will be tempted to violate it in good times Suppose now there are no state-contingent contracts... here the incentives to default is when output is low

21 Do countries default in bad times? Figure: Source Uribe Schmitt-Grohe (2015)

22 Do countries default in bad times? Tomz and Wright (2007, 2013): data from countries default during bad times but the evidence is weak Output is only 1.5% below trend In 40% of the defaults output is even higher than the trend How many countries were contracting? 75% occurs when output growth is below trend

23 Sovereign debt: some questions Figure: Source Uribe Schmitt-Grohe (2015) Consumption and investment fall Sudden stops: trade balance reversals Real exchange rate depreciates

24 Costs of Sovereign debt: financial exclusion Gelos et al (2011), Richmond and Dias (2009), Cruces and Trebesch (2013) Around 16 years to have full access to international markets (defined as asset flows of a magnitude larger than 1% of GDP) Around 5 year for partial access (positive flows) Here all the measures are fairly subjective Authors look at bonds and bank loans. However full exclusion was rarely the case

25 Costs of Sovereign debt: financial exclusion Gelos, Sahay and Sandleris (2011) Micro data on sovereign bond issuance and public syndicated bank loans They measure resumption of access to credit as the first year after (entering) default in which gov borrows from these sources and the stock of debt increases Mean exclusion is 4.7 years Richmond and Dias (2009) count years of exclusion starting from when the countries exit default: they have partial and full re-access. Use aggregate data: 5.7 and 8.4 years Cruces and Trebesch (2013) exclusion as a penalty

26 Costs of Sovereign debt: financial exclusion Figure: Source Uribe Schmitt-Grohe (2015)

27 Costs of Sovereign debt: output costs Default tends to be one feature of a macro crisis, hence it is difficult to isolate the costs of default Endogeneity is a second issue (growth, trade are endogenous) Chuan and Sturzenegger (2005), Boresztein and Panizza (2009), De Paoli, Hoggarth and Saporta (2006), Levy Yeyati and Panizza (2011) estimate growth regressions with default dummies They also identify permanent output losses Very simple regressions, more needs to be done From a growth accounting exercise Zarazaga (2012) computes a 13% output loss on average for Argentina for the period

28 Costs of Sovereign debt: output costs Boresztein and Panizza (2009) GROWTH i,t = α + βx i,t + γdefault i,t + ɛ i,t X are controls: On average default is associated to a decrease in growth of 1.2 percentage points per year Allowing for temporal structure of the default: one year lagged default only matters, higher lags are non-significant... Short lived effects (joint effect is 2.6 percentage points) Additional exercises to study the direction of causality, decompose default on factors and residual default and then regress output growh on the residual. Still find negative coefficient Alternative interpretations of the residual

29 Costs of Sovereign debt: more on exclusion Boresztein and Panizza (2009) RATING i = α + βx i + γdefault i + ɛ i Besides exclusions, do lenders penalize defaulters when they return to asset markets in terms of cost of credit? Is this effect long lasting? RATING measures average credit rating , DEFAULT measures previous default history Default history is negatively correlated with credit ratings Another exercise: direct impact of default on borrowing costs, default in t 1 has a large and significant effect on spreads (400 bp), t 2 is still sizeable, but earlier is not

30 Costs of Sovereign debt: trade sanctions Boresztein and Panizza (2009) Rose (2005): do defaults have negative effects on trade? Use info on renegotiations with Paris Club in a gravity trade model. Find 15 years negative trade effects, 8% per year Borensztein and Panizza (2006), we use industry-level data and find that sovereign defaults are particularly costly for export-oriented industries: short lived effects Which is the channel? Little historical record of countries imposing quotas or embargos on a country that falls in default A more likely scenario: deterioration in the credit quality of exporting firms after the default NTC is net trade credit scaled by international trade in country i in year t NTC i,t = α + µ i + βx i,t + γdefault i,t + ɛ i,t

31 Costs of Sovereign debt: trade sanctions Boresztein and Panizza (2009) NTC i,t = α + µ i + βx i,t + γdefault i,t + ɛ i,t Arellano and Bond (1991) GMM difference estimator allows to consistently estimate a fixed effect model that includes the lagged dependent variable Also allows to deal with endogeneity by instrumenting the explanatory variables with their lagged values Without lagged dependent variable, find that the coefficient of the default dummy is negative and statistically significant 0.13 Adding the lagged dependent variable, find that the effect is negative and large only in the first and second year of the default This result suggests that default does have a negative effect on trade credit but that this effect is short lived.

32 Costs of Sovereign debt: domestic banking system Boresztein and Panizza (2009) Key, I haven t seen much of this lately A default hurt domestic bondholders, specially banks may hold large amounts of sovereign debt Test if sovereign defaults lead to banking crises (deterioration of balance sheets) or a domestic credit crunch (confidence crisis), uncertainty build a banking crisis index using Glick and Hutchinson (2001), Caprio and Kingelbiel (2003), and DellâAriccia et al (2005) The probability of a banking crisis conditional on default is 14%, an 11 percentage point increase with respect to the unconditional probability Test if default generates credit crunch: use industry level data, test if defaults have a larger negative impact on sectors that require more external finance Defaults do not seem to have special effects on these sectors

33 Costs of Sovereign debt: growth accounting Zarazaga (2012) A key problem with all regressions is that output growth and defaults are both endogenous variables Zarazaga (2012) proposes an accounting exercise Argentine defaults 1982 and 2001 imply a peak in capital to output ratio in the run-up of the default, followed by a significant decline after the default k/y = 1.9 before default (in line with other emerging economies). After default it falls until 1.35 in what is the output per capita loss associated to this loss of capital?

34 Costs of Sovereign debt: growth accounting Zarazaga (2012) Assume y t = k 0.4 t h 0.6 t, this technology implies y t h t = ( ) 2/3 kt If capital to output ratio would have not fallen from 1.9 to 1.35, output per worker would be 26% larger in 2007 that what it was ((1.9/1.35) 2/3 1) 100) y t