Debt Management, Country Categorizations and Government-Financial Market Relations: Presentation for UNC Center for European Studies Fall Lecture Series 2012, Beyond the Euro Crisis

1. Debt Management, Country
Categorizations and
Government-Financial
Market Relations
Layna Mosley, UNC
October 26, 2012

2. Sovereign Debt in the EU
• Early 2000s: interest rate convergence among euro-zone
nations.
• Virtuous circle for sovereign borrowers.
• Default risk assumed to be non-existent, even among
highly-indebted EU nations.
• “No bailout” clause of Maastricht Treaty lacked
credibility.
• Little market response to violations of the Stability
and Growth Pact.
• Governments were able to suggest rules for investors.
• E.g. 3 percent fiscal deficit rule.

3. 25
Interest Rates on Benchmark Government Bonds
Australia
Austria
Belgium
Canada
20
Czech Republic
Denmark
Finland
France
Germany
15
Greece
Hungary
Iceland
Ireland
10 Italy
Japan
Korea
Netherlands
New Zealand
5 Norway
Portugal
Slovak Republic
Spain
Sweden
0
United Kingdom
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011

4. Government Bond Rates, 2005-2011
18
16
14 Austria
Belgium
12 France
Germany
10 Greece
Ireland
8 Italy
Netherlands
6 Portugal
Spain
4
United Kingdom
United States
2
0
2005 2006 2007 2008 2009 2010 2011

5. • With the crisis, a
renewed attention to 4.5
default risk, and to 4
differences across 3.5
countries. 3
2.5
• Reward for EMU 2
disappears, or at 1.5
least is reduced.
1
0.5
• Chart: variance in 0
government bond
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
rates among OECD
nations.

6. What should we learn from the
EU’s debt crisis?
(1) The return of default risk among developed
nations?
(2) All government debt is not created equal, or
managed equally.
• Affects sensitivity of governments to financial market
pressures.

7. Debt Management
• When debt matures
• Tradeoff between cost and rollover risk
• 2010: Ireland 5.9 years vs. UK 14.1 years

8. United States
United Kingdom
Spain
Average Time to Maturity of Government Debt, 2010
New Zealand
Netherlands
Italy
Ireland
Hungary
Germany
France
Finland
Estonia
Denmark
Canada
Austria
Australia
16
14
12
10
8
6
4
2
0

9. Average Time to Maturity of Marketable Government Debt, OECD
Countries
8
7
6
5
4 25th pctile
50th pctile
3 75th pctile
2
1
0
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010

10. OECD governments will need to
refinance about 30% of their
long term debt in the next 3
years.
Source: OECD Sovereign
Borrowing Outlook, No.
4, 2012

11. Source: OECD Sovereign Borrowing Outlook, 2012

12. Debt Management
• When debt matures
• Tradeoff between cost and rollover risk
• 2010: Ireland 5.9 years vs. UK 14.1 years
• Who holds debt
• Private vs. official creditors (e.g. central banks)
• Resident vs. non-resident investors
• 2010: 94% non-resident Greece vs. 56% Italy

13. United States
United Kingdom
Percentage of Marketable Debt Held by Non-Residents, 2009
Turkey
Sweden
Spain
Slovenia
Slovak Republic
New Zealand
Mexico
Italy
Hungary
Finland
Estonia
Denmark
Czech Republic
Canada
Austria
Country
90
80
70
60
50
40
30
20
10
0

14. Non-Resident Investment as % of Total Marketable
Debt, OECD Countries
60
50
40
30 25th pctile
50th pctile
75th pctile
20
10
0
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010

15. Debt Management
• When debt matures
• Tradeoff between cost and rollover risk
• 2010: Ireland 5.9 years vs. UK 14.1 years
• Who holds debt
• Private vs. official creditors (e.g. central banks)
• Resident vs. non-resident investors
• 2010: 94% non-resident Greece vs. 56% Italy
• In which currencies debt is denominated
• “Original sin” (Eichengreen and Hausman)
• Risk to investors vs. risk to governments
• Domestic vs. foreign currency debt

16. Debt Management Outcomes
• To what extent do these outcomes reflect strategic choices by
governments, versus willingness of investors to lend?
• Debt management outcomes as dependent variable
• What are the implications of debt management outcomes for
responses to crises and, more broadly, for government
policymaking autonomy?
• Debt management outcomes as independent variable.

17. Governing Debt
• Establishment of separate DMOs in many OECD nations in
the 1980s and early 1990s.
• Variation in location (MoF, CB)
• Variation in autonomy and mandates
• In emerging markets, greater attention in 2000s.
• Concerns about short term debt and foreign currency
denominated debt.
• Efforts to move from floating to fixed rate debt, and from
foreign- to domestic-issued debt.
• Facilitated by global market liquidity in mid-2000s.
• Attempts by IGOs to diffuse “best practices” for DMOs
• Main tradeoff: costs vs. risks
• Shorter maturity & foreign currency
• Longer maturity & domestic currency

18. Dependent Variable:
Time to Maturity of Public Debt
• Cross-sectional time series analyses
• Independent variables: moving average over last three years
(alternative: five years).
• Lagged dependent variable
• 1980-2009, OECD nations
• n=~300 (13 to 19 countries)
• Significant predictors of average time to maturity:
• Inflation (-)
• Government budget balance (-), or government debt (-)
• Effective number of political parties (-)
• Central bank independence (+; less robust)
• Not significantly associated with time to maturity:
• Left government
• Coalition (vs. single party) government
• EMU participation
• Location of debt management office (separate or not)

19. What should we learn from the
EU’s debt crisis?
(1) The return of default risk among developed nations?
(2) All government debt is not created equal, or managed
equally.
(3) Investors’ assessments of sovereign borrowers vary over
time and across countries – and not only because of changes
in country-specific economic or political fundamentals.

20. Assessing Sovereign Risk?
• When setting risk premiums for sovereign debt, at what do
investors look?
• If governments are attentive to market pressures, how do these
pressures operate?
• Investors’ primary concerns are currency, inflation and default
risk.
• 1990s and 2000s: investors treated developed and emerging
market countries differently.
• Assumed no default risk among developed nations, so consider
only macro-indicators (deficits, inflation)
• Worried more about default among developing nations, so looked
also at supply side policies, government ideology, elections.

21. Assessing Sovereign Risk
• Developing nations are therefore more constrained by
market pressures than developed ones.
• Also have greater need to attract foreign capital.
• And are more exposed to externally-induced volatility:
• Importance of push vs. pull factors to risk premiums.
• Commodity exporters
• The borrowing strategies pursued by emerging market
governments can exacerbate “ability to pay” concerns.
• After accounting for policy outcomes, emerging and
frontier market countries pay higher risk premiums than
developed nation borrowers.

22. Variation in Market Constraints
• The prices governments pay to borrow on international
markets vary markedly:
• Across countries (sovereign credit ratings, macroeconomic
fundamentals)
• Over time (liquidity, risk appetite, elections)
(Archer et al 2007; Bernhard and Leblang 2006; Cantor and Packer 1996; Hardie
2006; Jensen and Schmith 2005; Mosley 2003; Tomz 2007)
• In addition, different types of governments are differently
constrained by global capital markets:
 developed vs. developing
 commodity vs. manufacturing exporters
 borrowers from commercial banks vs. bond markets
(Campello 2012, Mosley 2003, Kaplan 2012, Wibbels 2006)

23. Are market constraints
interdependent?
• Does the sovereign risk premium paid by one country
systematically shape the way that investors assess sovereign
risk in other nations?
• If so, how, and through which channels do sovereign risk
assessments diffuse?
• One principal mechanism: country classifications
• Professional investors sort countries into “peer” groups

24. Are market constraints
interdependent?
• Optimal portfolio diversification dictates that professional
investors often manage highly dissimilar assets
• sovereign debt, corporate debt, equities, derivatives and cash
• and they often invest in a diverse range of locations
• They rely on information shortcuts to assess risk (heuristics)
• Summary indicators of fiscal and monetary outcomes.
• Budget deficit
• Debt ratios
• Inflation
• Also: categories
• “developed” vs. “developing” (Mosley 2003)
• peripheral Europe: “emerging Europe” in the 1990s  “eurozone” in
the 2000s  “PIIGS” in the 2010s

25. Sovereign Peer Groupings
• Investors may over- or under-estimate sovereign risk based on
the category into which the country is grouped:
• When investors are more optimistic about a given group of
countries, each country in that group will experience an
improvement in market access.
• When investors are more pessimistic about a category of
countries, a borrower within that category may suffer – even if
the country’s fundamentals do not warrant such pessimism.
• This implies that investors’ responses to domestic policy are
neither fixed, nor fully objective.

26. Interdependent Sovereign Risk
• Over the long term, sovereign risk assessments should vary
with domestic fundamentals.
• In the short term, sovereign risk may correlate across nations
due to contagion from crises or changes in global liquidity.
• But, even after controlling for these short- and long-term
effects, we expect an additional effect of “country category”
• Country risk premium will be significantly correlated with the risk
premiums paid by other borrowers in the same category.
(Hypothesis 1)

27. Comparing Countries
• We also expect that, when investors evaluate individual
sovereign borrowers, they do so in relative terms:
• The risk premium associated with a given fiscal deficit, for
instance, depends on what other countries are doing.
• Governments are evaluated relative to what they are expected to
do.
• Tomz 2007: stalwarts vs. lemons
• Here, peer categorizations matter because they define the
relevant comparison group.
• Sovereign risk premiums are associated, all else equal, with what
other borrowers in the same category are doing, in terms of
government budgets, and deficits. (Hypothesis 2)

28. Which Peer Groups Matter?
• We examine 3 types of investment categorizations:
1. Region:
• Asia, Western Europe, post-Communist Europe, Latin
America, Non-Latin Caribbean, Middle East and North
Africa, North America, South Asia, and Africa [World Bank
categories]
2. Market and Economic Development:
a. MSCI: Emerging Markets; Frontier; Developed
b. FTSE: Emerging Markets; Frontier; Developed; Advanced
Emerging; Secondary Emerging
3. Risk Rating:
• Fitch Long-Term Sovereign Credit Ratings
• (Coded as a 1-12 sovereign risk score)

29. Data and Method
• Dependent variables:
• Sovereign spreads (EMBI): monthly and annual data for 26
emerging market economies, 2001-2010.
• Credit default swap (CDS) prices, monthly data for 26 developed
and developing countries, 2000-2010.
• Independent variables:
• Domestic economic: government debt, government
consumption, budget balance, inflation, capital account openness.
• Domestic political: democracy, government ideology, opposition
party ideology, electoral cycle, presidential/parliamentary
• Global: US interest rate, US treasury bond yields, US stock market
returns.
• Peer group: average risk premium of those in the same category

30. Data and Method
• We estimate cross-sectional time series models, using an Error
Correction Model (ECM).
• This allows us to consider both the short-term and long-term
effects of the regressors.
• Generalized least squares estimator, country fixed effects, linear
time trend.
• Full results of the estimations are available in the paper
(Tables 1 through 5)

31. Main Findings
Blank cells indicate that the effect of peer category risk premium is not
statistically significant, controlling for all other independent variables.

32. Main Findings
• In terms of relative assessments (Hypothesis 2):
• The level of government debt in (regional) peer countries is
negatively and significantly associated with spreads.
• A country’s own debt level is, as we would expect, positively
and significantly associated with spreads.
• Categorization may provide governments with an opportunity
to outperform their peers, for which they are rewarded.
• If we take sovereign ratings as the dependent variable,
• When one’s rating category peers have better fiscal
outcomes, one’s rating is higher, all else equal.
• Again, a country’s own fiscal balance also is positively and
significantly associated with one’s sovereign rating.

33. Caveats and Conclusion
• Annual data obscures what are often shorter-term movements in
sovereign debt markets.
• Models suggest that almost all of the disturbances to equilibrium
in risk premiums are corrected within one year.
• Most of the political variables included in our
models, however, are measured on an annual basis.
• Sample: a limited set of countries (e.g. those for which there is a
CDS market or which are included in EMBI+)
• More generally, aggregate-level observational data may not be the
best way to assess our proposed causal mechanisms.
• Future work: survey experiment design.
• Our findings suggest, however, that the ways in which countries are
grouped – and changes in those groupings – may be important
determinants of governments’ capacity to access bond markets.

38. DV: Annual Δ_Spread Table 1. Explaining Annual Changes in Sovereign Debt Spreads
Peer Category: Region Risk Rating MSCI FTSE
1 2 3 4
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Spread t-1 -1.08 *** 0.07 -1.12 *** 0.06 -0.94 *** 0.07 -1.11 *** 0.07
Peer Diffusion
Peer Spread
t-1 0.21 ** 0.10 -0.20 * 0.11 0.22 * 0.13 -0.26 *** 0.09
Δ 0.37 *** 0.07 0.09 0.07 0.45 *** 0.10 -0.13 0.09
Domestic Politics and Economy
Gov Consumption
t-1 1.02 24.23 4.59 27.13 3.72 24.43 19.76 25.89
Δ -23.86 19.78 -4.07 21.34 -15.16 20.08 -36.84 * 22.33
Debt
t-1 4.37 ** 2.18 4.15 ** 1.88 2.77 2.24 5.51 *** 2.10
Δ 6.40 *** 2.26 5.78 *** 1.76 7.04 *** 2.13 7.88 *** 2.15
Maturity
t-1 1.15 5.10 -3.85 5.46 -7.16 5.38 -2.40 4.82
Δ -1.79 3.56 -6.35 * 3.62 -8.14 ** 3.66 -2.77 3.51
Inflation
t-1 0.33 0.46 0.07 0.48 0.68 0.46 0.63 0.52
Δ 0.02 0.33 -0.11 0.34 0.10 0.34 0.02 0.39
Budget Balance
t-1 -20.89 ** 10.57 -26.37 ** 10.98 -16.42 11.20 -20.98 * 11.21
Δ -31.89 *** 8.30 -25.73 *** 8.47 -25.46 *** 8.64 -23.93 *** 8.79
Democracy
t-1 11.86 14.17 20.07 14.42 12.47 13.52 15.10 12.90
Δ 22.40 22.37 25.12 21.99 5.68 23.53 32.87 22.03
KA Open
t-1 -80.68 ** 33.55 -65.33 * 35.37 -54.40 33.85 -59.09 * 33.26
Δ -59.39 * 36.41 -56.40 40.39 -58.81 40.33 -31.58 35.59
Years to Election
t-1 -32.76 ** 14.63 -29.64 * 15.49 -25.99 * 14.44 -25.52 ** 13.04
Δ -9.47 9.92 -12.17 9.98 -10.52 9.90 -7.66 8.88
Left
t-1 128.18 267.97 26.77 385.25 -30.57 440.41 28.30 337.53
Right
t-1 36.95 400.40 -673.79 * 399.72 -642.29 * 377.14 -650.50 * 346.34
Opposition Right
t-1 85.90 61.58 40.49 61.68 56.35 57.26 50.99 45.05
Opposition Left
t-1 116.75 * 70.25 104.07 68.31 93.53 67.58 54.50 55.99
System
t-1 -52.33 39.09 -7.75 39.86 1.05 39.45 -29.78 39.05
Common Shocks
US Prime Rate
t-1 29.75 ** 12.28 30.99 *** 12.44 34.93 *** 13.21 33.09 *** 13.00
Δ -25.89 * 14.80 -68.22 *** 14.88 -10.85 17.29 -70.54 *** 15.19
Time
t-1 -12.85 8.95 -22.22 *** 8.47 -9.43 9.64 -19.05 ** 8.42
Const. 25956 17988 44910 *** 16975 19192 19351 38372 ** 16896
N. obs 171 171 171 171
Wald chi2 454.59 1353.14 479.67 444.68
Prob > chi2 0 0 0 0
FGLS error correction model of annual change in Sovereign Stripped Spreads *** p<.01; ** p<.05; * p<.1