## _wp13164

## Source details

**Canonical URL:** [_wp13164](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13164.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13164.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13164.pdf.json)

---

### Introduction and context
- After an extended period of heightened volatility, emerging market sovereign spreads narrowed steadily in the second half of 2012.
- Key drivers of improved market sentiment:
  - European Central Bank announcement of the Outright Monetary Transactions (OMT) program in September (conditional purchase of Euro Area sovereign bonds in unlimited amounts at the secondary market).
  - The Federal Reserve started a new bond purchase program and committed to keep interest rates at exceptionally low levels at least until the mid-2015.
  - The Bank of Japan announced further monetary easing.
  - Favorable political/legal developments: outcome of Greek elections in mid-2012 and favorable ruling of the German constitutional court regarding the European Stability Mechanism in September.
- Market flows and spreads:
  - Emerging market bond funds experienced an inflow of almost US$ 1 billion per week on average during the second half of 2012–beginning of 2013.
  - The EMBI Global spread decreased by around 180 basis points between early June 2012 and late January 2013.
  - A subsequent small correction occurred due to increasing risk aversion related to uncertainty in Cyprus.

### Main analytic objectives and approach
- Three angles:
  - Disentangle effects of global and country-specific developments on emerging market debt spreads.
    - Finding: both matter for long-run spreads; global factors dominate short-run dynamics.
  - Test whether strength of fundamentals affects sensitivity of spreads to global factors.
    - Methods: (i) fixed effects panel with sample split into weak vs. strong fundamentals; (ii) pooled mean group estimation relating country-specific short-term coefficients to fundamentals.
    - Finding: countries with stronger fundamentals tend to have lower sensitivity to changes in global risk aversion.
  - Decompose changes in spreads across seven sub-periods over the last decade attributing movements to fundamentals, global factors, and unexplained residuals.
    - Novelty: decomposition splits residual changes into correction of initial misalignment and increases/decreases in misalignment during each period.

### Data, variables, and sample
- Sample and measure:
  - An unbalanced panel dataset of 147 monthly observations between January 2001 and March 2013 for 18 emerging markets in three regions.
  - Sovereign risk/spread: monthly averages of daily Emerging Market Bond Index Global (EMBIG) sovereign spreads from J.P. Morgan’s MorganMarkets.
  - EMBIG spread: market-capitalization-weighted average of spreads on US$-denominated Brady bonds, Eurobonds and traded loans issued by sovereign and quasi-sovereign entities.
- Country-specific fundamentals:
  - ICRG database used; 22 variables aggregated into three composite ratings:
    - Economic Risk Rating (ERR): range 0–50; weighted average of GDP per capita, real GDP growth, inflation, fiscal balance (percent of GDP) and current account balance (percent of GDP).
    - Financial Risk Rating (FRR): range 0–50; weighted average related to foreign debt (percent of GDP), foreign debt service (percent of exports), current account (percent of exports), official reserves (months of imports) and exchange rate stability (12-month appreciation/depreciation against the US$).
    - Political Risk Rating (PRR): range 0–100; weighted average of government stability, socioeconomic conditions, investment profile, internal conflict, external conflict, corruption, military in politics, religious tensions, law and order, ethnic tensions, democratic accountability, and bureaucracy quality.
  - Sample countries:
    - Asia: China, Indonesia, Malaysia, Pakistan, Philippines
    - Central and Eastern Europe: Bulgaria, Hungary, Poland, Russia, Turkey, Ukraine
    - Latin America: Argentina, Brazil, Chile, Colombia, Mexico, Peru, Venezuela
- Global factors:
  - VIX (Chicago Board Options Exchange Volatility Index) measures implied volatility of S&P index options; expected positive association with spreads.
  - U.S. Federal funds rate used as proxy for global liquidity conditions; lower Fed funds rate associated with higher liquidity and expected positive relationship with spreads.
  - VIX found stationary; U.S. Federal funds rate appears to follow a unit root.
- Stationarity and cointegration:
  - Im-Pesaran-Shin and Fisher-type ADF tests: mixed for EMBIG spread; reject unit roots for country-specific explanatory variables.
  - Westerlund (2008) cointegration test: no cointegrating relationship found among these variables.
  - Phillips and Moon (2000): pooled regression of two nonstationary but non-cointegrated variables yields consistent long-run average regression coefficients as N and T become large; proceed with fixed effects estimation.

### Empirical specifications and estimation techniques
- Fixed effects panel regression:
  - Dependent variable: ln(EMBIG spread).
  - Country-specific fundamentals vector k(1): ERR, FRR, PRR.
  - Global explanatory variables k(2): VIX and U.S. Fed funds rate.
  - Country fixed effects and Driscoll-Kraay robust standard errors used.
- Pooled Mean Group (PMG) estimator (Pesaran, Shin and Smith (1999)):
  - ARDL (p,q,...,q) reparameterized to allow short-run coefficients to differ across countries while constraining long-run coefficients to be homogeneous.
  - By setting p=1 and q=1, model estimated to obtain short- and long-run separation.
- Robustness checks:
  - Endogeneity: regressions with up to five lags of explanatory variables; signs, magnitudes and significance broadly unchanged except for U.S. Federal funds rate.
  - Multicollinearity: ERR and FRR strongly correlated; excluding FRR increases sensitivity of spreads to PRR and VIX and raises ERR coefficient four-fold (with weakened significance) and lowers model explanatory power; excluding ERR has limited effect.
  - Region and group splits to explore heterogeneity.

### Fixed effects estimation — key quantitative findings
- Sensitivity of spreads to country fundamentals (whole-sample):
  - FRR: a 1 percent increase (improvement) in FRR lowers spreads by 2.8–3.5 percent.
  - PRR: a 1 percent increase lowers spreads by 2.3–2.5 percent.
  - ERR: a 1 percent increase lowers spreads by 0.5–0.6 percent.
  - Interpretation: spreads most sensitive to FRR and PRR; investors pay more attention to political risk and liquidity (PRR and FRR) than to solvency indicators (ERR) in the short term.
- Global factors (fixed effects):
  - VIX: a 1 percent increase associated with a 0.7 percent increase in spreads (positive and strongly significant).
  - U.S. Federal funds rate: coefficient negative but becomes insignificant when VIX is included; significant with negative coefficient in Asia in regional splits.
  - 3-month and 10-year U.S. Treasury yields tested but not significant.
- Model fit and explanatory power:
  - Overall R-squared around 0.5–0.6 across three specifications.
- Heterogeneity:
  - Countries with higher economic and financial risk (low ERR and low FRR) have higher sensitivity to VIX.
  - VIX coefficient similar magnitude for low and high PRR groups.
  - Regional patterns:
    - ERR not significant in Europe, significant in other regions.
    - FRR more important in Asia and Europe; FRR and PRR roughly equal in Latin America.
    - VIX exposure highest for European countries.
    - U.S. Federal funds rate significant with negative coefficient in Asia.

### Pooled Mean Group (PMG) estimation — key quantitative findings
- Long run:
  - FRR and PRR: long-run coefficients negative and significant (stronger fundamentals associated with lower equilibrium risk premia).
  - ERR: coefficient changes sign compared to fixed effects suggesting stronger economic fundamentals associated with higher spreads in the long-term; short-term ERR coefficient negative but not significant.
  - VIX: significant and positive long-run effect on spreads.
  - U.S. Federal funds rate: not significant driver in long-run in specifications (1)-(3).
  - Error correction coefficient: significant but low (spread deviations from long-run equilibrium corrected at a slow rate).
- Short run:
  - Global factors are principal drivers of short-run dynamics.
  - VIX: positive short- and long-run effect.
  - U.S. Federal funds rate: significantly negative short-run effect (U.S. monetary tightening lowers spreads in short-run) but ambiguous and insignificant in long-run when country fundamentals included.
  - Country-specific fundamentals: ERR not significant in any PMG specification; FRR and PRR significant in only one specification each.
- Heterogeneity and sensitivity:
  - Country-specific short-term coefficients plotted against country-average risk ratings: countries with higher average risk ratings have lower sensitivity to VIX.
  - PMG split by region:
    - Long-run VIX significant in each region; highest coefficient in Europe.
    - U.S. Federal funds rate significantly positive in Asia in long run (U.S. monetary loosening associated with declining Asian spreads).
    - FRR significant only in Europe; PRR negative in Asia and Latin America, weakly positive in Europe.
    - Short run: VIX (positive) and U.S. Federal funds rate (negative) significant in each regional specification.
    - Error correction term indicates misalignment corrected fastest in Asia, then Europe, then Latin America.

### Decomposition of spread changes and misalignment (seven sub-periods)
- Misalignment defined as residual = actual spread − fitted spread.
- Visual patterns and sub-period narratives:
  - Pre-crisis (up to Aug 2007): models explain about half of tightening; both global and country-specific factors important; unexplained tightening driven by correction of earlier underpricing and accumulation of misalignment during boom years.
  - July 2007–late summer 2008: spreads increased somewhat; models explain 30–50% of increase via deteriorating global and domestic fundamentals; unexplained part due to elimination of initial overvaluation and emergence of undervaluation.
  - Post-Lehman (late 2008): sharp increases; models explain 76 and 93 percent of increase (FE and PMG), driven mostly by sharp deterioration in global factors and worsening country fundamentals.
  - Early 2009–April 2010: substantial tightening; models explain 70 and over 90 percent of tightening (FE and PMG), driven mostly by improved global market sentiment and some improvement in fundamentals; unexplained part mostly reduction in undervaluation.
  - April 2010–June 2012: substantial increase in spreads (Eurozone debt crisis); models explain around 40 percent of increases, with misalignment becoming more important from mid-2011 onward.
  - June 2012–end of sample (March 2013): substantial narrowing of spreads; FE explains around three quarters of decrease while PMG explains the entire tightening; tightening almost entirely driven by improvement in global factors; small unexplained part mostly correction of undervaluation and some small overvaluation in some countries.
- Regional examples of misalignment dynamics:
  - Latin America: Brazil, Colombia, Peru had higher-than-predicted spreads 2001–2006, turning to negative or near-zero residuals 2006–2012; Chile and Mexico had no sizeable misalignment; Argentina shows positive residuals during debt-restructuring and mixed patterns thereafter.
  - Asia: Indonesia and Malaysia no substantial misalignment; Philippines similar to Brazil/Colombia/Peru; China and Pakistan lower-than-predicted spreads pre-crisis and above-predicted spreads 2010–2012.
  - Central and Eastern Europe: Bulgaria, Hungary, Poland exhibited lower spreads than fitted in 3–4 years pre-crisis ("EU Halo"); since 2007 mixed results—Poland small misalignment, Hungary notable residuals post-Lehman and after 2010 due to unorthodox policies and rule-of-law concerns.

### Main conclusions
- Both country-specific fundamentals and global factors are important determinants of foreign currency-denominated sovereign bond spreads; global factors are relatively more important in the short run.
- Strength of fundamentals affects sensitivity of risk premia to global factors: countries with stronger economic and financial indicators have lower sensitivity to changes in global risk aversion.
- Decomposition findings:
  - Improvements in global factors and country-specific fundamentals explain just more than half of the pre-crisis tightening.
  - During periods of severe stress and lack of clarity on country-specific developments (e.g., Eurozone debt crisis), global factors tend to drive spread changes and misalignment tends to increase.
- 2012–early 2013 tightening largely driven by global factors (improved market sentiment and ample global liquidity, including ECB OMT announcement).

### Policy-relevant implications and recommendations
- Improving fundamentals delivers two benefits:
  - Lower sovereign spreads.
  - Lower sensitivity of spreads to adverse global market changes (premium from solid domestic fundamentals).
- Caution for countries experiencing large inflows and falling spreads:
  - (i) The fall in spreads may reflect improvements in fundamentals only to a small extent.
  - (ii) Spreads may be lower than implied by domestic and global conditions in some countries (overvaluation).
  - (iii) Sensitivity to global factors is high for countries with weak fundamentals, implying potential reversal if industrial-country monetary stimulus is withdrawn or sentiment reverses.
- Recommendation: continued focus on improving fundamentals to lower both the level and sensitivity of sovereign spreads.

*Source: _wp13164 (PDF chapter excerpts)._

### 1. Descriptive Statistics ..............................................................................................

### 1. Descriptive Statistics

### Introduction and context
- After an extended period of heightened volatility, emerging market sovereign spreads narrowed steadily in the second half of 2012.
- Drivers of improved market sentiment included:
  - European Central Bank announcement of the Outright Monetary Transactions (OMT) program in September that includes the conditional purchase of the Euro Area sovereign bonds in unlimited amounts at the secondary market.
  - The Federal Reserve started a new bond purchase program and committed to keep interest rates at exceptionally low levels at least until the mid-2015.
  - The Bank of Japan announced further monetary easing.
  - Favorable political/legal developments: the outcome of Greek elections in mid-2012 and the favorable ruling of the German constitutional court regarding the European Stability Mechanism in September.
- As a result of exceptionally low yields in the industrial countries and ample liquidity, emerging market bond funds experienced significant inflows:
  - During the second half of 2012–beginning of 2013 emerging market bond funds experienced an inflow of almost US$ 1 billion per week on average.
  - The EMBI Global spread decreased by around 180 basis points between early June 2012 and late January 2013.
- A subsequent small correction occurred due to increasing risk aversion related to uncertainty in Cyprus.

### Main analytic objectives (three angles)
- Disentangle the effect of global and country-specific developments on emerging market debt spreads.
  - Finding: both country-specific and global developments matter for long-run spreads, but global factors dominate short-run dynamics.
  - Economic rationale: country fundamentals change slowly, while global factors vary quickly and drive short-run asset price movements.
- Investigate whether strength of fundamentals affects sensitivity of spreads to global factors.
  - Methods: (i) fixed effects panel with sample split into weak vs. strong fundamentals; (ii) pooled mean group estimation relating country-specific short-term coefficients to fundamentals.
  - Finding: countries with stronger fundamentals tend to have lower sensitivity to changes in global risk aversion.
- Decompose changes in spreads across seven sub-periods over the last decade to attribute movements to fundamentals, global factors, and unexplained residuals.
  - Contribution beyond prior work: decomposition includes changes in the residual split into correction of initial misalignment and increase/decrease in misalignment during each period.

### Key high-level results from decomposition and dynamics
- Across all emerging markets and the sample period, seven sub-periods can be identified characterized by general decreases or increases in spreads relative to the previous period.
- Pre-crisis period (up to August 2007):
  - Models explain about half of the tightening of spreads.
  - Both global factors and country-specific fundamentals played an important role.
  - The unexplained tightening was driven both by correction of earlier mispricing (underpricing of emerging market debt) and accumulation of misalignment during boom years.
- Crisis dynamics:
  - Periods of widening and tightening are generally well-explained by the model and align with crisis phases:
    - 2007–early Fall 2008: crisis mostly contained to mature financial systems.
    - Post-Lehman (late 2008): “mature market crisis” evolved into a confidence and growth crisis, spilling into emerging markets.
    - 2009–early 2010: partial thaw and fundamentals improvement.
    - Spring 2010–mid-2012: recurring stress from Eurozone debt crisis.
    - Mid-2012–early 2013: spectacular improvement in global market sentiment as monetary authorities alleviated tail-risk concerns.
  - In periods of severe market stress and general lack of public understanding of country-specific developments (for example, intensive phases of the Eurozone debt crisis), global factors tend to drive changes in spreads and misalignment increases in magnitude and share of actual spreads.
- 2012 performance:
  - The strong performance of emerging market sovereign debt in 2012 was mainly driven by an improvement in global factors (risk perception and liquidity).
  - The small unexplained component mostly reflected correction of undervaluation of emerging market debt; some misalignment began to build by early 2013 but broadly disappeared due to increasing uncertainty related to Cyprus.

### Policy-relevant implication
- Solid domestic fundamentals provide a premium: they reduce sensitivity of sovereign spreads to global risk aversion and thus offer some cushion against sudden shifts in global market sentiment.
- Short-run stabilization of spreads during episodes of ample global liquidity can lift all markets temporarily, but countries with weaker fundamentals remain vulnerable to corrections when spreads revert toward long-term fundamentals.

### Methods summary and dataset
- Sample and measure:
  - An unbalanced panel dataset of 147 monthly observations between January 2001 and March 2013 for 18 emerging markets in three regions.
  - Sovereign risk/spread measure: monthly averages of daily Emerging Market Bond Index Global (EMBIG) sovereign spreads from J.P. Morgan’s MorganMarkets.
  - EMBIG spread definition: market-capitalization-weighted average of spreads on US$-denominated Brady bonds, Eurobonds and traded loans issued by sovereign and quasi-sovereign entities.
- Explanatory variables:
  - The study uses a set of country-specific fundamentals and global factors (listed in the paper’s data section).

### Relation to existing literature
- Confirms and extends prior findings:
  - Global factors (international interest rates, volatility, risk aversion, liquidity) and country-specific fundamentals both matter; global factors often dominate short-run dynamics.
  - The sensitivity of spreads to global factors is lower for countries with stronger fundamentals.
- Builds on methodologies used in the literature:
  - Fixed effects panel estimation and pooled mean group estimation to separate short- and long-term drivers and allow heterogeneity in short-term coefficients across countries.
  - Novelty: comprehensive seven-period decomposition including decomposition of changes in residuals into correction of initial misalignment and changes in misalignment within each period.

*Source: _wp13164 - 1. Descriptive Statistics (PDF chapter).*

### 1. Country-specific fundamentals:

### 1. Country-specific fundamentals

### Overview
- The analysis uses risk indicators from the International Country Risk Guide (ICRG) database, which contains monthly data on economic, financial and political risk in 140 countries.
- Based on 22 variables, three composite ratings are constructed: Economic Risk Rating (ERR), Financial Risk Rating (FRR), and Political Risk Rating (PRR).
- Low values of these composite ratings signal higher risk; these indicators are expected to have a negative relationship with spreads.

### Composite risk ratings: definitions and ranges
- Economic Risk Rating (ERR):
  - Defined as the weighted average of risk points assigned to GDP per capita, real GDP growth, inflation, fiscal balance (percent of GDP) and current account balance (percent of GDP).
  - Range: 0–50.
- Financial Risk Rating (FRR):
  - Defined as the weighted average of risk points related to foreign debt (percent of GDP), foreign debt service (percent of exports), current account (percent of exports), official reserves (months of imports) and exchange rate stability (appreciation/depreciation against the US$ over the most recent 12-month period).
  - Range: 0–50.
- Political Risk Rating (PRR):
  - Defined as the weighted average of the risk rating of the following components: government stability, socioeconomic conditions, investment profile, internal conflict, external conflict, corruption, military in politics, religious tensions, law and order, ethnic tensions, democratic accountability, and bureaucracy quality.
  - Range: 0–100.

### Sample countries included in the sample
- Asia: China, Indonesia, Malaysia, Pakistan, Philippines
- Central and Eastern Europe: Bulgaria, Hungary, Poland, Russia, Turkey, Ukraine
- Latin America: Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela

### Market infrastructure considerations
- Market infrastructure characteristics—such as depth, liquidity or turnover—could influence the sensitivity of spreads to global factors.
- This is addressed in part by the inclusion criteria for debt instruments applied in constructing the EMBIG spreads: only issuances larger than US$ 500 million are included, which in part takes care of the depth and liquidity.

### Advantages and methodology note
- The ICRG composite risk ratings are readily available and capture several economic, financial and political variables.
- A detailed description of the methodology is available at PSR Group’s website (http://www.prsgroup.com/PDFS/icrgmethodology.pdf).

*Source: 1. Country-specific fundamentals (excerpt).*

### 2. Global factors:

### 2. Global factors:

### Definitions of global variables and expected signs
- Global risk aversion: The Chicago Board Options Exchange Volatility Index (VIX) measures the implied volatility of S&P index options (source: Bloomberg). VIX is expected to be positively associated with spreads.
- Global liquidity conditions: The U.S. Federal funds rate is used as a proxy for global liquidity conditions (data from the Federal Reserve website). Lower Fed funds rate is assumed to be associated with higher liquidity and is expected to have a positive relationship with spreads.
- Interaction and overlap:
  - VIX partially captures global liquidity conditions, especially during crisis periods and near the zero-bound when non-conventional monetary policy is employed.
  - Massive liquidity provision via balance sheet expansion by the Fed, the ECB, BOE, and BoJ during the recent crisis helped moderate market risk aversion.
  - VIX is forward-looking and reacts to central bank announcements (example: VIX moderated after the ECB announced its OMT program).

### Descriptive statistics and correlations (summary)
- Tables referenced:
  - Table 1: Descriptive statistics of the variables (sources: J.P. Morgan, ICRG, Bloomberg, Fed, authors’ calculations).
  - Table 2: Pairwise correlation matrix (correlations calculated between logs; p-values in parenthesis; source: authors’ calculations).
- Key correlation findings:
  - EMBIG spreads are negatively correlated with each risk rating variable (ERR, FRR, PRR) and the U.S. Federal funds rate.
  - EMBIG spreads are positively correlated with VIX; correlations are significantly different from zero.
  - Correlation between risk rating indicators is always positive; correlation between FRR and PRR is insignificant.
  - VIX has a significant negative correlation with each risk rating variable (higher global risk aversion associated with worse country fundamentals).

### Theoretical model (equilibrium and spread formation)
- Equilibrium condition for a risk-neutral investor lending to a country with non-zero default probability (price-taker in global debt markets):
  - Equation (1): (notation preserved from source) relates default probability (݌), global risk-free rate (݅
כ), and country-risk premium (ݏ).
- Country-risk premium expressed as:
  - Equation (2): ݏ = function of ݌ and ݅
כ (country-risk premium is positively related to probability of default and the global risk-free interest rate).
- Default probability logistic specification:
  - Equation (3): ݌ = exp(∑ߚ௜ܺ௜) / (1 + exp(∑ߚ௜ܺ௜)), where ܺ௜ and ߚ௜ are determinants and coefficients.
- Spread equation in log form:
  - Equation (4): ln(ݏ) = ln(1 + ݅
כ) ߚ + ∑ܺ௜ߚ௜, with ܺ௜ capturing country-specific fundamentals and global market conditions.

### Empirical specification and estimation techniques
- Two estimation techniques used to estimate equation (4):
  - Fixed effects panel regression (equation (5)):
    - Dependent variable: ln(EMBIG spread)
    - Country-specific fundamentals vector (k(1)) includes ERR, FRR, PRR.
    - Global explanatory variables vector (k(2)) includes VIX and U.S. Fed funds rate.
    - Country fixed effects ߤ௜ and Driscoll-Kraay robust standard errors used (robust to heteroskedasticity, serial correlation, cross-sectional dependence).
  - Pooled Mean Group (PMG) estimator (equations (6)–(9)), following Pesaran, Shin and Smith (1999):
    - ARDL (p,q,...,q) framework reparameterized to allow short-run coefficients to differ across countries while constraining long-run coefficients to be homogeneous.
    - By setting p=1 and q=1, model (9) estimated.
- Stationarity and cointegration:
  - Im-Pesaran-Shin and Fisher-type ADF tests: mixed for EMBIG spread; reject unit roots for country-specific explanatory variables.
  - VIX found stationary; U.S. Federal funds rate appears to follow a unit root.
  - Westerlund (2008) cointegration test: no cointegrating relationship found among these variables.
  - Phillips and Moon (2000): pooled regression of two nonstationary but non-cointegrated variables yields consistent long-run average regression coefficients as N and T become large; proceed with fixed effects estimation following literature.

### Fixed effects estimation: main quantitative findings
- Sample: 18 emerging markets (database of 18 emerging markets used in analysis).
- Whole-sample results (Table 3; dependent variable: Log of EMBIG spread):
  - Country-specific fundamentals:
    - Financial risk rating (FRR): a 1 percent increase (improvement) in FRR lowers spreads by 2.8–3.5 percent.
    - Political risk rating (PRR): a 1 percent increase lowers spreads by 2.3–2.5 percent.
    - Economic risk rating (ERR): a 1 percent increase lowers spreads by 0.5–0.6 percent.
    - Interpretation: spreads are most sensitive to FRR and PRR; investors pay more attention to political risk and liquidity situation (PRR and FRR) than to solvency indicators (ERR) in the short term.
  - Global factors:
    - VIX: a 1 percent increase in global risk aversion is associated with a 0.7 percent increase in spreads (positive and strongly significant).
    - U.S. Federal funds rate: coefficient negative but becomes insignificant when VIX is included; significant with negative coefficient in Asia in regional splits.
    - 3-month and 10-year U.S. Treasury yields tested but not significant.
  - Model fit:
    - Overall R-squared around 0.5–0.6 across three specifications.
  - Endogeneity checks:
    - Regressed spreads on up to five lags of each explanatory variable; sign, magnitude and significance of variables (except U.S. Federal funds rate) remain broadly the same, suggesting no endogeneity.
  - Multicollinearity checks:
    - ERR and FRR strongly correlated; excluding FRR increases sensitivity of spreads to PRR and VIX and increases ERR coefficient four-fold (with weakened significance) and lowers model explanatory power.
    - Excluding ERR does not substantially affect other coefficients or model explanatory power.
  - Group and regional splits:
    - Split-sample by low/high risk ratings:
      - FRR and PRR remain significant with negative coefficients in each specification.
      - Countries with higher economic and financial risk (low ERR and low FRR) have higher sensitivity to VIX.
      - VIX coefficient similar magnitude for low and high PRR groups.
    - Regional splits (Table 6):
      - ERR not significant in Europe, significant in other regions.
      - FRR more important in Asia and Europe; FRR and PRR roughly equal in Latin America.
      - VIX exposure highest for European countries.
      - U.S. Federal funds rate significant with negative coefficient in Asia.

### Pooled Mean Group (PMG) estimation: main quantitative findings
- Whole-sample PMG results (Table 7; dependent variable: Log of EMBIG spread):
  - Long run:
    - FRR and PRR: long-run coefficients negative and significant in baseline specification (stronger fundamentals associated with lower equilibrium risk premia).
    - ERR: coefficient changes sign compared to fixed effects and suggests stronger economic fundamentals associated with higher spreads in the long-term (short-term ERR coefficient negative but not significant).
    - VIX: significant and positive long-run effect on spreads.
    - U.S. Federal funds rate: not significant driver in long-run in specifications (1)-(3).
    - Error correction coefficient: significant but low (spread deviations from long-run equilibrium corrected at a slow rate).
  - Short run:
    - Global factors are principal drivers of short-run dynamics.
    - VIX: positive short- and long-run effect.
    - U.S. Federal funds rate: significantly negative short-run effect (monetary policy tightening in the U.S. lowers spreads in short-run) but ambiguous and insignificant in long-run where country fundamentals included.
    - Country-specific fundamentals: ERR not significant in any PMG specification; FRR and PRR significant in only one specification each.
  - Heterogeneity and sensitivity:
    - Country-specific short-term coefficients plotted against country-average risk ratings (Figure 3): countries with higher average risk ratings (better fundamentals) have lower sensitivity to VIX.
    - PMG split by region (Table 8):
      - Long-run VIX significant in each region; highest coefficient in Europe.
      - U.S. Federal funds rate significantly positive in Asia in long run (U.S. monetary loosening associated with declining Asian spreads).
      - FRR significant only in Europe; PRR negative in Asia and Latin America, surprisingly positive (weakly significant) in Europe.
      - Short run: both VIX (positive) and U.S. Federal funds rate (negative) significant in each regional specification.
      - Error correction term indicates misalignment corrected fastest in Asia, then Europe, then Latin America.

### Decomposition of changes in spreads and misalignment
- Misalignment defined as difference between actual and fitted spreads (residual).
- Three main sub-periods identified (visual analysis of Figures A1 and A2):
  a) Pre-crisis period, before 2007: most EM had positive residuals except Bulgaria, Chile, China, Hungary, Poland, Ukraine (markets required lower spreads than fundamentals suggested).
  b) Crisis mostly confined to industrial country financial systems, 2007–fall 2008: most EM had negative residuals (markets put a positive premium on EM debt).
  c) Broad-based crisis, since late 2008: majority of EM had positive residuals (more cautious pricing).
- Regional narratives (selected highlights):
  - Latin America: Brazil, Colombia, Peru had higher-than-predicted spreads 2001–2006, turning to negative or near-zero residuals 2006–2012; Chile and Mexico had no sizeable misalignment; Argentina patterns inconclusive with positive residuals during debt-restructuring period and negative residual 2005–2011 then positive in 2012.
  - Asia: Indonesia and Malaysia no substantial misalignment; Philippines similar to Brazil/Colombia/Peru; China and Pakistan had lower-than-predicted spreads pre-crisis and above-predicted spreads 2010–2012.
  - CEE: Bulgaria, Hungary, Poland had lower spreads than fitted in 3–4 years pre-crisis ("EU Halo"); since 2007 mixed results—Poland small misalignment, Hungary notable residuals post-Lehman and after 2010 due to unorthodox policies and rule-of-law concerns.
- Decomposition across identified sub-periods (Tables 10 and 11; FE and PMG):
  - Pre-crisis (Jan 2001–Aug 2007): models explain just above half of average compression in EM spreads; FE attributes larger improvement to country-specific factors; PMG attributes more to global factors. Excess compression reflects correction of initial undervaluation and emergence of overvaluation during favorable global environment.
  - July 2007–late summer 2008: spreads increased somewhat; models explain 30–50% of increase via deteriorating global and domestic fundamentals; unexplained part due to elimination of initial overvaluation and emergence of undervaluation.
  - Post-Lehman (late 2008): sharp increases in spreads; models explain 76 and 93 percent of this increase (FE and PMG, respectively), driven mostly by sharp deterioration in global factors and worsening country fundamentals.
  - Early 2009–April 2010: substantial tightening; models explain 70 and over 90 percent of tightening (FE and PMG), driven mostly by improved global market sentiment and some improvement in fundamentals; unexplained part mostly reduction in undervaluation.
  - April 2010–June 2012: substantial increase in spreads (Eurozone debt crisis); models explain around 40 percent of increases, with highest importance of misalignment (increase in misalignment) during mid-2011 onward.
  - June 2012–end of sample (March 2013): spreads narrowed substantially; FE explains around three quarters of decrease while PMG explains the entire tightening. This tightening is almost entirely driven by improvement in global factors. Small unexplained part mostly reflects correction of undervaluation and some small overvaluation in some countries.
- Policy-relevant decomposition insight:
  - The decline in spreads in the second half of 2012 and into early 2013 was driven largely by global factors (improved market sentiment and ample global liquidity due to easy policy in developed economies, and ECB OMT announcement).

### Main conclusions and policy implications
- Both country-specific fundamentals and global factors are important determinants of foreign currency-denominated sovereign bond spreads; global factors are relatively more important in the short run.
- Strength of fundamentals affects sensitivity of risk premia to global factors: countries with stronger economic and financial indicators have lower sensitivity to changes in global risk aversion.
- Decomposition of changes in spreads:
  - Improvements in global factors and country-specific fundamentals explain just more than half of the pre-crisis tightening.
  - During periods of severe stress and lack of clarity on country-specific developments (e.g., Eurozone debt crisis), global factors tend to drive spread changes and misalignment tends to increase.
- Practical policy implications:
  - Countries benefit from improving fundamentals through both lower sovereign spreads and lower sensitivity of spreads to adverse global market changes.
  - Caution for countries experiencing large inflows and falling spreads:
    - (i) the fall in spreads may reflect improvements in fundamentals only to a small extent;
    - (ii) spreads may be lower than implied by domestic and global conditions in some countries (overvaluation);
    - (iii) sensitivity to global factors is high for countries with weak fundamentals, implying potential reversal if industrial-country monetary stimulus is withdrawn or sentiment reverses.
  - Continued focus on improving fundamentals is recommended to lower both the level and sensitivity of sovereign spreads.

*Source: authors’ calculations (extracted from _wp13164 - 2. Global factors:).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2013/_wp13164.pdf_