## _wp10283

## Source details

**Canonical URL:** [_wp10283](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10283.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10283.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10283.pdf.json)

---

### Key findings and contributions
- Net private debt flows to developing countries increased more than sevenfold by 2007 compared to levels two decades earlier and accounted for half of the international capital inflows to these economies throughout the 2000s.
- Most of the upsurge was in cross-border bank lending, while bond issuance remained highly concentrated in a few emerging economies.
- Loan and bond spreads are determined by borrower risk-profile factors (e.g., loan (bond) rating), borrower type (sovereign vs private), and industry.
- Country macroeconomic fundamentals such as international reserves and investment ratios play an important role in determining access to international debt markets.
- Determinants of private cross-border debt flows are significant overall, and for some determinants the effects differ across global credit cycle phases (expansions vs contractions; booms vs busts).
- This study updates analysis through the recent global economic downturn and compares effects of determinants of private debt spreads across different phases of the global credit cycle.

### Data and methodology
- Period covered: January 1993 through December 2009.
- Countries: 129 emerging and developing economies for syndicated loans and 76 for international bond placements.
- Micro-data source: Dealogic (loan and bond characteristics including spread, volume, borrower type, industry, maturity, credit rating, collateral).
- Macroeconomic data source: IMF’s International Financial Statistics.
- Syndicated lending represents approximately two-thirds of cross-border lending to developing countries.
- Dealogic loan dataset comprises commitments and announcements (not actual disbursements); it includes cross-border bilateral loans in addition to syndicated loans.
- Two levels of empirical analysis:
  - Disaggregated loan (bond)-level regressions using micro-data for 1993–2009.
  - Aggregated cross-country panel regressions adding country-specific macroeconomic factors.
- Cycle identification:
  - Expansions/contractions: twelve months preceding a peak in smoothed monthly series = expansion; twelve months following the peak = contraction. Three global expansion-contraction episodes identified: July 1996 through June 1998; July 1999 through June 2001; May 2006 through April 2008.
  - Booms/busts: smoothed monthly series above/below thresholds set at a half and a one standard deviation band around the historical mean.

### Loan and bond cross-section (disaggregated) findings
- Baseline loan cross-section (Table 6, baseline, full sample; Adjusted R-squared: 0.21; N = 15,837):
  - 1 = public borrower: coefficient -0.170***
  - 1 = financial borrower: coefficient -0.333***
  - 1 = manufacturing borrower: coefficient 0.080***
  - 1 = services borrower: coefficient 0.069***
  - 1 = presence of collateral: coefficient 0.162***
  - Log loan value: -0.153***
  - Log loan maturity: 0.061***
- Interpretation:
  - Public sector and financial borrowers pay on average about one fifth and one third less, respectively, on cross-border loans compared to private and non-financial counterparts.
  - Collateralized loans associated with spreads about 16 percent higher than non-collateralized loans (collateral may signal risk); effect stronger during expansions.
  - Including rating dummies makes the collateral effect negligible; being investment grade strongly lowers cost:
    - 1 = investment grade (loans, Table 7 column 1): -0.750***.
- Bond cross-section (Table 8 baseline, full sample; Adjusted R-squared: 0.25; N = 3,874):
  - 1 = public borrower: -0.283***
  - 1 = financial borrower: -0.361***
  - 1 = services borrower: 0.257***
  - 1 = presence of collateral: -0.616***
  - Log bond volume: 0.035
  - Log bond maturity: 0.036
  - Investment-grade effect (bonds, Table 9): 1 = investment grade: -1.041***.
- Additional cross-section notes:
  - Sector premia vary by cycle phase (e.g., manufacturing premium rises in good times, insignificant in bad times).
  - Loans to financial companies elicit much smaller cost increases in bad times relative to non-financial borrowers.

### Panel (country-level aggregated) findings
- Model: ln Sjt = ln Mjt β'' + Zjt β' + γj + αt + εjt
  - Sjt: volume weighted period average loan (bond) spread in borrowing country j.
  - Zjt: country-specific macro variables (Reserves/GDP, Investment/GDP, Trade/GDP, Growth, Change in the real exchange rate, Inflation).
- Summary statistics (Table 10):
  - Reserves/GDP: Observations 24,750; Mean 0.81; Standard deviation 0.76; Min 0.03; Max 5.00
  - Investment/GDP: Observations 22,500; Mean 0.23; Standard deviation 0.07; Min 0.07; Max 0.81
  - Trade/GDP: Observations 24,050; Mean 0.81; Standard deviation 0.59; Min 0.03; Max 5.66
  - Growth: Observations 23,980; Mean 0.01; Standard deviation 0.17; Min -0.94; Max 0.87
  - Change in real exchange rate: Observations 9,508; Mean 0.00; Standard deviation 0.08; Min -0.90; Max 3.62
  - Inflation: Observations 8,730; Mean 0.07; Standard deviation 1.41; Min -0.30; Max 126.97

- Cross-border bank lending panel (Table 11, Log spreads):
  - Full sample (column 1; Observations: 1,017; R-squared: 0.200):
    - Reserves/GDP: -0.402*** (0.113)
    - Investment/GDP: -2.223*** (0.471)
    - Trade/GDP: 0.199 (0.188)
    - Growth: -0.212 (0.263)
    - Change in real exchange rate: -0.166 (0.429)
    - Inflation: 0.153 (0.268)
    - Log maturity: -0.0288 (0.0299)
    - Constant: 5.498*** (0.189)
  - Expansion vs. contraction highlights (Observations: Expansion 251; Contraction 247):
    - Reserves/GDP: Expansion -1.035*** (0.204); Contraction -0.434** (0.214)
    - Investment/GDP: Expansion -1.619* (0.884); Contraction -1.344 (0.852)
    - Trade/GDP: Expansion 0.678*** (0.253); Contraction -0.418* (0.243)
    - Growth: Expansion 0.335 (0.533); Contraction -0.862* (0.507)
  - Boom vs. bust highlights (Observations: Boom 446; Bust 382):
    - Investment/GDP: larger reduction during bust times in some specifications (e.g., -1.515** in boom/baseline; -1.973*** in bust/baseline)
    - Log maturity: Boom 0.114** (0.0467); Bust -0.0143 (0.0439)

- Cross-border bond issuance panel (Table 12, Log spreads):
  - Full sample (column 1; Observations: 632; R-squared: 0.304):
    - Reserves/GDP: 0.0828 (0.175)
    - Investment/GDP: -1.009 (0.746)
    - Trade/GDP: -0.114 (0.319)
    - Growth: 0.00209 (0.395)
    - Change in real exchange rate: 2.605*** (0.769)
    - Inflation: 0.3185 (0.326)
    - Log maturity: -0.150** (0.0621)
    - Constant: 5.898*** (0.316)
  - Expansion vs. contraction highlights (Observations: Expansion 166; Contraction 133):
    - Change in real exchange rate: Expansion 1.970 (1.144); Contraction 2.139 (0.822)
    - Log maturity: Expansion -0.429*** (0.148); Contraction 0.0969 (0.108)
  - Boom vs. bust highlights (Observations: Boom 282; Bust 225):
    - Reserves/GDP: Boom -0.180 (0.221); Bust 0.643** (0.293)
    - Log maturity: Boom 0.0250 (0.101); Bust -0.291** (0.122)

- Panel-level interpretation:
  - Reserves/GDP and Investment/GDP have the largest and most consistent negative associations with loan spreads in the full sample (Reserves/GDP -0.402***; Investment/GDP -2.223***).
  - Trade/GDP effect changes by cycle phase: positive during expansions (Trade/GDP 0.678*** in expansion) and negative during contractions (Trade/GDP -0.418* in contraction).
  - Change in the real exchange rate raises bond spreads (Change in real exchange rate 2.605*** in full-sample bond regression), consistent with currency-mismatch and repayment concerns.
  - Maturity effects vary by episode and instrument:
    - Loan maturities: positive effect on spreads during boom times (Log maturity 0.114** in boom).
    - Bond maturities: associated with a larger cost reduction in expansion compared to contractions in some specifications (e.g., Log maturity -0.429*** in expansion).

### Robustness, limitations, and caveats
- Introducing rating dummies reduces the collateral effect and shows strong negative coefficients for investment grade in both loans and bonds; sample size drops when ratings are included.
- One-period lagged macro aggregates produce similar results except lagged real GDP growth loses significance.
- Limitations:
  - Changing composition of new loan and bond issues over time may affect average spreads; cross-country panel cannot fully control for composition effects.
  - Observed equilibrium spreads and volumes make it difficult to disentangle push (global supply/liquidity) vs. pull (country demand/creditworthiness) factors.
  - No quarterly series for debt-to-GDP or debt-service; country fixed effects used as proxy.
  - Potential sample selection/self-selection issues (e.g., only more creditworthy borrowers access markets during downturns) are not fully resolved.

### Policy-relevant conclusions and implications
- Higher international reserve holdings and higher investment ratios are associated with materially lower borrowing costs on average:
  - Reserves/GDP coefficient (loans, full sample): -0.402***.
  - Investment/GDP coefficient (loans, full sample): -2.223***.
- Investment ratios likely convey information about future growth potential and repayment capacity, reducing spreads relative to consumption-financed borrowing.
- Exchange-rate stability matters for bond issuance costs: changes in the real exchange rate significantly raise bond spreads (Change in real exchange rate 2.605*** in Table 12, full sample).
- Cycle-dependent effects imply policy focus should consider prevailing global lending conditions:
  - During expansions, trade openness (Trade/GDP) may raise loan spreads, potentially reflecting import-driven credit demand and shorter maturity trade credit.
  - During contractions, higher trade shares can reduce spreads, potentially reflecting export resilience.
- Credit quality and disclosure (ratings) are strong determinants of issuance cost reductions; strengthening credit information and obtaining investment-grade ratings can lower borrowing costs across episodes.

*Source: _wp10283 - References  ............................................................................................................*

### References  ............................................................................................................

### _wp10283 - References  ............................................................................................................

### Key findings and contributions
- Net private debt flows to developing countries increased more than sevenfold by 2007 compared to levels two decades earlier and accounted for half of the international capital inflows to these economies throughout the 2000s.
- Most of the upsurge was in cross-border bank lending, while bond issuance remained highly concentrated in a few emerging economies.
- Loan and bond spreads are determined by borrower risk-profile factors (e.g., loan (bond) rating), borrower type (sovereign vs private), and industry.
- Country macroeconomic fundamentals such as international reserves and investment ratios play an important role in determining access to international debt markets.
- Many determinants of private cross-border debt flows are significant overall, and for some determinants the effects differ across global credit cycle phases (expansions vs contractions; booms vs busts).
- This study provides an updated analysis through the recent global economic downturn and is the first, to the authors’ knowledge, to compare the effects of determinants of private debt spreads across different phases of the global credit cycle.

### Data and sample composition
- Period covered: January 1993 through December 2009.
- Countries: 129 emerging and developing economies for syndicated loans and 76 for international bond placements.
- Micro-data source: Dealogic (loan and bond characteristics including spread, volume, borrower type, industry, maturity, credit rating, collateral).
- Macroeconomic data source: IMF’s International Financial Statistics.
- The debt flows considered are concentrated mostly in middle-income countries and the East Asia Pacific, Europe and Central Asia, and Latin America regions.
- Syndicated lending represents approximately two-thirds of cross-border lending to developing countries.
- The Dealogic loan dataset comprises commitments and announcements (not actual disbursements); it includes cross-border bilateral loans in addition to syndicated loans.

### Empirical approach and methodology
- Two levels of empirical analysis:
  - Disaggregated loan (bond)-level regressions using micro-data on cross-border syndicated loans and international bond issuances for 1993–2009.
  - Aggregated cross-country panel regressions adding country-specific macroeconomic factors.
- Main empirical focus: determinants of loan and bond spreads and how these determinants change across cycle phases.
- Identification of cyclical fluctuations:
  - Series constructed by aggregating loan and bond volumes into monthly series; normalized by population (real flows per capita, using US GDP deflator) and by GDP (flows as a percentage of GDP).
  - Hodrick-Prescott filter applied to monthly series to smooth short-term fluctuations.
  - Expansions/contractions: the twelve months preceding a peak in the smoothed series defined as an expansion; the twelve months following the peak defined as a contraction. Using this method, three global expansion-contraction episodes are identified:
    - July 1996 through June 1998
    - July 1999 through June 2001
    - May 2006 through April 2008
  - Booms/busts: periods when the smoothed monthly series of lending volumes are above (boom) or below (bust) thresholds set at a half and a one standard deviation band around the historical mean.
- Rationale for two partitioning methods:
  - Expansions/contractions capture turning points where growth rates shift sign around peaks in lending.
  - Booms/busts capture unusually high or low levels of lending relative to the long-term average (positive/negative gaps).
  - The two methods classify the global lending series differently but are complementary for understanding cyclical fluctuations; no a priori expectation about how empirical relationships differ between methods.

### Literature context and empirical expectations
- Prior literature documents importance of country-specific macro variables and global conditions in determining credit spreads:
  - Edwards (1984): spreads increase with debt-to-GNP ratios and decrease with investment-to-GNP ratios.
  - Kamin and Kleist (1999): emerging market spreads declined up to the Asian crisis more than could be explained by fundamentals alone; spreads responded to high levels of global liquidity.
  - Studies find differences between bond and loan markets (bond spreads often higher than loan spreads).
- Global conditions such as global bond issuance (liquidity) and developed-economy interest rates influence volumes of bond issuance in emerging markets.
- Short-term debt is generally more volatile than medium- and long-term debt, especially during crises; crises can alter maturity and composition of flows.
- Sovereign crises can reduce foreign credit to private firms through supply-channel (risk perception), demand-channel, and exogenous shocks affecting sovereign-default probability; foreign credit to private sector can decline substantially (e.g., approximately 20 percent more than expected from fundamentals following a sovereign debt crisis).

### Specific empirical design notes
- Loan data from Dealogic includes some cross-border bilateral loans and represents commitments/announcements rather than disbursements.
- Gross commitments are used to assess borrowers’ market access because disbursements can mask active borrowers (large disbursements offset by large repayments).
- The paper estimates cross-section and panel regressions; disaggregated regressions exploit richer borrower-level characteristics (rating, maturity, collateral, borrower type, industry) complementing macro fundamentals.

*Source: _wp10283 - References  ............................................................................................................*

### 1. Loan (Bond) Level Cross-Section Estimation

### 1. Loan (Bond) Level Cross-Section Estimation

### Model specification
- Baseline cross-section equation (log-linear):
  - ln S_i = β2' ln X_i + β4' ln Y_i + β6' DMY + β8' Dx_i + β10' D_i + ε_i  (equation (1))
- Definitions of key variables:
  - S_i: the spread of loan (bond) i.
  - Dx_i: vector of dummy variables capturing loan (bond) attributes (e.g., borrower sector — public/private, financial/non-financial, manufacturing/services/energy — as well as debt collateral and quality).
  - Y_i: the value of the loan (bond) i.
  - D_i: set of regional and time dummies that control for regional and global shocks.
  - ε_i: error term.

### Estimation approach
- Estimation method:
  - Model estimated by OLS over the full sample and over sub-samples defined by cycle phases: expansion, contraction, boom, and bust.
- Instrument differentiation:
  - Regressions are estimated separately for loans and bonds because loans and bonds have different characteristics and foreign debt composition matters, particularly in times of crisis.
- Syndicated loans:
  - Syndicated loans may be easier to customize and renegotiate than bonds, which may impact the overall quality of borrowers that have access to loans.

### Testing differences across credit cycle phases
- Pooled interaction approach:
  - To test statistical differences in the effect of explanatory variables between phases of the credit cycle (expansion vs. contraction and boom vs. bust), the model is estimated on observations pooled from expansions and contractions (booms and busts, respectively) with interactions between an expansion (boom) dummy and all relevant regressors.
  - Statistically significant coefficients on the interacted variables indicate that the impact of exogenous variables differs across cycle phases.

### Booms and busts illustrations (figures referenced)
- Figure 6: "Booms and busts in private debt flows to emerging and developing countries - real flows per capita 1/"
  - Real private debt per capita
  - US$ 2007
  - Footnote: 1/ Half and one standard deviations bands around the historical mean of real private debt per capita.
- Figure 7: "Booms and busts in private debt flows to emerging and developing countries - flows to GDP 1/"
  - Private debt to GDP
  - Footnote: 1/ Half and one standard deviations bands around the historical mean of private debt to GDP.
- Axis/date markers shown in figures: Jan 93, Jan 94, Jan 95, Jan 96, Jan 97, Jan 98, Jan 99, Jan 00, Jan 01, Jan 02, Jan 03, Jan 04, Jan 05, Jan 06, Jan 07, Jan 08, Jan 09.
- Example vertical scale values (from figures): 0.1%, 0.2%, 0.3%, 0.4%, 0.5%.

*Source: _wp10283 - 1. Loan (Bond) Level Cross-Section Estimation*

### 2. Panel Estimation

### 2. Panel Estimation

### Methodology and specification
- Model specification (country j, quarter t):
  - ln Sjt = ln Mjt β'' + Zjt β' + γj + αt + εjt
  - Sjt: volume weighted period average loan (bond) spread in borrowing country j
  - Zjt: vector of country-specific macro variables
  - Mjt: period average of volume weighted loan (bond) maturities
  - γj: country-specific fixed effects
  - αt: time-specific fixed effects
  - εjt: error term
- Macroeconomic fundamentals used as explanatory variables:
  - Reserves/GDP
  - Investment/GDP
  - Trade/GDP (imports + exports as ratio to GDP)
  - Growth (growth rate of real GDP)
  - Change in the real exchange rate
  - Inflation
- Cycle-phase analysis:
  - Dummy variable = 1 during expansion (boom) quarters and interactions with each explanatory variable to detect differentiated effects across cycle phases.
  - Timing variables capture global lending pattern fluctuations; country fixed effects proxy for unobserved capacity-to-repay metrics (e.g., debt/GDP not available quarterly).

### Cross-section (individual loan and bond) findings
- Spreads vary across cycle phases:
  - Average cost increase from boom to bust for issuing international bonds is about 60 basis points more compared with obtaining a cross-border syndicated loan.
  - During contractions, bond issuance cost is about 15 basis points less than that of syndicated loans.
- Sector and borrower-type effects (loan cross-section, Table 6 baseline, full sample):
  - 1 = public borrower: coefficient -0.170***
  - 1 = financial borrower: coefficient -0.333***
  - 1 = manufacturing borrower: coefficient 0.080***
  - 1 = services borrower: coefficient 0.069***
  - 1 = presence of collateral: coefficient 0.162***
  - Log loan value: -0.153***
  - Log loan maturity: 0.061***
  - Adjusted R-squared: 0.21; N = 15,837
- Interpretation of loan cross-section results:
  - Public sector and financial borrowers pay on average about one fifth and one third less, respectively, on cross-border loans compared to private and non-financial counterparts (Table 6, column 1).
  - Collateralized loans are associated with spreads about 16 percent higher than non-collateralized loans (consistent with collateral signaling risk), with a stronger positive effect during expansions.
  - When a rating dummy is added (Table 7), collateral effect becomes negligible and being rated “investment grade” reduces borrowing cost by:
    - 1 = investment grade: coefficient -0.750*** (full sample, Table 7 column 1)
  - Loan maturity and volumes:
    - Maturity positively correlated with loan spreads; volume negatively correlated.
- Bond cross-section results (Table 8 baseline, full sample):
  - 1 = public borrower: -0.283***
  - 1 = financial borrower: -0.361***
  - 1 = services borrower: 0.257***
  - 1 = presence of collateral: -0.616***
  - Log bond volume: 0.035
  - Log bond maturity: 0.036
  - Adjusted R-squared: 0.25; N = 3,874
- Bond-specific insights:
  - Presence of collateral significantly reduces financing cost for bonds (contrast with loans).
  - Investment grade for bonds strongly reduces cost:
    - 1 = investment grade: -1.041*** (Table 9, full sample, rated specification)
- Additional cross-section notes:
  - Borrower sector premia can vary by cycle phase (e.g., manufacturing premium rises in good times, insignificant in bad times).
  - Loans to financial companies elicit much smaller cost increases in bad times relative to non-financial borrowers.
  - Caveats: limited ability to control for local bank participation, changing investor assessments across crises, selection into market during downturns.

### Panel (country-level aggregated) results
- Summary statistics (Table 10, panel variables):
  - Reserves/GDP: Observations 24,750; Mean 0.81; Standard deviation 0.76; Min 0.03; Max 5.00
  - Investment/GDP: Observations 22,500; Mean 0.23; Standard deviation 0.07; Min 0.07; Max 0.81
  - Trade/GDP: Observations 24,050; Mean 0.81; Standard deviation 0.59; Min 0.03; Max 5.66
  - Growth: Observations 23,980; Mean 0.01; Standard deviation 0.17; Min -0.94; Max 0.87
  - Change in real exchange rate: Observations 9,508; Mean 0.00; Standard deviation 0.08; Min -0.90; Max 3.62
  - Inflation: Observations 8,730; Mean 0.07; Standard deviation 1.41; Min -0.30; Max 126.97
- Panel regressions for cross-border bank lending (Table 11, Log spreads):
  - Full sample (column 1):
    - Reserves/GDP: -0.402*** (0.113)
    - Investment/GDP: -2.223*** (0.471)
    - Trade/GDP: 0.199 (0.188)
    - Growth: -0.212 (0.263)
    - Change in real exchange rate: -0.166 (0.429)
    - Inflation: 0.153 (0.268)
    - Log maturity: -0.0288 (0.0299)
    - Constant: 5.498*** (0.189)
    - Observations: 1,017; R-squared: 0.200
  - Expansion vs. contraction (columns 2–3 highlights):
    - Reserves/GDP coefficients:
      - Expansion: -1.035*** (0.204)
      - Contraction: -0.434** (0.214)
    - Investment/GDP:
      - Expansion: -1.619* (0.884)
      - Contraction: -1.344 (0.852)
    - Trade/GDP:
      - Expansion: 0.678*** (0.253)
      - Contraction: -0.418* (0.243)
    - Growth:
      - Expansion: 0.335 (0.533)
      - Contraction: -0.862* (0.507)
    - Log maturity:
      - Expansion: -0.0843 (0.0678)
      - Contraction: 0.0616 (0.0586)
    - Observations: Expansion 251; Contraction 247
  - Boom vs. bust (columns 4–5 highlights):
    - Investment/GDP shows a larger reduction during bust times in some specifications (e.g., coefficient -1.515** in boom/baseline; -1.973*** in bust/baseline).
    - Log maturity:
      - Boom: 0.114** (0.0467)
      - Bust: -0.0143 (0.0439)
    - Observations: Boom 446; Bust 382
- Panel regressions for cross-border bond issuance (Table 12, Log spreads):
  - Full sample (column 1):
    - Reserves/GDP: 0.0828 (0.175)
    - Investment/GDP: -1.009 (0.746)
    - Trade/GDP: -0.114 (0.319)
    - Growth: 0.00209 (0.395)
    - Change in real exchange rate: 2.605*** (0.769)
    - Inflation: 0.3185 (0.326)
    - Log maturity: -0.150** (0.0621)
    - Constant: 5.898*** (0.316)
    - Observations: 632; R-squared: 0.304
  - Expansion vs. contraction (columns 2–3 highlights):
    - Change in real exchange rate significant positive:
      - Expansion: 1.970 (1.144)
      - Contraction: 2.139 (0.822)
    - Log maturity:
      - Expansion: -0.429*** (0.148)
      - Contraction: 0.0969 (0.108)
    - Observations: Expansion 166; Contraction 133
  - Boom vs. bust (columns 4–5 highlights):
    - Reserves/GDP:
      - Boom: -0.180 (0.221)
      - Bust: 0.643** (0.293)
    - Log maturity:
      - Boom: 0.0250 (0.101)
      - Bust: -0.291** (0.122)
    - Observations: Boom 282; Bust 225
- Panel-level interpretation:
  - Reserves/GDP and Investment/GDP have the largest and most consistent negative associations with loan spreads in the full sample and across some cycle phases (e.g., Reserves/GDP -0.402***; Investment/GDP -2.223***).
  - Trade/GDP effect differs by cycle phase: increases credit spreads during expansions (Trade/GDP 0.678*** in expansion) and reduces them in contractions (Trade/GDP -0.418* in contraction).
  - Change in the real exchange rate increases bond spreads (Change in real exchange rate 2.605*** in full-sample bond regression), consistent with currency-mismatch and repayment concerns.
  - Maturity effects vary by episode and instrument:
    - Loan maturities: positive effect on spreads during boom times (Log maturity 0.114** in boom, Table 11), negative or muted in other phases.
    - Bond maturities: associated with a larger cost reduction in expansion compared to contractions in some specifications (Table 12).

### Robustness checks and caveats
- Rating-variable robustness:
  - Introducing rating dummies reduces the collateral effect and shows strong negative coefficients for investment grade in both loans and bonds.
  - Note: sample size drops when ratings are included; fewer observations for rated deals.
- Endogeneity checks:
  - One-period lagged macro aggregates produce similar results except lagged real GDP growth loses significance.
- Limitations explicitly noted:
  - Changing composition of new loan and bond issues over time may affect average spreads; cross-country panel cannot fully control for these composition effects.
  - Observed equilibrium spreads and volumes make it difficult to disentangle push (global supply/liquidity) vs. pull (country demand/creditworthiness) factors.
  - No quarterly series available for debt-to-GDP or debt-service; country fixed effects used as proxy.
  - Potential sample selection/self-selection issues (e.g., only more creditworthy borrowers access markets during downturns) are not fully resolved.

### Policy-relevant conclusions
- At the country level, higher international reserve holdings and higher investment ratios are associated with materially lower borrowing costs on average.
  - Reserves/GDP coefficient (loans, full sample): -0.402***.
  - Investment/GDP coefficient (loans, full sample): -2.223***.
- Investment ratios likely convey information about future growth potential and repayment capacity, reducing spreads relative to consumption-financed borrowing.
- Exchange-rate stability matters for bond issuance costs: changes in the real exchange rate significantly raise bond spreads (Change in real exchange rate 2.605*** in Table 12, full sample).
- Cycle-dependent effects imply policy focus should consider prevailing global lending conditions:
  - During expansions, trade openness (Trade/GDP) may raise loan spreads, potentially reflecting import-driven credit demand and shorter maturity trade credit.
  - During contractions, higher trade shares can reduce spreads, potentially reflecting export resilience.
- Credit quality and disclosure (ratings) are strong determinants of issuance cost reductions; strengthening credit information and obtaining investment-grade ratings can lower borrowing costs across episodes.

*Source: IMF Working Paper section "2. Panel Estimation" (content unit: _wp10283 - 2. Panel Estimation).*

### REFERENCES

### REFERENCES

### Cited works
- Abbas, A. and J. Christensen, 2007, “The Role of Domestic Debt Markets in Economic 
 Growth: An Empirical Investigation for Low-Income Countries and Emerging 
 Markets”, IMF Working Papers No.07/127. 

- Andrew, D. and S. Ishii, 1995, “The Mexican Financial Crisis: A Test of the Resilience of 
 the Markets for Developing Country Securities”, IMF Working Paper No.95/132. 

- Antzoulatos, A., 2000, “On the determinants and resilience of bond flows to LDCs, 1990-
            1995”,            Journal of International Money and Finance, Vol. 19, Issue 3, pp 399–418.  

- Arteta, C. and G. Hale, 2008, “Sovereign debt crises and credit to the private sector”, Journal 
 of International Economics, Vol. 74, Issue 1, pp. 53–69. 

- Berger, A. and G. Udell, 1990,”Collateral, loan quality, and bank risk”, Journal of Monetary 
            Economics, Vol. 25, Issue 1, pp. 21–42. 

- Edwards, S., 1984, “LDC Foreign Borrowing and Default Risk: An Empirical
            Investigation:            1976-1980”,            American Economic Review, Vol. 74, Issue 4, pp. 726-
            734.            

- Eichengreen B., and A. Mody, 1998, “What explains changing spreads on emerging market 
 debt: fundamentals or market sentiment?” , NBER Working Paper 6408. 

- Hale, G., 2007, “Bonds or Loans? The Effect of Macroeconomic Fundamentals”, Economic 
 Journal, Royal Economic Society, Vol. 117, Issue 516, pp. 196–215, 01. 

- Kamil, H., and K. Rai, 2010, “The Global Credit Crunch and Foreign Banks' Lending to 
 Emerging Markets: Why Did Latin America Fare Better?”, IMF Working Papers 
            No.10/102.            

- Kamin, S. and K.von Kleist, 1999, “The evolution and determinants of emerging 
 market credit spreads in the 1990s”, BIS Working Paper No.68. 

- Jeanneau, S. and M. Micu, 2002, ”Determinants of International Bank Lending to Emerging 
 Market Countries”, BIS Working Paper No. 112.  

- Laeven, L. and F. Valencia, 2008, “The Use of Blanket Guarantees in Banking Crises”, IMF 
 Working Paper No. 08/250. 

- John, K., Lynch, A., and M. Puri, 2003, “Credit Ratings, Collateral and Loan Characteristics: 
 Implications for Yield”, Journal of Business, Vol. 76, Issue 3, pp. 371–409. 

- Mendoza, E. and M. Terrones, 2008, “An Anatomy Of Credit Booms: Evidence From 
 Macro Aggregates And Micro Data”, IMF Working Paper No. 08/226. 

- Min, H., 1998, “Determinants of Emerging Market Bond Spread: Do Economic 
            Fundamentals            Matter            ?”,            World Bank Policy Research Working Paper No. 1899. 

- Min, H., Lee, D., Park, M. and S. Nam, 2003, “Determinants of Emerging Market 
 Bond Spreads: Cross-Country Evidence”, Global Finance Journal, Vol. 14, Issue 3, 
            pp.            271–86.            

- Nini, G., 2004, “The Value of Financial Intermediaries: Empirical Evidence from Syndicated 
 Loans to Emerging Market Borrowers”, Federal Reserve Board International Finance 
 Discussion Paper No. 820. 

- Ul Haque, N., Kumar, M., Mark, N. and D. Mathieson, 1996, “The Economic Contents of 
 Indicators of Developing Country Creditworthiness”, IMF Working Paper No.96/9. 

- Uribe, M. and V. Yue, 2006, “Country Spreads and Emerging Countries: Who Drives 
            Whom?”,            Journal of International Economics, Vol. 66. Issue 1, pp. 6–36 

- World Bank, 2009, Global Development Finance, Washington, D.C. 

- Wooldridge, P., 2002, “Uses of the BIS Statistics: An Introduction”, BIS Quarterly Review, 
 March, pp. 75–92.

*Source: REFERENCES (content unit).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2010/_wp10283.pdf_