## wpiea2021185-print-pdf - Section 3 discusses the overall result comparing the cost of borrowing in the dollar versus the euro for global firms outside of the two currency regions

## Source details

**Canonical URL:** [wpiea2021185-print-pdf - Section 3 discusses the overall result comparing the cost of borrowing in the dollar versus the euro for global firms outside of the two currency regions](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021185-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021185-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021185-print-pdf.pdf.json)

---

### Methodology: measuring unhedged and hedged relative borrowing cost
- Unhedged relative borrowing cost
  - Two panels of secondary market bond yield and price data for euro- and U.S. dollar-denominated bonds issued by firms located outside the U.S. and euro area.
  - Secondary market yields used as a proxy for primary market issuance yields under the assumption that spreads between primary and secondary markets are small for large global firms and do not vary systematically by currency.
  - Cross-sectional regression (estimated each date t):
    - y_it = α_t 1_{EUR,i} + β_{f t} + γ_{m t} + δ_{r t} + ε_{it}
    - α_{EUR,t} (ˆα_t) is the euro minus dollar yield differential at time t controlling for firm fixed effects, maturity bucket and rating bucket. Data are winsorized on the dependent variable with a 95% window at each month.
- Hedged relative borrowing cost (corporate covered interest rate parity basis ψ)
  - Define ψ_t ≡ (y^e_t − y^$_t) + (f_t − s_t), where y^e_t and y^$_t are risky euro and dollar bond yields, and f_t and s_t are forward and spot exchange rates.
  - Decompose ψ_t as:
    - ψ_t = (y^e_t − r^e_t) + [r^e_t − r^$_t + (f_t − s_t)] − (y^$_t − r^$_t)
    - Separates euro credit spread, cross-currency basis (risk-free CIP deviation), and dollar credit spread.
  - Construct CIP-adjusted credit spread per bond:
    - S^{FXHedged}_{it} ≡ y_{it} − r^c_t + 1_{EUR,i}[r^e_t − r^$_t + (f_t − s_t)]
  - Regress S^{FXHedged}_{it} on same covariates as unhedged regression:
    - S^{FXHedged}_{it} = α_t 1_{EUR,i} + β_{f t} + γ_{m t} + δ_{l t} + ε_{it}
    - Estimated ˆα_t measures the currency-hedged relative borrowing cost ψ_t.

### Data: samples, coverage, and construction
- Monthly bond price dataset (August 2003 to September 2020)
  - Source: Bloomberg secondary market yields.
  - Sample criteria: euro- and U.S. dollar-denominated fixed- and zero-coupon non-callable corporate bonds with at least $50 million notional and at least one year to maturity; issuers outside the U.S. and euro area; issuer has a qualifying bond trading in both currencies at some point.
  - Exclude issuers whose country of risk is the United States or euro area and exclude supranationals.
  - Final monthly dataset: 3,452 bonds issued by 172 firms.
- Daily bond price dataset (January 2014 to September 2020)
  - Source: Bloomberg BVAL evaluated and quoted prices; yields estimated using duration and coupon.
  - Final daily dataset: 4,485 bonds issued by 190 firms.
- Dealogic issuance data (2000–2020)
  - Coverage: 82,185 euro- and dollar-denominated bonds (maturity at issuance > 1 year) issued by firms with nationality outside the U.S. and euro area.
  - Sum notional value: $12.4 trillion, with $8.8 trillion issued in the dollar and $3.6 trillion issued in the euro.
- Risk-free and cross-currency curves
  - Linearly interpolated to match bond maturities at each observation date.

### Relative borrowing cost: aggregate findings
- Dollar issuance share and magnitude
  - Since the introduction of the euro, about 75% of euro and dollar corporate issuance outside the U.S. and euro area has been denominated in dollars for much of the post-GFC period.
- Unhedged and hedged basis time-series (August 2003–September 2020)
  - Unhedged basis: widened to about -180 basis points between euro launch and GFC (negative indicating dollar cheaper), fell as far as -320 basis points after the crisis, rose in 2019 and tightened to about -100 basis points during onset of COVID-19.
  - Hedged corporate basis: generally held consistently close to zero, dipping in March 2020.
- Fama–MacBeth regression aggregated results (Table 4 summary)
  - Hedged borrowing costs between the two currencies are similar on average:
    - Euro coefficient is -5.5 basis points and not significant at the 5% level.
  - Unhedged borrowing cost difference favors the euro on average:
    - On average, the euro borrowing cost is 103 basis points less than the dollar borrowing cost.
- Interpretation
  - Similarity in hedged yields implies that larger dollar issuance is met with sufficient demand so marginal cost of issuance is similar when hedged.
  - Comparing bonds of the same firm in different currencies helps isolate currency denomination effects from country/sovereign effects.

### Impact of dollar shortages during COVID-19 and crisis episodes
- March 2020 (COVID-19 onset)
  - Hedged relative borrowing cost: dollar became about 100 basis points more expensive than euro on a hedged basis (hedged basis dropped by about 100 basis points before rebounding by June 2020).
  - Asymmetric moves in credit spreads: dollar credit spread spiked sharply relative to euro credit spread and remained elevated through mid-April.
  - Mechanism: sudden stop in dollar funding and a "dash for cash" drove sell-off in dollar-denominated securities.
- Global Financial Crisis parallels
  - Similar rise in relative dollar borrowing cost when dollar funding markets seized up and investors sold risky dollar bonds for safe assets.

### Comparison to benchmark CIP deviations (risk-free)
- Benchmark (risk-free) CIP deviations after GFC
  - Benchmark CIP deviations persistently positive for the dollar against the euro, indicating it has been expensive to swap euro into dollar in money markets (dollar funding shortage).
  - Measure: cross-currency basis swap as indicator of CIP deviation in risk-free rates.
- Contrast with hedged corporate basis
  - Hedged corporate basis has fluctuated between negative and positive and mostly held near parity.
  - Decoupling interpretation: CIP deviations in money markets likely related to bank balance sheet frictions; global firms can arbitrage borrowing differentials in international debt markets more effectively than banks can in money markets.
- Relationship expressed (stylized)
  - Hedged corporate basis = (euro credit spread) − (dollar credit spread) + Risk-free CIP deviation
  - In severe dollar funding distress, hedge-adjusted corporate basis sign depends on relative magnitudes of dollar and euro credit spreads and risk-free CIP deviations.

### Dollar safety premium: evidence for safe bonds
- Safety characteristics and relative hedged borrowing cost
  - High-rated and short-maturity corporate bonds exhibit a cheaper borrowing cost in the dollar (dollar safety premium).
- Rating-sorted findings (Figure 6)
  - AAA corporate basis peaked at about 50 basis points (dollar cheaper) during the GFC.
  - Lower-rated corporates (e.g., BBB) basis sank as low as -100 basis points.
  - Gap between AAA and BBB widened to about 150 basis points during the European sovereign debt crisis.
  - During onset of COVID-19:
    - Monthly data: BBB basis plunged to -75 basis points at end-March 2020; AAA basis fell to -50 basis points (daily data).
    - Daily data: BBB basis plunged as low as -175 basis points; AAA basis fell to -50 basis points.
- Maturity-sorted findings (Figure 7)
  - Longer tenors experienced much larger negative corporate basis during crises:
    - During the GFC, corporate basis for longer tenors sank as low as -225 basis points, while short maturities stayed roughly zero.
    - Post-GFC: basis for longer tenors persistently negative; basis for short maturities zero or positive.
    - Using daily data: 10+ year bonds basis sank as far as -225 basis points; 1–3 year bonds basis slid to about -75 basis points.
- Comparison with sovereign basis
  - AAA corporate basis is typically higher and more often positive than the broad corporate basis.
  - Sovereign basis (5-year bunds vs U.S. Treasuries) is typically positive, indicating a dollar safety premium for U.S. Treasuries relative to German bunds; AAA corporate basis and sovereign basis show similar patterns of cheaper dollar borrowing during periods of global stress.

### Issuance currency choice: issuer behavior and equilibrium
- Firm-level relation between relative borrowing cost and issuance share
  - Construct firm-level hedged and unhedged basis as differences between firm’s euro and dollar bond yields and CIP-adjusted credit spread.
  - Regress monthly dollar issuance share on firm-level hedged basis, unhedged basis, and lagged dollar issuance share (accounts for equilibrium where relative borrowing cost and issuance quantity interact).
- Regression findings (Table 5 summary)
  - Cheaper relative borrowing cost in the dollar corresponds to a higher dollar share of issuance.
  - Hedged basis effects:
    - Regression (1) with time fixed effects: a 1 basis point increase in the hedged basis is associated with a 0.91 percentage point increase in the dollar issuance share.
    - Regression (4) with firm fixed effects: a 1 basis point increase in the hedged basis is associated with a 0.50 percentage point increase in the dollar issuance share.
  - When controlling for both hedged and unhedged basis in regression (6), the content is truncated and further detail is not provided in the source excerpt.
- Interpretation
  - Issuers flexibly adjust the currency mix of debt at the margin in response to relative hedged and unhedged borrowing costs, supporting the finding that issuance quantities equilibrate marginal borrowing costs across currencies.

### Key quantitative results (preserved exactly)
- Aggregate and regression coefficients (Table 4 and Table 5 excerpts)
  - Euro -5.551 ∗ (Hedged cost)
  - Euro -102.7 ∗∗∗ (Unhedged cost)
  - 4-6 years 28.23 ∗∗∗ (Hedged cost); 62.81 ∗∗∗ (Unhedged cost)
  - 7-9 years 54.29 ∗∗∗ (Hedged cost); 140.3 ∗∗∗ (Unhedged cost)
  - 10+ years 87.38 ∗∗∗ (Hedged cost); 218.3 ∗∗∗ (Unhedged cost)
  - AA 42.83 ∗∗∗ (Hedged cost); 42.99 ∗∗∗ (Unhedged cost)
  - A 65.91 ∗∗∗ (Hedged cost); 65.97 ∗∗∗ (Unhedged cost)
  - BBB 94.01 ∗∗∗ (Hedged cost); 96.23 ∗∗∗ (Unhedged cost)
  - High yield 132.9 ∗∗∗ (Hedged cost); 135.5 ∗∗∗ (Unhedged cost)
  - Not rated 74.13 ∗∗∗ (Hedged cost); 78.50 ∗∗∗ (Unhedged cost)
  - N 135,210 (Hedged cost); 134,986 (Unhedged cost)
  - R2 0.864 (Hedged cost); 0.922 (Unhedged cost)
  - Hedged basis 0.091 ∗∗∗ (col 1, Table 5)
  - Hedged basis 0.053 ∗ (col 2)
  - Hedged basis 0.050 ∗∗ (col 3)
  - Hedged basis 0.086 ∗∗∗ (col 4)
  - Unhedged basis 0.060 ∗∗∗ (col 1)
  - Unhedged basis 0.033 ∗ (col 2)
  - Unhedged basis −0.007 (col 3)
  - Unhedged basis −0.022 ∗ (col 4)
  - USD share t−1 0.237 ∗∗∗ (cols 1–3); 0.094 ∗∗ (cols 4–6); 0.090 ∗∗ (col 6)
  - N 2,682 (col 1); 2,679 (col 2); 2,666 (col 3); 2,682 (col 4); 2,679 (col 5); 2,666 (col 6)
  - Adjusted R2 0.103 (cols 1–3); 0.206 (col 4); 0.202 (col 5); 0.209 (col 6)
  - Notes: Firm clustered standard errors in parentheses. ∗ p <0.1, ∗∗ p <0.05, ∗∗∗ p <0.01. Data winsorized on dependent variable with a 95% window each month.
- Additional numeric points and samples
  - Monthly bond dataset sample period: 2003 to 2020 at a monthly frequency.
  - Daily bond dataset sample period: January 2014 to September 2020 at a daily frequency.
  - COVID-19 episode analysis sample: January 2020 to June 2020 at a daily frequency.
  - Table 1 monthly sample: August 2003 to September 2020 at a monthly frequency; Table 2 country summary totals: Total 3,032.3 (Amt. issued, billion U.S. dollars), 3,452 (Bonds), 172 (Firms).
  - Numeric point noted in source title: 0.86 percentage points.

### Policy-relevant implications and practical conclusions
- For policymakers and regulators monitoring global dollar dominance:
  - While safe dollar assets retain a premium, FX-hedged corporate debt markets exhibit near-equalized hedged borrowing costs across dollar and euro at the margin due to large dollar issuance and active issuer arbitrage.
- For corporate treasury and risk-management teams:
  - Currency hedging is integral for many large global firms; issuance currency choice responds to hedged and unhedged basis signals as evidenced by positive coefficients on hedged basis in firm issuance regressions.
- For market analysts:
  - Deviations in cross-currency basis and hedged corporate borrowing costs provide actionable signals for likely shifts in issuance patterns by large global issuers.

*Source: Section excerpts from wpiea2021185-print-pdf (IMF Working Paper content provided in the input).*

### Section 3 discusses the overall result comparing the cost of borrowing in the dollar versus

### wpiea2021185-print-pdf - Section 3 discusses the overall result comparing the cost of borrowing in the dollar versus the euro for global firms outside of the two currency regions

### Methodology: measuring unhedged and hedged relative borrowing cost
- Unhedged relative borrowing cost
  - Use two panels of secondary market bond yield and price data for euro- and U.S. dollar-denominated bonds issued by firms located outside the U.S. and euro area.
  - Secondary market yields are used as a proxy for primary market issuance yields under the assumption that spreads between primary and secondary markets are small for large global firms and do not vary systematically by currency.
  - Cross-sectional regression (estimated each date t):
    - y_it = α_t 1_{EUR,i} + β_{f t} + γ_{m t} + δ_{r t} + ε_{it}
    - α_{EUR,t} (ˆα_t) is the euro minus dollar yield differential at time t controlling for firm fixed effects, maturity bucket and rating bucket. Data are winsorized on the dependent variable with a 95% window at each month.
- Hedged relative borrowing cost (corporate covered interest rate parity basis ψ)
  - Define ψ_t ≡ (y^e_t − y^$_t) + (f_t − s_t), where y^e_t and y^$_t are risky euro and dollar bond yields, and f_t and s_t are forward and spot exchange rates.
  - Decompose ψ_t as:
    - ψ_t = (y^e_t − r^e_t) + [r^e_t − r^$_t + (f_t − s_t)] − (y^$_t − r^$_t)
    - This separates euro credit spread, cross-currency basis (risk-free CIP deviation), and dollar credit spread.
  - Construct CIP-adjusted credit spread per bond:
    - S^{FXHedged}_{it} ≡ y_{it} − r^c_t + 1_{EUR,i}[r^e_t − r^$_t + (f_t − s_t)]
  - Regress S^{FXHedged}_{it} on same covariates as unhedged regression:
    - S^{FXHedged}_{it} = α_t 1_{EUR,i} + β_{f t} + γ_{m t} + δ_{l t} + ε_{it}
    - Estimated ˆα_t measures the currency-hedged relative borrowing cost ψ_t.

### Data: samples, coverage, and construction
- Monthly bond price dataset (August 2003 to September 2020)
  - Source: Bloomberg secondary market yields.
  - Sample criteria: euro- and U.S. dollar-denominated fixed- and zero-coupon non-callable corporate bonds with at least $50 million notional and at least one year to maturity; issuers outside the U.S. and euro area; issuer has a qualifying bond trading in both currencies at some point.
  - Exclude issuers whose country of risk is the United States or euro area and exclude supranationals.
  - Final monthly dataset: 3,452 bonds issued by 172 firms.
- Daily bond price dataset (January 2014 to September 2020)
  - Source: Bloomberg BVAL evaluated and quoted prices; yields estimated using duration and coupon.
  - Final daily dataset: 4,485 bonds issued by 190 firms.
- Dealogic issuance data (2000–2020)
  - Coverage: 82,185 euro- and dollar-denominated bonds (maturity at issuance > 1 year) issued by firms with nationality outside the U.S. and euro area.
  - Sum notional value: $12.4 trillion, with $8.8 trillion issued in the dollar and $3.6 trillion issued in the euro.
- Risk-free and cross-currency curves
  - Linearly interpolated to match bond maturities at each observation date.

### Relative borrowing cost: aggregate findings
- Dollar issuance share and magnitude
  - Since the introduction of the euro, about 75% of euro and dollar corporate issuance outside the U.S. and euro area has been denominated in dollars for much of the post-GFC period.
- Unhedged and hedged basis time-series (August 2003–September 2020)
  - Unhedged basis: widened to about -180 basis points between euro launch and GFC (negative indicating dollar cheaper), fell as far as -320 basis points after the crisis, rose in 2019 and tightened to about -100 basis points during onset of COVID-19.
  - Hedged corporate basis: generally held consistently close to zero, dipping in March 2020.
- Fama–MacBeth regression aggregated results (Table 4 summary)
  - Hedged borrowing costs between the two currencies are similar on average:
    - Euro coefficient is -5.5 basis points and not significant at the 5% level.
  - Unhedged borrowing cost difference favors the euro on average:
    - On average, the euro borrowing cost is 103 basis points less than the dollar borrowing cost.
- Interpretation
  - Similarity in hedged yields implies that larger dollar issuance is met with sufficient demand so marginal cost of issuance is similar when hedged.
  - Comparing bonds of the same firm in different currencies helps isolate currency denomination effects from country/sovereign effects.

### Impact of dollar shortages during COVID-19 and crisis episodes
- March 2020 (COVID-19 onset)
  - Hedged relative borrowing cost: dollar became about 100 basis points more expensive than euro on a hedged basis (hedged basis dropped by about 100 basis points before rebounding by June 2020).
  - Asymmetric moves in credit spreads: dollar credit spread spiked sharply relative to euro credit spread and remained elevated through mid-April.
  - Mechanism: sudden stop in dollar funding and a "dash for cash" drove sell-off in dollar-denominated securities.
- Global Financial Crisis parallels
  - Similar rise in relative dollar borrowing cost when dollar funding markets seized up and investors sold risky dollar bonds for safe assets.

### Comparison to benchmark CIP deviations (risk-free)
- Benchmark (risk-free) CIP deviations after GFC
  - Benchmark CIP deviations persistently positive for the dollar against the euro, indicating it has been expensive to swap euro into dollar in money markets (dollar funding shortage).
  - Measure: cross-currency basis swap as indicator of CIP deviation in risk-free rates.
- Contrast with hedged corporate basis
  - Hedged corporate basis has fluctuated between negative and positive and mostly held near parity.
  - Decoupling interpretation: CIP deviations in money markets likely related to bank balance sheet frictions; global firms can arbitrage borrowing differentials in international debt markets more effectively than banks can in money markets.
- Relationship expressed (stylized)
  - Hedged corporate basis = (euro credit spread) − (dollar credit spread) + Risk-free CIP deviation
  - In severe dollar funding distress, hedge-adjusted corporate basis sign depends on relative magnitudes of dollar and euro credit spreads and risk-free CIP deviations.

### Dollar safety premium: evidence for safe bonds
- Safety characteristics and relative hedged borrowing cost
  - High-rated and short-maturity corporate bonds exhibit a cheaper borrowing cost in the dollar (dollar safety premium).
- Rating-sorted findings (Figure 6)
  - AAA corporate basis peaked at about 50 basis points (dollar cheaper) during the GFC.
  - Lower-rated corporates (e.g., BBB) basis sank as low as -100 basis points.
  - Gap between AAA and BBB widened to about 150 basis points during the European sovereign debt crisis.
  - During onset of COVID-19:
    - Monthly data: BBB basis plunged to -75 basis points at end-March 2020; AAA basis fell to -50 basis points (daily data).
    - Daily data: BBB basis plunged as low as -175 basis points; AAA basis fell to -50 basis points.
- Maturity-sorted findings (Figure 7)
  - Longer tenors experienced much larger negative corporate basis during crises:
    - During the GFC, corporate basis for longer tenors sank as low as -225 basis points, while short maturities stayed roughly zero.
    - Post-GFC: basis for longer tenors persistently negative; basis for short maturities zero or positive.
    - Using daily data: 10+ year bonds basis sank as far as -225 basis points; 1–3 year bonds basis slid to about -75 basis points.
- Comparison with sovereign basis
  - AAA corporate basis is typically higher and more often positive than the broad corporate basis.
  - Sovereign basis (5-year bunds vs U.S. Treasuries) is typically positive, indicating a dollar safety premium for U.S. Treasuries relative to German bunds; AAA corporate basis and sovereign basis show similar patterns of cheaper dollar borrowing during periods of global stress.

### Issuance currency choice: issuer behavior and equilibrium
- Firm-level relation between relative borrowing cost and issuance share
  - Construct firm-level hedged and unhedged basis as differences between firm’s euro and dollar bond yields and CIP-adjusted credit spread.
  - Regress monthly dollar issuance share on firm-level hedged basis, unhedged basis, and lagged dollar issuance share (accounts for equilibrium where relative borrowing cost and issuance quantity interact).
- Regression findings (Table 5 summary)
  - Cheaper relative borrowing cost in the dollar corresponds to a higher dollar share of issuance.
  - Hedged basis effects:
    - Regression (1) with time fixed effects: a 1 basis point increase in the hedged basis is associated with a 0.91 percentage point increase in the dollar issuance share.
    - Regression (4) with firm fixed effects: a 1 basis point increase in the hedged basis is associated with a 0.50 percentage point increase in the dollar issuance share.
  - When controlling for both hedged and unhedged basis in regression (6), the content is truncated and further detail is not provided in the source excerpt.
- Interpretation
  - Issuers flexibly adjust the currency mix of debt at the margin in response to relative hedged and unhedged borrowing costs, supporting the finding that issuance quantities equilibrate marginal borrowing costs across currencies.

*Source: Section excerpts from wpiea2021185-print-pdf (IMF Working Paper content provided in the input).*

### 0.86 percentage points.

### wpiea2021185-print-pdf - 0.86 percentage points.

### Key empirical findings
- The hedged and unhedged relative borrowing cost between euro- and dollar-denominated bonds is analyzed; a positive value indicates that euro-denominated bonds have a higher borrowing cost.
- After the cost of exchange rate hedges, the costs of comparable corporate debt in the dollar and the euro are similar.
- A dollar premium persists in a subset of bonds with the highest credit quality and shortest maturities, resembling safe sovereign bonds.
- Global corporate borrowers adjust issuance currency, switching between euro and dollar issuance in response to relative issuance costs, effectively arbitraging price fluctuations toward rough parity.
- Not all global firms hedge currency risk using financial instruments; some have natural hedges (example: under the petrodollar system, oil is priced almost exclusively in the U.S. dollar, so global oil firms receive most revenue in dollars and need not hedge dollar debt).
- Numeric point noted in source title: 0.86 percentage points.

### Conclusion and interpretation
- The dominance of the U.S. dollar and the "exorbitant privilege" (lower borrowing costs for relatively "safe" dollar-denominated assets such as U.S. Treasuries and AAA-rated corporates) remains at the margin for the safest assets.
- For FX-hedged corporate debt excluding sovereign and "quasi-sovereign" debt, the dollar advantage appears exhausted at the margin: aggregate demand for risky dollar debt lies to the right of demand for risky euro debt, but the much larger size of dollar issuance equalizes marginal prices.
- Large issuers in the global corporate debt market actively arbitrage deviations from rough price parity between hedged dollar and euro borrowing costs.

### Data, samples, and timing
- Monthly bond dataset sample period: 2003 to 2020 at a monthly frequency (monthly bond dataset).
- Daily bond dataset sample period: January 2014 to September 2020 at a daily frequency (Bloomberg BVAL).
- COVID-19 episode analysis sample: January 2020 to June 2020 at a daily frequency (daily bond dataset).
- Table 1 monthly sample: August 2003 to September 2020 at a monthly frequency; summary statistics reported include notional amount ($ billion), number of bonds per ultimate parent, maturity at issuance (years); data sourced from Bloomberg.
- Table 2 country summary totals: Total 3,032.3 (Amt. issued, billion U.S. dollars), 3,452 (Bonds), 172 (Firms).

### Selected quantitative results (preserved exactly from source)
- Table 4 (Fama-Macbeth Regression of Borrowing Costs and Returns), selected coefficients and statistics as reported:
  - Euro -5.551 ∗ (Hedged cost)
  - Euro -102.7 ∗∗∗ (Unhedged cost)
  - 4-6 years 28.23 ∗∗∗ (Hedged cost); 62.81 ∗∗∗ (Unhedged cost)
  - 7-9 years 54.29 ∗∗∗ (Hedged cost); 140.3 ∗∗∗ (Unhedged cost)
  - 10+ years 87.38 ∗∗∗ (Hedged cost); 218.3 ∗∗∗ (Unhedged cost)
  - AA 42.83 ∗∗∗ (Hedged cost); 42.99 ∗∗∗ (Unhedged cost)
  - A 65.91 ∗∗∗ (Hedged cost); 65.97 ∗∗∗ (Unhedged cost)
  - BBB 94.01 ∗∗∗ (Hedged cost); 96.23 ∗∗∗ (Unhedged cost)
  - High yield 132.9 ∗∗∗ (Hedged cost); 135.5 ∗∗∗ (Unhedged cost)
  - Not rated 74.13 ∗∗∗ (Hedged cost); 78.50 ∗∗∗ (Unhedged cost)
  - N 135,210 (Hedged cost); 134,986 (Unhedged cost)
  - R2 0.864 (Hedged cost); 0.922 (Unhedged cost)
  - Notes: Firm clustered standard errors in parentheses. ∗ p <0.1, ∗∗ p <0.05, ∗∗∗ p <0.01. Data winsorized on dependent variable with a 95% window each month.
- Table 5 (Firm-level issuance flows), selected coefficients and statistics as reported:
  - Hedged basis 0.091 ∗∗∗ (col 1)
  - Hedged basis 0.053 ∗ (col 2)
  - Hedged basis 0.050 ∗∗ (col 3)
  - Hedged basis 0.086 ∗∗∗ (col 4)
  - Unhedged basis 0.060 ∗∗∗ (col 1)
  - Unhedged basis 0.033 ∗ (col 2)
  - Unhedged basis −0.007 (col 3)
  - Unhedged basis −0.022 ∗ (col 4)
  - USD share t−1 0.237 ∗∗∗ (cols 1–3); 0.094 ∗∗ (cols 4–6); 0.090 ∗∗ (col 6)
  - N 2,682 (col 1); 2,679 (col 2); 2,666 (col 3); 2,682 (col 4); 2,679 (col 5); 2,666 (col 6)
  - Adjusted R2 0.103 (cols 1–3); 0.206 (col 4); 0.202 (col 5); 0.209 (col 6)
  - Notes: Clustered standard errors in parentheses. *p <0.1, **p <0.05, ***p <0.01. Sample: 2003 to 2020 at a monthly frequency; hedged and unhedged corporate bases calculated using the monthly bond dataset; dollar share of issuance calculated using data sourced from Dealogic.

### Figures and descriptive patterns (as presented)
- Figure 3: EUR-USD relative borrowing cost — hedged and unhedged relative borrowing cost, sample 2003–2020 monthly.
- Figure 4: Credit markets during COVID-19 — 4a hedged and unhedged relative borrowing cost during COVID-19; 4b median firm-level median credit spread for euro- and dollar-denominated bonds (January 2020–June 2020 daily).
- Figure 5: Hedged relative borrowing cost compared to the 5-year EUR-USD LIBOR cross-currency basis (mid price of cross-currency basis swap, Bloomberg).
- Figure 6: Hedged relative borrowing cost by rating (2003–2020 monthly).
- Figure 7: Hedged relative borrowing cost by maturity (2003–2020 monthly).
- Table summaries report medians, means, and SDs for notional amount ($ billion), maturity (years), N per firm, and counts by currency and country.

### Policy-relevant implications
- For policymakers and regulators monitoring global dollar dominance: while safe dollar assets retain a premium, FX-hedged corporate debt markets exhibit near-equalized hedged borrowing costs across dollar and euro at the margin due to large dollar issuance and active issuer arbitrage.
- For corporate treasury and risk-management teams: currency hedging is integral for many large global firms; issuance currency choice responds to hedged and unhedged basis signals as evidenced by positive coefficients on hedged basis in firm issuance regressions.
- For market analysts: deviations in cross-currency basis and hedged corporate borrowing costs provide actionable signals for likely shifts in issuance patterns by large global issuers.

*Source: wpiea2021185-print-pdf (excerpted content provided).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021185-print-pdf.pdf_