## The Dominant Currency Financing Channel of External Adjustment

## Source details

**Canonical URL:** [The Dominant Currency Financing Channel of External Adjustment](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023164-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023164-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023164-print-pdf.pdf.json)

---

### Empirical focus and sample
- Analysis focuses only on firms in the sample that ever imported.
- All measures of foreign currency exposure entering the regressions were standardized to have a mean of zero and have a unit standard deviation.

### Key measured shocks and summary statistics
- Average wealth shock (liquidity shock plus debt revaluation after 2015:Q3) = 0.8% of total assets.
- Some firms have negative values of the wealth shock.
- Weighted average original loan maturity ≈ five years (distribution shown in referenced Figure A2).

### Baseline specification (identification strategy)
- Natural experiment: unanticipated depreciation of the peso in 2014:Q3 to study the debtor currency funding (DCF) channel.
- Baseline estimating equation:
  - ln(1 + Y_ft) = β × FCE_f × 1(t ≥ t0) + controls_ft + ε_ft
  - Y = firm-level exports or imports of firm f at time t.
  - FCE_f ∈ {foreign currency leverage (FCL_f t in t = 2014:Q1), liquidity shock (LS_f t,t′ with t = 2014:Q1 and t′ = 2015:Q3), wealth shock (WS_f t,t′ with t = 2014:Q1 and t′ = 2015:Q3)}.
  - 1(t ≥ t0) = dummy equal to one in any quarter during or after the depreciation (t ≥ 2014:Q3).
- Standard errors double-clustered at firm and quarter level.
- Controls include firm and time fixed effects.
- β interpreted as differential effect of financial exposure to exchange rate shocks on imports (exports).

### Concerns about endogeneity and robustness checks
- Foreign currency debt and maturity structure are endogenous choices; addressed by exploiting an unanticipated, sudden depreciation driven by an oil price drop.
- Re-estimated regressions using inverse hyperbolic sine transformation of Y; results almost qualitatively and quantitatively identical to log specification (note: inverse hyperbolic sine ≈ log(2Y) for non-small Y).

### Exogeneity of the liquidity shock (high-frequency identification)
- Liquidity shock constructed exploiting firms’ payment dates during the depreciation and daily exchange rate differences (use of exact maturity date increases exogeneity).
- No statistically significant linear relationship between firms’ average loan maturity before depreciation and liquidity shock (Table A3 referenced).
- High-frequency exchange rate volatility can generate substantial variation in peso repayment amounts even with small differences in loan issue dates (illustrative example: one firm repays 2.413 billion pesos, another 2.294 billion pesos (5% less) due solely to timing).

### Identification, data, and robustness (additional strategies)
- Regression controls for loan maturity and uses day-of-week of loan issuance as exogenous variation.
- Tests show firms with larger liquidity shocks are balanced on observables (average maturity, profitability, age).
- Baseline coefficients remain stable after controlling for observables interacted with post dummy and sector-time fixed effects (Tables A3–A4 referenced).
- Placebo: replacing foreign currency leverage with domestic leverage yields no significant negative effects on imports or exports.

### Baseline empirical results on imports and exports
- Sample sizes:
  - All firms: 21,850.
  - Exporting firms: around 7,200.
- One standard deviation in foreign currency leverage = 3.2% foreign currency leverage.
- Imports:
  - One standard deviation increase in share of foreign currency debt → additional 10% decline in imports (all firms).
  - Non-exporters: effect roughly doubles → 20% stronger decline in imports for a one standard deviation increase.
  - Firms that both import and export: effect ≈ 1.8% and not statistically significant.
- Decomposition:
  - Liquidity shock: one standard deviation increase → ≈ 5% decline in imports (all firms).
    - Non-exporters: ≈ 7% decline (statistically significant).
    - Exporters: 2.5% decline (not statistically significant).
  - Wealth shock (includes revaluation of later-maturing debt): one standard deviation increase → non-exporters associated with 12% larger compression of imports; exporters ≈ 1% (not statistically significant).
- Exports:
  - Coefficients mostly negative but much smaller and not statistically significant.
  - Higher foreign currency leverage, larger wealth or liquidity shocks are not associated with a larger drop in exports.
  - Dominant currency revenues (exports priced in dollars) provide a hedge for exporters.

### Dynamics and panel evidence
- Event-study dynamics:
  - No differential pre-trends from 2012:Q3 to 2014:Q2.
  - After 2014:Q3 depreciation, non-exporters with larger financial exposure accumulate import compression over time; three years after depreciation effect reaches coefficient -0.4 (implying a 40% larger contraction for a one standard deviation larger exposure).
  - Exporters’ imports remain close to zero pre- and post-depreciation; estimates not statistically significant.
- Annual panel (2008–2018) confirms non-exporters shrink imports significantly in the year following a depreciation when they had higher lagged foreign currency leverage; exporters do not show the same response.

### Hedging and heterogeneity
- Hedging statistics (2013):
  - Hedged via dollar assets or derivatives: ≈ 13% of all firms.
  - Firms with FX derivative outstanding in 2013: 2.9% of all firms.
  - Firms with outstanding foreign-currency assets: 11.6% of all firms.
  - Hedging by export status: almost 30% of exporting firms hedged versus 5% of non-exporters.
- Hedging regressions (triple interaction FCE × Hedge × post):
  - Triple interaction consistently positive and offsets negative effect of foreign currency borrowing on imports.
  - Firms hedging via foreign currency assets or derivatives do not exhibit the import contraction associated with DCF; cushioning effect largest for non-exporters.
  - For exporters, additional hedging has an insignificant positive effect because exporters are naturally hedged via dollar export revenues.

### Lending, interest rates, and delinquencies (financial friction mechanism)
- Borrowing composition:
  - Average firm with non-zero foreign currency borrowing had roughly 20% of its debt denominated in foreign currency.
  - Such a firm would see a 3% decline in total borrowing after the shock when accounting for substitution into domestic currency loans.
- Loan-level interest rates:
  - Firms with larger financial exposure face higher interest rates after depreciation, driven entirely by non-exporting firms.
  - For exporters, interest-rate responses are statistically insignificant (sometimes negative).
- Delinquencies:
  - Firms with larger financial exposure more likely to fall behind on loan payments after depreciation; effect driven by non-exporters.
  - Quantitatively, a one standard deviation increase in foreign currency leverage for non-exporters → increase of approximately 60 basis points in delinquencies, around a 7% increase.

### Quantification and aggregate contribution
- Counterfactual exercise using Table 1 estimates:
  - If no firms had foreign-currency borrowing, peak-to-trough import contraction would have been around 17% smaller.
- Model-based calibration:
  - Calibrated partial-equilibrium model (lognormal demand, parameters in Table A13) yields DCF channel contribution = 13.7% import compression for non-exporters.
  - Reduced-form back-of-the-envelope suggests DCF accounts for around 17% of import compression (consistent with calibrated value).
- Net worth shock parameter in calibration: 15%.

### Theoretical model (parsimonious model with financial frictions)
- One-period model: firms start with net worth A = ad + af e; firms borrow B to finance imported input pM M = A + B.
- Verification cost parameter γ, bank funding return R, revenue function ρ(M/pM, φ)δ with monotonicity and concavity assumptions.
- Key implications:
  - Implication 1: Depreciation decreases imports more for firms that borrow more in foreign currency (consistent with Table 1).
  - Implication 2: Depreciation increases interest rates more for firms that borrow more in foreign currency (consistent with Table 10).
  - Implication 3: Depreciation does not decrease imports more for exporting firms that borrow more in foreign currency (consistent with Table 2).
  - Implication 4: Depreciation does not increase interest rates more for exporting firms that borrow more in foreign currency (consistent with Table 10).
- Pricing currency margin:
  - Dominant currency pricing (DCP, dollars) mutes DCF effects because export revenues rise in domestic currency and act as hedge.
  - Producer currency pricing (PCP, pesos) would lead to stronger contraction in imports; calibration shows DCP exporters insulated while PCP exporters would be affected by net worth shocks.

### Calibration and key parameters (reported exactly)
- σ = 5
- γ = 0.18
- R = 1.03
- Import share = 64%
- F = 1.1
- δ ~ LN(ν, −ν^2 / 2), ν = 3
- Nominal interest rate of 10.5%
- d_e = 40%
- A = 0.12
- d_A = 15%
- Contribution of Dominant Currency Financing Channel:
  - Exporters 0%
  - Non-exporters 13.7%

### Policy-relevant takeaways and macro implications
- Dominant Currency Financing (DCF) amplifies import contraction after a depreciation via increased debt revaluation, higher interest rates, and increased delinquencies, predominantly for non-exporting firms.
- Exporters are largely insulated due to dollar-priced export revenues (natural hedge) and greater use of derivatives and foreign assets.
- Hedging instruments (foreign assets or FX derivatives) materially cushion the DCF effect, especially for non-exporters.
- DCF channel contributes materially (empirical ≈ 17%, calibrated ≈ 13.7% for non-exporters) to import compression during a depreciation, alongside expenditure switching and valuation effects; implications for net foreign asset stability and policy design addressing currency mismatches and corporate hedging incentives.

*Source: wpiea2023164-print-pdf - section 5.*

### 2018. Their value in pesos increased annually throughout the period.

### wpiea2023164-print-pdf - 2018. Their value in pesos increased annually throughout the period.

### Empirical focus and sample
- Because import regressions are the main focus of the empirical exercise, the analysis focuses only on firms in the sample that ever imported.
- All measures of foreign currency exposure entering the regressions were standardized to have a mean of zero and have a unit standard deviation.

### Key measured shocks and summary statistics
- The average wealth shock, which equals the liquidity shock plus the debt revaluation after 2015:Q3, equals 0.8% of total assets.
- Some firms have negative values of the wealth shock (these firms benefited from the exchange rate movements before the depreciation started).
- Figure A2 (referenced) shows the distribution of maturity, with the weighted average original maturity at around five years.

### Baseline specification (identification strategy)
- The unanticipated depreciation of the peso in 2014:Q3 is used as a natural experiment to study the debtor currency funding (DCF) channel of a foreign exchange rate depreciation on trade.
- Baseline estimating equation:
  - ln(1 + Y_ft) = β × FCE_f × 1(t ≥ t0) + controls_ft + ε_ft
  - Y denotes either firm-level exports or imports of firm f at time t.
  - FCE_f is one of: foreign currency leverage (FCL_f t in t = 2014:Q1), liquidity shock (LS_f t,t′ with t = 2014:Q1 and t′ = 2015:Q3), or wealth shock (WS_f t,t′ with t = 2014:Q1 and t′ = 2015:Q3).
  - 1(t ≥ t0) is a dummy that equals one in any quarter during or after the depreciation (t ≥ 2014:Q3).
- Standard errors are double-clustered at the firm and quarter level.
- Controls include firm and time fixed effects:
  - β is interpreted as the differential effect of financial exposure to exchange rate shocks on imports (exports) between firms with different levels of foreign currency exposure.
  - Time fixed effects absorb the level effect on imports (exports) of all firms.
  - Firm fixed effects control for time-invariant firm characteristics (e.g., average reliance on imported inputs, size).

### Concerns about endogeneity and robustness checks
- DCF is a choice for firms; foreign currency debt and its maturity structure can respond to expectations regarding exchange rate movements, which may bias results.
- The study addresses endogeneity by studying an unanticipated, sudden depreciation unrelated to the Colombian economy (driven by a drop in the oil price).
- The logarithmic transformation of the dependent variable can introduce biases through the constant; regressions are re-estimated with an inverse hyperbolic sine transformation of Y and find almost qualitatively and quantitatively identical results.
  - Note: Except for small values of Y, the inverse hyperbolic sine transformation approximately equals log(2Y), or log(2) + log(Y).

### Exogeneity of the liquidity shock (identifying the causal impact)
- Construction of the liquidity shock exploits firms’ payment dates during the depreciation period and differences in the daily exchange rate to identify the mechanism.
- The approach uses high-frequency identification by using the date of maturity (rather than year or quarter), making exposure to exchange rate movements more exogenous.
- Comparison with other studies:
  - Almeida et al. (2012) and Duval et al. (2020) use the Global Financial Crisis and year/quarter-level identification, which may be confounded by other crisis-related factors (e.g., increased uncertainty).
  - The present approach argues that because the exchange rate depreciation was driven by an oil price drop and not by Colombian-specific factors, effects can more plausibly be attributed to the exchange rate.
- Cross-sectional regression results (Table A3, referenced) show no statistically significant relationship between the average maturity of firms’ loans before the depreciation and the liquidity shock in a linear regression.
  - While there is a constructed nonlinear relationship between maturity and the liquidity shock, the analysis rules out that firms that borrow shorter (on average) are also systematically faced with a larger liquidity shock.
- High exchange rate volatility during the depreciation can explain the absence of correlation between average maturity and the liquidity shock; an illustrative example is presented:
  - Two firms both need to repay $1 million on the day the loan matures; one borrowed on the 16th of December, 2009, the other on the 21st of December, 2009—only the day of the week differs, implying substantial variation from high-frequency exchange rate movements.

*Source: wpiea2023164-print-pdf - 2018. Their value in pesos increased annually throughout the period.*

### 2.413 billion pesos, while the other firm needs to pay back 2.294 billion pesos (5% less).  This example

### wpiea2023164-print-pdf - 2.413 billion pesos, while the other firm needs to pay back 2.294 billion pesos (5% less).  This example

### Identification, data, and robustness
- Identification exploits quasi-exogenous variation in timing of loan maturities (example values: 2.413 billion pesos versus 2.294 billion pesos (5% less)) to construct a liquidity shock driven solely by the depreciation and by the exact day the loan was issued.
- Regression strategy:
  - Controls for loan maturity and uses day-of-week of loan issuance as exogenous variation.
  - Tests for balance in pre-existing firm characteristics (average maturity, profitability, age) and finds firms with larger liquidity shocks are balanced on observables.
  - Controls for these observables interacted with the post dummy and sector-time fixed effects; baseline coefficients remain extremely stable (see referenced Tables A3–A4).
- Placebo: replacing foreign currency leverage with domestic leverage yields no significant negative effects on imports or exports, supporting the foreign-currency-specific channel.

### Baseline empirical results on imports and exports
- Sample sizes and measures:
  - All firms: 21,850.
  - Exporting firms: around 7,200.
  - One standard deviation in foreign currency leverage = 3.2% foreign currency leverage.
- Imports:
  - A one standard deviation increase in share of foreign currency debt is associated with an additional 10% decline in imports (all firms).
  - Non-exporters: effect roughly doubles → 20% stronger decline in imports for a one standard deviation increase.
  - Firms that both import and export: effect ~1.8% and not statistically significant.
- Decomposition into liquidity shock and wealth shock:
  - Liquidity shock: one standard deviation increase → around 5% decline in imports (all firms).
    - Driven by non-exporters: ~7% decline and statistically significant.
    - Exporters: 2.5% decline but not statistically significant.
  - Wealth shock (includes revaluation of debt maturing later): one standard deviation increase → for non-exporters associated with a 12% larger compression of imports; for exporters ~1% and not statistically significant.
- Exports:
  - Coefficients mostly negative but much smaller and not statistically significant.
  - Neither higher foreign currency leverage nor larger wealth or liquidity shocks are associated with a larger drop in exports.
  - Dominant currency revenues (exports priced in dollars) provide a hedge for exporters against the negative financial effects of depreciation.

### Dynamics and panel evidence
- Event-study dynamics:
  - No differential pre-trends between treated and control firms from 2012:Q3 to 2014:Q2.
  - After depreciation starts in 2014:Q3, non-exporters with larger financial exposure accumulate import compression over time; three years after the depreciation the effect reaches a coefficient of -0.4 (implying a 40% larger contraction for a one standard deviation larger exposure).
  - Exporters’ imports remain close to zero both before and after depreciation; estimates not statistically significant.
- Annual panel (2008–2018):
  - Confirms that non-exporters shrink imports significantly in the year following a depreciation when they had higher lagged foreign currency leverage; exporters do not show the same response.

### Hedging and heterogeneity
- Hedging statistics (2013):
  - Around 13% of all firms were hedged either through dollar assets or via derivative positions.
  - 2.9% of all firms had a foreign exchange derivative contract outstanding in 2013.
  - 11.6% of all firms have outstanding assets in foreign currency.
  - Hedging differs by export status: almost 30% of exporting firms hedged, versus 5% of non-exporters.
- Hedging regressions (Equation 6 specification):
  - Triple interaction (FCE × Hedge × post) is consistently positive and offsets the negative effect of foreign currency borrowing on imports.
  - Firms that hedge through foreign currency assets or through derivatives do not exhibit the import contraction associated with DCF; the cushioning effect is largest for non-exporters.
  - For exporters, additional hedging (derivatives or foreign assets) has an insignificant positive effect because exporters are already naturally hedged via dollar export revenues.

### Lending, interest rates, and delinquencies (financial friction mechanism)
- Borrowing composition and aggregate effect:
  - Average firm with non-zero foreign currency borrowing had roughly 20% of its debt denominated in foreign currency.
  - Such a firm would see a 3% decline in total borrowing after the shock when accounting for substitution into domestic currency loans.
- Loan-level interest rates (Equation 7):
  - Firms with larger financial exposure face higher interest rates after the depreciation, driven entirely by non-exporting firms.
  - For exporters, coefficients on interest-rate responses are statistically insignificant (sometimes negative).
- Delinquencies (Equation 8):
  - Firms with larger financial exposure are more likely to fall behind on loan payments after the depreciation; effect driven by non-exporters.
  - Quantitatively, a one standard deviation increase in foreign currency leverage for non-exporters is associated with an increase of approximately 60 basis points in delinquencies, around a 7% increase.

### Quantification and aggregate contribution
- Counterfactual exercise using Table 1 estimates:
  - If no firms had foreign-currency borrowing, the peak-to-trough import contraction would have been around 17% smaller.
- Model-based calibration and contribution:
  - Calibrated partial-equilibrium model (lognormal demand, parameters in Table A13):
    - DCF channel contributes 13.7% import compression for non-exporters in the calibrated model.
    - Reduced-form back-of-the-envelope suggests DCF accounts for around 17% of the import compression (consistent with calibrated value).
  - Net worth shock parameter in calibration: 15%.

### Theoretical model (parsimonious model with financial frictions)
- Framework:
  - One-period model where firms start with net worth A = ad + af e; firms borrow B to finance imported input pM M = A + B.
  - Verification cost parameter γ, bank funding return R, revenue function ρ(M/pM, φ)δ with usual monotonicity and concavity assumptions.
- Key implications derived:
  - Implication 1: A depreciation decreases imports more for firms that borrow more in foreign currency (consistent with Table 1).
  - Implication 2: A depreciation increases interest rates more for firms that borrow more in foreign currency (consistent with Table 10).
  - Implication 3: A depreciation does not decrease imports more for exporting firms that borrow more in foreign currency (consistent with Table 2).
  - Implication 4: A depreciation does not increase interest rates more for exporting firms that borrow more in foreign currency (consistent with Table 10).
- Pricing currency margin:
  - When exporters price in dominant currency (DCP, i.e., dollars), DCF effects are muted because export revenues rise in domestic currency and serve as a natural hedge.
  - If exporters priced in producer currency (PCP, pesos), the model predicts stronger contraction in imports due to lack of hedging; calibration shows DCP exporters are insulated while PCP exporters would be affected by net worth shocks.

### Policy-relevant takeaways and macro implications
- Dominant Currency Financing (DCF) is a notable financial channel of external adjustment:
  - DCF amplifies import contraction after a depreciation via increased debt revaluation, higher interest rates, and increased delinquencies, predominantly for non-exporting firms.
  - Exporters are largely insulated due to dollar-priced export revenues (natural hedge) and due to greater usage of derivative markets and foreign assets.
- Hedging instruments (foreign assets or FX derivatives) materially cushion the DCF effect, especially for non-exporters who lack natural hedges.
- Macro and stability implications:
  - The DCF channel contributes materially (empirical ~17%, calibrated ~13.7% for non-exporters) to import compression during a depreciation, in addition to expenditure switching and valuation effects.
  - Results have consequences for the stability of countries’ net foreign asset positions and for the design of policies addressing currency mismatches and corporate hedging incentives.

*Italic line: Source: IMF Working Paper (wpiea2023164-print-pdf).*

### References

### wpiea2023164-print-pdf - References

### References and cited literature (selected)
- Adler, Gustavo, Camila Casas, Luis Cubeddu, Gita Gopinath, Nan Li, Sergii Meleshchuk, Carolina Osorio-Buitron, Damien Puy, and Yannick Timmer (2020) “Dominant currencies and external adjustment”, IMF Staff Discussion Note. 6
- Amiti, Mary, Oleg Itskhoki, and Jozef Konings (2014) “Importers, exporters, and exchange rate disconnect”, American Economic Review, 104 (7), pp. 1942–1978. 14, 29
- Gopinath, Gita, Emine Boz, Camila Casas, Federico J Diez, Pierre-Olivier Gourinchas, and M Plagborgmøller (2020) “Dominant currency paradigm”, American Economic Review, 110 (3), pp. 677–719. 6
- Bernanke, Ben S, Mark Gertler, and Simon Gilchrist (1999) “The financial accelerator in a quantitative business cycle framework”, Handbook of macroeconomics, 1, pp. 1341–1393. 7, 24, 73, 74, 76
- Casas, Camila, Federico J Díez, Gita Gopinath, and Pierre-Olivier Gourinchas (2016) “Dominant currency paradigm”, Technical report, National Bureau of Economic Research. 6, 17, 29
- Amiti, Mary and David E Weinstein (2011) “Exports and financial shocks”, Quarterly Journal of Economics, 126 (4), pp. 1841–1877. 7, 17
- Korinek, Anton (2011) “Excessive dollar borrowing in emerging markets: Balance sheet effects and macroeconomic externalities”, Working Paper. 7

(References list in the source continues; above are selected entries reproduced exactly as in the content.)

### Empirical findings — key table estimates and statistics
- Table 1: Imports (ln(imports), quarterly)
  - Post × FCL = -0.106 ∗∗∗ (column (1), SE (0.025))
  - Post × FCL = -0.215 ∗∗∗ (column (2), SE (0.047))
  - Post × FCL = -0.018 (column (3), SE (0.023))
  - Post × LS = -0.046 ∗∗ (column (1), SE (0.020)); Post × LS = -0.069 ∗ (column (2), SE (0.034))
  - Post × WS = -0.059 ∗∗∗ (column (1), SE (0.020)); Post × WS = -0.120 ∗∗∗ (column (2), SE (0.032))
  - Sample sizes: N = 524,943; 350,934; 174,009 (various columns)
  - Notes: FCL, WS, and LS normalized to mean 0 and standard deviation 1. Post = 1 for period after 2014:Q2.

- Table 2: Exports (ln(exports), quarterly)
  - Post × FCL = -0.010 (SE (0.025))
  - Post × LS = -0.025 (SE (0.023))
  - Post × WS = -0.024 (SE (0.026))
  - N = 169,232

- Table 3: Placebo Domestic Leverage
  - Post × Domestic Leverage effects on ln(imports) and ln(exports) are small and statistically insignificant:
    - e.g., Post × Domestic Leverage = -0.006 (SE (0.004)) for ln(imports) (column (1); N = 523,719)

- Table 4: Panel Regression (annual)
  - FCL_{i,t-1} coefficients:
    - Column (1) ln(imports): 5.008 ∗∗∗ (SE (0.386))
    - Column (2) non-exporters ln(imports): 6.715 ∗∗∗ (SE (0.651))
    - Column (3) exporters ln(imports): 3.586 ∗∗∗ (SE (0.443))
    - Column (4) ln(exports): 2.098 ∗∗∗ (SE (0.329))
  - Interaction FCL_{i,t-1} × ∆ER_{t-1}:
    - Column (2): -5.044 ∗∗ (SE (2.225))

- Table 5: Descriptive Statistics, Hedging (2013)
  - Share of firms (2013) with foreign assets: Non-exporters 4.0%; Exporters 26.9%; All firms 11.6%
  - Hedge FX Derivatives: Non-exporters 1.6%; Exporters 5.6%; All firms 2.9%
  - Hedge (Foreign Assets or FX Derivatives): Non-exporters 5.1%; Exporters 28.7%; All firms 12.9%
  - Number of firms: Non-exporters 14,618; Exporters 7,232; All firms 21,850

- Hedging interaction results (Table 6, Table 7, Table 8 — quarterly ln(imports))
  - Table 6 (Hedge = foreign assets OR FX derivatives):
    - Post × FCL = -0.150 ∗∗∗ (column (1), SE (0.045))
    - Post × FCL × Hedge = 0.140 ∗∗ (column (1), SE (0.051)) — hedged firms attenuate the negative Post×FCL effect
  - Table 7 (Hedge = foreign currency assets):
    - Post × FCL = -0.142 ∗∗∗ (column (1), SE (0.039))
    - Post × FCL × Hedge = 0.126 ∗∗ (column (1), SE (0.047))
  - Table 8 (Hedge = foreign exchange derivative positions):
    - Post × FCL = -0.124 ∗∗∗ (column (1), SE (0.033))
    - Post × FCL × Hedge = 0.127 ∗ (column (1), SE (0.062))

- Table 9: Borrowing (quarterly, ln(FC borrowing) and ln(LC borrowing))
  - Post × FCL on foreign-currency borrowing: -0.587 ∗∗∗ (SE (0.117))
  - Post × FCL on local-currency borrowing: 0.104 ∗∗∗ (SE (0.026))
  - Post × LS on foreign-currency borrowing: -0.207 ∗∗∗ (SE (0.059))
  - Post × WS on foreign-currency borrowing: -0.391 ∗∗∗ (SE (0.097))
  - Firm FE and Time FE included; N = 524,943 (columns)

- Table 10: Interest Rate (bank-day to firm-day interest rates; bank×time FE used)
  - FCL × Post = 0.040 (column (1), SE (0.028)); in some columns 0.134 ∗∗∗ (SE (0.049))
  - LS × Post = 0.052 ∗ (SE (0.029)) or 0.074 ∗ (SE (0.041)) in alternate samples
  - Sample sizes: N = 364,928; 220,732; 143,774 (various columns)

- Table 11: Delinquency (Delinquent dummy; quarterly)
  - Post × FCL = 0.002 ∗∗∗ (column (1), SE (0.001))
  - Post × FCL = 0.006 ∗∗∗ (column (2), SE (0.002))
  - Post × LS and Post × WS also positive and significant in some columns
  - Large panel sizes: N = 3,635,780; 2,165,253; 1,470,378 (various columns)

### Counterfactuals and graphical results (figures, described)
- Figure 4 and Figure 5 present counterfactual series computed assuming firms in 2014:Q2 had zero foreign currency leverage:
  - Figure 4: actual (thick blue line) and counterfactual (dashed red line) values of Colombia’s imports.
  - Figure 5: actual (thick blue line) and counterfactual (dashed red line) values of Colombia’s net exports.
  - See subsection 4.6 in source for counterfactual exercise details.

- Figure 1–3: Estimated impact plots (point estimates with 95% and 99% confidence bands) of foreign currency leverage on:
  - Non-exporters’ imports (Figure 1)
  - Exporters’ imports (Figure 2)
  - Exports (Figure 3)
  - Notes: thin blue dotted line represents average peso-dollar exchange rate (right axis). Standard errors clustered at firm and quarter levels.

- Figure 6: Model comparison of dominant currency pricing (DCP) vs producer currency pricing (PCP) on imports response to net worth shock.

### Appendix: Additional tables and robustness (selected numeric highlights)
- Table A1: Descriptive Statistics, Firm Characteristics (all variables in thousands of dollars)
  - Assets mean: Non-exporters 8,810 (S.D. 154,309); Exporters 19,655 (S.D. 208,475); All firms 12,400 (S.D. 174,183)
  - Number of firms: Non-exporters 14,618; Exporters 7,232; All firms 21,850

- Table A2: Independent Variables percentiles
  - Foreign leverage mean 4.4%; percentiles: 10th 0.2; 25th 0.4; 50th 2.0; 75th 6.3; 90th 11.6%
  - Liquidity shock mean 0.1%; percentiles: 10th -0.2; 25th 0.0; 50th 0.0; 75th 0.0; 90th 0.6%
  - Wealth shock mean 0.8%; percentiles: 10th -0.1; 25th 0.0; 50th 0.2; 75th 0.9; 90th 2.6%

- Table A5: Imports: Quantities
  - Post × FCL (Units) = -0.338 ∗∗∗ (SE (0.073))
  - Post × FCL (Kilograms) = -0.317 ∗∗∗ (SE (0.074))

- Table A6: Extensive margin (unique countries and products)
  - Post × FCL on #Countries = -0.038 ∗∗∗ (SE (0.009))
  - Post × FCL on #Products = -0.075 ∗∗∗ (SE (0.017))

- Table A11: Investment (cross-sectional change in investment)
  - FCL coefficients on ∆Investment: e.g., Column (2) -0.101 ∗∗ (SE (0.045)); Column (5) -0.101 ∗∗ (SE (0.043))

- Table A12: New Borrowing (loan-level regressions, bank×time FE)
  - Foreign-currency loan FCL × Post = -0.284 ∗∗∗ (SE (0.059))
  - New borrowing results appear for foreign currency, local currency, and all loans (N = 364,911)

### Analytical model (appendix C) — key expressions and comparative statics
- Baseline follows Bernanke et al. (1999) with imported inputs and foreign currency borrowing.
- Key definitions and expressions (as in source):
  - Ψ( ̄δ ) = E[δ | δ < ̄δ ] F( ̄δ ) + ̄δ(1 − F( ̄δ ))  (Equation 27)
  - ζ( ̄δ ) = γ E[δ | δ < ̄δ ] F( ̄δ )  (Equation 28)
  - r ≡ (1 / p_M) ρ_M(M) / R  (Equation 38)
  - Actual interest rate on the loan: 1 + i = R / ( ̄δ (Ψ( ̄δ ) − ζ( ̄δ )))  (Equation 42)
  - Elasticity result: ε_{1+i, ̄δ} = 1 − ε_{Ψ−ζ,δ} with ε_{Ψ−ζ,δ} < 1 (Equation 43–44) implying ε_{1+i, ̄δ} > 0.
- Model with exporting: export profits π_e enter as pre-paid profits that increase net worth (Equation 45). Exporters with near-zero default probability do not react to exchange-rate balance-sheet channel.

### Calibration and quantitative contribution (Table A13)
- Calibrated parameters (reported exactly):
  - σ = 5
  - γ = 0.18
  - R = 1.03 (Risk-free interest rate of 3%)
  - Import share = 64% (Data)
  - F = 1.1 (Export intensity of 65%)
  - δ ~ LN(ν, −ν^2 / 2), ν = 3
  - Nominal interest rate of 10.5%
  - d_e = 40% (Data)
  - A = 0.12 (Net worth to sales ratio of 0.5)
  - d_A = 15% (Data)
- Results: Contribution of Dominant Currency Financing Channel
  - Exporters 0%
  - Non-exporters 13.7%

*Italic source attribution: Content reproduced from wpiea2023164-print-pdf - References*

### section 5.

### section 5.

### Document identification
- Page: 77
- Title: The Dominant Currency Financing Channel of External Adjustment
- Working Paper No.: WP/2023/164

*Source: wpiea2023164-print-pdf - section 5.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023164-print-pdf.pdf_