## _wp1697

## Source details

**Canonical URL:** [_wp1697](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1697.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1697.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1697.pdf.json)

---

### I. Introduction: nature and evolution of dollarization
- Dollarization as a diagnostic for weak macroeconomic fundamentals in many emerging market economies (EMEs) over the past 40 years.
- Origins and evolution:
  - 1970s in Latin America: “pure” currency substitution during high- and hyper-inflation bouts, with substitution of foreign currency (usually the US dollar) for domestic currency holdings.
  - Currency substitution typically gradual: domestic currency first loses medium-of-exchange function and then unit-of-account function for major-item transactions (real estate, machinery and equipment), producing dual currency economies.
- From currency substitution to financial dollarization:
  - Early 1990s: banks began taking deposits and lending in two or more currencies, producing asset substitution (financial dollarization) as local currency lost store-of-value function.
  - Financial dollarization grew in the 1990s through the early 2000s across regions, creating financial stability challenges.

### Financial stability risks and historical crises
- Key risks in dollarized financial systems:
  - Large mismatches between the share of dollar-denominated liabilities and the share of dollar-denominated assets.
  - Large income losses to banks when exchange rates fluctuate.
  - Runs on less-than-fully backed dollar-deposits due to fractional reserve banking.
  - Potential for generalized bank runs under not-so-uncommon pre-conditions (Rajan and Tokatlidis, 2005).
- Historical episodes: Turkey in 1994, Argentina in 1995, Russia in 1998, Argentina in 2001.

### Research gap and objectives
- Currency substitution widely studied; dynamics of financial dollarization less well understood.
- Existing explanatory strands include models with financial frictions, institutional weaknesses, and reduced-form panel regressions/VARs.
- This paper’s three questions:
  - How homogenous have patterns of financial dollarization and de-dollarization been across EMEs over the past fifteen years?
  - What factors contributed to recent de-dollarization of highly dollarized economies such as Peru?
  - What lessons does recent financial de-dollarization offer going forward?

### Stylized facts and cross-country patterns (2000–early 2015)
- Dataset and measure:
  - Financial dollarization measured as stock of deposits denominated in foreign currency (US dollar and/or the euro) as a share of total deposits, constructed from the Standardized Report Form (SRF).
  - Quarterly data for 28 emerging market economies over 2000:Q1–2015:Q1.
- Fact 1: Cross-country variation
  - Median dollarization ratio for emerging market economies: just over 17 percent.
  - Distribution positively skewed; several economies semi-dollarized (close to 50 percent) or highly dollarized (higher than 50 percent).
  - Regional patterns: Latin America and Europe (largely formerly transition economies) are significantly more dollarized than Asia/rest of the world.
  - Regional medians 95 percent confidence intervals: Latin America median [28.8, 35.6]; Europe median [27.0, 34.5].
  - Legal restrictions (e.g., Brazil and Chile) contribute to dispersion in Latin America.
- Fact 2: Persistence
  - Financial dollarization displays high persistence in autoregressive estimates:
    - Persistence coefficients (deposit dollarization ratio (lagged)):
      - MGFELSDV: 0.925*** [0.00966]
      - Panel FE: 0.946*** [0.0163]
      - LSDV: 0.974*** [0.00758]
  - Interpretation: dollarization tends to remain virtually unchanged over time; may reflect weak institutions, poor fiscal frameworks, nonexclusive credit contracts, heterogeneous agents.
- Fact 3: De-dollarization can be persistent
  - Mean deposit dollarization fell significantly in several emerging Europe and Latin America economies in the 2000–07 period.
  - 2000–07 associated with adoption/dissemination of inflation targeting (IT) regimes.
  - Conclusion: dollarization is persistent but not irreversible.
- Fact 4: Post-GFC stall or reversal
  - De-dollarization stalled or partly reversed following the Global Financial Crisis (2007:Q3–2008:Q4).
  - Emerging Europe and Latin America experienced mild reversal in 2008–09; continued reversal in emerging Europe during the 2009–13 European sovereign debt crisis (notably Serbia and Ukraine).
  - Latin America overall continued de-dollarization, with Peru and Uruguay showing substantive de-dollarization through 2015.
  - Persistence pre- and post-GFC (deposit dollarization ratio (lagged)):
    - Pre-GFC:
      - MGFELSDV: 0.837*** [0.0403]
      - FE: 0.914*** [0.0374]
      - LSDV: 0.977*** [0.0181]
    - Post-GFC:
      - MGFELSDV: 0.810*** [0.0327]
      - FE: 0.875*** [0.0277]
      - LSDV: 0.964*** [0.0170]
  - Note: slight reduction in persistence after the GFC but not statistically significant at 5 percent.

### The analytics of de-dollarization: drivers and channels
- Two sets of drivers: domestic (country-specific) and external (global).
- Domestic factors
  - Adoption of inflation targeting (IT)
    - IT hypothesized to strengthen domestic currency as store of value if it delivers low and stable inflation; potential drawback: exchange rate volatility and “fear to float” in highly dollarized economies.
  - Macroprudential policies
    - Caps on loan-to-value ratios, capital requirements, higher provisions for dollar lending; expected to discourage financial dollarization but may have negative medium-term effects on deepening/growth.
  - Differential reserve requirements (taxing dollar lending/deposits)
    - Increasing reserve requirements or reducing RR remuneration on dollar deposits equivalent to a tax on dollar liabilities; may reduce deposit dollarization but increase banks’ net external borrowing, leaving loan dollarization unchanged.
  - Expected real GDP growth (1-year ahead)
    - Ambiguous: higher domestic investment can increase demand for credit (possibly in dollars); higher net capital inflows can increase dollarization if intermediated by banks.
  - Relative price of non-tradables to tradables
    - Improvement generally expected to reduce credit dollarization; exceptions exist (e.g., expansion of dollar loans to real estate).
  - Mechanical valuation effect of exchange rate depreciations
    - Depreciation mechanically raises the share of dollar loans/deposits via valuation effects; commodity prices used as proxy for bilateral exchange rate—dollarization should fall when commodity prices rise.
- External factors
  - World interest rates
    - Lower world interest rates lower cost of U.S. dollar external funding encouraging dollar-denominated offshore borrowing; net effect on deposit and loan dollarization ambiguous.
  - Global risk aversion
    - Rising global risk aversion induces flight to safe-haven currencies like the U.S. dollar, raising financial dollarization; VIX used as proxy.

### Regression analysis — Peru (1993:Q1–2014:Q4)
- Data and approach
  - Longest quarterly time series for a country: Peru 1993:Q1–2014:Q4.
  - Dollarization measures: credit dollarization (dollar-denominated loans / total loans) and deposit dollarization.
  - Unit root tests: cannot reject unit roots in deposit and credit dollarization ratios; no robust evidence of cointegration with explanatory variables.
  - Estimation in first differences: dependent variable is quarterly change in dollarization ratio; seasonal dummies included.
- Credit dollarization regression highlights (Table 3 and Table 5 summaries)
  - Domestic controls:
    - Inflation targeting (IT) dummy: negative and significant in many specifications.
      - Example: column 3 IT coefficient -0.716 with note: -0.716*4 = -2.86 implying adoption of IT reduced credit dollarization by over 2½ percentage points a year in that example.
    - Higher provisions for foreign currency loans (dummy): negative and significant (temporary effect).
    - Higher capital requirement for forex exposure: mixed/insignificant coefficients (e.g., 0.272 [0.239] in column 1 of Table 3; other specifications show insignificance).
    - Marginal reserve requirement in foreign currency (difference): often insignificant (e.g., -0.035 [0.0262] in Table 3).
    - Real GDP growth (1-year ahead forecast): positive and sometimes significant (e.g., 0.283** [0.125] in Table 3).
    - Real exchange rate (logs, lagged difference): negative and significant in some columns (e.g., -12.21** [5.787]; -16.63*** [5.933] in Table 3).
      - One standard deviation increase in lagged change in the real exchange rate (about 0.023 percent) associated with a ¼ percentage point reduction in credit dollarization.
  - External factors:
    - Metal prices (logs real, difference): negative and significant (e.g., -5.714*** [1.584]; -6.584*** [1.194]).
      - One standard deviation increase in change of log real metal prices (0.085) associated with a reduction in credit dollarization of ½ percentage points.
    - Libor interest rate (3-month US, difference): positive and sometimes significant (e.g., 0.653*** [0.214]; 0.787*** [0.186]).
      - Interpreted as higher interest rate elasticity of demand for dollar loans than for local currency loans in Peru.
    - VIX (logs eop, difference): positive and significant (e.g., 1.294*** [0.440]; 1.148*** [0.401]).
      - A standard deviation increase in first difference of log VIX (0.274) associated with an over ¼ percentage point increase in credit dollarization.
  - Preferred specification (column 12, Table 3):
    - Combines domestic and external drivers; marginal reserve requirement effect remains insignificant.
    - R-square d 0.519; Adjusted R-square d 0.443.
    - Finding: better growth prospects significantly associated with an increase in credit dollarization.
- Deposit dollarization regression highlights (Table 4 and Table 5 summaries)
  - Signs and significance broadly similar to credit regressions with notable differences:
    - Marginal reserve requirement on dollar deposits appears to discourage deposit dollarization in most specifications (e.g., -0.0990*** [0.0341]; -0.0890** [0.0393]).
    - IT dummy is generally not statistically significant for deposit dollarization (coefficients include -0.18 [0.313], -0.0621 [0.323], etc.).
    - Higher provisions for foreign currency loans show strong negative and significant effects on deposit dollarization (e.g., -1.533*** [0.290]).
    - Metal prices (logs real, difference) remain strongly negative and significant for deposit dollarization (e.g., -7.706*** [1.522]).
    - Libor and VIX effects on deposit dollarization: Libor sometimes significant and positive (e.g., 0.882*** [0.333]); VIX positive and significant (e.g., 1.732*** [0.559]).
  - Conjectured behavioral differences:
    - Banks may be more sensitive than depositors to inflation risks when allocating assets between local currency and dollar loans.
    - Depositors may respond differently, for example being lured by higher local currency interest rates when inflation is volatile and when deposit maturities are shorter than loans.
    - Implication: marginal reserve requirements on dollar deposits reduce deposit dollarization but not necessarily credit dollarization; suggests RR policies should be part of a package that also penalizes other sources of dollar funding (including short-term debt).
- Robustness checks (Table 5 summary)
  - Examined exchange rate volatility and domestic and foreign inflation volatility (three-month non-overlapping standard deviations).
  - Nominal exchange rate volatility and domestic CPI inflation volatility: not economically or statistically significant for credit or deposit dollarization.
  - Some evidence that U.S. inflation volatility discourages credit dollarization (not robust across specifications and not significant for deposit dollarization).

### Main empirical findings and interpretation
- Cross-country evidence:
  - Broad de-dollarization across many EMEs in years preceding the global financial crisis; process stalled or reversed in many EMEs post-GFC.
  - De-dollarization persistence is not synonymous with irreversibility.
  - After a partial and short-lived reversal in 2008–09, de-dollarization continued in some highly dollarized economies (e.g., Peru); in some lower-dollarized EMEs, dollarization increased mildly over the past five years.
  - De-dollarization shaped by combination of domestic and global factors.
- Peru case study conclusions:
  - De-dollarization can be achieved through sound macroeconomic policies and some macro-prudential measures.
  - Strongest domestic factor behind credit de-dollarization appears to be introduction of inflation targeting in 2002.
  - Conservative fiscal stance since 2002 halved public debt since 2002, contributed to investment-grade status, and supported domestic debt market development.
  - External factors reinforcing de-dollarization:
    - Lower international interest rates contributed to a fall in loan dollarization.
    - Decline in global stock market volatility (pre-GFC and from 2010 onwards) weakened flight-to-safety effects.
    - Higher commodity prices in 2000–2007 and some post-GFC recovery helped de-dollarization by appreciating the local currency.

### Policy implications and recommendations
- Monetary regime and credibility
  - Adoption of credible inflation targeting frameworks can help de-dollarize credit (empirical Peru example: IT associated with a reduction in credit dollarization of over 2½ percentage points a year in a specific regression example).
  - IT may not suffice to reduce deposit dollarization; depositor behavior can differ from banks’ behavior.
- Macroprudential and regulatory toolkit
  - Measures that can support de-dollarization:
    - Higher loan provisioning for foreign currency loans (shown negative and significant for both credit and deposit dollarization).
    - Higher capital requirements on dollar loans (effects mixed; can discourage dollar lending).
    - Differential reserve requirements and lower RR remuneration on dollar deposits act like a tax on dollar liabilities and can reduce deposit dollarization (e.g., marginal RR coefficients -0.0990*** [0.0341]; -0.0890** [0.0393]).
  - Caveat: these measures can tax financial intermediation and may backfire if all sources of bank financing are not equally affected; banks may substitute toward non-deposit dollar funding (net external borrowing), leaving loan dollarization unchanged.
  - Recommendation: RR policies should be used as part of a package that also penalizes other sources of dollar funding (including short-term debt).
- External condition dependence and sequencing
  - De-dollarization benefits from:
    - abundant international liquidity,
    - strong commodity prices,
    - low global volatility.
  - Global risk aversion (VIX) and world interest rates can counteract de-dollarization; policymakers should account for external shocks and global financial conditions.
  - Growth trade-offs: faster expected growth can increase credit dollarization via higher investment demand and capital inflows; policymakers need to balance growth objectives with currency composition of credit.
  - Cost-benefit assessment required: some short-run de-dollarization measures (macroprudential controls) may have adverse medium-term effects on financial deepening and growth.

### Data notes and sample composition (Appendix)
- Cross-country sample in section II: 28 emerging market economies grouped in three regions:
  - Latin America: Argentina, Chile, Costa Rica, Dominican Republic, Guatemala, Mexico, Paraguay, Peru, Uruguay, and Venezuela.
  - Europe: Turkey, Ukraine, Czech Republic, Slovak Republic, Serbia, Hungary, Croatia, Slovenia, Poland, and Romania.
  - Asia and other: South Africa, Israel, Egypt, Indonesia, Korea, Malaysia, Pakistan, and Thailand.
- Peruvian database selected data sources and definitions:
  - Inflation targeting dummy: Central Bank of Peru; value of one from 2002:Q1 and zero otherwise.
  - Higher provisions for FC loans dummy: Garcia-Escribano, M., 2010; value of one during 2006:Q3-2007:Q2 and zero otherwise.
  - Higher capital requirement for FX exposure dummy: Choy, M. and G. Chang, 2014; value of one from 2012:Q4 and zero otherwise.
  - Libor rates on US dollar deposits (3-months): IFS.
  - CPI: IFS (for Peru and the United States).
  - Metal price index: IMF commodity prices.
  - Real GDP growth (1-year forecast): Consensus forecast/WEO.
  - VIX: FRED CBOE Volatility Index.
  - Real effective exchange rate: IMF.
  - Reserve requirement in dollar deposits (marginal): Central Bank of Peru.
  - Bilateral exchange rate (Soles per Dollar, Average): Central Bank of Peru (Interbank exchange rate).
  - Credit of financial system to private sector (foreign currency and total), liquidity, and quasimoney (local currency): Central Bank of Peru.

*Content based solely on _wp1697 - Section 4 concludes with a brief discussion of policy implications.*

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

### _wp1697 - References .............................................................................................................

### I. Introduction: nature and evolution of dollarization
- Dollarization has been a buzzword for weak macroeconomic fundamentals in many emerging market economies (EMEs) over the past 40 years.
- Origins and evolution:
  - 1970s in Latin America: a “pure” currency substitution phenomenon driven by high- and hyper-inflation bouts, inducing substitution of foreign currency (usually the US dollar) for domestic currency holdings.
  - Currency substitution process typically gradual: domestic currencies first lose their medium of exchange function and then their unit of account function, particularly for major-item transactions like real estate, machinery and equipment.
  - Local currency transactions generally did not disappear, producing dual currency economies in much of the region.
- From currency substitution to financial dollarization:
  - By the early 1990s banks began taking deposits and lending in two or more currencies, generating asset substitution (financial dollarization) as local currency lost its store of value function.
  - Financial dollarization grew in the 1990s through the early 2000s across distinct regions, creating financial stability challenges.

### Financial stability risks and historical crises
- Key risks inherent in dollarized financial systems:
  - Large mismatches between the share of dollar-denominated liabilities and the share of dollar-denominated assets.
  - Large income losses to banks when exchange rates fluctuate.
  - Possibility of runs on less-than-fully backed dollar-deposits due to fractional reserve banking.
  - These risks can trigger generalized bank runs under not-so-uncommon pre-conditions (Rajan and Tokatlidis, 2005).
- Historical episodes involving materialization of these risks:
  - Turkey in 1994, Argentina in 1995, Russia in 1998, Argentina in 2001.

### Research gap and paper objectives
- Currency substitution widely studied; dynamics of financial dollarization less well understood.
- Existing explanatory strands include:
  - Models with financial frictions such as asymmetric liquidation costs and imperfect information (Broda and Levy-Yeyati, 2006; Rapoport, 2009).
  - Institutional weakness and non-competitive banking in two currencies (Rajan and Tokatlidis, 2005; Catão and Terrones, 2000).
  - Reduced-form panel regressions and VARs (De Nicólo et al. 2003; Levy-Yeyati, 2006; Honohan, 2007; Antinolfi et al., 2007; García-Escribano and Sosa, 2011).
- This paper’s three questions:
  - How homogenous have been the patterns of financial dollarization and de-dollarization across the EMEs over the past fifteen years?
  - What factors have contributed to the recent de-dollarization of some highly dollarized economies such as Peru?
  - What are the lessons of the recent financial de-dollarization experience going forward?

### Stylized facts and cross-country patterns (2000–early 2015)
- Main stylized facts documented:
  - Most EMEs have de-dollarized since 2000.
  - Dollarization persistence—previously emphasized—has been broadly on the wane in most highly dollarized EMEs.
  - Global factors have been important drivers of de-dollarization.
  - The speed of de-dollarization has been far from homogeneous across countries, implying an important role for country-specific factors.

### Focus on Peru: method and findings
- Rationale for focusing on Peru:
  - Peru represents a striking and arguably successful experience of de-dollarizing from very high and persistent levels.
- Empirical approach:
  - Aggregate perspective with reduced-form estimates subject to a greater variety of controls and robustness checks than some prior literature.
- Broad picture emerging for Peru:
  - Global factors have been important contributors to de-dollarization.
  - Monetary and macro-prudential policies—particularly the move to an inflation targeting regime and the use of macro-prudential measures—have also been crucial.

### Policy implications and trade-offs
- Desirability and policy instruments:
  - If financial sector de-dollarization is desirable, which policy choices and instruments are preferable?
  - Adoption of an inflation targeting regime is generally non-distortionary.
  - Introduction of some prudential measures can be useful in the short run but may have distortionary effects if not properly designed.
  - A de-dollarization process relying only on prudential measures is more likely to hamper financial intermediation and growth.

### Structure of the paper
- Section I: Reviews international experience with financial sector dollarization from 2000 to early 2015 and highlights pertinent stylized facts.
- Section 2: Briefly reviews the main determinants of financial dollarization (de-dollarization).
- Section 3: Provides econometric evidence for Peru.

*Source: _wp1697 - References .............................................................................................................*

### Section 4 concludes with a brief discussion of policy implications.

### _wp1697 - Section 4 concludes with a brief discussion of policy implications.

### Stylized international facts
- Measure and dataset
  - Financial dollarization measure: stock of deposits denominated in foreign currency (US dollar and/or the euro) as a share of total deposits, constructed with data from the Standardized Report Form (SRF).
  - Dataset: quarterly data for 28 emerging market economies over 2000:Q1–2015:Q1.
- Fact 1: Cross-country variation
  - Median dollarization ratio for emerging market economies: just over 17 percent.
  - Distribution: positively skewed around the median; several economies semi-dollarized (close to 50 percent) or highly dollarized (higher than 50 percent).
  - Regional patterns: Latin America and Europe (largely formerly transition economies) are significantly more dollarized than Asia/rest of the world.
  - Regional medians confidence intervals: Latin America median 95 percent confidence interval [28.8, 35.6]; Europe median 95 percent confidence interval [27.0, 34.5].
  - Noted legal restrictions (e.g., Brazil and Chile) contribute to dispersion in Latin America.
- Fact 2: Persistence
  - Financial dollarization displays high persistence estimated via autoregressive processes.
  - Table 1 persistence coefficients (deposit dollarization ratio (lagged)):
    - MGFELSDV: 0.925*** [0.00966]
    - Panel FE: 0.946*** [0.0163]
    - LSDV: 0.974*** [0.00758]
  - Interpretation: dollarization tends to remain virtually unchanged over time; possible reflection of weak institutions, poor fiscal frameworks, nonexclusive credit contracts, heterogeneous agents.
- Fact 3: De-dollarization can be persistent
  - Mean deposit dollarization fell significantly in several emerging Europe and Latin America economies in the 2000–07 period.
  - 2000–07 period associated with adoption/dissemination of inflation targeting (IT) regimes.
  - Conclusion: dollarization is persistent but not irreversible.
- Fact 4: Post-GFC stall or reversal
  - De-dollarization stalled or partly reversed following the Global Financial Crisis (2007:Q3–2008:Q4).
  - Emerging Europe and Latin America experienced mild reversal in 2008–09; reversal continued in emerging Europe during the 2009–13 European sovereign debt crisis (notably Serbia and Ukraine).
  - Latin America continued de-dollarization overall, with Peru and Uruguay showing substantive de-dollarization through 2015.
  - Table 2 persistence pre- and post-GFC (deposit dollarization ratio (lagged)):
    - Pre-GFC:
      - MGFELSDV: 0.837*** [0.0403]
      - FE: 0.914*** [0.0374]
      - LSDV: 0.977*** [0.0181]
    - Post-GFC:
      - MGFELSDV: 0.810*** [0.0327]
      - FE: 0.875*** [0.0277]
      - LSDV: 0.964*** [0.0170]
  - Note: slight reduction in persistence after the GFC but not statistically significant at 5 percent.

### The analytics of de-dollarization
- Two sets of drivers: domestic (country-specific) and external (global).
- Domestic factors highlighted
  - Adoption of inflation targeting (IT)
    - IT is hypothesized to strengthen domestic currency as a store of value if it delivers low and stable inflation.
    - Potential drawback: exchange rate volatility and "fear to float" in highly dollarized economies.
  - Macroprudential policies
    - Examples: caps on loan-to-value ratios, capital requirements, higher provisions for dollar lending.
    - Expected to discourage financial dollarization, though medium-term effects on deepening/growth may be negative (cost-benefit analysis beyond this paper).
  - Differential reserve requirements (taxing dollar lending/deposits)
    - Increasing reserve requirements (RR) or reducing RR remuneration on dollar deposits is equivalent to a tax on dollar liabilities.
    - Effect may reduce deposit dollarization but could increase banks' net external borrowing, leaving loan dollarization unchanged.
  - Expected real GDP growth (1-year ahead)
    - Two channels: higher domestic investment (ambiguous effect on dollarization depending on income elasticities) and higher net capital inflows (can increase dollarization if U.S. dollar inflows intermediated by banks).
  - Relative price of non-tradables to tradables
    - Improvement in relative price of non-tradables is generally expected to reduce credit dollarization; exceptions exist (e.g., expansion of dollar loans to real estate).
  - Mechanical valuation effect of exchange rate depreciations
    - Local currency depreciation mechanically raises the share of dollar loans/deposits in totals via valuation effects.
    - Commodity prices used as proxy for bilateral exchange rate (expected negative association): dollarization should fall when commodity prices rise.
- External factors
  - World interest rates
    - Lower world interest rates lower cost of U.S. dollar external funding encouraging dollar-denominated offshore borrowing; net effect on deposit and loan dollarization ambiguous and empirical.
  - Global risk aversion
    - Rising global risk aversion induces flight to safe-haven currencies like the U.S. dollar, raising financial dollarization.
    - VIX index used as proxy for global risk aversion.

### Regression analysis (Peru, 1993:Q1–2014:Q4)
- Data and approach
  - Longest quarterly time series available for a country: Peru 1993:Q1–2014:Q4.
  - Dollarization measures used: credit dollarization (dollar-denominated loans / total loans) and deposit dollarization.
  - Unit root tests: cannot reject unit roots in deposit and credit dollarization ratios; no robust evidence of cointegration with explanatory variables.
  - Estimation in first differences: dependent variable is quarterly change in dollarization ratio.
- Credit dollarization regression highlights (Table 3)
  - Domestic dummies and controls:
    - Inflation targeting (IT) dummy: negative and significant in many specifications.
      - Example coefficient interpretation: column 3 IT coefficient -0.716 with note: -0.716*4 = -2.86 implying adoption of IT reduced credit dollarization by over 2½ percentage points a year.
    - Higher provisions for foreign currency loans (dummy): negative and significant (temporary effect).
    - Higher capital requirement for forex exposure: mixed/insignificant coefficients across specifications (e.g., 0.272 [0.239] in column 1).
    - Marginal reserve requirement in foreign currency (difference): coefficients often insignificant (e.g., -0.035 [0.0262]).
    - Real GDP growth (1-year ahead forecast): positive and in some specifications significant (e.g., 0.283** [0.125]).
    - Real exchange rate (logs, lagged difference): negative and significant in some columns (e.g., -12.21** [5.787]; -16.63*** [5.933] in later columns).
      - One standard deviation increase in the lagged change in the real exchange rate (about 0.023 percent) associated with a ¼ percentage point reduction in credit dollarization.
  - External factors:
    - Metal prices (logs real, difference): negative and significant (e.g., -5.714*** [1.584]; -6.584*** [1.194]).
      - One standard deviation increase in the change of the log of real metal prices (0.085) associated with a reduction in the credit dollarization ratio of ½ percentage points.
    - Libor interest rate (3-month US, difference): positive and, in some specifications, significant (e.g., 0.653*** [0.214]; 0.787*** [0.186]).
      - Interpreted as higher interest rate elasticity of demand for dollar loans than local currency loans in Peru.
    - VIX (logs eop, difference): positive and significant (e.g., 1.294*** [0.440]; 1.148*** [0.401]).
      - A standard deviation increase in the first difference of log VIX (0.274) associated with an over ¼ percentage point increase in credit dollarization.
  - Preferred specification (column 12, Table 3)
    - Combines domestic and external drivers; marginal reserve requirement effect remains insignificant.
    - R-squared about a little over 50 percent (R-square d 0.519; Adjusted R-square d 0.443) indicating reasonable explanatory power.
    - Finding: better growth prospects significantly associated with an increase in credit dollarization.
- Deposit dollarization regression highlights (Table 4 summary)
  - Signs and significance broadly similar to credit dollarization regressions, with differences:
    - Marginal reserve requirement on dollar deposits appears to discourage deposit dollarization in most specifications.
    - IT dummy is no longer statistically significant for deposit dollarization.
  - Conjecture on IT vs. deposits:
    - Banks may be more sensitive than depositors to inflation risks when allocating assets between local currency and dollar loans.
    - Depositors may respond differently (e.g., lured by higher local currency interest rates when inflation is volatile), especially when maturity gap between loans and deposits is large (lower average maturity of deposits).
    - Implication: marginal reserve requirements on dollar deposits reduce deposit dollarization but not credit dollarization; suggests RR policies should be part of a package that also penalizes other sources of dollar funding (including short-term debt).
- Robustness checks (Table 5 summary)
  - Examined exchange rate volatility and domestic and foreign inflation volatility (computed as three-month non-overlapping standard deviations of changes).
  - Nominal exchange rate volatility and domestic CPI inflation volatility: not economically or statistically significant for either credit or deposit dollarization.
  - Some evidence that U.S. inflation volatility discourages credit dollarization (not robust across specifications and not significant for deposit dollarization).

### Policy implications (as discussed within the analysis)
- Monetary regime and credibility
  - Adoption of credible inflation targeting frameworks can help de-dollarize credit (empirical evidence for Peru: IT adoption associated with a reduction in credit dollarization of over 2½ percentage points a year in the specific regression example).
  - However, IT may not be sufficient to reduce deposit dollarization; depositor behavior may differ from banks’ behavior.
- Macroprudential and regulatory toolkit
  - Macroprudential measures (caps on LTV, higher provisions for FX loans, capital requirements) can discourage dollar lending and reduce credit dollarization, though some effects may be temporary.
  - Differential reserve requirements and lower RR remuneration on dollar deposits act like a tax on dollar liabilities and can reduce deposit dollarization, but may induce banks to increase non-deposit dollar funding (e.g., net external borrowing), potentially leaving loan dollarization unchanged.
  - Policy mix recommendation: unconventional RR policies should be used as part of a package that also penalizes other sources of dollar funding by banks (including short-term debt) to prevent substitution toward non-deposit dollar liabilities.
- External condition dependence
  - De-dollarization is influenced by external factors: commodity price booms (e.g., metal prices) and favorable external conditions that appreciate the local currency can facilitate de-dollarization.
  - Global risk aversion (VIX) and world interest rates can counteract de-dollarization efforts; policymakers should account for external shocks and global financial conditions when designing de-dollarization strategies.
- Growth trade-offs and sequencing
  - Faster expected growth can increase credit dollarization through higher investment demand and capital inflows; policymakers need to balance growth objectives and currency composition of credit.
  - Cost-benefit considerations: some measures that reduce dollarization in the short run (e.g., macroprudential controls) may have adverse medium-term effects on financial deepening and growth; a comprehensive assessment is required beyond the scope of this paper.

*Italic: Content based solely on _wp1697 - Section 4 concludes with a brief discussion of policy implications.*

### Section II, this coefficient is ambiguously signed in theory. Another ambiguously signed coefficient is expected

### _wp1697 - Section II, this coefficient is ambiguously signed in theory. Another ambiguously signed coefficient is expected

### Determinants of Financial Sector Deposit Dollarization (Peru, 1993:Q1-2014:Q4) — main regression results (Table 4)
- Sample size: 88 observations.
- Dependent variable: change in the deposit dollarization ratio.
- Seasonal dummies included in all regressions.
- Key domestic factors (coefficients shown with robust standard errors in brackets; significance: * 10 percent, ** 5 percent, *** 1 percent):
  - Inflation targeting (dummy): coefficients reported across specifications include -0.18 [0.313], -0.0621 [0.323], -0.182 [0.353], -0.0142 [0.337], -0.182 [0.354], -0.163 [0.345], -0.015 [0.332], 0.167 [0.290].
  - Higher provisions for foreign currency loans (dummy): -1.533*** [0.290], -1.412*** [0.324], -1.498*** [0.330], -1.430*** [0.306], -1.475*** [0.328], -1.566*** [0.321], -1.290** [0.532].
  - Higher capital requirement for foreign exchange exposure (dummy): 0.642 [0.506], 0.515 [0.463], 0.707 [0.623], 0.59 [0.492], 0.592 [0.574], 0.081 [0.518].
  - Marginal reserve requirement in foreign currency (difference): -0.0990*** [0.0341], -0.0890** [0.0393], -0.0222 [0.0348].
  - Real GDP growth (1-year ahead forecast): -0.0501 [0.193], -0.0806 [0.200], 0.0487 [0.172].
  - Real exchange rate (logs, lagged difference): -14.41* [7.972], -11.18 [8.161], -17.15** [7.146].
- Key external factors:
  - Metal prices (logs real, difference): -7.706*** [1.522], -8.906*** [1.562], -8.812*** [1.848].
  - Libor interest rate (3-month US, difference): 0.353 [0.460], 0.882*** [0.333], 0.892*** [0.330].
  - VIX (logs eop, difference): 1.732*** [0.559], 1.341** [0.527], 1.318** [0.570].
- Goodness of fit (selected R-squared / Adjusted R-squared across specifications):
  - R-squared values include 0.0373, 0.0813, 0.0965, 0.161, 0.0973, 0.142, 0.191, 0.223, 0.0461, 0.12, 0.35, 0.455.
  - Adjusted R-squared values include -0.00914, 0.0252, 0.0295, 0.0874, 0.0183, 0.0666, 0.0974, 0.185, 0.000111, 0.0778, 0.302, 0.367.

### Determinants of Financial Sector Dollarization — additional variables and credit vs deposit results (Peru, 1993:Q1-2014:Q4) (Table 5)
- Sample size: 88 observations for reported specifications.
- Dependent variable: change in the loan dollarization ratio (or the change in the deposit dollarization ratio).
- Selected domestic factors (coefficients with robust standard errors in brackets):
  - Inflation targeting (dummy): -0.586** [0.278], -0.576** [0.284], -0.433* [0.239], -0.015 [0.332], -0.00736 [0.341], 0.165 [0.293].
  - Higher provisions for foreign currency loans (dummy): -1.763*** [0.388], -1.799*** [0.353], -1.612*** [0.513], -1.566*** [0.321], -1.595*** [0.398], -1.302** [0.566].
  - Higher capital requirement for FX exposure (dummy): -0.0055 [0.355], -0.0131 [0.344], -0.436 [0.323], 0.592 [0.574], 0.586 [0.565], 0.0914 [0.517].
  - Marginal reserve requirement in foreign currency (difference): -0.0165 [0.0288], 0.162 [0.155], 0.274** [0.129], -0.0890** [0.0393], -0.0841 [0.194], 0.0425 [0.168].
  - Real GDP growth (1-year ahead forecast): 0.164 [0.153], -0.0234 [0.0314], 0.0252 [0.0206], -0.0806 [0.200], -0.0998** [0.0407], -0.027 [0.0341].
  - Real exchange rate (logs, lagged difference): -12.14** [6.030], -11.86* [6.333], -16.86*** [6.114], -11.18 [8.161], -10.84 [8.637], -16.98** [7.518].
  - Inflation volatility (logs, difference): -0.135 [0.135], -0.0755 [0.113], -0.187 [0.265], -0.0583 [0.183].
  - Exchange rate volatility (logs, difference): 0.0299 [0.130], -0.0165 [0.108], 0.107 [0.189], 0.0539 [0.167].
- Selected external factors:
  - Metal prices (logs real, difference): -6.597*** [1.387], -6.571*** [1.393], -6.347*** [1.288], -8.906*** [1.562], -8.913*** [1.571], -8.636*** [1.766].
  - Libor interest rate (3-month US, difference): 0.653*** [0.214], 0.660*** [0.207], 0.758*** [0.192], 0.882*** [0.333], 0.880** [0.335], 0.855*** [0.300].
  - VIX (logs eop, difference): 1.004** [0.404], 0.810* [0.428], 0.910** [0.423], 1.341** [0.527], 1.394** [0.593], 1.321* [0.663].
  - US inflation volatility (logs, difference): -0.119 [0.0852], -0.146** [0.0621], 0.0324 [0.0860], -0.00698 [0.0730].
- Goodness of fit (selected R-squared / Adjusted R-squared):
  - R-squared values include 0.295, 0.305, 0.309, 0.324, 0.54, 0.191, 0.208, 0.35, 0.351, 0.457.
  - Adjusted R-squared values include 0.213, 0.204, 0.258, 0.265, 0.445, 0.0974, 0.0937, 0.302, 0.294, 0.344.

### Main empirical findings and interpretation (from text)
- Cross-country evidence:
  - There was financial de-dollarization across a broad range of emerging market economies in the years preceding the global financial crisis, but the process stalled in recent years in many—though not all—EMEs.
  - De-dollarization persistence is not synonymous with dollarization irreversibility.
  - After a partial and short-lived reversal in 2008–09, de-dollarization continued in some highly dollarized economies, like Peru; in some lower-dollarized EMEs, dollarization increased mildly over the past five years.
  - De-dollarization has been shaped by a combination of domestic and global factors.
- Peru case study:
  - De-dollarization can be achieved through sound macroeconomic policies and some macro-prudential measures.
  - The strongest domestic factor behind credit de-dollarization appears to be the introduction of inflation targeting in 2002.
  - A conservative fiscal stance since 2002 has halved public debt since 2002, contributed to investment-grade status, and spurred confidence in domestic macroeconomic management—supporting domestic debt market development.
  - External factors that reinforced de-dollarization:
    - Lower international interest rates lowered the cost of foreign borrowing and contributed to a fall in loan dollarization.
    - Decline in global stock market volatility (pre-GFC and from 2010 onwards) weakened flight-to-safety effects, contributing to de-dollarization.
    - Higher commodity prices in 2000–2007 and some post-GFS recovery helped de-dollarization by appreciating the local currency.

### Policy implications and recommendations (from text)
- Being open to global factors helps de-dollarization when:
  - international liquidity is abundant,
  - commodity prices are strong, and
  - global volatility is low.
- Macro-prudential measures can support de-dollarization, notably:
  - higher loan provisioning for foreign currency loans,
  - higher capital requirements on dollar loans.
  - Caveat: these measures can tax financial intermediation and may backfire if all sources of bank financing are not equally affected.
- Most effective approach:
  - credible macroeconomic policies combined with inflation targeting and supportive fiscal policies.
  - Such policies, when combined with favorable external conditions, can be effective even in highly dollarized countries (Peru experience).

### Data notes and sample composition (Appendix)
- Cross-country sample in section II: 28 emerging market economies grouped in three regions:
  - Latin America: Argentina, Chile, Costa Rica, Dominican Republic, Guatemala, Mexico, Paraguay, Peru, Uruguay, and Venezuela.
  - Europe: Turkey, Ukraine, Czech Republic, Slovak Republic, Serbia, Hungary, Croatia, Slovenia, Poland, and Romania.
  - Asia and other: South Africa, Israel, Egypt, Indonesia, Korea, Malaysia, Pakistan, and Thailand.
- Data sources for Peruvian database (selected):
  - Inflation targeting dummy: Central Bank of Peru; takes value of one from 2002:Q1 and zero otherwise.
  - Higher provisions for FC loans dummy: Garcia-Escribano, M., 2010; takes value of one during 2006:Q3-2007:Q2 and zero otherwise.
  - Higher capital requirement for FX exposure dummy: Choy, M. and G. Chang, 2014; takes value of one from 2012:Q4 and zero otherwise.
  - Libor rates on US dollar deposits (3-months): IFS.
  - CPI: IFS (for Peru and the United States).
  - Metal price index: IMF commodity prices.
  - Real GDP growth (1-year forecast): Consensus forecast/WEO.
  - VIX: FRED CBOE Volatility Index.
  - Real effective exchange rate: IMF.
  - Reserve requirement in dollar deposits (marginal): Central Bank of Peru.
  - Bilateral exchange rate (Soles per Dollar, Average): Central Bank of Peru (Interbank exchange rate).
  - Credit of financial system to private sector (foreign currency and total), liquidity, and quasimoney (local currency): Central Bank of Peru.

*Source: _wp1697 - Section II, this coefficient is ambiguously signed in theory. Another ambiguously signed coefficient is expected (IMF working paper content provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2016/_wp1697.pdf_