## 1. Private Sector Financial Dollarization

## Source details

**Canonical URL:** [1. Private Sector Financial Dollarization](https://www.imf.org/-/media/files/publications/wp/2019/wp1919.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wp1919.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wp1919.pdf.json)

---

### I. Introduction
- Dollarization defined as use of foreign currency as medium of exchange, store of value, or unit of account by residents; notable in Latin America, Eastern Europe, parts of Asia, and sub-Saharan Africa (SSA).
- Persistence: dollarization can persist as a legacy of past severe economic and political disruption despite improved macroeconomic stability and institutional credibility.
- Research gap: scant attention to differences in determinants between household and firm dollarization.
- Objective: provide systematic empirical analysis differentiating household and firm dollarization across countries to inform targeted de-dollarization policies.
- Data and scope: deposit and loan dollarization for households and firms from IMF IFS database, covering 2001 to 2016; caveats include incomplete country coverage and measurement issues in registries.

### II. Main findings (summary)
- Structural factors largely drive both household and firm deposit and loan dollarization rather than macroeconomic stability over 2001–2016.
- Impact of structural factors differs between households and firms, and between deposit- and loan-dollarization.

### III. Data, measurement, and samples
- Dollarization measures: share of FX deposits/loans of total deposits/loans for households and firms.
- Database: IMF International Financial Statistics (IFS).
- Sample period: 2001 to 2016.
- Effective econometric samples:
  - Households’ deposit dollarization: 52 countries.
  - Households’ loan dollarization: 39 countries.
  - Firms’ deposit dollarization: 49 countries.
  - Firms’ loan dollarization: 63 countries.
- Observations with 0 or 100 percent dollarization are excluded from effective samples.
- Financial sector composition (Average 2001-2016):
  - Other Financial Corporations: Deposits 5.2, Loans 2.0; Dollarization Deposits 19.2, Loans 16.3
  - State and Local Government: Deposits 1.2, Loans 0.8; Dollarization Deposits 9.3, Loans 8.0
  - Public Non-Financial Corporations: Deposits 6.5, Loans 5.2; Dollarization Deposits 23.7, Loans 20.9
  - Private Sector: Firms: Deposits 30.9, Loans 75.0; Dollarization Deposits 30.7, Loans 47.7
  - Private Sector: Households: Deposits 56.2, Loans 17.1; Dollarization Deposits 26.8, Loans 21.0
- Aggregate sector shares (text context):
  - Firms represent 87.1 percent of total deposits and 92.1 percent of total loans in the financial sector, excluding the Central Banks.
  - In terms of dollarization, they represented 57.5 percent of deposits and 68.7 percent of deposits and loans respectively.

### IV. Theoretical perspectives and micro evidence
- Three principal theoretical perspectives (Levy-Yeyati 2006):
  - Portfolio view: dollarization as optimal portfolio choice given currency returns; deposit and loan behavior should be analogous conditional on rate differentials and no regulatory restrictions.
  - Market development view: dollarization arises from market imperfections—limited current account openness, weak financial deepening (low M2/GDP), and lack of domestic investment opportunities.
  - Institutional view: weak institutions and political instability undermine monetary credibility and encourage FX holdings and FX borrowing.
- Micro evidence highlights heterogeneity:
  - Households: country studies show mixed borrower characteristics and motives (risk aversion, income in FX, remittances, euro adoption expectations).
  - Firms: FX borrowing linked to currency matching (exporters), firm size, age, sophistication, and access to foreign funds.

### V. Empirical strategy and key variables
- Baseline panel model:
  - Dependent variable: dollarization (deposits or loans in FX as percent of total) for households and firms.
  - Explanatory groups:
    - CURRENCY: inflation rate and nominal exchange rate depreciation vs USD.
    - PORTFOLIO: deposit or loan interest rate spreads to the US and relevant real interest rates.
    - MARKET DEVELOPMENT (MKTDEV): Chinn-Ito KA index, external debt to GDP, M2 to GDP, GDP per capita.
    - ACCESS TO FX: net exports to GDP.
    - INSTITUTIONS: exchange rate regime (defacto peg dummy), polity index.
    - CONTROLS: population; country fixed effects αi; iid error εi,t.
- Measurement highlights:
  - Inflation, depreciation, real interest rates, spreads measured in percent.
  - External debt, M2, net exports as percent of GDP.
  - KA index and polity as indices; GDP per capita in thousands of US dollars; population in millions.
- Estimation approaches:
  - Eight specifications for robustness: pooled OLS (with variations), Arellano-Bond, random effects, fixed effects with heteroskedasticity and autocorrelation corrections.
  - All regressors lagged by one period.
  - Variance inflation factors below 2.5; Condition Number 5.30 (deposit) and 5.98 (loan).
  - Wald tests to compare households vs firms.

### VI. Key econometric findings — deposits
- Market development variables are the most crucial drivers of deposit dollarization for both households and firms.
- Financial deepening (M2 to GDP):
  - One standard deviation increase in M2 to GDP (i.e., 21.7 percent) is associated with a reduction in both households’ and firms’ deposit dollarization of about 13 percentage points.
- Capital account openness:
  - Greater openness associated with larger levels of deposit dollarization for firms across specifications.
- External debt:
  - One standard deviation increase in external debt (i.e., 31.8 percent) increases deposit dollarization for households and firms (positive and significant).
- Net exports (access to FX):
  - Increasing net exports as share of GDP by one standard deviation (i.e., 16.8 percent) reduces deposit dollarization by 1.7 pps for both households and firms.
- Deposit-rate spread:
  - One standard deviation increase in the spread (i.e., 11.5 percent) decreases deposit dollarization by 1.4 pps, on average across households and firms.
- Institutional quality:
  - Better institutions consistently associated with lower household deposit dollarization; institutional quality seems irrelevant for firms’ deposit dollarization.

### VII. Key econometric findings — loans
- Currency substitution and portfolio variables largely insignificant for loan dollarization of both households and firms.
- Households:
  - M2 to GDP is the only variable consistently significant and negative across most specifications.
    - An increase in M2 to GDP by one standard deviation (i.e., 21.7 percent) decreases households’ loan dollarization by 3.4 pps.
- Firms:
  - Capital account openness:
    - An increase in capital account openness by one standard deviation (i.e., 1.4 units) represents, on average, an increase of 4.7 pps in firms’ loan dollarization.
  - Net exports:
    - An increase in net exports by one standard deviation (i.e., 16.8 percent) reduces firms’ loan dollarization by 3.2 pps in higher income countries and by 2.8 pps in lower income countries.
  - Exchange rate policy has no statistically significant impact on firms’ borrowing in FX.
- Persistence:
  - Arellano-Bond results indicate persistence: the lagged dependent variable is positive and significant.

### VIII. Income-level disaggregation (Average 2001-2016)
- Deposits — cross-income contrasts:
  - Deposits — Firms: Average 28,8 (Higher Income) / 37,2 (Lower Income); Diff -8.4 ***; S.D. 21,2 / 25,5; Obs. 774 / 552
  - Deposits — Households: Average 23,1 / 31,6; Diff -8.5 ***; S.D. 26,0 / 27,6; Obs. 817 / 589
  - Interpretation: lower income countries exhibit greater deposit dollarization for households and firms.
- Loans — cross-income contrasts:
  - Loans — Firms: Average 51,3 (Higher Income) / 37,5 (Lower Income); Diff 13.8 ***; S.D. 17,0 / 20,5; Obs. 931 / 772
  - Loans — Households: Average 15,9 / 21,2; Diff -5.3 ***; S.D. 17,8 / 20,1; Obs. 623 / 350
  - Interpretation: loan dollarization is significantly higher in higher income countries (firms), possibly reflecting FX-revenue firms and access to FX borrowing.
- Regression insights by income group:
  - Capital account openness tends to encourage higher firms’ deposit dollarization in higher income countries but reduce it in lower income countries.
  - Exchange rate regime: floating regime marginally discourages household deposit dollarization in richer countries; no impact in poorer countries.
  - Firms’ borrowing in FX positively related to income per capita in both richer and poorer countries.
  - Financial deepening effects on household loan dollarization stronger in some higher income-country specifications.

### IX. Evolution and contributors to changes (2001–2016)
- Median trends (excluding 0 and 100 percent countries):
  - Households reduced median deposit and loan FX shares between the first and second half of 2001–2016.
  - Firms’ deposit dollarization declined between halves, but firms’ FX loans increased.
- Drivers:
  - Post-crisis domestic currency depreciation and low global interest rates may have encouraged firms to borrow more in FX.
- Contributions decomposition (pooled OLS coefficients applied to observed variable changes):
  - Market development factors are the largest contributors to explaining changes in dollarization across agents and types during 2001–2016.
  - After market development, firms’ loan dollarization is explained more by access to foreign currency in higher income countries; firms’ deposit dollarization explained by currency substitution in lower income countries.
  - Institutions significantly explain changes in households’ loan dollarization in lower income countries.
  - A non-negligible portion of dollarization remains unexplained.

### X. Policy implications and recommendations
- Focus on structural factors to reduce dollarization rather than relying primarily on macroeconomic stability.
- Financial sector development:
  - Deepening local-currency lending and broadening access to local-currency finance can reduce household loan dollarization.
  - When the financial sector can lend in local currency, households tend to borrow in local currency.
- Capital account liberalization:
  - Should be carefully managed to avoid encouraging FX borrowing by non-FX-earning firms; monitor that only FX-earning companies borrow in FX.
- Institutions and credibility:
  - Strengthening institutions and monetary credibility remains important to reduce incentives for FX holdings and FX borrowing—especially for households.
- Consumer protection and regulation:
  - Household FX credit raises financial stability concerns because households typically earn in local currency and are less naturally hedged; regulatory measures and consumer protection (disclosure, limits on FX lending) may be warranted.
- Further research:
  - Micro-data with disaggregated household and firm information and inclusion of regulatory variables needed to better understand motives and refine sector-specific policies.

### XI. Collinearity and diagnostic statistics
- Variance Inflation Factors and Condition Number (selected):
  - Mean VIF: 1.47 (Loans), 1.36 (Deposits)
  - Condition Number: 5.98 (Loans), 5.30 (Deposits)
  - Individual VIFs (examples preserved): Inflation Average 1.46 (Loans), 1.53 (Deposits); M2 % of GDP 1.52 (Loans), 1.45 (Deposits); GDP p.c. 1.86 (Loans), 1.75 (Deposits)

*Source: wp1919 (IMF working paper content, using IMF IFS data for 2001–2016).*

### 1. Private Sector Financial Dollarization _____________________________________________ 12

### 1. Private Sector Financial Dollarization

### I. Introduction
- Dollarization is the use of foreign currency as a medium of exchange, store of value, or unit of account by residents, and is notable in low-income and emerging market countries of Latin America, Eastern Europe, parts of Asia, and sub-Saharan Africa (SSA).
- The paper emphasizes that while macroeconomic instability has receded in many countries and institutional credibility has improved, dollarization can persist and is often a legacy of past severe economic and political disruption.
- Research gap: the literature has focused on overall dollarization but paid scant attention to differences in determinants between household and firm dollarization.
- Objective: to provide systematic empirical analysis differentiating household and firm dollarization across countries to better inform targeted de-dollarization policies.
- Data and scope: deposit and loan dollarization for households and firms calculated from the IMF IFS database, covering the period 2001 to 2016; caveats include incomplete country coverage and measurement issues in registries.

### II. Main findings (as summarized in the source)
- Both household and firm deposit and loan dollarization tend to be largely driven by structural factors, rather than macroeconomic stability—though this may reflect the study period (2001-2016) when inflation and exchange rate policies were broadly kept in check in most countries.
- The impact of structural factors differs between households and firms, and between deposit- and loan-dollarization.

### III. Literature review — theoretical views and micro evidence
- Three principal theoretical perspectives on dollarization (Levy-Yeyati 2006):
  - Portfolio view: dollarization reflects optimal portfolio choice given returns in each currency; deposit and loan behavior should be analogous conditional on rate differentials and no regulatory restrictions.
  - Market development view: dollarization arises from market imperfections—limited current account openness, weak financial deepening (low M2/GDP), and lack of domestic investment opportunities—leading both households and firms to prefer FX assets and FX borrowing.
  - Institutional view: weak institutions and political instability undermine monetary credibility and encourage FX holdings and FX borrowing; institutional quality affects both households and firms.
- Microeconomic evidence on households:
  - Household FX borrowing is relatively understudied; available country studies (Hungary, Austria, CEECs) find heterogeneity in borrower characteristics:
    - Pellényi and Bilek (2009): FX borrowers in Hungary not statistically different in income, age, gender; FX borrowing driven mostly by macro factors (high interest spreads, stable exchange rate, foreign-owned bank competition); FX borrowers more risk averse and aware of currency risks.
    - Beer, Ongena and Peter (2010): Austrian FX borrowers tend to be less risk averse, older, better educated financially, and wealthier.
    - Fidrmuc et al. (2013): CEEC household FX loans driven by lack of trust in local currency/institutions, hedging factors (remittances, income in FX), and euro adoption expectations.
  - Existing household studies are micro-focused, country-specific, and pay less attention to macro and institutional factors.
- Microeconomic evidence on firms:
  - Firm FX borrowing often linked to currency matching (exporters with FX revenues) and firm sophistication:
    - Kedia and Mozumdar (2003); Brown, Kirschenmann and Ongena (2010): larger, older, less distressed firms demand more FX loans.
    - Brown et al. (2011): firm-level FX revenues more important than country-level interest differentials for FX borrowing.
    - Mora et al. (2013) (Lebanon): exporters and firms using formal finance more likely to have FX debt; profitable firms less likely to have FX debt.
  - Basso et al. (2011): in transition economies, increased access to foreign funds raises credit dollarization but reduces deposit dollarization; interest rate differentials matter differently for credit vs. deposits; household credit dollarization lower than corporate.
- Limitations of the existing literature include geographic concentration on Latin America and Transition economies and potential bias from EU accession and EMU expectations in CEECs.

### IV. Data and measurement (overview from the source)
- Dollarization measures: share of FX deposits/loans of total deposits/loans for households and firms (e.g., firms’ loans dollarization = firms’ loans in FX as percent of total firms’ loans).
- Database: IMF International Financial Statistics (IFS).
- Sample period: 2001 to 2016.
- Appendix contents referenced: data sources (appendix A2), country/sample details, and regression sample tables (A1.a–A1.d), summary statistics (A3), correlations (A4), collinearity analysis (A5).

### V. Implications for policy and further research (drawn from source reasoning)
- Targeted de-dollarization policies should differentiate between households and firms because:
  - Structural determinants play a major role and differ across sectors and across deposit vs. loan dollarization.
  - Firms’ FX borrowing is often linked to FX revenues (e.g., exporters) and to firm-level sophistication; policies must account for hedging needs and market structure.
  - Household FX credit raises financial stability concerns because households typically earn in local currency and are less naturally hedged; regulatory measures and consumer protection (disclosure, limits on FX lending) may be warranted.
- Policy emphasis on strengthening institutions and monetary credibility remains important to reduce incentives for FX holdings and FX borrowing.
- Market development (deepening local currency financial markets, expanding domestic savings instruments) can reduce FX demand for both deposits and loans.
- Further cross-country empirical work with disaggregated household and firm data is needed to refine sector-specific policy recommendations.

*Source: IMF working paper content (excerpt covering Introduction, Literature Review, and Dynamics sections, using IMF IFS data for 2001-2016).*

### 87.1 percent of total deposits and 92.1 percent of total loans in the financial sector, excluding

### 87.1 percent of total deposits and 92.1 percent of total loans in the financial sector, excluding the Central Banks

### Overview
- The analysis documents the composition and determinants of financial dollarization for households and firms using International Financial Statistics – IMF data and authors’ calculations, average 2001-2016.
- Effective econometric samples: 52 (39) countries for households’ deposit (loan) dollarization; 49 (63) countries for firms’ deposit (loan) dollarization. Observations with 0 or 100 percent dollarization are excluded from the effective sample.

### Financial sector composition and dollarization by sector (Average 2001-2016)
- Table 1: % of Total and Dollarization (%) (Deposits / Loans)
  - Other Financial Corporations: Deposits 5.2, Loans 2.0; Dollarization Deposits 19.2, Loans 16.3
  - State and Local Government: Deposits 1.2, Loans 0.8; Dollarization Deposits 9.3, Loans 8.0
  - Public Non-Financial Corporations: Deposits 6.5, Loans 5.2; Dollarization Deposits 23.7, Loans 20.9
  - Private Sector: Firms: Deposits 30.9, Loans 75.0; Dollarization Deposits 30.7, Loans 47.7
  - Private Sector: Households: Deposits 56.2, Loans 17.1; Dollarization Deposits 26.8, Loans 21.0
- Aggregate statements in the text:
  - Firms represent 87.1 percent of total deposits and 92.1 percent of total loans in the financial sector, excluding the Central Banks. (text context)
  - In terms of dollarization, they represented 57.5 percent of deposits and 68.7 percent of deposits and loans respectively (Table 1 text).

### Evolution of dollarization (2001–2016)
- Excluding countries with 0 and 100 percent dollarization:
  - Households reduced their median level of deposits and loans denominated in FX between the first and second half of 2001–2016.
  - Firms’ deposit dollarization also declined between halves, but firms’ FX loans increased.
- Possible drivers discussed:
  - Post-crisis depreciation of domestic currencies for many developing and transition economies could explain expansion in firms’ loan dollarization.
  - Low interest rate environment following the global financial crisis and “search for yield” may have encouraged local companies to borrow in foreign currency.

### Differences of dollarization by level of income (Average 2001-2016)
- Table 2: Financial Dollarization by Level of Income (Higher Income / Lower Income / Diff)
  - Deposits — Firms: Average 28,8 / 37,2 / Diff -8.4 ***
    - S.D. 21,2 / 25,5; Obs. 774 / 552
  - Deposits — Households: Average 23,1 / 31,6 / Diff -8.5 ***
    - S.D. 26,0 / 27,6; Obs. 817 / 589
  - Loans — Firms: Average 51,3 / 37,5 / Diff 13.8 ***
    - S.D. 17,0 / 20,5; Obs. 931 / 772
  - Loans — Households: Average 15,9 / 21,2 / Diff -5.3 ***
    - S.D. 17,8 / 20,1; Obs. 623 / 350
  - Significance: *** p<0.01, ** p<0.05, * p<0.1
- Interpretation:
  - Lower income countries show greater deposit dollarization for households and firms (likely reflecting hedging and store-of-value behavior following episodes such as hyperinflation).
  - Loan dollarization is significantly higher in higher income countries (possibly reflecting large companies with FX revenues and access to FX borrowing).

### Correlations with macroeconomic variables (Figures 3 and 4; avg. 2001–16)
- Firms:
  - Dollarization of deposits and loans mainly correlated with financial market development (proxied by M2 to GDP) for both higher and lower income countries.
  - Correlation between firms’ loan dollarization and inflation stronger for higher income countries; correlation of firms’ deposit dollarization and inflation stronger for lower income countries.
- Households:
  - Household loan dollarization shows larger correlations with macroeconomic variables in lower income countries than in higher income countries.
  - Household deposits dollarization appears strongly correlated with financial market development.
- Representative regression lines and R² values shown in figures (examples preserved in figures):
  - Firms Loans vs Inflation: y = -0.23x + 39.7, R² = 0.005 (Lower Income); y = -0.9x + 56.6, R² = 0.125 (Higher Income)
  - Firms Deposits vs M2 to GDP: y = -0.45x + 53.1, R² = 0.151 (Lower Income); y = -0.37x + 47.7, R² = 0.309 (Higher Income)
  - Households Loans vs M2 to GDP: y = -0.22x + 27.7, R² = 0.164 (Lower Income); y = -0.13x + 22.9, R² = 0.052 (Higher Income)
  - Households Deposits vs M2 to GDP: y = -0.55x + 53.4, R² = 0.177 (Lower Income); y = -0.54x + 51.6, R² = 0.390 (Higher Income)

### Determinants: data, baseline model, and estimation strategy
- Baseline panel model (notation preserved qualitatively):
  - Dependent variable: dollarization (deposits or loans in foreign currency as percent of total) for households and firms.
  - Key explanatory groups:
    - CURRENCY: inflation rate and nominal exchange rate depreciation against the US dollar (currency substitution dimension).
    - PORTFOLIO: deposit or loan interest rate spreads to the US dollar and relevant real interest rates (portfolio optimization considerations).
    - MARKET DEVELOPMENT (MKTDEV): Chinn-Ito index of capital account restrictions, external debt to GDP, M2 to GDP ratios, GDP per capita.
    - ACCESS TO FX: net exports to GDP (ease of access to foreign exchange).
    - INSTITUTIONS: exchange rate regime, political freedom indicators.
    - CONTROLS: population size; country fixed effects αi; iid error εi,t.
- Variable measurement:
  - Inflation, exchange rate depreciation, real interest rates, interest rate spreads measured in percent.
  - External debt, M2, net exports measured as percent of GDP.
  - KA index, polity as indices; income level as dummy.
  - GDP per capita measured in thousands of US dollars; population in millions.
- Estimation approaches:
  - Eight model specifications for robustness: pooled OLS (with time dummies, trend, income dummy, interaction), Arellano-Bond (model 6) including lagged dependent variable, random effects (model 7), fixed effects (model 8) with heteroskedasticity and autocorrelation corrections.
  - All regressors lagged by one period to reduce endogeneity concerns.
  - Variance inflation factors remain below 2.5; condition number tests 5.30 (deposit) and 5.98 (loan), suggesting no strong multicollinearity.
  - Wald tests used to compare coefficients between households and firms.

### Key econometric findings (deposit dollarization focus; households vs. firms)
- Market development variables are the most crucial drivers of deposit dollarization for both households and firms.
- Financial deepening (M2 to GDP) effect:
  - One standard deviation increase in M2 to GDP (i.e., 21.7 percent) is associated with a reduction in both households’ and firms’ deposit dollarization of about 13 percentage points.
- Capital account openness:
  - Less restrictions (greater openness) associated with larger levels of deposit dollarization for firms across specifications.
- External debt:
  - Larger levels of external debt as percent of GDP related to greater levels of deposit dollarization.
  - One standard deviation increase in external debt (i.e., 31.8 percent) increases deposit dollarization for households and firms (sentence in source ends detailing magnitude; results discussed as positive and significant).

### Interpretation and remaining gaps
- Goodness of fit measures are relatively low, indicating a significant portion of dollarization remains unexplained by the model.
- Micro-data factors (e.g., financial sophistication, age of economic agents) that likely influence deposit and loan dollarization are not captured in the model.
- Arellano-Bond results indicate persistence in dollarization: the lagged dependent variable is positive and significant, reducing the explanatory power of other slow-moving factors.

*Source: International Financial Statistics – IMF, author’s calculations*

### 3.2 pps. Net exports as percent of GDP, a proxy for access to foreign exchange, appears as an

### wp1919 - 3.2 pps. Net exports as percent of GDP, a proxy for access to foreign exchange, appears as an

### Key empirical findings on deposit dollarization
- Net exports as percent of GDP, a proxy for access to foreign exchange, is an important factor reducing deposit dollarization for both households and firms.
  - Increasing net exports as share of GDP by one standard deviation (i.e. 16.8 percent) reduces dollarization by 1.7 pps.
- The spread between domestic and US deposit rates appears significant in the non-Pooled OLS models.
  - A one standard deviation increase in the spread (i.e. 11.5 percent) decreases deposit dollarization by 1.4 pps, on average across households and firms.
- Currency substitution considerations are relevant in random and fixed effects models, but the currency substitution model does not appear to be an important explanatory factor in explaining deposit dollarization of households and firms.
- For firms, depreciation against the US dollar increases deposit dollarization significantly.
- For households, inflation appears relevant but results are not statistically strong; the period covered was broadly disinflationary, reducing inflation’s observable impact on deposit dollarization.
- Individuals may hedge through means other than FX deposits (e.g., buying real assets such as houses, cars).

- Institutional quality shows a divergence between agents:
  - For households, better institutions are consistently associated with lower levels of deposit dollarization across models.
  - For firms, institutional quality seems irrelevant for deposit dollarization, possibly because firms can hedge risks or because financial market development variables capture relevant institutional conditions for firms.

### Loan dollarization: contrasts between households and firms
- Currency substitution variables are statistically insignificant for loan dollarization of both households and firms, suggesting borrowing in FX is not directly determined by inflation or exchange rate depreciation.
- Portfolio view variables (relative cost of borrowing in FX) are largely insignificant for loan dollarization.
- Households:
  - M2 to GDP ratio is the only variable consistently significant and negative across almost all model specifications.
    - An increase in M2 to GDP by one standard deviation (i.e. 21.7 percent) decreases households’ loan dollarization by 3.4 pps.
  - Financial deepening (greater M2 to GDP) encourages borrowing in local currency.
- Firms:
  - Capital account openness exhibits explanatory power for firms’ loan dollarization.
    - An increase in capital account openness by one standard deviation (i.e. 1.4 units) represents, on average, an increase of 4.7 pps in firms’ loan dollarization.
  - Net exports as percent of GDP is significant and negatively signed for firms’ loan dollarization across models:
    - An increase in net exports by one standard deviation (i.e. 16.8 percent) reduces dollarization of firms’ loans by 3.2 pps in higher income countries and by 2.8 pps in lower income countries.
  - The exchange rate policy has no statistically significant impact on firms’ borrowing in FX.

### Income-level disaggregation: higher versus lower income countries
- Deposit dollarization:
  - Income-level dummy is not significant, though sign is negative (deposit dollarization lower in higher income countries).
  - Inflation significantly increases deposit dollarization of firms in lower income countries under most specifications.
  - The deposit-rate spread is negatively correlated with dollar holdings for both households and firms in both higher and lower income countries; coefficients are larger in higher income countries.
  - Capital account openness:
    - In higher income countries, more open capital account encourages higher dollarization of firms’ deposits.
    - In lower income countries, a more open capital account reduces firms’ deposit dollarization.
  - Exchange rate regime:
    - For households, floating exchange rate marginally discourages deposit dollarization in richer countries but has no impact in poorer countries.
- Loan dollarization:
  - Firms’ borrowing in FX is positively related to income per capita in both richer and poorer countries (as income rises, firms borrow more in FX); this is not the case for households.
  - Currency substitution and portfolio variables are largely insignificant across income groups.
  - Real lending rate plays a role for firms’ loan dollarization in lower income countries when country effects are considered.
  - Financial deepening (M2 to GDP) is statistically significant mainly for higher income countries’ households’ loan dollarization in some specifications.
  - Access to FX finance (net exports as percent of GDP) consistently determines firms’ loan dollarization in both higher and lower income countries.

### Contributions to changes in dollarization (2001–2016)
- Using pooled OLS coefficient estimates and observed variation in explanatory variables:
  - Market development factors are the largest contributors explaining changes in dollarization across agents and types (loans and deposits) in the period analyzed (2001-2016).
- Differences between income groups:
  - After market development factors, firms’ loans dollarization is explained more by access to foreign currency in higher income countries.
  - Firms’ deposits dollarization is explained by currency substitution factors mainly in lower income countries.
  - Institutions significantly explain changes in households’ loans dollarization in lower income countries; in higher income countries, non-market-development factors make small contributions.
  - Households’ deposit dollarization is explained by access to foreign currency mainly in higher income countries; in lower income countries portfolio model factors are also important.
- A non-negligible portion of dollarization remains unexplained by the variables considered, indicating unique features not captured by traditional explanatory variables.

### Main conclusions and policy implications
- Summary conclusions:
  (i) The currency substitution model (loan and deposit dollarization rising with higher inflation and nominal depreciation) is weak in explaining both household and firm dollarization in the sample.
  (ii) Structural factors measured by the market development model better explain dollarization behavior of households and firms, though impacts differ across agents.
    - Capital account liberalization tends to encourage deposit dollarization by firms (and to a lesser extent households); opening the capital account should be monitored to ensure only FX-earning companies borrow in FX.
  (iii) Financial sector development tends to have a negative impact on deposit dollarization of firms (and to a lesser extent households). Household borrowing in FX is negatively impacted by financial sector development.
    - When the financial sector can lend in local currency, households tend to borrow in local currency.
  (iv) Access to foreign exchange negatively impacts loan dollarization of firms, but not households.
  (v) A floating exchange rate regime has little impact on deposit dollarization for either agent, but it has a negative impact on household loan dollarization (discourages household borrowing in FX), while not affecting firms’ borrowing in FX.
  (vi) Institutional quality has no visible impact on firms’ deposit dollarization but significantly reduces households’ deposit dollarization; better institutions encourage firms to borrow in FX but do not affect household FX borrowing.
- Policy implications:
  - To reduce dollarization of deposits and loans, focus on structural factors—such as developing the financial sector—rather than relying primarily on macroeconomic stability.
  - Capital account liberalization should be carefully managed to avoid encouraging FX borrowing by non-FX-earning firms.
  - Deepening local-currency lending and broadening access to local-currency finance can reduce household loan dollarization.
  - Additional research including regulatory variables is needed to better understand motives behind household and firm dollarization.

*Source: wp1919 (selected section).*

### Appendix A

### Appendix A

### Effective regression samples

- Table A1.a. Effective Regression Sample for Households Deposits Dollarization
  - Haiti; Nepal; Congo, Dem. Rep. of; Mozambique; Rwanda; Sierra Leone; Tanzania; Uganda; Cambodia; Indonesia; Mongolia; Solomon Islands; Armenia; Kyrgyz Republic; Moldova; Tajikistan; Ukraine; Bolivia; Guatemala; Honduras; Nicaragua; Egypt; Morocco; Bangladesh; Bhutan; Pakistan; Sri Lanka; Cape Verde; Ghana; Kenya; Nigeria; Zambia; Fiji; Malaysia; Albania; Azerbaijan, Rep. of; Belarus; Bosnia & Herzegovina; Bulgaria; Georgia; Macedonia, FYR; Turkey; Costa Rica; Dominican Republic; Guyana; Jamaica; Mexico; Paraguay; Venezuela, Rep. Bol.; Algeria; Angola; South Africa

- Table A1.b. Effective Regression Sample for Households Loans Dollarization
  - Haiti; Nepal; Congo, Dem. Rep. of; Mozambique; Rwanda; Tanzania; Uganda; Indonesia; Mongolia; Armenia; Kyrgyz Republic; Moldova; Tajikistan; Ukraine; Bolivia; Guatemala; Honduras; Nicaragua; Egypt; Morocco; Sri Lanka; Zambia; Malaysia; Albania; Azerbaijan, Rep. of; Belarus; Bosnia & Herzegovina; Bulgaria; Georgia; Macedonia, FYR; Colombia; Costa Rica; Dominican Republic; Jamaica; Mexico; Venezuela, Rep. Bol.; Algeria; Angola; South Africa

- Table A1.c. Effective Regression Sample for Firms Deposits Dollarization
  - Haiti; Nepal; Congo, Dem. Rep. of; Mozambique; Rwanda; Sierra Leone; Tanzania; Uganda; Cambodia; Indonesia; Mongolia; Solomon Islands; Armenia; Kyrgyz Republic; Moldova; Tajikistan; Ukraine; Bolivia; Guatemala; Honduras; Nicaragua; Egypt; Morocco; Bangladesh; Pakistan; Sri Lanka; Cape Verde; Ghana; Kenya; Nigeria; Zambia; Fiji; Malaysia; Albania; Azerbaijan, Rep. of; Belarus; Bosnia & Herzegovina; Bulgaria; Georgia; Macedonia, FYR; Turkey; Dominican Republic; Guyana; Jamaica; Mexico; Venezuela, Rep. Bol.; Algeria; Angola; South Africa

- Table A1.d. Effective Regression Sample for Firms Loans Dollarization
  - Haiti; Nepal; Benin; Burkina Faso; Burundi; Comoros; Congo, Dem. Rep. of; Guinea-Bissau; Mali; Mozambique; Niger; Rwanda; Senegal; Sierra Leone; Tanzania; Togo; Uganda; Indonesia; Mongolia; Solomon Islands; Armenia; Kyrgyz Republic; Moldova; Tajikistan; Ukraine; Bolivia; Guatemala; Honduras; Egypt; Morocco; Bangladesh; Bhutan; Sri Lanka; Cameroon; Cape Verde; Congo, Republic of; Cote d'Ivoire; Kenya; Lesotho; Swaziland; Zambia; Fiji; Malaysia; Albania; Azerbaijan, Rep. of; Belarus; Bosnia & Herzegovina; Bulgaria; Georgia; Macedonia, FYR; Brazil; Colombia; Costa Rica; Dominican Republic; Guyana; Jamaica; Mexico; Venezuela, Rep. Bol.; Algeria; Angola; Botswana; Gabon; South Africa

### Sources of variables (Table A2)

- Deposits/loans dollarisation (%) — International Financial Statistics (IFS)—IMF
- Inflation (%) — World Economic Outlook Database—IMF
- Nominal depreciation against USD (%) — International Financial Statistics (IFS)—IMF
- Real deposit/lending rate (%) — International Financial Statistics (IFS)—IMF
- Deposit/lending rate spread to USA (%) — International Financial Statistics (IFS)—IMF
- Capital Account Openness Index—KA Index — Chinn-Ito Index—2015
- External debt (% of GDP) — World Economic Outlook Database—IMF
- M2 (% of GDP) — International Financial Statistics (IFS)—IMF
- GDP per Capita (thousand USD) — World Economic Outlook Database—IMF
- Net exports (% of GDP) — World Economic Outlook Database—IMF
- Defacto Peg (dummy) — AREAER Database—IMF
- Polity index — Polity IV database
- Population (millions) — World Economic Outlook Database—IMF

### Collinearity analysis (Table A5)

- Variance Inflation Factors and Condition Number (columns: Loans, Deposits)
  - Inflation - Average: 1.46, 1.53
  - Depreciation against USD - average: 1.33, 1.32
  - Lending rate spread to USD: 1.88, 
  - Real lending rate: i(t) - infl(t+1): 1.94, 
  - Deposit rate spread to USD: , 1.30
  - Real deposit rate: i(t) - infl(t+1): , 1.41
  - Chinn-Ito index: 1.16, 1.14
  - External Debt % of GDP: 1.18, 1.17
  - M2 % of GDP: 1.52, 1.45
  - GDP p.c.: 1.86, 1.75
  - Net Exports % of GDP: 1.50, 1.45
  - Dummy: Defacto floating regime = 1: 1.26, 1.26
  - Polity Score: 1.15, 1.16
  - Population: 1.36, 1.34
  - Mean VIF: 1.47, 1.36
  - Condition Number: 5.98, 5.30

### Other tables noted

- Table A3. Summary Statistics of Variables (listed, no numeric content presented)
- Table A4. Independent Variables Correlation Matrix (listed, no numeric content presented)

*Appendix A, wp1919 - Appendix A*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wp1919.pdf_