## wpiea2019211-print-pdf

## Source details

**Canonical URL:** [wpiea2019211-print-pdf](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019211-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019211-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019211-print-pdf.pdf.json)

---

### Main findings and summary
- Interest rate restrictions reduce growth by about 0.4-0.7 percentage points, with the effect being larger in economies with larger financial systems.  
- Full liberalization is necessary to significantly increase growth; changes in interest rate restrictions short of full liberalization have a limited impact.  
- Regional differences: effect appears strongest in sub-Saharan Africa (SSA), the Middle East and North Africa (MENA), and in transition countries; insignificant in Asia and in advanced economies.  
- Financial repression reduces the probability of crisis (short-term stability benefit), but this positive effect is much smaller than the adverse direct effect on growth; on net, financial repression has a significant adverse effect on growth.  
- Aggregate-sample implications:
  - Financial repression may have reduced real per capita growth, on average, by 26-27 basis points (Table 13, total effect on sample averages).  
  - Alternative augmented approach yields estimated average reduction in growth of 17-18 basis points, of which 11-12 basis points may be due to the indirect effect through crisis risk (Table 14).

### Definition, forms, and objectives of financial repression
- Definition: direct government intervention that alters the equilibrium in the financial sector, typically to provide cheap loans to companies and governments by lowering returns to savers below market levels.  
- Major forms listed:
  - Interest rate controls (ceilings or floors on lending and deposit rates);  
  - Directed lending (mandatory allocations);  
  - Restrictions on international capital movements;  
  - Restrictions to entry into the banking sector and state-owned banks;  
  - Unconventional monetary policies flattening the interest-rate curve.  
- Stated policy objectives:
  - (a) public financing through seigniorage by maintaining real interest rates below market equilibrium;  
  - (b) subsidizing sectors/industries via low-cost credit from “captive” savers;  
  - (c) maintaining financial stability by creating a predictable credit environment.

### Stylized theoretical mechanisms (three-agent model and Appendix II)
- Agents: savers, borrowers, financial intermediaries. Core relations (preserved notation):
  - D_L = a_L − b_L i_L  (A.1)  
  - S_D = a_D + b_D i_D  (A.2)  
  - L_S = k D_D with k = 1 − r  (A.3)  
  - i_L − i_D = p + q L  (A.4)
- Key effects of loan-rate ceiling i_L^C (binding if i_L^C < i_L*):
  - Excess demand for loans, reduced deposit rates, lower deposit supply, credit rationing (A.11–A.14).  
  - Deposit-market derived demand becomes kinked; binding ceiling induces financial disintermediation and reduced savings channeled through banks.  
- Key effects of deposit-rate ceiling i_D^C:
  - Capped deposit supply (A.15), capped loan supply (A.16), banking rents due to spread (A.17–A.19), rationing in deposit market, non-interest competition, deadweight losses from withdrawals.  
- Dual ceilings:
  - Both bind only if i_L^C lies within specified range (A.20); rents split between banks and approved borrowers depending on i_L^C position.  
- Distributional and welfare implications:
  - Transfers from savers to beneficiary borrowers and banks (quasi-fiscal); deadweight losses from excluded depositors and borrowers; misallocation and rent-seeking.

### Data, measurement, and sample
- Primary IRC variable: index of “interest rate controls” (IRC) with values 0–3:
  - 0 = strictest controls; 1 = extensive but not universal; 2 = binding constraints apply to a significant share; 3 = banks essentially free (full liberalization).  
- Binary Financial Restrictions Index (FRI) mapping:
  - FRI = 0 if IRC = 3; FRI = 1 if IRC < 3.  
- Coverage: annual data for 90 economies over 45 years, 1973 to 2017 (89 countries and one Special Administrative Region). Notes:
  - China data available from 1981.  
  - For 18 countries that became independent or joined Fund between 1990 and 1993, data available for 24-27 years.  
- 2017 average IRC scores reported:
  - G-7 countries: 3  
  - LA-5 countries: 2.8  
  - sub-Saharan countries: 2.4
- Other data sources: IMF WEO, IFS, Penn World Tables, IMF Financial Soundness Database, World Bank WDI, ICRG, IMF Strategy and Policy Review Department debt crisis indicator.  
- Crisis definition: high inflation, high risk premia on debt, arrears on external debt, debt restructuring, or receiving emergency official financial assistance.

### Descriptive patterns and episodes
- Historical liberalization wave: roughly 1984–1996:
  - Until 1984, three-fourth of jurisdictions had some form of interest rate restrictions; by 1991 ratio fell below one-half; after 1995 less than one-quarter maintained restrictions; by 1999 ratio stabilized around 17-18 percent.  
- Episode identification (Gourinchas and Obstfeld 2012 method): seven-year window centered on t = 0 when IRC increases from <3 to 3 (liberalization) or declines 3→<3 (introduction of controls).  
- Episodes summary:
  - 71 jurisdictions experienced episodes of full liberalization; 9 maintained restrictions throughout; 10 always liberal.  
  - Identified 84 episodes of full liberalization, 568 observations out of 3,718, mostly between 1985 and 1995.  
- Unconditional descriptive contrast (across all observations):
  - Per capita growth has been 0.52 percentage points higher in the absence of interest rate restrictions.  
- Selected average characteristics by group (jurisdictions that liberalized at some point; restricted throughout; liberal throughout):
  - Number of jurisdictions: 71; 9; 10.  
  - Real GDP growth (percent): 3.4; 5.0; 2.1.  
  - Real per capita growth (percent): 1.9; 3.3; 2.3.  
  - Inflation (percent): 44.3; 32.7; 73.7.  
  - Excluding high-inflation cases: 7.0; 6.5; 4.5.  
  - Public debt (percent of GDP): 31.6; 25.3; 37.7.  
  - External debt (percent of GDP): 46.9; 31.3; 54.2.  
  - Per capita income (U.S. dollars): 9,161; 1,177; 19,246.

### Outcomes around liberalization episodes (3 years before vs. 3 years after)
- Key comparisons (three preceding years / three following years / liberalized regime / restricted regime where indicated):
  - Real per capita growth (percent): 0.6 / 2.3 / 2.3 / 2.0.  
  - Net capital formation (percent of GDP): 20.3 / 21.5 / 22.1 / 20.2.  
  - Savings (percent of GDP): 20.3 / 20.2 / 21.8 / 18.7.  
  - Private sector credit (percent of GDP): 28.7 / 30.8 / 56.9 / 28.4.  
  - Net capital formation (real annual growth in percent): -13.5 / 5.6 / 46.8 / 4.9.  
  - Private sector credit (real annual growth in percent): 97.9 / 9.7 / 7.9 / -2.0.  
  - Inflation (percent): 143.8 / 22.2 / 19.5 / 54.3.  
  - Probability of the start of a recession (percent): 16.2 / 13.5 / 9.8 / 12.4.  
  - Percent of years in a recession: 34.5 / 23.4 / 16.5 / 28.5.  
  - Probability of the start of a debt crisis (percent): 5.0 / 2.5 / 2.8 / 1.2.  
  - Percent of years in a debt crisis: 10.0 / 15.6 / 9.9 / 3.8.  
  - Fiscal deficit: -1.6 / -1.4 / -1.7 / -1.1.  
  - Public debt: 19.3 / 25.9 / 46.8 / 14.5.  
  - Current account balance: -3.2 / -3.0 / -1.0 / -3.1.  
  - External debt: 45.3 / 39.3 / 55.9 / 34.5.  
  - Per capita income (U.S. dollars): 4,313 / 5,685 / 15,900 / 2,253.
- Patterns noted:
  - Growth increases markedly after liberalization (unconditional and conditional estimates).  
  - No significant change in savings, net capital formation, or private sector credit as percent of GDP; real growth in net capital formation appears to increase while private sector credit declines.  
  - Liberalization associated with lower inflation; many liberalizations occurred during very high inflation episodes.  
  - A crisis is more likely to start in the three years preceding liberalization than in the three years following it; about three-fourth of crises in sample were accompanied or preceded by a change in interest rate regime (usually easing).

### Introduction of controls (tightening episodes)
- Episodes: 30 episodes in 30 jurisdictions, total 203 observations; episodes cluster around 1981-85, 2001-08, and 2013-15; sample includes 29 countries and one Special Administrative Region.  
- Macroeconomic behavior around introduction of controls:
  - Inflation, fiscal deficit, external debt, and the probability of crisis increase after introduction of interest rate controls.  
  - Private sector credit increases in percent of GDP after introduction of controls, but real net capital formation (investment) appears to slow.  
  - Net change in real per capita growth visible at the median, less evident at top and bottom quartiles.  
  - Probability that a crisis may start (or continue) increases—at least temporarily—after controls are introduced.
- Key numeric observations (three preceding years / three following years / values outside episodes / restricted vs liberalized regime):
  - Real per capita growth (percent): 3.3 / 2.9 / 2.3 / 1.7.  
  - Net capital formation (percent of GDP): 21.4 / 22.4 / 22.0 / 20.0.  
  - Savings (percent of GDP): 20.7 / 20.5 / 21.7 / 18.9.  
  - Private sector credit (percent of GDP): 37.2 / 43.1 / 53.5 / 27.5.  
  - Net capital formation (real annual growth in percent): 8.1 / 7.2 / 40.4 / 1.9.  
  - Private sector credit (real annual growth in percent): 19.9 / -2.3 / 8.3 / 14.5.  
  - Inflation (percent): 16.2 / 38.7 / 30.3 / 70.3.  
  - Probability of the start of a debt crisis (percent): 1.1 / 3.5 / 3.0 / 1.5.  
  - Percent of years in a debt crisis: 5.7 / 8.2 / 11.1 / 4.4.  
  - Fiscal balance: -0.9 / -1.2 / -1.7 / -1.1.  
  - Public debt: 26.6 / 27.4 / 44.4 / 14.4.  
  - Current account balance: -1.5 / -3.1 / -1.3 / -3.2.  
  - External debt: 32.8 / 37.1 / 54.3 / 36.2.  
  - Per capita income (U.S. dollars): 3,927 / 4,951 / 14,807 / 2,410.

### Panel data estimates — direct effects on growth and specification
- Empirical strategy: augmented growth regressions on panel data (90 jurisdictions, 1973–2017) with macro and institutional controls; episode-based event dummies and conditional regressions with control variables.  
- Unconditional model: y_ct = a_c + sum_{z=-3}^{3} b_z d_zt + epsilon_ct.  
- Conditional model adds K controls: y_ct = a_c + sum_{z=-3}^{3} b_z d_zt + sum_k gamma_k x_kct + epsilon_ct.  
- Main panel results:
  - Presence of interest rate controls (FRI = 1) reduces growth by about 0.6-0.7 percentage points (equivalent to 28-33 percent of the average growth rate of the entire sample, 2.1 percent).  
  - Liberalization in the previous three years has milder impact (~0.5 percentage points), implying lagged effects.  
  - Changes among degrees of restrictions matter less than changes between presence and absence (binary transition).  
  - Debt/crisis dummy: GDP growth on average 2 percentage points lower at times of crises.  
  - Inflation negative effect; population growth negative (coefficient above -1); investment, trade openness, fiscal balance positive; external debt negative.
- Table 6 (entire sample, fixed effects, 1973-2017) summary:
  - Observations: 2,880.  
  - Number of jurisdictions: 86.  
  - R-squared range across specifications: 0.1935 to 0.1986.  
  - Interest rate controls explain about 0.3-0.4 percent of the variance (about 1.5-2 percent of explained variance).  
  - Within-country variance explained: about 19-20 percent; entire-sample variance explained: about 14-15 percent.
- Notable coefficient magnitudes (example column (1)):
  - FRI (lagged): -0.6485*** (standard error 0.2017).  
  - Debt crisis: -2.0014*** (standard error 0.2586).  
  - Inflation: -0.0058*** (standard error 0.0007).  
  - Population growth (percent): -0.9705*** (standard error 0.0975).  
  - Fixed capital formation: 0.1677*** (standard error 0.0148).  
  - Public debt: 0.0089*** (standard error 0.0027).

### Heterogeneity and robustness checks
- Stronger negative relation between repression and growth since early 1990s; in some specifications controls associated with about 0.8 percentage points lower real GDP growth.  
- Effects strongest and most significant in SSA, MENA, and transition economies; insignificant in Asia and advanced economies.  
- Robustness: results generally robust across fixed/time effects, random effects, Arellano-Bond dynamic models, five-year averages, Penn World Tables specifications, and treatment-effect matching—though statistical significance varies.  
- Adding World Governance Indicators reduces sample size and weakens significance but FRI coefficients remain negative and generally significant.

### Transmission channels (Table 10)
- Savings (share of GDP) is significantly negatively related to presence of interest rate controls.  
- Private sector credit (percent of GDP and real annual growth) is negatively related to financial repression and to recent removal of controls.  
- Growth in total factor productivity adversely related to presence of interest rate controls, but not to short-term regime changes.  
- Interpretation: long-term harm via reduced savings, reduced credit, and misallocation (lower TFP); medium-term easing may not immediately boost savings and can temporarily reduce private credit.

### Financial repression and crises — indirect effects
- Probit panel estimates: FRI significant and negative in probit for debt crisis probability, suggesting restrictions reduce debt crisis risk in a period.  
- Estimated marginal effect: financial repression could have reduced probability of a debt crisis by about 1 percentage point (computed at sample averages).  
- Indirect effect on growth via crisis channel: about 0.02 percentage points (positive), negligible relative to direct negative effect of 0.6-0.7 percentage points.  
- Alternative augmented approach (adding estimated impact of repression on crisis risk per observation into growth regression) yields average reduction in growth of 17-18 basis points, of which 11-12 basis points may be due to the indirect effect through crisis risk.  
- In some specifications adding indirect impact renders direct coefficient on financial repression statistically insignificant.

### Case studies: Kenya and Bolivia
- Kenya (controls reintroduced September 2016; law ruled unconstitutional March 14, 2019, with 12 months to reconsider):
  - Lending rate ceiling: 4 percent above a reference rate; floor on time deposits equal to 70 percent of the reference rate.  
  - Documented effects:
    - Stock of credit to SMEs fell by about 10 percent in one year.  
    - Banks increased lending to corporates and households while reducing SME lending.  
    - Stock of credit of small banks declined by about 5 percent within a year.  
    - Shift of credit toward public sector; shift from time deposits to demand deposits.  
    - Average profit margins on private sector lending declined further and turned negative.  
    - Policy appears to have contributed to cyclical increase in non-performing loans by incentivizing short-term lending.  
  - Law under revision following court ruling; controls remained during reconsideration.
- Bolivia (Financial Services Law, 2013; implementing decrees 2015 onwards):
  - Regulated lending rates, lending targets (productive sectors and social housing), discretionary deposit-rate floors for accounts ≤ Bs.70,000 (about $10,000).  
  - Ceilings in sectors: social housing (5.5–6.5 percent), productive sectors (6–7 percent), microfinance (11.5 percent).  
  - Portfolio quotas: full-service banks to commit at least 60 percent of loan portfolios to social housing and productive sectors (50 percent for SMEs) by end-2018.  
  - Observed effects:
    - Rapid credit growth in 2015-17 as banks met quotas; credit growth moderated later.  
    - Declining bank profitability (earnings on assets and equity) as spreads squeezed.  
    - Declining capital adequacy ratios (partly offset by profit-retention requirement); ratios still appear adequate.  
    - Non-performing loans low and well provisioned, but increases in restructured loans warrant monitoring.  
    - Number of borrowers < $5,000 decreased by 10 percent from Sept 2014 to June 2018; average loan amounts increased—suggesting some small borrowers excluded despite inclusion goals.

### Policy-relevant implications and conclusions
- Financial repression operates as a quasi-fiscal operation taxing some agents to finance public objectives; rational only if benefits exceed costs and compared with alternatives.  
- Main costs summarized:
  - Compresses return to savers → suboptimal savings → reduces funds for investment and access to financing.  
  - Increases demand for credit via lower lending rates → credit rationing on non-economic criteria → inefficiencies.  
  - Reduces bank profitability (if deposit floors), impairing capital accumulation and potentially raising stability risks.  
  - Distorts investment allocation → lower average quality and rate of return; encourages rent-seeking and possible corruption.  
- Empirical summary:
  - On average, financial repression reduces growth by about 0.4-0.7 percentage points (robust across models).  
  - Effects stronger in SSA, MENA, LAC, and transition countries; insignificant in Asia and advanced economies.  
  - Partial easing short of full liberalization has limited impact on growth; full reform appears necessary to materially enhance growth.  
  - Financial repression reduces period probability of debt crisis (short-term stability gain), but this indirect positive effect is far smaller than the direct negative effect on growth.  
  - Partial loosening without full liberalization may raise crisis risk—arguing to strengthen supervision prior to partial reforms.  
- Final judgment: restrictions have a significant adverse effect on growth and on access to financing; in the long term, countries would generally be better off without financial repression.

### Appendices and literature highlights
- Appendix I — Review of the Literature:
  - Traces evolution from 1960s Keynesian support for intervention to McKinnon-Shaw critique in 1970s, to information-frictions literature (Stiglitz-Weiss) in 1980s, endogenous-growth perspectives in 1990s, and mixed/conditional empirical findings since 2000.  
  - Measurement evolution: from binary indices to multi-component indices covering credit controls, interest-rate controls, entry barriers, regulation, capital flow restrictions, privatizations, securities policy, and prudential regulations.  
  - Literature remains inconclusive on net welfare and crisis effects; debate on whether prudential safeguards are necessary precursors to liberalization.
- Appendix II — Stylized model: presents linear demand/supply relations, equilibrium expressions (A.5–A.10), and analytical results for loan and deposit ceilings (A.11–A.20) including rent decomposition and welfare implications.
- Appendix Table IV.4 (Penn World Tables data) — selected regression notes:
  - Interest rate controls coefficients reported across columns, e.g., -0.0098*** (0.0016) in column (1) and -0.0150*** (0.0029) in column (8).  
  - Growth of real physical capital coefficients range from 0.3545*** (0.0444) to 0.7795*** (0.0631) across columns.  
  - Interaction terms and significance codes preserved; observations vary across specifications (e.g., 3,244; 902; 1,247; 2,944).  
  - Standard errors reported in parentheses; significance codes: *** p<0.01, ** p<0.05, * p<0.1.

*Source: wpiea2019211-print-pdf — chapter and appendices summarized above.*

### References _________________________________________________________________________ 62

### wpiea2019211-print-pdf - References

### Figures and Tables Inventory
- Figures listed:
  - 1. Restricted and Unrestricted Equilibrium in Credit and Deposit Markets
  - 2. Jurisdictions with Interest Rate Controls, 1973-2017
  - 3. Average GDP Growth by Average FRI (By Year)
  - 4. GDP Growth and Interest Rate Regime, 1973-2017
  - 5. Economic Growth During Liberalization Episodes
  - 6. Conditions at Around Liberalization Episodes
  - 7. Episodes of Introduction of Controls, 1973-2017
  - 8. Economic Growth After the Introduction of Controls
  - 9. Probability of Crisis (Around the Introduction of Controls)
  - 10: Kenya: Selected Financial Indicators
  - 11. Bolivia: Selected Financial Indicators
- Tables listed:
  - 1. FRI Score by Region and Year
  - 2. Correlation: Financial Repression Index (FRI) and Selected Macroeconomic Variables
  - 3. Full Liberalization Episode
  - 4. Liberalization Episodes: Before and After
  - 5. Tightening Episodes: Before and After
  - 6. Panel Data Estimates (Entire Sample, Fixed Effects, 1973-2017)
  - 7. Impact of Financial Repression on Growth—FRI Coefficient by Period
  - 8. Impact of Financial Repression on Growth—FRI Coefficient by region
  - 10. Transmission Channels
  - 11. Probability of Crisis
  - 12. Probit Panel Data Results
  - 13. Probability of Crisis—Cumulative Effects on Growth
  - 14. Estimates of Direct and Indirect Effect (Excluding Hazard Term)

### Appendices
- I. Review of the Literature
- II. A Stylized Model of Financial Repression
- III. Description of Variables
- IV. Robustness Analyses
- V. Additional Tables

### Introduction — Definition, Historical Evolution, and Study Focus
- Definition and scope:
  - Financial repression is defined as direct government intervention that alters the equilibrium reached in the financial sector, usually aiming at providing cheap loans to companies and governments by lowering returns to savers below the rate that otherwise would prevail.
  - Forms include ceilings on interest rates, directed credits to certain industries, or constraints on the composition of bank portfolios.
  - Often accompanied by restrictions on financial activity such as controls on international capital movements.
- Historical evolution of views:
  - Theoretical arguments for financial repression reference market failures (Stiglitz 1989, 1993, 2000) and information frictions (Espinosa-Vega and Smith 2001).
  - Some papers justify restrictions when combined with industrial policies (Yulek 1997), citing Japan and Korea.
  - From the 1960s onward an initial consensus supported government intervention; over time this shifted toward viewing government as an “impartial referee,” promoting liberalization since the 1980s.
  - The global financial crisis of 2008-09 rekindled debate on government role, including expanded regulation and the possibility of financial repression to mobilize seigniorage revenue to reduce government debt.
- Recent developments:
  - Renewed calls for explicit government intervention include measures such as quantitative constraints on credit allocation and ceilings on bank interest rates.
  - Some calls have not resulted in effective action (example countries: Tanzania and Azerbaijan); a few measures have been adopted (referenced as Section IV).
  - There have also been recent cases of liberalization.
- Costs and channels of distortion:
  - By compressing returns to savers, financial repression leads to a suboptimal savings rate, increasing scarcity of funds for investment, disintermediation, and reduced access to financing.
  - If costs are placed on banks by imposing a floor on deposit rates, financial stability risks may increase.
  - Weakening price signals distorts allocation of investment, reducing average investment quality and rate of return.
  - Awarding rents to beneficiaries encourages wasteful rent-seeking and potentially corruption.
- Research question and data:
  - The study focuses on government-mandated limits on interest rates that commercial banks can apply to deposits and loans.
  - The study assesses the impact of financial repression on per capita real GDP growth and on the probability of crisis.
  - The database of Abiad, Detragiache, and Tressel (2008) covering the period 1973-2005 was extended (coverage discussed in text) to analyze effects.

*Source: wpiea2019211-print-pdf - References*

### 2017. While their database covers a variety of indicators, its extension here is limited to the

### wpiea2019211-print-pdf - 2017. While their database covers a variety of indicators, its extension here is limited to the

### Main findings and summary
- Interest rate restrictions reduce growth by about 0.4-0.7 percentage points, with the effect being larger in economies with larger financial systems.
- Full liberalization is necessary to significantly increase growth; changes in interest rate restrictions short of full liberalization have a limited impact.
- Regional differences: effect appears strongest in sub-Saharan Africa (SSA), the Middle East and North Africa (MENA), and in transition countries.
- Financial repression reduces the probability of crisis (a beneficial effect for growth), but this positive effect is dwarfed by the larger adverse direct effect; on net, financial repression has a significant adverse effect on growth.

### Stylized analysis of interest rate restrictions (mechanisms)
- Objectives of financial repression include:
  - (a) public financing through seigniorage (by maintaining real interest rates below their market equilibrium levels), often aimed at facilitating the reduction of public debt;
  - (b) subsidizing particular sectors or industries (by ensuring access to low-cost credit financed by “captive” savers);
  - (c) maintaining financial stability in the medium term by creating a predictable credit environment protected from competition.
- Major forms of financial repression listed:
  - Interest rate controls (ceilings or, less frequently, floors on bank lending and deposits rates);
  - Directed lending (mandatory instructions to banks to allocate a minimum amount of loans to specific beneficiaries);
  - Restrictions on international capital movements;
  - Restrictions to entry into the banking sector;
  - Direct government intervention in the financial sector (establishing and operating state-owned banks);
  - Unconventional monetary policies that keep the interest rate curve artificially flat.
- Interaction of interventions:
  - Ceilings on lending interest rates lead to credit rationing and non-interest allocation criteria; often accompanied by directed lending.
  - Ceilings on deposit rates yield rents to banks and deposit rationing; encourage entry barriers.
  - Ceilings on deposits or loans encourage international capital movements, inducing capital controls.
- Quasi-fiscal effect: ceilings on loan interest rates transfer losses from depositors to beneficiary borrowers (a quasi-fiscal operation), while exclusion and withdrawal losses are deadweight losses.
- Motivations for authorities include lobbying/corruption, support for infant financial sectors, favoring needy/strategic borrowers, or financing investments with externalities.

### Theoretical illustration (three-agent model)
- Economy agents: savers, borrowers, financial intermediaries (banks).
- Under binding administrative loan rate ceiling i_L^C (below market clearing i_L^*):
  - Excess demand for loans; banks reduce supply by lowering deposit rates; deposit supply falls; credit rationing results (Figure 1 schematic).
  - Deposit market demand becomes kinked; at lower volumes deposit supply becomes flat where banks cannot offer higher rates without loss.
- Bank borrower selection under rationing:
  - Circumvent via higher non-interest charges (fees/commissions);
  - Favor less risky projects, long-term customers, or non-economic criteria (connections), leading to inefficient allocation and potential justification for directed lending.
- Distributional effects:
  - Ceiling benefits selected borrowers at expense of (a) borrowers excluded by rationing, (b) depositors receiving lower rates, (c) depositors withdrawn from market.
  - Losses to (b) are transferred to beneficiary borrowers; losses to (a) and (c) are deadweight losses.

### Data and measurement
- Primary variable: index of “interest rate controls” (IRC) representing administrative or legal controls on commercial bank deposit and loan rates.
- Coverage: annual data for 90 economies over 45 years, from 1973 to 2017 (89 countries and one Special Administrative Region). Notes:
  - For China, data available from 1981.
  - For 18 countries that became independent or joined the Fund between 1990 and 1993, data available for 24-27 years.
- IRC index values: four possible numerical values, 0 to 3:
  - 0 = strictest controls on interest rates;
  - 1 = extensive but not universal controls, binding constraint to dominant share of market;
  - 2 = binding constraints apply to a significant share, dominant share remains free or subject to loose constraints;
  - 3 = banks essentially free to set own interest rates (full liberalization).
- For computation, IRC mapped into binary Financial Restrictions Index (FRI):
  - FRI = 0 if IRC = 3 (full liberalization);
  - FRI = 1 if IRC < 3 (significant restrictions present).
- 2017 average IRC scores reported:
  - G-7 countries: 3
  - LA-5 countries: 2.8
  - sub-Saharan countries: 2.4
- Other data sources: IMF World Economic Outlook, International Financial Statistics, Penn World Tables, IMF Financial Soundness Database, World Bank WDI, International Country Risk Guide, IMF Strategy and Policy Review Department debt crisis indicator.
- Crisis definition: high inflation, high risk premia on debt, arrears on external debt, debt restructuring, or receiving emergency official financial assistance.

### Descriptive and correlational observations
- Major liberalization wave roughly between 1984 and 1996:
  - Until 1984, three-fourth of jurisdictions had some form of interest rate restrictions.
  - By 1991, ratio fell below one-half.
  - After 1995, less than a quarter maintained restrictions.
  - By 1999 ratio stabilized around 17-18 percent, increasing toward 25 percent only in recent years.
- Across all observations, per capita growth has been 0.52 percentage points higher in the absence of interest rate restrictions.
- Correlations with FRI:
  - Per-capita GDP: strong negative correlation on entire sample (-40 percent) and in annual averages (-34 percent); not significant in jurisdiction averages across periods.
  - FRI negatively correlated with public and external debt and money/GDP ratio.
  - FRI positively correlated with inflation.

### Empirical strategy and episodes analysis
- Empirical approach: augmented growth regressions on panel data covering 90 jurisdictions and 45 years with macroeconomic and institutional controls.
- Episode definition (Gourinchas and Obstfeld 2012 method): seven-year window centered on year t = 0 when restrictions fully removed (IRC increases from <3 to 3, FRI 1→0) or reintroduced (IRC declines 3→<3, FRI 0→1).
- Sample composition:
  - 71 jurisdictions experienced episodes of full liberalization;
  - 9 maintained interest rate restrictions throughout the period;
  - 10 always maintained a liberal regime.
  - Identified 84 episodes of full liberalization, concerning 568 observations out of 3,718, mostly between 1985 and 1995.
- Characteristics (average values) by group (jurisdictions that liberalized at some point; jurisdictions that maintained a restricted regime throughout; jurisdictions that maintained a liberal regime throughout):
  - Number of jurisdictions: 71 (liberalized), 9 (restricted), 10 (liberal)
  - Real GDP growth (percent): 3.4 (liberalized), 5.0 (restricted throughout), 2.1 (liberal throughout)
  - Real per capita growth (percent): 1.9, 3.3, 2.3
  - Inflation (percent): 44.3, 32.7, 73.7
  - Excluding high-inflation cases: 7.0, 6.5, 4.5
  - Public debt (percent of GDP): 31.6, 25.3, 37.7
  - External debt (percent of GDP): 46.9, 31.3, 54.2
  - Per capita income (U.S. dollars): 9,161, 1,177, 19,246

### Outcomes around liberalization episodes (three years before vs. three years after)
- Selected comparisons (three preceding years; three following years; liberalized regime; restricted regime; Values outside episodes where indicated):
  - Real per capita growth (percent): 0.6 (three preceding years), 2.3 (three following years), 2.3 (liberalized regime), 2.0 (restricted regime)
  - Net capital formation (percent of GDP): 20.3, 21.5, 22.1, 20.2
  - Savings (percent of GDP): 20.3, 20.2, 21.8, 18.7
  - Private sector credit (percent of GDP): 28.7, 30.8, 56.9, 28.4
  - Net capital formation (real annual growth in percent): -13.5, 5.6, 46.8, 4.9
  - Private sector credit (real annual growth in percent): 97.9, 9.7, 7.9, -2.0
  - Inflation (percent): 143.8, 22.2, 19.5, 54.3
  - Probability of the start of a recession (percent): 16.2, 13.5, 9.8, 12.4
  - Percent of years in a recession: 34.5, 23.4, 16.5, 28.5
  - Probability of the start of a debt crisis (percent): 5.0, 2.5, 2.8, 1.2
  - Percent of years in a debt crisis: 10.0, 15.6, 9.9, 3.8
  - Fiscal deficit: -1.6, -1.4, -1.7, -1.1
  - Public debt: 19.3, 25.9, 46.8, 14.5
  - Current account balance: -3.2, -3.0, -1.0, -3.1
  - External debt: 45.3, 39.3, 55.9, 34.5
  - Per capita income (U.S. dollars): 4,313, 5,685, 15,900, 2,253
- Key empirical patterns:
  - Growth increases markedly in the wake of liberalization; immediate acceleration after liberalization appears in unconditional and conditional estimates.
  - No significant change in savings, net capital formation, or private sector credit as percent of GDP, though real growth in net capital formation appears to increase after liberalization while private sector credit declines.
  - Liberalization associated with lower inflation; many jurisdictions liberalized during periods of particularly high inflation and accompanying reforms reduced inflation rapidly.
  - A crisis is more likely to start in the three years preceding liberalization than in the three years following it.
  - If a crisis starts, it tends to last longer in the wake of liberalization; unconditional probability of a crisis is higher in the wake of liberalization, while the probability that a crisis may start is somewhat lower in the years following liberalization.
  - About three-fourth of crises in the sample were accompanied or preceded by a change in the interest rate regime (usually easing of restrictions).
  - Liberalization seems associated with an increase in public debt (consistent with financial repression enabling seigniorage extraction), while external debt appears to decline in years following liberalization though generally higher in liberal regimes.

### Econometric specification (for episodes)
- Unconditional model (Equation 1):
  - y_ct = a_c + sum_{z=-3}^{3} b_z d_zt + epsilon_ct
  - where a_c is country-specific constant, d_zt are dummies for distance z from liberalization event.
- Conditional model (Equation 2) adds K control variables x_k:
  - y_ct = a_c + sum_{z=-3}^{3} b_z d_zt + sum_k gamma_k x_kct + epsilon_ct

*Source: Authors’ estimates based on the dataset and analysis presented in the document.*

### Introduction of Controls

### Introduction of Controls

### Episodes and timing
- 30 episodes in the sample, in 30 different jurisdictions, for a total of 203 observations.  
- Episodes cluster around: 1981-85, 2001-08, and 2013-15.  
- Sample includes 29 countries and one Special Administrative Region.  

### Macroeconomic behavior around introduction of controls
- Inflation, fiscal deficit, external debt, and the probability of crisis increase after introduction of interest rate controls.  
- Absolute changes in per capita growth and public debt are smaller than during episodes of liberalization.  
- Tightening episodes exhibit a stronger net change (decline) in the current account balance.  
- Private sector credit increases in percent of GDP after introduction of controls, but real net capital formation (investment) appears to slow.  
- Net change in real per capita growth visible at the median, less evident at top and bottom quartiles.  
- The probability that a crisis may start (or continue) increases—at least temporarily—after interest rate controls are introduced.

Key numeric observations from tightening episodes (three preceding years / three following years / values outside episodes / restricted vs liberalized regime in table format present in source):
- Real per capita growth (percent): 3.3 / 2.9 / 2.3 / 1.7 (rows shown in source table).  
- Net capital formation (percent of GDP): 21.4 / 22.4 / 22.0 / 20.0.  
- Savings (percent of GDP): 20.7 / 20.5 / 21.7 / 18.9.  
- Private sector credit (percent of GDP): 37.2 / 43.1 / 53.5 / 27.5.  
- Net capital formation (real annual growth in percent): 8.1 / 7.2 / 40.4 / 1.9.  
- Private sector credit (real annual growth in percent): 19.9 / -2.3 / 8.3 / 14.5.  
- Inflation (percent): 16.2 / 38.7 / 30.3 / 70.3.  
- Probability of the start of a debt crisis (percent): 1.1 / 3.5 / 3.0 / 1.5.  
- Percent of years in a debt crisis: 5.7 / 8.2 / 11.1 / 4.4.  
- Fiscal balance: -0.9 / -1.2 / -1.7 / -1.1.  
- Public debt: 26.6 / 27.4 / 44.4 / 14.4.  
- Current account balance: -1.5 / -3.1 / -1.3 / -3.2.  
- External debt: 32.8 / 37.1 / 54.3 / 36.2.  
- Per capita income (U.S. dollars): 3,927 / 4,951 / 14,807 / 2,410.

### Panel data estimates — direct effects on growth
- Presence of interest rate controls (IRC index below 3, or FRI equal to 1) has a significant adverse impact on growth.  
- Interest rate controls reduce growth, on average, by 0.6-0.7 percentage points (equivalent to 28-33 percent of the average growth rate of the entire sample, 2.1 percent).  
- Liberalization (or its opposite, the introduction of controls) in the previous three years has a milder impact (about 0.5 percentage points), implying significant lagged effects.  
- Net change in IRC in the preceding three years does not appear significant; changes between presence and absence of restrictions matter more than changes among degrees of restrictions.  
- Debt/crisis dummies: GDP growth is on average 2 percentage points lower at times of crises.  
- Inflation has a negative effect; population growth has a negative coefficient (above -1).  
- Investment, trade openness, and fiscal balance have positive impacts on growth; external debt has a negative impact.

Table 6 model summary statistics (entire sample, fixed effects, 1973-2017):
- Observations: 2,880.  
- Number of jurisdictions: 86.  
- R-squared range across specifications: 0.1935 to 0.1986.  
- Interest rate controls explain about 0.3-0.4 percent of the variance (about 1.5-2 percent of the explained variance).  
- Estimates explain 19-20 percent of the within-country variance and about 14-15 percent of the variance of the entire sample.

Notable coefficient magnitudes (selected, as reported):
- FRI (lagged) coefficient example: -0.6485*** (standard error 0.2017) in column (1).  
- Debt crisis coefficient: -2.0014*** (standard error 0.2586) in column (1).  
- Inflation coefficient: -0.0058*** (standard error 0.0007) consistently across columns.  
- Population growth (percent) coefficient: -0.9705*** (standard error 0.0975) in column (1).  
- Fixed capital formation coefficient: 0.1677*** (standard error 0.0148).  
- Public debt coefficient: 0.0089*** (standard error 0.0027) in column (1).

### Heterogeneity and robustness
- The negative relation between financial repression and growth strengthened since the early 1990s; in recent decades controls associated with about 0.8 percentage points lower real GDP growth in some specifications.  
- The link strongest and most significant in SSA and MENA regions, and in transition economies; insignificant in Asia and in advanced economies.  
- Results remain generally robust across fixed/time effects, random effects, Arellano-Bond dynamic models, five-year averages, Penn World Tables specifications, and a treatment-effect matching model—though statistical significance varies.  
- Adding institutional quality indicators (World Governance Indicators) reduces sample size and weakens statistical significance but estimated coefficients on financial repression remain negative and generally significant.

### Transmission channels (Table 10 findings)
- Savings (share of GDP) is significantly negatively related to the presence of interest rate controls.  
- Private sector credit (percent of GDP and real annual growth) is negatively related to financial repression and to recent removal of controls (FRId3) or easing (IRCd3).  
- Growth in total factor productivity is adversely related to the presence of interest rate controls, but not to changes in regime in the short term.  
Interpretation offered: long-term harm via reduced savings and credit and misallocation (lower TFP); medium-term easing may not immediately boost savings and can temporarily reduce private credit.

### Financial repression and crises — indirect effects
- Probit panel estimates: financial repression (FRI) is significant and negative in the Probit for debt crisis probability—suggesting restrictions reduce the risk of a debt crisis in a given period.  
- Estimated marginal effect: financial repression could have reduced the probability of a debt crisis by about 1 percentage point (computed at sample averages and across observations).  
- Indirect effect on growth via crisis channel (multiplying average impact by crisis dummy coefficient): small, about 0.02 percentage points (positive), which weakens only marginally the direct negative effect of 0.6-0.7 percentage points.  
- Aggregate sample implication: financial repression may have reduced real per capita growth, on average, by 26-27 basis points (Table 13, total effect on sample averages).  
- Alternative augmented approach (adding estimated impact of repression on crisis risk per observation into growth regression) yields estimated average reduction in growth of 17-18 basis points, of which 11-12 basis points may be due to the indirect effect through higher risk of debt crisis (Table 14).  
- When the indirect impact is added, the direct coefficient on financial repression can become statistically insignificant in some specifications.

### Case studies — Kenya and Bolivia
Kenya (interest rate controls reintroduced September 2016; law later ruled unconstitutional on March 14, 2019, with 12 months to reconsider):
- Lending rate ceiling: 4 percent above a reference rate; floor on time deposits equal to 70 percent of the reference rate.  
- Effects documented (Alper et al. 2019; Safavian and Zia 2018):
  - Sharp decline in bank credit to micro-, small-, and medium-sized firms; stock of credit to SMEs fell by about 10 percent in one year.  
  - Banks increased lending to corporates and households while reducing SME lending—possible reallocation toward less risky borrowers.  
  - Stock of credit of small banks declined by about 5 percent within a year.  
  - Shift of credit away from private sector toward public sector.  
  - Shift from time deposits to demand deposits.  
  - Average profit margins on private sector lending activities declined further and turned negative.  
  - Policy appears to have contributed to the cyclical increase in non-performing loans by incentivizing short-term lending.  
- Law under revision following court ruling; controls remained in place during reconsideration period.

Bolivia (Financial Services Law, 2013; implementing decrees 2015 onwards):
- Law aimed to promote financial inclusion and stability, but included regulated lending rates, lending targets (productive sectors and social housing), and discretionary floors on deposit rates for accounts not exceeding Bs.70,000 (about $10,000).  
- Ceilings in specific sectors: social housing (5.5–6.5 percent), productive sectors (6–7 percent), microfinance (11.5 percent).  
- Portfolio quotas: full-service banks directed to commit at least 60 percent of loan portfolios to social housing and productive sectors (50 percent for SMEs) by end-2018.  
- Observed effects:
  - Rapid credit growth in 2015-17 as banks accelerated lending to meet quotas; credit growth moderating toward trend later.  
  - Declining bank profitability ratios (earnings on assets and equity) as spreads squeezed by controls.  
  - Declining capital adequacy ratios (offset in part by requirement to retain at least 50 percent of profits to increase capital); ratios still appear adequate.  
  - Non-performing loans low and well-provisioned, but increases in restructured loans warrant monitoring of asset quality.  
  - Number of borrowers of less than $5,000 decreased by 10 percent from September 2014 to June 2018; average loan amounts increased—suggesting some small borrowers excluded from formal system despite inclusion goals.

### Conclusions and policy-relevant implications
- Financial repression is a quasi-fiscal operation that taxes some agents to finance public objectives; rational only if benefits exceed costs and compared with alternatives.  
- Main costs:
  - Compresses return to savers → suboptimal savings rate → reduces funds for investment and access to financing.  
  - Increases demand for credit via lower lending rates → rationing based on non-economic criteria → inefficiencies.  
  - Reduces bank profitability (if deposit floors), impairing capital accumulation and potentially raising financial stability risks.  
  - Distorts allocation of investment, reducing average quality and rate of return; encourages rent-seeking and possible corruption.  
- Empirical summary:
  - On average, financial repression reduces growth by about 0.4-0.7 percentage points (robust across models).  
  - Effects stronger in SSA, MENA, LAC, and transition countries; insignificant in Asia and advanced economies.  
  - Partial easing (short of full liberalization) has limited impact on growth; full reform appears necessary to materially enhance growth.  
  - Financial repression reduces the probability of a debt crisis in a period (short-term stability gain), but this indirect positive effect on growth is much smaller than the direct negative effect.  
  - Partial loosening without full liberalization may raise crisis risk—arguing for strengthening financial sector supervision prior to partial reforms.  
- Case-study lessons:
  - Interest rate controls can disrupt financial stability and reduce access for small enterprises (Kenya: SME credit decline; Bolivia: potential asset quality risks and reduced small-borrower access).  
- Final judgment: Restrictions have a significant adverse effect on growth and on access to financing; in the long term, countries would generally be better-off without financial repression.

*Source: Authors’ estimates and analysis contained in the chapter "Introduction of Controls" (wpiea2019211-print-pdf).*

### APPENDIX I. Review of the Literature

### APPENDIX I. Review of the Literature

### A. The 1960s: Early Literature on the Finance and Growth Nexus
- Dominant Keynesian approach (Keynes 1936; Tobin 1965) favored active government intervention in credit markets, including controls on interest rates and credit composition to support investment and growth.
- Rationale for financial repression: distinguish capital used for productive purposes from capital used for speculation; reduce effects of uncertainty on private investment decisions.
- Policy motivations: reach full employment and mobilize resources to repay large post-war debt; banking sector provided capital and advice for reconstruction and industrialization.
- Finance-growth causality frameworks:
  - Gerschenkron (1962) and Patrick (1966): two patterns — “demand following” (finance expands in response to development) and “supply leading” (finance leads investment at earlier stages).
  - Cameron (1967): financial systems can be both growth-inducing and growth-induced; quality and efficiency of intermediation critical to allocate savings to risk-prone entrepreneurs and mobilize unproductive wealth into productive uses.

### B. The 1970s: The McKinnon-Shaw Approach/Paradigm
- McKinnon (1973) and Shaw (1973) critiqued financial repression as harmful to long-run growth because it reduces funds available for investment; defined financial repression as combination of indiscriminate nominal interest rate ceilings and high and accelerating inflation.
- Core prescription: allow real interest rates to reach market-clearing levels to increase savings and investment; avoid seigniorage through inflation.
- Mechanisms of harm from repression:
  - Deposit rate ceilings widen lending-deposit spreads, create rents for banks, raise entry barriers to banking.
  - Loan rate ceilings produce credit rationing, allocation based on transaction costs, political influence, reputation, loan size, and corruption, reducing average efficiency of investment.
  - Banks become conservative, favoring “safe” government bonds over risky loans.
- Extensions and evidence:
  - Mathieson (1980), Fry (1980): deposit rate repression reduces real money demand, credit availability, and real GDP growth.
  - Kapur (1976), Mathieson (1980): reserve requirements widen spreads and reduce money demand.
  - Galbis (1977): two-sector model — liberalized deposit rates channel savings to modern, higher-return sector and raise average investment efficiency.
- Policy recommendations: abolish interest rate ceilings and directed credit, reduce reserve requirements, and promote financial sector competition.

### C. The 1980s: Critiques of Financial Liberalization Policies
- Neostructuralist critique: in presence of “curb” (unorganized) money markets, raising real deposit rates can shift assets to formal markets subject to reserve requirements and reduce financial intermediation; high rates may increase propensity to save and trigger cost-push inflation with adverse short-term demand effects.
- Microeconomic critiques emphasize asymmetric information, adverse selection, and moral hazard in credit markets:
  - Stiglitz and Weiss (1981): market-clearing high interest rates attract risky borrowers and induce riskier borrower behavior, causing credit rationing even without government intervention.
  - Mankiw (1986): if loan applicant pool is too risky, raising interest rates can collapse credit markets.
  - Diamond (1984): monitoring costs justify delegation to intermediaries; Williamson (1987) shows resulting equilibrium credit rationing and inefficient investment allocation.
- Policy implication: market failures imply markets are not inherently efficient; properly designed public interventions can achieve second-best outcomes.

### D. The 1990s: Finance and Endogenous Growth
- Endogenous growth framework emphasizes productivity as driver of long-term growth; finance can sustain growth by enhancing aggregate investment efficiency and enabling innovation.
- Mechanisms:
  - Bank intermediation allows financing of illiquid, high-productivity assets while preserving liquidity (Bencivenga and Smith 1991).
  - Financial sector channels savings to productive uses via information collection and analysis; expansion raises savings and induces competition and technical efficiency through “learning-by-doing” (Berthélemy and Varoudakis 1996), potentially generating multiple equilibria and traps at low financial development.
- Policy debate on intervention:
  - Some studies argue distortions (deposit rate ceilings, high reserve requirements) reduce innovation (Roubini and Sala-i-Martin 1992).
  - Romer (1986) notes positive externalities of investment and potential role for subsidies to achieve socially efficient outcomes.

### E. The Most Recent Studies (since 2000)
- New focus areas: measurement of financial repression; whether liberalization should be accompanied by prudential reforms; whether crises encourage reforms; whether liberalization increases or decreases crisis risk; whether high public debt post-2008-09 encourages return to financial repression.
- Measuring financial repression:
  - Early binary indices captured presence/absence of restrictions (e.g., interest rate controls).
  - Later indices accounted for degrees of repression across multiple components: credit controls, interest rate controls, entry barriers, regulation, international capital flows, privatizations, securities market policy, prudential regulations (Abiad and Mody 2003, 2005; Abiad, Oomes, and Ueda 2004; Abiad, Detragiache, and Tressel 2008).
- Liberalization-growth empirical evidence: inconclusive
  - Studies reporting positive impact: Levine, Loayza, and Beck 2000; Tornell, Westermann, and Martinez 2004; Bonfiglioli and Mendicino 2004; Bonfiglioli 2005; Rancière, Tornell, and Westermann 2006; Lee and Shin 2008; Romero-Àvila 2009; Burmann 2013.
  - Studies reporting negative impact: Eichengreen and Leblang 2003; Bashar and Khan 2007; Ahmed 2013.
  - Studies reporting ambiguous or insignificant results or conditional effects: Arestis and Demetriades 1997; McLean and Shreshta 2002; Dawson 2003; Boot 2000; Bussière and Fratzscher 2008; Ben Gamra 2009 (impact varies with intensity of liberalization).
  - Model-based nuance: Yulek (2017) two-sector model suggests repression can be welfare-improving if it internalizes positive externalities in modern sector (no empirical evidence presented).
- Financial liberalization and prudential regulations:
  - Argument that prudential safeguards are necessary precursors to liberalization to contain crisis risk.
  - Reforms often lack proper prudential safeguards due to technical complexity and vested interests (Walter 2003; Hlaing and Kakinaka 2018).
- Crises and reforms:
  - Crises tend to encourage financial liberalization and regulatory reforms; crises often the most frequent stimulus to reform (Bates and Krueger 1993; Lora and Olivera 2004).
  - Empirical findings: Waelti (2015) — crisis origin determines reform type; Masciandaro and Romelli (2017) — crises increase central bank role in supervision; Hlaing and Kakinaka (2018) — crises encourage liberalization but often result in “incomplete” reforms lacking stronger prudential regulation.
- Liberalization and crisis risk: mixed empirical evidence
  - Findings that liberalization increases crisis probability: Tornell, Westermann, and Martinez 2004; Rancière, Tornell, and Westermann 2006.
  - Findings that institutional strength or further liberalization may reduce crisis risk: Demirgüç-Kunt and Detragiache 1998; Hartwell 2017.
  - Findings that liberalization reduces crisis likelihood: Loizos 2018; Lee, Lin, and Zerng 2016; Barrell, Karim, and Ventouri 2017.
  - Transitional dynamics: destabilizing short-run effects but long-run institutional improvements (Kaminsky and Schmuckler 2002).
- Financial repression and public debt:
  - Post-2008-09 public debt surge renewed interest in repression as a tool to mobilize seigniorage and reduce debt.
  - Reinhart and Rogoff (2013): historical evidence that advanced countries have used financial repression.
  - Reinhart and Sbrancia (2015): financial repression acted as a tax on bondholders and savers via low real interest rates and contributed to debt reduction after World War II; could be necessary (but not sufficient) to restore sustainability.
  - Chari, Dovis, and Kehoe (2016, 2018): argue repression can be optimal when it enables credible government borrowing in difficult circumstances (e.g., wars).
- Other focused studies and meta-analyses:
  - Country- and variable-specific studies: bank efficiency (Hermes and Meesters 2015); capital allocation efficiency (Bhaduri and Bhattacharya 2018 for India); consumer credit (Brissimis, Garganas, and Hall 2014 for Greece).
  - China-focused studies: Zhang, Zhu, and Lu (2014) — trade and financial openness improved financial efficiency and competition but negatively impacted financial sector development with provincial variation.
  - Africa and developing-country studies: Obademi and Elumaro 2014 for Nigeria; Zhou et al. 2018 for Togo.
  - Other topics: financial repression and “populism” (Norkina 2018); disintermediation (Mertens 2008).
  - Meta-analyses and deep surveys: Bumann, Hermes, and Lensink 2013; Arestis, Chortareas, and Magkonis 2015; Valickova, Havranek, and Horvath 2015; Loizos 2018 synthesis emphasizing a “Post-Keynesian attempt to take an institutional perspective within a globalized financial and economic environment.”
- Overall assessment: empirical and theoretical literature remains inconclusive; the “jury is still out.”

*Source: APPENDIX I. Review of the Literature*

### APPENDIX II. A Stylized Model of Financial Repression

### APPENDIX II. A Stylized Model of Financial Repression

### Model setup: behavioral relations and parameters
- Linear demand for bank loans (L_D):
  - D_L = a_L − b_L i_L  (A.1)
- Linear supply of deposits (D_S):
  - S_D = a_D + b_D i_D  (A.2)
- Definitions and constants:
  - i_L is the interest rate on loans; i_D is the interest rate on deposits.
  - a_L, a_D, b_L, b_D are constant parameters.
  - Constant loan/deposit ratio, independent of volumes or interest rates:
    - (1) S_D L  r D  kD = − = , where r is the reserve requirement (share of deposits), and k = 1 − r. (A.3)
  - Linear interest spread relation:
    - S_LD i i p q L − = + , where p and q are constant parameters. (A.4)

### Derived bank demand and supply functions (unrestricted)
- Derived deposit demand function (combining borrower demand, spread relation, loan market equilibrium L_S = L_D, and L_S = k D_D):
  - 1 D L D LL a i p k q D = − − + / (b_L b_b?)  (A.5) 
    - (Equation as presented in the source text: 1/(b_b?) formatting preserved from original.)
- Derived supply of loans by banks (combining saver deposit supply, spread, and loan/deposit relations):
  - 1 S D L DD a i p q L b k b b = − ++  (A.6)
- Unrestricted deposit-market equilibrium (equating demand and supply in deposit market):
  - Equilibrium volume D*:
    - * L D D L D L D L D a b a b p b b D kb b b kqb b b + − = ++  (A.7)
  - Equilibrium deposit interest rate i_D*:
    - () 1 * L D L L D D L D L aa kqbpb i kbbkqb b − + − = ++  (A.8)
- Unrestricted loans-market equilibrium (equating demand and supply in loans market):
  - Equilibrium volume L*:
    - * D L L D L D D L D L a b a b b b p Lk kb b kb b q + − = ++  (A.9)
  - Equilibrium loan interest rate i_L*:
    - ( ) 1 * L D D D L D L D L D L a kb q ka kb p i kb b kb b q + + − = ++ . (A.10)

### Effect of a ceiling on loan interest rates (i_L^C)
- When authorities impose a ceiling i_L^C on loan interest rates, derived demand for deposits where the ceiling binds:
  - A_C D D L i i p q kD = − − . (A.11)
- The ceiling is binding whenever:
  - 1 D C D L L LL a p k q D i p q kD bb (expression comparing derived and ceiling-demand)  − −  (A.12)
  - Define D_C as the value of D_D that satisfies (A.12) as an equality.
- Implications and restricted equilibrium in deposit market:
  - Derived demand for deposits becomes kinked: equal to (A.5) when D < D_C and to (A.11) when D ≥ D_C (Appendix Figure II.3).
  - New deposit-market equilibrium volumes and rates when ceiling binds:
    - Volume: ( ) 1 C D D L A D a b i p D b kq + − = +  (A.13)
    - Interest rate: 1 1 A C D DL DD a kq i i p b kq b ( ) = − − + . (A.14)
  - Since i_L^C < i_L*, both the new volume and the new interest rate are lower than in the free market equilibrium: a binding loan-rate ceiling induces financial disintermediation and reduces savings channeled through the banking system to borrowers for investment. (footnote 26 provides decomposition of surplus transfers.)

### Effect of a ceiling on deposit interest rates (i_D^C)
- When authorities impose a ceiling i_D^C on deposit interest rates, deposit supply is capped:
  - C C D D D a b i = +  (A.15)
- The supply of loans is therefore capped at:
  - C C C D D D L kD ka kb i = = + . (A.16)
- Loans-market clearing interest rate at this capped volume:
  - C C B L L D D D L LL a L a ka kb i i bb − − − == . (A.17)
- Minimum rate demanded by banks:
  - ( ) min 1 B C C C L D D D D i i p q kD p qka qkb i  + + = + + +  (A.18)
- Banks earn a rent equal to:
  - ( ) ( ) min C B B C D L L R i i i L  −  . (A.19)
- Implications:
  - Deposit-rate ceilings create rationing in the deposit market; reduced deposits translate into reduced loan supply, higher loan interest rates, and lower volumes.
  - Banks earn rents by paying low deposit rates and charging high loan rates; they engage in non-interest competition to attract scarce deposits (e.g., free checks, perks).
  - Costs are borne by:
    - (a) depositors receiving lower (capped) rates;
    - (b) depositors withdrawing from the market due to low rates;
    - (c) borrowers paying higher loan rates;
    - (d) borrowers withdrawing from the market.
  - Items (a) and (c) finance the bank rent (a quasi-fiscal operation benefiting banks/intermediaries); items (b) and (d) are deadweight losses.
  - Deposit-rate caps are likely to be accompanied by measures restricting entry or size in the banking sector.

### Ceilings on both loan and deposit rates
- Both ceilings bind only if i_L^C lies within the range:
  - ( ) min , B B LL ii . (A.20)
  - Below this range: demand for deposits falls below D_C and the deposit-rate ceiling ceases to bind.
  - Above this range: the loan-rate ceiling exceeds borrowers’ willingness to pay and is not binding.
- When both ceilings bind and i_L^C is inside the range:
  - The rent R(i_D^C) is split:
    - Component still accrued by banks: (i_L^C − i_L^B min) L_C
    - Component accrued by borrowers whose loans are approved: (i_L^B − i_L^C) L_C
  - The rent is partially paid by depositors who earn lower interest rates than in market equilibrium. (footnote 28 quantifies pieces of transfers and liquidity buffers.)

### Mechanisms of surplus transfer and welfare implications (notes summarized)
- Loan-rate ceilings:
  - Borrowers who obtain loans gain surplus equal to (i_L^A − i_L^C) L_A, where i_L^A is the rate at which demand equals L_A; part of this gain is financed by reductions in bank intermediation and reserve costs as volumes fall; the remainder is a transfer from savers who remain in market and receive lower deposit rates.
- Deposit-rate ceilings:
  - Banks capture rents funded by lower deposit rates and higher loan rates; rationing and non-interest competition arise; deadweight losses from market withdrawals occur.
- Dual ceilings:
  - Rents are split between banks and borrowers depending on where i_L^C lies relative to thresholds.

### Key equations and expressions (selected)
- D_L = a_L − b_L i_L  (A.1)
- S_D = a_D + b_D i_D  (A.2)
- L_S = k D_D with k = 1 − r  (A.3)
- i_L − i_D = p + q L  (A.4)
- Derived deposit demand and loan supply expressions: (A.5), (A.6)
- Unrestricted equilibria: (A.7), (A.8), (A.9), (A.10)
- Restricted derived demand under loan-rate ceiling: (A.11) and binding condition (A.12)
- Restricted deposit-market equilibrium under loan ceiling: (A.13), (A.14)
- Capped deposit supply and resulting loan supply: (A.15), (A.16)
- Loans-market clearing rate and rent under deposit ceiling: (A.17), (A.18), (A.19)
- Condition for both ceilings to bind: (A.20)

*Source: APPENDIX II. A Stylized Model of Financial Repression, wpiea2019211-print-pdf*

### Appendix Table IV.4. Panel Data Estimates (entire sample, Penn World Tables data)

### Appendix Table IV.4. Panel Data Estimates (entire sample, Penn World Tables data)

### Key regression coefficients (variable — columns (1)–(8))
- Interest rate controls (FRI)
  - -0.0098*** (0.0016), -0.0081*** (0.0014), -0.0022 (0.0021), -0.0101*** (0.0022), -0.0076*** (0.0015), -0.0089** (0.0040), 0.0005 (0.0033), -0.0150*** (0.0029)
- Growth of real physical capital
  - 0.5773*** (0.0285), 0.5799*** (0.0248), 0.3545*** (0.0444), 0.6843*** (0.0436), 0.5638*** (0.0270), 0.4909*** (0.0554), 0.5085*** (0.0587), 0.7795*** (0.0631)
- Growth of employment
  - 0.4506*** (0.0261), 0.4293*** (0.0244), 0.7166*** (0.0415), 0.3567*** (0.0390), 0.3721*** (0.0264), 0.2579*** (0.0565), 0.1551*** (0.0466), 0.5906*** (0.0611)
- Growth of human capital
  - 0.0436 (0.1264), 0.0818 (0.1120), 0.6459** (0.2797), 0.0212 (0.1490), 0.0654 (0.1231), -0.0474 (0.3210), 0.0180 (0.1685), 0.0938 (0.2570)

### Interactions with the FRI index (interaction term — columns (1)–(8))
- growth of real physical capital
  - 0.0012** (0.0005), 0.0011** (0.0005), -0.0007 (0.0009), 0.0005 (0.0008), 0.0003 (0.0005), -0.0007 (0.0013), 0.0001 (0.0013), 0.0008 (0.0012)
- growth of employment
  - -0.0007 (0.0015), -0.0005 (0.0015), -0.0001 (0.0019), 0.0001 (0.0022), 0.0001 (0.0014), 0.0001 (0.0040), 0.0011 (0.0031), -0.0011 (0.0031)
- growth of human capital
  - -0.0182*** (0.0061), -0.0181*** (0.0061), 0.0129 (0.0103), -0.0022 (0.0094), 0.0007 (0.0059), 0.0287 (0.0250), -0.0011 (0.0133), -0.0038 (0.0146)

### Constants, sample and significance
- Constant
  - 0.0061*** (0.0018), 0.0054*** (0.0016), 0.0020 (0.0021), 0.0035 (0.0027), 0.0059*** (0.0017), 0.0160*** (0.0057), 0.0171*** (0.0036), -0.0091** (0.0040)
- Observations: 3,244, 3,244, 902, 1,247, 2,944, 772, 559, 688
- R-squared: 0.2182, 0.3918, 0.2375, 0.2075, 0.1409, 0.1708, 0.3132
- Number of jurisdictions: 868, 622, 317, 319, 141, 7
- r2_w: 0.218, 0.218, 0.392, 0.238, 0.208, 0.141, 0.171, 0.313
- r2_b: 0.682, 0.695, 0.823, 0.762, 0.725, 0.619, 0.799, 0.829
- r2_o: 0.278, 0.279, 0.470, 0.314, 0.277, 0.158, 0.269, 0.339

### Notes on inference and estimation
- Standard errors are reported in parentheses.
- Significance codes: *** p<0.01, ** p<0.05, * p<0.1.
- FE = fixed effects; RE = random effects.
- See Appendix III for variable description.
- Source and data: Authors’ estimates; data from Feenstra, Inklaar, and Timmer (2015).

*Source: Appendix Table IV.4. Panel Data Estimates (entire sample, Penn World Tables data) — wpiea2019211-print-pdf.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019211-print-pdf.pdf_