## wpiea2022137-print-pdf

## Source details

**Canonical URL:** [wpiea2022137-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022137-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022137-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022137-print-pdf.pdf.json)

---

### I. Introduction
- India's financial sector experienced a rapid increase in non-performing assets (non-performing loans, NPLs) after the global financial crisis (GFC) and the 2018-2019 run on non-banking financial companies (NBFCs).
- Real GDP growth averaged 6.7 percent from 2011 to 2018, then moderated to 3.7 percent in 2019 (NBFC crisis) prior to the COVID-19 crisis.
- Study channels:
  - Cyclical effects on near- and medium-term GDP growth using a growth-at-risk (GaR) approach.
  - Long-term effects via bank balance sheets, credit growth, and growth using bank-level panel regressions covering banks that account for about 85 percent of total banking sector assets.
- High-level findings:
  - Higher credit and lower NPLs are associated with higher GDP growth in the near- and medium-term.
  - More favorable credit conditions matter especially during periods with low growth.
  - For private banks, the level of capitalization is strongly correlated with credit growth; the relationship is weaker for public banks (PSBs).
  - Higher real GDP growth is observed when better-capitalized banks extend more credit, conditional on those banks not having excessive NPLs.

### II. Cyclical Financial Conditions and Near-Term Growth

#### Data and stylized facts
- Quarterly macro-financial database: 2000Q1 to 2021Q3; variables include GDP growth, inflation, policy rate, bond yields, sovereign spreads, stock prices, credit growth, credit to GDP gap, the NPL ratio, world growth, oil prices, exchange rates, and other macroeconomic variables.
- Credit and leverage facts:
  - Credit-to-GDP ratio peaked at around 106 percent in 2012 and declined to around 90 percent in 2021.
  - Bank-credit-to-GDP ratio currently stands at around 55 percent.
  - Credit-to-GDP gap turned negative from 2012.
  - Corporate credit growth slowed from a peak of close to 30 percent in 2008 to zero at its trough.
  - NPL ratio peaked at around 11 percent in 2017 and has since come down to around 8 percent.

#### Methodology
- GaR approach following Adrian et al. (2019) and Prasad et al. (2019).
- Five macro-financial partitions aggregated via principal component analysis (PCA):
  1. Domestic prices: policy interest rate, 10-year treasury bond yield, sovereign bond spread, change in stock prices.
  2. Credit and leverage: credit growth, credit to GDP ratio, credit to GDP gap, NPL ratio.
  3. Macroeconomic vulnerabilities: inflation, current account balance to GDP ratio, short-term external debt to reserve ratio.
  4. External prices: changes in oil prices and exchange rates.
  5. External macro: world GDP growth.
- Quantile regression for h-quarter ahead GDP growth across quantiles q ∈ {10 percent, 25 percent, 50 percent, 75 percent, 90 percent}:
  - y_{t+h}^q = α^q + β1^q X_{1,t} + β2^q X_{2,t} + β3^q X_{3,t} + β4^q X_{4,t} + β5^q X_{5,t} + γ^q y_t + ε_{t+h}^q
- Horizons considered: 4 quarter ahead to 16 quarter ahead.
- GaR uses estimated quantile regressions and a t-skew distribution to derive the probability density of future GDP growth and enables scenario analysis.

#### Results and scenario analyses
- PCA and indicators:
  - Credit and leverage partition: credit to GDP ratio, credit to GDP gap, and credit growth have positive loadings; NPL ratio has a negative loading.
  - First principal component of credit and leverage partition captures 77 percent of the variance.
  - Credit and leverage indicator peaked in 2005/2006 and trended downward since 2011/2012.
  - Domestic prices partition: 10-year treasury yields, policy interest rate, and sovereign yields have positive loadings; change in stock prices has a negative loading.
  - Macroeconomic vulnerabilities peaked in 2012/13 and trended downward thereafter.
- GaR shock results (two standard deviation shocks):
  - Two standard deviation negative shock to the credit and leverage partition (principal component):
    - 4 quarter ahead GDP growth mode declines from 7.6 percent to 5.3 percent.
    - 5 percent GaR shifts from -5.7 percent to -12.2 percent.
    - Probability of growth below -5.7 percent increases from 5 percent to 11 percent.
  - Two standard deviation positive shock to the domestic prices partition (tightening in price-based financial conditions):
    - 4 quarter ahead GDP growth mode declines from 7.6 percent to 7 percent; distribution shifts slightly left.
  - Two standard deviation positive shock to the macroeconomic vulnerability partition:
    - 4 quarter ahead GDP growth mode declines from 7.6 percent to 5.9 percent; distribution shifts left with higher tail risks.
- Term structure and quantile effects:
  - High credit and low NPLs have a positive and significant impact on GDP growth across horizons (4, 8, 12, 16 quarters).
  - Impact is larger at lower quantiles (i.e., when GDP growth is lower), implying favorable credit conditions matter disproportionately during weak growth periods.
- Robustness:
  - Alternative specification focusing on bank credit to GDP ratio and bank credit growth yields robust results: negative shock to credit and leverage shifts GDP growth distribution left; higher bank credit and stronger credit quality particularly supportive when growth is weak.

#### Investment-at-Risk and Consumption-at-Risk
- Two standard deviation negative shock to the credit and leverage partition shifts distributions of investment and consumption growth left.
- Investment outcomes:
  - 4 quarter ahead investment growth mode falls from 20 percent to 10.4 percent.
  - Prior to the shock, there was a 5 percent probability that investment growth could be below -10.9 percent.
  - After the shock, there is a 5 percent probability that investment growth could be below -19.2 percent.
- Consumption outcomes:
  - Mode of the 4 quarter ahead consumption growth declines from 12.5 percent to 6.7 percent.
  - 5 percent Consumption-at-Risk shifts from -5 percent to -15.6 percent.
- Estimation note: results estimated using data from 2001Q4 to 2021Q3. Credit and leverage partition captures credit to GDP ratio, credit to GDP gap, credit growth, and NPL ratio (first three with positive loadings; NPL ratio negative).

#### Policy implications from GaR results
- Empirical implication:
  - Higher credit and lower NPLs are associated with higher near- and medium-term GDP growth.
  - A negative shock to credit and leverage (lower credit and higher NPL ratio) shifts distributions of GDP, investment, and consumption growth left, lowering expected growth and increasing downside risks.
- Policy measures emphasized:
  - During low economic growth, policies to support credit growth and to strengthen balance sheets are particularly important.
  - Credit guarantee schemes for MSMEs and loan restructuring schemes for COVID-affected borrowers supported credit growth and cushioned the pandemic’s economic impact.
  - Going forward, make support measures more targeted and facilitate the exit of non-viable firms.
  - Financial regulators should ensure loans benefiting from COVID-related restructuring schemes are closely monitored and properly provisioned to safeguard financial sector balance sheets and support the economic recovery.

### III. Financial Sector and Long-Term Growth (Bank-level analysis)

#### Literature and scope
- Poor capitalization and weak asset quality negatively impact banks’ ability to provide credit.
- Citations summarized include Gambacorta and Shin (2018), Muduli and Behera (2021), Blattner et al. (2019).
- Study focus: role of Indian banks’ balance sheets on credit growth and overall output growth, with particular focus on differential roles of public and private banks.

#### Data and sample
- Bank-level variables and controls (FitchConnect): cost of funding, growth of debt funding, credit growth, bank capitalization, non-performing assets, return on assets.
- Sample details:
  - Annual frequency from 1998-2021.
  - Includes only public banks and private banks; excludes NBFCs, foreign banks, and development banks.
  - Sample accounts for about 85 percent of total assets in the Indian financial sector in any given year.
- Macro-level data from Reserve Bank of India via Haver and CEIC.

#### Bank-level capitalization, balance sheets, and channels
- Capitalization measures (alternative definitions):
  - Common equity over total assets.
  - Total equity over total assets.
  - Regulatory Tier 1 capital over total assets.
  - Capital adequacy ratio (Tier 1 regulatory capital over risk weighted assets).
- Trends:
  - Median bank capitalization volatile and slightly higher for private banks in earlier years; since 2010 the gap between private and public banks widened, though both trended upwards in recent years.
  - Notable upward shift in capital adequacy ratio since 2012 when India announced intended adoption of Basel III recommendations (recommending a 9 percent capital adequacy ratio), aimed to be implemented in 2018-19.
- Asset–equity correlation:
  - For private banks correlation between assets and equity is close to one even after controls — hypothesis of unit elasticity cannot be rejected.
  - For PSBs the correlation is much weaker once time fixed effects are included.

#### Three-step empirical strategy (funding-cost and funding-level channels)
- Steps:
  1. Test whether bank capitalization reduces cost of funding (cost of non-equity funding defined as total interest paid over total debt excluding equity and reserves).
  2. Test whether capitalization is associated with an increase in debt funding (growth of debt funding).
  3. Test whether capitalization is associated with lending growth.
- Simple correlations suggest:
  - Higher bank capitalization associated with lower funding costs.
  - Higher capitalization associated with greater debt funding growth.
  - Positive correlation between lending growth and capitalization — interpretation: better capitalized banks lend more, possibly via cheaper debt funding.

#### Bank-level empirical results (selected coefficients and interpretations)
- Estimation approach: dynamic GMM (Arellano and Bond, 1991); robustness checks include panel fixed-effects and alternative capitalization measures.
- Sample periods:
  - Dynamic GMM panel sample period: 1998-2021.
  - Full macro regressions sample: 1990-2021.
  - Subsample analysis for lending growth: 2010-2021.

- Table 2 (Bank capitalization and cost of debt funding) — selected reported coefficients:
  - Leverage Ratio (total equity/assets) -0.0241 ; -0.0521***
  - Leverage Ratio (common equity/assets) -0.0206 ; -0.0494***
  - Leverage Ratio (tier 1/assets) -0.0248 ; -0.0531***
  - Capital Adequacy Ratio -0.0177 ; -0.0260***
  - N 373 / 157 / 408 / 229 / 402 (various specifications)
  - Year FE Yes; Bank controls Yes
  - Interpretation: Higher capitalization associated with lower debt funding costs, especially for private banks. For PSBs the negative association is weaker on average.

- Table 3 (Bank capitalization and debt funding growth) — selected coefficients:
  - Leverage Ratio (total equity/assets) 1.393*** ; 2.276***
  - Leverage Ratio (common equity/assets) 1.151*** ; 2.205***
  - Leverage Ratio (tier 1/assets) 0.529 ; 4.755***
  - Capital Adequacy Ratio 0.947*** ; 1.393***
  - N 373 / 157 / 406 / 228 / 399 (various specifications)
  - Interpretation: Private banks with greater capitalization associated with large and significantly greater debt funding growth. For public banks the relationship is less robust but positive and smaller.

- Table 4 (Bank capitalization and lending growth) — selected coefficients:
  - Leverage Ratio (total equity/assets) 0.371 ; 0.630
  - Leverage Ratio (common equity/assets) 0.259 ; 0.639
  - Leverage Ratio (tier 1/assets) 0.0703 ; 1.970**
  - Capital Adequacy Ratio 0.369 ; 0.805**
  - N 373 / 157 / 406 / 228 / 399 (various specifications)
  - Interpretation: Positive relationships for private banks in specific specifications; weaker evidence for PSBs in full sample.

- Table 5 (Bank capitalization and lending growth, 2010-21) — selected coefficients:
  - Leverage Ratio (total equity/assets) 2.681** ; 2.200***
  - Leverage Ratio (common equity/assets) 3.024** ; 1.970***
  - Leverage Ratio (tier 1/assets) -0.101 ; 1.957***
  - Capital Adequacy Ratio 0.432 ; 1.369***
  - N 166 / 130 / 198 / 175 / 197 (various specifications)
  - Interpretation: Restricting to 2010-21 yields an even stronger positive relationship between capitalization and lending for private banks. For PSBs no such evidence in full sample.

#### Macro-level correlations between credit growth and GDP growth
- Table 6 (Real GDP Growth and Credit Growth) — selected coefficients:
  - Sample period: 1990-2021.
  - Credit Growth coefficient: 0.0421*** (private banks) and 0.0105 (PSBs).
  - Controls shown in excerpts:
    - Inflation -0.293*** to -0.620*** (various columns).
    - Real effective exchange rate (RBI) -0.199*** to -0.343*** (various columns).
    - World GDP 0.318*** to 0.406*** (various columns).
  - Definitions:
    - Low NPL ratio = 1 if bank i’s NPL ratio in year t is below the sample mean.
    - High credit growth = 1 if credit growth is above (winsorized) mean; credit growth winsorized at the 1st and 99th percentile.
  - Interpretation:
    - Strong positive correlation between higher credit growth and real GDP growth, but only for banks with low NPL ratio.
    - Result appears driven entirely by private banks; public banks show no relationship between credit growth and real GDP growth regardless of NPL level.
    - Bank size not related to whether credit growth is associated with higher GDP growth.
    - Regression identifies correlations, not causal relationships; procyclicality or reverse causation cannot be excluded.

#### Robustness and limitations
- Results robust to panel fixed-effects and alternative sample definitions; detailed robustness tables in Annex Tables 3–7.
- Limitations:
  - Methodology cannot identify causal effects given potential endogeneity.
  - Analysis abstracts from lending by non-banks, which represent a large share of credit in India.
  - Foreign bank sample relatively small (results in Annex Table 7).

### IV. Main substantive conclusions and policy implications
- Empirical conclusions:
  - Better capitalized banks—particularly private banks—tend to:
    - Pay less for debt funding (lower cost of non-equity funding).
    - Raise more debt funding (higher growth of debt funding).
    - Lend more (higher lending growth), especially during 2010-21.
  - Aggregate correlation: higher credit growth from banks with low NPLs is associated with higher real GDP growth; relationship driven by private banks.
  - Public banks historically provide the largest share of aggregate credit but show weaker or no detectable relationship between their credit growth and real GDP growth in these correlations.
- Policy implications (as stated):
  - Ensure adequate credit growth and improve bank balance sheets, particularly by reducing NPLs, to support growth.
  - Ensure private banks are well capitalized (via new equity issuance or reduced cash dividends) to enable access to more and cheaper debt funding that supports lending.
  - Prioritize efforts to clean up bank balance sheets and boost capitalization—especially for private banks—to boost credit growth and GDP growth over the medium term.
  - Given differences between private and public banks, further work to understand drivers of those differences and address them could help promote growth.

*Source: Financial Sector and Economic Growth in India — Working Paper No. WP/22/137 (wpiea2022137-print-pdf).*

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

### wpiea2022137-print-pdf - References

### I. Introduction
- India's financial sector faced a rapid increase in non-performing assets (referred to hereafter as non-performing loans, NPLs) after the global financial crisis (GFC) and the 2018-2019 run on non-banking financial companies (NBFCs).
- Real GDP growth averaged 6.7 percent from 2011 to 2018, then moderated to 3.7 percent in 2019 (NBFC crisis) prior to the COVID-19 crisis.
- The paper examines the nexus between India’s financial sector and economic growth through two channels:
  - Cyclical effects on near- and medium-term GDP growth using a growth-at-risk (GaR) approach.
  - Long-term effects via bank balance sheets, credit growth, and growth using bank-level panel regressions covering banks that account for about 85 percent of total banking sector assets.
- Key high-level findings summarized in the introduction:
  - Higher credit and lower NPLs are associated with higher GDP growth in the near- and medium-term.
  - More favorable credit conditions matter especially during periods with low growth.
  - For private banks, the level of capitalization is strongly correlated with credit growth; the relationship is weaker for public banks.
  - Higher real GDP growth is observed when better-capitalized banks extend more credit, conditional on those banks not having excessive NPLs.

### II. Cyclical Financial Conditions and Near-Term Growth

#### A. Data and Stylized Facts
- Quarterly macro-financial database for India covering 2000Q1 to 2021Q3; variables include GDP growth, inflation, policy rate, bond yields, sovereign spreads, stock prices, credit growth, credit to GDP gap, the NPL ratio, world growth, oil prices, exchange rates, and other macroeconomic variables.
- Data sources include Haver Analytics, Reserve Bank of India, Central Statistics Office, International Monetary Fund, Bank of International Settlements, Ministry of Statistics and Programme Implementation, Bombay Stock Exchange, Energy Information administration/Chicago Mercantile Exchange, and Bloomberg.
- Definitions and detailed data sources referenced in Annex Table 1.
- Credit and leverage facts:
  - The credit-to-GDP ratio peaked at around 106 percent in 2012 and declined to around 90 percent in 2021.
  - The bank-credit-to-GDP ratio currently stands at around 55 percent.
  - The credit-to-GDP gap turned negative from 2012.
  - Corporate credit growth slowed from a peak of close to 30 percent in 2008 to zero at its trough.
  - The NPL ratio peaked at around 11 percent in 2017 and has since come down to around 8 percent.

#### B. Methodology
- GaR approach (following Adrian et al., 2019 and Prasad et al., 2019) used to estimate severity and likelihood of sharp economic slowdowns by exploiting information in financial prices and aggregates.
- Five macro-financial partitions constructed and aggregated via principal component analysis (PCA):
  1. Domestic prices: policy interest rate, 10-year treasury bond yield, sovereign bond spread, change in stock prices.
  2. Credit and leverage: credit growth, credit to GDP ratio, credit to GDP gap, NPL ratio.
  3. Macroeconomic vulnerabilities: inflation, current account balance to GDP ratio, short-term external debt to reserve ratio.
  4. External prices: changes in oil prices and exchange rates.
  5. External macro: world GDP growth.
- Quantile regression specification estimated for h-quarter ahead GDP growth across quantiles q ∈ {10 percent, 25 percent, 50 percent, 75 percent, 90 percent}:
  - y_{t+h}^q = α^q + β1^q X_{1,t} + β2^q X_{2,t} + β3^q X_{3,t} + β4^q X_{4,t} + β5^q X_{5,t} + γ^q y_t + ε_{t+h}^q
  - X_{1,t} .. X_{5,t} denote the five partitions above.
- Horizons considered include 4 quarter ahead to 16 quarter ahead.
- GaR uses the estimated quantile regressions and a t-skew distribution to derive the probability density of future GDP growth and enables scenario analysis.

#### C. Results
- PCA loadings and summary indicators:
  - For the credit and leverage partition, the credit to GDP ratio, credit to GDP gap, and credit growth have positive loadings on the first principal component, while the NPL ratio has a negative loading.
  - The first principal component of the credit and leverage partition captures 77 percent of the variance.
  - The credit and leverage indicator peaked in 2005/2006 and has trended downward since 2011/2012, coinciding with the negative credit-to-GDP gap.
  - For domestic prices, 10-year treasury yields, policy interest rate, and sovereign yields have positive loadings; change in stock prices has a negative loading.
  - Macroeconomic vulnerabilities: short-term external debt and inflation have positive loadings; current account balance has a negative loading. Macroeconomic vulnerabilities peaked in 2012/13 and trended downward thereafter.

- Scenario analyses (GaR):
  - Two standard deviation negative shock to the credit and leverage partition (principal component):
    - The 4 quarter ahead GDP growth mode declines from 7.6 percent to 5.3 percent.
    - The 5 percent GaR shifts from -5.7 percent to -12.2 percent.
    - Probability of growth below -5.7 percent increases from 5 percent to 11 percent.
  - Two standard deviation positive shock to the domestic prices partition (tightening in price-based financial conditions):
    - The 4 quarter ahead GDP growth mode declines from 7.6 percent to 7 percent; distribution shifts slightly left.
  - Two standard deviation positive shock to the macroeconomic vulnerability partition:
    - The 4 quarter ahead GDP growth mode declines from 7.6 percent to 5.9 percent; distribution shifts left with higher tail risks.

- Term structure and quantile effects:
  - Quantile regression coefficients for the credit and leverage partition evaluated at horizons of 4, 8, 12, and 16 quarters.
  - High credit and low NPLs have a positive and significant impact on GDP growth across all horizons.
  - The impact is larger at lower quantiles (i.e., when GDP growth is lower), indicating that favorable credit conditions matter disproportionately during weak growth periods.

- Robustness:
  - Alternative specification focusing on bank credit to GDP ratio and bank credit growth yields robust results: negative shock to credit and leverage shifts GDP growth distribution left; higher bank credit and stronger credit quality particularly supportive when growth is weak (see Annex Figures 1 and 2).

- Investment- and Consumption-at-Risk:
  - A two standard deviation negative shock to the credit and leverage partition shifts the distributions of investment and consumption growth left.
  - The 4 quarter ahead investment growth mode falls from 20 percent to 10.4 percent.
  - The 5 percent Investment-at-Risk shifts from -10.9 percent to (text cutoff in source).

*Source: IMF Staff estimates; data and analysis as presented in the source PDF.*

### 19.2 percent. In  other words, there was a 5 percent probability that investment growth could be

### wpiea2022137-print-pdf - 19.2 percent. In  other words, there was a 5 percent probability that investment growth could be

### Shock results: Investment-at-Risk and Consumption-at-Risk
- A two standard deviation negative shock to the credit and leverage partition shifts the distribution of near- and medium-term growth outcomes.
- Investment tail outcomes:
  - Prior to the shock, there was a 5 percent probability that investment growth could be below -10.9 percent.
  - After the shock, there is a 5 percent probability that investment growth could be below -19.2 percent.
- Consumption outcomes:
  - The mode of the 4 quarter ahead consumption growth declines from 12.5 percent to 6.7 percent.
  - The 5 percent Consumption-at-Risk shifts from -5 percent to -15.6 percent.
- Notes on estimation:
  - Results estimated using data from 2001Q4 to 2021Q3.
  - The credit and leverage partition captures: the credit to GDP ratio, the credit to GDP gap, credit growth, and the NPL ratio.
  - The first three variables have positive loadings on the principal component; the NPL ratio has a negative loading.
  - Source: IMF Staff estimates.

### Policy discussions and recommendations
- Empirical implication:
  - GaR results suggest higher credit and lower NPLs are associated with higher GDP growth in the near- and medium-term.
  - A negative shock to credit and leverage (lower credit and higher NPL ratio) shifts distributions of GDP, investment, and consumption growth to the left, lowering expected growth and increasing downside risks.
- Policy measures emphasized:
  - During low economic growth, policies to support credit growth and to strengthen balance sheets are particularly important.
  - Credit guarantee schemes for MSMEs and loan restructuring schemes for COVID-affected borrowers were important to support credit growth and cushion the pandemic’s economic impact.
  - Going forward, make support measures more targeted and facilitate the exit of non-viable firms.
  - Financial regulators should ensure loans benefiting from COVID-related restructuring schemes are closely monitored and properly provisioned to safeguard financial sector balance sheets and support the economic recovery.

### Financial sector and long-term growth (literature and scope)
- Literature summary:
  - Poor capitalization and weak asset quality negatively impact banks’ ability to provide credit.
  - Gambacorta and Shin (2018): well-capitalized banks enjoy lower costs of debt financing and higher annual credit growth.
  - Muduli and Behera (2021): in India, a positive correlation between bank equity and credit growth; affects monetary policy transmission.
  - Blattner et al. (2019): less-capitalized banks cut lending in response to higher capital requirements, potentially contributing to weaker productivity growth.
- Study focus:
  - Examines the role of Indian banks’ balance sheets on credit growth and overall output growth.
  - Particular focus on the differential role of public and private banks in driving credit growth.

### Data and stylized facts
- Bank-level variables and controls (from FitchConnect):
  - Cost of funding, growth of debt funding, credit growth, bank capitalization, non-performing assets, return on assets.
- Sample details:
  - Annual frequency from 1998-2021.
  - Sample includes only public banks and private banks; excludes non-bank financial companies, foreign banks, and development banks.
  - The sample accounts for about 85 percent of total assets in the Indian financial sector in any given year of the sample.
- Macro-level data sources:
  - GDP growth and various India-level or global controls from the Reserve Bank of India via Haver and CEIC.

*Source: wpiea2022137-print-pdf (excerpt provided).*

### Annex Table 2.

### Annex Table 2

### Bank-level capitalization definitions and trends
- Main explanatory variable: bank-level capitalization, defined alternatively as:
  - common equity over total assets,
  - total equity over total assets,
  - regulatory Tier 1 capital over total assets,
  - capital adequacy ratio (Tier 1 regulatory capital over risk weighted assets).
- Empirical notes:
  - Focus on domestic banks only.
  - Figure 7: median bank capitalization volatile and slightly higher for private banks in earlier years; since 2010 the gap between private and public banks widened, though both trended upwards in recent years.
  - Notable upward shift in the capital adequacy ratio since 2012 when India announced intended adoption of Basel III recommendations (recommending a 9 percent capital adequacy ratio), aimed to be implemented in 2018-19.

### Bank balance sheet composition and asset–equity correlation
- Capital is a relatively small share of balance sheets for both private banks and PSBs (Figure 8).
- Correlation analysis (Equation (2)) between total assets and total equity:
  - Model: log(total assets_it) = α_i + γ_t + β log(total equity_it) + δ X_it + a_it.
  - Results (Figure 9): for private banks correlation between assets and equity is close to one even after controls — hypothesis of unit elasticity cannot be rejected. For PSBs the correlation is much weaker once time fixed effects are included.

### Funding-cost and funding-level channels (three-step empirical strategy)
- Steps:
  1. Test whether bank capitalization reduces cost of funding (cost of non-equity funding defined as total interest paid over total debt excluding equity and reserves).
  2. Test whether capitalization is associated with an increase in debt funding (growth of debt funding).
  3. Test whether capitalization is associated with lending growth.
- Simple correlations (Figure 10) suggest:
  - Higher bank capitalization associated with lower funding costs.
  - Higher capitalization associated with greater debt funding growth.
  - Strong positive correlation between lending growth and capitalization.
  - Interpretation: better capitalized banks lend more, possibly via cheaper debt funding.

### Aggregate credit and macro linkages
- Aggregate patterns:
  - Figure 11: Throughout the period public banks have been responsible for the largest share of credit to the economy.
  - Since around 2013, private bank credit growth has been much faster than public bank credit growth.
- Figure 12: Aggregate correlation between real GDP growth and credit growth is relatively strong and positive for both bank types.
- Macro regression specification (Equation (6)) to relate real GDP growth to credit growth:
  - Controls include inflation, real effective exchange rate, world GDP growth.
  - NPLs specified as dummy = 1 if bank i’s NPL ratio in year t is below the sample mean.
  - Credit growth also defined as dummy for high credit growth (above sample mean).
  - Credit growth variable winsorized at the 1st and 99th percentile.

### Methodology and estimation
- Bank-level funding-cost regression (Equation (3)) and funding-level regression (Equation (4)) estimated via dynamic GMM (Arellano and Bond, 1991).
- Lending regressions (Equation (5)) similarly estimated.
- Robustness: panel fixed-effects estimations and alternative capitalization measures. Results reported in Annex Tables 3–6.
- Sample periods referenced:
  - Dynamic GMM panel sample period: 1998-2021 (tables for funding costs and debt funding growth).
  - Full sample for macro regressions: 1990-2021.
  - Subsample analysis for lending growth focused on 2010-2021.

### Key empirical results (bank-level)
- Table 2 (Bank capitalization and cost of debt funding) – selected reported coefficients and sample details as presented:
  - (1)(2)(3)(4)(5)(6)(7)(8)
  - Leverage Ratio (total equity/assets) -0.0241 -0.0521***
  - Leverage Ratio (common equity/assets) -0.0206 -0.0494***
  - Leverage Ratio (tier 1/assets) -0.0248 -0.0531***
  - Capital Adequacy Ratio -0.0177 -0.0260***
  - N373 373 157 363 408 408 229 402
  - Year FE Yes Yes Yes Yes Yes Yes Yes Yes
  - Bank controls Yes Yes Yes Yes Yes Yes Yes Yes
  - Standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001. Sample period from 1998-2021. Model estimated using the dynamic GMM panel.
- Interpretation in text:
  - Higher capitalization associated with lower debt funding costs, especially for private banks. For PSBs the negative association is weaker on average.
- Table 3 (Bank capitalization and debt funding growth) – selected reported coefficients and sample details as presented:
  - (1)(2)(3)(4)(5)(6)(7)(8)
  - Leverage Ratio (total equity/assets) 1.393*** 2.276***
  - Leverage Ratio (common equity/assets) 1.151*** 2.205***
  - Leverage Ratio (tier 1/assets) 0.529 4.755***
  - Capital Adequacy Ratio 0.947*** 1.393***
  - N373 373 157 363 406 406 228 399
  - Year FE Yes Yes Yes Yes Yes Yes Yes Yes
  - Bank controls Yes Yes Yes Yes Yes Yes Yes Yes
  - Growth in debt funding — PSBs and Private banks results reported. Standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001. Sample period from 1998-2021. Model estimated using the dynamic GMM panel.
- Interpretation in text:
  - Private banks with greater capitalization associated with large and significantly greater debt funding growth. For public banks the relationship is less robust but positive and smaller.
- Table 4 (Bank capitalization and lending growth) – selected coefficients as presented:
  - (1)(2)(3)(4)(5)(6)(7)(8)
  - Leverage Ratio (total equity/assets) 0.371 0.630
  - Leverage Ratio (common equity/assets) 0.259 0.639
  - Leverage Ratio (tier 1/assets) 0.0703 1.970**
  - Capital Adequacy Ratio 0.369 0.805**
  - N373 373 157 363 406 406 228 399
  - Year FE Yes Yes Yes Yes Yes Yes Yes Yes
  - Bank controls Yes Yes Yes Yes Yes Yes Yes Yes
  - Standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001. Sample period from 1998-2021. Model estimated using the dynamic GMM panel.
- Table 5 (Bank capitalization and lending growth, 2010-21) – selected coefficients as presented:
  - (1)(2)(3)(4)(5)(6)(7)(8)
  - Leverage Ratio (total equity/assets) 2.681** 2.200***
  - Leverage Ratio (common equity/assets) 3.024** 1.970***
  - Leverage Ratio (tier 1/assets) -0.101 1.957***
  - Capital Adequacy Ratio 0.432 1.369***
  - N166 166 130 165 198 198 175 197
  - Year FE Yes Yes Yes Yes Yes Yes Yes Yes
  - Bank controls Yes Yes Yes Yes Yes Yes Yes Yes
  - Standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001. Sample period from 2010-2021. Results for sample period 1990-2010 available in appendix or upon request.
- Interpretation in text:
  - For private banks there is evidence of a positive relationship between capitalization and lending; for PSBs no such evidence is found in the full sample. Restricting to 2010-21 yields an even stronger positive relationship between capitalization and lending for private banks.

### Key empirical results (macro-level)
- Table 6 (Real GDP Growth and Credit Growth) — selected presentation as shown:
  - Dependent variable: Real GDP Growth
  - Sample period 1990-2021.
  - Reported coefficient excerpts and sample sizes:
    - Credit Growth 0.0421*** (private banks) and 0.0105 (PSBs)
    - N807 807 824 588 309 279 221 (various columns)
    - R-sq 0.035 0.051 0.069 0.381 0.410 0.412 0.382
  - Noted controls and coefficients in table excerpt:
    - Inflation -0.293*** -0.336*** -0.411*** -0.620*** (various columns)
    - Real effective exchange rate (RBI) -0.199*** -0.263*** -0.0682 -0.343***
    - World GDP 0.348*** 0.358*** 0.318*** 0.406***
  - Low NPL ratio defined as NPL ratio below the sample mean. High credit growth defined as credit growth above (winsorized) mean. Credit growth winsorized at the 1st and 99th percentile.
- Interpretation in text:
  - Columns (3) and (4): strong positive correlation between higher credit growth and real GDP growth, but only for banks with low NPL ratio.
  - Result appears driven entirely by private banks (column (5)); public banks (column (6)) show no relationship between credit growth and real GDP growth regardless of NPL level.
  - Bank size (column (7)) not related to whether credit growth is associated with higher GDP growth.
  - The regression identifies correlations, not causal relationships; procyclicality or reverse causation cannot be excluded.

### Robustness and limitations
- Results robust to alternative estimation approaches (panel fixed effects) and alternative sample definitions; detailed robustness tables in Annex Tables 3–7.
- Limitations explicitly noted:
  - Methodology cannot identify causal effects between bank capitalization, funding, lending, and GDP growth given potential endogeneity.
  - Analysis abstracts from lending by non-banks, which represent a large share of credit in India.
  - Foreign bank sample is relatively small (results in Annex Table 7).

### Main substantive conclusions
- Better capitalized banks—particularly private banks—tend to:
  - Pay less for debt funding (lower cost of non-equity funding).
  - Raise more debt funding (higher growth of debt funding).
  - Lend more (higher lending growth), especially during 2010-21.
- Aggregate correlation: higher credit growth from banks with low NPLs is associated with higher real GDP growth; this relationship is driven by private banks.
- Public banks:
  - Have historically provided the largest share of aggregate credit.
  - Show weaker or no detectable relationship between their credit growth and real GDP growth in these correlations.
  - Possible explanations include different lending objectives (e.g., directed lending), implicit government support, or different funding models.

### Policy implications (as stated)
- Ensure adequate credit growth and improve bank balance sheets, particularly by reducing NPLs, to support growth.
- Ensure private banks are well capitalized (via new equity issuance or reduced cash dividends) to enable access to more and cheaper debt funding that supports lending.
- Prioritize efforts to clean up bank balance sheets and boost capitalization—especially for private banks—to boost credit growth and GDP growth over the medium term.
- Given differences between private and public banks, further work to understand drivers of those differences and address them could help promote growth.

*Source: Annex Table 2, wpiea2022137-print-pdf*

### References

### wpiea2022137-print-pdf - References

### Key references cited
- Adrian, Tobias, Nina Boyarchenko, and Domenico Giannone, 2019. “Vulnerable Growth.” American Economic Review, Vol 109(4): 1263–89
- Adrian, Tobias, Federico Grinberg, Nellie Liang and Sheheryar Malik, 2018. “The Term Structure of Growth-at-Risk.” IMF Working Paper 2018/180.
- Ang, Andrew, Monika Piazzesi, and Min Wei (2006). “What does the yield curve tell us about GDP growth?," Journal of Econometrics, vol. 131(1-2), pages 359-403.
- Carvalho, Daniel, 2014, “The Real Effects of Government-Owned Banks: Evidence from an Emerging Market,” Journal of Finance, 69(2), pp. 577-609.
- Claessens, Stijn, Ayhan Kose, and Marco Terrones. 2011a. “Financial Cycles: What? When? How?” CEPR Discussion Paper 8379.
- Claessens, Stijn, Ayhan Kose, and Marco Terrones. 2011b. “How Do Business and Financial Cycles Interact?” CEPR Discussion Paper 8396.
- Demirguc-Kunt, Asli and Ross Levine, 2018. “Finance and Growth”, Volume 1, Edward Elgar Publishing.
- Gambacorta, Leonardo and Hyun Song Shin, 2016. “Why bank capital matters for monetary policy”, BIS Working Paper 558.
- Goodhart, Charles and Boris Hofmann, 2008. "House prices, money, credit, and the macroeconomy," Oxford Review of Economic Policy, vol. 24(1), pages 180-205.
- International Monetary Fund, 2017. “Global Financial Stability Report.” Washington, October 2017.
- La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer, 2002, “Government ownership of banks,” Journal of Finance 62, pp. 265–302.
- Levine, Ross, 2005. "Finance and Growth: Theory and Evidence," Handbook of Economic Growth, chapter 12, pages 865-934.
- Muduli, S. and Behera, H., 2021. Bank capital and monetary policy transmission in India. Macroeconomics and Finance in Emerging Market Economies, pp.1-25.
- Prasad, Ananthakrishnan, Selim Elekdag, Phakawa Jeasakul, Romain Lafarguette, Adrian Alter, Alan Xiaochen Feng and Changchun Wang, 2019. "Growth at Risk: Concept and Application in IMF Country Surveillance," IMF Working Papers 2019/036.
- Seth, Gaurav, Supriya Katti and B.V. Phani, 2022. “Stock Price Reaction on the Announcement of Basel Implementation: Evidence from Indian Banks” Reserve Bank of India Working Paper No. 1

### Annex I — Definitions and data sources (selected variables)
- Real GDP Growth: Real GDP at Market Prices, % Change - YoY — Haver Analytics / Central Statistics Office
- Policy Rate: Repo Rate (EOP, % per annum) — Haver Analytics / Reserve Bank of India
- Treasury bill yields (10 year): 10-Year Government Bond Yield (EOP, % per annum) — Haver Analytics / Reserve Bank of India
- Sovereign spreads: JPSSGINB Index — Bloomberg
- Stock price change: Stock Prices: BSE Sensex/BSE 30 Index (% YoY) — Haver Analytics / Bombay Stock Exchange
- Inflation Rate: Consumer Price Index % Change - YoY — Haver Analytics / Ministry of Statistics and Programme Implementation
- Current account deficit: BOP: Current Account Balance / Real GDP at Market Prices — Haver Analytics / Central Statistics Office and Reserve Bank of India
- Short term external debt to reserve ratio: Short-Term Gross External Debt / Intl Liquidity Reserves — Haver Analytics / Reserve Bank of India
- NPL ratio: Non-Performing Loans to Total Gross Loans (EOP, %) — Haver Analytics / International Monetary Fund
- Credit growth: Adj Credit by All Sectors to Nonfin Priv Sector (% YoY) — Haver Analytics / Bank of International Settlements
- Credit to GDP Ratio: Adj Credit to the Private Nonfinancial Sector (% of GDP) — Haver Analytics / Bank of International Settlements
- Credit to GDP gap: Private Nonfinancial Credit to GDP Gap (EOP, %) — Haver Analytics / Bank of International Settlements
- World GDP growth: Real GDP, seasonally adjusted, % YoY, World — International Monetary Fund, World Economic Outlook
- Oil price change: West Texas Intermediate ($/Barrel) (% YoY) — Haver Analytics / Energy Information Admin/Chicago Mercantile Exch
- Exchange rate change: India: Rupee/US$ Exchange Rate (AVG) (% YoY) — Haver Analytics / Reserve Bank of India

Additional bank-level variables (definitions and sources):
- Leverage ratio (total equity): Total equity divided by total assets (%) — FitchConnect / Reserve Bank of India
- Leverage ratio (common equity): Total common equity divided by total assets (%) — FitchConnect / Reserve Bank of India
- Leverage ratio (Tier1): Tier 1 capital divided by total assets (%) — FitchConnect / Reserve Bank of India
- Capital adequacy ratio: Tier 1 capital divided by risk-weighted assets (%) — FitchConnect / Reserve Bank of India
- Cost of funding: Total interest expense divided by total debt funding excluding derivatives — FitchConnect / Reserve Bank of India
- Debt funding growth: Growth rate of debt funding — FitchConnect / Reserve Bank of India
- Growth of gross loans: Growth rate of gross loans (%) — FitchConnect / Reserve Bank of India
- Return on assets: Net income divided by total assets (%) — FitchConnect / Reserve Bank of India
- NPL: Total impaired loans divided by gross loans — FitchConnect / Reserve Bank of India
- GDP growth: Real GDP growth (%) — International Monetary Fund
- Policy rate: Repo rate (average %) — Haver / Reserve Bank of India
- Real effective exchange rate: Real effective exchange rate against 10 currency basket — Haver / Reserve Bank of India
- Exchange rate: Rupee/USD exchange rate, nominal — Haver / Reserve Bank of India
- US Policy rate: Effective Fed Funds Rate — Haver

### Annex III — Selected regression and panel results (preserving reported coefficients and statistics)
- Annex Table 3 / Cost of funding (PSBs, Private Banks) — sample period from 1998-2021:
  - Leverage Ratio (total equity/assets): -0.0399** (0.0189) ; -0.0208** (0.00975)
  - Leverage Ratio (common equity/assets): -0.0299* (0.0170) ; -0.0185* (0.0102)
  - Leverage Ratio (tier 1/assets): -0.0249 (0.0567) ; -0.0605*** (0.0196)
  - Capital Adequacy Ratio: -0.0205 (0.0144) ; -0.0229** (0.00999)
  - N 392 392 181 384 434 434 248 426
  - R2 0.877 0.876 0.875 0.872 0.777 0.777 0.819 0.783
  - Year FE Yes; Bank FE Yes; Bank controls Yes
  - Robust standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001.

- Annex Table 4 / Growth in debt funding (PSBs, Private banks) — sample period from 1998-2021:
  - Leverage Ratio (total equity/assets): 1.035** (0.477) ; 1.821 (1.212)
  - Leverage Ratio (common equity/assets): 0.791* (0.468) ; 1.762 (1.186)
  - Leverage Ratio (tier 1/assets): 2.100* (1.253) ; 4.794*** (1.430)
  - Capital Adequacy Ratio: 0.649* (0.339) ; 1.212 (0.835)
  - N 392 392 181 384 433 433 248 426
  - R2 0.580 0.578 0.736 0.579 0.293 0.292 0.577 0.290
  - Year FE Yes; Bank FE Yes; Bank controls Yes

- Annex Table 5 & 6 / Growth of gross loans (PSBs, Private banks) — sample period from 1998-2021 and subset 2010-21:
  - For full period:
    - Leverage Ratio (total equity/assets): 0.143 (0.463) ; 0.666 (0.605)
    - Leverage Ratio (common equity/assets): 0.166 (0.457) ; 0.682 (0.581)
    - Leverage Ratio (tier 1/assets): 2.145 (1.346) ; 1.826* (1.026)
    - Capital Adequacy Ratio: 0.431 (0.327) ; 0.741* (0.434)
    - N 392 392 181 384 433 433 248 426
    - R2 0.640 0.640 0.716 0.645 0.239 0.239 0.575 0.236
  - For 2010-21 subset (smaller N reported):
    - Leverage Ratio (total equity/assets): 2.149* (1.121) ; 2.276*** (0.624)
    - Leverage Ratio (common equity/assets): 3.499*** (1.280) ; 2.112*** (0.644)
    - Leverage Ratio (tier 1/assets): 1.870 (1.572) ; 1.990*** (0.675)
    - Capital Adequacy Ratio: 0.0812 (0.927) ; 1.407*** (0.299)
    - N 185 185 157 184 224 224 196 222
    - R2 0.701 0.710 0.707 0.692 0.648 0.644 0.673 0.656
  - Year FE Yes; Bank FE Yes; Bank controls Yes; Robust standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001.

- Annex Table 7 / Foreign banks (dynamic GMM panel; sample period from 1998-2021):
  - Leverage Ratio (total equity/assets): -0.0275 (0.0173) ; 6.603*** (1.186) ; 73.88*** (20.12)
  - Leverage Ratio (common equity/assets): -0.0253 (0.0173) ; 6.570*** (1.181) ; 73.40*** (20.17)
  - Leverage Ratio (tier 1/assets): -0.139** (0.0581) ; 2.980** (1.480) ; 2.085 (1.390)
  - Capital Adequacy Ratio: 0.00960 (0.0207) ; -0.127 (1.473) ; -7.719 (19.36)
  - N 868 655 837 979 527 879 795 278 (various columns)
  - Year FE Yes; Bank controls Yes
  - Standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001. Model estimated using the dynamic GMM panel.

### Annex II — Figures (notes on Growth-at-Risk robustness checks)
- Annex Figure 1: Robustness Check - Growth-at-Risk: Shock to Credit and Leverage (Bank Credit)
  - Results capture a two standard deviation negative shock to the credit and leverage partition (bank credit only), based on a growth-at-risk approach estimated using data from 2001Q4 to 2021Q3.
  - The credit and leverage partition captures the bank credit to GDP ratio, bank credit growth, the credit to GDP gap, and the NPL ratio.
  - The first three variables have positive loadings on the principal component, while the last variable has a negative loading.
  - Source: IMF Staff estimates.

- Annex Figure 2: Robustness Check - Growth-at-Risk: Term Structure for Credit Indicators (Bank Credit)
  - Notes: The horizons refer to quarters. The x-axis refers to the different quantiles q of the quantile regression. The y-axis refers to the coefficient β2q for the partition of credit and leverage (bank credit only) in the quantile regression.
  - The credit and leverage partition captures the bank credit to GDP ratio, bank credit growth, the credit to GDP gap, and the NPL ratio.
  - The results are based on the growth-at-risk approach estimated using data from 2001Q4 to 2021Q3.
  - Source: IMF Staff estimates.

*Financial Sector and Economic Growth in India — Working Paper No. WP/22/137*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022137-print-pdf.pdf_