Macroprudential regulation and bank behaviour: theory and evidence from a quasi-natural experiment

Abstract

The article examines the impact of macroprudential policies on bank credit growth. Towards this end, we develop a model of bank behaviour which examines the possible impact of such policies. The testable propositions of the model are empirically examined using a natural experiment for India. The results appear to suggest that macroprudential policies interact with bank ownership to moderate the severity of the credit cycle.

Keywords:

• macroprudential policies
• banking
• India
• risk weights
• provisions

JEL classification:

• G21
• G28
• E51

Acknowledgements

I would like to thank, without implicating, Prof. Romar Correa and especially two anonymous referees for the insightful comments on an earlier draft which greatly improved the analytics and exposition.

Notes

1. As one referee pointed out, given the portfolio of loans, the interest spread may be reduced when cross-elasticities of demand between bank loan products are taken into account, yielding possible diversification benefits to the bank. For purposes of analytical tractability, we chose to ignore such strategic substutability/complementarity between loans.

2. Since the focus is primarily on MPPs, we do not complicate the framework using differential rate for insured and uninsured debt (see, e.g., Blum 2002).

3. Accordingly, θ can be thought of as a proxy for aggregate credit risk.

4. Following from our earlier discussion, (1−θ₁) > (1−θ₂) and likewise, (1−q₁) > (1−q₂). In other words, riskier loans necessitate higher standard provisions.

5. The implicit assumption is that the returns on L₁ and L₂ are statistically independent.

6. Formally, min {[(1 + r₁)*L₁, (1 + r₂)*L₂] + (1 + i)*B} > (1 + r_D)*D > (1 + i)*B.

7. To see this, we re-arrange Equation (7) to obtain: r_D* = i/(1 + δ_D), where 1/(1 + δ_D) is the ‘mark-down’ on deposits.

8. Define K = w₁*(L₁/L) + w₂*(L₂/L), so that the capital constraint E ≥ w₁*L₁ + w₂*L₂ can be rewritten as (dividing and multiplying the RHS by L): E ≥ K*L. This formulation emphasizes the fact that the ‘average’ capital requirement depends on the composition of the loan portfolio.

9. Note that making an additional unit of loan to a type j borrower induces the bank to raise its equity base by ω_j thereby raising the opportunity cost of equity.

10. As one referee pointed out, in general, provisions are done for expected loan losses. Furthermore, the bank managers through capital management can enhance their credit growth on the basis of this provision (as part of tier-II capital), which may exacerbate the adverse selection and risk taking of banks. To examine this further, prior to our empirical analysis, we test the income smoothing and capital management practices through loan loss provisions. Accordingly, we test the following specification for bank b at time t:

where LLP = loan loss provisions, CRAR = capital adequacy ratio, RoA = (net profit/asset), Log Asset = log of bank asset (to control for scale economies), FEE = fee income to total income, TD = time dummies (to control for year effects), OD = ownership dummies and ε is the error term. The bank-specific variables are lagged one period in order to address endogeneity concerns. The coefficient φ₁ < 0 would be indicative of capital management, whereas φ₂ > 0 would be supportive of income smoothing.

The results in the following table appear to provide weak evidence in support of capital management (negative coefficient on CRAR, which is significant at the 0.10 level), although the evidence regards income smoothing (positive coefficient on RoA) is less compelling.

Table

11. MPPs were also undertaken for ‘Other Retail’ wherein the provisions were altered, although the RWs were kept unchanged.

12. MPPs were imposed beginning December 2004.

Additional information

Notes on contributors

Saibal Ghosh

Saibal Ghosh is working as an officer with the Reserve Bank of India. His primary research interests are in the areas of banking and financial stability.