(Extract from page 48 of BIS Quarterly Review, December 2015)
A useful template for an analysis of the macroeconomic costs and benefits of the new leverage ratio (LR) requirement is the BCBS's 2010 long-term economic impact (LEI) report, which investigates the impact of the core Basel III capital and liquidity requirements.
The LEI methodology separates the assessment into two steps by: (i) assessing the expected benefits of higher capital requirements in the new steady state, specifically the reduction in the expected output losses from systemic crises; and (ii) comparing this with the expected costs of higher capital requirements in terms of forgone output. In deriving these estimates, the LEI adopts an explicitly conservative approach - making assumptions that tend to raise cost estimates and downplay expected benefits, introducing a downward bias into the estimates of expected net benefits. The key finding is that even large increases in bank capital requirements from their pre-crisis levels are unlikely to result in macroeconomic costs that outweigh the associated benefits in terms of reduced crisis costs.
Expected benefits. Conceptually, the expected benefits are based on multiplying the probability of systemic financial crises, given different minimum capital ratios, by the expected macroeconomic costs of such crises should they occur. To derive a link between crisis probabilities and different capitalisation levels, the LEI uses a range of probit models as well as portfolio credit risk analyses that treat the banking system as a portfolio of banks. Averaging the results from these models, it then derives a schedule with diminishing marginal returns (ie the extra effect of additional capital declines as the capital level increases). Later studies broadly confirm these results (eg Junge and Kugler (2013)).
Crisis cost estimates, in turn, are derived from academic studies of historical crisis experiences. The LEI study found that the median cost of systemic banking crises in these studies is 63% of GDP in net present value terms. But the variation in these cost estimates is large, and later studies have generated both higher and lower estimates (see Romer and Romer (2015) for an example at the lower end). A shortcoming of these studies of the cost of financial crises is that they rely only on pre-2007 data, missing the impact of the most recent crisis episode. An exception is Haldane (2010), who estimates the present value of output losses from the recent crisis to be between 90 and 350% of world GDP, depending on the strength of permanent effects. More recently, Ball (2014) confirms these results, with estimates implying that the weighted average cumulative loss across all OECD countries amounts to about 180% of pre-crisis GDP. Ball also finds that the growth rate in potential output has declined by 0.7 percentage points per year since the crisis and that, so far, there has been no reversal. If this decline in potential output is permanent, this would significantly increase potential crisis costs and would strengthen the case for action to prevent them.
Expected costs. If higher bank capital requirements raise banks' costs, banks may respond by raising their lending spreads to counterbalance the decline in their return-on-equity (RoE). As a result, real economy borrowing costs may rise, translating into lower investment and equilibrium output. To estimate the magnitude of this effect in the long run, the LEI assumes that banks maintain a constant RoE by passing on to their customers all additional costs that are due to higher capital requirements. The estimated increase in lending spreads is then fed into a variety of macroeconomic models (that is, the dynamic structural general equilibrium models and semi-structural and reduced form models in use at participating central banks) to assess the resulting impact on GDP.
The headline result of this exercise is that a 1 percentage point increase in the CET1/RWA ratio translates into a 0.12% median decline in the level of output relative to its baseline (with the corresponding value for the liquidity requirements being a one-off 0.08% decline in the output level).
In addition, a companion study (BIS (2010)) found that the macroeconomic impact of the transition to higher capital requirements was expected to be limited. By design, these results are likely to overstate the true costs, given the underlying assumption that the Modigliani-Miller theorem
is violated even in the long run and that the cost of issuing equity does not change with capital levels. In fact, banks' required RoE can be expected to decline as their balance sheet leverage and the riskiness of their equity fall.
For more details, see BCBS (2010). The original result of 0.09% is stated in terms of tangible common equity over Basel II RWA, which corresponds to about 0.12% in terms of CET1/RWA given our estimated conversion factor of about 0.78. The Modigliani-Miller theorem states that, under certain assumptions (such as the absence of taxes, bankruptcy costs, agency costs and asymmetric information), the value of a firm is unaffected by how that firm is financed (Modigliani and Miller (1958)).