The recent financial crisis and economic contraction that followed highlighted the crucial role that banks play in facilitating the extension of credit and enabling economic growth. This underlies the economic rationale for imposing regulations on the banking industry, including minimum capital requirements designed to mitigate risks banks would not otherwise account for in their behaviour.
Macroprudential policy also includes provision for dampening cyclical over-exuberance through a regime of capital buffers on top of prevailing microprudential regulatory capital requirements. Such a ‘countercyclical capital buffer’ is part of the Basel III accord and also part of the toolkit of the Bank of England’s Financial Policy Committee (FPC). The capital buffer could be increased in a credit boom in order to generate greater self-insurance for the system as a whole and act as a restraint on overly exuberant lending. A sectoral capital requirement also allows the FPC to change capital requirements on exposures to specific sectors of the economy.1
These developments have raised the issue of how increases in regulatory capital ratios are likely to affect the broader macroeconomy. There is a high degree of uncertainty as to how banks might respond to future increases in macroprudential capital ratio requirements, the effect of such responses on the real economy, and how this might vary depending on the prevailing economic circumstances and state of the business cycle. For example, in periods when there are concerns about the strength of financial institutions, an increase in macroprudential capital requirements will likely support resilience and lending.
Conversely, in an environment where market participants perceive risks to the financial system to be small, banks may be able to borrow at a rate that is relatively insensitive to how much capital they have. In that case, an increase in macroprudential capital requirements could cause banks’ cost of funding to rise. Banks might pass this increase in funding costs on to their borrowers by raising interest rates on loans, and/or reduce the quantity of credit they extend. This might, at least in the short term, lead to a tightening in credit conditions for the real economy, helping to arrest the build-up of vulnerabilities created by an overextension of credit (see Tucker et al. 2013).
Estimating the effect of a countercyclical capital buffer on the macroeconomy
Estimating the effect of the future operation of a countercyclical capital buffer on economic variables is also complicated by the fact that such a policy tool has never before been used. There are, moreover, very few changes to aggregate regulatory capital requirements observable in past data. And for those changes in regulatory capital that have occurred, it is difficult to isolate how much of the change in bank lending behaviour was as a result of regulation, rather than broader macroeconomic developments affecting the prospects for banks or the health of their balance sheets.
The existing literature proposes two broad methods for surmounting this problem. One strand of literature attempts to estimate the impact of future macroprudential policy by explicitly representing the dynamics of banks' balance sheets using dynamic stochastic general equilibrium (DSGE) models (BIS 2010 provides a summary). A second seeks to proxy the effect of future changes in macroprudential requirements by performing a ‘bottom-up’ estimation of the effect of past changes in observable microprudential ‘Pillar 2’ regulatory capital requirements (Aiyar et al. 2011, Bridges et al. 2014). But neither is without caveats. In particular, there are reasons to believe that such positive shocks to individual Pillar 2 capital requirements are an imperfect proxy for increases in capital requirements affecting all banks simultaneously, not least given how in the latter case, lending could less easily shift to other banks (or to shadow banks – see Meeks et al. 2014).
A ‘top-down approach’
In Noss and Toffano (2014), we offer an alternative approach based on a study of the ‘top-down’ joint dynamics of the aggregate capital ratio across all UK-resident banks and a set of macro-financial variables, including lending growth. This attempts to identify shocks in past data that match a set of assumed directional responses of other variables to future changes in aggregate bank capital requirements (a similar technique has been used in the recent monetary policy literature aimed at disentangling the effect of credit demand and supply shocks (see, for example, Barnett and Thomas 2013).
This approach assumes that an increase in banks’ aggregate regulatory capital has a negative effect on the provision of bank lending, at least in the short run (this follows from literature examining the effects of shocks to credit supply (see discussion in Hristov et al. 2011). In order to identify this type of credit supply shock, an increase in regulatory capital is also assumed to be associated with an increase in issuance of bonds by non-financial firms (as firms substitute their borrowing away from that from banks), and a decrease in the return on bank equities relative to that of the rest of the market, reflecting a decline in the profitability of banks as they forego otherwise profitable lending opportunities.
To the extent that policymakers concur with the directional response of macroeconomic variables to changes in macroeconomic capital requirements assumed in this model, its outputs may – in certain states of the economic cycle – provide a plausible ‘upper bound’ on the short-term effects of future increases in aggregate capital requirements – intended to increase resilience in the face of a credit boom.
Figures 1 and 2. Median impulse responses of a 15 basis-point (one standard deviation) structural shock to the change in UK bank capital-to-asset ratios (equating to a 12 basis-point permanent increase in its level).
Note: Dotted lines show 16th/84th percentile (+/- one standard deviation) responses.
Figure 1. Response of M4 Lending growth
Figure 2. Response of GDP growth
Figures 1 and 2 show the impulse response functions for M4 lending and GDP growth resulting from a 15 basis-point increase in the first difference of the capital-to-asset ratio of banks operating in the UK. This is associated with a 0.25 percentage-point reduction in quarterly lending growth after two quarters, the effect of which fades to zero after around 20 quarters. The effect is significant, at the 16th/84th percentiles, for around five quarters.
The impact on quarterly GDP growth is insignificant, though the median impulse response function reaches its maximum on impact of -0.08 percentage points after two quarters. This insignificant impact on output is consistent with firms substituting away from bank credit and towards that supplied via bond markets, in response to a reduction in bank lending as a result of an increase in regulatory capital.
Also of interest are the cumulative impacts of the shock to regulatory capital on the levels of other variables, as well as their growth rates. Figures 3 and 4 show the cumulative effect on the levels of M4 lending and GDP. The level of lending is reduced by approximately 1.4% after around 17 quarters while GDP is cumulatively reduced by 0.25%.
Figure 3. Level of M4 Lending
Figure 4. Level of GDP
Note: Dotted lines show 16th/84th percentile (+/- one standard deviation) responses.
It is also important to note, however, that the set of sign restrictions assumed here – whereby an increase in capital requirements has a contractionary effect on lending – is likely to apply only during a boom in the extension of credit, such as that witnessed pre-crisis. Indeed, if we adapt the specification by omitting the sign restriction on lending growth, the response of lending to an increase in capital ratios is weakly positive. This suggests that the results are highly contingent on the state of the economic cycle.
In particular, they may not match the response of banks to regulation post-crisis, where, for example, high levels of macroeconomic uncertainty might have led market participants to be very concerned about banks’ vulnerabilities to economic shocks, rendering banks’ borrowing costs highly sensitive to the amount of capital used to finance their lending. Banks may be reluctant to raise capital unilaterally and may not be sufficiently profitable to generate capital organically. But in such circumstances an increase in macroprudential capital levels could improve investor confidence in the health of banks, allowing their cost of funding to fall, and thus rendering them able to increase their level of capital without decreasing their lending. Evidence of this can be seen in the bank capital raising that followed the recent US Supervisory Capital Assessment Program (SCAP), which appeared to increase confidence in the banks concerned and allowed them to increase their level of capital and increase their level of lending.
Estimating counterfactual paths of lending and output
Given these results, it is also possible to compute counterfactual paths of variables had regulatory capital been increased pre-crisis. This is an interesting simulation exercise, in that it offers policymakers an estimate of how their intervening to increase aggregate capital requirements pre-crisis might have affected lending and output.
The scenario examined here is that in which a macroprudential policymaker had intervened to maintain bank capital ratios at their 2006Q1 level (of just over 6%). This counterfactual capital ratio is greater than that actually witnessed pre-crisis, but less than that observed since the crisis. It is illustrated in Figure 5. Intuitively, an increase in macroprudential capital requirements during 2006–7 would have increased bank resilience and reduced the severity of the subsequent crisis.
Figures 5–10. Counterfactual paths of M4 and growth, had banks’ regulatory capital requirements been held at 6.1% from 2006:1
Note: Solid lines show actual data; dotted lines show counterfactuals corresponding to the 50th, 16th and 84th percentile responses.
Figure 5. UK-operating bank capital ratio
Figure 6. Credit:GDP ratio
Figure 7. M4 lending growth
Figure 8. GDP growth
Figure 9. Level of M4 lending
Figures 7–10 show the range of effects on the growth in, and level of, lending and GDP corresponding to such a policy experiment. The effect of this counterfactual increase in capital ratios is to reduce median quarterly lending growth by up to 1.5 percentage points in 2006–7, but for this effect to be reversed from mid-2008, with lending growth taking a value up to 2.7 percentage points greater than that actually witnessed. The overall effect on growth is far less marked. The median counterfactual ratio of credit-to-GDP (Figure 6) is less than its actual outturn during 2006–9; but the resilience of lending means that it shows no decline since.
Under the assumption that a macroprudential shock works in the way specified in the paper, these results indicate that intervention by a macroprudential policymaker that ‘smoothed-through’ declines and increases in aggregate bank capital ratios, can smooth the peaks and troughs in the lending and credit-to-GDP cycles.
An extension to sectoral lending
In Noss and Toffano (2014), we also extend our results to consider how the effect of aggregate bank capital requirements differs across lending to different sectors. The effect on lending to households appears far weaker than that to private non-financial corporates (PNFCs), with the maximum reduction in the growth of lending to PNFCs being around three times that to households. The difference in effect between the two sectors may arise because the risk weights on banks’ lending to households, which tends to have a lower write-off rate of loans where the borrower cannot repay, are lower than those on lending to corporates, meaning that a change in macroprudential regulation causes banks to reduce household lending less than that to other sectors.
In Noss and Toffano (2014), we provide an estimate of the possible effect of future changes in aggregate regulatory capital requirements – such as those instigated by a macroprudential policymaker – on the macroeconomy during an upswing in the credit cycle. We use a top-down approach to identify shocks to capital in the past data whose associated movement in other macro variables matches one set of possible priors as to the expected response to future changes in macroprudential capital requirements during a credit boom.
This analysis concludes that an increase of 15 basis points in aggregate capital ratios of banks operating in the UK (unweighted for risk) is associated with a median reduction of around 1.4% in the level of lending after 16 quarters. These results may be of use to a macroprudential policymaker, and may offer an ‘upper-bound’ as to the short-term effect of future changes in system-wide capital requirements. The impact on quarterly GDP growth is statistically insignificant, a result that is consistent with firms substituting away from bank credit and towards that supplied via bond markets.
Aiyar, S, C W Calomoris, and T Wieladek (2012), “Does Macropru Leak? Empirical Evidence from a UK Policy Experiment”, Bank of England Working Paper 445.
Bank of England (2013), “The Financial Policy Committee’s powers to supplement capital requirements”, Draft Policy Statement.
Barnett, A and R Thomas (2013), “Has weak lending and activity in the United Kingdom been driven by credit supply shocks?”, Bank of England Working Paper 482.
BIS Macroeconomic Assessment Group (MAG) (2010), “Assessing the macroeconomic impact of the transition to stronger capital and liquidity requirements – Final Report”, group established by the Financial Stability Board and the Basel Committee on Banking Supervision, August.
Bridges, J, D Gregory, M Nielsen, S Pezzini, A Radia, and M Spaltro (2014), “The Impact of Capital Requirements on Bank Lending”, Bank of England Working Paper 486.
Hristov, N, O Hulsewig, and T Wollmershauser (2011), “Loan Supply Shocks during the Financial Crisis: Evidence for the euro Area”, CESifo Working Paper 3395.
Meeks, R, B Nelson, and P Alessandri (2014), “Shadow banks and macroeconomic instability”, Bank of England Working Paper 487.
Noss, J and P Toffano (2014), “Estimating the impact of changes in aggregate bank capital requirements during an upswing”, Bank of England Working Paper 494.
Tucker, P, S Hall, and A Pattani (2013), “Macroprudential policy at the Bank of England”, Bank of England Quarterly Bulletin, 53(3): 192–200.