Stress tests and the countercyclical capital buffer: the UK experience

A pedestrian shelters under a Union Flag umbrella in front of the Bank of England, in London, Britain August 16, 2018.  REUTERS/ Hannah McKay - RC1927B08FA0
Editor's note:

The remark was delivered at the 50th Anniversary Conference of the Money, Macro & Finance Research Group at London School of Economics on September 4, 2019.

Unlike its US counterparts at the Federal Reserve, the Bank of England’s Financial Policy Committee (FPC) has made active use of countercyclical capital buffers in its effort to keep the resilience of the financial system in line with the risks to it, Donald Kohn, the Robert V. Roosa Chair in International Economics at Brookings and a member of the FPC, said in a lecture delivered at the London School of Economics on Sept. 4, 2019.

A countercyclical capital buffer (CCyB) is an extra layer of capital that authorities can require that banks hold in periods when credit growth is strong, so capital is available to absorb losses when times turn bad. In the UK, the CCyB rests in the region of one percent in normal or standard risk environments. The CCyB is raised in good times as credit growth picks up and lending terms tend to get easier; it supplements other bank capital requirements which are based on the riskiness of banks’ assets and can actually reduce minimum capital requirements in good times. The CCyB may be cut when risks materialize to encourage lending.

In the UK, stress tests are an important input into FPC’s decisions on the CCyB. UK and US stress tests differ in a number of dimensions, but share the goal of assuring that banks will have enough capital to continue to lend in bad times as well as good, avoiding a situation like 2008-09 when bank tightening of credit made a bad situation worse.

The FPC has adjusted the CCyB, partly based on stress test results. It began to raise it toward one percent in early 2016, based partly on the 2015 stress test results and the judgment that risks were in the standard range. It cut the CCyB to zero as financial risks began to materialize after the Brexit referendum in June 2016 and encouraged banks to use the capital released to support lending to UK households and businesses. As markets settled down, the CCyB was subsequently raised to one, as the risk environment was seen as standard apart from Brexit. (The CCyB in the US is currently set at zero.)

Stress tests are based on scenarios that are actively countercyclical. When the unemployment rate is low, the stress test scenarios foresee a larger increase than when the unemployment rate is already high. When house prices are high, the scenarios foresee steeper drops than when house prices are low. Projected losses from credit and market risks have risen from 2016 to 2017 to 2018 as the economy has grown and credit has loosened to a degree, especially for mortgages. But factors other than scenario design (for example, loan loss accounting rules) account for some part of that increase.

Stress test results have also been applied to gauge the effects of Brexit on the banking system. Even a disorderly Brexit scenario was seen to be less severe and produce fewer losses than the 2018 stress test, so the FPC could say that the UK banking system would be resilient to even a worst-case disorderly Brexit. Stress tests have also been used to drill down on particular types of loans—for example, consumer loans, where the microprudential supervisor used the results to make sure capital matched risk in particular banks.

Research and modelling are important and developing inputs to the stress tests and countercyclical macroprudential policy. The Bank of England has an active program to check on bank models and to capture contagion and feedback in the financial system as well as to identify tail risks. But much remains to be done. Kohn concludes by encouraging the academic audience to join the efforts to model contagion channels and the interaction of the financial sector with the real economy to support policy to protect financial stability.

Read the full remark here»