Regulation of Financial Services

Finance services regulation is difficult to get right – knee jerk reactions often lead to unintended consequences and potentially the roots of the next bubble or crisis.

The UK financial services sector employs more than two million people in the UK with each one contributing nearly twice the value of an average worker to the economy, according to research from TheCityUK. Unfortunately the industry is blighted by a history of boom and bust cycles that have systemic repercussions through the economy as a whole. Banks, asset managers, or for that matter consumers, display herding behaviour with strong evidence that individual choices are often overwhelmed by group pressures to compete and survive.

Financial regulation is therefore properly targeted to control the “rules of the game” to minimise any damaging volatility in markets and yield positive outcomes for the economy as a whole, but not necessarily for all the individuals within it. This is not to argue there should be no safety net for individuals in our economy, just that financial prudential regulation is not the natural means to provide it. Contrary to much political comment over the past few years market failures cannot be wished away by hoping for better “moral” behaviour from the participants. The solution to market failures must lie with regulators – as a market participant you either play the game or you leave the market.

At a UK, European and International level, regulators have developed a range of responses to the financial crisis of recent years, and many of these rely on complex mathematical models of credit risk. Our company, 4Most Europe Ltd, was founded in 2011 to help UK banks, typically those with millions of retail customers, to implement credit risk models and prepare for increasing regulatory demands. It has been a busy time and the experience has provided us with insight into which regulatory developments are most likely to achieve their aims.

Good regulation vs rules with unintended consequences

  • Regulations which aim to increase the understanding of management, investors and the public of risks are effective, self-reinforcing and are likely to change emergent behaviour. Market wide stress testing initiatives have been particularly powerful in this regard – by requiring banks to consider explicitly the worst scenarios and publish the outcomes, the market has been driven to protect and plan for those events.
  • By contrast, regulatory interventions which simply represent a knee jerk reaction to minimise aspects of the market that are troubling, are likely to have unintended side-effects. For example the accounting rules to be introduced in 2018 that cover reporting of credit losses in banks financial statements (IFRS9) have the explicit objective of recognising a banks’ losses more fully and earlier in the economic cycle. This is predicated on the idea that on average the market can forecast the next recession, or indeed the green shoots of recovery.
  • Unfortunately economic models are much more useful to understand the past and present than forecast the future – as a result the side effects of the new IFRS9 regulations in practice will be to increase banks’ recognised losses in the early phases of a recession and thereby potentially increase the likelihood of bank failure (see “The Significance of IFRS9 for Financial Stability and Supervisory Rules” - Zoltán NOVOTNY-FARKAS study for the European Parliament, September 2015). To counteract this, banks will need to be more conservative in their lending in economically sensitive segments (e.g. small businesses and mortgage lending) and this could reduce efficiency and act as a drag to the financial services and the economy in general. 

The proper role of mathematical risk models in regulation

  • Models are a simplified representation of aspects of the real world. If the use of a model within regulation does not encourage participants to question the assumptions and provide insight, then the output of the model is of less value.  Requirements for reverse stress testing, where banks are asked to identify under how severe a stressed scenario they could cope before they would need to be wound up, are showing promise in this regard.  

Key Terms:

Stress testing – is a type of analysis that considers what the financial impact on a bank would be in terms of balance sheet and capital requirements should the economy fall into a specific stressed economic scenario. These tests have been used widely by US, European and other regulators to provide investor confidence in a bank’s robustness and ability to withstand systemic shocks.

Reverse stress testing – This is similar to stress testing but turns the problem on its head; specifically it asks banks to determine which scenario is most likely to result in a bank breaching certain levels of capital ratio in a recession.

Vasicek Model – this is a mathematical model that describes how the credit losses over a 12 month period for an idealised portfolio would be expected to be concentrated over time due to variation of a single hypothetical economic factor. The model is used with some standard, regulatory determined parameters, to calculate sophisticated Internal Ratings Based banks capital requirements under Pillar 1 of the Basel Accord (Basel II and III).

Tail risk – Banks lending money expect to lose a proportion of it on average due to customer financial difficulties. This can be priced into the lending product and is called the Expected Loss. Over time, banks will sometimes lose more than the expected loss and sometimes less – the spread of loss is called the Unexpected Loss. It is found that in most types of lending over a wide range of economic conditions the occurrence of abnormally large losses while small, is more likely than abnormally small losses.  This skew to the risk distribution is commonly described as a “non-normal” tail risk.

By contrast the formulation of the models underlying Pillar 1 of the advanced approaches to credit risk under the Basel Accord (and subsequently incorporated in EU law) were found to be less effective in the recent downturn and have needed to be augmented.

The underlying Vasicek model used in the fixed Pillar 1 regulatory model is implausible; the fact that tail risks are not normally distributed was popularised in Nassim Taleb’s “The Black Swan: The impact of the highly improbable”, 2007). Furthermore, the key regulatory correlation parameters are deliberately set by regulators to be a single value by asset class across the world.

Finally, it arbitrarily looks to assess maximum losses purely in a 12 month window whereas real world losses may span multiple years.  Nonetheless banks are blindly required to calculate a number that is neither comparable amongst institutions (the correlations in reality could vary by type of business) nor meaningful in its own right.  The objective was to have a risk sensitive measure of capital requirements that would be a level playing field for all institutions. Its failure during the banking crisis has led to the introduction of additional backstops (leverage ratio requirements) and the use of the far blunter instrument of simply increasing the quantity and quality of capital required which ultimately reduces investment and economic activity.

Conclusion

Countries face many challenges to ensure growth and prosperity for their citizens. For the UK, with its concentration in financial services, the risks are there but certainly no more challenging than those faced by resource or manufacturing based national economies. To throw away our advantage in this field in favour of more easily replicated expertise in manufacturing or technology would be foolish. We can enhance our regulation and utilise our mature legal infrastructure as a competitive advantage to be an attractive location for financial services. Increasingly this will need to use and reflect the insights provided by analysing the vast quantities of data that the industry generates.