Over the course of the last twelve months there have been increasing signs that the largest banking groups are well on track to achieve their long-term target capital levels.1
This is good news and consistent with our view that bank balance sheets have significantly de-risked such that a bank-led financial crisis is very unlikely in the short run.
However, we have been examining a key post-crisis theme of Balkanisation which we believe highlights a much sharper differentiation between the prospects for the top tier banking groups and the also rans. Furthermore it also highlights the benefit of being able to operate in a simple legal structure.
Balkanisation is a term used to describe the impact of a variety of regulatory changes which demand more capital to be locked up in the jurisdiction in which the relevant business activity takes place. This can take many forms. For example:
• ring-fencing in the UK
• new holding company requirements for non-US banks operating in the US
• similar requirements proposed for Europe
• consumer protection rules requiring that marketing of product can only be undertaken by local regulated banks; or
• exchange control rules in emerging markets.
This week you may also have seen discussion on whether the withdrawal of EU-passporting means that European banks will need to create UK subsidiaries to continue to operate in London – the flip side of the more obvious concern about UK banks operating in Europe post-Brexit.
We consistently believed that the dramatic increase in these requirements was going to be a major headache for certain banks and one of our common investment themes is that markets do not fully understand the consequences of Balkanisation.
Specifically, we have been studying how the limitations on deploying and repatriating capital will impact traditional diversification effects. Historically, the allocation of capital from one business to another was mainly an internal accounting process fully within the control of the bank’s executives. However, as banking groups become siloed, there are now legal restrictions, delays and the need for regulatory approval preventing the free movement of capital from one part of the group to another.
We have analysed the volatility of quarterly returns (RoE) for a group of the top 30 global banks over the last 10 years. We grouped the 30 banks into three peer groups based on their post-crisis volatility of RoE (the top peer group had an average standard deviation of quarterly RoE of 2.7% compared to the bottom group which had 10.3% standard deviation).
We assumed that the volatility trends will continue but we assumed that all of the banks would generate an average return on equity of 8% in the coming years (in line with what the top tier groups have generated since the start of 2016 – a very generous assumption for the bottom tier group as we will explore in a forthcoming article).
Using these assumptions, we ran Monte Carlo simulations of the capital position of a hypothetical two-subsidiary banking group and compared to an identical bank which can operate on a fully integrated basis.2
The following chart shows the probability that the simulated banks with return characteristics of the different peer groups would exceed their capital buffers over the course of a year if operated in two groups or as a single entity.
Whilst we expected this to be an issue for certain banking groups, we were surprised by how stark the differentiation is.
Whilst this may sound obvious, the best run banks with the lowest volatility of returns have major competitive advantage. They can run safely with significantly lower capital buffers. A bank with low volatility of returns can run 1-2% lower capital buffers than its more volatile competitors and this would mean a RoE outperformance of c8-17% (ie RoE of 8% vs 6.9% for its more volatile peer) all other things being equal.
Equally the effect of operating across two subsidiaries is a major drag on returns. To operate with a similar likelihood of consuming capital buffers a two-entity group would need 1-2% higher level of capital buffer compared to an equivalent fully integrated group.
- For example the recent Bank of England Stress Tests included some large economic shocks and all banks passed the test.
- We have necessarily had to make some other significant simplifying assumptions. For example we have assumed that bank quarterly RoEs are normally distributed. Whilst this is almost certainly not true in practice our analysis going back to the end of 2008 shows that, as a group, the returns are actually very close to normally distributed with a moderate fat tails on the downside.
- We have also assumed that banks’ different business units profits are 50% correlated with each other – an examination of this is well beyond the scope of this article but what academic evidence there is suggests that the returns across different business lines do have a degree of correlation. We also ignore the fact that returns of a given bank are likely to be auto-correlated. Adjusting for this would increase the winners vs losers effect of the results.
- Underlying data was sourced from CreditSIghts
- Some other observations we draw from this analysis are:
- Large retail franchises in each subsidiary are a significant benefit as they have historically reduced volatility meaningfully
- The effect is the biggest issue for banks where the various subsidiaries are of similar sizes or material in absolute terms. Whilst a bank with large operations in one country and much smaller in another would have some impact from these effects they are likely to be manageable within the overall group buffers.
- We have modelled based on the assumption of RoE being the only variable and balance sheets remaining static. Banks will always operate with some capital buffer and in practice have a range of management actions they can take to manage short falls but none of these are easy or costless (eg sell assets, reduce trading risk, cut costs etc).