The Banking Response to the Brexit Vote: Caution but not Closure
Dom Smith, Head of Debt Analytics
Among the many financial market considerations being speculated upon in the wake of Brexit is the appetite for bank lending to commercial real estate. A fall in capital values and consequent rise in LTVs on existing lending could push up capital requirements, making historic lending less profitable and diminishing the availability of funds for future lending. However, in analysing some of the more adverse anticipated market scenarios, we find that any capital shortfall is likely to be minimal – a minute fraction of that facing the industry in the 2008/09 crisis. So, Bank lending to commercial real estate will not be significantly curtailed by capital constraints brought on by post-vote market jitters. Furthermore, recent action by the Bank of England to reduce the counter-cyclical capital requirement will in effect act to dampen the impact of rising RWAs on balance sheets. Of course, other factors may influence lending behaviour on the part of the banks, but restructuring in the UK lending market means that borrowers are far less exposed to one type of lender than was the case a decade ago.
Modelling Impact on UK Bank CRE Lending of Different Market Scenarios
Following the vote for a Brexit, markets expect a modest decline in capital values, as investors generally move to a “risk off” approach causing buyers to retrench or to seek (though not necessarily be given) price chips. It is too early to say whether capital values will in fact fall, or by how much. However for the purposes of selecting an indicative range of outcomes to analyse, we can look at recent REIT share price falls. These suggest that evaluating the impact of a fall in the range of -5% to -15% would be reasonable – a significant change, but nowhere near something of the scale witnessed in the 2007-09 crash when peak-to-trough value decline was more of the order of -45% (and the largest quarterly decline in value was -15%). We therefore take three scenarios, of one-off declines of -5%, -10% and -15%, and model their impact on existing lending. To perform this analysis, it has been necessary to carry out a detailed study into the composition of the UK bank CRE loan portfolio. For a full understanding of this work, please see the Methodology section.
Figure 1 shows the impact on key LTV points, marking the assumed boundaries of the Slotting categories, of the three market decline scenarios outlined above.
[Note: Although we make no distinction between CRE sectors here due to the lack of reliable data on LTV by property type or characteristic, it is important to acknowledge that any decline would not be felt universally; some sectors may be more exposed to price decline from yield or rental growth factors than others, and an increase in risk aversion may favour some property characteristics (e.g. lease length) over others].
This shift of lending into Slotting categories with higher RWA has an obvious impact on increasing the amount of RWA both overall, and the proportion accounted for by weaker Slots. The one slight exception to this is when lending moves from the Weak to the Default Slot; here, RWA is freed up, as the RWA held against loans in Default is 0% (versus 250% in Weak). The drawback of course is that lenders have to make a provision of 50% of the balance of loans entering the Default Slot. So, in seeking to understand the change in RWA under different scenarios therefore, it is necessary to break out the increase in RWA, the value of provisions, and the RWA “reclaimed” from loans falling into Default. As Figure 2 and Figure 3 show:
- Under the Base scenario, total RWA for UK bank CRE lending is estimated at £63.5bn.
- When assuming a -5% decline in values, this figure rises by £1.5bn to £65.0bn, with a provision of £0.3bn for loans entering Default.
- Under the -10% value decline scenario, RWA rises by £3.1bn to £66.6bn, with a provision of £0.5bn for loans entering Default.
- The rise in RWA is much more significant under the -15% value decline scenario; here RWA increases by further £10.1bn to £73.6bn, with a provision of £0.9bn for loans entering Default.
So, if market values decline by -15%, we estimate that RWA would increase by 16% relative to the Base scenario. Under smaller declines of -5% and -10%, the increase in RWA would be much smaller at 2% and 4% respectively. Provisions for loans falling into Default Slotting treatment would range from £0.3bn to £0.9bn.
Estimating Change in Performance – The Impact on Return on RWA
With higher RWA under the various scenarios under consideration, one of the key measures of UK bank loan performance, Return on RWA (RoRWA), will come under pressure. Little data is available on the RoRWA of UK bank CRE lending specifically, so we model the impact of our various forecast RWA scenarios on RoRWAs ranging from 2-4%, as shown in Figure 4. We also, in Figure 5, express for this range of RoRWAs the proportion of one year’s income accounted for by the level of provisions associated with each value decline scenario.
- The impact on RoRWA is arguably minimal under both the -5% and -10% market decline scenarios; RoRWA of 3.00% in a Base scenario falls to 2.93% and 2.86% respectively.
- Under the -15% market decline scenario however, the impact is more noticeable; a 3.00% RoRWA dips by c40bps to 2.59%.
- Provisions as a proportion of one year’s income are also, as would be expected, more significant under the more drastic -15% market decline scenario, where they account for 30-59% of one year’s income, and 39% at the 3.00% RoRWA level.
- For the -5% and -10% market value decline scenarios, the proportion of one year’s income accounted for by provisions is much lower, ranging from 10-20% and 19-38% respectively.
- At the 3.00% RoRWA level, provisions account for 26% of one year’s income under the -10% market decline scenario, and just 13% of one year’s income under the -5% market decline scenario – or 3.0 and 1.5 months’ income respectively.
This is not 2008. CRE Lending Markets Should Remain Resilient
Our analysis of the three market value decline scenarios outlined above suggests that as long as any decline in values does not materially exceed 10% the impact on RWA associated with UK banks CRE lending should be, if not minimal, then certainly manageable. Provisions on the £76.6bn book will be of the order of a quarter to a half a billion pounds, potentially representing less than a quarter of one year’s income from CRE lending overall, and a miniscule fraction of the sorts of write offs endured following the crash in 2008 and 2009, when the total amount of write offs approached 10% (or c£25bn) of the value of loans at the peak of the market. Returns to lenders will obviously be slightly lower, but again, only marginally so.
So goes the theory. In practice of course, things may be different. Wary of recent history, it would be understandable if UK banks took a much more wary attitude to CRE lending. Even if volumes were unaffected, there is likely to be a strong desire among lenders to try to offset falls in profitability with improved margins on new lending, and to counter the need for increased RWAs on existing lending with lower RWAs on newer lending. The latter desire will mean that lending appetites are likely to become more conservative, both in terms of seeking to lend at lower LTVs and also preferring stronger property collateral to ensure new lending falls into the Strong or at worst Good Slots. As far as lending terms go, margins are likely to come under some upward pressure, though falling swap rates may offset the impact on borrowers, keeping total borrowing costs lower by historic standards.
To some extent, the above analysis applies to a Regulatory vacuum; in fact, the response from the Bank of England has been swift, with the announcement of the removal of the counter-cyclical capital buffer freeing up £5.7bn of capital requirements on banks. In effect, this acts as a damper to the rise of RWAs and, all things being equal, will enable banks’ Return on Capital to be protected, even if RoRWA diminishes. Whatever the impact on lending appetite may have been without such intervention, it is likely to have been influenced positively by the BoE’s move.
Borrowers may be sheltered further from (perceived) problems in this part of the lending community. UK banks remain the largest single source of lending to CRE, but they are no longer dominant; lenders other than UK banks accounted for 66% of originations in 2015. The last few years have seen a significant rebalancing of the UK CRE debt market away from an over-reliance on one single type of participant and towards a broader base. Insurers for example have been attracted to the sector by its capacity for liability matching and the relatively high returns on offer versus other forms of credit, in particular Gilts. These are advantages which will not disappear in the short-term and which have, in the latter case, improved since the vote to Brexit – 10 year Gilt yields have fallen by c50bps in the last week.
While there may be much noise therefore around the vulnerability of the UK CRE lending market to the more volatile times we now find ourselves in, we believe that fundamentally, provided value declines do not approach the teens, the sector will remain robust.
Methodology: Understanding the UK Bank CRE Lending Portfolio
In order to calculate the change in capital requirements for lenders under different market scenarios, a large amount of loan level information would be required, including LTV, duration, collateral details and current Slotting treatment and policy. At present however, only a limited amount of aggregated data is available on UK commercial real estate (CRE) loan books, and combining market-level statistics to provide additional granularity often requires making significant assumptions. This may change in the future, most obviously if work on the PIA’s Loan Database is fruitful, but for now we rely on an element of interpretation.
At the end of 2015, drawn funding to CRE provided by UK banks and building societies (“UK banks”) stood at £76.6bn according to DeMontfort University’s “Commercial Property Lending Report”. This is the element of the total lending market – £168.4bn of drawn funding to CRE – that will be the focus of this analysis, representing as it does the best available estimate of the size and nature of the UK banks’ CRE lending exposure. Without precise, granular detail on Slotting treatment, the next best thing would be for this £76.6bn to be split out by LTV, length of outstanding loan term, and property criteria such as sector, geography, lease length etc. Unfortunately, while data is available on the repayment profile, mix of LTVs and geography/sector of underlying collateral, this data is frequently on a differing sample, and moreover is siloed, making more detailed investigation involving overlaying one level of disaggregation (e.g. repayment profile) over another (e.g. LTV split) problematic.
Figure 6 illustrates these issues, and shows how we have attempted to make the best of the available data. Beginning with the £76.6bn UK bank CRE lending, we use the available data on repayment profile (Chart 1), which details the proportion of debt scheduled for repayment in each of the five years to 2020, in the five years collectively thereafter, and beyond 2025. This breakdown refers to £157.1bn of lending from all lenders; it is therefore necessary to make some assumptions as to what this tells us about the subset of UK bank CRE lending. Our key assumptions – which admittedly are likely to be a slight over-simplification of reality – are to assume that UK banks CRE lending is restricted to lending of five years or less, and that the distribution of repayment profile matches that of the 2016-2020 repayment profile of the whole lender community. Thus, we estimate £17.0bn of UK bank CRE debt scheduled for repayment in 2016, £12.8bn in 2017, £13.0bn in 2018, £15.4bn in 2019 and £18.4bn in 2020 (Chart 2). For the purposes of assessing RWA under Slotting, this equates to £36.3bn of debt repaying in less than 2.5 years, and £40.3bn of debt repaying in more than 2.5 years.
The next stage of analysis is to attempt to overlay current LTV onto this repayment profile. Data is available on the broad distribution of current LTVs (Chart 3) on a subset of £141bn of lending by banks, building societies and insurance companies (irritatingly, a different subset again from the two already used). If it is assumed that the £76.6bn of UK bank CRE lending has an identical LTV profile to this wider subset, then this debt can be assessed through a combination of repayment and LTV (Chart 4). Thus, the UK bank CRE lending due for repayment in 2016 is broken down into £6.3bn at <50% LTV, £8.5bn at 51-70% LTV, £1.0bn at 71-85% LTV, £0.3bn at 86-100% LTV, £0.3bn at 101-120% LTV and £0.5bn at >120% LTV. Similar breakdowns are also produced for 2017-20 repayments.
We acknowledge that the above will be an over-simplification of then picture; UK bank CRE lending is likely to include some debt dated beyond 2020 (though not as much as would be expected to be found in say an insurer’s lending portfolio), while the LTV profile may be skewed such that higher LTV lending is more likely to be of shorter duration (it will presumably have been originated at more usual levels and will have “needed” longer to have deteriorated). In the absence of more detailed data however, we do not wish to make assumptions that contradict what little evidence is available.
A further desirable step in adding to our understanding of UK bank CRE lending would be to introduce to the above a sector and geography dimension, splitting out the LTV and repayment pieces by type and location of property collateral. Again, very high level data is available from the DeMontfort study splitting out the whole lending sample by sector (retail, office, industrial) and region (six regions), as shown in Figure 7, but again it is limited by the sample sets being different, and by one measure not overlaying the other; there is no regional dimension to the sector data, making it impossible to understand whether, for example, Central London accounts for more or less than 43% of office lending. We therefore judge that it would be a set of assumptions too far to apply a sector/geography split to the repayment and LTV.
The final stage of understanding the existing UK bank CRE loan portfolio is to attempt to break out the book by Slotting category. It is important to understand that what follows is, once again, a necessary simplification; a wide range of variables go into determining Slotting category of which LTV will be just one. However, in the absence of any granular detail on other factors, we have here assumed that these would be neutral across the sample and in effect used LTV as the single determinant in setting Slotting category. Figure 8 sets out the RWAs ascribed to each Slotting category by the Regulator together with our assumptions, for the purposes of this model, on the LTV boundaries for UK bank CRE lending. We assume that lending at sub-50% LTV will fall into the Strong category, 51%-65% LTV lending will be classified as Good, 66%-80% as Satisfactory, 81%-100% as Weak and above 100% as Default.
By making a final assumption that within each LTV bracket lending is evenly spread, these boundaries can thus be imposed onto the profile of the book arrived at above (Figure 6), to break out lending by LTV, repayment and Slotting category, as shown in Figure 9. This in turn allows an estimate of RWA for the whole UK bank CRE loan portfolio to be calculated, as shown in Figure 10. Thus, we arrive at the following estimates for UK bank CRE lending:
- £63.5bn of RWA on lending of £76.6bn,
- £16.6bn of RWA on £27.4bn of Strong lending,
- £22.4bn of RWA on £27.8bn of Good lending,
- £16.4bn of RWA on £14.3bn of Satisfactory lending,
- £8.2bn of RWA on £3.9bn of Weak lending.
Note that no RWA is assumed on lending in Default, where a provision of 50% of the value of the loan should have been made; we assume that this has already taken place.
To find out more about our Debt Analytics specialism and how we can help you, contact Dom Smith, Head of Debt Analytics on +44(0)207 182 2369.