Before getting onto the rationale behind publishing more market data, and having spent the first few years of my career building financial models based on such market data for banks & lenders, I thought I’d start by giving a brief introduction to modelling for those who are not familiar – it’s not as complicated as you might think!
I’ll use real estate as an example of the simplest sector to model, but the same basic principles apply to others. Note all data is for illustration purposes only!
- The request: someone wants to buy The Gherkin for £700m, and they want a loan of 60% on this, for 10 years – would you say Yes or No?
- I’d start with the “tenancy schedule” of the building, a list of each of the occupiers & their key rental terms. In the diagram below each of the 4 tenancies expires within the loan term, and modelling assumes a replacement tenancy after a void period (typically 6 months to a year, depending on the location and quality of the space)
- The model continues to assume replacement tenancies until the end of the loan period, and a few years afterwards also, since the lender must be comfortable that the building is well-occupied and so can be easily refinanced / sold at the end of the loan
Current tenancy schedule for the Gherkin, followed by modelled future tenancies
- All these tenancy inputs are guided by market experience. For example, in the diagram the top floors are retail (restaurant) which typically has a shorter tenancy length and void period than the office floors in the rest of the building. The expected inputs will then be “stressed” or subjected to more harsh inputs or “downside scenarios”, e.g. each tenant doesn’t pay as much rent as the previous, or it takes longer than anticipated to find a new tenant etc.
- This forms the asset’s future income projection over the life of the loan – but not all of the rental income above is profit, there are usually various costs associated with acting as landlord e.g. tenancy agency fees, service charges, and rent incentives – the level of these costs varies over time. Here there are £3m costs in the first year, so a net income of £27m:
- Finally, we add on the debt cashflows. The request is for a 60% loan on a building worth £700m i.e. loan of £420m. Let’s say the rate of interest is 2% p.a. + loan capital repayments of £4m p.a. = a total “Debt Service” of £18m in the first year
- Now we can answer the key question – can the asset’s cashflows support the debt payments? In this case, YES, since the net income is at the lowest £20m p.a., even with our harsh downside scenario assumptions, which is enough to pay the debt costs of £18m p.a. But this is quite a tight “coverage ratio” (£20m / £18m = 1.1x) i.e. there is only just enough, so a cautious lender should build in loan protections
Whilst (as noted above) the same basic principles apply to other sectors, in fact the modelling of other sectors can get much more sophisticated than this. In asset classes where the data is more readily available, there are various methods of simulation that can be performed to more accurately emulate multiple possible future cashflow paths all at once and consider their likelihoods, based on historic data and calibrated by expected input parameters. The most common of these methods is Monte Carlo simulation, and is well-explained elsewhere: http://www.investopedia.com/articles/07/montecarlo.asp
Chart of different net rental income paths, simulated using Monte Carlo methods
Real estate in the UK, however, is a part of the financial market that is relatively stunted by its lack of data. There are various reasons for this:
- They are often huge in size and as physical assets, they take longer to buy/sell: compared to say a £420m bundle of index stocks, in which you can take comfort that there will be some natural diversity, in the above example the £420m investment is entirely dependent on the performance of 1 building in 1 location and in 1 asset class! Therefore a lot of extra due diligence must be performed in advance to ensure the investment is a safe one, and the legal documentation cannot be easily standardised.
- …so they are less frequently traded and as a result, a less well developed system is in place to record these relatively infrequent deals.
- There is also a privacy angle: the people involved in the sale are often sensitive about publicising the details, since they try to limit potential competition when it comes round to refinancing, and also if the lender wants to later sell part of their loan on, it can help if the end buyer is unaware how much they’re making on it!
All this has led to the UK real estate market being very data-hungry, and while this need is now broadly starting to be met on the asset side by the major property firms such as CBRE and JLL, the UK real estate debt market is still very data-light.
In the US the real estate data shortage is less critical, with loans being recorded more regularly on a centralised basis. The Bank of England is starting to wake up to the need for enforcing such a system here in the UK also, as per their discussion paper of May 2014 (http://www.bankofengland.co.uk/publications/Documents/news/2014/dp300514.pdf). They aim to use the system to spot warning signs of upcoming overheating and when they might need to intervene. This will inevitably take a few years to come to fruition…
In the meantime, the UK investment management and advisory business, Laxfield Capital, has been in a unique position in the commercial real estate market, by having a well-spread selection of lenders that they act for, and also having many connections on the borrower side. They have made good use of this by rigorously recording and updating key facts about all the deals that they have assessed or seen in the market since 2012.
This in turn led to Laxfield asking me to produce a semi-annual market-wide publication, the “Laxfield UK CRE Debt Barometer” (www.laxfieldcapital.com/laxfield-uk-cre-debt-barometer), based on the aggregated trends in this data (so as to not breach an individual deal’s privacy) and applying market knowledge to interpret the insights noted. It tracks a few key metrics (loan size, loan-to-value %, region, sector i.e. office vs shops etc, loan length, loan purpose and pricing) over time, and suggests a rationale for the changes.