Understanding 'true LTV'

Risk management is now top of the boardroom agenda for mortgage lenders across the globe. Mark Witherspoon of UKValuation explains why understanding the concept of 'true loan-to-value' is a key element of any lender's approach to risk management

It would be something of an understatement to say that the UK mortgage market has been going through a tough time in recent months. Mortgage professionals throughout the industry are beginning to face up to the fact that we are in the midst of full-scale slowdown. As the prospect of homeownership begins to recede for first-time buyers, and existing borrowers attempt to cling on to the property ladder - let alone climb it - it is worth remembering how we found ourselves in this precarious position to begin with and what we can do to avoid making the same mistakes again.

As we are all painfully aware, the problems in the mortgage market arose because of a failure of confidence in the creditworthiness of banks, caused by the subprime saga in the US. The costs of interbank lending rose sharply, which had an impact on liquidity, the lifeblood of the mortgage industry in recent years, and the fallout has been there for all to see. This in turn has had knock-on effects for the wider economy and the industry as a whole with mortgage lenders having to cut costs, tighten their belts and unfortunately pass on these woes to the borrower in the shape of rate rises and more stringent lending criteria.

There are those in the industry who may suggest that tightening up lending criteria now conjures up images of stable doors closing while horses disappear over the horizon in a cloud of dust and stampeding hooves, but whether you subscribe to this view or not it is clear that the industry needs to take a more proactive approach to managing risk - and gaining a more accurate sense of how much exposure they actually have.

If you add to this picture the tenets of the Basel II accord, whose stated aims are to ensure that "capital allocation is more risk sensitive", to separate "operational risk from credit risk and quantify both" and to "align economic and regulatory capital more closely to reduce the scope for regulatory arbitrage", and it is clear that mitigating risk is a most pressing issue.

However, prevalent as these concerns are, the current marketplace is demanding that cloth be cut accordingly, heads be lowered below the parapet and costs reduced - meaning that the use of all risk-related tools and processes is under review. Perhaps you would expect this to give rise to dissenting voices against the use of automated valuation models (AVMs) from people who feel that they are merely a 'fashionable' luxury that should be scaled back until liquidity returns to the market. In fact, in our experience the opposite is true. Many lenders feel the accuracy of information when it comes to the AVM's ability to mitigate risk to be invaluable.

A multi-purpose tool

AVMs are not just an origination tool but a good way to reassure stakeholders that lenders are doing all they can to understand their mortgage books and their capital bases and are taking appropriate steps to manage risk. When you add to this AVMs' ability to check valuations retrospectively - providing the opportunity to check the quality of a lender's book - and the fact they can help with the audit panel management process, it is clear that in the current climate AVMs serve several important purposes and are proving a vital weapon in the lender's armoury.

So the question must be: if the mortgage industry is going through a process of retrenchment and introspection and AVMs are proving successful not just at point of sale but also in the auditing process, what can the rest of the market learn from the way that many lenders have embraced and employed them?

As AVM technology improves and continues to build on what was once simply mortgage origination, more cost-effective products can be offered to other professionals within the mortgage industry who are keen to audit their own processes, including surveyors. There is an increasing demand for a solution to provide an auditing service around the point of origination. Risk management presents a number of tangible benefits for surveyors, allowing them to provide quality assurance by implementing more stringent processes, illustrating a commitment to accuracy in valuations and keeping on top of quality control. On a more practical level, it can also lead to a reduction in premiums for professional indemnity insurance.

Comparison

The process is simple to understand. The model runs an AVM alongside the surveyor's valuation. Where large discrepancies are uncovered, potential problems are flagged to the interested parties allowing them to check whether the cause is sloppy processes, a flawed value made under duress, genuine mistakes or - the worst-case scenario - fraud. The need for surveyors to be able to provide extra guarantees around the accuracy and validity of their valuation is vital at a time when the market is jittery and lenders are looking for assurances before dusting off their wallets.

In the UK, we have witnessed tremendous growth in the use of AVM technology and yet, we also have some curious anomalies. We have a situation whereby most mortgage lenders have a policy for AVM use that is driven by cost recovery on remortgage products. How strange that usage has yet to make significant inroads into house purchase transactions? I do not believe that this is a risk play but rather one of economics, in that for many remortgage transactions the cost of making the loan is met by the lenders and the customer is far more time-sensitive. Therefore, every penny saved makes a lending margin contribution and every hour saved is valuable. Whilst on house purchase transactions, the customer pays, so why save them money? The whole process is going to take ages anyway, so why the rush? This also creates a dilemma in terms of how much to charge for a valuation, particularly in these 'treating customers fairly' days.

Over the last couple of years AVM developers have witnessed widespread market recognition of the fact that AVM technology has much to offer outside of the initial loan application process. By this we mean the application of the same core methodologies to whole loan books or portfolios. In this context, AVMs are without equal. An AVM is capable of crunching through an entire book in a matter of hours and generating an accurate analysis at various levels - the value of the whole book, the creation and valuation of tranches within the book, and the identification of anomalies within a book. This facility also enables AVM suppliers to enhance the data output side to include supporting risk data, and as such AVM suppliers have carved out a new role in support of Basel II, whole loan debt sale, and purchase and balance sheet management techniques. There is also a lot more to come out of the ability to run vast data sets through complex modelling processes - and so, as they say, watch this space.

I have also been asked how an AVM would perform through a downturn in the housing market. Mortgage fraud is raising its head as a serious issue in the UK market: a number of brokers have been fined by the Financial Services Authority. Before we come on to the thorny issue of 'true LTV', it makes sense for us to doublecheck that AVMs are as robust when prices are falling as when they are rising.

In the UK, towards the latter part of 2007, the leading house price indices were indicating property price reductions across many parts of the country. This trend has continued through 2008, and it has raised a pertinent question for AVM users: can they continue to trust AVM valuations in a declining market? I believe that they can - and these are the reasons why.

Objective correlative

One of the key benefits of an AVM is objectivity. This objectivity applies not just to the way in which the valuation is statistically derived, but also to the way in which the AVM tracks house price trends. The UKValuation AVM tracks the market using several hundred different indices which correspond to specific property types in specific localities. These indices are the starting point, which allow the AVM to objectively track the market trends in a consistent manner, regardless of whether the market is trending up or down.

Data for the AVM is fused from many trusted UK market information sources, including the Ordnance Survey, CACI Acorn data, Royal Mail, Land Registry, surveyors and mortgage lenders. It is data from these last two sources that is most important for ensuring timely updates to the house price indices, and this data is received on a monthly basis. This therefore introduces an approximately one-month lag into the process. However, this issue is not specific to AVMs: a time-lag is built into all of the leading indices including those from Halifax, Nationwide, DCLG and Land Registry. This means that there are several further points to consider.

Firstly, even a surveyor's valuation will rely on comparables. When surveyors are required to index these comparables over a period of time the question arises: using which index? And are these indices consistently applied? Secondly, accepting that no index can ever be completely up-to-date, a surveyor will make subjective adjustments based on their understanding of market conditions. The AVM will not attempt to do this and will instead assume a flat price trend from the last known data points. Both approaches have their strengths and limitations, but it goes without saying that the application of subjective adjustments or forecasts by surveyors can create challenges from a risk perspective regarding their basis and consistency.

True LTV

We've established that AVMs work in a declining market - and therefore are not inherently 'risky' as a valuation method when house prices are falling. What we need to do next is to use the AVM methodology to start to understand 'true loan-to-value'.

Before we do, let's just remind ourselves of the basics. LTV is a mathematical calculation which expresses the amount of a mortgage as a percentage of the total value of real property. For instance, if a borrower wants £130,000 to purchase a house worth £150,000, the LTV ratio is £130,000/£150,000 or 87%.

LTV is one of the key risk factors that lenders assess when qualifying borrowers for a mortgage. The risk of default is always at the forefront of lending decisions, and the likelihood of a lender absorbing a loss in the foreclosure process increases as the amount of equity decreases. Therefore, as the LTV ratio of a loan increases, the qualification guidelines for certain mortgages become much more strict. Lenders can require borrowers of high LTV loans to buy mortgage insurance to protect the lender from the buyer default, which increases the costs of the mortgage.

Low LTV ratios (say, below 80%) may carry with them lower rates for lower-risk borrowers and allow lenders to differentiate them from higher-risk borrowers, such as those with low credit scores, previous late payments in their mortgage history, high debt-to-income ratios, high loan amounts or 'cash-out' requirements, insufficient reserves and/or no income documentation.

Higher LTV ratios are primarily reserved for borrowers with higher credit scores and a satisfactory mortgage history. The full financing, or 100% LTV, should be reserved for only the most creditworthy borrowers (but as we have seen, this golden rule of mortgage lending has often been forgotten in recent years by lenders and borrowers desperate to join the booming UK housing market!).

Many lenders manage their risk exposure based on relatively simple LTV ratios. But the situation may be more complex than this. Let's take a simple example. Borrower A has a property worth £200k. He has a first charge mortgage of £150k, giving him an LTV of 75%. This is considered safe territory for any lender, even in the tightened market of today. The lender assumes the level of risk and arrears as per a 75% LTV, which is what the Basel II capital adequacy requirements demand. So far, so good.

However, much if not all of this good analysis and scorecard building is applied only at the time of loan origination. Building on our scenario above, Borrower A subsequently takes out second and third charges with two other lenders of £25k each, totalling £50k. He also builds up unsecured credit card debts of £20k. However, his main mortgage lender continues to treat him like a 'safe bet' 75% mortgage borrower - but he's not. His 'true LTV' is 110%. He is an imposter and the original lender is oblivious to his true identity.

So, I would contend that this loan will behave more like the 110% loan, not a 75% LTV loan, because now only the borrower knows his true level of debt. In a slowing housing market, the critical question would be: is he more likely to default on the mortgage payments than the real Mr. 75%?

To truthfully answer this question, lenders should consider a systematic process under which they record and monitor second and subsequent charges and regularly review against a current estimate of market value for the property.

So, the challenge for UK lenders now is to start to understand how close their assessment of the LTV of any property - or portfolio of properties - is to the truth. This variance gives us the gap between perceived LTV and true LTV, which in turn tells us how much risk is really attached to a loan or portfolio of loans. Look at those loans originated in the last three to five years and effectively re-underwrite the risk. But as described this 'true LTV' must embrace both unsecured and secured debt to give the full picture. Thus, true LTV will give the market a clear and transparent understanding of exactly how loans are likely to behave; as such, it's a vital risk management tool for lenders to consider.

Mark Witherspoon is CEO of UKValuation, which provides AVM services and residential property valuations for UK financial services providers.

www.ukvaluation.co.uk

UKValuation is a member of the First American CoreLogic group of companies, which itself is a subsidiary of the First American Corporation (NYSE:FAF).

AVMs: a summary

- AVM house price indices treat appreciating and depreciating markets identically.

- Due to the nature of the data sources there is a data time-lag of approximately one month.

- In a falling market this lag can lead to a consistent overvaluation but the effect is both small and quantifiable.

- The maximum size of potential overvaluation can be estimated at just over 1.5%, which is not significant compared with the natural volatility associated with any form of valuation.

- Overall, the analysis shows that a falling market will not significantly affect AVM accuracy and therefore we recommend that AVM criteria do not need to be restricted for this reason.

The time lag effect: A quantifiable impact?

The time lag effect arises from the fact that the AVM implicitly makes the assumption that the index remains flat until there is hard data available to determine the direction of the market.

As can be seen from the scale of the charts, the effects of the lag are very small. Critically, they are also quantifiable. The size of the lag can be estimated several ways:

- If house prices increase or decrease by 20% pa, that equates to a compound monthly change of 1.5% - hence a monthly lag should be no more than this.

- Between Feb 2004 and Oct 2006 (a period of prolonged market growth in the UK), the largest observed undervaluation of UKValuation test portfolios was just 0.5%.

- Parts of the US have now experience a sustained period of market decline and so they are uniquely placed to analyse the impact on AVM accuracy. A recent study by First American CoreLogic, UKValuation's parent company, demonstrated that in regions with fast-falling house prices, a maximum overvaluation of 1.7% was recorded in the second and third quarters of 2007.

It can therefore be surmised that, unless truly exceptional market conditions occur, the time-lag effect will account for between 1.5% and 2% maximum error either way, in the worst case. This market-dependent, consistent bias is not significant when viewed against the natural variation in individual property valuations (which is observed regardless of whether the valuation is derived from an AVM or a surveyor).

AVMs are able to respond to both rising and declining markets and, although the time-lag effect can create small, consistent over- or undervaluations (depending on the market conditions), these variances are very small when compared with the natural variations associated with individual property valuations.

Furthermore, the average over- or undervaluation of the AVM resulting from market trends is highly quantifiable, and this is in contrast with the potential variability associated with surveyor valuations in volatile markets - especially in such times when there is a degree of uncertainty about the future direction of the market. The market can never know with certainty where prices will move next, and different surveyors from different companies will all have their own interpretations of the prevailing market conditions.

The AVM sidesteps the problem by accepting the time-lag and only making price adjustments once sufficient data has become available.

In conclusion, we can state that a declining market does not significantly affect AVM accuracy. This leads us to recommend that there is no need for mortgage lenders to restrict AVM-specific lending policy rules or criteria in the current market environment.