Insecure About Secured Lending
By John Kapitan, Managing Director – Consulting, and Brannan Johnston, Director – Analytical Services, of ERisk

In this month’s ERisk Report we show how modeling recovery rate risk improves the management of secured lending businesses through the cycle.

Secured lending activities such as commercial real estate (CRE) lending, leasing, asset-based lending, and loans with guarantees all look like great businesses, as long as the “security” functions as intended.

The new Basel II regulations will look favorably upon these businesses – very low Loss Given Default statistics in the regulatory calculations will more than make up for typically lower credit quality (higher Probability of Default) in Basel II’s Internal Ratings Based capital formula.

But when certain risk factors in a secured lending sector all line up, these businesses can suddenly seem more like heavily leveraged bets on how bad things might get. They can generate losses that are far worse than expected, as happened with CRE lending in the early 1990s or, more recently, with aircraft leasing.

We can see a demonstration of this in Figure 1, which shows the relationship between auto financing charge-offs (line) and changes in used car prices (columns) from 1994 to 2002. The two peaks in the chargeoff series occur in 1997 and 2002, both during periods of unexpectedly small rises (or decreases) in used car prices.

Specialist secured lenders intuitively understand how best to monitor collateral and dispose of assets in the markets underlying their business. But only formal models of the behavior of collateral in stressed markets and the resulting volatility in recovery rates can help an institution price properly for risk, so that shareholder value is created over the course of a full credit cycle.

Institutions that realize that risk modeling offers competitive advantages might be tempted to use shortcuts like the new Basel II regulatory capital approach. Unfortunately, Basel II recovery rates are static and model only the risk of extreme default rates: this misses the interplay between default rates and recovery rates in a stressed scenario that is so crucial to properly estimating secured lending risk (and being able to foresee potential high-risk threats before they manifest themselves).

Let’s look at some key risk issues in three types of secured lending – leasing, asset based lending, and transactions with third-party guarantees – to show how a risk model that is up to the task can provide more insight.

Separation anxiety

Leasing professionals often say that the main source of risk in their business lies in the value of collateral or residual assets, whether these are airplanes, forklifts, or office equipment. But to understand the deeper cyclical risks in the leasing industry we must understand how this “market risk” interacts with the Probability of Default to produce a volatile mixture of both market risk and credit risk.

In any lease, as the value of the remaining lease payments amortizes over time, and the value of the underlying equipment depreciates, there is generally an unsecured exposure for the difference between the two. Most lessors understand that there is risk associated with the gap between the value of the remaining lease payments and the value of the equipment during the lease. But not enough analysis goes into understanding the volatility of the market value of the equipment both during and at the end of the lease, assuming the lessor takes ownership of the equipment at the end of the lease.

Figure 2 illustrates the difference between the present value of future cash flows (green line) and the value of the leased equipment (blue line). The gap between the green and the blue line represents the unsecured credit risk exposure. In the worst circumstance (illustrated by the lower dashed blue line in Figure 2 at a one in a hundred confidence level), we can see that there is considerable credit risk exposure for most of the deal and that the leasing company may be left with an asset with an impaired residual value at the end of the lease.

The size and volatility of this unsecured lending gap must be carefully measured as it should be factored into the pricing of the transaction, and might require changes in the structure of the deal – for example, the introduction of risk mitigants such as collateral guarantees or a transfer of the risk to the lessee. Across an entire portfolio, deals like this will drive the risk profile of the company.

Such risk analyses also have important product design and strategic implications. For example, auto-leasing companies that suffered from tremendous residual value losses in the mid- to late 1990s at the end of short-term leases (24 or 36 months) sought to reduce these losses by marketing longer-term leases (48 and 60 months) at attractive rates. However, many of these companies found that they had simply replaced market risk with credit risk, as the unsecured exposure gap increased significantly during the course of the longer-term leases.

With no numbers attached to the economic value of the additional credit risk, it was impossible to work out whether the strategy made sense or not.

This tradeoff between credit risk and market risk provides an interesting comparison point against a straight-up loan to finance the buyer’s purchase of the same asset (as described in the Box opposite).

Kicked while you’re down

Something else compounds the complexity of the risk analysis: when things go wrong in secured lending, they have a tendency to go wrong together.

The same factors that cause default rates to climb in an industry that is in a downturn can send the values of equipment into a tailspin, as we’ve seen recently in the aircraft leasing industry. Secured lenders then worsen any cyclical trough by creating a glut of assets on sale at just the wrong moment in the sector cycle.

So when the firm in Figure 2 considers the volatility of its recovery rates, it also has to consider the chance that worst-case recovery rates will coincide with worst-case default rate projections across its portfolio.

This is why lenders must turn away from their habit of basing business decisions (strategy, pricing, financial reporting) on expected losses and make a more formal estimation of the volatility around the expected loss rates.

To do this effectively, they must make sure that operating risks do not muddy the picture.

Illusion of control

The most vivid example of how operating and credit risk factors feed on each other is asset-based lending. In this kind of secured lending, the collateral takes the form of dynamic assets such as accounts receivables or inventory.

The credit offered to the customer varies in line with the receivables; as the receivables pay down, the amount of credit extended falls away. In theory, the loan is over-collateralized at all times so the lender focuses on assessing and monitoring collateral value. Unfortunately, whenever a borrower defaults it’s almost always the case that the receivables either don’t exist or are worth much less than the loan – otherwise the borrower wouldn’t be defaulting in the first place.

It’s tempting to classify this kind of loss as a one-off fraud, or as some kind of “operational risk” rather than as a credit risk. In fact, this simply disguises the true risk of this business, which is that there is an intrinsic credit risk associated with the borrower irrespective of the type of collateral.

In some cases, the recovery rate will indeed be 100%. But we must incorporate volatility into our estimate of the recovery rate to recognize that in a significant number of cases the collateral will not exist or will be worth much less than expected.

Fear of rejection

Something similar happens when we look at the volatility of loss rates associated with loan guarantees – an important risk factor for small business, trade finance, and even residential mortgage loans sold to agencies. In this type of lending, guarantees of a customer’s performance by some other more creditworthy company or the government act as a kind of collateral. But such loans are by no means risk free: guarantors often find reasons not to honor their promise.

This risk is often tracked as an operational risk because it typically arises from problems associated with loan underwriting. But let’s take a deeper look at the two risk drivers:

- The defect rate: the rate at which the lender books loans that could later be said to be out of guidelines, or makes errors in the documentation that could cause the guarantor to put the loan back to the lender.

- The discovery rate: the rate at which the guarantor discovers a defect, given that one exists.

The defect rate is likely to be a mostly stable rate that isn’t very correlated with economic drivers of the default rate. But the discovery rate of catching a defect is neither stable nor uncorrelated. Guarantors have no incentive to look for defects on loans that are performing; if they’re clever, they will focus their scrutiny on loans that have defaulted so they can hand the risk back to the lender. Therefore, the discovery rate is likely to be strongly correlated to the default rate of the borrowers, causing the risk on guarantees to be best measured as a credit risk.

This credit risk measurement can be accomplished through a careful modeling of the recovery rate, which is dependent on the defect rate and the percentage of the loan that is guaranteed. Thus, for loans that are 80% guaranteed (eg, an SBA loan), one should apply an unsecured Loss Given Default statistic to the 20% that is not guaranteed, and apply the defect rate times the unsecured loss rate to the guaranteed portion.

This modeling technique can also be applied to loans sold without recourse such as mortgages sold to agencies, but where errors in documentation or loans outside of underwriting standards can allow the purchaser to put the defaulted loans back to the originating institution. (It’s worrying to think that most mortgage banks probably don’t know their exposure to this risk since they don’t track loans after they’re sold.)

Shine a light on the monster under the bed

Proper modeling of recovery rates is the only way to fully understand and price for the risk in secured transactions including residential mortgages, commercial real estate, leasing, asset-based lending, and loans with guarantees.

It’s easy to fall into the trap of over-reliance on security if you don’t carefully think through what might go wrong with collateral when the borrower defaults and when many borrowers are defaulting.

Lenders frequently misunderstand the type and potential magnitude of these risks by neglecting the volatility and correlation of default rates with recovery rates, and mistakenly characterizing risks as operational risks when they are best modeled as credit risks.

The solution is to make the economic effect of risk interactions transparent by carefully adapting credit risk methodologies – particularly Economic Capital approaches – to the unique nature of secured lending.

The information can then be fed into a range of critical management decisions such as pricing, product design, and business leverage to make sure the firm can survive any cyclical downturn – and generate real risk-adjusted profits during the (supposedly) credit risk-free period of any lending cycle.

This article was contributed by John Kapitan, Managing Director – Consulting, and Brannan Johnston, Director – Analytical Services, of ERisk, who welcome your comments at jkapitan@erisk.com.

For more an extensive library of resources on risk and capital management, visit the ERisk website at www.erisk.com.

©2008 Sungard. All rights reserved. Legal Information