Stress-testing has had a profound effect on risk management practices within banks. In the wake of the global financial crisis, regulators around the world instituted policies designed to substantially remove the possibility of large-scale bank failures.
Capital reserves at banks have been appreciably raised, and regulators are applying a considerably greater amount of scrutiny to the way banks monitor and predict risk.
Although stress-testing is a young discipline and will surely change radically over the next decade, the procedures being put in place hold the potential to transform the way financial institutions around the world are managed.
In the form implemented by U.S. regulators, stress-testing involves trying to infer the likely performance of a portfolio assuming that a specific, dire macroeconomic scenario starts to unfold.
Most large banks have active auto loan portfolios. Therefore, a great deal of effort has been applied to the problem of stress-testing default probabilities, credit losses, loan recovery rates, lending volumes, and revenues associated with the management of auto portfolios.
Leases, although rarer in the banking industry, have also demanded increased analytical scrutiny. The shift into leasing has added residual forecasting and return rate probability to the mix of factors to be stress-tested by the auto industry.
The purpose of this article is to consider some of the lessons learned from research conducted to stress-test bank auto portfolios. This is not meant to be a general discussion of stress-testing, per se, but a summary of some elements that are quite unique to the world of auto finance. Auto loans and leases, like mortgages, are secured by collateral.
Unlike property, however, autos are mass produced, depreciating products with a relatively short life span. These features change the way auto loans and leases behave in the context of a downside macroeconomic event.
This article will set out by considering, at a high level, how auto loans tend to perform in recessions. Data for the 2001 and 2008-2009 events will be compared and contrasted with a view toward trying to predict how auto portfolios might be expected to behave in a future recession.
We will consider as part of this discussion the effect of fiscal stimulus packages such as cash for clunkers and whether the auto industry could expect a repeat performance in future recessions. We will conclude by asking whether a subprime-mortgage style crisis could conceivably affect the future auto industry.
Pro and counter cyclicality
The 2001 and 2008-2009 recessions were different beasts. The 2001 event was tamer, having its origins in the dot-com bust and 9/11; the Great Recession was triggered by a banking sector crisis, the likes of which had not been seen since the 1930s. The auto sector, it is safe to say, was not a proximate cause of either event.
The performance of the new-car industry was markedly different in each event. Coming into the 2001 recession, new-car sales were climbing steadily, reaching a peak of close to 19 million annualized units around the turn of the millennium.
As the recession started to bite, generous incentive packages offered to customers, coupled with a strong “nesting” instinct in the wake of 9/11, acted to keep unit sales of new vehicles relatively high. If new-car sales were the only economic indicator you considered, you would be unaware that the 2001 recession even happened.
The Great Recession, by way of contrast, saw new-car sales crater. In early 2009, during the immediate aftermath of the Lehman Brothers collapse, monthly sales fell to around 9 million annualized, barely half the precrisis level. Some attribute the sales decline to problems at GM and Chrysler, and these events certainly did not mitigate against the dire impact of the recession.
It can be noted, though, that unaffected manufacturers also suffered steep reductions in sales. It is fair to say that Toyota and Ford could have sold more to unsatiated GM customers had the demand for their vehicles actually been present in the economy. The new-car industry has been fighting back over the past six years but it took until 2014 before pre-Great Recession sales numbers were observed once more.
Although consumers continued to refresh their vehicles during the 2001 recession, foregoing things such as vacations and international travel for a nicer ride, Great Recession consumers clearly opted to delay the purchase of new or newer vehicles. This had a decidedly negative, deep impact on auto manufacturers and new-car dealers.
For financiers, meanwhile, the recession was a relatively mild, short-lived event. By early 2010, auto default rates were already at prerecession levels.
Recoveries on those loans that were in default were low because used-vehicle prices were rising at breakneck speed. At least one bank in the knowledge of the present author saw recoveries exceed gross credit losses during the latter days of the Great Recession.
It behooves us to ask why the recession was so mild and short for some auto financiers.
The fact that the 2008-2009 recession started in mortgage was, in a roundabout way, a positive for the auto finance industry. With house prices falling rapidly in most parts of the country, consumers’ normal resilience in paying their mortgage through tough times crumbled rapidly.
If we consider delinquency rates on mortgages, car loans and credit cards it becomes immediately clear that, prior to 2009, mortgages consistently had the lowest rate of nonpayment, followed by auto and credit cards in third place.
The subprime crisis led to a huge rise in mortgage delinquency and default, to the point where, by far, mortgage was the most defaulted form of consumer loan (in terms of percent of dollar volume outstanding) by late 2009. With house prices falling, troubled consumers overwhelmingly decided to favor the car loan over the house. You would get terrible gas mileage if you ever tried to drive the house to work.
It turns out that if home foreclosure ever becomes inevitable, it gives consumers an unexpected source of disposable income that can be applied to other loans.
It is a fairly callous form of accounting, but a missed $1,600 mortgage payment represents four $400 car payments or 16 $100 credit card installments for a troubled mortgagee.
The second relevant point is that the decline in new-car sales in late 2008 and early 2009 inevitably led to a dearth of used-car supply in early 2010. This phenomenon is well known in the auto industry, though our view is that the time lapse between new-car sales and the used-car price surge is a fair bit shorter than the view commonly held in the industry. If the dominant form of transaction, for example, is a three-year lease, it might be reasonable to expect the gap between these events to be about 36 months.
The empirical data, however, show that new-car sales lead used-car prices by around 12 to 18 months; this pattern has been consistent over the past few decades. These supply dynamics are overwhelmingly responsible for the surging used-car prices seen in 2010 and the thin losses observed by many in the auto finance sector during the same period.
In a hypothetical future recession, new-car sales is the key variable to track. If sales remain aloft during the recession, financiers and remarketers should brace for a period of low used-car prices about a year after the onset of recession.
If new-car sales crater, however, the industry can take succor in the fact that used-vehicle supply will contract, and prices will rise, thus cushioning the back end of the recessionary period.
A short side note relates to the issue of fiscal stimulus during a recession. Whether or not you agree with the concept of the government taking on the role of “spender of last resort” during a deep recession, it is still valid to consider the appropriate form of stimulus should such an intervention ever be deemed necessary.
During the last recession, the auto industry was supported by bailouts of GM and Chrysler and by the $3 billion cash four clunkers scheme that triggered some to update their ancient vehicles during the deepest part of the recession.
The empirical evidence on the success of cash for clunkers is thin. A cynical analyst could line up the 2009 scheme and the 2010 price surge and say that X caused Y.
A less cynical, data-driven person certainly could not reject this assertion; future empiricists may be able to weigh in on the issue of exactly how much of the 2010 price surge was due to cash for clunkers market activity. Equally certainly, we can say that the scheme did not hurt used-car price dynamics during this time.
The Chrysler and GM bailouts were certainly positive for auto industry employment during the recession, though the longer-term effects of the bailouts will likely be highly negative.
Our view is that if fiscal stimulus is to be enacted, the auto industry is an excellent candidate for support. Car use is close to universal in the U.S., and the auto sector employs many millions of Americans across an enormous range of household income levels.
It is difficult to find other industries with better stimulus credentials in terms of progressivity or heavy employment concentration.
The next time recession looms therefore, assuming pro-stimulus legislators are in place, it is completely reasonable to believe that stimulus funds will flow in the direction of the auto sector.
Can it happen to us?
The discussion until now has focused on the effects of externally triggered recessions on the auto industry. History shows that recessions are caused by structural imbalances or bubbles in key sectors of the economy and that economic pain is highest in the sector where the imbalances first appeared. It would therefore be prudent of us to consider the likelihood that the auto industry will be the root cause of the next recession.
As mentioned already, the auto industry is certainly big enough to cause the U.S. economy to topple. Some banks are exposed to auto portfolios that are consequential enough to trigger bank failure. The auto sector, like the mortgage industry, uses securitization in high volume; it was these forces that many blame for the subprime crisis that triggered the Great Recession.
Despite the presence of these preconditions, our view is that an auto-induced recession is highly improbable. One of the commonalities observed in previous financial recessions was the existence of speculative asset price bubbles.
During the 2005-2006 housing boom, for instance, many individuals and companies were buying real estate with zero yield on the expectation of reaping capital gains upon sale of the asset.
House prices, after all, never, ever fall! At a basic level, this form of psycho-pathology is not plausible in the auto sector.
Although it is true that some highly specialized companies exist to arbitrage minor market inconsistencies, most sane people do not expect to make a profit when buying and selling used vehicles unless they happen to be auto dealers. Cars are mass-produced, depreciating assets that do not lend themselves to speculation regarding potential capital gains.
Some may argue that speculation is possible by people operating within the world of asset-backed securities. This is true, but only up to a point.
Someone analyzing a securitized deal could quite easily and accurately determine a theoretical maximum value of the collateral that cannot be breached under any sensible circumstances. If an investor were to pay such a price, or anything remotely similar, they would cease to be investors after a short time.
Given the size of the industry, one does not want to completely rule out the possibility of an auto-caused recession. If such a black swan exists, it is far more likely that the production and manufacturing side of the industry is its natural habitat. The auto-induced recession swan does not swim on the financial side of the lake.
The auto sector has many characteristics that make a subprime-style collapse unlikely.
The shorter half-life of cars relative to houses, the fact that vehicle production can be initiated or called off relatively quickly, that auto markets are inherently non-speculative, and the fact that used-vehicle price dynamics tend to be countercyclical all support the notion that auto recessions are unlikely to linger.
This is not to say that the early days will not be dark—they will be. The evidence, though, suggests that recovery will tend to be robust, ongoing and earlier than that of the rest of the economy.
The broader point is that the onset of bank stress-testing has led to a deeper and clearer understanding of the macro forces acting on the auto industry. Expect future stress tests to yield many interesting insights for auto industry insiders.
Tony Hughes is managing director of credit analytics at Moody’s Analytics.