If ever there were a business the attractiveness of which was in dispute among bankers, it is indirect auto lending.
Dealer lending is an activity with weak growth fundamentals and somewhat volatile profitability levels that today are below the hurdle rate for most banks. Yet, as in most banking businesses, there are star performers - institutions that almost always earn more, and sometimes appreciably more, than the needed 14% to 15% after tax. Typically, the keys to success are a detailed understanding of all relevant costs (and therefore the required pricing) and a capacity to control risk.
The primary costs in indirect auto, as in any lending business, are the cost of money, the cost of expected loan losses, and functional expenses. To build an appropriate cost of funds into loan pricing, banks must first calculate the expected cash flows of the loan asset. These cash flows must be funded with a series of strips or money tranches, the present-value- weighted average of which sums to the loan's duration. In the case of indirect auto, the duration, inclusive of expected prepayment, approximates 18 months. This corresponds to a two-year average life.
The match-funded cost of money is actually composed of three elements:
1. A given short rate - usually Libor.
2. The cost of swapping this short rate into the appropriate long rate (swapping across the maturity spectrum creates what is called the Libor swap curve); and
3. A liquidity premium, which reflects the fact that the cost of borrowing cash at the relevant maturity is actually higher than the rate shown in the swap yield curve (about 10 basis points higher for a borrower rated single-A at the two-year point).
Unfortunately, many banks ignore the term-liquidity component in their cost of funds. This often causes them to understate true costs and therefore to agree to pricing that is lower than that required by shareholders. The failure to consider the term liquidity premium can overstate reported return on equity by one to two percentage points.
Another cost area where a sharp-penciled perspective is necessary but often overlooked is loan loss provisioning. Many banks set the provision, which is the charge for expected loan losses that must be reflected directly in the loan price, by dividing current chargeoffs by current balances. But in any consumer lending business, losses start out low and then rise to a peak in about a year and a half. Hence, in periods where the portfolio is growing, the above provisioning procedure will often understate the level of the needed coverage by some 20 to 30 basis points, leading to a 2% to 4% overstatement of ROE.
To prevent this occurrence, First Manhattan Consulting Group suggests that banks examine the projected loss pattern in a discrete loan cohort and then set an annual expected loss charge whose present value over the life of the loan is just equal to the present value of the anticipated losses. This type of change can be called the "steady-state" or economic provision. It will be higher than actual chargeoffs in the first year of the life of the loan, when losses are low, and lower than chargeoffs in the second and third years of this life, when losses tend to rise. But over the full loan maturity, this type of provision will fairly compensate the bank for expected or average losses.
The third cost item is functional expenses. Fully loaded, these average about 150 basis points per dollar of managed assets in indirect auto. But there is a substantial variation among banks, depending on scale of operations and capacity to monitor and control dealer risk.
Indirect auto lending traces an 85% to 90% scale curve. This means that every doubling of volume will result in a decline in unit costs of approximately 10% to 15%.
Even more important than scale of operations is dealer quality. Dealers that attempt to serve the more marginal customers create excess origination and eventually loan collection expenses for their banks. Hence risk control blends into cost control. Banks need to set and make known to dealers their policies as to acceptable grades of risk and the pricing appropriate to each. Failure to do so will unfavorably impact both functional costs and loss provisions.
Expenses also can be reduced by raising the ratio of loans closed to applications received. There is a wide variation among dealers in this ratio, depending on the amount of their rate-approval and credit-approval shopping. Banks need to develop ties with dealers that have histories of low application fallout.
The more astute banks measure and array loan profitability by individual dealer. They typically discover that nearly half the dealer population consistently provides paper that is low risk and therefore highly profitable, while an approximately equal proportion consistently provides higher-risk paper that yields little or no profit. Good banks attempt to improve their supply channels by isolating the characteristics peculiar to the former group of dealers. They then reward their best suppliers with better pricing and/or a more favorable "reserve" policy (in indirect auto lending "reserves" represent the difference between the rate charged to the customer and the bank's buy rate, the lion's share of which is routinely upfronted to the dealer).
Those banks that understand how to build superior supply channels become caught up in a "virtuous" as opposed to a "vicious" circle. That's because the benefit of having good supply channels is not only the capacity to operate with less functional cost per loan but often also with less economic, though not regulatory, capital per dollar of loan asset.
While the loss provision exists to cushion the impact of expected losses, the capital account serves as a buffer against unexpected losses - i.e., the variance of losses around their mean value. That variance tends to widen as average losses increase and narrow as they decrease, much as the variance of individual prices in the economy increases (decreases) as the rate of inflation rises (falls). Otherwise put, the mean and the standard deviation of loan losses are roughly proportional. (They are not linearly proportional, however. The ratio of unexpected to expected losses drops somewhat as the average rises. Nevertheless, a high average level of losses is typically associated with a high loss variance.) With good dealer selection, therefore, average losses and loss volatility (and therefore needed capital) drop.
The economic capital required to backstop any bank business is a function of loss volatility and institutional risk tolerance, as proxied by the target credit rating. Most regional banks aspire to about a strong single-A or weak double-A credit rating, which means that they must earmark enough capital to produce a probability of institutional default no greater than one in 300 or no greater than one in 2,000. Given existing loss volatilities in indirect auto lending, that amount of capital varies by institution, but typically is in the range of between 5% and 8%. Unfortunately, with this capital amount, the typical bank earned only an 8% ROE in 1994. (This is in sharp contrast to the 16% to 18% return posted in the higher spread environment on 1993.)
Today's depressing arithmetic reinforces the need to follow the procedures suggested in this article - those that enable some banks, by dint of correct pricing, cost reduction, and capital economies, to record at least double the average ROE in low- as well as high-spread periods. in summary, these procedures are:
1. Understand the true cost of money and the required loss provision, as these help govern minimally acceptable product pricing.
2. Keep a firm grasp on the two interrelated variables of risk and expense by isolating the characteristics that define superior dealers, educating dealers as to the risk and pricing tradeoffs acceptable to the bank, and cementing dealers' loyalty through preferential pricing or service.
Mr. Zizka is a managing vice president of First Manhattan Consulting Group, New York. Mr. Rose, formerly senior columnist for this newspaper, is also associated with the firm.