Last week’s news about Wells Fargo’s retail sales tactics provides excellent evidence that all of the required ethics and UDAAP (Unfair, Deceptive, or Abusive Acts or Practices) training to which bankers are subjected is no match for poorly-designed incentive schemes and unobservant (or complicit) supervisors.
We have two main points to make regarding this well-publicized fiasco, including guidelines on how one should structure incentive schemes. However, we start by noting that Wells isn’t the only bank attempting to cross-sell to existing customers. So, it’s likely that this past weekend wasn’t the greatest for a lot of banking executives, and it’s possible that we’ll see similar CFPB actions against other banks. (Update: Per October 25, 2016 edition of The Wall Street Journal, it seems that the OCC is indeed investigating other banks.)
A Brief History
In the eighties, the focus was on having relationships with corporate customers (with the main tactic being the offering of low-rate loans as loss-leaders). It seemed that every large bank was following the same strategy: “let’s give the customer (or be part of group that gives the customer) a cheap loan. Once the customer sees how wonderful we are, we’ll establish a long-term relationship, and be able to charge more (for our charm?) and earn outsized profits.” Corporate borrowers took full advantage of these strategies and the banks’ eagerness, but had no reason to make any commitment. As it turns out, one bank’s money is no better than another’s. Ergo, few long-term relationships were formed, and, therefore, for the banks, no value was generated. It reminded us of the old dating advice, relayed as a question, that mothers frequently gave to daughters when the girls were old enough to marry: “why buy the cow, when the milk is free?”
Later in the 20th century, the industry’s realization that federally-insured, below-market-rate deposits have value seemed to induce more interest in creating and managing retail banking relationships.1 However, we think an aspect of The Financial Crisis played a larger role in the recent emphasis on cross-selling: during and after the crisis, many banks observed that customers with multiple products defaulted less frequently than marginal borrowers.
Next tab: A Few Questions
A Few Questions
A Few Questions and Correlations with Cross Selling
Please note that we are not arguing that cross selling is a poor tactic; however, there are costs associated with it, and in many cases, the premise behind arguments for its superiority seem to be lacking. Let’s restate the logic behind the impetus on cross-selling: “our ‘best’ customers defaulted less; so, let’s loan more to them (and try to provide more services, too). How can we lose?” (The question is purely rhetorical.)
First, note that at most banks, these customers, who survived the crisis at higher percentages than other, less-involved customers, weren’t necessarily identified as the “best” customers prior to the crisis. It’s an ex post definition. Also, focusing on observed default or loss rates, only, especially during a period of stress, ignores the benefits associated with possibly charging higher rates to new marginal customers, including less creditworthy ones. There is a risk-reward relationship to consider.
The crisis was extraordinary. The question is: how can you make the most in ordinary times while still being protected in an extraordinary bad time? Scenario analysis and stress testing help answer that question.
Second, providing more loans and services to a smaller base (“cross selling,” by definition, means to existing customers) increases the correlations between and among (1) losses across portfolios and (2) reductions in fee and service revenue (PPNR) during times of stress. (The reader would be amazed by how many people choose to ignore or don’t understand this simple mathematical fact.) In other words, your “best” customers last time might not be your best customers this time, especially if they owe more than they did last time and can’t pay.
Previous tab: A Brief History
Next tab: An Aside
Aside
Aside: Sometimes, Poor Execution is a Godsend
As we mentioned above, Wells isn’t the only financial institution with a heavy emphasis on retail cross-selling. (In fact, some places have that strategy because of Wells’ perceived early success at it. “Look at all those new accounts Wells is getting!”) So, we suspect that executives at other banks are wondering if similar, pathological (and illegal) behavior exists within their own firms.
It is possible that those firms haven’t been able to execute as swiftly as Wells, and that reminds us of something that is often overlooked regarding The Financial Crisis. Some of the banks that survived the crisis were actually attempting to do the same stupid shit that wound up severely damaging or destroying others, but for a variety of reasons, were unable to execute on their plans.2 Thus, because of their inability to execute, and for that reason, alone, those firms were spared losses and embarrassment. This illustrates a different type of math where two negatives don’t sum to something negative or multiply to something positive but together map to zero.
Of course, no organization has admitted that its resiliency was based on its inability to execute a dubious plan. In that light, we encourage readers to search “narrative fallacy,” but only after they finish reading about incentives.
Previous tab: A Few Questions
Next tab: Improper Incentives
Bad Incentives
Consider any “work center” that processes or converts generic input(s) into output(s). For an incentive problem to exist the processor must have hidden information (adverse selection) or take hidden effort in a random setting (moral hazard) or both. The difference between improper and proper incentives is the not-so-subtle difference between (1) simplistically measuring inputs or outputs (like new accounts) or combinations thereof and (2) knowing what you want and then understanding (a) how the desired tasks, processing, or judgment should occur to convert inputs to outputs in the desired way and (b) what signals are available to indicate that the desired action(s) have been taken.
For example, if you simplistically think you want “more accounts,” and then you (explicitly or implicitly) strongly induce employees to generate more accounts, then you’ll get more accounts. However, they may not be the types of accounts that you really, truly want but didn’t (or couldn’t) articulate, especially if those new accounts are fictitious or illegal. That’s because you have ignored the environment, constraints, and all of the ways that employees can create more accounts. Yes, context matters.
Previous tab: An Aside
Next tab: Solving Incentive Problems
Aligning Incentives
How to Think about Incentive Problems
Much of what we discuss below refers to moral hazard or hidden effort problems like the one that Wells faced. They arise in decentralized environments where–for whatever reason–it was determined to be too expensive to observe or monitor each worker’s actions or decisions and there is at least a bit of chance or randomness.
So, what is the “desired way” that one should want workers to take? It’s the method of processing or working that maximizes the organization’s objective, e.g., creating long-term value, when the following items and their interactions are considered simultaneously.
- Available alternative courses of action and their effects on outcomes, a.k.a. production functions in microeconomics;
- The likely marginal benefits and costs of each of the those possible tactics or actions;
- The environment, including constraints like laws, and randomness that are outside of your control; and
- Incentive costs to induce the desire behavior given employee preferences.
Note that incentive costs are more than just compensation costs. They also include the costs of sub-optimal or dysfunctional behavior that can’t be eliminated without destroying the scheme’s overall efficacy. In other words, if you try to eliminate all dysfunctional behavior associated with a particular scheme and you’ll eliminate all the benefits, too.
Now, to be sure, incentives problems are challenging to solve because the solution involves considering how other people–employees, not inanimate objectives–optimally respond to the forces applied to them while simultaneously considering the (random) environment and the other factors mentioned above. However, as Wells discovered, it is extremely rare that the solution found by ignoring those items and their interrelationships will turn out well.
Previous tab: Improper Incentives
Next tab: An Alogrithm to Design an Incentive Scheme
Algorithm
How to Design an Incentive Scheme
So, how should you design an optimal incentive? Ask and answer the following questions. If you don’t, then you better be lucky. If your consultants don’t, then they shouldn’t be paid.
- What do you want the person to do and/or not to do (to assist in achieving the organization’s objective)?
- What signals are available about the person’s desired (and undesirable) effort(s) and/or decisions to maximize the objective?
- What are the characteristics of those available signal(s)? In particular, how sensitive, precise, objective, and manipulable are the signals to the person’s actions and decisions?
- Given the answers to the first three questions, which combination of signals–along with other, observable environmental variables–should be used as performance measures, and how should they be weighed to induce the person to maximize the organization’s net benefits (however defined) within his or her work center?
That’s all there is to it! Actually, these are difficult questions to answer, especially the last two, and that’s why thought and consideration and understanding the business process so crucial!
Note that in most cases,
-
- Expected gross benefits–excluding expected compensation costs–won’t be as great as if the employee’s effort were observable because it’s usually too expensive to motivate that behavior through incentive pay.
- Despite that, expected compensation costs will be greater than the less-variable (more fixed) compensation that could have been paid if effort were observable.
- However, monitoring costs should be substantially lower without the need for close supervision, i.e., to a certain extent, the pay scheme is substituting for supervision.
If you have questions or seek more information, please use our contact form or call us at 1-844-773-7626, and we would be happy to speak with you.
Previous tab: Solving Incentive Problems