Most bankers are familiar with the concept of a risk-return trade-off. This is the principle that potential return rises with an increase in risk. Low levels of risk pays low potential returns, whereas high levels of risk pay high potential returns. Further, bankers are taught early in their career that they are in the business of taking risk and without taking risk banks would not earn profits. We believe that every business takes risk from dry cleaners to bottlers, to retailers - and banking is no different. However, empirical evidence and common sense have taught us that credit risk in banking is a different animal and that the willingness in banking to trade yield for credit risk credit must be reconsidered.
Every year we analyze the relationship between net interest margin (NIM) and ROA for all banks in the country. For all banks between $100mm and $10B, the graph appears below. The correlation between the NIM and ROA for this group of banks for 2016 was 0.14 (correlation coefficient). While the relationship is statistically significant (but just barely), its explanatory power (or degree of influence) is very minimal. This helps underscore the fact that targeting NIM may not necessarily translate into a higher ROA.
Why NIM Is Not An Ideal Metric For Performance
Let’s consider why this is the case. The low correlation between NIM and ROA actually supports the risk-return tradeoff. It indicates that banks that want additional yield must take additional risk. By taking additional risk, banks take on more future credit loss and thus a lower ROA in the long run. If you believe the analysis above, you would be indifferent to individual loan decisions. Because your world view is that loans are efficiently priced, and each loan will translate to the same ROA (portfolio effects excluded). Each loan would earn the bank the market clearing ROA. Your view of the market would be supported by the risk-return relationship shown below – each loan is priced on the red line, and in the long run, all banks would earn the same ROA (excluding certain operational differences).
The reality is quite different. The markets (i.e. bankers) are not pricing every loan efficiently, and many loans are priced above or below the red line shown above. Smart bankers want to get loans above the red line and avoid loans below that red line. Here is where the analysis gets more interesting; bankers do not seem to make mistakes evenly above and below the line. The risk-return trade-off mispricing occurs much more frequently on more risky credits than less risky credits.
Bankers tend to underprice higher risk loans to a greater extent than lower risk loans. This makes sense for a number of reasons:
- Riskier credits have a wider dispersion in performance (the dispersion around the expected probability of repaying is wider), leading to more possibility for mistakes.
- Credit risk is a distant cost, while yield is an immediate gain. Humans, tend to favor immediate gratification at the expense of a distant cost.
- Bankers are typically paid incentives for higher yield, but the bank’s shareholders must pay for future credit mistakes. This is a classical agency/principal dilemma.
- Competition is constantly leading us astray in pricing. The adage in the industry has been that you can only perform as well as your dumbest competitor. If your competition is underpricing risk, the borrower is more than willing to take that capital, and your only choice is to lose that loan. It is more likely that some lender will inevitably make a risk analysis mistake on the riskier loan than the less risky loan (that comes with lower yield).
Because of the above factors, the risk-return trade-off actually looks more like the graph below. Between Zone 1 (less risk credit with lower return) and Zone 2 (higher risk credit with higher return), banks want to land in the green semi-circle and avoid the red semi-circle. Unfortunately, the red semi-circle in Zone 2 is quite large and many banks get trapped in the game of wanting to outwit the market (gain additional NIM without taking additional risk to demonstrate higher ROA).
Unfortunately, the strategy of competing for riskier credits has not worked for the vast majority of the banks in the industry. When we look at the correlation between NIM and ROA across a number of years, we find that that relationship is even less explanatory when we consider it over a business cycle. The NIM/ROA correlation coefficient approaches zero over a business cycle, and that correlation coefficient is negative during credit downturns for obvious reasons.
It would appear that bankers would be better off to compete for less risky credits because those credits are more efficiently priced. One obvious reason that certain banks continue to focus on more risky credits is that of operational inefficiencies. These banks cannot make their business model work at lower yields because of overhead costs. However, in the long run, through an economic cycle, the business model of underwriting riskier credits leads to suboptimal ROA.
It appears that loans are not priced efficiently in the market, but further, riskier loans are priced much less efficiently at the disadvantage of lenders and the advantage of borrowers. A bank’s long-term ROA would be optimized by deemphasizing underwriting higher risk credits and concentrating on operational efficiencies to make the business model profitable with lower yielding assets. It appears that for the average bank, and in the long run, the risk-return trade-off for credit risk erodes shareholder value.
Submitted by Chris Nichols on March 07, 2017