A Simple Exercise to Gauge Agricultural Banks’ Susceptibility to Stress
Safety & Soundness Update - September 2016
Published October 12, 2016 | September 2016 issue
Credit losses at agriculturally concentrated banks continue to be a top concern for the Ninth District. Commodity prices remain depressed despite recent gains, and nationwide average farm income is half of what it was two years ago. As a result, agricultural land values are declining throughout the Midwest, and nonperformance rates on loans have begun increasing at agriculturally concentrated banks.
In the September 2015 issue of Banking in the Ninth, Ron Feldman described work done with Joseph Smith in Minneapolis to understand the risk that falling agricultural land values would pose for agriculturally concentrated banks. We have done some simple follow-up work investigating how agricultural banks fare individually when hypothetically faced with loan loss rates from the farm crisis of the early 1980s. In this note, we describe how our initial findings for Ninth District banks suggest that many banks would suffer large capital declines when tested against extreme credit losses. That said, a large majority of agricultural banks would not fail despite the very extreme losses we impose on them in the exercise.
Key features of the agricultural bank exercise
Our analytical exercise is very simple and, in some sense, implausible. We assume banks with relatively large volumes of agricultural loans will face losses akin to some of the worst seen during the farm crisis of the early 1980s. We make a few assumptions about income and payouts and then determine the level of capital these banks would have after the stress scenario.
The stress scenarios we engineer do not represent our expectation about future losses and capital impacts at Ninth District banks; for all of these institutions to experience such severe loan losses simultaneously would be almost impossible. What, then, is the point? Picking an extreme scenario helps identify the individual banks that are susceptible to “tail” stress in the agricultural sector. It also sets out one extreme end of the potential results of a very stressful event for agricultural banks.
For the stress scenarios, we pick levels of loan loss based on those seen at the height of the farm crisis period (1984-87). We choose loss rates for agricultural loans and all other loans independently. They are mapped to banks based on the state in which they are headquartered. For instance, Nebraska’s historical loss rates would be applied to a bank headquartered in Omaha. The loss rates are then applied to the current loan balances of banks.
We also make choices about other factors that would affect bank capital after accounting for loan losses. For instance, we might specify a suspension of dividend payments and equity buybacks, or set these values at the previous year’s levels.
For our initial analysis, we subject our Ninth District banks to two scenarios:
- Severe scenario
- 75th percentile agricultural loan loss rates from farm crisis period (state-by-state)
- 75th percentile losses on other loans from farm crisis period (state-by-state)
- Payouts (e.g., dividends and equity buybacks) are halted
- Extremely severe scenario
- 95th percentile agricultural loan loss rates from farm crisis period (state-by-state)
- 95th percentile losses on other loans from farm crisis period (state-by-state)
- Payouts continue at the same level as in 2015
To be clear about what the severe scenario implies, the 95th percentile loss rate is the loss rate that only 5 percent of banks in that particular state faced or exceeded during the farm crisis. This means 95 percent of banks in that state had a lower loss rate. As noted, this scenario reflects a very extreme case: We are stressing all agricultural banks in a state against losses that only 5 percent of banks faced in the farm crisis.
Initial findings for the Ninth District
We apply the loan loss rate and capital calculation choices to bank regulatory reporting data from the fourth quarter of 2015. We then examine the resulting changes in the tier 1 leverage ratio (tier 1 capital to average total capital).
End-state capital ratios reflect the severity of the scenario. After loan loss stress is applied, the average tier 1 leverage ratio drops from 10.6 percent to 10.1 percent in the mild scenario and to 5.0 percent in the severe scenario. In the severe scenario, approximately 45 percent of Ninth District banks’ tier 1 leverage ratios drop below 4 percent, which is the Federal Reserve’s current adequate capitalization threshold for this metric. Thus, we would classify 45 percent of these banks as being susceptible to crossing this regulatory threshold under very severe circumstances. The figure above shows the dramatic capital decline in the severe scenario across all the banks being examined. The figure is called a “box and whiskers” plot. The middle of the box shows the capital level for the “middle” or 50th percentile bank. The top of the box shows the 75th percentile, while the bottom of the box shows the 25th percentile. The top and bottom of the line shows the 95th and 5th percentile, respectively. In short, the current tier 1 leverage ratio (left) is much higher and less dispersed than the post-stress tier 1 leverage ratio (right).
This analysis has a number of limitations. It relies on data from the farm crisis, which may not reflect how stress in the agricultural sector would present itself today. It makes a number of simplifying assumptions about bank operations and performance. Finally, it does not take into account multiyear stressed scenarios.
Despite these limitations, it is a simple and transparent test that can identify banks that are susceptible to a worst-case scenario in the agricultural sector. Indeed, choosing an extreme case shows how large the losses have been in the past. That said, the graph suggests that a substantial majority of banks would survive this extreme test with positive capital, and likely would be able to continue supporting their community.