A simple graph of agricultural land values in the Midwest over time (see below) must give Ninth District readers pause for four related reasons.
- Land values have reached record levels after a very steep climb.
- The factors associated with rising agricultural land values, such as output prices for major commodities, have reversed.
- Many banks in the Ninth District have significant exposure to the agricultural sector.
- Many banks failed the last time agricultural asset values fell after such a rise.
What would a fall in agricultural land values mean for banks today? How much would loan losses go up? How far would capital fall? What would happen to the chances of bank failure?
I have worked with my colleague Joseph Smith to try to answer these questions. Our initial take is that most “agricultural banks” (defined below) would suffer manageable losses if agricultural land values fell a lot. A smaller number of banks particularly sensitive to a fall in land values could experience much larger losses.
These initial results are very uncertain in two ways. First, we base the answers on statistical analysis that relies on the best but very imperfect data we have. The results come with the inherent large range of possible outcomes that such analyses produce. Second, our results depend significantly on the degree to which relationships between variables such as agricultural land values or agricultural income and bank loan performance today are the same as those relationships from the last agricultural banking crisis or have relaxed. If the relationship has not changed, losses will be higher. I offer more details on our analysis after providing a few facts on agricultural banks in our district.
Ninth District Agricultural Banks
49 percent of banks in the Ninth Federal Reserve District, or 279 institutions, are “agricultural banks.” That is, more than 25 percent of their loans are secured by agricultural real estate or are made for the purpose of agricultural production. We view 26 percent of Ninth Federal Reserve District banks as being very concentrated in agriculture, with more than 50 percent of their loans for agricultural production or secured by agricultural real estate.
Two additional factors set the context for the exposure of these banks to changes in the agricultural economy. First, lending by Ninth District agricultural banks has been growing at a steady pace. For the median district agricultural bank, lending has grown at an annual rate of 3.6 percent since 2010 versus 0.3 percent for the median nonagricultural bank. Second, capital levels of agricultural banks and nonagricultural banks are roughly the same regardless of concentration. The median capital level (Tier 1 leverage ratio) for nonagricultural banks was 10.1 percent. For banks with more than 25 percent but less than 50 percent of their lending to agriculture, it was 9.7 percent. Finally, for banks with more than 50 percent of their lending to agriculture, the median capital level was 10.3 percent. Thus, there does not seem to be a stable relationship between capital and concentration. None of these facts means that a fall in agricultural land values leads to big problems. They do suggest good reasons to be cautious.
Estimating the Effect of a Fall in Land Values
We have to make many assumptions to estimate how a fall in agricultural land values translates into potential problems for the agricultural banking sector. I describe a few. We first have to assume how far land values fall. We pick several potential declines but will focus in this summary on a 25 percent fall in one year. Such a very large fall mirrors the worst two years in the prior agricultural crisis, and thus represents a “tail event.”
We then have to choose a particular statistical approach. We rely on two methods that economists at the Federal Reserve have used for other types of quantitative “what-if” analyses of banks. One method focuses on the “average” bank, while the other allows for examination of banks particularly sensitive to a given bad event, like a fall in real estate values.
We then have many decisions to make about which data to use in the analysis. To list just two:
- What happens to other variables (e.g., the condition of agricultural producers or agricultural output prices) when land values fall?
- How do we incorporate banking experience from the agricultural banking crisis into the analysis if we think underwriting of loans and economic conditions today are not
Space limitations prevent a full description of choices we had to make in this note, but we will present a full paper on our analysis at the Federal Reserve and Conference of State Banking Supervisors Community Banking in the 21st Century, Research and Policy Conference. The statistical approach suggests results for the “representative” bank but cannot project the impact of falling land values on any particular bank. Management at each agricultural bank should assess the potential effect of a decline on their institution and take appropriate measures to control their risk exposure.
What do we find?
- We find relatively low losses for the average agricultural bank, even when we structure our analysis to maximize the downside effect of a fall in land values (e.g., by focusing on the experience in the last agricultural banking crisis). We estimate that the net charge-off rate of the average bank would jump from 0.2 percent to between 1.22 percent and 1.77 percent at its worst. We also estimate a decline in capital of around 134 to 183 basis points over two years, assuming the 25 percent fall in agricultural land values and related bad outcomes.
- In contrast, there are banks that the data suggest are more vulnerable to a downturn in the agricultural land values than the average agricultural bank. Think of these banks as being at the “far end” (i.e., 95th percentile) of the distribution of agricultural banks in terms of the outcomes they will suffer when land values and producer income fall. We estimate that the net charge-off rate of these outlier banks would jump from 1.1 percent to between 3.0 percent and 4.5 percent. In addition, we estimate declines in capital of 163 to 298 basis points
over two years, assuming the 25 percent fall in agricultural land values and related bad outcomes. These banks have the worst 5 percent experience in the hypothetical agricultural downturn which, by definition, makes them unusual. But
5 percent of all U.S. agricultural banks (as defined in our study) represents 92 institutions, a number that is significant from the standpoint of bank system stress.
Again, I must emphasize the uncertainty of any prediction along the lines I just described. That said, I am confident that banks can protect themselves from bad outcomes through
effective risk management of their agricultural exposure,
a topic covered well in prior Federal Reserve guidance
(see “SR 11-14”, 2011).