How Banks Can Really Diversify Their Loan Portfolios

Cross correlations for bank credit loan portfolios

You can slice and dice your credit portfolio all you want, but if you are not paying attention to cross-correlations your efforts could be sub-optimal. For example, many banks separate their multifamily exposure away from their single family exposure. In some markets, these two subsectors are almost 80% correlated. A drop in housing prices usually occurs at the same time as a drop in multifamily values and in similar fashion delinquencies at banks usually move in lock-step. While segmenting your portfolio may serve to check a regulatory box and may help to comfort your management, banks should not draw too much solace. In this post, we explore the top customer segments that CenterState has found to help true diversification.


Separating Risk And Correlations


It is important to recognize the difference between risk and correlations. A sector can be risky and still present a low or negative correlation to the rest of a bank’s portfolio. For example, loans which are tied to the performance of alternative energy, such as solar, are risky given their default history and cash flow volatility but is one of the major lending sectors that exhibit very little correlation to the general economy. Thus, a bank that structures and prices this risk right helps offset the volatility and adding it to the general portfolio serves to decrease the risk of the overall portfolio. In this manner, one plus one can be less than two. 


Impact of cross-correlations on bank loan reserves / allowances / ALLL


This misunderstanding has often gotten in the way to properly understanding and putting into practice the utilization of cross-correlations when designing a bank credit portfolio. In the example above, when you combine equal exposure of commercial real estate (CRE) and commercial loan to a consumer electronics company, the average weighted expected loss is 51 basis points in our example. However, because these risks have low correlations to each other they help offset the risk. Real estate tends to follow the general economy while consumer electronics does not. As such, when the expected loss increases for CRE, the odds are that the expected loss for the consumer electronic firms does not. Thus, the combined expected loss, on average, is less than the weighted average by about 2 basis points in our example.


Our Top 25 Low Correlative Industries


Here is our latest research on the cash flow and market capitalization of the 25 industries that have the lowest correlations to the general economy. While we get specific, in general, healthcare, technology, consumer products, agriculture, food, and telecomm are some general industries that we focus on.


This means that we can adjust pricing and structure to be more attractive to these focused industries. It also means that we can have more of a sales effort and be more tactical in our marketing. The more we bank schools, water utilities or technology firms, the more exposure we can handle in real estate. Alternatively, the other way to look at this is the lower our cross-correlations are across our credit portfolio, the less reserves and/or capital we need to hold.


Below is a list of our top 25 sectors with the lowest correlations to the general economy. In addition, we have included a volatility measure as a proxy for risk letting you know what areas you might need to move with caution on. 


25 industries with the lowest cross correlation to the economy


Putting Cross-correlations Into Action


Banks should attempt to correlate income, property values or market values to have an understanding of how various sectors interact with each other. We correlated the above sectors to the general economy, but in the past, we published how different real estate subsectors were correlated with each other (HERE) and how a bank’s real estate holdings are correlated to their investment portfolio.


There are market indices such as commercial real estate REITs, transportation, utility and others that can at least get your credit team in the right ballpark of how different sectors move together. Bankers can use the industries that are applicable to the bank and as long as your data is representative and homogenous enough, bankers can create a correlation matrix in Excel (using the “correl” function) in order to view the relationship of every sector on every other sector. In this manner, lending and credit teams can focus on those sectors with low correlations. Weigh each sector by the outstanding loan amount and apply the cross-correlations and bankers now have a more sophisticated tool for managing their loan portfolio.


Breaking your credit portfolio into various loan types is interesting, but may not be instructive enough to serve to actually manage risk. Understand and construct your loan portfolio using correlation analysis and your bank can gain a significant strategic advantage against community banks that are not paying attention to how their credit risks align.