Yesterday, we covered a set of economic indicators that have proven to be unreliable at predicting the future of rates, credit, loan or deposit growth. The subject is topical as many banks are working through their budget forecasts and instead of just relying on history, many banks seek to increase the accuracy of their predictions by utilizing these indicators. One way to do this is to incorporate forecasts of these economic indicators and then use that as the basis for fine tuning bank budget variables.
The most common, is GDP. While true GDP has some predictive power, due to large revisions and other “noise,” that power only is accurate to about one quarter forward. In out-of-sample testing (predicting utilizing data that was not previously used to form your conclusions), GDP only helped explain about 9% of loan growth one quarter out. Worse, past one quarter, the predictive power of GDP, even when lagged, dropped almost to zero.
While GDP is not a great predictor of the future, these other indicators are:
Rate Spread: One of the most accurate indicators of both loan growth and credit quality also happens to be one of the simplest. If you take the spread between the 10-year Treasury (assuming no default) and the 3-month Treasury Bill (currently about 2.6%), that is a product of how steep the yield curve is. The yield curve, it turns out is one of the most accurate predictors of the future beating all indicators and most every economist. When the curve is steep, the collective wisdom of the market foretells growth ahead. This market indicator turns out to be about 30% correlated to loan growth which is about as good as it gets in the econometrics business. Almost as important, this accuracy holds to about six quarters forward (however the power of the indicator drops with each forward quarter). Presently, the current spread predicts modest loan growth ahead for 2014.
Unemployment: If you want to forecast credit quality, projected unemployment tends to be one of the most accurate predictors of credit and the related allowances and losses. This inverse relationship is based on the concept that when people are out of work, it is a sign that the economy is having problems. Conversely, when employment is improving, quality is looking up. At present, employment is set to increase for the next two quarters before leveling off through 2015.
Global Purchasing Managers Index: The export component of this index has been found by major banks to be highly accurate predictor of trade. A favorite of both the Bank of England and the Fed, this indicator explains approximately 91% of global trade activity and about 72% of US production. The problem is that unless your bank derives revenue clients that focus on international activity, this indicator probably lacks predictive power for the average community bank. Currently, this index is at 55, a near term high since 2011, and foretells of increasing international activity and economic expansion.
Branch Profitability Set: Sometimes groups of indicators together can increase the accuracy. For example, by combining the NYSE monthly average stock index with the Spread indicator mentioned above, the combined indicator can be made even more accurate, particularly one year out or less. In similar fashion, when it comes to predicting branch profitability, utilizing a set of indicators, has been found to explain almost half of branch profitability, accurate up to about 18 months out. These indicators include household formation, business growth (the number of new businesses starting up or moving into the area), household income and business mix. No surprise, but this set comes down to the function that profitable growth around the branch tends to be a predictor of profitability. Oddly, while the number of competing branches tends to influence profitability it is not a material factor. In other words, you would rather open or keep a branch in an area with good growth in profitable customers and products and lots of competition than the opposite.
Most of these indicators point to modest growth looking over the next two years so anything more than that baseline growth, means banks may have to find other ways to steal market share. It is usually a sign of an inaccurate bank model that has both loan growth expanding that a faster than economically feasible clip AND profitability improving. Usually, you don’t get both.
For example, in many metro areas, the above indicators point to about 2.5% growth and slightly improving credit quality. For a bank to forecast 5% growth, means that the extra growth has to come from taking a loan or deposit from another bank. As such, the bank either needs to have a way to determine which potential customers are mispriced in the market or must present an alternative of a better service package or more favorable pricing than where it is today. Either way, this likely means higher cost for the bank. As such, to be successful AND accurate in forecasting, banks need to forecast either greater sales/marketing expense or pricing that is MORE favorable than today’s already competitive rates.
Forecasting is a tricky business but these econometric data sets can help banks improve their accuracy and see deeper into the future. The best thing a bank can do is to play around with national, regional and state indicators in order to see what works best on a historical or in-sample basis. Once a bank has an indicator or indicators narrowed down, then it can use it to predict data out-of-sample to further hone the accuracy. By starting with the indicators included here, banks can be well on their way to being more accurate predictors of the future.
Submitted by Chris Nichols on October 17, 2013