Posted on July 10, 2014 by Ben Connard

We have an immense amount of data at our fingertips both on a macro level and on a company level. We can find the change in housing prices over time in almost any developed country, export the data from Bloomberg and make a chart in less than 10 minutes. This access dramatically improves our research capabilities, but leads to two questions. What data is actually worth analyzing and how do we analyze it?

This situation arose this past quarter when we were researching international banks. After reviewing dozens of banks across multiple countries, we purchased a Singapore bank, the United Overseas Bank Limited. Our goal was to find a bank that makes loans which rarely default and had sufficient deposits to fund those loans. We felt that the United Overseas Bank did this exceptionally well and had enough capital to absorb unforeseen losses.

We arrived at this conclusion by analyzing data. There is a wealth of available information on banks due to intensive reporting requirements so we first had to decide which ratios were the most important. For example, banks record a provision for loan losses, which is money set aside to cover losses they estimate will result from defaulted loans. This number seems important, but we felt the actual loan losses were more important. A bank may set aside 1% of loans as a provision, but if it’s actually writing off 2%, what good is the provision? And what does this say about the bank management?

Eventually, we found a set of criteria we wanted a bank to meet, screened on the criteria and did a more in-depth dive on the banks that met our strict criteria. We think we found a strong bank with solid management and good growth prospects. Getting the data was easy. Analyzing and understanding the data was the hard part.