Moving forward requires a clear understanding of where you have been. Take a look back and then chart a clearer, well-informed path for the future. David Rogers’ career in risk has spanned more than 25 years, so this isn’t the first recession or crisis he has weathered. The Risk Management Knowledge Exchange asked Rogers for a short interview to see if perspectives have changed somewhat now that the crisis is less pressing and regulations are beginning to immerge.
WAYNETTE TUBBS: Could you talk about the situation many banks and financial services firms found themselves in and also the aspects of the underlying scoring process that perhaps some businesses had not fully appreciated?
DAVID ROGERS: Prior to the crisis, banks across Europe, and I believe many banks globally, had spent a long time in a growing market with very benign default patterns. This made it quite difficult to capture a sufficient default population from which to build their statistical models. So in a lot of cases, the people at the model development sites took short cuts that – from a numerical sense – were better than nothing, but from a business perspective had dire implications. Generally, their population choices might not have been the best choices from a business perspective. It was only as a more integrated view of the data collection and data usage spread more thoroughly throughout the organization that the affects of these analytical and pre-processing decisions became more obvious.
For example, default models should have been built using populations of delinquent customers of more than 90 days. But, in some cases banks were unable to find a sufficient number of borrowers who fit that profile. So, the modelers would look for default populations of more than 60 days, but a model using that data doesn’t quite predict the groups you want. Banks also found problems when the regional default pattern started to come through on the data. The banks were in no position to explain what was happening.
There are numerous examples of naïve approaches, which at the time appeared conservative but in hindsight were less than adequate. Collateral backed securities are an example of this. The challenge was keeping up with the pace of business while ensuring that new areas were tackled with the appropriate consideration.
What would you recommend as the key areas banks or insurance companies need to improve to make their scoring models more applicable to their business?
ROGERS: That’s a difficult question to answer. I have always felt that there are two things that banks need to watch at a board level: a change in the risk profile of the existing business and a change in the profile of existing customers. These are quite dependent on each other. I also believe that banks and insurance firms need to ensure that everything is risk-based, including using risk-weighted assets to drive the business. There is a sense in the smaller European countries that the reason it got so bad was because one or two banks were making tremendous profits by capturing the new part of the economy that was quickly growing while the other banks held to their standard approaches for a number of years. Eventually these more traditional banks gave in to shareholder pressure and went after the expanding, less understood, market. By abandoning their position – more conservative banking – a large number of banks across Europe ended up essentially nationalized.
Earlier, you talked about shortcuts in data sampling to find a default population for modeling. The data aspect is always an important element at the beginning of these projects: Do you think that the lessons learned are necessarily understood across the business at the highest level of the organization?
ROGERS: I don’t think that there’s any visibility at the higher level of the organization of these approaches to modeling. Those short cuts might have been taken in the mid-2005 to 2006s when people didn’t have a more integrated view of what was going on and couldn’t see the danger. I don’t think that is the case as of 2007 and 2008. By then, population visibility went across the business including even the lenders. An illustration would be where a firm was doing a credit score where it was able to return details about the characteristics of the individual alone and in the context of the population. So that was pushing the information that was available from the risk systems out onto the lender’s work bench in a much more detailed fashion so that they could make a more risk-aware decision.
Do you see a change in who is making the decisions about what data to collect and how to collect it?
ROGERS: Yes. In the last four or five years, the risk teams would have made the risk-related data decisions. However, as you look at those decision-making teams today, you see a tendency to bring the business people into the process of deciding how to collect the data. That also includes the people doing the lending.
When we talk about the crisis and what went wrong, were the banks missing expertise that would have helped them avoid this trouble?
ROGERS: That’s an interesting question when you consider the hubbub related to quants making the mistakes that brought down Wall Street, but yes, in those areas where banks got into trouble, they probably didn’t have the skills. You can see that when you take a look at some of the banks that are quite strong in more traditional banking but seem to be exposed in the recent construction bubble.
Across Europe and globally, there has been a lot of movement of risk professionals in the last 12 to 18 months. I’m wondering what effect you think this risk practitioners’ musical chairs has all had on the remaining practitioners and on the organizations?
ROGERS: In many cases, the people who moved would have been at the center of the risk teams in the place they were previously working. Now, you have new risk people, new risk structures in a lot of the organizations and with a lot of the old ways of doing things gone: I believe this can be an opportunity. More progressive firms will embrace new risk techniques and look to see what they can develop in terms of a risk-based approach – not to take unwarranted risks as has been happening in the past, but to be more prepared for what they need to do in terms of business issues and to ensure success.
There is also a growing adoption of industry wide credentials through organizations such as GARP and PRMIA where the skills of risk personnel can be compared with those across other financial institutions. This is an extremely positive step that educates risk people about the broader risk subject and informs them about risks that might be coming down the line.
Do banks have a better understanding of their data today?
ROGERS: I think certainly those banks that have more recent versions of a banking intelligence solution on the credit scoring side can track the movements in the populations. They are very well equipped to at least identify where the trouble is starting. Given the way the financial services sector has collapsed globally it is difficult to argue that any single area caused, or would have prevented, the meltdown. In a globalized market it seems that awareness of all elements of the business, and how those elements interact, is a vital component of a robust business.
If you have questions for Rogers, you can find his contact information on his author bio page or you can write your questions in the comments section below. For more information about making your data into the corporate asset that your organization deserves, download and read this white paper: Data as a Board-Level Issue.