Skip navigation

Many unintended consequences in business & finance occur because upside (profitability) and downside (loss) risk have very different characteristics.

In 2007 I sifted through 330,000 RMBS (Residential Mortgage Backed Securities) assets or loans with the purpose of developing a default & loss econometric model that would enable my then employer differentiate between good and bad RMBS bonds.

I had to do a lot of data structuring and clean up to standardize incoming data from about 20 different servicers. I should say data structuring and data scrubbing. And after that sift through this asset data set to filter out the bad data. So you can imagine the amount of work that went into this.

My strategy to identify the loss quality of a bond was to use the origination and close data to forecast future losses. Why? Because once you have purchased the bond you are stuck with whatever losses that unfold.  If this was possible we could very quickly eliminate the poor quality bonds, and use cash flow methodologies (more on this in a later blog) to evaluate the better quality bonds. We did not want to expend effort valuing poor quality bonds.

Dichotomy of Metrics
Well, at least that was the theory. What went wrong? In residential mortgages FICO scores are a major component, if not the primary driver for assessing future default behavior. I found that for residential mortgage FICO scores at origination were poor predictors of future defaults. How? Build two probability distributions of FICO scores. The first is a distribution of FICO scores of the borrowers at origination for all assets. The second is the distribution of the FICO scores at origination for defaulted assets only. You could not tell the difference between the two distributions!

Bewildering. Was there an analogy in other areas of finance? Yes. Company pro forma profitability analysis are good predictors of future profitability, but they are poor predictors of severe losses or bankruptcy risk. To get a handle on future company bankruptcy risk one requires a tool like Altman’s Z-Scores.

Here we observe a dichotomy in forecasting characteristics. On the one hand pro forma informs us of future profitability and on the other hand Altman’s Z-Scores informs on bankruptcy risk. We concluded that FICO scores behave like pro forma profitability, that they are good predictors of success but poor predictors of defaults.

I must add that unlike investment bankers and fund managers who are focused on upside risk or profitability, my career in financial services has been focused on downside risk. Therefore I did not test if FICO scores are good predictors of success.

When we recognized this dichotomy we abandoned RMBS bonds as a viable investment vehicle. But this story does not end here. I have an extensive business reengineering background and investigated how FICO scores are generated.

How Credit Cards Drive FICO Scores
From a business process perspective recent FICO scores are generated by your credit card issuer. Really. You the card holder use your card to transact purchases. In so doing you create debt. How much of this debt you pay down, how quickly, etc., is reported to the rating agencies. This reporting provides an insight into your debt payment characteristics which the rating agencies convert into FICO scores. FICO scores appear to predict defaults because they are adjusted downwards as your credit card payments deteriorate. That is, in my opinion, FICO scores have a limited window of about 90-days within which they work, but by then it is too late for the card issuer. Or no credit card visibility no FICO adjustments.

Here is the weak link in this whole business process. Without your credit card spending and payments reported by your card issuer, the rating agencies would not have a clue what your recent credit rating should be.

Take 2 people John & Jay. John has house and car payments and uses his credit card for all his purchasing transactions. Jay also has house and car payments, but he only uses cash for his purchasing transactions.

Now both have suffered a partial loss of income. They both continue to make house and car payments. Therefore the ratings agencies cannot tell from these payment that there has been a partial loss of income. Both reduce their spending to manage within their new realities. But John has now slowed his credit card payments. The rating agencies see this and are able to adjust John’s FICO scores accordingly but not Jay’s.

The use or not of credit cards provides asymmetric information to rating agencies. Now guess what happens as credit card companies ramp up interest rates, consumers pay down their card debt and stop using their cards.

Unintended Consequences
All manner of unintended consequences occur.

1. The banking industry loses it recent FICO information for a much larger proportion of the population.

2. Consumers do get penalized with lower FICO scores for not having recent payment histories.

3. First time home buyers will face higher rates because they don’t have sufficient recent history.

4. FICO scores are no longer as “effective” as they were “supposed” to be.

5. Because of this additional “FICO risk” for the same “risk” bonds with credit card debt should trade at a discount to bonds with residential mortgages which in turn should trade at a discount to CMBS bonds.

6. As I had stated in my Feb/02/09 post Risk-Reward is Non-Linear, increased rates will increase defaults. The unintended consequence of this is that banks that have rushed to increase rates to 30% before the new laws come into effect, will have locked themselves into a population demographics that experiences a higher default rate. Therefore we can expect to see an increase in credit card losses.

I’m sure there are a lot of very smart people in the credit card and rating industries, but… This situation is not dissimilar to what I observed as a management consultant reengineering manufacturing companies – one may have the best, the brightest and the most experienced managers but surprisingly when they come together at their management meetings they do strange things.

There needs to be a better way to address bank losses on credit cards and a better way to treat customers. Remember your customers are not your enemies.


Disclosure: I’m a capitalist too, and my musings & opinions on this blog are for informational/educational purposes and part of my efforts to learn from the mistakes of other people. Hope you do, too. These musings are not to be taken as financial advise, and are based on data that is assumed to be correct. Therefore, my opinions are subject to change without notice. This blog is not intended to either negate or advocate any persons, entity, product, services or political position.


One Comment

  1. It is these people who thought they know econometrics and statistics yet don’t really understand how a model should be used that make such innocent comments to confuse innocent people. sigh… and you think behavioral science with quant approaches is the same as the problems in manufacturing firms… ha ha ha

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: