How Consumer Demand Drives Credit Risk, Jan 29 2012

This is the first update from Prescient Models. I know your time is valuable, and I do not want to fill your inbox, but I thought you might like to hear of some recent developments. These newsletters will be rare, so that their content can be kept to the important topics.

In researching the origins of the US mortgage crisis, we uncovered not just the first signs of that crisis, but a measure of a credit cycle that has impacted retail lenders several times over the last couple decades.

What we discovered was not only Demand vs Credit Riska clear credit cycle (shown at right), but also that it correlated strongly to shifts in consumer demand as reported in the Federal Reserve Board''s Senior Loan Officer Opinion Survey. The same survey asks lenders if they are loosening or tightening lending standards. Our analysis showed no correlation between lending standards and industry-wide credit risk.

Underwriting is a process of selecting the best loans from those who apply. But what if none of the good consumers apply? That is why estimating consumer demand is so important. When consumers want loans, they will be good. When consumers do not want loans, those that you write will be much higher risk.

The full story will be published in RMA Journal in March, and the academic paper was presented at the Conference on Credit Scoring and Credit Risk in 2010 in Edinburgh. All of our papers are available on our website.

The problem here is that the government survey of consumer demand is quarterly, released a month after the end of the quarter. That means the information is many months out of date. Prescient Models is now providing free weekly estimates of mortgage demand and consumer loan demand (auto and card) as an early warning indicator to lenders. You can find them on our forecasts page. We will not send newsletter announcements to minimize noise.

Social Media Index of Nonfarm Payroll

Yes, we've been busy. We have also created an estimate of the level of US nonfarm payroll using information from social media. By gauging the level of interest in such things as jobs and unemployment benefits, we can create a reasonably accurate estimate of nonfarm payroll.

Changes in nonfarm payroll are one of the most important inputs to credit risk forecasts and stress test models, so we thought that having a weekly measure could be quite valuable to lenders.

You will find it free on our forecasts page.

What else are we doing?

New loan-level modeling techniques, new data sources, training courses and more. If you need advice on how to create an in-house forecasting solution, help in creating one, or just need someone else to do it, let us know.

Blog Next