The recession has played havoc with the credit card portfolios of the major bank card issuers.
As the U.S. economy deteriorated, credit card delinquencies skyrocketed and writeoff rates for some of the largest portfolios more than doubled, to 6% or 7% of credit card outstandings. Some have hit double-digit levels.
The financial impact of these writeoffs on a bank's bottom line can be significant. For a $5 billion portfolio, a doubling of gross writeoffs from 3% to 6% implies an extra $150 million annual direct hit to net income.
A Profit Center
Given that most credit card operations were solidly profitable before the recession, rapidly escalating losses can become a significant problem.
In addition, an extremely weak economic recovery and high levels of consumer debt imply that credit card writeoffs will not decline quickly.
Fortunately risk managers, sophisticated modeling tools can now accurately predict writeoff levels in detail, both on a regional basis and by account cohort. These tools measure risk for individual portfolios and identify problem sets of accounts and problem regions.
In addition, these techniques can be used in the portfolio acquisition process to identify regional and fundamental risks before acquisition.
Risk managers first used vintage, or life-cycle, analysis to predict credit card losses during the 1980s. Rating agencies such as Standard & Poor's Corp. also use vintage analysis to evaluate the potential risks for asset-backed credit card securities created by large bank issuers.
A form of vintage analysis has even been used recently by Edward Altman at New York University's Stern School of Business to examine the loss experience on junk bonds.
Mr. Altman has shown that, contrary to popular belief, the default risk for a fixed set of bonds does not steadily increase as the bonds age. Instead, losses peak and then begin to subside, following a kind of life cycle.
Grouping by Similarities
Simply put, vintage analysis relies on segmentation to analyze a group of accounts with similar credit characteristics. This segmentation can be organized by either marketing cohort or account vintage (age of the account).
The analysis thus focuses on a relatively homogeneous set of accounts that were added to the portfolio at the same time.
The rationale underlying vintage analysis is that the "bad" credit accounts will tend to become delinquent and written off more quickly than the "good" accounts.
As these accounts are written off and are pulled from each portfolio cohort, a cleansing takes place that increases the average quality of the remaining credit card accounts.
Deadbeats Filtered Out
Each cohort or vintage of accounts is essentially being filtered of its deadbeat borrowers. Aside from external changes in the economic environment, like a recession, the accounts that remain on the books slowly become the cream of the crop.
Recent analysis undertaken on a large national credit card portfolio has clearly demonstrated that individual cohorts of credit card accounts follow a natural aging process.
This analysis of writeoff behavior shows that, independent of economic forces, writeoff rates-gross writeoffs divided by outstandings - tend to rise rapidly during the first 20 to 24 months of a vintage's lifetime.
They peak at about two years and then slowly decline toward a long-run stable writeoff rate. Credit card vintages therefore follow a stable rising and falling pattern - unit a recession hits.
Economic Forces Matter
As many bankers have discovered, analyzing credit card writeoff behavior independent of the influence of economic forces like recession is an inherent shortcoming of the simplified method described above.
This is because in some cases even older, "clean" groups of accounts with declining writeoff rates before the recession tended to have a large jump in account delinquencies and rising, not falling, writeoff rates.
In times of recession, even credit card holders with superb credit histories can become unemployed and fail to make timely payments. In general, their mortgage debt, home equity loans, and auto loans get priority.
A White-Collar Recession
Credit card delinquencies have also increased in prevalence during the current recession for traditionally recession-proof white-collar workers.
To improve life-cycle analysis and make it more empirically based, regional economic behavior must be integrated into risk management models.
Forecasting models can then capture both the life-cycle behavior and the deviations from the life cycle caused by the recession.
Risk managers will thus be able to better predict the true downside risk in a credit card portfolio as well as the long-run writeoff outlook for when the economy improvements.
Two Variables Studied
To improve life-cycle analysis, we modeled the individual portfolio cohorts as a function of two kinds of variables.
First, we modeled the pure life-cycle behavior that was independent of economic forces using a time variable to capture the natural aging process in each cohort.
Then we added regional economic variables such as unemployment, wage and income growth, and consumer confidence.
We found that as unemployment jumped in New England, for example, the contribution to higher losses in the entire portfolio from New England accounts became significant.
On the other hand, lower unemployment in the Southeast translated into a smaller jump in writeoffs in that region.
The added accuracy gained by disaggregating into regions using regional economic variables in these life-cycle models was extremely significant. These variables strongly explained how credit card writeoff rates deviated from their pure life-cycle pattern for individual cohorts and regions.
And they did so based on the timing of the recession in each region. This means that forecasts of unemployment and income growth that predict the timing of the end of each region's recession can foretell how quickly writeoff rates will decline toward normal levels.
Two additional twists were undertaken. First, we were able to quantify and compare the overall credit quality within each cohort. This was done by looking at each life-cycle peak writeoff rate adjusted for the affects of the recession.
A cohort of lower quality might have a peak writeoff rate 15% to 20% above the peak rate for a higher-quality cohort.
Identifying relative credit quality for each cohort allows the risk manager to focus on those groups of accounts with lower overall credit quality. Credit limits can be controlled, authorizations for new credit charges can be scrutinized, and marginal accounts can be monitored more closely.
Second, we were able to compare relative writeoff behavior across regions, independent of the recession's influence. New England and California are two cases in point. While writeoffs rose rapidly in New England, it was clear that the region's rapid economic deterioration was the primary culprit.
Independent of the recession, however, the peak writeoff rate in New England was actually quite low. This is consistent with other default behavior in New England, on mortgage loans for example, and seems related to the region's traditional conservatism.
A Different Story
In contrast, the California economy up to now has fared better than New England's, but adjustments for the recession show that peak writeoff rates in California are significantly higher than in New England.
This may be related to a more liberal financial lifestyle, along with legal differences that make it easier to declare bankruptcy in California.
Two years ago, when the economy began to deteriorate, credit card products for most banks were a strong source of income net of writeoffs, which could partially offset the problems in commercial loan and highly leveraged transaction portfolios. Unfortunately, recent loss behavior shows that credit card portfolios can be highly sensitive to recession.
With the entry of nonbank credit card providers such as American Telephone and Telegraph Co., competition on interest rates and annual fees is heating up, as shown by Citibank's recent policy changes on pricing.
Danger of Interference
Regulatory intervention by the government has the potential to further erode the profits banks earn from credit cards. In this environment, understanding and controlling risk will become the key to preserving credit card profitability.
The benefits of early-warning signals on regional credit deterioration can be significant. In the current slow-growth environment, the identification and anticipation of credit problems will improve the credit process, and the best tools available will incorporate differences in the impact of regional economic behavior on card portfolios.