Credit analysis is done to determine the likelihood a debt can be repaid. Often, you will hear about how analysts concern themselves with the 5 Cs of Credit, or particularly the level of cash flow. No matter what analysts focus on, it all seems highly analytical.
While there is no doubt a method to the madness, there is much more method than one may realize. All analysis can be broken down into two realms: cross sectional analysis and time series analysis.
Cross sectional analysis is the study of a particular aspect in a single point in time. For example, say you are concerned with the market rent for an apartment building, and you go out and gather information on what other apartments are renting for. That is cross sectional analysis. When you spread a guarantor’s personal financial statement, which is also cross sectional analysis. The key here is we are not examining how the details change over time, we are only examining the details at hand today.
Time series analysis, on the other hand, examines how a particular aspect changes over time. If you want to know how average rent for an apartment in a city has changed over the past three years, which is time series analysis. When you spread a guarantor’s tax returns for the past three years to examine the consistency of their income; you are conducting time series analysis.
Time series analysis tends to be more challenging, because it is the responsibility of the analyst to create standard time periods, and then the analyst is further tasked with determining which events belong in each time period. Often people struggle to keep it all sorted out.
For example, we may look at a credit report and find a monthly car payment, and habitually, we create an annualized required payment by multiplying by 12 months. But, if a car loan pays off in 6 months, it should not be counted as a recurring debt service requirement for years into the future. And in the present year, it should only be counted as payment for 6 months and not 12.
In many cases, the application of both time series and cross sectional analysis is well warranted. Take for example average hotel revenue per room (RevPAR). First, it is prudent to examine the RevPAR of a similar class of properties (e.g. budget hotel RevPAR if a budget hotel) to determine if projections reflect the current market conditions. This is cross sectional analysis.
But then, you will want to check if RevPAR has been increasing or decreasing year over year. That is time series analysis. And, you may find that RevPAR is higher in summer and lower in winter, making monthly comparison of RevPAR troublesome. An analyst would be wise to create an annualized average RevPAR, which can then be compared to annualized projections or to historical annual tax returns.
When to use cross sectional analysis and when to use time series analysis is a tough call, but for the most part, it is best to use both if at all possible. For example, cash flow is often looked at as a time series, with historical financials compared to debt service. But equally important is where that cash flow is coming from. If cash flow depends on a single source or payer, there is concentration risk which must be investigated. The analysis should shift to a cross sectional understanding of who that payer is, rather than a time series of how successful debt service coverage has been in the past.
Instead of asking yourself whether it makes more sense to apply time series analysis rather than cross sectional analysis, I think ultimately you must ask whether it is okay not to do one or the other. If you cannot produce a defensible answer as to why to ignore using one method, then there is probably a need to do it!