The time element in default rate calculations

This blog post is part of a “Beyond the default rate” series, which investigates what the default rate actually means, how it affects you and why you should know more than simply the number you see on some report.

If you haven’t yet read the previous post in the series, please do that before reading this one. This post will be building upon the foundation we set in there.

As a reminder, I repeat the default rate formula here again:

Where:

D – default portion of the portfolio

LP – the loan portfolio

D% – default rate of the portfolio

The LP part of the formula

As if the sum vs count part wasn’t confusing enough already, there’s also a time element at play when you consider calculating for defaults. This is what we’ll look at now – the LP or loan portfolio part of the formula.

Let’s say you want to know what the current default rate for EST loans on Bondora platform is. Do you simply take all the defaults and figure out its proportion out of all EST loans issued so far?

You could, but you’d end up with some almost random result. The correct answer would be no. The not so obvious thing is though, what should you look at instead?

Cutting the end portion of dataset

One thing you should always exclude from your calculations are the loans issued in the last few months, as these don’t even have a technical possibility to default by now even in cases where they are certain to do so. If you were to include these, then simply issuing more loans will reduce your default rate, although in reality nothing has changed and you’re simply investing more funds.

There’s no certain rule on how many months you should exclude, however, since a decent proportion of defaults happen around the 4th-5th month, excluding the last 4-6 months of loans may give you a decently accurate result already.

Cutting the beginning

Another thing to consider is that things have changed a lot over the years. Probably the default rates for loans issued in 2009 have not much to do with the loans issued today.

In addition, if the recovery is high and/or default rates are low for a decent amount of loans at one period, then including these loans in your calculation can skew your data to show a lot lower default rate than actual current rate is. This may lead you to invest into loans that actually are higher risk than you think.

Same happens in the opposite direction if the default rate has been high before and now is significantly lower (like it seems to be the case with some Finnish loans). By including the loans from earlier periods in your calculations, you will avoid investing into a profitable segment with plenty of investment opportunities due to negatively skewed results.

Again, there’s no certain rule on how to proceed with this issue. Use your best judgment. However, keep in mind that this may also be difficult to do in some cases due to very small sample sizes. For example, if there are only 50 EST loans issued within your selected period, then you won’t be able to get any meaningful results out of your analyses.

Keeping changes in mind

In certain cases it will be important to also keep changes on the platform or underwriting processes in mind when doing your analyses. For example, if you’re simply analyzing all the EST loans, then you should perhaps keep in mind that in the early years Bondora also issued loans with credit score of 500 (active payment problems) or at one point loans were additionally granted to businesses.

Including these in your analyses can likely skew your data in whichever direction and can in some cases make your results close to meaningless.

I have collected the timeline of changes from the weekly newsletters in the post here to help you find relevant information faster and easier:
https://rahafoorum.ee/en/bondora-timeline-changes-and-returns/

Again, there’s no exact rule on what affects defaults and how much, and thus you may just as well be wasting your time in some cases when trying to account for those, but it’s your own decision whether you want to better be safe or think the results will be close enough.

Simplest option would be to just include the latest portion of loans where least amount of largely meaningful changes have occurred.

Yearly default rates

Although the above-mentioned approach is one of the most popular and usually is the one that most people should use, the most accurate and meaningful measure would be a yearly or monthly default rate. By this, I mean the default rate of loans issued within a certain period, after a certain period.

For example’s sake, let’s say loans issued within a month and the default rate of these loans after 6 months and after 12 months.

With this approach you could then map any changes over time and any increases or decreases in issued loans won’t have an effect on your calculations.

However, this approach is the most time consuming and is likely overkill for most purposes if you’re an individual investor without a huge portfolio or great automation skills.

In addition, this is often not even feasible, since there is too little data to get any meaningful results if you’re trying to look at any more specific segments than simply total portfolio or all the loans together that were issued in different countries.

Conclusion

Even though in this case there are also several different options to use, each have their own flaws and virtues. If you didn’t yet realize it after reading the first post of the series, hopefully you’ll understand by now that calculating the default rate is not as straightforward and easy as it might seem in the first place.

In the next post we’ll revisit the default portion of the formula again and will look into a few nuances related to this, including a typical mistake new investors make when thinking about this topic.

To make sure you get the notification about the new post when it is published, subscribe below:

Read the next post in the series:

https://rahafoorum.ee/en/initial-defaults-vs-current-default-rate/