Is Bondora giving away your money?

NB! Bondora has released some additional details in regards to the process. Last updated 26.03.2016 8 AM. Relevant sections have been marked accordingly.

Since the information available is far from complete, any comments on possible misunderstandings or mistakes in the post below are more than welcome and I’ll gladly improve on the analysis accordingly.

IN SHORT

• Apparently yes.

Introducing the DCA to the mix

Before analyzing the latest update from Bondora on the topic, let’s take a step back and review the timeline of changes that have occurred so far in the debt collection area.

• Update 26.03.2015 10AM: June 10th, 2015 – My attention was brought to another blog post on the topic that I wasn’t able to find the first time around. There’s a mention of a 3-6 month test project to start sending loans to DCAs. No mention that this would be covered by investors though.
• August 12th, 2015 – Bondora publishes a blog post titled Legal structure: What are we changing?, where the last paragraph mentions vaguely about using collection agencies due to new requirements arising from some changes in laws and possible cost to investors:
  

  Other changes are related to the improvements made in our debt recovery process. We have started using collection companies since companies we have now shortlisted have been able to prove their ability to recover debt faster than through courts. Since in different countries the regulations governing consumer debt collection vary, there might be a need to deduct third party debt collection fees from recovered cash flow whilst the claim is reduced by the full amount paid by the debtor to the collection company. The LGD parameter used in risk pricing already accounts for such deductions as it uses net cash flows from defaulting customers as basis of the calculation.

• September 23rd, 2015 – Bondora publishes a follow-up post on the topic explaining the process and revealing some indicative DCA (Debt Collection Agency) cost ranges that investors will have to bear compared to previous “borrower pays” policy.

  

• September 24th, 2015 – In the comments of the abovementioned post, the back then CMO Jevgenijs mentions first time in public that the new process is applied to loans that have defaulted after 8th June 2015. This is of course a bit weird, considering that first mention of anything about this topic was published in August, a more than 2 entire months after the mentioned deadline here.

  

• Sometime in the beginning of December or end of November 2015, a lot (if not all) of previously Stage 3 (Decision received from court, proceeding accordingly), Stage 2 (Court case filed) and even some Stage 4 (Passed to the bailiff/recollection) loans were passed in secrecy to DCAs without notification to investors. Based on my portfolio, these included loans that defaulted as far back as April 2014, more than a year before the previously mentioned June 8th, 2015. While it’s unlikely that collection agency is better at getting investments back than bailiff, it does add additional cost for me as investor if there is any recovery.
• December 3rd, 2015 – In a reply to support query, Bondora said that their statistics show that DCAs work better than the previous process and this was the reason why these loans were moved to the collection agency. When asking to see said statistics, I was told that they are working to bring this information to investors as well. This is yet to happen as of 25.03.2016.
• First half of March (?) 2016 – Bondora decided to send all loans to DCAs as soon as they are overdue for 7 days. While initially it was possible to see from investment details whether a loan has been sent to DCA or it’s still in-house, then it now has been hidden under a general term “collection”. My earliest “collection” started at 11.03.2016.
• March 23, 2016 – Bondora announced the change to start sending loans to DCAs already after 7 days of overdue for a “more aggressive collection strategy”. This seems to come roughly about a week after the process has actually been implemented into the system. This process seems to be applied for all loans, irregardless if they were issued back in 2010 or will be issued next week.

How much DCAs cost?

More than 9 months have passed since the new process was introduced and one would expect to have some information and statistics on how much the collection fees have been and what’s the success rate for the experiment conducted at investors’ expense.

However, you’d be wrong. As of 25.03.2016., there is still no data provided on this, the fees are still not highlighted anywhere in the UI nor datasets as far as I know.

As a result, besides the fact that we have no idea how effective the collection has been with the DCAs compared to previous process, investors have no idea whatsoever how much fees THEY have paid to DCAs. It’s been over 9 months of a HIDDEN service with a HIDDEN cost.

Bondora HAS mentioned that on average the DCA cost has been around 15%, but that is relatively useless knowledge when you consider that:

• different countries have different fee levels;
• different DCAs have different fees;
• investors have no idea what fees are applied to their specific loans.

Buying and selling ghosts

In addition to previously mentioned problems, this has caused another issue related to Secondary Market transactions with defaulted loans. People are buying and selling claims that don’t exist, while Bondora is misleadingly showing them as if they actually exist.

The easiest way to explain this is with an example.

Let’s say that investor A has a loan of €100 that defaulted and was thus passed on to DCA. DCA will bring in some decent recovery and investor A will receive a repayment of €50. Investor A then puts it to market and sells it with -20% discount to investor B who will pay €40 for it.

Recap:
– Defaulted amount: €100
– Recovered: €50
– Sold principal amount: €50
– Sales price: €40

So investor A ended up with a total loss of €100 – €50 – €40 = €10 and investor B received a claim of €50 for €40 at a -20% price, right? Nope…

So what actually happened here is that borrower didn’t repay €50. Borrower actually paid €50 + DCA fee, which at 15% would be ~€58.8 total. This in turn means that investor A is actually selling a claim of €41.2 and investor B actually bought this defaulted claim of €41.2 at -3% discount, instead of the €50 at -20% discount shown in the UI by Bondora (and probably in income report, return and account value calculations and everywhere else…some additional capital gains and increases for XIRR, anyone?).

Of course, if the DCA fee was one of the higher ones at 35%, then investor B actually paid a mark-up of 73% for this defaulted claim, which is almost double the maximum technically allowed mark-up on the Secondary Market.

The only thing that at least somewhat reduces the atrociousness of the situation, is the fact that this only affected defaulted loans that have had some recovery through the DCAs. So it might not have been too many such transactions. Of course, 9 months is a pretty long time to pile some of those up still and according to Bondora the recovery through DCAs has been rampant.

On a more positive note, since March 2016, Bondora has removed this limitation, so you’ll be sure to meet this situation of ghostly claims even when buying/selling overdue and even current loans in the future.

Good luck!

All investments to DCAs after 7 days overdue

Bondora was again very vague in their announcement of the new process, but after many questions and comments from investors, Pärtel has provided some additional information to understand this under the comments:

While it’s better than the previous vacuum in regards to the relevant info on the topic, it is still somewhat confusing.

Update 25.03.2016 4PM: Based on the additional information published, it seems that collection fees are only applied until loan recovers back to “current” status. In other words, if I understood this correctly, once this 7 day overdue loan pays back the debt, then next payment won’t have the DCA cost anymore. Unless the loan goes back to 7 days overdue again of course. This is good news in terms of the scale of the cost for investors, but won’t affect the profitability calculation.

Update 25.03.2016 11PM: Apparently, if a borrower repays directly to Bondora, ignoring the DCAs request, then DCA won’t be paid. Not sure how that would work, but ok. As a result the following DCA profitability calculations below are assuming 100% of borrowers paying through DCAs. In real life, there could be some borrowers paying directly, but there’s no way for us to find this out, so we’ll assume for now that it’s very unlikely.

Update 25.03.2016 4PM: Pärtel has now claimed that the percentage is instead 58% in Estonia, 66% in Finland and 87% in Spain, not 58% total as previously mentioned in his blog comment. While there’s no information what data is included in these figures, I’d assume it also includes 2014 loans. It is lower  for 2015 loans.

Since the fee mentioned here is 15%, we can assume that investors will pay the same fee for recovery loans that are a week late as for those that have defaulted. Which is already the first alarm signal, since it’s actually a lot easier to recover something from a loan that’s been only a week late and where (according to Pärtel’s comment) 42% of loans actually recover even without any effort from DCAs. Shouldn’t the pricing of the service also reflect this?

Anyway, let’s look at what the success rate for recoveries should be for DCAs to be worth the expense.

Since Bondora has provided the average fee (which again, is actually as useless as it was when we covered this figure in the previous section above) and the rate at which loans default, we can relatively easily calculate how much of this 58% should DCAs recover to cover their costs for investors.

The result is that at minimum, 7,4% should be recovered additionally pre-default to the 42% that recover with already existing procedures to reach the level where additional recovery supposedly pays for itself.

Grab the DCA profitability calculator to run your own calculations

(You’ll be added to a non-active list for the delivery system to work and then deleted. If you want to receive further updates from the blog, subscribe from the end of the post.)

In short, the current situation of 58% defaulting after being 7 days overdue, should be changed to 50,6% defaulting instead. So, Bondora is expecting DCAs to effectively stop roughly additional 13% of the defaults from happening at minimum.

I personally would actually assume there’s a bigger margin for error here before Bondora considered implementing this system, because merely getting to zero would be pretty stupid reason for adding another unpredictable element to the equation. I would imagine that Bondora expects DCA to bring a profit (covering its expenses and bringing back more, so in total investor will actually earn more because of this process).

Shall we then say that the reduction of 7 days overdue to default should be by roughly 19% (down to 39% from 58% of loans defaulting after 7 days overdue) for it to be a meaningful addition to the collection process? This would give investors a 10% added pre-default recovery after DCA fees.

Of course, that’s a simplified calculation where we assume that all loans and all fees are equal and these recoveries would NOT occur anytime soon if loans were passed to DCAs only after defaulting.

According to Bondora’s claims about the efficiency of the DCA process used after loans have defaulted, the latter assumption should be very much incorrect. This means that the difference should be significantly higher than the 19% additional loans recovered to actually bring any benefits compared to the previous system.

What I mean here, is that if a loan with previous system defaulted and started repaying 1 month after default, then the win with new system would be relatively worthless if the new system caused the repayments to simply start coming in a month or two earlier. It definitely wouldn’t be worth paying the success fee for the other 42% of loans.

In short, any recovery coming from the 7 day overdue loans pre-default, can’t be considered purely as additional recovery out of context, without accounting for what the result would have been without this more aggressive collection process. The same way as you can’t really claim that the initial DCA process has increased recoveries, without comparing it to the process without DCAs.

Update 25.03.2016 4PM: Bondora now claims to have run some prior experiments with this process, while even covering the costs themselves, mind you. Really? And this is not mentioned in the argumentation for why this new process is being implemented? Unfortunately the conclusion is very vague without giving even sample sizes and selection (was this 10 loans or 100? was this random loans or only certain criteria? was there a control group or all sent to DCAs?), how long the experiment lasted and when was it done (Bondora already has admitted it was too short period to make real conclusions, but has there been time for the loans to relapse back to 7 day overdue or was this test last week? was there even time for the loans to default?).

There’s also no comparison made in regards to the previous process, which would be needed to actually see whether adding this early intervention provides a better or worse outcome for investors.

If there are some actual analysis of this experiment, Bondora should share the results and methodology in detail, but committing to give away 15% of all loans recovering from 7 days of overdue based on a “wasn’t long enough to conclusive” type of experiment, is not my cup of tea.

Update 25.03.2016 8PM: Previously I compared the outcome of this process directly to what it should amount to as recovery rate. Unfortunately this can’t actually be done this easily because the pre-default recovery only affects a single payment, but after default recovery rate is calculated based on all defaulted loans. In other words, the 19% additional reperforming loans would equal to a recovery where more than 30% of defaulted loans make a monthly payment equal to the payment in their schedule (essentially recover in the sense that they’ll pay the loan according to original schedule, but this can’t be highlighted by simply looking at recovery rate as a whole). From the other aspect the pre-default recovery also is a “softer” recovery, meaning that a loan that starts reperforming may easily stop performing the next month again, making the values not as easily comparable.

Update 25.03.2016 8PM: This also highlights the meaningless of a one time increase in the reperforming rate from 7 days overdue, which one would get by doing a short-term experiment, as highlighted above, to test the effect of this type of interventions, as this tells us nothing about the actual size of the effect if we don’t know how permanent this reperformance actually is. If the borrower goes to 7 days overdue every month or several months before defaulting, this process may easily be less efficient than letting the loans simply default and become fully collectible.

How many loans go 7 days overdue?

So you might be saying now something in the lines of: “Who actually cares? This only affects the loans that are overdue for at least 7 days.”

Let’s look at some stats to see how many loans actually reach the 7 day overdue status during their lifetime.

So it’s not some insignificant tiny portion of loans that are affected, as one may have thought by the way we were notified about this process change. In fact, it seems like more loans will be sent to DCA than not sent.

There’s one issue with that table though. Since it includes all loans, then things might have changed by quite a bit. Especially when we consider that risk-based pricing was introduced in 2015 and the non-EST loans from 2014 were pretty disastrous in several segments.

To clean it up, let’s look at the same figures, but for only loans issued in 2015. It’ll underestimate the number of loans that’ll reach this status, due to the segment being relatively fresh and early in the loan cycle, but at least we’ll remove most of the trashy loans from 2014 from the segment.

Though there has been less than a year on average for these loans to mature, the percentage of loans reaching 7 days overdue status at some point is quite similar to the whole portfolio.

What happens after 7 days of overdue?

Bondora gave us a figure that on average 58% of loans default after going overdue for 7 days. Unfortunately the data available for us isn’t probably as good as the one Bondora uses directly from the source, but we can still do similar analysis.

While the results are better in 2015 than in total portfolio in regards to the percentage that have defaulted, then even with the whole portfolio, Estonian loans were at 48%. In other words, less than half of Estonian loans that reach 7 days overdue, actually have defaulted in the past and the percentage seems to have become smaller or at minimum, there seems to be more pre-default recovery coming in even without the aggressive DCA tactics.

Spanish loans look the worst in this context, with roughly same proportion reaching 7 days as Finnish loans, but a lot more of them actually defaulting.

In general the lesson here is that platform level averages are relatively useless piece of information when there are huge differences between the actual values between countries and DCA fees.

Judging by the proportion of loans defaulting in Spain, it might even make sense to use that more aggressive 7 day process there to see whether it has any beneficial effect or not. Since the recovery without DCAs there is pretty low compared to other countries, it could be a lot easier to reach profitability at even considerably lower success rates.

In case of other countries though, the actual needed outcome just became even more unrealistic. In fact, with 32% of EST loans defaulting, it would be close to impossible to reach profitability after a 15% fee.

A total of 80% pre-default recovery rate would provide roughly zero benefit from the DCA process, so this is already stepping into the realms of science fiction.

P.S. Since we only have one instance of 7 Days Overdue value in loan dataset, then the results in Table 3 actually underestimate the likelihood of a loan recovering from any 7 days overdue instance. I can’t account for the situations where borrower has reached 7 days overdue for several times at different points in time and recovered more than once before defaulting or continuing payments according to schedule until maturity. All of these cases will now be taxed with DCA fees.

Which loans default, which recover?

As we saw before and as you may have already guessed, I don’t really like averages that much. They too easily hide the actual story behind that average. Thus it shouldn’t come as a too much of a surprise that I have created this table for Ratings as well.

While we can see a pretty clear pattern for Estonian loans where lower percentage of loans reach 7 days overdue status in the lower risk Ratings and it increases as the risk Rating increases, then this pattern pretty much doesn’t exist for Spanish loans and is not that robust for Finnish loans.

I’m guessing this has to do with the fact that previous Rating model was too optimistic when estimating the risk levels of Spanish and Finnish loans in 2015.

There’s a pretty clear pattern that higher risk loans that reach 7 days overdue, also default more likely than lower risk loans. It’s also clear that there are obviously more higher risk loans defaulting in general, which is expected.

In short, what this means, is that if you are investing at a lower risk level, you will be penalized more by this new DCA system than the investors who are investing into higher risk loans.

This also means that if it was unrealistic to reach profitability on the whole segment level, then it’s even more unrealistic for lower risk loans.

At the same time, it might be somewhat more realistic for higher risk loans, at least by simply numbers, if we assume that higher risk loans recover at same pace as lower risk ones (which may be a possibility).

Of course, for EST and FIN the rates aren’t higher than the 58% mentioned by Bondora, so not too much hope for profitability in either of those countries in any segments if you ask me.

If we also consider that according to Bondora, the Rating V2.0 model was updated to eliminate a considerable amount of the first payment default loans, which Spain is so famous for, then I wouldn’t be too certain that even in Spain this process can be profitable with these conditions going forward.

Conclusion

Here are some of the key takeaways:

• By adding this new aggressive collection process, Bondora has changed the value of your investments and your potential return without notifying you before implementing this change. This is also affecting the investments you have made previously with the assumption that the return will be better (including purchases from Secondary Market).
• Bondora is continuing to hide the actual DCA costs and success rates more than 9 months after the DCA process was introduced.
• Bondora has knowingly led investors to buy and sell claims for months already that don’t exist or are actually in different amounts and mark-ups/discounts than misleadingly shown by Bondora in the UI and other reporting. This is leading to direct monetary losses for the parties making the purchases.
• Before 26.11.2015, Bondora has also likely claimed 1.5% commission on these transactions based on the incorrect sales price from both, the seller and buyer.
• Roughly half of all loans issued in 2015 have gone overdue for 7 days at some point already with likely more to do so in the coming years.
• Most loans, with the exception of Spanish ones, recover fully from being 7 days overdue or make at least partial payments before defaulting.
• Even at the default proportion of 58% of all 7 day overdue loans given by Bondora, the recovery from default would have to be insanely high for it to be beneficial or even cover the expenses related to it. With the rates we’re seeing for 2015 loans (with the exception of Spain), it may be impossible even if we pretend to be in a science fiction movie.
• The negative effect from this new more aggressive collection process will affect most the risk averse investors who have been investing into lower risk loans. Higher risk loans will be less affected, but in most cases it is a losing deal even for these loans at the 15% success fee level.

In even shorter conclusion:

• Bondora has one-sidedly implemented a new process, which will significantly reduce the return for more risk averse investors and will reduce the return in majority of higher risk cases as well.

The process seems to be adding a significant risk, a practically guaranteed additional loss at 15% fee level for investors, without much room for actual benefit even in case of an extremely optimistic scenario.

As an investor, I can only conclude that Bondora is causing me direct monetary loss with this new process to provide monetary gains for the DCAs, without providing me any additional benefit.

For this reason I find this new service to be directly harmful to Bondora’s investors and violating investors’ rights, by automatically deducting money from their account for a service that’s not in their best interests, that they don’t want, that was implemented in secrecy prior to any announcements or updates to Terms of Use (which has been updated on 22.03.2016 after the new process was already implemented) and that will almost certainly cause financial loss for the investors based on the information available.

On a more positive final note though, it is now possible to withdraw without the withdrawal fee, so you don’t really have to wait for a large amount to be collected on your account to start voting with your money.

I tried to take a look at the topic from a different perspective a well. Make sure to check out the article here.