Summary of our Bondora visit

A couple of weeks ago some Estonian bloggers got invited to Bondora to talk a bit about future plans, to get a chance to ask our questions and discuss overall better communication. I’m going to write down a slightly chaotic list based on the notes I made, please notice that this is what I wrote down and understood, any errors are wholly mine.

realestate

Company restructuring

The business has been seriously reorganized to separate people who work with borrowers and people who work with investors. This is largely due to how little overlap there is with the products, meaning in the long run better service and quality for both sides.

Rework of investor communications

Pärtel was honest in admitting that investor relations have not been working at the level they should have, and complaints have been justified. There is a whole new investor relations team, and I have already had some experience with them, and I’ve seen far more professional answers and contact than from general customer service before.

Rework of business logic and other details

2015 was a year of big change, including both reworking the models for rating clients and reworking a lot of the legal framework. The legal changes mean that now an investor essentially purchases a claim for a piece of a loan, but isn’t directly tied to the loan taker – all loans for a moment go through Bondora’s records, but legally it’s set so that even in case of Bondora’s bankruptcy investors keep their claims. The legal framework has also been set to include institutional investors.

Banking license in the future

Due to practical reasons of trying to enter new markets, a banking license is inevitable due to how much easier it makes accessing new markets. It would also allow Bondora to include more products for people, for example a set interest CD that would be backed by the loan book. Bondora itself investing partially into the loans was also mentioned. It would definitely allow for more flexibility in terms of products and new markets.

But active retail investors?

Retail investors are here, and will remain. Active ones who manage their own portfolio are however a rarity and there just isn’t enough money in their hands to support expansions into new countries (this is already known based off all other big P2P sites). This means that institutional investors will have to come on board to support expansion.

Institutional investors

Pärtel mentioned that the “ballpark” for institutionals is that they’d be able to provide something in the range of 100 million per year to help keep up loan supply. It might seem like much, but expansion plans that are going around are mentioning 8 new countries by 2017. Institutional investors are also more interested in Spanish loans due to higher returns. (Retail investors are always scared that they’re going to lose access to EST loans, but that’s too tiny of a market for any institutional investor.)

Bondora rating

Overall the rating has proven itself. There was a tiny update in December to improve the current model. We were shown a lot of charts & data and overall conclusion is that the rating has failed to predict well for Spanish B, D, C, however those combined are at 2% of the portfolio, and the new model was set to fix it. Risk/recovery are currently on plan.

Recovery

There will be open data in the near future about how the DCA (debt collection agency) have done in recovering, but Pärtel said that the data is better than expected. The reason why they use DCA at start instead of courts is that 1) often people fear DCA’s more, 2) courts take a very long time to act 3) courts are inconsistent in their rulings. Hope to see a blog post by them about this at some point.

Buyback

One of the bloggers, Jaak, raised the topic of a buyback guarantee and allowing investors to accept the loss, if they so desired from defaulted loans. Pärtel stated being very much against it due to the black box that it creates. (I agree on that). An investor can accept the loss when selling on the secondary market though, since now that the loans are legally restructured you can sell defaulted loans as well, though of course the discount has to be rather steep.

Secondary market

Pärtel also briefly introduced plans to work on secondary market pricing, to include a sort of a “fair price”, allowing people to assess sm loans better, and eventually use the autobidder to buy and sell on the sm. This is probably very far in the future – you can still manually price loans, but if they fall outside of the limits of the fair price that Bondora estimates then people need to manually purchase them.

Primary and secondary market

Are not going anywhere, but will also not be receiving any significant upgrades or additions. Currently no plans to phase them out.

Sharing data

One of the topics we discussed was official communication channels. The Bondora forum exists but is rather useless, it’s used by ~100 people, the loudest of whom often no longer investors, meaning that new users get no useful info there, since Bondora doesn’t actively track it. The current official communication channel is the blog that’s been getting a lot more action recently. We also passed along the suggestion for regular Q&A sessions.

API

First third party applications are already available and in use. Looking at other P2P portals, the pricing of the API tends to  come to 0,2-0,5%. They are following the market to see what’s happening. We briefly discussed the issued of API bids being last in line to make a bid and that this might be a bigger problem in the future. The issue is understood, and can be tracked better now that the old PMs are closed, it will create a clearer picture for how well loans fill. Also, in the future it would be possible (once more institutional investors are on board) that a set % of a loan is always set aside for retail investors.

Future fixes

Essentially what we’ve been waiting for forever – more data & charts on the web (I’ve sent them my wishes like 3x I think). There should be a fix for the cash flow page finally as well in addition to overall quality of life fixes. (The first – bid amount per loan for autobidder users is already live.) Looking forward to seeing all the fixes (hopefully in the near future).

Summary

Overall, while I complain a fair bit about some of the decisions and communication blackouts that Bondora has had (and that complaining has been justified) then I have continuously invested into Bondora due to the returns offered.

I am happy that they are finally starting to work on many issues that have been problematic for a long time (years). Of course, as usual – with many things I’ll have to see them to believe them, but even the fact that we were invited to meet shows a level of openness we haven’t seen before.

Overall, hoping for many positive changes in 2016 both in terms of web use comfort, better access to data, better communication and overall growth of the company.

21 thoughts on “Summary of our Bondora visit

  1. “Pärtel stated being very much against it due to the black box that it creates. (I agree on that)”

    Could You explain this sentence a bit more? What is black box?

    1. “Black box” refers to the problem that you don’t know what is going on “inside” something (in this case the buyback mechanic – what happens when they buy the defaults back), making it much more risky. Example being – Trustbuddy guaranteed a 12% return while not letting investors know what was happening inside, which ended up with a lost 5 million euros.
      If you have to guarantee a certain % of money for buyback then the question becomes – 1) where does that money come from 2) what happens in a crisis when defaults spike 3) if it’s profitable to buy the loans back at x%, then it would be just fine for investors to keep them and get the returns in the long run.

  2. Thanks for the coverage and for doing your part in keeping Bondora on their toes!

    Despite all the frustration and lost trust, I too hope they can find long term, sustainable success without having to sacrifice individual investors.

  3. “Pärtel stated being very much against it due to the black box that it creates. (I agree on that)””

    “Trustbuddy guaranteed a 12% return while not letting investors know what was happening inside, which ended up with a lost 5 million euros.”

    I may add my (sarcastic) comment: Does Pärtel’s and your view also apply to Bondora’s Portfolio Manager where you have three slightly different options to invest in multiple risk level and multiple countries and have to rely on Bondora’s credit rating as well as – to be proven- recovery performance for non-estonian countries?

    I mean come on this is funny.

    I am also very amused that institutional investors seek spanish loans. Either private investors are so risk averse and freak out on every default and can not understand the benefit in spanish loans or this is just a BIG FAT LIE. How can an insitutional investor want spanish loans. Look on the default rates for spanish loans: http://www.bilder-upload.eu/show.php?file=f56575-1449519455.png

    57% of all spanish loan volume is still in default for 2013
    50% for 2014
    and for all loans up to May 2015 it is already 42%!!!!

    I would be happy if Bondora or those insitutional investors bought up all spanish loans at par and they can reap the interest until maturity of these loans. This would proof that they believe it what they do. I guess that this will never happen.

    I can also remember Pärtel’s roadmap (late back in 2012/2013) to add 6 countries per quarter. And he also stated in an interview that he wanted to expand to USA in 2015. Non of this has happend. Now it its 7 countries until 2017. I understand that plans may change but Bondora’s keeps promising on multiple occasions and does not deliver.

    1. Hey! I do not like the autobidder personally, I have stated that multiple times, and do my own bidding via API. However, I do speak to enough investors to have a good understanding of how many people simply do not care or want to spend the time working on the strategy – the amount of people who have actually turned on the autobidder is living proof of that. Whether or not the bidders are properly balanced, that’s a whole different discussion topic that needs a lot of statistical work.

      I’m honestly a bit unsure about the topic of recovery. From what I see in my own portfolio – recovery is happening. Slowly, but it will. Also, even if I fully charge off all defaulted loans my portfolio still isn’t a disaster.
      I have seen many different reports and analytical posts that different bloggers have done, and the biggest problem with all of them – they are getting different results. This is one of the reasons I hope that Bondora will start giving out more data to start clearing up all the myths and rumours (or alternatively verify the facts). I find it hard to believe that such a huge business with such international backers that has both internal and external audits running is simply lying – in addition to all the moral problems that would actually be a criminal violation by Estonian law.

      As far as expansion goes – I’m happy that they decided to clear up they back rooms before expanding. Adding new countries on top of non-working infrastructure and non-working communication systems would’ve been a recipe for disaster. They have expanded to include a lot of investors from different countries, but as far as not adding additional markets – I’m happy that they’re working on the credit rating fixes instead.

      1. Hey Kristi, you are right that it is a point for Bondora that many investors turned the autobidder on. They may believe what they see on the site to be their RoI. I (and you may also) know that Bondora’s calculation is tricky and the rules of proper accounting (Transparancy!!) are bent.

        I did my own analysis (see the link above) and my findings basically are the same with the analysis done by Taavi http://rahafoorum.ee/en/performance-of-bondora-rating/ –> Non-Estonian Loans are a more or less big in trouble.

        What other analysis have you seen ? Care to share the link?

        1. I’m hoping for more recovery data from Bondora, since Pärtel stated that DCA results were better than expected. I haven’t dug that deep into Taavi’s report, however I do know that Bondora sent a very long letter to him asking to correct some things and explaining their calculations. Seems like there is bad blood there :/
          I do think that you shouldn’t rely on the on-site XIRR, I also calculate it manually, but I see the point of showing it – at start predicting losses and recovery is pretty difficult to do in a way that doesn’t paint a very depressing picture.
          I’ve seen some links in the Bondora Facebook group & Bondora forums, I assume you’ve seen the same things that I have.

          1. I did send the analysis to Bondora before publishing it with hopes of finding out if there are some errors in my calculation, data or logic somewhere, because the results seemed quite pessimistic.

            I also received a long e-mail with 12 points from Bondora.

            All the relevant points were taken into account as much as possible. Some of them were unclear, but Bondora refused to clarify on what they meant by those exactly so perhaps those were not fully accounted for.

            The analysis was also reviewed by some people who know a thing or two about statistics/data analysis and investing and their feedback also taken into account.

            Additionally, all the analysis process is described in the post and all the data files are available for anyone to double-check the results or point out any specific step in the analysis that you think is incorrect.

            I urge people to do it and let me know how it went. I’ll gladly fix mistakes if there are any left in there and clarify about the process used if any questions should arise. Everyone’s welcome to download the dataset and comment on the post.

            What I didn’t do, is follow Bondora’s “request” to add these 12 points as the most prominent and first thing in my blog post in the top with an image of XIRR results which had nothing to do with my analysis or the performance of Bondora Rating and was somehow missing Slovakia from it entirely. I’m guessing it’s the same table that was shown to Kristi.

            All the relevant feedback that has reached me, has been taken into account and I will gladly accept and consider any additional relevant feedback that Bondora or anyone else might have about the analysis. Comments are open under the post.

    2. As much as I appreciate Taavi’s efforts and time put into the calculations, I do believe that to actually prove something, one would need to use proper statistical tests and p-values etc, not just look at the comparisons. Some groups seem to have very low sample sizes and the higher risk groups are prone to large variance and deviations from the average anyway, hence stats is key here.

      1. Yes, some groups have low sample sizes. Feel free to sum them up together within a country, or simply look at the row of country totals. In general though, a mathematical rule of thumb is that if you sum up two or more negative numbers, the outcome is still a negative number. In other words, even if you sum them all up together and only look at the countries as a whole, the conclusion is still the same.

        I also agree that you would expect some variance from the result. It’s obvious that the result will not be exactly same as the EL% shown in the Rating. They would not be the same anyway, as the loans are not even close to fully matured and thus you’d expect practically all of the outcomes to be better than expected figures at the time of the analysis.

        Part of variance also means that the numbers could be skewed in one direction or the other. In this instance there’s not much happening in the other direction.

        Also, Bondora Rating includes this variance in the pricing as unexpected loss and country risk. In other words, that variance is partly the reason why interest rates are set higher than they would with simply expected loss.

        However, if you look at the results in Table 5 in the total figure for all non-EST loans, you will see that with the default rates at that point in time, the loss one could expect after the assumed recovery rate, was already 23% higher than expected initially. Considering that one of the highest possible Expected Return (interest rate – Expected Loss) is around 20-22%, you can make some conclusions about the outcome and whether the expected loss and the variance has been accounted for accurately or not.

        As for the higher risk and variance comment, weirdly enough, with this type of loan product, the variance is actually supposedly reduced as the risk level increases. While it sounds a bit counter-intuitive, then it kind of makes sense I guess, when you consider that 50% is already going to default and you have accounted for this in your pricing, so you have a lot less left that could perform worse than expected compared to a loan segment where only 2% is expected to default.

        The latter is of course assuming that a proper diversification level has been reached as in when analyzing a large enough number of loans together. I think you would still expect more variance when picking only a few loans from the risky segment compared to few from lower risk ones, but I could be wrong on this one.

  4. Thanks for sharing this.

    I really hope this additional data and the quality of life updates happen ASAP :p

    One question for you: Did Pärtel or anyone else give any specifics when talking about new countries? e.g. which countries, EU/non-EU, any plans for Latvia/ Lithuania/ Scandinavia?

    Cheers

    1. Sadly no hints. However, I’d guess Latvia and Lithuania would be out (one too small & enough competition, the other has legal issues with P2P). I’d guess bigger markets – Poland? Has a currency issue though. Same problem with Sweden.

  5. Such a disapointment, being forced to use a 3rd party company and pay for a service which was free before…Is bondora still different from other finance companies ?

    1. While I do understand the disappointment – depends on how the API developers monetize – was this a surprise? Growth means more regulation, stricter system, access to institutional investors and that includes third party applications. Sadly, yes, this plays into the hands of people who have more money.

  6. Interesting. I like Pärtel’s thinking about buyback guarantees, they’ve always been a no-go for me. I don’t want to buy an insurance that covers up my reals risk.

    I hope they stay committed to openly providing all data about the portfolio (and risk class) performance.

    But as we are at it… what I find fishy is the comment about institutional investors buying specifically Spanish loans. Bondora’s own data shows that Spanish loans overall are making a loss, I don’t see an institutional investor knowingly invest into a lossy asset class. They seem to be confident that the performance will improve, which is obviously good, do they provide any data supporting that?

    1. From what I understand then the Spanish ratings are the ones that got fixed the most. I am hoping for recovery info soon, I assume that is where their optimism is coming from.
      Also, for institutional investors the issue is scale – there just isn’t enough Estonian loans (or even Finnish tbh) to bother with them, hence why they keep trying to work on Spain I suppose.

      1. I fully understand the scale issue but even as an institutional you don’t want to invest into a lossy asset class, not even at scale :) Unless you are in Catch 22….

        So let’s hope they show us better ratings soon.

        What disappointed me is that they discontinued to report the performance of the score classes in the weekly reports when things started to go south, that was an extremely bad sign but at least they still report the raw data, although I personally don’t like to compile statistics myself, I think transparency is the job of the platform.

        I also believe it’s a fallacy to think that covering up risks somehow helps. For me, I’m not scared of risk but if a platform lacks transparency I am out. I know other investors think differently and prefer “guarantees” but I think that’s just lossbait, you can build a scam on these investors but not a long-term business.

        1. There’s nothing wrong inherently with the guarantees per se. Essentially it’s just a way to optimize taxes and earn a higher return in the process at same risk level, if done correctly.

          For example, if Bondora’s return for a group of loans was 20% at a 50% interest rate, I would easily take a guarantee of 15% as a private individual investor since that would end up as a much higher return after taxes. I will not get this same discount on taxes by selling these loans on secondary market though, as suggested in the post.

          Problem with those guarantees is that you would need totally different information to actually see whether those guarantees make sense or not and whether they’re plausible in the long-term.

          None of the platforms offering such buy-back guarantees have published such data as far as I know, so currently they are black boxes in the sense that you have no idea how much buffer there actually is and how much additional stress will collapse that buy-back system.

          Essentially it’s the same thing that Bondora has mentioned before with their Bond type offering. Not sure how the “black box” argument given by Pärtel would be different in that case though.

          1. There is a second issue with a buyback guarantee since it changes the risk profile.

            Three cases:
            1. Best case: it’s an insurance and the guarantee is given by a (hopefully solvent) insurance company. That’s the best case and indeed, the tax gains could offset the costs.

            2. It’s an informal guarantee. Here, I still have tax optimizations but I might lose the guarantee if scores turn out to be too optimistic which – frankly – they have been for almost every p2p credit platform I have seen so far, including Bondora (Zopa being a noteworthy exception).

            3. It’s a formal guarantee. Then I create a risk concentration at the issuer which amplifies any issues arising from any problems with the risk model. I essentially drive the platform towards a high risk of insolvency which is actually the last thing I want to see when there is trouble collecting loans.

            So a formal guarantee not given by a third party requires a rock solid risk model and I haven’t seen a lot of people in this business having one.

            One of the good things in p2p lending is that systemic platform risk is relatively low. Sure, this comes at the price of a moral hazard that has indeed be seen on most platforms but at least the platform has the chance to survive and continue servicing loans, even if the risk model was initially wrong.

            I think if you really want to address the tax issue you either need a third party for the guarantee or you need to do securitization of loan bundles (sell shares of a portfolio which is big enough to average out losses, ideally a portfolio of loans in the same score class and ideally without lenders selecting the loans to avoid moral hazard on the lender side).

    1. I started a company portfolio for both Omaraha and Bondora.
      I will reinvest some of the earnings from Bondora (20%) and cash out some (80%) to move that money into stocks. (more tax efficient as a private investor)

Comments are closed.