Our Credit Director about the October track record

4 years after the launch of the platform, many of you are asking us questions about our default rate. We asked the most representative questions to Marc Sebag, our European Credit Director, in a very open interview. With his answers, Marc also shares insights into the improvements we have made in our risk scoring model and on our recovery methods.

Some context: how our default statistics are calculated & insights into our approach

The default statistics are available on our website through this link and are updated every month.

In case of default we apply a provision to the outstanding capital of the project. A larger provision is applied on the outstanding capital with the number of days that a payment is late (the longer the borrower is in default, the less likely to repay). Capital reserves in case of default are done as follows:

  • Loans which are less than 30 days late: 40% haircut,
  • Loans which are between 30 and 120 days late: 80% haircut,
  • Loans which are subject to a legal recovery procedure or more than 120 days late: 100% haircut.

Default provisions also affect your internal rate of return post-default. Provisions affect your gains on your Portfolio page: we consider the reserve of capital as a potential loss. Please keep in mind that it is a provision, even in case of default it might still be possible to recover the amount due by the company, partially or even totally.

We think that taking provisions as soon as a loan is late is the best way to be transparent about the health of your portfolio at any given time. 

Question: why did the default rate increase in 2018 and 2019? 

Marc: It is normal, after 48 months of activity, that the portfolio experiences defaults as it matures. The question is whether this is happening as expected or not. There are 4 main drivers of unexpected defaults:

  • Fraud: 7 projects are affected by fraud, for a total amount of € 5.9M.
  • Acquisition finance: 18 projects in default share the same goal, the financing of the acquisition of business assets or a company. These projects account for a total outstanding amount of €6,1M.
  • Industry risk: we have identified 3 industries with a higher risk than expected:
    • The construction industry (8 projects/ €1,6M),
    • Bakeries (3 projects/ €1,1M),
    • Catering (2 projects/ €0,6M).
  • Geographic risk: we have identified specific regions with a higher risk than expected: 5 projects are in default for a total outstanding amount of €0,5M.

I would also like to say that lending to SMEs is not without risks and we want you to be aware of this. Therefore, we repeat again and again that diversification and equiponderation of your portfolio are critical.

Question: what do you qualify as fraud? 

Marc: We speak of fraud when the borrower (usually the manager, but the CFO and the auditor can also be involved) misrepresented the truth in the contract signed with October. A misrepresentation can involve: wrongful or hidden information about the health of the company, false declaration, internal fraud (money laundering) or false or altered financial statements intentionally provided.

In general, fraud affects not just October lenders, but all the borrower’s stakeholders (banks, creditors, clients, suppliers, employees, etc.). In such a situation, almost systematically, the funding is not used for the intended project and thus not generating the revenues planned in our analysis. As a result missed repayments occur very early.

97% of the amount in default of A projects is linked to 1 fraudulent project. Without this project the default rate is 0,22% (in €). It explains why the default rate for A projects is higher than for B projects.

Question: what are the October recovery methods when we speak of fraud? 

Marc: Out of the 7 fraudulent projects:

  • We have started criminal proceedings for 4 of them. For 3 projects, we have already managed to fully recover the outstanding capital for the lenders. For 2 of this 3 projects the managers are in jail.
  • We have recovered 1 project by amicable means (without having to go as far as criminal proceedings).
  • For the 2 remaining we are still working with our lawyers.

Question: do you see any sectors that have a higher default rate?

Marc: There are 3 sectors that we have identified with a higher risk and I will explain why:

  • The construction industry: it is a cyclical sector. Big companies are better equipped to manage the invoicing of their clients. Smaller companies are totally focussed on their day-to-day business and forget to invoice their clients, which leads to liquidity stress.
  • The bakery and the catering sectors: customers are faithful to the management of business from this sector (we talk about clientele intuitu personae). In the case of acquisition financing the amount of customers could be misrepresented.

It is worth noting that these industries are riskier in the general population as well, construction companies have a higher default rate with traditional finance providers too. You can find the official statistics of the Banque de France here. We took that into account in our analysis; but what we have learned is that the risk of these sectors in the population which looks for financing on a platform is even more pronounced.

Question: do you see any type of financing with a greater default rate? 

Marc: As mentioned in question 1, acquisition finance. Why did these projects go into default?

  • Operational risk: what we call operational risk, is the risk ofaccidents. For example non-performing equipment, a flood, a fire, decrease in traffic due to protests (strike and yellow jackets), etc. These are unexpected. They are not and cannot be ‘calculated’ in the risk score given to the borrower at the moment of the analysis, because they cannot be predicted. 11 projects have been concerned by this for a total amount of € 2,5M.
  • Misrepresentation or deceit from the seller: for 6 projects for a total amount of €3,6M a conflict between the selling party and the buyer led to cashflows difficulties for the company. These conflicts are systematically due to a misrepresentation of the assets bought: fiscal and social debts hidden by the seller, the sale of a client portfolio while the seller is still handling the clients with another business entity, social tensions within the company, etc.

Question: is there any correlation between geography and default? 

Marc: there are several higher risk regions for us where:

  • The payment behaviors of small companies is uncertain,
  • There are long payment deadlines,
  • The recovery process is very difficult.

Question: do you plan to implement guarantees on borrowers or offer insurance to lenders? 

Marc: We are already taking securities on some projects (and did this since the beginning of October) such as blockage of shareholder loans, first demand guarantee and subordination of convertible bonds.

Since April 2018, the Italian State Guarantee Fund (Fondo di Garanzia) applies to some of the Italian October Projects. In case of default, the guarantee covers at least 40% of the outstanding amount of the loan. It offers some protection against the risk of the loan, helping SMEs access credit and decreasing the risk profile for lenders. You can read more in this tutorial.

Since September 2018 we have decided to systematically take guarantees for acquisition financing such as pledges on shares or ongoing concern, in order to align the interests of the lenders and the borrower. When this is the case, it is specified in the lender loan contract and we have decided to make it also visible in the project description.

We are not currently working on offering a type of insurance to lenders for 2 reasons:

  • Our past experience has led us to believe that it cannot be easily activated, because the eligibility criteria are very strict. In France, in 2016, we acquired the business assets of a competitor who had put in place an insurance on part of their loan portfolio. When managing their workout portfolio we were not able to activate this insurance for companies that were covered, not even once. Why? The insurance contract was built in such a way that it rendered improbable to trigger the guarantee.
  • It is very costly.

Question: what are the correcting measures you took on the risk selection? 

Marc: on acquisition financing, along with the guarantees, we have developed additional internal guidelines since September 2018. Our credit analysis takes into account:

  • New governance risk with zero year of history: if the buyer is not part of the company he/she is taking over, the analyst downgrades the rating. This applies even if the buyer has a long experience in the industry. Taking over a new business can be compared to a new start.
  • Higher repayment capacity to absorb the operation risk: the minimum repayment capacity of the borrowers willing to acquire a new business / asset must be higher of at least 20% that the ratio we usually ask to our clients. We want to deal with companies profitable enough to absorb post acquisition surprises. 

We have implemented stricter measures on the sectors mentioned earlier. 

  • In the construction industry: we only deal with B-rated companies that have a minimum turnover of several million euros because we believe bigger companies manage their clients more efficiently. We have also set a maximum duration of 3 years in order to mitigate the cyclicality of the sector.
  • The eligibility criteria for the financing acquisitions of bakeries and catering companies are stricter than before (on the turnover mainly, on rentability margin and the level of assets).

On geographies, we have taken comparable corrective measures:

  • The size of the company (in terms of number of employees, turnover, level of assets) must reach a certain minimum.
  • We have stopped lending to companies belonging to certain industries in these regions.

On fraud, we regularly conduct internal workshops, to identify patterns in order to be less exposed. We also collaborate actively with government efforts to make information sharing on fraud efficient, while respecting potential borrowers’ right to privacy. It is difficult to share these results as it would give ideas to the fraudsters on the criteria we deep dive.  Read more about our risk selection and what October already does to flag fraud

Question: what is the outcome of these correcting measures? 

We have back tested the portfolio with our corrective actions and taking out the fraudulent projects. The default rate goes from 7,49% to 2%. What does backtest mean? It means we have applied our new selection criteria to the portfolio to see if we would have financed the projects or not.

Question: how far will October go to recover our money?

Marc: I personally (along with the management of October) lend automatically and systematically to all projects. The management has committed €8.85M in total for all the October Funds. Therefore we are fully aligned with lenders interests.

We never give up, this is not an option. Especially in fraud cases. We go to court and engage judicial fees at our own cost. If we are not able to recover the capital, these costs are for October. They are never charged to the lenders. Some lenders have asked me why –  because I want my money back as much as you do and because I want to give a clear message to future fraudsters: we will chase you. If you want to read more about our collection methods you can read this tutorial.

Question: what is the impact of the default on the yield of the October portfolio? 

Marc: You can always check the IRR post-default on our statistic page as well:

You can see that, for a portfolio that would have lent the same amount to all projects ever published on October, the IRR post-default is of 4.59% (vs. an IRR before default of 5.99%). The difference is of -1.40% between the initial IRR and post-default IRR (which remains positive).

There are 2 conclusions out of this table:

  • Again diversification and equiponderation (lending the same amount to all projects) are key,
  • Though the default rate has reached an unexpected level the net IRR of the global October portfolio remains largely positive.