The coronavirus crisis has created a high level of economic uncertainty. As a result, lenders in general are less willing to provide credit. In fear of a credit crunch and eventually the bankruptcy of SMEs, governments are reducing the risk for lenders with state guarantees. With a state guarantee the government covers part of the loss in case a loan defaults. In other words, if a company cannot repay the loan, the state will partially repay on behalf of the company. As such governments make it saferfor lenders to keep lending in this crisis.
October can distribute state guaranteed loans in France, Italy and The Netherlands. In order to increase efficiency and distribute more loans, October is digitizing its credit analysis further for projects under €250,000 and covered by a state guarantee. In this article we explain how we have been making a shift to data-driven credit analysis which enables us to take faster decision: will this company get a loan or not?
Our current credit analysis
Our credit analysis is performed in-house by our local credit teams. Our financial analysts are in charge of collecting the quantitative and qualitative information necessary to assess the risk of a project. For example, they get information from annual reports and insert it in our scoring model. They would fill the scoring model, automatically or manually, with a lot more information about a company’s financial results, market outlook and management team. The scoring model generates a credit score and an interest rate that would be offered. This is what we call underwriting.
Underwriting is the largest part in any credit analysis, but not the only part. Basically our process can be summarized in 5 steps:
- Borrower or financial intermediary filing an application online.
- Analysis and (online) meeting with borrower by October’s analysts.
- Loan offer generation, underwriting and compliance by October’s analysts.
- Approval by October’s international Credit Committee.
- Fraud prevention by October’s operations team.
Discover all the details behind this 5 steps in our tutorial
Introducing Magpie
Over the last years we have developed an automatic borrower loan eligibility assessment model. We call it “Magpie”. Magpie is built using machine learning algorithms on large amounts of data collected by October over the past 5 years across different countries. It is constantly improving as we collect more data and apply cross-country learnings. This algorithm has been developed by our Data and Tech teams.
Magpie estimates the probability of payment default (PD) of a company and scores it from a scale going from 1 to 5. A Magpie score of 5 means that the probability of default is very high. Normally, Magpie score 1 equals a credit score of B and Magpie score 2 is a C, while others are not eligible for a loan and thus do not get a grade.
How does Magpie work?
Let’s assume you are the CEO of a company and request €100,000 for expanding your delivery channel. What will happen?
1. First you will test your eligibility to an October loan with our Quickscan, where we ask you to enter your company name, loan amount requested and the reason for your request.
2. Based on the company name:
- Magpie gathers data from public and private data providers, such as the chamber of commerce, to check if your company matches our eligibility criteria: at least €250,000 turnover, profitable and 2 years in existence.
- Magpie automatically analyses documents submitted by you, such as bank statements and annual reports.
- Magpie compares your company and request to similar companies, in terms of size, activity and location.
3. Magpie predicts your capability to repay the loan, i.e. your probability of default.
When your company gets a Magpie score of 1, 2 or 3, one of October’s analysts will take over. They will gather extra information, conduct an interview with a representative of your company and do the rest of the underwriting as explained above.
Our analysts have already been basing their work on the Magpie scores since 2019 in France and January 2020 for Italy.
Comparison between the credit analyses:
| Step | Standard Projects | Instant Projects |
| 1 | Borrower or financial intermediary filing an application online. | Borrower or financial intermediary filing an application online. |
| 2 | Analysis and (online) meeting with borrower by October’s analyst. | Analysis and loan offer generation by Magpie. |
| 3 | Loan offer generation and underwriting. | Compliance checks by October’s analysts. |
| 4 | Compliance checks by October’s analysts. | Approval by October’s international credit committee. |
| 5 | Approval by October’s international credit committee. | Fraud prevention by October’s operation team. |
| 6 | Fraud prevention by October’s operation team. |
With the new process we’re able to save a significant amount of time in steps 2 through 4. By omitting human credit analysis, we are able to provide a credit decision within minutes. We estimate that the process from applying for a loan, to the payout of the loan will be 20% quicker.
Why are we introducing Instant Projects?
The longer and more complex it is to get financed, the less time companies can work on their projects. We want to offer companies a product that is fast and simple. We want to give them time to concentrate on the work that really matters. Within a matter of minutes a company will know whether they can get an instant loan and at what costs.
At the same time, we will be able to do more loans at lower costs. The scalability makes it worthwhile to do smaller projects and frees personnel to work on larger and complex loan applications. We will use human capital more effectively and use our technology to work efficiently.
We have been working towards digitizing and streamlining our credit analysis for a while. Magpie has replaced our eligibility test for a year in France and few months in Italy and has been performing well.
On top of that, the current environment provides a unique opportunity to introduce our new credit process. State guarantees in France, Italy and The Netherlands cover a large part of the risk. By making faster credit decisions we’re able to distribute more loans and help more business in this crisis than we would have been without technology.
Consequences for lenders
The Instant projects will not replace the traditional projects. There will be 2 kinds of projects on the platform: instant and non-instant. You can recognize an Instant project by the project description. The project description of Instant projects is standardized and will contain less information, because this information is normally written by our financial analysts. Instant projects will always have a credit score of B or C, depending on their Magpie score.
These Instant projects will be co-financed by retail lenders and the October fund, containing institutional investors and the management of October, like any other project.
The goal of Instant projects is also to distribute more loans. As a lender you will have more investment opportunities on the platform. You can use these to diversify your investment as much as possible.
For now only French Instant projects covered by a State Guarantee will be open to retail lenders.