This article describes Quantexa's new SME (Small or Medium Sized Enterprise) classifier. Our classifier has been informed by European Commission rules. It has been implemented to be able to scale for large legal hierarchies.
Introduction
The world we know today is built on the premise of successful organisations and businesses. At a fundamental level, organisations are a community of like-minded people working and contributing towards a similar goal, and this remains true across the business landscape. Despite this, organisations that share this lifeblood often differ in a variety of ways; the investment management arm of a multinational bank is categorically different from the self-employed tree surgeon. Although these organisations differ in many ways, it is the difference in their size and scale that is perhaps the most obvious. Programmatically discerning these differences at scale will form the focus of this article.
The size of an organisation has a profound impact on the manner of its operations. Large organisations often have access to larger pools of capital, can more easily expand operations, and are generally more resilient to periods of economic instability. Smaller businesses (Small or Medium Sized Enterprises; SMEs) are traditionally deficient in these pillars, and we only need look towards recent ‘relief packages’ offered to SMEs (globally) during the Covid-19 pandemic as evidence of these challenges.
SMEs serve a major role in society as they account for the majority of businesses worldwide. They make up 99% of all enterprises in the EU’s business economy and two thirds of its private sector jobs [1]. In addition, they make up 50% of the total revenue generated by U.K. businesses [2]. The importance of SMEs within an economy therefore cannot be understated, and understanding inter-business relations with SMEs is a continued area of interest for organisations in every sector.
As well as being highly significant at the societal level, SMEs and their business interests are of great relevance across a number of different industries. Some examples include:
- Credit Risk: Understanding capital allocation and where capital relief could be applied for financial instruments; Accurately identifying and classifying the right industry for better risk modelling.
- Customer Intelligence: Identifying SMEs for tailored services.
- Fraud: Appropriate monitoring of risk associated with SMEs.
Given the importance of SMEs, many organisations fall short of the mark when attempting to identify them with precision. This is often due to both the variability in defining SMEs, and the difficulty in verifying the required information stipulated in such definitions.
In this article, we demonstrate Quantexa’s SME classifier. This tool can accurately classify an enterprise as being an SME or not. It leverages Quantexa’s Network Generation capabilities to provide the crucial contextual information required for identifying SMEs. It also works at scale, allowing quick classification across a given portfolio of enterprises.
Before a deep-dive into the inner workings of the classifier, the next section will provide some important definitions for what precisely constitutes an SME.
What are SMEs?
An SME is a Small or Medium Sized Enterprise. Organisations must meet certain conditions to have SME status. Although other definitions may exist, those described in guidance published by the European Commission [3] have been used at Quantexa to inform the SME classifier.
The European Commission SME definition was selected because it is widely used in Europe, but also beyond – as other jurisdictions (such as Hong Kong and the USA) base their identification of SMEs on comparable metrics, albeit with different thresholds. The definition within the European Commission guidance centres on the Headcount, Turnover, and Balance Sheet total of an enterprise, in addition to its relevant shareholdings, interests, and owners. Relevant enterprises in the legal hierarchy around the enterprise of interest must have their Headcount, Turnover, and Balance Sheet totals included in the final calculation. This is due to the fact that enterprises that are connected to large organisations may have access to significant additional resources and assistance than others that do not. See below for a worked example of the some of the SME logic used in the European Commission’s definition.
Figure 1: The legal hierarchy of enterprises A and B. Enterprise B owns 100% of Enterprise A.
In the example above, we wish to determine the SME status of Enterprise A, which is 100% owned by Enterprise B. Under the European Commission SME definition, the headcount, turnover, and balance sheet total of Enterprise B must be added to the headcount, turnover, and balance sheet total of Enterprise A. In combination with other logic described by the European Commission SME definition, the combined totals must ultimately be compared to the thresholds given in Figure 2 below.
Figure 2: Required thresholds for SME classification. If an enterprise’s combined totals satisfy the given threshold requirements, then, in addition to some other checks,the enterprise can be classified as an SME.
Despite Enterprise A’s Headcount, Turnover and Balance Sheet totals falling comfortably under the SME thresholds on its own, it cannot be classified as an SME since Enterprise A (3 employees) and B (255 employees) have a combined total of 258 employees. Other checks must also be performed, but in this example we would conclude that Enterprise A is not an SME since the combined 258 employees exceeds the allowable thresholds (< 250) defined in Figure 2.
The advantages of accurately classifying SMEs
Accurately classifying SMEs can bring many advantages for a number of different use-cases.
One such example is within the Credit Risk space; Credit Risk concerns itself with understanding the risk of default on debt that may arise from a failure to make required payments. Accurately classifying SMEs, and in doing so considering the legal hierarchy of an enterprise, allows lenders to better understand an organisation's financial stability in informing the probability of default. Additionally, lending to SMEs is encouraged via financial authority initiatives that enable lenders to reduce their capital requirements for Credit Risk on exposures to SMEs, an initiative known within Europe as the ‘SME Supporting Factor’. In simplistic terms, this means that banks can free up capital resources that can be redeployed in the form of new loans.
Another example is in Customer Intelligence, which focuses on an organisation's ability to identify the most effective ways to interface and interact with their customers. Understanding that SMEs are classified within the context of their legal hierarchy not only enables organisations to offer tailored services to support SMEs, but also exposes other organisations within an enterprise’s legal hierarchy that may also benefit from such support.
Finally, risk monitoring with respect to fraud benefits from the accurate classification of SMEs. SMEs are sadly seen as soft targets, perhaps due to limited awareness or resources available to mitigate the risk of fraud taking place. Indeed, the types of fraud to which SMEs are vulnerable include phishing scams and invoice and payroll fraud. With a 2023 study highlighting a 151% increase in the value of fraud cases between 2021 and 2022, ensuring that SMEs, businesses for which a disproportionately large fraud loss would be detrimental, are suitably protected is paramount [4]. Accurately classifying SMEs affords financial institutions the ability to encourage and incorporate more comprehensive fraud monitoring controls to safeguard SMEs from the increasing threat of external or internal malicious actors.
The Difficulty in classifying SMEs and how Quantexa can help
Businesses can often be misclassified as being a SME because of a lack of visibility of their associated legal hierarchy. As noted above, the Headcount, Turnover, and Balance Sheet totals of both the enterprise in question and other relevant enterprises in its legal hierarchy must be considered before classification against the thresholds described in Figure 2. It is therefore crucial to gain a holistic view of an enterprise’s legal hierarchy. At Quantexa, we perform fast graph-traversal on big data at scale. This allows us to expand multiple hops away from enterprises of interest in order to understand complex legal hierarchies - ultimately enabling us to incorporate as much relevant information as possible in determining SME classification. As noted within this article, the Headcount, Turnover, and Balance Sheet totals of the enterprise in question and other relevant enterprises within its legal hierarchy must be considered as part of the SME classification assessment described in Figure 2. Because of this, businesses can often be misclassified as an SME due to a lack of visibility of other enterprises within the associated legal hierarchy.
See below for an example of a more complex legal hierarchy.
Figure 3: Complex legal hierarchies can be analyzed quickly. The highlighted enterprise (in orange) has been expanded up to 3 hops away.
Information about enterprises up to 3 hops away can still be relevant in forming the final SME classification of a given enterprise of interest. The number of calculations to perform at each hop can increase exponentially as the graph expands. It can therefore be seen that the ability to traverse outwards efficiently is of great importance, particularly as legal hierarchies grow in complexity.
Another reason why SMEs can be difficult to classify is due to the complexity of the European Commission definitions. For example, when assessing an enterprise with multiple business owners, the rules dictate that it may become necessary to modify certain ownership percentages on the input graph. This occurs when business owners are linked together, which indicates within the European Commission SME guidance that owners are acting jointly in some capacity and are comparable to that of a singular enterprise.
Below is a diagram to show this in action.
Figure 4: The above diagram describes a small legal hierarchy. The highlighted enterprise (in orange) has 3 owners. The owners own x, y, and z% of the orange enterprise. These owners are linked due to continued upstream ownership exceeding >50%. This means that each of the x, y and z% need to be summed together as the 3 owners of the orange enterprise are considered to be acting jointly.
All enterprises within a legal hierarchy must be assessed for ownership reassignment as exemplified in Figure 4. This can result in many calculations on larger and more complex legal hierarchies – a task that is optimised within Quantexa to avoid bottlenecks and long processing times, and yield the most contextually accurate perspective.
Another key component of the European Commission SME definition centres on how much of the enterprise in question is owned by a Public Body (e.g. a local authority).
The Definition states:
“An enterprise is not an SME if more than 25% of its capital or voting rights are directly or indirectly owned or controlled, jointly or individually by one or more public bodies.”
To solve this problem we have created an algorithm that can calculate indirect and direct ownership. Below is an example of this in action:
Figure 5: Public Body 1 directly owns some of the highlighted enterprise (in orange). Public Body 2 indirectly owns some of it as well, through the 2 enterprises marked with a *.
In the above example, the indirect ownership of the highlighted enterprise (in orange) by Public Body 2 can be calculated by multiplying ownership percentages through the legal hierarchy. The formula is
(a*d) + (b*c)
This must then be added with the direct ownership of Public Body 1 of the highlighted enterprise, denoted e. If (a*d) + (b*c) + e > 25%
then the highlighted enterprise is not an SME. On a very large legal hierarchy this can result in many calculations being necessary. We have optimised these calculations to speed up processing times. In addition, we note here that it is a calculation worth performing early on as we can stop any other rule checks early if Public Body ownership is established. This is because Public Body ownership above the 25% threshold guarantees that an enterprise is not an SME, regardless of the outcomes of the other rules and definitions. Stopping early in this way can greatly speed up processing times, and is an important reminder of the necessity of applying these rules in a logically sequential manner.
The above examples all rely on a complete and accurate legal hierarchy being generated for each enterprise of interest. Quantexa’s entity resolution can be leveraged to construct these hierarchies. We are able to resolve together all businesses and individuals from company registry data. We can then link these together into a complete hierarchy by referencing ownership data and shareholding percentages within our available data. Finally, we can traverse through these hierarchies at scale by leveraging Quantexa’s Q Knowledge Graph algorithm.
Conclusion
In conclusion, we have taken the European Commission SME Definition and implemented an accurate, scalable, and quick algorithm for SME classification. Classifying SMEs accurately can potentially bring a lot of cost savings to Banks or other businesses under current legislation, and paves the way for future opportunities for SMEs driven by this innovation. The complexity of this problem comes from how large the surrounding legal hierarchies can be, and how many complex rules must be applied over them. There are many more rules defined in the European Commission SME Definitions that we have not covered above, but we have nevertheless implemented in our product. We have therefore showcased Quantexa’s innovative approach to this challenge, and achieved an important milestone to support businesses accurately identifying SMEs. As rules and regulations adjust to the ever-changing landscape of challenges faced by SMEs, Quantexa is well-prepared to adapt to whatever changes come our way, giving our clients regulatory confidence when it comes to SME classification.
[1] https://ec.europa.eu/commission/presscorner/detail/en/qanda_23_4411
[2]
https://www.fsb.org.uk/uk-small-business-statistics.html
[3]
https://ec.europa.eu/docsroom/documents/42921
[4]
https://www.sjp.co.uk/individuals/news/how-important-is-fraud-detection-for-small-businesses