Predicting unlisted SMEs' default: Incorporating market information on accounting-based models for improved accuracy
Introduction
Credit risk refers to the risk created by unexpected changes in the credit quality of a counterparty or issuer and its quantification is one of the major challenges in modern finance.1 Traditionally, it is measured as the firm's likelihood of default on its contractual or required obligations and the monetary losses imposed if such default occurs. Although SMEs are the most active economic units representing the backbone of a nation's economy, due to their special characteristics, their credit and operational risks are perceived to be higher. These risks are especially prevalent during periods of prolonged financial crises as they pose a significant threat to the real economy given the potential negative impact on companies' profits, sales and investment (Claessens, Tong, & Wei, 2012).
In a recent study, Gupta and Gregoriou (2015) show that the proportion of US SMEs under financial distress in their sample had increased from 19% in 1990 to 31.61% in 2013. At the same time, the proportion of actual bankruptcies had fallen from 1.39% to 0.5% for the corresponding period with micro-SMEs being the companies worst affected.2 A similar picture is portrayed in the UK where according to the Office for National Statistics (ONS), the average death rate for new businesses in the period 2009–14 is reported as 10.35% with a five-year survival rate of only 41.7% (ONS, 2015). In an attempt to explain this trend, Guariglia, Spaliara, and Tsoukas (2016) demonstrate a statistically significant link between debt-financing cost and corporate survival rates for young and non-exporting firms, especially during the period of the 2007–09 financial crisis when borrowing rates had increased dramatically. This reduced survival rate for SMEs was amplified by the considerable difficulty in obtaining the necessary finance for their operations. Government statistics in the immediate aftermath of the 2007–09 financial crisis show that only 74% of those SMEs seeking finance in the year 2010 indeed managed to obtain some form of it as compared to a rate of 90% in the period 2007/08 (BIS Economics Paper, 2012). This decreased ability of SMEs to access short-term capital is mainly attributed to the supply of bank lending as banking and financial institutions became more risk averse as well as being legally obliged, by new financial services regulations, to increase their liquidity by holding more capital.3 Extant literature shows that unlike their large-size counterparts, SMEs' applications for financing tend to be rejected more frequently, with the evidence for this being consistent across the entire spectrum of SME activity, industrial classification and their ability to innovate (Lee, Sameen, & Cowling, 2015).
The severity of this problem in the UK setting has been extensively highlighted over the years in a series of governmental reports. For example, Cruickshank (2000) suggests the existence of systemic problems affecting the quality of lending services for SMEs such as the possible overcharging for such services by banks, the lack of available information regarding alternative banking products, and the existence of significant weaknesses in the systems of redress when things go wrong for the borrowing firms. As the report concludes, these problems are exacerbated by the presence of a complex monopolistic structure in the UK banking environment (Cruickshank, 2000, pp.161–167). A more recent joint-study by the Competition and Markets Authority (CMA) and the Financial Conduct Authority (FCA) further supports such claims. It points out that high barriers to entry and expansion in the supply of business current accounts (BCAs) and general business loans to SMEs for newer and smaller credit providers are a major weakness of the UK bank lending system (CMA & FCA, 2014). This study also reports a significant widening of the difference in the interest rate charged by banks to SMEs for term loans and the Bank of England (BoE) base rate. For example, from an annual average of less than 2.5% in the period prior to the financial crisis, interest rates have increased to 4% or more in the period 2008–2012. Hence, all prior policy literature unanimously highlights the need for improving market competition in UK banking. Allowing smaller credit providers to enter the market will most certainly improve lending services, thereby reducing both the cost of such services for SMEs and the rate of insolvencies for such firms.
This large increase in SME insolvencies during and after the 2007–09 financial crisis, alongside the introduction of the Basel II capital requirements for banks, have led to a renewed interest in the SME credit risk assessment literature with the main focus on accounting-based default prediction models designed specifically for SMEs.4 Nonetheless, although these models are easy to use in a practical sense, they ignore the important role of market factors in predicting potential default events. In terms of credit risk quantification using market-based information, one of the most widely-used models in academic literature and practice alike is the one introduced by Merton (1974) and further developed by the KMV Corporation.5 Although this model has significant advantages over the accounting-based approaches, including better predictability during financial crises and over short-term horizons, the unavailability of market information in the case of unlisted companies deems it inapplicable for the majority of SMEs (Bilderbeek & Pompe, 2005; Lin, Ansell, & Andreeva, 2007; Richardson, Kane, & Lobingier, 1998). To the best of our knowledge, there is no prior literature regarding the effect of market-based factors on the accuracy of the unlisted SMEs' default prediction models and especially no elaborate default prediction models that can use market information for predicting their potential default.
Hence, our study addresses this important gap in the literature by introducing a new modelling approach that can utilise market information for predicting unlisted SMEs' default using an average sample of 181 UK listed SMEs (L-SMEs thereafter) and 19,681 unlisted SMEs (U-SMEs thereafter) over the period 2004–2013. This is accomplished by combining market and accounting information in a way that takes into account the association between U-SMEs’ accounting ratios and the Merton's distance-to-default (DD) for L-SMEs. Our hybrid default prediction model exhibits superior predictive power compared to its existing accounting-based counterparts when tested across our entire sample of U-SMEs.
The rationale behind this approach is simple. Prior studies in the field of corporate finance highlight the benefits of adopting a comparability method in equity valuation of unlisted/private companies using industry-level data (Alford, 1992; Baker & Ruback, 1999; McCarthy, 1999).6 In a similar manner, we show that such a method has important benefits in default prediction. If the use of a market-based valuation approach is a reliable way for deriving firm value for unlisted/private companies (Alford, 1992; Baker & Ruback, 1999), then there is no theoretical reason why this approach cannot also be used for the purpose of default prediction in the case of such firms. Bhojraj and Lee (2002) further suggest that the use of market-based valuation multiples usually functions as a “satisficing” device for the professional community, trading off methodological complexity and completeness for the purpose of convenience and cost efficiency. Likewise, from the perspective of a large and highly-diverse bank engaging primarily in transaction lending technologies, lending to informationally ‘opaque’ SMEs poses significant risks and costs. Assuming a decision to lend to a L-SME or an U-SME is mutually exclusive, a bank will most likely be inclined to lend to the former type of company as the cost of accessing all relevant ‘soft’ information for the U-SMEs can be relatively high.7 Our proposed approach solves this problem, by allowing U-SMEs to be treated as if they are of a ‘quasi-listed’ status and be judged in terms of credit ability on an equal basis to the former type of firms. Furthermore, under conditions of market efficiency8 the use of market-based information can be considered as a credible source of anticipated economic conditions, as the latter will affect not only the L-SMEs but the U-SMEs as well.9
Our results indicate that the new methodological approach can forecast U-SMEs’ default better than the traditional way of empirically using a set of accounting ratios, as the use of the proxied market information significantly increases the accuracy of prediction. Moreover, our new hybrid model appears to be superior in predicting U-SMEs’ default events during the financial crisis and within a short-time span, a vital aspect for all banks engaging in U-SMEs' lending activities as part of their day-to-day operations.
The paper proceeds as follows: section 2 reviews the market-based default prediction models literature. Section 3 presents the data collection procedure, the estimation of the relevant modelling variables and introduces our new hybrid model. Section 4 discusses the empirical findings on its performance. Section 5 concludes.
Section snippets
Lending technologies
Traditional bank lending approaches are typically distinguished between transaction-based lending and relationship-based lending. As prior literature suggests, both technologies appear to have important benefits but their adoption is typically determined by factors such as the size of the borrowing firm, its financial history, the transparency of the borrower's financial statements, but also the size and organisational structure of the lender (Baas & Schrooten, 2006; Bartoli, Ferri, Murro, &
Data
Our study employs a sample of listed (L-SMEs) and unlisted SMEs (U-SMEs). All L-SMEs are selected based on the Basel definition for small firms, i.e. firms with total turnover value of less than £42 million. All operating and financial indicators are obtained from Thomson's Datastream. The final sample of L-SMEs is 198 companies, all of them in the manufacturing sector covering the period from 2004 to 2013. With regards to the UK U-SMEs’ sample, our dataset consists of approximately 20,000
Default prediction for the L-SMEs sample using the Merton-KMV model
As it is assumed in Merton-KMV model, the asset value is subject to normal distribution and the default distance reflects the standard deviation from the company's default. Thus, firms' expected default probability () can be calculated using the normality function of distance to default (DD). According to Table 4, the average value for the sample is approximately 1.7 percent (1.695) while the minimum value is 0.000 and the maximum is 27.847 percent. By observing individual EDFs,
Conclusion
An accurate U-SMEs’ default prediction model is crucial for all relevant stakeholders, including banks and other credit providers, the companies' owners, government agencies, accounting professionals, etc. It primarily helps banks to assess U-SMEs’ financial prospects with a higher level of accuracy, reducing the lenders' potential losses due to credit misallocation. It also enables banks to estimate the capital requirement for their U-SMEs’ lending portfolio more accurately by truly reflecting
References (65)
- et al.
Comparing the performance of market-based and accounting-based bankruptcy prediction models
Journal of Banking & Finance
(2008) - et al.
Availability and cost of credit for small businesses: Customer relationships and credit cooperatives
Journal of Banking & Finance
(1998) - et al.
35 Years of studies on business Failure: An overview of the classic statistical methodologies and their related problems
The British Accounting Review
(2006) - et al.
SME financing and the choice of lending technology in Italy: Complementarity or substitutability?
Journal of Banking & Finance
(2013) - et al.
Small and medium-size enterprises: Access to finance as a growth constraint
Journal of Banking & Finance
(2006) - et al.
The performance of hybrid models in the assessment of default risk
Economic Modelling
(2016) - et al.
Extending the Merton model: A hybrid approach to assessing credit quality
Mathematical and Computer Modelling
(2007) - et al.
The consolidation of the financial services industry: Causes, consequences, and implications for the future
Journal of Banking & Finance
(1999) - et al.
Bank size, lending technologies, and small business finance
Journal of Banking & Finance
(2011) - et al.
A more complete conceptual framework for SME finance
Journal of Banking & Finance
(2006)
Relationship banking: What do we know?
Journal of Financial Intermediation
The impact of SME's pre-bankruptcy financial distress on earnings management tools
International Review of Financial Analysis
Alternative bankruptcy prediction models using option-pricing theory
Journal of Banking & Finance
Credit risk measurement and early warning of SMEs: An empirical study of listed SMEs in China
Decision Support Systems
From the financial crisis to the real economy: Using firm-level data to identify transmission channels
Journal of International Economics
Family firms, soft information and bank lending in a financial crisis
Journal of Corporate Finance
Relationship lending within a bank-based system: Evidence from European small business data
Journal of Financial Intermediation
Redefinition of the KMV model's optimal default point based on genetic algorithms - evidence from Taiwan
Expert Systems with Applications
Access to finance for innovative SMEs since the financial crisis
Research Policy
The effect of the set of comparable firms on the accuracy of the price-earnings valuation method
Journal of Accounting Research
.Financial ratios, discriminant analysis and the prediction of corporate bankruptcy
The Journal of Finance
Modelling credit risk for SMEs: Evidence from the U.S. market
Abacus
.The value of non-financial information in small and medium-sized enterprise risk management
The Journal of Credit Risk
Relationship banking and SMEs: A theoretical analysis
Small Business Economics
Estimating industry multiples
Have financial statements become more informative? evidence from the ability of financial ratios to predict bankruptcy
Review of Accounting Studies
Small business credit availability and relationship lending: The importance of bank organisational structure
Economic Journal
Forecasting default with the Merton distance to default model
Review of Financial Studies
Who is my peer? A valuation-based approach to the selection of comparable firms
Journal of Accounting Research
The prediction of bankruptcy of small- and medium-sized industrial firms
Journal of Business Venturing
SMEs access to external finance
The pricing of options and corporate liabilities
Journal of Political Economy
Cited by (22)
The role of feature importance in predicting corporate financial distress in pre and post COVID periods: Evidence from China
2023, International Review of Financial AnalysisResearch and development investments, development costs capitalization, and credit ratings: Exploratory evidence from UK R&D-active private firms
2022, Journal of International Accounting, Auditing and TaxationCitation Excerpt :However, Kreß et al. (2019, p. 674) point out that it is “questionable as to whether [their] results are transferable to smaller non-public firms with more severe risk–benefit profiles”. Indeed, findings from studies concerning public firms may not be unrestrictedly transferable to the context of private companies, as they are perceived to exhibit higher credit and operational risks (Andrikopoulos & Khorasgani, 2018); moreover, their R&D activities suffer from higher uncertainty. More specifically, one can argue that when compared with smaller (private) companies, larger (public) firms can benefit from economies of scale and economies of scope.
Deep reinforcement learning imbalanced credit risk of SMEs in supply chain finance
2024, Annals of Operations ResearchFirm failure prediction for small and medium-sized enterprises and new ventures
2024, Review of Managerial ScienceForecasting SMEs’ credit risk in supply chain finance with a sampling strategy based on machine learning techniques
2023, Annals of Operations Research