Diminishing efficacy of Credit Score for SMEs and rise of AI in SME lending

Introduction

‘Access to finance’ is indispensable for the growth and sustenance of Small and Medium-sized Enterprises (SMEs), those are exerting a profound influence on the economies across the world. Undoubtedly, SMEs are pivotal drivers of economic development, contributing significantly to job creation (67% globally), innovation, and GDP growth. However, access to capital remains a critical challenge for many SMEs, as 85 % are unserved or underserved, debilitating their future potential. Insufficient financing impedes their ability to invest in new technologies, scale operations, penetrate new markets and may constrain execution. Consequently, this hampers their growth trajectory and hinders their capacity to contribute to overall economic prosperity.

Their struggle for seeking timely capital is aggravated by ground realities – stretched payment cycles for SMEs. For instance, In Figure 1, the data clearly shows that average debtor days of the MSMEs has been strikingly high - consistently running over 90 days. Such delays in receiving payments put pressure on their cash flow, further exacerbating the financing challenges faced by SMEs.

Figure 1: The average debtor days from 1997-98 to 2017-18. The database considered has 15,000-20,000 companies (out of which 1500-2000 companies are in the smallest size.

Additionally In Figure 2, the gross working capital cycle (days) has also been estimated for these firms, consistently running over 300 days, indicating prolonged delays in converting assets into cash. Factors such as higher debtor days, low inventory turnover ratio, and limited creditor turnover worsen the situation. Therefore, addressing the financing needs of SMEs is essential for fostering entrepreneurship, driving economic growth, and promoting job creation.

Figure 2: The gross working capital cycle (days)

Understanding Credit Scoring Models

Credit scoring plays a crucial role for financial institutions when considering credit approval for applicants (Chen and Chiou, 1999). In the financial sector, credit typically involves providing financial resources to individuals or organizations under agreed terms and conditions for both parties. Credit scoring allows financial institutions to evaluate borrowers' capacity to repay loans promptly.

Unfortunately, credit assessment process hasn’t changed since the inception of ‘lending’ practices 2000 years ago. The process of generating credit scores relies heavily on financial data, largely revolving around ‘ability to payback’ as indicated by the parameters commonly used in statistical models. Also, the key metric credit scoring revolves around ‘record of payments’; hugely downplaying the real creditworthiness, potential or intent of the borrower.

Common financial and non-financial parameters used in these models may include:

• Payment history: This includes factors such as the timeliness of previous payments, frequency of late payments, and any history of defaults or bankruptcies.
• Credit utilization ratio: This refers to the percentage of available credit that is currently being used. High utilization ratios may indicate financial strain and increased risk.
• Length of credit history: The longer a borrower's credit history, the more data available for analysis. A longer credit history may be associated with lower risk.
• Types of credit: Lenders may consider the mix of credit types, such as credit cards, instalment loans, and mortgages, in assessing creditworthiness.
• Recent credit inquiries: Multiple recent inquiries for new credit may suggest financial instability and could impact credit scores.
• Debt-to-income ratio: This compares a borrower's total debt to their income and helps assess their ability to manage additional debt.
• Employment history and stability: Stable employment may be viewed favorably by lenders as it indicates a reliable source of income.
• Demographic factors: Some models may consider demographic variables such as age, income level, and residential stability.
• Public records: Information from public records, such as tax liens, judgments, and bankruptcies, may also be included in credit score calculations.

These parameters are used collectively in statistical models to assess credit risk and generate credit scores that help lenders make informed decisions about extending credit to borrowers.

Why what we have done to assess credit is not enough?

Globally SME lending gap is at ~ USD 10 trillion (5.7 trillion in emerging markets as per World Bank estimates) and is growing in leaps and bounds year on year. The traditional reliance on financial data for assessing the creditworthiness of small firms is often inadequate due to asymmetries of data (Bank of England Paper 2020) and even if data gap is bridged factors such as unreliable financial reports or a lack of expertise in preparing them can become impediments. As a result, lenders may have stringent credit models, and may not be comfortable in lending to borrowers who don’t have proven track record of paying back. This eliminates large percentage of promising companies with strong intent and integrity as they don’t fit into the classic criteria of ‘high credit score’.

The following data are not available for unlisted firms, particularly SMEs,

• Market data: Market-related information, such as stock prices and trading volumes, is typically unavailable for unlisted firms.
• Financial statement data: Unlisted firms may be granted concessions regarding the amount of financial statement data they are required to file. This means that some of the accounting ratios used in studies of listed companies' failures may not be available for SMEs.
• Specific accounting ratios: Due to the concessions granted to unlisted firms, certain accounting ratios used in studies of listed companies' failures may not be calculable for SMEs.
• Other financial data: Various financial data points used in bankruptcy prediction models, such as those employed in multi-logit approaches or non-parametric models, may not be readily available for unlisted SMEs.

These limitations in data availability highlight the challenges in conducting comprehensive analyses of the financial health and failure risks of SMEs, particularly those that are unlisted. Also, the paradox of SME lending involves a mandatory ‘in person meeting’ even after the financials have been submitted. As a result, alternative methods that utilize non-financial data become essential for solving ‘access to finance’ for SMEs.

“Artificial intelligence and generative AI may be the most important technology of any lifetime.” Marc Benioff, chair, CEO, and co-founder, Salesforce

Fresh perspective with AI – ‘Alternate Data’ is new oil.

Historically, the utilization of non-financial data in credit decisions has been limited, primarily due to challenges in gathering and quantifying such information. With the advent of the Internet, accessing and analyzing non-financial data has become more feasible and cost-effective. In the following sections, we outline our approach for leveraging various non-financial measures available online to predict the creditworthiness of firms. This shift towards utilizing AI for financial as well as non-financial data represents a significant advancement in assessing the creditworthiness of small businesses, offering new insights and opportunities for more accurate evaluations.

AI-Powered Innovations in SME Lending

A new wave of FinTech and neobanks has disrupted the traditional SME lending ecosystem by leveraging digital technology, data analytics, and AI. These advancements enable lenders to accelerate payment processing, automate credit decision-making, and offer personalized experiences tailored to individual SME needs. By automating data collection, risk assessment, and pricing, lenders can provide real-time approvals, ensuring quick access to funds for SMEs.

According to a global technology leader Temenos “There is a pivotal moment for banks to redefine their approach to serving the SME sector. It emphasizes the transformative potential of leveraging technology, particularly Artificial Intelligence (AI), to innovate and enhance the SME customer experience. Temenos advocates for the adoption of design-centric and data-driven products and services, with a focus on digital experiences and data utilization.”

New age NBFC and fintech

The emergence of alternative finance (alt-finance) companies fills the longstanding funding gap for SMEs overlooked by traditional banks. Alt-finance lenders, unencumbered by the constraints of major financial institutions, are at the forefront of adopting new strategies and technologies, including AI. These lenders embrace flexibility and personalization, incorporating innovative criteria such as projected revenue and market potential into the lending process, thus catering to the unique needs of SMEs.

Benefits of AI in SME Lending

AI-powered innovations revolutionize risk assessment and credit scoring in SME lending. By analyzing vast datasets from diverse sources including traditional credit history, transactional data, and social media activity, AI algorithms provide more accurate and robust credit evaluations, particularly for businesses with limited credit histories. Additionally, AI automates manual underwriting processes, reducing assessment times and enabling lenders to handle larger application volumes without compromising risk analysis quality.

Challenges and Considerations

Despite its transformative potential, AI in lending presents challenges such as ethical considerations, data privacy compliance, and the risk of perpetuating biases. However, alt-finance lenders are well-positioned to address these challenges, leveraging their ideal data sets and smaller customer bases to monitor and mitigate biases within AI systems.

Future Projections

The global market for AI usage in fintech is projected to reach $61.3$ billion by 2031, driven by increased productivity and tech adoption accelerated by the COVID-19 pandemic fallout. AI-powered innovations will continue to enhance efficiency, accuracy, and inclusivity in SME lending, facilitating faster access to funds and proactive risk management. Successful integration of AI in SME lending fosters a tech-first mindset that enhances human capabilities rather than replacing them, ultimately benefiting both lenders and SMEs.

AI use cases in SME lifecycle

Generative AI has the potential to bring significant advancements and transform business functions. AI in finance can help in five general areas:?

• Personalize services and products
• Manage risk assessment and fraud
• Enable transparency and compliance
• Automate operations, scale, and reduce costs

Here are AI use cases in the SME lifecycle from the perspective of banks:

Banca AideXa’s clients provide consent to provide access to their current transaction history of business accounts.?Instead of relying solely on figures from financial reports, the bank utilizes thousands of data points as a basis for decision-making on each credit application. This bypasses the bottleneck traditional financiers face in SME lending.

Lead Management

• AI-driven lead scoring: Banks use AI algorithms to analyze data from various sources, such as transaction history, financial statements, and online behavior, to assess the creditworthiness of SMEs seeking financing. This helps banks prioritize leads and target their marketing efforts effectively.

Credit Decision

• Credit risk assessment: AI models analyze SMEs' financial data, industry trends, and macroeconomic factors to evaluate credit risk accurately. This assists banks in making informed decisions about extending credit to SMEs and setting appropriate terms and conditions.
• Automated underwriting: AI-powered underwriting processes automate the evaluation of loan applications, reducing processing times and improving efficiency. Banks can use AI algorithms to assess SMEs' creditworthiness, verify information, and generate credit decisions quickly and accurately.

Portfolio Management :

• Risk-based portfolio optimization: AI algorithms analyze banks' loan portfolios and identify risk concentrations, industry exposures, and potential vulnerabilities. This helps banks optimize their portfolios by adjusting asset allocations, diversifying risks, and improving overall risk-adjusted returns.
• Predictive analytics for asset quality: Banks use predictive analytics to forecast loan defaults, identify early warning signs of credit deterioration, and proactively manage non-performing assets. AI models analyze historical data, market conditions, and borrower characteristics to predict future credit losses and mitigate risks effectively.

Collections

• AI-powered collection strategies: Banks leverage AI algorithms to develop customized collection strategies based on borrower behavior, payment patterns, and risk profiles. AI models segment delinquent accounts, prioritize collection efforts, and optimize resource allocation to maximize debt recovery while minimizing costs and customer churn.
• Sentiment analysis and customer engagement: Banks use AI-driven sentiment analysis tools to monitor customer feedback, social media interactions, and communication channels. This enables banks to identify early signs of financial distress, engage with SMEs proactively, and offer personalized assistance to address their needs and concerns.

Overall, AI technologies empower banks to enhance lead management, credit decision-making, portfolio management, and collection processes for SME financing. By leveraging AI-driven insights and automation, banks can improve operational efficiency, mitigate risks, and deliver superior customer experiences throughout the SME lifecycle.

As per a report by UK based firm Digilytics, by 2027, AI software and tools could eliminate the administrative workload on banking staff?by 2.4 hours a day for each employee and even better in capital markets to save by 2.9 hours a day per employee.

Conclusion

In conclusion, "access to finance" stands as a critical determinant of the growth and sustainability of Small and Medium-sized Enterprises (SMEs) globally. While SMEs serve as pivotal drivers of economic development, contributing significantly to job creation, innovation, and GDP growth, the challenge of accessing capital remains prevalent, with a staggering 85% of SMEs being unserved or underserved. Insufficient financing hampers their capacity to invest in new technologies, scale operations, and penetrate new markets, ultimately hindering their contribution to overall economic prosperity.?? Traditional credit scoring methods, while integral to financial institutions' decision-making processes, often fail to capture the true creditworthiness, potential, or intent of SMEs, leaving many promising businesses excluded from access to finance. However, the advent of AI-powered innovations in SME lending presents a fresh perspective, leveraging non-financial data and advanced analytics to provide more accurate and transparent credit assessments. By embracing AI-driven solutions, banks can revolutionize their approach to SME financing, offering personalized services, mitigating risks, and fostering economic growth and entrepreneurship on a global scale - and they must embrace AI with an open mind.