City Research Online

Abstract

This thesis includes four essays describing and investigating different aspects of small and medium enterprise (SME) credit scoring, which is associated with expansion of credit to SMEs, a group with historically restricted access to finance. Chapter 1 introduces SMEs and their importance and historical challenges with accessing finance, SME lending and its characteristics, and credit scoring and its foundation, readily available data predictive of borrower likelihood of loan repayment. Chapter 2 focuses on evaluating the quality of credit reporting systems, a key source of data of an effective credit score. Chapters 3 and 4 focus on credit scoring uses and the credit risk data and methods used for the models. Through different approaches, each study seeks to contribute to successful application and advancement of the credit scoring technology.

Credit reporting systems are the foundation of an effective credit score. Chapter 2 introduces a method for formal evaluation of indices of credit reporting system quality. I assess the correlation of lender perceptions of the usefulness of the credit reporting systems (private credit bureaus and public credit registries) in their country and the only globally available measurement of credit reporting infrastructure quality, the World Bank’s Doing Business Credit Information Index (CII). In the analysis I combine bank-level responses from the European Bank for Reconstruction and Development’s (EBRD’s) second Banking and Performance Environment Survey (BEPS II) and country-level CII data. I find that the CII is somewhat correlated with lender perceptions of credit reporting system utility but could be improved by incorporating credit information system adult coverage rates and distinguishing between private credit bureaus and public credit registries.

Chapter 3 identifies factors associated with lender success with use of credit scoring for SME lending. I conduct a global survey of financial institutions (FIs) across 19 countries and correlate success levels with a variety of factors (e.g., credit reporting infrastructure, size of the institution, data sources, customer type, usage, FI’s model and credit risk management and reporting). The FIs reporting the highest levels of success were more likely to use credit scoring models with data sources related to repayment history and deposit information and more likely to rely on credit scores for existing customers. On average, lenders considered credit performance with the institution to be the data source with the highest predictive value.

Chapter 4 examines the diffusion of recent credit scoring innovations (incorporation of alternative data and/or use of new credit scoring methods, such as machine learning) among lenders providing Retail and SME lending. For this analysis I incorporate bank-level data from a variety of data sources, including the BEPS III (2020), FI balance sheet and income statement metrics with country level data (credit reporting infrastructure, lender protections, extent of alternative finance, micro business and SME number and financing gap). Larger, more profitable FIs are more likely to be using alternative data and/or new credit scoring methods, as are FIs that have an ongoing relationship with a Fintech company. Country-specific factors such as financial and credit infrastructure, population size, and the prevalence of alternative finance in the market are also instrumental and correlated with usage. Majority foreign-bank owned FIs are less likely to be currently using these innovations.

Collectively, the three studies contribute to the literature by presenting a method to facilitate improvement of credit reporting systems and by identifying factors associated with credit scoring success and innovation.

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HG Finance
Departments:	Bayes Business School > Bayes Business School Doctoral Theses Bayes Business School > Finance Doctoral Theses

Export

Downloads