UK SMEs: quantifying their pandemic risk and credit risk exposures in the wake of the COVID-19
Lead Research Organisation:
University of Nottingham
Department Name: Nottingham University Business School
Abstract
Small and medium-sized enterprises (SMEs) constitute a critical pillar of the UK economy. More than 99% of the roughly 6 million businesses in the UK are SMEs and they employ more than 16 million workers. As the impact of the COVID-19 pandemic becomes clearer, it is evident that SMEs are facing serious and unprecedented challenges, including declining revenues, defaulting on loans, inability to retain employees and postponing growth plans. However, many SMEs in the UK find it extremely difficult to obtain funding through standard banking channels as the lack of financial information about SMEs makes it difficult to evaluate SMEs' credit risk and debt repayment capacity. Hence, to meet all these pressing needs, it is critical to develop an efficient protocol to assess SMEs' pandemic risk exposure and SMEs' resilience towards funding shortages caused by COVID-19.
This project will use Artificial intelligence (AI) techniques including Machine Learning (ML), Deep Learning (DL), and Big Data to develop two novel analytical tools:
1) The Pandemic Risk Index of UK SMEs (PRI):
In this strand, the project will develop a novel Pandemic Risk Index (PRI) to model the potential economic, financial, and reputational effects of COVID-19 on UK SMEs in the short and long run. The academic and professional literature emerging in the wake of the COVID-19 crisis has considered several factors in isolation. However, this index aims to combine as many COVID-19- relevant variables as possible into one holistic multidimensional set of metrics. This is to have a better informed understanding of the big picture by accounting for and explaining the various weights and interrelationships of these variables. The main variables (but not exclusively) of this index would be (all of them are at the firm-level): exposure to global supply chains, exposure to international capital markets, corporate governance, financial flexibility, and geographical proximity to COVID-19 hotspots.
2) AI-based Programme Suite to assess the Credit Risk of Borrowing UK SMEs (AI_CREDIT):
In this strand, the project will develop an effective AI-based Python programme suite (AI_CREDIT) using Machine Learning (ML) and Deep Learning (DL) to provide policymakers in the UK government and financial intermediaries with an accurate and timely evaluation of an SME borrower's credit risk profile. With this, policymakers and lenders can make prompt decisions in providing appropriate emergency loans to SMEs to overcome their funding shortages and mitigate the impact of COVID-19. Based on the cutting-edge application of ML/DL to corporate credit risk, this project will develop a novel programme suite by integrating innovative methods. The innovations introduced by this project will extend the application of ML/DL in the estimation of SMEs' credit profiles by training ML/DL with a large amount of seemingly irrelevant data about large firms.
The research impact of this project is relevant to many stakeholders. Policymakers and lenders can directly benefit by gaining access to novel tools to allocate funds and support SMEs efficiently. Other financial institutions including Insurance companies and private equity funds will benefit from the tools in assessing the risk related to SMEs in terms of insurance policies and investment decisions, respectively. All these are likely to lead to efficient allocation of funds and reduction of cost of funds allocated to SMEs which in turn will help SMEs to survive and thrive the current and any future pandemic disruptions.
The planned project is UK wide, and it will be applicable to all UK SMEs. The project is in collaboration with the Bank of England and the Confederation of British Industry (CBI). CBI is a leading business lobby group that promotes business interests within public bodies and deals with the impact of policy on businesses in the UK. The engagement with the project partners and other stakeholders is crucial to scale up the implement
This project will use Artificial intelligence (AI) techniques including Machine Learning (ML), Deep Learning (DL), and Big Data to develop two novel analytical tools:
1) The Pandemic Risk Index of UK SMEs (PRI):
In this strand, the project will develop a novel Pandemic Risk Index (PRI) to model the potential economic, financial, and reputational effects of COVID-19 on UK SMEs in the short and long run. The academic and professional literature emerging in the wake of the COVID-19 crisis has considered several factors in isolation. However, this index aims to combine as many COVID-19- relevant variables as possible into one holistic multidimensional set of metrics. This is to have a better informed understanding of the big picture by accounting for and explaining the various weights and interrelationships of these variables. The main variables (but not exclusively) of this index would be (all of them are at the firm-level): exposure to global supply chains, exposure to international capital markets, corporate governance, financial flexibility, and geographical proximity to COVID-19 hotspots.
2) AI-based Programme Suite to assess the Credit Risk of Borrowing UK SMEs (AI_CREDIT):
In this strand, the project will develop an effective AI-based Python programme suite (AI_CREDIT) using Machine Learning (ML) and Deep Learning (DL) to provide policymakers in the UK government and financial intermediaries with an accurate and timely evaluation of an SME borrower's credit risk profile. With this, policymakers and lenders can make prompt decisions in providing appropriate emergency loans to SMEs to overcome their funding shortages and mitigate the impact of COVID-19. Based on the cutting-edge application of ML/DL to corporate credit risk, this project will develop a novel programme suite by integrating innovative methods. The innovations introduced by this project will extend the application of ML/DL in the estimation of SMEs' credit profiles by training ML/DL with a large amount of seemingly irrelevant data about large firms.
The research impact of this project is relevant to many stakeholders. Policymakers and lenders can directly benefit by gaining access to novel tools to allocate funds and support SMEs efficiently. Other financial institutions including Insurance companies and private equity funds will benefit from the tools in assessing the risk related to SMEs in terms of insurance policies and investment decisions, respectively. All these are likely to lead to efficient allocation of funds and reduction of cost of funds allocated to SMEs which in turn will help SMEs to survive and thrive the current and any future pandemic disruptions.
The planned project is UK wide, and it will be applicable to all UK SMEs. The project is in collaboration with the Bank of England and the Confederation of British Industry (CBI). CBI is a leading business lobby group that promotes business interests within public bodies and deals with the impact of policy on businesses in the UK. The engagement with the project partners and other stakeholders is crucial to scale up the implement
| Description | For the Pandemic Risk Index (PRI), the key findings are: 1) Principal Component Analysis (PCA) has been used as a dimensionality reduction technique to construct firm-level and regional indices using a large set of variables. 2) Our firm-level indices (financial and non-financial) have a strong predictive power of post-Covid-19 solvency of the UK SMEs as measured by credit scores and county court judgements. 3) Our regional social Covid-19 impact index is constructed using five variables: a) proportion of people who suffer from medium to high Covid-19-related anxiety, b) proportion of people who feel they receive enough Covid-19-related information, c) proportion of people who have to self-isolate due to Covid-19, d) proportion of people who can afford an unexpected but necessary expense of £850 after Covid-19 began, and e) proportion of people who are either uncertain about or reject the notion that life will return to full normality after Covid-19. This social index has a strong predictive power of post-Covid-19 county court judgements. 4) The UK SMEs in regional Covid-19 hotspots suffer from a more severe deterioration in their creditworthiness. 5) Our sector-level Covid-19 exposure index proves to consistently predict the deterioration in post-Covid-19 solvency of UK SMEs. 6) Overall, our indices constructed by using PCA, governmental health data, ONS social data, and stock market data enhance the prediction models of post-Covid-19 solvency of UK SMEs. AI_CREDIT is designed to evaluate the implications of Covid-19 on credit-risk-related issues of UK SMEs. As the risk profiles of SMEs are diverse, it is essential to evaluate different SMEs' risk profiles thoroughly during the pandemic to facilitate efficient funding allocation. Below details the new findings from our AI_CREDIT analyses: 1) We demonstrate that machine learning (ML) methods predict SMEs' credit scores well and enrich the analysis of SME credit profiles. The estimation residuals and prediction errors of ML methods show that the data we obtained is sufficient for achieving stable prediction performance of credit scores. 2) Popular XGBoost and LightGBM methods achieve superior performance in credit assessment compared with other ML methods. LightGBM is highly efficient even for big data. 3) When the LightGBM method was used to conduct an analysis of UK firms' credit scores in the 12 NUTS1 regions individually, London has a relatively even distribution of credit scores while most other regions show two fat tails. 4) All of the residuals of the estimations demonstrate series like white noises and the information criteria of these residuals are zeros, indicating that the models estimated for the 12 regions are adequate descriptions. 5) Our methods provide consistent high accuracies and stable performance across the 12 regions of the UK. 6) ML findings show the importance of various firm characteristics in the model estimation-the firm features for the 12 regions are similar, where the likelihood of failure (the firm age) has the most significant negative (positive) effects on credit scores. Our project has initiated a constructive collaboration with Experian PLC, a market-leading credit referencing agency, which has access to an enormous volume of credit risk-related data on UK companies. Our engagement with the company has initially attracted funding from the University of Nottingham's ESRC IAA NPIF/ABC Postdoctoral Engagement Fellowship scheme. This had enabled us to place our post doc with the company for 6 months which ended in September 2021. Via this collaboration, we have been utilising Experian data to strengthen our analyses and benefit from their expert advice and knowledge sharing with the company. We have conducted various analyses using company credit scores provided by Experian, to improve, expand and validate our findings on UK SMEs credit risk exposure after the pandemic. The collaboration and our research outcomes have attracted attention within the company, encouraging them to provide further funding to extend the contract of our post from October 2021 till the end of March 2022. The proposed research plan for the extended period includes further analyses to complement our UKRI project as well as exploring additional areas of research that carry importance for the university and the company. |
| Exploitation Route | The project has a clear and innovative dissemination plan that involves research outputs to academic and non-academic world. The dissemination of the research outcomes is achieved via: 1) The Project's Workshops: The three workshops were utilised to share the research outcomes with academic and non-academic audience. The workshops are serving as a vehicle to receive feedback from the relevant stakeholders, conduct capacity building for PhD students and Early Career Researchers (ECRs), and explore areas for further collaborations with UK and international policymakers. 2) The Project's Website: This blog is utilised to share the research outcomes with the relevant stakeholders (subject to Non-disclosure Agreements signed with our project partners and collaborating institutions). The blog will have an interface of our AI_CREDIT platform that can be used for free by SMEs, financial institutions, and the public to get quantifiable estimations of credit risk of UK SMEs in the wake of Covid-19 pandemic. 3) The Project's Discussion Paper: The detailed academic theories, methods and results are written and presented as an academic paper. The paper will be presented in international academic conferences. This paper will be submitted to a high-quality academic journal. |
| Sectors | Financial Services, and Management Consultancy |
| URL | https://blogs.nottingham.ac.uk/business/2021/11/10/uk-smes-quantifying-their-pandemic-risk-and-credit-risk-exposures-in-the-wake-of-the-covid-19-crisis/ |
| Description | The project prompted a wide interest from policymakers and industry practitioners who demonstrated a comprehensive understanding of both the project's outcome and its contributions to addressing the economic impact of the pandemic. In collaboration with our key stakeholders including, the Bank of England (BoE) and Confederation of British Industry (CBI), we organised multiple dissemination events that attracted the interest of various national and international organisations concerned with policies and financing targeted at small and medium enterprises (SMEs). We regularly met with the BoE to discuss the scope of our project and to share our initial findings with the relevant policy committees. The project complements, with a richer dataset, a very recent BoE-internal work on SME financing, (see Bank of England Working Paper No. 924). Going forward, in early December 2021, we organised a high-profile policy workshop with the BoE involving research papers, a panel of four experts (2 distinguished academics and 2 speakers from major policy institutions in this field) and one keynote speech by a BoE Financial Policy Committee member. Our engagement and collaborations with the BoE and the CBI continued during 2022. We responded to the interest expressed by international stakeholders by organising another high-profile SMEs Policy Workshop on 22 - 24 March 2022. This event provided an opportunity to disseminate our research findings to both national and international audiences: https://www.nottingham.ac.uk/news/rebuilding-business-resilience-in-the-wake-of-covid-19. The focus of the workshop was on the challenges and opportunities in the SME sector across UK, Europe, Asia and the globe. During the workshop, our project team a) showcased our UKRI-funded SMEs' Pandemic Risk Index (PRI) research findings, b) held panel discussions involving academic and non-academic participants, and c) delivered training and capacity building for early career researchers (ECRs) and PhD students on research-for-policy. The event was supported by Research England's QR-PSF Funding. This dissemination workshop attracted major national stakeholders such as the BoE, CBI, the British Business Bank (BBB), Experian UK, Innovate Finance, and UK Finance and international stakeholders, such as the Asian Development Bank Institute (ADBI), the Bombay Stock Exchange (BSE), Bpifrance, the European Central Bank (ECB), Liberis, the Organisation for Economic Co-operation and Development (OECD), and the United Nations Industrial Development Organisation (UNIDO). The success of the workshop helped to draw attention to our project and work and its significance and originality in terms of the research landscape on SMEs both nationally and internationally. In addition, this success paved the way for the launch of the new SMEs Seminar Series which is supported and jointly organised by national and international organisations. |
| First Year Of Impact | 2022 |
| Sector | Financial Services, and Management Consultancy |
| Impact Types | Economic |
| Description | Research for policy training for PhD students and Early Career Researchers (ECRs) |
| Geographic Reach | National |
| Policy Influence Type | Influenced training of practitioners or researchers |
| Description | ESRC-NPIF Accelerating Business Collaboration Postdoctoral Engagement Fellowship |
| Amount | £21,000 (GBP) |
| Organisation | University of Nottingham |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 04/2021 |
| End | 09/2021 |
| Description | Research Collaboration & Postdoctoral Engagement with Experian PLC |
| Amount | £10,000 (GBP) |
| Organisation | Experian |
| Sector | Private |
| Country | Ireland |
| Start | 01/2022 |
| End | 03/2022 |
| Description | Research Collaboration & Postdoctoral Engagement with Experian PLC |
| Amount | £10,000 (GBP) |
| Organisation | University of Nottingham |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 10/2021 |
| End | 12/2021 |
| Description | Research England QR Policy Support Fund |
| Amount | £15,000 (GBP) |
| Organisation | University of Nottingham |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 12/2021 |
| End | 03/2022 |
| Description | Business Collaboration Engagement with EXPERIAN |
| Organisation | Experian |
| Country | Ireland |
| Sector | Private |
| PI Contribution | We have initiated a constructive collaboration with Experian PLC, a market-leading credit referencing agency, which has access to an enormous volume of credit risk-related data on UK companies. We had a successful post doc placement with Experian using the funding scheme, ESRC-NPIF Accelerating Business Collaboration Postdoctoral Engagement Fellowship. Our post doc was placed at Experian for 6 months (April-September 2021) to perform compilation and analysis of the proprietary data provided by the company. Due to the success of the collaboration and effectiveness of the work done, it was extended for another six months (October 2021 - March 2022) with the funding received from the company and the Faculty of Social Sciences at the University of Nottingham. |
| Collaborator Contribution | This project was conducted in collaboration with the Bank of England (BoE), Confederation of British Industry (CBI) and Experian PLC. BoE has contributed its specialised knowledge about credit risks and lending arrangements for UK SMEs and organised a policy workshop (the second workshop of this project) in December 2021 to help with disseminating the outcomes of the project to the academic and non-academic stakeholders. CBI has provided comprehensive business survey data for UK firms (including SMEs) which are: 1) Industrial Trends Survey, 2) Distributive Trades Survey, 3) Services Sector Survey, and 4) Financial Sector Survey for the period (2000 - 2021). CBI has also contributed to the second workshop of the project by delivering presentations and exploring areas for further collaboration and funding with UK and international policymakers. Experian PLC has provided proprietary data on the financial and non-financial metrics of UK firms (including SMEs) and borrowing arrangements made by these firms with UK and international lenders. |
| Impact | The Experian PLC supported our multi-disciplinary project with important data on SME's historical credit and current account information. The research produced explains the propagation of dissolution risks, which yields important practical implications, as it helps financial institutions to allocate resources more effectively and SMEs to assess and hedge their dissolution risks. The disciplines involved in this collaboration involves economics, finance, econometrics, and machine learning: Economics: The research analyses the economic consequences of Covid-19 for UK SMEs. It also considers the moderating effects of macroeconomic indicators such as inflation and unemployment in this context. Finance: The research looks into the financial effects of Covid-19 on UK SMEs. More specifically, it analyses the changes in credit risk profiles of SMEs in different sectors and regions after the pandemic outbreak in early 2020. Econometrics: The research employs advanced econometrics methods of data reduction such as Principal Components Analysis (PCA) and Factor Analysis to reduce the dimensions and maximise the informative power of analysed variables which constitute the Pandemic Risk Index (PRI) and its regional and sectoral sub-indexes. Machine Learning: The research utilises Machine Learning (ML) techniques, such as Popular XGBoost and Light methods, to model the correlations and spillovers of Covid-19 effects across different UK regions and sectors. We have set up a blog to share our key highlights and outputs with public in general: https://blogs.nottingham.ac.uk/business/2021/11/10/uk-smes-quantifying-their-pandemic-risk-and-credit-risk-exposures-in-the-wake-of-the-covid-19-crisis/ |
| Start Year | 2021 |
| Description | Keynote talk at an international conference organised by African Accounting & Finance Association |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Policymakers/politicians |
| Results and Impact | Keynote talk at an international conference The PI was invited to give keynote talk at the 10th African Accounting and Finance Conference, organised by African Accounting & Finance Association, on 2-3 September 2021. She had shared the project aims and the initial findings. |
| Year(s) Of Engagement Activity | 2021 |
| Description | Keynote talk at an international conference organised by Indonesia Stock Exchange and Indonesian Finance Association (IFA) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Policymakers/politicians |
| Results and Impact | Keynote talk at an international conference The PI was invited to give keynote talk at The 7th Indonesian Finance Association International Conference, organised by Indonesia Stock Exchange and Indonesian Finance Association (IFA), 6-7 October 2021. She had shared the project aims and the initial findings. |
| Year(s) Of Engagement Activity | 2021 |
| Description | Press release on Rebuilding business resilience in the wake of COVID-19 |
| Form Of Engagement Activity | A press release, press conference or response to a media enquiry/interview |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Media (as a channel to the public) |
| Results and Impact | The press release was published on Monday, 21 March 2022, one day before our project's 3rd workshop which was held on 22-24 March 2022. The press release included quotes from the PI and the external partners in the project from Bank of England and Confederation of British Industry. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.nottingham.ac.uk/news/rebuilding-business-resilience-in-the-wake-of-covid-19 |
| Description | SFIC-CHASM Workshop on the Economic Response to Covid-19 |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Policymakers/politicians |
| Results and Impact | SFIC-CHASM Workshop on the Economic Response to Covid-19 (online, 20th July 2021) The team was invited to give a presentation about the project at the workshop organised by the University of Birmingham. The team had constructive discussions with the audience including other speakers from University of Kent, LMU Munich University of Hagen and National Institute of Economic and Social Research (NIESR). |
| Year(s) Of Engagement Activity | 2021 |
| Description | The First Dissemination Workshop of the Project |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Policymakers/politicians |
| Results and Impact | 1) The First Dissemination Workshop of the Project (online, 28th January 2021): This workshop aimed to introduce the project and share the initial findings of the project with academics and policymakers. The project team has benefited from discussions and debates with representatives from our project's institutional partners (Bank of England and Confederation of British Industry) and several academics and practitioners. |
| Year(s) Of Engagement Activity | 2021 |
| URL | https://blogs.nottingham.ac.uk/business/2021/11/10/uk-smes-quantifying-their-pandemic-risk-and-credi... |
| Description | The Second Dissemination Workshop of the Project |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Policymakers/politicians |
| Results and Impact | The Second Dissemination Workshop of the Project (online, 2nd December 2021): The Bank of England took the lead to run the second policy workshop, in collaboration with the consortium and the CBI. The theme was on SMEs during the COVID-19 crisis. The workshop had more than 150 participants registered for the event, from all corners around the globe. The updated findings of the project on the Pandemic Risk Index (PRI) and AI_CREDIT (an Artificial Intelligence tool to quantify credit risk exposure) in UK SMEs have been presented in the workshop. Valuable feedback and insightful comments have been received from the workshop participants reflecting the academic and non-academic perspectives. The workshop has also been an excellent opportunity to exchange knowledge with other participants and presenters and explore areas for further collaborations with respect to conducting research, applying to research grants, and capacity building. |
| Year(s) Of Engagement Activity | 2021 |
| URL | https://blogs.nottingham.ac.uk/business/2021/11/10/uk-smes-quantifying-their-pandemic-risk-and-credi... |
| Description | The Third Dissemination Workshop of the Project |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Policymakers/politicians |
| Results and Impact | During 22-24 March 2022, the research team organised the third SMEs policy workshop in collaboration with the Bank of England and the Confederation of British Industry. The goal of this workshop was to disseminate the most recent findings of the research project, conduct capacity building by providing research training to PhD students and Early Career Researchers (ECRs), and exchange knowledge and explore potentials for further collaborations with UK policymakers (such as the BBB) and international policymakers (such as the ADBI, UNIDO, the OECD, and the ECB). The workshop was designed to leverage the research team's existing partnerships with the UK and international policymakers to invite up to 250 delegates to support global SMEs. The participants included business practitioners, policymakers, academics, and PhD students, all interested in learning about the latest research on SMEs in the context of the pandemic and credit risk. Overall, the workshop organised in March was an important opportunity for the research team to engage directly with a diverse group of stakeholders, including policymakers, academics, and practitioners, and share the findings of the project. |
| Year(s) Of Engagement Activity | 2022 |
| URL | https://www.nottingham.ac.uk/news/rebuilding-business-resilience-in-the-wake-of-covid-19 |