The paradox of progress: Technological advancements in banking and the dual impact on SME bank borrowing

Hryckiewicz, Aneta; Korosteleva, Julia; Kozłowski, Łukasz; Rzepka, Malwina; Wang, Ruomeng Working Paper The paradox of progress: Technological advancements in banking and the dual impact on SME bank borrowing ADBI Working Paper, No. 1468 Provided in Cooperation with: Asian Development Bank Institute (ADBI), Tokyo Suggested Citation: Hryckiewicz, Aneta; Korosteleva, Julia; Kozłowski, Łukasz; Rzepka, Malwina; Wang, Ruomeng (2024) : The paradox of progress: Technological advancements in banking and the dual impact on SME bank borrowing, ADBI Working Paper, No. 1468, Asian Development Bank Institute (ADBI), Tokyo, https://doi.org/10.56506/DOZI7138 This Version is available at: https://hdl.handle.net/10419/305426 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/igo/ ADBI Working Paper Series THE PARADOX OF PROGRESS: TECHNOLOGICAL ADVANCEMENTS IN BANKING AND THE DUAL IMPACT ON SME BANK BORROWING Aneta Hryckiewicz, Julia Korosteleva, Lukasz Kozlowski, Malwina Rzepka, and Ruomeng Wang No. 1468 July 2024 Asian Development Bank Institute The Working Paper series is a continuation of the formerly named Discussion Paper series; the numbering of the papers continued without interruption or change. ADBI’s working papers reflect initial ideas on a topic and are posted online for discussion. Some working papers may develop into other forms of publication. The article has been sponsored by the National Science Center (NCN) in Poland under 2021/41/B/HS4/03586. The Asian Development Bank refers to “China” as the People’s Republic of China. Suggested citation: Hryckiewicz, A., J. Korosteleva, L. Kozlowski, M. Rzepka, and R. Wang. 2024. The Paradox of Progress: Technological Advancements in Banking and the Dual Impact on SME Bank Borrowing. ADBI Working Paper 1468. Tokyo: Asian Development Bank Institute. Available: https://doi.org/10.56506/DOZI7138 Please contact the authors for information about this paper. Email: aneta.hryckiewi[email protected], [email protected], j.korostelev[email protected]c.uk Aneta Hryckiewicz is a Visiting Scholar and Associate, Said Business School, University of Oxford, and an Associate Professor at Kozminski University, Warsaw, Poland. Julia Korosteleva is a professor at University College London, United Kingdom (UK). Lukasz Kozlowski is an associate professor at Kozminski University, Department of Banking, Insurance, and Risk, Warsaw, Poland. Malwina Rzepka is a PhD candidate, Economic Institute for Empirical Analysis, Warsaw, Poland. Ruomeng Wang is a PhD candidate, University College London, London, UK. The views expressed in this paper are the views of the author and do not necessarily reflect the views or policies of ADBI, ADB, its Board of Directors, or the governments they represent. ADBI does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use. Terminology used may not necessarily be consistent with ADB official terms. Discussion papers are subject to formal revision and correction before they are finalized and considered published. The authors are grateful for all comments received from the participants of the research seminar at the Saïd Business School, University of Oxford (2022); the IFABS conference in Naples (2021); and the Workshop on SME Performance under Uncertainty (2024) in Tokyo. The authors do not report any conflict of interest. All opinions are their own and should not be associated with any related institution. Asian Development Bank Institute Kasumigaseki Building, 8th Floor 3-2-5 Kasumigaseki, Chiyoda-ku Tokyo 100-6008, Japan Tel: +81-3-3593-5500 Fax: +81-3-3593-5571 URL: www.adbi.org E-mail: [email protected] © 2024 Asian Development Bank Institute ADBI Working Paper 1468 A. Hryckiewicz et al. Abstract This study delves into the impact of technological bank innovations on small and medium-sized enterprise (SME) borrowing across the European Union. By analyzing a comprehensive dataset of 179,921 SME-bank lending relationships from 2009 to 2019, we explore the mechanisms through which technological advancements in banking reshape traditional lending practices. Our empirical analysis documents that banks’ technological innovations have a more substantial impact on SMEs’ long-term credit growth than on their short-term growth, indicating the usefulness of these technologies in providing data that not only reduce information asymmetry but also enhance long-term decision channels. Specifically, blockchain and automation play a crucial role in expanding bank credit to SMEs. However, we also identify a paradoxical dual effect: while technological advancements facilitate credit access, they simultaneously increase the cost of borrowing for SMEs. This finding highlights a complex interplay where technological progress in banking presents both opportunities and challenges, especially for more opaque firms seeking financing. Our study contributes to the understanding of the nuanced role of innovation in banking, offering insights into the dualistic nature of the impact of technology on SME financing. Keywords: SME financing, bank technology, innovation, collateral, cost of intermediation JEL Classification: G21, O16, O33, G23, G32 ADBI Working Paper 1468 A. Hryckiewicz et al. Contents 1. INTRODUCTION .......................................................................................................... 1 2. DATA AND METHODOLOGY ...................................................................................... 4 2.1 Sample Description .......................................................................................... 4 2.2 Methodology ..................................................................................................... 7 3. EMPIRICAL RESULTS .............................................................................................. 11 3.1 Banks’ Digitalization and SME Borrowing ...................................................... 11 3.2 Banks’ Technology and SMEs’ Borrowing: A Difference-in-Difference Approach ........................................................................................................ 13 3.3 The Role of Different Types of Bank Technology in SME Borrowing ............. 15 4. THE CHANNELS OF THE TECHNOLOGICAL EFFECTS ON SME BORROWING . 18 4.1 Reduction in Information Frictions and Reliance on Collateral ....................... 18 4.2 Credit Access and Cost of Intermediation ...................................................... 20 5. ROBUSTNESS CHECKS ........................................................................................... 21 5.1 Endogeneity Concerns and Instrumental Variable Regression (IV) ............... 21 5.2 Identification Strategy ..................................................................................... 25 6. CONCLUSIONS ......................................................................................................... 29 REFERENCES ...................................................................................................................... 30 ADBI Working Paper 1468 A. Hryckiewicz et al. 1 1. INTRODUCTION In the face of rapid technological evolution, the banking industry has been at the forefront of embracing change. From the convenience of electronic payments to the insightful world of big data analytics and the intelligence of AI solutions and blockchain, banks have harnessed these innovative technologies to revolutionize their decisionmaking processes. These recent advancements in technology can introduce new lending paradigms that could extend novel opportunities to traditionally underserved customers, like SMEs. Small and medium-sized enterprises (SMEs) play a vital role in promoting sustainable economic growth worldwide via fostering innovation and competition, and creating employment. In Europe, SMEs account for 99% of the enterprise population and for more than half of its GDP and employment.1 Despite their well-acknowledged importance for the economy, SMEs receive a disproportionately small share of credit from financial institutions, and such a trend persists across both developed and developing countries (Beck and Demirgüç-Kunt 2006; ECB SAFE Survey 2020; World Bank 2018, 2023). These financing constraints are rooted in the inherent information opacity of SMEs, which exacerbates the asymmetry between lenders and borrowers and leads to credit rationing, as proven by Stiglitz and Weiss (1981). Additionally, the lack of valuable collateral and unproportionally high costs of bank financing for SMEs exclude the latter from bank financing even further (Beck and Demirgüç-Kunt 2006; De Blick, Paeleman, and Laveren 2023; Harrison et al. 2022; Yaldiz Hanedar, Broccardo, and Bazzana 2014). In this research, we investigate whether the most recent technological innovations adopted by banks can resolve the existing challenges faced by SMEs when they seek bank financing, and which technological advancements can contribute to SMEs’ improvement in accessing bank credit. Using the Amadeus firm-level panel dataset, our sample includes 179,921 SMEs across the EU, from 2009 to 2019, paired with data from 54 major European banks. We focus on SMEs with existing bank relationships to discern the additive value of technology in providing data to banks. We then merge bank data with information on technological solutions implemented by each bank affiliated with an SME. More importantly, for each bank, we can identify a technological innovation that a bank has implemented as well as the year of its implementation. Thus, by aggregating all the technological innovations implemented by a bank between 2009 and 2019, we can evaluate the level of the bank’s technological innovation and track its progression over time. We retrieve technological data from the Crunchbase and CBInsights databases and supplement this information by web-mining processes allowing us to identify banks’ announcements on bank technological product acquisition and/or development as well as its nature. Additionally, we also use alternative measures for bank technological development, such as: (i) the number of filed applications by a bank; (ii) the number of patents granted by a bank; and (iii) the number of deals a bank has been involved with as a venture capitalist (VC). We retrieve this information at a bank-year level from GlobalData. However, a certain caution to these measures must be granted as they are available only for 3,500 public companies, with the banking sector being highly underrepresented. We use these alternative measures of bank technological innovativeness to test the robustness of our results. 1 https://ec.europa.eu/growth/smes_en. ADBI Working Paper 1468 A. Hryckiewicz et al. 2 Our methodology involves two-way fixed-effect regression models, including interaction effects, difference-in-difference (DID) estimations, and two-stage instrumental variable (2SLS IV) techniques to address potential endogeneity between the adoption of bank technology and SME borrowing. We show that the Second Payment Services Directive (PSD2) adopted by the European Commission at the end of 2015 has caused an exogenous shock, significantly speeding up the digitization of the entire financial sector in Europe afterwards. The occurrence of this shock has extended the access to different technologies for banks and provided a solid foundation for our DID estimations. In contrast to existing literature that examines FinTech and BigTech firms’ role in catering to underserved customers, our study offers a unique perspective on the role of technological innovation at banks in accessing credit for opaque borrowers. While there are some academic studies documenting how FinTech and BigTech companies extend financial services to overlooked or underserved customers (for example, Balyuk (2022); Beaumont et al. (2021); Berger et al. (2021); Cornelli et al. (2023); Gambacorta et al. (2019); Gopal and Schnabl (2022); Jagtiani et al. (2021); Jagtiani and Lemieux (2018, 2019);, Ouyang (2022); Palladino (2021)), little is known about how recent banking technological innovations are changing the lending framework toward opaque customers. Though the role of bank digitalization in supplying superior credit has been recently evidenced, mainly during the pandemic crisis (see, for example, Branzoli et al. (2021; Ferri et al. (2019); Kwan et al. (2021); or more generally D’Andrea and Limodio (2023)), these studies are mainly silent concerning the role of technology in supplying credit to underserved customers. To the best of our knowledge, the only study that attempts to look at this question is that of Sedunov (2017), who uses US bank data for the period between 2001 and 2008 – long before the real FinTech development area began. Similarly, Sheng (2021) examines the role of FinTech institutions in providing bank credit to SMEs in the People’s Republic of China (PRC); however, the author uses macro data rather than bank-level data. With our study, we close the gaps in the existing literature by utilizing the most recent technological solutions adopted by individual banks to assess their role in improving access to bank credit for underserved customers. Our study reveals that the recent adoption of technological innovations by banks has enabled SMEs to overcome certain barriers and get better access to bank credit. However, we also found that a certain level of technological development at banks is necessary to effectively address the challenges associated with the information opacity of SMEs. Importantly, we observe a more pronounced impact of bank technology on increased SMEs’ long-term borrowing compared to short-term borrowing. This suggests that technological innovation not only mitigates asymmetric information problems but also enhances information efficiency, affecting other channels of bank credit decision-making. This underscores the multifaceted role of different technological innovations in assisting banks with credit decisions for opaque companies. Additionally, our analysis enriches the literature by examining the role of specific types of technological innovations in banks’ credit decision-making for SMEs. We explore the complexity and innovativeness of technological development at banks, including the adoption of electronic payments, online lending, personal finance solutions, data analytics, regulatory technology, blockchain, and automation. Subsequently, we test how these technologies address the unique challenges faced by opaque borrowers. The uniqueness and complexity of our dataset significantly distinguish us from previous academic studies that examine the impact of the general level of bank digitalization, mostly measured by bank IT spending (Branzoli, Rainone, and Supino 2021; D’Andrea and Limodio 2023; Kwan et al., 2020; Martinez Peria et al. 2022; Pierri and Timmer ADBI Working Paper 1468 A. Hryckiewicz et al. 3 2022) or by access to the internet (D’Andrea and Limodio 2023). The gradualness of our data allows us to investigate the contribution of each technological solution to address the problems of SMEs and their access to bank credit. Our research underscores the pivotal role of blockchain technology in mitigating the difficulties encountered by SMEs in obtaining bank credit. In particular, our regression analysis reveals that blockchain technology significantly broadens the spectrum of credit options available to SMEs. Through its ability to collect and process large sets of data, it allows banks to reduce the asymmetric information problem in an efficient way, improving access to bank credit for opaque firms. Moreover, we also find that data analytics and automation also appear to be highly statistically significant in improving SMEs’ access to credit. Consequently, our regressions reveal that access to data and efficiency in collecting and processing them seem to be the most important factors in extending credit to opaque customers. Moreover, our study offers novel and comprehensive insights into various channels through which technological innovation in banks can lead to increased SME borrowing. We focus on how such technology influences the easing of collateral requirements and the cost of intermediated bank credit. While a few studies, such as those by Buchak et al. (2018), Jagtiani and Lemieux (2018), and Beaumont et al. (2021), compare the cost of credit offered by FinTech companies to that of banks, the precise impact of technological advancement on the cost of bank intermediation remains underexplored. These aspects have been primarily discussed in FinTech and BigTech literature, often without definitive conclusions, but are less widely covered in banking literature. Similarly, the current literature has not addressed how bank technology affects the role of collateral in opaque companies accessing bank credit. Although Holmstrom and Tirole (1997) and Gambacorta et al. (2023) argue that greater access to hard data could ease bank requirements for collateral, the existing literature has not empirically verified the link between firm collateral and bank credit. Our data create a great testing ground to test this relationship. Our regression results document that banks with more innovative technology require less valuable collateral from SMEs than less digitalized banks. In other words, SMEs associated with technologically advanced banks can access credit with lower collateral requirements than those associated with less digitalized banks. This finding seems to suggest that transactional data may also improve the screening processes at banks. At the same time, our regression results challenge the presumption that technological innovation leads to lower financing costs. In turn, we find that more technologically advanced banks tend to charge SMEs higher interest rates. Our study may indicate that the technology might not perfectly replace the relational soft data that banks accumulate over time, and the ease of the collateral forces banks to charge a higher risk premium. In our study, we rigorously address various potential biases and endogeneity concerns across various specifications, measures of bank technological development, and the nature of SME-bank relationships. Firstly, we employ OLS regression with an interaction term as well as DID to test the potential issues related to the assumptions of DID. Secondly, we utilize TWFE DID staggered with timing as well as a single treatment period to test the robustness of our DID results. Thirdly, to address potential endogeneity related to banks’ individual features and their technological development, we employ two-stage instrumental variable (2SLS IV) regression using country-level variation related to the adoption of PSD2 as an exogenous instrument for banks’ technological development. However, to enhance the technological advancement of banks, we conduct additional regressions to address the identification problem. These regressions involve examining the relationship between individual SMEs and banks, ADBI Working Paper 1468 A. Hryckiewicz et al. 4 such as firm digital intensity or industry type, which could potentially introduce bias into our estimated results. All our robustness analyses highlight the crucial role of bank technological innovation in shaping SMEs’ access to credit. Our paper is structured as follows. The next section discusses the data and methodology, while Sections 3 and 4 present the results, which are further tested for their robustness in Section 5. Section 6 offers conclusions and policymaking implications. 2. DATA AND METHODOLOGY 2.1 Sample Description To investigate our research questions, we assemble a variety of data, such as SME data, bank-level data, including information on each bank-implemented technological solution, and macroeconomic country-level data. Our data collection process starts with the construction of the SME panel sample. For this, we use the Amadeus database provided by the Bureau van Dijk, which is the major source for EU-comparable financial and accounting data on firms. The process is comprised of two stages: (i) construction of the key financial indicators and firm-level controls for those firms with unconsolidated accounts; and (ii) gathering of the firms’ bank affiliation information. To identify SMEs, we follow the European Commission’s definition of SMEs – which is also used by Eurostat – as having fewer than 250 persons employed and an annual turnover of up to EUR50 million or a total balance sheet of no more than EUR43 million. The Amadeus database is also a primary source of information allowing us to link SMEs with their affiliated banks. We thus restrict our database to only those firms for which the information on their bank affiliation was available in the database. In total, we can identify 179,921 firms from 15 countries, i.e., Austria, Croatia, Denmark, Estonia, France, Germany, Greece, Hungary, Ireland, Latvia, Poland, Portugal, Slovenia, Spain, and the UK. Table 1 gives an overview of the sample structure by the year and by the number of banks affiliated with a firm. It also provides the features of SMEs used in our analysis. Table 1: Sample Structure This table presents a sample structure based on the observations employed in regressions from Specification 1 in Table 4. Panel A. Sample structure by year Year Countries Observations % of Observations 2009 10 7,395 0.7 2010 12 30,061 3.0 2011 12 74,473 7.4 2012 12 68,994 6.9 2013 13 103,770 10.3 2014 13 116,038 11.6 2015 13 119,877 11.9 2016 14 129,938 12.9 2017 14 133,456 13.3 2018 14 134,228 13.4 2019 15 85,178 8.5 All years 15 1,003,408 100.0 continued on next page ADBI Working Paper 1468 A. Hryckiewicz et al. 11 Table 3: Descriptive Statistics This table presents descriptive statistics for the sample. Variable Observations Firms Mean Std. Dev. Min. 1st Quart. 2nd Quart. 3rd Quart. Max. A. Dependent variables DEBT.GR 1,003,408 179,921 –0.004 0.116 –0.547 –0.042 –0.016 0.006 0.856 INT.COST 629,578 128,922 0.086 0.274 –0.057 0.007 0.029 0.064 4.003 LT.DEBT.GR 1,001,487 179,830 –0.008 0.098 –0.547 –0.036 –0.016 0.003 0.763 ST.DEBT.GR 1,002,321 179,867 –0.008 0.069 –0.536 –0.024 –0.012 0.003 0.756 B. Other firm-level variables PROFIT 1,003,408 179,921 0.026 0.159 –2.000 0.007 0.028 0.070 0.600 FIXED.ASSETS 1,003,408 179,921 0.299 0.259 0.000 0.075 0.231 0.473 1.000 LOW.COLLAT 977,667 176,325 0.499 0.500 0.000 0.000 0.000 1.000 1.000 EQUITY 1,003,408 179,921 0.466 0.264 0.000 0.246 0.450 0.679 1.000 ASSET.TURN 1,003,408 179,921 1.679 1.514 0.000 0.750 1.303 2.102 14.999 FIRM.SIZE 1,003,408 179,921 –0.268 1.688 –10.125 –1.392 –0.229 0.835 3.912 LN.FIRM.AGE 1,003,408 179,921 2.724 0.818 0.000 2.398 2.890 3.219 5.541 DIGITAL.FIRM 1,731,608 173,161 0.030 0.171 0.000 0.000 0.000 0.000 1.000 HIGH.CAPITAL 1,731,608 173,161 0.014 0.117 0.000 0.000 0.000 0.000 1.000 C. Country–level variables PRI.CREDIT 1,003,408 179,921 1.081 0.366 0.324 0.937 1.112 1.306 1.921 GDP.GROWTH 1,003,408 179,921 0.017 0.022 –0.143 0.007 0.020 0.029 0.084 GDP.PC 1,003,408 179,921 35.385 5.202 21.024 31.305 35.969 38.906 86.550 UNEMPL 1,003,408 179,921 0.155 0.067 0.031 0.097 0.153 0.214 0.275 D. Bank fundamentals BANK.SIZE 1,003,408 179,921 12.073 1.502 8.252 10.920 12.301 13.288 14.625 BANK.LOANS 1,003,408 179,921 0.597 0.093 0.131 0.561 0.595 0.655 0.863 BANK.EQUITY 1,003,408 179,921 0.080 0.031 0.011 0.063 0.070 0.081 0.224 BANK.DEPO.GR 1,003,408 179,921 0.055 0.105 –0.424 –0.003 0.034 0.085 1.311 E. Financial innovations at a bank AUT.SOFT 1,003,408 179,921 0.223 0.375 0.000 0.000 0.000 0.500 1.000 BLOCKCHAIN 1,003,408 179,921 0.153 0.338 0.000 0.000 0.000 0.000 1.000 ANALYTICS 1,003,408 179,921 0.134 0.301 0.000 0.000 0.000 0.000 1.000 LENDING 1,003,408 179,921 0.191 0.348 0.000 0.000 0.000 0.333 1.000 PAYMENTS 1,003,408 179,921 0.266 0.422 0.000 0.000 0.000 0.500 1.000 PERSON.FIN 1,003,408 179,921 0.089 0.260 0.000 0.000 0.000 0.000 1.000 REGULAT 1,003,408 179,921 0.238 0.387 0.000 0.000 0.000 0.500 1.000 BANK.INNOV 1,003,408 179,921 1.294 1.770 0.000 0.000 0.000 2.000 7.000 HIGH.DIGITAL 1,731,608 173,161 0.138 0.345 0.000 0.000 0.000 0.000 1.000 MAX.DIGITAL 1,643,512 164,351 2.750 2.096 0.000 1.000 3.000 4.000 7.000 3. EMPIRICAL RESULTS 3.1 Banks’ Digitalization and SME Borrowing We start our analysis by investigating the general impact of bank technological advancement on SME borrowing over time. Table 4 presents the regression results on whether, and if so how, bank technological innovation (BANK.INNOV) is correlated with different types of credit growth at SMEs. More specifically, we examine the role of technology in overall credit growth, as well as long-term and short-term growth, presented in Specifications (1)–(3), respectively. As discussed in the previous section, the PSD2 has fostered innovation and technological advancements through the entry of new firms offering innovative financial products and services, thereby increasing the technological development of many other financial institutions, including banks. To see whether we could also notice a change in SME borrowing before and after a potential shift in bank technological innovation, ADBI Working Paper 1468 A. Hryckiewicz et al. 12 we interact the BANK.INNOV variable with the year dummies equal to one for the years after the adoption of PSD2 (2016–2019) and zero otherwise (YEAR2016_DUMMY). The results are presented in Specifications (4)–(6). All results from this section are presented in Table 4. Table 4: The Role of Bank Technological Innovations in SMEs’ Credit Growth The table presents the regression results for firmand year-fixed-effect panel models of bank technological innovation on SMEs’ credit growth. DEBT.GR represents the growth of SMEs’ combined short-term and long-term bank debt at time t, divided by the previous year’s total assets (inflation-adjusted). LT.DEBT.GR indicates the growth of SMEs’ long-term credit at time t, while ST.DEBT.GR denotes the growth of the SMEs’ short-term credit. BANK.INNOV is a measure of a bank’s technological innovation, defined as the sum of all technological solutions adopted by bank i at time t. YEAR2016_DUMMY takes one for the period after PSD2 adoption (years 2016–2019). In the interest of brevity, we do not present coefficients for the year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** refer to significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) (5) (6) Variables DEBT.GR LT.DEBT.GR ST.DEBT.GR DEBT.GR LT.DEBT.GR ST.DEBT.GR L. BANK.INNOV 0.00109*** 0.000984*** 0.000240** 0.00119*** 0.000921*** 0.000824*** (0.000166) (0.000137) (0.000101) (0.000281) (0.000232) (0.000163) YEAR2016_DUMMY* BANK.INNOV 0.000530** 0.000462** –1.91e-06 (0.000221) (0.000185) (0.000122) L.FIX_A –0.0573*** –0.0634*** 0.00627*** –0.0572*** –0.0633*** 0.00631*** (0.00171) (0.00145) (0.000856) (0.00171) (0.00145) (0.000856) L.EBIT_S –0.00573*** –0.00643*** 0.000909 –0.00579*** –0.00648*** 0.000870 (0.00120) (0.00104) (0.000639) (0.00120) (0.00104) (0.000639) L.EQUITY 0.135*** 0.0909*** 0.0393*** 0.135*** 0.0910*** 0.0395*** (0.00144) (0.00119) (0.000747) (0.00144) (0.00119) (0.000747) L.TAT 0.0140*** 0.00819*** 0.00508*** 0.0140*** 0.00818*** 0.00507*** (0.000283) (0.000213) (0.000151) (0.000283) (0.000213) (0.000151) L.LN_SALES –0.0100*** –0.00628*** –0.00282*** –0.0100*** –0.00627*** –0.00281*** (0.000461) (0.000375) (0.000252) (0.000461) (0.000375) (0.000252) L.LN_FIRM_AGE –0.0135*** –0.00906*** –0.00456*** –0.0132*** –0.00884*** –0.00437*** (0.000958) (0.000757) (0.000516) (0.000960) (0.000759) (0.000517) PRICREDIT 0.0165*** 0.0234*** 0.00394*** 0.0187*** 0.0252*** 0.00522*** (0.00218) (0.00179) (0.00119) (0.00219) (0.00180) (0.00121) GDPGROWTH 0.331*** 0.286*** 0.215*** 0.358*** 0.307*** 0.232*** (0.0150) (0.0120) (0.00939) (0.0155) (0.0125) (0.00978) GDPPCPPP –0.000544* –0.000439* –0.00167*** –0.000470 –0.000414* –0.00158*** (0.000324) (0.000251) (0.000189) (0.000322) (0.000250) (0.000188) UNEMPL –0.223*** –0.194*** –0.177*** –0.226*** –0.198*** –0.177*** (0.0172) (0.0137) (0.00944) (0.0172) (0.0136) (0.00949) L.BANK_LN_ASSETS –0.00122 –0.00176* –0.00554*** –0.00235* –0.00258*** –0.00635*** (0.00120) (0.000990) (0.000698) (0.00121) (0.00100) (0.000708) L.BANK_LOANS –0.00919** –0.00696* –0.0140*** –0.0147*** –0.0109*** –0.0175*** (0.00436) (0.00377) (0.00246) (0.00445) (0.00386) (0.00254) L.BANK_EQUITY –0.0299** –0.000445 –0.0289*** –0.0269** 0.00141 –0.0273*** (0.0129) (0.0107) (0.00790) (0.0130) (0.0107) (0.00796) L.BANK_DEPO_GR 0.00253* 0.00277** 0.00285*** 0.00286** 0.00283** 0.00344*** (0.00137) (0.00112) (0.000880) (0.00135) (0.00110) (0.000866) 0.0335* 0.0384*** 0.142*** 0.0452** 0.0481*** 0.149*** Constant (0.0175) (0.0148) (0.00953) (0.0176) (0.0149) (0.00961) Observations 1,003,408 1,034,658 1,011,309 1,003,408 1,034,658 1,011,309 R-squared 0.043 0.036 0.041 0.043 0.036 0.041 Number of FIRM_ID 179,921 183,557 180,751 179,921 183,557 180,751 Time FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes ADBI Working Paper 1468 A. Hryckiewicz et al. 13 Our findings offer interesting insights into SME borrowing and bank technological advancement. Primarily, we see that bank technological advancement is positively correlated with all forms of SME credit growth. More specifically, we find that SMEs affiliated with more technologically advanced banks experience higher credit growth. This gives us a first insight into the role of bank technological innovation in extending access for SMEs to bank credit. Interestingly, we find that the impact of bank technological innovation appears to be more pronounced on long-term than on shortterm credit. The SMEs affiliated with a bank that has one additional solution experience a 0.09% higher credit growth in long-term credit, whereas the effect on short-term credit growth is only 0.024%. This result is economically valid as the mean for the credit growth at SMEs for the whole sample is negative. The result means that SMEs could offset the negative market trend associated with more technologically advanced banks. These findings are particularly promising as they suggest that technological solutions not only reduce the asymmetric information problems to facilitate SME short-term credit but probably also improve other channels affecting banks’ long-term lending decisions. The regression results on interaction presented in Specifications (4)–(6) offer additional insight into our analysis. They document that the effect of bank technological innovation on SME borrowing did not occur homogeneously over time. The regression results seem to suggest that when bank technological innovation has sped up, we can see higher credit growth at SMEs. This conclusion is supported by the positive and statistically significant coefficients of interaction variables between BANK.INNOV and YEAR2016_DUMMY. At the same time, we notice that BANK.INNOV variables remain highly statistically significant with a positive coefficient across all specifications. This finding is in line with other academic findings on the role of payment data in banks’ lending decisions (Ghosh, Vallee, and Zeng 2021; Ouyang 2022). Our results document that the periods of these technological developments coincided with higher SME credit growth, potentially suggesting that these solutions could support banks in their lending decisions by increasing the efficiency in collecting and processing information. 3.2 Banks’ Technology and SMEs’ Borrowing: A Difference-in-Difference Approach In this section we compare SME borrowing before and after the introduction of the PSD2. Subsequently, we also group banks affiliated with SMEs based on the level of their technological innovation depending on the number of adopted solutions to be able to compare the SME borrowing affiliated with highly digitized and less digitalized banks as described in the Methodology section. Table 5 presents the regression results for DID estimations. Our results present interesting insights. First, we note that the results from the DID render the same conclusions as from the linear regression. This could suggest that endogeneity related to the bank-SME relationship might not be present. More specifically, we see that SMEs associated with more technologically advanced banks have experienced much higher credit growth, both short-term and long-term, than those associated with less innovative banks. This finding has significant implications, suggesting that banks that have followed the technological revolution since the PSD2 have probably improved their access to different data and their processing. This finding is consistent with Angelini, Tollo, and Roli (2008), Bazarbash (2019), Fuster et al. (2019), Jagtiani and Lemieux (2019), Khandani et al. (2010)documenting the ADBI Working Paper 1468 A. Hryckiewicz et al. 14 importance of big data, data sharing, AI, and other automated procedures in improving banks’ credit scoring processes. Table 5: The Role of Bank Technological Innovations in SMEs’ Credit Growth The table presents the DID estimations for the firmand year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ credit growth. The treatment period commenced in 2016 and continued onwards. Treated banks are those that adopted more than four technological solutions in a given year (HIGH.DIGITAL) or banks that have adopted any technological solution since 2016, while the maximum number is taken between 2016 and 2019 as a measure of bank technological development (MAX.DIGITAL). DEBT.GR represents the growth of an SME’s combined short-term and long-term bank debt at time t, divided by the previous year’s total assets (inflation-adjusted). LT.DEBT.GR indicates the growth of an SME’s long-term credit at time t, while ST.DEBT.GR denotes the growth of the SME’s short-term credit. For the sake of brevity, we do not present coefficients for the constant term and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) (5) (6) Variables DEBT.GR LT.DEBT.GR ST.DEBT.GR DEBT.GR LT.DEBT.GR ST.DEBT.GR YEAR2016_DUMMY* HIGH.DIGITAL 0.00341*** 0.00273*** 0.00118*** (0.000578) (0.000455) (0.000364) YEAR2016_DUMMY* MAX.DIGITAL 0.00108*** 0.000918*** 0.000468*** (0.000137) (0.000115) (8.06e-05) L.FIX_A –0.0573*** –0.0634*** 0.00626*** –0.0572*** –0.0633*** 0.00629*** (0.00171) (0.00145) (0.000856) (0.00171) (0.00145) (0.000856) L.EBIT_S –0.00570*** –0.00641*** 0.000914 –0.00573*** –0.00644*** 0.000900 (0.00120) (0.00104) (0.000639) (0.00120) (0.00104) (0.000639) L.EQUITY 0.135*** 0.0908*** 0.0393*** 0.135*** 0.0909*** 0.0394*** (0.00144) (0.00118) (0.000747) (0.00144) (0.00119) (0.000747) L.TAT 0.0140*** 0.00819*** 0.00508*** 0.0140*** 0.00818*** 0.00508*** (0.000283) (0.000213) (0.000151) (0.000283) (0.000213) (0.000151) L.LN_SALES –0.0100*** –0.00629*** –0.00282*** –0.0101*** –0.00631*** –0.00283*** (0.000461) (0.000375) (0.000252) (0.000461) (0.000375) (0.000252) L.LN_FIRM_AGE –0.0135*** –0.00909*** –0.00455*** –0.0133*** –0.00893*** –0.00444*** (0.000957) (0.000757) (0.000516) (0.000959) (0.000758) (0.000517) PRICREDIT 0.0153*** 0.0224*** 0.00358*** 0.0180*** 0.0247*** 0.00470*** (0.00219) (0.00180) (0.00119) (0.00220) (0.00180) (0.00121) GDPGROWTH 0.325*** 0.280*** 0.215*** 0.339*** 0.293*** 0.222*** (0.0149) (0.0120) (0.00927) (0.0151) (0.0122) (0.00940) GDPPCPPP –0.000716** –0.000600** –0.00169*** –0.000763** –0.000645*** –0.00170*** (0.000320) (0.000247) (0.000187) (0.000319) (0.000246) (0.000186) UNEMPL –0.235*** –0.205*** –0.179*** –0.227*** –0.199*** –0.175*** (0.0170) (0.0135) (0.00931) (0.0170) (0.0135) (0.00941) L.BANK_LN_ASSETS –0.000302 –0.000963 –0.00532*** –0.00180 –0.00220** –0.00597*** (0.00119) (0.000991) (0.000692) (0.00120) (0.000995) (0.000707) L.BANK_LOANS –0.00649 –0.00450 –0.0136*** –0.0143*** –0.0110*** –0.0171*** (0.00431) (0.00373) (0.00243) (0.00449) (0.00390) (0.00257) L.BANK_EQUITY –0.0317** –0.00205 –0.0286*** –0.0298** –0.000674 –0.0273*** (0.0129) (0.0107) (0.00790) (0.0129) (0.0107) (0.00789) L.BANK_DEPO_GR 0.00140 0.00167 0.00272*** 0.00153 0.00179 0.00286*** (0.00135) (0.00110) (0.000864) (0.00134) (0.00110) (0.000860) Constant 0.0305* 0.0361** 0.141*** 0.0499*** 0.0522*** 0.149*** (0.0175) (0.0149) (0.00954) (0.0176) (0.0148) (0.00965) Observations 1,003,408 1,034,658 1,011,309 1,003,408 1,034,658 1,011,309 R-squared 0.043 0.036 0.041 0.043 0.036 0.041 Number of FIRM_ID 179,921 183,557 180,751 179,921 183,557 180,751 Time FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes ADBI Working Paper 1468 A. Hryckiewicz et al. 15 Similarly, as in the previous regressions, we also find that bank technological development renders a different effect on short-term versus long-term credit growth at SMEs. We notice that technology impacts long-term SME credit more significantly than short-term SME credit. Similarly, as in the linear regression, this could suggest that technology is efficient in reducing the information asymmetry by improving data collection and processing. This could be highly beneficial, especially for the short-term nature of credit. However, for long-term credit our finding might suggest that bank technologies also improve other decision channels (such as, for example, credit scoring models or screening procedures). These advantages are particularly significant for banks when making long-term rather than short-term credit decisions. Interestingly, while the statistical effects remain constant across different definitions of bank treatment group (HIGH.DIGITAL and MAX.DIGITAL), we observe some heterogeneity in terms of economic effects. As anticipated, the economic influence of bank technological development on SME credit growth is less prominent when we categorize banks into a treatment group, defined as those with any adopted solution after 2015, while the technological progress of these banks is gauged as the maximum number of solutions adopted between 2016 and 2019 (Specifications (4)–(6)). This definition introduces additional heterogeneity across banks, as banks with one solution as well as those with seven solutions enter the treatment group, rendering different effects on SMEs’ credit growth. At the same time, HIGH.DIGITAL variable includes highly technologically advanced banks. Interestingly, the regression results indicate that our effects apply to both short-term and long-term SME credit, though the effect on long-term SME borrowing is again more pronounced. These results unambiguously tend to suggest that SMEs affiliated with more technologically advanced banks experience higher credit growth than those affiliated with less innovative banks. At the same time, this highlights the transformative potential of bank technological solutions in reshaping the landscape of SME borrowing. As regards other control variables, we find that most coefficients are strongly statistically significant and exhibit the expected signs. For example, unsurprisingly we observe that higher bank debt growth is reported by younger (LN.FIRM.AGE) and smaller (FIRM.SIZE) companies with the capacity to increase the role of debt in their financing structure. However, firms with a high asset turnover (ASSET.TURN) or limited share of fixed assets in total assets (FIXED.ASSETS) are more likely to be on the point of reaching their production capacity limits and, as a result, may be more inclined to raise them through investments financed with additional debt. Interestingly, we also observe that more profitable firms (PROFIT) are less likely to incur more debt, which is in line with pecking order theory: Firms first finance their investment out of retained earnings, which is the cheapest and most readily available alternative, then out of debt, and lastly by issuing equity, seen as the most expensive option for firms. 3.3 The Role of Different Types of Bank Technology in SME Borrowing The impact of bank technological innovativeness on firm borrowing seems to be a multifaceted issue. So far, our results have documented that the level of bank technological advancement might improve an SME’s situation with regard to bank credit due to improved access to technologies supporting banks with the data collection and processing, thereby affecting the credit scoring models and screening procedures. Yet, the exact role played by different individual technological innovations in shaping a bank’s credit decisions on SMEs’ credit growth remains unclear. To investigate this, we next evaluate the impact of banks’ technological solutions on SMEs’ short-term and ADBI Working Paper 1468 A. Hryckiewicz et al. 16 long-term borrowing. This allows us to assess the value of each solution for SMEs’ borrowing. Table 6 presents the results of our regression analyses. Panel A considers the effects on long-term borrowing while Panel B of Table 7 focuses on short-term borrowing. In addition to a bank technological solution type, we include the general level of a bank’s technological innovation as a separate control variable in all specifications. We also report the results by including all solutions in the same regression model (Specification (8) of Panels A and B). Table 6 (PANEL A): The Role of Individual Technological Innovations in SMEs’ Credit Growth The table presents the regression results for firmand year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ credit growth. LT.DEBT.GR indicates the growth of SMEs’ long-term credit at time t. BANK.INNOV is a measure of a bank’s technological innovation, defined as the total number of all technological solutions adopted by bank i at time t. For brevity reasons, we do not present coefficients for firm- (PROFIT, FIXED.ASSETS, EQUITY, ASSET.TURN, LN.FIRM.AGE, and FIRM.SIZE), country- (PRI.CREDIT, GDP.GROWTH, GDP.PC, and UNEMPL), or bank-level control variables (BANK.SIZE, BANK.LOANS, BANK.EQUITY, and BANK.DEPO.GR), the constant term, and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) (5) (6) (7) (8) VARIABLES LT.DEBT.GR LT.DEBT.GR LT.DEBT.GR LT.DEBT.GR LT.DEBT.GR LT.DEBT.GR LT.DEBT.GR LT.DEBT.GR L. ELECTRONIC. PAYMENTS –0.00219*** –0.000 (0.000691) (0.000677) L. ONLINE.LENDING –0.00309*** –0.00179** (0.000735) (0.000897) L. PERSONAL_FIN 0.000702 –0.000635 (0.000690) (0.000858) L. ANALYTICS –0.000165 0.00175** (0.000728) (0.000738) L. REG_TECH –0.000231 0.00108 (0.000746) (0.000813) L. BLOCKCHAIN 0.00309*** 0.00457*** (0.000629) (0.000648) L. AUTOMATIZATION 0.000738 0.00213*** (0.000641) (0.000663) L.BANK.INNOV 0.00144*** 0.00146*** 0.000919*** 0.00100*** 0.00102*** 0.000491*** 0.000901*** (0.000195) (0.000180) (0.000151) (0.000159) (0.000191) (0.000166) (0.000156) Constant 0.0416*** 0.0349** 0.0379** 0.0383*** 0.0383*** 0.0321** 0.0402*** 0.0339** (0.0149) (0.0149) (0.0149) (0.0149) (0.0148) (0.0150) (0.0150) (0.0152) Observations 1,034,658 1,034,658 1,034,658 1,034,658 1,034,658 1,034,658 1,034,658 1,034,658 R–squared 0.036 0.036 0.036 0.036 0.036 0.036 0.036 183,557 Number of FIRM_ID 183,557 183,557 183,557 183,557 183,557 183,557 183,557 0.036 Time FE Yes Yes Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Yes Yes ADBI Working Paper 1468 A. Hryckiewicz et al. 17 Table 6 (PANEL B): The Role of Individual Technological Innovations in SMEs’ Credit Growth The table presents the regression results for firm– and year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ short-term credit growth. ST.DEBT.GR denotes the growth of the SMEs’ short-term credit. BANK.INNOV is a measure of a bank’s technological innovation, defined as the total number of all technological solutions adopted by bank i at time t. In the interest of brevity, we do not present coefficients for firm- (PROFIT, FIXED.ASSETS, EQUITY, ASSET.TURN, LN.FIRM.AGE, and FIRM.SIZE), country- (PRI.CREDIT, GDP.GROWTH, GDP.PC, and UNEMPL), or bank-level control variables (BANK.SIZE, BANK.LOANS, BANK.EQUITY, and BANK.DEPO.GR), the constant term, and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) (5) (6) (7) (8) Variables ST.DEBT.GR ST.DEBT.GR ST.DEBT.GR ST.DEBT.GR ST.DEBT.GR ST.DEBT.GR ST.DEBT.GR ST.DEBT.GR L. ELECTRONIC. PAYMENTS –0.00493*** –0.00400*** (0.000503) (0.000495) L. ONLINE.LENDING –0.00139** 0.00107 (0.000555) (0.000728) L. PERSONAL_FIN 0.000702 0.000712 (0.000690) (0.000629) L. ANALYTICS 0.000928* 0.00253*** (0.000542) (0.000558) L. REG_TECH 0.000732 –0.000577 (0.000711) (0.000818) L. BLOCKCHAIN 0.00209*** 0.00334*** (0.000456) (0.000497) L. AUTOMATIZATION 0.00167*** 0.00131** (0.000474) (0.000520) L.BANK.INNOV 0.00126*** 0.000456*** 0.000919*** 0.000145 0.000115 –9.26e-05 5.16e-05 (0.000141) (0.000143) (0.000151) (0.000123) (0.000165) (0.000119) (0.000110) Constant 0.150*** 0.141*** 0.0379** 0.143*** 0.142*** 0.138*** 0.146*** 0.146*** (0.00960) (0.00955) (0.0149) (0.00953) (0.00953) (0.00959) (0.00971) (0.00978) Observations 1,011,309 1,011,309 1,034,658 1,011,309 1,011,309 1,011,309 1,011,309 1,011,309 R-squared 0.041 0.041 0.036 0.041 0.041 0.041 0.041 180,751 Number of FIRM_ID 180,751 180,751 183,557 180,751 180,751 180,751 180,751 0.041 Time FE Yes Yes Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes Yes Yes The regression results provide compelling evidence. We observe that blockchain technology stands out as a predominant force, demonstrating the most significant economic impact among the technologies evaluated. Its capacity to provide a wide range of real-time data seems to be especially advantageous in reducing asymmetric information and improving credit scoring models, as supported by recent literature (Yang, Abedin, and Hajek 2023; Zheng et al. 2022). We also find that automation solutions play an important role in SMEs accessing long-term financing. This might suggest that the efficiency of information collection and processing is extremely important (Garg et al. 2021). Not surprisingly, our regression results document that blockchain has the greatest effect on the improved access of SMEs to long-term credit. Interestingly, our findings reveal that online lending solutions are negatively associated with long-term credit growth, suggesting that banks may still prioritize a mix of soft and hard information over purely data-driven insights for long-term loan decisions. However, at the same time, we note that the BANK.INNOV variable measuring the general level of bank technological innovation seems to be highly statistically and economically significant, suggesting that the general level of bank technological development, i.e., a mix of different technologies adopted by banks, is important while ADBI Working Paper 1468 A. Hryckiewicz et al. 18 considering the long-term nature of SME borrowing. These results confirm the complex nature of bank long-term decisions requiring different technologies supporting banks’ credit decisions. In contrast, our findings related to short-term SME borrowing (Panel B) illuminate the significance of specific technological applications such as automation, data analytics, and blockchain technology. Interestingly, the aggregate level of a bank’s technological advancement seems to be less important. This can be attributed to the nature of shortterm loans, which are characterized by smaller sums and shorter durations, requiring less exhaustive data and a simplified risk evaluation process. For such financial products, banks leverage automated technologies to reduce information asymmetry and expedite credit issuance efficiently. Interestingly, advanced payment solutions negatively impact short-term funding at SMEs, potentially due to a crowding-out effect. Meanwhile, well-developed electronic payment systems at banks seem to improve clients’ liquidity management, reducing their short-term borrowing needs, in line with findings from Carbó-Valverde, Cuadros-Solas, and Rodríguez-Fernández (2020). 4. THE CHANNELS OF THE TECHNOLOGICAL EFFECTS ON SME BORROWING 4.1 Reduction in Information Frictions and Reliance on Collateral So far, our results suggest that bank technological advancements provide banks with a wide spectrum of different data, which significantly seems to mitigate the asymmetric information problems, reducing credit risk for banks. Furthermore, access to a large set of real-time data could allow banks to switch to advanced credit scoring models, thus reducing banks’ demand for collateral.3 Therefore, technological innovation could lead to relaxed collateral requirements for SMEs, which remains one of the major barriers to loan access identified in the literature (Beck and Demirgüç-Kunt 2006; Chan and Thakor 1987; Yaldiz Hanedar, Broccardo, and Bazzana 2014; Niinimäki 2018). In this section, we examine the impact of bank technology on collateral requirements by interacting a bank’s level of technological innovation (BANK.INNOV) and the value of an SME’s collateral (COLLATERAL). A statistically significant coefficient of the interaction term (COLLATERAL*BANK.INNOV) would indicate a moderating effect of bank technology on the reliance of collateral for SME borrowing (Specification (1)). Furthermore, we consider SMEs with lower-value collateral (LOW.COLLATERAL) – those whose fixed asset to total asset ratio is below the median – and introduce it into the model as another interaction term (LOW.COLLATERAL*BANK.INNOV). The regression results are presented in Table 7. 3 This is in line with the BASEL III requirements. According to these, banks that use advanced credit scoring models may ease the requirements for collateral without imposing additional capital (BIS 2017). ADBI Working Paper 1468 A. Hryckiewicz et al. 19 Table 7: The Role of Collateral in Accessing Bank Credit for SMEs The table presents the regression results for firmand year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ credit growth. DEBT.GR represents the growth of an SME’s combined short-term and long-term bank debt at time t, divided by the previous year’s total assets (inflation-adjusted). LT.DEBT.GR indicates the growth of an SME’s long-term credit at time t, while ST.DEBT.GR denotes the growth of the SME’s short-term credit. BANK.INNOV is a measure of a bank’s technological innovation, defined as the total number of technological solutions adopted by bank i at time t. COLLATERAL refers to the value of a firm’s fixed assets as a proportion of its total assets at time t. LOW.COLLATERAL is a binary variable that takes a value of one if the firm’s fixed asset value is below the sample median, and zero if it is above the median. The interaction term (COLLATERAL*BANK.INNOV or LOW.COLLATERAL*BANK.INNOV) includes a one-period lagged collateral and a measure of bank technological innovation (BANK.INNOV) at time t. For the sake of brevity, we do not present coefficients for firm- (PROFIT, FIXED.ASSETS, EQUITY, ASSET.TURN, LN.FIRM.AGE, and FIRM.SIZE), country- (PRI.CREDIT, GDP.GROWTH, GDP.PC, and UNEMPL), or bank-level control variables (BANK.SIZE, BANK.LOANS, BANK.EQUITY, and BANK.DEPO.GR), the constant term, and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) Variables DEBT.GR DEBT.GR LT.DEBT.GR ST.DEBT.GR COLLATERAL*BANK.INNOV 0.00606*** (0.000406) LOW.COLLATERAL*BANK.INNOV 0.00177*** 0.00198*** –3.80e-06 (0.000171) (0.000136) (0.000105) L.LOW.COLLATERAL –0.0263*** –0.0267*** 0.000304 (0.000635) (0.000524) (0.000345) BANK.INNOV –5.04e-05 0.000868*** 0.000432*** 0.000823*** (0.000205) (0.000197) (0.000166) (0.000117) L.COLLATERAL –0.0615*** –0.0772*** –0.0833*** 0.00656*** (0.00177) (0.00192) (0.00163) (0.000949) Observations 1,003,165 1,003,408 1,034,658 1,011,309 R-squared 0.043 0.046 0.041 0.041 Number of FIRM_ID 179,904 179,921 183,557 180,751 Time FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes The findings reveal that the importance of a bank’s technological innovativeness is neutralized when a firm has adequate collateral (Specification 1). The interaction term (COLLATERAL*BANK.INNOV) emerges as positive and significant, suggesting that technology complements firm collateral in facilitating funding access for SMEs. However, the negative coefficient of collateral alone hints at a potential reduction in the impact of collateral on borrowing when bank technology is considered, particularly in the context of more technologically advanced banks. Interestingly, we also observe a positive coefficient for the interaction of low-value collateral and bank technological innovation (LOW.COLLATERAL*BANK.INNOV) (Specification 2), suggesting that firms with less collateral benefit from higher credit growth when aligned with technologically adept banks. The bank’s technological innovativeness variable itself is positively correlated with firm borrowing, while a low collateral value remains a significant negative factor. This implies that technological innovation may help firms with limited collateral obtain external funding by counterbalancing the negative effects of low collateral value. ADBI Working Paper 1468 A. Hryckiewicz et al. 20 Distinct differences are noted when segregating the effects of short-term and long-term SME borrowing. The positive impacts of bank technological advancement on collateral requirements are particularly pronounced for long-term credit, resonating with academic findings that emphasize the critical role of collateral in securing long-term financing (Schmalz, Sraer, and Thesmar 2017). These results underscore the potential for bank technology to aid opaque firms in obtaining long-term funding, an essential component of sustainable business growth. 4.2 Credit Access and Cost of Intermediation The potential of technology to enhance information collection and processing efficiency is significant, which in turn might influence the cost of intermediation – a notable barrier for SMEs seeking external funding in Europe. Thus, understanding the effect of a bank’s technological advancement on credit costs could provide vital insights. This section presents regression results reflecting the impact of bank technological innovation on intermediated bank credit to SMEs. Table 8 presents the regression results. Specifications (1)–(2) utilize standard linear models with interaction variables, while Specifications (3)–(4) apply a DID approach. Table 8: The Impact of Bank Technological Innovations on the Cost of Credit for SMEs The table presents the regression results for firmand year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ cost of credit. INT.COST represents the sum of all interest payments made by an SME at time t on its average value of short-term and long-term bank debt, adjusted for inflation. Specifications (1) and (2) detail the regressions on the interaction between bank technological innovativeness (BANK.INNOV) and year dummies for periods after 2015. Specifications (3) and (4) present the results of difference-in-difference regressions, where the treatment effect began in 2016 and continues onwards. Here, treated banks are identified as those that adopted more than four technological solutions in a given year (HIGH.DIGITAL) after 2015. Alternatively, digitalized banks are defined as those adopting any technological solution after 2015, with the level of digitalization measured by the maximum number of solutions adopted throughout the total sample period (MAX.DIGITAL). In the interest of brevity, we do not present coefficients for firm- (PROFIT, FIXED.ASSETS, EQUITY, ASSET.TURN, LN.FIRM.AGE, and FIRM.SIZE), country- (PRI.CREDIT, GDP.GROWTH, GDP.PC, and UNEMPL), or bank-level control variables (BANK.SIZE, BANK.LOANS, BANK.EQUITY, and BANK.DEPO.GR), the constant term, and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) (4) Variables INT.COST INT.COST INT.COST INT.COST YEAR2016_DUMMY*BANK.INNOV 0.00150** (0.000657) BANK.INNOV 0.000441 –0.00185** (0.000468) (0.000851) YEAR2016_DUMMY*HIGH.DIGITAL 0.00342** (0.00156) YEAR2016_DUMMY*MAX.DIGITAL 0.000834** (0.000406) Observations 634,770 634,770 634,770 634,770 R-squared 0.012 0.012 0.012 0.012 Number of FIRM_ID 129,387 129,387 129,387 129,387 Time FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes ADBI Working Paper 1468 A. Hryckiewicz et al. 27 interaction, would indicate that banks’ move towards digitalization has an inherent value in facilitating SME credit growth that is separate from any preexisting relationships. This distinction is critical, as our HIGH.DIGITAL definition specifically includes banks that have undergone digitalization since 2015. The methodology, therefore, allows us to account for historical SME borrowing patterns before 2016, and observe the evolution of credit relationships following the digitalization shift. The results presented in Table 12 capture these dynamics and provide insights into the influence of bank technological advancement on SME financing. Table 12: Robustness: The Role of SMEs’ Digitalization in the Access to Bank Credit The table presents the regression results for firmand year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ credit growth. DEBT.GR represents the growth of an SME’s combined short-term and long-term bank debt at time t, divided by the previous year’s total assets (inflation-adjusted). LT.DEBT.GR indicates the growth of an SME’s long-term credit at time t, while ST.DEBT.GR denotes the growth of the SME’s short-term credit. BANK.INNOV is defined as the number of technological solutions adopted by bank i at time t. DIGITAL.FIRM refers to all firms operating in the digital sector, classified according to NACE codes. This category includes Computer programming activities (NACE: 6201); Computer consultancy activities (NACE: 6202); Computer facilities management activities (NACE: 6203); Other information technology and computer service activities (NACE: 6209); Data processing, hosting, and related activities (NACE: 6311); Web portals (NACE: 6312); Publishing of computer games (NACE: 5821); Other software publishing (NACE: 5829); and Retail sale via mail order houses or via the internet (NACE: 4791). HIGH.DIGITAL denotes highly digitalized banks, defined as those exceeding four technological solutions adopted. For the sake of brevity, we do not present coefficients for firm- (PROFIT, FIXED.ASSETS, EQUITY, ASSET.TURN, LN.FIRM.AGE, and FIRM.SIZE), country- (PRI.CREDIT, GDP.GROWTH, GDP.PC, and UNEMPL), or bank-level control variables (BANK.SIZE, BANK.LOANS, BANK.EQUITY, and BANK.DEPO.GR), the constant term, and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) Variables DEBT.GR LT.DEBT.GR ST.DEBT.GR YEAR2016_DUMMY*HIGH.DIGITAL 0.00341*** 0.00284*** 0.00110*** (0.000582) (0.000459) (0.000365) YEAR2016_DUMMY*HIGH.DIGITAL*DIGITAL.FIRM 0.000217 –0.00400 0.00309 (0.00360) (0.00249) (0.00260) Observations 1,003,408 1,034,658 1,011,309 R-squared 0.043 0.036 0.041 Number of FIRM_ID 179,921 183,557 180,751 Time FE Yes Yes Yes Firm FE Yes Yes Yes The econometric analysis yields no evidence to support the hypothesis that firms with a pronounced digital presence secured more credit in the aftermath of 2015, coinciding with an era of intensified bank digitalization. Additionally, the empirical data do not affirm the predilection for digitally advanced firms to establish credit relationships with similarly digitalized banking institutions. Contrarily, the variable HIGH.DIGITAL, which is indicative of a bank’s technological advancement, exhibits a statistically significant and positive association with the growth in SME credit. This robust correlation underscores that the digitalization of banks serves as a more critical determinant in augmenting SMEs’ access to external financing than the digital attributes of the SMEs themselves. ADBI Working Paper 1468 A. Hryckiewicz et al. 28 Lastly, we employ a reverse approach using a randomized sample method, where we create a sample comprising companies from the least digitalized industries. We hypothesize that these companies are likely to be from capital-intensive industries. We classify them based on their NACE codes,6 assigning a dummy variable of one to companies assigned to these codes (HIGH.CAPITAL). If a bank’s technological development favours more digitalized companies than companies operating in less digital environment, we could see a statistically significant effect of the interaction term between dummies identifying more capital-intensive companies (HIGH.CAPITAL) and more digitalized banks (HIGH.DIGITAL) with a negative sign after 2016. Table 13 presents the regression results on interaction and HIGH.DIGITAL as a separate control variable. Table 13: Robustness: Randomized Control Sample Using Capital-Intensive Firms The table presents the regression results for firmand year-fixed-effect panel models examining the impact of bank technological innovation on SMEs’ credit growth. DEBT.GR represents the growth of an SME’s combined short-term and long-term bank debt at time t, divided by the previous year’s total assets (inflation-adjusted). LT.DEBT.GR indicates the growth of an SME’s long-term credit at time t, while ST.DEBT.GR denotes the growth of the SME’s short-term credit. BANK.INNOV is defined as the number of technological solutions adopted by bank i at time t. HIGH.CAPITAL denotes all firms operating in the nondigital sector, classified according to NACE codes. This category includes codes such as: Construction of residential and nonresidential buildings (NACE: 4120); Freight transport by road (NACE: 4941); Restaurants and mobile food service activities (5610). HIGH.DIGITAL denotes highly digitalized banks, defined as those exceeding four technological solutions adopted. For reasons of brevity, we do not present coefficients for firm- (PROFIT, FIXED.ASSETS, EQUITY, ASSET.TURN, LN.FIRM.AGE, and FIRM.SIZE), country- (PRI.CREDIT, GDP.GROWTH, GDP.PC, and UNEMPL), or bank-level control variables (BANK.SIZE, BANK.LOANS, BANK.EQUITY, and BANK.DEPO.GR), the constant term, and year dummy variables. Standard errors, clustered at the firm level, are shown in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. (1) (2) (3) Variables DEBT.GR LT.DEBT.GR ST.DEBT.GR YEAR2016_DUMMY*HIGH.DIGITAL*HIGH.CAPITAL 0.00295 0.00231 0.000641 (0.00194) (0.00159) (0.00102) HIGH.DIGITAL 0.00317*** 0.00254*** 0.00113*** (0.000594) (0.000464) (0.000380) Observations 1,003,408 1,034,658 1,011,309 R-squared 0.043 0.036 0.041 Number of FIRM_ID 179,921 183,557 180,751 Time FE Yes Yes Yes Firm FE Yes Yes Yes The regression results present interesting findings. We notice that the interaction term of HIGH.CAPITAL and HIGH.DIGITAL is statistically insignificant. This finding suggests that less digitalized firms do not seem to experience lower credit growth than any other firms when they are associated with more digitized banks. At the same time, we can see that the HIGH.DIGITAL variable is highly statistically significant, indicating a positive sign. The regression results support our previous finding that this is a bank technological development that extends access to funding for SMEs and not any 6 These industries and respective NACE codes are: 4120 – construction of residential and nonresidential buildings; 4941 – freight transport by road; 5610 – restaurants and mobile food service activities. ADBI Working Paper 1468 A. Hryckiewicz et al. 29 individual features of the companies or banks. More specifically, we also note that bank technological development supports long-term credit growth in particular, and to a lesser extent short-term SME borrowing. 6. CONCLUSIONS The recent digitalization of financial services has raised a lot of public and academic debate on the role of bank technology in addressing financing constraints faced by SMEs. For years this specific group of companies has been underfunded by traditional banks due to their information opaqueness. The recent adoption of technological innovations by banks has brought some hope of increasing data access and its efficient processing, which should lead to better credit availability for these companies. This study investigates the impact of bank technological innovations on borrowing by SMEs and the way they transform traditional lending frameworks of banks. The research utilizes a comprehensive dataset encompassing 179,921 SME-bank lending relationships across the European Union from 2009 to 2019. Our results emphasize that bank technological innovations have a more substantial impact on long-term borrowing by SMEs than on short-term borrowing, indicating the usefulness of these technologies in providing data that not only reduce information asymmetry but also enhance long-term decision channels, such as banks’ credit scoring processes. Moreover, our regression results highlight the pivotal role of blockchain technology and automation as technologies that enhance the efficiency of data collection and processing, thereby mitigating the difficulties encountered by SMEs in obtaining bank credit. The study also highlights a dual effect of technology on SME borrowing. While it eases collateral constraints, it paradoxically raises the cost of credit for SMEs. This is especially observable since the introduction of the European PSD2, which caused a significant upsurge in bank digitization. The results indicate that relaxing the requirements for SMEs to access bank credit may challenge banks to increase the risk premium due to a potentially looser relationship or a relaxed attitude toward the collateral. Our investigation into the impact of bank technological innovation on SME borrowing suggests targeted policy interventions. To improve the situation of SMEs in their bank credit access, regulators should incentivize the adoption of specific technological solutions. Moreover, they must also ensure that such innovations do not disproportionately raise the cost of credit for SMEs. Supporting competitive pricing, especially for the most opaque firms, and maintaining the benefits of relationship banking amidst digitalization seem to be key. Oversight of banks’ pricing strategies is essential to prevent the undue transfer of technology investment costs to SME borrowers. Ultimately, policies should aim to balance technological advancement with equitable credit access, reinforcing the foundation for sustainable economic growth driven by robust SME financing. ADBI Working Paper 1468 A. Hryckiewicz et al. 30 REFERENCES Angelini, E., G. di Tollo, and A. Roli. 2008. A Neural Network Approach for Credit Risk Evaluation. Quarterly Review of Economics and Finance 48: 733–755. Babina, T., S. Bahaj, G. Buchak, F. De Marco, A. Foulis, W. Gornall, F. Mazzola, T. Yu. 2024. Customer Data Access and Fintech Entry: Early Evidence from Open Banking, NCBR Working Paper 32089. Baker, A. C., D. F. Larcker, and C. Y. Wang. 2022. How Much Should We Trust Staggered Difference-in-Differences Estimates? Journal of Financial Economics 144: 370–395. Balyuk, T. 2022. FinTech Lending and Bank Credit Access for Consumers. Management Science 69: 555–575. Bazarbash, M. 2019. FinTech in Financial Inclusion: Machine Learning Applications in Assessing Credit Risk, IMF Working Papers 2019/109. Beaumont, P., H. Tang, and E. Vansteenberghe. 2021. The Role of FinTech in Small Business Lending. Working Paper. https://paulhbeaumont.github.io/pdfs/bvt.pdf (accessed 20 December 2022). Beck, T., and A. Demirgüç-Kunt. 2006. Small and Medium-Size Enterprises: Access to Finance as a Growth Constraint. Journal of Banking and Finance 30: 2931–2943. Beck, T., A. Demirgüç-Kunt, L. Laeven, and V. Maksimovic, 2006. The Determinants of Financing Obstacles. Journal of International Money and Finance 25: 932–952. Berger, T., V. Burg, A. Gombović, and M. Puri. 2021. On the Rise of Fintechs: Credit Scoring Using Digital Footprints. Review of Financial Studies 33: 2845–2897. Branzoli, N., E. Rainone, and I. Supino. 2021. The Role of Banks’ Technology Adoption in Credit Markets during the Pandemic. SSRN Electronic Journal. Available at SSRN: https://ssrn.com/abstract=3878254. Buchak, G., G. Matvos, T. Piskorski, and A. Seru. 2016. Fintech, regulatory arbitrage, and the rise of shadow banks. Journal of Financial Economics 130: 453–483. Carbó-Valverde, S., P. J. Cuadros-Solas, and F. Rodríguez-Fernández, 2020. The Effect of Banks’ IT Investments on the Digitalization of their Customers. Global Policy 11: 9–17. Chan, Y.-S., and A. V. Thakor. 1987. Collateral and Competitive Equilibria with Moral Hazard and Private Information. Journal of Finance 42: 345–363. Chiu, J., and T. V. Koeppl. 2019. Blockchain-Based Settlement for Asset Trading. Review of Financial Studies 32: 1716–1753. Coombs, C., D. Hislop, S. K. Taneva, and S. Barnard. 2020. The Strategic Impacts of Intelligent Automation for Knowledge and Service Work: An Interdisciplinary Review. Journal of Strategic Information Systems 29: 101600. Cornaggia, J., Y. Mao, X. Tian, and B. Wolfe. 2015. Does Banking Competition Affect Innovation? Journal of Financial Economics 115: 189–209. Cornelli, G., J. Frost, L. Gambacorta, P. R. Rau, R. Wardrop, and T. Ziegler. 2023. Fintech and Big Tech Credit: Drivers of the growth of digital lending. Journal of Banking and Finance 148: 106742. D’Andrea, A., and N. Limodio. 2023. High-Speed Internet, Financial Technology, and Banking. Management Science 70: 671–1342. ADBI Working Paper 1468 A. Hryckiewicz et al. 31 De Blick, T., I. Paeleman, and E. Laveren. 2023. Financing Constraints and SME Growth: The Suppression Effect of Cost-Saving Management Innovations. Small Business Economics 62: 961–986. ECB SAFE Survey. 2020. Survey on the Access to Finance of Enterprises in the Euro Area: April to September 2020. European Commission, November (September). Ferri, G., P. Murro, V. Peruzzi, and Z. Rotondi. 2019. Bank Lending Technologies and Credit Availability in Europe: What Can We Learn from the Crisis? Journal of International Money and Finance 95: 128–148. Fuster, A., M. Plosser, P. Schnabl, and J. Vickery. 2019. The Role of Technology in Mortgage Lending. The Review of Financial Studies 32: 1854–1899. Gambacorta, L., Y. Huang, Z. Li, H Qiu, and S. Chen. 2023. Data Versus Collateral. Review of Finance 27: 369–398. Gambacorta, L., Y. Huang, H. Qiu, and J. Wang. 2019. How Do Machine Learning and Nontraditional Data Affect Credit Scoring? New Evidence from a Chinese Fintech Firm. BIS Working Papers No. 834. Garg, P., B. Gupta, A. K. Chauhan, U. Sivarajah, S. Gupta, and S. Modgil. 2021. Measuring the Perceived Benefits of Implementing Blockchain Technology in the Banking Sector. Technological Forecasting and Social Change 163: 120407. Ghosh, P., B. Vallee, and Y. Zeng. 2021. FinTech Lending and Cashless Payments. SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3766250. Gopal, M., and P. Schnabl. 2022. The Rise of Finance Companies and FinTech Lenders in Small Business Lending. The Review of Financial Studies 35: 4859–4901. Harrison, R., Y. Li, S. A. Vigne, and Y. Wu. 2022. Why Do Small Businesses Have Difficulty in Accessing Bank Financing? International Review of Financial Analysis 84: 102352. Holmstrom, B., and J. Tirole. 1997. Financial Intermediation, Loanable Funds, and the Real Sector. The Quarterly Journal of Economics112: 663–691. Jagtiani, J., and C. Lemieux. 2018. Do Fintech Lenders Penetrate Areas that Are Underserved by Traditional Banks? Journal of Economics and Business 100: 43–54. ———. 2019. The Roles of Alternative Data and Machine Learning in Fintech Lending: Evidence from the Lending Club Consumer Platform. Financial Management 48: 1009–1029. Khandani, A., E., A. J., Kim, and A. W. Lo. 2010. Consumer Credit-risk Models via Machine-Learning Algorithms. Journal of Banking & Finance 34: 2767–2787. Kwan, A., C. Lin, V. Pursiainen, and M. Tai. 2021. Stress Testing Banks’ Digital Capabilities: Evidence from the COVID-19 Pandemic. SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3694288. Lerner, J., A. Seru, N. Short, and Y. Sun. 2023. Financial Innovation in the 21st Century: Evidence from U.S. Patents. Journal of Political Economy forthcoming. Martinez Peria, M., Y. Timmer, N. Pierri, and K. Modi. 2022. The Anatomy of Banks’ IT Investments: Drivers and Implications. IMF Working Papers 2022(244). Ouyang, S. (2022). Cashless Payment and Financial Inclusion. SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3948925. ADBI Working Paper 1468 A. Hryckiewicz et al. 32 Palladino, L. M. 2021. The Impacts of Fintech on Small Business Borrowing. Journal of Small Business and Entrepreneurship 33: 639–661. Pierri, N., and Y. Timmer. 2022. The importance of technology in banking during a crisis. Journal of Monetary Economics 128: 88–104. Polasik, M., A. Huterska, R. Iftikhar, and Š. Mikula. 2020. The Impact of Payment Services Directive 2 on the PayTech Sector Development in Europe. Journal of Economic Behavior and Organization 178: 385–401. Preziuso, M., F. Koefer, and M. Ehrenhard. 2023. Open Banking and Inclusive Finance in the European Union: Perspectives from the Dutch stakeholder Ecosystem. Financial Innovation 9: 1–27. Roth, J., C., P. H., A. Bilinski, and J. Poe. 2022. What’s Trending in Difference-inDifferences? A Synthesis of the Recent Econometrics Literature. Journal of Econometrics vil. 235: 2218–2244. Schmalz, M. C., D. A. Sraer, and D. Thesmar. 2017. Housing Collateral and Entrepreneurship. The Journal of Finance 72: 99–132. Sedunov, J. 2017. Does Bank Technology Affect Small Business Lending Decisions? Journal of Financial Research 40: 5–32. Sheng, T. 2021. The Effect of Fintech on Banks’ Credit Provision to SMEs: Evidence from China. Finance Research Letters 39: 101558. Spring, M., J. Faulconbridge, and A. Sarwar. 2022. How Information Technology Automates and Augments Processes: Insights from Artificial-Intelligence-Based Systems in Professional Service Operations. Journal of Operations Management 68: 592–618. Woodridge, M. J. 2010. Econometric Analysis of Cross Section and Panel Data. Second Edition, MIT. World Bank. 2018. Improving Access to Finance for SMEs: Opportunities through Credit Reporting, Secured Lending and Insolvency Practices. World Bank Group, May 2018. ———. 2023. Small and Medium Enterprises (SMEs) Finance: Improving SMEs’ Access to Finance and Finding Innovative Solutions to Unlock Sources of Capital. World Bank Group Working Paper, May 2023. Yaldiz Hanedar, E., E. Broccardo, and F. Bazzana. 2014. Collateral Requirements of SMEs: The Evidence from Less-Developed Countries. Journal of Banking and Finance 38: 106–121. Yang, F., M. Z. Abedin, and P. Hajek. 2023. An Explainable Federated Learning and Blockchain-Based Secure Credit Modeling Method. European Journal of Operational Research, forthcoming. Zheng, K., L. J. Zheng, J. Gauthier, L. Zhou, Y. Xu, A. Behl, J. Z. Zhang. 2022. Blockchain Technology for Enterprise Credit Information Sharing in Supply Chain Finance. Journal of Innovation and Knowledge, 7: 100256. Websites: https://ec.europa.eu/growth/smes_en. BIS (2017): https://www.bis.org/bcbs/publ/d424.pdf.