線上融資平台與傳統商業銀行企業放款徵信方法之差異-以星展(台灣)銀行及OnDeck為例 - 政大學術集成

全文

(1)國立政治大學商學院國際經營管理英語碩士學位學程 International MBA Program College of Commerce National Chengchi University. 碩士論文政治. 大. 立Master’s Thesis. ‧. ‧ 國. 學 sit. y. Nat. 線上融資平台與傳統商業銀行企業放款徵信方法 n. er. io. 之差異-以星展(台灣)銀行及OnDeck為例 a l of lending servicei vbetween traditional A comparative analysis n Ch U engchi banks and online lending platforms. Student: Eric Chi Advisor: Professor Shari S. C. Shang. 中華民國一〇七年十一月 November 2018. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(2) 線上融資平台與傳統商業銀行企業放款徵信方法之差異以星展(台灣)銀行及OnDeck為例 A comparative analysis of lending services between traditional banks and online lending platforms. 研究生：齊孝昕. 學. ‧ 國. 指導教授：尚孝純立. Student: Hsiao-Hsi, Chi 政治大 Advisor: Shari S. C. Shang. 國立政治大學. ‧. 商學院國際經營管理英語碩士學位學程. A Thesis. er. io. a. sit. y. Nat. 碩士論文. n. iv Submitted l to International MBAnProgram. Ch. U. i e n g c hUniversity National Chengchi. in partial fulfillment of the Requirements for the degree of Master in Business Administration. 中華民國一〇七年十一月 November 2018. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(3) Acknowledgement IMBA was a life changing experience to me. I was inspired by Professor Lee who encouraged us to expand our comfort zone and to explore new opportunities. Without his inspiration, I could not make decision to test my capability. Besides Professor Lee, my advisor Shari who always opens her time to let me discuss my concept with her. IMBA is not the end of my study. All the wonderful people I met, no matter in class or out of class, enabled me to be a more educated person. We exchange ideas in class and conquer many class projects together. Those precious. 治政 experiences will become my lifetime nutrition. 大立 ‧ 國. 學. Last but not least, I would like to thank my wife, Sally Yen. Without her, I could not finish my. ‧. IMBA degree and this thesis. Manners maketh man. Sally always brings me to a higher standard. sit. y. Nat. and helps me become better. My life is different because of her and I hope I could elevate myself. n. al. er. io. to a better man and a person with discipline that Sally and my family are proud of.. Ch. engchi. i. i n U. v. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(4) Abstract A case study focuses on comparative analysis of traditional bank and online lending platform By Eric Chi Deposit and lending is a fundamental function of a bank. A core competiveness of a banker is. 政治大. the ability of minimizing a borrower’s credit risk. In current traditional banking system, lending. 立. decision is a time-consuming process and requires a mountain of documents.. ‧ 國. 學. However, with the advancement of information technology and the use of smart data analytics,. ‧. various data could be collected from public authorities and social media and analyzed from different perspectives. The process of decision-making for credit approval has changed greatly. y. Nat. n. er. io. al. sit. in some online lending platforms.. v. This study strives to answer the questions of “Why banks and online lending platforms perform. Ch. engchi. i n U. differently in operation?” Areas in the size of facility, risk management and pricing quotation will be examined. The study results revealed that banks have better capability in risk management and complicated deals can be handled. On the other hand, online lending platforms have better efficiency but charge a relatively higher interest margin and facility fee. Keywords: Online Lending Platform, Lending decision, OnDeck, DBS Bank, Comparative analysis. ii. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(5) TABLE OF CONTENTS 1. Introduction ...........................................................................................................................1 1.1. Research Motivation ........................................................................................................1 1.2. Research Objectives .........................................................................................................2. 政治大. 1.3. Research Processes ..........................................................................................................3. 立. 2. Literature Review .................................................................................................................4. ‧ 國. 學. 2.1. Introduction of P2P lending .............................................................................................4. ‧. sit. y. Nat. 2.2. Introduction of SME ........................................................................................................5. er. io. 2.3. Introduction of Online Lending Platform for SME .........................................................5. al. n. v i n C hLocal Lending Program 2.4. The Current Method of DBS e n g c h i U for SME .....................................6 2.5. Lending to SME in Taiwan ..............................................................................................7 2.6. Market Size of SME lending............................................................................................8 2.7. Regulation of Banking and Credit in Taiwan...................................................................9 2.8. Value Proposition Model................................................................................................10 2.9. Platform Revolution .......................................................................................................10. iii. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(6) 3. Research Methodology .......................................................................................................12 3.1. Design ............................................................................................................................12 3.2. Data Collection ..............................................................................................................13 3.3. Data Analysis .................................................................................................................13. 政治大. 4. Case Study ...........................................................................................................................14. 立. 4.1. DBS ................................................................................................................................14. ‧ 國. 學. 4.2. On Deck Capital (“OnDeck”) ........................................................................................21. ‧. sit. y. Nat. 5. Results and Analysis ...........................................................................................................26. er. io. 5.1. Cross Case Analysis .......................................................................................................26. al. n. v i n Ch 5.2. Analysis and Findings .................................................................................................... 28 engchi U 5.3. Keys to improve efficiency of DBS local lending program ..........................................29. 6. Research summary..............................................................................................................30 6.1. Research contribution and limitation .............................................................................32. 7. Reference .............................................................................................................................33. iv. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(7) List of Figures and Tables Figure 1 The Research Process ................................................................................................3 Figure 2 Credit approval process of DBS .............................................................................16 Figure 3 Required Data Set for DBS Lending Decision ......................................................17 Figure 4 The work flow of 中小企業星易貸 ........................................................................18. 立. 政治大. Table 1 Number of SMEs in different area .............................................................................7. ‧ 國. 學. Table 2 Years of Establishment of SME in 2016.....................................................................8. ‧. Table 3 Regulation of Banking and Credit in Taiwan ...........................................................9. sit. y. Nat. Table 4 Current Major OLP for SMEs .................................................................................13. er. io. Table 5 Sales breakdown by segment of DBS in FY2017 ....................................................15. al. n. v i n Table 6 The Default Chance of C a given J21 Scoring as h e n g c h i Uof 2012. .........................................19. v. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(8) . 1. Introduction Lending is an ancient business activity but lending decision is still very complicated in current commercial banks. Lending decision is highly relied on information analysis. However, the credit process of lending to a large corporate and lending to a small and medium size (SME) corporate is very different. The major difference derives from the quality of data and the professionalism of credit officer.. 政治大. A standard credit process requires more than two months because the borrower has to. 立. prepare more than ten required documents and fill in various applications to fulfill government. ‧ 國. 學. regulation and banks’ internal credit policy. Under credit review process, the credit approver will raise several questions to the borrower based on credit application provided by the credit. ‧. analyst or the relationship manager. The review process requires one week to one to two month.. y. Nat. io. sit. 1.1. Research Motivation. n. al. er. Fintech is the most popular topic in recent years. There are many researches paying. Ch. i n U. v. attention to Peer-to-Peer online lending platform (林博仁, 2016; 梅驊, 2016). However, the. engchi. research on online lending platform (“OLP”) lending to corporate borrower is limited. Fintech is a trend of the future which enables borrowers to have chance to apply for credit line through an easier and faster method. Why it is a hot topic that Fintech changed entire financial system? Lending to corporate borrowers in a bank requires a very complex and long process. Typically, bankers need to understand five important information, which were People, Purpose, Repayment, Perspective and Protect. These were so-called 5P. To monitor and mitigate a bank’s credit risk, credit analysts are required to review and analyze all these five factors on a credit application. A traditional credit process requires one month for credit application preparation.. 1. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(9) . On the other hand, lending to corporate borrower through online platform only required few minutes and no complicated documentation was required. While the required documents were less than traditional commercial bank, the default rate for those online lending platforms were as low as a bank. New online platforms were disrupting the traditional market for small business loans. (Mills and McCathy, 2014) The efficiency of lending decision between online lending platform and commercial bank. 政治大 has better efficiency compared to traditional commercial bank. 立. makes a huge difference. This study would like to discuss why and how online lending platform. ‧ 國. 學. 1.2. Research Objectives The aim of this paper is therefore twofold.. ‧. sit. y. Nat. (a) The first purpose is to understand the current development of online lending platform. io. er. on corporate lending.. (b) The second purpose is to make a comparative analysis for lending policy in DBS. al. n. v i n C h policy of onlineUlending platform. Bank (Taiwan) and the lending engchi. Given the theoretical positions of the study as mentioned above, the study aims to provide an answer to the following questions: (a) Why online lending platform are with better efficiency? (b) Which industry or corporate is a target for online lending platform? (c) In what areas can DBS improve by learning from those online lending platforms? (d) Why can those lending platforms maintain such low default rate?. 2. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(10) . 1.3. Research Processes This research focuses on the current development of online lending platform and makes a comparative analysis for lending to small and median size company of DBS Bank (Taiwan). Thus, a deep and comprehensive understanding on the current online lending platform is required.. Research Purpose and Motivation. 立. 政治大 Problem Statement. ‧ 國. 學. Background and Literature Review. y. ‧. Nat. n. er. io. al. sit. Research Methodology. Ch. Case Study. engchi. i n U. v. Result and Analysis. Conclusion and Recommendation Figure 1 The Research Process. 3. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(11) . 2. Literature Review In this chapter, the research includes a comprehensive literature review on OLP for SME corporate. Besides literature review, a further study on OLP’s business model will also be discussed in this chapter.. 2.1. Introduction of P2P lending Peer-to-peer lending is a new financing method dated back to 2005, initiated by Zopa.com. 治政 (Zopa, 2018). A typical P2P lending established on an online 大 platform. Lenders can lend money 立 to non-specific borrowers through online marketplace. In the P2P lending platform, lenders are ‧ 國. 學. investors who receive interests as returns from the deal.. Han, 2012). ‧. In the past research, four characteristics of P2P lending had been mentioned. (Chen and. y. Nat. n. al. er. io. the marketplace.. sit. (a) Transparency: Under information symmetry, any borrowers or lenders are able to join. Ch. i n U. v. (b) Low operating cost and low entry barrier: Thanks to low operating cost, P2P lending. engchi. is attractive for borrowers who need small amount of money. (c) Diversified risks: P2P lending diversified risks by sharing borrowers’ risks to a group of investors. (d) Credit investigation and authorization by third party: P2P lending cooperates with third party organization, such as rating agency or government credit agency to support credit investigation. Therefore, references about borrowers are more accessible to the lenders.. 4. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(12) . 2.2. Introduction of SME The definition of SME is different in different markets or countries. Since this study focused on research in Taiwan, the study defined SME based on Small and Medium Enterprise Administration Ministry of Economic Affairs’ definition1. The benchmark for SME has two principles (i) Paid-in Capital and (ii) Number of employees. However, the standard varies in terms of the nature of business and industry. (a) The enterprise is a company in manufacturing, construction, mining or quarrying. 政治大. industry with either paid-in capital of TWD80 million or less, or less than 200 regular. 立. employees.. ‧ 國. 學. (b) The enterprise is a company in the industry other than any of those mentioned in the previous paragraph and either had its sales revenue of TWD100 million or less in the. Nat. sit. y. ‧. previous fiscal year, or less than 100 regular employees.. io. er. Based on the definition, the sales turnover is defined by the figures approved by the tax authorities in that fiscal year or the previous fiscal year. On the other hand, the number of. al. n. v i n C hmonthly number ofU employees for whom is registered in employees is defined by the average engchi labor insurance.. 2.3. Introduction of Online Lending Platform for SME Online lending platform is a fully automated funding marketplace to small and medium business, which shortens lending decision from weeks to minutes by using data generated through business activities, including financials, online sales order, shipping records and other online reference. 1. Standards for identifying small and medium-sized enterprise. https://www.moeasmea.gov.tw/ct.asp?xItem=8261&CtNode=530&mp=2 (Review: 1 February 2018). 5. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(13) . 2.4. The Current Method of DBS Local Lending Program for SME DBS Bank (Taiwan) has developed a lending program especially for enterprises in Taiwan to accommodate to the local regulation requirement and try to differentiate from severe competition in banking industries. There are 1.4 millions of SMEs in Taiwan, which was around 97.7% of registered companies in Taiwan and there are 78.2% of employees are working in SMEs. (經濟部中小企業處, 2017). Due to the large amount of SMEs in Taiwan, DBS Bank (Taiwan) has launched a lending program to customize lending program in Taiwan. The local. 政治大. lending program starts from 2012 and the program keeps being updated to adapt to the changing. 立. local market. The program includes three types of risk acceptance criteria. Financial criteria. With the change of the program, the importance of financial. ‧ 國. 學. (i). criteria has decreased. This is because the quality of financial statements of SMEs. ‧. is not accurate enough and financial figures are biased by different industries and. Nat. sit. y. different accounting standards. It’s getting difficult for banks to rely on financial. n. al. er. io. data to analyze a SME’ solvency. Besides the inferior data accuracy, financial. i n U. v. data should not be totally relied on because of three major points, including (i). Ch. engchi. Time lag: Financial statements are outdated indicators; (ii) Model restriction: The inferior capital structure is a negative signal but not the only factor to make a credit decision. (iii) Disintegrate management: Managements have responsibility to prepare financial statements. Thus, the management’s integrity is very important. (林妙宜, 2001) (ii). Non-financial criteria. Non-financial data is a vital benchmark for program lending because small and medium size companies could be adversely impacted by external environment easily. Therefore, a promoter’s integrity in business activity, family relationship, litigation record and personal financial stance could 6. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(14) . be an important information on lending decision. However, as there is many judgmental information from non-financial reference, a comprehensive check list results in long and inefficiency lending decision. The non-financial information includes (i) JCIC (Joint Credit Information Center) record; (ii) litigation record; (iii) search for recent news; (iv) land and property registration documents, etc. (Lin, 2016). 政治大. 2.5. Lending to SME in Taiwan. 立. In 2016, there are 1.4 millions SMEs in Taiwan. Geographically speaking, northern Taiwan. Southern Taiwan Eastern Taiwan Other Area Total. al. 342. Ch. y. er. 657. Percentage of Total SMEs. sit. Number of SMEs (Unit:’000). n. Central Taiwan. io. Northern Taiwan. Table 1 Number of SMEs by areas. Nat. Area. ‧. ‧ 國. 學. accounted for almost half of SMEs. The detailed distribution is tabulated as below.. 355. e34n g c h i U. v ni. 46.3% 24.3% 25.2% 2.4%. 20. 1.5%. 1,408. 100%. In terms of number of SMEs, Northern Taiwan accounted for more than 40% of SMEs, indicating that the number of SMEs was associated with business activities in Taiwan. Nevertheless, in terms of years of establishment, around 30% of SMEs’ were less than 5 years.. 7. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(15) . Table 2 Years of Establishment of SME in 2016 Year(s) of Establishment. Percentage. Accumulated Percentage. <1. 6.8%. 6.8%. 1-2. 7.0%. 13.8%. 2-3. 5.9%. 19.7%. 3-4. 5.9%. 25.6%. 4-5. 4.9%. 30.5%. 5-10. 18.3%. 48.8%. 10-20. 25.8%. 74.6%. >20. 25.4%. 100.0%. 政治大. The short years of establishment also implied a higher default risk for SMEs. (Small and. 立. Medium Enterprise Administration, 2015; 經濟部中小企業處, 2017). ‧ 國. 學. 2.6. Market Size of SME lending. ‧. While large corporate only accounted for 2.3% of registered companies in Taiwan but. sit. y. Nat. 71.8% of bank loans were lent to large corporates, indicating the SMEs only consumed smaller. al. n. TWD11,503 billion.. er. io. portions of total bank debts. In FY2015, the size of commercial lending amounted to. Ch. engchi. i n U. v. In FY2016, First Commercial Bank is a leading player in SME lending, followed by Taiwan Cooperative Bank, Hua Nan Commercial Bank and Taiwan Business Bank. In FY2016, the top 10 banks shared 71.5% of SME’s loan outstanding and most of them were governmentsupported banks in Taiwan. Main players: First Commercial Bank, Taiwan Cooperative Bank, and Taiwan Business Bank. Most of them are government-supported banks. The size of the market: TWD 3,243 billion Percentage of market size control by government-supported banks: 63.2% of market share. 8. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(16) . 2.7. Regulation of Banking and Credit in Taiwan Banking is a highly regulated industry. This industry is both governed by international regulation and local regulation. Banks are generally subject to minimum capital requirement based on an international set of capital standards, known as Basel Accords2. A bank is a financial institution that collect deposits from depositor and lend money to borrowers. Lending activities can be directly financed by bankers or indirectly financed through capital markets. Therefore, banks are highly regulated in Taiwan because its importance of. 政治大. financial stability.. 立. ‧ 國. Regulation of Banking and Credit in Taiwan. ‧. Executive Yuan. Financial Supervisory Commission. n. Foreign Exchane Business. Ch. Banks. engchi U. er. io. al. Financial Supervisory Commission Banking Bureau All banking busines. sit. y. Nat Central Bank of Taiwan. 學. Table 3 Regulation of Banking and Credit in Taiwan. v ni. Loan Interset Rate. Civil Law. Central Bank of Taiwan is another authority of regulation for banks in Taiwan because Taiwan is a highly regulated market especially for exchange rate of New Taiwan Dollar. Banks in Taiwan are required to follow Central Bank of Taiwan’s instruction toward foreign exchange business, such as trade loan, currency exchange or all foreign currency-related business. The purpose of such regulation is to maintain TWD’s exchange rate in a reasonable range to stabilize the economy. . 2. https://en.wikipedia.org/wiki/Basel_Accords. 9. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(17) . According to Civil Law, section 205, the maximum annualized loan rate is 20 percent. That is to say, the maximum loan margin is capped at 20 percent and the maximum loss is a hundred percent.. 2.8. Value Proposition Model The Value Proposition Model (Osterwalder, Pigneur, Bernarda, & Smith, 2014) is a tool which helps you examine whether your product or service creates the value that your customers are looking for.. 立. 政治大. ‧ 國. 學. The value of a product or service can be categorized into two basic types: gain creator and pain reliever. Gain creators produce outcomes or benefits that the customer expects, such as. ‧. functional utility, social gains, positive emotions, and cost savings. Pain relievers eliminate or. sit. y. Nat. reduce the unwanted happenings that customers experience before, during, or after they. io. n. al. er. accomplish a task or that prevent them from doing so.. i n U. v. A successful value proposition would find the fit between what a company offers and what. Ch. engchi. customers want. Therefore, it is very important to identify the target customers, what task they are trying to accomplish, their gains and the pains, and analyze whether your product or service will meet their needs.. 2.9. Platform Revolution As the book, Platform Revolution (Parker, Alstyne, & Choudary, 2016) suggests, the platform is a new business model that uses technology to connect people, organizations, and resources in an interactive ecosystem in which value can be created and exchanged. Platforms use data-based tools to create community feedback loops that traditional companies cannot. 10. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(18) . realize. Platforms like Amazon, YouTube, and Airbnb use feedback loops to gather community comments about the quality of content (on Amazon and YouTube) or the reputation of service providers (on Airbnb), increasing the efficiency and reliability of subsequent market interactions. Feedback from other consumers makes it easy to find books, or videos, or rental properties that are likely to suit their needs. They also mention network effects, which refers to the impact that the number of users of a. 政治大 large, well-managed platform community to produce significant value for each user in the 立. platform has on the value created for each user. Positive network effects refer to the ability of a. ‧ 國. 學. platform, which are the main source of value creation and competitive advantage in a platform. business. In the case of Uber, two sides of the market are involved: riders attract drivers, and. ‧. drivers attract riders. In the case of Airbnb, landlords attract tenants, and tenants attract. sit. y. Nat. landlords. The effects for network growth is so great that platform businesses will often spend. io. er. money to stimulate participants to one side of the market. If one side is attracted to join the platform, the other side will follow.. n. al. Ch. engchi. 11. i n U. v. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(19) . 3. Research Methodology To have a comprehensive understanding of OLP for SME and credit policy of DBS Bank (Taiwan), case study is the main research method of this research. The purpose of this research is exploring and comparing the current lending policy between OLP for SME and DBS Bank (Taiwan). Thus, by doing case study, the research will be able to compare the pros and cons as well as summary the current situation of OLP for SME and local lending policy of DBS.. 政治大. 3.1. Design. 立. The purpose of the research is to study the current impact of OLP for SME and to compare. ‧ 國. 學. the pros and cons with DBS’s lending process. Thus, the research deploys the comparative analysis of OLP and DBS Bank (Taiwan) as well as their lending decision process. Usually, the. ‧. quantitative research methods are carried out to address a specific hypothesis, while the more. sit. y. Nat. open-ended qualitative method are designed to allow novel, unexpected findings to emerge. al. n. learning from OLP.. er. io. (Bryman, 2012). This study exams if DBS Bank (Taiwan) could improve its lending policy by. Ch. engchi. i n U. v. Industry research is designed to benchmark the market condition for SME lending in Taiwan. By conducting the industry analysis and comparative analysis, the differences between OLP and DBS Bank (Taiwan) are found.. 12. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(20) . 3.2. Data Collection According to HBR (Karen Gordon Mills and Brayden McCathy, 2014), the major OLPs for SME are OnDeck, Kabbage, Fundera, Biz2credit, Enova, etc. Given the limitation of public information of privately held platforms, such as Kabbage, Fundra, Biz2Credit, this study selects OnDeck as a target to conduct analysis. This is because OnDeck is (i) a listed company in NYSE (NYSE:ONDK) and (ii) a platform specialized in SME lending through OLP. Table 4 Current Major OLP for SMEs Name. Listed (Y/N). OnDeck. Y. Kabbage. N N. USA. https://platform.biz2credit.com/. Y. USA. http://ir.enova.com/welcome. N. Taiwan. https://www.bznk.com/. N. Singapore. https://www.validus.sg/. al. n. 3.3. Data Analysis. y. sit. io. https://www.fundera.com/. Nat. Validus. USA. er. Bznk. N. ‧. Enova. https://www.kabbage.com/. ‧ 國. Biz2Credit. USA. 學. Fundera. 立. 政治大 Website USA https://www.ondeck.com/. Major Market. Ch. engchi. i n U. v. To study the difference between commercial bank and online lending platform, the research looks into four different dimensions, including decision flow of lending process, required data sets, technology platform, and logic of rating model.. 13. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(21) . 4. Case Study Case study enables this research to compare between traditional bank and online lending platform. By comparing lending policy, data collection, decision criteria and all stakeholders, this research is able to pin point the key differences of traditional banking system and online lending platform.. 4.1. DBS. 政治大. Company Background. 立. DBS Bank incorporated in 1968 by Government of Singapore. Today, DBS is the largest. ‧ 國. 學. bank in South East Asia in terms of assets and among the larger banks in Asia, with total assets. ‧. of SG$518 billion as of December 31st, 2016. DBS operates across 18 markets and around 280 branches around the world. DBS Bank provides full functions of financial service. The service. y. Nat. io. sit. includes consumer banking division (CBG), institutional banking division (IBG), treasury. n. al. er. markets and others. CBG provides consumer banking and wealth management. The product. Ch. i n U. v. and services include current and savings account, fixed deposits, loans and home finance, cards,. engchi. payments and investments. IBG focuses on broadening and deepening of customer relationships. Products and services comprises credit facilities, and working capital financing as well as specialized lending. Treasury markets include structuring, market making and trading across a broad range of treasury products. In FY2017, total income increased 4% to SG$11,924 million and net profit increased 4% to SG$4,390 million. Institutional banking division accounted for 44.2% of total income.. 14. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(22) . Table 5 Sales breakdown by segment of DBS in FY2017 (Unit: SGD million) Business Segment. Total Income (FY2017). % of sales (FY2017). CBG. $4,671. 39.2%. IBG. $5,275. 44.2%. Others. $1,978. 16.6%. Total. $11,924. 100%. Lending Decision Flow. 治政 In DBS, a lending decision is a complicated, time-consuming and subjective decision. 大立 Before building up a line of credit, a standard credit process usually requires two months to go ‧ 國. 學. through the whole credit decision flow. A longer lending decision flow meant a higher operating. ‧. cost to obtain a credit line. Additionally, within the lending decision flow, every step will. n. er. io. al. sit. Nat. requires sufficient communication to reach a group consensus.. y. include two or more stakeholder departments to involve in the decision of approval, which. Ch. engchi. 15. i n U. v. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(23) . 政治大. 立. ‧. ‧ 國. 學 er. io. sit. y. Nat. al. n. Figure 2 Credit approval process of DBS. Ch. Required Data Set for DBS Lending Decision. engchi. i n U. v. The lending decision for SME in DBS requires to collect various financial and non-financial data sets. Every front desk banker or credit analyst is required to review and analyze the data and information based on personal experiences and subjective judgement. Since this process includes many subjective interpretations, any conclusion and any determination might be biased by different subjective standards.. 16. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(24) . 立. 政治大. ‧. ‧ 國. 學 sit. y. Nat. al. n. Technology Platform. er. io. Figure 3 Required Data Set for DBS Lending Decision. Ch. engchi. i n U. v. To attract a large amount of SMEs customers in Taiwanese market, DBS has launched a technology platform and new credit product, 星展中小企業星易貸3. By shortening lending decision to 5 days, DBS would like to differentiate its market positioning as a faster and easier bank from a complicated and time consuming traditional bank. The platform requires the borrower to fill out an application with their basic company background. However, the lending process is still the same. To approve a line of credit within 5 working days is more like a marketing slogan. According to the platform, eleven information are required to be filled out 3. https://www.dbs.com.tw/sme-zh/dbs-forms/microloan-form.page. 17. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(25) . by applicants, including (1) Company name; (2) Tax ID; (3) Registered City; (4) Type of industry; (5) Existing company or not; (6) Years of establishment; (7) Business scale; (8) Contact person; (9) Mobile phone; (10) TEL and (11) Mail address. The key difference of this lending program is that it focuses on domestic company with tax ID and it narrows down a target company’s sales within TWD800mn so that bankers are able to assess credit risk with an easier and faster method.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Figure 4 The work flow of 中小企業星易貸. 18. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(26) . The logic for rating model of 中小企業星易貸 (“the Program”) A risk rating is given as a benchmark for each target customer. By using the risk rating, bankers are easier to know the default risk so that bankers are able to shorten credit process by using less required information. That is to say, bankers could know the credit risk based on credit rating instead of subjective and collective lending decision. Every credit rating indicated a certain default risk. Thus,. 政治大. bankers are able to determine pricing quotation easier and facility size faster given a certain risk. 立. rating.. ‧ 國. 學. For the Program, DBS adopted joint credit information center’s (JCIC) J21 as an external. ‧. risk benchmark. Together with some in-house pre-screening checking criteria, bankers could. sit. y. Nat. assess credit risk easier.. n. al. Risk Range <=1420. er. io. Table 6 The Default Rate of a given J21 Scoring as of 2012. The Default Chance of J21. Ch. Default rate. engchi. 33.49%. i n U. v. Accumulated Sample Size 0.4%. >1420-1460. 13.41%. 1.0%. >1460-1500. 9.34%. 2.6%. >1500-1540. 4.09%. 7.0%. >1540-1580. 1.97%. 14.9%. >1580-1620. 0.77%. 30.8%. >1620-1640. 0.39%. 52.7%. >1660-1700. 0.12%. 71.1%. >1700-1740. 0.06%. 87.3%. >1740-1780. 0.04%. 96.1%. >1780-1820. 0%. 98.7%. >1820. 0%. 100.0%. 19. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(27) . Source: Join Credit Information Center According to historical non-performing loan (NPL) rate of SME borrowers in DBS Bank (Taiwan), the non-performing loan against total loan outstanding was 2.19%. (DBS, Annual Report, 2017). The number not only perfectly equaled to possibility of default in terms of number of customers in DBS Bank (Taiwan) SME segment but it also indicated how this bank managed its portfolio. The total non-performing loan against total loan outstanding was around. 政治大 portfolio was higher than the total NPL of DBS Bank (Taiwan). Given the fact, we can interpret 立 1.08% in DBS Bank (Taiwan) in FY2017. Based on the above two figures, NPL for SME. ‧. ‧ 國. segment.. 學. that DBS Bank (Taiwan) had an inferior customer portfolio or poor risk management in SME. Therefore, to compare non-performing loan rate in DBS Bank (Taiwan) , 2.19%, with. Nat. sit. y. potential default rate with JCIC scoring, DBS could improve risk management while. n. al. er. io. implementing JCIC scoring into lending decision. The JCIC 21 scoring of 1580 or above. i n U. v. represented an acceptable risk given 0.77% of default rate for the given risk range.. Ch. engchi. 20. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(28) . 4.2. On Deck Capital (“OnDeck”) On Deck Capital was incorporated in 2006 based in U.S.A. OnDeck had underwritten more than USD7 billion in loans to SME in US, Canada and Australia. These loans were granted based on short-term working capital needs. By investing in the platform, OnDeck has created a differentiated small business lending experience that is simple and fast. OnDeck used automation and advanced analytics on small business lending. The data points include bank activities shown on the customers’ bank statements, business and personal credit bureau reports,. 政治大. government filings, tax and census data. In addition, OnDeck also analyzed reputation and. 立. social data. (OnDeck, 2017). ‧ 國. 學. Thus, small business are able to apply for a term loan or line of credit, 24 hours a day, 7. sit. y. Nat. approved, the fund could be transferred as fast as the same day.. ‧. days a week from OnDeck’s website. OnDeck made a lending decision immediately and, if. n. al. er. io. As of December 31, 2017, 3% of loans were funded via OnDeck’s online platform. i n U. v. purchasers, 58% of loans were financed via OnDeck’s facilities, 30% of loans were funded via. Ch. engchi. proceeds raised from OnDeck’s securitization transaction and 9% of loans were financed via OnDeck’s owned capital. For FY2017, OnDeck reported USD334.6 million of sales and USD11.5 million of net losses.. 21. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(29) . Customer Profile As of December 31, 2017, OnDeck’s customers had a median annual revenue of approximately USD631,000, with 90% of customers ranging from USD162 thousand to USD3.9 million in revenue and have been in business for a median of 8 years, with 90% in business between 2 and 30 years. During 2017, the average size of a term loan was approximately USD57 thousand and the. 政治大. average size of a line of credit extended to OnDeck’s customers was approximately USD23. 立. thousand. Based on OnDeck’s annual report, the net interest margin has been at 29.7% for the. ‧. ‧ 國. 學. io. sit. y. Nat. n. al. er. past three years.. Ch. engchi. 22. i n U. v. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(30) . Lending Decision Flow OnDeck is an online working capital finance provider. According to most of accounting textbook, the definition of working capital means current assets deduct current liabilities. For the lending decision flow of Ondeck, four types of information is required, including purpose of loan, company background, and financial information and banking relationship.. ‧. n. 8. Business address 9. Zip code. y. basis. name. Ch. term financing. balance on daily. 7. Legal business. al. 17. Any other short-. 16. Average bank. 6. E-mail address. io. needed. annual revenue. number. Nat. money is. relationship. 15. Estimated gross. 5. Telephone. Banking. sit. 立. 2. Size of a loan. 3. How soon the. 政治大. 學. loan. Financial information. 4. Name. ‧ 國. 1. Purpose of a. Company background. er. Purpose of a loan. engchi. i n U. v. 10. Business phone 11. Years of business 12. Legal entity type 13. Where the business incorporated 14. Business tax ID. 23. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(31) . Regarding applicant’s personal information, the below information is required. 1.. Name. 2.. Home address. 3.. Zip code. 4.. Home phone. 5.. Date of birth. 6.. Percentage of business you own. 7.. Social security number. 立. 政治大. ‧ 國. 學. OnDeck reviewed not only a promoter’s personal credit but also a borrowing company’s business condition. Sometimes, OnDeck required additional information, such as most recent. ‧. three-month bank statements.. sit. y. Nat. io. er. By using the above information, a company is able to apply for a working capital loan within a few minutes and receive funding within one to two calendar days.. n. al. Technology Platform. Ch. engchi. i n U. v. Unlike traditional commercial bank, OnDeck is a purely branchless and online lending platform. Borrowers are required to register and input their basic information by themselves, which enabled OnDeck to reduce cost of employees while doing data input. Besides this cost advantage, OnDeck was able to transfer operational risk of data input from employees to borrowers. OnDeck could also lower its operating cost in renting a costly office in a business area by approving commercial loan online.. 24. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(32) . Given the business nature of OnDeck, OnDeck sourced capital through multiple channels, including debt facilities, securitizations, and OnDeck marketplace, a platform where enabled individual investors to invest with their money. With banks increasing digitalized business, the opportunity to market OnDeck-as-a-Service platform to banks has increased. OnDeck’s platform touched every aspect of the customer life cycle, including customer acquisition, sales, scoring and underwriting, funding and servicing and collections.. 政治大. The logic for rating model of OnDeck. 立. ‧ 國. 學. OnDeck Scores, the rating model of OnDeck, analyzed thousands of data points from dynamic, disparate data source, and the relationships among those attributes, to assess the credit. ‧. worthiness of small business rapidly and accurately. According to OnDeck’s annual report, the. sit. y. Nat. data points include customer bank activity shown on their bank statements, business and. io. er. personal credit bureau reports, government filings, tax and census data. Besides data points above, OnDeck also analyze reputation and social data under certain circumstances. OnDeck. al. n. v i n Score enable borrowers to have aCfunding, a day, 7 days a week on OnDeck website h e n24g hours chi U in minutes.. For a lending decision, OnDeck Score planned to achieve target 6 to 7 percent provision rate range. In addition, the gain on sale of loans through OnDeck Marketplace has declined from $14.4 million, or 5.0% of gross revenue in 2016 to $2.5 million, or 0.7% of OnDeck’s gross revenue in 2017 reflecting a lower percentage of OnDeck’s term loans sold into OnDeck Marketplace.. 25. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(33) . 5. Results and Analysis In this section, the research conducts cross-case analysis. By comparing the difference of SME lending in DBS Bank and online lending platform, OnDeck, we could understand the advantages and disadvantages of these two systems.. 5.1. Cross Case Analysis In this section, this research summarized the key differences of DBS local lending program. 政治大. and OnDeck.. 立. DBS Local Lending Program. Dates required to. In few minutes. y. Nat. io. provide. borrowers but DBS Bank. al. n. Data required to. Ch. required bankers to assess. engchi. sit. 11 information input by. er. a line of credit. In 5 days. ‧. obtain the approval of. OnDeck. 學. ‧ 國. Item. 24 information input by. v iborrowers n U. more data externally.. Median: USD631 thousand Borrowers range from USD300 The size of a borrower thousand to USD16 million.. More than 90% of borrowers range from USD162 thousand to USD3.9 million.. 26. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(34) . Average size of Short-term loan: Size of a loan ranges from. USD23 thousand. The size of line of USD400 thousand to USD1.0 credit. Average size of long-term loan:. million.. USD 57 thousand 9% from OnDeck’s owned capital;. 政治大 30% from OnDeck’s. 立. Bank’s balance sheet. 58% from OnDeck’s facilities;. ‧. 3% from online platform. sit. investors.. n. al. 2.19%4. debts Net interest margin. er. io. or provision for bad. y. Nat. Average rate of NPL. securitization transaction;. 學. ‧ 國. The source of capital. Ch. engchi. 3%-5%. v to 7% provision rate i6% n U 29.7%. 4. According to DBS Bank (Taiwan) annual report of FY2017, DBS Bank (Taiwan) recognized TWD894 million. of NPL out of TWD40,782 million of secured loans.. 27. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(35) . The First Loan: 2.5%-4% of loan amount; The Second Loan: 1.25%-3% of Facility fee. 1%-1.5% of loan amount loan amount; After the Third Loan: 0-3% of. 立. 政治大. loan amount. 5.2. Analysis and Findings. ‧ 國. 學. Based on this research, OnDeck had a better efficiency in lending decision but the size of borrower and size of a loan were smaller. Compared to DBS Local Lending Program, OnDeck. ‧. has a higher NPL or provision rate. However, OnDeck also applied for a higher pricing for both. Nat. sit. y. facility fee and net interest margin. Traditional banks remain competitive in this coming. er. io. revolution because commercial banks and online lending platforms targeted different markets.. al. n. v i n Efficiency and pricing wereCa h trade-off. This research e n g c h i U found that the worst efficiency. translated into a lower default rate or provision rate. Meanwhile, OnDeck charged higher. interest rate to borrowers and it has better efficiency. Size of a facility is also associated with efficiency and risk management. With better risk management or better KYC (know-yourcustomer), lenders were willing to grant a larger size of facility. This study found that OnDeck has better efficiency because it adopts fully automated lending decision and collects data from borrowers or public information. Comparing to DBS Local Lending Program, when certain degree of automation is applicable, validation or. 28. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(36) . information input by bankers are still required. Besides, OnDeck used its own algorism to grade its borrower instead of referring risk rating to external reference. Benefited by less humandriven decision, OnDeck has better efficiency than DBS. This research also found that small-sized company suited online lending platform better because most of the time small-sized companies ask for simple working capital loan. As a company grows bigger, it requires a more structured product, which online lending platform. 政治大 merger and acquisition or other corporate finance related deals. Therefore, small-sized 立. cannot provide. Those products include documentary trade service, letter of credit, guarantee,. ‧ 國. 學. companies with stable and recurring cash flow are the optimal target customer for online lending platform.. ‧. An automated lending system is under development by DBS. By learning from OnDeck,. Nat. sit. y. DBS could focus on small-sized business and try to implement automation for its lending. n. al. er. io. decision. To reach a better balance of risk and return, DBS could increase interest charge or. i n U. v. facility fee for small-sized customers, who utilize DBS Online Lending Program.. Ch. engchi. 5.3. Keys to improve efficiency of DBS local lending program The key difference of DBS local lending program and OnDeck Scoring was the degree of automation of a lending process. OnDeck Scoring required fewer data input by borrowers and mainly relied on algorithm to do lending decision. However, OnDeck has to bear a higher provision rate and borrowers are required to pay a higher net interest margin compared to lending from DBS.. As a lender, it is very difficult to balance risk and efficiency given. information asymmetry. Even though OnDeck is able to approve a loan in minutes, OnDeck had not been profitable since establishment due to NPL.. 29. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(37) . 6. Research summary In this research, a cross case analysis of lending decision process of DBS Local Program and OnDeck was conducted. Four key findings are listed as below. Data driven lending decision. By comparing DBS Local Lending Program and OnDeck, this research found both bank and online lending platform adopted a certain degree of automation. This research discovered that more and more information are to. 政治大. be input manually directly by borrowers or could be automatically retrieved from public information.. 立. 學. ‧ 國. Reducing labor cost and relying less on subjective decision. In the traditional. ‧. lending decision, bankers have to validate and analyze all information and then bankers submit their analysis to credit approvers. That is to say, the traditional lending. y. Nat. io. sit. decision takes a long time to reach consensus by maker and checker. Also, the process is time-consuming and detail-oriented. Comparing to traditional lending decision. n. al. er. 1.. Ch. i n U. v. process, lending by program costs less and has better efficiency.. engchi. Data integrity becomes important. In the traditional lending process, experience of individuals is more important than data quality but this situation is different from the online lending platforms or lending decision in the future. By adopting automation and relying on data inputted by applicants, how to validate data cleansing and maintain clean database becomes a new topic in new the current lending decision. With less subjective decision and higher degree of automation, clean and correct database becomes a key issue of lending decision.. 30. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(38) . 2.. Risk and efficiency are still difficult to balance. Information asymmetric is the most challenging topic while doing a commercial loan. Bankers are able to mitigate credit risk by doing deeper due diligence with borrowers or giving a tailor-made product. Nevertheless, to some extent, a better risk control means to sacrifice efficiencies. Credit charge could reduce gradually. To balance risk and efficiency, OnDeck charged a relatively high interest expense and facility fee to borrowers. However, the credit. 政治大 second year. This design deters borrower from using this platform year after year 立. charge have no significant difference when borrowers renew their line of credit in the. ‧ 國. 學. when the borrowers have accumulated some track records in the platform.. Learning from traditional banks, where the credit charge is determined by either. ‧. borrowers’ credit risk or borrowers’ bargaining power in loan market. To keep. Nat. sit. y. customers who have good repayment record in OnDeck, OnDeck could learn from. n. al. er. io. traditional banks. That is to say, credit charge could reduce slightly in the second year. i n U. once credit rating and other conditions are status quo. 3.. Ch. engchi. v. Online lending platforms focus on small-scale businesses. With a large number of small businesses, the cost of acquiring a new customer is relatively high but return on acquiring a new customer is relatively low. Thus, online lending platform diversified credit risk by increasing number of customers. Besides, small-scale business only requires less complicated product. Working capital loan is sufficient for small-scale business to expand business or maintain daily operation. Thus, small-scale business is more suitable for online lending platforms or program lending. 31. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(39) . 4.. DBS Local Lending program could focus on smaller-sized business. Given around 1.4 million of small and median-sized corporations in Taiwan, DBS Bank (Taiwan) could consider to redesign the segmentation and try to apply DBS Local Lending Program for smaller sized companies, such as those whose sales is within TWD50 million. The smaller-sized borrowers enables DBS Bank (Taiwan) to manage risk exposures at. 政治大 of borrowers. The automation of lending decision is feasible when the program is able 立 a smaller amount and reach economy of scale by increasing and diversifying number. ‧ 國. most suitable customer segment for program lending.. 學. to make the same decision as credit approvers and the small-sized borrower is the. ‧. 6.1. Research contribution and limitation. Nat. sit. y. Commercial loan means to put money or resource to a right place. Finding a right company. n. al. er. io. and a proper way to put money are the key to make a loan decision. With the rising discussion. i n U. v. on Fintech, bankers are also thinking of using data to improve efficiency. However, after doing. Ch. engchi. this research, due to information limitation for OnDeck, there is room for further study as below. 1.. Why does OnDeck fail to become profitable?. 2.. What is OnDeck’s mythology of rating algorithm?. If this research could have a deeper understanding about OnDeck Score’s rating algorism, this research could have more contribution on lending decision automation for DBS Bank (Taiwan) or all online lending platforms. In the next phase of the research, the efficiency could be further examined by statistical test.. 32. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(40) . 7. Reference BrymanA. (2012). Social Research Methods. New York: Oxford University Press Inc. Chen, D., and Han, C. . (2012). A Comparative Study of Online P2P Lending in the USA and China. Journal of Internet Banking and Commerce . DBS. (2017). Annual Report. Taipei: DBS Taiwan.. 政治大. DBS. (2018). 中小企業星易貸 . https://www.dbs.com.tw/sme-zh/dbs-forms/microloan-. 立. form.page. ‧ 國. 學. Karen Gordon Mills and Brayden McCathy. (2014). The State of Small Business Lending:. ‧. Credit Access during the Recovery and How Technology May Change the Game.. y. Nat. er. io. sit. Harvard Business School.. LinKeng-han. (2016). Reserach on Taiwan commercial bank's detection of financial fraud in. n. al. Ch. engchi. i n U. v. China: the use of public information. 國立政治大學. OnDeck. (2017). Annual Report. OsterwalderAlex, PigneurYves, BernardaGreg, & SmithAlan. (2014). Value Proposition Design. Hoboken, New Jersey: John Wiley & Sons, Inc. ParkerG.Geoffrey, AlstyneW. VanMarshall, & ChoudaryPaulSangeet. (2016). Platform Revolution. New York: W. W. Norton & Company, Inc. Small and Medium Enterprise Administrationof Economic AffairsMinistry. (2015 年 3 月 30. 33. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(41) . 日). Standards for Identifying Small and Medium-sized Enterprises. Small and Medium Enterprise. Administration,. Ministry. of. Economic. Affairs:. https://www.moeasmea.gov.tw/ct.asp?xItem=8261&CtNode=530&mp=2 Wikipedia. (2018). Basel Accords. Wikipedia: https://en.wikipedia.org/wiki/Basel_Accords YenChiLi. (2017). JKOS Pay: Challenges of Marketplace Adoption in the Mobile Payment Space. 國立台灣大學.. 立. 政治大. Zopa. (2018). https://www.zopa.com/.. ‧ 國. 學. 林妙宜. (2001). 公司信用風險之衡量. 國立政治大學.. ‧. 林博仁. (2016). Loan default prediction：analyzing P2P lending data on the spark platform.. y. Nat. er. io. sit. 台北: 國立政治大學.. 梅驊 . (2016). The Discussing of the Future Development of P2P Lending in Taiwan by. n. al. Ch. engchi. i n U. v. Analyzing the Worldwide P2P Industry. 台北: 國立台灣科技大學. 經濟部中小企業處. (2017). 2017 中小企業白皮書. 經濟部.. 34. DOI:10.6814/THE.NCCU.IMBA.004.2019.F08.

(42)