Trust in financial products and institutions is widely recognised as being essential for financial markets to function efficiently. For example, since the very origins of banking, trust has played a foundational role in banks with regard to their safekeeping and depository functions. Recently, financial markets have evolved, with the rapid emergence of financial technology (fintech) firms which engage in many of the same credit functions that banks perform; indeed, many have argued that these new fintech firms may displace the traditional role of banks in financial markets (He et al. 2017). However, even with these new non-bank firms, the concept of trust has emerged in policy discussions regarding the potential impact of non-intermediated credit on banks and the credit market, with practitioners and market participants understanding that trust is important in enabling fintech firms to compete with banks.
In a recent paper, we theoretically analyse how the notion of trust may affect credit markets, and we argue that trust in financial institutions may have a first-order impact on whether non-bank (fintech) firms can survive when competing against traditional banks (Thakor and Merton 2018). The starting point of our analysis is the observation that the lending functions that banks and non-bank fintech firms perform are the same – both provide debt financing to clients. Our theory then rests upon three main building blocks. First, banks raise significant financing through deposits, and depositors are customers who want their services insulated from the bank’s credit risk (Merton 1993, 1995, 1997, Merton and Thakor 2018). Second, non-bank/fintech lenders are ‘all-equity’ financed (Philippon 2015, 2016). Third, trust in lenders – which we model via a belief updating process that is partly Bayesian and partly non-Bayesian – interacts with these features.
Two dimensions of trust
Trust has two dimensions: (i) being trustworthy, and (ii) being competent. Trustworthiness is about intent, whereas competence is about skills. A completely trustworthy entity that is incompetent can make bad decisions that are as ruinous as those made by an entity that is untrustworthy. Our focus is on the first dimension of trust, so we assume that banks and fintech lenders are equally skilled in collecting and processing relevant credit information about loan applicants. In part, our view is that in the real world, technology is a competitive industry, and any entity can buy the information technology or hire the people it needs to best perform credit analysis. Indeed, while fintech firms may have introduced new technology into the marketplace, big banks have enormous capital resources that other smaller players do not, which allows them to eventually adopt the new technology. Thus, there is no a priorireason to expect one kind of lender to have a technological advantage over another in the long run.
Modelling trust in lenders
More specifically, our modelling of trust utilises the ‘hypothesis testing representation’ (HTR) framework from Ortoleva (2012). Agents operate under one of two ‘paradigms’ or models of the world. In Model I, a lender is viewed as trustworthy, in which case it will always choose to make a good loan. In Model II, a lender is viewed as self-interested, in which there is the possibility that it may choose to make an inefficient (private-benefit) loan. Uncertainty about the correct model of the world (i.e. is the lender trustworthy or self-interested?) is captured by a prior over priors, while within-model uncertainty (if self-interest, is the lender still worth financing?) can be viewed as a reputational effect and is captured by beliefs that are revised in a normal Bayesian manner. We initially assume that all agents adopt Model I as the model of the world, in which case they trust that lenders will always choose to make a good loan – in other words, they do not contemplate the possibility that lenders will make bad loans. A reason for trust being placed in these types of lenders in the first place is likely due to a combination of the technologies being opaque and yet working well to begin with. Indeed, if such technologies were transparent, then there may be no need for trust for them to be adopted.
Trust can be either maintained or lost, depending on the realisation of loan outcomes. In particular, if a borrower repays the lenders, then trust will be maintained. If a borrower defaults on a loan, then whether trust is lost depends on the state of the economy. If the state of the economy is relatively weak, then agents will attribute the loan default to economic conditions (since a worse economic state lowers the probability of default) and not to the lender’s choice of a bad loan. In this case, trust will be maintained, and the lender’s funding costs will appear insensitive to defaults, thus insulating lenders from the adverse reputational consequences of loan defaults and making the pricing of credit seem disassociated from risk in environment. However, if a borrower defaults and the economy is strong, then this may constitute strong evidence that is inconsistent with Model I being correct. Thus, trust may be lost as agents switch to Model II. The result is a sharp and discontinuous rise in funding costs for the lender. We also show that trust is asymmetric in the sense that it is harder to regain than it is to lose – once trust is lost, good loan outcomes will be assigned to reputation, making it difficult to shift from Model II to Model I.
We believe that this framework of trust is consistent with a number of facts from the financial crisis. For example, Gorton and Metrick (2012) document that the average haircut on bilateral (non-US Treasury) repo transactions rose from zero in early 2007 to almost 50% at the peak of the crisis in late 2008, with several classes excluded entirely from being used as collateral. Similarly, Iyer et al. (2013) document an unexpected and sudden freeze of the European interbank market in August 2007. These are examples of discontinuities in pricing where agents believed in a particular model of the economic environment and then, faced with unexpected news, switched to a different model.
What happens when trust is lost?
With this framework of trust in hand, we show that, conditional on having lost trust (Model II), banks have a stronger reputational incentive to make good loans. This stems from banks’ access to insured deposits, and customers’ willingness to provide deposits at a below-market interest rate, which gives banks a valuable funding advantage and higher profitability that generates a stronger incentive to maintain that advantage. Financiers see that banks have these incentives, while fintech firms do not. That is, financiers recognise the stronger reputational incentives of banks to make good loans in an environment in which lenders are not unconditionally trusted and are viewed as self-interested. Thus, when trust is lost and reputation becomes important, the cost of funding rises more for fintech firms than for banks. For fintech firms, this reduces their profits from good lending enough to make bad lending seem more attractive. Investors will therefore not finance fintech firms without trust. This means trust is essential for fintech lenders to operate – banks may be able to survive a loss of trust, but fintech lenders will be forced to shut down. This suggests that the popularity of fintech firms may be halted if there is a major event that erodes trust.
In our model, the key difference between banks and fintech firms is on the funding side. We focus on this institutional difference because we believe it is of first-order importance in developing a theory of trust in lending. There are other institutional features, to be sure, such as differences in regulation and information gathering processes. However, introducing these differences will only strengthen our results. Prudential regulation of banks – like regulatory supervision and capital requirements – is intended to increase trust in banks, so unless one believes that all regulatory costs are a net waste of resources, introducing this feature is likely to make banks more trustworthy. Further, since both banks and fintech firms have access to the same technology but banks also have access to soft information about borrowers in their role as relationship lenders, introducing this feature will give banks a competence advantage over fintech lenders, so they will dominate them on both dimensions of trust: trustworthiness and competence.
Implications for informational transparency
Our analysis of trust also has interesting implications for informational transparency related to banks. If investors can understand the nature of information (e.g. loans), then information disclosure may substitute for trust. However, if loans are ‘inherently opaque’ (i.e. cannot be made transparent), then transparency is not a feasible substitute for trust. This provides a different perspective on bank opacity from that of earlier research. For example, Bhattacharya and Chiesa (1995) propose that banks choose to be opaque in order to protect their borrowers’ information. Dang et al. (2017) argue that banks are opaque because disclosing information would expose depositors to unwanted risk. Our point differs in that, when information can be efficiently processed by investors, trust and transparency are partial substitutes and the market will segment into trusted lenders who are opaque and other lenders who are transparent. But when information is costly to process, the value of trust is high, and all lenders will remain opaque, with trusted lenders having a significant funding cost advantage.
Our focus on trust is complementary to a growing literature – both empirical and theoretical – which examines the role of trust in financial markets (e.g. Guiso et al. 2008, Gennaioli et al. 2015a,b). We differ in our focus on the role of trust in determining the cost and availability of financing to lenders, and the resulting nature of competitive interactions between banks and fintech lenders. Our modelling of trust also allows us to distinguish between trust and reputation, and to provide new insights such as how trust is asymmetric in the sense that it is harder to re-gain than it is to lose.
References
Bhattacharya, S, and G Chiesa (1995), "Proprietary information, financial intermediation, and research incentives", Journal of Financial Intermediation 4(4): 328-357.
Dang, T V, G Gorton, B Holmström, and G Ordonez (2017), "Banks as secret keepers", American Economic Review 107(4): 1005-29.
Gennaioli, N, A Shleifer, and R Vishny (2015), "Money doctors", The Journal of Finance 70(1): 91-114.
Gennaioli, N, A Shleifer, and R Vishny (2015), "Neglected risks: The psychology of financial crises", The American Economic Review 105(5): 310-314.
Gorton, G, and A Metrick (2012), “Securitized banking and the run on repo”, Journal of Financial Economics 104(3): 425-451.
Guiso, L, P Sapienza, and L Zingales (2008), "Trusting the stock market", The Journal of Finance 63(6): 2557-2600.
He, D, R Leckow, V Haksar, T Mancini-Griffoli, N Jenkinson, M Kashima, T Khiaonarong, C Rochon, and H Tourpe (2017), "Fintech and Financial Services: Initial Considerations", IMF Staff Discussion Note: Fintech and Financial Services: Initial Considerations 17, no. 05.
Iyer, R, J-L Peydró, S da-Rocha-Lopes, and A Schoar (2013), "Interbank liquidity crunch and the firm credit crunch: Evidence from the 2007–2009 crisis", The Review of Financial Studies 27(1): 347-372.
Merton, R C (1993), “Operation and Regulation in Financial Intermediation: A Functional Perspective”, in P Englund (ed.), Operation and Regulation of Financial Markets, The Economic Council, Stockholm: 17-67.
Merton, R C (1995), "A Functional Perspective of Financial Intermediation", Financial Management 24(2): 23-41.
Merton, R C (1997), "A Model of Contract Guarantees for Credit-Sensitive, Opaque Financial Intermediaries", European Finance Review 1(1): 1-13.
Merton, R C and R T Thakor (2018), "Customers and investors: a framework for understanding the evolution of financial institutions", Journal of Financial Intermediation.
Ortoleva, P (2012), "Modeling the change of paradigm: Non-Bayesian reactions to unexpected news", The American Economic Review 102(6): 2410-2436.
Philippon, T (2015), "Has the US finance industry become less efficient? On the theory and measurement of financial intermediation", The American Economic Review 105(4): 1408-1438.
Philippon, T (2016), “The fintech opportunity”, NBER Working Paper No. 22476.
Thakor, R T and R C Merton (2018), “Trust in Lending”, NBER Working Paper No. 24778.
