«Alejandro Drexler, Antoinette Schoar ∗ This version August, 2011 Abstract This paper provides evidence that shocks to the relationship between loan ...»
Does Soft Information Matters? Evidence From Loan Officer
Alejandro Drexler, Antoinette Schoar ∗
This version August, 2011
This paper provides evidence that shocks to the relationship between loan oﬃcers
and their borrowers aﬀects the credit decisions of the bank as well as customers’
repayment and borrowing behavior. When a loan oﬃcer unexpectedly has to be
absent from the job, the existing borrowers of the absent loan oﬃcer are less likely to take on a new loan from this bank and are more likely to miss a payment. The reduction in the borrowing is explained by a lower number of loan applications and a reduction in the application approval rate. This ﬁndings suggest that clients are loyal to their loan oﬃcer, that the bank reduces lending when soft information is less available, and that loan oﬃcers have an important role in monitoring the clients.
Alejandro Drexler is at the Mccombs School of Business at The University of Texas at Austin, Antoinette ∗ Schoar is at MIT, NBER, and ideas42. We are grateful for the comments and suggestions of the participant in the seminar at The University of Texas at Austin, and Universidad Catolica de Chile, specially Daniel Paravisini for a great discussion. We thank Bank Estado, especially Roxana Aravena, Jose Luis Arriagada, Pablo Coto, Enrique Errazuriz, Carlos Hernandez, Soledad Ovando, Hector Pacheco, Oliver Prostran, Marco Sambuceti, German Texido, Emilio Velez, Victor Vera, and Pamela Zalduando for providing the data and making us familiar with the internal HR processes of the bank. We also thank Manasee Desai, and Katherine Gordon for their help collecting and organizing the data. Finally we thank Rouzhna Nayeri for excellent work in editing the paper. All remaining errors are our own.
Introduction Most credit programs are based on extensive interactions between loan oﬃcers and the businesses they lend to. This relationship based approach to lending is especially widespread for small and more opaque borrowers, where formal documentation of proﬁts and record keeping is less reliable. The loan oﬃcer has the diﬃcult role of solving the informational gap between the bank and the borrower by gathering soft information about potential borrowers.
The relationships between loan oﬃcers and their clients often extends beyond information collection, and many times loan oﬃcers help borrowers assessing the ﬁnancial needs of their business or even help ensuring that clients repay. The importance of relationship lending has been proposed in a myriad of theory papers, see for example Rajan (1992), Petersen, and Rajan (1994), Petersen, and Rajan (1995), Berger, and Udell (2002), Berger, Miller, Petersen, Rajan, and Stein (2005). However, there has been only little empirical research to document the role of loan oﬃcers in mitigating information asymmetries or moral hazard between the bank and its clients. A few notable exceptions are Herzberg, Liberti, and Paravisni (2010), and Liberti, and Mian (2009).
The novel contribution of this paper is that we study (exogenous) shocks to the loan oﬃcer-client relationship: Their impact on the credit provision to borrowers as well as the borrowers’ behavior. Speciﬁcally the shocks we rely on are loan oﬃcer absentee spells due to sickness, pregnancy, resignation or layoﬀs. We work with a bank in Chile, BancoEstado, which lends to small businesses in the informal sector where credit screening relies mostly
on the loan oﬃcers but also on the entire client portfolio each loan oﬃcer manages (client characteristics and repayment borrowing behavior).
Overall we ﬁnd that loan oﬃcer absenteeism leads to signiﬁcant changes in the borrowing and repayment behavior of client and the credit provision of the bank. In particular, when the original loan oﬃcer is absent we observe a 0.9% reduction in the probability of taking up a new loan from the bank (13% reduction as a fraction of the unconditional probability of taking up a new loan from the bank). This reduction is explained by changes in both the client application rate, and the bank approval rate. Speciﬁcally, on the client side, the application rate decreases in 0.68% (9% reduction as a fraction of the unconditional probability of applying for a new loan), and on the bank side the approval rate per application decreases in 4.3% (5.2% reduction as a fraction of the unconditional probability of approving a loan application). This switch in credit access is particularly interesting since we do not see a change in credit terms after a loan oﬃcer leaves, e.g. interest rates and loan maturity is unchanged on average. This is of course contingent only on the borrowers who do choose to take up a new loan. The eﬀects on repayment behavior and borrowing outside the bank are diﬀerent for diﬀerent type of leaves. For example while a steep increase in borrowing outside the bank is observed after a loan oﬃcer gets sick, this reaction is not signiﬁcant when the absenteeism is related to a loan oﬃcer pregnancy, layoﬀ or resignation.
The fact that reactions to diﬀerent type of absentee spells are diﬀerent is expectable.
resignation might be correlated with the prior performance of the loan oﬃcer’s portfolio.
Laid oﬀ loan oﬃcers might be let go due to the particularly poor performance of their portfolio; while resigning loan oﬃcers might be poached away by competitors due to their above average skills or performance. Pregnancies diﬀer from the other absentee spells in that there is a long lead time which allows the bank and the loan oﬃcer to prepare the clients for the loan oﬃcer’s leave in order to prevent potential problems. Therefore the most exogenous source of absenteeism in our sample are major sickness periods of loan oﬃcers. These spells are largely unexpected for both the bank and the loan oﬃcer, and are independent of the loan oﬃcers’ portfolio characteristics.
We therefore separately study the eﬀects of the diﬀerent absentee spells on the loan oﬃcer’s client portfolio. When only looking at sickness spells, we ﬁnd that clients whose loan oﬃcer has to take a sickness leave are 1.2% less likely of renewing their loan with the bank during the months that the original loan oﬃcer is on leave (19% as a fraction of the unconditional probability of renewing the loan). These clients also show a 2.1% increase in the probability of borrowing outside the bank (13% increase as a fraction of the unconditional probability of borrowing outside the bank), and an increase of 1.1% in the probability of missing a payment (10% as a fraction of the unconditional probability of missing a payment). Interestingly, when looking at the credit portfolio of loan oﬃcers who were ﬁred we see a much stronger drop in the likelihood of starting a new loan, a spike in
that clients of loan oﬃcers who were resigned (in most cases because they were hired away) do not see a change in their loan renewal probability. But they do experience an increase in missed payment. And ﬁnally, loan oﬃcers who leave due to pregnancy see no increase in missed payments. However, these clients show a drop in loan renewals during the time of the loan oﬃcer’s leave. This drop in loan renewal rates during pregnancy leaves is almost entirely explained by a reduction in the number of loan applications, we conjecture that this decrease in the likelihood of applying for a new loan might be a form of ‘loyalty’ by the clients, who wait for the new loan until their loan oﬃcer is back from maternity leave.
Overall these results suggest that the relationship between loan oﬃcers and their clients has ﬁrst order eﬀects on the borrowing behavior and the access to credit.
We also investigate whether there is an interaction eﬀect between the characteristics of the borrowers in the loan oﬃcer’s portfolio and the eﬀect of loan oﬃcers leaving. In particular we are interested in client characteristics that proxy for the importance of soft information for the lending decision, such as credit score and average loan sizes of the borrower prior to the current loan, and length of the relationship between the loan oﬃcer 1 See also Hertzberg, Liberty, and Paravisini (2010), who show that incoming loan oﬃcers have strong incentives to report bad news about the portfolio of a predecessor loan oﬃcer. While in our set up the ﬁred loan oﬃcers could not suppress information about non payment they could have manipulated default rates by renewing loans for clients that are experiencing economic distress.
Looking at the interaction eﬀects for sickness leaves we observe that ﬁrms oﬀset the reduction in lending by borrowing from other banks. The only exception are big ﬁrms with poor credit score. This shows that relationship lending is particularly important for ﬁrms with low credit score, where creditworthiness is more diﬃcult to asses. We also observe in this table that small ﬁrms with good credit score do not show a deterioration in their repayment behavior. An interesting ﬁnding is that big ﬁrms with good credit score still show a deterioration in their repayment behavior, which suggests that their quality may be lower than their actual credit score shows.
When looking at the interaction eﬀects for pregnancy and layoﬀ leaves we ﬁnd that big companies with good credit score not only show a deterioration in their repayment behavior, but also the deterioration is steeper for these ﬁrms compared to ﬁrms with worse credit score. This ﬁnding is consistent with the ﬁndings in Hertzberg, Liberty, and Paravisini (2010). In fact loan oﬃcers will have strong incentives to suppress bad news about large companies, because disclosing these news will strongly aﬀect their wage. Furthermore, by hiding this information, the loan oﬃcer lets big companies in ﬁnancial distress to keep good credit scores. As suggested by Hertzberg, Liberty, and Paravisini, when the loan oﬃcer has to leave the bank (either permanently or temporarily) the replacing loan oﬃcer will have strong incentives to disclose the real situation about these ﬁrms.
Overall our results suggest that the relationship between loan oﬃcers and their clients is
mit ‘soft’ information to a colleague. When loan oﬃcers have to go on leave unexpectedly, in particular due to sickness, we see that their clients are less likely to get a new loan within the bank. While small borrowers and high score borrowers are able to substitute the loss in credit access by taking on new loans outside the bank, big clients with poor credit score are not able to get outside funding. In addition borrowers show a deterioration in repayment behavior when their loan oﬃcer is absent. This might suggest that loan oﬃcers also play a role in reducing moral hazard behavior especially for small and opaque ﬁrms. Big ﬁrms show a steeper deterioration in their repayment behavior suggesting the loan oﬃcer also hide bad news from the bank by renewing loans to bad clients.
These results also shed an interesting new dimension on the pricing behavior of banks.
We observe that clients do not experience an increase in interest rates when their loan oﬃcer leaves. So it seems that loan oﬃcers do not use soft information they have on borrowers to hold them up for higher margins. We also see that the deterioration in the repayment behavior is not accompanied by an increase in interest rates. This suggest that interest rates, at least for this segment of the market, react slowly to changes in the probability of default.
We analyze the credit characteristics and repayment behavior of micro entrepreneurs of a large local bank operating in Chile, as well as how these characteristics and repayment behaviors change when the loan oﬃcer is absent for one month or longer. We study all of the clients borrowing from the micro-credit division of the bank. The micro credit division operates independently of the rest of the bank, and has its own lending technology, specially designed for micro credit businesses. The micro credit division operates in the branches of the bank but has separate personnel and oﬃce space. Only clients with yearly sales below US$ 110,000 can borrow from the micro-credit division, clients exceeding this limit must borrow through the regular lending process of the bank. The micro credit division of the bank has 210,000 clients of which 187,000 were borrowers (had non zero debt) at some point during the period of the study.
The bank as well as its micro credit division is organized in 3 Zones: North of Chile, Metropolitan area, and South of Chile. The Metropolitan area consists of the capital city and the counties nearby. North of Chile consists of the rest of the counties located north of Santiago, and South of Chile consists of the rest of the counties located south of Santiago.
A branch that oﬀers micro credit services must have at least one loan oﬃcer, and may have one or more loan oﬃcer assistants. Loan oﬃcers assistants can only process pre
oﬃcers can issue pre-approved loans as well as regular loans. In this study we will focus on loan oﬃcers, because they have decision power in the lending process.
The allocation of loan oﬃcers to clients starts when the client chooses his branch. Clients can freely choose their branch but will usually choose the branch that is closest to their business. In addition, clients rarely switch branches unless they relocate their home and/or business. However, some clients prefer to go to a bigger branch even if it is located further away from their home or business. In particular, the main branch located in downtown Santiago is very popular and has many clients that do not live particularly close to the main oﬃce. Once the client has chosen his branch the allocation of new clients to loan oﬃcers works as follows: The clients goes to the branch, new clients are serviced in a ﬁrst come ﬁrst serve basis and are allocated to the loan oﬃcer that becomes available. Old clients, on the contrary, wait until their already assigned loan oﬃcer becomes available.
Given this protocol, the allocation of new clients to loan oﬃcers is random within branches.