Financial Institutions I: The Economics of Banking Prof. Dr. Isabel Schnabel Gutenberg School of Management and Economics Johannes Gutenberg University Mainz Summer term 2011 V4 1/30
I. Introduction II. Do Financial Institutions Matter? III. Why Do Banks Exist? III.1. Transaction costs III.2. Liquidity Insurance III.3. Delegated Monitoring IV. Credit Rationing 2/30
I. Introduction II. Do Financial Institutions Matter? III. Why Do Banks Exist? III.1. Transaction costs III.2. Liquidity Insurance III.3. Delegated Monitoring IV. Credit Rationing 3/30
III.3. Delegated Monitoring Reading material: Freixas/Rochet, p. 30 34; Diamond (Review of Economic Studies, 1984) Diamond (1984) models the risk transformation by banks Banks grant risky loans Depositors hold safe deposits Central question: Why can banks offer a better return-risk trade-off than the market? 4/30
Basic Idea of the Model Firms (borrowers) suffer from an incentive problem and have to be monitored Financial intermediaries allow for a saving in monitoring costs if they monitor firms instead of each individual household (so far this is a pure transaction cost argument) But: Who monitors the monitor? Solution: If many projects are financed at the same time, the costs of delegation can be reduced due to diversification (in the limit to zero) Result: Involvement of a financial intermediary dominates a solution with direct financing 5/30
Asymmetric Information in Financial Relationships 3 types of informational asymmetries: 1. Ex-ante: Quality uncertainty at the beginning of the financial relationship, hidden information Adverse selection 2. Ex-interim: Behavioral uncertainty during a financial relationship, hidden action Moral hazard 3. Ex-post: Lack of verifiability of returns at the end of a financial relationship Costly state verification 6/30
Solutions to Asymmetric Information 2 types of solutions: 1. Reduction of informational asymmetry, e. g. through monitoring 2. Incentive compatible contracts: Contracts are written in a way that gives the informed party an incentive to behave properly (i. e., as desired by the uninformed party) Both solutions are subject to information costs 1. Monitoring costs 2. First-best solution often does not satisfy incentive compatibility conditions Second-best solution Information costs = difference between first- and second-best solution 7/30
Costs of Asymmetric Information Costs of asymmetric information are generally borne by both sides of the financial transaction Example 1: Lender has to incur costs to prevent undesirable behavior by the borrower Monitoring costs are shifted (at least partly) to the borrower (depending on market power) Example 2: Borrower has an incentive to exert too little effort or to take excessive risks This is anticipated by lenders and will therefore be reflected in loan rates In some circumstances, the financial transaction is not carried out at all even though it would have been beneficial for both sides 8/30
Asymmetric Information in Diamond (1984) Ex-post uncertainty: Lenders can observe realized project returns only if they incur monitoring costs Costly state verification (Gale/Hellwig, 1985) Borrowers have an incentive to understate realized returns to avoid a repayment of the loan Solutions: 1. Monitoring Monitoring costs 2. Incentive compatible contract, here: loan contract with non-monetary penalties when repayments are too low Penalties are costly from a welfare perspective because they have to be imposed even if the entrepreneur truly reports that returns were low 9/30
Assumptions of the Model N risk neutral entrepreneurs need one unit of funds each to carry out risky projects A large number of risk neutral lenders (> N m) want to invest 1/m units each Loans supply > loan demand Result: Lenders can be pushed to their reservation utility Alternative investment gives an expected return of R Result: Lenders will provide funds to an entrepreneur only if the return is at least R Due to an excess supply of loans, lenders obtain an expected return of exactly R 10/30
Investment Projects of Entrepreneurs t = 0: Investment of one unit t = 1: Stochastic return ỹ 0 ỹ ȳ < E(ỹ) > R, i.e., it is efficient to carry out the project under full information Projects of different entrepreneurs are stochastically independent Asymmetric information: Only the entrepreneur can observe the realization of ỹ Types of financing contracts: (a) Direct financing (b) Indirect financing with financial intermediary 11/30
(a) Direct Financing Information problem makes certain financial contracts unfeasible Example: Contract where the repayment is proportional to (unobservable) project returns Entrepreneur will always claim that the project return was zero Repayment = zero Lenders would anticipate this and would not provide any loans in the first place 12/30
Solutions to Asymmetric Information with Direct Financing Note: In this model, the costs of asymmetric information are always borne by the entrepreneurs (lenders always receive an expected return of R) Entrepreneurs are happy to write a contract that prevents them from misbehaving Two possibilities: 1. Incentive compatible contract with non-monetary penalties 2. Monitoring 13/30
1. Incentive Compatible Contract with Non-monetary Penalties Non-monetary penalties: Penalties can be carried out even if the entrepreneurs does not have any wealth (or claims not to have any wealth) Penalties hurt the entrepreneur, but do not generate any direct benefit for lenders Penalties are used only to set incentives for the entrepreneur to behave properly (deterrence) Hence, penalties as such reduce social welfare (but they have positive incentive effects) 14/30
Non-monetary Penalties Examples for non-monetary penalties: Prison Loss of reputation Costs of job search Time spent in bankruptcy procedure (Physical punishment) 15/30
Optimal Financial Contract z(y) = Actual repayment by entrepreneurs φ(z(y)) = Penalty at a repayment z(y) φ(z(y)) = 0 if the entrepreneur repays the required amount Note: The penalty function cannot depend on y! Optimal penalty function: Penalty must be sufficiently high to set incentives for the desired behavior Penalty has to be as small as possible to avoid welfare costs 16/30
Optimal Financial Contract φ (z(y)) = max[h z(y),0], where h is the agreed repayment Penalty has to be equal to the difference between agreed and actual repayment Then the sum of repayment and penalties is always equal to h for the entrepreneur, no matter how high the repayment The entrepreneur is then indifferent between lying and not lying By assumption, the entrepreneur will not lie in such a situation 17/30
Optimal Financial Contract How to choose h? h is set such that E[z(y)] = R In an incentive-compatible contract, we get z(y) = min[h,y] Entrepreneur passes on the entire returns of the project to the lender as long as y < h, otherwise he repays h Like standard credit contract! At the optimum, h > R: Due to the possibility of small project returns, the borrower has to pay a risk premium 18/30
Welfare Costs of Financial Contract with Penalties Welfare costs = expected penalty E[φ (z(y))] Ex post, penalizing the entrepreneur lowers welfare Punishment occurs whenever the repayment is smaller than the agreed repayment h But then, the actual repayment is also below h! Entrepreneur is punished even though he is telling the truth 19/30
2. Contract with Monitoring Alternative solution: Monitoring by each lender Costs of monitoring: m K (m = number of lenders per firm) Implicit assumption: Repayment by entrepreneurs to the monitoring agent cannot be observed (otherwise, delegated monitoring would not cause any problems) 20/30
Monitoring vs. Contract with Penalties Under what conditions is monitoring better (from a social welfare perspective) than a contract with non-monetary penalties? Condition: m K < E[φ (z(y))] Monitoring is expensive if m is large Contract with penalties is expensive if bad returns happen with a high probability Conclusion: Monitoring is preferable when the number of lenders is small and low project returns are very likely 21/30
(b) Indirect Financing with Financial Intermediary Idea: If monitoring is delegated to a financial intermediary (FI), monitoring costs can be reduced from m K to 1 K Problem: How can we make sure that the FI truthfully passes on the collected information? Information problem has not been solved, but it has been shifted to another level Additional delegation costs If the FI also has to be monitored, monitoring costs even increase to (m+1) K Central contribution of this paper: Delegated monitoring can entail lower costs and can therefore be welfare-increasing 22/30
Delegated Monitoring By involving a FI, there arises an additional information problem Optimal contracts between (1) the entrepreneur and the FI, and (2) the FI and lenders: Monitoring of the FI by lenders cannot be optimal because this would cause monitoring costs of m K (same as with direct financing) Financial contracts with non-monetary penalties between the FI and the entrepreneur cannot be optimal because this would cause the same costs as direct financing Optimal solution: Contract between lender and FI = contract with penalties and monitoring of firms by FI But: This cannot work if FI finances only one firm (costs: E[φ (z(y))]+k) What happens when FI deals with several firms? 23/30
Delegation Costs and Diversification Idea: Risk neutral FI finances N entrepreneurs with stochastically independent project returns (a sufficient condition is that returns are not too strongly correlated) FI monitors entrepreneurs at costs N K N m lenders Between lenders and FI there is a contract with non-monetary penalties Expected penalties depend on the probability of returns below the agreed repayment h 24/30
When is Delegation Optimal? Central point of the paper: Due to diversification of the bank portfolio, this probability goes to zero for N! Law of large numbers: For large N, the return of a diversified bank portfolio converges to the expected project value (and hence is approximately certain) Welfare costs of financing through FI K (delegation costs converge to 0) A perfectly diversified FI is optimal if K < min{m K,E[φ (z(y))]} K < E[φ (z(y))] In many cases, a contract with delegated monitoring dominates direct financing 25/30
Conclusion If many lenders are needed to provide a loan to an entrepreneur, monitoring costs can be reduced by delegating monitoring to a financial intermediary When the correlation among loans is small and FI finances many firms, delegation costs are small and decrease in the number of loans Conclusion: Delegated monitoring can be optimal even though this creates an additional layer and therefore new agency problems 26/30
Critical Assessment Important paper that explains the existence of banks through their risk transformation function Predictions of the model: Indirect financing with financial intermediaries can be optimal Banks play a role if information problems are important Financing of banks through debt Deposits are relatively safe in spite of the high risk of single loans, reason: diversification Banking business entails economies of scale, here: diversification benefits 27/30
Critical Assessment In reality, financial intermediaries are far from being perfectly diversified, they hold many non-diversifiable risks In the model, banks ideally should be infinitely large, but: high concentration in the banking sector may not be desirable for other reasons (especially competition) Model offers no description of the banking system, banking competition does not play a role in the model In reality, there also exist many small banks, diseconomies of scale in banking? 28/30
Critical Assessment Bank capital does not play any role in this model even though it is an important determinant of the riskiness of bank debt (including deposits) Model focusses on the lending business of banks, but abstracts from the characteristics of the deposit contract (other than Diamond/Dybvig 1983) Considered information problem is a bit special, other types of informational asymmetries may be more important, e. g. moral hazard in banks risk choices, adverse selection of borrowers No discussion of reputation effects This would require a dynamic model 29/30
Program of Next Lecture IV. Credit rationing, Model by *Stiglitz/Weiss (American Economic Review, 1981) 30/30