The IMF staff (Benes et al. 2014) recently unveiled a new model that “has been developed at the IMF to support macrofinancial and macroprudential policy analysis”. In introducing the model they argue that “such new analytical frameworks require a major revamp of the conventional linear dynamic stochastic general equilibrium (DSGE) models”. We agree with Benes et al. that the workhorse extant models used in policy analysis are not well suited to this task and that “an area that requires particular attention is the special role played by banks, most importantly the role of bank balance sheets.” But, while we share their motivation and goals, we suggest a slightly different modelling approach. The purpose of this note is to present our vision for how such macroprudential modelling efforts might best proceed. We begin with some philosophic observations about the key ingredients that should be included in this kind of analysis and then describe a few findings from a model we have developed.
Modelling financial instability
Eric Rosengren (2011) provides an excellent starting point for any modelling exercise of this sort by defining what is meant by financial stability and instability. He writes:
“Financial stability reflects the ability of the financial system to consistently supply the credit intermediation and payment services that are needed in the real economy if it is to continue on its growth path.
Financial instability occurs when problems (or concerns about potential problems) within institutions, markets, payments systems, or the financial system in general significantly impair the supply of credit intermediation services – so as to substantially impact the expected path of real economic activity.”
This characterisation is brilliant in several respects. First, he is appropriately expansive in recognising the multiple roles that the financial system plays in the economy. Second, he correctly ties the importance of financial stability to supporting the real economy. Promoting financial stability should not be an abstract goal; it matters because instability adversely impacts consumption and savings decisions. Finally, he perceptively points out that not only do actual problems matter, but potential problems that are yet to materialise can also be critical.
In Kashyap et al. (2014a) we discuss how these definitions can be used to guide model building. Rosengren’s definitions fit well with the existing academic literature on intermediation. In particular, prior work suggests three theoretical channels through which intermediaries can improve welfare. One strand of thinking emphasises the role that they play in extending credit to certain types of borrowers (e.g. Diamond 1984). A second views them as vehicles for improving risk-sharing. They can do this through the liability side of their balance sheet by creating safer and risky claims (e.g. deposits and equity) against the assets that they hold (Benston and Smith 1976, Allen and Gale 1997). And the third perspective supposes that they specialise in creating liquid claims that are backed by illiquid assets (Diamond and Dybvig 1983).
- Our first principle is that it is imperative to start with a general model where the financial system plays all three of these roles.
This is not true of most of the new literature on macroprudential regulation. For instance, Benes et al. (2014) have no role for liquidity provision in their set up – to be fair, very few macro models do. Without this contribution to the financial system, certain forms of instability are ruled out; for instance, funding runs can only happen if we suppose that there is a maturity mismatch between assets and liabilities.
Likewise, regulation to fix potential runs, such as proposals for narrow banking (e.g. Cochrane 2014), also appears to be especially appealing. But if the fragility that creates the possibility of runs is not valuable on its own, of course, eliminating it would be desirable! The more challenging question is what happens if there is a fundamental underlying reason why maturity mismatches create value for some parties. In that case, regulation that stops runs for certain organisational forms may simply move activity to other unregulated entities (because of the underlying legitimate demand for liquidity creation). Or, if the regulation is fully effective at preventing runs, then the social value of the liquidity provision is lost. A suitable model should be capable of assessing these trade-offs.
- Our second principle is that intermediaries should operate in an environment where the savers who use them are forward looking, and the prices the intermediaries face adjust (endogenously) to the regulatory environment.
For example, much of the Basel Committee’s regulatory agenda has focused on raising capital requirements for banks. The consequences of such regulations are very different when the banks have to offer attractive enough returns to entice savers to buy equity, than when the price of equity is unrelated to the size of a bank’s balance sheet and/or the risks on that balance sheet.
We are unaware of any existing models that satisfy these two principles. So, in Kashyap et al. (2014b), we have constructed one by modifying the classic Diamond and Dybvig’s (1983) framework. In their original model, banks only provide liquidity insurance to savers (by allowing depositors the option of withdrawing early), so there is no other risk-sharing or additional lending that takes place because of banks. Our modifications are designed so that banks also provide these services.
There are four specific changes that we make.
- First, savers can buy equity in a banking sector and save via deposits.
- Second, the banks choose to invest in safe assets or to fund entrepreneurs who have risky projects.
Together, these changes make banks’ asset and liability structure interesting, and create a situation where there is some fundamental risk that cannot be diversified away.
- Third, the banks and the entrepreneurs face limited liability.
- Finally, there is a probability of a run, but unlike in Diamond and Dybvig, the decision whether to run is governed by the banks' leverage and mix of safe and risky assets.
In particular, when the banking system has more loans relative to safe assets, or more deposits relative to equity, a run is more likely. The possibility of the run reduces the incentive to lend and take risk, while limited liability pushes for excessive lending and risk-taking. So, the baseline economy suffers from two problems: The destructive aspects of runs (where savers can lose money, banks can fail, and borrowers have their loans pulled), and the moral hazard problems that come from limited liability.
The banks in this world not only offer liquidity insurance with their deposits, but they offer savers a better alternative to making direct loans to entrepreneurs. This occurs because the banks start with some initial equity, which serves as a buffer in the case the entrepreneurs’ projects fail; so, banks are pooling and trenching risks, and by doing so they can attract more funds to lend than the savers would be willing to risk if they could not access the banks.
The bulk of our paper explores the portfolio choices that savers and banks make in this kind of environment. One nice feature of this model is that it can be used to explore how capital regulation, liquidity regulation, deposit insurance, loan to value limits, and dividend taxes alter allocations and change the degree of run-risk and total risk-taking. Rather than focusing on very specific findings about how these interventions can matter, we mention here a few findings that seem likely to carry over to other models that respect our two principles.
Some general findings of the model
- First, as a benchmark, we compute the portfolio allocations that a central planner would make. We find that approximating the planner's allocations with just one regulation is impossible. In this model, it takes at least two tools to overcome the two distortions.
- Second, the way that the various regulations change behaviour is very different, and combining some of them leads to very little improvement.
Put differently, it is not correct to conclude that combining any two tools is necessarily enough to correct the two externalities in the model.
- Third, the interactions among the regulations are sufficiently subtle that it would be hard to guess which combinations prove to be optimal in this model.
We do not want to claim that our model is sufficiently general that the findings necessarily apply in all other models. But, attempting to assess different regulations (and to calibrate how they should be set) would be very difficult to do without consulting a range of models. Intuition helps, but at some point it runs out.
- Finally, coming up with regulations that simultaneously eliminate runs and shrink total lending (and risk-taking) is hard.
This happens because the usual interventions that make runs less likely either create opportunities for banks to raise more funds or take more risk, or so severely restrict the savers, banks or borrowers that one of them is made much worse-off.
We hope that these ideas will lead others to move away from small perturbations of existing DSGE models and instead consider much more fundamental changes.
Disclaimer: The views in this column are solely the responsibility of the authors and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System or of any other person associated with the Federal Reserve System.
Allen, F and D Gale (1997), "Financial Markets, Intermediaries, and Intertemporal Smoothing", Journal of Political Economy, 105(3), pp. 523-546.
Benes, J, M Kumhof and D Laxton (2014), “Financial Crises in DSGE Models: A Prototype Model”, IMF Working Paper 14/57.
Benston, G J and C W Smith (1976), “A Transactions Cost Approach to the Theory of Financial Intermediation”, Journal of Finance, 31(2), pp. 215-231.
Cochrane, J H (2014), “Toward a run-free financial system”, working paper, University of Chicago.
Diamond, D W and P H Dybvig (1983), “Bank Runs, Deposit Insurance and Liquidity”, Journal of Political Economy, 91 (3), pp. 401–419.
Diamond, D W (1984), “Financial Intermediation and Delegated Monitoring”, Review of Economic Studies, 51(3), pp. pp. 393-414.
Kashyap, A K, D P Tsomocos and A P Vardoulakis (2014a), "Principles for Macroprudential Regulation", Banque de France Financial Stability Review, No. 18, April 2014, pp. 173-181.
Kashyap, A K, D P Tsomocos and A P Vardoulakis (2014b), “How does macroprudential regulation change bank credit supply?”, National Bureau of Working Paper 20165.
Rosengren, E S (2011), “Defining Financial Stability, and Some Policy Implications of Applying the Definition” Keynote Remarks at the Stanford Finance Forum, Graduate School of Business, Stanford University, 3 June.