Monetary economists have frequently expressed the view that the financial system is an important source of and propagation mechanism for cyclical fluctuations. Indeed, Keynes (1936), Simons (1948), Friedman (1960) and many others have argued that the free and unregulated operation of financial markets can lead to indeterminacy of equilibrium and "excessive economic fluctuations," even in the absence of shocks impinging on the rest of the economy. In modern terms, this argument claims that the financial system itself is a source of endogenously arising economic volatility.
This view has a long empirical foundation. Most of the pre-World War II recessions were associated with substantial transfers of resources out of the banking system and into other assets. For instance, most of the pre-World War II recessions described by Friedman and Schwartz (1963) were associated with increases in the currency-deposit ratio. Particularly severe recessions were associated with particularly sharp increases in this ratio (that is, with bank panics). And even in the last three decades, several recessions have been accompanied by phenomena termed "disintermediation" or "credit crunches." In all of these episodes the volume of bank-extended credit declined, and "credit crunches" have often been associated with the increased incidence of non-price rationing of credit.
Why do we observe sharp fluctuations in the volume of intermediated credit? Can these fluctuations "cause" business cycles, or are they merely a symptom of them? Is credit market activity a potential or (as many have argued) a necessary source of endogenously arising excess fluctuations? Are there economic configurations which dictate that credit market activity must be associated with endogenous economic volatility?
This paper considers the relationship between credit and production in a simple model of dynamic general equilibrium, namely the non-monetary overlapping generations economy with production introduced by Diamond (1965). We modify that economy in only two respects: we introduce some intragenerational heterogeneity, and we force some kinds of capital investment to be credit financed. Under complete public information about the characteristics of potential borrowers, we show that both of these modifications are purely cosmetic, and make no qualitative difference to the properties of competitive equilibria. Specifically, given any positive initial capital stock, the economy monotonically approaches a non-trivial steady state equilibrium, which is unique under our assumptions. Thus there is no scope either for indeterminacy of equilibrium, or for endogenous fluctuations.
Heterogeneity makes a lot of difference when there is private information about borrower characteristics (here ex ante information about loan repayment probabilities). In the presence of an adverse selection problem, lenders will seek to elicit from borrowers information regarding loan repayment probabilities. Lenders will do so by structuring the loan contracts they offer which specify both loan quantities and interest rates in order to induce potential borrowers to self-select or, in effect, to reveal their type. Thus all loan contracts offered must be incentive compatible. Incentive compatibility can be attained in two distinct ways. In one, the full-information exchange of credit may be consistent with self-selection in the loan market, in which case it can be duplicated under private information by appropriately chosen loan contracts. Alternatively, the exchange of credit that occurs under full information may not be incentive compatible, in which case self selection constraints will bind on the choice of loan contracts. In this case lenders will use credit rationing as a means of inducing self-selection.
Interestingly, there is a range of values for the current capital stock (equivalently, for factor prices) under which the full information allocation of credit is incentive compatible. Here it is feasible for the equilibrium law of motion of the capital stock to coincide with that which obtains under perfect information. There is also a range of capital stocks for which incentive constraints necessarily bind; here credit must be rationed and the law of motion of the capital stock yields lower values relative to the situation of complete information. In addition, under conditions we describe, there is a non-trivial closed interval of current capital stocks for which either binding or non-binding incentive constraints are consistent with equilibrium in the loan market. When this happens, the economy can be in either of two regimes: a Walrasian regime in which competitive markets allocate credit in the standard way, and a regime of credit rationing. If credit is rationed, resources leave the banking system and the allocation of investment becomes less efficient in a manner we will make precise. Since each of these regimes is consistent with equilibrium, equilibrium is indeterminate: the economy can follow either the full information or the private information law of motion for the capital stock. Moreover, as we will demonstrate, the economy can switch from one law of motion to the other in either a deterministic or a stochastic manner. The result is that there will be fluctuations in output and the capital stock, and these fluctuations need not dampen over time. Thus both indeterminacy of equilibrium and "excessive" fluctuations can be observed when agents are privately informed about loan repayment characteristics.
Perhaps more interestingly, our model economy generates endogenous reflective barriers, a floor below which incentive constraints cannot bind in equilibrium, and a ceiling above which Walrasian credit allocations cannot be incentive compatible. We then use the presence of these barriers to examine two possible configurations of equilibria.
Since our model economy generates two distinct laws of motion for the capital stock (one corresponding to Walrasian allocations in the credit market, and one corresponding to credit rationing), it may possess two distinct steady state equilibria: one corresponding to Walrasian outcomes and another corresponding to credit rationing. If these two steady states lie inside the reflective barriers they are both equilibria, as are various paths that approach either of them monotonically. In addition, there is a large set of equilibria in which the economy endogenously switches between the two laws of motion. Indeed, we demonstrate the existence of equilibria in which m periods of expansion are followed by n periods of contraction (and so on) for every pair of integer values (m,n). Moreover, all such equilibria are asymptotically stable, so that the economy generates a high dimensional multiplicity of equilibria displaying asymmetric cyclical fluctuations when m n.
It is also possible that the two steady state equilibria lie outside the reflective barriers, in which case neither one constitutes a legitimate equilibrium. In this case, the economy does not admit any equilibrium with monotonic time paths in the capital stock or output: all equilibrium sequences have turning points, and fluctuations must be observed along every equilibrium path. We derive necessary and sufficient conditions for the existence of deterministic (m,n) cycles (m periods of expansion followed by n periods of contraction). We also propose an algorithm which identifies all such periodic cycles, and isolate the largest pair (m,n), which we associate with the maximally persistent periodic orbit. Asymmetric cycles with m > n (expansions are longer and shallower than contractions) are found to exist if the "floor" of the economy is tighter than its "ceiling".
What happens when these fluctuations occur? When the economy transits from an expansion phase to a contraction phase, resources leave the banking system and are allocated to less efficient uses. This occurs because depositors (correctly, as it turns out) conjecture that rates of return offered by the banking system will be low. As banks lose depositors, less credit is available, and banks are forced to ration it. Hence contractions are associated with disintermediation and "credit crunches." Moreover, when steady state equilibria lie outside the economy's reflecting barriers, these phenomena cannot be avoided. For economies with this property, indeterminacy and excessive fluctuations are not merely possible but actually unavoidable.
To summarize, dynamic equilibria with adverse selection satisfy an unusual, set-valued, discontinuous difference equation which contains two well-defined, partially overlapping domains with distinct structural regimes. At intermediate values of economic activity the two regimes overlap, permitting the economy to switch between them. The resulting indeterminate equilibria may be indexed by a stochastic process that governs an unobserved regime-switching variable in the manner proposed by Hamilton (1989, 1990). The economic interpretation of this state variable is to regard it as an index of savers' expectations about credit market conditions. An expanding body of evidence suggests that such nonlinear mechanisms accord with the behavior of many aggregate time series.
The remainder of the paper proceeds as follows. Section 2 lays out the environment and the nature of trades, and describes the equilibrium conditions that obtain when credit is and is not rationed. Section 3 shows how the economy can transit between Walrasian regimes and regimes of credit rationing, while Sections 4 and 5 examine the existence of cyclical equilibria which display undamped oscillations. Section 6 contains a variety of numerical examples of such equilibria, and Section 7 concludes.
Download
PDF Ebook Financial Intermediation And Regime Switching In Business Cycles
