تفاوت های بین کشوری در اجرای سیاست های پولی و پول نوسانات نرخ بازار

Volatility patterns in overnight interest rates display differences across industrial countries that existing models—designed to replicate the features of individual countries’ markets—cannot account for. This paper presents an equilibrium model of the overnight interbank market that matches cross-country differences in patterns in interest volatility by incorporating differences in how central banks manage liquidity in response to shocks. Our model is consistent with central banks’ practice of rationing access to marginal facilities when the objective of stabilizing short-term interest rates conflicts with another high-frequency objective, such as an exchange rate target.

The overnight market for unsecured interbank loans plays a key role in the financial structure of most industrial countries. It anchors the term structure of interest rates and serves as the channel through which monetary policy is executed and liquid funds are funneled into the non-bank sector. To make models tractable for term-structure analysis and asset-pricing, most studies of this market have side-stepped the role of monetary policy in determining interest rates. The few studies that have taken the role of policy into explicit account have attempted to replicate features of individual markets—typically, the U.S. federal funds market. However, cross-country evidence (for instance, BIS, 1997; Ball and Torous, 1999; Prati et al., 2003) points to patterns in interest rates that call for a more integrated framework than available from models tailored to fit individual markets. In this paper, after building on previous empirical studies to present more comprehensive evidence on the historical behavior of short-term interest rates in the main industrial countries, we present a theoretical model of the overnight interbank market that replicates the main patterns in interest volatility highlighted in our empirical work. Specifically, our agenda is to explain theoretically the following stylized facts: (i) In countries relying on periodic reserve requirements, short-term interest rates display sharp cyclical volatility behavior; (ii) In the United States and in other countries whose central banks are committed to stabilizing interest rates at high frequency, interest rate volatility falls as market rates depart from their target level and/or approach rates on official marginal facilities; by contrast, interest rate volatility rises as market rates approach rates on marginal facilities, in countries where access to such facilities is rationed, for instance, because interest rate targeting is subordinated to exchange rate targeting; and (iii) Interest rate volatility rises during periods in which a central bank may be reluctant to supply or drain funds to and from the market, for instance, because the country's exchange rate is veering away from its target level. In our effort to explain these patterns theoretically, our goal is to present a unified framework that captures the main cross-country differences along a key dimension of policy execution: A central bank's willingness to offset high-frequency liquidity shocks by injecting or draining liquidity into and from the market, at both intra-marginal and marginal liquidity-management facilities. This willingness plays a central role in our analysis as it does in actual policy execution in the industrial world. For instance, many central banks routinely ration access to official lending and borrowing when the goal of stabilizing interest rates conflicts with another high-frequency objective, such as an exchange rate target. Conversely, other central banks react forcefully to deviations of interest rates from target. Our model captures this difference by parameterizing the amount of funds banks can trade with the central bank in response to shocks, and shows this degree of freedom to help replicate the main features of short-term interest rate volatility in the main industrial countries. To place our effort into context, we should note the extent to which this study adds to previous research on interbank markets, including our own previous work. Empirically, we are not the first ones to investigate fact (i)—which, for U.S. data, has been in the public domain at least since Spindt and Hoffmeister (1988)—and fact (iii)—which empirical research has tried to document (generally without success; see, for instance, Artis and Taylor, 1994) at least since Svensson (1991). Fact (ii) was first investigated in Prati et al. (2003), though that analysis should be viewed as preliminary, for two main reasons. First, Prati et al. (2003) offers no model selection procedure, in contrast with the formal general-to-specific selection procedure followed here. Second, that study links interest rate volatility to the level of target interest rates, which are rarely available and often vary insufficiently to allow identifying non-linear volatility effects. By contrast, here we link volatility to (lagged) market rates, which is more consistent with standard time-series models of interest rates and allows us to investigate non-linear volatility behavior for all countries, rather than only for countries that announce targets and vary them within interest rate corridors. These changes increase our estimation effort by an order of magnitude, but lead to a description of interest rate behavior that is both more satisfactory and more consistent with our subsequent theoretical effort. Theoretically, our work contributes to recent research on the equilibrium determination of short-term interest rates and on the micro-foundations of monetary policy execution. The standard reference for this research is Hamilton (1996), whose model, however, allows only for bank-level uncertainty and deterministic interest rates. Also, as most other models in the literature (for instance, Clouse and Dow, 1999 and Clouse and Dow, 2002; Hayashi, 2001; and, earlier, Ho and Saunders, 1985 and Campbell, 1987), Hamilton focuses only on the demand side of the money market, leaving no role for central bank intervention. Models allowing for central bank intervention are offered by Nautz (1998), Bartolini et al. (2002), and Valimaki (2003). These models, however, are tailored to the specific features of German, U.S., and euro area markets, respectively, and lack the flexibility needed to explain cross-country differences in interest rate behavior. In particular, they assume that banks can borrow and lend funds freely at official marginal facilities. Our main innovation is to allow for rationed provision of liquidity as market rates depart from target and approach rates on marginal borrowing and lending facilities, showing this feature to be important to capture the main cross-country differences in the behavior of interest rate volatility. Finally, Farnsworth and Bass (2003) assume, as we do, the central bank to intervene to target short rates only imperfectly. Beyond technical differences, their model focuses on the dynamics of target rates, while we focus on higher-frequency dynamics of market rates around targets, viewing targets as determined, at lower frequency, by central banks’ response to changes in inflation, output, and other macro-variables. Farnsworth and Bass also abstract from institutional details such as periodic reserve requirements and the distinction between marginal and intra-marginal liquidity provision. We include these details explicitly and assign to them a key role in shaping the behavior of the main countries’ interbank markets.

A simple model of an interbank market in which a central bank intervenes—with limited commitment—to stabilize interest rates around a target, can replicate the main qualitative properties of money market rates’ volatility in the world's largest interbank markets. These properties include a tendency for interest rate volatility to behave cyclically in countries with reserve averaging, a tendency for volatility to rise in periods when a central bank is less likely to accommodate shocks to banks’ liquidity, and—most interesting—a non-linear response of volatility to movements of market rates within official interest rate corridors. To replicate these properties, our model relies only on parameterized differences in central banks’ commitment to interest rate smoothing at high frequency, pointing to this aspect of monetary policy execution as a key factor shaping the behavior of money markets and short-term interest rates around the world. Our agenda for further research is topped by two items. First, allowing for heterogeneity in banks’ behavior and exposure to shocks should be useful to capture features of the inter-bank trading environment that our representative-agent model with instantaneous trading is not designed to emulate. Chief among these features are a tendency for small banks to be lenders and large banks to be borrowers in the market, and the sharp changes in interest volatility observed in intra-day data (including a systematic rise in volatility at day's end). A second goal should be to incorporate the high-frequency, institution-driven perspective of our and of related studies, into a framework more suitable for asset pricing. At this stage, our contribution should be seen as complementary to those of Piazzesi (2001) and Farnsworth and Bass (2003), whose focus on asset pricing accompanies a much more stylized description of banks’ trading environment than that provided by our study. A first step for future research bridging these two perspectives would be to parameterize the non-linearities in interest rate behavior that we document and explain here, use them to augment otherwise- standard pricing models, and see if this improves the empirical performance of the model against short-term bond price data.