EFFICIENCY AND BARGAINING POWER IN THE INTERBANK LOAN MARKET

• Author: Matthew Shum,James Chapman,Federico Echenique,Jason Allen
• Date: 01 May 2016
• DOI: http://doi.org/10.1111/iere.12173
• Published date: 01 May 2016

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 2, May 2016

EFFICIENCY AND BARGAINING POWER IN THE INTERBANK LOAN MARKET∗

BYJASON ALLEN,JAMES CHAPMAN,FEDERICO ECHENIQUE,AND MATTHEW SHUM1

Bank of Canada, Canada; California Institute of Technology,U.S.A.

Using detailed transactions-level data on interbank loans, we examine the efficiency of an overnight interbank

lending market and the bargaining power of its participants. Our analysis relies on the equilibrium concept of the core,

which imposes a set of no-arbitrage conditions on trades in the market. For Canada’s Large Value Transfer System,

we show that although the market is fairly efficient, systemic inefficiency persists throughout our sample. The level of

inefficiency matches distinct phases of both the Bank of Canada’s operations as well as phases of the 2007–8 financial

crisis. We find that bargaining power tilted sharply toward borrowers as the financial crisis progressed and (surprisingly)

toward riskier borrowers.

1. INTRODUCTION

Multilateral trading markets are endemic in modern economies with well-known examples

such as the bargaining over tariffs and similar trade barriers among WTO countries, monetary

and fiscal policy making among European Union countries, co-payment rate determination

among hospital and insurance company networks, and even trades of players among professional

sports teams. Our article presents a novel approach to empirically assess the efciency of these

markets and the bargaining power of the different agents in the market. We study the Canadian

interbank market for overnight loans.

A serious impediment to the analysis of efficiency and bargaining power in real-world trading

environments is the complexity of the markets themselves. The players are engaged in a com-

plicated game of imperfect competition, in which some of their actions are restricted by trading

conventions, but where the players may communicate and send signals in arbitrary ways. Even

if we could write down a formal model that would capture the interactions among players, it

would be difficult to characterize the equilibrium of such a game—a prerequisite to any analysis

of bargaining and efficiency. Moreover, the outcome of such a game greatly depends on the

assumed extensive form. For example, outcomes can vary according to the sequencing of offers

(who is allowed to make an offer to whom and when) as well as the nature of information

asymmetries among the players. For these reasons, a complete “structural” analysis of such

imperfectly competitive bargaining environments seems out of the question.

In this article, we take a different approach. Instead of modeling the explicit multilateral

trading game among market participants, we impose an equilibrium assumption on the final

outcome of the market. Our approach is methodologically closer to general equilibrium theory

than to game theory: We use the classical equilibrium concept of the core. The core simply

imposes a type of ex post no-arbitrage condition on observed outcomes; it requires that the

outcome be immune to defection by any subset of the participating players. Many alternative

∗Manuscript received June 2013; revised January 2015.

1The views expressed here are those of the authors and should not be attributed to the Bank of Canada. We thank the

Canadian Payments Association. We thank Lana Embree, Matthias Fahn, Rod Garratt, Denis Gromb, Scott Hendry,

Thor Koeppl, James MacKinnon, Antoine Martin, Sergio Montero, Mariano Tappata, and James Thompson as well as

seminar participants at the University of Western Ontario, Renmin University of China, the Bank of Canada workshop

on financial institutions and markets, the FRBNY, IIOC (Arlington), and Queen’s University for comments. Any errors

are our own. Please address correspondence to: Matthew Shum, HSS, Caltech, Mailcode 228-77, 1200 E. California

Blvd., Pasadena, CA 91125. Phone: 626-395-4022. Fax: 626-432-1726. E-mail: mshum@caltech.edu.

691

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

692 ALLEN ET AL.

equilibrium concepts would imply outcomes in the core, but the advantage for our purposes

is that the core is “model free,” in the sense that it does not require any assumptions on the

extensive form of the game being played. As we shall see, the relatively weak restrictions of

the core concept nevertheless allow us to draw some sharp conclusions about how efficiently

the Canadian interbank market functioned in the years preceding and during the most recent

economic crisis.

Subsequently, for outcomes that are in the core, we define a simple measure of how much the

observed outcomes favor particular market participants: specifically, borrowing versus lending

banks in the interbank market. We use this measure as an indicator of bargaining power and

analyze its relationship to characteristics of the market and its participants. Thus, in our article

efciency means the degree to which the absence of arbitrage conditions imposed by the core

are satisfied, and bargaining power results from the position of the outcomes in the core. If

the outcome is relatively more favorable to some agents, we shall say that these agents have

enjoyed greater bargaining power.

We study the Large Value Transfer System (LVTS) in Canada, which is the system the Bank

of Canada uses to implement monetary policy. Throughout the day, LVTS participants send

each other payments and at the end of the day have the incentive to settle their positions to

zero. If there are any remaining short or long positions after interbank negotiations these must

be settled with the central bank at unfavorable rates. Participants are therefore encouraged to

trade with each other in the overnight loan market. This market is ideal for study for various

reasons: First, the market operates on a daily basis among seasoned players, so that inexperience

or na¨

ıvete of the players should not lead to any inefficiencies. Second, there is a large amount

of detailed data available on the amount and prices of transactions in this market. Finally, the

LVTS is a “corridor” system, meaning that interest rates in the market are bounded above and

below, respectively, by the current rates for borrowing from and depositing at the central bank.

This makes it easy to specify the outside options for each market participant, which is a crucial

component in defining the core of the game; at the same time, the corridor leads to a simple

and intuitive measure of bargaining power between the borrowers and lenders in the market.2

Several researchers have explicitly modeled the decision of market participants in environ-

ments similar to LVTS. For example, Ho and Saunders (1985), Afonso and Lagos (2011), Duffie

and Gˆ

arleanu (2005), Duffie et al. (2007), and Atkeson et al. (2013) examine the efficiency of the

allocation of funds in the Federal funds market or over-the-counter markets, more generally.3

The systems, markets, and agents under study in this article have previously been examined in

Chapman et al. (2007), Hendry and Kamhi (2009), Bech et al. (2010), and Allen et al. (2011).

Moreover, as previously mentioned, the core imposes, essentially, no-arbitrage conditions

on the trades in the interbank market, so that inefficient outcomes—those that violate the

core conditions—are also those in which arbitrage opportunities were not exhausted for some

coalition of the participating banks. Thus, our analysis of the interbank market through the lens

of the core complements a recent strand in the theoretical finance literature exploring reasons

for the existence and persistence of “limited arbitrage” in financial markets (see Gromb and

Vayanos, 2010, for a survey of the literature).

A market outcome is the result of overnight lending between financial institutions at the end

of the day: The outcome consists of the payoffs to the different banks. We (1) check if each

outcome is in the core (this can be done by simply checking a system of inequalities), and (2)

measure the degree to which outcomes are aligned with the interests of net borrowers or lenders

in the system: our measure of bargaining power. We proceed to outline our results.

2Since Canada operates a corridor system, outside options are symmetric around the central bank’s target rate, and

changes to the target do not arbitrarily favor one side or another of the market. In contrast, in overnight markets

without such an explicit corridor, both the outside options and bargaining power are not as convenient to define. Many

central banks use a corridor system—e.g., the ECB. The Federal Reserve and Bank of Japan, however, use reserve

regimes. Corridor systems rely on standing liquidity facilities whereas reserve regimes rely on period-average reserve

requirements. See Whitsell (2006) for a discussion.

3An interested reader can find a book length treatment of the economics of OTC markets in Duffie (2012).