شوک نقدینگی و دلاری سازی سیستم بانکداری

This paper shows how uncertainty about liquidity demand can lead to a high degree of dollarization in the banking system. I study a model where the demand for currency in each period is random, and where it is easier for banks to borrow in local currency in times of crisis than in dollars. Banks choose a portfolio composed of local currency, dollars, and real loans. Compared to the anticipated transactions demand for each currency, I show that the bank will hold a relatively large amount of dollars and a relatively small amount of local currency. I also show the existence of a dollarization multiplier: as the anticipated transactions demand for dollars increases, the dollarization of the banking sector increases more than proportionately.

Much of the recent literature on dollarization has focused on the characteristics of partially dollarized economies, where dollars and local currency each account for a substantial fraction of economic transactions. This literature has identified the degree of dollarization of the banking system as being a particularly important variable.1 The effects of monetary policy, for example, as well as the reaction of the economy to external shocks, appear to depend critically on the degree to which domestic banks choose to denominate their transactions in dollars. This raises the important question of what determines the level of dollarization in the banking system. Most of the existing explanations focus on “liability dollarization,” in which depositors choose to have their deposits denominated in dollars. The literature is therefore concerned largely with explaining why people would choose to open dollar-denominated accounts (see Broda and Levy Yeyati, 2003, Calvo, 1999, Catão and Terrones, 2000 and Savastano, 1992; among others).2 Other explanations have concentrated on network externalities that appear when banks’ transactions are made in foreign currency. A lot of work in this area has been done with regard to Bolivia, which is one of the main examples of a partially dollarized economy (see for example Peiers and Wrase, 1997, Reding and Morales, 1999 and Cuddington et al., 2002).3 This work is closely related to the well known phenomenon of “hysteresis,” which was first introduced to the study of dollarization issues by Uribe (1997). In this paper, I concentrate on the other side of a bank’s balance sheet. I show that uncertainty about liquidity demand will also tend to push a banking system toward becoming highly dollarized. The model I develop is a generalization of Champ et al. (1996).4 I transform their one-currency model into a two-currency model, introducing the possibility for banks to hold reserves in either local or foreign currency. Hereafter I will refer to these currencies as pesos and dollars, respectively. In each period, the demand for currency is stochastic. A fraction of the agents who demand currency will need pesos, and the remaining fraction will need dollars. I make the natural assumption that in times of high liquidity demand, it is easier for a bank to borrow in local currency than in dollars. I show that this assumption implies that, given some anticipated transactions demand for each currency, banks will choose to hold a relatively large amount of dollars reserves and a relatively small amount of reserves in local currency. Uncertainty about liquidity demand therefore leads to a form of “asset dollarization” in addition to the liability dollarization studied in other papers.5 During a liquidity crisis, a bank will be able to meet the demand for pesos, since it can borrow pesos from the Central Bank. However this will not be true for dollar demand. This asymmetry leads a bank to hold a “precautionary” stock of dollars and the model will show how this is done. In fact it will be shown that the demand for dollar reserves is in large part driven by “rare events” or “extreme events”. I concentrate first on the worst-case scenario, where there is no possibility for banks to borrow dollars from the Central Bank. I then generalize the results by introducing the possibility of borrowing in dollars at a positive interest rate. The main idea of the model, that banks are self-insuring, is not exclusive to a dollarized economy or banking system. For example, Antinolfi and Keister (2006) show, in a single currency model, how different Central Bank policies affect the equilibrium levels of cash reserves and of real investment held by banks. However, the question of how banks choose to dollarize their assets has not been modeled yet in a clear and useful manner, and this paper aims to fill that gap. The remainder of the paper proceeds as follows. In Section 2, the two-currency model of money and liquidity demand is developed. Section 3 studies the equilibria of the model when banks cannot borrow any dollars from the Central Bank. The “dollarization multiplier” is defined and derived. In Section 4, I generalize the results to a setting where it is possible for banks to borrow some dollars from the Central Bank. Finally, in Section 5, I present some concluding remarks.

I have studied a pure-exchange economy in which spatial separation, limited communication and random relocation combine to create an environment for analyzing the dollarization of the banking system. It has been shown that banks will hold relatively large positions in dollars compared to the transactions demand for dollars. The results are the same whether the banks are unable to borrow dollars from the Central Bank or this type of borrowing is possible but costly. In either case, banks will choose to hold enough dollars to cover the worst case scenario; in other words, the first line of defense, in case of a liquidity crisis, will be provided by the banks themselves. It has been shown also that dollarization of the banking system increases faster than the demand for dollars, and this is an equilibrium as long as View the MathML sourceθ∈[θ¯,1]. If θ is much larger than View the MathML sourceθ¯, banks could be holding a lot of pesos. The model shows, however, that the dollarization of the banking system will be surprisingly high relative to the dollarization of the whole economy. That, is 1 − θ could be interpreted as a measure of what transactions are carried in dollars. In theory, it could be small or large, but whatever 1 − θ is, the banking system will be more dollarized than that because of the dollarization multiplier. The model has been presented assuming that banks cannot borrow any dollars from the Central Bank, or can borrow but at a penalty rate. The latter case allows for more general results. In a crisis situation, when there is a shortage of dollars, it seems natural for a “dollar premium” to emerge (that is, for dollars to be at least slightly more expensive for a bank to get), especially if the Central Bank is using up all (or almost all) of its reserves in order to act as a liquidity insurer for the banking system. One striking feature of the model is that it shows that high domestic inflation is not necessary to generate a tendency toward dollarization. In the steady-state equilibria I study, the supply of pesos and the domestic price level are both constant. If, instead, the government were to increase the money supply in each period, steady-state inflation would result. This inflation would lead banks to shift resources away from pesos and toward other assets (the well-known Mundell-Tobin effect), including dollars. In this way, high domestic inflation could be another, complementary cause of asset dollarization. The point of this paper, however, is that even in the absence of inflation, high levels of dollarization can result from uncertainty about liquidity demand.12 Finally, I must mention that the model can be compared using country data. If someone looks at the data generated by the model, in each period, she would observe the realized demand for pesos and the realized demand for dollars. Just looking at that data, the banking system will probably appear to be “overly” dollarized. That is, it is likely that there will be realizations of the variable π that are above π∗, so that sometimes the bank will run out of pesos and need to borrow. But there will probably not be any realizations of π that are very close to 1. So even though the bank is occasionally running out of pesos, the bank will always have a stock of dollars around that it is not using. Why is the bank so “attached” to holding dollars even when there are regular shortages of pesos? The model tells us the answer: the demand for dollars is driven by “rare events” or the worst-case scenario. If we observe data from a period in which the worst case did not happen, the data will be “biased” in a sense. Dollarization will be costly to the banks, insofar that they have to maintain more liquidity in dollars. It would be interesting to see how this could affect financial intermediation, the quantity of loans given, and the level of economic activity. These issues along with how steady inflation could affect the results of the model are left for future research.