بازارهای دارایی تقسیم بندی شده و رژیم نرخ ارز مطلوب

This paper revisits the issue of the optimal exchange rate regime in a flexible price environment. The key innovation is that we analyze this question in the context of environments where only a fraction of agents participate in asset market transactions (i.e., asset markets are segmented). Under this friction alternative exchange rate regimes have different implications for real allocations in the economy. In the context of this environment we show that flexible exchange rates are optimal under monetary shocks and fixed exchange rates are optimal under real shocks.

Fifty years after Milton Friedman's (1953) celebrated case for flexible exchange rates, the debate on the optimal choice of exchange rate regimes rages on as fiercely as ever. Friedman argued that, in the presence of sticky prices, floating rates would provide better insulation from foreign shocks by allowing relative prices to adjust faster. In a world of capital mobility, Mundell's (1968) work implies that the optimal choice of exchange rate regime should depend on the type of shocks hitting an economy: real shocks would call for a floating exchange rate, whereas monetary shocks would call for a fixed exchange rate. Ultimately, however, an explicit cost/benefit comparison of exchange rate regimes requires a utility-maximizing framework, as argued by Helpman (1981) and Helpman and Razin (1979). In such a framework, Devereux and Engel (2003) reexamine this question in a sticky prices model and show how results are sensitive to whether prices are denominated in the producer's or consumer's currency. On the other hand, Cespedes, Chang, and Velasco (2004) incorporate liability dollarization and balance sheets effects and conclude that the standard prescription in favor of flexible exchange rates in response to real shocks is not essentially affected. An implicit assumption in most, if not all, of the literature is that economic agents have unrestricted and permanent access to asset markets.3 This, of course, implies that in the absence of nominal rigidities, the choice of fixed versus flexible exchange rates is irrelevant. In practice, however, access to asset markets is limited to some fraction of the population (due to, for example, fixed costs of entry). This is likely to be particularly true in developing countries where asset markets are much smaller in size than in industrial countries. Table 1 shows that even for the United States, the degree of segmentation in asset markets is remarkably high. The table reveals that, as of 1989, 59% of U.S. households did not hold any interest bearing assets (defined as money market accounts, certificates of deposit, bonds, mutual funds, and equities). More strikingly, 25% of households did not even have a checking account as late as in 1989. Given these facts for a developed country like the United States, it is easy to anticipate that the degree of asset market segmentation in emerging economies must be considerably higher. Since asset markets are at the heart of the adjustment process to different shocks in an open economy, it would seem natural to analyze how asset market segmentation affects the choice of exchange rate regime.4This paper abstracts from any nominal rigidity and focuses on a standard monetary model of an economy subject to stochastic real and monetary (i.e., velocity) shocks in which the only friction is that an exogenously-given fraction of the population can access asset markets. The analysis makes clear that asset market segmentation introduces a fundamental asymmetry in the choice of fixed versus flexible exchange rates. To see this, consider first the effects of a positive velocity shock in a standard one-good open economy model in the absence of asset market segmentation. Under flexible exchange rates, the velocity shock gets reflected in an excess demand for goods, which leads to an increase in the price level (i.e., the exchange rate). Under fixed exchange rates, the adjustment must take place through an asset market operation whereby agents exchange their excess money balances for foreign bonds at the central bank. In either case, the adjustment takes place instantaneously with no real effects. How does asset market segmentation affect this adjustment? Under flexible rates, the same adjustment takes place. Under fixed exchange rates, however, only those agents who have access to asset markets (called “traders”) may get rid of their excess money balances. Non-traders – who are shut off from assets markets – cannot do this. Non-traders are therefore forced to buy excess goods. The resultant volatility of consumption is costly from a welfare point of view. Hence, under asset market segmentation and in the presence of monetary shocks, flexible exchange rates are superior to fixed exchange rates. Asset market segmentation also has crucial implications for the optimal exchange rate regime when shocks come from the goods market. We show that when output is stochastic, non-traders in the economy unambiguously prefer fixed exchange rates to flexible exchange rates because pegs provide a form of risk pooling. Under a peg, household consumption is a weighted average of current period and last period's output which implies that the consumption risk of non-trading households is pooled across periods. Under flexible rates, however, the real value of consumption is always current output which implies no intertemporal risk sharing. Trading households, on the other hand, prefer flexible exchange rates to fixed exchange rates since maintaining an exchange rate peg involves injecting or withdrawing money from traders which makes their consumption more volatile under a peg. However, trading households' access to asset markets ensures a much smaller increase in their consumption volatility relative to the reduction in consumption volatility of non-trading households. Using a population share weighted average of the welfare of the two types, we show that under fairly general conditions, the non-traders' preferences dominate the social welfare function. Hence, when output is stochastic, an exchange rate peg welfare dominates a flexible exchange rate regime. In sum, the paper shows that asset market segmentation may be a critical friction in determining the optimal exchange rate regime. Our paper is related to an older literature on exchange rate regimes. Perhaps the closest paper is Fischer (1977) who showed that in an economy with no capital mobility, fixed exchange rates produced better outcomes than flexible exchange rates when shocks are real while flexible exchange rates are better when shocks originate in the money market. There are two main differences between Fischer (1977) and our paper. First, we solve a micro-founded optimizing model while Fischer obtained his results in the context of a reduced-form model. The reduced-form nature of the model made his analysis unsuitable for a choice-theoretic welfare analysis. Second, we analyze an economy with heterogenous agents whereas Fischer did not. In our model, agent heterogeneity is key to understanding the role of monetary policy in affecting real outcomes. In our framework, monetary policy can react to output disturbances by redistributing resources from one agent to the other. This channel is critical in achieving the first-best in our model and is missing completely in Fischer's model. In fact, it is this heterogeneity of agents which also differentiates our work from the work of Helpman and Razin (1982). Helpman–Razin studied an environment with uncertainty and incomplete markets to show that flexible exchange rates can produce higher welfare than fixed exchange rates. However, the key feature of our model is incomplete market participation — some agents are absent from asset markets. In a previous version of this paper we have shown that our results carry over to the complete asset markets case. Hence, what is central for our results is incomplete market participation, not incomplete asset markets. The paper proceeds as follows. Section 2 presents the model and the equilibrium conditions while Section 3 describes the allocations under alternative exchange rate regimes and derives the optimal regime under monetary and output shocks. Section 4 studies the optimal, first-best monetary policy rule. Finally, Section 5 concludes. Algebraically tedious proofs are consigned to the Appendix.

The determination of the optimal exchange rate regime for an open economy is one of the oldest issues in international economics. While there exists a very long and old literature on this topic, most of the work in this area has been conducted in the context of environments with some sort of nominal rigidity – either in wages or in prices. In this paper we have studied an entirely different environment wherein the key friction is in asset markets. In particular, we have analyzed a model in which only a fraction of agents trades in assets. In this environment fixing exchange rates entails central bank interventions in the asset market where only a fraction of agents is present. Hence, monetary shocks (shocks to velocity in our context) under fixed exchange rate regimes cause redistributions across agents thereby generating consumption volatility. On the other hand, when exchange rates are flexible, monetary shocks cause changes in the price level which insulate agents' real balances. Thus, asset market segmentation causes an inherent welfare bias towards flexible exchange rate regimes when shocks are monetary. On the other hand we have also found that when the economy faces output shocks, under fairly general conditions, fixed exchange rates unambiguously welfare dominate flexible rates. It is worth nothing that while we have derived the optimal state contingent money growth rule which implements the first-best allocation in this economy, we have not undertaken a detailed qualitative and quantitative comparison of state-contingent rules with optimal non-state contingent rules which would also include the Friedman rule. This is an interesting research topic and is the focus of attention in a related paper (Lahiri et al., 2004). We have also ignored the issue of endogeneity of market segmentation. In particular, one would expect that agents endogenously choose to be traders or non-traders with the choice being sensitive to the cost of participating in asset markets as well as the prevailing exchange rate and/or monetary regime. However, we see no reason to believe that this would change our key results. As should be clear from the intuition provided in the paper, what matters for our results is that, at every point in time, some agents have access to assets market while others do not. Which particular agents have access to asset markets and whether this group changes over time should not alter the essential arguments. A formal check of this conjecture is left for future work.