(b)  Flexible Exchange Rates as a Cause of Shocks

The major theoretical development in this literature in the last decade has concerned the working of asset markets when financial markets are incomplete. This issue is important for the optimal currency area literature because it suggests and formalises reasons why a separate currency can be welfare reducing. The main insight is that if the value of a currency is determined by factors unrelated to economic fundamentals, the use of the currency distorts the ability of agents to enter contracts aimed at reducing the fundamental risk they face. If the “noisiness” of the currency is sufficiently great, agents forced to use the currency will be worse off than if they were able to use a different currency. This idea, expressed as a concern that flexible exchange rates are a cause of economic shocks as well as a means of adjusting to economic shocks, is one of the main reasons forwarded by the European Commission for forming a monetary union.

There are four components to this literature:

1. reasons why financial markets are incomplete;
2. reasons why financial asset values can deviate from fundamentals;
3. reasons why noisily priced assets can be welfare reducing;
4. reasons why such assets would continue to be used even when they are welfare reducing in the aggregate.

Each step relies on some combination of transaction costs, information asymmetry, or bounded rationality, and thus involves some reasonably complex arguments and modelling techniques.

(i) Incomplete Markets[36]

It is quite obvious that financial markets are incomplete - that there do not exist financial assets that agents can use to contract for the delivery of different types of goods in all possible different states of the world in all future time periods. It is also somewhat obvious that the incompleteness of these markets is the result of people’s limited ability to deal with uncertainty, their limited ability to process information, and the expense of writing and enforcing contracts. Market incompleteness means that people can not perfectly insure themselves against future random events, but they can enter some financial contracts, and accumulate different types of monetary assets, to provide themselves with limited insurance. The combination of limited rationality and information and the potential difficulty of enforcing contracts with others means that agents tend to use only those assets which are the easiest to use, about which they can most confidently predict the future state contingent values, and whose consequence for the parties involved can be predicted with relative accuracy.[37] In practice this means using a small number of financial contracts to provide nominal purchasing power in the future, and then using spot commodity markets to convert purchasing power into goods.

One of the prime contracts that agents use to allocate income into the future are simple monetary contracts that deliver one nominal unit at a specified time. The real value of these contracts in different locations will of course depend on prices in different locations in the future. These prices will depend on various aspects of the state of the world, including the geographic distribution of the contracts entered into by other people. For this reason, the relative value of different currencies in different states of the world will clearly have a major effect on agent’s welfare. Consequently, forward planning agents who are trying to choose the mix of contracts that best suits their risk preferences in the face of different types of shocks will be best served if the values of different currencies in different circumstances are reasonably predictable. Put differently, the usefulness of different currencies will be enhanced if their actual value in different states of the world is as close as possible to their predicted values in these states. Consequently, a desirable quality for a currency is that its value can be predicted as a function of various fundamental macroeconomic variables.

(ii) Noisy Asset Values

In the last two decades there has been considerable debate among economists as to whether the values of a variety of types of assets can be adequately explained by fundamental considerations, or whether asset values are in some sense excessively volatile. While it is not possible to prove one way or the other, there is a growing consensus among economists that exchange rates are excessively volatile, and that there is little short term relationship between exchange rates and economic fundamentals even if exchange rates eventually reflect fundamental factors in the longer term. Three types of evidence are presented:

1. here is almost no useful relationship between fundamentals such as relative inflation rates or interest rate differentials in the short and medium terms even though in the long term ¾ a period five years or more - some fundamentals appear to explain long term exchange rate movements;[38]
2. macroeconomic models appear incapable of out-forecasting a random walk model at short and medium term horizons, even when actual future values of the macroeconomic variables are used in the forecasts[39];
3. the volatility of all real variables except the real exchange rate seem to be unaffected by the exchange regime.[40].

These points are amplified in Appendix 1.

If macroeconomic models cannot be used to forecast the exchange rate, except in the longer term, and if the volatility of observed real macroeconomic variables except the real exchange rate are unaffected by the exchange rate regime, it is difficult to believe that observed macroeconomic variables are the cause of most exchange rate movements. Frankel and Rose (1995) deduce from these two observations that either unobserved macroeconomic shocks cause exchange rate volatility, or that exchange rate movements are unrelated to macroeconomic fundamentals. Dismissing the former explanation, they seek a solution in the microstructure of the foreign exchange market, including the factors that explain participants’ heterogeneous expectations.

The heterogeneity of exchange rate expectations is a further fault with macroeconomic models of the exchange rate, as is the huge volume of foreign exchange rate trade that occurs in practice. In a standard rational expectations model, all participants have the same expectations and there is little trade because prices adjust to the level where there are few non-fundamental gains from trade. Survey evidence suggests convincingly that participants have a wide range of expectations about the future values of exchange rates, and that there is extensive trade based on these beliefs. The diversity of expectations about the exchange rate tends to be larger than the diversity of beliefs about macroeconomic variables.

A model of flexible exchange rates parameterised to empirical facts about the world should be able to explain the observed diversity of exchange rate expectations, the volume of trade, the high short term volatility, and the role played by macroeconomic fundamentals in the long term. For a model to be able to do this successfully, Frankel and Rose (1995) argue it will need to:

1. have short term dynamics arising from the trading process which generate volatility which swamps fundamentals;
2. explain why market participants have limited “rational expectations” arbitrage; and
3. provide a role for fundamentals which limits the extent to which the exchange rate can deviate from fundamentals in the short and medium terms.

Fortunately, several models of the microstructure of asset markets have been developed over the last decade which provide general foundations for this work. These models were largely developed to explain why the share market is so volatile, but are being transformed to incorporate institutional details of the foreign exchange market to see if they are relevant to explain the process by which exchange rates are determined. These models have concentrated on three features of the market – the process of expectations formation; the dissemination of information through the market; and the role of risk aversion. Some fairly general results have emerged. Of these, possibly the most important is the demonstration that an asset price can deviate from fundamentals for long periods of time without there being a strong incentive for “fundamentals speculators” to drive the asset price back to its fundamental value. Moreover, it can be shown that price speculation can be destabilising without impoverishing those agents who do the destabilising, a result counter to normal economic intuition. Two classes of these models are briefly discussed in Appendix 2.

This large literature provides both an empirical and a theoretical basis for believing that exchange rates can deviate from their fundamental values in the short to medium term. This potentially reduces the usefulness of having a separate local currency as a unit of account in which to denominate contracts, for the real value of this currency may not be particularly predictable.

(iii) Conditions when a Separate Currency can be Welfare Reducing.

It is usually argued that a separate and flexible currency is a useful tool for macroeconomic stabilisation, because it enables relative prices to change in response to adverse economic shocks when wages and prices are sticky. This is correct as stated. However, it can also be argued that excessive volatility in the exchange rate can be a cause of economic shocks to the rest of the economy. This argument has recently been formalised within an incomplete markets setting by Neumeyer (1998). The basis of the argument is that when there are incomplete markets, an additional currency should increase the insurance possibilities open to agents through trade in nominal assets, because the different assets will have different state-contingent values. Ordinarily this would mean ...

“the loss of monetary independence entailed by fixed exchange rate regimes, or monetary unions, is socially costly because it makes the real payoff of assets denominated in different currencies equivalent, effectively reducing the number of financial instruments with which economic agents can share risks.” (Neumeyer, p 247)

However, when the value of an additional currency is not closely related to economic fundamentals, this result is not true. The noise associated with the asset means that its value will not be highly predictable in different economic states of the world. This is not just problematic for those undertaking contracts in the particular asset; because the random outcomes of the value of this asset affect income distribution among all who use it, prices and therefore the real values of all other assets will be affected as well. In this case the value of insurance possibilities available from all other assets is reduced, because their real value will be contaminated by the randomness of the additional currency. Eliminating the asset will increase the insurance value of all remaining assets, as the predictability of their real value in the face of economic shocks increases.

This result is rather important. An excessively volatile currency is not just bad for those forced to use it, but it can be bad for others as well, because it can disrupt the working of the entire financial system.