A Macroeconomic Analysis of EU Accession under Alternative Monetary Policies*

JCMS 23 Volume 41. Number 5. pp. 941 64 A Macroeconomic Analysis of EU Accession under Alternative Monetary Policies* MICHAEL B. DEVEREUX University of British Columbia Abstract This article provides an analytical discussion of the adjustment to EU accession for an economy under alternative assumptions about monetary policy rules. The postaccession phase is characterized by rapid capital inflows and real exchange rate appreciation. If accession is combined with membership of the euro area and a pegged exchange rate, then the post-accession period exhibits excessive foreign borrowing, high wage inflation, an excessive stock market boom, and much too rapid growth in the non-traded sector at the expense of the exportable goods sector. Alternative monetary policies can be used to eliminate the inefficiencies of the post-accession adjustment, but some bring real costs in terms of lower growth and unemployment. We find that the best policy is one of flexible inflation targeting with some weight on exchange rate stability. In the absence of exchange rate adjustment, fiscal policy could be used, but this requires complicated time-varying expenditure taxes. While the analytical discussion emphasizes the benefits of exchange rate adjustment, a later section of the article explores some more recent arguments regarding non-traditional costs of exchange rate volatility. Introduction The accession countries of central and eastern Europe (CEECs) face a rapid process of adjustment as they approach the time of entry into the EU. Their experience is in part quite similar to that of other emerging market economies in Latin America and East Asia. Like those countries, the CEECs have to deal with rapid capital inflows from the rest of the world, and structural change within their domestic economies. But they are unlike the typical emerging market economies in other ways. For one thing, many of the CEEC econo- *I thank SSHRC and the Bank of Canada Fellowship program for research funding. The views expressed here do not represent those of the Bank of Canada., 96 Garsington Road, Oxford OX4 2DQ, UK and 35 Main Street, Malden, MA 2148, USA

942 MICHAEL B. DEVEREUX mies were already industrialized before the beginning of the transition from central planning, and have had to deal with the transformation to a marketbased industrial economy. But a second key feature of the experience of the CEECs that differentiates them from the conventional emerging market economies is that, as a counterpart of accession to the EU, they will face severe limitations on the use of domestic macroeconomic policy instruments. Since these countries are likely to become members of ERM II immediately upon accession or very soon afterwards, their monetary policies will be tightly constrained by the need to maintain a peg to the euro. Likewise, although revisions to the Stability and Growth Pact are likely to be made in the next few years, it is still reasonable to assume that full membership of the euro area will impose some restrictions on independent fiscal policies. At the same time, however, these countries will be facing a period of substantial real adjustment, which would in general require the independent use of monetary and fiscal policies. This article will examine the implications of post-eu accession experience for the CEECs, and in particular highlight the nature of the post-accession adjustment problems that may arise due to the inability to use independent monetary and fiscal policies. The approach is to develop this analysis within a structural dynamic general equilibrium model of a small open economy (adopted from Devereux and Lane, 23b). Using this model, the process of post-accession adjustment within a variety of scenarios is tracked. In particular, the adjustment process is first examined in a frictionless economy, where there is no independent need for monetary or fiscal policy tools to guide the adjustment. This is then contrasted with an economy with nominal frictions, where the adjustment process may be hampered by the inability of internal prices and wages to adjust sufficiently quickly following accession. In such an environment, the monetary policy rule becomes of critical importance to the adjustment process. Two alternative monetary rules are contrasted. In the first, the adjustment process is constrained by immediate adoption of the euro, so the nominal exchange rate cannot change at all during adjustment. In the second, the country may follow an inflation-targeting monetary rule, allowing for the exchange rate to move during the adjustment process. What is the nature of the adjustment that the CEECs will experience? The focus is on two types of structural shifts that follow EU accession movements in country-specific risk premia (the cost of foreign borrowing), and movements in productivity within the domestic exportable sector. Roughly speaking, these can be viewed as capital inflow shocks, and internal productivity shocks. adjustment to both shocks requires a real exchange rate appreciation. The capital inflow shock requires real appreciation because

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 943 the economy experiences a rise in domestic consumption and investment demand. The productivity shock requires a real appreciation through the familiar Balassa-Samuelson mechanism. 1 Section III of the article shows that the post-accession adjustment constrained by a fixed exchange rate (i.e. euro area membership) deviates significantly from the process that would occur in a frictionless economy, where prices could adjust immediately. The key difficulty is that neither the real exchange rate nor real wages can adjust quickly enough with nominal rigidities. Under the fixed exchange rate constraint, the combination of capital inflows and export sector productivity growth gives rise to excessive foreign borrowing, high domestic wage and price inflation, and an excessive boom in the domestic stock market, relative to the efficient adjustment process. In addition, there is an excessive boom in the domestic economy. The general equilibrium, welfare-based modelling strategy provides ways in which to evaluate alternative paths of adjustment for the economy. From a welfare standpoint, a boom is excessive if economic activity is higher than it would be in a frictionless economy. 2 By contrast, if monetary policy is unconstrained, then a policy of inflation targeting can be used to avoid the inflationary boom that follows accession. If wage rates were flexible, then a policy of strict inflation targeting (stabilizing the price of non-traded goods) could be used so as to achieve exactly the efficient adjustment process in the post-accession period. But with slowly adjusting wage rates, strict inflation targeting eliminates the post-adjustment inflation only at the cost of a domestic recession and a rise in unemployment. In this case, a type of flexible inflation and exchange rate targeting is shown to be much more preferable, closely tracking the path of the frictionless economy in the post-accession environment. But a basic prerequisite (for either type of inflation targeting) is that the nominal exchange rate undergoes an immediate appreciation following accession. A number of possibilities are also discussed for post-accession adjustment that allows for exchange rate stability, but a faster response of the real exchange rate. One suggestion is to have a one-time nominal appreciation at the time of accession. Another is to use domestic tax policy to prevent the immediate over-expansion in the non-traded goods sector following accession. Expenditure taxes on non-traded goods can be used to achieve this. But this 1 The Balassa-Samuelson mechanism refers to the necessity for real exchange rate appreciation (an increase in the domestic price level relative to that in the rest of the world) in response to a high rate of productivity growth in the export, or traded goods sector (see Sinn and Reutter, 21, for discussion of this effect in Europe). 2 Thus, the methodology can judge a boom to be excessive even within a region of a currency area (e.g. the case of Ireland), where there is no danger of high growth and inflation leading to a subsequent currency crisis.

944 MICHAEL B. DEVEREUX case must follow a complicated, time-varying path for taxes, and it is unclear that this would be administratively feasible. While the analytical model emphasizes the benefits of independent monetary policy and nominal exchange rate adjustment, many recent writers have questioned the application of this type of analysis to emerging market economies. In addition, as noted by Calvo and Reinhart (22), many emerging economies seem to be distinctly averse to exchange rate variability. In a later section of the article, a number of recent non-traditional arguments in favour of stable exchange rates are discussed. In each case, the implication of these alternative perspectives is that less importance be given to domestic monetary policy and more importance be put on exchange rate stability. Section I outlines the model, while Section II discusses some details of the solution. Section III contains the main analytical discussion. Section IV covers some alternative arguments about the costs of exchange rate flexibility, while Section V concludes. I. Monetary Policy in a Small Open Economy Here a simple framework for the analysis of post-accession macroeconomic experience for a small economy is developed. The assumption is that accession to the EU precipitates two main macroeconomic shocks: a fall in the country risk premium, and a rise in the level of productivity in the export goods sector. Both require a large increase in the real exchange rate. 3 But the immediate adjustment in the short run depends critically on the stance of monetary policy. Two alternative monetary policy rules are contrasted: one where the country immediately joins the euro area and therefore must keep its nominal exchange rate fixed, and a second rule which involves a policy of domestic inflation targeting. Model Structure The structure is a standard two-sector dependent economy model. Two goods are produced: a domestic non-traded good, and an export good, the price of which is fixed on world markets. The analytical structure of the model is described in detail in Devereux and Lane (23b). Here, only the broad features of the model are described. A central underpinning is the presence of nominal rigidities. Slow nominal price and wage adjustment means that the 3 In the long run in the model, the real exchange rate is driven solely by supply-side factors. A rise in domestic productivity of export goods generates a long-run real appreciation because of Balassa- Samuelson effects (see, e.g., Obstfeld and Rogoff, 1996). The fall in the country risk premium causes a long-run real appreciation because it leads to a permanent fall in the cost of capital, and export production in the model is relatively capital intensive.

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 945 monetary and exchange rate policies during the post-accession period are of critical importance for macroeconomic outcomes. But, unlike a large closed economy analysis, the source of nominal rigidities must be different in this analysis. Few small economies would be expected to have much market power in export or import sectors. Rather, the prices of both goods will be determined by international competition. By contrast, in the domestic non-traded sector, firms are sheltered from competition, and there may exist substantial nominal rigidities. Hence, it is assumed that prices of non-traded goods can adjust only slowly over time following a macroeconomic shock. In addition, it is assumed that domestic wages are also pre-set, adjusting slowly in response to shocks. There are four sets of domestic actors in the model: consumers, firms, a monetary authority and a fiscal authority. In practice there is no real differentiation between the monetary and fiscal authority, as will become clear in the description below. In addition, there is a rest of world sector where foreigncurrency prices of export and import goods are set, and where lending rates are determined. Consumers Consumers supply labour to firms in both the non-traded sector and the export sector and consume domestic, non-traded goods as well as an imported good. In addition, they have access to international capital markets, where they can lend or borrow, as well as accumulating claims on domestic physical capital. There are adjustment costs of investment based on the Tobin q hypothesis. 4 As a result, the model has an endogenous domestic stock market. Foreign borrowing is subject to a country-risk premium, however, which is assumed to be exogenous from the point of view of households and the government of the small economy. The consumer s decision determines the demand for non-traded goods (arising from both consumption and investment demand), demand for imports and labour supply. Wages are assumed to be pre-set, and adjust slowly over time as each wage contract comes up for renewal. 5 Firms Production is carried out by firms in the non-traded and export sectors. Firms combine labour and capital to produce output with constant returns to scale 4 According to the Tobin q approach to investment, a firm will undertake new net investment whenever the value of physical capital in production exceeds the replacement cost of that capital, where all costs (including depreciation and interest costs) are used to compute the replacement cost. 5 Barry et al. (23) argue that wage rigidity is an important part of the post-accession adjustment process in the Czech Republic.

946 MICHAEL B. DEVEREUX production functions. The sectors differ in their production technologies (to be discussed further in the calibration section). While export-sector firms are competitive and take prices as given, non-traded firms are imperfectly competitive, and set prices in advance. Price setting is staggered, so that at any given date only a fraction of price contracts come up for renewal. This ensures that the adjustment to a domestic or external shock in the small economy takes place at a gradual rate. The cost-minimizing behaviour of firms determines the demand for factors of production. It is assumed that labour is instantaneously mobile across the two sectors, but capital is not. It takes time for capital accumulation to take effect and for capital to move into each sector. Monetary Policy Rules In this model, the monetary authority uses a short-term interest rate as the monetary instrument. Subject to some technical conditions that ensure a unique price level, an interest rate rule provides a complete characterization of monetary policy in an open economy. 6 The details of the interest rate rule are discussed in the Appendix. The essential features of the rule involve adjusting the nominal (short-term) interest rate, which is the monetary authority s instrument of control, in order to meet one or a number of monetary policy objectives. One such objective is a fixed exchange rate. In this case, the authorities increase the interest rate automatically in response to any deviation of the exchange rate from a target peg. An alternative monetary rule is inflation targeting whereby the authorities adjust the interest rate upwards in response to any deviation of the inflation rate from its target level. This could be defined in terms of the inflation rate of domestic (i.e. non-traded) goods, or the overall consumer price index (CPI). The focus is mainly on the former, so that the price inflation target is domestic inflation. In this model, because there is immediate pass-through of exchange rate changes into traded good prices, and foreign prices are exogenously given, CPI inflation targeting is very close to a fixed exchange rate because, given that non-traded goods prices are sticky, such a rule essentially implies fixing the exchange rate. Finally, two types of inflation targeting are compared. Strict inflation targeting aims to keep the price level completely stabilized, while more flexible targeting adjusts interest rates roughly according to a Taylor rule. 7 6 See Woodford (1999) for conditions on interest rate rules required for uniqueness in the price level. In addition, see Clarida et al. (1999). 7 A Taylor rule is defined as a monetary rule that increases the nominal interest rate more than one for one with inflation, or forecasted inflation. See Taylor (1993) for the original exposition of this type of monetary rule. This type of rule is widely seen as a reasonable description of monetary policy-making in inflation targeting countries (see, e.g., Bernanke et al., 21).

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 947 This analysis will compare the performance of the model along a number of dimensions, as the economy adjusts to the accession. Since the main reason to be concerned with the monetary policy rule is the ability to cope with the presence of slow nominal adjustment, the way in which the economy would adjust with no nominal rigidities at all, but where the monetary authority followed a policy of inflation targeting, is used as a benchmark for comparison. This is in essence equivalent to a post-accession environment where prices and wages are fully flexible. This benchmark is compared to the actual adjustment predicted by the model when price and wage rigidities do exist, and the monetary rule can be one where the exchange rate is either kept fixed (e.g. the country immediately takes on ERM II membership) or the monetary authority follows an inflation-targeting rule. Model Equilibrium As a full dynamic general equilibrium framework, the model solution implies a path for inflation, interest rates and exchange rates, as well as consumption, investment, employment, output and the current account. In general, however, there is no explicit mathematical solution for the model equilibrium. Instead, following standard literature, numerical solution techniques are used. In order to proceed some assumptions need to be made regarding the choice of parameter values and functional forms. II. Solution and Calibration The model is explicitly constructed as an application to monetary policy in central and eastern European accession countries. In this light, in calibrating the model, both parameter values and shocks in the relevant range are used. One of the key parameters concerns the size of the non-traded goods sector. If the non-traded goods sector comprises a very small part of the economy, then price rigidities are less important, and the whole question of monetary policy rules is then less pressing. Estimates for the importance of non-traded goods can be obtained by computing the magnitude of the service and construction sector as a proportion of total GDP. Using IMF statistical estimates it is estimated that the non-traded sector comprises 65 per cent of GDP in Poland and about 6 per cent in the Czech Republic. Hence the share of nontraded goods in total GDP is set at.6. There is no direct evidence on the factor intensity of each of the sectors, but it is reasonably assumed that the non-traded sector is more labour intensive than the export sector. In the absence of estimates of other parameter values, such as the depreciation rate of capital, the costs of adjustment in investment, and the elasticity of labour supply, these are chosen to equal the values common in the recent

948 MICHAEL B. DEVEREUX literature (see Devereux and Lane, 23b). In addition, it is assumed that both wages and the prices of non-traded goods are adjusted on average every four quarters. This introduces a degree of wage and price stickiness that is commonly used in models of the US and European economies. Post-Accession Shocks The focus is on two of the most important factors in the discussion of the effects of EU accession. The first is the increasing availability of external capital. There are many ways in which this could come about: an elimination of capital controls (either official or unofficial); an increase in trade flows directly stimulating foreign direct investment; or an increase in the creditworthiness of domestic firms. Any of these effects may be roughly captured by a fall in the external country risk premium. A fall in the external risk premium will stimulate borrowing and capital inflows into the domestic economy. Baldwin et al. (1997) argue that one of the key effects of EU accession will be to reduce the country risk premium on external borrowing. A permanent fall in the cost of capital will increase the size of the traded goods sector and, given labour-intensive non-traded goods, imply a long-run real appreciation. A second effect of EU accession is on productivity. Buiter and Grafe (22) argue that eastern European countries are likely to exhibit rapid convergence in traded goods productivity in the run-up to accession to EMU. This naturally requires a real exchange rate adjustment via the Balassa-Samuelson mechanism. A productivity shock can be regarded as representing this channel of convergence. Thus, the analysis provides an insight into the appropriate monetary policy rules to follow in the face of productivity convergence. The way in which these two structural changes affect the model will naturally reproduce many of the key features that are emphasized in the transition experience of eastern European and other emerging market economies. The fall in the external risk premium will generate an endogenous rise in capital inflows and a boom in the domestic economy, and the rise in export goods productivity will lead to a jump in domestic investment and the stock market. Since quantitative estimates of the two shocks are very difficult to obtain, the risk premium shock and the productivity shock are scaled so that they have equivalent effects on long-run output. III. The Adjustment to EU Accession Here a comparison of the adjustment process under alternative monetary policy strategies after EU accession is presented. Note that the whole exercise presupposes that the country does not immediately become a member of the euro area upon accession, and so can exercise some independence in monetary

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 949 policy. This section may therefore be thought of as a comparison of whether the country immediately joins ERM II and follows a policy of tightly pegging the exchange rate to the euro, or (less likely) delays joining ERM II and follows an independent policy of inflation targeting. In either case, the aim is to give an analytical outline of the macroeconomic consequence of accession under alternative monetary rules. The results are presented in a series of panels in Figures 1 and 2. These describe the impact of the two shocks on various aspects of the economy under different monetary rules. In Figure 1, each panel contains three loci which represent, respectively, the response of the economy with flexible wages and prices, the response under a fixed exchange rate, and the response under an inflation targeting rule. The response in the economy without wage and price rigidities is first described. This represents a benchmark for comparison. Note that, in the absence of other sources of distortion, this outcome is also the desired or optimal macroeconomic response of the economy. 2 1.5 1 Inflation targeting Exchange rate peg.5.5.5 1 1.5 Inflation targeting Exchange rate peg.5 Figure 1a: Output 2 Figure 1b: Real Exchange Rate.3 1.2.1 Inflation targeting Exchange rate peg 1 2 3 Inflation targeting Exchange rate peg.1 Figure 1c: Inflation 4 Figure 1d: Trade Balance

95 MICHAEL B. DEVEREUX 2.5 4 2 1.5 1 Inflation targeting Exchange rate peg 3 2 1 Inflation targeting Exchange rate peg.5 1.5 Figure 1e: NT Output 2 Figure 1f: Employment 3 1 2 1 Inflation targeting Exchange rate peg.5 Inflation targeting Exchange rate peg 1 Figure 1g: Stock Market.5 Figure 1h: Wage Inflation Adjustment in a Frictionless Environment The first line in each panel of Figure 1 describes the adjustment to accession in an economy where there are no nominal frictions, and the monetary rule is one of strict inflation targeting. In this case, the real adjustment is independent of the monetary policy rule. The combination of higher productivity in the export sector and the fall in the external cost of borrowing leads to an immediate rise in the trade deficit, as the economy borrows both to consume and invest. GDP grows little in the beginning, but gradually increases to its new higher long-run level. The boom in capital inflows to finance both investment and consumption leads to a boom in the non-traded goods sector. Total employment moves little, increasing only slightly over time. This is because employment is subject to conflicting influences. On the one hand, the increase in export goods productivity and rise in capital accumulation in both sectors directly increases the demand for labour (since exporting firms are unrestricted by domestic aggregate demand). But the rise in overall wealth tends to reduce the supply of labour. Since, in this case, wages are fully flexible, the labour market clears, and overall employment is relatively unchanged.

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 951.4.5.3.2.1 Flexible inflation targeting.5 1 1.5 Flexible inflation targeting Figure 2a: Output 2 Figure 2b: Real Exchange Rate.5 1.4.3.2 Flexible inflation targeting 1.1 2 Flexible inflation targeting.1 Figure 2c: Inflation 3 Figure 2d: Trade Balance Note that an efficient adjustment of the real economy involves a sharp initial real exchange rate appreciation. Since the domestic monetary policy is one of strict non-traded goods inflation targeting, this means that the nominal exchange rate appreciates, and hence there is negative CPI inflation. This is mirrored by a fall in nominal wages (although real wages clearly rise). The figure also shows that there is a stock market boom, as the price of capital (Tobin s q) rises to reflect the increase in investment following the accession. In the long run, the trade balance deficit is reversed, and the economy shifts towards trade surplus, as the rise in external debt is repaid. In the process, the real exchange rate depreciates over time, but settles at a higher steady state level (relative to the initial conditions), to reflect the permanently higher level of productivity in the export sector. Adjustment under a Fixed Exchange Rate Figure 1 also illustrates the post-accession adjustment in the presence of slow price adjustment in the non-traded sector and sticky wages, assuming that the adjustment must occur without any changes in the nominal exchange rate.

952 MICHAEL B. DEVEREUX.6.2.4.1.2 Flexible inflation targeting.1 Flexible inflation targeting.2.2 Figure 2e: NT Output.3 Figure 2f: Employment 1.5.4 1.5 Flexible inflation targeting.2.2 Flexible inflation targeting.5 Figure 2g: Stock Market.4 Figure 2h: Wage Inflation The presence of wage and price rigidities has a major impact on the adjustment dynamics of the economy. Non-traded goods prices can rise only gradually in response to the increase in demand fuelled by the capital inflows. As a result, the immediate real exchange rate appreciation is muted. This leads to a much bigger boom in the non-traded goods sector, and an immediate boom in aggregate GDP. Given the slow adjustment of wages, employment rises sharply, much more than in the flexible wage-price economy. Since the nominal exchange rate is prevented from adjustment, real appreciation has to take place via an increase in inflation in the non-traded goods sector. This causes an increase in wage inflation. Since absorption is higher than with flexible wages and prices, the deterioration of the trade balance is greater. Hence, the postaccession economy with wage and price rigidities experiences excessive capital inflows under a fixed exchange rate. Finally, there is an excessive rise in the stock market, relative to the flexible wage and price economy, reflecting the higher demand for capital goods in both sectors. Note that the long-run equilibrium adjustment in the economy with nominal rigidities is effectively the same as that in the economy with efficient

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 953 adjustment. The differences arise only with respect to the consequences of servicing a higher long-run level of net foreign debt in the economy with price rigidities. This case therefore allows us to outline in precise form the nature of the temporary adjustment problem in the post-accession economies that give up the possibility of nominal exchange rate adjustment. When the nominal exchange rate must be kept constant, there is an excessive boom in the domestic economy following accession. This boom is financed by overborrowing and over-investment, and is characterized by excessive wage inflation, and an excessive boom in the domestic stock market. How might authorities react to this boom? In the next sub-section, the constraint that the nominal exchange rate must be kept constant in the adjustment to accession is removed. Adjustment under Inflation Targeting The previous sub-section makes clear that a post-accession policy which keeps the nominal exchange rate fixed is associated with inflation in wages and domestic (non-traded) goods. The high domestic inflation is the only way that a real exchange rate appreciation can materialize. But this may involve an inflation rate significantly higher than the European average, endangering the accession country s eligibility for euro area membership. To avoid this inflation, accession countries might focus on targeting domestic inflation rather than the exchange rate. Now assume that the monetary authority follows a policy of strict inflation targeting in the non-traded goods sector. Hence, interest rates adjust so as to keep non-traded goods inflation constant. If wages were fully flexible, this policy would in fact replicate the outcome of the economy with flexible nontraded goods prices. This result parallels the recent literature on optimal monetary policy in closed and open economy models (e.g. Clarida et al., 1999; Benigno and Benigno, 21). By adjusting monetary policy to target the price level (in this case the price of non-traded goods), the monetary authority can completely undo the underlying constraint due to the inability of non-traded goods prices to adjust quickly. But with slow wage adjustment, the monetary policy problem is more complicated, and a policy of strict inflation targeting cannot completely undo the effects of nominal rigidities. Figure 1 illustrates the adjustment under a strict inflation targeting monetary policy. The policy is by construction successful at avoiding domestic inflation following accession. Instead, the country experiences a sharp nominal exchange rate appreciation, which achieves the necessary real appreciation. But eliminating domestic inflation does not come without some costs. Strict inflation targeting requires a contractionary monetary policy that acts so as to

954 MICHAEL B. DEVEREUX offset the expansionary stimulus of the post-accession shocks. In contrast to the experience under a fixed exchange rate, GDP falls, as the economy is pushed into recession after accession. Employment falls, rather than rising sharply, as would take place under fixed exchange rates. The key problem can be seen from the movement of wages in Figure 1 under strict inflation targeting. The post-accession shock causes a sudden nominal exchange rate appreciation. Export goods firms would then need a fall in nominal wages to remain competitive. In the economy with flexible wages, this takes place immediately (as evidenced by the discussion above). But with sticky wages, this is not possible. Hence real wages for the export sector rise sharply, and export goods output falls, reducing both overall GDP and employment, given the slow adjustment of wages. Wage inflation also falls, but not nearly quickly enough to offset the negative effect of the appreciation. The non-traded goods sector is essentially flat initially, and then gradually expands so as to achieve the path that it would take under fully flexible wages and prices. A strict inflation targeting policy therefore avoids the post-accession inflation that would take place under fixed exchange rates. But it does so by dampening real demand so much that the economy experiences a recession and a rise in unemployment. The economy does avoid an excessive boom and an excessive rise in the trade deficit, price and wage inflation, and stock prices, but the policy is one of overkill in the sense that economic activity is less than would be achieved with flexible wages and prices. If wage rates were instead flexible, a strict inflation targeting policy would be much more palatable, as discussed above (in fact, it would replicate the equilibrium without sticky prices). But in reality, slow nominal wage adjustment must be taken into account in the monetary policy problem. The dilemma that policy-makers face is clear. If the exchange rate is pegged following accession, there are excessive capital inflows and high domestic inflation. A policy of strict inflation targeting avoids the over-borrowing and domestic inflation, but it does so at the cost of a recession and a loss of domestic employment. Is there an intermediate monetary policy which would do better in moving the economy along the path that would be generated by the flexible wage and price equilibrium? In the model, the answer is yes. A two-part policy of flexible inflation targeting and flexible exchange rate targeting can be designed so as to get very close to the flexible wage and price outcome. Figure 2 repeats Figure 1, except now the inflation targeting policy is based on a type of Taylor rule, adjusting the interest rate in response to nontraded goods inflation so as to stabilize inflation, without trying to keep it at exactly zero and, in addition, adjusting interest rates upwards slightly in response to a depreciation in the exchange rate from the target level. As the figure makes clear, this hybrid policy involves a much lower inflation rate

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 955 than under the pegged exchange rate, but at the same time leads GDP almost exactly to track that of the flexible wage and price economy. Moreover, employment is completely stabilized on impact, and very closely follows the flexible wage and price outcome thereafter. The effective benefit of this flexible inflation and exchange rate targeting is two-fold. First, it allows for some domestic inflation, ensuring that the necessary real exchange rate appreciation takes place. But also it limits the initial exchange rate appreciation, thus reducing the real wage pressure on exporting goods firms. The combination of these two things means that the economy adjusts in almost exactly the desired manner. Adjustment with Initial Revaluation A notable aspect of the model dynamics in the flexible wage and price postaccession period is that the relative price changes take place quickly. The economy requires a big initial real exchange rate appreciation, followed by a gradual real depreciation to the new steady state real exchange rate. This suggests a natural strategy for post-accession adjustment, while still remaining within the EMU constraints (i.e. still keeping the exchange rate pegged). On accession, the country could undertake a one-time nominal exchange rate appreciation, and thereafter follow a fixed exchange rate. This essentially involves selecting a higher entry level exchange rate to begin ERM II membership than the initial market rate at accession. This would achieve a relative price change without the need for a high rate of post-accession inflation. Is this policy beneficial? The answer depends on the degree of flexibility in wages. If wages are fully flexible, then an initial nominal appreciation could be used to achieve the desired real exchange rate. But with sluggish nominal wages, this would not work, because the one-time nominal appreciation, while achieving the desired real exchange rate (relative price of domestic to foreign goods), would generate an increase in the real wage for the export sector, and have the same problems as a policy of strict inflation targeting. More generally, however, this policy is likely to be politically infeasible, given the strict rules of nominal convergence applied to prospective euro area members. So, even if wages were flexible enough that there were efficiency arguments for a one-time nominal appreciation immediately on accession, the need to manage financial market expectations in order to ensure a smooth adjustment to membership of the single currency area, combined with the difficulty in pinpointing the required degree of nominal appreciation, would seem to rule out this two-part strategy of adjustment.

956 MICHAEL B. DEVEREUX Fiscal Policy in the Adjustment Process While accession countries may lose control of monetary/exchange rate policy, they will still retain control of fiscal policy. Can fiscal policy be utilized to smooth the adjustment process, in the absence of nominal exchange rate adjustment? This is quite a different question from the usual one of using fiscal policy as a stabilization tool when a fixed exchange rate rules out the use of monetary policy. The problem facing the post-accession countries will not be to stabilize their economies around a constant (or slowly growing) natural rate of output, but rather to ensure upwards adjustment in the natural rate without high inflation and excessive booms in asset prices and the real economy, and excessive foreign borrowing. A natural fiscal candidate is the use of tax policy. Given that the key distortion is the inability of non-traded goods prices and wage rates to adjust quickly enough, there is in fact a unique sequence of taxes that can be implemented that will exactly sustain the efficient path of the economy, in the absence of any nominal exchange rate adjustment. These would be structured as indirect taxes (or expenditure taxes), but falling on non-traded goods and wages. By choosing a path of taxes that exactly replicates the time path of price of non-traded goods and wage rates that is attained in the flexible wage and price economy under a fixed exchange rate, the authorities can ensure that the economy adjusts exactly as in the efficient case, with no adjustment required either in the exchange rate, wages, or the price of non-traded goods. While this type of tax-transfer scheme is correct in principle, again in practice it is questionable whether it is administratively feasible, since it involves a substantial degree of fiscal fine-tuning. Nevertheless, the general principle that expenditure taxes might be used in a broad way to avoid the initial excesses of the post-accession boom does have considerable merit. The key failing of the adjustment period is that there is an over-expansion in both sectors of the economy. This can be substantially offset with an initial tax increase that is gradually phased out over the adjustment period. By judiciously designing tax policy, the fiscal authority can avoid the excesses of the post-accession phase, while still maintaining a fixed exchange rate. Application to Central and Eastern European Countries The central and eastern European accession countries who enter the euro area of course give up all option of using the exchange rate as a macroeconomic adjustment mechanism. How costly is this likely to be? How appropriate is a single currency that includes these countries based on the above analysis? Table 1 illustrates real GDP growth rates for Germany, France, Spain and Italy, as well as Estonia, Latvia, Poland, Lithuania and Hungary, over the

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 957 Table 1: GDP Growth and Volatility 199 2 GDP Growth GDP Volatility France 1.8 1.3 Germany 2.9 2.8 Italy 1.6 1.1 Spain 2.5 1.7 Estonia a 3.7 4. Latvia b.8 7.4 Lithuania b.9 8.9 Poland 2.2 6.1 Hungary c.5 5.4 Notes: a since 1994; b since 1993, c ends 1998. 199s. GDP growth was clearly much more volatile for these countries. This suggests that underlying macroeconomic shocks are substantially higher in this group of countries. This is not surprising. All emerging market economies display substantially higher volatility in GDP and other macroeconomic aggregates than OECD economies. Despite the greater exposure to macroeconomic shocks and the apparently greater need for real exchange rate adjustment, it has been widely noted that emerging market economies for the most part do not allow their currencies to float freely. Calvo and Reinhart (22) observe that the high frequency exchange rate volatility in most developing countries is substantially less than that between the US and Japan, or the US and the UK. For many Latin American countries, for instance, exchange rate stability with the US dollar has historically been sought after. For the central and eastern European countries, and especially the accession countries, stability with the euro is a high priority. But this tends to go against the tenor of the results above. Can we understand why these countries put such a high emphasis on maintaining exchange rate stability, when a key aspect of a balanced monetary policy, according to the theory, is to allow exchange rate adjustment? In the next section, some basic arguments are developed that have been raised in the recent literature and that attempt to come to grips with such an explanation. Their policy implications are then discussed.

958 MICHAEL B. DEVEREUX IV. Extensions and Qualifications There are many factors that qualify the arguments of the last section, and question the supposed benefits of exchange rate flexibility. Space constraints do not allow coverage of the very large literature on fixed versus floating exchange rates. Instead some issues are addressed that are of particular importance to emerging market countries and can be encompassed within the modelling framework of the previous section. The first issue is concerned with the belief that flexible exchange rates, far from acting as a stabilizing mechanism, may themselves create excessive volatility. Excessive Volatility Contrary to the message of the analytical model above, a large body of empirical work has failed to uncover robust relationships between the exchange rate and standard economic fundamentals such as GDP, productivity growth and monetary policy. Flood and Rose (1995) suggest that, as regards the experience of the G-7 countries, the exchange rate seems to have a life of its own, exhibiting high and persistent volatility apparently independent of any macroeconomic fundamentals. Jeanne and Rose (22) develop this theme further, constructing a model of noise trader speculation in foreign exchange markets that can rationalize an excessive degree of real exchange rate volatility in a floating exchange rate regime. How important is this for emerging markets? While there are a number of other important reasons why emerging market economies might want to avoid exchange rate flexibility, one in particular is that flexible exchange rates are likely to generate excessive volatility of the real exchange rate. It is not hard to see how this might happen. Emerging markets are subject to much more volatile capital flows than industrial countries. In the analysis above, an efficient flexible exchange rate rule requires very high real exchange rate volatility. This may in and of itself generate significant uncertainty, especially if, as is highly likely, the precise cause of exchange rate volatility cannot be observed by market participants. In that case, markets cannot distinguish between exchange rate fluctuations generated by an optimal response to differential productivity movements or a fall in the country risk premium, and a policy-generated movement in the exchange rate. Furthermore, as suggested by Calvo (2), financial market (or capital flow) shocks are themselves likely to be endogenous to the regime. In particular, if an efficient monetary rule requires a significant amount of exchange rate adjustment, this requires some degree of discretion on the part of the policy-maker, leaving the currency vulnerable to confidence-driven capital market shocks.

EU ACCESSION AND ALTERNATIVE MONETARY POLICIES 959 The notion that financial market shocks are endogenous to the exchange rate regime has recently been implemented in a structural model of a small economy by Kollman (21). Kollman observes that interest rate differentials between industrial economies, controlling for exchange rate movements, display large deviations from uncovered interest rate parity (UIRP) under floating exchange rates. 8 In other words, there are large and persistent shocks to the UIRP relationship. But these shocks were not observed in the earlier fixed exchange rate era. The source of these shocks is not clear deviations from rationality, speculative behaviour, noise traders, etc., all represent candidate explanations. Kollman provides an estimate of departures from UIRP at quarterly frequencies based on US data relative to a basket of European economies. Comparing welfare across regimes, he finds that, in some circumstances, a fixed exchange rate may be desirable if it eliminates these financial market shocks, even in an environment where a floating exchange rate (using an inflation targeting rule) would otherwise be optimal. In terms of altering the policy conclusions of the previous section, the presence of endogenous, regime-dependent exchange rate volatility would suggest that monetary policy in the accession countries should pay more attention to exchange rate stability, and less attention to inflation targeting objectives than the above structural model suggests. Liability Dollarization The traditional analysis of the costs and benefits of exchange rate adjustment pays scant attention to the structure of financial markets. In the model above, the way in which production and investment spending is financed is of no importance. This abstraction from the institutional details of financial markets represents the conventional methodology in macroeconomics it reflects the conclusion of the Modigliani-Miller theorem that financing does not matter for real outcomes. But the mass of evidence that we have from emerging market crises over the last decade would seem to fly in the face of this view. In many, if not all of these crises, financial factors and financial institutions were at centre stage. Exchange rates crises were often preceded by bank failures. Capital inflows were associated with huge growth in domestic bank lending and stock market booms. The crises led to a sharp collapse in both. Countries that relied heavily on short-term bank loans were more prone to crises than other countries, where capital inflow was more heavily weighted towards foreign direct investment. After the Mexican and Asian crises, many economists argued that the traditional analysis of the trade-off between fixed and flexible exchange rates 8 Uncovered interest rate parity predicts that domestic interest rates will equal foreign interest rates plus the expected depreciation of the exchange rate.

96 MICHAEL B. DEVEREUX ignored completely the central role that the exchange rate plays in the financial system of emerging market economies (Calvo, 2; Krugman, 1999; Aghion et al., 21). Their main argument was that, contrary to the conventional framework where financing does not matter, the presence of financial distortions, moral hazard and other informational problems means that the balance sheet position of banks and other financial institutions becomes central to the macroeconomic adjustment mechanism in emerging market economies. One of the key aspects of this is the presence of substantial foreign currency denominated debt, or liability dollarization in the economies of these countries. Given that assets are mostly denominated in domestic currency, but liabilities are substantially in dollars, there is a currency mismatch between assets and liabilities for a bank in an emerging market economy. An exchange rate depreciation can generate a large fall in net worth. Because net worth represents a constraint on investment and production financing, this introduces a different, credit channel effect for the exchange rate that can possibly mitigate or even completely offset the direct stabilizing benefits of exchange rate depreciation in the traditional model. How does this alter the cost benefit analysis of exchange rate rules described above? In Devereux and Lane (23a), a simplified version is constructed of the small open economy model applied to the analysis of the credit channel effect of exchange rate depreciation. The credit channel enters because domestic financial institutions are constrained by net worth considerations in borrowing to finance intermediate imports, and their liabilities are in the form of foreign currency. It is shown that this may substantially alter the welfare case for adjusting the exchange rate to respond to terms of trade shocks. If the effect is big enough, the economy may be better off without any exchange rate adjustment, even if it is exposed to large terms of trade fluctuations. Devereux and Lane (23a) also show that, empirically, countries that have large net liability positions denominated in a given currency tend to pursue a more stable exchange rate vis-à-vis that currency. Therefore, when financial market distortions are brought into consideration, the standard calculation of the benefits of exchange rate adjustment may be very significantly altered. For economies with underdeveloped financial institutions, constrained by balance sheets in the presence of dollarized liabilities, the exchange rate may play a completely different role to what it would in the adjustment process for an industrial economy with mature financial markets. To this extent, the option of using the exchange rate may not be worth very much at all for many emerging markets. Indeed, by entering into a single currency area, the country may in fact speed up the convergence of financial practices towards the levels of the richer economies.