Internal Liberalization as a Barrier To Export-led Recovery in Central European Countries Preparing For EU Accession.

Frensch, Richard

I. Introduction

The EU's Agenda 2000 allows for the Union's eastern enlargement, for which ten Central and East European countries have applied so far: the seven current CEFTA members (Bulgaria, the Czech Republic, Hungary, Poland, Romania, Slovakia, and Slovenia) and the three Baltic states (Estonia, Latvia, and

Lithuania). For prospective new EU members the Agenda 2000 requires (a) an established democratic system, (b) a workable market economy, (c) the ability of the respective economies to withstand competitive pressures from the EU once goods and factor markets are actually opened, and (d) the acceptance of the EU's legal and regulatory framework. As the applicants' current GDP levels average roughly one third of the EU's in purchasing power parity terms, fast pre-accession growth would tremendously ease the enlargement process.

With 1997 GDP shares of exports and imports of goods and services combined ranging from 56% (Poland) to 121% (Czech Republic), these economies are textbook examples of small, open economies. But in spite of an hitherto unprecedented liberalization of prices and foreign trade regimes, export sector contributions to their GDP growth have generally not exceeded the experience of OECD countries to a significant extent in recent years.

This essay tries to answer this puzzle by combining two transition-specific trends. First, accompanying the decline of formerly oversized industrial sectors, the share of services in both employment and GDP has been gradually increasing since the early nineties. This economy-wide structural change represents a reallocation of resources out of former excess supply sectors into former excess demand sectors, i.e., out of agriculture and industry into services following internal liberalization (price liberalization and ease of market entry). This in turn is largely a reallocation out of the tradables into the non-tradables sector and thus diametrically differs from resource flows envisaged in standard trade liberalization and structural adjustment programs elsewhere in the world, which aim at redirecting resource flows into the opposite direction [see, e.g., Sachs and Larrain (1993, pp. 667ff]. This transition-specific reallocation has helped to generate and keep in motion a second transition-specific trend: a continuous appreciation of the real exchange rate, which is independent of the measure and varies with the chosen exchange rate regime only as a matter of degree. This real trend appreciation has so far acted as a significant barrier to a strong export-led recovery from the transition recession, which should have been expected as a consequence of the unprecedented price and foreign trade liberalization in these countries.

The general idea is thus very simple: (1) The higher the extent of transition-specific reallocation of employment in the economy following internal liberalization, the higher the real appreciation. (2) The higher the real appreciation, the lower the export sector real growth contribution. Combining (1) and (2) yields: An extensive transition-specific reallocation of resources acts as a barrier to export-led recovery from the transition recession. The following sections try to disentangle this rough picture: Section II describes the hitherto unprecedented speed and scale of the external sector liberalization in selected transition countries, while section III outlines their surprisingly unspectacular recent export sector contributions to growth. Section IV links export sector growth contributions to real appreciation, and sections V and VI close the argument by relating real appreciation to economy-wide transition-specific reallocation. Section VII concludes. While the argument in sections II - V is supported mainly by graphical evidence, section VI contains a regression analysis of the dynamics of real exchange rates during transition, establishing a cet. par. positive link between real exchange rates and transition-specific reallocation. This procedure together with data availability dictates the selection of country experiences presented in the various sections: While the illustrative medium term presentation in sections II-V is confined to the CEFTA-6 group (i.e., the current CEFTA members without Bulgaria), estimations in section VI employ panel data available between 1991 and 1998 from all ten EU applicant countries mentioned above to provide evidence on the short run impact of reallocation on real appreciation.

II. External liberalization

Tables 1A and B describe the process of foreign trade liberalization in CEFTA-6 economies. Although the speed of liberalization has not been the same everywhere--the rather gradual development in Romania contrasts with the shock approach taken in former Czechoslovakia--, speed and extent of this process have been unprecedented so far: Starting out with a centrally planned external sector in the late eighties, all CEFTA-6 countries have an essentially liberal foreign trade regime by now.(1)

Table 1A: External sector liberalization indices for CEFTA-6 economies, 1989-94, according to de Melo et al. (1996)

 1989 1990 1991 1992 1993 1994

Czech Republic, CR 0.0 0.0 0.8 0.9 0.9 0.9
Slovakia, SR 0.0 0.0 0.8 0.9 0.8 0.8
Poland, PL 0.2 0.9 0.9 0.9 0.9 0.9
Hungary, HU 0.5 0.7 0.9 0.9 0.9 0.9
Romania, RO 0.0 0.1 0.3 0.6 0.7 0.8
Slovenia, SLO 0.5 0.7 0.8 0.9 0.9 0.9

Table 1B: External sector liberalization indices for CEFTA-6 economies, 1994-97, according to EBRD (1994-1997)

 1994 1995 1996 1997

Czech Republic, CR 4 4* 4* 4+
Slovakia, SR 4 4* 4* 4
Poland, PL 4 4* 4* 4+
Hungary, HU 4 4* 4* 4+
Romania, RO 4 4* 3 4
Slovenia, SLO 4 4* 4* 4+

Notes:

(1.) The EBRD definition of external sector liberalization has changed slightly over the years, but has remained comparable over time; 1 marks the minimum, 4+ the maximum value. In de Melo et al. (1996) 0 denotes the lowest, 1 the highest possible value.

(2.) Both classification systems are mutually comparable, the classification in de Melo et al. (1996) is partly based on the EBRD classification system [cf. the discussion in de Melo et al. (1996, Appendix)].

Sources: de Melo et al. (1996) and EBRD (1994-97).

The effects of this unprecedented foreign trade liberalization visible so far have not been restricted to the immediate impact upon domestic relative prices, which changed into the direction of international relative prices. Foreign trade liberalization swiftly resulted in a geographic reorientation of trade: Taken as a whole, CEFTA-6 countries now experience a trade intensity with the EU that almost matches intra-EU trade [see, e.g., UNECE (1998, No. 1, p. 134)].

This geographic reorientation is increasingly--albeit slower in pace and to a different degree among countries--accompanied by a change in the commodity structure of trade [for the following, cf. Eichengreen and Kohl (1998)]: While the Czech Republic and Hungary have already been rather successful at exporting technologically more sophisticated products, Slovakia has so far retained its specialization in less sophisticated products. Poland, Romania, and Slovenia represent intermediate cases, with structural improvements varying strongly across industrial export branches.

Summing up, it is by now the prevalent opinion that external sector liberalization in transition countries has been unprecedented in scale and speed with tangible effects on geographic orientation and commodity composition of foreign trade. As a result, external liberalization has so far cet. par. proved significantly growth enhancing for transition economies [as demonstrated, e.g., by Fischer et al. (1998)] with a growth effect in the same order of magnitude as that of privatization and institutional reforms (i.e., price liberalization and competition).(2)

III. Export sector growth contributions

In the light of the liberalization recorded in the previous section, one might expect strong positive impulses from the external sector for CEFTA-6 countries recovering from their early transitional recession that should significantly exceed the experience of "normal" economies when recovering from "normal" business cycle troughs. To render such a comparison, we examine the "cumulative contributions of exports of goods and services to real GDP growth" for two distinct groups of countries, i.e. OECD versus CEFTA-6 countries, over a recovery period of three years following either the most recent cyclical trough (for OECD countries) or the last year of the transitional recession (for CEFTA-6 countries). We define cumulative growth contributions of exports as total real GDP growth in percentage points over these three years holding all other GDP components constant over the same three years.(3) Figure 1 shows three years cumulative real growth contributions of exports of goods and services for OECD and CEFTA-6 countries against their degree of international openness (measured by the GDP share of exports plus imports).

[Figure 1 ILLUSTRATION OMITTED]

As Figure 1 illustrates, export sector real growth contributions increase with the degree of international openness for OECD countries. For CEFTA-6 countries--with a sample of only six observations--this relationship cannot be confirmed. Even more importantly, Figure 1 shows that deviations of export sector real growth contributions (normalized by the degree of openness) from the OECD average (as measured by the indicated regression line) differ considerably across CEFTA-6 countries: Significant large scale positive deviations may be detectable for Romania but certainly not for the country group as a whole. One might thus seriously question the a priori expectation that all CEFTA-6 countries' thorough external sector liberalization necessarily resulted in extraordinary, positive export sector real growth contributions. Obviously, the undisputed cet. par. growth effects from external sector liberalization, stated in the previous section, have--to a different extent among CEFTA-6 countries--been "eaten up" by some other adverse effect on export growth contributions; the most likely candidate, of course, is an appreciation of the real exchange rate.

IV. Real exchange rate appreciation and export sector growth contributions

External sector developments are closely related to real exchange rate developments. This relationship may also be illustrated in the framework of cumulative export sector growth contributions outlined above. To that purpose Figure 2 contrasts cumulative export sector real growth contributions (normalized by the degree of openness) over the first three years of recovery from the transitional recession against real appreciation over the same period. The real exchange rate is proxied by two popular measures, i.e. (1) the nominal exchange rate against the DM deflated on the basis of consumer prices, and (2) local wages expressed in US-$ (dollar wages). Independent of the real exchange rate measure, cumulative export sector growth contributions appear to decrease with the extent of real appreciation in this country group.

[Figure 2 ILLUSTRATION OMITTED]

V. Links between internal liberalization, reallocation and real appreciation

Two transition-specific trends dominate recent macroeconomic developments in Central European economies preparing for EU accession. First, accompanying the decline of formerly oversized industrial sectors, total employment in services has been gradually increasing since the early nineties. Figure 3 illustrates this trend singling out the two CEFTA-6 economies for which it has been most pronounced, i.e. the Czech Republic and Slovakia.

[Figure 3 ILLUSTRATION OMITTED]

This economy-wide transition-specific structural change represents a reallocation of resources from former (i.e., under central planning) excess supply into former excess demand sectors, especially from agriculture and industry into services.(4) Figure 4 exemplifies the (factor-) price sensitivity of this process for the Czech Republic and Slovakia, the two CEFTA-6 economies that appear to have experienced this structural change most prominently: The reallocation of labor between sectors is accompanied by corresponding changes in relative wages.

[Figure 4 ILLUSTRATION OMITTED]

The second major trend these economies have experienced in recent years is a continuous real exchange rate appreciation. Figure 5, again singling out the Czech Republic and Slovakia, reveals this second trend using the same real exchange rate proxies as Figure 2.

[Figure 5 ILLUSTRATION OMITTED]

Abstracting from unemployment (cf. footnote 4), one might relate these two transition-specific trends, following an argument similar to Grafe and Wyplosz (1997). Internal liberalization in a transition economy, i.e., price liberalization and ease of market entry, encourages the emergence and growth of the formerly under-supplied services sector. As the services sector is at the same time a former excess demand sector as well as the largest part of the economy's non-tradables sector, this process is accompanied by a rise in the price of non-tradables relative to tradables, i.e. a real appreciation. As shown in the Appendix, our real exchange rate measures (nominal exchange rates deflated by consumer prices and wages in foreign currencies) are closely linked to the ratio of non-tradables to tradables prices in transition countries if one assumes that the ratio of sectoral prices is proportional to the ratio of sectoral unit labor costs.

The ratio of unit labor costs in the non-tradables sector relative to those in the tradables sector is closely connected to the sectoral reallocation of labor and increases during transition mainly for two reasons: Starting from a disequilibrium, the reallocation of labor aider internal liberalization is accompanied by a contemporaneous increase in the relative wage in the non-tradables sector (as, e.g., illustrated in Figure 4);(5) also [as argued e.g. in Grafe and Wyplosz (1997)], the sectoral reallocation of labor induces a lagged restructuring response of firms in the tradables sector implying decreasing unit labor costs in the tradables sector relative to the non-tradables sector. This induced productivity growth in the tradables sector should be differentiated from exogenous sectoral productivity differentials in the Balassa-Samuelson approach. The resulting trend real appreciation during transition effectively curbs any export-led transformational recovery.

It should of course not go unnoticed that the internal liberalization only revealed the disequilibrium in the sectoral allocation of labor: This implies that real exchange rates of centrally planned economies have been systematically undervalued prior to transition. Accordingly, the reallocation effect on exchange rates described above is a fundamental, equilibrium source of real exchange rate developments.

VI. The short-term impact of sectoral reallocation on real appreciation during transition

Naturally, any relationship between real appreciation and its fundamental sources should be expected to become effective in the medium and long term rather than in the short term, when nominal shocks typically dominate real exchange rate movements. However, the estimations presented in this section confirm that transition-specific employment reallocation exerts a significant influence on short-term, i.e. year-to-year, changes of the real exchange rate.

Estimations below are based on the common hypothesis that long-run real exchange rates follow real labor productivity developments relative to abroad [see, e.g. Halpern and Wyplosz (1997)], while in the short nm the impact of nominal shocks dominate. While such a hypothesis might usually call for an error-correction model that combines long-run equilibrium relationships with short-run deviations, such a model nevertheless does not seem appropriate in the situation under consideration: First, 6 years of available data are rather short for testing long-run relationships; second, the transition-specific relationship between intersectoral reallocation and real appreciation hypothesized above implicitly rules out that real exchange rates "on average" realize their equilibrium level during transition. Consequently, only a short-term estimation approach is taken under the hypothesis that even in the short run real exchange rate movements during transition react to productivity changes and employment reallocation as well as, of course, to nominal shocks. As derived in the Appendix, estimation equations for different measures of the real exchange rate can therefore be reduced to:

(1) [rer.sub.i](j,t)= [c.sub.o] + [c.sub.1] [multiplied by] cpi(j,t-1) + [c.sub.2] [multiplied by] a (j,t) + [c.sub.3] [multiplied by] ls(j,t) + [c.sub.4] [multiplied by] ls (j,t-1) + [c.sub.5] [multiplied by] ERR (j,t) + e(j,t),

for country j in year t, where:

[rer.sub.i], = 1,..,4: Real exchange rate; the four real exchange rates measures used are the nominal exchange rates against the DM and US-$ deflated on the basis of consumer prices and dollar and DM wages,

cpi: Consumer price index,

a: Aggregate labor productivity, i.e., GDP per total employment (including self-employed),

ls: Share of service sector employment in total employment (including self-employed),

[rer.sub.i], cpi, and a are yearly rates of change measured in percent; ls measures the change of shares in percentage points.

ERR: Exchange rate regime dummy; ERR = 1, for periods and countries with a fixed exchange rate or controlled floating, ERR = 0, otherwise (i.e., for free floating or crawling pegs).

e: error term.

Estimations of equation (1) employ panel data from all ten EU applicant countries named in the introduction, i.e., the seven CEFTA countries plus the Baltic countries, for the period 1991-98. Episodes of more than double-digit inflation were excluded from the sample; exchange rate data for Baltic countries are available only for 1993 onwards (for Estonia) or 1994 onwards (Latvia and Lithuania). Estimation is done by OLS enriched with exchange-rate regime specific dummy variables. Due to the limited number of observations, especially per individual country, exchange rate regimes are classified into only two groups with comparable effects upon real exchange rate developments, i.e. fixed exchange rates and controlled floating versus crawling pegs and free floating. To deal with remaining panel data specific heteroskedasticity problems, White's correction for heteroskedasticity is employed.

The results, presented in Table 2, are compatible with the apriori expectations formulated in the Appendix. Especially, they confirm a positive and (with one exception) significant short-run influence of intersectoral employment reallocation upon real exchange rate movements, a faster impact of reallocation on wage-based real exchange rate measures than on price-based real exchange rate measures, and a stronger short-run impact of total productivity changes on wage-based real exchange rate measures than on price-based real exchange rate measures.

Table 2: Yearly rates of change of four real exchange rate measures: all ten EU applicant countries, 1991-98

 Nominal Nominal DM wages US-$
 exchange exchange wages
 rate aga- rate against
 inst the the US-$
 DM (cpi- (cpi-
 deflated) deflated)

constant -5.7(*) -13.80(***) -12.49(**) -20.29(***)
 (-1.70) (-3.94) (-2.05) (-3.44)
cpi(-1) 0.25(***) 0.41(***) 0.27(**) 0.42(***)
 (2.81) (6.57) (2.40) (4.46)
a 0.24 0.18 1.78(**) 1.78(**)
 (0.64) (0.47) (2.66) (2.59)
ls 2.46(*) 3.53(***)
 (1.87) (3.12)
ls(-1) 0.88 1.91(**)
 (1.04) (2.63)
ERR 8.06(***) 9.35(***) 13.20(***) 15.01(***)
 (3.00) (3.44) (3.37) (3.97)

# of obser-
 vations 53 53 52 52
[R.sup.2] 0.34 0.56 0.36 0.49
[R.sup.2]
 adj. 0.28 0.53 0.31 0.45
DW 1.83 2.23 1.71 1.78

Notes: 1. OLS with heteroskedasticity-consistent standard errors and covariances; 2. t-values in parentheses. 3.(*) ((**), (***)) denote significance at 10 (5, and 1) per cent levels, respectively.

Data sources: Cf. Figures 2 and 3.

VII. Conclusion: Intersectoral employment reallocation as a barrier to export-led recovery in transition

The paper links two trends observable in transition economies, i.e., continuous resource reallocation and trend real exchange rate appreciation: An extensive transition-specific reallocation of resources from former excess supply sectors, such as industry and agriculture (i.e., mostly tradables production), to former excess demand sectors, specifically market services (i.e., mostly non-tradables production), acts as a barrier to export-led recovery from the early transitional recession in Central and Eastern European countries preparing for EU accession. As it was the internal liberalization that put transition-specific reallocation of employment into motion, it is consequently internal liberalization that cet. par. limits the growth stimulus from external liberalization. Of course, as internal liberalization only revealed the previous disequilibrium in the sectoral allocation of labor, one might soften this view by holding the (extent of the) previous disequilibrium responsible rather than liberalization.

To some extent, observers have concentrated on the microsphere when looking for reasons behind real exchange rate appreciation during transition and potentially resulting currency overvaluation. Especially in the case of the Czech currency crisis of May 1997 the most popular suspect is an allegedly misspecified privatization strategy: the Czech voucher privatization scheme of the early nineties which failed to result in effective corporate control in privatized industry, thus allowing an ever widening gap between wages and productivity to open up, is held responsible for real exchange rate appreciation.

While this paper argues that transition-specific intersectoral resource reallocation may significantly be responsible for at least some portion of real exchange rate trend appreciation in transition economies, in the sense of Blanchard's (1997) definition of transition as reallocation and restructuring this is no contradiction to the quoted micro-based view; transition firms' micro problems with equalizing wage pressures and productivity changes (differing in extent across countries and industries) may rather be attributable to a transition-specific employment reallocation that is not equally strong across transition countries.

Finally, it has repeatedly been stated that macroeconomic policy measures (including price and trade liberalization) are not sufficient for a successful transition. With its special emphasis on foreign sector growth contributions, this view is strongly supported in this paper: Again resorting to Blanchard (1997), one might conclude that the first transition phase, i.e., resource reallocation out of former excess supply into excess demand sectors, works as an inevitable barrier against a post-transition export-led recovery, given the initial disequilibrium allocation of labor. Permanent growth impulses from foreign trade should thus be expected only from the second, presumably longer transition phase of firm-level restructuring.

Notes

(1.) For an interesting attempt at relating this fast liberalization to a transition-specific political perspective, see Wunner (1998).

(2.) In Fischer et al. (1998) external liberalization, privatization, and other institutional reforms are measured by three respective liberalization indices provided in de Melo et al. (1996), where the external liberalization index is the one shown in Table 1A of this paper. Alternative regressions of each of these (highly correlated indices) plus other explanatory variables on yearly GDP growth rates of 25 transition economies between 1992 and 1995 yield estimated coefficients for the three indices that are of the same order of magnitude.

(3.) The formula for calculating three years cumulative real growth contributions of real exports E is thus CGC(E) = (E(j+3) -E(j))/GDP(j), with j denoting the year of the most recent cyclical trough (for OECD countries) or the last year of the transitional recession (for CEFTA countries).

(4.) We abstract from unemployment during transition. Rather, we stress direct reallocation between (tradable and non-tradable) sectors, where we broadly identify these sectors with state-owned or privatized agriculture and industry versus a newly emerging private services sector. So far, most transition models [most prominently Blanchard (1997)] have assumed that the major source of the emerging private sector employment is unemployment from industry restructuring. But, as noted e.g. in Brixiova (1997), this would either imply a quickly emerging private sector together with short-lived unemployment or a slowly emerging private sector together with persistent unemployment; contrary to this, transition economies have experienced quickly emerging private sectors combined with persistent unemployment, which is compatible with the explanation that private sector firms hire directly from industry, i.e. that the most significant labor flows are directly from industry into services.

(5.) For a sample of fourteen OECD countries, Strauss (1997) is able to show that for periods of less than five years wage differentials between the non-tradables and the tradables sector can raise intersectoral price differentials and therefore cause real appreciation, thus invalidating the assumption of intersectoral wage equalization featuring prominently in the Balassa-Samuelson approach to real exchange rate movements. For countries in transition, one might therefore expect intersectoral wage differentials to cause a continuous real appreciation at least as much as for OECD countries.

(6.) Cf., e.g., Halpern and Wyplosz (1997, p. 441) or Strauss (1997, p. 392).

(7.) Again, see Halpern and Wyplosz (1997, pp. 438ff.)

References

Blanchard, Olivier. 1997. The Economics of Post-Communist Transition, Oxford, Clarendon Press.

BMWi (German Federal Ministry of Economy). 1995-99. Wirtschaftslage und Reformprozesse in Mittel- und Osteuropa. Sammelbande 1995-99, BMWi-Dokumentationen, Berlin.

Brixiova, Zuzana. 1997. "On the speed of transition in Central and Eastern Europe: Does on-the-job search matter?," IMF Working Paper 97/102, Washington, D.C., August.

De Melo, Martha, Cevdet Denizer, and Alan Gelb. 1996. "From plan to market: Patterns of transition," Policy Research Department Working Paper No. 1564, The World Bank, Washington, DC.

EBRD. 1994-97. Transition Reports 1994-1997, London.

Eichengreen, Barry, and Richard Kohl. 1998. "The external sector, the state, and development in Eastern Europe," CEPR Discussion Paper No. 1904, London.

Fischer, Stanley, Ratna Sahay, and Carlos A. Vegh. 1998. "From transition to market: Evidence and growth prospects," IMF Working Paper No. 98/52, Washington, DC, April.

Grafe, Clemens, and Charles Wyplosz. 1997. "The real exchange rate in transition economies," CEPR Discussion Paper No. 1773, London, December.

Halpern, Laszlo, and Charles Wyplosz. 1997. "Equilibrium exchange rates in transition economies," IMF Staff Papers, 44, 4, December, pp. 430-61.

International Financial Statistics on CD-Rom (IFS), IMF, Washington, DC, various issues.

Prasad, Eswar S., and Jefferey Gable. 1998. "International evidence on the determinants of trade dynamics," IMF Staff Papers, 45, 3, September, pp. 401-39.

Sachs, Jeffrey D., and Felipe Larrain B. 1993. Macroeconomics in the Global Economy, Englewood Cliffs, NJ, Prentice Hall.

Strauss, Jack. 1997. "The influence of traded and nontraded wages on relative prices and real exchange rates," Economics Letters, 55,3, pp. 391-5.

Wunner, Norbert. 1998. "Trade liberalization and political support in transition economies," Economics of Transition, 6, 2, pp. 409-25.

United Nations Economic Commission for Europe (UNECE). 1998. Economic Survey of Europe, No. 1, New York and Geneva.

Appendix

In equation (1) in section VI we regress rates of change of various exchange rate measures on nominal disturbances (proxied by lagged inflation), aggregate productivity growth and sectoral reallocation of labor (proxied by the change of the share of services employment). To motivate regression equation (1), we define price-based measures of the real exchange rate as

(A-1) [RER.sub.P] = P [multiplied by] e/P*,

where P (P*) is the home (foreign) price level and e is the nominal exchange rate (expressed as units of foreign per home currency, i.e., an increasing e denotes a nominal appreciation of the home currency). In general, P may be consumer (CPI) or producer prices (PPI). In our context, foreign countries are the US and Germany.

Wage-based measures of the real exchange rate are simply

(A-2) [RER.sub.W] = W [multiplied by] e,

where W is nominal wages at home. [RER.sub.W] is thus measured as US-$ or DM wages.

We assume that

(i) Prices (CPI or PPI) may be expressed as a combination of tradables and non-tradables prices according to(6)

(A-3) ln P = (1-[Alpha]) ln [P.sub.T] + [Alpha] ln [P.sub.NT],

T (NT) denoting tradables (non-tradables).

(ii) All prices are proportional to unit labor costs, i.e.

(A-4) [P.sub.T] = [[Beta].sub.T] [W.sub.T]/[A.sub.T] and [P.sub.NT] = [[Beta].sub.NT] [W.sub.NT]/[A.sub.NT],

where A is labor productivity.

(iii) Movement of non-tradables prices is unrestricted by purchasing power parity; for tradables, purchasing power parity is relaxed due to the existence of a product quality differential,

(A-5) [P.sub.T] = [Kappa] [multiplied by] [P.sub.T] */e,

where k [is less than] 1 describes product quality differences between goods produced in transition economies versus abroad.(7) We assume kappa to grow with aggregate labor productivity,

(A-6) [Kappa] = [A.sup.[Gamma], [Gamma]] [is greater than] 0

(iv) Foreign prices are treated as constant. This simplification reflects the fact that during the period under consideration inflation rates in transition countries are of a different order of magnitude than in the US or Germany. As we are interested in rates of change only, foreign prices are neglected in the following.

Accordingly, from (A-1), assumption (iv) and inserting (A-5),

ln [RER.sub.P] = ln P + ln e = ln P + ln kappa - ln [P.sub.T].

From (A-3) and (A-6),

(A-7) ln [RER.sub.P] = [Gamma] ln A + [Alpha] ln([P.sub.NT]/[P.sub.T]).

Using (A-4), we similarly derive,

ln [RER.sub.W] = ln W + ln e = ln P + ln A + ln [Kappa] - ln [P.sub.T] or--again via (A-3) and (A-6),

(A-8) ln [RER.sub.W] = (l+[Gamma]) ln A + [Alpha] ln([P.sub.NT]/[P.sub.T]).

In general, the [P.sub.NT]/[P.sub.T]-ratio in (A-7) and (A-8) should respond to (1) exogenous sectoral productivity differentials (assuming sectoral wage equalisation as in the Balsassa-Samuelson approach), (2) sectoral reallocation (assuming wage differentials), and (3) nominal shocks and their different impacts on [P.sub.NT] versus [P.sub.T] depending upon the chosen exchange rate regime (assuming weak PPP for tradables, as in assumption (iii)). As argued already in section V, sectoral reallocation is deemed to have influenced the [P.sub.NT]/[P.sub.T]-ratio much more pronouncedly than exogenous sectoral productivity differentials in transition countries so far: By (A-4), the price ratio [P.sub.NT]/[P.sub.T] is proportional to ([W.sub.NT]/[A.sub.NT])/([W.sub.T]/[A.sub.T]) and is thus closely linked to the sectoral reallocation of labor during transition: At the beginning of transition, excess wages in the unrestructured tradables sector imply ([W.sub.T]/[A.sub.T]) [is greater than] ([W.sub.NT]/[A.sub.NT]), as the non-tradables sector was quickly privatized wiping out potential excess wages there. The relationship ([W.sub.NT]/[A.sub.NT])/([W.sub.T]/[A.sub.T]), however, increases during transition mainly for two reasons: We know that--starting from a disequilibrium--liberalization induced reallocation of labor is accompanied by a contemporaneous increase in the relative wage in the non-tradables sector [W.sub.NT]/[W.sub.T] (as, e.g., illustrated in Figure 4); also [as argued e.g. in Grafe and Wyplosz (1997)], sectoral reallocation of labor induces a lagged restructuring response of firms in the tradables sector implying a decrease in [W.sub.T]/[A.sub.T] relative to [W.sub.NT]/[A.sub.NT]

Proxying nominal short-run disturbances by lagged inflation, aggregate productivity growth by the rate of change of GDP per total employment and sectoral reallocation of labor by the change of the share of services employment, while capturing the exchange rate regime with a dummy variable, we arrive at the regression equation (1) in section VI.

In general, we should expect a faster impact of labor reallocation on wages and thus on wage-based real exchange rate measures than on prices and price-based real exchange rate measures. In addition, comparing (A-7) and (A-8), we should apriori expect a stronger short-run impact of total productivity changes on wage-based real exchange rate measures than on price-based real exchange rate measures. The results in Table 2 fulfill these expectations.

Richard Frensch

Department of Economics Osteuropa-Institut Munchen

Without implicating, the author is grateful to Andrea Boltho, Laszlo Halpern, Michael Keren, Holger Wolf, an anonymous referee and the editor for critical comments on earlier versions of the paper.