Book
Share
Chapter

III. Sectoral and Employment Effects of the Opening Up of Eastern Europe

Author(s):
International Monetary Fund
Published Date:
August 1997
Share
  • ShareShare
Show Summary Details
Author(s)
Peter Dittus and Palle S. Andersen

Introduction

“Policy makers will face great pressure to subsidize, or failing that, to protect the manufacturing sectors under threat. It would be a critical mistake to yield to this pressure…. For the developing countries, being squeezed out of their export markets would stymie their growth potential. Such a setback would expose the vulnerability of new democracies in Eastern Europe” (Brown and Julius 1993).

“In our view, given the advantages for Europe of a successful economic transition combined with political stability in the East, the EC would have been justified in moving far more rapidly than it currently appears willing to do” (Hughes and Hare 1992).

The above quotes illustrate a dilemma policymakers are currently facing. On the one hand, both economic theory and empirical studies suggest that the expansion of free trade will provide large gains (Rollo and Smith 1993). In fact, the reforming countries in Eastern Europe consider trade rather than aid the most important form of help that Western Europe can give to their transformation. On the other hand, unemployment in Western Europe is currently close to 11 percent, and employment in the manufacturing industries, where the competitive advantage of Eastern Europe is most apparent, has been falling for several years. Because the removal of trade barriers is certain to engender additional adjustment pressures in the short run and the gains may appear only with some lag, there is some reluctance in industrial economies to liberalizing East-West trade; in fact, the trade agreements concluded so far have been rather restrictive, and the additional export revenue from tariff reductions has been estimated at less than 15 percent of exports (Messerlin 1993; Mastropasqua and Rolli 1993). Moreover, about half of Eastern Europe’s exports to the West remain subject to nontariff barriers, most of which are found in the sectors considered “sensitive” by Western Europe, including agriculture and food, textiles and apparel, chemicals, and iron and steel.

This paper attempts to illustrate both sides of the dilemma. It is divided into two parts and a summary that attempts to derive tentative policy conclusions. The objective of the first part is to provide a general perspective on the dilemma facing policymakers in Western Europe by reviewing the literature on the possible effects of open trade on aggregate employment, the composition of employment, and the structure of wages. More specifically, the first part attempts to answer two questions:

  • Can the falling shares of manufacturing in output and employment seen so far be ascribed to foreign trade, or have they been induced by shifts in domestic demand?

  • What are the likely effects on the composition of labor demand as opposed to aggregate employment, and on the structure of wages as opposed to aggregate real wages?

The second part provides another perspective on the fears of Western policymakers by focusing on Western Europe’s sensitive sectors and estimating the impact of expanding free trade on employment in these sectors.1 The methodology is kept transparent and simple, and the estimates are best understood as indicating broad orders of magnitude only. There are three sections. The first section illustrates a basic asymmetry in the importance of trade in the sensitive areas: market access is extremely important to Eastern Europe but much less so to Western Europe. This section also considers the elements relevant to estimating the impact of increasingly free trade on West European employment. The second section presents several trade and employment scenarios for the Organization for Economic Cooperation and Development (OECD) countries in Europe and compares the impact on Austria with the effects on Spain and Portugal. The third section provides a historical perspective on the predicted employment changes and briefly considers the likely employment impact on Eastern Europe.

Overall—and very much in contrast to popular misperceptions—the paper finds that the effects of improved market access on employment in the West are rather small under a range of assumptions. Analytical arguments and empirical estimates clearly support granting more market access to Eastern Europe.

1. Manufacturing Employment, Foreign Trade, and Unemployment

The fact that manufacturing employment in Europe has declined while unemployment has increased has given rise to fears about deindustrialization (Table 1). The European unemployment rate appears to be continuing its trend upward, with no reversal in sight (see Appendix 1). The impression that this development may be linked to heightened import competition has provided the background to resistance against trade liberalization. But are external causes really to blame? The following two sections take a closer look at this issue.

Table 1.Developments in Manufacturing and Unemployment in Selected Countries(In percentages)
Countries1970197519801985199019911992
I. Manufacturing output/GDP, current prices
United States125.122.721.719.918.4
Japan36.030.229.229.529.129.0
Germany38.434.332.331.730.529.2
United Kingdom 134.731.026.823.622.121.0
Canada19.818.517.917.216.72
Austria33.78.627.826.926.125.9
II. Manufacturing output/GDP, constant prices
United States120.019.519.218.918.9
Japan25.124.626.829.531.532.0
Germany35.433.332.631.729.829.2
United Kingdom130.028.525.223.624.023.3
Canada20.018.817.917.816.215.4
Austria26.725.626.726.927.427.4
III. Manufacturing employment/aggregate employment
United States24.621.320.418.016.215.715.3
Japan27.125.824.725.024.124.324.4
Germany33.330.228.326.126.025.925.5
United Kingdom33.530.528.022.820.019.518.51
Canada20.718.817.215.714.913.613.0
Austria30.030.129.528.227.026.825.51
IV. Unemployment/total labor force
United States5.08.37.27.15.36.77.4
Japan1.21.92.02.62.12.12.2
Germany0.64.03.38.06.06.36.7
United Kingdom2.43.66.111.65.98.310.0
Canada5.76.97.510.58.110.311.3
Austria1.11.51.53.63.23.53.6
Sources: OECO National Accounts, Vols. I and II; OECD Labour force Statistics (annual and quarterly); and national data.

Partly estimated.

1989.

Sources: OECO National Accounts, Vols. I and II; OECD Labour force Statistics (annual and quarterly); and national data.

Partly estimated.

1989.

a. Causes of the decline in manufacturing employment

While the fall in manufacturing employment in Western Europe is an indisputable reality, it is difficult to say whether it has been caused by domestic factors or by weakened competitiveness and increased imports from developing countries. Judging by the concerns widely expressed in Europe about high labor costs and the rise in imports of manufactured goods from the newly industrialized economies (NIEs), foreign trade has been the major cause of rising unemployment. However, these concerns seem to be somewhat overstated, for several reasons.

  • First, competitiveness is a relative and not an absolute concept and, by definition, cannot be a major cause of rising unemployment in all industrial countries.

  • Second, most developing countries, and especially the European countries in transition, do not expand exports to build up a current account surplus but to be able to import more. Even though the import penetration ratios of the NIEs in OECD Europe have increased at annual rates of 6.25-13 percent during the 1980s (Martins 1993), their imports (in volumes) have increased by over 10 percent per year during the same period, leaving a trade surplus with OECD Europe of only $500 million in 1992. Moreover, despite the rise in import penetration, the share of these economies in aggregate demand has remained below 2 percent in all European countries. Similar trends may be observed for United States; Krugman and Lawrence (1993) estimate that the swing in the manufacturing trade balance accounts for less than 1 percentage point of the fall in the share of manufacturing output in total GDP between 1970 and 1990.

  • Third, most empirical studies using factor contents of trade (see below) in assessing the employment effects of trade between industrial Europe (the North) and developing European countries (the South) find rather small effects. The cumulative employment loss of North-South trade may be less than 1 percent of manufacturing employment in the North, while the corresponding employment gain in the manufacturing sectors of the South, at around 5 percent, is somewhat higher.

  • Finally, when imports of manufactured goods were included in the extended Okun equations (see Appendix 1), the coefficient was always negative. This result may, of course, be due to the presence of manufacturing employment, which is likely to be correlated with imports of manufactured goods. However, positive coefficients were obtained by regressing the share of manufacturing employment on imports of manufactured goods.

There are, however, also studies pointing to larger actual or potential effects. Revenga (1992), using manufactured import prices rather than imports, finds that lower import prices reduced manufacturing employment in the United States by 4.5-7.5 percent (depending on the industry) during 1980-85, when the U.S. dollar appreciated. During the same period, manufacturing wages declined by 2 percent in response to increased import competition. Moreover, arguing that most studies based on the factor contents of trade tend to understate employment effects because they do not take noncompeting imports into account, Wood (1993) presents alternative estimates, with significantly larger effects.

Wood’s adjustment consists of (i) replacing the vectors that measure the effects of import displacements with counterfactual vectors based on the factor prices of the trading partner, and (ii) taking account of substitution elasticities in the sectors concerned.2 The results (Table 2) are startling, as the employment loss for the industrial countries increases from 0.8 million to almost 13 million, with unskilled workers accounting for most of the fall. For the developing countries, the changes are less dramatic, as the net employment effect merely triples and growing demand for unskilled workers accounts for all of the increase. Yet it may be asked whether Wood’s method should also be applied at the margin. More specifically, since a possible opening of trade with Eastern Europe would come on top of a long period of adjustment to trade with developing countries, actual rather than counterfactual factor quantities in the sectors most affected may be more relevant.

Table 2.Impact of North-South Trade on Employment(In million person-years)
Industrial countriesDeveloping countries
ExportImportNet effectExportImportNet effect
I. Conventional method
Total4.4-5.2-0.819.9-15.94.0
II. Counterfactual method
Skilled labor2.2-2.4-0.22.6-3.2-0.5
Unskilled labor2.2-14.7-12.517.3-3.214.1
Total4.4-17.1-2.719.9-6.413.6
% of labor force in manufacturing-12.312.9

It is also possible that international competition, combined with advances in communication technology and increasingly mobile direct investment, affects wage demands and, indirectly, the demand for labor in industrial countries. As argued in MacFarlane (1993), Australian workers are increasingly faced with a choice between reducing their wage claims or seeing their jobs relocated to one of the emerging Asian countries, where labor costs are much lower and productivity is rapidly advancing toward Australian levels. It is not inconceivable that workers in exposed sectors in Western Europe could be faced with a similar choice regarding Eastern Europe.3 Perhaps the earlier wage norm, whereby real wage growth in line with productivity was seen as sufficient to maintain a stable level of unemployment, will have to be replaced by a revised norm that takes into account wage and productivity increases in low-wage developing and reforming countries.4

On balance, it would seem that earlier studies ascribing most of the decline in manufacturing employment to a combination of above-average productivity growth, low relative prices, and price and income elasticities of domestic demand of less than unity may have exaggerated somewhat the impact of domestic factors by not taking sufficient account of noncompeting imports and the loss of jobs through relocation (Krugman and Lawrence 1993). Nonetheless, the evidence suggests that import competition—up to now, at least—has not been a major factor in either the declining share of manufacturing in GDP or the rise in unemployment.

b. The impact of trade on employment composition and wage structures

Even if its total effects on employment are small, trade can affect the composition of labor demand and the structure of relative wages. According to the Heckscher-Ohlin-Samuelson theorem, international trade generates a tendency toward the convergence of factor proportions and relative factor prices across countries. Output changes induced by shifts in the terms of trade will, under certain assumptions, also lead to a fall in relative income for the owners of the previously scarce factor of production (Stolper-Samuelson theorem) and can even cause an absolute loss. Thus, if some opening up of markets takes place, it will undoubtedly be accompanied by claims for compensation. Under even more restrictive assumptions, there may be a real income loss for the economy as a whole (Johnson and Stafford 1992). In this case, compensation will not suffice and political arguments against liberalizing trade are likely to develop.

However, the convergence of factor proportions and relative factor prices may be difficult to detect if other factors (such as non-neutral technical progress favoring skilled workers, or changes in the labor supply) dominate or the underlying assumptions (identical technologies across countries and industries and a gradual removal of barriers to trade) are not satisfied. Moreover, changes in relative factor prices appear to be highly dependent on the structure of the industries involved and the extent to which rent sharing is present. Further, the empirical studies are difficult to compare. Some focus on the international convergence of factor prices, while others are concerned with changes in relative factor prices within individual countries. The models applied also differ widely, especially with respect to the degree of disaggregation and the definition of industrial characteristics.

Nonetheless, most empirical studies point to rather large compositional and distributional effects, notably in periods of large shifts in the net trade balance. Using manufacturing data for 12 OECD countries, Martins (1993) finds that import penetration tends to reduce relative wages in sectors such as textiles and clothing that are characterized by a high degree of fragmentation and a low degree of product differentiation. By contrast, relative wages tend to rise in very segmented sectors with a high degree of product differentiation, since firms in such sectors typically have a great deal of market power and entry costs are high. This type of rent sharing is also documented by Grey (1993), who looks at wage differentials in Canadian manufacturing and finds relatively high wages in export-oriented industries but comparatively low wages in sectors faced with import competition.5

Borjas and Ramey (1993) analyze changes in relative wages (using import penetration, immigration, and industrial structures as the principal determinants) and obtain results that partly contradict those reported by Martins. Based on a high negative correlation between the trade deficit for durable manufactured goods and a comparison of the wages of college and high school graduates, the researchers set up a model in which the rents found in highly concentrated industries favor primarily employees with relatively low educational levels.6 Most industries producing durable manufactured goods (primary metals, farm and office equipment, and automobiles and aircraft) are characterized by a high degree of concentration and a large proportion of high school graduates who receive higher wages than employees with a similar education employed in other less concentrated industries. Because of rent sharing, increased imports of durable manufactured goods have a more adverse effect on the relative wages of high school graduates than increased imports of, for instance, textiles and apparel, which are produced in less concentrated industries (Borjas and Ramey 1993). The researchers’ empirical evidence supports this hypothesis. Borjas and Ramey also find that immigration tends to depress wages of less-educated workers, but this finding may be particular to the United States, where most immigrants have low skills—especially compared with the recent immigrants to Germany.

Murphy and Welch (1992) also identify the effects of trade by looking at the ratio between the wages of college and high school graduates and, like Borjas and Ramey, find a strong correlation between the rise in this ratio and the widening trade deficit for durable manufactured goods. Using a model that allows them to distinguish between the effects of demand and supply, they find that in the 1970s most of the changes in the wage ratio can be ascribed to domestic supply shifts. In the 1980s, however, when the ratio rose substantially, unemployment shocks and the rising deficit for durable manufactured goods had a significant and positive influence.

Katz and Murphy (1992) report even stronger results using data that are more finely disaggregated with respect to industries and levels of education. Using the conventional method regarding the factor contents of trade, they find that the rising U.S. trade deficit reduced the relative labor demand for high school dropouts by 1.5-4.0 percent during 1979-85 but increased demand for college graduates by 1.5 percent.7 In earlier years, however, the trade effect on U.S. wage differentials appears to have been relatively small, and even during the period when the rise in the deficit was most pronounced, for most groups the effects of trade accounted for less than one third of the change in overall relative demand.

Some studies, however, cast doubt on the convergence hypothesis or find that convergence is dominated by other factors. For instance, Krugman and Lawrence (1993) and Lawrence and Slaughter (1993) see no evidence of convergence of factor proportions (that is, a change in favor of the previously scarce but now less expensive factor) in U.S. manufacturing data. They conclude that most of the shifts in the composition of employment and the wage structure reflect domestic changes favoring the demand for nonproduction or skilled workers. However, while the empirical evidence looks robust and the influence of domestic demand changes (in particular non-neutral technical change) should not be underrated, the arguments and underlying model may not take sufficient account of production relocation (or the threat of relocation) and differences in technologies among sectors.

Davis (1992) tests the Heckscher-Ohlin-Samuelson theorem on relative wages in manufacturing industries in 19 industrial and developing countries. An initial assumption that relative factor prices converge toward a world average is decisively rejected by the data. However, when taking account of changes in average trade ratios and country-specific effects, Davis finds some support for the Heckscher-Ohlin-Samuelson theorem, though the tendency toward convergence is modest and more than offset by country-specific factors.

Moreover, convergence is not found for developing countries, and there is a clear asymmetry between changes in exports and imports, as shifts in imports have a larger and more significant effect than the former. As regards the possibility that trade may lead to an absolute income loss for total manufacturing as well as for the aggregate economy, the model used by Johnson and Stafford (1992) assumes that the erosion of comparative advantage induced by rapid productivity gains in other countries will lead not only to a reduction in the rent of the “scarce factor” but also to an income loss for the economy as a whole. To meet the assumptions of the model, this real income loss appears as a deterioration in the overall terms of trade (the previously scarce factor is employed in other sectors at a lower real wage).

At first glance, the relative prices presented in Table 3 appear to support this hypothesis. Except for the two oil producers (the United Kingdom and Canada) the terms of trade have declined since 1970, so that real disposable income has fallen relative to real output. Moreover, export prices for manufactured goods have fallen compared with the overall output deflator for manufacturing, indicating that firms may have been forced to cut prices and margins in foreign markets. However, the rise in oil prices in the 1970s is the principal cause of the decline in the aggregate terms of trade and oil, and other supply shocks are not relevant to the issue of whether foreign trade reduces real incomes. Moreover, a decline in the terms of trade can be induced by above-average productivity growth as well as by forced cuts in margins and prices. In fact, the developments shown in Panel II of the table, combined with the constant output shares in constant prices (Table 1), point to the forced cuts as a more plausible explanation. Finally, and in marked contrast to the Johnson-Stafford model, Panel IV of Table 3 suggests that the industrial countries have gained from trade in manufactured goods with the developing countries. Similar results are reported by Lawrence and Slaughter (1993). Contrary to the popular perception that increased trade erodes real wages, Lawrence and Slaughter show that the terms of trade improved during the period when real wages declined and that a major cause of the decline was a relatively large increase in the prices of nontraded goods, notably housing.

Table 3.Developments in Relative Prices(1980=100)
Memo item: %
Countries19701975198519901992change, annual rate
1970-801980-92
1. Terms of trade, aggregate economy
United States140120115111115-3.31.2
Japan1811371009290-5.8-0.9
Germany11210995108110-1.10.8
United Kingdom10390101101103-0.30.3
Canada85949699971.6-0.2
Austria10710699104105-0.70.4
II. Output deflators: manufacturing/aggregate economy
United States1101039388-1.0-1.3
Japan1321129285831-2.7-1.7
Germany1091041011031011-0.90.1
United Kingdom31041061018181-0.4-1.9
Canada10098979820.0-0.2
Austria1211079691911-1.9-0.9
III. Export prices, manufactured goods/output deflator
United States76921011062.70.6
Japan9410099878710.6-1.3
Germany96101102898810.4-1.1
United Kingdom396989911611010.40.9
Canada9892958020.2-2.5
Austria1051091018480-0.5-2.0
IV. Export prices, manufactured goods: industrial
United States10597134132135-0.52.5
Japan95381031241410.52.9
Germany7995811241242.31.8
United Kingdom7275861201253.41.9
Canada125104114110102-2.30.1
Austria8710185118114-1.41.1
Sources: OECD National Accounts, Vols. I and II; Monthly Bulletin of Statistics and National Data, United Nations.

1991.

1989.

Estimated.

Sources: OECD National Accounts, Vols. I and II; Monthly Bulletin of Statistics and National Data, United Nations.

1991.

1989.

Estimated.

2. The Impact of Increased Trade with Eastern Europe on Employment in Western Europe

The first part of the paper has documented the pronounced upward drift in unemployment in Europe and argued that while the competitive pressure exerted by rising imports is unlikely to be the major driving force behind this increase, it cannot be excluded as a contributing factor. This background helps to explain fears in Western Europe that the removal of trade barriers with Eastern Europe could engender strong adjustment pressures and add to unemployment, in particular in sensitive sectors (see, for example, Busch and Frölich 1993). These sectors are important to Western Europe, accounting for 22 million jobs, or 15 percent of total employment (see Appendix Tables 1 and 2). In Spain and Italy, one fifth of total employment is in these sectors, and in Portugal one third. This part of the paper attempts to put these fears into perspective by estimating the impact on employment in Western Europe of growing trade with Eastern Europe.8

a. Determinants of the employment impact

There is a basic asymmetry in trade between Eastern and Western Europe, in particular in the sensitive sectors already mentioned. Sensitive goods account for a high proportion of Eastern Europe’s total exports but constitute only a small percentage of the sensitive goods Western Europe imports.

In Eastern Europe, these exports account for half of total exports to the European Union (EU) (Table 4). During 1990-92, exports of sensitive goods rose twice as fast as total exports, with most of the increase coming from textiles and apparel. Even faster was the expansion of semimanufactures and miscellaneous consumer goods, which are not constrained by nontariff barriers (NTBs). The analysis in the following pages nevertheless focuses on sensitive sectors because NTBs are prevalent in these sectors, and—in the short run at least—there is the potential to expand these exports substantially. Table 4 also shows that some 85 percent of East European exports to the EU come from the Czech Republic, Hungary, Poland, and the Slovak Republic, the “Central European four” (CE4).

Table 4.Merchandise Exports from Central and Eastern Europe to the European Union, 1990-92
BulgariaFormer

Czechoslovakia
HungaryPolandRomaniaTotal
Total merchandise1.27.25.39.21.924.7
Percent of area exports to EU4.729.121.337.47.5100.0
Export structure in 1992
Food18.94.819.812.94.911.7
Raw materials and ores9.17.16.68.62.77.1
Fuels1.73.11.67.22.84.2
Metal products10.911.65.111.97.610.0
Chemicals8.28.89.87.14.98.1
Other semimanufactures6.717.39.212.99.212.8
Machinery and transport12.120.719.113.48.316.3
Textiles and apparel21.512.717.016.935.416.9
Other consumer goods10.411.911.59.123.611.6
Sensitive sectors59.537.951.847.852.846.7
Total100.0100.0100.0100.0100.0100.0
Change in percent (value), 1990-92
Food-0.523.17.9-15.575.0-1.4
Raw materials and ores96.332.113.149.313.637.5
Fuels-63.6-9.4-22.2-15.7-87.2-35.2
Metal products4.158.0-22.029.9-18.422.6
Chemicals2.128.629.82.821.317.4
Other semimanufactures47.2142.737.378.3-9.078.6
Machinery and transport-54.533.1-3.93.9-52.60.9
Textiles and apparel146.1114.145.180.031.970.1
Other consumer goods89.171,128.828.77.636.5
Sensitive sectors58.891.731.924.844.641.3
Total7.661.215.319.3-16.223.1
Sources: GATT (1993), International Trade; and authors’ calculations.
Sources: GATT (1993), International Trade; and authors’ calculations.

In contrast, OECD Europe’s imports of sensitive goods from the CE4 account for less than 2 percent of total imports of these goods (the highest share is textiles, at 3.2 percent) and for less than 0.5 percent of gross production (Table 5). This basic asymmetry is clearly an important determinant of the impact of improved market access on employment. Additional factors are considered below.

Table 5.Selected Indicators of Sensitive Sectors in OECD Europe
Imports from CE4TotalLabor intensity
% of total

imports
% of gross

production
employment

(In millions)
(% of total

manufacturing)
Agriculture/Food2.10.312.8155.9
Textiles3.21.03.9161.3
Chemicals1.10.33.762.9
Metal products1.80.91.892.7
Total of above sectors1.70.422.1120.6
Sources: OECD foreign Trade by Commodities, Series C, 1990; OECD National Accounts, 1991; UN National Accounts Statistics: Main Aggregrates and Detailed Tables, 1990, Part I; and authors’ calculations and estimates.
Sources: OECD foreign Trade by Commodities, Series C, 1990; OECD National Accounts, 1991; UN National Accounts Statistics: Main Aggregrates and Detailed Tables, 1990, Part I; and authors’ calculations and estimates.

The first element for consideration is the time horizon of the analysis, which is a major factor in determining the impact of increased trade on employment. In the short to medium term, the level and structure of trade are determined mainly by the existing capital stock and the degree of market access. In the short run, there may be some scope for expanding exports in the so-called sensitive sectors, given existing capacity. Beyond the very short run, however, it seems unlikely that exports of iron and steel and chemicals will expand much, since it is not clear that the CE4 countries have a comparative advantage in these highly capital-intensive industries. Over a longer period, trade may also be constrained by the speed with which production capacity can be expanded in areas of comparative advantage. Agricultural products, food, and textiles and apparel offer scope for a more substantial expansion of exports over the medium term. The approach taken here, however, focuses on the short term and the expansion of exports that can be achieved with existing production capacity.

The next element to consider is the level to which trade may rise if obstacles are eliminated. Two methods have been used in the literature. The first one is the gravity model, which determines trade flows using country characteristics thought to influence trade. The most prominent variables are geographic distance and income level. Hamilton and Winters (1992) and Baldwin (1993) have used this approach to predict the level of trade between Western and Eastern Europe. Baldwin’s estimates of CE4 exports to the EU suggest that Hungary and Poland have already reached the maximum medium-term potential level of exports, while there is still room for expansion in the Czech and Slovak Republics (Table 6). The long-term potential is clearly much higher but is conditional on a catch-up scenario that lies beyond the scope of this paper.

Table 6.Potential and Actual Exports, 1992(In billions of U.S. dollars)
Baldwin: Exports to ECCollins/Rodrik: Total exports
CountryActualMedium-termLong-termActualMedium-termLong-term
Former Czechoslovakia7.214.971.411.536.169.2
Hungary5.35.526.610.720.348.9
Poland9.29.646.614.051.9148.2
Total21.730.0144.636.2108.9266.3
Sources: Baldwin (1993) and Collins and Rodrik (1991) converted into 1992 figures by multiplying with the nominal growth rate of world trade between the year for which estimate was done and 1992.
Sources: Baldwin (1993) and Collins and Rodrik (1991) converted into 1992 figures by multiplying with the nominal growth rate of world trade between the year for which estimate was done and 1992.

This second approach is basically counterfactual: how would trade have developed had the communist interlude not happened? Based on prewar trade matrices, the estimates by Collins and Rodrik in Table 6 suggest a much higher potential for export growth than Baldwin’s. How can the difference between these two approaches be explained? First, the predictive power of both approaches is not very high, and rough orders of magnitude are the most that can be expected. Nonetheless, the differences are strikingly large and probably due to different estimates for income (see Table 7), since Collins and Rodrik use purchasing power parity (PPP) estimates, while Baldwin uses an average of PPP and exchange rate-based measures. Despite the large differences, the researchers appear to agree that there is some scope for a further increase in exports from the CE4, although the potential is perhaps not as large as commonly thought.9 However, further information is clearly necessary in order to judge how much exports will increase if nontariff barriers are lifted. There are two possible approaches: one based on existing exports in relation to the production capacity in the CE4, and one based on the demand effects of eliminating NTBs.

Table 7.Estimates of Per Capita Income(In U.S. dollars)
AuthorYearCzechoslovakiaHungaryPoland
Based on purchasing power parity
CIA19897,9006,0904,560
EFTA19897,8806,1104,570
PlanEcon19887,6006,4905,450
Heston & Summers1985 (1980 dollars)7,4245,7654,913
Based on current exchange rates
CEPR1990/91 rate3,4002,6001,900
Credit Suisse First Boston1988 rate3,5003,0002,000
IBRD1990 rate avg.3,1402,7801,690
BIS1990/91 rate2,4002,7001,900
Average PPP7,7006,1144,873
Average current exchange rats3,1102,7701,873

Statistics on domestic sales and exports by industrial sector are not readily available; only Hungary publishes such figures on a regular basis (Table 8). On average, exports in sensitive sectors account for two fifths of total sales and, in some sectors, for one half. Of course, these figures provide no indication of total production capacity, which is likely to be higher than current sales (given the decline over the last few years). Nonetheless, the figures suggest that the potential for increasing exports rapidly, without adding to existing capacity or reducing domestic consumption dramatically, is rather limited. The situation appears to be similar in Poland (Misala 1994).

Table 8.Production and Exports in Hungary, 1993

(Q1 to Q3 at annual rate)

SectorDomestic salesExportsExports (% of
Class(In millions of US$)total sales)
Food154,17065713.6
Textiles1767921023.6
Apparel1815324461.4
Footwear, etc.1911211550.7
Chemicals241,21477939.1
Basic Metal2762439638.8
Total nonfood2,7831,74538.5
Source: Central Statistical Office, Statistical Bulletin 10/1993.
Source: Central Statistical Office, Statistical Bulletin 10/1993.

How much more in additional imports could OECD Europe absorb if NTBs were eliminated? One way to answer this question is to calculate the ad valorem tariff equivalents on CE4 exports. Eliminating NTBs can then be seen as equivalent to abolishing the tariff equivalents. Assuming the highest import elasticities found in the literature for these goods, it is possible to calculate the import increase that would result from a reduction in the price. Valid tariff equivalents of antidumping actions and other NTBs are notoriously difficult to calculate, and any estimate should be taken with a grain of salt. Table 9 shows the estimates by Messerlin (1993), which suggest an increase in the exports of sensitive goods of between 13 percent (for textiles) and 178 percent (for agricultural products).

Table 9.The Maximum Impact on CE4 Exports If EU NTBs Are Eliminated
IndustriesNTB ad valorem tariff

equivalent on CE4

exports (%)
Maximum import

elasticity in

literature
Export increase if

NTBs are eliminated

(% change)
Agriculture804.0178
Textiles101.413
Apparel183.046
Chemicals163.041
Steel224.082
Sources: Messerlin (1993), Table 5; and Aghion and others (1992). The export increase is estimated by applying the highest import elasticities in the literature to the reduction in import prices that results from the elimination of NTBs on CE4 exports.
Sources: Messerlin (1993), Table 5; and Aghion and others (1992). The export increase is estimated by applying the highest import elasticities in the literature to the reduction in import prices that results from the elimination of NTBs on CE4 exports.

The last element is the method used to estimate the employment impact. An aggregate macroeconomic model might give some indication, but the focus on sensitive sectors necessarily requires a disaggregated model. A partial industry equilibrium model could be used; Rollo and Smith (1993) use this method, although they do not focus on the employment impact. Most desirable would be a general equilibrium model with a suitable disaggregation. The approach taken in this paper is less ambitious and much more preliminary. Three simple scenarios are developed to provide a broad perspective on orders of magnitude. A brief description follows; a more detailed explanation is given in Appendix 2.

Three scenarios for export expansion in sensitive sectors provide the starting point for the analysis. It is assumed that these additional exports crowd out OECD Europe production one-for-one—an assumption that is bound to overstate the likely impact, because price changes and real income gains will lead to increased OECD demand and allow additional imports to be absorbed without corresponding supply reductions. The decline in value added in sensitive sectors is assumed to be proportional to the decline in gross production. The total decline in value added attributable to increased imports in sensitive sectors is, however, not confined to these sectors. Reduced production in textiles, for example, also leads to reduced demand for transportation, machinery, and services. The total effect is taken into account in the simulations and converted into employment losses, using sectoral productivity data.

It is highly likely that large increases in imports from CE4 countries will occur without corresponding increases in exports to those countries. After all, a major reason for the increased exports from the CE4 countries to OECD Europe is precisely to allow these countries to import more and to modernize capital stock without resorting to heavy external borrowing. To the degree that exports from OECD Europe to the CE4 are also rising, employment is created. The assumptions here are that exports will consist of a bundle of manufactures and that data for the manufacturing sectors as a whole are used to translate increases in value added into employment gains. In practice, the gains may be higher, since some of the additional value added is likely to come from service industries, where productivity is lower. Consequently, employment gains are probably understated and employment losses overstated, so that the calculations are clearly biased, showing higher employment losses in OECD Europe than are likely. It should be noted that based on these assumptions, trade growth will lead to employment losses in OECD Europe even if trade remains balanced, because import industries are more labor intensive than export industries. As long as interindustry trade dominates, these assumptions are realistic and consistent with analyses of the factor content of North-South trade (Wood 1993).

b. Scenarios of the employment impact

This section analyzes the impact of increased trade on employment in OECD Europe and attempts to quantify the differential impact on Austria and on Spain and Portugal. The analysis is divided into three scenarios. For each scenario, two sets of impact figures are shown (Table 10). The first shows the impact on gross employment in the sensitive sectors as well as in the rest of the economy, assuming unchanged OECD Europe exports. The second assumes either that the trade deficit of the CE4 deteriorates by $4.5 billion (scenario I) or that incremental trade is balanced (scenarios II and III).

Table 10.The Employment Impact in OECD Europe of Increased Trade with Eastern Europe: Three Scenarios(In thousands of U.S. dollars)
Scenario II:
Scenario I:Quotas/NTBsScenario III:
Sector1989-92abolishedImports quadruple
NetNetNet
Gross(trade

deficit)
Gross(balanced

trade)
Gross(balanced

trade)
Agriculture/Food-34-34-60-60-101-101
Textiles/Apparel-40-40-12-12-119-119
Chemicals-10-10-4-4-30-30
Metal-15-15-11-11-46-46
Other-140-178-14867-421247
Total Sensitive-99-99-87-87-296-296
Total-23979-201-19-717-49
(% of total employment in sector)
Agriculture/Food-0.3-0.3-0.5-0.5-0.8-0.8
Textiles/Apparel-1.0-1.0-0.3-0.3-3.0-3.0
Chemicals-0.3-0.3-0.1-0.1-0.8-0.8
Metal-0.9-0.9-0.6-0.6-2.6-2.6
Other-0.10.1-0.10.1-0.30.2
Total Sensitive-0.4-0.4-0.4-0.4-1.3-1.3
Total-0.20.1-0.10.0-0.50.0
Source: Authors’ estimates (using the procedure described in Appendix 2).
Source: Authors’ estimates (using the procedure described in Appendix 2).

The first scenario extends the 1989-92 period. It assumes that exports from the CE4 double in sensitive sectors to OECD Europe and that the trade balance ($4.5 billion) deteriorates (Table 11). Here 100,000 jobs are lost in the four sensitive sectors but about 200,000 jobs are created in other sectors. According to this scenario, growth in trade since the fall of the Berlin Wall may have helped to create about 100,000 net new jobs in OECD Europe and will do so again if this trend is extended.

Table 11.CE4 Trade with OECD Europe(In billions of U.S. dollars)
19891992Change
Exports12.624.7+96%
Imports13.229.8+126%
Balance-0.6-5.1-$4.5bn
Source: OECD Monthly Statistics of Foreign Trade, Series A.
Source: OECD Monthly Statistics of Foreign Trade, Series A.

The second scenario assumes that EU/European Free Trade Association (EFTA) nontariff barriers are lifted and that trade is balanced as far as changes are concerned.10 With the exception of the Czech Republic, CE4 countries are subject to balance of payment constraints, making further increases in trade deficits unlikely indeed. Assuming that the estimated tariff equivalents are reasonable, 100,000 jobs will be lost in sensitive sectors, for a net total loss of 20,000 jobs. If imports are quadrupled, 300,000 jobs will be lost in sensitive sectors, for a net loss of 50,000. The third scenario simply assumes, again under conditions of balanced trade, that CE4 exports to OECD Europe quadruple. This scenario is unlikely during the next few years and merely serves to provide an upper limit to the possible impact of trade on employment. Increases of this size are not improbable over the medium term if additional foreign direct investment helps build up capacity. However, with the existing capital stock, such increases are not anticipated.

The second panel of Table 10 scales these estimates with respect to sectoral and total employment and shows that the estimated impact is truly small. The losses in sensitive sectors between 1989 and 1992 may have amounted to 0.4 percent of employment in these sectors, against a net gain of 0.1 percent of total employment in OECD Europe. The elimination of NTBs may lead to a gross employment loss of 0.4 percent in sensitive sectors, but the impact on the total economy is negligible. Even in the unlikely quadrupling scenario, total employment losses are hardly noticeable, while sectoral losses rise to 1.3 percent. These figures suggest that employment losses in sensitive sectors resulting from the elimination of NTBs are very small indeed and that the impact on total employment in OECD Europe is minuscule. Taking trade growth since 1989 together with the complete elimination of NTBs in 1994, for example, still leaves an overall positive “jobs balance,” not to speak of wider benefits such as the gains from trade and the fact that these gains may help to maintain some measure of political stability in Eastern Europe.

The distribution of employment losses or gains within OECD Europe is probably at least as interesting as the overall picture. It has been argued that the countries closest to Eastern Europe are most likely to be affected because of their geographical proximity (Holzmann, Thimann, and Petz 1994). Others have thought that countries like Spain or Portugal should feel more adjustment pressure, because their share of employment and value added in sensitive sectors is larger (Hochreiter 1993b). A detailed analysis of these effects is clearly beyond the scope of this paper. It would have to take into account not only that the Southern countries are competitors with Eastern Europe in third markets like OECD Europe but also that products from the South (like fruits and vegetables) may be particularly well suited to Eastern Europe.

The exercise pursued here is simple and should be understood as a first illustration of some of the influences involved. While we feel that the estimates of the overall impact presented above are quite robust, in the sense that more refined methods and data are unlikely to alter the basic picture, we regard the following analysis as much more tentative and preliminary.

The idea is to capture both the geographical proximity factors and the sectoral composition factor, again assuming that exports from the CE4 crowd out OECD Europe production. The geographical factor is taken care of by comparing typical trade shares derived using counterfactual or gravity model analysis with actual trade shares. The bigger the gap, the higher the potential for catch-up. Using the figures of Collins and Rodrik (1991), the gap appears largest for Portugal and Spain and rather small for Austria (Table 12). Thus, the geographical distance argument points to a rather low impact on Austria because its share of trade is already closer to the typical level than the shares of Spain and Portugal—contrary to what might be expected. If actual exports reach their predicted ratios over a period of ten years, exports to Austria will have to grow at an annual rate of 8 percent, compared with 22 percent for Spain and 33 percent for Portugal. Spain and Portugal, as we have noted, are also much more exposed in sensitive sectors.

Table 12.The Regional Employment Impact of Increased CE4 Exports: A Comparative Scenario(In percent)
AustriaSpainPortugal
Employment in sensitive sectors
(% of total employment)10.919.033.9
Share in CE4 exports in 1989
Actual3.70.50.0
Predicted (Collins/Rodrik)8.23.60.6
Ratio Predicted/Actual2.27.417.4
Annual export growth if predicted ratio reached in 10 years8.022.033.0
Scenario: sensitive exports of CE4 double
Gross employment impact if export shares reach predicted ratio in 10 years
% of employment in sensitive sectors-1.7-0.1-0.1
% of total employment-0.20.00.0
Required export growth for same employment impact in all countries7.0200.0600.0
Sources: Authors’ estimates
Sources: Authors’ estimates

At first glance, the impact on gross employment of a doubling of CE4 exports in sensitive sectors is paradoxically much higher in Austria than in the other two countries, from both the sectoral perspective and the point of view of overall employment. The reason for this difference is that since Spanish and Portuguese imports in sensitive sectors start from a small base, even quite substantial import growth has very little effect on employment in these sectors; only over time do the effects become larger. Starting from present import levels, import growth in Spain and Portugal will need to be a large multiple of Austria’s in order to generate a similar employment impact (last line in Table 12). Thus, despite structural factors limiting the initial impact, Austria will continue to be much more affected than the Southern countries.11

c. Employment impacts: historical trends and unemployment in Eastern Europe

While the employment impacts calculated above are small relative to current employment in the sensitive sectors, they may still be rather large in comparison with past changes in these sectors, creating what are felt as strong adjustment pressures and a corresponding resistance to liberalizing trade. Moreover, there may still be substantial gains in Eastern Europe as a result of increased trade. Table 13 shows annual employment changes in OECD Europe by sector and for the entire economy, shedding some light on the first of the above issues. Between 1980 and 1991, employment in the sensitive sectors declined at an annual rate of 2.2 percent, a figure that rises to around 3 percent when food is excluded. Hence, even the gross employment impact of Scenario III (1.3 percent), which assumes an adjustment over several years, is well below the average annual change.

Table 13.Sectoral Unemployment Changes in OECD Europe, 1980-91(Percentage change, annual rates)
Sector1980-91
Agriculture-3.1
Food-1.0
Textiles-3.0
Wood-2.1
Paper-0.5
Chemicals-0.1
Mineral-1.9
Basic metal-2.5
Machinery-0.7
Manufacturing-1.2
Other-0.9
Sensitive sectors-2.2
Total0.4
Source: OECD National Accounts.
Source: OECD National Accounts.

Table 14 shows potential reductions in current unemployment in Eastern Europe, using the various scenarios presented in Table 10. Because of existing differences in productivity, it has been assumed that for each job lost in the West, two jobs will be gained in the CE4. On this basis, the gross effects of Scenarios I and II amount to a decline in the number of unemployed of 10-12 percent. The gross effects of Scenario III reduce unemployment to only 25 percent of the present level, but (as noted earlier) given existing capacities, these effects are unfeasible without additional investment. As in the case of OECD Europe, the employment effects appear to be rather small in the more realistic scenarios that assume balanced trade (Table 14, Scenarios II and III). Consequently, even though the potential effects are somewhat larger in Eastern than in Western Europe, they provide only limited relief to the problem of rising unemployment.

Table 14.Potential Unemployment Effects in the CE4: Three Scenarios
Scenario IScenario IIScenario III
GrossNet (DEF)GrossNet (B)GrossNet (B)
Changes in total employment-23979-201-19-717-49
(in thousands)
% change in unemployment-12.31.0-10.3-1.0-73.7-1.5
Sources: Table 11, and authors’ estimates.
Sources: Table 11, and authors’ estimates.

3. Summary and Conclusions

Even though East European countries consider increased trade the most effective form of support for their reforms, the EC/EFTA countries continue to restrict market access for half of Eastern Europe’s exports. In the public debate, fears of additional employment losses in a period of rapidly rising unemployment feature prominently. This paper has attempted to determine whether these fears are justified, using two approaches. First, we review the literature on the impact of trade on employment. Second, we develop scenarios to calculate the employment losses in OECD Europe as the result of reduced trade barriers for East European exports. The two parts of the paper come to mutually reinforcing conclusions. The short-run effects on employment of reduced trade barriers are very small, so small in fact that they are within the normal error range of the calculations. Moreover, once the long-run income gains on both sides of increased trade are taken into account, it is evident that fears of employment losses can be understood only as misperceptions.

The conclusion that the impact on employment of increased trade with Eastern Europe is likely to be small should perhaps not come as a surprise. Recent studies of the employment consequences of the North American Free Trade Agreement (NAFTA) have also found only very small effects (Hufbauer and Schott 1993), and a much earlier study of the relationship between trade and employment by the U.S. Department of Labor concluded that “the modest magnitude of these impacts in total is striking and especially so in the view of assumptions that tend to bias results in the direction of exaggerating the adverse impacts” (Dewald 1978, p. 6).

Part I of the paper describes how, over the last 20-25 years, the share of manufacturing in total employment has declined in most industrial countries. Over the same period, unemployment has increased in OECD Europe, and the United States has experienced a combination of stagnating aggregate real wages and falling relative wages for the unskilled. The last 20-25 years have also witnessed a significant rise in the share of developing countries in world trade, notably in laborintensive manufactured goods. According to standard trade theory, the opening of trade between countries with different factor endowments leads to a convergence of factor proportions and relative factor prices. When applied to trade between industrial and developing countries, this theory posits a shift in the distribution of factor income in favor of capital and skilled workers. Moreover, because increases in trade will be accompanied by sectoral and regional changes in output and employment, a rise in frictional unemployment is to be expected.

Empirical tests of the predictions of standard trade theory provide only mixed to weak evidence even for trade between countries with very different factor endowments. One reason may be that despite spectacular growth rates, the shares of developing countries in world trade and in industrial countries’ total demand are still very low. Another reason is that the applied models vary widely and do not always satisfy the assumptions of standard trade theory. Summarizing this varied and mixed picture, we find that increased trade and more import competition have had some effects on aggregate employment and on the structure of employment and wages in the industrial countries but that these effects are small compared with the changes brought about by developments in domestic demand and in the nature of technological progress.

Part II is empirical, developing three scenarios of the impact on employment of increased trade between OECD Europe and Eastern Europe. The focus is on a reduction of NTBs in the so-called sensitive sectors, which account for half of East European exports. Even with assumptions that clearly bias the estimates toward overstating the consequences, the employment effects in OECD Europe are very small, both in relation to current employment and unemployment levels and compared with historical changes. This finding is perhaps surprising in view of the current debate, but it is entirely consistent with international experience and underscores the fact that trade-induced employment effects are small and dominated by changes related to domestic developments. The employment effects in Eastern Europe are also small, but the calculations for East European countries may understate the implicit consequences of not gaining greater market access. A generally recognized condition for successful reform is higher investment, but so far the ratio of investment to GDP has remained insufficient in most of these countries. It is difficult to calculate the investment/GDP ratios needed for sustainable growth, but it seems fairly evident that domestic saving will not be sufficient, at least in the early phase. It is therefore important to create incentives for capital imports, in particular foreign direct investment. Increased market access and more liberal domestic content rules could act as powerful stimulants for such inflows.

By contrast, if trade remains restricted and investment/GDP ratios stay depressed, a growing number of unemployed workers will no doubt start looking for jobs in Western countries. Considering the U.S. experience with immigration (mostly of unskilled workers), such a development could lead to adjustment pressures very similar to those estimated for expanded trade, except that the risk of social tensions and the need for additional residential construction will be much higher. Furthermore, in the long run Western Europe is likely to gain in terms of income as well as employment. Analytical arguments as well as empirical estimates clearly support granting more market access to Eastern Europe. What remains is to change popular misperceptions of trade.

Appendix 1 Manufacturing employment and unemployment

In their predictions of the demise of manufacturing in OECD countries, Brown and Julius (1993) draw on the negative trends seen for agriculture during most of this century. One interesting feature of these negative developments is that they were not accompanied by a rise in aggregate unemployment, because workers released from agriculture were able to find employment in industry or the services sectors. By contrast, during the more recent period of falling employment in manufacturing, most industrial countries have experienced a marked rise in unemployment. Consequently, there may be a risk that the loss of jobs in manufacturing will be accompanied by a permanent rise in unemployment and that the prospects of additional job losses induced by trade liberalization will generate strong resistance to expanded trade.

This question is further pursued in Appendix Tables 1 and 2, using the United States and Germany as representatives of OECD countries. As can be seen from Table 1, there is no long-run positive or negative trend for unemployment from 1900-92, when agricultural employment declined, and the Dickey-Fuller test rejects the null hypothesis of a random walk. By contrast, over the last 30 years, the random walk hypothesis can be rejected only when a positive time trend is added. The trend is almost five times steeper for Germany than for the United States, but analyzing unemployment on the basis of the Okun equation hardly changes this impression. As can be seen from columns (1) and (3) of Appendix Table 2, the output gap has a significant influence on unemployment in both countries but, in addition, unemployment has a positive trend (which again is much steeper for Germany than for the United States). In other words, while actual output tends to return to potential output following a negative shock, unemployment remains at a higher level or returns to the initial level only with a considerable time lag. It is a moot point whether employment shocks lead to a shift in the equilibrium rate of unemployment (hysteresis), or whether it merely takes a long time for actual unemployment to return to the earlier equilibrium (high degree of persistence).

Appendix Table 1.Unemployment: Autoregressions 1(Annual data)
CountriesΣU-iTrR2ADF2ADF3
1900-92
United States0.82-0.20 (0.2)0.79-0.17 (3.1)
Germany0.80-0.22 (0.2)0.75-0.20 (3.2)
1960-92
United States0.593.31 (2.5)0.96-0.27 (2.3)-0.41 (3.0)
Germany0.737.41 (2.4)0.94-0.04 (0.9)-0.23 (3.0)

Note: Critical values for rejecting the null hypothesis of a random walk are respectively 2.60 (1 percent significance level) and 1.95 (5 percent).

U = rate of unemployment (dependent variable)

ΣU-i = the sum of coefficients for U-i (i = 1.4)

Tr = a linear time trend (t-ratios in brackets)

ADF = the coefficient β in the regression dU = βU-1 + αi dU-i (i = 1.4; t-ratios in brackets)

Without trend.

With trend.

Note: Critical values for rejecting the null hypothesis of a random walk are respectively 2.60 (1 percent significance level) and 1.95 (5 percent).

U = rate of unemployment (dependent variable)

ΣU-i = the sum of coefficients for U-i (i = 1.4)

Tr = a linear time trend (t-ratios in brackets)

ADF = the coefficient β in the regression dU = βU-1 + αi dU-i (i = 1.4; t-ratios in brackets)

Without trend.

With trend.

Appendix Table 2.Explaining Changes in Unemployment 1
VariablesGermanyUnited States
1. Extended Okun equations
(1)(2)(3)(4)
(Quarterly data: 1960; Q1-1993; Q3)
C-1.1724.8720.102
Tr7.9021.902-0.102
Y/Y*-0.572-0.292-0.662-0.562
MAN-0.572-0.502
NWAGE0.7221.372
INV (1)0.1320.14
INV (2)0.252
LU0.102
WCV0.062
URANGE0.332
R20.850.960.530.91
SEE1.250.651.060.50
DW0.090.140.100.44
II. Contributions to changes in unemployment
Period1973-751980-851973-751979-82
Y/Y*1.20.82.93.3
MAN1.21.51.51.6
NWAGE0.91.00.20.6
INV-0.6-0.2-0.2-0.1
LU0.3
URANGE0.41.9
WCV0.1
Tr-0.6
Σ3.15.03.94.3
dU4.15.84.15.0

Notation: Y/Y* = ratio of actual to potential GDP, with the latter estimated using a Hodrick-Prescott filter;

MAN = ratio of manufacturing employment to total employment;

NWAGE = ratio of non-wage labor costs to wages;

INV1 = ratio of gross fixed investment to GDP;

INV2 = ratio of investment in buildings and structure to investment in machinery and equipment;

LU = ratio of long-term unemployed to all unemployed;

WCV = coefficient of variation for compensation per employee in 8 manufacturing industries;

URANGE = range of unemployment rates across states (50 for the United States and 11 for Germany);

Tr = time trend; and

C = intercept.

Coefficients significant at 1 percent level of significance.

Notation: Y/Y* = ratio of actual to potential GDP, with the latter estimated using a Hodrick-Prescott filter;

MAN = ratio of manufacturing employment to total employment;

NWAGE = ratio of non-wage labor costs to wages;

INV1 = ratio of gross fixed investment to GDP;

INV2 = ratio of investment in buildings and structure to investment in machinery and equipment;

LU = ratio of long-term unemployed to all unemployed;

WCV = coefficient of variation for compensation per employee in 8 manufacturing industries;

URANGE = range of unemployment rates across states (50 for the United States and 11 for Germany);

Tr = time trend; and

C = intercept.

Coefficients significant at 1 percent level of significance.

While adding a time trend considerably improves the fit of the Okun equations, it is obviously not very satisfactory to explain the rise in unemployment by a time trend. Moreover, even with a trend, the Durbin Watson (DW) statistics remain very low, suggesting that the equations are misspecified and/or that some important variables have been left out. Consequently, in a rather ad hoc way and without attempting to construct a comprehensive labor market model, we added various variables proposed in the literature; the results are shown in columns (2) and (4) of Appendix Table 2. Despite the high R2s, the equations are still not very satisfactory, as the DW statistics remain far too low and the exogeneity of some of the variables is questionable. Yet there is one striking feature: in both countries, the declining share of manufacturing in total employment is strongly correlated with the increase of unemployment, accounting for about one third of the rise during the two periods when the increase was most pronounced. Moreover, since the share of manufacturing employment has not returned to its earlier level after the two recessions, deindustrialization may be a reason for the random walk behavior of aggregate unemployment. If this interpretation is valid, it would provide a basis for the likely reactions to the further threat to the manufacturing sector that will follow the opening of trade with Eastern Europe.

Appendix 2 Estimating the employment impact of increased trade between OECD Europe and Eastern Europe

The employment impact in OECD Europe of increased trade with Eastern Europe is estimated with a simple back-of-the-envelope calculation. Increased imports from Eastern Europe in a particular sector are assumed to lead to a complete crowding out of OECD Europe production:

with P representing total production and M representing imports. The subscript i represents the sensitive sectors, so i = 1, 2, 3, or 4 (1 = agriculture/food, 2 = textiles/apparel, 3 = chemicals, 4 = metal products). This calculation amounts to assuming inelastic East European supply combined with inelastic Western demand and/or perfectly elastic Western supply.

The corresponding decline in value added Y is given by

and, finally, the employment E impact in the sensitive sector is, assuming fixed coefficients:

Combining (1) and (3) yields a simple equation that expresses employment changes in the sensitive sector i as a function of import changes:

Total employment, however, changes by more than it does in the sensitive sector only. The imports that do not reduce value added in the “sensitive” sector reduce value added (again assuming full crowding out of domestic production) in other parts of the economy (sector k). Analogous to the considerations above, this impact is given by

The total gross employment impact of increased imports from Eastern Europe is then given by:

The employment effect of exports is calculated in an comparable manner. The two estimates are then combined to yield net figures for the employment impact of increased trade.

Value added and employment by sector are taken from the OECD National Accounts. These statistics cover only employees, not the self-employed, and thus are likely to lead to an underestimation of the employment in the agricultural sector. The ratio between gross output and value added is the average of the ratios for Germany and France as reported in the United Nations National Accounts. Trade data are from the OECD Trade Series C, 1990. Data for Eastern Europe refer, unless otherwise noted, to the Czech Republic, Poland, Hungary, and the Slovak Republic.

Appendix Table 3.Value Added in Sensitive Sectors, 1991(In percent of GDP)
AgricultureFoodAgriculture &

Food
Textiles &

Apparel
ChemicalBasic SteelTotal

Non-food
Total

Sensitive
Austria2.83.46.21.53.61.46.512.7
Belgium1.83.14.91.55.42.19.013.8
Denmark3.53.36.80.82.30.23.310.1
Finland4.82.47.20.62.10.83.510.6
France3.12.85.91.24.01.16.312.2
Germany1.33.24.51.25.52.28.913.3
Iceland10.16.416.60.90.90.72.519.1
Italy3.32.25.53.32.80.87.012.5
Luxembourg1.42.23.60.82.44.88.011.6
Netherlands3.92.96.80.65.00.56.112.9
Norway2.93.05.90.31.61.02.88.7
Portugal5.86.011.87.22.11.210.522.2
Spain4.04.58.42.14.91.38.216.7
Sweden2.12.24.30.42.10.73.27.5
Greece14.13.617.72.52.60.65.623.3
United Kingdom1.52.13.61.32.20.74.27.8
OECD Europe2.72.95.61.63.81.26.712.2
Sources: OECD National Accounts; UK National Accounts, 1991.
Sources: OECD National Accounts; UK National Accounts, 1991.
Appendix Table 4.Employment in Sensitive Sectors, 1991(In percent of total employment)
AgricultureFoodAgriculture &

Food
Textiles &

Apparel
ChemicalBasic SteelTotal Non-foodTotal Sensitive
Austria0.93.24.12.72.41.66.710.9
Belgium2.62.75.32.42.61.46.311.6
Denmark5.63.59.01.31.90.23.412.5
Finland9.02.411.51.31.70.73.715.2
France5.72.68.31.82.41.05.313.5
Germany3.32.96.21.93.92.48.114.3
Iceland10.49.620.01.60.80.83.223.2
Italy9.51.611.24.92.00.77.618.8
Luxembourg3.11.84.90.52.95.79.114.0
Netherlands4.93.17.91.02.60.64.112.0
Norway6.12.58.60.51.90.93.412.0
Portugal20.33.323.68.31.50.510.333.9
Spain9.93.213.23.31.90.65.919.0
Sweden1.31.93.10.61.81.03.36.5
Greece
United Kingdom1.22.43.71.91.41.64.98.5
OECD Europe6.22.68.82.62.41.36.215.0
Sources: OECD National Accounts; UK Employment Gazette, December 1993.
Sources: OECD National Accounts; UK Employment Gazette, December 1993.
References

    AghionP. and others. 1992. “Towards the Establishment of a Continental European Customs Union.” In Trade Payments and Adjustment in Central and Eastern Europeed.J.Flemming and J.M.C.Rollo.London: Royal Institute of International Affairs and the European Bank for Reconstruction and Development pp. 157-80.

    BaldwinR.E.1993. The Potential for Trade between the Countries of EFTA and Central and Eastern Europe. EFTA Occasional Paper 44. Geneva: European Free Trade Association.

    BorjasG. and V.Ramey.1993. “Foreign Competition, Market Power and Wage Inequality: Theory and Evidence.NBER Working Paper 4556. Cambridge, Mass.: National Bureau of Economic Research.

    BrownR. and D.Julius.1993. “Manufacturing in the New World Order.” In Finance and the International Economyed.R.O’Brien.Oxford: Oxford University Press pp. 7-17.

    BuschB. and H.P.Fröhlich.1993. “Aufbruch im Osten: Anpassung im Westen. Europa nach dem Fall des Eisernen Vorhangs.Beiträge zur Wirtschafts und Sozialpolitik 211. Köln: Institut der deutschen Wirtschaft.

    CollinsS.M. and RodrikD.1991. Eastern Europe and the Soviet Union in the World Economy. Policy Analyses in International Economics 32. Washington, D.C.: Institute for International Economics.

    DavisS.J.1992. “Cross-country Patterns of Change in Relative Wages.” In NBER Macroeconomics Annual.Cambridge, Mass.: National Bureau of Economic Research pp. 239-300.

    DewaldW.G.ed.1978. The Impact of International Trade and Investment on Employment. Washington, D.C.: U.S. Department of Labor.

    GreyA.1993. “Interindustry Wage Differentials in Manufacturing: Rent and Industrial Structure.Canadian Journal of Economics pp. 525-33.

    HamiltonC.B. and L.A.Winters.1992. “Trade with Eastern Europe.Economic Policy 14 pp. 77-116.

    HochreiterE.ed.1993a. “The Impact of the Opening of the East on the Austrian Economy: A First Quantitative Assessment.Working Paper 11. Vienna: Österreichische Nationalbank.

    HochreiterE.1993b. “Pressure for Adjustment in the West: Some Macroeconomic Consequences of the Opening of Eastern Europe: Comment.Vienna: Österreichische Nationalbank. Mimeo.

    HolzmannR.C.Thimann and A.Petz.Pressure to Adjust, Consequences for the OECD Countries from Reforms in Eastern Europe,Empiricaforthcoming.

    HufbauerG.C. and J.J.Schott.1993. NAFTA: An Assessment. Washington, D.C.: Institute for International Economics.

    HughesG. and P.Hare.1992. “Trade Policy and Restructuring in Eastern Europe.” In Trade Payments and Adjustment pp. 181-209.

    JohnsonG. and F.Stafford.1992. “Models of Real Wages and International Competition.Ann Arbor: University of Michigan, Department of Economics.

    KatzL.F. and K.M.Murphy.1992. “Changes in Relative Wages, 1963-87: Supply and Demand Factors.Quarterly Journal of Economics pp. 35-78.

    KatzL.F. and L.Summers.1989. “Industry Rents: Evidence and Implications.Brookings Papers on Economic Activity pp. 209-75.

    KrugmanP. and R. LawrenceR.1993. “Trade, Jobs and Wages.NBER Working Paper 4478. Cambridge, MA: National Bureau for Economic Research.

    LawrenceR.Z. and M.J.Slaughter1993. “International Trade and American Wages in the 1980s: Giant Sucking Sound or Small Hiccup?Brookings Papers on Economic Activity Microeconomics 2 pp. 161-226.

    LevyF. and R.J.Murnane.1992. “U.S. Earnings Levels and Earnings Inequality: A Review of Recent Trends and Proposed Explanations.Journal of Economic Literature pp. 1333-81.

    MacFarlaneI.1993. “Forecasting the Australian Economy,Reserve Bank of Australia BulletinSeptember.

    MartinsJ.O.1993. “Market Structure, International Trade and Relative Wages.OECD Working Paper 134. Paris: Organization for Economic Cooperation and Development.

    MastropasquaC. and V.Rolli.1993. “Industrial Countries’ Protectionism with Respect to Eastern Europe: The Impact of the Association Agreement Concluded with the EC on the Exports of Poland, Czechoslovakia and Hungary.Bank of Italy Discussion Paper 188. Rome: Bank of Italy.

    MesserlinP.A.1993. “The EC and Central Europe: The Missed Rendezvous of 1992?Economics of Transition Vol. 1 pp. 89-109.

    MisalaJ.1994. “Development Tendencies of Polish Foreign Trade in 1990-93.Foreign Trade Research Institute Discussion Paper 52.

    MurphyK.M. and F.Welch.1992. “The Structure of Wages.Quarterly Journal of Economics pp. 39-69.

    OstryS.1993. “The Threat of Managed Trade to Transforming Economies.EBRD Discussion Paper 3. London: European Bank for Reconstruction and Development.

    RevengaA.L.1992. “The Impact of Import Competition on Employment and Wages in US Manufacturing.Quarterly Journal of Economics pp. 255-84.

    RichterS.1993. “East-West Trade under Growing Western Protectionism.Forschungsberichte 198. Vienna: Wiener Institut für Internationale Wirtschaftsvergleiche.

    RolloJ. and A.Smith.1993. “EC Trade with Eastern Europe.Economic Policy 16 pp. 139-81.

    WangZ.K. and L.A.Winters.1993a. “Liberalising EC Imports of Footwear from Eastern Europe.CEPR Discussion Paper 836. London: Center for European Policy and Research.

    WangZ.K. and L.A.Winters.. 1993b. “EC Imports from Eastern Europe: Iron and Steel.CEPR Discussion Paper 825. London: Center for Economic Policy and Research.

    WoodA.1993. “The Factor Content of North-South Trade in Manufactures Reconsidered.Weltwirtschaftliches Archiv pp. 719-743.

The data for these estimates cover OECD Europe and the Czech Republic, Poland, Hungary, and the Slovak Republic, which account for the bulk of trade with OECD Europe.

The traditional method of calculating employment effects can be summarized in the following two equations:

where c is a vector of factor quantities for, respectively, the North and South; X is the total value of North-South trade; cx,n and cx,s represent vectors of factor quantities in exports from, respectively, the North and South; and cm,n and cm,s represent vectors of factor quantities in the sectors of, respectively, the North and South, displaced by imports from, respectively, the South and North.

The loss of jobs induced by relocation is difficult to model precisely when looking at import data alone. To the extent that current imports include the effects of production relocations in the past, the estimates of the impact of current imports on jobs is subject to the same criticism as Wood’s argument with respect to noncompeting imports. More important, perhaps, is the need to take account of potential job losses through relocation when estimating future imports following trade liberalization, although, for U.S. manufacturing, outsourcing and relocation have so far had only a marginal effect on factor proportions (Lawrence and Slaughter 1993).

If real and relative wages in the West do not adjust to shifts in labor demand, it is not sufficient to derive future employment effects from current trade flows, since the loss of jobs through relocation should also be taken into account.

Similar results were previously found for U.S. manufacturing; see Katz and Summers (1989).

Analyzing relative employment and wage effects on the basis of trade balances or trade flows is a practice actually rather weakly grounded in standard trade theory but nevertheless widely used (Lawrence and Slaughter 1993).

Using a similar method but looking at relative supply shifts and including the endowment effects of immigration as well as net trade, Levy and Murnane (1992) report findings that arrive at higher figures. According to these calculations, the supply of high school dropouts relative to the supply of college graduates increased by one quarter to one third for men and by almost one half for women.

An early assessment of the effects on Austria is contained in Hochreiter (1993a).

Both approaches focus on total trade. For the exercise at hand, it would be preferable to exclude energy trade, though the differences are unlikely to be large.

Less ambitious measures to liberalize trade with Eastern Europe in a multilateral context are discussed by Ostry (1993).

It may be of interest to note that these tentative estimates contradict those presented by Holzmann, Thimann, and Petz (1994), who find an extremely strong impact on Austria, though it is not quite clear what lies at the root of their result.

    Other Resources Citing This Publication