Week-8

Economic Policy October 1989 PrinUd in Great Britain

EFTA and the internal European market

Victor Norman

Summary

The EFTA countries must come to grips with the EG's single market. This article contrasts two options. Under the first option, the EFTA countries do not change their current arrangements with the EG. Under the second option, they join the EG's common market. Since the EG is the EFTA countries' main trading partner, significant changes in cost, demand and market structures within the EG must substantially affect EFTA. The simulation results - focusing on the motor vehicle and pharmaceutical industries - confirm that 1992 is important for EFTA countries, not so much because of the threat that an internal EG market represents, but rather because of the opportunities that a larger European market offers. Generally, the EFTA countries would not seem to lose much if they decided to stay outside the internal market but they could share much of the gains if they chose to integrate themselves. Quantitatively, the gain accru- ing to EFTA is likely to be greater than the gain to the typical EG country. This is because EFTA home markets are small, so that the tradeoff between scale economies and competition is more severe than for the larger EG countries.

EFTA and the internal European market Victor D. Norman Norwegian School of Economics and Business Administration and CEPR

1. Introduction

The EFTA countries - Austria, Finland, Iceland, Norway, Sweden and Switzerland - must soon decide how to react to the establishment of the internal market within the European Community. They have several options. At one extreme, they could do nothing - trusting that the free trade arrangements they already have with the Community will protect them from overly damaging effects, and hoping that they may perhaps benefit indirectly from higher efficiency and demand within the EC. At the other extreme, they could join the Community or enter into individual or joint agreements with the EC about a larger 'European space' establishing an internal market of 18 countries in Europe. Between the two extremes are a number of less radical alternatives; for example, EFTA might selectively appeal to the Commission s reciprocity principle to become part of the European market. Only key products would be concerned and the EC directives for those goods would be copied. The purpose of this paper is to assess the consequences for EFTA countries of the establishment of the EC internal market under two extreme alternatives: the EFTA countries do not take part in the integration process at all and current trade impediments between EFTA and EC remain; or else the EFTA countries become part of the internal market, and trade impediments between EFTA and EC are reduced to the same extent as inside the EC.

At the micro level, Emerson et al. (1988) distinguish three principal effects of an internal European market: lower real trade costs, more I I I am very grateful to Alasdair Smith and Anthony Venables who have developed the model used in a subsumial portion of this paper, and who have given me the data used in their application of the model to EC industries. I also want to thank Harry Flam, Henrik Horn, Jan HaaUnd and Anthony Venables, as well as members of the Economic Policy panel and seminar participants at the Norwegian School of Economics, for valuable comments. Finally, thanks to Charles Wyplosi for both valuable comments and editorial assistance.

EFTA and the internal European market 425

aggressive competition and lower unit production costs. Lower trade costs will result from the reduction or abolition of border controls and the harmonization of product standards. More aggressive competition will automatically follow: lower trade costs will promote exports, so that all producers will have less of a shelter at home, while also exporting more. The effect will, however, be significantly greater if the internal market brings to an end intra-EC price discrimination, i.e. if markets become truly integrated. Evidence of substantial price discrimination suggests that firms exploit large market shares in their home markets by charging higher prices at home, and selling at lower prices in export markets. If markets become fully integrated, such discrimination will no longer be possible and firms will not be able to capitalize on a dominant position in their home markets. Finally, more aggressive competition could reduce average production costs, because high-cost producers will lose market shares, and because a more competitive market could foster larger (possibly fewer) firms which take better advantage of scale economies.

Some of these benefits can accrue in all industries simultaneously; for example, trade costs can be reduced across the board or sales can be reallocated from export to domestic markets without affecting other markets. However, increases in industry output require a reallocation of resources in favour of the industries concerned. Such increases can clearly not occur in all industries simultaneously. Instead, we expect reallocation from industries where the competition effects of 1992 are small, to industries where they are large. Such general equilibrium repercussions should therefore dampen, but not eliminate, the benefits from the internal market.

If EFTA countries become part of the integrated market, they should qualitatively derive the same benefits as the EC countries. Quantitatively, the gain accruing to EFTA is, however, likely to be greater than the gain to the typical EC country; the reason is that EFTA home mar- kets are small, and as a result EFTA countries experience a more severe tradeoff between scale economies and competition than the larger EC countries. The EFTA countries are very open, and their trade with EC countries is, in relative terms, far greater than trade be- tween large EC countries like the UK and France. As a result, if EFTA countries remain outside, changes in cost, demand and market struc- tures within the EC could have substantial spill-over effects on EFTA countries.

European integration will have two conflicting effects on the exports of non-member countries: on the one hand, greater efficiency and correspondingly higher real income within the EC could raise demand for non-EC country exports. As the EC is by far the most important

426 Victor Norman

export market of EFTA, such a demand effect could be significant. On the other hand, more aggressive competition in EC markets and lower transaction and production costs within the Community, could reduce the market shares and depress the export prices of non-EC producers. The two effects correspond roughly to the ideas of 'trade creation' and 'trade diversion' known from the literature on customs unions: lower trade barriers between some countries will generate more trade ('trade creation'), but will also divert trade from third countries to the members of the union ('trade diversion').

EC integration could also have spill-over effects in the home markets of non-EC countries. If, for example, EC and EFTA markets are less than perfectly segmented, increased competition within the EC will reduce profit margins in EFTA markets as well, to the advantage of domestic consumers. Similar effects will occur if 1992 lowers the mar- ginal cost of EC producers. On the other hand, there could be adverse product-selection effects, because increased competition in export markets might force EFTA firms to abandon some of their products. As an example, we could imagine that Swedish car manufacturers would be forced to abandon some models. To the extent that Swedish cars have particular appeal to Swedish consumers, the loss of consumer surplus could be significant. The net effect on EFTA countries if they remain outside is, therefore, in principle unclear, but likely to be negative.

The paper is organized as follows. Section 2 gives some background information on the structure of EFTA production and trade. Section 3 presents the method of investigation adopted here, namely industry simulations. The following three sections present and interpret the results of a number of simulations for the motor vehicle and phar- maceutical industries. These simulation experiments deal with the reduction in trade costs, the integration of previously segmented markets, and a reduction in production costs. Seaion 7 considers the role of possible entry and exit of firms. Seaion 8 investigates the sensitivity of the welfare results to the modelling of market shares. In this respect, the respective weight given to non-tariff trade barriers and consumer preferences in explaining difference in market shares turns out to be important. Because these results are derived from paitial equilibrium simulations, they may miss some important economy-wide effeas. Seaion 9 indicates that general equilibrium interaction dampens the effeas which were previously identified, but that does not alter the main conclusions. Finally, in Seaion 10, some policy implications are pointed out.

At the outset, it is also worth noting that the discussion focuses on the gains from integration which arise from improvements in industry

EFTA and the internal European market 427

structure and performance. It is based on the recent literature on trade and trade policy dealing with imperfectly competitive markets. An introduction to this literature is given in Helpman and Krugman (1989). Other sources of integration gains - notably gains relating to the integra- tion of European factor markets - are ignored. The same is partly true for gains arising from comparative advantage - some comparative advantage effects are implicit in the general equilibrium analysis in Section 9, bu^ a more detailed analysis of how 1992 could affect the exploitation of comparative advantage in Europe is beyond the scope of this paper. (For a study of comparative advantage between EFTA and EC countries, see Haaland and Norman, 1987). The paper draws on several recent studies of 1992 and EFTA countries. Krugman (1988) sketches many of the qualitative market structure effects analyzed more closely here. Flam and Horn (1989) give a detailed survey of possible effects for Sweden. Some of these effects are discussed in greater detail by Lundberg (1989), who also gives much relevant data and presents a highly interesting study of Sweden s comparative advantage vis-a-vis the EC.

2. The structure of EFTA production and trade

There are three aspects of EFTA production and trade to keep in mind when looking at the relationship between EFTA countries and the EC. The first is size: EFTA countries are very small. Taken together, all the EFTA countries add up to an economy only half the size of West Germany. The implication is that the EC is far more important to EFTA countries than EFTA is to the EC. The second is that intra-EFTA trade, with one important exception, is insignificant relative to the trade of individual EFTA countries with the EC. The third is that EFTA trade with the EC reflects significant inter-industry specialization, the Scan- dinavian EFTA countries in particular having substantial net exports of metals and forest products, and net imports of other manufactures.

The second characteristic is evident from Table 1, which gives the percentage distribution of EFTA trade across trading partners. The EFTA countries have been split into Alpine EFTA (Austria and Switzer- land) and Nordic EFTA (Iceland, Finland, Norway and Sweden). It appears that there is hardly any trade between the two EFTA subgroups: only 1.5% of the total trade of EFTA is between the Alpine and the Nordic countries. Trade between Austria and Switzerland is not par- ticularly important either. On the whole, the Alpine EFTA countries have 10 times more trade with the EC than with EFTA countries. Intra-Nordic trade is somewhat more important. Still, even the Nordic

428 Victor Norman

Table 1. Composition of EFTA trade, 1986 (% of total EFTA trade)

Exports to from

Nordic EFTA Alpine EFTA EC ROW Total

Nordic EFTA

4.4 0.8

13.2 7.1

25.5

Alpine EFTA

0.7 1.2

17.8 5.5

25.2

EC

13.8 12.6

ROW

8.1 7.6

Total

27.0 22.2

Source: EFTA Trade 1986 (EFTA, Geneva).

EFTA members have about six times more trade with the EC than with each other.

The lack of significant intra-EFTA trade suggests that interaction between EFTA countries can be ignored when looking at the effects of 1992. One can thus concentrate on the bilateral interaction between the EC and each individual EFTA country. The one important excep- tion is Sweden. Of the intra-Nordic trade, the dominant flows are between Sweden and Finland and between Sweden and Norway, the latter being particularly important. Sweden is Norway's largest trading partner, and the Norwegian market is the single most important expert market for Sweden. When looking at the effects of 1992 on Sweden or Norway, the interaction between the two can thus not be ignored. The same is true, but to a lesser extent, of Sweden and Finland.

EFTA exp

EFTA imp

N-EFTA N EFTA exp imp

A-EFTA A-EFTA exp imp

Other manufactured goods Paper and metals

Figure 1. EFTA trade with the EC, 1986 ($b) Source: EFTA Trade 1986 (EFTA. Geneva).

EFTA and the internal European market • 429

The third characteristic concerns intersectoral specialization and trade. The exports of EFTA countries in general, and the Nordic EFTA countries in particular, reflect their comparative advantage in energy and raw materials-intensive production, with very substantial exports of metals and forest products. This is shown in Figure 1. Such specializ- ation could be an important element of the benefits accruing to EFTA following the 1992 programme; indeed, if market integration has its primary direct effect on manufactures other than metal and forest products, we should expect resources within the EC to be reallocated into production of these other manufactures. In turn, more of the EC metals and paper markets will be left to EFTA producers.

3. The approach: industry simulations

3.1. The Smith-Venables model

One objective of this paper is to assess empirically the importance of the various effects that European integration could have on EFTA countries. To address this question, one methodology consists of cali- brating a model to represent particular industries. The effect of integra- tion is then simulated. For the EC countries. Smith and Venables (1988) have carried out a number of such simulation experiments. To make comparable assessments for EFTA countries, my model, presented in Appendix A, will be largely identical to the one they use. In the model, an industry is described by a number of firms, each of which produces various goods (which can be interpreted as brands or models). Each firm has a monopoly on its own brands, but different brands are close substitutes, so the market is one of monopolistic competition with economies of scale in the production of each variety, and economies of scope across varieties. Firms decide on the quantity of each variety that they will produce and on the quantity they want to sell in different markets. In the Smith-Venables study, firms can decide on the number of product varieties and there is free entry and exit. In what follows, I set the number of firms and the number of brands per firm. This enables me to study the effert of entry and exit explicitly. Each firm behaves as if the number of models and quantities produced by other firms were given and all firms sell in all markets. To capture the implications on intra-EFTA trade, I distinguish two EFTA markets, namely Sweden and Norway; the EC is divided into five identical submarkets, and the rest of the world is divided into three identical submarkets. This setup permits the effects of EC integration to be captured without having to model in detail each EC national market. There are national preferences, transport costs and other non-tariff barriers to trade between markets.

430 • Victor Norman

3.2. Two industries

The model describes two industries: motor vehicles (using 1985 data) and pharmaceutical products (using 1981 data). Motor vehicles provide an interesting case study for several reasons. First, at the European level, there is both significant product differentiation and oligopolistic interaction, so that the establishment of an internal EC market could have important effects. Second, because Sweden is a substantial expor- ter, the car market provides a good illustration of the effects of 1992 on EFTA exports to the EC. Third, other EFTA countries are significant net importers from the EC. Accordingly, motor vehicles should provide a good case study regarding the implications for EFTA imports. Fourth, Sweden is a major supplier of motor vehicles to other EFTA countries, and the interaction between EFTA and EC produaion in EFTA markets can thus be studied. Pharmaceuticals were chosen because they provide wholly different insights. First, in this market we observe both more product differentiation and more producers than in the motor vehide industry. Accordingly, oligopolistic interaction should be less important. Second, international trade is very small compared to total production, suggesting that substantial gains can be reaped from integration. Finally, trade is mostly of an intra-industry nature, so that all countries will presumably experience the effect of integration both in the home and export market.

3.3. Calibration

Details of the calibration procedure are given in Appendix A. Only one point should be mentioned here, since it has an important bearing on the results. It concerns initial trade barriers. We typically find high market shares for domestic firms and brands, and low market shares for imports. In the absence of clear differences in production costs, such a pattern can reflect either consumer preferences for home- produced goods, or substantial non-tariff trade barriers. As both expla- nations are equally consistent with the data, an arbitrary choice has to be made. Practically, both trade restrictions and national preferences are captured as a tariff equivalent. This is nothing but the percentage ad valorem tariff which would produce the same effect. Following Smith and Venables, I arbitrarily attribute to non-tariff barriers a 10% tarifF equivalent, the remainder of the difference in market shares being ascribed to consumer preferences. In Section 8, however, I assume that all of the market share differences are due to non-tariff trade barriers. Data describing the initial industry structures are shown in Table 2. The table also gives the tariff equivalents of trade restrictions and

EFTA and the internal European market 431

Tahle 2. Calibration of industry simulations

Sales ($m) Purchases from

Norway Sweden EC ROW

Tariff equivalents For imports from

Norway Sweden EC ROW

Market struaure Suhmarkets Domestic firms Brands per firm

Pharmaceuticals (1981)

Nor

79 44 64 22

Swe

8 225 199 95

of national

— 37 67 76

1 3 1

32

53 64

1 6 2

EC

14 134

20,772 1,815

ROW

13 116

4,528 50,920

preferences and

47 40 46 57

5 78

9

45 38 40 44

3 60 23

1

Nor

0 230 751 355

AotOT vehicles (1985)

Swe

0 3,483 1,413

633 real trade costs 1

— 20 34 42

1 0

—

—

40 47

1 2 2

EC

0 1,464

119,287 10,740

[%)

— 25 18 35

5 18 4

ROW

0 2,466

24,849 285,533

— 21 25 16

3 12 13

Nole: The following values are given to technical coefficients: elasticity of substitution: 5.9 in Pharmaceuticals, 11.1 in motor vehicles; price elasticity: 0.8 for Pharmaceuticals, 1.5 for motor vehicles; ratio of marginal to average costs: 0.78 for Pharmaceuticals, 0.84 for motor vehicles.

national preferences. As is seen, the tariff equivalents are quite high, emphasizing the importance of deciding whether market share differen- ces are due to preferences or trade barriers.

4. Reduction in trade costs

4.1. Single industry efifects

The first simulation experiment deals with a redurtion in real trade costs, like, for example, transport costs or the costs of customs for- malities. We need to have a key to interpret the simulation results. Hence, we shall first trace out the efferts of a reduction in trade costs on a single industry. By so doing, we abstract from some important, but analytically difficult, aspects of inter-market competition. In our context, the reduction of trade costs within the EC will unambiguously lower prices. Indeed, foreign goods become cheaper, their sales and market share increases, the overall supply is raised and accordingly prices fall and the market as a whole is enlarged. Three main efferts actually take place. First, some resources are saved: less is spent on unprodurtive trade costs, so that more resources are available to produce

432 Victor Norman

and buy goods, thereby improving the overall allocation of resources. Second, competition is enhanced as the disadvantage of foreign pro- ducers due to trade costs is lessened. Finally, and conversely, the domestic firms' market share is reduced since part of their initial competitive advantage, due to the trade costs incurred by foreign producers, is eroded.

It is clear from this decomposition of effects that there should be an overall gain, but that it will not be evenly distributed among the various parties involved. The appropriate welfare analysis is presented formally in Box 1. Quite obviously, the domestic consumers are net beneficiaries as they now purchase more goods at a lower price. The foreign pro- ducers also tend to benefit. Indeed, they appropriate part of the reduc- tion in trade costs and they increase their sales and market share at the expense of domestic firms. As a result of the overall market expansion, price will, however, fall. Still, the negative effect of the price reduction falls short of the positive ones. Naturally, domestic firms are losers both because of the fall in prices and because of their loss of market share. It is not obvious, therefore, that the home country benefits from the reduction in real trade costs, as it depends on whether the consumers' gain more than offsets the producers' loss. It is not clear that the world at large benefits either, because total trade costs might increase. Indeed, the increase of the foreign firms' market share means that more goods are supplied by high cost producers given that trade costs are not incurred by domestic firms. This is but one instance where distortions, trade costs in the case at hand, may have negative effects on otherwise welfare enhancing measures.

The general result is that exporting countries are likely to gain from reductions in real trade costs while importing countries may gain or lose, and the world as a whole is likely to gain. A reduction of real trade costs between EC countries should, therefore, be beneficial to all countries involved; each will gain in their export markets, they could lose in their home markets but export gains should outweigh any losses at home, thereby giving each country a share in the total EC gains.

The situation is different for the EFTA countries. The position of EFTA firms selling in EC markets is analogous to the position of home market firms described above: they lose market shares and their profits decline since their marginal cost is unaffected by a reduction in internal EC trade costs. What about EFTA home markets? A reduction in internal EC trade costs need not have any effect in EFTA markets at all. If the two sets of markets are completely segmented, the population of firms is unchanged, and trade costs between EC and EFTA countries remain unchanged, there will be no domestic effect in EFTA countries.

EFTA and the internal European market 433

Box 1. The eflFects of a reduction in real trade costs The effects of a redurtion in real trade costs within the EC are best illustrated for the case of a homogeneous good which is sold in seg- mented markets. Such a market is depicted in Figure 2 where competi- tion is of the Cournot type (firms set output quantities). Two types of firms, foreign and domestic, operate in the market. Domestic suppliers produce at marginal cost bh, and foreign firms at marginal cost bf. Part of the marginal cost of foreign firms consists of real trade costs. For these initial conditions, there is an equilibrium with price p^ total sales x", and sales by domestic firms x?. Next, suppose that real trade costs are reduced, thereby lowering the marginal cost of foreign firms to b]. The effect will be lower market price (/>'), increased total sales (x'), and reduced sales by domestic firms {x\). The welfare implications are straightforward: domestic consumers will gain - consumer surplus increases by {A-^B + C). Domestic firms will lose - their profits fall by

). Foreign firms will most likely gain-their profits rise by —.B). This can be decomposed as follows; G is the reduction

in real trade costs at initial imports, D is made of profits shifted from domestic firms as the result of a stronger market position by foreign firms; and (F —i3) is the profit effect of increased total sales.

The net effect on the home country is {B + C — D — H), which may be positive or negative. The net effect on world welfare is (C + F + G — H). With trade cost reductions which are significant relative to the initial level of trade barriers, however, the net effect on world welfare will be positive.

Demand

xO

Figure 2. The effecu of a reduction in real trade costs

434 Victor Norman

In that case, therefore, there will be no domestic market gain compensat- ing the EFTA firms' loss in EC markets.

In practice, spill-over effects are likely. Markets are not perfectly segmented and the marginal cost of EC sales to EFTA could be affeaed by what happens within the EC. We know, for example, that if border delays are reduced within the EC, transport costs to countries outside the EC will be reduced as well. If EFTA countries remain outside the internal market, such spill-over effects will of course be of a smaller order of magnitude than the effects in EC markets. Moreover, if there are spill-over effects to EFTA markets, there will also be spill-over effects to third markets where EFTA and EC countries compete. The gains that EFTA countries could obtain through spill-over efferts in their domestic markets could, therefore, easily be outweighed by some fall in their exports to third country markets. On the whole, it seems quite likely that EFTA countries - if they remain outside the internal market - will lose from a reduction of trade costs within the EC, at least so long as we ignore demand effects and general equilibrium interactions. If EFTA countries join the internal market, they should benefit.

4.2. Simulation results

As in Smith and Venables, the cost reduction is set at 2.5% of the initial value of intra-EC trade, which is a modest estimate. The simulation assumes that markets remain segmented, and that there are no changes in marginal production costs or in the number of firms. Accordingly, EFTA countries will not benefit from lower trade costs within the EC if they remain outside. The results of the simulation are given in Table 3. (The effects on EC welfare are somewhat larger than those obtained by Smith and Venables - for pharmaceuticals 0.39-0.41 %, compared to their estimate of 0.29%; for motor vehicles 1.39-1.42%, compared to 0.83%. The differences are probably due to market specification.)

Two remarks should be made about the results. First, and not surpris- ingly, Scandinavian participation matters very little to the EC - if Nor- way and Sweden join the internal market, the EC could make an extra gain of 0.02-0.03% of initial consumer expenditure. Second, real trade cost reductions are important to the Scandinavian countries. If they remain outside, lower trade costs within the EC would give them a modest loss. If they join, and we get comparable reductions in the real trade costs of Scandinavian exports and imports, they would reap quite significant benefits; for pharmaceuticals, the gains would be three times as large (in relative terms) as in the EC, whereas for motor vehicles gains would be of the same order of magnitude as in the EC. Table 3 also gives the reduction in real trade costs as a percentage of the total

EFTA and the internal European market 435

Table 3. EfiFects of 2.5% lower real trade costs within the EC

Pharmaceuticals Motor vehicles

Scandinavia outside Produrtion {%) Exports {%) Imports (96) Profits ($in) G>ns. surplus ($m) Welfare ($m) Welfare

(% of expenditure) Trade cost reduaion

(% of welfare gain)

Scandinavia inside Produaion (%) Exports (%) Imports (%) Profits ($m) Cons, surplus ($m) Welfare ($m) Welfare

(96 of expenditure) Trade cost reduction

(96 of welfare gain)

Nor

-0.32 -0.88

0 -0.1

0 -0.1

-0.02

0

0.14 9.44 4.33 -1.0

3.8 2.8

1.25

101

Swe

-0.61 -0.56

0 -0.5

0 -0.5

-0.09

0

2.68 6.02 4.58

0.9 7.3 8.2

1.46

64

EC

0.47 3.98 7.39 -3.5 21.4 17.9

0.39

61

0.49 4.90 7.79 -3.7 22.6 18.9

0.41

61

ROW

-0.08 -0.31

0 -2.4

0 -2.4

-0.01

0

-0.10 -0.62

0 -3.0

0.0 -3.0

-0.02

0

Nor

0 0 0 0 0 0

0

0

0 0

3.31 0.0

30.1 30.1

2.03

91

Swe

-2.68 -2.82

0 -17.5

0 -17.5

-0.30

0

2.91 5.35 9.39

-24.7 75.2 50.4

0.88

57

EC

3.10 8.86

10.51 -17.6 403.0 385.4

1.39

56

3.11 10.10 10.83 -21.6 415.1 393.5

1.42

56

ROW

-0.49 -0.55

0 -43.2

0 -43.2

-0.04

0

-0.55 -0.82

0 -37.4

0.0 -37.4

-0.04

0

welfare gain. For the EC, lower real trade costs account for around 60% of the total gain; the rest correspond to the induced efficiency gain explained in Section 4.1. It is worth noting that the induced efficiency gain to EFTA countries (if they become part of the internal market) might represent a much smaller fraction of their total gain. In the case of Norway and Pharmaceuticals, there is actually a (small) efficiency loss, so the reduction in real trade costs amounts to more than 100% of the welfare gain.

5. Effects of market integration

5.1. Single industry effects

The second potentially important effect stems from the removal of international price discrimination. As before, an analysis of the case of a single industry is presented to serve as a background for the simulation experiment. Emerson et al. (1988) present convincing evidence of substantial price differences between EC countries. The differences between EC and EFTA countries are likely to be of at least the same

436 Victor Norman

order of magnitude. Consequently, a potentially important effect of 1992 is the integration of previously segmented markets. To assess the consequences of market integration, it is important to specify the struc- ture of international price differences. The following analysis explicitly assumes that firms charge higher prices at home than abroad. This assumption is based on the reciprocal-dumping theory of intra-industry trade formulated by Brander and Krugman (1983). This theory suggests that because market shares are larger at home, the demand for their products is less elastic. Consequently, it is in their interest to exploit their domestic market power and discriminate across markets; they will charge higher prices at home than in foreign markets where they face a higher demand elasticity because of their lower shares.

Casual observation tends to confirm the Brander-Spencer hypothesis for many products. Fiat cars are sold in Germany at a price so much lower than that in Italy that significant reimportation to Italy takes place. Norwegian Jordan toothbrushes are sold in Hong Kong at half the price charged in Norway. Finnish Arabia tableware is 10-20% more expensive in Finland than in other Scandinavian countries. According to the producer, the typical sales price (to the producer) for a Norwegian DBS bicycle sold in Norway is NOK 2,000. The average export price is NOK 1,375. A similar price differential exists for Norwegian phar- maceutical products.' Still market share is likely to be only one of many determinants of price sensitivity of demand. For many goods, consumer product perceptions may be more important. The Swedish furniture chain IKEA is an example. In Scandinavia, it is perceived as a mass- consumer, low-price, medium-quality product. In North America it caters to a limited segment of young, design-conscious buyers. As a result, price sensitivity is likely to be higher at home than abroad. It is not surprising, therefore, that IKEA often charges higher prices in France and the US than in Scandinavia.^ Yet IKEA and similar pro- ducers will be regarded as an exception: prices will be assumed to reflect market shares and thus to be higher at home than abroad.

Market integration then has two distinct effects. First, firms reallocate sales from market segments where they initially had to charge a low price (export markets) to market segments where they initially were able to charge a high price (home markets). Second, the market power

I ' Information about the price differentials for Fiat was provided by Luigi Spaventa. The DBS

bicycle figures were taken from an unpublished student paper by Ingunn Lftnning at the University of Oslo; and the information on Norwegian Pharmaceuticals from an unpublished student paper at the Norwegian School of Economics and Business Administration. The casual data on toothbrushes and china is based on own observations.

' Information on IKEA prices was provided by Steinar Vagsud at the Centre for Applied Researd*. Bergen, who is currently engaged in a study of international price differences.

EFTA and the internal European market 437

enjoyed by all firms will be weakened, margins will fall and total output will increase.

The first effect, namely the reallocation of output, is illustrated in Box 2. If as a result of market integration firms must charge the same price on domestic and foreign markets, they will lower prices at home and increase them abroad. Domestic consumers will benefit and their foreign counterparts will lose. The firms are necessarily worse off: they could have charged the same price in both markets before integration but had chosen not to do so. (Under segmentation the loss from their reduced market share abroad is more than offset by the bigher price than tbey can cbarge on tbe domestic market.) Globally, however, tbere is a net welfare improvement. Tbe gain to tbe domestic consumers outweigbs tbe combined losses of foreign consumers and domestic firms, so tbat tbe exporting country benefits at tbe expense of tbe importing country. If botb countries are simultaneously exporting to eacb otber, botb benefit if tbeir exports are of tbe same magnitude. If tbere is a trade imbalance, tbe outcome migbt favour tbe country witb positive net exports.

So far, in order to concentrate on tbe market reallocation effect, it bas been assumed tbat total sales are given. However, market integration will also bave an effect on total sales by eacb firm. Tbe direction of cbange is, bowever, unclear. Tbe ability to discriminate could induce firms to produce more wben tbey face segmented markets tban wben tbey must price uniformly across different countries. A monopolist discriminating between different buyers in a single market would do exartly tbat. On tbe otber band, because tbe market power of any individual seller in a large, integrated market is smaller tban in seg- mented markets, margins sbould be lower in an integrated market, and accordingly total production sbould be bigber. In tbe bencbmark case wbere domestic and foreign demands are equally sensitive to price cbanges, tbe only source of price discrimination must come from differences in market power at bome and abroad: tbe effect of integra- tion will be to reduce tbe market power of tbe dominant seller in eacb market segment, resulting in lower average prices, and larger total sales.

5.2. Simulation results

Tbe results presented in Table 4 describe a simulation wbicb allows for market integration (i.e. price equalization across markets) in addition to reductions in real trade costs. Again, tbe numbers for tbe EC are somewbat different from tbe corresponding estimates of Smitb-Venables; for pbarmaceuticals, I obtain 1.7% compared to 1.1%, wbereas for motor vebides, tbe Smitb-Venables estimate is bigber tban tbe 2.8% gain

438 Victor Norman

Box 2. The effects of market integration

Given that total output amount to x*, a firm has to allocate sales between the home (xO and export (x,) markets. This firm faces a home-market demand curve Dh and an export market demand curve D,. with corre- sponding marginal revenue curves MR^ and MR,. If the home and export markets are fully segmented, the firm will charge a price p^ at home and p, abroad, with corresponding sales given by xf and (x* - xf). If price discrimination is impossible, the same price, p' will be charged in both markets. Domestic sales will amount to x^, the rest being shipped abroad.

Home consumers will gain (A + B) from integration. Foreign con- sumers will lose (D + E). The firm will obtain {E-F) from its export sales; from its home sales it will lose A but gain {C + D + F); the net effect on the firms profits is, therefore, (C + D + £ — >l)-which is necessarily negative (otherwise firms would not have undertaken price discrimination to start with). Summing all of these effects, we see that the world will gain (B + C) from an end to market segmentation. Since foreign consumers lose, however, more than 100% of the gain goes to the exporting country-its total net gain is (B + C + D + E). Of course, if the market is one of intra-industry trade only, and if exports and imports balance, each country will gain on its exports and lose on its imports. Export gains should outweigh import losses and each country should end up with a net gain equivalent to (B + C).

Figure 3. The effects of market integration

EFTA and the internal European market

Table 4. E£Fects of market integration and lower real trade costs

439

Scandinavia outside Production (%) Exports (%) Imports (%) Profits ($m) Cons, surplus (Jm) Welfare ($ni) Welfare

(% of expenditure) Trade cost reduction

(% of welfare gain)

Scandinavia imide Production (96) Exports (%) Imports (%) Profits ($m) Cons, surplus ($m) Welfare ($m) Welfare

(% of expenditure) Trade cost reduction

(% of welfare gain)

Nor

- 3 . 3 4 - 9 . 3 2

0 - 0 . 6

0 - 0 . 6

- 0 . 2 6

0

6.51 -6.61 - 6 . 5 3

- 2 . 6 11.5 8.9

3.97

47

Pharmaceuticals

Swe

-6.41 - 5 . 9 5

0 - 5 . 6

0 - 5 . 6

- 0 . 9 9

0

2.53 -4.22 -7.66

- 9 . 7 22.5 12.8

2.28

68

EC

6.14 -2.97

-16.18 -173.7

253.5 79.8

1.72

39

6.16 -3.27

-16.24 -175.7

255.4 79.7

1.71

40

ROW

- 0 . 8 6 -3.27

0 - 2 5 . 2

0 - 2 5 . 2

- 0 . 1 3

0

- 0 . 9 0 - 4 . 1 2

0 -26.7

0.0 -26.7

- 0 . 1 4

0

Nor

0 0 0 0 0 0

0

0

0 0

3.82 0.0

30.0 30.0

2.02

- 6

Motor

Swe

- 8 . 0 8 -8.51

0.00 - 5 2 . 4

0 - 5 2 . 4

-0.91

0

7.26 -22.69 -47.39 -181.4

528.6 347.1

6.03

32

vehicles

EC

11.78 - 5 . 6 0

-15.47 -662.3 1,446.4

784.2

2.83

66

12.07 - 7 . 0 5

-15.90 -699.1 1,471.2

772.1

2.78

68

ROW

- 1 . 4 8 - 1 . 6 6

0 -129.1

0 -129.1

- 0 . 1 2

0

- 1 . 6 6 -2.51

0 - 1 3 9 . 9

0.0 -155.5

- 0 . 1 3

0

given in Table 4. For motor vehicles, they however assume that there are some economies of scale at the margin, while the estimate in Table 4 assumes constant marginal cost. The results indicate very significant Scandinavian gains from participation in European market integration. For Pharmaceuticals, the net gain from being inside rather than outside (the sum of the gain from participation and the loss from non-participa- tion) amounts to 3.4% for Norway and 4.2% for Sweden. For motor vehicles the net gain is 2.0% for Norway and as high as 6.9% for Sweden.

6. Effects of lower EC production costs

Production costs can be affected through several channels. First, if marginal costs differ across firms, high-cost firms will lose market shares, because of more intense competition. A^ a result, the effective marginal cost of industry output will be lower. Second, larger total output, for a given number of firms, gives longer production runs per firm. If there are some economies of scale at the margin, lower marginal cost will result. This effect could be reinforced by the exit of firms from the industry, an issue directly studied in Sertion 7. Third, increased compe- tition could reduce X-inefficiency. To capture the effects of lower

440

Table 5. Effects of

Scandinavia outside Produaion (%) Exports (%) Imports (%) Profits ($m) Cons, surplus ($in) Welfare ($m) Welfare

(% of expenditure)

Scandinavia inside Production (%) Exports (%) Imports (%) Profits ($m) Cons, surplus ($m) Welfare ($m) Welfare

(% of expenditure)

1.5% EC cost reduction,

No EC cost

Nor

0 0 0 0 0 0

0

0.00 0.00 3.82

0.0 30.0 30.0

2.02

Swe

-8.08 -8.51

0 -52.4

0 -52.4

-0.91

7.26 -22.69 -47.39 -181.4

528.6 347.1

6.03

motor vehicles

t reduction

EC

11.78 - 5 . 6 0

-15.47 - 6 6 2 . 3 1,446.4

784.2

2.83

12.07 - 7 . 0 5

-15.90 -699.1 1,471.2

772.1

2.78

ROW

- 1 . 4 8 - 1 . 6 6 -

0 - 1 2 9 . 1

0 -129.1

- 0 . 1 2

- 1 . 6 6 - 2 . 5 1

0.00 - 1 3 9 . 9

0.0 - 1 3 9 . 9

- 0 . 1 3

1.5%

Nor

0 0

1.50 0

13.3 13.3

0.89

0.00 0.00 5.83

0.0 46.1 46.1

3.10

Victor Norman

EC cost

Swe

-11.12 -12.27

5.02 - 9 7 . 9

40.5 - 5 7 . 3

- 1 . 0 0

4.59 -25.45 -41.86 -215.3

548.2 332.8

5.78

reduaion

EC

17.79 2.55

-14.58 -547.4 1.856.3 1,308.8

4.72

18.03 1.02

-14.92 -587.2 1,882.4 1,295.2

4.67

ROW

-2.86 -3.57

2.07 -376.0

222.9 -153.1

-0.14

-3.01 -4.36

2.07 -390.0

222.9 -167.0

-0.15

production costs, I simulate an exogenous reduction of 1.5% in EC marginal cost. For non-EC countries, no cost reduction is assumed. Since non-EC production goes down or increases by much less than EC output, the assumption of a reduction in EC costs alone seems reason- able. The results, for motor vehicles, are shown in Table 5. One observes that a reduaion in EC costs raises EC welfare by an additional 1.9% of consumer expenditure. The total effect of a reduaion in trade costs, market integration and cost reduaion then adds up to a gain of 4.7%. The corresponding Smith-Venables estimate is 4.1%. In this case, Nor- way would reap an additional gain of roughly 1 % of consumer expen- diture, which originates from a fall in the price of cars in the Norwegian market. For Sweden, the net effect is small. Swedish producers get lower profits, which are almost exactly offset by gains to Swedish con- sumers from cheaper imports.

7. E£Fects through entry and exit

The simulations presented so far predia a considerable reduction in profits. In the integration case, for example, profits fall by some 3% of sales. As a result of this, firms might exit from the industry. Such exit will reduce, but not eliminate, the competitive effea of integration. At the same time, it will give further benefits in terms of lower average cost to EC countries. If EFTA countries remain outside the internal

EFTA and the internal European market

Table 6. E£Fects of market integration, pharmaceuticals ($ni)

441

Profits

Consumer surplus

Entry/exit

None Free

None Free

Norway

Outside

-0.6 0.0

0.0 -0.2

Inside

-2.6 0.0

11.5 7.2

Sweden

Outside

-5.6 0.0

0.0 -2.3

Inside

-9.7 0.0

22.5 14.3

market, the effects will again be ambiguous. EFTA firms will still lose in EC markets. In their home markets, however, they will face fewer EC competitors, and their domestic profits should rise. The overall effect on EFTA profits is, therefore, unclear. With free entry and exit, the effect on the number of EFTA competitors would accordingly be ambiguous. So is the effect on EFTA consumers. Having fewer EC products available entails a loss, but that could be offset by more domestic produas if there is room for new EFTA entrants. Similarly, the degree of competition in domestic markets could rise or fall. Table 6 illustrates the effects of exit. It gives the welfare effects of market integration (reduction in trade costs plus an end to market segmentation, but without cost reductions) for pharmaceuticals, with or without £ree entry and exit. In the no entry/exit case, the number of firms is fixed at the initial level. With free entry and exit, the number of firms adjusts so as to restore zero profits.^ The results are very much as expected: without entry or exit, there will be a substantial loss in profits, which, if the Scandinavian countries become part of the internal market, is more than offset by consumer gains. With free entry and exit, both Scandinavian and EC firms will leave the industry, restoring zero profits. However, the consumer gains from integration become correspondingly smaller; indeed, exit dampens the pro-competitive effea of integration and the consumer surplus is reduced as some brands disappear. The net effect of entry and exit is small.

8. Non-tariff trade barriers and demand composition effects

When there are trade restrictions, in the form of tariffs or non-tariff barriers, second-best effects must also be taken into account: trade restrictions produce a wedge between the domestic price and the import

The integer constraint on the number of firms is ignored.

442 Victor Norman

price of a good. This Vk̂ edge is the difference between the marginal value of an extra unit of imports and its marginal cost to the importing country. An extra unit of imports will, therefore, give a welfare gain equal to the wedge. As a result, we shall not be indifferent between an increase in consumption of imports and an equivalent increase in the consumption of home-produced goods. Second-best effeas of this type need not be considered if European integration involves the complete elimination of trade barriers. That seems unlikely, however.

This second-best effect raises particular problems when it comes to numerical assessment, because the trade impediments are unobservable and generally indistinguishable from (true) national preferences for own goods. An example can illustrate the issue. Casual observation suggests that Swedes buy more Volvos, and fewer Opels, than the Germans. Two interpretations are plausible. One is that Swedes prefer Volvos, while Germans prefer Opels. The other is that there are non- tariff barriers to imports of Opels to Sweden (and to Volvos to Germany). At the same time, it seems difficult to identify precisely the importance of non-tariff barriers. For example, how does one ascertain whether purchases of Volvos as company cars by Swedish corporations are due to true preference or to a feeling that Swedish officials will look more favourably on a company which 'buys Swedish'? Hence, in the absence of accurate information on non-tariff barriers, there is no way we can decide which interpretation is correct. The problem clearly only arises when dornestic and imported goods are imperfect substitutes; if they are homogeneous, any difference in Volvo's market share in domestic and foreign markets must be due to transport costs or artificial trade barriers. Still, if we cannot separate the effect of national preferences from the effert of non-tariff barriers, we are not able to assess the second-best effect of European integration and, therefore, we are left with an important ambiguity.

Table 7 compares the welfare effects of integration in two cases: in the first case, (as in Table 4) market shares differences across countries are explained by consumer preferences in addition to a 10% tariff equivalent representing non-tariff barriers; in the second case, con- sumer preferences are identical and all of the differences in market shares are due to non-tariff barriers. The differences, particularly for Sweden, are striking. To understand why, let us illustrate the effects on Sweden of Scandinavia becoming part of the internal market. If Swedish purchases of Swedish cars reflect a real preference, Swedes will strongly benefit from an integration process which brings down the domestic price of their favourite cars. If consumption patterns are driven instead by non-tariff barriers, the benefit to consumers is much reduced. On the other hand, firms will experience a much more

EFTA and the internal European market 443

Table 7. Consumer preferences versus non-tariif barriers: effect on the motor vehicle industry (change in welfare as % of initial consumer expenditure)

1992 without Scandinavia Discrimination due to real preference Discrimination due to trade bai-riers 1992 with Scandinavia Discrimination due to real preference Discrimination due to trade barriers

Norway

0 0

2.0 1.0

Sweden

-0.9 -0.7

6.0 -0.1

EC

-2.8 -0.6

2.8 0.5

ROW

-0.1 -0.1

-0.1 -O.I

dramatic fall in profits following integration. Indeed, the rents accru- ing to firms are directly proportional to their ability to exploit consumer preferences. Finally, if initial trade patterns reflect non-tariff barriers, someone earns a pure rent from the wedge between consumer and producer prices, and that someone will lose when the wedge is reduced. Put differently, if high home-market shares reflect true preferences, there will be considerable welfare gains to be made from policies which lower the prices charged by firms in their bome market and market integration has exactly tbat effect. If, on tbe otber hand, dominance by home-market firms reflects non-tariff barriers, the important gains come from increased consumption of imports; to reap those, imported goods should become cheaper. Market integration, however, contributes to the opposite. With segmented markets, firms charge lower prices in the markets where they face trade barriers than they do at home. Integration forces them to charge the same price everywhere, so that imports become more expensive relative to home-produced goods.

9. General equilibrium effects

Market integration delivers three types of benefits. The first benefit is the direct saving of trade costs. The second benefit stems from a better allocation of resources within an industry. For example, as international price discrimination is reduced, sales will be reallocated from foreign to domestic buyers, and that could (but need not) be beneficial. Industry restructuring, which lead to fewer but larger firms could be another example, given that with scale economies a larger total output can be produced for a given amount of inputs. The third benefit results from better allocation of resources between industries. In general, imperfectly competitive industries tend to be too small, and highly competitive industries too large, by comparison with an efficient allocation of resour- ces across sectors. The internal EC market should make all industries more competitive, but the impart should be greatest for those industries

444 Victor Norman

which are least competitive initially. As a result, we should expect a shift of resources in favour of less competitive industries.

We have found that a considerable part of the gains in the two industries considered above is associated with increased total output, and thus with reallocation of resources into these industries from the rest of the economy. For the estimation of these gains to be correct, two key assumptions need to be satisfied.^ First, the rest of the economy must be perfectly competitive, so that marginal costs reflect the oppor- tunity value of the extra resources used. Second, the industries concer- ned must be small, so that their expansion does not affect input prices. None of these assumptions is acceptable in the context of the 1992 programme. Indeed, the internal market will have significant effects only if market imperfections are widespread. That violates the implicit assumption that a couple of industries are imperfectly competitive, while the rest of the economy is perfectly competitive. It also means that 1992 will stimulate a large number of industries simultaneously. This will clearly have effects on input prices. If a large number of industries are deregulated simultaneously, income effects could also arise, which would affect the composition of demand and indirectly the allocation of resources.

A complete assessment of 1992 and its effects on EC and non-EC countries thus requires general equilibrium calculations capturing the interaction between simultaneous changes in market structure in all industries. No such calculations have been made. Emerson et al. (1988) have attempted to assess economy-wide effects for the EC; their approach is, however, far from a consistent general equilibrium evaluation.

9.1. A numerical general equilibrium model

To grasp the issues and to get a feeling for potential magnitudes, I have constructed a simple and highly aggregated general equilibrium model for the EC and EFTA. Even though the model is highly sim- plified, it should capture the more important general equilibrium effects of 1992. A systematic comparison of different models with tbe same structure and the same data base as the one used here does, however, indicate that general equilibrium predictions of trade policy effects are highly sensitive to the way in which imperfect competition is captured (see Norman, 1989). In particular, the relative weights assigned to prod- uct differentiation and oligopolistic interaction matter substantially.

* Strialy speaking, a third assumption is required as well, namely that the country pursues optimum trade policies for other industries. That need not concern us here, however.

EFTA and the internal European market 445

This observation should serve as a general warning to treat the numbers in this section with great caution.

The model (presented in Appendix B) has three sectors of produc- tion, namely the production of metals and forest produrts, the produc- tion of other manufactures, ind the produrtion of non-tradeables (mainly services). Competition is perfert in metals and forest products and non-tradeables. Elsewhere, there is product differentiation and quantity competition much as in the partial equilibrium models dis- cussed above. There are two sets of countries, namely the EC and EFTA. Each set consists of five identical countries. This is meant to represent Austria, Switzerland, Finland, Sweden and Norway in EFTA and UK, France, West Germany, Italy, and the remaining countries, for the EC. Markets for non-tradeables are purely national. There is one European market for paper and metals, and (initially) nationally segmented markets for the other manufactures. Quantities traded with the rest of the world are exogenous. There is free and costless trade in paper and metals. For other manufactures there are non-tariff barriers, set arbitrarily (as in the partial equilibrium models) as a tariff equivalent of 10%. All remaining differences in market shares are attributed to real preferences. The model is calibrated to actual 1984 production and trade data for the EC and EFTA (see Table 8).

9.2. Welfare effects: partial versus general equilibrium

I present simulations for the same experiments as those performed in Sections 4 and 5, namely a reduction in real trade costs and full market integration. As before, I look at the polar cases of EFTA remaining outside and joining the internal market. The welfare efferts that I obtain are given in Table 9 and compared to the equivalent partial equilibrium efFerts (i.e. ignoring income effects in demand and general equilibrium effects on marginal cost and production structure). The main conclusion is that general equilibrium interartions significantly dampen the welfare efferts of the internal market but substantial net gains remain. The general equilibrium calculations give estimated gains from full market integration which are around 60% of the corresponding partial equih- brium estimates. In the case of trade cost reductions only, the general equilibrium estimates are 50-55% of the corresponding partial equili- brium estimates. The estimated losses to EFTA countries if they remain outside the internal market are roughly the same. Similarly, general equilibrium interactions do not affect the relative gains to EC and EFTA. The conclusion that EFTA countries have more to gain from European integration than the EC countries is, therefore, still valid.

446 VicTor Norman

s

i n

- I 00

EFTA and the internal European market 447

Table 10. General equilibrium effects on production, trade and prices (% cbange)

2.5% reduction in real trade costs Trade

Metals and paper**' Exports, other manufartures Imports, other manufactures

Production Metals and paper Other manufactures Other goods and services

Prices Other manufactures Other goods and services Faaor inputs

EFTA

EFTA

1.58 -0.36 -0.34

0.10 -0.04

0.00

0.00 -0.05 -0.05

Market integration (incL 2.5% reduction in real Trade

Metals and paper'"^ Exports, other manufaaures Imports, other manufactures

Production Metals and paper Other manufactures Other goods and services

Prices Other manufactures Other goods and services Factor inputs

17.04 -4.14 -3.93

1.09 -0.40

0.04

0.05 -0.52 -0.54

outside

EC

1.84 -0.19 -0.10

-0.05 0.33

-0.01

-0.36 0.02 0.03

trade costs)

19.85 -2.20 -1.19

-1.03 3.55

-0.61

-3.99 0.21 0.52

EFTA

EFTA

0.36 3.91 3.93

-0.02 0.77

-0.05

-1.11 -0.01

0.01

-16.30 -10.28

-8.90

-1.97 4.74

-1.03

-5.52 0.48 1.00

inside

EC

0.44 2.20 1.13

-0.02 0.39

-0.01

-0.42 0.00 0.01

-18.93 -4.98 -2.96

-0.24 3.60

-0.62

-4.44 -0.19

0.12

Note: (a) Change in EFTA net exports and EC net imports.

9.3. General equilibrium e£Fects on production and trade

As mentioned above, general equilibrium calculations ensure con- sistency in welfare assessments, when deregulation occurs simul- taneously in a number of industries. These calculations also indicate the effects of the internal market on the inter-industry pattern of production and trade. Table 10 shows that the inter-industry effects are quite large and highly dependent on how the internal market develops. If European markets remain segmented, and 1992 only invol- ves a reduaion in real trade costs for 'other manufactures', it will generally stimulate trade within Europe. Some of this trade stimulus will spill over into goods (paper and metals in the model) which are not directly affeaed by the 1992 programme. If the EFTA countries remain outside, EFTA trade with the EC will shift from products directly

448 Victor Norman

affected by the single market to some which are not directly affected. This accords with intuition and in particular, EFTA exports of paper and metals will increase.

If European markets are fully integrated, the effeas are numerically larger and perhaps less obvious. Consider first the case where EFTA remains outside the internal market. Table 10 indicates a substantial increase in EFTA production and exports of paper and metals, and a corresponding decline in EFTA trade (exports and imports) in other manufactures. There are two forces at work. First, EC production of other manufactures being stimulated, the demand for factor inputs in the EC increases. Accordingly, the price of factor inputs rises and EC resources are reallocated from paper and metals production into pro- duction of other manufactures. Higher EC input prices also make EC products less competitive in EFTA markets, explaining the reduaion in EFTA imports of other manufactures. Reduced EC production of paper and metals contributes to increased EFTA exports of those goods. The other force is the loss of EFTA market shares in EC markets for other manufactures. The excess resources shift into the production of paper and metals, while input prices decline.

If EFTA becomes part of integrated European markets, production of other manufactures is stimulated both in EFTA and the EC, raising input demand and prices. At the same time, however, firms shift their sales from export to home markets, so that trade in other manufactures declines and resources formerly absorbed by trade (transport and other real trade costs) become available. Some of these resources are used in the EC production of metals and paper, largely eliminating the initial negative impact. This causes a substantial decline in the volume of trade in metals and paper, reinforced by the reduction in demand due to a fall in the relative prices of other goods.

10. Conclusions

The numerical simulation results presented in the paper show that 1992 is important for EFTA countries, not so much because of the threat that an internal EC market represents but rather because of the oppor- tunities that a larger European market offers. Generally, the EFTA countries would not seem to lose much if they decide to stay outside the internal market; the simulations suggest losses of less than 1% of initial consumer expenditures. The losses could be greater for those EFTA countries which export few produas, like Finland with its high concentration in forest products, Norway with oil and metals, or Iceland with fish products. These countries could experience a significant deterioration in their terms of trade because they may be forced into

EFTA and the internal European market 449

even greater reliance on their staple products. Nevertheless, it does not seem that there is a basis for arguing that EFTA countries must partici- pate in the European integration process to maintain their high real income levels. The EFTA countries could, however, gain significantly by becoming part of the internal market. The industry simulations indicate much larger potential gains from European integration for EFTA countries than for the current EC members. The gains to the average EFTA country could be two to three times as large as for the average EC country. In absolute terms, the typical EFTA gain could easily amount to some 2-4% of initial consumer expenditure.

Resource constraints could prevent EFTA countries from simul- taneously reaping such sizeable gains in a large number of industries. General equilibrium interactions lower the estimated average gain by some 30-40%. Nevertheless, a fairly significant increase in real income would accrue to EFTA countries if they participated in a successful internal market. The effects on industrial structure are also worth noting. Should EFTA countries remain outside a fully integrated EC market, they would be forced into greater reliance on their traditional exports of metals, paper products and other semi-manufactured goods. If they participate in European integration, we should expect their industrial growth to be concentrated in other areas of manufacturing.

It should be emphasized that the calculations include only some of the effects that European integration can have on EFTA countries. Three omissions may be particularly important. First, the internal market might have an impact on firms' location. If EFTA countries remain outside the internal market, there could be large-scale exodus of firms. Second, we have assumed that some goods are non-tradeables and remain so. If 1992 successfully creates European markets for previously non-traded goods and services, there would be substantial additional gains. Finally, the industry calculations assume that European countries have already exploited their comparative advantage fully so that European integration will give no further 'classical' gains from trade. The general equilibrium simulations capture some comparative advantage effects. Still, a more detailed analysis of Comparative advan- tage within Europe and its implications for the effects of European integration remains to be done.

Discussion Henrik Horn University of Stockholm

Norman's article assesses the possible consequences of the completion of the internal market for Norway and Sweden, using for the main

450 Victor Norman f^

part the methodology developed by Smith and Venables (1988). One i' particularly interesting aspect is the analysis of the motor industry since fi motor vehicles are a major Swedish export and make up a significant ^ share of private consumption. Another appealing aspect of the paper fi is that it goes some way towards incorporating general equilibrium it effects in the assessment of the internal market programme. Quite <l importantly too, the paper provides a very clear exposition of the issues f> at hand and in particular of partial equilibrium effects. I find that simulation models are ingenious constructs, which provide a useful tool. t However, I do wonder whether these models should be used in the i policy debate regarding the pros and (more seldom) the cons of the « 1992 programme. In particular, I am sceptical about the validity of n the predictions obtained from these models. a

We know from the work of Smith and Venables that the internal K market may lead to a significant increase in the welfare of EC countries : if market segmentation is eliminated. One of the things we learn from 3 Victor Norman s paper is that large gains could accrue to Sweden and 0 Norway, at least for the two industries under review. Still, two conditions î must be met actually to realize these benefits. First, firms currently 1 discriminate in price between national markets to an important extent, ji and second they will stop doing so as a result of the creation of the i internal market. As argued in the paper, there is strong empirical support for the first condition: there appear to be substantial price j differences between different national markets in Europe. Yet, the r analysis is also based on the crucial assumption that discrimination takes t a particular form, namely that producers systematically charge higher ^ prices in their home markets than in their export markets. Norman provides some examples in support of this assumption. More systematic j evidence is, however, called for. In fact, it is relatively easy to find j counter examples even in one of the industries under review; in par- j ticular, automobiles like BMW, Mercedes, SAAB and Volvo often sell at a higher price in the US than in the domestic markets of their producer. At the same time American automobiles are (or at least have been) substantially more expensive in Europe than in the US.

Even if price discrimination does in fact take the particular form assumed in the paper, one can still wonder whether the 1992 pro- gramme will eliminate it. Of course, a number of plausible arguments can be put forward; in particular, it will be easier for consumers and firms to arbitrage, taking advantage of price differentials across national markets. Still, we do not know the extent to which the observed price discrimination is the result of the obstacles to trade that the internal market is supposed to remove. As a result we cannot predict to what extent prices should converge. At the same time, firms have an interest

EFTA and the internal European market 451

in segmenting markets by means of implicit or explicit agreements. If anything, the internal market will also increase the incentives for firms to reach such agreements. In short, the assumption that the 1992 programme will turn segmented national markets into a fully integrated area is questionable. It is natural from an analytical point of view but yields extreme predictions for the real-world consequences of the 1992 programme.

Next, one can also wonder whether the market structure which is assumed in most studies of European integration is an appropriate description of real world competition. For analytical simplicity, these models consider that firms choose quantities, and sometimes prices, once-for-all. This assumption might not be valid for many of the indus- tries which are likely to be affected by the internal market; in those industries a limited number of firms interaa. These firms have some- times co-existed for many years, and it seems reasonable to assume that they expea to co-exist with at least some of their competitors for some time in the future. One should thus expect the one-shot Nash equilibrium to be a poor description of such markets. The repeated character of the interaction should make other outcomes feasible and perhaps more likely. Accordingly, repeated games should be preferred as descriptions of firms' behaviour.

Another alleged source of welfare gain from the 1992 programme is a reduction of 'X-inefficiency' While in the present paper this is only brieffy referred to, it often plays a more prominent role in the policy debate. The concept should, however, be clarified and one should question the existence of the 'inefficiency' in the first place. Indeed, organizational slack, presumably taken as a sign of inefficiency, could be the result of an employment contract that takes into account employees' disutility of effort. Could we then be certain that a product market event that reduces this slack by making people exert more effort is welfare improving?

Finally, it seems that one should also consider how the trade relations and policies between EFTA and the rest of the world would be afferted if EFTA joined the EC. The motor vehicles industry is a good case in point. The EC is more protertionist against Japan than is Sweden when it comes to imports of cars. First, the EC tariff on Japanese cars is 10% while the corresponding Swedish tariff is 6.2%. More importantly, many EC countries employ VERs against Japanese exports whereas Sweden does not. As a result the market shares of the Japanese producers are substantially higher in Sweden than in the EC countries: in 1987 the market share of Japanese producers in terms of newly registered cars was around 10% in EC9 and 21.7% in Sweden. The question then arises as to whether Sweden would have to align its trade policy vis-a-vis the

452 Victor Norman

rest of the world with the EC policy. The answer is probably yes. It is ' unclear, at this point, what such an alignment would entail, given that the future trade policies of Sweden and the EC (despite declarations « by the European Commission), are not known. It is also unclear to what extent Sweden would be affected by, say, a 10% quota for Japanese cars for the whole of the internal market. Yet, it is not unrealistic to envisage a scenario where the internal market results in a reduction of trade barriers between Sweden and the EC, but where Sweden has to conform to more restrictive import policies vis-a-vis South Korea and Japan.

On the whole, one should stress that the numbers coming out of Norman s analysis are extremely imprecise for predictive purposes. It is unfortunate that this observation, which is obvious to economists, is lost on laymen, who often take them as authoritative.

Horst Siebert Kiel Institute of World Economics

Victor Norman s paper presents an interesting analysis of the impact that the single European market will have on EFTA in two polar cases, namely that EFTA joins the EC or remains outside. In studying the effect of integration, Norman focuses on imperfectly competitive markets. He discusses the case of a given firm selling in two markets. This approach to market integration is somewhat narrow. At the same time, it can produce fairly odd results, such as the conclusion that imports become more expensive relative to home produced goods. One can think of two other important effects of market integration which are not analysed in the paper, namely the exploitation of comparative advantage and locational arbitrage by firms and capital. Indeed, the single market will provide more scope for exploiting comparative advan- tage, with, for instance, relatively labour-abundant countries specializ- ing on labour-intensive production (see Haaland and Norman, 1987). In addition, the movement of capital and the relocation of firms will reduce factor price differentials among countries much more quickly than trade specialization. Capital movements and the relocation of firms respond to comparative advantage so that trade and location are sub- stitute mechanisms of arbitrage in space.

The dynamic effeas of factor movements should also be considered. Assume that a foreign country has a large endowment of some factor of production which is relatively immobile. For instance, one might think of a large endowment of unskilled labour, resulting in low wages.* This country may be able to attract new capital and new technology as they flow more freely across the Community. As a result, the overall cost position of this country might be improved. Hence, with the mobility

EFTA and the internal European market 453

of some factors of production, competitiveness of a region for its export commodity may be improved. The efficiency gain might be larger than suggested by a static analysis.

The relocation of firms and capital outflows may be particularly relevant for the EFTA countries if they stay outside. EFTA countries might thus forgo an improvement in competitiveness. At the same time, EFTA specializes in products which are intensive in skilled labour, and it might be affeaed by the product cycle; some of the technologies used in manufacturing these products, once standardized, will be applied by lower wage countries. In response to cost differentials, and market segmentation, EFTA firms might thus like to locate inside the EC. EFTA countries would then lose out.

With respect to the partial equilibrium simulation undertaken for Pharmaceuticals and motor vehicles, I wonder whether the effects of lower trade costs, market integration and lower production costs can actually be added up to obtain a complete estimate of the benefits. The author provides no guidance on this matter.

Next, it is interesting to note that in a general equilibrium context welfare efferts are lower by comparison to what partial equilibrium analysis suggests. The reason is of course that not all industries can expand at the same time. The expansion of an industry raises factor prices, thereby reducing benefits in other sectors. The expansion of some industries with economies of scale also implies an appreciation of the currency, thereby deteriorating the competitiveness of other sertors. As an extreme case, if all sectors improved their competitive position by the same percentage, only the exchange rate would change. Similarly, considering capital outflow out of the EFTA might significantly affect the empirical results.

The effect of market integration on capital flows and competitiveness, should thus be very significant for overall welfare estimates. The main thrust of the paper is that EFTA may not* lose too much by not joining relative to the starting position, hut could gain significantly by joining. This assertion can be backed by looking at the development of the last 20 years. Indeed, we observe that EFTA countries have not achieved a degree of integration with EC countries comparable to the degree achieved by EC countries between themselves. Figures 4 and 5 show the development of trade since 1970 between the 12 EC countries and the 6 EFTA countries. Trade has been influenced by several waves of integration. First, in 1977, a free trade area was established between EFTA and EC countries. This is basically a free trade area in manufac- tures dealing with tariffs and excluding other forms of segmentation such as regulated entry conditions. Second, the EC was successively enlarged from six to nine (UK, Denmark and Ireland, 1973), 10 (Greece

454 Victor Norman

30

20

10

exports

imports ,

it

1970 1975 1980 1985

Figure 4. EFTA's share in EC exports to (imports from) EC and EFTA members

1981) and then 12 (Portugal and Spain, 1986) members. Figure 4 suggests that the EC's exports to the six EFTA countries have grown less rapidly than intra-EC exports and exports to EFTA taken together. Hence, it seems that EFTA has not been integrated with respect to the EC's exports to the same extent as EC members with respect to total EC and EFTA exports. As for EC's imports from EFTA, the proportion has not changed. EFTA has not lost competitiveness for its exports. One also observes (Figure 5) that EC exports to EFTA have declined relative to total EC exports. Judging from the absorption of EC exports

70

60

50

10

M

intra-EC exports in total EC exports - ^

V - - ' intra-EC imports in total EC imports

40

30

20

EFTA's share in total EC exporu

EFTA's share in total EC imports

1970 1975 1980 1985

Figure 5. EFTA's share in toUl EC exports (imports) and share of intra-EC exports (imports) in toUl EC exporU (imports).

EFTA and the internal European market 455

by EFTA, we see that integration was, therefore, reduced. EC imports from EFTA in proportion to total imports rose slightly^ so that by this measure integration has somewhat increased. However, intra-EC trade (exports went up from 53.36 to 58.66% and imports from 50.29 to 58.78%) clearly has increased more, so that the EC countries have integrated between themselves more than with EFTA. One may thus conclude that the establishment of a free trade area between the EC and the EFTA in 1977 has not been as successful in fostering trade between these areas as a full integration could have been. This outcome may be due to the fact that, in spite of the abolition of tariff barriers various non-tariff barriers to trade such as import quotas, subsidies or national regulation still exist between the EC and the EFTA.

Norman analyses the polar cases of EFTA joining or remaining out of the EC. However, potential forms of cooperation and integration between the EC and EFTA range from full-fledged membership to loose forms of consultation. Cooperation can also take place with EFTA as a whole or with specific EFTA members. In fact, a full-fledged membership of EFTA seems unrealistic because EFTA countries are not unanimously willing to allow the free movement of people, com- modities, services and capital. For example, Switzerland is not inclined to change its agricultural and immigration policies. Sweden uses its exchange rates to make real wages flexible, which would not be possible in the EMS. Both countries would also like to continue their status of neutrality. Consequently, some mechanism of consultation may be found in which information on integration measures will be exchanged. Such a process, even making consultation compulsory, will still not affert the legislative and executive implementation of the internal market programme.

The paper does not deal with the process through which Europe is becoming integrated. This process is important for EFTA countries. Indeed, the European Commission has found that ex ante harmonization is widely impractical. As a result, integration occurs through competition between national institutional arrangements triggered by the principle of mutual recognition. This principle has been established by the verdia of the European Court on the famous case of the cassis cie Dijon. In other words, harmonization occurs ex post. Competition among institu- tional arrangements is only possible if households and consumers have the option of spatial arbitrage. Relocation of firms forces the political

J Interestingly, EFTA's share in total EC imports in manufaauring has declined slightly since 1970. This outcome is all the more surprising as tariff barriers in manufaauring between the EC and EFTA were abolished in 1977.

456 Victor Norman

process to harmonize the institutional arrangements. One may conjee- ' ture that EFTA members will take different attitudes with respea to institutional competition between national regulations. Thus, besides ' full-fledged membership of one or two countries, we can expect that ' specific countries will adopt those EC rules where they have a particular stake. The following areas can be distinguished. First, it seems that the * border controls necessary to administer the differences in the tax system (value added tax) are likely to continue (except maybe in Austria and Norway). Second, technical standards, when harmonized in the EC, ' may be easily adopted by the EFTA countries. Indeed, EFTA countries ' export capital-intensive products where norms are particularly relevant * and, therefore, have a particular stake in this matter. Next, the removal of entry barriers in the form of regulation in specific industries (electricity, banking and insurance) might be a matter of reciprocal negotiation between the EC and EFTA. For instance, creating new property rights with respect to transmission lines for electricity may require common action of all countries involved. But again, a country may try to establish a niche, for instance in the banking industry (Switzerland).

Finally, transportation infrastructure will presumably be affected in two ways. EFTA countries in the periphery will have an interest to improve their links to the heart of the EC area. Here the competitive process operates. For the Alps, representing a barrier to North-South trade, a bargaining solution will have to be found in which the EC's demand for improved traffic flow and the opportunity costs to the country involved have to be balanced.

General discussion

Can price differences across countries for a given product be taken as a good indicator of the lack of market integration? Or equivalently, can we anticipate that market integration will bring about a high degree of price convergence? This question was debated by various panel mem- bers. The matter is important because Victor Norman's estimates of the welfare gains from integration is very sensitive to the degree of price convergence that is assumed. Richard Baldwin first indicated that in the US there were substantial price differences across states and sometimes across nearby cities. He concluded that the large price differences observed in Europe today would possibly remain even after the completion of the internal market programme. Norman argued, however, that price differences in the US had been documented by the EC and turned out to be much lower than in Europe, where accordingly there is a scope for price convergence. In addition, price differences

EFTA and the internal European market 457

in the US occur between different economic units and not between administrative units, as in Europe. This suggests that price differences in Europe indeed stem from a lack of market integration and could, therefore, narrow down. Jacques Melitz also pointed out that flexible (or even quasi-fixed) exchange rates between European currencies would always maintain some price disparities.

Norman's analysis stresses that the welfare gains from market integra- tion and price convergence crucially hinge on whether these price differences are due to a disparity of preferences or to the existence of non-tariff barriers. However, according to Anthony Venables, the dis- tinction between preferences and non-tariff barriers is not fundamental; what matters is whether or not barriers involve real resource costs. Indeed, a difference in demand has the same effect as a non-tariff barrier involving a real cost. In turn, these will differ from barriers which simply generate pure rents. Still, the problem remains of assessing the respective role of both types of barriers in explaining the observed price differences. Venables argued that the main entry barrier in foreign markets stems from the sunk cost of setting up distribution networks. Since real resources are used in setting up distribution networks, bar- riers involving resources should a priori be considered as relatively more important. Jorge Braga de Macedo added that one should be careful about the endogeneity of the implicit barriers stemming from differences in demand; he thought that as Europe becomes more integrated, tastes and cultures will also converge, thereby reducing demand differences.

Damien Neven wondered about the choice of the pharmaceutical industry, as an interesting benchmark to estimate the benefits of integration; indeed, prices in this industry are subject to government control and arbitrage is inherently difficult because a third party is usually paying for the products. Norman argued that arbitrage could still take at the wholesale, rather than retail, level where this latter difficulty does not appear. Stephen Breyer added that certification procedures would still be a significant impediment to competition.

With respect to the general equilibrium analysis, Baldwin indicated that the estimates of the welfare gains could be significantly biased downwards; indeed, the model assumes that the supply of factors is fixed and that factors are fully utilized. In this context, an increase in the output of one industry happens entirely at the expense of output in another industry. Consequently, the general equilibrium estimates of the welfare gains tend to be very much lower than what a partial equilibrium suggests. Still, most European countries experience high

458 Victor Norman

unemployment rates and the labour supply could be further expanded. As a result, outputs in different industries are not direct substitutes and the total welfare effect of integration is likely to be higher than what the model suggests.

Jean-Pierre Danthine also reflected on the magnitude of the welfare effects in general; he thought that the estimates of the gains were very small, at least by comparison with the perception that the general public holds about them. The fact that the general public seems to be misled was, in his opinion, worrying because companies are today taking strategic decisions on the basis of wrong perceptions. Norman concur- red, indicating that according to opinion surveys, a large majority of businessmen seems to expect increased profits from the 1992 pro- gramme. Surely, the marketplace will impose some consistency in their views and lead to a more sober approach.

Appendix A. Structure and calibration of the industry models

The partial equilibrium model used is similar to the Smith-Venables (1988) model, so only a sketch will be given here.

A l . Equilibrium in each market segment

Demand in a market segment is derived from an expenditure funaion defined over the price index for the industry, p, and the consumer utility level u.

' + u (Al)

where the price index is a CES aggregate of individual product prices

(A2)

where n, is the number of country-j producers selling in the market, TWy is the number of brands per firm from country ;, and />, is the price charged by country-j producers. The elasticity of substitution is a.

Total industry demand will be

(A3)

and the demand for a country-; product variety becomes

(A4)

EFTA and the internal European market 459

A country-j firm has a cost function

bi = b'{Xj,mj) (A5)

displaying both economies of scale and economies of scope, i.e. (letting subscripts denote partial derivatives) with

In the simulations, the cost function is assumed to be linear both in m and X.

A country-y firm faces trade barriers in the form of ad valorem tariff equivalents t,. Thus, its profits will be

'ixj, mj) (A6)

For given m,, the firm chooses x, so as to maximize (A6), taking the number of models and quantities produced by other firms as given, giving the condition that marginal revenue equal marginal cost. If we let 5, denote the market share of firm j (i.e. the share that all the models of firm j constitute in the industry aggregate), we can write the condition as

O - - 1

We have a Cournot market equilibrium for a given number of models when (A7) is satisfied for all firms.

A2. Market segments; simultaneous equilibrium in all segments

There are H market segments, each described by equations like (Al)- (A7). Letting x* denote sales by a firm from country j in market h, and using corresponding notation for other variables, the full set of equi- librium conditions can then be written as:

c* = (l-e'')A''(^'')-^* (A8)

(A9) J

(AlO)

460 Victor Norman

Equations (A8)-(A 11) give us H{2n + 2) equations to determine prices and outputs from each firm in each market, and aggregate sales and the industry price index for each market.

A3. Calibration

For Pharmaceuticals, production data were taken from the official statistics for Norway, Sweden, the EC and the US (the latter as a proxy for the rest of the world; ROW production being assumed proportional to US production), while trade data were taken from UN trade statistics. For motor vehicles, Norwegian and Swedish data were taken from official statistics, while GATT data were used for the EC and the rest of the world. Data on market structure in the EC (Herfindahl indexes) were provided by Smith and Venables; as were the price and substitution elasticities used. Market structure data for Sweden and Norway are guesstimates, based on different sources.

Calibration involves solving a fairly complex set of simultaneous equations. In the description given here, the simultaneity aspect is ignored and the process is described as a sequence of steps. The reader, therefore, should not take the description literally; the purpose is only to convey the important issues involved in calibrating the model.

Data on the number of firms can be obtained from Herfindahl indexes of industry concentration - if all national firms were identical, the Herfindahl index would simply be the inverse of the number of firms; so the inverse of the Herfindahl index can be used as a proxy for the number of domestic producers. From production and trade statistics we have data on the value, at producer prices, of sales by different countries in different markets. Combining the two, we get data on sales per firm in different markets. To proceed to data on sales per firm per model, we need to know the number of models produced by each firm. Model data do not exist, but we can normalize by setting the number of models produced per firm in a particular country equal to one, and then proceed by assuming that total output per model is the same for firms in all countries. In this way, we can arrive at figures for sales per model for a firm from country j in market h; i.e. the revenue per model that firm ; receives in market h.

The other essential piece of information we need is marginal cost. Again, that is generally not known. We know total revenue and total output per firm per model, however; and thus average revenue. If we assume zero profits, we then know average cost (= average revenue). Using empirical estimates of economies of scale in the industry, we can then compute marginal cost.

The industry price elasticity, e* is assumed to be the same in all markets, and is taken from literature studies; market shares are known

EFTA and the internal European market 461

from production and trade statistics; so (Al 1) can be used to compute the producer prices il-t^)p^ consistent with marginal revenue = marginal cost, as a function of the elasticity of substitution, a''. But we can, by appropriate choice of units, set the average producer price (across all firms and markets) equal to one; assuming the elasticity of substitution to be the same in all markets, we can then find the common value of o-*. All producer prices are then found from (All).

Consumer prices could be found directly from the demand functions (AlO), if we knew the preference parameters a*: from (AlO) we could find relative consunjer prices (p^/p'l) from relative demands (xy/x^) and the ratios of preference parameters (oy/af); with no trade costs for home-produced goods the consumer price of the home-produced prod- uct varieties would equal the producer price, so one consumer price in each market would already be known, fixing the levels of all prices.

Unfortunately, we do not know a'- All we can obtain from (AlO), therefore, are the products a'- {p'ly^

Thus, all we know are the producer prices, (1 -t'-)p'l. and oj{p'j)~'̂ From these we can find (1 - t'j)(a'-y''^ We also know that there are no trade costs for home-produced varieties, so t^ — 0. By appropriate choice of units in the industry demand functions (A8) we can always set a* = 1 for one set of products; it is natural to do so for the home-produced varieties. If we do that, we can interpret (I — t'Dia'-)""^ as a measure of 'effective discrimination against goods of origin ; in market h: with no trade costs and no preference for national goods, tj would be zero and ay one, so (1 —<y)(oy)'''' would be one. The higher the trade costs, or the lower the preference for good i relative to the home good, the lower will (1 - ty )(ay )'^" be. Generally, therefore, [1 - (1 - ty )(ay)''"] can be used to measure the tariff equivalent of trade costs and national preferences-it says how much lower the producer price of a country-) product must be if it is to sell the same quantity as a home-produced product variety. These tariff equivalents are all that can be determined through calibration.

Appendix B. Structure of the general equilibrium model

Each country has a production possibilities frontier defined over total outputs from the three industries and production of transport services. Demand at the industry level is generated from a Cobb-Douglas utility function defined over the three sets of outputs, while demand for individual products in the imperfectly competitive industry reflects a constant elasticity of substitution between product varieties. The model is closed by the requirement that demand expenditure in each country

462 Victor Norman

equal the value of production, less exogenous trade surpluses (included to calibrate the model to data).

The general equilibrium model differs from standard CGE models only in assuming imperfect competition in some of the production sertors. The structure of the model is very similar to a family of such models discussed in Norman (1989).

Bl. Preferences and technology

Let c? denote consumption of good i in country Ji, ^t its (consumer) price, and M* the (one-consumer) utility level of country h. Demand in each country is then generated from a Cobb-Douglas expenditure function

(y' (Bl) with (compensated) demand functions c? = (de'^/dp'l). Uncompensated demands are found from the requirement that expenditure equal income, 3)̂

«" = / (B2)

Goods 1 (paper and metals) and 3 (non-tradeables) are homogeneous. Good 2 (other manufactures) is an aggregate of differentiated products, similar to the industry aggregate in the partial equilibrium model. Thus, product differentiation is captured by making the price of good 2 a CES price index defined over the prices of different product varieties; dual to this is a proper quantity aggregate of consumption of the differentiated products.

The production technology is one of constant returns - apart from fixed costs in the production of good 2 - so each firm faces constant marginal cost. Through implicit, general-equilibrium effects on relative factor prices, however, the industry marginal cost curves are upward- sloping. This is captured through a CET production frontier:

IiA,(xo+/o)"^u,-t;,, (B3) i

where Xj, is output of good i in country ;, fij are any fixed input requirements associated with the production of good t in country ;, and (y — 1) is a measure of how convex the production frontier is; i.e. how sensitive the opportunity cost of good i is with respect to the quantity produced of the good. On the right-hand side, Vj can be interpreted as a fixed resource constraint in the country; while v^ denotes the resources that go into transportation or other sources of

EFTA and the internal European market 463

international trade costs. The production frontier gives general equi- librium cost functions for good i in country ;':

bii = iWi/y)Aii{Xii+fii)^ (B4)

where Wj is the resource shadow price in country j .

B2. Product markets

Goods 1 and 3 are produced in perfectly competitive sectors. There is free, costless trade in good 1, and European markets for that good are fully integrated. It is used as numeraire. There is no trade in good 3. Good 2 is produced in oligopolistic industries, identical to the one described in the partial equilibrium model; markets for good 2 are (initially) segmented; and there are real trade costs associated with it. Trade costs take the form of ad valorem tariff equivalents t'lj that have to be paid by country-; producers of good i selling in market h.

Thus, product market equilibrium conditions for sertors 1 and 3 consist of straightforward market clearing conditions and conditions that marginal cost equal price; while the product market conditions for good 2 are identical to those in the partial equilibrium model (except for an added subscript denoting good 2). The exact conditions need not be spelled out.

The trade costs incurred by country-;' producers will be:

y^^ (B5)

B3. Factor markets and income definitions

Factor market equilibrium obtains when (B3) holds as an equality i.e. for that Wj which gives:

^ ^ . = ̂ i (B6)

where

v, = ̂ (B7)

are the input requirements associated with the trade costs 6,,. The model is closed by the income identity

+ profits^ (B8)

•r 464 Victor Norman ''

B4. Calibration .,

EC data were taken from the official EC statistics, while EFTA data were constructed on the basis of production and employment statistics « for the Nordic countries, employment statistics for Austria and Switzer- ^ land and EFTA trade statistics. We have data on values of consumption and production, and thus on values of factor inputs. The number of firms in a representative good-2 industry is set exogenously, as are tariff equivalents of real trade costs. Choosing units so as to give initial marginal costs of 1, we obtain consumer prices from marginal costs and tariff equivalents. Knowing marginal costs and consumer prices, we obtain quantities produced and consumed, and quantities employed of factor inputs. What remain are demand function parameters (the distributional shares a? in the Cobb-Douglas expenditure function, and the substitution elasticity a-^ and the preference parameters Og, in the CES sub-utility function for good 2); and the parameters of the produc- tion frontier (y, and Aij).

Of these, the substitution and transformation elasticities (T\ and % are set exogenously. The value of yj is set arbitrarily at 1.5, which in the competitive case gives a price elasticity of supply of 2. The good-2 elasticity of substitution is set at 4.

The Cobb-Douglas shares are obtained directly from income and expenditures. The CES preference parameters, given exogenously specified tariff equivalents, follow from the computed quantities and prices, in the same way as in the partial-equilibrium model. The produc- tion frontier parameters Aij can be found from quantities and prices, using the definition of marginal cost.

When calibrating the model to actual production and trade data, third-country trade must be accounted for as well. That is done in the simplest possible manner, by assuming exogenous trade flows, at fixed (world market) prices between third countries and EFTA and the EC. Exogenously given trade surpluses are then also included in the income- expenditure equations.

References

Brander, J. A. and P. R. Krugman (1983). 'A reciprocal dumping model of international trade', Journal of Intemalional Econimics.

Emerson, M., with M. Aujean, M. Catinat, P. Goybet and A. Jacquemin (1988). 'The Economics of 1992', European Economy.

Flam. H. and H. Horn (1989). 'Ekonomiska konsek marknad", in Svemk ^kmomi oth £uro^inl- egrationen, Stockholm.

Haaland, J. I. and V. Norman (1987). 'EFTA and the world economy', EFTA Occasional paper. No. 19.

EFTA and the internal European market 465

Helpman, E. and P. R. Krugman (1989). Trade policy and market structure, MIT Press. Krugman, P. R. (1988). 'EFTA and 1992', EFTA Occasional Papers, No. 23. Lundberg, L. (1989). 'Svenskt naringsliv och den europeisk integratione', Svensk ekonomi och

Europaintegrationen, Stockholm. Norman, V. (1989). 'Assessing the trade and welfare effeas of trade liberalization', mimeo (paper

prepared for the 1989 International Seminar on Macroeconomics). Smith, A. and A. J. Venables (1988). 'Completing the internal market in the European Community',

European Economic Review.