- ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~0 POLICY RESEARCH WORKING PAPER 1293 Product Standards, Modeling the static and steady-state effects on trade, Imperfect Competition, and production, and market _M * ' t * s F 1 ~~~~~~~~~~~~structure of completion of the Completion of the Market Mt,eo opeino h European Union's internal in the European Union market. Glenn Harrison Thomas Rutherford David Tarr The World Bank International Economics Department International Trade Division April 1994 POLICY RESEA-.CH WORKING PAPER 1293 Summary findings Harrison, Rutherford, and Tarr model the static and country differences because of the greater countrv steady-state effects on trade, production, and market disaggregation. structure of completion of the European Union's (EU's) The additional effect of the program of standards en internal market. consumer demand elasticities increases the competition The impetus for change comes from the removal of and reduces markups in imperfectly competitive border costs and the costs of producing to different industries. Then there are additional gains from national standards. It also comes from consumers' rationalization, as well as consumer efficiency gains in greater ability to substitute among the products of imperfectly competitive secto.s, that result in an increase producers in different EU countries, once the European in the estimated gains to about 1.2 percent of GDP Union adopts its program on standards. (again with wide differences across EU countries). In the analysis of the static scenario, removing border The steady-state results let the capital stock in eacn costs and the costs of supply-side standards improves the country adjust to its new higher equilibrium value, which welfare of ElI countries by only about 0.5 percent of acts as an additional endowment of capital, allowing the GDP. Results vary greatly across the countries of the Eurovean Union to produce a higher level of income. European Union, however, because the benefits to a The gains to the European Union then rise to about 2.6 country are roughly proportional to its share of intra-EU percent of GDP. trade in its GDP. This is the first model to identify these This paper - a product of the International Trade Division International Economics Department - is part of a larger efforr in the department to assess the impact of changes in the global trading environment on developing countries. The study was funded by the Bank's Research Support Budget under the research project "The Impac. of EC 1992 and Trade Integration in Selected Mediterranean Countries" (RPO 675-64). Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Nellie Artis, room N10-037, extension 38010 (42 pages, plus 46 pages of appendices). April 1994. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' oun and should not be attributed to the World Bank, its Executive Board of Directors, c any of its member countries. Produced by the Policy Research Dissemination Center Product Standards, Imperfect Competition and Completion of the Market in the European Union by Glenn W. Harrison, Thomas F. Rutherford and David G. Tarr' Dewey H. Johnson Professor of Economics, Department of Economics, College of Business Administration, University of South Carolina; Assistant Professor, Department of Economics, University of Colorado; and Principal Economist, The World Bank. The authors would like to thank acknowledge he4iful comments from Alan Winters, John Whalley and participants ef seminars held at: the EC Commission;, the U.S. International Trade Commission; the World Bank; the 4th Annual Computable General Equilibrium Modeling Conference, Waterloo Ontario; and Georgetown University. This study is part of the World Bank's research on "The Impact of EC 1992 and Trade Integration in Selected Mediterranean Countries", supported under grant RPO #67564. Table of Contents 1. Introduction .................................................... 1 2. A Multi-Regional Trade Model ...................... S General Model Features ......................................... 5 Modelling the Reduction in Border and Standards Costs of Intra-EC Trade .... ..... 8 Theory and Evidence Against Uniform Pricing ........................... 9 Modelling the Price Effects of Integration ......... ................... 11 Impact of Standards on Substitution ............................. 11 Modelling the Impact of Increased Substitution. 15 3. The Static Effects of Completion of the Market .20 Static Welfare Effects .21 CRTS Results and Comparison with IRTS .21 The Impact of IRTS .27 Intra-EC Trade .33 Comparing the Static Effects ..34 4. The Steady-Stats Effects of Completion of the Market ..37 Modelling the Steady State Welfare Effects .37 Steady-State Welfare Effects ...................................... 39 Explaining the Steady State Effects ........... ................ 39 5. Conclusion .... ....................... 40 References ................................. 40 1. Introduction How large are the static welfare benefits of completion of the internal market in the European Community? What are the economic mechanisms underlying these estimates? Are the long-run estimates significantly different from the short-run effects? The size of existing estimates, and their rationale, vary a great deal. First consider the static welfare effects. The official forecasts were that the w:fare gains would be between 4.3% and 6.4% of 1988 GDP, albeit based on heroic extrapolations of partial equilibrium calculations for a handful of industries.' Academic numerical general equilibrium calculations2 tend to be much lower, ranging from 0.25% to 1.35% of GDP. These studies produce a range of estimates varying, most importantly, with the nature of the assumed market structure and pricing policy. When countries within the EC are assumed to have segmented markets, lower estimates are derived (about 0.25 to 0.50 percent of GDP). In this segmented markets case the benefits of EC integration derive from the removal of border costs and the reduction in costs from producing to a common standard. A commonly held view, however, is that EC92 will have its strongest impact in imperfectly competitive markets by increasing competition and inducing rationalization. The higher estimates derive from models which assume that completion of the market will induce a single integrated market in which international price discrimination is impossible, i.e., a uniform EC-wide price is imposed. These larger estimates are sometimes characterized as hypothetical because no clear rationale is articulated for how the integration process will induce a common price. That is, if arbitrage fails to eliminate price differences (net of transportation, border and standards costs) prior to 1993,' it is not I This range is reported in Cecchini et al. [1988; p.83] and Emerson et al. [1988; Table 10.1.1, p.203], which are extrapolations of the work of Smith and Venables [1988] for a number of sectors which experience incresing returns to scale. Winters [1992; p. 105] has appropriately characterized these extrapolations as 'heroic'. I Due to Gasiorek, Smith and Venables [19921, Haaland and Norman [19921 and Mercenier [1992], all of which are summarized below. I We use the date "1993' to refer to the nominal year in which completion of the market is to begin. The delys in national legislative efforts to implement necessary statutes, as well as other delays in application of those law, makes this date a matter of pedagogic convenience rather than a prediction about when the EC will in fact have completed the single-market program. clear why it would eliminate price differences (net of transportation costs) after 1993.' We briefly review the empirical work testing for price discrimination in the United States. It shows that even in the United States, which is more fully integrated than the EC is likely to become in the near future, price discrimination appears to exist in a wide variety of n,arkets. Moreover, a number of monopolistic competition models have recently been developed that indicate that price discriminatikn may exist even when market power arises only from monopolisticaliy competitive product differentiation. Thus, both theory and empirical work orn the relatively integrated markets in the United States indicate the implausibility of the modelling exercises that assume that no price discrimination is possible after the implementation of the EC92. In addition, we note that the imposition of uniform pricing results (due to the elimination of reciprocal dumping implied by uniform pricing) in the counterintuitive result that intra- EC trade declines subsequent to the EC92 program.5 Thus, we believe that it is necessary to model market integration within the EC92 program with an approach that captures the idea that EC92 will increase competition and rationalization in imperfectly competitive markets without imposing complete elimination of firm level price discrimination across the EC. We model the effects of completion of the internal market in the EC on trade, production and market structure. The impetus for change in our framework comes from two interrelated effects. One effect is the, now standard, reduction of border and standards costs (analogous to the segmented markets approach of previous studies, where the benefits of reduced standards costs is limited to the impact on improved economies of scale). In addition, we maintain that the single market program on standards will have the impact of increasing the ability and willingness of buyers to substitute among the products of 'Karp [1992; p.63] makes the same point: 'If it were possible to price discriminate before 1992, then lowering transportation costs (broadly defined) might either increase or decrease the incentive for price discrimination.' Smith and Venables [ 1988; p. 1523] note that '... it is not obvious that there exist feasible changes in EC trade policy and competition policy that could impose such a change. In practice policy may be expected to be some combination of our two experiments.' 5 See, for example, Smith and Venables [1988; pp. 1520-1523]. - 2- producers in different countries. We review the theory and evidence regarding the impact of standards which shows that the impact of standards, especially in the manner implemented in the EC92 program, sould be to increase the substitution possibilities of EC buyers for the products of EC producers. In many cases, such as common interconnection in telephone equipment, it is because the products are physically altered so that the products of different producers become interchangeable. In other cases it is because protectior'ist barriers are eliminated, as occurred with German beer purity restrictions. For other products a quality assurance standard may reduce the advantages of product differentiation, such as occurred in the United States as a result of SAE standards for motor oil and brake fluid. These effects are appropriately modelled as a change in the elasticity of substitution of consumers in EC countries for the varieties of output of other EC firms. We simulate this change along a continuum from an initial situation, characterized by firm level product differentiation where consumers regard the output of other EC firms are equally substitutable with other non-EC imports, to where consumers regard the varieties of all other EC firms as equally substitutable with home varieties. A monopolistic competition model is developed in which each firm's post-1992 markup is endogenously determined as a function of the substitutability of the outputs of different EC firms in the preference structure of national consumers. We show, however, that even in the extreme case of no national preferences, pricing differences and market segmentation are not eliminated. As long as arbitrage is imperfect, price differences (net of transportation costs) may persist, and firms determine their optimal markups in each national market based on perceived elasticities that vary with (among other things) the firm's market share in each national market.6 Thus, an important contribution of this paper is that we provide a rationale for 6 Haaland and Wooton L1992] have shown that if trade barriers or national preferences continue to exist in the post-1992 environment, the uniform EC-wide pricing that results from complete elimination of segmentation can, contrary to the conventional wisdom, reduce welfare. The reason is that if firms are forced to charge a uniform price across the EC, but national preferences are retained, they will raise prices on their intra-EC exports and this may dominate the effects of lower markups on their domestic sales. In other words, EC92 could reduce welfare due to the loss of the beneficial aspects of a type of reciprocal dumping described by Brander and Krugman [19831. A similar result is shown analytically by Malueg and Schwartz [forthcoming]. The Haaland-Wooton model does not, - 3- the reduction of price discrimination occasioned by the 1992 program that we numerically implement, in which the extent of remaining price discrimination is endogenously determined. Our estimates are that the reduction of border and standards costs associated with the EC92 program will result in welfare gains of about 0.5% of EC welfare, a nontrivial welfare increase, but not overwhelming given that increasing returns to scale sectors are in the model. Allowing for the additional effects of standardization on consumer substitution elasticities (our version of integration), however, more than doubles the estimated welfare gains. Turning to the estimates of long-run welfare effects, the benefits appear to be much greater. In a widely cited study Baldwin 11989; pp. 266, 2691 employed estimates from comparative statics models of the effects cf EC92 as an exogenous input into a single sector Solow-type growth mcdel. He suggests that the long run effects will be at least double the static effects and probably will be as much as four times larger. We also estimate the steady state growth effects of EC92. Our approach differs fro.n Baldwin [1989; 19921 in that we integrate our multisector approaclh and the steady state effects into a single model. Consequently, the interaction between long run growth effects and comparative static effects at any initial income level is determined endogenously in our model. We estimate that the benefits in the steady state more than quadruple the estimate of the welfare benefits (about 2.4% of EC GDP) compared to the segmented markets comparative statics estimate, explaining some of the anticipatory euphoria surrounding the completion program even before it is fully enacted. Our general equilibrium model is described in section 2, and the main policy simulations discussed in section 3 (static welfare effects) and section 4 (steady state welfare effects). We compare our model and simulations lb some detail with the previous literature, pointing out the differences in our however, explain why EC92 would force uniform pricing in markets which were previously characterized by price discrimination. The problems that they identify, however, have led Winters [1992; p.20] to call for further research to distinguish trade barriers and consumer preferences as a rationale for home market bias. -4 - approach. Compared with earlier estimates our greater regional disaggregation reveals sharp differences of expected benefits from EC92 among EC countries based on intra-EC trade intensities. We emphasize: the endogenous determination of the extent of reduction in price discrimination deriving from the impact of the EC92 product standardization program on consumer perceptions of substitutability, the estimation of steady state growth effects in a unified endogenous framework, and the considerably greater regional and commodity disaggregation in our model. In most other respects our model is of the same genre as earlier general equilibrium work. 2. A Multi-Regional Tracle Model Generai Model Features In this subsection we focus on general modelling features. Due to their importance in modelling the process and effects of integration in the EC, we discuss in some detail below how we model (i) the reduction of border and standards costs as a result of the EC92 program, (ii) market integration as an endogenous consequence of other aspects of the EC92 program and (iii) the extension of static estimates of welfare effects to steady state estimates. Appendix A provides a det-i1ed algebraic formulation of the model and the decomposition algorithm employed for solution (see Rutherford [1992a,bI). The model separately identifies nine regions of the EC: Belgium, Denmark, France, Germany, Italy, the Netherlands, Portugal, Spain, and the United Kingdom. It also identifies Morocco, Turkey, and a residua! Rest of World (ROW), making twelve regions in all. The present version identifies twenty-six sectors per region, listed in Table 1. Twelve sectors in each region are characterized by increasing returns to scale (IRTS), as identified in Table I in the column reporting the Cost Disadvantage Ratio (CDR). The CDR, which is defined as the ratio of fixed to total costs, provides a measure of the importance of scale -5- Table 1: Sectors of the Model ID s| _m,~ It |. V. Add" CDR (1) AGR A&kWftm 010 35 FOO Feed 310,5 350 3 I1 BET ewiee Md Tobacc 35. 390 2 EKE &-V WO. 0%). 7... . - Uill Udus00 2 STE 1gW dI 130 1 5 NMM Nw.MeI MetFl Pn,ducow IS0 1 6 | CHM Cbb-thelmi _ 2 e MET Mei Prodw 190 2 S &MA 1agh meab1we' 210 2 6 OMA Off le htry 230 1 4 E3 Ebctical toodo 250 3 5 VEH maw Vdiciw 20 2 It 01! Oqb T,_oqt E13410 290 14 TXC Tiih ad C1hq 410. 430 2 __ . WOO Wo 450 I PAP PaprmPvond Pr50 2 it RPL Rubbw ad PFedi 490. 510 4 CON Ccmts¶d and Repair 53D. 5S _ TRA Tr__e SX 12 FIN P _mce Servi 6M. 710, 73 15 TRN TgAOgt ad Cmminkmi 610. 6M, 00, 650 6 6 HEBA Hed& S ei. 750 M 4 EDU HAaW Swyie M UM 4 SUR Odbw Su SW.90, ? 6 PUB OuMWe Pubboc Seo.i Ito, P3D to economies.7 The market structure in IRTS sectors is monopolistically competitive, with free entry and exit. All other sectors exhibit constant returns to scale (CRTS) and have a competitive market structure. 'The methodology for estimation of the CDR is discussed in Appendix B. - 6 - Table I also shows the original Eurostat sectors from which our aggregation was obtained, and the aggregate EC share of each sector in value added. We adopt a relatively detailed level of sectoral and regional disaggregation in our mcdel, since disaggregation has (as we show below) an important impact on the estimated effects of EC92. As Gasiorek, Smith and Venables (19921 (GSV) note, regional aggregation entails a nontrivial assumption in these exercises since it is primarily through the reduction in border and standards costs of exporting that the EC92 prograrn will have its impact. Regional aggregation in effect converts exports, on which a cost reduction would be achieved, into domestic production. Moreover, we show below that greater regional disaggregation is fundamental in revealing the considerable differences in the welfare benefits that will accrue to the EC countries with different intra-EC trade intensities. Table 2 lists the nine EC countries in our model and thei- intra-EC trade intensities. Regarding sectoral aggregation, previous studies have employed reasonable aggregation structures for IRTS sectors, but have employed one or at most two large residual CRTS sectors representing at least two-thirds of the economy. Table 8 offers a summary of the main model features of previous studies. Our 26 sector model allows us to keep separate virtually all of the sectors which have significant IRTS as estimated at the 44-sector level. It also allows Ls to maintain some disaggregation of CRTS sectors, along the lines that might be expected to influence results (viz., traded versus non-traded, labor-intensive versus capital-intensive). By received standards in the literature this makes for a "big model". The rest of our static model is relatively standard. We have multiple price-wedge distortions, such as factor taxes in production, value-added taxes, import tariffs, export subsidies, voluntary export restraints (represented as ad valorem equivalents) and non-tariff barriers (also represented as ad valorem equivalents). Productiorwentails the use of intermediate inputs and primary factors. Primary factors are mobile across sectors within a region, but are internationally immobile. Each region has a single representative consumer, as well as a single government agent, -7 - Modelling the Reduction in Border and Standards Costs of Intr:A-EC Trade The two types of real trade costs which constitute the "sand in the gears" in our model are border costs and standaraization costs on the supply side. Border costs represent the costs of undertaking trade, such as administrative costs of transacting and transporting over international boundaries. Standardization costs on the supply side are due to differences in technical specifications and regulations across national boundaries, being associated with the production costs of fulfilling technical regulations in foreign EC markets. Standardization costs, which would include the costs on the demand side of fewer or less adequate substitutes as discussed below, are viewed as being much more of a barrier to intra-LiC trade than explicit border costs (see Emerson et al. [1988; p.32ff.]), and their removal was one of the most ambitious aspects of the EC92 program. It is also the aspect of the program that has naturally been the hardest to enforce, given t' ; welter of national legislation and regulatory directives involved. No observer of intra-EC trade would doubt that substantial standardization costs will continue to exist in the EC for many years to come. Virtually all studies of the welfare effects of EC92 use a joint ad valorem production cost for these items of 2.5%. We follow this assumption, but decompose this value into an amount attributable to border costs and an amount attributable to production standardization costs. The reason for doing this is that we model the effects of each in different ways. Border costs are modelled as additional purchases of the domestic "transportation' good in each region. This good represents the activity of shipping, handling and warehousing for customs purposes. Standardization costs on the supply side are more likely to be reflected in extra costs of producing the specific good itself, rather than the purchase of inputs of any specific good. Thus we model supply side standardization costs as additional value-added in each sector in which trade takes place. These costs are treated as falling on imports to, and exports from, the region. They do not fall on domestic sales, and hence should be interpreted as the differential costs of transacting with other EC countries. -8 - Estimates of border costs for many EC regions and sectors can be extracted from Cawley and Davenport [1988; Tables B2, A31. The weighted averages of these border costs for trade with other EC regions is shown in Table 2. The EC average is about 1.7%, based on these estimates. If we assume that the real trade costs from border costs and supply side standardization costs amounts to 2.5% for each region, the region-specific standardization costs are derived residually and are shown in Table 2. These estimated supply side standardization costs are then assumed to apply to all traded goods within the indicated region. Theory and Evidence Against Uniform Pridng Previous studies have modelled completion of the market in two stages, with the first stage being the same as ours. In the second stage, previous studies have attempted to capture the presumption that EC integration will have its strongest impact by increasing competition in IRTS sectors. Their second stage involves a shift in the pricing behavior of national firms as the result of completion. Pricing is originally undertaken with "segmented markets", and with monopolistic competition in IRTS sectors this results in lower prices in foreign markets than at home.' With completion of the market these segmented markets are replaced with an integrated market in which price discrimination between home and foreign markets is assumed to be impossible. We claim that it is inappropriate either on theoretical or empirical grounds to model EC integration as a process that imposes complete pricitig uniformity. First consider the theory. Although the possibility of price discrimination by a monopolist selling in segmented markets has long been accepted in economic theory, recent theoretical work has shown the existence of price discrimination equilibria in a variety of multi-firm settings, even with free entry (e.g., Katz [19841, Borenstein [19851 ' This result awsumes, plausibly enough, that foreign matet shares for a given firm are smaller than domoatic market shares. Hence the perceived demand elasticity is different in the two markts, resulting in diffences in ,.ark-ups. -9- Table 2: Border, standard costs ar. intra-EC trade intensities for EC regions (in percentage) Country Border costs Standardization costsa Intra-EC" trade intensity Belgium 1.692 0.808 35.6 Denmark 1.695 0.805 14.9 France 1.717 0.783 11.1 Germany 1.807 0.693 11.9 Italy 1.576 0.924 10.1 Netherlands 1.623 0.877 29.1 Portugal 1.540 0.960 12.0 Spain 1.663 0.837 08.8 United Kingdom 1.903 0.597 11.9 a. Border plus standard costs sum to 2.5 percent. b. Defined as one-half intra-EC ex,ports plus imports divided by GDP. and Holmes [19891)9. These analyses suggest that price discrimination may arise even when market power over price arises only from monopolistically competitive product differentiation. On empirical grounds, consider the United States market, which is considerably more integrated than the EC is likely to become in the immediate future. Although there are generally alternative theoretical explanations for the observed pricing practices,"0 there are a wide variety of observed pricing practices in the United States that are consistent with price discrimination. These include: (1) two-part tariffs, in which a lump sum fee is charged and a usage charge per unit applies;" (2) discount coupons, which allow those consumers with lower opportunity cost of time to obtain a lower price; (3) selling the I See Varian (1989] for a review of the theory of price discrimination under non-monopoly conditions. 10 See Carlton and Perloff [19901 for alternate explanations. The welfare economics of these practices is a separte matter which we do not address. " Tie-in sales, where the purchaer of the product is required to purchase related products, is one type of two part tariff. For example, IBM, in requiring purchasers of its computers to buy its tabulating cards was able to obtain a higher price for the machine from the more intensive users. Xerox, is charging a per unit rental fee for copies that exceeded the marginal maintenan costs, price discriminated against intensive users. See Scherer and Ross (1990]. - 10 - same product under different brand names and charging a premium for the label as occurs in supermarkets;'2 (4) discounts on products that face stiff competition, for example in the shoe machinery (Kaysen [19561) and computer markets (Fisher, McGowan and Greenwood [1983]); and (5) geographic price discrimination under single or multiple basing points, with some absorption of transportation charges by the producer.13 A further highly counterintuitive result of the uniform pricing assumption is that intra-EC trade declines as a result of EC integration. This is because firms charge higher markups on their domestic sales than on export sales prior to integration. Uniform pricing in the EC causes firms to raise export markups and lower domestic markups, which reduces intra-EC trade. Modelling the Price Effects of Integration ImDact of Standards on Substitution. As a result of the concerns raised above, reservations have been expressed regarding the manner in which integration has been modelled in previous studies of EC92." Our approach is based on a review of the theoretical and empirical literature on standards which is presented in Appendix C. We conclude that the impact of the standards component of the single market program (which is regarded by businessmen as the most important component of the single market 12 See Wolinsky [1987]. '3 Some form of basing point system has been adopted by many U.S. industries including the steel, cement, lead and wood pulp industries. See Scherer and Ross (19901 for an elaboration. In addition, recent studies have found evidence of price discrimination in monopolistically competitive markets. Using data from retail gasoline markets, Borenstein [19911 used a model of spatial competition in which differences in the willingness of buyers to switch stations results in price discrimination between leaded and unleaded gasoline. As gasoline stations which offered leaded gasoline diminished, the relative price of leaded to unleaded gsoline rose. Shepard [1991] found the price difference between full service and self-service gasoline in stations that offered both services was consideably greater than the price diffetence across different stations that only offered one type of service. Based on a similar model, Borenstein and Rose [1991] found price discrimin&tion in the U.S. airline industry. Pashigian and Bowen [1991] have found that the greater use of sa:es in the pricing of apparl in retail stores is consistent with a price discrimination theory of pricing (as well as greater uncertainty), in which consumers who are more intered in being in fashion have ISS elastic demand and pay higher prices early in the season. "' For example, see Karp [1992; p.63], Smith and Venables [1988; p.1523], Merenier [1992; p.1] and Winters [1992; p.20]. *11 - program) will be to substantially increase the ability and willingness of EC buyers to substitute among the products of EC suppliers. Thus, one of the key features of our modelling approach is that the increased competitiveness in IRTS sectors comes about through the greater ability and willingness of EC buyers to substitute among the products of EC producers due to the single market program on standards. Standards arise from a variety of reasons. Taking the United States as an example, there are literally thousands of industrywide product standards. Most of these are voluntary, arise from the demand from buyers, and may be classified into uniformity standards and quality standards.'5 There arc also situations in which suppliers desire product standards, most notably for anticompetitive reasons.16 We explain why the single market program on standards should increase the ability of buyers to substitute among the products of competing suppliers in each of these cases. First consider uniformity standards. The most important aspect of these is interchangeability standards, which are designed for the express purpose of allowing buyers greater subst .Ation possibilities among different suppliers. Well known interchangeability standards are those in the auto industry which were imposed on parts suppliers by the Society of Automotive Engineers (SAE) on behalf of auto assemblers; these standards were designed so that assemblers would be able to substitute among different suppliers of parts such as wheel rims, spark plugs and screws and bolts. Other examples are standardized socket sizes in lamps that allow substitution among light bulbs. A photographer may interchangeably choose among various films, cameras, tripods, lens, filters and exposure meters due to standardization. To facilitate replacement, bricks sizes were standardized under government regulation. Operating controls for forklift trucks are almost all standardized so that training of drivers is interchangeable among different manufacturers products. Despite the enormous variety of paper towels, virtually all are eleven inches wide Is See Hemenway [1975] for an excellent and more elaborate treatment of the theory and evidence on industrywide voluntary standards. 16 Another prominent reason for suppliers to desire standards is to gain economies of scale. This was discussed above and included in the model. - 12 - to fit the standardized dispenser. Regarding quality standards, these also increase the substitution possibilities of consumers because the reduce the product differentiation advantages of producers. In the spirit of the Akerlof [ 1970] model, problems of degradation of product qual i.y often arise in markets where buyers have less information than sellers and it is costly to produce higher quality products. Brand names will often develop to resolve the degradation of product quality problem, but then consumers are reluctant to substitute among competing suppliers due to product differentiation. Grading of products then will allow greater substitutability among competing brands. The auto industry succeeded in imposing quality standards on most of its raw material suppliers such as steel, rubber, petroleum and machine products. Among these industries, opposition from steel suppliers was strongest because steel suppliers were reluctant to lose product differentiation advantages associated with brand names. Similarly oil viscosity standards for auto motor oil were imposed on reluctant refineries. In general, products that are graded for quality have little product differentiation advantages." Standards are sometimes imposed by suppliers for anticompetitive purposes. For example, fearing anticompetitive standards, the U.S. Federal Communications Commission mandated standardized interconnection for terminal equipment and open network architecture (ONA), which means that the components of the telephone system are made available on an unbundled basis so that competing supplier services can be combined in any manner desired. These standards have been highly successful, at least judged by the wide variety of independent supplier equipment and services that have developed." As part of its single market program, EC policy is attempting to implement similar reforms in standards. The EC green paper on telecommunications policy indicates that through the promotion of Europe-wide standards-it shall provide equal access to all market participants (EC [1987; p.5]). This includes milk, eggs, meat, lumber, soybeans, diamonds and mushrooms. ' See Besen and Saloner [1987] for further details. - 13 - Moreover, ONA has been mandated by the Maastricht treaty partly to insure the uniform interpretation of essential requirements across the Community in order to limit the possibilities for the imposition of restrictive licensing conditions." Probably the best known example of supplier-based anticompetitive standards is national standards that have the force of law for the purpose of protection, which the EC refers to as technical regulations. For example, EC courts interpreted German beer purity laws as protectionist and required that beer manufactured in any member state could be imported into Germany. This is an application of the mutual recognition principle in which products produced in one member state may be sold throughout the EC. In the absence of specific EC legislation however, EC states may still block intra-EC imports if certain national interests are involved. So the EC has attempted to achieve harmonization of technical regulations in which EC directives indicate mandatory requirements for national regulations.' In either event, the European consumer will be able to consume products from other EC states that previously were illegal to import. It is likely that considerations such as these led businessmen who were surveyed in the "Costs of non-Europe" project to conclude that they would have greater access to new regional markets and there would be increased price and non-price competitiveness as a result of the single market (Nerb [1988; p.27]). Moreover, the survey led some observers to conclude that "there is a dynamic of dema I associated with the learning process of consumers and enterprises which is released or accelerated when barriers are removed" (Nerb (1988; p.26]). Thus, the theory and empirical work on standards indicates that the impact of the EC92 program of standards will be to substantially increase the ability of buyers to substitute among the products of EC suppliers, justifying our modelling focus. "See XIII, September 1993, p.12 (the magazine of DO XIII of the EC) and Emerson et al. [1988; p.861. DA manufacturer may meet the technical requirements either by producing a product that meets the defined essential requirements (which allows manufacturer variety subject to the constraint of meeting the essential requirements of the product) or by producing a product in conformit) with European standardization bodies (Emerson et al., 1988, p.40). - 14 - Modelling the Impact of Increased Substitution. We employ a flexible demand structure in which consumers may have preferences for Ehe products of firms depending on region of origin. As a result of standardization Et buyers may substitute more readily among products from different EC firms. Then the perceived elasticity of demand of EC firms on intra-EC exports increases, and intra-EC export price margins are reduced endogenously. Consequtently, as markets become more integrated intra-EC trade generally expands. Firms in CRTS sectors will not directly be led to change their prices by this change in EC consumer perceptions, but there will be indirect effects due to the general equilibrium interactions induced by changes in the prices of IRTS sectors.2" We discuss this aspect of the model in greater detail below. More formally, prior to the EC92 program we enviszge consumers as possessing weakly separable utility functions that allow multiple stage budgeting of their choice decisions. First, based on a Cobb-Douglas utility function at the top level, they choose between the 26 different composite goods which are listed in Table 1, such as "motor vehicles" and "textiles and clothing". Having chosen how much to spend on each aggregate commodity, they then choose between domestic and imported composites of this commodity: between "domestic autos" or "imported autos", for example. This decision is based on a CES sub-utility function. Having decided how much to allocat; to imported autos, consumers then decide how much to allocate to imports from different regions in the model. This decision is also based on a CES sub-utility function. Finally, given the decision on how much to spend on autos from each country, consumers allocate expenditures on the different varieties from each country based on the lowest level CES sub-utility function.22 Figure 1 displays this structure of preferences (after the 21 A CRTS sector might use as an intermediate input the output of an IRTS sector whose price changes, therby changing the costs of the CRTS sector. Altematively, the CRTS sector may be a substitute in demand for the IRTS sector, and therefore experience a change in demand due to the change in relative prices between the two. I Given that the elasticities of substitution for these last two levels of choice with respect to foreign country and foreign firm variety (within a country) are the same, the structure is equivalent to firm level competition ammng all imports. A special case of this structure, with all elasticities of substitution equal, yields firm level competition among all firms that is independent of the country of origin. - 15 - FRENCH IMPORTS ¢ormak Lm Japanese O Other Imports autoamobilam from: Germon fIrms Jspnese firms firma from other region* Figure 1: Structure of Preferences Before the EC92 Program A EC Non-EC DD 1U O French ~O ermea Other EC Japanese U. Other Non-EC U from Sreack irma German rwme Japanese firma U.S firms Figure 2: Structure of Preferences After the EC92 Program - 16 - initial Cobb-Douglas choice of how much to spend on the aggregate commodity). After the EC92 program is completed, we envisage consumers have a different preference structure to reflect the notion that, due to EC92 programs regarding standards, EC-produced goods are now better substitutes than before. They are better substitutes for other EC-produced imports, and they are better substitutes for domestic varieties. We simulate this preference change along a continuum where in the limit consumers regard the products of all EC firms as equally substitutable. The latter preft:ence structure is shown in Figure 2. Specifically, in the fully integrated scenario the preference structure of consumers is directly analogous to the segmented scenario prior to EC92. Having decided how much to spend on any aggregate commodity, such as autos, consumers allocate that income among imported and domestic varieties of the commodity. However, now the notion of a "domestic variety" includes all other EC-produced varieties, even if they are not produced in the EC country in which the consumer lives. That is, we shift from treating other EC products as being "foreign varieties" and treat them as if they were 'domestic varieties".3 These two preference structures represent extremes, reflecting the situation before EC92 and the situation after the successful completion of the EC92 program. It is then a simple matter to model the path towards completion of the market as a weighted average of these two preference structures. Thus a 25% weighting on the pre-EC92 preference structure could be interpreted as saying that a quarter of the (homogeneous) consumers in that country perceive no change in the substitutability of EC-produced n This representation of preferences as changing 'structure' just reflects presuned changes in the off-diagonal elements of an implicit Slutsky substitution matrix between domestic and foreign varieties. It is possible to represent this change explicitly in a CGE model, using flexible functional forms such as the non-separable CES structure developed by Perroni and Rutherford 11992], but we opt for the simpler more easily interpreted formulation presented here. The crucial modelling feature here is not the use of different utility function structures; rather it is the correct characterization that EC goods are now better substitutes for home goods. - 17 - goods.' In most circumstances, economists are reluctant to consider changes in preferences as the basis for a policy exerc'ise, in large part because the created difficulties in evaluating welfare measures, such as Hicksian equivalent or compensating variation. In the present model, however, we modify preferences in such a way that indices of welfare change remain weli defined. This is because the changes in preferences do not alter the optimality of the initial allocation. In the absence of monopolistic pricing, the change in preferences alone do not disrupt the benchmark equilibrium. The key elasticities of substitution are denoted ODM. ODD and aOmM and reflect the substitutability between domestic (D) and imported (M) goods, alternative domestic varieties, and alternative foreign countries (and varieties). These elasticities are crucial to the effects of EC92. Our priors are that aDM < a,M < oDD' Our priors derive from our above discussion and appendix on standards which indicates that products produced in the same country will be more substitutable among themselves than products from different countries. This gives us the relationship that OK, < aDD; this inequality plays an important role in the analysis below.' Although it does not play an important role in the analysis, we also posit that domestic consumers are less willing to substitute foreign varieties for domestic varieties than they are among different varieties from foreign sources. In the absence of data-based estimates of these elasticities we initially specify aDM = 5, aw = 10, and ODD = 15. Our priors for the three key elasticities of substitution remain the same when applied to pre or post EC92 preference structure. A formal derivation is provided in Appendix A, but intuition into how market integration affects 24 We assume that this fraction is constant across all countries, reflecting the overll proges twrds completion of the EC92 program. It would be a simple matter to let it vary across EC countries, perhaps reflecting the differential speed with which EC countries are accepting EC-wide standards and are removing border cosa on EC imports. There is some anecdotal evidence of such differentials, at least as measured by the speed of enacting national legislation to implement the EC92 program (e.g., The Economrst, July 3-9, 1993, Survey on the European Community). U It also gives us cDMV < aDD' - 18 - markups and intra-EC trade can be obtained through exarnination of the markup equations. Define the markup for firms from one EC country (r) selling into another EC country (r') in the segmented market situation ;s m,,, and in the fully integrated equilibrium as m,,. It follows from equations (47) and (49) of Appendix A that the markup under segmented markets is: m,, - I + I - Il + I 1 aM an °w,V| N,6,m aD,, N, and the markup under fully integrated markups is: OD I OD+D I N I E, |w + I N (2) where rEEC and 0,,, denotes the market share of region r firms in region r'. Then the change in markup (defined as the fully integrated minus the segmented markup) simplifies to - in,,, = (3) LODD aUMM Nr L EC OrDM aDD J 5'4 ODM aAMM When this expression is negative, markups on intra-EC trade will fall and trade will tend, cetens paribus, to increase. As we have argued above, there is likely to be greater similarity and substiutabiity among the products of domestic producers than among the products of foreign producers, and still less between domestic and foreign, and we have chosen elasticities accordingly, i.e., aDg - 5, ,aA - 10, and aDD a 15. The first term in this simplified expression is negative when ODD > aw4 , and for our specific values equals -0.033; but there is some ambiguity with respect to the second term. Given the - 19 - ranking of the elasticities of substitution, the second term will be positive the smaller is 0c in relation to O., i.e., the smaller the share domestic firms have of the domestic market relative to non-EC imports in the domestic market.' Given our specific values, the change in the markup equals -0.033 + 6,, [(. 133)(l)1 N, [ @, 9C,, ] Numerically, we find that the sign of the second term is positive in most cases, but considerably smaller than the first term bec c, it is multiplied by the ratio of a share (0,,,) divided by the number of firms, where for example for Belgium's markup in Germany we have the share of Belgian exports in the German market divided by the number of Belgian firms. Thus, we find that the change in the markup is uniformly negative, where in most cases the decline in the markup is between 2 and 3.3 percent. We discuss how we model the steady state effects of completion of the market in section 4. 3. The Static Effects of Completion of the Market Static Welfare Effects Table 3 displays the results of completion of the market on aggregate EC welfare. In effect, this measure of the welfare effects adopts a utilitarian social welfare function for the EC, such that we simply add up the welfare effects for individual EC countries. In Table 3 we display a matrix of results where we simulate partial and full removal of border and supply side standards costs and well as partial and complete change in the elasticities of substitution (referred to in the table as percent of integration). Although the steady state results are presented in Table 3, we defer discussion of them to section 4. 2 Specifically, the second term will be positive if and only if 4 ,,, + E O,,, > 30,,,,. rRtEC rfe5C - 20 - Table 3: Welfare effects of EC92 on the EC Percent Removal of Border and Standards Costs 0% 25% 50% 75% 1I0% Steady Steady Stfea; Steady Stea4 Percent Static saxw Statk sate Statk stati Static state Static swe ntegration (1) (2) (3) (4) (5) (6) (7) 8) /9) (10) 0% IRTS 0.00 0.00 0.12 0.38 0.25 0.85 0.38 1.33 0.52 1.79 CRTS 0.00 0.00 0.11 0.34 0.22 0.74 0.34 1.15 0.46 1.56 25% IRTS 0.11 0.30 0.24 0.72 0.37 1.14 0.51 1.57 0.66 2.03 CRTS 0.00 0.00 0.11 0.33 0.23 0.70 0.35 1.08 0.47 1.48 50% IRTS 0.23 0.59 0.36 0.98 0.50 1.38 0.65 1.80 0.81 2.25 CRTS 0.00 0.00 0.11 0.30 0.23 0.64 0.35 0.99 0.49 1.37 75% IRTS 0.36 0.83 0.50 1.20 0.65 1.59 0.81 2.00 0.98 2.44 CRTS 0.00 0.00 0.11 0.27 0.23 0.57 0.36 0.90 0.50 1.24 100% IRTS 0.50 1.04 0.65 1.38 0.82 1.76 0.99 2.16 1.18 2.60 CRTS 0.00 0.00 0.11 0.23 0.23 0.49 0.36 0.79 0.30 1.10 a. Welfare effcts are in each case the aggregatc equivalent vaiation as a percent of aggregate EC GDP. Source: Model esimta. First, consider the static welfare effects with IRTS in Table 3. These estimates imply that removal of internal trade barriers and market integration are complementary to each other, which is consistent with the received wisdom from previous studies. Full integration with complete removal of internal trade barriers results in an aggregate welfare gain of 1.139% of GDP per annum. The relatve contribution of market integration and removal of trade barriers appears to be roughly additive and, as it turns out, somewhat symmetric. That is, x% integration and y% removal of sand achieves about the same welfare increase as y% integration and x% removal of sand, provided neither x nor y are zero. - 21 - CRTS Results and Comparison with IR[S. Despite the fact that we have argued strongly that it is appropriate to model 12 of the 26 sectors (about 21 percent of value-added) as being subject to IRTS, it is helpful in urnderstanding the results to consider the results in the counterfactual case in which all sectors are subject to CRTS. We implement this by setting the CDR equal to zero for all sectors. Given that there are no rationalization gains to be realized from improved scale efficiency in a CRTS model, we would expect that the welfare effects of EC92 would be much smaller, and indeed they are. In Table 3 the effect of assuming IRTS and monopolistically competitive pricing in our model is seen to increase the welfare gains by almost nothing (if we assume zero price integration) or by as much as 100% (if we assume full price integration). In Table 4 we preserL results showing the distributional effects across countries. We find considerable disparity in the welfare effects across EC countries, even though all gain. Table 4 shows the effect on welfare in percentage and in absolute terms (equivalent variation as a percent of GDP and in billions of 1985 U.S. dollars, respectively), the percentage change in the real wage and the real price of capital, and finally the percentage change in the real cost of living. The numbers without parentheses are for the IRTS fully integrated scenario in which all border and standards costs "sand" are removed. The CRTS results are in parentheses. Examining Table 4 for the distribution of welfare gains across the countries of the EC in the CRTS case, one sees that Belgium and the Netherlands are the countries which experience the largest increase in welfare as a percent of (iDP. The key to understanding this is to recognize that we have modelled the impact of the EC92 program on standards and borders costs as costs which will be reduced on intra-EC exports onlly, i.e., there are no cost reductions on production for the domestic market. Border costs require resources of the domestic transportation sector in our model and supply side standards costs are a component of value-added to the extent that the good is exported. Since the single market program is assumed to reduce the border and standards costs of intra-EC exporting by 2.5 percent, the first order - 22 - Table 4: Country composition of the welfare effects of EC92a: 100% integration and removal of border and standards costs e Per-ca chmge in: Regi-n We(fare I Welfa,r in dbuare Real wages Real prVce' Real co aw.r qfcapita . prces STa Steady Static Steady Szauic Steady S&tac Swuiy state 5*41 sA Belgium (BE) 3,37 6.39 3.32 6.31 5.2 6.7 3.3 1.4 1.0 (1.53) (1 .51) (2.6) (1.3) (0.7) Germny (DE) 1.10 2.03 7.27 13.39 1.0 1.6 0.9 0.1 0.1 (0.44) (2.90) (0.4) (0.3) (0.1) Dmamrk (DK) 1.82 3.78 1.21 2.51 2.4 3.0 1.7 0.7 0.4 (C.81) (0.54) (1.2) (0.3) (0.4) Spain (ES) 0.80 1.96 2.19 5.36 0.8 1.4 1.2 0.4 0.0 (0.36) (0 93) (0.4) (0.5) (0.2) France (FR) 1.13 2.47 7.07 IS.44 __ 1. 1.4 03 0.1 (0-30) (3.14) (0.5) (0.7) (0.2) taly (T) 1.05 2.03 4.62 8.8 1.0 1.5 0.8 0.0 40.1 (0.46) (2.03) (0.4) (0.3) 0 Toe Netherlas 2.48 7.73 3.41 10.63 3.0 4.9 4.0 1.2 1.1 (NL) (1.20) (1.65) (1.3) (2.5) (0.7) Portugal (C 1.04 1.72 0.53 0.7 0.5 0.6 1.3 -0.2 .0.6 (0.42) (0.21) (0.3) (0.3) 0 Uuited Iingdom 0.80 1.49 4.31 797 0 1 0.4 1.0 -0.3 .0.3 (UK) (0.29) (1.57) (-0 2) (0.4) (-0.3) Rest of Woid -0.00 0.05 -0.35 4.06 -0.00 0.0 0.3 0 O.0 (ROW) I-0.005) 1 (-0.42) (0.1) (01) 0 s bNumiben without paentheses are for IRTS in fully integrated scenario with *11 border and standards costs rmoved. Results for experiment with all sectors sbject to constant retums to scale are in parentheses. b Equivalnt variation as a percent of GDP. el Equivak.t variation in billions of 1985 U.S. dollarm d. Rea price of capita is fixed in the stady mate scenario. effect on welfare (as a percent of GDP) will be approximately equal to the share of intra-EC trade in GDP times 2.5 percent. But there will be additional gains from removing distortion costs. In Figure 3 we depict the welfare economics of a 2.5% reduction in the costs of intra-EC exporting under CRTS (in partial equilibrium). We take the example of Belgian exports of steel to Germany. There is a rectangle of benefits from the reduction of the costs of exporting Belgian steel to - 23 - FIGURE 3: Distribution of EC 92 Benefts under CRTS: Case of Belgian Steel In Germany \ ~~~~~~~~~1.026 EX J ~~~~~~~~~~~~EX sPo IPi j ~~~~~~~~~~~D '8 / I o, I , Quantity of Belgian steel In Gewrany NOTE: The Ina equilibrium for Belgian stee" in Gemarn Is at (Po, Qa), determined by the Intersecton of the export supply curve of Belgian steel producers to Gerrany (1.025 E9. and tie demand by German consurners for Belgian steel (0). Followin the changes of EC92, the export supply curve shifts down by 2.6 percent to EX. resulting a new equilibuum (abstracting from general equuilbrium effects) at (Pi, Q,). There are rectwnles of benefits A (to Germany) pius a (to Belgium), the sum of whbich Is 2.6 percent of the Initial exports; plus trangls of benefits from reduced distorion Costs equal to C (to Gerrnany) and D (to Belgium). - 24 - Germany, equal to 2.5% of the initial costs of exporting. Provided neither German demand for steel, nor Belgian supply of steel, are limiting elasticity cases, the initial rectangle of benefits will be shared between the exporting and importing country (A to German consumers and B to Belgian producers in the figure). In addition, the border and standards costs are analogous to distortions to trade, which when removed allow a reallocation of resources with an increase in intra-EC trade. That is, there are 'triangles' of benefits that augment the benefits from the reduced costs of exporting. In Figure 3 triangle C is a benefit to German consumers and triangle D is a gain to Belgian exporters. These realiocation benefits will be increased or decreased to the extent that the country is responsive to changes in relative prices of imports and exports. Figure 3 also reveals an important feature which impacts on the sharing of the benefits of EC92 across the EC countries. The more elastic is a country's demand or supply curves in relation to the other countries of the EC, the smaller will be its share of the benefits of EC92. In T2b!le 2 we list the intra-EC trade intensities of the EC countries in our model. The three countries that trade most intensely within the EC are Belgium, Netherlands and Denmark, and these are the three countries that gain the most from the EC92 program in the CRTS case. For the remaining countries there is little difference in their intra-EC trade intensities (ranging from 12% to 8.8%), and there is also not a large difference in their welfare gain as a percent of GDP (ranging from 0.5% to 0.29%). The benefits for all countries under CRTS exceed the first order effect of the lowering of the cost of production. In particular, 2.5% times the trade intensity ratio yields the following first order welfare effect in percent of GDP: Belgium 0.89, Netherlands 0.37, Denmark 0.37, Germany 0.30, Portugal 0.30, U.K. 0.30, France 0.28, Italy 0.25, and Spain 0.22. Other than for the U.K. and Denmark, the welfare benefits are between 1.4 and 2.2 times this "first order' effect, suggesting that the triangles in Figure 3 - 25 - are quite large.' By the standards of CRTS models, we have assumed rather large trade elasticities in our benchmark equilibrium, so the triangles in figure 3 are almost as large as the rectangles.2' Examining Figure 3 reveals that the benefits would be closer to the rectangles the more inelastic are the demand elasticities." When we counterfactually reduce all three import demand elasticities for all countries in our model to one-fourth of their original values, the benefits for most EC countries are reduced to between 1.17 and 1.28 of the rectangle effect.30 Thus, the intensiveness with which a country engages in trade is of first order importance in explaining the benefits that are likely to be achieved from CRTS sectors. More aggregated models have aggregated small countries such as Belgium with larger countries. In so doing they have produced a region which is closer to average with respect to trade intensities, which will mask important distributional effects across the EC countries. In column 9 of Table 3 the CRTS welfare gain to the aggregate EC from complete removal of border and standards costs barriers to intra-EC trade varies from 0.46% to 0.5% with CRTS, depending on the degree of market integration. Market integration, which in our model is greater substitutability among EC products, has only a very small effect on welfare in a CRTS model (by inducing slightly more resource movement) because it does not affect markups or entry and exit. 27 For the U.K. the benefits are close to the first order effect; this indicates that the U.K. is obtaining a relatively small share of the benefits of integration. As discussed above, this suggests that U.K. import demand or export supply is more elastic than that of the other EC countries. 23 Similar to the approach of Gasiorek, Smith and Venables [1992], we choose to calibrate with elasticities to be consistent with price-cost margins given by estimates of the CDR subject to a zero profit model assumption. Nonetheless, the trade elasticities are small in relation to those used in their models. 29 We show below, however, that the higher elasticities reduce the additional welfare gain from rationaliution and consumption efficiency gains in IRTS sectors. Thus, the high elasticities have offsetting effects regarding their welfare impact. w Exceptions are Denmark, which remains an outlier, obtaining benefits equal to 1.78 times the rectangle effect, and Germany and the U.K. which obtain less than the rectangle effect. Note also that the large demand elasticities that we have employed, however, result in somewhat higher estimates of the welfare effects and of the adjustment across industries in both the CRTS and I1P S cases. - 26 - The Impact of IRTS. Examination of Tables 3 and 4 reveals that the welfare benefits of the complete removal of border and standards costs roughly double. both for the aggregate EC as well as for individual EC regions, when the impact of IRTS with full market integration is incorporated. The key to explaining the impact of IRTS is the role of market integration on elasticities, markups and entry and exit. To clarify this point first compare the CRTS and IRTS results with 0% integration (row I of Table 3). Then the static effects of removing border or standards costs are virtually identical between IRTS and CRTS. That is, without additional effects from market integration, the beneficial effects of removing border and standards costs can be attributed to "traditional' efficiency gains found in a standard CRTS framework. The key to understanding the impact of IRTS in our model is that integration increases the elasticity of demand on intra-EC exports, thereby inducing markup declines. The equation for the markup was discussed earlier. A decline in the equilibrium markup can only be achieved through an improvement in the realization of scale efficiency. Despite the fact that we employ equivalent variation in general equilibrium, to isolate the crucial variables in IRTS sectors, in Figure 4 we present a partial equilibrium, consumers surplus interpretation of a symmetric monopolistically competitive industry with a given level of fixed costs per firm and constant marginal costs (MC). An additional simplification in Figure 4 is that it presumes a homogeneous output, so the elasticity increase is represented simply as a rotation of the market demand curve.' Originally there are nO firms charging price PO and producing q0. We assume zero profits, so rectangles A+ C equal fixed costs per firm. As a result of an increase in the market elasticity of demand, the perceived marginal revenue of the firm will equal marginal costs at a larger output level. In the absence of general equilibrium effects which chan-e actor costs, the marginal and average cost curves for the "' The analysis of Figure 4 extends naturally to our firm level product differentiation case by defining the market output as the CES aggregate of the output of the individual firms, and the market price as the price dual to the CES quantity aggregate. - 27 - FIGURE 4: RatlonszaUon mad Corhumpton EMoloy Gabm *hm Inlnmr d Ematm REPRESINTATNE FIRM AMRKET De p~~~~~~~~~~~~~~p Pt ~~~~~~~~~~~~~Pt D MC~~~~~~~~~~~~~M. '>~~~~~~~~~ ' 1 q . Mm M NOTE: bIer_as mnwcut danwnd ebalty kv*jd rpmrtlvs fms to br prim and soma to edt b,kyaIn hit rww zero pr equdun. 1V* weat gain equas Xt ares urKde tie dernwd mtrdn abve MC, Lo., Xi omrxnpion tfiknry gaki B pkus to m1lnJon gan noD, whem neD - ri - 28 - representative firm remain unchanged. Then, due to the zero profit constraint, exit must occur such that the new perceived demand curve of the representative firm, d,, is tangent to the average cost curve at the new lower markup. It follows that P, < P. and industry output increases. Welfare effects are shown in the panel of Figure 4 labelled Market, where D. and D, are the initial and new market demand curves. Since price exceeds marginal costs in the initial equilibrium, the expansion of output increases welfare; but the welfare gain is decomposed info two parts. First there is the typical consumption efficiency triangle B (as would occur for example with removal of a tax).' Unlike a tax, however, there is an additional gain of consumers' surplus equal to the rectangle n, that has no offsetting cost or lost tax revenue. The additional consumers' surplus gain derives from the rationalization gain of spreading fixed costs. That is, the expansion of output costs society resources at the rate of MC per unit but is valued as the area under the demand curve.33 These consumption efficiency and rationalization gains from increased elasticities in IRTS sectors are gains from the single market program over and above the gains from the reduction of the costs of exporting that were characterized in Figure 3.3 A similar interpretation will apply even when general equilibrium effects are incorporated that will induce shifts in the market demand curves. For IRTS industries that experience output decline or only small output increase, exit occurs. For those that experience output increases, entry can occur with a markup decline only with a significant output expansion. All of these scenarios result in rationalization 32 Following the procedure of Burns [1973], the efficiency triangle is obtained by connecting the pre and post equilibria by a straight line. 33 Since fixed costs per firm are unchanged A+C = C+D, i.e., A=D and n,4 = nj). Then the welfare increase equals the area under the demand curve between nq0 and n,q, that lies above industry marginal costs. 34 The greater the absolute value of the decline in the markup, the larger will be the welfare benefits from rationalization and consumption efficiency. A proportional reduction in the elasticities of substitution in the model, however, will increase the absolute value of the decline in the markup. That is, given benchmark elasticities, OD = 15, or'. = 10, the dominant first term in equation 3 is (1115)-(1I10) = -1/30. If the elasticities of substitutionwere scaled down, for example by 1/5, then the first term inequation3 would equal (1/3)-(1/2)= -5/30. This would result in a larger drop in the new equilibrium price in Figure 4 and greater welfare gains. - 29 - in the use of fixed costs and, since price-marginal costs markups decline, a decrease in consumption deadweight loss. These are benefits above what would occur in a CRTS sector. In Table 5 we present some of the key data for the IRTS sectors in three representative countries: Belgium, France and Spain. We focus on these countries because: Belgium is the country that gains the most from the EC92 program; France is a representative large country; and Spain is a country with relatively small trade shares in Germany. The table presents results for the scenario of 100 percent integration and 100 percent removal of border and standards costs. In the first three columns we present the rprcent decline in the 'Lerner" markup ([price-marginal costs]/price). The equation for the change in the Lerner markup is presented in equation 3. Note that for the industries in all three countries, their markups on their export sales in Germany change between -1.0 and -3.3 percent. For some products there are no exports to Germany in the benchmark data, hence the markup equation is not applicable. Since the Spanish share of the German market is small, when the markup equation applies, the decline in the markup is close to -3.3 percent, which is the value of the first term in the change in the markup equation. In columns 4-6 we present the percent change in entry and exit, and in columns 7-9 we present the percent change in output per firm. The key differences appear in the output per firm columns. We see that rationalization is much greater in Belgium than in the other two countries. Again the reason is that the intra-EC trade intensity of Belgium is much greater than the other two countries in the table. Since export sales constitute a much larger percentage of output for Belgium, the same percentage decline in markup on exports to the various EC markets induces a much larger reallocation of resources in Belgium's IRTS sectors. Columns 10-12 -how the percent change in output in the IRTS sectors, and the numbers in parentheses show the percent change in production when these sectors are presumed to operate under CRTS. One can see that different IRTS scenarios expand and contract in the different countries (a full - 30 - Table 5: Changes in IRTS sectors in Belgium, France and Spain: 100% integration and removal of border and s.:andards costs Percentage change in: Mark up in Germany Number f firnw OPSut per fnrm prod**on' BE FR SP BE PR SP BE FR SP BE Fa SP (1) (2) (3) (4) (5) (6) (2) (8) (9) (10) (i1) (12) Food -3.3 -3.2 -3.3 -19.6 -7.7 -1.0 30 11 6 4.7 2.9 5.1 (.5.2) (1.8) (1.6) Steel -1.6 -2.9 -3.3 4.5 0 -6.2 54 12 4 60.9 12.3 -2.7 (32.1) (6.0) (.7.3) Nonmetallic NAb NA NA -33.1 -11.0 2.4 17 5 9 -21.6 -6.6 i.1 minerals (-11.3) (01.1) (0.5) Chemicals -2.8 -3.1 NA -4.1 -4.6 -8.6 28 7 -1 22.8 1.8 -9.8 (4.3) (1.0) (4.5) Metal NA -3.3 -3,2 -27.2 1.4 1.4 5 3 5 -23.9 4.9 6.9 products (-11.8) (1.7) (-1.5) industrial -2.8 -3.1 -3.3 -10.9 -5.3 -11.2 33 -1 0 18.5 -5.9 -11A1 nachinery (11.3) (.4.4) (-12.4) Office NA -2.6 -3.1 -29.6 -8.8 -0.1 13 15 1 -20.7 5.3 0.6 macbinery (-12.1) (2.6) (0.0) Electrical -1.0 -3.2 -3.3 -22.9 -8.3 -12.5 44 1 2 11.1 -7.4 -10.7 goods (9.7) (-3.0) (-0.4) Motor -3.1 -3.2 -3.3 49.5 -16.3 -12.1 60 27 20 139.4 6.7 5.4 vehicles (45.5) (2.7) (-6.1) Otder NA -3.0 NA -29.1 -6.7 -12.6 20 12 8 -14.8 4.8 -5.3 transport (-12.3) (0.4) (0.0) equrpment Paper -3.3 -3.3 -3.3 -17.0 -11.2 -7.8 51 8 11 25.2 -4.5 2.8 (.0.5) (-2.0) (3.8) Rubber and NA -3.3 -3.3 -10.2 -2.7 -2.8 46 0 0 30.8 -2.8 -2.3 plastics _ _ _ I I _ (25.0) (-2.7) (-8.7) a, Value in parentheses is percent change in production under CRTS. b. NA = Not applicable since there are no exports to Gcermany from this country of this product in the benchmark. presentation of the output changes by industry and region are presented in the Appendix). The interesting pattem is that the sign of the output change in the IRTS scenario is generally the same as the sign of the output change in the CRTS scenario. The magnitude of the change in absolute value, however, is generally larger with IRTS. Thus, traditional determinants of resource allocation, factor intensities and remaining tax wedges, play a key role in influencing industrial structure. Shifts in relative costs that occur - 31 - in a CRTS world provide an impetus for output decline or expansion. If relative costs decline and output expands due to a change in factor intensities, in an IRTS world the output expansion induces a further decline in average costs and price, which magnifies the output increase. Returning to Table 4, note that despite the fact that markups decline, and real incomes35 increase, with integration, real prices also increase. For a sharp example, in Belgium real incomes increase while real prices also increase by 1.4% despite some significant declines in specific IRTS sectors.36 The key is that the relative price of labor increases by 5.2%, and the relative price of capital by 3.3%, so factor earnings increase at a faster pace than commodity prices." Factor earnings increase so much in Belgium because resources are being allocated more efficiently, hence the value of the marginal product of each factor is greater than with the previous allocation of resources. In addition to the benefits that occur under CRTS as discussed above, there are significant gains from rationalization and markup declines. All of these gains are dependent on the intra-EC trade intensity of the country. Intra-EC Trade. In the Appendix we present the percentage change in exports and imports to EC countries and non-EC countries, respectively. We have argued that a counterintuitive feature of modelling EC integration as uniform pricing is that intra-EC trade declines as a consequence. With our approach to integration, however, there is a strong increase in trade among the EC countries, and a modest decline in trade between EC countries and the rest of the world. We also note, without showing in a table, that 35 Our numeraire is a basket of final consumption in the Rest of the World. 36 For example, the following IRTS sectors experience price declines: Food by 3.3%, Iron & Steel by 4.7%, Chemicals by 3.2%, Industrial and Office Machinery by about 4.5%, and Vehicles by 5.5%. Most CRTS sectors in Belgium experience an increase in relative prices (e.g., Beverages & Tobacco by 2.4%, Utilities by 2.6%, and virtually all service sectors by about 3%). In the CRTS version of the model the Belgian increase in the price level is only about 0.7%. The structural pattern of price increases and decreases is approximately the same as in the IRTS version, even though no sector has IRTS in this case. 37 The same qualitative effect occurs in the CRTS model, with the relative price of labor increasing by 2.6% and the relative price of capital by 1.3%. - 32 - domestic sales of EC firms typically decline. Although domestic markups increase for EC firms, the largest declines in domestic sales are not explained by differences in markup changes; rather the largest declines in domestic sales occur in those countries that have the highest intra-EC trade intensities. Thus, overall we have a picture of EC countries trading much more with each other, relying less on their domestic markets for sales, and, since their economies become more trade intensive, there is only a slight decline in trade with countries outside of the EC. Regarding the decline in EC trade with the rest of the world, we have modelled EC92 as a shift in relative costs for EC countries toward trading with EC countries rather than non-EC countries. That is, the 2.5 % cost decrease on exports to EC countries accrues only to firms in EC countries. It is possible that some, .f not all, of the border and standards costs reductions that accrue to EC firms will also benefit non-EC firms.' To the extent that this occurs then there would be an increase in non-EC exports and imports to and from the EC for ROW relative to what we have reported. This would also allow a welfare increase for the rest of the world in the static model since then a picture similar to figure 3 would also apply to the rest of the world exports to EC countries. Comparing the Static Effects All four of the previous studies summarized in Table 6 followed Smith and Venables [1988] by first simulating a ^.5% reduction in border and standards costs, and then by simulating the same reduction in costs along with uniform pricing across EC markets (what they call "market integration"). The different results are partly explained by the different structural features of their models. We emphasize that there were good reasons for these features and aggregation schemes to be chosen for the purposes of the original studies; we caution, however, that this can affect the economics of the analysis 31 For .xample, a Moroccan truck delivering vegetables to Belgium should have reduced border costas it pauses between Spain and France and between France and Belgium. - 33 - Table 6: Key modelling assumptions and results of GE studies Harrison, Gasiorek, Rutherford Smith Haaland & & Tarr & Venablts Norman Mercenier (present study) (1992) (1992) (1992) Modelling assumptions Percent of economy subject to 21 34 23 6 IRTS' Weighted average CDRb 1.6% 2.4% 1.9% 1.1% (8%) (7%) (8%) (19%) Number of EC (non-EC regions) 9 (3) 7 (1) 1 (4)' 5 (1) Number of IRTS (CRTS) sectors 12 (14) 14 (1) 11 (2) 5 (4) Welfare effects, static model Remove border & standard costs 0.52 0.44 0.25 n.a. ('sand in the gears')2 Market integration 1.18 1 .35 0.48 0.87j . We calculate sector res based on value-add estimae for EUR12 fron Emerson eS al. 11988; table C2, pp. 271-2751 b. Weighted avenge CDRc are CDR values weighted by the shae of total valueadded of the ector. Weighted averge CDR vulu in parentheses are CDR values weighted by the sector's share of value-added within IRTS ectors. c. l4siand ad Nonran (19921 decomnpoae each of the EC and EFA into 6 identical region. d SC-wide welfare effects are a GDP-weighted average of the individual country welfare effects A reduction of real trade cost of 2.5% is simulated. c. Hasaand and Norman [1992; Table 3.41 report welfare gSains s a percent of expenditures on tradeable goods, wbich are 25% of their total expenditures. We scele their estimates (by 1/4) to plac. them on a comparable basis. f. Mercenier's (19921 Counot competition, free entry ad interrstionsl factor nobility case (Table 6d). Source: Authors' calculations. dramatically. All of the models in Table 6 included at least one sector with IRTS. Although for different reasons in our model versus the others, in general the single market program results in a rationalization of resources in the IRTS sectors across the EC, which increases welfare due to the greater realization of scale efficiency. Then the larger the share of the IRTS sectors and the greater the extent of unrealized scale economies in those sectors, Lhe larger the benefits. The first row of Table 6 shows the combined share of value-added of the IRTS sectors in the - 34 - respective models. Ceteris paribus, we would therefore expect GSV to find the largest gains from the single market program. The extent of unrealized scale economies in the IRTS sectors is measured here by the CDR applied in the initial equilibrium. As shown in row 2 of Table 6, Mercenier [19921 assumed higher values for the CDR in his IRTS sectors than the others, and thus there were considerable gains from scale to be realized in his IRTS sectors."9 Thus, despite the fact that the Mercenier [19921 model has a smaller share of the economy subject to IRTS than the Haaland and Norman [1992] model, it generates larger welfare estimates. In aggregate, our static results would appear to be most closely comparable to those of GSV. However, closer examination will show that the way in which EC92 has it's effects on the economy is quite different. In addition to the fundamentally different way in which integration in IRTS sectors is approached, which we have emphasized above, our aggregation schemes are quite different. First consider the regional aggregation questions. Table 6 shows that GSV offer the most regionally disaggregated model within the EC. GSV correctly note that the choice of regional aggregation can be relevant to estimates of welfare effects. They emphasize that since it is the elimination of barriers between regions within the EC that one gets the benefits of the single market program, the fewer the number of EC regions identified in the model structure, the smaller is the opportunity for sand in the gears to be present and to play a role in the simulations. We have shown above, however, that the role of trade intensities is crucial in explaining the different impacts across countries. The relatively homogen ous distribution of benefits found in GSV is a natural consequence of a regional aggregation scheme that aggregates over regions with diverse trade intensities. Haaland and Norman [19921 refer to their model as the "twin" to GSV, with greater disaggregation of the non-EC regions, but they treat the EC and EFTA as six identical regions. Obviously "'Me weighted average CDRs are calculated two ways. In the first, the sector weight is its share of value-added in the total economy; in the second, the sector weight is its share of value-added within the sectors with IRTS. - 35 - their approach will not allow an analysis of the distribution of benefits within the EC. We now turn to sectoral aggregation questions. The reason that previous studies choose their aggregation schemes is that lumping together two IRTS sectors with different CDRs (or an IRTS sector and a CRTS sector) will result in a weighted average CDR for the aggregate sector between the CDRs of the separate sectors, the exact value depending on the relative size of the aggregated sectors. Examination of the database on CDR estimates assembled in Appendix B, employing the original 44- sector Eurostat classification, reveals considerable variance in the CDR estimates across sectors. Given that returns to scale are presumed to be the driving force behind the larger welfare gains of EC92, previous studies have chosen correctly to keep as many IRTS sectors separate as computational considerations permitted, and to lump the remaining sectors into one residual CRTS sector. This way the resulting model reflects well the "returns to scale relief" of the original 44-sector database. Theoretically, however, it is possible that one can cause similar biases by excessive aggregation of the CRTS sectors. For example, these biases may originate from the differences in factor intensities that are being "lost" as one aggregates. Assume for the sake of argument that IRTS sectors expand due to EC92 and that they are relatively capital-intensive. The ability of these IRTS sectors to expand will depend on the factor intensities of other sectors, since they are the ones that will have to release resources to allow the IRTS sectors to expand. Aggregation over two CRTS sectors, one of which is capital intensive and the other of which is labor intensive, will result in an aggregated sector which is a weighted average of the two regarding factor intensities. Since different regions will specialize in different sectors, the results at the regional level could be very different.' These welfare effects will also depend on whether or not the CRTS sectors are the beneficiaries of reduced trade barriers. Whether or not sectoral A similar point applies with respect to aggregations over sectors with disparate trade policies when one is studying the effects of movements towards uniform policies; the greater is the model aggregation, the less welfae benefits would such uniformity policies be expected to generate, as emphasized in Harrison, Rutherford and Tar [1993; p. 195] in the context of a single-open economy model. - 36 - aggregation of CRTS sectors makes much of a difference to the estimated impact of EC92, however, is an open issue. To investigate this we have aggregated all 14 of our CRTS sectors into a single large sector, and consider a version of our model with 12 IRTS sectors and one CRTS sector. At the aggregate EC level, the welfare improvement is equal to 1.17% of GDP in the case of 100 percent removal of border and standards costs and 100 percent integration. This amounts to little difference compared to analogous 26 sector experiment where 1.18 percent welfare increase was obtained. For individual countries, however, there are significant differences with the following welfare increases (as a percent of GDP) by country in the 13 sector model: BE 2.93, DE 1.13, DK 1.29, ES 0.44, FR 1.11, IT 1.33, NL 2.88, PT 0.52, and UK 0.81. For the Netherlands, this constitutes an increase of 0.8% of GDP compared with the analogous scenario in which the CRTS sectors are disaggregated, placing it almost equal to Belgium as the country with the largest percentage gain in welfare. The gain for Portugal is reduced by a significant 0.5%. We conclude that at the level of specific regions there are significantly different results introduced by the aggregation bias. 4. The Steady-State Effects of Completion of the Market Modelling the Steady State Welfare Effects It is apparent that the political rhetoric surrounding major policy initiatives such as the EC92 program, as well as other regional trade liberalizations in process, revolves around the steady state effects rather than the static effects. This rhetoric has fed on arguments and calculations based on static effects, but has not been checked by the extension of those arguments and calculations to properly quantify the possible extent of steady state effects. We propose doing so for EC92 with a simple extension of our model. The present goal is not to describe the path of the EC after EC92, nor even to quantify the net long-run welfare effects of that path taking into account the costs of transition to a higher steady state - 37 - growth rate. Rather, we attempt to show how one can extend a relatively rich, detailed static model to address the upper bound welfare effects of EC92 in the long run." Our steady state calculations build on our static calculations in a simple way. In the static calculation we allow the price of capital to vary within each country, while holding constant the aggregate stock of capital in each country. The steady state calculation essentially reverses this: we allow the capital stock in each country to be endogenously determined while holding constant the price of capital in each country. This approach is in the spirit of the equilibrium concept proposed by Hansen and Koopmans 11972] and Dantzig and Manne 119741 for multisectoral planning models: solve for a time-invariant capital stock. An invariant capital stock equilibrium is a set of prices, production and investment levels for which the economy is able to grow at a steady rate with constant relative prices. In our model the optimal capital stock is defined as the stock such that the cost of investment, including depreciation and interest, is exactly equal to the capital rental rate. This can be viewed as a multisectoral version of the 'golden rule" equilibrium from older growth theories. The most important difference between our steady state calculation and the Hansen and Koopmans [1972] approach is that we do not explicitly measure the commodity-composition of investment. Hence there is some uncertainty regarding the determination of the steady-state capital price. We simply assume that the price of capital, within each region, is identical to the price of a basket of consumption goods. When we further assume that the benchmark capital stock is optimal, then the steady-state calculation reduces to fixing the capital price and permitting the capital stock to find an endogenous level. We emphasize that this calculation measures an upper bound on potential welfare gains in the long run. In the public finance literature there are many examples in which steady state gains are large but in the corresponding intertemporal model the gains are virtually offset by adjustment costs. After all, the " We would also argue that such a calculation, while being of interest in its own right, is also an important (numerical) input into any calculation of the longer run growth path that took account of transitional dynamics. - 38 - capital stock can only be produced through investment, and that requires reduced consumption along the transition path. For sufficiently high discount rates, the cost of the foregone consumption could easily outweigh the longer-run benefits of the capital accumulation it allows. We believe that our approach is defensible because it provides a meaningful upper bound on the potential gains from classical (Solow-type) growth effects. Of course, the values calculated in this way could fail to capture growth effects arising from induced improvements in productivity or innovation (so called "learning by doing"). Steady-State Welfare Effects Table 3 shows the baseline steady-state welfare effects of the EC92 program. The gains are slightly more than double those from the comparable static model. Completion of the program results in an aggregate steady-state welfare gain of 2.38% of GDP for the EC. The distribution of these welfare gains, and related economic effects, is shown in Table 4. Explaining the Steady State Effects In large measure the steady state effects are a reflection of the static effects. Examination of detailed production, trade, and pricing patterns reveals the same story as described earlier. The exception, of course, is that the capital stock is free to grow or contract to that level that keeps the price of capital at its benchmark value. Since we generated an increase in the relative price of capital in all regions in our static model, this implies that there must be an expansion of the capital stock in each region in the steady state. This expansion of the capital stock then works through in the model like an "endowment effect", generating larger welfare gains since there are more resources to be employed. - 39 - 5. Conclusion We model the static and steady state effects of completion of the internal market in the European Community on trade, production and market structure. The impetus for change comes from the removal of border costs and standard costs, as well as an increase in the perceived similarity of national products due to increased acceptance of common technical standards. Removing the border and standards costs of intra-Community trade results in relatively small welfare gains. The additional effect of standardization of consumer perceptions on market structure more than doubles the estimated benefits; and the steady state growth effect more than quadruples the welfare gains, explaining some t f the anticipatory euphoria surrounding the completion prograrr even before it is fully enacted. References Akerlof, G., "The Market for 'Lemons:' Quality Uncertainty and the Market Mechanism," Quarterly Journal of Econoaucs, 84, 1970, 489-500. Baldwin, Richard, 'The growth effects of 1992", Economic Policy, October 1989, 248-270. Baldwin, Richard E., 'Measurable Dynamic Gains From Trade", Journal of Politcal Economy, 100(1), February 1992, 162-174. Besen, Stanley and Garth Saloner, *The Economics of Telecommunications Standards," in R. Crandall and K. Flanun (eds.), Changing the Rudes: Technological Change, International Competition and Regulation in Communications (Washington, D.C.: The Brookings Institution, 1987). Borenstein, Severin, 'Price Discrimination in Free-Entry Markets," Rand Journal of Economics, 16, Autumn 1985, 380-397. Borenstein, Severin and Rose, N.L., "Competition and Price Dispersion in the U.S. Airline Industry," Working Paper #3785, National Bureau of Economic Research, July 1991. Brander, James A., and Krugman, Paul R., "A Reciprocal Dumping Model of International Trade," Journal of International Economics, 15, 1983, 313-321. Brooke, Anthony; Kendrick, David, and Meeraus, Alexander, GAMS: A User's Guide (Redwood City, CA: The Scientific Press, 1988). - Burns, Michael, "A Note on the Concept end Measure of Consumers' Surplus", American Economic Review, 63(3), 1973, 335-344. Carlton, Dennis and Perlof, Jeffrey, Modem Indusrial Organization (New York: Harper Collins, 1990). Cawley, Richard, and Davenport, Michael, "Partial Equilibrium Calculations of the Impact of Intenal Market Barriers in the European Community". Economic Paper No. 73, European Commission, Brussels, 1988. - 40 - Ceoechini, Paolo; Catinat, Michel; and Jacquemin, Alexis, 7he European ChaUenge: 1992 (Aldershot, UK: Wildwood House, 1918). Chipman, John, "Homothetic Prmferenees id Aggregation", Journal of Economic Theory, S, 1974, 26-38. Dantzig, George, and,Manne, Alan S.. "A Compiementarity Algorithm for an Optimal Capital Path with Invariant Proporaoi", Journal of Econormic Theory, 9, 1974, 312-323. Eisenberg, E., "Aggregation of Utility Functions", Managenwnt Science, 7, 1961, 337-350. Emerson, Michael; Aujeen, Michel; Catinat, Michel; Goybet, Philippe, and Jacquemin, Alexis, The Economics of 1992: 7he E.C. Commission's Assessment of the Economic Affects of Completing the Internal Market (Oxford: Oxford University Press, 1988). European Community, Towards a Dynarnic European Economy: Green Paper on the Development of the Common MaItfor Telecommunications Equipment (Brussels: EC Commission, 1987). Fisher, Franklin; McGowan. John; and Greenwood, Joel, Folded, Spindled and Mutilated: Econonic Analysis and U.S. v. IBM (Cambridge, MA: MIT Press, 1983). Gasiorek, Michael; Smith, Alasdair; and Venables, Anthony J., '1992: Trde and Welfare; A General Equilibrium Model", in Winters, L.A. (ed.), Trade Flows and Trade Policy After '1992' (New York: Cunbridge University Press, 1992). Haaland, Jan 1, and Wooton, Ian, "Market Integration, Competition and Welfare', in Winters, L.A. (ed.), Trade Flows and Trade Policy After '1992' (New York: Cambridge University Press, 1992). Hansen, Terje, and Koopmans, Tjalling, "On the Definition and Computation of a Capital Stock Invariant Under Optimizaion', Journal of Economic Theory, 5, 1972, 487-523. Harrison, Glenn W.; Jones, Richard; Kimbell, Larry J., and Wigle, Randall, 'How Robust Is Applied General Equilibrium Modelling?", Journal of Policy Modelling, 14, 1992. Hanison, Glenn W.; Rutherford, Thomas F., and Tarr, David G., 'Trade Reform in the Partially Liberalized Economy of Turkey", The World Bank Economric Review, 7(2), may 1993, 191-217. Harrison, Glenn W.; Rutherford, Thomas F., and Wooton, Ian, 'An Empirical Database for a General Equilibrium Model of the European Communities", Empirical Economics, 16, 1991, 95-120. Hemenway, David, Industrywide Voluntary Product Standards (Cambridge, MA.: Ballinger, 1975). Holmes, T.J., "The Effects of Third Degree Price Discrimination in Oligopoly," American Economic Review, 79, March 1989, 244-250. Katz, Michael, 'Price Discrimination and Monopolistic Competition," Econometn a, 52, November 1984, 1453-1471. Kaysen, Carl, United States versus United States Shoe Machinery Corporation (Cambridge, UK: Cambridge University Prs, 1956). Karp, Larry, 'Discussion of Gasiorek, Smith and Venables," in Winters, L.A. (ed.), Trade Flows and Trade Policy After '1992' (New York: Cambridge Univerity Pres, 1992). Krugman, Paul, "Endogenous Innovations, Intenational Trde and Growth,' Unpublished Mamuscnpt, 1988. Malueg, David, and Schwauzz, Marius, "Pamllel Imports, Demand Dispersion and lemational Price Discrimination", JodwVa of International Economics, forthcoming. Mclo, Jaime de, and Tarr, David, General Equilibrium Analysis of U.S. Foreign Trade Policy (Carnbridge, MA: MIT Press, 1992). - 41 - Mermenier, Jean, 'Completing the European Internal Market: A General Equilibrium Evaluation Under Alternative Market Structure Assumptions", Worldng Paper 0892, Department of Economics, Univernity of Montreal, March 1992. Nerb, G., "The Completion of the European Market: A Survey of European Industry's Perception of the Likely Effects," The Costs of Non-Eurqpe (Luxembourg: Office for Official Publications of the European Communities, 1988, Volume 3). Norman, Victor, "EFTA and the Internal European Market", Econonuc Policy, October 1989, 423465. Pashigian, Peter and Bowen, Brian, "Why Products are Sold on Sale?: Explanations of Pricing Regularities," Quarterly Journal of Economics, 106, November 1991, 1015-1038. Perroni, Carlo, and Rutherford, Thomas F., "Regular Flexibility of Nested CES Functions", European Economuic Renew, forthcoming. Rutherford, Thomas F., "Extensions of GAMS for Complementauity and Variational Problems Arising in Applied Economics", Econonies Working Paper 92-9, Department of Econorices, Univernity of Colorado at Boulder, 1992a. Rutherford, Thomas F., "Sequential Joint Maximization", Econonics Wor/ang Paper 92-8, Departnent of Economics, University of Colorado at Boulder, 1992b. Scherer, F.M., and Ross, David, Industrial Market Structure and Economic Performance (Boston: Houghton Mifflin, 1990). Shepard, Andrea, "Price Discrimination in Retail Configuration," Journal of Poluical Economy, 99, February 1991, 30-53. Smith, Alasdair, and Venables, Anthony J., "Completing the Internal Market in the ELropean Comnunity: Some Industuy Simulations", European Economic Revew, 32, 1988, 1501-1525. Solow, Robert M., "A Contribution to the Theory of Economic Growth", Quarterly Journal of Econoiucs, 70, February 1956, 65- 94. Tarr, David, "A Permanent Income Approach to the Growth of Incone by Sectors in a Leontief Economy," Quarterly Journal of Economics, Febrwary 1975, 89, 145-150. Tarr, David, "On Distributed Lags, Morishina Matrices, and the Stability of Economic Models," Econometica, May 1976, 44, 597-600. Tarr, David, "The Steel Crisis in the United States and the European Community: Causes and Adjustments," in R. Baldwin, C. Hamilton and A. Sapir (eds.), Issues in U.S. -EC Trade Relations (Chicago: University of Chicago Press for the NBER, 1988). Tarr, David, "Comments on Trade Policy Under Imperfect Competition," in A. Srnith and P. Krugman (eds.), Strategic Trade Policy (Chicago: University of Chicago Press, forthcomning). Varian, Hal R., "Price Discrimination," in R. Schmalansee and R. Willig (eds.), Handbook of Industrial Organization (Amnierdan: North Holland, 1989). Winters, L. Alan, "The Welfar and Policy Implications of the Intenational Trade Consequences of '1992'", Anerican Econonic Review (Papers A Proceedings), 82, May 1992, 104-108. Winten, L. Alan (ed.), Trade Flows and Trade Po*icy After '1992' (New York: Cambridge University Prss, 1992). Wolinsky, Asher, Brmnd Names and Price Discrimination," Journal of bidstral Economics, 35, Mach 1987, 255-268. - 42 - Table of Contents Appendices Table Al Effects of EC92 on Output: IRTS compared with CRTS Table A2 Change in the Value of Exports to and Imports from EC Countries, by Region and Product Table A3 Change in the Value of Exports to and Imports from non-EC Countries, by Country and Product Appendix A: Algebraic Formulation of the Model A-1 I Model Structure - An Overview A-1 1. I Key Features A-2 1.2 Markets and Prices A-2 1.3 Equilibrium Surmnary A-3 2 Equations for the Central Model A-4 2.1 Markets - Primal Form Equations A-4 2.2 Profit conditions A-7 3 Pricing Equations for Monopolistic Industries A-9 3.1 CES Aggregators (primal form) A-11 3.2 Associated price indices (CES cost functions) A-11 3.3 Associated demand functions A-11 3.4 Firm behavior A-12 3.5 The perceived elasticity of demand A-12 3.6 Application to the EC model A-14 4 Formulation and Solution using Dual Joint Maximization A-15 4.1 Integrability with homothetic preferences A-16 4.2 Joint maximization with primal variables A-17 4.3 Joint maximization in price space A-18 4.4 Dual joint maximization with price distortions A-19 4.5 Dual joint maximization formulation for MRT A-20 5 Single commodity submodel for monopolistic sectors A-20 Appendix B: Calibrating the Cost Disadvantage Ratio B-1 Appendix C: Standards and Buyers' Increased Substitution Possibilities C-5 The Auto Case C-6 Uniformity Standards C-8 Quality Standards C-8 Other Motivation for Standards C-9 Small Firm Quality Assurance C-9 Anticompetitive Motivation C-10 Conclusion C-11 Appendix D: Data and Benchmarking D-1 1 Assembly and Reconciliation of 10 and Trade Data D-1 2 Calibration of the Monopolistic Submodel D-4 3 The Benchmark Equilibrium D-8 Table A1 Effw.tu of EC92 on output. IRTS compared with CRTS' (percentage change) tor BE DE DK ES FR IT AL PT UK ROW AGR 2.2 -5.0 18.2 2.3 3.3 -4.3 11.9 2.1 0.1 0.50 (-2.0) (-3.0) t7.2) (O0n7 _ (2.6) (-2.6) (2.5) (I.-1) (40 4) (0.2) ..,,'4.7 -7.2 27.8 5.1 2.9 -6.2 32.7 4.4 -2.1 0.0 ,, ., t-5 (-2.5) (8.9) (2 0) (1.8) (1.6) (-1.5) (0.L0) 3ET -6.6 1.8 1.3 -0.5 3.6 -2.3 -6.2 10.5 0.0 0.0 (- 1.0) (1.4) (O.S8) (40 T? (3 6)_ O (-3. 5) (-3.S ) (S.0(.6) (0.0) NE 2.4 -5.3 0.8 2.7 -0.3 -0.2 5.9 8.6 4.4 0.2 (5.4) (_ 5.6) (3.6) (1.7) (_ 1.7) (-0.4) (10.3) _6._ ) (3.6) (40.2) TI 7.6 0.6 1.7 0.5 1.5 -2.0 4.1 -1.2 0.3 -0.1 (4.2) (0.1) (1.1) (0.4) (1 2) _,(-1.8) (1.4) (-2.2) (0.0) (0.0) F 2.1) (0.2) (1.2) (-0.8) (6.0) (-6.2) (21.6) (-7.3) (-2.8) (-1.6) w4 ~ g >21.6 -3.0 -13.8 11.1 -6.6 -2.1 -14.5 -2.3 17.7 0.1 .f....f3' .11.3) (4 .2) (-5.0! (3 .7) (-I . I ) (0.0) (-S .S) (0 .S) (4 .6) (0 . I ) w~r '" Z2.8 1.7 -17.2 -9.8 1.8 -4.8 23.8 -13.0 4.2 -0.6 N 23.9 -0.7 18.5 6.9 4.9 4.9 2.7 -0.8 -4.3 -0.3 1..211 I 18S -0.6 -4.2 -11.1 -5.9 5. 1 10.3 -28.9 -9.2 -0.4 ' 11 3) ( ) (-4.1) (4. -4.4) (2.9) (3.2) (-12.4) (-4,4)- (4024) . 20.7 3.8 -12.7 -0.6 5.3 -3.5 -6.0 0.0 2.6 -0.3 .. ' , '.12. 1)_ (2.3) (-7.4) (-1.7) (2.6) 2 (-4.8) (0.0) (2.0) (4 . 12 ) - W 11.1 7.2 -11.1 -10.7 -7.4 3.8 31.3 8.2 -3.9 -0.4 ,4 '9.7) (3,1) (-1.7) (4.7? 0-0oL (0.7 (20.1) (4.4) (-2.7? (4.1) i 39.4 4.4 -40.0 5.4 6.7 0.6 19.6 -6.6 -11.3 -0.2 . 7!., . 45.5) (2.7) (-15.9) (1.4) (2.7) (-2.4) (-0.6) (-6.1) (-6.3) (-O.1) . M 414 3 6.6 -1.4 -5.3 4.8 0.1 16.4 0.0 4.9 -0.9 '12.3) (2.3) (-8, 1) (-2.3) (0.4) (2.0) (0.2) (0.0) (2.4) ( 4.2 ) . C -0.5 1.1 -1.9 0.6 -1.5 3.9 7.0 1.5 -'.5 0.1 (9.2) (0.6) (2.0) (-0.5) (-1.2) (3.4) (9.4) (0.8) (-2.0) (0.0) A/00 8.8 0.5 34.8 0.2 -6.3 4.3 -17.5 14.8 -0.4. 0.1 _ 16.7) (-0.1) (37.0) (0.2) (-7.4) (3.1) (-14.3) (13.6) (-2.4) (0.0) Wz,, . < .0z.2S.2 9.4 3.5 2.8 4.5 4.5 -1.6 10.7 -4.6 -0.1 ¢ ."Z,S4.S) (3.7) (-0.2) (0.7 (-2.0) (-O.1) (-4.3) (3.8) (-1.2) (0. 0) .. ,',',. ,p0.8 -7.0 -9.4 -2.3 -2.8 25.5 -16.9 -15.0 -0.8 -0.2 .0) (-2.7) (-4.8) (-1.3) (-2.7) (10.5) (-10.5) (-8.7) (0.2) (0.0) ON' 1.7 1.0 1.1 0.6 0.8 0.3 1.9 0.7 1.2 0.0 (0.5) (0.4) (0.3) (0.3) (0.4) (0. 1) (0.7) (0.2) (0.6) (0.0) -5.4 0.4 0.0 0.4 1.2 0.1 -1.3 0.4 0.0 0.1 (-3 .2) (0.2) (-O. 1) _ (0.2) (0.8) (-O. 1) _ -I.2) (0.2) _ (-O. 1) (.1 FINl -3.7 0.7 -1.6 0.8 0.0 0.9 0.7 -1.0 1.0 -0.1 (-1.6) (0.3) (-I1.7) (0.3) (40.2) (O.S) _(40.3) (-1.0) (0.6) (0.0) TRN -6.4 -2.3 -3.' -0.4 -1.1 -0.8 -18.7 0.5 2.1 0.4 (-11.2) (-2.S) (-2.2) (40.6) (-1.6) (40.9) (13.4) (40.9) (I.0) (0.2) HEA 1.7 0.7 0.9 0.4 0.5 0.4 0.9 0.3 0.0 0.0 (0.8) (0.2) (0.4) (0.2) (0.2) (0.2) (0.4) (0. 1) (0.0)(0) DDU 0.2 0.5 0.6 0.1 0.4 0.2 0.2 0.3 0.0 0.2 (0.0) (0.2) (0.2) (0. 1) (0.2) (0. 1) (0. 1) (0. 1) (0.0) (0.0) ER 0.4 1.0 0.6 0.3 0.5 0.4 -4.6 0.0 0.4 0.0 (0.2) (O.) (. (0(01) (0.2) (0.2) (-.) (-O.1)(03(.) UB 0.9 0.8 0.8 0.6 0.7 0.6 -0.9 0.6 0.6 0.2 (0.3) (0.3) (0.4) (0.3) (0.3) (0.3) ( (03) (0.2) (0. 1) a. CRTS reault are in panthc. Shaded secon uLe those subject to IRTS. Table A2 Chane im the value of exporb tod hlmporb fromn EC oousbries, by regiom and product (in percentage) | Elr IM~~L EXb Dl EX lM EX im EX IM EX IM EX lM EXI IM EX IIM EX Mf AGR 49 ~~52 54 93 38 26 35 156 41 Bs 42 148 18 52 0 0 35 17 3 -1 FOO 132 ~135 Ila 792 11II 35 117 32 IJH 105 126 309 97 14 140 38 102 168 Q -1 BET 2 S 8 27 51 46 45 62 48 O O O 34 65 84 5 32 40 5 -9 ee 17 0 0 0 35 39 40 36 49 63 50 51 20 18 66 53 14 7 1 o 13 _35 _6 .56 45 51 69 32 41 13 80 569 41 66 63 88 24 180 9 -1 STE 74 41 66 70 6 47 St 97 81 64 87 261 64 40 107 214 38 45 -10 2 NM I 121 44 64 47 159 104 3 61 160 63 84 Sl 124 81 102 K3 39 -5 -5 CM 66 35 68 62 O O O O 76 71 89 206 62 31 o o 68 57 -4 o MEb 89 2161 76 100 60 III 5S I IS 47 1 16 63 1 10 126 144 169 104 285 -3 2 IMA SO 30 67 56 89 91 78 107 74 77 89 53 so 57 96 434 34 107 -8 2 OMA 0 0 36 24 0 0 72 84 41 37 53 96 0 0 0 o 0 o -1 o- ELG 34 28 71 49 27 5 7 92 203 86 159 103 93 52 24 119 95 69 116 -4 0 VEH 132 69 S8 53 -17 552 71 65 63 60 79 96 73 64 10S 262 39 143 S5 4 15 I 103 56 43 8S 165 61 .130 75 60 34 45 871 49 O 48 48 -6 2 TX 7 12 24 25 13 21 121 24 121 30 12S_ 12 22 23 2R IS 21 41 1 -2 WCOo 42 22 23 28 76 21 31 34 23 87 52 35 25 1120 36 2 41 47 3 O PP 78 So 94 SS 99 10 10S 92 94 140 109 124 83 I1 IS o8 41 8S 210 -2 O RL 37 17 O O O O 113 167 96 .133 103 23 O O 62 313 O O -2 O CON -4 113 22 S O O O _ O 17_ _16 20 65 7 19 30 14 42 2 2 46 70 _ 0 O O O 35 12 24 -2 312 to 23 19 O SO 2 7 -899 FI 19 ,37 16 12 9 709 17 -7 21 22 23 9 -6 6 O O 36 7 4 -4 TN -2S 28 9 iS 4 _3_ _ 5 14 9 17 3 -11 64 23 10 35 1 1 -27 ED O O O O O O O O O O O -7 O O O O O 6 -26 slER~~ ~~ ~~~~~ _ _ _ _ 17 _ _- - 10 42-O-O 5 7 4-8 4 _ _ UB O- O 52 -17 37- _ ° O O O 56 10- - 10 40 0 O O O 8a 4 a. See tble I for sedor nae. Table A3 Change In the value of expoeb to mnd Imports fronm on-EC oountries, by oountry end product CaM percege) BE DE DKES FR ir tT |PlT r UK R RW E. l 1m L. LV EX | D lL - - EX | EX Er EX 0 EX W AGR 13 _ i 26 22 _ 6 4-2 -6 11 46 2158 5 -4 O FOO O O -3 -3 2 4 -W - -8 -5 20 0 0 W -8 I _L BFL 0 O _3W4 O _ 0 -1_ _ _ _3 3 3 _3 1 O 3 1 -I 33 _ A 0 0 O 0 0 - O AL O 0 O _ 3 O Ull 19 57 2 1 -2 3 O Q 0 3 O O O O 0 O -16 0 Q O4 4 -2 5 - 2 1 - 6 - -1 5 A I N-2 O _ =1 -3 === =4 =2- 2 - 0 2 4 =2 11 == -7 0 -I -I 1 -321 O -I I I -I -3 5 -2 6 4 1 4 IA-2 -4 L1 1 -1 _4 0 6 _ -S -1 2 - 7 560 -3 _ 01MA 2 0 _ O _ 4 -08 1 -5 -6 12 0 0 O 0 ELG -7 -23 ~ ~2 -1 O ° -I -9 -2 -3 3 -5 -1 B2 O . 5 -9 2 O VEH ~25 178 O -I ° 0 4 -I 2 -4 3 -S 2 -27 9 -6 6 -13 - 0 01E 14 -1Q _ _ 1 -° - -2 1 -I -2- - 5-4 0 2 A 0 0 W 1 I 4 O 4 _ -2 1 0 O 3 -2 2 -7 WOO O O O O O O O 0 O O 0 2 O O Q O 2 -25 0 O .0 0 I 0 0 0 -2 -2 -2 -7 0 -4 -7 -30 2 -8 10 -4 15 -2 0 - 4 0 - 2 5 -4 - 5 -8 5 -2 2 - f1 5 W -2 1 4 9 0 = W 0 - -9 0 = W W 0 2 0 | TRA -24 2 O Q 0 -2 11 -3 O O O -14 I O O 6 O 3 14 I 191 0 0 - 0 - 0 4 2 - 3 _ 0 0 0 2 1 9 |LTRN -22 O -I ° 8 ° ° I Q -3 O O 0 15 1 3 O 5 O 4 2 £EL W ° 0 p O O 07 O O W Q O = O OO 0 - O 0 FM 0 AL AL AL -10 AL A A -2 2L 0 -16 0 0 0 AL AL 0 A PUB 0 0 0 0 0 0 0 0 0 0 6 0 -10 0 0 0 0 0 0 Appendix A: Algebraic Formulation of the Model This appendix presents the mathematical structure of a multi-regional general equilibrium model which has been developed for trade policy analysis. We refer to the model as MRT. These notes provide a detailed reference which complements the model's GAMS source code. Details of data preparation and calibration are provided separately in Appendix D. Section 1 presents a general overview of the equilibrium structure. Section 2 presents a more detailed algebraic representation of equilibrium conditions for the central model. Section 3 develops the pricing equations for monopolistic sectors. Section 4 describes how the central equilibrium structure is represented using dual joint maximization. Section 5 summarizes the partial equilibrium submodels which determine price-cost margins and the number of firms in monopolistic sectors. 1 Model Structure - An Overview 1.1 Key Features The main features of the model are described in the text. They are: * 12 regions: Belgium, Denmark, Spain, France, Italy, Netherlands, Portugal, United Kingdom, Morocco, Turkey and Rest-of-World. * 26 traded commodities. * Monopolistic competition with increasing returns to scale in a subset of 12 industries. * A flexible Armington structure in which commodities are distinguished by country and/or region of origin. Market integration is represented by a change in the preferences of consumers within the EC so that intra-EC goods become closer substitutes. This induces changes in producer pricing for sectors subject to monopolistic competition. A-1 * Two types of resource-consuming non-tariff barriers: "borders costs" and "standards costs", both of which are affected by the EC92 program. * Multiple price-wedge distortions: factor taxes in production, tariffs, export subsidies, voluntary-export restraints (represented by ad valorem equivalents), non-tariff import restrictions (also represented by ad valorem equivalents). * Full general equilibrium structure with input-output tables and fully endogenous primary factor markets in all regions. 1.2 Markets and prices The following notational conventions are adopted: ij are used to index goods. r,r' are used to index countries/regions. f is used to index primary factors (labor and capital) P, refers to a market price index. Unless otherwise noted, the benchmark value for any market price is unity. X' refers to the benchmark value of a quantity variable X. When reference prices are unity, all quantities are measured in billions of 1985 U.S. dollars. The equilibrium structure of the model is based on the following market prices: PUr Price index for final consumption in region r. PSj, Consumer price for the Armington aggregate of good i in region r, inclusive of all applicable tariffs, border costs and monopolistic markups. PY;, Supply price (marginal cost) of good i from region r, excluding fixed costs associated with the production of goods in industries subject to increasing returns. A-2 PVi, Price index for factor inputs (value-added) in sector i, region r. This price is gross of all applicable factor use taxes. PM, Aggregate price index for imports of good i into region r - applicable when region r has not been integrated. PEi, Aggregate price index for imports and domestic supplies of good i from all EC countries in an EC region r - applicable only when EC markets are integrated. PNi, Aggregate price index for imports of good i from non-EC countries within an EC region r - applicable only when EC markets are integrated. PFf, Price for factorf in region r. 1.3 Equilibrium Summary Final demand in each region arises from a single representative agent maximizing a Cobb-Douglas utility function subject to a budget constraint. Income consists of returns to primary factors (labor and capital) and tax revenue rebated to the consumer in a lump sum. Within each region, final and intermediate demand are expressed for the same Armington composite of domestic and imported varieties of individual goods. In the non-EC regions (and in the EC regions prior to market integration), the composite "supply" is a nested CES aggregate in which at the first level domestic varieties trade off with imported varieties while at the second level there is substitution between imports from different regions. After integration, within EC regions, the structure of this Armington aggregation changes to reflect improved substitutability between commodities from different EC member states. Production has ia standard structure with fixed coefficients for intermediate inputs and a CES aggregate of individual primary factors comprising value-added. A-3 Two types of costs are associated with inter-regional trade. First, we introduce "standards costs" through which a product variety produced in one country requires additional value-added in order to be sold in a second country. Second, we account for "borders costs" representing costs of transporting goods between different regions. Both of the costs are zero for domestic sales. Some of the commodities are produced subject to increasing returns to scale with free entry driving profits to zero. In this framework price-cost margins are determined by Cournot competition (with fixed conjectural variations). In equilibrium firms enter the market to the point that revenues from markups on marginal cost exactly balance the fixed costs of production. Changes in market share or elasticities of substitution in demand affect the optimal Cournot markup, so our modification of intra-EC elasticities (which would have not effect in a model of perfect competition) here causes significant changes in supply prices and market share. 2 Equations for the Central Model 2.1 Markets - Primal Form Equations (i) Regional output: Y-, Xi,,, (1) I,, where Yi, is output of good I in region r, X,,,. is exports of good i from region r to r'. (ii) Regional demand: Si, = + aYj, + T, a2) where Si, is total supply (production plus imports), C1, - is total final consumption, a,, is intermediate demand coefficient, and T,l is demand for good i in transport costs. A4 (iii) Value-added: Vj, - aj,',, + jA,, + fi, N., (3) ,:S,~~~~~~~~~I where <, is total sector i value-added, ay,, is value-added demand coefficient, a',rr' is the standards cost coefficient for shipments of good i from region r to region r., f, is the fixed cost per firm, and Ni, is the number of firms (for IRTS sectors). (iv) Primary factor markets: Ff E a& (4) where F, is the endowment of factorf in region r, and aF fiis the rrice-i-esponsive demand coefficient for factor in sector i. (v) Armington supply for non-integrated regions r: S,F ~ ~ ~ 'i H I, where Sj, is the benchmark supply, r?,r is the value-share of domestic supply X,,,, is benchmark exports of good . from region r to region r', OM,' is the benchmark value share of region r exports in region r' imports, and PDMI,PMZM are determined by the Armington elasticities of substitution, aDM and mm pm. a v- 1 A-5 (vi) Armington supply for integrated regions r: Si, o. f X~ + (1~aC 1 ': 6 I Ir'C E , IC fAfI, where aec is the benchmark value-share of EC imports, PC,. is the value share of imports from region r in all imports from EC countries, and 96/x. is the value share of imports from region r in all imports from non-EC countries. (vii) Value-added supply: F vi, = VP,x E iflr 4| where V., is benchmark value-added, aSi, is the benchmark value share of factorf, a, is the benchmark input coefficient, and p, is determined by the elasticity of substitution. (viii) Border/transport cost: O i-if where r is the index of the single commodity employed, for transport services, and is the transportation cost coefficient. A-6 (ix) Welfare index: w, ' . [ cl (9) where is benchmark final demand for good i, region r. 2.2 Profit conditions (i) Value-added: PVv, l J l (10) where is the ad-valorem factor tax rate PVs, is the benchmark (tax-inclusive) price (ii) Marginal cost: PYa, - aPVL. + EafPSj, (11) (iii) Armington composite supply price for non-integrated regions: PS,, - j a ( [- ,') | (12) PDI;;I' PM,,J in which: PDD - (O+A,,)PY,, (13) and PM,,- [ G[, (1 +lA)( + i)(PYw,+A,,PT,i+astPV,i)] | (14) and PT,, - PSI, (15) where is the markup on marginal cost on sales of good i from a firm in region r in region r.. A-7 is the ad-valorem tax rate which incorporates import tariffs, export subsidies and various non-tariff barriers, PDj, is the benchmark supply price for goods rrom domestic producers, PM;, is the benchmark supply price for imports. (iv) Armington composite supply price for integrated regions: { E PE1., 1 ~"rPN 1" (16) PS,, = + (I-cg,,) | | | PEi, PN J in which: ,E1, 4 ,EC [(I +Ai1,,)(1 + ,)(PY,+#,,.PT,+asPVJ] | (17) and PNM, |1. 9,, +(l ")(, 1+!,)(Pi+,,,PT,+apJPVir,)) (18) where PEj, is the benchmark supply price for imports from EC countries, and PN, is the benchmark supply price for imports from non-EC countries. A-8 (v) Regional income, including factor income, tax revenue and profits from monopolistic sectors: 1, - FfrPFf, + income from primary factors E 9,,, ii,,, (pY,,,+#,,,PT,+ai,,1PVi,,) X.,, + net tax revenue and NTB rents on imports S(1-6 ,) , (Prf,+f ,,PT,+as,,Pv,) X,., + i'd (19) net tax revenue and NTB rents on exports 4, PFf, aœr, V, + factor tax revenue £ E [ Al, (I +11 7) (Pi? + ,PT+ajPV.; X,, - PK. ji, Nj, monopoly markup revenue less fixed costs where 0!., is the share of trade taxes accruing to the importing country, r. In most calculations, free entry assures that monopoly profits are zero. (vi) Final demand: ci, -L, l, (20) (vii) Free entry zero profit condition for monopolistic firms: N [ F;,0 (I+t;,,,) (PAj,+#j,,fPTf +as V,) Xj (21) Nj, s ~~~~~PV,l f,l 3 Pricing Equations for Monopolistic Industries The following features characterize the model representation of industries producing under IRTS: A-9 * Goods are distinguished by firm, by region and by area of origin. An area may contain one or more regions. Within a given country, for example, an area of origin may be "imported" or "domestic1'. The first of these contains all regions other than the importing country itself. The second contains only a single region. * Demands arise from a nested Constant-Elasticity-of-Substitution (CES) structure. At the lowest level is a CES aggregate of supply from tirms in a single region r. At the next level, region r supply trades off with other regions from the same area, and at the top level, consumers choose between aggregate commodities from different areas. For example, in the EC following integration, consumers distinguish between EC and non-EC commodities. In all other countries as well as within the EC prior to integration, consumers only distinguish domestic and imported varieties. Demand for the final composite arises from consumer maximization of a Cobb-Douglas utility function, so that the Marshallian demand function exhibits a unitary price elasticity of demand. * Producers compete in quantities based on a Cournot model with fixed conjectural variations. Markups on over marginal cost are based on maximization of profit. Profits fall to zero, however, due to free entry. Fixed costs at the firm level are exogenous, and as markup revenues change, so too does the number of firms producing within a given region. Algebraic relations for supply of a single commodity to a single market will now be outlined. One such set of equations applies to each market for commodities produced subject to increasing returns to scale. The following notation applies: X Aggregate demand; Yk Supply from area k; S, Supply from region r; qf. Supply from firmf in region r; P Price index for aggregate demand; A-10 Pk Price index for supply from area k; w, Price index for supply from region r; and 7rf, Sales price for supply from firmf in region r. 3.1 CES Aggregators (primal form) X Ci [ ir a^"; (22) x. s [ ,s1/ S-] j:T(23) &'= [Eg' ] (24) 3.2 Associated price indices (CES cost functions) w, = (Fah p<)T (27) 3.3 Associated demand functions ~~~~ (28) [Pk (29) SI =F IP | Yk for k-k,. (29) ) A-1l 3.4 Flrm behavior Consider the optimization problem facing firm f in region r selling into a given market. Profit is given by: flr(q) - wlrq - Cf>(q) (31) First order conditions for profit maximization are: ar=,- air q + xft - c>,(q) (32) aq aq~ wu~q in which cf, represents the marginal cost of supply which is treated as exogenous in this derivation. The first order conditions may be written: cf,= (K-r,A,) (33) in which mr, is the markup over marginal cost (expressed on a gross basis): ,r,, 1 I _ at,, qf, (34) - 8.ef, awf, tr In this equation e,, is the "perceived elasticity of demand". The pricing equations presented here are easy to interpret, but they are somewhat more difficult to derive. The expression for e,, arises from the nested CES structure of demand taken together with the Couraot assumptions concerning the response of other producers. 3.5 The perceived elasticity of demand Begin with the inverse demand function: I= S w (35) We then compute the derivative: aw,, lw,, + " 5 as, + ,aW, (36) q,, f qM e S, qf, w, aqf, Under Cournot conjectures' I For simplicity, the equations presented here are based on unitary conjectural variations. Conjectures departing from one are a simple extension. A-12 as, s S,r (37) aQft lJqf' J dw, and the term aw' is computed by applying the chain rule a second time: aqfr dw, -w, as, (38) aq', as, aq,, Combining, we have: -If -- + --I-I + qfq S,] aw, (39) aqf,pfr, f e5 q, w, qf, as, Make the substitution: Iqf-1/, (40) to obtain: 1 1 r qft dw, S, 7rfrqf, (41) ef, e w,S, oS, w, w,S, Apply similar steps at the next level to obtain: Ow, 5, I w,S, +aPk Y W,S, (42) as, W,r n t1 7k Yk OYkPk PY Apply the operations again, taking into account the fact that demand elasticity for aggregate X is unity: aPk Yk I I PAY, PkYA - m -- PX- -~ (43) ay, Pk a a PX PX We then may assemble all of these equations to obtain an expression for the optimal Cournot markup. To simplify notation, we introduce: -t = Wp,S for k-k, (44) and =X ' pk (45) Px We then have: A-13 ah . -_ + _-_ (46) 3.6 Application to the EC model In our modelling we begin with two distinguished supply areas in each market: domestic and imports. There are three associated elasticities of substitution: ODD is the elasticity of substitution between goods supply by different domestic firms mm jis the elasticity of substitution between any two foreign supplies, whether or not they are firms within the same region, and ODM is the elasticity of substitution between goods from any pairs of domestic and foreign firms. Let r,,, denote the market share of region r firms in region r'. In the benchmark we have only domestic and imported firms distinguished. Applying the formulae above we therefore have two different markup equations (one for sales into the domestic market and the other for export sales): = ~~+fiiJ+[.iJzrr I ODD | aDM ODD | N, ODM| N, (47) I + + I- 1r laMM tODM umJ NO [, ODM] N, where e M 6,9 (48) F,,# In the counterfactual experiment which we interpret as "market integration," we modify the structure of demand for consumers within EC countries. In the modified functions, firms within the EC area are distinguished from non-EC firms. The elasticity aDD determines the trade-off between any pairs A-14 of EC firms (whether or not they are in the same country), while aDM determines substitution possibilities between EC and non-EC supplies. The elasticity of substitution between two non-EC firms remain equal to OMM- The mark-up equations tor firms selling into the EC market become: I + I - I -r + I - I | rE EC aDD aDU DD N,oEC OMD N, (49) + 0;."]N~'c + [i- ..~.rOEC Ut 5 OUM I Dm a. O is feasible for anyx x E T 2 Justification: 1: Implied by the definition of e() 2 Follows from the Cobb-Douglas structure of the aggregation function. 3 Follows from linear algebra. 4 From the definition of u^. A-17 A competitive equilibrium in this economy is characterized by a vector of market prices, w, and a vector of activity levels, y, which satisfy the following conditions: Market clearance: Ey xf + w; 2 d, j h A Zero profit: _rTXi 2 0, vJ Budget balax.ce: 7rT78 irTdh, vh In addition, the choices by consumers and producers must be independently rational. Hence, dl maximizes Uh(d) subject to budget balance, and x. maximizes unit profit (wTx) subject to technology constraints (x E7). The equilibrium is supported as a solution of the joint maximization problem: max W(U) = f UA(dA)& h S.t. (50) xi + wŽ 2 sd', vi A competitive equilibrium is represented when the Negishi weights are chosen so that 0b is proportional to irTwA. The SJM algorithm searches for 0b values by successive approximations in which th;e Lagrange multipliers for the previous iteration are used to form an estimate for the welfare function budget shares in the subsequent iteration. 4.3 Joint maximization in price space Let eb(r), the unit expenditure function for consumer h, be defined as above. Let oj(ir) be the analogous "unit revenue function" for sectorj, defined as: oj(r) a max 7rTx s.t. x E Ti Sheppard's lemma characterizes consumer demand and producer netputs: T dA(lr) = Ve,(7r) e ) eA(-r) A- 18 and xj(7r) - Vj(Ir) Using these relations and identifying the Lagrange multipliers as the activity vector y, it is straightforward to show that the Karush-Kuhn-Tucker conditions for the following nonlinear program correspond to a competitive equilibrium for the underlying economy: max 1 'A log(e,(ir)) - irrw, s.t. .(wr) s 0 vj The SJM iterations follow a sequence of maximization problems in which, for iteration k, C, = (-,, 4.4 Dual joint maximization with price distortions In the general equilibrium structure price distortions can, without loss of generality, be applied only to the supply side of the economy. Suppose that xi is chosen to solve max *Tx s.t. x E TJ in which * is a vector of tax-distorted prices (users costs), for example: *.i = iri(l+t~,). When ta" distortions are present, the tax revenue returned per unit operation of sectorj is given by (7r-*)Tx. In order to accommodate price distortions in the dual joint maximization procedure, the tax distortions are introduced into the constraints, and the tax revenue effects are treated symmetrically with factor endowments - using lagged values for production activities. The generic dual-form optimnization problem is: max E 4( 1. log(e,(w)) - irT w, - O e9j Vx (7- r) s.t. X,(*) 5 0 A-19 in which qh is the fraction of sectorj tax revenue which accrues to household h and where Y and £are lagged values for the sectorj activity level and netput vector, respectively. 4.5 Dual joint maximization formulation for MRT The MRT model is formulated as a joint maximization problem in the space of prices. In every major iteration, the following nonlinear programming problem to refine estimates of prices and quantities (activity levels and trade flows are determined by the value of Lagrange multipliers on the corresponding zero profit constraints): maxE 1, log |lPS:' J - s.t. (10)-(18) In this calculation Ir(X, V) stands for what appears on the right-hand-side of equation (19), evaluated using the current estimates for trade flows and value-added, JZ,,, and t,. For purposes of determining sectoral prices and incomes, monopoly markup rates (i,,,.) are held fixed. A consistent equilibrium 's obtained through a recursive procedure in which calculations of an equilibrium for the central model alternate with single sector simulations for individual markets to determine optimal markups and numbers of active firms. 5 Single commodity submodel for monopolistic sectors The equi ibrium conditions for markets where technologies exhibit increasing returns to scale and firms price above marginal cost do not fit into a Negishi-type computational framework. We have therefore adopted a decomposition procedure to overcome this problem. In our computations, separate systems of nonlinear equations generate consistent approximations of market shares, numbers of firms and markups over marginal cost for each commodity. Demands and supplies for all regions are included A-20 in these systems of equations, although factor markets, income effects and intersectoral linkages are ignored. In every iteration, regional demand functions are calibrated to the most recently calculated general equilibrium solution. The single commodity models are formulated with marginal costs assumed constant. This seems appropriate given the degree of disaggregation (no monopolistic sector commands more than 2% of aggregate expenditure or value-added). Given constant marginal costs, the selling price is then determined solely through the markup equations. The single commodity models include four classes of equations: * Inverse demand functions which translate markups and exogenous marginal costs into sales prices. * Equations relating the number of firms to sales volumes, trade flows and (exogenously specified) fixed costs. These equations embody the assumption that free entry drives profits to zero. * Markup equations, as derived in section 3, which define markups over marginal cost as a function of market share and iumbers of firms. * Ordinary demand functions which translate sales prices into trade flows. A-21 Appendix B: Calibrating the Cost Disadvantage Ratio We assume that total costs at the level of the representative firm may be specified as: c = f + mq wheref is fixed costs, m is constant marginal costs, and q is firm level output. Average costs are then: ac f +m q Assuming zero profits, data in the initial equilibrium provide information on the industry total costs (C) and industry output (Q). If there are n representative firms initially, then nc, - C, and nq, = Ql, where the subscript I refers to the data of the initial equilibrium. Since c' ncfl C, q, nq, Q' the initial data provides one point on the firm's average cost curve: c,, f + m. q, q, Given the specified functional form of the average cost curve, it may be calibrated if a second point is known. Suppose we have an estimate that if output declines to aq, then average costs increase to ( I) where 0 < a < 1, , > I and where ia < I is required for marginal costs to be non- negative. This provides a second point on the industry average cost curve: C f + m. q, atq, Multiplying numerator and denominator in the last two equations by n, we obtain equations involving industry output and costs, on which data is available. Thus we have: C, F Q, Q, and B-l Q , r where F is industr-y fixed costs. Solving these equations for industry fixed cost and industry marginal cost yields: F =Cl 1 - I) - t and Since the cost disadvantage ratio (CDR) is defined as which by symmetry equals F we c C know that at the initial equilibrium: CDR = ($-1)cr i -'i We obtain the values of a and a primarily from engineering studies summarized by Pratten [1987], a synposis of which is available in Emerson et al. [ 1988]. We enter data for these two parameters at the 44-sector level corresponding to the sectors in the EC input-output tables, and then aggregate to the sectors of our model. The mapping and sources of our CDR estimates are provided in Table B-i. The final, aggregated, CDR estimates used in the model are listed in Table I in the text. Following others such as Gasoriek, Smith and Venables [19921, we assume that the initial output level is at the minimum efficient scale (MES). This is not an unreasonable assumption, since firms should have difficulty competing at less than MES. However, given our assumed functional form, at MF.S further cost savings can be obtained by expanding output. To the extent that output is less than MES, we have indeed underestimated the CDR, since the slope of the average cost curve increases in absolute value for decreases in output. B-2 Table B-i: Source of Data on CDR Values Sec¶r Sh of M3 * Perceg Coti . 0W4" (a) lecrima at OuVut lfpbli CDR S