2 Trade liberalization and the industrial structure of production

View the most recent version.

Canadian economists have long argued that tariff protection resulted in plant sizes and production runs within Canadian manufacturing plants that were too short to fully exploit economies of larger scale production. Safarian's (1966, 7) pioneering survey on the relative costs of foreign multinationals operating in Canada reported that most foreign affiliates operating in Canada had higher unit costs than parent companies' plants located in the United States. These higher costs were attributed to a variety of sources; but shorter production runs was the most common response for those reporting higher unit costs.

It is the effect of tariffs on these structural characteristics that provided the foundation for early analyses of the potential consequences of free trade between Canada and the United States (Wonnacott and Wonnacott 1967) and proved to be one of the most important drivers of increases in welfare predicted to result from free trade (Harris 1984; Cox 1994).

Falling trade barriers were seen to lead to longer production runs, either through increases in the size of plants or through the production of fewer varieties of products within plants. Empirical analyses of Canadian manufacturing plants suggested Canadian plants suffered from both suboptimal plant size and excessive production line diversity (Daly, Keys and Spence 1968; Caves 1975), and that falling barriers to trade would result in larger plant sizes (Economic Council of Canada 1967, 1975; Royal Commission on Corporate Concentration 1978). Earlier work by Baldwin and Gorecki (1986) examined the extent to which suboptimal plant scale in Canada was a factor behind Canada–United States productivity differences.

Output diversity in Canadian plants arises not from scope economies, but despite scope diseconomies when the Canadian manufacturing sector is modelled as an import-competing sector serving a small national market (Eastman and Stykolt 1967). Explanations for an industrial structure that produces plants and production runs of suboptimal size have relied on models treating manufacturing industries as containing a small (oligopolistic) number of producers charging a price pegged to the tariff-ridden price of imports. This number of sellers is represented as a free-entry equilibrium, where each (identical) incumbent earns normal profits, or perhaps a bit more, but the entry of one more firm would make all of them unprofitable. Each firm (plant) is treated as facing a downward-sloping demand curve and produces an output that does not exhaust available scale economies. When scale economies are not fully exploited, firms have the incentive to add product lines, even when there are scope diseconomies, if the reduced costs, due to the exploitation of plant-scale economies, offset the disadvantage of scope diseconomies.

In this world, the predicted effect of trade liberalization on firms' diversification choices depends on how competition is modelled. The effect of size on the organization of production depends on firms competition, product differentiation, the supply of potential entrants and other factors. However, under reasonably general conditions, the enlargement of the market should be expected to induce some combination of increased output per firm and increased numbers of competitors, accompanied by a lowered equilibrium price. Given the assumed structure of costs, this change reduces the firm's incentive to pack diversifying products into a plant in order to spread plant fixed costs. As a result, diversification falls. Together, increased plant size and reduced diversification lead to longer production runs.

Baldwin, Caves and Gu (2005) consider the effect of trade policy in this world—one with plant diversification in a market with pervasive product differentiation, resulting in each producer facing a downward-sloping demand curve. The production of each supplier is assumed to be subject to scale economies, and costs curves differ with regards to the level of average costs and the extent of plant scale economies (scale economies for plants with the industry's output as their principal product). Imports and potential imports are similarly differentiated and supplied by price-setting producers. A Nash equilibrium is assumed with each producer (and importer) taking its rivals' prices as given.

In this world, the high protection experienced by Canadian manufacturing industries removes importable varieties from the market, lessening the substitution possibilities that face the typical domestic producer, lowering the elasticity of its individual demand curve and raising the average domestic price. As a result, more entry of domestic producers occurs and, under certain conditions, the typical domestic producer's equilibrium output shrinks and the incentive to pack the domestic plant with diversifying outputs intensifies.

In this case, unilateral tariff reductions by Canada should reduce plant diversity. Simultaneous reductions in U.S. tariffs reinforce this tendency. In this model, when a small country's (e.g., Canada) producers gain access to external markets in which prices now exceed their marginal costs, they are likely to face highly elastic demand curves, thanks to large markets for their exports. They are also likely to select large plant scales that remove the incentive for plant diversification. Other domestic producers with high costs that deny them access to exporting either shut down or expand their production for the domestic market (if the elasticity of the demand that they face has increased). Either way, their actions contribute to a reduction of diversity for the industry's average plant.

Recent analytical work by Baldwin and Gu (2008) develops an extension of this approach using a world of monopolistic competition and heterogeneous producers. This model combines themes regarding product differentiation found in Ottaviano and Thisse (1999) to build on Melitz's (2003) model of heterogeneous single-product plants. It departs from the latter by assuming that firms have multiple products.

Consumers' preferences are represented by a quasi-linear utility function that is defined over a continuum of differentiated varieties. The model assumes that production exhibits economies of scale within varieties and economies of scope across varieties. To enter the differentiated product sector, a firm must bear fixed costs of entry E, regardless of the size of its product range, thus implying that economies of scope are present.

An entrant learns about the marginal cost of the production of a variety after commencement of operations. In this world, there are a finite number of multiproduct firms and each firm controls a non-negligible set of varieties. As such, firms are assumed to behave like oligopolists. When choosing its product range and the length of production runs, a firm no longer neglects its impact on the market as in monopolistic competition models of trade.¹ The firm must account for the impact of its choice on the demand for its varieties through its effect on total market demand. Using a two-stage game, a firm chooses its product range Ωi in the first stage and then the quantity and price of its varieties q_i and p_i in the second stage.

A free-market zero-profits equilibrium determines the number of firms, the variety produced and the average length of a production run. Compared with an average firm in a smaller market, the one in a larger market supplies a larger number of varieties (with a higher degree of product diversification). It has a longer production run and sets a lower price for its product varieties. It is larger, more productive and has higher profits. There are more product varieties and more firms in a larger market.

The analytical model also provides intuitive results on the impact of scale and scope economies on product diversification, length of production run, firm size and firm profits. The existence of strong scale economies within individual products is related to higher product specialization, longer production runs and higher profits. However, it has no effect on firm size and productivity. The existence of strong scope economies at the firm level is related to higher product diversification, larger firm size, lower productivity and higher profits. But it has no effect on the lengths of production runs for individual products.

The model can also be used to generate predictions about the nature of the free-entry equilibrium when trade with other regions is permitted. Firms in this world not only have to decide on offerings for the domestic market but they also have to decide whether they will serve the export market. In this case, firms will divide into non-exporters and exporters—with the more cost- efficient serving both the domestic and the export market. These firms differ in that non- exporters have higher unit costs and their response to declines in profits can be different from that of exporters, who have lower unit costs and have more product diversity than do non- exporters. Exporters also respond differently to tariff cuts. Tariffs are incurred to penetrate foreign markets and consist of taxes plus transportation costs. The latter are larger for outlying regions. The overall impact of tariff reductions is gauged by the average change coming from the two different segments—exporters and non-exporters.

The model generates a number of testable implications concerning how the number of products, product diversification, plant size and the length of production runs should change when trade liberalization occurs.

Hypothesis 1: Trade liberalization should lead to a decline in the number of products supplied by individual plants. This decline will be larger for plants with larger home markets and lower transportation costs to access export markets.

For non-exporters, trade liberalization will have a similar effect on measures of the number of products that take into account the distribution of shares across products—such as the entropy measure used here. It has an ambiguous effect on the measure of existing and new exporters.

Hypothesis 2: Trade liberalization has an ambiguous effect on overall plant size since it reduces the size of non-exporters and has an ambiguous effect on the size of current and new exporters. But if the effect is positive, it should be higher for plants with larger home markets and with lower transportation costs to access foreign markets. It should be noted that Baldwin and Gu (2008) find the impact at the national level for exporters to be positive.

Hypothesis 3: Trade liberalization increases the production-run lengths of individual products at existing exporters and new exporters and has no effect on the length of production runs for non- exporters. On average, the length of production runs should increase and it should be higher for plants with larger home markets and lower transportation costs associated with exporting for foreign markets.

Consistent with their analytical predictions, Baldwin and Gu's empirical analysis (2008) shows that declining tariff barriers resulted in increased production-run lengths for Canada as a whole as a result of fewer products being produced within plants.

To summarize, the analytical and empirical literature suggests that smaller regions and those facing higher tariffs will have fewer producers, that they will have less diversification and lower production-run lengths than those facing lower tariffs and with larger populations. They also predict that trade integration will result in increasing specialization and longer production runs. The effect of trade on plant size is more ambiguous. Regardless of the type of adjustment— product diversity, plant size or production-run length—the effects of trade integration will be stronger in those places with larger home markets and those located closer to export markets and, therefore, will incur lower transportation costs.

2.1 Implications

Various analytical models described previously help to inform our examination of differences in plant size and specialization across Canadian regions and the impact of trade liberalization on them. These regions are Atlantic Canada, Quebec, Ontario and Western Canada. These four regions differ substantially from one another in terms of population (Table 1); Ontario had over 11 million inhabitants in 1999, while the smallest region had around 2 million. Over the study period, Ontario and Western Canada grew by over 3 million, Quebec by only about 1 million and Atlantic Canada by less than 200,000.

The regions also differ in terms of their proximity to the focal point of market potential in North America—the U.S. manufacturing belt. Ontario abuts this geographic area that stretches from the mid-Atlantic States to the Midwest, while Atlantic and Western Canada are on the margins of the North American market. Quebec, while not as centrally located as Ontario, is better positioned than Atlantic and Western Canada within the North American market. Transportation costs from these regions act as a type of tariff on products that are exported to this market.

Table 1
Regional population estimates

Declining barriers to trade affect regions differently because tariffs are higher for outlying regions, due to higher transportation costs. The model predicts small changes in industrial structure for those regions facing higher tariffs, which consist of taxes plus transportation costs. The reaction of producers in a market to reductions in tariffs depends upon the density of those markets and the distance a region is from its export markets.

Distance to market acts as a tariff on the exports of a region. Empirical studies that model the volume of trade between countries or regions (Frankel, Wei and Stein 1997) have consistently found distances to be one of the most important determinants of the volume of trade. For instance, Brown and Anderson (2002) find the elasticity on distance is around one for the majority of two-digit manufacturing export sectors for interstate and province-state trade in North America. For a continental economy like Canada's, distance has a non-trivial effect on market access and, therefore, location relative to the broader North American market may help to determine how producers react to falling barriers to trade.

To understand this phenomenon, it is helpful to contrast the relative locations of Ontario and Western Canada with respect to the U.S. market. Ontario is within one day's trucking of 100 million people in the United States. There is no location in Western Canada that is comparable in terms of its market access. Therefore, even though both regions may experience the same absolute decline in tariff barriers, this decline serves to increase the total market faced by Ontario firms by far more than it does for firms located in Western Canada. This implies that, ceteris paribus, falling tariff barriers from the North American Free Trade Agreement will have a stronger effect on producers in Ontario, specifically Southern Ontario, than in Western Canada.

Placing these results within a geographic context implies the effects of growing integration will vary, depending on the location of a region relative to larger continental markets. These effects should be felt most in those regions that are at the nexus of the North American market. In practical terms, this means producers in Ontario and possibly Quebec will change how they organize their production the most, while Atlantic Canada and Western Canada will experience more muted gains. In the section below, we test these expectations.

 

¹ In monopolistic competition models of trade in differentiated products, each firm produces one variety because there are no economies of scope across varieties. In these models, each firm correctly neglects its impact on the market.