Canada's Agricultural Trade in North America: Do National Borders Matter?

The Canada—United States Trade Agreement (1989) or CUSTA, and the Uruguay Round Agreement on Agriculture or URAA (1995) are the two major trade agreements affecting the level of agricultural trade between Canada and the United States.¹ The CUSTA removed most of the tariff and nontariff barriers on agricultural trade between the two countries. Some of the more contentious trade issues, such as the operation of the supply-managed system and the Canadian Wheat Board, were left to the URAA negotiations. Barichello et al. reviewed the effect of CUSTA on agricultural trade between the United States and Canada and concluded that trade was far from frictionless, even though many tariffs were lowered and some quantitative restrictions removed.

The CUSTA went into effect on January 1, 1989, and was aimed at increasing trade between Canada and the United States by removing tariff and nontariff barriers. CUSTA did not result in a set of harmonized external tariffs for Canada and the United States and thus created a Free Trade Area (FTA) rather than a Customs Union or a Common Market. Those sectors deemed able to compete in the FTA had their tariff protection removed immediately while others had it eliminated over a five- or ten-year period. Trade restrictions were left on some products such as sugar, dairy, and poultry. The CUSTA allowed for a snapback provision on fruits and vegetables in the form of a price-based tariff for a period of twenty years. The reduction in tariff and nontariff barriers was to lead to greater integration and reduced home consumption bias in the two economies. CUSTA's provisions were fully implemented as of January 1, 1998.

The North American Free Trade Agreement (1994) or NAFTA replaced the CUSTA to include Canada, Mexico, and the United States. The NAFTA will eliminate most tariff barriers and many nontariff barriers affecting U.S.—Mexico and Mexico—Canada trade over fourteen years, ending on January 1, 2008. The NAFTA did not make additional tariff reductions to U.S.—Canada agricultural and food trade (Appleton).

The URAA (1995) is the second trade agreement affecting the flow of agricultural goods and services between the United States and Canada. The URAA (1995) is a derivative of the General Agreement of Tariffs and Trade or GATT (1994). The multilateral GATT (1994) covers a wide range of goods and services but excludes most trade in agriculture products. With the URAA (1995), agriculture was partially brought under the discipline of the GATT (1994). The World Trade Organization (WTO) was created on January 1, 1995, by the member nations of the GATT. The function of the WTO is to deal with the rules of trade between nations. The debate over a further multilateral reduction in agricultural trade barriers will occur within the WTO.

The URAA (1995) reduced tariffs on agricultural products and the level of trade-distorting subsidies, particularly those deemed to be export subsidies. It also removed the quantitative restrictions on agricultural trade allowed under Article 11(2)c of the GATT (1948). This Article protected the Canadian supply-managed system from imports. Article 11(2)c was replaced with a system of tariff—rate quotas (Skully). Josling, Tangermann, and Warley found that the URAA (1995) only partially liberalized agricultural trade because it left some restrictive border measures in place, such as tariffs on dairy products.

Notwithstanding the trade agreements between Canada, Mexico, and the United States, a home consumption bias in Canada remains. McCallum found substantial bias in Canada when examining the level of aggregate merchandise trade between the United States and Canada for 1988. Using the 1993 U.S. Commodity Flow Survey (CFS), Hillberry tested the robustness of McCallum's finding. Hillberry states “…the CFS estimates of the “border effect” are almost exactly equivalent to those estimated by McCallum” (p. 1). In 1998, Helliwell (1998) reported that the home consumption bias in Canada had declined since the CUSTA, but it remains larger than what trade theory would predict. Anderson and van Wincoop suggest that substantial gains in trade and welfare are attainable from a further reduction in North American tariff and nontariff barriers. While they did not explicitly model trade in agricultural commodities, they report that large trade restrictions still exist at the Canada—U.S. border.

For 1984 and 1989, Cyrus reports that tariff and nontariff barriers have large Canada—U.S. border effects for most agricultural commodities. While the CUSTA and the URAA lowered tariff and nontariff barriers, trade friction remains an issue at the Canada—U.S border. The recent action of the Ranchers Cattlemen Action Legal Foundation (RCALF), the U.S.—Canada wheat disputes, and the tomato wars are three examples of the trade friction between the two countries. In 2000, RCALF brought both an antidumping and countervailing suit against Canadian live cattle imports into the U.S. market. While no permanent trade barrier resulted from this case, it demonstrates the pressures that lobby groups place on governments to restrict agricultural trade across the U.S.—Canada border. To date, no one has measured the Canadian home consumption bias in agricultural trade with Mexico.

This article provides an estimate of the border effect for agricultural trade between the United States and Canada, and Mexico and Canada in 1992–98. It is important to know how restricted agricultural trade is between countries and thus the potential welfare gains from further trade liberalization.

Tariff barriers, while the most visible, are not the only source of trade restriction. Nontariff barriers are also important and need to be reduced so that goods and services can move efficiently across national boundaries. We report large border effects for trade in some agricultural goods, such as dairy products, between Canada and the United States. A surprising result is that the Canadian border effects with Mexico are much lower than with the United States.

A second issue examined in this article is the symmetry of the border effect between agricultural imports into Canada and agricultural exports from Canada. This symmetry is examined for both Canadian—American and Canadian—Mexican trade. U.S. producers claim that Canada maintains agricultural marketing boards, which promote exports and restrict imports. Our conclusion is that the Canada—U.S. border effects are more pronounced for imports than exports (i.e., the two border effects are not symmetric in magnitude). Symmetry is more likely in the case of Canadian—Mexican agricultural trade.

We provide a short overview of Canada—U.S. and Canada—Mexico agricultural trade, followed by a brief discussion of the gravity model and the data used to estimate the gravity equation. The econometric results are presented along with a discussion of the estimates. We compare our results with the work by Cyrus, and examine how the border effect has changed for agricultural trade since the introduction of the CUSTA and NAFTA.

Agricultural Trade Patterns

Since 1992, the United States has increased its market share of Canadian agricultural imports and exports (table 1). Grain is the one Canadian agricultural export for which the United States is not the primary market. This is not surprising given that both the United States and Canada are major grain exporters. Canada does import a small amount of grain, mostly corn and rice, and the United States is the principal supplier.

With such crops as citrus and winter vegetables, California and Florida are the largest exporters of agricultural products to Canada. Massachusetts is the destination for the largest share of Canadian agricultural exports. In Canada, Ontario is the largest province in terms of agricultural imports and exports. On a per capita basis, however, the Prairie Provinces are the largest exporters of agricultural commodities.

Inter-provincial agricultural trade over 1992–98, was larger than agricultural trade with the United States, averaging Cdn$17 billion per year versus Cdn$14 to the United States (Statistics Canada). This is surprising given the perceived wisdom that north—south trade is the most important to Canadian farmers and processors. Most of the Canadian provinces produce similar agricultural products for export, such as meat and grains, and import products like citrus, nuts, and winter vegetables.

Canadian—Mexican agricultural trade was much smaller than Canadian—American agricultural trade (table 2). Canada exports a small quantity of grain and oilseeds to Mexico and imports agricultural products such as winter vegetables from Mexico. Since the NAFTA, Canadian imports of fruits and vegetables from Mexico have doubled in percentage terms,² highlighting the importance of examining the agriculture trade flows between Canada and Mexico.

Gravity Trade Model

Tinbergen was the first to use the gravity model to explain international trade flows. However, it was Anderson that developed a solid theoretical foundation for the gravity model. Feenstra, Markusen, and Rose provide an excellent review of the theoretical foundations of the gravity model. The gravity model has been used to measure the impact of policy changes on trade flows including regional trade agreements (Bayoumi and Eichengreen), the economic effects of open versus closed trade blocs (Wei and Frankel), the relationship between increased trade and economic growth (Frankel and Romer), and the impact of uncertainty on trade and home bias (Anderson and Marcouiller). The gravity model has also been used to estimate the home bias in the European Union countries (Nitsch) as well as between Canada and the United States (McCallum; Helliwell (1998); Stein and Weinhold). However, there remains a concern with how well the gravity model explains trade flows when monopolistic competition exists (Hummels and Levinsohn); we do not investigate this topic.

In the simplest form of the gravity model, the quantity of trade s_ij between two regions i and j is a function of the size of the importing region i (GDP_i), the exporting region j (GDP_j), and the distance between the two markets, DIST. The model is represented as follows:

(1)

where the shipments between the two regions and the size of the importing and exporting region are measured in value terms (in this case, dollars). The distance variable is a proxy for the cost of transportation between the markets. When examining Canadian inter-provincial trade, Helliwell and Verdier show that the gravity model may underestimate the size of the border effect for small regions. This concern also arises when we consider trade between Canadian provinces and U.S. states, which are often of very different economic size. The error term is assumed to be normally distributed with a constant variance (Helliwell 1998).

The border effect can be estimated using the above model by incorporating a dummy variable, labeled BORDER as:

(2)

The border variable takes on a value of 0 when trade occurs within the home country and 1 otherwise. If the border has no effect on trade flows, β₄ will not be significantly different from 0. The anti-log of β₄ is an estimate of the border effect.

Home consumption bias in Canada can occur because imports into Canada and/or exports from Canada are restricted at the border. Canadian tariff and nontariff barriers restrict imports. Exports are restricted by other-country tariff and nontariff measures. To determine which of these two effects is most persuasive, we estimate equation (3), written as:

(3)

where the variables I and E are dummy variables for imports into Canada and exports from Canada, respectively. If in equation (3), β₄=β₅, then the border effect is symmetric for both imports and exports; if not, we have an estimate of which direction trade is most restricted.

Data

The majority of the data (all data observations are annual) for this study were obtained from different sources within Statistics Canada. The first major data set, inter-provincial trade flows, was obtained from the Input-Output Division of Statistics Canada. This is a unique data set because no other country keeps track of trade among its own regions. The data consist of input—output matrices of trade flows among the provinces and the rest of the world for a wide range of product groups, including agriculture.

Statistics Canada removed all transshipments within Canada from the input—output tables, i.e., wheat destined for Japan (through the Port of Vancouver) from Manitoba does not show up as wheat exported from Manitoba to British Columbia. Four categories of agricultural trade were available from the inter-provincial data set for 1992–98 and were used in this study. They are: (a) grains; (b) meat, fish, and dairy; (c) fruits, vegetables, and other food; and (e) other agriculture (see table 1 footnote for an explanation of the categories).

Only annual input—output data are available so we cannot break out seasonal trade effects for such commodities as fresh vegetables and fruit. This level of aggregation does not allow a completely clean measure of the border effect for some products. For example, processed meat and dairy trade data are aggregated by Statistics Canada, which makes an accurate measure of the border effects for these commodities nearly impossible. On occasion, Statistics Canada omitted entries within a particular matrix because there were too few observations (confidentiality) or the trade flows were near zero. In order to maintain as large a data set as possible, on the advice of Statistics Canada, the RAS method³ (Stone) was used to fill in these values.

The second major data set consisted of the value of trade among each of the Canadian provinces and each of the 30 selected American states, and Mexico.⁴ These data were obtained using software from Statistics Canada.

The data set for predicting the border effect between the United States and Canada for each of the four category groups contained 9 (trade flows) × 10 (provinces) = 90 observations provincially,⁵ plus 10 (provinces) × 30 (U.S. states) × 2 (trade flows) = 600⁶ observations for U.S. state data. The data set used for predicting the border effect between Mexico and Canada contained 9 (trade flows) × 10 (provinces) = 90 observations provincially, plus 1 (country) × 10 (provinces) × 2 (trade flows) = 20 observations for Mexico. This was not always the case though, as will be shown in the estimated regressions because cases of zero trade with both the United States and Mexico were omitted.⁷

Distance data for inter-provincial trade and province—state trade were calculated using the great circle method, which is the weighted average of the latitudes and longitudes of the three most populous cities (Cyrus). This was done to create an economic center for each of the regions.⁸

The final data set, gross domestic product (GDP), was obtained from Statistics Canada's publication (Provincial Economic Accounts 1998 Annual Estimates) for the Canadian provinces. The data for the individual states of the United States were obtained from the United States Department of Commerce's (Bureau of Economic Analysis) online publication. Mexican GDP data were obtained from the International Monetary Fund's CD-ROM, (International Financial Statistics 1999). All the data collected in value terms were converted to Canadian dollars using official exchange rates and deflated to 1992 values.

Econometric Results

Conclusions

The results of this study show that the border effects for agricultural trade between the United States and Canada and between Mexico and Canada have declined over the period 1992–98. Canadian agricultural trade has a lesser home consumption bias than in 1984, with the highly protected supply-managed sector being the one major exception.

The first surprising result is that the border effect for agricultural trade between Canada and the United States is larger than between Canada and Mexico. More work is needed to explain this difference. As mentioned earlier, one explanation may be the rent-seeking behavior of the lobby groups and the politicians on both sides of the United States—Canadian border.

A second surprising result is that Canadian exports move across the U.S.—Canada border more easily than imports into the Canadian market from the United States. This result suggests that Canada has nontariffs barriers to restrict trade with the United States. The state trading marketing boards are the prime suspects. Both the United States and Canada have effective domestic industry groups that lobby politicians for economic protection. This economic protection often comes in the form of border measures that reduce trade.

The Canadian border is still important. Agricultural trade with the United States and Mexico is by no means free and home consumption bias does exist in Canada. It is not surprising that NAFTA did significantly reduce the border effect for agricultural trade. The critics who claim NAFTA did not increase agricultural trade appear to be wrong. Progress has been made toward freer agricultural trade in North America but there are still miles to go before the market is a seamless trading entity.