November 1998 SCSB# 390

TRADE, POLICY, AND COMPETITION:
FORCES SHAPING AMERICAN AGRICULTURE PROCEEDINGS


Chapter 8
The Canada–U.S. Free Trade Agreement:
Competitive Tradeoffs Between Foreign Direct Investment and Trade


Ravi Munirathinam, Mary A. Marchant, and Michael R. Reed

Introduction

The Canada–U.S. Free Trade Agreement (CUSTA) was implemented on January 1, 1989, creating the world's largest free-trading bloc between the world's largest trading partners (Dixit and Roningen 1989, pp. 1023-33). This trade agreement also helped in formulating the North American Free Trade Agreement (NAFTA), which went into effect in January 1, 1994. According to CUSTA, all agricultural tariffs between the United States and Canada would be phased out over a 10-year period, in addition to improved market access for products from both countries and limited use of subsidies.

Since its enactment, CUSTA impacted U.S. and Canadian agriculture in different ways. For example, even though total agricultural exports increased dramatically in both countries, their export market share changed. Agricultural exporters in the United States increased their export share of consumer oriented processed food products to Canada, while Canadian exporters shifted their exports to the United States toward bulk and intermediate products. Additionally, U.S. direct foreign investment in Canadian food processing industries increased from $2.0 billion to $3.6 billion between 1989 and 1995. During this same period, Canadian investment in U.S. food processing industries increased from $894 million to $5.4 billion in 1994 (all values in this paper are reported in current U.S. dollars).

The primary objective of this research is to analyze the impact of CUSTA on U.S. and Canadian trade and foreign investment in the food processing industry. Attention will focus on activities of U.S. multinationals in Canada and Canadian multinationals in the United States. This paper first provides an overview of the effects of CUSTA. Second, it provides an empirical analysis of specific effects of CUSTA on U.S. exports as well as sales of U.S. food processing affiliates, operating in Canada. Third, it presents a similar analysis for Canadian exports. Analyses of other relationships are in progress, specifically exports of other types of agricultural products (bulk and intermediate), and the behavior of Canadian food processing affiliates operating in the United States.

U.S. Exports and Investment in Canada

Following the enactment of CUSTA, two way trade between the United States and Canada increased substantially. Total agricultural exports from the United States to Canada increased from $2.2 billion in 1989 to $5.7 billion in 1995, an increase of 159 percent (Table 8.1 and Figure 8.1). Examination of trade composition shows that between 1989 and 1995 exports of bulk-, intermediate-, and consumer-oriented processed food products from the United States to Canada increased by 54 percent, 80 percent, and 215 percent, respectively.

FIGURE 8.1 U.S. EXPORTS OF AGRICULTURAL PRODUCTS TO CANADA
 

The United States, which exports primarily bulk commodities with low value added, has fallen behind the rest of the world in exporting (consumer-oriented) processed food products with high value added. In this sector, growth in world trade greatly exceeds growth of U.S. exports. This decline in U.S. market share is partially explained by the tendency of large U.S. firms to invest in foreign countries rather than export (Bredhal, Abbott, and Reed 1995).

In contrast, U.S. exports of processed foods to Canada present a different picture. Figure 8.2 shows that the U.S. export share for consumer-oriented commodities to Canada increased from 61 percent in 1989 to 74 percent in 1995. During the same period, the export share for bulk commodities to Canada declined from 12 percent to 7 percent. Similarly, the share for intermediate products declined from 27 percent to 19 percent.

FIGURE 8.2 U.S. EXPORTS OF AGRICULTURAL COMMODITIES TO CANADA
 

Looking at investment, the stock of the U.S. investment in Canadian food industries increased from $2.02 billion in 1989 to $3.6 billion in 1995, an increase of 78 percent. Figure 8.3 compares U.S. investment in Canada, sales of Canadian affiliates of U.S. food processing firms, and U.S. exports of consumer-oriented processed food products between 1982 and 1993. All categories increased with the greatest changes occurring in the export and affiliate sales categories. For example, U.S. exports of consumer-oriented products increased from $1.1 billion to $3.9 billion (a 256 percent increase) during this time period, while sales of Canadian affiliates of U.S. firms increased from $5.2 billion in 1982 to $8.5 billion in 1993, an increase of 64 percent. Trade theory suggests that with market liberalization and free trade agreements, multinational firms may choose to export to foreign countries (in this case, Canada), rather than invest in processing facilities. Our empirical analysis will examine whether these strategies compete with or complement each other.

FIGURE 8.3 PERFORMANCE OF U.S. FOOD CORPORATIONS IN CANADA
 

Canadian Exports and Investment in the United States

Canadian agricultural exports to the United States increased from $2.9 billion in 1989 to $5.5 billion in 1995, an increase of 91 percent (Table 8.2 and Figure 8.4). Examination of trade composition shows that Canadian exports of bulk-, intermediate-, and consumer-oriented products to the United States increased by 110 percent, 103 percent, and 77 percent, respectively. Figure 8.5 shows that the Canadian export share of consumer oriented processed foods imported by the United States declined from 52 percent in 1989 to 48 percent in 1995. Concurrently, Canadian exports of bulk and intermediate products to the United States increased marginally.

FIGURE 8.4 CANADIAN EXPORTS OF AGRICULTURAL PRODUCTS TO THE U.S.
 


FIGURE 8.5 CANADIAN EXPORTS OF AGRICULTURAL COMMODITIES TO THE U.S.
 

Canadian investment in U.S. food industries increased from $894 million in 1989 to $5.4 billion in 1994. The sales of Canadian food processing affiliates operating in the United States grew steadily throughout the 1980s and were particularly strong from 1988 to 1990–peaking at more than $6 billion in 1990 (Figure 8.6). However, these sales fell to $5.21 billion in 1993 (data were not available for 1991-1993 due to disclosure problems). It is during this later period that Canadian exports of bulk-, intermediate-, and consumer-oriented products increased (Figure 8.4).

FIGURE 8.6 PERFORMANCE OF CANADIAN FOOD CORPORTATIONS IN THE U.S.
 

Notice that the time-related effects of CUSTA appear to differ between the United States and Canada. There was an immediate dramatic increase in U.S. agricultural exports to Canada in 1989 (the first year of the CUSTA), especially for intermediate and consumer oriented products (Figures 8.1 and 8.3). Yet, for Canada, there appears to be a two-year lag before exports were positively affected–Canadian exports to the United States didn't increase appreciably until 1991–and the increase was more gradual (Figures 8.4 and 8.6). Additionally, a comparison of Figures 8.3 and 8.6 shows that sales by foreign affiliates are the largest component for both the United States and Canada, especially relative to exports. Thus, firms appear to be using both trade and foreign investment strategies to develop foreign markets.

Literature Review

According to Malanoski (1994), international trade agreements such as the General Agreement on Tariffs and Trade (GATT) and NAFTA, which lower tariffs and other trade barriers, will have a limited influence on U.S. food manufacturers' decision to locate production facilities abroad. Malanoski states that the major factors which motivate corporations to move production facilities abroad are delivery costs, availability of raw ingredients, and tailoring products to foreign needs; tariffs and labor costs are not usually important factors. Labor cost results are corroborated empirically by Reed and Ning (1996).

Ruppel and Harris (1993) found that U.S. firms are directly investing in overseas processed food plants. However, leading U.S. multinational food processors are clearly expanding U.S. exports, even as they increase their investment in foreign food processing facilities. The authors reported that sales from foreign affiliates rose 56 percent between 1988 and 1993, while exports from their U.S. plants grew even faster, at 143 percent.

According to Handy and Henderson (1994), most large food manufacturers rely more heavily on foreign investment than on exports as their major strategy to access foreign markets. Sales from U.S. foreign affiliates are nine to 10 times larger than are exports from their U.S. parent firms. This highlights the importance of using both trade and foreign affiliate sales data in assessing the international competitiveness of food manufacturing firms.

Finally, the determinants of direct foreign investment are classified into demand-side and supply-side factors. Demand-side (pull) factors include political stability, cultural similarity, host market size, growth rate, wage rate, investment incentives, and tariff levels. Supply-side (push) factors are more firm specific, such as economies of scale, market power, product life cycle, intangible assets, and internationalization (Ning and Reed 1995, pp. 77-85).

Empirical Model

In a liberalized trade environment, producers use both exports and foreign direct investment strategies to operate in the international market. The equilibrium level of exports and foreign direct investment can be estimated by using systems of simultaneous equations. However, this approach is not possible in this study because of severe data limitations. For examples, the data on U.S. investment in Canada, and Canadian investment in the United States are not available for the years before 1983. Due to these data limitations, investments and exports are estimated individually using two separate models, described below.

Trade Model

There are many factors which determine the demand for U.S. exports to Canada, and Canadian exports to the United States. Equation (1) specifies our empirical model, where the dependent variable is U.S. exports of consumer-oriented processed products to Canada (EXPORTS), comprised of 16 agricultural industries. This variable (originally expressed in U.S. dollar terms) was deflated by a U.S. export price index so that it would be denominated in quantity units, rather than dollars.

EXPORTS i = a0 + a1GDP i + a2XPRICE i + a3DUMMY i + a4SALES i + ei (1)

The independent variables include real gross domestic product (GDP) of Canada as a measure of national income, which should have a positive coefficient; the deflated export price of U.S. exports (converted into Canadian dollars) as a price variable (XPRICE), which should have a negative coefficient; and foreign direct investment sales (SALES) by Canadian affiliates of U.S. food processing corporations to capture synergistic effects between trade and foreign direct investment. A positive coefficient on SALES indicates trade and foreign direct investment are synergistic marketing strategies; while a negative coefficient indicates that they are competing strategies. Additionally, a dummy policy variable (DUMMY) was used to capture the impact of CUSTA; 0 before 1989 and 1 after 1989.

Ordinary least squares (OLS) regression analysis using a covariance model on panel data was conducted with SAS software (Kmenta 1986, pp.630-35; SAS, 1988). The 16 industries were treated as separate observations and dummy variables were included so a covariance analysis could be performed (the model pooled time series and cross sectional data by food industry, but allowed different coefficients for all independent variables by industry).

Similar analyses were conducted on Canadian exports of consumer-oriented processed food products to the United States. The dependent variable was Canadian exports to the United States, deflated by a Canadian export price index. The independent variables included the U.S. real gross domestic product, the Canadian export price index, a dummy variable to capture the impact of CUSTA; and sales of U.S. affiliates of Canadian corporations.

U.S. Direct Foreign Investment in the Canadian Food Sector

Using the Ning and Reed (1995, pp. 77-85) classification of the determinants of direct foreign investment described in the literature review, this study focuses on demand-side factors. The dependent variable selected is U.S. direct foreign investment (DFI) in Canadian food industries. The independent variables include income generated from Canadian operations (INC), the real effective exchange rate between Canada and the U.S. (EXCH), and a dummy variable to capture the CUSTA influence (DUMMY).

A parallel study could not be conducted for Canadian investment in the United States due to lack of data stemming from disclosure problems in Canadian business transactions. Again, ordinary least squares estimation using SAS software was used to obtain the coefficient estimates. Both income and the dummy variable are expected to have positive coefficients, while the exchange rate (Canadian $ / U.S. $) is expected to have a negative coefficient. For example, as the U.S. dollar strengthens relative to the Canadian dollar, U.S. exports should contract since Canadians can now buy relatively fewer U.S. goods; however U.S. investment in Canada should expand since U.S. investors can buy relatively more Canadian assets for their investments. The functional form of this investment model is presented below:

DFI i = a0 + a1INC i + a2EXCH i +a3DUMMY i + ai (2)

Empirical Analysis

Data Sources

The data were collected for the period 1976-1994. Data on U.S. agricultural exports to Canada were collected from the Trade and Marketing Analysis Branch of the Foreign Agricultural Service, U.S. Department of Agriculture. Gross domestic product (GDP), export and import price indices, and the real effective exchange rate data were collected from the International Financial Statistics published by International Monetary Fund, Washington D.C. Foreign direct investment, income from operations in Canada, and direct investment data were collected from the Survey of Current Business published by the Bureau of Economic Analysis (BEA). Data on sales of processed foods by foreign affiliates of their U.S. parent companies in addition to data on sales of processed foods by the U.S. affiliates of Canadian companies were collected from the following BEA publications: U.S. Direct Investment Abroad: Operations of U.S. Parent Companies and Their Foreign Affiliates and Foreign Direct Investment in the U.S.: Operations of U.S. Affiliates of Foreign Companies.

U.S. Exports of Consumer-Oriented Products to Canada

Two sets of regression analyses were performed on equation (1) using an OLS covariance model on panel data to determine the behavior of U.S. agricultural exports to Canada (Kmenta 1986, pp.630-35). One analysis (model 1) used all data available (1976-95), but excluded the U.S. affiliate sales in Canada (which were only available for a limited number of years). The other analysis (model 2) used the fully specified model, but with fewer observations (1982-93), due to data limitations. Both regressions pooled observations from the 16 agricultural industries within the consumer-oriented processed food category. These 16 industries of processed foods include snack foods; breakfast cereals; red meats–fresh and prepared; poultry; processed dairy products; eggs and egg products; fresh fruits; fresh vegetables; processed fruits and vegetables; fruit and vegetable juices; tree nuts; wine and beer; nursery products and cut flowers; pet foods; and "other processed foods," which is used as a base industry for our covariance model using the above pooled cross-section and time-series observations (Kmenta 1986, pp.630-5). Empirical results reported in Table 8.3 and Table 8.4 are for the base industry in the covariance analysis, which was the "other processed foods" category described above. The full results for the remaining 15 industries are available from the authors, though they are discussed briefly below.

Empirical results for model 1 (the regression without affiliate sales; Table 8.3) indicate that the CUSTA policy coefficient (DUMMY) was of the correct sign and significantly different from zero at the 1 percent level. Thus, indicating that CUSTA did in fact have an impact on U.S. exports to Canada. The coefficient indicates that CUSTA alone increased U.S. exports of "other processed foods" by $2.3 billion per year. The coefficients on the CUSTA dummy variables for the other 15 industries were mostly negative (11 of 15), but all were smaller in absolute value than the coefficient in Table 8.3; thus, the net effect for this covariance analysis indicates that the CUSTA effect was positive for each food industry. The Durbin-Watson statistic indicated that no serial correlation was present at the 1 percent level of significance. Additionally, no multicollinearity was detected, since all variance inflation factors were less than 10 (Gujarati 1995, pp. 300-2.; Kennedy 1992, pp 183-7).

Additionally, the Canadian income coefficient (GDP) had the correct sign and was significantly different from zero at the 1 percent level in our base industry. Income coefficients for the remaining 15 processed foods industries were generally negative and significantly different from zero (indicating that the income effect for those other industries was less than for "other processed foods" category). The size of the coefficients for many of these 15 industries were such that the combined effect would be very close to zero, indicating that income effects were minimal for many industries.

The export price coefficient was positive (which was not expected), but not significantly different from zero at the 5 percent level for our base industry. Export price coefficients for the other 15 industries were negative, though only three were significantly different from zero. Thus, it does not appear that the export price has much impact on U.S. exports to Canada.

Results presented in Appendix 8.1 correspond to model 1. This explains the behavior of U.S. exports of consumer-oriented products excluding direct foreign investment (DFI) sales. Results presented in Appendix 8.1 indicate that income (GDP) coefficient was positive and significant for the following industries: breakfast cereals, red meats, poultry meat, dairy products, and wine and beer. The export price coefficient was negative (as expected) and significant for the industries, namely red meat, poultry meat, dairy product, egg and product, fresh fruit, fresh vegetables, tree nuts, and nursery products. CUSTA (dummy) coefficient was positive and significant for all the industries except snack foods, red meats, and processed fruit and vegetables.

Next we examined the empirical results for model 2, which included the full model specification (with the U.S. affiliate sales variable; Table 8.4), however, with fewer observations. Looking first at our "other processed foods" base industry, empirical results indicate that affiliate sales (SALES) were in fact positively related to exports and the coefficient was significantly different from zero at the 1 percent level, thus, indicating that U.S. exports and U.S. direct foreign investment in Canada are synergistic market strategies, rather than competing strategies.

In this model, the coefficients for the export price and the CUSTA dummy variables had the correct signs, and were significant at the 1 percent and 10 percent levels, respectively. The results indicate that CUSTA increased U.S. exports of "other processed foods" by $551 million per year, a much smaller amount than in model 1. Further, the income coefficient turned negative and significantly different from zero. As with model 1, the Durbin-Watson statistic indicated that no serial correlation was present at the 1 percent level of significance. Additionally, no multicollinearity was detected.

Turning to empirical results for the remaining 15 agricultural industries, income coefficients for all of the other 15 processed food industries were positive (with one exception) and most coefficients were similar in absolute value to the "other processed foods" result, reported above. This means that for most food industries the income effect was close to zero. All of the export price coefficients for the other 15 processed food industries were also positive, but all were smaller in absolute value than the export price coefficient for the "other processed foods" base category.

Detailed results for all the industries (model 2) are presented in Appendix 8.2. Looking at these results, we conclude that CUSTA coefficient was positive for all fifteen industries but significant for only three industries, namely red meats (fresh), fresh fruit, and fresh vegetables. Income coefficient was positive and significant only for the wine and beer industry which has high income elasticity compared to other industries included in this study. The export price coefficient was negative and significant for almost all the industries except snack food, fresh fruit and vegetables, and processed fruit and vegetables. DFI sales coefficient was positive and significant for almost all the industries. The only exceptions were snack foods, fresh vegetables, processed fruits and vegetables. These results lend support to the idea that DFI and trade are complementary and no support to the idea that DFI and trade are substitutes.

U.S. Direct Foreign Investment in Canadian Food Industries

Empirical results for the DFI model specified in equation (2) are shown in Table 8.5. This model explains aggregate U.S. direct investment in Canadian food industries. The OLS regression results indicated that all coefficients had correct signs and all were significantly different from zero at the 5 percent level, with the coefficients on income from existing operations and the CUSTA dummy significant at the 1 percent level. Results indicate that CUSTA seems to have stimulated American investment in Canada by $1.06 billion per year. Further, a $1 increase in U.S. affiliate income in Canada stimulates a $3.30 increase in investment. The model was tested for serial correlation, and multicollinearity; none was detected.

Canadian Exports of Consumer-Oriented Products to the U.S.

In addition to the above U.S. trade model assessing the impact of CUSTA on U.S. exports to Canada, a parallel analysis was conducted for Canada. Regression analyses based on the specification of equation (1) used OLS to explain the behavior of Canadian exports of consumer-oriented processed food products to the United States. Canadian exports of consumer-oriented products to the United States was the dependent variable, while independent variables included real U.S. GNP, the deflated export price of Canadian products, a dummy variable to capture the effect of CUSTA, and sales of U.S. affiliates of Canadian firms. As with the above U.S. trade model, data limitations resulted in two sets of regression analyses. One analysis (model 1) used all data available (1976-95), but excluded the Canadian affiliate sales in the United States. The other analysis (model 2) used the fully specified model, but with fewer observations (1982-93), due to data limitations. The results are reported in Table 8.6 and Table 8.7 with the "other processed food" category again serving as the base industry.

Results for model 1 (without affiliate sales) indicate that GDP and the CUSTA coefficient (DUMMY) were of the correct sign and significantly different from zero at the 1 percent level of significance. The export price coefficient was positive (which was not expected), but not significant. These results reinforce the influence of CUSTA on Canadian exports to the United States. The income and CUSTA coefficients were significant for all the 16 processed food industries included in this study. The Durbin-Watson statistic indicated that there was serial correlation problem in the original estimated model, which was corrected using the Cochran-Orcutt transformation (Kmenta, 1986, pp.298-334), and final results are reported in Table 8.6.

Appendix 8.3 presents the detailed results for all the industries for model 1. Results indicate that the income coefficient was significant only for the red meat (fresh) industry in this model whereas export price coefficient was not significant for any of the industries included in this model. The CUSTA coefficient was positive and significant for eight of the 16 industries represented in this model.

Results for model 2 (including affiliate sales) indicate that the GDP and the CUSTA coefficient were of the correct signs and significantly different from zero at the 1 percent level (Table 8.7). The export price coefficient was negative and significant at the 5 percent level of significance. Affiliate sales (SALES) were negatively related to exports, and the coefficient was significant at 1 percent level. This indicates that Canadian exports and Canadian investment in the U.S. are competing market strategies. This finding is in contrast to the strategy adopted by U.S. producers to access the Canadian food market. No serial correlation or multicollinearity were detected in this model.

Detailed results for model 2 are presented in Appendix 8.4. Looking at these results, we can confirm that the income coefficient was positive and significant for only five industries namely red meats (prepared), fresh fruits, tree nuts, nursery products, and roasted coffee. Export price coefficients were not significant for any of the industries included in this study. The CUSTA coefficient was positive and significant for almost all the industries indicating that the free trade agreement has influenced the flow of consumer-oriented products from Canada to the United States Foreign direct investment (FDI) sales coefficient was negative and significant for six industries.

Summary and Conclusions

This research contributes to the knowledge base of the potential impacts of the Canada U.S. Free Trade Agreement, enacted in 1989. Analysis of the data indicated a definite change in trade patterns for both the United States and Canada after the implementation of CUSTA. Total agricultural exports dramatically increased for each country–U.S. exports to Canada more than doubled (2 ½ times), while Canadian exports to the United States nearly doubled between 1989 and 1995. The rate of these effects varied by country, where U.S. industries adjusted immediately to the benefits of trade liberalization, while Canadian industries took two years before witnessing a dramatic increase in export volume.

Additionally, although total exports increased for each country, the product mix of exports changed. For the United States, the export share of consumer oriented (processed food) products to Canada increased significantly, from 61 percent in 1989 to 74 percent in 1995, corresponding to a decline in the export share for both bulk and intermediate products. Whereas, Canada's export share for processed foods declined slightly from 52 percent to 48 percent, as its export share for both bulk and intermediate products slightly increased.

Econometric analysis indicates that CUSTA did, in fact, impact U.S. exports to and investment in Canada for processed foods. The CUSTA policy variable was statistically significant at the 1 percent level in all estimations, e.g., both models of U.S. exports to Canada, as well as the U.S. direct foreign investment model. Additionally, results indicate the synergistic effects of export and direct foreign investment strategies by the United States, while Canada appears to use exports and foreign direct investment as competing strategies. Further research will be performed to determine how CUSTA has impacted exports of bulk and intermediate products for both the United States and Canada.

References

Appendixes


Document Prepared by:
Leigh H. Stribling, lstribli@acesag.auburn.edu
Alabama Agricultural Experiment Station
Auburn University

 return to top of page

  return to contents