Gravity, Borders, and Regionalism: A Canada–US Sub-National Analysis

ABSTRACT

This article examines the extent to which trade costs influence the magnitude and direction of both east-west and north-south trade in Canada and the United States. With the aid of an alternative framework which pays attention to key estimation issues in the gravity literature, we garner further evidence in support of a decline, over time, in the home bias syndrome. Our results uphold the Linder hypothesis but refute the Heckscher-Ohlin factor endowment proposition. In light of the recently modernized trilateral trade agreement in North America, we conclude with policy lessons on buffering the Canadian economy from asymmetric trade shocks from its southern neighbor.

KEYWORDS:

• Border effect
• home bias
• trade policy
• gravity model

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ The USMCA was signed on November 30, 2018.

² After more than 13 months of intensive negotiations, NAFTA was replaced by the USMCA.

³ While Canada’s borders extend northwards to Greenland, the Arctic, and Alaska, about 90% of Canadians live within 160 kilometers of the US border. As such, 3.9 sq. km population density gives an average density for all of Canada, while the cold weather farther north means the region is, in fact, sparsely populated.

⁴ http://www.conferenceboard.ca/hcp/provincial/economy/income-per-capita.aspx.

⁵ Krugman (1980), Helpman and Krugman (1985), Bergstrand (1990), Markusen and Wigle (1990), Deardorff (1998), Anderson and van Wincoop (2003), Evenett and Keller (2002), and Helpman, Melitz, and Rubinstein (2007).

⁶ Since they tend to be more trade-dependent.

⁷ A type of nonlinear relationship that has a firm grounding in economic theory. By implication, the effect of a change on any explanatory variable X_i on Y depends on the levels of the other X’s in the equation.

⁸ This is because the FGLS is robust to any form of heteroscedasticity, in addition to being able to weigh the observations in our data sets based on the square root of their variances.

⁹ The appendix is available online at www.tandfonline.com/uitj.

¹⁰ Characterized by a fall of 1% in global GDP, compared to a 10% decline in global trade.

¹¹ The CFS, undertaken through a partnership between the US Census Bureau and the Bureau of Transportation Statistics is conducted every 5 years as part of the Economic Census.

¹² Neither the GDP nor trade data are deflated. The reason for this is simple: the gravity model is an expenditure function which merely allocates nominal GDP into nominal imports. As explained by Baldwin and Taglioni (2006), the problem of spurious correlation might occur when the price deflators are correlated as a result of international inflation.

¹³ See Wei (1996); Wolf (1997); Helliwell and Verdier (2001); Head and Mayer (2002); Chen (2004); Mayer and Zignago (2005); and Antweiler (2007).

¹⁴ This method is popular because it provides a seamless means of calculating the stock of fixed assets whenever direct information is missing.

¹⁵ Statistics Canada uses 3.58% and 3.65% as the old and new capital stock annual growth rates, respectively.

¹⁶ https://web.csulb.edu/~syamarik/.

¹⁷ This, in a way, reflects the effects of NAFTA’s full implementation, and the progress made on Canada’s Agreement on Internal Trade.

¹⁸ Refers to the elasticity of substitution, under common assumptions, between varieties of a particular good and the price elasticity of each individual variety. See Helpman and Krugman (1985) for details.

¹⁹ See Rose and van Wincoop (2001) and Evans (2003).

²⁰ Upon ratification by the respective legislatures of all three countries in 2019.

²¹ Events leading to the signing of the USMCA clearly corroborate this.