Border Effects Before and After 9/11: Panel Data Evidence Across Industries

• Author: Zhiqi Chen,Anindya Sen,Horatiu A. Rus
• Date: 01 October 2016
• DOI: http://doi.org/10.1111/twec.12413
• Published date: 01 October 2016

Border Effects Before and After 9/11:

Panel Data Evidence Across Industries

Zhiqi Chen

1

, Horatiu A. Rus

2

and Anindya Sen

2

1

Department of Economics, Carleton University, Ottawa, ON, Canada and

2

Department of Economics,

University of Waterloo, Waterloo, ON, Canada

1. INTRODUCTION

Apersistent and very surprising empirical result in the international trade literature is the

existence of substantial ‘border effects’. The conundrum consists of the fact that intran a-

tional trade is typically found to be a multiple of international trade, even after controlling for a

diverse array of barriers such as distance, economic size, explicit trade obstacles and a host of

other country-specific characteristics, such as common language, adjacency, remoteness or alter-

native trading opportunities. The seminal study by McCallum (1995) compares trade among

Canadian provinces to corresponding flows with American states in 1988, in a gravity equa-

tion framework and using aggregate trade volumes. The key finding is that interprovincial trade

is roughly 20 times larger than flows between provinces and states. In a much cited paper, Obst-

feld and Rogoff (2001) view McCallum’s results as one of the six major puzzles in international

macroeconomics. Subsequent studies have generally found border effects of smaller but compa-

rable magnitudes in North America and Europe.

1

In another seminal paper in a related sublitera-

ture, using price variation across cities in Canada and the United States, Engel and Rogers

(1996) find nominal price stickiness to account for part of a similarly substantial borde r effect.

More recently, Gorodnichenko and Tesar (2009) point to the challenge of estimating border

effects when significant heterogeneity exists between countries with respect to the distribution of

price differentials. Our paper contributes to the first strand of the literature, using trade volumes

at the industry level, rather than price differentials, to study the evolution of border effects.

We construct a comprehensive industry-specific (at the three-digit NAICS level) database

that covers intranational and international trade flows within Canada and between Canada and

the United States between 1992 and 2005. By incorporating state-to-state trade volumes, in

addition to province-to-province and province-to-state flows, these data allow us to analyse

the effects of borders in international trade from a dual perspective. The long time-series of

our data is also important as most studies have typically relied on single-year cross-sections.

The use of disaggregated industry-level data over multiple years enables us to reliably assess

We acknowledge assistance with data collection from Linda Jonker and Derrick Khoo. This paper would

not have been possible without some outstanding research assistance from Wai Hong Choi. We thank

Marc Duhamel, participants to the 2010 Carleton Library Series Workshop on Industrial Organization

and Market Structure in Canada, an anonymous referee and the editor of this journal for constructive

comments. Chen gratefully acknowledges financial assistance from the Social Sciences and Humanities

Research Council of Canada.

1

See Helliwell and McCallum (1995), Wei (1996), Helliwell (1998), Head and Mayer (2000), Nitsch

(2000), Hillberry (2002), Anderson and van Wincoop (2003), Evans (2003, 2006), Chen (2004) and

Chen and Novy (2011). Particularly notable are estimates by Anderson and van Wincoop (2003) that

reduce Canada’s borders effects to a range between 14 and 16. See the Appendix for a slightly more

detailed review of this literature.

©2016 John Wiley & Sons Ltd

1456

The World Economy (2016)

doi: 10.1111/twec.12413

The World Economy

the existence, persistence and variability of border effects across sectors and to control for

potentially confounding year-specific shocks. Exclusive reliance on aggregate trade effectively

imposes a uniform response not only of province–state pairs, but also across the trading sec-

tors of the two economies. The time span of these data allows us to evaluate the effects of

two significant shocks that theoretically should have impacted the magnitude of border effects

between Canada and the United States. Specifically, the 1994 North American Free Trade

Agreement (NAFTA) and the increase in border security in the aftermath of the tragic 11

September 2001 (or 9/11) events.

2

The econometric estimation is based on a theoretical

justification for the sector-level gravity equation which we also develop in the paper.

Why do we see any significant border effects at all? As noted by Goldfarb (2007), an obvious

consequence of enhanced border security is longer and less predictable waiting times, which con-

tributes to increasedexpenditures on variable inputs such as fueland drivers hours, as well as more

rapid depreciation of trucks and related capital equipment. Manufacturers might respond to possi-

ble delays by increasing average inventory levels or by shipping goods earlier in order to ensure

timely arrival. Border delays are especially critical for perishable commodities. All else being

equal, border delays increase the costs of international trade and enhance the attractiveness of

domestic substitutes. The possibility of this resultingin lower international trade flows is evident.

Exploiting the effects of NAFTA and 9/11 may shed some light on the source, economic

nature and composition of border effects. For example, if non-tariff trade barriers are partly

responsible for the border effects, then we expect to see a decrease after NAFTA, above and

beyond the decrease recorded as a result of the 1988 Free Trade Agreement , which largely

dismantled tariff barriers. On the other hand, there also exists some evidence of a ‘9/11 effect’

in bilateral trade flows between Canada and the United States. In the wake of the terrorist

attacks on the United States, a series of measures aimed at tightening border security have

been adopted, and these are largely considered to have led to a substantial ‘border thickening’

and thereby to a slowing down, if not a reversal, of closer economic integration between

Canada and the United States. Negotiations to speed up the flow of goods and people across

the border are currently under way between Canada and the United States. The currently pro-

posed security perimeter and regulatory cooperation agreement

3

is aimed at promoting trade,

economic growth and jobs on both sides, while maintaining a high security level. These

efforts are based on the belief that the border did thicken. This paper is one of the first studies

that present empirical evidence of such impact.

4

The raw data certainly exhibits features of both events. Based on data from 25 (three digit)

NAICS level, predominantly manufacturing industries, Figure 1 reveals that in absolute

terms, imports by Canadian provinces from the United States increased by about 70 per cent ,

while exports rose by over 90 per cent over the pre-2001 period included in our sample.

However, both bilateral exports and imports fell rather sharply over a short duration following

2

As elaborated in Section 3, our database covers more commodities and a longer period than Anderson

and Yotov (2010).

3

For an overview of the Canada-United States Beyond the Border Action Plan, see Government of

Canada (2015). The latest development in the Beyond the Border project is a custom pre-clearance

agreement between the two countries that would allow customs agents to work in each others’ countries

(Panetta, 2015).

4

Anderson et al. (2011) also find that the US border thickened after 9/11 for some of the nine service

sectors in their study. Our paper complements theirs since the sectors in our sample are mostly manufac-

turing.

©2016 John Wiley & Sons Ltd

BORDER EFFECTS BEFORE AND AFTER 9/11 1457