Employment and location patterns of advanced services in non-urban Canada

Abstract

This article examines the location pattern and role of advanced producer services outside the metropolitan core. Will such services migrate to and stay in peripheral regions and boost growth as would be necessary if such regions are to benefit from the growth of the advanced services sector? Our results for Canada suggest real limits to the upside potential. We reach this conclusion from a descriptive analysis of employment and establishment data, and from a statistical test of agglomeration behaviour of professional, scientific and technical services establishments in non-urban Canada.

Acknowledgement

This research forms part of a larger project sponsored by Industry Canada for which financial assistance is gratefully acknowledged. Thanks are due to Sven Illeris and Bill Beyers for comments on an earlier version presented to the 2003 Conference of RESER, Mons, Belgium. We are also grateful to Mario Polèse and John Ries for helpful commentary. Alvin Simms of the GEOIDAL Centre at Memorial University of Newfoundland enthusiastically shared his cartographic expertise, and Joelle Aucoin provided able research assistance.

Notes

Jorgenson [2003] argues convincingly that differences in methodology and data aside, the remarkable behaviour of ICT prices provide the key to the productivity findings.

Known variously as knowledge-intensive, high-order, high-quality, and business services traded primarily as intermediate input services.

See Wernerheim and Sharpe [2004].

This refers not to the rubber boots, garbage bags, and hockey sticks manufacturing of yesteryear, but to the techno-savvy PST firms that even a former premier believed would be sufficiently footloose to come if broadband Internet access was extended to rural Newfoundland.

See e.g. Rounds [1993]. See also Martinelli [1991] for a relevant study for Italy.

See e.g. Luger and Shetty [1985]; Moomaw [1988]; Coughlin, Terze and Arrondee [1991], Woodward [1992]; Head, Ries and Swenson [1995]; Maurel and Sédillot [1999].

See also Coffey [1996]; Shearmur and Alvergne [2002].

These papers are discussed in Wernerheim and Sharpe [1999]. See also Bandt [1991] for a discussion of the problems with introducing services in the standard theorems of international trade.

See also Wasylenko and McGuire [1985].

See Appendix A for a detailed description.

The CBP data does not allow us to distinguish between firms and establishments. We use these terms interchangeably for the present purpose.

See Appendices B and C for definitions and more details.

In practice this process is complicated by contiguity and boundary issues that affects the number of urban cores identified. See Appendix C for more details.

See e.g., Coffey [1996]; Coffey and Shearmur [1997].

The CMA of Quebec, for example, incorporates all or part of 8 CDs; that of Ottawa, 4; Toronto, 7; and Montreal, 16. The larger amount of non-core area in these CDs gives rise to a higher non-core/core ratio.

The observations for 1997, 1999, and 2000 indicating a zero unemployment rate are more likely due to missing data.

This is the (in)famous independence-of-irrelevant-alternatives assumption (IIA) that the effects of unobservable attributes of plants and location choices are uncorrelated.

See Ellison and Glaeser [1997: 893–5] for details.

The {x _i } are taken as exogenous, and the {s _i } are taken as endogenously determined by s _i  = Σ _k z _k u _ki where z _k is the kth plant's exogenously fixed share of the industry's employment, and u _ki is an indicator variable equal to 1 if plant k chooses to locate in area i, and zero otherwise. The Herfindahl index is calculated as H = Σ _k z _k ².

The trigger values used by EG and MS are as follows: γ < 0.02; 0.02 ≤ γ ≤ 0.05; γ ≥ 0.05 imply, respectively, low, medium and high concentration (localisation).

This is consistent with the result for PST establishments in metropolitan areas reported by Wernerheim and Sharpe [2003].

I am grateful to Bill Beyers and John Ries for drawing my attention to the sensitivity of the EG index to the various parameter values.