Finance, insurance and real estate firms and the nature of agglomeration advantage across Canada and within metropolitan Toronto.
Subject: Book publishing (Surveys)
Gross domestic product (Surveys)
Corporations (Surveys)
Universities and colleges (Canada)
Universities and colleges (United Kingdom)
Universities and colleges (Surveys)
Unemployment (Canada)
Unemployment (United Kingdom)
Unemployment (Surveys)
Real property (Surveys)
Author: Meyer, Stephen P.
Pub Date: 12/22/2007
Publication: Name: Canadian Journal of Urban Research Publisher: Institute of Urban Studies Audience: Academic Format: Magazine/Journal Subject: Social sciences Copyright: COPYRIGHT 2007 Institute of Urban Studies ISSN: 1188-3774
Issue: Date: Winter, 2007 Source Volume: 16 Source Issue: 2
Product: Product Code: 2731000 Book Publishing; 9900010 Corporations; 8220000 Colleges & Universities; E229000 Unemployment; 6500000 Real Estate NAICS Code: 51113 Book Publishers; 61131 Colleges, Universities, and Professional Schools; 53 Real Estate and Rental and Leasing SIC Code: 2731 Book publishing; 8221 Colleges and universities
Organization: Company Name: Oxford University Press (Oxford, England)
Geographic: Geographic Scope: Canada; United Kingdom Geographic Code: 1CANA Canada
Accession Number: 179315119
Full Text: Abstract

This study compares the spatial patterns of finance, insurance and real estate firm concentration at inter-urban (for urban areas across Canada) and intra-urban (for the Toronto census metropolitan area) scales and attempts to ascertain to what extent agglomeration economies advantages are contributing to the spatial arrangement of these firms. Urban areas that specialize in finance firm activity, while rare in Canada, will typically possess a high proportion of university degree holders, a wide variety of business establishments, a comparatively balanced population age structure, a low unemployment rate and a large proportion of people employed in higher order service occupations. Real estate firm specialization is most likely to occur in growing and wealthy urban areas that are also diverse in terms of establishment type and population age structure. There are many urban areas in Canada that feature insurance firm specialization but overall this FIRE sub-sector is poorly predicted by agglomeration economies criteria. The highest densities of FIRE firm activity in metropolitan Toronto are found in the City of Toronto's south-central 'downtown' district. Based on a survey of large FIRE firms, it is demonstrated that this central location remains important because of various agglomeration advantages such as the ability to benefit from face-to-face contacts.

Keywords: finance, insurance, real estate, Canadian FIRE firms, urban specialization, agglomeration economies


Cette etude compare la distribution et la concentration spatiales des societes de finance, d'assurance et immobiliere a l'echelle interurbain (pour les zones urbaines a travers le Canada) et intra-urbain (pour la region metropolitaine de recensement de Toronto). L'article vise a determiner dans quelle mesure les avantages d'economies d'agglomeration contribuent a la concentration spatiale de ces compagnies. Les zones urbaines qui se specialisent dans l'activite financiere, bien qu'elles soient rares au Canada, possedent typiquement une haute proportion de detenteurs de diplome universitaire, une large variete d'etablissements d'affaires specialises, une population comparativement equilibree sur le plan de la structure d'age, un bas taux de chomage et une grande proportion des gens qui sont employes dans des occupations de service specialises. La specialisation des compagnies immobilieres se produit, plus souvent qu'autrement, au sein de zone urbaines privilegies et en pleine croissance marques par la diversite de type d'etablissement et sa population (structure d'age). Il y a plusieurs zones urbaines au Canada ou se concentre les societes d'assurance, mais en general ce sous-secteur est difficile a identifier selon les criteres d'economies d'agglomeration. La densite la plus elevee d'activite des societes de finance, d'assurance, et immobilieres se retrouve dans la region metropolitaine de recensement de Toronto, plus precisement dans le secteur sud-central du centre ville de Toronto. Notre enquete de ces grandes societes demontre que le centre-ville demeure un facteur de localisation important a cause des avantages d'agglomeration divers, tel la capacite de profiter des contacts face-a-face.

Mots cles: Societes canadiennes de finances, d'assurances et immobilieres, economies d'agglomeration


Finance, insurance and real estate (FIRE) activities have far-reaching influence on individuals and businesses through a multitude of transactions carried out by banks, trust companies, credit unions, investment firms, insurance companies, real estate firms and other related institutions. In addition, Canadian FIRE companies are significant contributors to the nation's gross domestic product, provide substantial government tax revenues, are important employers (in terms of both professional/quaternary occupations and tertiary jobs) and generate local spin-off activity.

With reference to the 2002 North American Industry Classification Structure (NAICS), Canada's FIRE sector can be more precisely defined as follows.

1) Monetary authorities or activities inherent to the central bank.

2) Credit intermediation: this includes all personal, commercial, corporate and institutional banking activities, credit unions, and other financial transactions processing.

3) Securities, commodity contracts and other financial investment and related activities which involve all forms of securities and commodity exchanges.

4) Funds and other financial vehicles: such as pension funds, open-end investment funds, mortgage investment funds and segregated funds.

5) Insurance carriers and related activities which include direct individual life, health, medical, property and automobile insurance and reinsurance activities.

6) Real estate: such as lessors of real estate, offices of real estate agents and brokers, and real estate property managers (Statistics Canada 2005).

Continuing advancements in information and communications technology have enabled all firms (FIRE institutions included) to complete more of their transactions from a distance but, in general, geography still biases location decisions. Simply, economic activity clusters in space (as is evident within cities and regionally both intra- and inter-nationally) and this suggests that agglomeration economies persist in this era of the Internet, teleconferencing and the like. Agglomeration economies, or the benefits and/or cost savings that firms derive by clustering, are commonly denoted in two ways: localization economies and urbanization economies (Hanink 1997). Localization economies occur when similar firms doing similar things share a location and benefit from proximity advantages. These benefits include being near specialized labour pools and inputs and the ability to participate in industry-specific knowledge exchange. As Myrdal (1957) originally explained with his 'circular and cumulative causation' ideas, over time, a cluster of comparable firms can encourage additional similar and complementary firms to locate in the vicinity, which in turn enhances the cluster's agglomeration benefits and reputation. Urbanization economies are scale advantages that extend to all firms in urban areas (such as a varied workforce and a large local market). Therefore, localization economies are gains that are external to the firm but internal to the cluster of related industry and urbanization economies are external to both the firm and industry, but internal to the urban area (Moonaw 1988; Feldman 2000). Admittedly, there is some debate over the ease with which agglomeration economies can be classified; however, it is generally agreed that localization economies imply specialization (or industry-specific) benefits and urbanization economies are linked with urban density and diversity advantages (Shearmur and Polese 2005).

While it may be tempting to assume that financial activities are the least affected by geography when compared to other economic sectors, evidence continues to show that "the business of finance still thrives upon dose inter-firm and interpersonal relationships" (Tickell 2000: 236). In fact, the benefits of agglomeration advantage (particularly in forging trust through face-to-face contacts) may be even more important today as the FIRE sector becomes increasingly competitive in this time of restructuring. For instance, in Canada, legislation developments in the early 1990s allowed financial institutions to diversify in function. This has blurred the boundaries between firms (banks and insurance companies, for example, offer increasingly similar services) and has encouraged the ongoing consolidation of the financial sector in Canada via mergers and acquisitions (Daniel 2003). Internationally, regulations on FIRE companies still vary from country to country but the overall trend has been to encourage a more competitive environment through "the large-scale dismantling of regulatory structures such as exchange and capital controls and cross border investment rules and increased foreign investment" (Laulajainen 2000: 216). While it is probable that FIRE companies continue to be influenced by agglomeration advantage, the nature of these benefits needs to be explored in this era of more distant inter-connections and greater FIRE sector competitiveness.

FIRE-Related Activity in Canada: Evidence from the Geographic Literature

The spatial propensities of finance, insurance and real estate activity in Canada has typically been described in a wider context under the umbrella of producer services, higher order services and/or office activity research and the scale of assessment has been Canada-wide, intra-provincial and intra-metropolitan. While somewhat dated, Coffey (1995) provides an excellent literature review of producer service research in Canada.

Regarding Canada-wide findings, Coffey and Shearmur (1997) determined that FIRE employment grew continuously from 1971 to 1991 (but at a rate less than all producer services combined) and showed both provincial/regional and temporal variation. There were also differences in the growth of FIRE employment across urban centers and these variations could not be explained by population size alone. Correspondingly, outstanding examples of FIRE sector specialization (in various financial sub-sectors such as banking, securities and insurance) include several small urban centers (such as Swift Current, Stratford, Collingwood and others) along side Toronto, Montreal and Vancouver. In a more summarized fashion, Katz and Bordt (2004) provide the top three industrial specializations (derived by employment-based location quotients) for larger Canadian centers. Cities that feature a relative propensity to emphasize FIRE activity are Moncton (investment), Quebec (investment and insurance), Peterborough (real estate), Toronto (financial), Kitchener (insurance and real estate), London (insurance) and Barrie (real estate).

Coffey (2000) found that, in comparison to overall service sector employment, finance, insurance and real estate remains very concentrated in Canada. In absolute numbers, about half of all FIRE-related employment in 1991 was in Canada's four most populous metropolitan areas (Toronto, Montreal, Vancouver and Ottawa-Hull) and 66.9% in the ten largest census metropolitan areas (CMAs). Despite some decentralization (in 1971 the ten largest CMAs had 68.2 percent of all FIRE employment), the overall pattern of large center dominance remains clear. In efforts to better understand the dynamics of the Canadian economy, Coffey employed a regression analysis. Dependent variables (urban employment growth rates for various economic sectors) were tested against a series of independent variables measuring: regional influence, population size, metropolitan proximity, firm competition, diversity of economic structure and socio-economic attributes. Over two time-periods (1971 to 1981 and 1981 to 1991), urban employment growth in the FIRE sector was associated with the growth rates of the wider area (regional influence), diversity in economic structure and, for 1981 to 1991 only, metropolitan proximity.

Eberts and Randall (1998) considered the intra-provincial trends in producer services employment in Saskatchewan (for 1991) and found that FIRE and business services were under-represented in the province's two largest cities (Regina and Saskatoon), whereas services to primary producers and to transportation and communication functions were over-represented. The authors deduce that the distribution of producer services is affiliated with the underlying industrial structure of the region and is not related to the urban hierarchy in a completely linear fashion. Another intra-provincial study was provided by Bowles (2000) who estimated the vulnerability of various communities in British Columbia to bank mergers. It was calculated that if two of the four major banking institutions combined, it would be rural communities that would experience the greatest reduction in accessibility to financial services.

Due predominately to Coffey's influence, the spatial characteristics of (and explanations for) intra-urban high order services patterns in the Montreal metropolitan area have been quite comprehensively addressed. In a 1996 publication, Coffey contrasts the forward and backward linkages inherent to FIRE and business service firms and, based on a detailed survey of 324 firms, was able to describe the supply and demand tendencies of producer services firms operating in Montreal. Several themes are explored but most of the study's main conclusions center on comparing FIRE service firms to business service establishments. For FIRE firms in particular, spatial proximity to both suppliers and customers within the Montreal metropolitan area appears to be crucial.

In a complementary study, presumably using the same sample of 324 firms, Coffey, Drolet and Polese (1996) ascertained that FIRE establishment are highly unlikely to move, especially those situated in Montreal's CBD, and this contrasts with other high order producer service firms which tend to be more mobile. Survey tabulations and a logistic regression analysis suggest that good accessibility to clients, a prestigious location and land costs/rental prices are important to FIRE firms in general but these criteria are more consequential to FIRE firms stationed outside of Montreal's CBD.

Coffey and Shearmur (2002) found that high order service employment (from 1981 to 1996) had grown in absolute terms throughout the Montreal CMA but that relative decentralization (generally from the central business district to suburban and peripheral 'poles') had occurred. Notably, financial and legal employment growth rates were highest in the CBD and this provided the only exception to the wider trend. The authors do not see this process as 'scattered' dispersion, but one of 'polycentricity' in which a small number of 'employment poles' are outstanding in terms of high order service employment concentration, growth and specialization. This leads the authors to deduce that face-to-face contacts (and other localization economies) must still be shaping this function-specific polycentric pattern of high order service employment, despite advances in telecommunications technology.

Intra-urban publications on Canadian FIRE activities beyond the case of Montreal are rare, but research by Shearmur and Coffey (2002) provides one exception and two studies by Gad (1985; 1991) are of general relevance to this literature review. Gad (1985) found that despite decades of suburbanization of economic activity, Toronto's central district (as of the mid- to late-1980s) remains a key location for many of Canada's dominant head offices and high order business services (such as financial and legal activities). Although it is difficult to directly compare the Toronto pattern with other large Canadian cities, Gad (1991) observed that specialized office space concentrations persist (such as with finance in Toronto, resource industries in Vancouver, oil companies in Calgary and large industrials in Montreal). The author ascertains that the office location decision process, and whether a core or suburban site is chosen, is driven by such factors as: the cost of office space, accessibility to other firms and consumers (in which face-to-face contact may be important), labour availability and various amenities (such as visibility and prestige). In a more recent study, Shearmur and Coffey (2002) used employment data (from 1981 and 1996) to compare sector-spatial patterns within Toronto, Montreal, Vancouver and Ottawa-Hull. In terms of FIRE employment, it was found that census tracts and/or municipalities located outside of the Toronto (and, to a lesser extent, the Ottawa) CBD are rivaling the traditional pattern of inner city dominance. In contrast, Montreal's CBD is actually increasing its FIRE specialization. The pattern in Vancouver has largely remained status quo with steady FIRE employment growth throughout the CMA.

Even though previous studies have provided invaluable information on FIRE concentration, specialization and change across Canada, the potential influence of agglomeration economies on FIRE specialization has not been explicitly addressed. Moreover, since the study of agglomeration economies relates directly to firm location, the use of actual firm information (rather than employment data) would be a necessary addition to the Canada-wide literature. In comparison, research on FIRE activity at the intra-urban scale has been more closely aligned with agglomeration economies theory, but applications beyond the case of Montreal are scarce.

Study Outline and Data Sources

This study is an attempt to supplement the literature on agglomeration/cluster analysis in Canada and to further our understanding of the spatial tendencies of HRE firms. It contains five main components. First, the spatial pattern of FIRE firms across Canada is illustrated and particular attention is given to depicting which urban areas specialize in FIRE firm presence (as measured by location quotients). Second, in efforts to explain this inter-urban pattern, several agglomeration economy advantages are isolated (through a regression analysis) as significant predictors of FIRE firm specialization. Third, at the intra-urban scale, a specific study of the Toronto CMA was pursued and FIRE firm kernel density measures show 'hot spots' of activity. Fourth, a spatial autocorrelation analysis (using the Moran's I index) provides statistical substantiation that the most influential FIRE firms (in terms of sales and employment) in the Toronto CMA continue to cluster in south-central Toronto. Fifth, the cases that were significant in the spatial autocorrelation analysis comprised a sample of firms that were surveyed by telephone to ascertain the advantages and disadvantages of a downtown location and the extent to which face-to-face contacts still matter to prominent FIRE firms.

Two data sets were essential to this study and both are, in all likelihood, the most comprehensive Canadian sources available for firm level information. At the inter-urban level, census agglomerations and census metropolitan areas were compared, and location quotients derived, by using information from Statistics Canadas (2004; 2001) 'Business Register'. This extensive data source includes all Canadian companies that: have an employee workforce for which payroll remittances are submitted to the Canada Customs and Revenue Agency, have at least $30 000 in annual sales revenue and/or have filed federal corporate income tax within the past three years. Provided that one has access to Statistics Canada's census tables, the 'Business Register' is universally available. For the addresses and characteristics (including sales and employee figures) of FIRE companies within the Toronto CMA, InfoCanada's (2004) 'Canadian Business Directory' was utilized. Telephone directories constitute the foundation of InfoCanada's company database, but many additional sources are used in the collection process (such as annual reports and government records). Also, InfoCanada contacts each business by telephone to verify records and to collect supplementary information. The 'Canadian Business Directory' must be purchased directly from InfoCanada. Due to the comprehensiveness of both the Statistics Canada and InfoCanada data sets, virtually any FIRE firm is equally likely to be represented in the samples (although the rare case of extremely small operators, earning less than $30 000 in annual sales, may not be enumerated in the 'Business Register'). Overall, it is believed that this analysis is free of any appreciable systematic bias.

The variables used in the regression analysis were derived via Statistics Canada's 'Business Register' (2001) and from various '2001 Census of Canada' tables, also provided by Statistics Canada (2003a; 2003b; 2003c; 2003d; 2003e; 2003f). Two digital maps were used: a boundary file of the Toronto CMA at the census tract level (Statistics Canada 2002a) and a road network file for metropolitan Toronto (Statistics Canada 2002b). SPSS version 10 was utilized to compute location quotients and to carry out the regression analysis. ArcGIS (version 9.0) was used to geocode FIRE firm addresses in metropolitan Toronto, to derive FIRE firm density measures and to complete the spatial autocorrelation analysis.

The Location of FIRE Firms in Canada and Possible Agglomeration Economies Explanations

As previous studies (using employment data) have indicated, there is strong and persistent polarization of FIRE sector activity over Canadian space (Coffey 2000) and this is reinforced here (see Table 1). For all FIRE establishment enumerated in each year's sample, over 87% are located in an urban area and this high proportion remains for the finance (monetary authorities, credit intermediation, securities, commodities and other financial investment and funds-related activity), insurance and real estate sectors individually. Of the FIRE sub-sectors, real estate firms are the most common in urban areas (with over half classified as such) and certainly the financial sector has also contributed a notable share. Although a large sample of insurance companies was compiled (roughly 12 000), this amounts to the smallest proportion of FIRE activity (5.3% in 2004 and 5.7% in 2001). In comparison to all establishments in Canadian urban areas, FIRE firms comprise a sizeable component of approximately 13% for both 2001 and 2004.

In absolute numbers, the Toronto CMA overwhelms the rest of Canada as the most important location for FIRE firms. As shown on Table 2, over one-fifth of Canada-based FIRE companies are located in Toronto (for 2001 and 2004) and this is almost double Montreal's second place accumulation. The top ten CMAs in Canada amassed over 63.0% of the total sample (for both years) and this is congruent with population size. However, when FIRE specialization is considered, via the location quotient, correspondence with urban population is not obvious (Coffey and Shearmur (1997) and Katz and Bordt (2004) illustrate similar findings with employment-based data). Within this context, the location quotient for Collingwood (for example) was calculated as:

(Number of FIRE firms in Collingwood / Number of firms in all sectors in Collingwood) / (Number of FIRE firms in Canada / Number of firms in all sectors in Canada).

A location quotient over one indicates a propensity to specialize in the FIRE sector and a value over 1.20 would indicate a clear relative concentration (Coffey and Shearmur 2002). The location quotients listed in Table 2 show that there is a mix of large metropolitan areas (Toronto, Vancouver and Montreal), intermediately sized CMAs (such as Winnipeg, Windsor and Kitchener) and smaller consolidated urban areas (Collingwood, Cobourg, Kelowna and the like) that feature FIRE firm specialization. While there is some variation in the top twenty ranking between 2001 and 2004, most of the centers that attract a disproportionate amount of FIRE firms, vis-a-vis the total urban economy, tend to persist.

There are interesting variations in the frequency of urban center specialization (or lack thereof) amongst the FIRE sub-sectors (see Table 3). Financial firm activity shows a striking pattern of polarized specialization amongst Canadian urban areas. In 2004, only Toronto, Winnipeg, Collingwood (ON), Montreal, Victoriaville (PQ), Val-d'Or (PQ), Saint-Hyacinthe (PQ), Vancouver and Charlottown (in that order) have location quotients of 1.20 or greater. Moreover, 80% (in 2004) and 82.2% (in 2001) of Canada's urban centers do not emphasize financial firm activity. This 'no specialization' proportion is considerably higher for finance firms in comparison to the insurance and real estate sub-sectors and the FIRE sector collectively. Correspondingly, the mean finance firm location quotient for all Canadian urban areas (for both 2001 and 2004) is well below one.

Location quotient values for insurance companies provide strong contrast to the finance firm example. Over two-fifths of Canada's urban areas show a distinct bias to be specialized in insurance firm activities (for 2004, this amounts to 65 of 140 centers). Interestingly, it is smaller urban areas, in often more peripheral locations, that top the 2004 list: Thompson (SK), Westaskiwin (AB), Grand Falls-Windsor (NF), Lachute (QC), Owen Sound (ON), Saint John (NB), Cobourg (ON), Kentville (NS), Moncton (NB) and Matane (QC). Fifteen metropolitan centers have insurance firm location quotients greater than 1.19 bur the highest ranking (Winnipeg at 1.51) is only 14th on the list. Notably, Canada's largest CMAs (Toronto, Montreal, Vancouver, Ottawa-Hull, Calgary and Edmonton) are absent from the 'specialized' category.

Real estate activity shows a fairly balanced distribution across the three categories with mean location quotient values at or near unity (1.00 for 2004 and 1.01 for 2001). It is noteworthy that of the 24 centers with location quotients greater than or equal to 1.20 in 2004, 11 are located in southern Ontario and 9 in southern British Columbia. It would be tempting to link the relative presence of real estate firms in these communities to strong population growth that was common to many southern Ontario and southern BC locations within the early part of this decade. The next section will more formally explore why various FIRE-related specialization tendencies have occurred in Canada's urban locations with reference to the potential role of agglomeration economies. As discussed, agglomeration economies benefit firms in several important ways (see Feldman (2000) for a summary of the literature). Urbanization economies are advantages that firms derive in cities and are based on scale, density and diversity. Transaction costs are reduced for many firms operating in dense urban environments due to the availability of markets, employees and businesses. Diversity of economic activity (Jacobs 1969; Glaeser, et al., 1992) and the structure of the social economy (Fuller and Jonas 2003) may also influence firms in urban environments. Localization economies occur for reasons that are akin to urbanization economies, but the benefits are more specific to the group of similar or complementary firms clustered closely in space. These advantages include: the development and availability of specialized labour pools, markets and industry-specific intermediate inputs, technology spillovers and access to information and knowledge (which may create performance incentives through competitive pressures and/or complementary exchange of tacit knowledge between firms) (Krugman 1991; Porter 2000). Whether the agglomeration advantages are industry-specific or more generally linked to the urban environment, it can be hypothesized that FIRE firm specialization may be related to: 1) scale (population size), 2) diversity (in terms of firms, labour and social characteristics), 3) industry-specific requirements (which include specialized requirements such as labour and service inputs) and 4) tacit knowledge (industry-specific know-how that may, at least in part, be conditioned by proximity to similar or complementary firms to form face-to-face relationships) (see also Moonaw 1988; Hanink 1997; Coffey and Shearmur 2002; Shearmur and Polese 2005; et al.).

In all, four stepwise regression equations were created: one for each dependent variable measuring firm specialization in finance, insurance, real estate and FIRE collectively. This specialization was expressed as location quotients for 135 Canadian urban areas in 2001. The independent variables that were tested as significant predictors of FIRE firm specialization echo scale, diversity and industry-specific advantages and the results provide a basis for understanding which of these agglomeration economies benefits are most relevant for the specific case of FIRE firm specialization in Canadian urban areas. Tacit knowledge and the relevance of face-to-face contacts were explicitly explored in the intra-urban analysis of metropolitan Toronto. In sum, there were ten independent variables tested in each of the four regression runs.

1) Scale was measured by population count and would be one indicator of urbanization economies advantage.

2) Diversity, another sign of urbanization economies, was implied by the Herfindahl Index and was measured in five distinct ways: amongst enterprises, labour force, income, age and mother tongue categories (1) (see Appendix). The Herfindahl Index is a measure of concentration and is calculated by squaring the percentage share of each subject (firm, person or etcetera) in each category and then summing the squares. The higher the number, the more concentrated is the subject across the categories, whereas a lower number indicates more balance or greater diversity across the categories. (2)

3) Industry-specific requirements were estimated in a number of ways. A skilled and educated workforce is potentially important to FIRE firms, so the proportion of degree holders from university was included in this analysis. 'Prosperity-measuring' indicators were also tested: population change, median income and the unemployment rate. Plainly, the FIRE sector is synonymous with 'money' and it can be argued that areas exhibiting prosperity and affluence are necessary requirements for finance, insurance and real estate firm specialization. Even though these independent variables may also imply general urbanization economies advantages, in the context of this study, it is reasoned that urban prosperity is more industry-specific in nature and suggestive of localization economies.

A number of other independent variables were initially tested in the regression runs, but ultimately eliminated because of high multicollinearity. Those variables were: senior management occupations, college degree holders, degree holders in commerce (all expressed as rates) and median house value. The results reported on Table 4, then, represent the most 'successful' regression equations in terms of the number of significant independent variables, maximum r-squared values and minimized multicollinearity. (3)

A strong case can be made for the notion that agglomeration advantages influence finance firm and overall FIRE sector specialization in Canadian urban areas. In that, if 50% (or better) of the variation in finance and FIRE firm specialization is explained by a few agglomeration economy-measuring variables (given that other non-agglomeration economy factors must also be contributing to the pattern) then one could safely claim some success. Real estate firm specialization is also reasonably well predicted by agglomeration economy characteristics as the procedure amassed four significant independent variables and a moderate r-squared value. Conversely, insurance company specialization in Canadian urban areas appears to be better explained by factors that lie outside of this analysis.

The inversely related 'establishments Herfindahl' is the only independent variable that was significant in all four of the regression equations. As this index decreases (and becomes less concentrated across the twenty sector categories), finance, insurance, real estate and FIRE firm specialization increase in Canadian urban areas. This indicates the importance of economic diversity to FIRE firm specialization. A balanced urban age structure is another frequent association as the 'age Herfindahl' independent variable was significant in three of the four equations. However, not all types of urban diversity encourage FIRE firm specialization. Variety in terms of ethnicity or income class is not significantly associated with any type of FIRE firm specialization.

The finance model features a number of agglomeration economy-measuring variables. The significance of the 'university degree holders' independent variable indicates the need for a highly educated labour force in areas featuring pronounced financial firm concentration. Diverse urban areas, in terms of age mix and establishments, appear to attract a greater relative presence of financial sector activity as well. This may indicate the need for market diversity: the greater the variety of firms and people of different ages, the greater the demand for a wide range of financial services which is met by a disproportionately high number of financial firms. The significant outcome of the scale variable (population count) suggests that financial firm specialization is most likely to occur in larger urban areas. The unemployment rate is inversely related to finance firm specialization and, in this context, probably indicates the attraction of a prosperous local economy (and lucrative market) to financial firms. Finally, the significance of the 'labour force Herfindahl' independent variable and its positive association with the finance dependent variable indicates that an increasingly concentrated labour force predicts greater financial firm specialization. A supplementary Pearson's correlation analysis was run between the 'finance firm location quotient' variable and the twenty labour force categories (each expressed as percentages of the total) (see the Appendix). The labour force classifications that were significantly (two-tailed, [infinity] = .01) and positively correlated with financial firm specialization were: finance and insurance (.495), professional, scientific and technical services (.430), information and cultural industries (.398), administrative and support, waste management and remediation services (.276) and management of companies and enterprises (.239). These moderate correlation values suggest that the labour force structure of areas that emphasize financial firm activity would feature concentration in quaternary and higher order tertiary occupations (or, more generally, high order services).

Real estate firm specialization was also significantly predicted by both industry-specific and more general urban-related location attributes. The importance of the population change variables is not surprising as growing areas will typically exhibit more house and office construction, greater residential and business mobility and, therefore, more economic opportunities for real estate firms. The significance of the median income variable associates urban prosperity with disproportionate real estate firm activity. As with the finance sector regression equation, real estate firm specialization is also linked with population age and establishment variety; which illustrates the importance of a diverse residential and business client pool to service.

The Spatial Pattern of FIRE Firms in the Toronto CMA and the Persistence of Agglomeration Advantages

To more completely assess the potential impact of agglomeration economies on FIRE firms, the specific case of metropolitan Toronto was evaluated. A sizable sample of 6655 Toronto CMA-based FIRE firms were extracted from InfoCanada's (2004) 'Canadian Business Directory'. Ultimately 5970 (2622 finance, 1085 insurance and 2263 real estate) of these firms were geo-referenced as points on to Statistics Canada's road network digital map using 'US Streets style' address geocoding in ArcGIS and these points formed the basis for this analysis. The majority of these cases contain employment (99.1%) and sales (86.6%) estimates for 2004. All map layers, and the data contained therein, were transformed by an equal distance projection ('North American equidistant conic') to ensure that distance measurements between points were accurate. This is particularly relevant for the spatial autocorrelation analysis.

Figure 1 provides an overview of the locations of FIRE firms throughout the Toronto CMA. Since some companies are at (or given the scale of this map, appear to be at) the same location, each point represents at least one firm. Nevertheless, the general pattern of FIRE firm attraction to larger municipalities within the Toronto CMA is apparent with the City of Toronto containing the most spatially compact collection of cases. To more clearly understand this pattern, kernel density measures were computed distinctly for finance, insurance and real estate firms. Refer to the larger maps on Figures 2, 3 and 4; the spatial autocorrelation results will be discussed later. In ArcGIS, the kernel density function calculates the density of points within a search radius and then allocates these values to each output raster cell. A smooth surface is created because the density value is highest at a given point and then diminishes towards the edge of the search radius. ArcGIS calculated the resolution (raster cell size) based on the shortest distance between input points (for financial firm points, for example, the resolution was 321.62 meters). The search radius was entered as one kilometer to represent roughly ten minutes walking distance. As shown, each map contains four categories of 'firms per kilometer squared' and the three non-zero classes were derived using the 'natural break (Jenks)' method of classification in ArcGIS.

The vast majority of the 'extremely dense' areas in the Toronto CMA are located in the City of Toronto's south-central region. For finance and insurance firms, the highest density category is represented by a collection of cells clustered in one 'downtown' agglomeration. The heart of this duster contains the intersections of York, Bay and Yonge (streets running north and south) with Front, Wellington, King and Adelaide (streets running east to west). Real estate firms display three clusters of extremely dense cells: one that shares roughly the same spatial extent as the finance and insurance downtown clusters and two that are to the north (but still within the City of Toronto). Cells allocated to the 'dense' category tend to encircle the 'extremely dense' clusters for finance firms and, except for two clusters in Mississauga and one in Oakville, the same is true for the real estate pattern. On the other hand, dense clusters of insurance firm activity can be found throughout the Toronto CMA with comparatively large agglomerations in Richmond Hill and the City of Toronto. The low to moderate density category, for all three FIRE sub-sectors, largely follows the pattern of population density throughout the Toronto CMA.

At least four general observations can be drawn from these density results. First, the highest concentrations of finance and insurance firm activity are located in a comparatively small area within the City of Toronto's south-central downtown district and this would appear to corroborate this area's reputation for being the most important financial district in Canada. Second, while extremely dense real estate clusters display a more polycentric pattern, these cells are still contained within the City of Toronto. Third, notable insurance firm concentrations are found throughout metropolitan Toronto (as shown by the 'dense' category) and this relatively dispersed spatial pattern corresponds to what was shown earlier when analyzing Canadian centers at the inter-urban scale. Fourth, while FIRE firm activity is quite ubiquitous throughout the Toronto CMA (as indicated by the many 'low to moderate' density clusters in all municipalities), the City of Toronto's importance is displayed yet again with an almost continuous covering of non-zero densities for all three sub-sectors. Mississauga, Richmond Hill and Markham are also noteworthy within this regard.


Overall, the south-central area of Toronto remains the most outstanding example of FIRE firm agglomeration and, as the following spatial autocorrelation analysis reveals, it is 'elite' firms that appear to be driving this cluster. In general, spatial autocorrelation can be either positive or negative; the former implying that points (or polygons) with similar attributes (weights) are clustered in space, the latter meaning that similar cases are dispersed. While tests for global spatial autocorrelation will indicate overall spatial patterns on a map, local spatial autocorrelation procedures identify 'hot-spots' of activity or areas of spatial heterogeneity (Boots and Tiefelsdorf 2000). The output from the ArcGIS version of the 'Anselin Local Moran's I' test provides an Index value and a standardized Z-score for each weighted case. In this application, six spatial autocorrelation tests were administered: finance, insurance and real estate firm locations were evaluated separately using both employment and sales weights. Those points that had Z-scores greater than 1.96 (for a 95% confidence threshold) are presented on Figures 2, 3 and 4. In other words, these firms are similar to their neighbours in terms of employment and/or sales attributes and this likeness is significantly greater than one would expect by chance.




As Table 5 shows, firms isolated as 'significantly clustered' (or cases of positive local spatial autocorrelation with Z-scores greater than 1.96) are large enterprises. Both employment and sales estimates for the significantly clustered cases collectively are far higher than the average for the Toronto CMA and this is true for all three FIRE sub-sectors. Moreover, the overall magnitude of these firms is apparent. For instance, the 26 finance firms that surpassed the 1.96 Z-score threshold employ roughly half of the CMA's workforce in finance. When financial firm location was weighted by year-end sales, the influence is equally impressive: 42.8% of all sales generated from financial firms in the Toronto CMA came from only 34 significantly clustered firms. And, as Figure 2 shows, all of these firms (weighted by employment and/or sales) are located in the City of Toronto's south-central region. (4) The influence of significantly clustered insurance firms is even more striking: 61.8% and 51.4% of the CMA's jobs and sales (respectively) in the insurance sector are generated from firms in the south-central area of Toronto. This contrasts with earlier observations. Insurance firms do concentrate in several other communities in the Toronto CMA, and indeed in smaller municipalities throughout Canada, but the largest firms in metropolitan Toronto are clearly attracted to the downtown agglomeration. Significantly clustered real estate firms in the south-central area captured the smallest proportions of employment and sales, but these percentages are still quite impressive (27% and 28.9%) given that a relatively small number of firms have amassed these totals. It should also be reported that 12 cases in the Toronto CMA were statistically significant in terms of negative local spatial autocorrelation (with Z-scores less than -1.96) but only two of these cases, real estate weighted by employment, are in the south-central region.

Therefore, statistically significant cases of local spatial autocorrelation are quite rare within the Toronto CMA with the notable exception of the clustering of 'elite' FIRE firms in the City of Toronto's south-central district. Given that these large FIRE firms would have the resources to locate virtually anywhere within metropolitan Toronto, but continue to choose this small area, implies that agglomeration advantage still biases location decisions despite continuing advances in information and communications technologies. To more completely assess this deduction, a short telephone survey was administered to this sample of large FIRE enterprises confirmed to be significantly clustered in Toronto's downtown area. The survey was conducted from August 4 to 19, 2005 and was directed towards managers or administrators working at each location. The questionnaire contained three open-ended queries pertaining to the advantages and disadvantages of a 'downtown Toronto' location and the importance of face-to-face contacts given technological improvements. In all, 40 responses (from the total number of 91 significantly clustered firms) were summarized and the results are presented on Table 6.

Amongst several 'textbook' localization economies benefits, good accessibility to complementary or competitive firms was the most frequently cited advantage of the south-central location. The idea of similar firms doing similar things at a given location which in turn is catalyst for attracting more related activity has been documented as crucial to the prosperity of local clusters in other places and this seems evident here as well. In addition, several of the firms' representatives indicated that the area's central location and high prestige gave the impression of success. This not only encourages FIRE firms to remain downtown but contributes to the high volume of customers and potential employees to frequent the district because it is 'the place to go' for those seeking FIRE services or FIRE employment. Overall, the cited advantages point to a common theme of accessibility: to other firms, customers, employees, public transit and entertainment. By comparison, the firms polled list location disadvantages with far less frequency, yet the downside of being located in Toronto's south-central region relate to time costs (for workers and clients commuting from the suburbs) and expense (particularly in terms of high office rent and parking costs).

Regarding face-to-face requirements in dealing with individuals and firms, only 7 of the 40 respondents indicated that face-to-face contact are less important today and no one considered modern technology to be an absolute replacement. The vast majority of the firm's representatives consider these contacts to be vital in building trust and this continues to be a primary reason for being in this centrally located downtown area. Indeed, many indicated that as the FIRE sector becomes more competitive, the need for forging trusting business relationships has increased and the best way to do this is by 'seeing the whites of eyes'.

The results of this survey for the specific subset of large FIRE firms significantly clustered in Toronto's south-central region suggest the continuing importance of agglomeration advantage (particularly in the form of localization economies and the exchange of industry-specific tacit knowledge). In this sense, earlier findings for higher order service activity in Montreal (Coffey, Drolet and Polese 1996; Coffey and Shearmur 2002) and for office space in Toronto (Gad 1991) are corroborated.


Empirical research on agglomeration economies and cluster formation has been biased towards the manufacturing sector and, more recently, high technology activity. But given that FIRE firms have shown similar inclinations to cluster in space and benefit the surrounding area with services, jobs and spin-off activity, research that examines the potential link between agglomeration economies and FIRE firm location choice is certainly warranted. Thus far, there have been few Canadian applications. Research that compares the location habits of finance, insurance and real estate activity and utilizes actual firm level data has been limited, particularly beyond the specific case of metropolitan Montreal. This paper appends the literature by providing results on the agglomeration tendencies of Canadian FIRE firms at both the inter-urban and intra-urban (for metropolitan Toronto) scales.

In absolute numbers, FIRE firms are most numerous in Canadas largest cities (with Toronto being the clear leader with over one-fifth of the nation's total in both 2001 and 2004). Yet, FIRE firm specialization is not only limited to the largest areas and the relative emphasis of insurance, real estate and, to a lesser degree, finance firms, vis-a-vis all economic sectors, can be found through the Canadian urban hierarchy. Urban area specialization in finance and real estate firm activity, as measured by location quotients, is reasonably well predicted by 'agglomeration economies' independent variables. For areas featuring finance, a high proportion of university degree holders is important but so is a diverse market that is well represented with firms from many sectors and individuals from several age cohorts. Finance sector specialization is also allied with low unemployment and is statistically associated with larger urban areas. The analysis also suggests that a labour force rich in quaternary and higher order tertiary occupations is associated with disproportionate financial firm activity. Diversity in establishment types and population age cohorts are also significant predictors of real estate firm specialization, yet urban areas emphasizing real estate firm presence tend to be places of growth and wealth as well. Of the FIRE sub-sectors, insurance company specialization exhibits the most dispersed inter-urban pattern across Canada (often emphasized in some very remote urban areas) and somewhat correspondingly is not well predicted by agglomeration economies-based independent variables.

Despite fantastic suburban and peripheral-metropolitan growth and the continuing development of transportation and communications networks over the last several decades, FIRE firm intensity remains most striking within a relatively small area in the City of Toronto's south-central district. A spatial autocorrelation analysis confirms that a number of large and influential finance, insurance and real estate firms are statistically clustered in the densest areas of FIRE firm activity in south-central Toronto. Based on a sample of these significantly clustered firms, it can be said that the accessibility, prestige and aesthetics of this downtown location outweigh the costs associated with operating in a high traffic area. Moreover, the firms' representatives were almost completely united in their continuing believe that face-to-face contacts remain vital in doing business and that this central location provides the best opportunities to meet with individuals and firm representatives in the same room or at a nearby restaurant.

Toronto's south-central business area is known not only for its financial district, but also as home to many of Canada's dominant headquarters (Meyer and Green 2003). Therefore, the significant clustering of large FIRE firms in this area and the apparent agglomeration economies influences may also be the result of specialized needs inherent to head offices. It would be interesting to contrast the agglomeration tendencies of FIRE head offices versus FIRE subsidiaries within the Canadian urban environment. Such future research may also shed light on why large FIRE firms in the suburbs (which are more likely to be branches) did not demonstrate significant clustering in the spatial autocorrelation analysis.

In sum, FIRE firm concentrations appear to be attracted to, and benefit from, areas of prosperity, accessibility and (depending on how it is measured) diversity. Areas that do not exhibit these traits may not only experience a decline in services, but also will not benefit from the added employment, tax base and spin-offs that FIRE clusters bring. Whether these conditions can be effectively enhanced by municipal governments is debatable, but it certainly would be useful for those involved in economic development planning to be mindful of the conditions that seem to encourage FIRE clusters and specializations. Future research could be directed towards understanding what factors explain the comparatively more dispersed pattern of insurance company specialization. As well, the spatial pattern of FIRE firms of various sizes could be explored more precisely. An explicit assessment of the benefits that FIRE clusters bring cities and neighbourhoods would be another interesting area of empirical study.


Boots, B., and M. Tiefelsdorf. 2000. Global and local spatial autocorrelation in bounded regular tessellations. Journal of Geographical Systems 2:319-348.

Bowles, P. 2000. Assessing the impact of proposed bank mergers on rural communities: a case study of British Columbia. Social Indicators Research 51: 17-40.

Coffey, W.J. 1995. Producer services research in Canada. Professional Geographer 47(1): 74-81.

--. 1996. Forward and backward linkages of producer-services establishments: evidence from the Montreal metropolitan area. Urban Geography 17(7): 604-632.

--, R. Drolet and M. Polese. 1996. The intrametropolitan location of high order services: patterns, factors and mobility in Montreal. Papers in Regional Science 75(3): 293-323.

--, and R.G. Shearmur. 1997. The growth and location of high order services in the Canadian urban system, 1971-1991. Professional Geographer 49 (4): 404-418.

--.2000. Canadian cities and shifting fortunes of economic development. In Canadian Cities in Transition: The Twenty-First Century, ed. T. Bunting and P. Filion. Toronto: Oxford University Press.

--, and R.G. Shearmur. 2002. Agglomeration and dispersion of high order service employment in the Montreal metropolitan region, 1981-96. Urban Studies 39(3): 359-378.

Daniel, F. 2003. Recent changes to Canadas financial sector legislation. Bank of Canada Review Winter 2002-2003: 3-16.

Eberts, D., and J.E. Randall. 1998. Producer services, labor market segmentation and peripheral regions: the case of Saskatchewan. Growth and Change 29(Fall): 401-422.

Feldman, M.P. 2000. Location and innovation: the new economic geography of innovation, spillovers and agglomeration. In The Oxford Handbook of Economic Geography, ed. G.L. Clark, M.P. Feldman and M.S. Gertler. Toronto: Oxford University Press.

Fuller, D., and A.E.G. Jonas. 2003. Alternative financial spaces. In Alternative Economic Spaces, ed. A. Leyshon, R. Lee and C.C. Williams. London: Sage.

Gad, G. 1985. Office location dynamics in Toronto: suburbanization and central district specialization. Urban Geography 6(4): 331-351.

Gad, G. 1991. Office location. In Canadian Cities in Transition, ed. T. Bunting and P. Filion. Toronto: Oxford University Press.

Glaeser, E., H.D. Kallal, J.D. Scheinkman and A. Shleifer. 1992. Growth in cities. Journal of Political Economy 100:1126-52.

Hanink, D.M. 1997. Principles and Applications of Economic Geography. Toronto: John Wiley & Sons.

InfoCanada. 2004. Canadian Business Directory (2004.2 edition). Mississauga, ON: InfoCanada.

Jacobs, J. 1969. The Economy of Cities. Toronto: Random House.

Katz, S., and M. Bordt. 2004. Community innovation: industrial specialization in Canadian cities. Statistics Canada Working Paper, Cat. No. 88F0006XIE.

Krugman, P. 1991. Geography and Trade. London: MIT Press.

Laulajainen, R.I. 2000. The regulation of international finance. In The Oxford Handbook of Economic Geography, ed. G.L. Clark, M.P. Feldman and M.S. Gertler. Toronto: Oxford University Press.

Meyer, S.P., and M.B. Green. 2003. Headquarters in Canada: an analysis of spatial patterns. Urban Geography 24(3):232-252.

Moonaw, R.L. 1988. Agglomeration economies: localization or urbanization? Urban Studies 25:150-161.

Myrdal, G. 1957. Economic Theory of Under-Developed Regions. New York: Methuen.

Porter, M.E. 2000. Location, clusters and company strategy. In The Oxford Handbook of Economic Geography, eds G.L. Clark, M.P. Feldman and M.S. Gertler. Toronto: Oxford University Press.

Shearmur, R., and W.J. Coffey. 2002. A tale of four cities: intrametropolitan employment distribution in Toronto, Montreal, Vancouver and Ottawa-Hull, 1981-1996. Environment and Planning A 34: 575-598.

--, and M. Polese. 2005. Diversity and employment growth in Canada, 1971-2001: can diversification policies succeed? The Canadian Geographer 49(3): 272-290.

Statistics Canada. 2001. Establishments counts by census agglomeration and census metropolitan area, showing NAICS sub-sectors and employee size ranges (June 2001 Canadian business patterns). Machine readable data file. Ottawa, ON: Statistics Canada Statistics.

--. 2002a. Census of Canada, 2001: cartographic boundary file map of census metropolitan areas: Toronto, Ontario at the census tract (CT) level. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2002b. Census of Canada, 2001: road network file map of census metropolitan areas: Toronto, Ontario. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2003a. Census of Canada, 2001. Table 95F0300XCB01004--CMACA: age (122) and sex (3) for population, for Canada, provinces, territories, census metropolitan areas and census agglomerations, 2001 census--100% Data. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2003b. Census of Canada, 2001. Table 97F0020XCB1001- CMACA: total income groups (22) in constant (2000) dollars, sex (3), age groups (7) and historical highest level of schooling (6) for population 15 years and over, for Canada, provinces, territories, census metropolitan areas and census agglomerations, 1995 and 2000--20% sample data. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2003c. Census of Canada, 2001. Table 97F0007XCB01009--CMACA: detailed mother tongue (160), sex (3), knowledge of official languages (5) and age groups for population, for Canada, provinces, territories, census metropolitan areas and census agglomerations, 1996 and 2001 censuses--20% sample data. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2003d. Census of Canada, 2001. Table 97F0012XCB01010--CMACA: industry--1997 North American industry classification system (23), class of worker (6) and sex (3) for labour force 15 years and over for Canada, provinces, territories, census metropolitan areas and census agglomerations, 2001 census--20% sample data. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2003e. Census of Canada, 2001. Table 95F0490XCB01004- profile of labour force activity, class of worker, occupation, industry, place of work, mode of transportation, language of work, and unpaid work, for census metropolitan areas and census agglomerations, 2001 census--20% sample data. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2003f. Census of Canada, 2001 (geo-suite). Population (2001 and 1996), dwelling counts (2001), and land area for census metropolitan areas and census agglomerations, 2001 census--100% data. Machine readable data file. Ottawa, ON: Statistics Canada.

--. 2004. Establishments counts by census agglomeration and census metropolitan area, showing NAICS sub-sectors and employee size ranges (June 2004 Canadian business patterns). Machine readable data file. Ottawa, ON: Statistics Canada.

--.2005. The Economy. Canada e-Book, 001_e.htm. Last accessed May 30, 2005.

Tickell, A. 2000. Finance and Localities. In The Oxford Handbook of Economic Geography, ed. G.L. Clark, M.P. Feldman and M.S. Gertler. Toronto: Oxford University Press.

Stephen P. Meyer

Department of Geography

Laurentian University


(1) Fuller and Jonas (2003) emphasize the importance of the local social environment from which credit unions in Britain derive their uniqueness and identity and thus provide justification for testing income, age and mother tongue diversity in this analysis. Enterprise and labour force diversity (as examples of urbanization economies) have been discussed more frequently in the literature, at least since the work of Jacobs (1969).

(2) The Herfindahl Index (H) is the sum of the squares of the category shares (s):

H = [n.summation over (i=1)] ([S.sup.2.sub.i])

This index is very robust and is not highly sensitive to the number of categories chosen. For illustrative purposes, the 'age Hertindahl' variable was calculated for the 135 urban areas using 21 (five year classes) and 11 (10 year classes) categories. The correlation between the two indexes was .998, indicating that it makes little difference which categorizing approach is used. As a result, the divisions used were the major sub-classes as defined by Statistic Canada (2003a).

(3) Three highly collinear variables, 'college degree holders', 'degree holders in commerce' and 'senior management occupations', were run (one at a time) in place of 'university degree holders'. In the two relevant equations (finance and FIRE), 'college degree holders' was insignificant and while 'degree holders in commerce' and 'senior management occupations' were both significant replacements for 'university degree holders', both produced lower r-squared results. Similarly, when the collinear 'median house value' variable replaced 'population change' it was significant but yielded a lower r-squared value.

(4) Since some firms share (or because of map scale appear to share) a location, the dots represent statistically significant locations more so than each significant firm.
Appendix: Categories Used in the Computation of Herfindahl Indices

Herfindahl categories for   1) Agriculture, forestry, fishing
establishments and for         and hunting
labour force                2) Mining and oil and gas
                            3) Utilities
                            4) Construction
                            5) Manufacturing
                            6) Wholesale trade
                            7) Retail trade
                            8) Transportation and warehousing
                            9) Information and cultural
                            10) Finance and insurance
                            11) Real estate and rental and
                            12) Professional, scientific and
                                technical services
                            13) Management of companies and
                            14) Administrative and support,
                                waste management and remediation
                            15) Educational services
                            16) Health care and social assistance
                            17) Arts, entertainment and recreation
                            18) Accommodation and food services
                            19) Other services (expect public
                            20) Public administration

                            These categories correspond with the
                            North American Industry Classification

Herfindahl                  1) English
categories                  2) French
for mother tongue           3) Aboriginal languages
(single response)           4) Romance languages (other than
                            5) Germanic languages (other than
                            6) Celtic languages
                            7) Slavic languages
                            8) Baltic languages
                            9) Finno-Ugric languages
                            10) Greek
                            11) Armenian
                            12) Turkic languages
                            13) Semitic languages
                            14) Indo-Iranian languages
                            15) Dravidian languages
                            16) Japanese
                            17) Korean
                            18) Sino-Tibetan languages
                            19) Tai languages
                            20) Austro-Asiatic languages (not
                                otherwise listed)
                            21) Malayo-Polynesian languages
                            22) Asiatic languages (not otherwise
                            23) Niger-Congo languages
                            24) Africa languages (not otherwise
                            25) Creoles
                            26) Other languages

                            These categories correspond with
                            Statistics Canada's mother tongue
                            classifications for major language

Herfindahl                  0) 0 - 4
categories                  2) 5 - 9
for age                     3) 10 - 14
                            4) 15 - 19
                            5) 20 - 24
                            6) 25 - 29
                            7) 30 - 34
                            8) 35 -39
                            9) 40 - 44
                            10) 45 - 49
                            11) 50 - 54
                            12) 55 - 59
                            13) 60 - 64
                            14) 65 - 69
                            15) 70 - 74
                            16) 75 - 79
                            17) 80 - 84
                            18) 85 - 89
                            19) 90 - 94
                            20) 95 - 99
                            21) 100 and over

                            These categories correspond with
                            Statistics Canada's age cohorts.

Herfindahl                  1) Under $2000
categories                  2) $2000 - $4999
for income                  3) $5000 - $6999
                            4) $7000 - $9999
                            5) $10000 - $11999
                            6) $12000 - $14999
                            7) $15000 - $19999
                            8) $20000 - $24999
                            9) $25000 - $29999
                            10) $30000 - $34999
                            11) $35000 - $39999
                            12) $40000 - $44999
                            13) $45000 - $49999
                            14) $50000 - $59999
                            15) $60000 - $74999
                            16) $75000 and over

                            These categories correspond with
                            Statistics Canada's income ranges.

Table 1: Finance, Insurance and Real Estate (FIRE)
Establishments in Canadian Urban Areas (Census Metropolitan
Areas (CMA) or Census Agglomerations (CA))

                                     2004 Sample        2001 Sample

Total FIRE establishments in           263 387            239 099

Percent of FIRE in urban areas           87.6               87.2

Percent of finance in urban              88.7               88.0

Percent of insurance in urban            84.5               84.0

Percent of real estate in urban          87.3               87.1

Total FIRE in all urban areas      230 838 (100.0%)   208 483 (100.0%)

Monetary authorities -- central       15 (0.0%)          12 (0.0%)

Credit intermediation and           16 397 (7.1%)      15 747 (7.6%)
related activities

Securities, commodity contracts     64 564 (28.0%)     62 156 (29.8%)
and other financial investment
and related activities

Funds and other financial            2 633 (1.1%)       1 810 (0.9%)

Financial sector (total of above    83 609 (36.2%)     79 725 (38.3%)
four rows)

Insurance carriers and related      12 135 (5.3%)      11 965 (5.7%)

Real estate                        135 094 (58.5%)    116 793 (56.0%)

Total establishments in all urban     1 818 778          1 598 333

Percent of establishments               12.7%              13.0%
classified as FIRE

Number of urban areas (CMA or CA)        140                135

Table 2: Leading Urban Areas in Terms of FIRE Firms: Showing Percent
of Canadian Total and Location Quotients (LQ

                           FIRE Percent                    FIRE Percent
                           of Total                        of Total
    CMA/CA                 2004          CMA/CA            2001

1   Toronto                22.6          Toronto           22.2
2   Montreal               12.7          Montreal          13.4
3   Vancouver              10.0          Vancouver         10.2
4   Calgary                4.7           Calgary           4.4
5   Edmonton               3.6           Edmonton          3.5
6   Ottawa-Hull            2.8           Ottawa-Hull       2.9
7   Winnipeg               2.0           Winnipeg          2.2
8   Quebec                 1.9           Quebec            2.0
9   Hamilton               1.9           Hamilton          1.8
10  Kitchener              1.4           London            1.3
    Total above            63.6                            63.9
    All CMAs/CAs           87.6                            87.2
    Total FIRE
    firms in
    Canada                 263 387                          239 099

                           FIRE LQ 2004                    FIRE LQ 2001

1   Collingwood (ON)       1.44          Cobourg (ON)      1.42
2   Cobourg (ON)           1.34          Collingwood (ON)  1.37
3   Vancouver (BC)         1.29          Toronto (ON)      1.36
4   Toronto (ON)           1.28          Vancouver (BC)    1.34
5   Winnipeg (MB)          1.20          Winnipeg (MB)     1.26
6   Windsor (ON)           1.19          Tillsonburg (ON)  1.23
7   Victoria (BC)          1.19          Stratford (ON)    1.22
8   Kelowna (BC)           1.18          Victoria (BC)     1.20
9   Vernon (BC)            1.17          Montreal (QC)     1.19
10  Montreal (QC)          1.17          Windsor (ON)      1.19
11  Elliot Lake (ON)       1.16          Kitchener (ON)    1.16
12  Brantford (ON)         1.15          Brandon (MB)      1.16
13  Kitchener (ON)         1.15          Norfolk (ON)      1.15
14  Charlottetown (PE)     1.15          London (ON)       1.14
15  Joliette (QC)          1.15          Woodstock (ON)    1.14
16  Saint-Hyacinthe (QC)   1.15          Kelowna (BC)      1.13
17  Lethbridge (AB)        1.14          Joliette (QC)     1.13
18  Brandon (MB)           1.14          Guelph (ON)       1.13
19  Sault Ste. Marie (ON)  1.14          Sudbury (ON)      1.13
20  London (ON)            1.13          Halifax (NS)      1.12

Table 3: Canadian Urban Areas in Various Location Quotient
(LQ) Categories

                   Specialized      Slight           No
                   Centers (LQ   Propensity to  Specialization
                    [greater      Specialize    (LQ [less than
                  than or equal  (LQ > 1 and      or equal
                    to] 1.2)        < 1.2)          to] 1)      Mean LQ

Finance 2004          6.40%         13.60%          80.00%        .83
Finance 2001          5.90%         11.90%          82.20%        .81

Insurance 2004       46.40%         16.40%          37.20%       1.16
Insurance 2001       45.90%         17.80%          36.30%       1.16

Real estate 2004     17.10%         37.90%          45.00%       1.00
Real estate 2001     20.70%         34.10%          45.20%       1.01

FIRE 2004             3.60%         37.10%          59.30%        .95
FIRE 2001             5.90%         35.60%          58.50%        .94

Note: in 2004 there were 140 urban areas (census metropolitan areas
and consolidated areas) and in 2001 there were 135.

Table 4: Stepwise Regression Results

Dependent           Significant Independent         Beta      Adjusted
Variable                    Variables           Coefficients  R Square

Finance location   1) University degree             .384
quotient             holders (rate)
                   2) Age Herfindahl               -.286
                   3) Establishments               -.273
                   4) Population count              .278
                   5) Unemployment rate            -.185
                   6) Labour force                  .138        .500

location quotient  1) Establishments               -.225        .043

Real estate        1) Population change (rate)      .269
location quotient  2) Establishments               -.374
                   3) Age Herfindahl               -.424
                   4) Median Income                 .373        .387

FIRE location      1) University degree             .312
quotient             holders (rate)
                   2) Age Herfindahl               -.431
                   3) Establishments               -.393
                   4) Unemployment rate            -.156
                   5) Population count              .183
                   6) Median income                 .221
                   7) Labour force Herfindahl       .127        .550


1) All of the independent variables listed in the table were
significant using a 95% confidence interval and are ordered by their
importance in the stepwise process.

2) All dependent and independent variables are measured using 2001
data (except population change which measures population difference
from 1996 to 2001).

3) In all cases, n is 135 Canadian urban areas (CMAs and CAs).

Table 5: Local Spatial Autocorrelation Results

                                 Average                      Clustered
                 Number of       Value of                   Firms' Value
                Significantly  Significantly    Average     as a Percent
                 Clustered      Clustered       Value of      of Toronto
                    Firms          Firms       Toronto CMA        CMA

Finance --
employment       26 of 2586       963.46          19.41         49.90%

Finance --
sales            34 of 1886       $135.00         $5.68         42.80%

Insurance --
employment       17 of 1080       1079.41         27.48         61.80%

Insurance --
sales            26 of 1048        $52.69         $2.54         51.40%

Real estate --
employment       17 of 2251       314.71          8.81          27.00%

Real estate --
sales            12 of 2236       $116.25         $2.16         28.90%


1) Significant firms have a standardized local Moran's I score of
greater than 1.96 (and therefore are statistically significant at a
95% confidence interval). A positive Z-score indicates local
clustering of cases with similar attributes.

2) all significantly clustered cases are located in 'south-central'
Toronto (see Figures 2, 3 and 4).

3) Sales are measured in millions of dollars.

Table 6: Agglomeration Impacts in South-Central Toronto (Results from
a Survey of Large FIRE Firms)

                                  FIRE      Finance  Insurance   Real

Location advantages

Access to complementary or
competitive firms              26 (23.8%)     15         5        6

Access to customers            17 (15.6%)     10         4        3

Benefit from public transit    16 (14.7%)      6         6        4

Meeting place of the city
(central location)             16 (14.7%)      6         4        6

Prestige/in the 'financial
hub'                          (15 (13.8%)      6         5        4

shopping                       11 (10.1%)      4         6        1

Access to employees             8 (7.3%)       4         3        1

Total responses               109 (100.0%)    51        33        25

Location disadvantages

Commuting time costs
(for employees or customers)   12 (21.8%)      2         5        5

Congestion                     12 (21.8%)      5         4        3

Expensive rent                 10 (18.2%)      4         3        3

Expensive parking/lack of
parking                        10 (18.2%)      3         2        5

Expensive (in general)         8 (14.5%)       3         2        3

Pollution                       2 (3.6%)       2

Cannot own building
lack of autonomy                1 (1.8%)                 1

Total responses               55 (100.00%)    19        17        19

Face to Face Contacts (with
clients or other firms) vs.
Technological Advancements

Face to face contacts are
vital                          33 (82.5%)     16        10        7

Face to face contacts are
important but less so with
time                           7 (17.5%)       2         3        2

Technology has replaced the
need for face to face
contact                         0 (0.0%)       0         0        0

Total responses                 4 (100%)      18        13        9

Number of firms contacted          40         18        13        9
Gale Copyright: Copyright 2007 Gale, Cengage Learning. All rights reserved.