The Extents of the Canadian Labour Market before 1930 March 8, 2000 J.C. Herbert Emery Department of Economics University of Calgary Calgary AB T2N 1N4 hemery@ucalgary.ca Abstract I analyse real wages series for 12 Canadian cities for the period 1901 to 1926. The analyses of these series reveal that Canada’s regional labour markets were integrated by 1926 and regional real wages in Canada were converging, but not towards equality. By 1926 a clear regional structure to real wages in Canada is apparent and the regional structure in real wages looks remarkably similar to the regional profile of per capita incomes for the period 1926 to 1962. Thus, it would appear that regional income disparities are an inherent trait of Canada’s heartland-hinterland structure. The reasons for the persistence of unequal real wages are not known, but it appears that high wage labour markets were Protestant labour markets while Catholic workers were proportionately more important in the lower wage markets. Whether the relationship between religious affiliations of workers and real wage levels across regions reflects human capital differences, or discrimination, it appears as though Canada had spatially integrated, but socially segregated, national labour markets by 1926. The existence of income differentials across Canada’s regions and provinces has been a persistent feature of the Canadian economy since at least 1890 (Green 1971). Further, the disparities in regional incomes have been unaffected by dramatic changes in market forces and government policy over the last century. More often than not, the Canadian consensus as to the proximate cause of the persistence in regional income disparities has been a dysfunctional Canadian labour market. Whether the cause of the dysfunction in the labour market is structural, a mismatch of job demands in one region and the skills of the supply of workers available from another region, or that policy related, government transfers discourage labour mobility, Canadian workers are viewed as not fully exploiting available arbitrage opportunities in Canadian labour markets. Consequently, as D. Green (1994) points out, Canadian policy recommendations for eliminating regional income differentials in Canada have focused on encouraging an expansion of the extents of Canada’s labour markets. Under a simple linear model of progressive labour market integration, government should identify and remove the main obstacles to arbitrage. Efforts to remove barriers to arbitrage have included efforts to improve information about job opportunities in other regions, and to aid in the matching of workers in one region with jobs in another region. For example, before 1930, Canadian governments established labour exchanges to facilitate the movement of under-employed workers to other provinces to find work. Today, HRDC operates an on-line National Job Bank and an on-line Electronic Labour Exchange that appear to be revivals of the preDepression approach to reducing unemployment. The roles of government programs and transfers to individuals have been viewed as a source of the labour market dysfunction. For example, Courchene (1981) highlights the role of Canadian policies like 1 unemployment insurance in reducing the degree of inter-regional labour mobility that has contributed to the persistence of regional income differentials. Consequently, reforms to Canada’s unemployment insurance system increased the emphasis on retraining displaced workers. Similarly, Coulombe (1999) argues that federal transfer payments, by enriching the levels of public goods and services in lower productivity Canadian regions, reduce the incentive for workers with high levels of human capital investment to move. Needless to say, policy efforts aimed at extending the geographic margins of regional labour markets and improving the operations of labour markets through EI reforms, and the expanded investment in training for displaced workers, have not eliminated regional income disparities. At this point there are two candidate explanations for this outcome. First, despite the best intentions and efforts of government, regional labour markets remain weakly integrated or non-integrated, leading to no or slow convergence to equal per capita incomes across regions. Second, it is possible that the failure of government initiatives to reduce income disparities through encouragement of market forces reflects that regional labour markets in Canada are adequately integrated and the regional income disparities are “equilibrium income differentials”. They are not arbitraged away because the differences in regional incomes do not reflect arbitrage opportunities. Our inability to distinguish between these two competing hypotheses at this point reflects that labour market dysfunction is a diagnosis asserted in light of regional income disparities as a symptom of non-integrated markets in a standard general equilibrium trade model. This diagnosis has been maintained despite the long time recognition of alternative models of the Canadian economy that would support different incomes in 2 equilibrium for integrated regional labour markets. A direct test of the links between Canada’s regional labour markets has not been provided. Given the existence of regional income differentials since at least 1890, an important contribution towards an understanding of the proximate reasons for the differentials would be to develop better information about the structure, evolution and operations of Canadian labour markets early in Canadian economic development. That is the purpose of this paper. By examining trends in regional real wages and regional real wage growth rates for the period 1901-1926, I seek to determine whether regional income disparities reflect that Canadian workers do not exploit arbitrage opportunities, or, whether regional income disparities reflect equilibrium income differentials that will not be eliminated through market forces. If the former is the better description, and nonintegration of regional labour markets is the source of regional income disparities, then efforts of Canadian policy makers to date are well targeted. On the other hand, if the latter description is the better of the two, then more attention should be paid to long term factors in Canadian development as possible reasons for the income disparities. For example, regional specializations in production and/or the early industrial leadership of Ontario could be causes of regional income disparities that exist in equilibrium that will not be removed by market forces since they do not reflect an opportunity for arbitrage. This paper presents analyses of real wage series for several building trades occupations in 12 Canadian cities over the period 1901 to 1926 to determine the geographic extents of Canadian labour markets. Emery and Levitt (2000) follow Bertram and Percy’s (1979) and Allen's (1994) work to construct inter-urban inter-temporal price indexes to construct real wage series for 12 Canadian cities. Analyses of the real wage 3 series suggest that by 1926 Canada’s regional labour markets were integrated to an extent that constitutes a national labour market. Real wages in Canadian cities were converging, but not to equality. Cities in Ontario, British Columbia and Alberta emerged as high wage labour markets while the Maritime provinces, Quebec, Manitoba and Saskatchewan emerged as lower wage labour markets. At this point I offer some speculative insights into the reasons for the equilibrium real wage differentials that point to the Canadian labour market as spatially integrated but socially segregated. 1. Labour Mobility, Income Disparities and the Canadian Labour Market Economic theory relating to adjustments in regional inequalities highlights interregional flows of people, goods or funds (Mansell 1975, 5). Heckscher-Ohlin theories of trade between nations, or regions, demonstrate that factor prices (wages paid to labour, and the return to capital) are equalized through commodity trade. Since per capita incomes that are often analysed in convergence studies reflect factor earnings, factor price equalization across regions would suggest that regional per capita incomes would also converge. If trade in commodities between regions is limited, but labour and capital move freely between regions, then I would expect labour and capital to migrate between regions until wages and the return to capital are equalized in both places. Barry (1996) points out that in traditional trade models, labour mobility between regions should equalize both GDP and GNP per head in both regions. An alternative to the trade models are neo-classical growth models. In the standard two-factor growth models (labour and capital), factor mobility always promotes convergence in per capita output (Taylor 1999). 4 In the context of the convergence predictions of standard trade models and neocclassical growth models, Mansell (1975) suggests that a puzzling aspect of the lack of convergence in income levels across Canada’s provinces is that Canada has had a high degree of inter-regional factor mobility and essentially free inter-regional trade. Further, Mansell suggests that an important challenge is to explain how observed wage and income inequality in Canada could have been maintained over time when there have been large differences in rates of development and growth. Inter-regional factor mobility and inter-regional trade in goods are unquestionable features of Canadian history. Consider the years 1896 to 1914, a key period in Canadian economic development. The Canadian prairies were settled, wheat was exported, and the Canadian economy grew rapidly. Under the traditional Wheat Boom story for this period, prairie wheat exports "transformed the static and isolated regions (of Canada) into an integrated and expanding national economy." (Rowell-Sirois Book I, 93). While the term "national economy" is vague as it combines notions of goods market integration and factor market integration, the emergence of a national labour market is part of the story: The development of the West was a national achievement and the participation of all areas in a common effort fostered a new sense of nationhood. Sons and daughters of the Maritimes and Central Canada migrated to the plains and built up the West, thus forging innumerable links between older Canada and the new." (pp. 91-92) A popular perception of the Canadian labour market at the turn of the century which suggests the existence of an integrated national labour market comes from Avery (1979, 8) where in February a labourer might find himself "a lumber worker in Iroquois Falls, Ontario; in June a railroad navvy along the National Transcontinental; in August a harvester in Grenfell, Saskatchewan; in November a coal miner in Fernie, British 5 Columbia." In a similar vein, Voisey (1988) describes the high degree of mobility that was typical of the pioneers who settled Vulcan Alberta before 1910. One 31 year old Vulcan settler in 1905 left Ontario for Winnipeg; moved onto Calgary; laboured throughout British Columbia; panned for gold in the Klondike and soldiered in South Africa before moving to Alberta. The mobility of workers into Canada, and across Canada’s regions is not only implied by anecdotal evidence. Quantitative evidence shows that 1890 to 1910 was a period of massive population redistribution across Canada's regions. Green (1994, 156157) relates the population movements to the integrating influence of the rise of prairie wheat exports: The Western economy was the key factor in Canadian development between 1900 and 1930. At the turn of the century it was largely unsettled and was only Weakly integrated into the Canadian economy as a whole. By 1930 this situation was completely reversed… At the center of this transformation was a large movement of people. The immigrants came not only from eastern regions in Canada but from Europe, especially Britain, and a significant number came from the United States… The rapidity of this transformation from uninhabited lands to a settled viable economy remains one of the central dynamic features of Canadian development during this three-decade period. Beyond the observed population and goods movements, there are other reasons to believe that Canada could have had a national labour market before 1930. Canada's rapid economic development before World War I coincided with what Williamson (1995,1996) characterizes as an emergence of a global economy with convergence of real wages across countries. By 1890, Canada had a transcontinental railway to link the prairie region with the East. Good information about economic opportunities on the prairies was available. Struthers (1983) describes in detail how the Federal government created and 6 operated "labour exchanges" to aid the dissemination of information to enhance labour flows in Canada. Further, immigrants responded to labour market conditions in deciding where in Canada they would settle (Green and Green 1993). While the "necessary conditions" for the existence of an integrated national economy seem to have been met before 1930, some evidence would appear to be at odds with the existence of an integrated national labour market. If regional economies are integrated, one expects convergence in income levels, the opposite of what Green (1967, 1971) finds over the Wheat Boom period. Green (1971) suggests the divergence of provincial incomes could reflect a Kuznets Curve description of Canadian economic development. In the early stages of development, inequality of regional income distribution increases and over time decreases as regional incomes converge. Williamson (1965) finds that with the exception of Canada, this pattern has been typical for the history of developed nations. For Canada, both Green (1971) and Williamson (1965) find that any convergence as part of the Kuznets relationship was weak at best. McInnis (1968) argues that Canada did not display any Kuznets effect making its development process an exception to the usual development process. Norrie and Owram (1996, 345) interpret movements of labour (and capital) over Canada’s history as evidence that factor flows necessary for factor price equalization across regions were taking place. They pose the question that remains; why were these flows insufficient to remove regional income differences? One possibility is that World War I, World War II and the Depression interrupted the convergence of regional incomes in Canada as Williamson (1995) suggests these events interrupted the development of the 7 global labour market. If this is the case I have to address why the convergence process did not resume after World War II. Another possibility for the existence of regional income disparities is that despite the observed movement of labour and goods, Canada’s regional economies were not integrated to a large enough extent. Struthers (1983) recognizes the existence of factor and commodity movements across Canada, but he sees the Canadian economy as somewhat less than a national market: "Staples also helped to create an economy that was more regional than national. Until the development of the Canadian West as a major wheat-producing region between 1896 and 1930, few economic links existed to bind the diverse regions of Canada together. Trade was more often with an external metropolis than with another Canadian province. As a result, the economic fortunes of Canada's regions, particularly where secondary manufacturing was Weak, could and did vary widely, depending on the health of the export staples. Unemployment rates east of the Ottawa River have traditionally been double those of Ontario and the West, and wage levels between regions have varied widely as Ill. Until the 1920s, labour markets in Canada flowed more frequently along a northsouth than an east-West axis, with the United States providing a major focal point for the talented and the jobless…To the extent that a `national' labour market did exist, it revolved around Western wheat which, in the words of the RoIll-Sirois report, 'transformed the static and isolated regions into an integrated and expanding national economy." (page 5) Boyer and Hatton (1994, 84) point out that there is no necessary link between migration and the extent of market integration; "analyses of the pattern and extent of migration movements shed little light on the issue of integration. Markets could be perfectly integrated but exhibit little migration or they could exhibit high rates of migration but be poorly integrated." Boyer and Hatton suggest that patterns of labour market integration resulting from economic development reflect the existence of established channels of migration between different regions. Rosenbloom (1996) 8 emphasizes similar notions with his discussion of labour market institutions that influenced migration patterns such as chain migration where current migrants follow the same paths as their predecessors. Unlike a convergence in per capita incomes across regions, a lack of convergence in wages paid for labour despite large migrations would not be unique to Canadian history. Boyer and Hatton (1994, 84) find that in Victorian Britain, during a time of "considerable migration opportunities for arbitrage were not fully exploited." Similarly, Rosenbloom (1990, 98) finds that the persistence of large regional real wage differentials in the US after the Civil War suggests that "significant variations in the relative scarcity of labor persisted over more than three decades and that potential opportunities for arbitrage went unexploited." Allen (1994, 125) finds that "despite massive migration both ways across the border with the United States, Canada preserved a distinctive wage structure." In contrast to Williamson's (1995) view of integrated international labour markets, Allen (1994) feels that the persistence of distinctive wage structures despite massive migrations makes it impossible to believe that labour markets were well integrated internationally. A. Green (1994) similarly finds that rural-urban wage ratios for the Canadian prairies moved quite differently from rural-urban wage ratios for the United States. If Canada could remain non-integrated with international labour markets, in particular that of the US, and it is the national level of analysis rather than the international level, then would it not follow that that regions within Canada could also be non- integrated? As D. Green (1994, 32) "If a different wage structure can develop in Canada relative to the United States in spite of massive migration, why cannot a different 9 structure develop in the Atlantic region relative to the rest of Canada, again in spite of a massive migration?” If Canada’s regions developed different wage structures, then we also need to know why they did. Are the wage differentials transitory but the rates of convergence are slow? Are the wage differentials equilibrium differentials resulting from the regional structure of the Canadian economy? Or are the differentials resulting from non-integration of regional labour markets? 2. Labour Market Integration: Concepts and Measures An economist’s notion of integration of geographically distinct markets is based upon the notion of the law of one price. In the absence of transportation costs, if buyers and sellers of a homogeneous good have complete information of the price of that good in the two markets, then in equilibrium the equilibrium price for the good is the same in both markets (Stigler and Sherwin 1985). If demand for the good rises in market A, then the price of the good will rise at A triggering additional supply into A from market B. Equilibrium is restored once the expansion of supply is large enough to restore the equality of the price in market A and B. This concept of one price can be generalized such that if there are transportation costs, or local amenities or disamenities associated with a given location, then prices can differ in equilibrium, but by no more than the amount of transportation costs or “compensating differentials”. In applying this notion of market integration to labour market integration, Rosenbloom (1990) characterizes a completely integrated labour market as one where workers at every location are aware of all employment opportunities elsewhere and through migration can offer their services to employers elsewhere. If there are no sitespecific amenities or disamenities, labour market equilibrium will have all workers of 10 equal ability doing identical work receiving the same real wage. If workers are not indifferent to the location where they work, migration between locations will cease when real wages differ in equilibrium only to the extent that the individuals are indifferent between working in the two locations. Labour market integration is most likely to be less than complete and determining the degree of labour market integration of two locations is primarily an empirical problem. Complete integration may be prevented by imperfect information about employment opportunities in distant locations; by institutional and financial constraints that impinge on a potential migrant’s ability to act on information as to employment opportunities in other places, and given that migrating is costly and potentially irreversible, uncertainty as to the persistence of favorable labour market conditions in other markets may lead workers to delay or forego acting on information about employment opportunities. For our purposes, I am interested in determining the degree of labour market integration across Canada’s regions. A popular approach to this problem has been to calculate the correlation coefficient of wage changes in two labour markets (Stigler and Sherwin 1985). The higher the correlation coefficient, the more closely integrated two markets are assumed to be. Boyer and Hatton (1994) point out that an important shortcoming of this approach is that one cannot distinguish between a strong tendency for instantaneous arbitrage (market integration) and common shocks to labour demand and supply at each location (spurious correlation). A second approach is to search for trends over time in measures of wage dispersion (e.g. the coefficient of variation) across several markets. Diminishing dispersion of wage rates is interpreted as evidence of an increased 11 degree of labour market integration. As with the correlation coefficient of wage changes across locations, the measure of dispersion approach has limitations for identifying labour market integration; an absence of a distinct trend in the measure of dispersion could reflect either very well integrated markets and very poorly integrated markets (Boyer and Hatton). I follow the approach of Boyer and Hatton (1994) who apply an error-correction modeling approach that is based on a structural model of the forces that determine the relationship between wage rates in two locations. Boyer and Hatton present a model where migration is the key force for wage convergence between two labour markets i and j. They express the net migration of labour from region j to region i at time t as: W m jit c log it 1 k W jt 1 c is a measure of the responsiveness of migration from j to i to a given wage differential between markets i and j at time t-1. The overall size of net migration from j to i depends upon the size of the wage differential and on the responsiveness of migration to the wage differential. c measures the degree of integration between labour markets j and i. As c approaches infinity, labour is perfectly mobile and the two markets are perfectly integrated. k represents a long run equilibrium wage ratio between i and j due to factors such as migration costs, differential labour productivity due to regional specialization in 12 production or different production functions, or non-wage advantages of market j to market i.1 Migration influences the wage ratio for the two markets through the adjustment of labour supply in each market. Suppose that a positive labour demand shock affects market i so the Wi rises relative to Wj. If c>o, hence migration of labour links the two markets, then labour will flow from j to i until the equilibrium wage ratio is restored. For example, if log(Wi/Wj)>k, then labour moves from j to i which will tend to lower Wi and/or possibly raise Wj. The migration continues until log(Wi/Wj)=k. The speed at which the equilibrium is restored depends upon the magnitude of c. The closer c is to 0, the longer the time to convergence to the equilibrium wage ratio. As c tends to infinity, adjustment is instantaneous and there is no meaningful sense in which a demand shock can affect i but not j since they are the same market. The long run tendency is towards the wage ratio Wi/Wj =ek. If k=0, then the long run equilibrium wage ratio is Wi/Wj=1; real wages are equal in equilibrium. If k0, then I would observe convergence to an equilibrium wage differential. In terms of identifying the possible sources of regional income disparities in Canada that may result from labour market integration, I are interested in three possibilities. First, if c=0, hence two labour markets are not integrated, possibly for reasons such as imperfect information or government intervention, then it is possible that real wage differentials reflect that arbitrage opportunities Were not exploited. Second, if c0 and k=0, then markets are integrated and real wages are equal in the long run. Failure to observe convergence in real wages between markets could reflect that the rate 1 In the latter case, an example would be if market i is in a small isolated place, then workers may require higher wages to be indifferent between working in i as opposed to j. In equilibrium labour will only migrate from j to i if the wage in i is at least k percent higher than the wage at j. 13 of convergence is slow, and/or market disruptions hamper the convergence tendency. A third scenario is possible where government initiatives to encourage an expansion of the extents of Canada’s labour markets will do nothing to eliminate regional wage disparities. If c0 and k0, then labour markets are integrated but convergence is to a long run equilibrium differential. If k0, then we cannot expect market forces to eliminate regional wage disparities. To generate an empirical model, Boyer and Hatton introduce a simple model of labour demand in two markets (which also allows for migration to and from third markets)2 that allows them to express the relation between two labour markets in terms of the wage alone. The model yields the following estimable relationship: W log( Wit ) d0 d1 log( W jt ) d 2 log it 1 vt W jt 1 d1 reflects the degree to which common forces affect markets i and j. d2 is a measure of the degree of integration between the two markets and reflects the size of the unobserved mobility parameter c. If c=0, then d2=0. Thus, the sign and size of d2 implies the direction and strength of migration flow between the two markets. This type of model is an error correction model. Boyer and Hatton highlight that the two wage rates are related in changes but the error correction term d2log(Wi/Wj) prevents them from diverging over time in levels if d2 is negative. 2 Integration of two Canadian labour markets could reflect the influence of a common third market. For example, workers in Halifax and Montreal may not view the other city as a possible place to move for work, but workers in both cities respond to wage changes in the Northeast US. It could also reflect the influence of immigrants from Europe who view Montreal and Halifax as alternative locations for work. Further, in a given urban labour market the potential supply of labour will reflect the responsiveness of labour outside the region to positive local wage shocks and the responsiveness of the workers in the rural hinterland for the city. 14 This empirical specification also yields an estimate of the long run equilibrium wage ratio for markets i and j. A long run stationary equilibrium is characterized by log(Wit)= log(Wjt)=vt=0. In this long run equilibrium, log(Wi/Wj)=-d0/d2, thus –d0/d2 is an estimate of k. Finally, d2 can be used to express the speed of adjustment of the wage in region i to a shock to the equilibrium wage ratio holding Wj fixed. The predicted lag between an initial shock and a return to equilibrium is: Ta 1 d2 d2 This measure reveals that if d2=-1, then market are perfectly integrated since the return to equilibrium is instantaneous. Finally, the regression equations also include a time trend variable and a variable which interacts the time trend variable with log(Wi/Wj). As such, I will be able to characterize d0 and d2 in 1901 and 1926. 3. Wage Data and an Inter-Urban Intertemporal Price Index To estimate the error correction model introduced in the previous section of the paper, I need observations on real wages for different labour markets in Canada. Emery and Levitt (2000) construct real wage series for 12 Canadian cities with data from Department of Labour supplement to the Labour Gazette on hours and wages. From this source I obtained nominal hourly wage rates for carpenters, bricklayers and builders’ labourers for the period 1900 to 1926. The cities are: St. John, Halifax, Quebec City, 15 Montreal, Ottawa, Toronto, Hamilton, Winnipeg, Regina, Calgary, Edmonton and Vancouver.3 Figure 1 shows Emery and Levitt’s (2000) inter-urban price indexes that yield real wages that reflect the purchasing power of the wage in city j at time t in terms goods in the base city (Toronto) in the base year (1913). This intertemporal inter-urban index is obtained by dividing the price of commodity m in city j at time t by the price of commodity m in Toronto in 1913 to create the price relatives that are then aggregated together using the budget expenditure weights in Bertram and Percy (1979). For the years between 1900 and 1905 and between 1905 and 1909 I have used linear interpolation to get the price indexes values. Before 1914 there are clear differences in price levels between cities West of the Lakehead and east of the Lakehead. The higher price levels prevailing in the Western cities reflect that food prices were somewhat higher than food prices in the east, but rental costs for housing were substantially higher. High rents for housing and high food costs were a characteristic of the Canadian prairies in the frontier stages according to Voisey (1988). Through World War I, prices in all cities increased dramatically. After World War I, price levels fell and stabilized across Canada and the Western/eastern division in cost of living had disappeared somewhat. Despite MacKinnon’s (1996) concerns over the usefulness of the Department of Labour wage data, I chose to focus our analysis on them for several reasons. First, the Department of Labour data provides us with a larger set of cities than the CPR wage data set. Second, the CPR wage data are primarily useful for general wage trends before 1918. After 1918, institutional factors specific to the determination of railway wages become important. Finally, our analyses of the MacKinnon CPR data suggested that I may be observing a wage policy of a large firm that may or may not be representative of the broader labour market. 3 16 Before World War I nominal wages for these occupations were highest in Vancouver, the prairie cities and the Ontario cities than they were in the Quebec and Maritime cities. Nominal wages increased substantially through the war, fell and then stabilized after the war. By 1926, there is a split in the levels of nominal wages East of Ottawa and West of Montreal for carpenters. A similar divergence also holds for carpenters. MacKinnon’s CPR wage data shows a similar division in the levels of nominal wages between the prairies and BC and Ontario and Quebec, at least until after World War I at which time the regional structure of wages for various railway occupations disappear for institutional reasons specific to railways. As with the Historical Statistics wage data, nominal wages increased dramatically during World War I, fell and stabilized after 1920. While nominal wages in the West of Canada Were higher than in the east, it was Ill known that the cost of living was higher in the West, and the early conditions of life in the West made it a less desirable place to settle than the east.4 Figure 2 shows the nominal wage series for builders’ labourers deflated by the interurban price index. Real wages were highest in Vancouver, Hamilton and Toronto. Despite high nominal wages in prairie cities, real wages on the prairies before World War I are as low as real wages in Quebec and the maritime cities. For the labourers, real wages in Regina and Winnipeg remain below those of Vancouver and the Ontario cities. In contrast, real wages paid in Calgary and Edmonton after World War I compare favorably to those paid in Ontario and For example, Voisey (1988, 179) observes that “Even though rural schoolboards offered higher wages than might be obtained back east, the teachers seldom gained, for they also faced proportionately higher living costs. Many did not wish to work in an isolated frontier schoolhouse in any event. And even in prosperous times school-boards did not always pay on time and in full.” Thus an interesting issue is how real wages compare across Canadian cities over this period. 4 17 BC. For carpenters, the story is somewhat different as the real wages in the prairie cities rise to similar levels as Vancouver and the Ontario cities after 1907. For both the labourer and carpenter occupations, clear regional divisions in real wage levels are apparent by 1926. The maritime and Quebec cities have the lowest real wages while Vancouver and the Ontario heartland cities have the highest real wages. MacKinnon’s CPR wage data provides a slightly different picture than the Department of Labour wage data. Deflating the CPR wages suggests that real wages Were lowest for Quebec, and highest for BC. Real wages in Ontario and the prairies were roughly equal and lay in between the Quebec and Vancouver levels. There is still a clear regional structure to real wage levels (at least before 1918) but the disadvantage of the prairie work location relative to Ontario is not apparent. For CPR labourers, real wages for Ontario based workers are higher than Prairie based workers until 1907. For the CPR fitters and machinists, there is no systematic difference in the wage levels between regions. The Error Correction Estimates The error correction model for wage changes in two labour markets was estimated for each of 132 city pairs for the wage data from the Department of Labour sources. The observations of real wages for all three building trades are pooled for the estimations. Occupation dummy variables are included to allow for different constant terms but the d2 is common to all three trades. The detailed regression output has not been included in this draft due to the volume of numbers to report. Instead, the statistics relating to the 18 degree of market integration between city pairs and the implied equilibrium wage ratios are discussed exclusively.5 Figure 3 presents matrices of labour flows from one city to another implied by the estimate of d2 in the error correction specification for 1901 and 1926. The cities in the rows of the matrices are the cities receiving labour, while the cities in the columns are the cities sending labour. To help interpret the numbers, shading of cells reflects the size of the d2 estimates. The lighter the shading in the cell, the stronger is the implied integration of the two cities. The upper panel of Figure 3 presents the estimates of d2 for all of the city pairs. The estimates suggest that labour markets in Canada in 1901 were small and regional in nature. Only 25 of 132 estimates of d2 are significant at size 5 percent. With the exception of Montreal, the estimates suggest some integration of all cities Toronto and East of Toronto. Hamilton appears to have been neither a recipient of Maritime workers, nor was Hamilton a source of workers for the Maritimes. The estimates for 1901 also suggest that Toronto and Winnipeg were strongly integrated with two way flows of labour. Winnipeg was receiving labour from the Maritime cities, Quebec city and Toronto. Regina also received labour from the same cities, and Ottawa. The estimates suggest that labour responded to opportunities within its own region and labour in the 5 The detailed regression results are available upon request. A time trend variable is also included in the model specification. The error correction models Were also estimated using the CPR wage data. These results are not reported here. The CPR wage data for 1901-1926 and 1901-1914 indicate that Toronto, Winnipeg and Montreal Were one labour market; labour flowed to Vancouver from Toronto and Montreal but not from Vancouver to those cities. Winnipeg and Vancouver Were integrated markets for CPR workers. The estimations reveal that real wages were converging to equilibrium differentials that reflect the regional wage structures revealed in Figure 7. In summary, compared to the Dept. of Labour data, the degrees of integration are higher but the overall story of integrated labour markets with convergence to equilibrium real wage differentials is the same.. 19 eastern markets was responding to opportunities in Western Canada. The strength of integration was generally weak however. The lower panel of Figure 3 presents that estimates of d2 for city pairs in 1926. With some notable exceptions, the strength of labour market integration has generally increased over that observed for 1901 as the estimates of d2 for 1926 are larger. In addition, 113 of 132 estimates of d2 are significantly different from zero at size 5 percent. By 1926 Canada had a national labour market. The lower values for d2 for some city pairs reflect the conditions of some markets by 1926. For example, workers appear to be shunning Regina by 1926, while workers in Regina were responding to opportunities in any city West of Montreal. Similarly, by 1926, only workers in Montreal appear to have been interested in wages in Winnipeg. Workers in Winnipeg appear to have been moving to any city other than Regina and Quebec City. Reflecting the movement of the frontier of settlement beyond Saskatchewan to Alberta, by 1926, Alberta was receiving workers from across all of Canada. Finally, the Maritime and Quebec cities are largely a source of labour for other regions and less of a destination for labour from other regions. Figure 4 presents the estimates of the long run real wage ratios between the set of cities and Toronto in 1901 and 1926. The profile reveals that in 1901 Vancouver and the Ontario cities had the highest real wages in the country. The prairie, Quebec and Maritime cities had lower and comparable real wages. The prairie labour markets were not lucrative labour markets early in the wheat boom period. At the same time, land purchase may have been an important attraction for migrants and that may explain the links between Toronto, Winnipeg and Regina despite the low prairie wages. 20 By 1926, once price levels across the regions have converged, the equilibrium wage profile has undergone some changes from that of 1901, but still real wages are not converging to equality. The maritime city wages relative to Toronto wages are unchanged from 1901. In the prairie cities, the ratio of real wages to Toronto real wages has increased. Winnipeg and Regina remain lower wage labour markets, but wages are higher than in the maritime cities. In contrast, Calgary and Edmonton have real wages that are as high, if not higher than those paid in Toronto. By 1926 regional real wages converged in a way that Canada had a distinct regional wage structure that suggests that the absence of complete factor price equalization was not the result of non-integrated, or weakly integrated regional labour markets. The long run tendency of the national Canadian labour market was to have low wage regions and high wage regions. The question that remains is why these long run real wage differences exist across the regions. Models that support equilibrium real wage ratios typically present situations in which labour is not perfectly mobile. This imperfect mobility must then be reconciled with the view that Canada had an integrated national labour market where labour did move across regions in sufficient amounts to exhaust arbitrage opportunities. My speculation at this point is that the more accurate portrayal of Canada’s labour market is one of geographically integrated, but socially segregated labour markets. After examining several characteristics of the populations of the 12 cities one feature of the populations appears to have some explanatory power for the nature of the real wage profile. Figure 5 presents the proportions of city populations that were Catholic in 1911 and 1931. Other than Montreal and Quebec, which are overwhelmingly 21 Catholic, the more catholic was a city’s population, the lower was the real wage in the city. The highest wage cities of Toronto and Vancouver Were dominated by Protestants. Ethnicity, age structures and other features of the city populations do not provide the same contrast across the low and high wage regions. Why would religious affiliation matter? There are at least two stories that could explain the correlation between Catholicism and real wage levels. First, Catholics on average had lower investments in human capital than Protestants, and on average, Catholics in western city populations were more likely to be recent immigrants to Canada than Protestants. Thus, it could be the case that low wage regions reflect lower human capital endowments in the region than for high wage regions. This story would be consistent with Coulombe’s (1999) argument that post-1945 convergence in provincial incomes reflects the investments in education in the eastern provinces that promoted the mobility of Maritimers to other labour markets. A human capital story is also supported by the observation that within given cities, or regions, there is often found to be little difference in the socio-economic status of Protestants and Catholics. Thus, high human capital Catholics can compete in high wage regions, but the lower human capital Catholics prefer to take advantage of the greater opportunities for them in low wage markets. Galbraith (1985) argues a similar story when he suggests that it was only the less successful citizens of Elgin County moved to prairies. A second story is a discrimination story. Catholic workers may have found limited opportunities in Protestant dominated labour markets. In Ontario, organizations like the Orange Order and the Protestant Protective Association (whose members took an oath not to where a Catholic if a protestant was available) raised the costs of entry into 22 the local labour markets for Catholic workers. In this story, Protestant workers move across regions and take advantage of job opportunities open to Protestants, while Catholic workers moved across regions to take advantage of job opportunities available to Catholics. Ultimately, there should be testable implications of these two stories that will allow me to distinguish which is the more plausible story. At this point I will leave the paper with a speculative conclusion that Canada had more than one national labour market. The markets were spatially integrated, but socially segregated from one another. Conclusions I have constructed real wages series for 12 Canadian cities for the period 1901 to 1926. Our analyses of these series reveal that Canada’s regional labour markets were integrated enough to constitute a national labour market. Real wages in Canada were converging, but not towards being equal across regions. By 1926 a clear regional structure to real wages in Canada was apparent and the regional structure in real wages looks remarkably similar to the regional profile of per capita incomes for the period 1926 to 1962. Interestingly, the low wage regions are those which had a higher proportion of Catholics in the population. Catholics on average had lower investments in human capital and often were more recent immigrants to Canada. Catholics also may have faced lesser opportunities in labour markets dominated by Protestant workers. Thus, discriminatory barriers in Ontario directed Catholics to the prairies and Maritimes. In this sense Canada had two national labour markets that remained socially segmented. 23 References Panos C. Afxentiou and Apostolos Serletis (1999) “Convergence Across Canadian Provinces,” University of Calgary Department of Economics Discussion Paper 99-03, forthcoming in Canadian Journal of Regional Science. R.C. Allen (1994) “Real Incomes in the English Speaking World, 1879-1913,” in George Grantham and Mary MacKinnon eds. Labour Market Evolution: The Economic History of Market Integration, Wage Flexibility and the Employment Relation (London and New York: Routledge). Gordon Bertram and Michael B. Percy (1979) "Real Wage Trends in Canada 1900-26: Some Provisional Estimates," Canadian Journal of Economics XII(2), 299-312. D.M. Cameron (1981) "Regional Economic Disparities: The Challenge to Federalism and Public Policy," Canadian Public Policy VII(4), 500-505. Gerald A. Carlino and Leonard O. Mills (1993) "Are U.S. Regional Incomes Converging? A Time Series Analysis," Journal of Monetary Economics 32, 335-346. R.H. Coats (1915) The Rise of Prices and the Cost of Living in Canada, 1900-1914, Board of Inquiry into Cost of Living, Vol. II (Ottawa: King’s Printer). Serge Coulombe (1999) Economic Growth and Provincial Disparity: A New View of an Old Canadian Problem, Commentary 122 (C.D. HoI Institute, March). Serge Coulombe and Frank C. Lee (1995) “Convergence Across Canadian Provinces, 1961 to 1991,” Canadian Journal of Economics XXVIII (4a), 886-898. Thomas J. Courchene (1981) "A Market Perspective on Regional Disparities," Canadian Public Policy VII(4), 506-518. Department of Labour, Canada Labour Gazette. Department of Labour, Canada, Wages and Hours of Labour in Canada. Price V. Fishback (1998) "Operations of `Unfettered' Labor Markets: Exit and Voice in American Labor Markets at the Turn of the Century," Journal of Economic Literature XXXVI, 722-765. George Grantham and Mary MacKinnon eds. (1994) Labour Market Evolution: The Economic History of Market Integration, Wage Flexibility and the Employment Relation (London and New York: Routledge). Alan G. Green (1967) “Regional Aspects of Canada’s Economic Growth, 1890-1929,” Canadian Journal of Economics and Political Science XXXIII, 232-245. 24 Alan G. Green (1971) Regional Aspects of Canada's Economic Growth (Toronto: University of Toronto Press). Alan G. Green (1994) "International Migration and the Evolution of Prairie Labor Markets in Canada, 1900-1930," in Hatton and Williamson eds. Migration and the International Labor Market, 1850-1939 (London and New york: Routledge). Alan G. Green and David A. Green (1993) "Balanced Growth and the Geographical Distribution of European Immigrant Arrivals to Canada, 1900-1912," Explorations in Economic History 30, 31-59. David A. Green (1994) “Bridging the Gap Between Labour Economics and Economic History: A Labour Economwest’s Perspective,” in George Grantham and Mary MacKinnon eds. Labour Market Evolution: The Economic History of Market Integration, Wage Flexibility and the Employment Relation (London and New York: Routledge). George Boyer and Timothy J. Hatton (1994) “Regional Labour Market Integration in England and Wales, 1850-1913,” in George Grantham and Mary MacKinnon eds. Labour Market Evolution: The Economic History of Market Integration, Wage Flexibility and the Employment Relation (London and New York: Routledge).Grantham and MacKinnon. Timothy J. Hatton and Jeffrey G. Williamson (1991) "Integrated and Segmented Labor Markets: Thinking in Two Sectors," Journal of Economic History 51(2), 413-425. Timothy J. Hatton and Jeffrey G. Williamson eds. (1994) Migration and the International Labor Market, 1850-1939 (London and New York: Routledge). Harold A. Innis (1956) Essays in Canadian Economic History, Mary Q. Innis ed. (Toronto: University of Toronto Press). Harold A. Innis (1940) "The RoIll-Sirois Report," Canadian Journal of Economics and Political Science VI, 562-571. Sukkoo Kim (1998) "Economic Integration and Convergence: U.S. Regions, 18401987,” Journal of Economic History 58(3), 659-683. Paul Krugman (1991) "Increasing Returns and Economic Geography," Journal of Political Economy 99(3), 483-499. Paul Krugman (1991) "History and Industry Location: The Case of the Manufacturing Belt," American Economic Review 81(2), 80-83. 25 Paul Krugman and Anthony J. Venables (1995) "Globalization and the Inequality of Nations," Quarterly Journal of Economics CX(4), 857-879. Paul Krugman (1998) "Space: The Final Frontier," Journal of Economic Perspectives 12(2), 161-174. Peter H. Lindert and Jeffrey G. Williamson (1985) "Growth, Equality and History," Explorations in Economic History 22, 341-377. Mary MacKinnon (1996) "New Evidence on Canadian Wage Rates, 1900-1930," Canadian Journal of Economics XXIX(1), 114-131. William L. Marr and Donald G. Paterson (1980), "Regional Growth and Retardation" Chapter 13 in Canada: An Economic History (Toronto: Gage Publishing). John N. McDougall (1991) "North American Integration and Canadian Disunity," Canadian Public Policy XVII (4), 395-408. Robert L. Mansell (1975) “Canadian Regional Inequality: The Process of Adjustment,” Unpublished Dissertation, University of Alberta. Marvin McInnis (1968) "The Trend of Regional Income Differentials in Canada," Canadian Journal of Economics 1(2), 440-470. Kenneth Norrie and Douglas Owram (1996) A History of the Canadian Economy Second Edition (Toronto: Harcourt Brace & Company). Kevin O'Rourke and Jeffrey G. Williamson (1994) "Late Nineteenth-Century AngoAmerican Factor-Price Convergence: Were Heckscher and Ohlin Right?" Journal of Economic History 54(4), 892-916. Harvey S. Perloff, Edgar S. Dunn Jr., Eric E. Lampard and Richard F. Muth (1960) Regions, Resources, and Economic Growth (Lincoln: University of Nebraska Press). Richard Pomfret (1993) The Economic Development of Canada Second Edition (Scarborough: Nelson Canada). Raymond Robertson (1997) "Wage Shocks and North American Labor Market Integration," Syracuse University The MaxIll School Department of Economics Working Paper #84. Joshua Rosenbloom (1990) "One Market or Many? Labor Market Integration in the Late Nineteenth Century United States," Journal of Economic History L(1), 85-107. Joshua L. Rosenbloom (1991) "Occupational Differences in Labor Market Integration: The United States in 1890," Journal of Economic History 51(2), 427-439 26 Joshua Rosenbloom (1996) "The Extent of the Labor Market in the United States, 18501914," NBER Hwestorical Paper 78. Joshua L. Rosenbloom (1996) "Was There a National Labor Market at the End of the Nineteenth Century? New Evidence on Earnings in Manufacturing," Journal of Economic History 56(3), 626-656. Matthew J. Slaughter (1997) "Per Capita Income Convergence and the Role of International Trade," American Economic Review 87(2), 194-199. George J. Stigler and Robert A. Sherwin (1985) “The Extent of the Market,” Journal of Law and Economics 28, 555-585. William A. Sundstrom and Joshua L. Rosenbloom (1993) "Occupational Differences in the Dispersion of Wages and Working Hours: Labor Market Integration in the United States, 1890-1903," Explorations in Economic History 30, 379-408. Alan M. Taylor (1999) “Sources of Convergence in the Late Nineteenth Cnetury,” European Economic Review 43, 1621-1645. Paul Voisey (1988) Vulcan: The Making of a Prairie Community (Toronto: University of Toronto Press). Jeffrey G. Williamson (1965) "Regional Inequality and the Process of National Development: A Description of the Patterns," Economic Development and Cultural Change XIII (4), 3-85. Jeffrey G. Williamson (1995) "The Evolution of Global Labor Markets Since 1830: Background Evidence and Hypotheses," Explorations in Economic History 32(2), 141196. Jeffrey G. Williamson (1996) "Globalization, Convergence and History," The Journal of Economic History 56(2), 277-306. Edward N. Wolff (1991) "Capital Formation and Productivity Convergence Over the Long Term," American Economic Review 81(3), 565-579. 27 FIGURE 1: Intertemporal Inter-Urban Price Indexes for 11 Cities, (Toronto 1913=100) 200 180 160 TORONTO 140 VANCOUVER EDMONTON 120 REGINA WINNIPEG 100 HAMILTON OTTAWA 80 MONTREAL QUEBEC SAINT JOHN 60 HALIFAX 40 20 0 1895 1900 1905 1910 1915 1920 1925 1930 28 FIGURE 5: Real Wages of Builders' Labourers, 1901-1926 0.8 0.7 0.6 TORONTO VANCOUVER 1913 Toronto $/ hour 0.5 EDMONTON REGINA WINNIPEG 0.4 HAMILTON OTTAWA MONTREAL QUEBEC 0.3 SAINT JOHN HALIFAX 0.2 0.1 0 1895 1900 1905 1910 1915 1920 1925 1930 29 d21901 To/From St John Halifax Quebec Montreal Ottawa Toronto Hamilton Winnipeg Regina Calgary Edmonton Vancouver St John 1 0.19 0.13 0.13 0.18 0.29 0.14 0.41 0.17 0.28 0.17 0.32 Halifax 0.11 1 0.48 0.03 0.48 0.38 0.08 0.24 0.13 0.17 0.16 0.14 Quebec 0.12 0.59 1 0.17 0.8 0.32 0.23 0.36 0.32 0.17 0.13 0.21 Montreal 0.23 0.2 0.35 1 0.29 0.05 0.02 0 0.02 0.4 0.09 0.11 Ottawa 0 0.19 0.4 0 1 0.23 0.16 0.05 0.18 0 0 0 Toronto 0.14 0.32 0.24 0 0.17 1 0.18 0.57 0.23 0 0.06 0.09 Hamilton 0.02 0.01 0.15 0 0.09 0.38 1 0.15 0.1 0 0 0.19 Winnipeg 0.49 0.41 0.4 0 0.19 0.62 0.19 1 0.08 0 0.03 0.24 Regina 0.37 0.33 0.49 0 0.5 0.47 0.37 0.4 1 0 0.29 0.3 Calgary 0.21 0.25 0.23 0.27 0.13 0.07 0.16 0 0 1 0.05 0.21 Edmonton 0.28 0.58 0.23 0 0 0.23 0.1 0.04 0.02 0 1 0.13 Vancouver 0.32 0.13 0.19 0 0.14 0.31 0.3 0.2 0.15 0.12 0 1 d21926 To/From St John Halifax Quebec Montreal Ottawa Toronto Hamilton Winnipeg Regina Calgary Edmonton Vancouver St John 1 0.69 0.44 0.62 0.38 0.39 0.47 0.41 0.23 0.38 0.49 0.37 Halifax 0.63 1 0.35 0.58 0.42 0.39 0.5 0.47 0.32 0.48 0.43 0.35 Quebec 0.39 0.41 1 0.24 0.51 0.53 0.29 0.3 0.35 0.46 0.31 0.25 Montreal 0.54 0.64 0.22 1 0.34 0.41 0.47 0.58 0.35 0.53 0.38 0.51 Ottawa 0.47 0.56 0.66 0.52 1 0.83 0.57 0.56 0.7 0.67 0.58 0.53 Toronto 0.42 0.57 0.63 0.47 0.86 1 0.45 0.53 0.6 0.76 0.6 0.53 Hamilton 0.48 0.17 0.29 0.6 0.55 0.38 1 0.52 0.44 0.64 0.44 0.37 Winnipeg 0.25 0.26 0.11 0.43 0.29 0.28 0.34 1 0.47 0.51 0.39 0.31 Regina 0.06 0.13 0.02 0.24 0.22 0.34 0.21 0.34 1 0.39 0.13 0.11 Calgary 0.38 0.52 0.34 0.31 0.53 0.57 0.63 0.74 0.66 1 0.32 0.49 Edmonton 0.69 0.61 0.47 0.62 0.61 0.79 0.6 0.91 0.57 0.63 1 0.71 Vancouver 0.36 0.44 0.32 0.49 0.63 0.46 0.37 0.61 0.39 0.71 0.54 1 FIGURE 3 0-0.2 0.2-0.4 0.4-0.6 0.6-0.8 0.8-1.0 FIGURE 4: Equilibrium Real Wage Ratios for 3 Building Trades 1901 and 1926 for 11 Cities Relative to Toronto 1.4 1.2 1 0.8 1901 1926 0.6 0.4 0.2 Va n co ed uve r m on to ca n lg ar y re gi W na in ni p ha eg m ilt To on ro nt ot o ta M wa on tre Q al ue be H c al if St ax Jo hn 0 FIGURE 5: Catholic Proportion of Total Populations of 12 Canadian Cities, 1911 and 1931 1.2 1 0.8 1911 0.6 1931 0.4 0.2 Halifax StJohn Quebec Montreal Ottawa Toronto Hamilton Winnipeg Regina Calgary Edmonton Vancouver 0 1 2 3 4 5 6 7 8 9 10 11 12 2