The Impact of United States Immigration Quotas on

The Impact of United States Immigration Quotas on
Migration to Canada During the 1920s
Byron Lew* and Bruce Cater
10th May 2002
*Department of Economics
Trent University
1600 West Bank Drive
Peterborough, Ontario K9J 7B8
Canada
email: [email protected]
The authors gratefully acknowledge the help of Herb Emery, Alan Green and Marvin McInnis,
all of whom read a draft and provided comments. Also the helpful comments of participants of
the fall 2000 meeting of the Canadian Network in Economic History, the Queen’s Economic
History workshop, and two anonymous referees, are much appreciated.
1
The Impact of United States Immigration Quotas on Migration to Canada During
the 1920s
Abstract
European emigration shifted dramatically toward Canada after the imposition of U.S.
quotas in the 1920s. Both countries formed policy to address the potential influx of unskilled
European labor. The context for Canadian policy formation differed in that there was a
potential substitution effect as the U.S. imposed quotas. An augmented gravity model of
immigrant settlement choice across North America using census stock data broken down by
period of arrival is used to establish the degree to which U.S. quotas shifted the pattern of
European emigration to North America.
Introduction
The U.S. immigration quotas of the 1920s have tended to be viewed by their impact on the
domestic U.S. economy.1 They are also indicative of the retreat from globalization characteristic
of the period from the late-nineteenth century through the recovery after World War II. 2 What is
not discussed, however, is impact the U.S. immigration quotas might have had on the supply of
immigrants to other new world countries. Canada, particularly in the early twentieth century, was
likely viewed by European immigrants as a close substitute for the U.S. If Canada was a
substitute, even if imperfect, then the imposition of U.S. quotas should have increased demand for
Canada as a destination. Though the era turned out to be shortened due to the Depression, the
1920s was a period of large immigrant inflows to Canada. On a per capita basis, immigrant inflow
rates were second only to those of the Wheat Boom (1896-1914). As well, per capita immigration
rates to Canada in the late 1920s exceeded the peak U.S. per capita inflow rates of the prewar era. 3
1
Goldin, “The Political Economy,” points out that the political debate over immigration restrictions was not a
recent phenomenon and that the imposition of the quotas occurred when the political forces opposed to immigration,
particularly labor, finally gained sufficient support. There were earlier attempts at restrictions through literacy tests
and other means in both Canada and the U.S. Green, “A Comparison.”
2
Timmer and Williamson, “Immigration Policy,” look at the impact of tightening of restrictions from the late 19th
century until the Depression.
3
At their peaks before 1914, U.S. per capita inflow rate of immigrants was on the order of 1.2% whereas Canada’s
rate peaked at just under 5%. While much lower during the 1920’s, the per capita inflow rate of immigrants to Canada
peaked at 1.6% in 1929.
2
The increased immigration to Canada could certainly have been due to a substitution of Canada
for the U.S. by European emigrants. However, the story is not only about U.S. quotas. Canadian
policy also played a supporting role. Canada aggressively relaxed its war era restrictions, but did
so selectively. In effect, Canada, like the U.S., imposed certain restrictions based on country of
origin. So, for example, while immigration rates to Canada from eastern European countries
increase dramatically over this period, immigration from the Mediterranean countries barely
changes in spite of dramatic cuts in the numbers admitted to the U.S. Canadian policy
undoubtedly followed the U.S. lead. Canadian policy makers, aware that a change in access to the
U.S. could imply a deflection of immigrants north, instituted a policy that in terms of its ranking
of European sending countries mirrored the U.S. policy. That it was not identical is due to the
differences in economic circumstances. Canada had recently settled its western provinces and had
experienced a sharp increase in economic growth. Immigration, by allowing for the settlement of
the West and increasing the size of the domestic Canadian market, had played an important role
and therefore policy towards immigration was relatively more liberal.
The focus of study of the impact of U.S. immigration restrictions on Canada has been its
impact on the emigration of the Canadian born. That the Canadian born left Canada for the U.S.
in large numbers in this period is well known.4 It is also been shown to have persisted for much of
the latter third of the 19th century.5 There has even been a debate in the past regarding the impact
of immigration into Canada on the emigration choice of the Canadian-born, with one subtopic
being the increased attractiveness of the U.S. to the Canadian-born after the imposition of the
U.S. restrictions.6 This study offers a test of whether or not Canada did in fact receive a boost in
European immigration after the U.S. ceased to be an open destination, a presumption of some of
the earlier debates.
In the next section, the details of policy changes of the 1920s in both countries will be
4
See Lavoie, L’emigration and Vedder and Gallaway, “Settlement Patterns.”
McInnis, “Immigration and Emigration.”
6
Percy, “American Immigration Legislation,” focusses directly on the impact of the U.S. quotas on Canadian
emigration, though he looks only at Quebec so his conclusions may not necessarily be robust for all of Canada. He
focusses on Quebec because the U.S. Census merely identifies where emigrants have come from but not where they
were born. To assess only the Canadian born, he therefore looks at those classified by the U.S. Census as French
Canadian. Because immigration to Canada from France was very small, he can be sure that virtually all those
appearing in the U.S. as French Canadian were born in Canada. However, French Canadian emigration patterns may
have looked quite different from that of English Canadians not only in terms of the match between choice of
destination and characteristics of the emigrant, but also because immigration into Quebec, unlike elsewhere in the
country, was almost entirely an urban phenomenon – mostly to Montreal.
5
3
examined. Patterns of immigration to the two countries and changes taking place over the 1920s
will be illustrated. The policy of both countries will be discussed in terms of their different
approach to the same problem: a potentially large influx of unskilled labor from Europe and its
consequent depressing effect on wages. Some mention will then be made of the data available and
the restrictions on its utilization. Estimates of the diversionary impact of U.S. quotas will be
presented. It will be shown that indeed U.S. quotas and their discriminatory effect altered the
supply of immigrants to Canada as hypothesized.
Immigration and Immigration Policy of the 1920s
U.S. quotas on Europeans immigrants were introduced as preliminary policy in two rounds in the
early 1920s, before being implemented in their final form again in the late 1920s (see Table 1).
All versions set annual immigration rates for each sending country in Europe based on their share
of immigrants living in the U.S. The Quota, or Per Centum Act of 1921 set annual quotas on
immigration from each country as 3% of the number of foreign-born of that country recorded in
the census of 1910. The Immigration Act of 1924 merely tweaked the quota. While the total was
reduced to 2%, the more important change was a revision of the base census year from 1910 to
1890.7 This revision in the census year was clearly discriminatory toward eastern and southern
Europe. While the 1924 quotas on Britain and Germany fell to about 75% of the 1921 quotas, and
quotas on Scandinavian countries dropped to about 50% of their 1921 level, quotas for Italians
and Russians as well as other countries of recent emigration fell to less than 10% of their 1921
quota (Table 1). As the 1921 quotas were simply based on the counts by origin in the recent
census, the distribution of quotas by country was essentially politically exogenous. The 1924
revision clearly endogenized the ethnic mix and shifted the source of immigrants into countries
more likely to contribute skilled labor.
The impact of the quotas can be seen by comparing immigrant flows into the U.S. before and
after the war (Table 1)8 . For example, while immigration to the U.S. from northern countries like
7
The 1924 Immigration Act did include several nonquota categories such as religious personnel, university
professors, and wives (sic.) and unmarried children under 18 of citizens born in quota countries. Hutchinson,
Legislative History, p. 471.
8
The immigration figures are by country of last permanent residence. As Czechoslovakia, Finland, Poland and
Yugoslavia were not separately recorded for the prewar years, but included under Austria, Germany, and Russia, the
decline in immigration from Austria and Russia appears larger as it includes the re-assignment component as well.
These are also gross flows so they do not account for return migration which was likely large for Italy. Gould,
4
Norway and Sweden over the period 1900-1914 is in the range of 10,000-20,000 per year, these
rates are cut by half from 1924 on. These declines are modest, however, compared with the
impact of the quotas on immigration from eastern and southern European countries. For example,
Italian immigration rates into the U.S. which had averaged around 200,000 per year before the
war, fall to 4,000 per year after 1924.9 Relative to the decline in eastern and southern European
immigrants, the decline in immigrant inflows from Britain, Germany and the Scandinavian
countries are relatively modest in comparison.
Canada followed the U.S. lead by adopting formal restrictions in 1923. 10 Canadian policy was
similar to U.S. restrictions, ranking immigrants by country of origin. They differed, however, in
that no explicit numerical limit was imposed. Like the U.S. system, European countries were
ranked – though not based on representation in the population – and conditions of admission were
made more restrictive the lower the ranking. British immigrants were to be admitted without
restriction beyond establishing their physical and mental health to insure they would not end up as
public charges. The remainder of the northern European countries were divided into two groups.
Immigrants from the preferred (as they were officially termed) countries of northwestern Europe
— the Scandinavian countries, Holland, Belgium, France, Switzerland, also Finland, and after
1926, Germany — were admitted subject to the same health and vagrancy restrictions as applied
to British immigrants. The difference between policy toward British immigrants and those
classified as preferred was the use of financial incentives to attract British agricultural labor.
Otherwise, the two groups were treated similarly. Immigrants from the rest of Europe —the
non-preferred countries — could be admitted only if destined for farm work in western Canada,
or if granted a permit by the Minister in response to a specific application by an employer. 11 The
conditionality on immigrants from the non-preferred countries was the means by which Canada
was to control the immigration of unskilled labor into Canada. As such, its implementation can be
understood in light of the great shift in demand for Canada induced by the U.S. quotas. Given the
similarity in the ranking assigned by both governments to European countries, it would appear
that Canadian and U.S. policy were motivated by similar factors. This is particularly evident after
the 1924 U.S. Immigration Act which simply set much more restrictive quotas on the countries
“European Inter-Continental Emigration.”
9
Averages for 1925-1930 may exceed the quota due to the admission of non-quota’s immigrants from these
countries, usually dependents of U.S. citizens.
10
Green, “A Comparison.”
11
Single females could be admitted as domestic servants.
5
Canada had already classified as non-preferred.12
However, even if the imposition of the tighter U.S. quotas in 1924 had a diversionary impact,
the Canadian immigration authority opened the border still wider. From 1925-1930, under what is
termed the railway agreement, the Canadian government authorized the two Canadian railway
companies to act as its agents for the admission of immigrants.13 The railway companies were
given authority to screen immigrants subject to the restrictions in place. Essentially, they were
allowed to recruit agricultural labor in Eastern Europe for the Canadian West. The result, evident
in Figure 1, was a large increase in immigration from the countries in which agents of the
railways were active, primarily central and eastern Europe.14
Both the shift in source countries of immigration to Canada and the timing of that switch are
coincident with the quotas of 1924, and with the implementation of the railway agreement. These
effects can be seen in the substantial shift in the distribution of immigrants between Canada and
the U.S. from the first to the second half of the 1920s shown in Table 2. 15 The abrupt changes in
source and destination would be hard to discern by only comparing the distribution of total
immigrants at census intervals of ten years. Therefore, a breakdown by year-of-arrival is used.
Canada’s share of immigration from every major European sending country increases in the latter
half of the 1920s, but the greatest increases are from the countries of central and eastern Europe.
For example, the greatest relative increase is for Hungarian immigrants, for whom Canada’s share
jumps from under 10% in the first half of the 1920s to over 70% in the second half, but the
increase is of similar proportion for most of the central and eastern European countries facing the
tightest quotas (relative to prewar immigration rates). Canada’s share of northwestern European
immigrants increases more modestly. Overall, the increase is about a doubling in Canada’s share
12
The last column of Table 1 shows the change in each country’s U.S. immigration quota as a percentage of the
total allowed immigration. All countries whose share of the quota does not decrease with the changes in 1924 are
those classified by Canada as preferred, with the exception of Bulgaria whose share stays the same because Bulgaria
receives the minimum quota of 100 in 1924. All other (non-preferred by Canada) countries’ percentage quota share
allotment fall, with Italy and Russia showing the greatest decline.
13
Avery, Reluctant Host, argues that the railway agreement was introduced at the behest of capital to keep wages
low. However, Green suggests that by providing a flow of labor to the prairies, farmers certainly would have benefited
through higher land rents and lower wages. Green, “International Migration,” and “In Search of Labour.”
14
Countries covered by the railway agreement were Austria, Czechoslovakia, Danzig, Estonia, Germany (withdrew
in 1928), Hungary, Latvia, Lithuania, Poland, Romania, Russia and Yugoslavia; although the railways did not have
offices in all these countries. Canada “Minutes.” Italy, as the origin of the second largest population of immigrants in
North America at the time, is conspicuously absent from this list.
15
The total flows for both halves of the 1920s are also reported, illustrating the quantitative impact of the quotas.
The data, drawn from each country’s census, are to be discussed below in the discussion of the statistical analysis.
The Census records immigration by year of arrival.
6
for these countries.
It should be noted, however, that as Canada had been a major destination for British immigrants
over the first three decades of the twentieth century – its share averaging about 50% –
immigration from the British Isles, particularly from England and Scotland, should be viewed
differently.16 And while there was an increase in Canada’s share of British immigrants from the
first to the second half of the 1920’s, this is only because of a relative decline in British
immigrants destined for Canada in the first half of the 1920’s. By the standards of the prewar
period, Canada’s share of British immigrants after the imposition of the U.S. quotas shows
essentially no change. The same is true of Irish immigration, except that Irish immigration to
Canada was considerably lower than the British. However, unlike the British, Irish immigration
flows increase to both countries in the second half of the 1920s.
It is the flow of Italian immigrants in particular that illustrates the restrictive nature of Canadian
immigration policy. Italian immigration to both countries declines absolutely in the latter half of
the 1920s. Canada had not attracted many Italian immigrants before the war – about 3% of the
North American flow. After the imposition of the U.S. quota, the proportion of Italian immigrants
to North America arriving in Canada increases, though still only to a modest 7% in spite of the
relative severity of the U.S. quota.17 So on a share basis Canada had become more attractive to
Italian immigrants, yet Italy is the only country from which immigrant flows to Canada decline
over the 1920s.18
It is clear that the U.S. quotas had a substantial impact on immigrant flows into the U.S. This is
well-known. There is also evidence that immigration flows to Canada increased as European
immigrants shut out of the U.S. substituted Canada. This strength of this substitution effect differs
by sending country depending on the relative impact of their quota. The U.S. quota reductions
were relatively greater on the countries of the new immigration of Eastern and Southern Europe.
This would have induced a greater relative increase of immigrants from these regions to Canada.
This appears to be the case for eastern European immigrants. However, Canada also managed its
immigrant admissions, and therefore the strength of the substitution effect also depends on
16
The term ’British’ will be used to refer to immigrants from Great Britain. The term’s meaning at the time might
have meant any citizen of the British Empire, including the Canadian-born. In the context used here, immigrants
from the rest of the British Empire are not included.
17
It is unlikely that the lack of substitution of Canada for the U.S. by Italian emigrants is due solely to the lack of
attractiveness of Canada. Italian immigration to Canada increases dramatically in the 1950s and early 1960s as Italy
becomes the second largest source of immigrants after the United Kingdom.
18
Immigrant flows from Iceland also decline, but the actual numbers are small.
7
Canada’s ranking of the sending countries. When Canada opens up its otherwise more restrictive
policy toward immigrants from Eastern Europe with the introduction of the Railway Agreement,
eastern European immigration rates increase dramatically. However, there is also evidence that
immigrants from Southern Europe did not find Canada a substitute for the U.S. This is a result of
Canadian policy which explicitly excluded them.
Agriculture and Canadian Immigration Policy
Evidence suggests that U.S. quotas along with the Canadian rankings shifted immigration to
Canada. Canadian immigration policy had an additional feature; immigrants were explicitly
diverted to the West as agricultural workers. If the general motivation behind restricting
immigration was the impact of immigrant labor on the wages of the domestic-born in low-skilled
industrial employment, then restrictive policy should be observed in all countries to which this
vulnerability applied. While Canada had an industrial workforce, it also had a much larger farm
sector. The policy of sending immigrants West was intended to respond to farmers needs while
simultaneously reducing downward pressure on wages in unskilled industrial employment
large-scale immigration might imply.19
The yearly immigration flow data show a substantial shift in destination of those admitted from
the non-preferred countries leading observers to conclude that while immigration patterns during
the Wheat Boom period were balanced between the agricultural West and the
manufacturing-based central provinces, immigration during the 1920s was more an agricultural
phenomenon.20 Yet there is also anecdotal evidence suggesting that the draw of the agricultural
West was not very strong. Critics of a liberal immigration policy in general and the railway
agreement in particular believed that immigrants were not settling in the West but were instead
settling in the industrial and resource centres in Ontario. The railways were criticized for not
discriminating on this criterion and admitting those not suited for agriculture and therefore less
likely to remain in the West.21 The crux of the problem is that agricultural labor demand on the
19
While this appears to be significant divergence in Canadian and U.S. policy, in fact U.S. policy was not so
dissimilar. In the U.S. Immigration Act of 1924, within each country’s quota, priority was to be given to agricultural
labor. Hutchinson, Legislative History, p. 496-97.
20
Green and Green, “Balanced Growth”; Green “International Migration”.
21
The railway agreement was renewed in 1927 and immigrants were required to have secured employment before
arrival. As well, a parliamentary committee review of immigration policy was ordered. See Canada “Minutes”.
8
prairies is highly seasonal. Those intending to establish homesteads would be more likely to
remain in the West. Those simply employed seasonally and then released were unlikely to remain.
It is important to understand the role of land policy of the prewar Wheat Boom on the politics
of the 1920s. The railways had been endowed with millions of acres of public land as a subsidy to
build transcontinental lines to enable western settlement. The railways, in order to encourage
settlement and therefore land sales, had offered financing to homesteaders; mortgage financing
was thin in this period as banks were not permitted to lend on real estate. During the 1920s, the
railways found themselves holding land from defaults on mortgages, and were therefore interested
in bringing settlers in to purchase these farms. So while the railways were pushing for open
immigration, the anti-immigration sentiments were assuaged with the agriculture-only restriction
on immigrants from the non-preferred countries. It is still an open question whether immigrants
did indeed tend to settle in the West in greater proportion than in the earlier Wheat Boom period.
In order to assess the extent to which immigrants were diverted to the West, evidence from the
Census is used. While the flow data on yearly immigrant admissions indicate a substantial shift to
the West, this is at best a short-run phenonmenon. The long-run response of immigrants to
Canada are shown in Table 3. Here the distribution of immigrants to Canada by region, origin,
and period of immigration are presented. For comparison, immigrants are grouped into the
following four categories: British Isles, northwestern Europe, central and eastern Europe, and
southern Europe. The Table shows the percentage of immigrants in each of the four categories
settling in each region, by period of arrival. All data are drawn from the 1931 Census of Canada
and represent location as of April 1931.22
Table 3 illustrates that there was no strong shift in permanent migration to the Prairies in the
1920s. For all four categories, the proportion living on the prairies declines in the postwar period,
with most of the shift being toward Ontario. Of those immigrating to Canada from central and
eastern Europe, 67% of those arriving during the first decade of the twentieth century settled on
the Prairies, while only 46% of those arriving during the latter half of the 1920s end up settling on
the Prairies. Except for southern Europeans, the increase in the proportion settling on the Prairies
from the first to the second half of the 1920s is modest despite the large increase in the yearly
flow immigrants westward.
Just as there was a shift by recent immigrant arrivals toward Ontario in Canada, so too in the
22
The data are merely categorized by the year in which immigrants arrived in Canada regardless of where they
initially settled. The Census only records their location at the time of the Census, not where they were living when
they arrived.
9
U.S. there was a shift by recent immigrant arrivals toward the Middle Atlantic states of New York,
New Jersey and Pennsylvania and away from all other regions. This was true of immigrant from
all source regions, though for British and northwestern European immigrants there was also a
somewhat smaller but positive shift into the East North Central states.23 Overall, though, there
was an obvious shift in the destination choice of recently arrived immigrants out of the traditional
agricultural regions.
The decline in the attractiveness of the agricultural West is also apparent from the increasing
urbanization of immigrants. The agriculture-only clause was presumably introduced in
recognition of this tendency, and in anticipation of it increasing with the imposition of the U.S.
quotas. Table 4 shows the percentage of immigrants to both the U.S. and Canada residing in
urban locations by year of immigration.24 By 1930, immigrants were more urbanized than the
native-born, and the more recently-arrived were the most urbanized, at least for most regions of
the U.S. (except the South Atlantic). The preference for agricultural workers in Canada may not
have been entirely ineffective, though. The data in Table 4 do indicate that, in contrast to the U.S.,
immigrant arrivals in Canada in the 1920’s were slightly less urban than earlier arrivals for all
region. Even immigrant arrivals of the 1920s settling in Ontario appear to have been slightly more
rural than earlier arrivals. This seems to suggest that though immigrants admitted with farm skills
and sent West ended up in Ontario, they did tend to end up in rural locations.
As policy was discriminatory on origin, it would be interesting to compare urbanization of
immigrants by country of origin by period of arrival. However, the data do not exist for Canada.
Instead a comparison of urbanization of immigrants by region of origin for two census years,
1920/21 and 1930/31, is presented in Table 5.25 Immigrants to the U.S. become more urban
across all regions from 1920/21 to 1930/31, with the exception of New England. 26 Even
23
Data not reported, but available from author.
Urban means any population center of 2,500 or more (with a few exceptions in New England) for the U.S.
census. In general, the Census of Canada classifies urban based on incorporation status, but for a more consistent
comparison with the U.S. census, only the totals for centers of 1,000 or more as reported in the Census of Canada are
used as the definition of urban in Canada. This is particularly important on the prairies where very small urban
centers proliferate. Further, it is a comparison of trends that is of interest so the difference in definitions between the
countries is not quite as significant.
25
The 1921 Census of Canada does not separate urban centers of less than 1000 people; therefore, the data for
1931 are for total urban rather than urban centers of 1000 people or more to facilitate comparison over time. Note
that the percentage urban for the Canadian native-born will higher in this table than in Table 4 as urban is here
defined more broadly.
26
The definition of urban was modified from the 1920 to the 1930 census, a change only for small centers in New
England, so that a decline in urbanization more likely reflects the change in definition. As a control, urbanization of
24
10
immigrants to the South Atlantic states become more urban excepting those from the British Isles.
Comparing immigration to Canada and the U.S., immigrants to Canada were less urban than
immigrants to the U.S., though, and increases in urbanization of immigrants over the decade 1920
to 1930 were greater in the U.S. Immigrants to Canada were becoming modestly more urban as
well, though not to the extent of the native-born. However there are exceptions, particularly in
Ontario and on the prairies. Immigrants to Ontario from northwestern Europe appear to have
become substantially less urban; whereas all other immigrants to the prairies except those from
northwestern Europe become less urbanized. The decline in urbanization among immigrants to
the prairies from central and eastern Europe are relatively modest, though, and this is the group
admitted in large numbers if destined for agriculture. Furthermore, as the share of central and
eastern European immigrants settling on the prairies declines in the 1920s, the decline in
urbanization among this group may reflect the increase in the share of prairie population
accounted for by those who arrived much earlier, during the Wheat Boom period. As there is a
larger decline in urbanization for British immigrants to the prairies suggesting that there were
agricultural opportunities on the prairies, there is at best only very weak support for the
effectiveness of Canadian policy of the 1920s in directing the non-preferred immigrants to the
West, at least in terms of their settling in the West. Moreover, the decline in urbanization among
northwestern European immigrants in Ontario suggests that many of those coming to Canada for
agriculture had greater opportunities in Ontario. Of those from central and eastern Europe, most
who left the prairies were likely remployed outside of agriculture.
In summary, while the U.S. imposed quotas in 1921, and tightened them in 1924, Canada
adopted a system of preference by ethnic origin in 1923. Canada then explicitly opened its
borders, relative to the U.S. in not adopting quotas, and absolutely through implementation of the
railway agreement in 1925. While imposition of the U.S. quotas appears to have diverted
immigrants to Canada, the Canadian immigration authority wanted to discriminate among that
flow for those with human capital appropriate to agriculture as a way of checking the potential
flood but simultaneously meeting the demand for labor. The railway agreement was intended to
implement these goals. The data suggest, however, that while the railway agreement may have
diverted the flow of immigrants to western Canada, it was only a temporary diversion. 27
Compared to the immigrant flows of the Wheat Boom period of 1896-1914, a greater proportion
the native-born is also reported and it too shows a decline for New England only.
27
Green has argued that from the perspective of farmers in the Prairies, it was the maintenance of the yearly flow
that was important. See Green, “International Migration” and “In Search of Labour.”
11
of the immigrant arrivals of the 1920s ended up in Ontario (Table 3). While this conclusion does
appear to contradict the observation that immigration to Canada in the 1920s was an agricultural
phenomenon, the shift into Ontario as the preferred destination of immigrants was also more rural
than urban. In other words, while it may be correct to characterize immigration to Canada in the
1920s as agricultural, it was not in the context of prairie expansion.
While the increased attractiveness of Canada as a destination for European immigrants has
been illustrated, it is possible that Canada offered better opportunities. A more formal analysis of
the distribution of immigrants across North America will be undertaken to control for the relative
attractiveness of each destination. A test of whether the U.S. quotas diverted immigrants into
Canada can then be performed.
Gravity Model and Settlement Patterns
The impact of the U.S. quotas of the 1920s will be explored using evidence on the behavior of
immigrants by period of arrival. The goal is to determine if Canada did draw more immigrants
than would be expected given the economic draw of its regions. Both the U.S. and Canadian
censuses report number of immigrants by destination, by birthplace and by period of arrival by
five-year intervals for the 1920s, and then by decade back to 1900. These data will be used to
infer differences in settlement choices among immigrant groups who arrived under different rules.
Immigrants arriving in earlier periods will have made their choice of destination under different
circumstances, so direct comparison between recent and earlier arrivals will include both the
effects of the rule changes under which the immigrant was admitted —the effect of interest —
and of any other differences over time, including ability to adapt, age of the immigrant, changes
in ’push’ factors in Europe, etc. Therefore comparisons among groups differentially affected by
immigration restrictions will be made to partly control for this problem. Analysis will be done by
individual country of birth as recorded in the Census.
As with previous studies of immigrant destination choice, an immigrant’s preference for
destination will be analyzed based on the characteristics of the destination. 28 Here, however,
immigrant destination choice is from among all U.S. states and Canadian provinces. Since the
28
Dunlevy, “Regional Preferences,” as the most recent example, contains a then current review of this literature.
There have been few studies since that make use of aggregated data because the more recent studies like Green and
Green, “Balanced Growth,” utilize microdata and therefore include individual characteristics of the immigrant as
well.
12
numbers of immigrants from different countries varied substantially, to control for possible
heteroskedsticity the dependent variable will be expressed as the proportion of immigrants to
North America per each major sending country settling in a destination. More formally, the
proportion of foreign-born from country settling in state/province arriving in period is
where is the stock of immigrants arriving
from country , settling in state/province ,
29 30
arriving in period as reported in the Census.
Settlement in any state or province is then explained as
where
is a vector of variables, discussed below, each representing the attractiveness of
state/province .The variables making up vector
capturing the attractiveness of a state/province
are
employment.
!"
personal income per capita.31
$#%"
improved acreage potential, taken as either the difference between the maximum improved
acreage in the future less the improved acreage at that census, or zero if the improved
acreage declines thereafter.
&(' ! percentage of farms operated by owner, part or full.
#)!*
percentage of employment in manufacturing.
annual mean temperature of a state/province’s principal city in Fahrenheit.32
29
United States, Fifteenth Census; Canada, 1931 Census.
The sending countries (birthplace) are: Austria, Belgium, Bulgaria, Czechoslovakia, Denmark, England,
Finland, France, Germany, Greece, Holland, Hungary, Iceland (1930/31), Ireland, Italy, Lithuania (1930/31), Norway,
Poland, Romania, Russia, Scotland, Spain (1930/31), Sweden, Switzerland, Wales, Yugoslavia.
31
Income is for 1930. Data for the United States for 1930 from United States, Personal Income; for Canada for
1930 from Leacy, Historical Statistics.
32
Canada, Canadian Climate Normals; U.S. Local Climatological Data. Also used for other climate variables.
30
13
heating degree days of a state/province’s principal city
The employment variable is intended to capture both opportunities available in a destination,
as well as to account for sheer size differences. It is hypothesized to correlate positively with
immigration. The elasticity is hypothesized to be greater for those with greater skills, and so
would be greater for those coming from northwestern Europe. Immigrants from countries where
average skills are lower may be more inclined to go into agriculture so the elasticity should be
lower. It has been used in other studies and found to perform well. 33 It will be very closely
correlated with population but because it is a ranking of labor market size, it will probably be
better associated with an immigrant’s choice of destination.
!"
The per capita income variable
is the direct ranking of all destinations’ economic draw. 34 It
is hypothesized to be positively correlated with immigration. There is a debate, however,
regarding the extent to which migrants moved from regions of low to high per capita incomes,
with evidence suggesting that real wages may be a better predictor. 35 A complete cross section of
representative wages by state and province comparable across countries is not available. As well,
since the income measure used is not GNP but personal income, wage differences will have a
greater weight.
To capture the attractiveness of a region for farming, the potential for future agricultural
# "
expansion is measured with
. Potential future acreage is used rather than current acreage to
control for the simultaneity between immigration and improved acreage if in fact immigrants
were attracted to the farm. The elasticity of immigration on potential future acreage is
hypothesized to vary substantially by country. The elasticity will be positive and larger for
immigrants from countries motivated by the attraction of agriculture, particularly the
Scandinavian countries. For others, it is unclear as there is no unequivocal ranking of immigrants
by country with regard to their choice of occupation. It would probably not be significantly
correlated to immigration for immigrants from the British Isles.
Since many immigrants may have been attracted to farming to become farmers and not merely
!
hired hands, the variable &('
is also used. One large difference between northern and southern
agriculture in this period is the large difference in the percentage of farms operated by an owner. 36
33
Green and Green, “Balanced Growth.”
Dunlevy, “Regional Preferences”; and Green and Green, “Balanced Growth,” discuss their need to construct a
proxy measure, gross value added, due to the lack of per capita income variable for their period of interest.
35
See for example Margo, “Regional Wage Gaps.”
36
Dunlevy, “Regional Preferences,” used other variables to explain a lack of immigration to the U.S. South in the
34
14
As with the variable measuring response to improved acreage, this variable should also be more
positively correlated for those from countries with immigrants attracted to agriculture.
Recognizing that immigrants in the early twentieth century may have been attracted by
#%! *
industrial expansion, , the percentage of the labor force employed in manufacturing is used.
Substantial variation is also likely across countries for this variable as well. It is hypothesized to
be positively correlated with migration for immigration from countries not displaying a strong
correlation with the agricultural variables.
Two variables are used to capture the general impact of climate. The average temperature has
been used in other studies and captures the basic climatic differences between regions. 37 But
because the mean disguises the difference in the variability of the climate which is likely greater
in the North, heating degree days is also included. It is measured as the sum over all days in a
year of the difference in the day’s average temperature from 65řF if below 65řF.38 Heating degree
days will be greater to the extent that climate is more variable.
It is hypothesized that immigrants will want to choose a destination more like that from which
they came, either because of comfort and familiarity, but also because agricultural human capital
will differ by agricultural region. Therefore, it is hypothesized that the elasticity on temperature
will be more negative the farther north the sending country. It should be noted, though, that
several countries like France, Germany and Russia may span several climatic zones so the
correlation for immigrants from these countries may be weaker than for those from countries with
less variation in climate. Assuming that immigrants prefer less variability in climate, both for
comfort and for agriculture, the elasticities on heating degree days should be negative.
One variable not included is distance. It is commonly used in studies of this kind. While some
have found it to be significant, others have not. Given the decline in transportation costs of the
late-nineteenth century, it is unlikely that distance from port posed much of a problem. 39
late-nineteenth century, however, some, like percentage black, is not recorded for Canada, though it would have been
barely positive.
37
Dunlevy, “Regional Preferences.”
38
If the day’s average temperature is above 65řF, it is added into cooling degree days. Both heating and cooling
degree days are used as rough indicators of energy cost differences. While cooling degree days may also reflect
comfort, it won’t reflect additional cost in this period. Cooling degree days was tested in preliminary work, but did
not add statistically to the explanation.
39
Even in studies of settlement patterns of European immigrants in the U.S. in the mid-nineteenth century, distance
does not perform particularly well. Dunlevy, “Regional Preferences” and Green and Green, “Balanced Growth,” also
find distance insignificant in explaining European immigration settlement patterns in the United States in the late
nineteenth and Canada in the early twentieth centuries respectively. Vedder and Gallaway, “Settlement Patterns,”
looking at Canadian migration to the U.S., however, do find distance is a significant explanatory variable. This might
15
Empirically, it proved to be insignificant. This is consistent with the view that immigrants choose
their destination within the continent based on the match between their personal attributes, human
capital, and the attributes of the region, rather than in a two-stage migration pattern. 40
All data come from the censuses of the two countries, except for personal income and climate
variables as noted above. There are several other potentially interesting variables that are reported
in the census of one country only. For example, a variable on ethnicity would probably be better
than country of birth as those reporting Poland or Russia may come from very different parts and
be influenced by very different factors. Ukrainians emigrating from Galicia will likely not be
influenced by the destination choice of Jews from northwestern Russia. While the Census of
Canada has immigrants as well as the native-born characterized by detailed ethnicity, the U.S.
Census records only by very broad racial categories. The U.S. Census records the native-born by
country of birth of both parents, but the Canadian Census does not. The Canadian Census
classifies the native-born by language first spoken whereas the U.S. Census is only interested in
whether or not the native-born speak English. Using the language of the foreign-born as an
explanatory variable is inappropriate because for most countries, the two are almost identical. The
number of those speaking Dutch by state/province is almost identical to the number of
immigrants born in Holland. As well, certain problems with using language arise because not all
languages are associated with a unique country. Those speaking Yiddish, for example, could be
from Russia, Poland, Germany or other central European countries. Religion might also prove
useful, and while the Canadian Census records details on religion, the U.S. Census does not.
All variables, dependent and independent, are logarithmically transformed. In a few
sub-periods, for a few countries, there are no immigrant arrivals for Prince Edward Island, so
! )
is approximated as .41
The major difference between the two countries’ censuses is the date at which each census is
taken. The U.S. Census was taken in April, 1930 while the Canadian Census was taken in June,
1931. Because the Canadian Census classifies immigrants by period of arrival, the arrivals for the
first five months of 1931 are summarized separately and can therefore be subtracted, so the for Canada will be for arrivals through December 1930 while for the U.S. it will be through
March 1930. While this will not fully control for the difference in dates at which the observations
suggest that the migration choice of those born on the continent differ from those arriving from overseas.
40
Green and Green, “Balanced Growth.”
41
The dependent variable is expressed as a proportion, so a number sufficiently small is needed and was chosen to
be smaller than the next smallest proportion. The results are not particularly sensitive to this choice.
16
are taken, the dependent variable is a stock variable and so cannot change too rapidly. Further,
immigration to Canada was virtually halted when restrictions were enacted in 1930, so additions
to totals were small.42 The only major potential bias is the ability of those in regions hardest hit
by the Depression to relocate over that year between the taking of the two censuses. Since in
Canada the prairies were hardest hit in terms of decline in income, 43 it can be assumed that if the
Census had been taken in 1930, there would have been more people, including immigrants, living
on the prairies and fewer in British Columbia and Ontario.
Since the test is of changes in immigration in the latter half of the 1920’s, the dependent
variable is further dis-aggregated by period of arrival. Both censuses report arrivals by five year
periods for the previous two decades, then for the decade 1901-1910, and then for all arrivals
before 1901.44 45 This yields in effect six dependent variables describing arrivals under different
sets of conditions. Each is expressed as a proportion of all foreign-born, per country, living in a
state/province. Because the choice of destination is also strongly correlated where immigrants of
a like ethnic class have gone, a dependent variable of the previous period’s immigration is used.
But because it is likely that destinations with more recent arrivals will be more influential over a
current immigrant’s choice, only the previous period’s immigration will be used. In other words,
the proportion of German immigrants arriving in 1920-24 that ended up in Pennsylvania will be
used as an independent variable to explain the proportion of German immigrants arriving in
1925-30 that ended up in Pennsylvania. Since arrivals before 1901 are not further categorized,
patterns of arrivals before 1901 will not be used as a dependent variable. 46
The 1930 U.S. Census, however, does not report year of arrival data for immigrants from all
countries of birth. It appears that if for any sub-period there were no arrivals in state from
42
Only immediate family of those already in Canada, and anyone with sufficient capital, were admitted. The mix
of immigrants to Canada after 1930 shifts dramatically toward women and children.
43
Safarian, The Canadian Economy.
44
Again, because of the difference in census dates, the sub-periods do not match exactly. The Census of Canada
reports arrivals for the periods 1926-30, 1921-25, ..., while the U.S. Census reports for 1925-30, 1920-24, 1915-19,
1911-14, 1901-10, <1901.
45
By year of arrival, the U.S. Census reports foreign-born white population rather than total foreign-born. The
number of non-white European immigrants is very small, much less than 1% of the total foreign-born for virtually all
countries, and about 1% for England and Austria.
46
Regardless of the year of arrival, immigration is recorded by place of residence at the time of the Census.
Therefore, even if all immigrants from a particular country initially settled in New York, but then relocated to Ohio
sometime before the 1930 Census, they will appear as immigrants to Ohio but will be classified by year of arrival in
the U.S. Therefore, the variable explaining the chain migration effect is based only on the numbers actually living in
the location where the new immigrant has also settled.
17
country , then immigrant arrivals to that state from country are simply not reported in the
breakdown by period of arrival. Only Germany seems to have had at least one immigrant arriving
in every period in every state. For consistency of comparison, the same subset of states is used
across all estimates allowing for ten countries to be included
with thirty-five observations each —
nine Canadian provinces plus twenty-six U.S. states.47 48 The states not included are mostly
southern and the smaller western states.
Results
The hypothesis tested is whether response by immigrants to Canada changed after the U.S.
imposed quotas. Results of regressions of the proportion of immigrants settling in each province
will be reported as a test for such an effect. A separate regression will be run on the dependent
" #)!
, set equal to one for a Canadian
variable for each period of arrival. A dummy variable,
province and zero for a U.S. state is included in each regression. The null hypothesis is that this
dummy variable will not change across periods. This would be the case if, in response to the U.S.
quotas, immigration to both Canada and the U.S. declines. If, on the other hand, immigration to
Canada does not decline in proportion to the decline in immigration to the U.S., the hypothesis of
Canada as a substitute destination would be supported. As each regression equation explaining
settlement in period contains as an independent explanatory variable settlement in period ,
the coefficient on the dummy variable indicates the strength of the change in the proportion of
immigrants from any country settling in Canada. A positive and significant coefficient indicates
greater settlement than predicted given variables explaining the draw of a state/province and given
the proportion of immigrants from country settling in state/province in the period immediately
previous.
One of the independent variables, employment, is endogenously determined. As immigration
increases to a state/province, so does employment. Therefore it is appropriate to use an
instrumental variables approach with employment of the native-born ( 47
! ) as the instrument. 49
The set was chosen to maximize (roughly) both the number of countries and the number of observations subject
to the constraints of including countries from the three categories, British, Preferred, non-Preferred, as well as
including countries covered by the railway agreement.,
48
The individual states included are: Maine, Massachusetts, Rhode Island, Connecticut, New York, New Jersey,
Pennsylvania, Ohio, Indiana, Illinois, Michigan, Wisconsin, Minnesota, Iowa, Missouri, Nebraska, Kansas,
Maryland, Florida, Texas, Montana, Wyoming, Colorado, Washington, Oregon, California.
49
Dunlevy, “Regional Preferences.”
18
As the estimation involves multiple equations, use of the potential correlation among error terms
across equations to improve the efficiency of the estimates is a possibility. However, the complete
system of 50 equations resulted in a singular matrix. This is likely caused by the independent
variables being common across all equations. It is possible to consider a system of a subset of
equations, but it is unclear whether it is appropriate to use a system per period of arrival, or by
country. In either case, parameters and standard errors do not change by much regardless of
which method is used. As a test of the hypothesis that the quota diverted immigrants to Canada,
the use of a single-equation model in its relative inefficiency, provides a stronger test of the
hypothesis. The ultimate acceptance or rejection of the null hypothesis is unchanged regardless of
estimation method. Therefore the results reported are for instrumental variables only without use
of cross-equation correlations.
The key results are the behavior of the dummy variables across periods. Because there are five
regressions per country and ten countries tested for a total of fifty regressions, as a test of the
hypothesis only the dummy variables themselves are reported. The hypothesis itself is supported
or rejected based on these dummy variables. These results are reported in Table 6. The complete
set of regression estimates are presented and discussed in the appendix.
A significant coefficient on the dummy variable for Canada indicates that settlement patterns
have changed significantly from the previous period. For example, the coefficient on the Canada
dummy variable for England for settlement in the period 1925-1930 is 0.85, significant at the 1%
level. Therefore, it appears that there was a diversion of English settlement to Canada. However,
the coefficients on the Canada dummy variables for three of the other four periods are also
significant. Therefore the change in settlement in the latter half of the 1920s is not unusual given
the shift toward Canada displayed by English immigrants over several periods. As well, though
insignificant, the coefficient on the Canada dummy variable is negative for the first half of the
1920s, suggesting that at least some of the increase in English immigration to Canada in the latter
half of the 1920s might reflect a catch up. Scottish immigration shows an almost identical pattern.
There is a shift in the latter half of the 1920s, but also a general tendency of shifting toward
Canada over the previous decades. These results can be interpreted as suggesting a modest shift
of English and Scottish immigrants to Canada, though the effect is slight and could also be part of
a general shift toward Canada by immigrants from these countries.
The hypothesis that the quotas diverted immigrants to Canada would be supported if the effect
was greatest for countries facing severe quota restrictions. Since the countries Greece, Italy,
19
Russia and Poland faced the most significant quotas among the countries used for the analysis, the
diversion effect on immigrants from these countries is predicted to be largest. While the
coefficients on the Canada dummies are large for Poland and Russia, they are somewhat smaller
for Italy and Greece. However, because of the discriminatory impact of the Canadian restrictions
as they were implemented through the railway agreement, it is expected that there should be
greater diversionary impact on Polish and Russian immigrants and a somewhat milder impact on
Italian and Greek immigrants.50 The increase in Canada’s share of Italian immigrants is due
mostly to the relatively large decline in Italian immigrants to the U.S. Italian immigration declines
to Canada as well indicating the effect of the Canadian restrictions.
For Italy, Greece and Russia, the diversionary effect in the latter half of the 1920s appears large
given that there does not appear to have been any large previous shifts toward Canada. Italy
shows a slight increase in immigration to Canada in the half-decade before the war, but then so do
almost all the countries. The coefficients on the Canada dummy for the first period, 1901-09;
however, are either insignificant, in the case of Italy, or even negative, in the case of Greece.
There is a shift for Russian immigrants apparent in the early 1920s as well, yet the effect is still
greatest for Russian immigrants in the latter half of the 1920s. The U.S. quotas appear to have had
the largest diversionary impact on Russian immigrants, followed closely by Polish.
As a counter example, Irish immigration to Canada appears to shift progressively toward the
U.S. over time. As Ireland’s quota is not onerous relative to the pre-quota flows, like the German
quota too, any shift toward Canada would have been expected to be mild. While Irish immigrant
arrivals of the first decade of the century display a strong shift to Canada, the effect diminishes
with each half-decade period into the future. Quantitatively there is an increase in the Canada
effect from the first to the second half of the 1920s for Irish immigrants, though the change is
insignificant statistically.
The two countries used to illustrate the behavior of the northwestern European countries with
more modest quota restrictions are Norway and Sweden. These are also examples of countries
labeled ’preferred’ by the Canadian immigration authority. So, for immigrants from these
countries, a modest decline in access to the U.S. was reinforced by a relatively open Canadian
50
There are several countries with even greater changes in immigration to Canada over this period, like Austria,
Czechoslovakia, Hungary, and Yugoslavia, but they cannot be used in the regression analysis because the U.S. Census
does not provide sufficient coverage of arrivals by year for these countries. The diversion hypothesis can really only
be tested using Poland and Russia and the countries with the greatest relative shift to Canada. The actual increase of
Russian immigrants heading to Canada from the first to the second half of the 1920s is small.
20
border. The diversionary effect is predicted to moderate for these two countries, as the data
support. Swedish emigrants in particular seems to display only the most modest shift toward
Canada in the latter half of the 1920s. For Sweden, the effect is statistically stronger in the first
half of the 1920s. Norway shows a slightly greater shift, but then Norway faced a modestly more
restrictive quota given the changes in flows to the U.S. (Table 1). Finally, Germany shows a
relatively large shift toward Canada. German immigration increases substantially to both the U.S.
and Canada in the second half of the 1920s, yet if appears that the U.S. quotas did cause some
diversion as well. German immigrants had shown a relatively large affinity for Canada in the first
decade of the century, but this declines until the second half of the 1920s. 51
In order to test the effect of the Canadian policy to settle the West, the regressions were rerun
replacing the single Canada dummy variable with four dummy variables representing the regions
of Canada: Maritimes, Central, Prairies, and Pacific. While the degrees of freedom begin to
decline with their inclusion, they do reinforce the tendency illustrated early for immigrants to
settle in Central Canada as much as or more than they do on the Prairies in the period when many
were supposed to have headed West. These results are reported in Table 7.
The general conclusions reported in Table 6 are still apparent. There is a definite shift toward
Canada in the latter half of the 1920s. But that shift involves all regions of Canada. Russia is the
only country for which the coefficient on the Prairie dummy variable is largest in the 1920s. For
all others, the Prairies are a less significant draw. For Norway, for example, the shift in
immigrants settling in Canada over this period is the weakest for the Prairies. As northerners, this
was one group the Canadian immigration authority had hoped would settle on the Prairies. The
evidence also bears out the observation drawn from the raw data that the Prairies displayed the
strongest draw in the first decade of the century only. For those from countries that did show a
propensity to settle in Canada in this period of the Wheat Boom, all tended to show the greatest
shift toward settling the Prairies. By the second half of the 1920s, while still drawing immigrants,
there is not increased tendency to settle in the Prairies over other regions in the country.
51
This decline includes the prewar half-decade, though perhaps growing anti-German sentiment in the British
world contributed to make Canada less attractive in the years leading up to the war.
21
Conclusions
In response to a potential increase in European emigration after World War I and its resultant
impact on unskilled wages, both the U.S. and Canada imposed restrictions. The motivations were
the same though the policy was implemented differently in each country. The solution for the
U.S. was rigid quotas based on discrimination by ethnic origin. Canada adopted the
discrimination by ethnicity, but did not impose quotas. Given the impact of the settlement of the
West, and immigration more generally, on economic growth in Canada during the Wheat Boom,
the immigration authority wanted to remain flexible and admit immigrants as labor market
conditions permitted. In addition, those from Eastern Europe were admitted if destined for the
West, and later only if they could pre-arrange employment. The motivation was too address the
chronic labor shortage in Western agriculture and relieve downward pressure on unskilled wages
in the industrial and resource regions that a large scale immigration might imply.
In addition, Canada faced the prospect of an increase in immigration simply due to the
substitution effect after the U.S. imposed quotas. To that end, tests for the attractiveness of
Canada as a destination for European immigrants after the imposition of U.S. quotas have been
performed. The evidence suggests that U.S. quota restrictions did divert immigrants to Canada,
and this diversion was relatively greater for immigrants from those countries most affected by the
U.S. quotas. This diversion was also partly influenced by Canadian policy.
It appears that there was substantial diversion of virtually all European immigrants to Canada
in the latter half of the 1920s after the imposition of the much tighter quotas of the U.S.
Immigration Act of 1924. But even though diversion did take place, the degree varied among
countries in proportion to the impact of the 1924 quotas. In general, immigrants from the British
Isles displayed a mild degree of diversion to Canada. Immigrants from the northwestern
European countries of Germany and Norway – though not Sweden – showed a slightly greater
tendency to be diverted to Canada. Immigrants from the countries of eastern Europe were most
readily diverted to Canada. And unlike the northwestern Europeans, immigrants from eastern
Europe arriving in earlier periods had shown no preference for Canada, making the shift in the
1920s particularly strong. Finally, immigrants from the southern European countries of Italy and
Greece also display diversion to Canada, mild for Italian immigrants and stronger for Greek
immigrants. It is suggested that Italian immigrants coming from a country not covered by the
railway agreement recruitments faced tighter restrictions, though because of the evidence of
diversion, restrictions into Canada were not as tight as U.S. quotas. It is not clear why Greek
22
immigrants displayed a greater degree of diversion to Canada, but they were a smaller group.
There is also evidence that the Canadian policy of diverting immigrants within Canada had
some impact, though that impact was probably very modest. For the period 1925-1930, it is only
immigrants from Russia who display an obvious preference for the Canadian prairies. For the
other groups for whom admission to Canada was predicated upon a destination in western Canada
— in this sample Greece, Italy, and Poland — the attraction of the prairies over central Canada is
not significantly greater, and even significantly smaller for Greece. It is probably more correct to
simply generalize that while Canada appeared to have become more attractive to European
immigrants in the late 1920s after imposition of U.S. quota restrictions, the degree to which
Canadian policy redirected these flows within Canada was slight.
Canada, then, remained more open to immigrants than did the U.S. and this was by design. The
Canadian authority imposed restrictions after the U.S. initiated its quotas of 1921, but these
restrictions were much looser. The Canadian authority appeared to be committed to maintaining a
supply of farm labor. It is possible that the Canadian policy of discrimination by region was partly
formed with this in mind. It was the northwestern European immigrants of the wheat boom era
that displayed the greatest tendency to settle on the prairies. Perhaps the ease with which they
were admitted partly reflects this. Lacking adequate supply, eastern Europeans were allowed in. It
is unclear why the Canadian authority was biased against Italian immigrants as their long-run
response to western settlement did not differ from those allowed in under the railway agreement.
Policy differed because the economies of the two countries differed. Canada was still relatively
agricultural. The wheat boom had meant substantial capital inflows awaiting repayment and the
key to repayment was export earnings, of which wheat was the most important. Further the
country had undergone rapid industrial expansion much more recently, with a major impetus
being the Wheat Boom in the decade before World War I. As Canada was small and, while
perhaps not actively extending, was at least not rapidly altering a policy of import substitution, 52
population growth was seen then as a key to expanding the limited domestic market. Policy
makers in Canada, unlike those in the United States, retained the option to allow in labor as
needed hoping to direct as much as possible to agriculture.53 This was true of British immigrants
52
An attempt at Reciprocity with the United States failed in 1911 when the Liberal government supporting freer
trade was defeated. The Liberals returned to power in the 1920’s, but rather than pursuing a similar bilateral trade
agreement chose instead to revise tariff schedules piecemeal to better suit politics. See Beaulieu and Emery, “Pork
Packers,” and Percy et al., “Reciprocity” on the political landscape and the failure of Reciprocity.
53
Again, this did appear in the U.S. Immigration Act of 1924 with a preference for agricultural labor within a
quota.
23
as well, though of course British immigrants were not barred if not destined for the prairies.
However, the relatively large flow of labor to the Canadian West was, for most, only temporary. It
appears that with only one exception, most immigrants regardless of origin, and therefore
condition of entry, were able to redistribute themselves across the country in roughly similar
fashion.
References
Avery, Donald. Reluctant Host: Canada’s Response to Immigrant Workers, 1896-1994.
Toronto: McClelland and Stewart, 1995.
Beaulieu, Eugene and J.C. Herbert Emery. “Pork Packers, Reciprocity, and Laurier’s Defeat
in the 1911 Canadian General Election.” This JOURNAL 61, no. 4 (2001): 1083-1101.
Canada. Atmospheric Environment Service. Canadian Climate Normals, 1961-1990.
Ottawa: Minister of Supply and Services, 1993.
Canada. Dominion Bureau of Statistics. Census of Canada 1931. Ottawa: J. O. Patenaude,
I.S.O., 1934.
Canada. Dominion Bureau of Statistics. Census of Canada 1921. Ottawa: 1924.
Canada. Parliament. Select Standing Committee on Agriculture and Colonization. Minutes
of Proceedings and Evidence and Report. Ottawa: F.A. Acland, 1928.
Canada. Report of the Department of Immigration and Colonization. Ottawa, various dates.
Dunlevy, James A. “Regional Preferences and Migrant Settlement: On the Avoidance of the
South by Nineteenth-Century Immigrants.” In Research in Economic History, 8, edited by
Paul Uselding, 217-51. Greenwich, CT: JAI Press Inc., 1982.
Easterlin, Richard A. “State Income Estimates.” In Population Redistribution and Economic
Growth United States, 1870-1950, edited by Everett S. Lee et al., 703-59. Philadelphia: The
American Philosophical Society, 1957.
24
Goldin, Claudia. “The Political Economy of Immigration Restrictions in the United States,
1890-1921.” In The Regulated Economy: A Historical Approach to Political Economy,
edited by Claudia Goldin and Gary D. Libecap, 223-57. Chicago and London: University of
Chicago Press, 1994.
Gould, J. D. “European Inter-Continental Emigration The Road Home: Return Migration
from the U.S.A.” Journal of European Economic History 9 (September 1980): 41-112.
Green, Alan G. “International Migration and the Evolution of Prairie Labor Markets in
Canada, 1900-1930.” In Migration and the International Labor Market, 1830-1939, edited
by Timothy J. Hatton and Jeffrey G. Williamson, 156-74. London: Routledge, 1994.
_____. A Comparison of Canadian and U.S. Immigration Policy in the Twentieth Century.
In Diminishing Returns: The Economics of Canada’s Recent Immigration Policy, edited by
Don J. DeVoretz, 31-64. Toronto: C.D. Howe Institute, 1995.
_____. In Search of Labour: The Political Economy of Canadian Immigration Policy,
1900-1930. Unpublished Manuscript, Queen’s University, 1996.
Green, Alan G. and David A Green. “Balanced Growth and the Geographical Distribution of
European Immigrant Arrivals to Canada, 1900-1912.” Explorations in Economic History 30
(January 1993): 31-59.
Green, Alan G. and Gordon Sparks. “Population Growth and the Dynamics of Canadian
Development: a Multivariate Time Series Approach.” Explorations in Economic History 36
(January 1999): 56-71.
Green, Alan G. and Malcolm C. Urquhart. “New Estimates of Output Growth in Canada:
Measurement and Interpretation.” In Perspectives on Canadian Economic History, edited by
Douglas McCalla, 182-99. Toronto: Copp Clark Pitman, 1997.
Hutchinson, E.P. Legislative History of American Immigration Policy 1798-1965.
Philadelphia: University of Pennsylvania Press, 1981.
Lavoie, Yolande. L’emigration des Canadiens aux Etats-Unis avant 1930: mesure de
phenomene. Montreal: Les Presses De L’Universite de Montreal, 1972.
25
Leacy, F. H., ed. Historical Statistics of Canada, Second Edition. Ottawa: Statistics Canada,
1983.
Margo, Robert. 1999. “Regional Wage Gaps and the Settlement of the Midwest.”
Explorations in Economic History 36 (April 1999): 128-43.
McInnis, R. Marvin. “Immigration and Emigration: Canada in the Late Nineteenth
Century.” In Migration and the International Labor Market, 1830-1939, edited by Timothy J.
Hatton and Jeffrey G. Williamson, 139-55. London: Routledge, 1994.
Percy, Michael B. “American Immigration Legislation in the 1920’s and its Impact on
Canadian Emigration to the United States.” manuscript, University of Alberta, 1982.
Percy, Michael B., Kenneth H. Norrie, and R.B. Johnston. “Reciprocity and Canadian
General Election of 1911.” Explorations in Economic History 19 (October 1982): 409-34.
Safarian, A.E. The Canadian Economy in the Great Depression. Toronto: McClelland and
Stewart, 1970.
Timmer, Ashley S. and Jeffrey G. Williamson. “Immigration Policy Prior to the 1930’s.”
Population and Development Review 24 (December 1998): 739-71.
U.S. Bureau of the Census. Fifteenth Census of the United States. Washington, DC: GPO,
1933.
U.S. Bureau of the Census. Fourteenth Census of the United States. Washington, DC: GPO, .
U.S. Department of Labor. Bureau of Immigration. Annual Report of the Commissioner
General of Immigration to the Secretary of Labor. Washington, DC: GPO, various dates.
U.S. Department of Commerce. Office of Business Economics. Personal Income by States
Since 1929. Washington, DC: GPO, 1956.
U.S. National Oceanic and Atmospheric Administration. Local Climatological Data Annual
Summary. Washington, D.C., various dates.
Vedder, R.K. and Gallaway, L.E. 1970. “Settlement Patterns of Canadian Emigrants to the
United States, 1850-1960.” Canadian Journal of Economics 3: 476-86.
26
Figure 1: Immigrants to Canada by Status of Country of Origin
80000
60000
Brit
Pref
40000
Rail
South
20000
0
1921
Source: Canada. Report
1923
1925
1927
27
1929
1931
Table 1: U.S. Immigration Quotas, 1921 and 1924
Quota
Act of
1921
Immigration
Act of
1924
Ratio
1924:1921
Immigration,
Average
per Year*
Change in
Proportion
Quota
Percent
Quota
Percent
1905−09
1910−14 1925−30
Austria
7,451
2.1%
785
0.5%
0.11
−1.6%
106,483
115,170
938
Belgium
1,563
0.4%
512
0.3%
0.33
−0.1%
4,930
5,690
628
Bulgaria
302
0.1%
100
0.1%
0.33
−0.0%
5,990
4,964
14,282
4.0%
3,073
1.9%
0.22
−2.1%
−
−
Denmark
5,694
1.6%
2,789
1.7%
0.49
0.1%
6,661
6,694
2,376
England
26,332
7.4%
18,182
11.0%
0.69
3.7%
50,135
43,753
13,576
Finland
3,921
1.1%
471
0.3%
0.12
−0.8%
−
−
France
5,729
1.6%
3,954
2.4%
0.69
0.8%
8,949
8,601
3,424
Czechoslovakia
Germany
(col 4 − col 2)
101
2,928
487
68,059
19.0%
51,227
31.1%
0.75
12.1%
34,759
32,239
44,914
Greece
3,294
0.9%
100
0.1%
0.03
−0.9%
20,437
26,442
183
Hungary
5,638
1.6%
473
0.3%
0.08
−1.3%
137,113
110,760
552
Ireland
40,568
11.3%
37,430
22.7%
0.92
11.4%
35,612
27,482
28,161
Italy
42,057
11.8%
3,845
2.3%
0.09
−9.4%
218,410
220,967
3,999
3,607
1.0%
1,648
1.0%
0.46
−0.0%
5,436
7,147
1,749
Norway
12,202
3.4%
6,453
3.9%
0.53
0.5%
18,993
11,416
5,480
Poland
25,827
7.2%
5,982
3.6%
0.23
−3.6%
−
Portugal
2,520
0.7%
503
0.3%
0.20
−0.4%
7,083
10,380
480
Romania
7,412
2.1%
603
0.4%
0.08
−1.7%
4,023
2,570
742
34,284
9.6%
2,248
1.4%
0.07
−8.2%
187,335
210,922
2,042
7,832
2.2%
4,845
2.9%
0.62
0.8%
15,698
15,678
15,352
912
0.3%
131
0.1%
0.14
−0.2%
3,364
5,726
190
Netherlands
Russia
Scotland
Spain
Sweden
−
5,909
20,042
5.6%
9,561
5.8%
0.48
0.2%
19,555
17,843
8,009
Switzerland
3,752
1.0%
2,081
1.3%
0.55
0.2%
3,568
3,762
1,881
Wales
2,475
0.7%
2,002
1.2%
0.81
0.5%
2,173
2,274
1,468
6,426
1.8%
671
0.4%
0.10
−1.4%
Yugoslavia
Total Europe
357,803
164,667
−
907,796
−
904,834
Note: Columns do not sum as not all countries are included.
*Data on immigration by country of last permanent residence is classified according to the borders extant at the time of
arrival. The countries Czechoslovakia, Finland, Poland, and Yugoslavia were not independent over this period so the
totals included in the data are for the countries to which they belonged at the time (Austria, Germany, and Russia).
Source: U.S. Annual Report.
28
644
247,194
Table 2: Canadian Share of North American Immigration, by Year of Immigration
Canadian Share of Immigration to North America
1930
Immigration by Year of Arrival
1920
all years 1925−30 1920−24 1915−19 1910−14 1901−10 <1901 all years
Canada
US
1925−30 1920−24
1925−30 1920−24
Austria
0.091
0.451
0.064
0.086
0.128
0.082 0.043
0.091
8010
2653
9741
38668
Belgium
0.209
0.588
0.198
0.317
0.233
0.161 0.073
0.175
4505
2465
3154
10013
Bulgaria
0.134
0.356
0.128
0.072
0.157
0.058 0.026
0.088
427
194
771
1327
Czech
0.044
0.483
0.051
0.036
0.014
0.011 0.004
0.012
16071
2623
17187
48325
Denmark
0.087
0.452
0.157
0.114
0.100
0.052 0.011
0.036
8787
2625
10638
14055
England
0.471
0.588
0.439
0.670
0.703
0.571 0.206
0.457
79031
81520
55424
104248
Finland
0.175
0.785
0.378
0.126
0.147
0.076 0.024
0.075
13292
6472
3638
10671
France
0.109
0.115
0.079
0.155
0.210
0.168 0.055
0.111
1652
1144
12652
13380
Germany
0.024
0.071
0.028
0.045
0.064
0.035 0.008
0.015
15471
3796
203563
129467
Greece
0.031
0.103
0.028
0.022
0.038
0.022 0.021
0.021
1186
792
10378
27757
Holland
0.074
0.326
0.126
0.080
0.099
0.052 0.004
0.042
4140
2073
8555
14421
Hungary
0.093
0.704
0.094
0.030
0.021
0.025 0.014
0.019
19682
3408
8288
32880
Iceland
0.673
0.366
0.350
0.495
0.763
0.812 0.670
−
97
135
168
251
Ireland
0.103
0.175
0.165
0.234
0.235
0.117 0.044
0.082
23173
14874
109622
75411
Italy
0.023
0.071
0.028
0.035
0.033
0.017 0.008
0.022
6232
7684
82140
265684
Lithuania
0.028
0.536
0.063
0.025
0.014
0.010 0.005
−
3259
546
2819
8060
Norway
0.085
0.288
0.135
0.155
0.208
0.103 0.009
0.060
9393
3821
23250
24555
Poland
0.118
0.637
0.129
0.107
0.094
0.075 0.047
0.054
64539
16423
36777
110554
Rumania
0.215
0.661
0.136
0.181
0.255
0.174 0.136
0.181
9872
3304
5059
21043
Russia
0.100
0.522
0.158
0.082
0.114
0.066 0.053
0.074
26028
23110
23829
122951
Scotland
0.439
0.432
0.380
0.672
0.665
0.532 0.211
0.471
48619
46673
63841
76125
Spain
0.010
0.022
0.004
0.007
0.017
0.012 0.011
−
100
62
4480
15587
Sweden
0.054
0.166
0.090
0.105
0.137
0.082 0.010
0.042
7012
4184
35104
42196
Switzerland
0.051
0.186
0.091
0.090
0.094
0.049 0.009
0.028
1936
1356
8463
13598
Wales
0.269
0.544
0.373
0.561
0.528
0.370 0.052
0.170
6551
2761
5488
4646
Yugoslavia
0.074
0.639
0.073
0.043
0.020
0.014 0.007
0.011
11896
2211
6725
28168
Source: Census of the United States, 1920, 1930; Census of Canada, 1921, 1931.
29
Table 3: Regional Distribution of Immigrants to Canada, by Year of Immigration and Birthplace,
1931
Origin Destination
Stock
1931
by Year of Arrival, 1931
1925−30
1920−24
1915−19
1910−14
Stock
1901−09
<1901
1921
British Isles
Maritimes
0.023
0.040
0.018
0.025
0.016
0.018
0.033
0.026
Quebec
0.091
0.119
0.106
0.083
0.087
0.076
0.091
0.083
Ontario
0.452
0.488
0.522
0.460
0.441
0.389
0.497
0.440
Prairies
0.274
0.254
0.222
0.264
0.287
0.325
0.215
0.301
B.C.
0.160
0.098
0.132
0.167
0.169
0.191
0.164
0.150
Maritimes
0.021
0.028
0.014
0.023
0.022
0.014
0.027
0.024
Quebec
0.094
0.097
0.092
0.119
0.095
0.077
0.118
0.105
Ontario
0.159
0.236
0.248
0.159
0.104
0.089
0.118
0.310
Prairies
0.555
0.467
0.415
0.556
0.628
0.660
0.567
0.434
B.C.
0.171
0.172
0.232
0.143
0.151
0.160
0.170
0.127
Maritimes
0.010
0.009
0.010
0.016
0.013
0.010
0.006
0.013
Quebec
0.117
0.138
0.131
0.176
0.109
0.096
0.063
0.095
Ontario
0.280
0.333
0.347
0.397
0.281
0.181
0.177
0.211
Prairies
0.540
0.461
0.455
0.362
0.558
0.669
0.666
0.623
B.C.
0.053
0.058
0.057
0.049
0.038
0.044
0.087
0.058
Maritimes
0.021
0.021
0.023
0.022
0.023
0.017
0.029
0.032
Quebec
0.228
0.184
0.180
0.237
0.240
0.257
0.298
0.224
Ontario
0.525
0.593
0.581
0.544
0.519
0.467
0.418
0.493
Prairies
0.092
0.072
0.084
0.090
0.102
0.109
0.061
0.118
B.C.
0.134
0.130
0.132
0.108
0.117
0.150
0.194
0.133
Maritimes
0.019
0.023
0.016
0.024
0.016
0.016
0.026
0.023
Quebec
0.102
0.128
0.114
0.100
0.098
0.085
0.090
0.092
Ontario
0.383
0.392
0.456
0.436
0.386
0.317
0.397
0.376
Prairies
0.366
0.370
0.295
0.288
0.365
0.429
0.339
0.382
B.C.
0.130
0.087
0.119
0.153
0.136
0.152
0.148
0.127
Northwestern Europe
Central & Eastern Europe
Southern Europe
Total Immigrants
Note: All columns refer to 1931 Census except column ’Stock 1921’ from 1921 Census.
Source: Census of Canada, 1921, 1931.
30
Table 4: Urbanization by Year of Immigration By Region, Immigrants and Native-born, 1930/31
Foreign Born
All years
1921−30
1911−20
Native
1901−10
<1901
born
Panel A: U.S.
New England
0.853
0.865
0.861
0.859
0.842
0.749
Middle Atlantic
0.885
0.921
0.885
0.880
0.867
0.748
East North Central
0.827
0.919
0.883
0.847
0.737
0.639
West North Central
0.475
0.630
0.592
0.531
0.409
0.413
South Atlantic
0.732
0.727
0.690
0.724
0.759
0.353
East South Central
0.725
0.759
0.718
0.711
0.733
0.278
West South Central
0.577
0.772
0.696
0.608
0.491
0.357
Mountain
0.470
0.502
0.423
0.444
0.491
0.389
Pacific
0.717
0.762
0.716
0.706
0.696
0.669
Panel B: Canada
Maritimes
0.578
0.483
0.631
0.664
0.589
0.355
Quebec
0.883
0.889
0.901
0.898
0.796
0.561
Ontario
0.718
0.694
0.742
0.743
0.683
0.548
Prairies
0.329
0.307
0.376
0.317
0.294
0.285
British Columbia
0.571
0.516
0.611
0.587
0.538
0.542
Notes: Maritimes are the provinces of Prince Edward Island, Nova Scotia, and New Brunswick.
The Prairies are Manitoba, Saskatchewan, and Alberta.
Source: Census of the United States, 1930; Census of Canada, 1931.
31
Table 5: Urbanization by Origin, 1920/21 and 1930/31
British
Isles
1920
Northwestern
Europe
1930
1920
Central/Eastern
Europe
1930
Southern
Europe
Native−born
1920
1930
1920
1930
1920
1930
Panel A: U.S.
New England
0.91
0.88
0.83
0.81
0.87
0.84
0.91
0.90
0.76
0.75
Middle Atlantic
0.86
0.88
0.82
0.84
0.86
0.88
0.89
0.91
0.72
0.75
East North Central
0.79
0.86
0.66
0.73
0.80
0.83
0.86
0.91
0.58
0.64
West North Central
0.57
0.61
0.35
0.39
0.45
0.47
0.78
0.83
0.37
0.41
South Atlantic
0.70
0.66
0.61
0.61
0.72
0.71
0.67
0.73
0.30
0.35
East South Central
0.70
0.74
0.57
0.60
0.69
0.73
0.67
0.74
0.22
0.28
West South Central
0.65
0.72
0.54
0.59
0.38
0.45
0.61
0.71
0.28
0.36
Mountain
0.52
0.56
0.40
0.43
0.40
0.44
0.37
0.46
0.36
0.39
Pacific
0.75
0.79
0.61
0.64
0.67
0.73
0.67
0.68
0.62
0.67
Total U.S.
0.81
0.84
0.58
0.64
0.78
0.81
0.85
0.87
0.48
0.49
Panel B: Canada
Maritimes
0.61
0.52
0.57
0.42
0.80
0.74
0.73
0.84
0.35
0.37
Quebec
0.91
0.91
0.85
0.85
0.96
0.95
0.95
0.95
0.53
0.61
Ontario
0.73
0.72
0.50
0.42
0.72
0.73
0.80
0.82
0.54
0.58
Prairies
0.52
0.47
0.21
0.21
0.28
0.26
0.57
0.54
0.35
0.36
British Columbia
0.51
0.61
0.33
0.42
0.31
0.39
0.54
0.62
0.46
0.56
Total Canada
0.65
0.65
0.34
0.35
0.46
0.48
0.77
0.80
0.48
0.53
Source: Census of the United States, 1920, 1930; Census of Canada, 1921, 1931.
32
Table 6: Coefficients, Canada Dummy Variable
1925−30
1920−24
1915−19
0.86 ***
1910−14
0.85 ***
−0.49
Germany
1.56 ***
0.62
−0.66 **
0.28
Greece
1.17 ***
0.08
−0.18
0.83 ***
−1.09 ***
Ireland
0.56
0.12
0.98 ***
1.13 ***
Italy
0.99 ***
0.12
−0.07
0.44 ***
0.35
Norway
1.10 ***
0.28
0.09
0.93 ***
0.87
Poland
1.83 ***
0.72 **
0.38
0.78 ***
Russia
1.97 ***
1.31 ***
Scotland
1.16 **
Sweden
0.77 *
0.48 **
−0.07
−0.66
0.70 **
0.57 ***
1901−09
England
0.58 **
1.37 ***
1.32 **
1.42 ***
−0.12
0.91 **
0.71 ***
1.37 ***
0.26
0.25
1.70 **
***=Significant at the 1% level.
**=Significant at the 5% level.
*=Significant at the 10% level.
Notes: Values are coefficients on dummy variables. Regressions themselves are reported in the
Appendix. Values without asterix are not significant at the 10% level.
Source: See the text.
33
Table 7: Coefficients on Regional Canada Dummy Variables
1925−30 1920−24 1915−19 1910−14 1901−09
1925−30 1920−24 1915−19 1910−14 1901−09
England
Norway
Maritimes
1.25 *** −0.55
1.34
0.40 *
1.18 ***
1.37 **
−0.13
0.75
0.59 *
0.13
Central
0.85 **
−0.31
0.61
Prairies
1.19 **
−0.36
0.83
−0.03
0.40 *
1.47 ***
1.27 **
−0.55
0.05
1.13 ***
2.28 *
2.19 ***
0.97
−1.62
0.60
1.11 ***
Pacific
0.85 **
−0.66
0.56
0.34
4.90 ***
1.74 ***
0.87 *
0.07
0.85 ***
2.20 **
0.89 **
Germany
Poland
Maritimes
1.74 ***
0.69
−1.45 ***
1.14 ***
1.25
1.29 ***
1.09
0.42
0.34
1.37 ***
Central
1.57 ***
0.23
−0.63 *
0.12
0.85
1.71 ***
0.63 **
0.76 *
0.94 ***
1.64 ***
Prairies
1.36 **
1.10
−0.83 *
0.19
2.67 ***
2.19 ***
1.48 **
0.10
1.30 **
Pacific
1.57 ***
0.71
−0.63 *
0.38 *
1.78 **
2.51 ***
0.92 *** −0.33
0.63 ***
1.32 ***
Greece
Maritimes
1.45 *** −0.28
Central
1.01 ***
Prairies
0.75 *
Pacific
1.30 ***
−0.33
−0.53
Russia
1.41 *** −0.97 *
1.56 ***
1.18 ***
0.36
0.44
0.04
0.25
0.64 *** −1.04 ***
2.46 ***
1.08 *** −0.23
0.40
−0.19
0.52
1.16 *** −1.04 *
4.00 ***
2.78 *** −0.31
−0.02
1.04 *** −1.14 ***
2.47 ***
1.75
0.15
−0.54 **
Ireland
−0.03
0.80 **
−0.02
0.96 **
0.47
−1.13 ***
Scotland
Maritimes
1.76 ***
0.30
0.55 *
0.87 ***
0.43
2.06 *** −0.66
1.30 ***
0.50 **
0.82
Central
0.57
0.20
0.74 ***
1.02 ***
1.07 ***
0.83 *
−0.25
0.67
0.73 ***
1.42 **
Prairies
1.00
0.38
0.73 *
1.02 ***
2.15 ***
1.23 *
−0.01
0.64
0.48
2.34 ***
Pacific
0.20
0.01
0.34
1.01 ***
1.50 ***
0.90 *
−0.52
0.39
0.69 ***
1.94 ***
Italy
Sweden
Maritimes
0.65 *
0.22
−0.17
0.20
Central
0.84 *** −0.05
−0.24
Prairies
1.16 *** −0.64
0.57
Pacific
1.37 ***
0.30
0.05
−0.65
0.69
0.82
0.06
−0.27
0.68
0.52 ***
0.41
0.58
0.53
0.63
−0.24
1.69
0.19
0.82
0.45
0.42 *
0.95
−0.55
3.13 **
0.41 **
0.54
0.96
0.83 ***
0.04
0.63
2.21 **
***=Significant at the 1% level.
**=Significant at the 5% level.
*=Significant at the 10% level.
Notes: Values are coefficients on dummy variables. Regressions themselves are reported in the Appendix. Values
without asterix are not significant at the 10% level.
Source: See the text.
34