Why do Wages Differ Across Countries? Lessons from migrants to Canada in the 1920s

By Frank Lewis, Professor Emeritus, Department of Economics, Queen’s University

A chambermaid in Canada has a wage more than twice that of a chambermaid in South Korea; a pharmacist earns four times what a pharmacist in India earns; and a registered nurse receives six times the earnings of a registered nurse in the Philippines. These approximate purchasing-power-parity comparisons typify the large wage gaps between Canada and many other countries; and lead to the question: why doesn’t the chambermaid in South Korea, the pharmacist in India, and the registered nurse in the Philippines move to Canada? 

There is the airfare, hardly a serious impediment; but immigrants must be pass through Canada’s immigration system and, depending on the occupation, they may require certification in Canada. In the 1920s prospective European immigrants faced the cost of travelling to Canada and there was a short period of lost earnings. Otherwise migration was effectively unrestricted and indeed supported in some cases by government and private agencies.  Yet the wage gaps for similar occupations were not much different from those we see today.

In the late 1920s a mason in Canada earned three times the wage of a mason in the Netherlands, a carpenter received four times the wage of a carpenter in Italy, and the wage of unskilled labourer was more than twice that of an unskilled labourer in Poland. By analysing immigration to Canada from five European countries in the 1920s : Ireland, Italy, the Netherlands, Poland, and Sweden, Alex Armstrong and Frank Lewis (2016) shed light not just on the wage gaps but on the process of immigration generally.

The inability to borrow to pay for migration was a serious barrier.  It meant that prospective migrants first had to save, delaying the time when they could migrate.  Median age of migration was in the late 20s. It also meant that emigrants faced an initial period of low consumption. The inability to smooth consumption seriously affected the desirability of migrating. Immigrants, especially from non-English-speaking countries, faced an initial period of low earnings as they acquired the background needed to enter the occupation for which they had trained. And nearly all immigrants suffered from the loss of contact with culture and community when they moved to Canada. Armstrong and Lewis introduce all these aspects into a formal life-cycle model, where a prospective migrant chooses optimal migration age and, subject to that choice, compares lifetime utility if they do or do not migrate.

An important element is explaining migration is the loss of the amenities associated with home country. This is affected not just by the country of origin, but also but also by wage of the prospective migrant. Treating amenities in the utility function as additive, it follows that higher wage workers will require a larger proportionate increase their wage in order to migrate. This is reflected in the wage comparisons: the higher the skill the greater was the proportional wage differential between Canada and the home country. Another factor that strongly affected amenity considerations were views about status. Prospective immigrants cared not just about their income, but also about their place on the income distribution. Per capita income in Canada was higher than in each of the five European countries, but the gap was especially great between Italy and Poland. Migrants from these countries required greater compensation.

Once these factors are taken into account, the experience of workers in the five European countries shows that, far from being excessive, the wage gaps are found to have been a reasonable reflection the various (utility) costs of settling in Canada rather than staying at home.

Table 1 provides the decomposition. An unskilled worker in Canada earned 71 percent more than one in Ireland. Of this differential, 15.3 percentage points represented the migration cost that included gradual convergence to the Canadian wage. Low initial consumption due to the inability to borrow accounted for 5.7 percent, loss of status due lower per capita income in Ireland and described here as the “Roy” effect accounted for 23.8 percent, and the remaining “taste” effect, reflecting the amenity value of living in Ireland, was 26.2 percent. That is, an unskilled worker in Ireland needed a 26.3 percent increase in annual consumption to be indifferent between moving and staying. Note that because of diminishing marginal utility of consumption the skilled worker needed a 43.9 percent increase in their consumption.

Table 1. Decomposing the wage gaps between Canada and Europe, 1925-1929 (percent)

Actual Migration
Borrowing Constraint “Roy” Effect Taste
   unskilled 71 15.3 5.7 23.8 26.2
   skilled 103 29.4 12.7 16.7 43.9
   unskilled 210 47.7 42.8 50.5 68.9
   skilled 314 84.9 83.8 37.4 108.2
   unskilled 72 21.2 7.0 11.6 32.2
   skilled 119 38.0 17.6 8.9 54.8
   unskilled 204 48.9 39.9 50.3 64.9
   skilled 220 59.9 42.7 29.0 88.7
   unskilled 32 12.2 1.5 6.5 11.9
   skilled 82 29.1 10.4 4.9 37.1

The results for Ireland correspond closely to those for the other the relatively high income countries, the Netherlands and Sweden. Per capita incomes in Italy and Poland were much lower and the wage gaps much greater.  Because wages were also much lower in these countries, emigrants had to reduce consumption to low levels to save for migration, which had to be compensated by very much greater consumption later on. Thus, for example, 42.8 percent of the more than 200 percent gap in the unskilled wage for Italy is explained by the borrowing constraint. The status or “Roy” effect was also much greater because workers in Italy in the same occupation as those in Canada were very much higher on the income distribution. Although the amenity (“taste”) effects are calculated to be somewhat greater than for the higher-income European countries, the overall message is that equilibrium wage differentials as high as 200 and 300 percent can be accounted for.

These findings and the related Armstrong-Lewis approach to self-selection in migration have broad implications that correspond to what was observed in the 1920s and in many way reflect what happens currently.

First, immigrants will not normally migrate at the start of their working lives. The need to accumulate capital will lead to a considerable delay. Second, because of the need to save immigrants require an immediate source of income after arrival. In the 1920s the vast majority of immigrants began as farm workers, but then quickly shifted to their intended occupation. Today, many immigrants start by driving taxis. Third, the borrowing constraint leads to a distorted consumption stream. The inability to fully smooth consumption means that immigrants must be compensated with much greater consumption over their lifetime. Fourth, workers with higher wages in the home country need a greater increase in their consumption abroad to compensate for the amenity benefit of remaining in the home country. This follows directly from diminishing marginal utility of consumption and is reflected in the generally greater proportional wage gaps for higher-skilled workers.  Finally, status, which has been found to have an important impact on well-being, applies to emigrants. This means that those in low-income must be compensated with much higher wages to offset the effect shifting to a lower rank on the income distribution.

All these effects and others become evident once the migration decision is recognized as depending not on lifetime income, but rather on the pattern of lifetime consumption.  As well the formal use a taste parameter, and recognition of status as an important influence opens many avenues for studying and explaining migration patterns.


Armstrong, A., and F. Lewis (2016) “Transatlantic Wage Gaps and the Migration Decision: Europe-Canada in the 1920s,” Cliometrica (online, March 24, 2016 – print to follow)

Photo of workers in the Firestone Tire & Rubber Company of Canada in Hamilton, 1923. Public domaine.