The impact of the 1962 repatriates from Algeria on the French labor market.
Previous studies of immigration have found only small effects of immigrants on labor market outcomes of natives, but few have addressed the problem of simultaneity between immigrant location choice and local labor market conditions. The importance of this study lies in its examination of a natural experiment in immigration, similar to the Mariel Boatlift studied by Card (1990). The timing of the immigration was independent by French economic conditions, and there was no selection bias among immigrants, as almost all those of European origin left Algeria at the end of the war. Cultural and climatic considerations had a large influence on the repatriates' choice of region, which was independent of local unemployment and wage rates. As these considerations led the repatriates to settle disproportionately in the south of France, the regional variation necessary for a cross-sectional study was established.
This paper examines the effect of the repatriates on unemployment of non-repatriates, on the wages of different occupations, on the labor force participation of non-repatriates, and on the migration decisions of other groups. Census data from France's 90 departments (provinces) are used to estimate cross-sectional equations for 1968 for the first three outcomes. Using additional data from the 1962 census, differenced equations are also estimated to take departmental fixed effects into account: a 1962 "pre-arrival" equation is subtracted from the 1968 "post-arrival" equation.
The fact that the influx was one of skilled labor suggest that lessons learned from this study might be most usefully applied to the western part of the Federal Republic of Germany, which has been receiving large flows of skilled workers from eastern Europe, including the area of the former German Democratic Republic. The results could also have some application to Britain, in view of the 1.1 million white South Africans entitled to right abode in Britain (The Economist 1990:60).
French nationals first began returning from overseas possessions in the 1950s. The loss of Tunisia and Morocco in 1956 led to particularly large numbers of repatriations. Some Europeans returned from Algeria before independence in 1962. Figure 1 shows the contribution to population growth of natural increase, repatriates from all overseas possessions, and foreigners.(1)
The Accords of Evian, which promised independence to Algeria, were signed in March 1962, and approved by the French electorate in April 1962. In May trickle of Europeans leaving Algeria became an exodus. The vast majority returned to France, where they settled disproportionately in the warmer southern departments, as the earlier repatriates had done. The first column of Table 1 and the map in Figure 2 show what percentage of the 1968, labor force the later repatriates constituted in the 21 regions of France. Column 2 of Table 1 gives equivalent information for the earlier repatriates (for whom participation rates are unavailable).(2)
[TABULAR DATA OMITTED]
The 330,000 repatriates from Algeria in the labor force in 1968 represented 1.6% of the French labor force.(3) By comparison, 500,000 foreign immigrants and 150,000 other French immigrants had arrived since 1962. The labor force was additionally expanded by the entrance of demobilized soldiers and baby boom cohorts. The 1968 labor force participation rate of the repatriates, 36.7%, was lower than the 41.1% rate for France as a whole. The repatriates were younger than the population living in France, but older than other migrants; they were especially strongly represented in the cohort aged 25-39 in 1968 (thus 19-33 upon arrival in France): 22.6% of repatriates were in this cohort, as opposed to 18.8% of the general population (Desplanques 1975).
It is generally considered that the reintegration of the repatriates was relatively painless (see, for example, Baillet 1976). The repatriates did have some advantages not normally available to immigrants. They received a special benefit lasting up to a year to support them during the job search, in addition to other benefits such as a lump sum toward housing costs. Further, the government in October 1962 and March 1963 asked all firms to provide a list of vacancies, and the repatriates were given first priority for these jobs. Contrary to what one might think, the repatriates possessed little capital of their own, most of it being tied up in property in Algeria for which no compensation was received. According to Baillet (1976), each had, on average, 5,000 francs, less than the average annual production worker's salary of 7,300 francs in 1962.(4)
Information from the 1968 census shows that the repatriates were over-represented in skilled occupations. They formed 2.9% of the upper-level professional (cadres superieurs) category (liberal professions, managers, engineers), compared with 1.6% of the labor force. The repatriates were over-represented among middle-level professionals and office workers (cadres moyens and employes) and the army and police, but under-represented among production workers (of whom 1.2% were repatriates) and among the self-employed.
Figure 3a shows the number of unfulfilled requests for work and the number of vacancies registered in employment bureaus at the end of each month. The vacancies series illustrates the favorable climate at the time of the repatriation and also the subsequent downturn. The decomposition of unsatisfied request for work into those filed by repatriates and those filed by others, shown in Figure 3b, appears to support the view that the repatriates were absorbed reasonably quickly. The numbers by region confirm that the shock was borne principally by the south, but a plot of average annual salaries for the period shows no dip in the southern regions relative to the national average. (These regional graphs are not shown but both are available from the author.) The 1968 census, however, reveals that six years after arriving, the repatriates had more than double the unemployment rate of non-repatriates at the national level. Unemployment in France was 1.7% in 1954 and 1.0% in 1962, and the rate for non-repatriates in 1968 was 2.1%, whereas for repatriates it was 4.5%. The right-hand columns of Table 1 give the figures by region.
The construction undertaken to house the repatriates may well have stimulated the local economies in which they settled, and may have attracted workers from other regions or countries (see Tapinos 1975:56).
Theoretical Background and
Theoretical predictions about the effects of immigration on the labor market depend on assumptions about trade, substitutability of natives and immigrants, and sizes of elasticities associated with different responses. In an autarkic economy, the increase in labor supply may lead to a fall in wages for natives if the immigrants are substitutes for the natives. In the short run there will be some unemployment due to non-instantaneous matching of workers and jobs, and this period of unemployment may be prolonged if wages are sticky for some reason. Conversely, the stimulus given to aggregate demand might, in fact, raise natives' wages. Changes in regional wages and unemployment could affect regional migration patterns.
In an open economy model with factor-price equalization, wages in the long run will remain at the world level. If time is needed to reallocate capital to the labor-intensive sector, the wage will be lower for a time. If wages are sticky, there will instead be a period of unemployment.
Most studies of immigration to the United states have found small or no effects on the labor market outcomes of natives (see, for example, Altonji and Card 1991; card 1990; LaLonde and Topel 1990). Altonji and Card, using 1970 and 1980 data, found that a one percentage point increase in the fraction of immigrants in the population reduced wages by 1.2% and did not affect unemployment. The labor market outcomes considered have included wages, unemployment, employment, weeks worked, and participation rates. Filer (1991) found, using 1975 and 1980 data, that natives in the United States were discouraged from moving to areas receiving inflows of foreigners. He speculated that this pattern could explain the small size of labor market effects of immigrants found in studies relying on cross-sectional variation in the concentration of immigrants.
Puig (1981) estimated equations for gross immigration between French regions of members of the labor force under 30 years of age, in the periods 1962-68 and 1968-75. The high correlation between the proportion of young repatriates from Algeria and the youth unemployment rate, either of which had a negative and significant coefficient when the other was excluded from the 1962-68 equation, coupled with the insignificance of youth unemployment in the 1968-75 equation, led him to conclude that the repatriates caused unemployment and discouraged internal immigration to regions where repatriates were concentrated. Further investigation of net rather than gross immigration, using departmental rather than regional observations, seems warranted, as does a direct examination of the repatriates' impact on unemployment.
The primary sources of data in the analysis are the censuses of 1962 and 1968, published by INSEE (Institut National de la Statistique et des Etudes Economiques). Because the 1962 census was taken in March, just before the exodus from Algeria began, it provides data describing conditions before the arrival of the repatriates. The next census in 1968 was also taken in march. The 1968 data used in this study are a one-quarter sample of the census forms.
The census definition of repatriates from Algeria for the 1962-68 period is persons in Algeria at the same time of the 1962 census and in France at the time of the 1968 census whose family name was not Arab or Berber.
The data on salaries are those furnished by employers to the tax service along with the number of employees, their occupations, and the periods worked. INSEE publishes average annual earnings (total money earned in the year divided by person-years of work) from a 1/25 sample. Employers whose information is used are those in the private and semi-public sectors, regardless of size. "Semi-public" refers to certain nationalized enterprises such as the SNCF (the railroad company) and Renault. Domestic servants and the self-employed are excluded. Salary data are the only data available by department or region in the years between 1962 and 1968.
The salary data have several deficiencies, which are discussed in detail in the data appendix. In particular, the appendix explains my decision to use 1962 and 1967 salary data. The appendix also contains summary statistics for the variables used in the empirical analysis in the next section. The analysis uses 88 observations, one from each department (province) except Corsica, which was excluded, and the Paris region departments, which were aggregated (see the data appendix).
Empirical Models and Results
Location of the Repatriates
In evaluating the extent to which the case at hand can be viewed as a good natural experiment, the characteristics of regions that attracted large absolute numbers of repatriates are less germane than those of regions where repatriates formed a large percentage of the work force. The repatriates arriving between 1962 and 1968 are assumed to have based their location decisions on 1962 variables. The equation estimated is (1) [repatriates.sub.1968,i]/[labor force.sub.1962,i]= f(early [repatriates.sub.1962,i], [temperature.sub.i], [unemployment.sub.1962,i], In [w.sub.1962,1], gov't [incentive.sub.i], department [structure.sub.1962,i])
The i subscript indicates that the estimation is a cross-section across departments.
The dependent variable is the number of repatriates in the labor force in 1968 as a proportion of the 1962 labor force. The early repatriates variable on the right-hand side of the equation is the percentage of the 1962 total population represented by the earlier repatriates from Algeria (who arrived in the years 1954-61). This variable is included to see if repatriates preferred to settle close to their compatriots.
Temperature is included to test the assertion that warmer areas had a larger proportion of repatriates. The temperature used is the average annual temperature in the administrative center of the department. The unemployment rate and log wages gauge the responsiveness of the repatriates to the main economic incentives. The "government incentive" is a variable representing government incentives (2,500- or 5,000-franc lump sums) for the repatriates to move to certain departments with a labor shortage. It equals 2 for departments in which a lump sum payment of 5,000 francs was offered, 1 for departments in which the lump sum offered was 2,500 francs, and 0 for departments in which no incentive was offered. Department economic structure is proxied by the employment shares of different sectors. The omitted sector is agriculture.
The result of the weighted least squares estimation as shown in Table 2; the weights are the 1962 labor forces in the different departments. An estimation using ordinary least squares yields very similar results.
Table 2. Determinants of the Location of Repatriates in the Labor Force, 1968. (Standard Errors in Parentheses) Coefficient Independent Variable and Std. Error Repatriates from Algeria, 1954-62, 22.44(**) as % of 1962 Population (0.18) Temperature 0.253(**) (0.048) Unemployment Rate, 1962 0.325(**) (0.152) Log Salary, 1962 -1.16(**) (0.54) Government Incentive(a) 0.107(*) (0.056) Services, 1962(b) -0.017 (0.043) Commerce and Banking, 1962 -0.002 (0.030) Mining, 1962 0.004 (0.012) Other Industry, 1962 0.019(**) (0.006) Construction, 1962 0.115(**) (0.029) Public Sector, 1962 0.060(**) (0.020) Transport, 1962 -0.028 (0.032) Adjusted R(2) 0.97 Note: The dependent variable is repatriates in the labor force in 1968 as a proportion of the 1962 labor force. The sample size is 88. The equation is estimated using GLS, with the 1962 labor forces as weights. (a) The government incentive variable has values of 0, 1, or 2; see text for explanation. (b) Sectoral variables are employment shares in % and proxy the economic structure of the department. Agriculture is the omitted sector. (*) Statistically significant at the .10 level; ** at the .05 level (two-tailed tests).
Clearly, the predominant factor in explaining the location of the repatriates is the location of earlier repatriates from Algeria. The coefficient of the earlier repatriates implies that a department with one percentage point more early repatriates attracted 2.4 percentage points more later repatriates to the labor force. It is unlikely that all of his effect is causal. There is anecdotal evidence, however, that established repatriates helped the new arrivals to find jobs, and the desire to retain as much of their "Pied Noir" culture as possible probably led to the earlier repatriate centers acting as seeds: the earlier repatriates attracted the first 1962 arrivals, who in turn made their area more attractive for those arriving in the following months. It may also be that the earlier and later repatriates had similar human capital to which the returns were high in the chosen regions. It is interesting to note that the location of earlier repatriates from Tunisia or Morocco does not predict nearly as well where the later repatriates from Algeria would settle as does the location of earlier repatriates from Algeria. (These results are not shown but are available from the author.)
The second important factor is the temperature. A one-degree (Celsius) increment in annual average temperature is associated with a 0.25 percentage point increment in the proportion of repatriates. The range of average temperatures in France is from 9.4[degrees] C to 15.2[degrees] C (about 49-59[degrees] F). It is worth noting in passing that the average temperature is 17.5[degrees] C (63.5[degrees] F) in Algiers and 15.7[degrees] C (60.3[degrees] F) in Constantine.
The signs of the coefficients on average salary and unemployment are counter-intuitive. The coefficient on salaries might be reflecting wage falls at the end of 1962 in response to the repatriates, but the coefficient on unemployment cannot be explained in this way. The repatriates variable appears to pick up an unmodeled fixed effect: the repatriates went, ceteris paribus, to areas with high levels of unemployment and low salaries. In subsequent regressions, this problem is addressed by differencing the 1968 and 1962 cross-sectional equations.
Now the impact of the repatriates on labor market outcomes may be investigated. First, the equation for the unemployment rate of non-repatriates is estimated. The basic form of the equations estimated assumes that unemployment may be modeled using independent variables that are reasonably exogenous. The 1968 cross-sectional equation estimated is
(2) non-repatriate [unemployed.sub.1968,i]/non-repatriate labor [force.sub.1968,i] =
f([repatriate.sub.1968,i], [education.sub.1968,i], [age.sub.1968,i], department [structure.sub.1968,i]
The repatriate variable is expressed as a proportion of the labor force in 1968. The education variable is the proportion of the population no longer studying whose highest qualification is a brevet superieur or baccalaureat (non-vocational secondary school qualifications) or university degree. The age variable represents young workers entering the labor force, and is equal to the proportion of the labor force aged 15 to 24 inclusive. This variable is endogenous to some extent, as young people are fairly mobile and migrate to areas with good opportunities, but this endogeneity is ignored.
The results of weighted least squares are shown in column 1 of Table 3. The weights are the non-repatriate labor forces.(5) Seven regional dummies are included on the right-hand side of the equation. The repatriates appear to have a very significant positive effect on the unemployment of non-repatriates. The coefficient means that a one percentage point increase in repatriates as a proportion of the labor force would cause a 0.34 percentage point increase in the unemployment rate of non-repatriates.
[TABULAR DATA OMITTED]
The first check on the cross-sectional equation is the inclusion of the repatriates variable from 1968 on the right-hand side of the 1962 unemployment equation. (The unemployment rate for 1962 is for the labor force as a whole.) If the cross-section effect were genuine, this variable would not influence 1962 unemployment. Column 2 shows, however, that the repatriates variable is still significant and positive, confirming that at least some of the significance in the 1968 equation is due to an unmodeled fixed effect. (The 1968 repatriates variable is also significant and positive in the 1954 cross-sectional equation.)
The second check is the estimation of the differenced equation (1962 is subtracted from 1968), which eliminates departmental fixed effects.(6) Although it is somewhat inconsistent with the idea of differencing, regional dummies are included also on the right-hand side of the differenced equation in column 3 to allow for different regional trends. They only slightly affect the repatriates variable.
The difference results show that the repatriates variable remains significantly positive, although the coefficient is reduced. A one percentage point increase in the proportion of repatriates would lead to a 0.12 percentage point increase in the non-repatriate unemployment rate.
In column 4, the repatraites variable of the column 3 equation has been instrumented with temperature and the proportion of the 1962 population represented by 1954-62 repatriates as excluded instruments. This specification rises both the significance and coefficient of the repatriates variable. The chi-squared statistic (with one degree of freedom) yielded by the Generalized Method of Moments Specification Test for orthogonality of instruments and residuals is 0.014.(7) Although this test is very weak, the failure to reject orthogonality is clear.
This analysis has ignored wages, participation rates, and the concentration of foreign immigrants, which are clearly all related to unemployment. The difficulty of finding good instruments for the differences of these endogenous variables prohibits identification of a differenced equation that takes them into account. If these differenced variables are simply added to the regression of column 3, however, the repatriates coefficient and standard error change little (these results are not shown).(8)
The regional dummies may be representing regional sensitivities to the business cycle that have not been captured by the other variables, and to the extent that they do so imperfectly, the repatriates variable may also reflect regional trends. The general fall of unemployment from 1954 to 1962 and the rise in unemployment from 1962 to 1968 were particularly marked in the departments of the south. The 1968 repatriates variable is negative and significant when included in the 1954-62 differenced equation if no regional dummies are used. The repatriates variable is less significant if regional
dummies are included, and the sign of the dummies is the opposite of that in the 1962-68 analysis.
Many different methods were used to attempt to capture this business cycle effect in the 1962-68 differenced equation, but none of them changed the results significantly. Most important, the industry variable was split up in several different ways, but the repatriates variable (and the regional dummies) remained significant. If the three departments with the highest unemployment increases from 1962 to 1968 and the highest proportion of repatriates are excluded from the 1962-68 differenced equation, the coefficient on the repatriates variable is no longer significant at the 5% level.
In conclusion, it appears that unemployment has not been modeled very successfully, but that the repatriates probably increased unemployment somewhat. The upper bound of the coefficient is 0.20, but there is evidence that the repatriates are picking up a business cycle effect, so the upper bound is probably too high. This magnitude implies that for France as a whole the arrival of repatriates representing 1.6% of the labor force increased non-repatriate unemployment by up to 0.3 percentage points and the repatriates thus accounted for up to 29% of the increase in unemployment in the period 1962 to 1968. In the department of the Var, the repatriates were 7.1% of the labor force in 1968, and thus increased unemployment by up to 1.4 percentage points. Unemployment in the department in 1962 was 2.9% and in 1968 among non-repatriates was 5.5%. Thus, the repatriates may have caused up to 54% of the increase in unemployment in the department of the Var.
Analysis analogous to that used for unemployment was used to assess the influence of the repatriates on the participation rates of non-repatriates, and no effect was found. (The tables presenting these results are not shown but are available from the author.)
The analysis of wages has three limitations. First, it is not possible to distinguish composition effects from effects on the indigenous French, as no salary data are available on the repatriates separately. Second, there exist no regional price indices for the period under examination, and thus nominal wages rather than real wages must be used in the estimation.(9) Third, although it would have been desirable to include union membership information, none exists, even at the national level.
A cross-section for 1968 (although with 1967 salaries) is estimated as follows:
(3) In [w.sub.1967,ik], =
The k subscript indicates that the wage and repatriates variables may be broken down by occupational category. No regional dummies were used for the wage equations; rarely were more than one of the regional dummies significant in any equation, and in addition the inclusion of the dummies reduced the degrees of freedom too greatly. The centralized system of setting minimum wages by occupation may account for this lack of regional variation after sectoral shares are controlled for.
The weighted least squares results are shown in column 1 of Table 4; the weights are the salaried labor forces, and all coefficients are multiplied by ten. The results suggest that the repatriates had a negative effect on wages. The results of the unweighted equation are similar.
[TABULAR DATA OMITTED]
When the variable for repatriates arriving 1962-68 is included on the right-hand side of 1962 salary equation (column 2), it is significantly negative. This result suggests that the repatriate variable is picking up an unmodeled regional fixed effect in the 1968 equations. Alternatively, it may be that salaries had already adjusted to some extent during 1962, and that this adjustment is the cause of the negative coefficient. Again the unweighted result is similar.
The results of the differenced equations are in columns 3-5. Both weighted and unweighted results are presented, since they differ. The dependent variable in column 3 was calculated using the unadjusted 1962 salary variable, and the results are for weighted least squares. Columns 4 and 5 show weighted and unweighted results for the dependent variable, which uses the corrected 1962 salaries (see the data appendix). Only in the weighted equation with the corrected salary are the repatriates significant. A one percentage point increase in repatriates in the labor force implies a 0.51-0.80% decrease in wages.
Several other equations were also estimated to check the robustness of the results. (The tables are not shown, but are available from the author.) The repatriates coefficient has similar or lower significance when temperature and early repatriates are used as instruments for the repatriates in the differenced equations, when differenced unemployment, foreigners in the labor force, and participation rate were added to the right-hand side of the differenced equations, when the regressions were run for three occupational categories, and when 1968 salaries were used.
From these many equations emerges only weak evidence of any fall in wages in response to the arrival of the repatriates. Still, wages may have been up to 1.3% lower in France in 1967 than they would have been in the absence of the repatriation, and 5.7% lower in the Var. These results may have been biased toward zero by the use of 1962 rather 1961 salaries.
Migration of Other Groups
Equations similar to Filer's (1991) are estimated to investigate whether repatriates discouraged immigration, either from abroad or from within France, to their chosen departments. If this were the case, the true effect of the repatriates would be underestimated by the cross-sectional approach. The two equations have the same independent variables:
(5) net internal [immigrants.sub.1968,i]
= labor [force.sub.1962,i]
ln [w.sub.1962,i], department [structure.sub.1962,i])
(6) gross international [immigrants.sub.1968,i]
= labor [force.sub.1962,i]
ln [w.sub.1962,i], department [structure.sub.1962,i])
International immigrants are any members of the 1968 labor force who were not in France in 1962 (and who were not repatriates). Similarly, internal migrants are labor force members who resided in different departments in 1962 and 1968. Unfortunately, the data do not permit calculation of net international immigration.(10)
The coefficients of the weighted regressions are in Table 5; the weights are the 1962 labor forces. Unweighted results are very similar.
[TABULAR DATA OMITTED]
The single stage results in columns 1 and 3 show the repatriates variable to have an insignificant coefficient in the equation examining internal migration, and a positive and significant coefficient in the equation examining international immigration. Instrumenting the repatriates variable with temperature and 1954-62 repatriates does not change the results greatly (columns 2 and 4). The coefficient in the international immigration equation is large: a one percentage point increase in repatriates is associated with a 0.4-0.6 percentage point increase in immigrants from abroad.
Unemployment and salaries clearly play an important role in the migration decisions of non-repatriates, and insofar as the repatriates affected these two factors, they affected migration patterns. If the significant coefficients on the repatriates and unemployment variables in both equations are considered, however, the total impact of the repatriates is seen to be about zero. Even if the possible salary change is included, the effect is very small.
The influx of repatriates from Algeria to France in 1962 represents a good natural experiment for examining the effects of immigration on the labor market. Although they suffered a high unemployment rate themselves, the repatriates had little impact on the unemployment of others. A one percentage point higher proportion of repatriates implied an increase in non-repatriate unemployment of at most 0.2 percentage points in 1968. Although this result indicates only a 0.3 percentage point increase in non-repatriate unemployment due to the repatriates at the national level, it implies a 1.4 percentage point increase in the department with the highest proportion of repatriates, the Var.
There is only weak evidence that the repatriates exerted downward pressure on wages; at most, a one percentage point higher proportion of repatriates is associated with wages that were 0.8% lower in 1967. This result implies that repatriates may have caused a 1.3% reduction in wages at the national level and a 5.7% reduction in wages in the Var. The relative movements of departmental wages in years between censuses do not suggest the existence of a larger effect that had disappeared by 1967.
The labor force participation rates of non-repatriates were unaffected by the presence of the repatriates.
The small magnitude of the elasticities found in this study is consistent with results for immigration to the United States (although I find more adjustment in unemployment and less in earnings than did Altonji and Card in their 1991 study). The explanation suggested by Filer (1991) for the small elasticities found in the American studies - namely, that internal migration compensates for the immigration from abroad - is, however, not applicable to this French study. Although internal migrants were discouraged from moving to areas where repatriates had increased the unemployment rate, international immigrants were attracted to areas with many repatriates, and the two effects offset each other.
Table A1 summarizes the data. There are several problems with the salary data. First, the average annual earnings by department for 1961, the last full year before the large influx of repatriates, are irretrievable. Thus, 1962 must be used as the "pre-arrival" year, in the hope that there may have been some lag in any adjustment of wages.
[TABULAR DATA OMITTED]
Second, the "post-arrival" year should be 1968, the year of the census. Social upheaval, however, beginning after the census had been taken, led to wage accords granting large pay increases to less skilled workers. The effects of this development on the results of the present study are unpredictable, and I therefore thought it best to use the salary data from 1967.
Finally, the 1962 data are only for workers who spent the whole year with the same employer, whereas the 1967 data are for all workers. The two are thus not comparable. It is possible to correct for this problem to some extent: since the 1962 salaries for both groups of workers (all occupations together) are known by region,(11) it is possible to use the ratio of the two to correct the 1962 salaries by department (all occupations together) fairly well.
The statistics on vacancies and unsatisfied requests for work are published by the Ministere du Travail, and come from the departmental employment bureaus. Registering at the employment bureau was a requirement for receiving unemployment benefits.
Use of census data leads to slightly misleading indications of change in the employment mix between 1962 and 1968, as employment in industry is most sensitive to the fact that 1962 was a boom year while 1968 was not.
The 1968 census contains information exclusively on the 1962 wave of repatriates, so it is not possible to know by department the 1968 stock repatriates that had arrived since 1954. In the differenced equations it is therefore assumed that there were no repatriates before March 1962; the repatriates variable is used as both a stock and flow.
The repatriates were, in general, French citizens. A small number, however, were foreigners, and the repatriates and foreigners variables thus overlap. This overlap is very small, and is ignored.
In 1964, the Paris region, previously divided into three departments, was divided into seven. Only one of the old departments is distinguishable in the new salary data, and the rest must therefore be aggregated. Corsica was excluded from the analysis due to problems with its census data. These adjustments resulted in 88 observations.
(1) Note that since different groups had very different labor force participation rates, a chart showing the contributions of these groups to the increase of the labor force would look different. (2) 100,000 Arabs and Berbers and 100,000 Jews also fled Algeria in 1962. They were accorded neither the status nor the privileges of the repatriates, and no data are available on them. It is known, however, that the Arabs and Berbers did poorly in the labor market (see Baillet 1976). (3) Except where otherwise specified, "repatriates" henceforth refers to those returning to France in the years 1962-68. (4) In 1962 and 1967 there were 4.9 French francs to the American dollar. (5) All regressions of Table 5 are weighted. The unweighted results are very similar. (6) The repatriates variable does not change upon differencing. See the data appendix. (7) In this test the two-stage least squares residuals are regressed on the instrumental variables. The [R.sup.2] from this regression has asymptotic chi-square distribution when multiplied by the degrees of freedom from the original equation. See Newey (1985). (8) A correction is made to the 1962 salary variable used to calculate the difference. For an explanation, see the data appendix. (9) A recent study, using regional price indices for 21 cities in 1977 and 1985, found that different costs of living accounted for approximately one-third of the difference in nominal wages between Paris and the rest of France that remained when gender and occupational composition had been accounted for. Adjusting for the cost of living also caused some alteration in the rank-ordering of wages in cities outside Paris. Unfortunately, such analysis is not possible before 1977 (see Centre d'Etude des Revenus et des Couts 1989). (10) In principle, foreign immigrants (apart from Algerians, to whom special rules applied) could only enter France by passing through a central office that matched them with employers who had requested foreign workers. By the time of the period in question, however, the majority of foreign immigrants were by-passing this procedure, and were thus able to make location choices themselves. (11) Recall that there are 21 regions, each of which contains several departments.
Altonji, Joseph, and David Card. 1991. "The Effects of Immigration on the Labor Market Outcomes of Natives." In John Abowd and Richard Freeman, eds., Immigration, Trade and Labor. Chicago: University of Chicago Press, pp. 201-34. Baillet, Pierre. 1976. Les Rapatries d'Algerie en France. Paris: La Documentation Francaise. Baudelot, Christian, and Anne Lebeaupin. 1979. Les Salaires de 1950 a 1975. Paris: INSEE. Card, David. 1990. "The Impact of the Mariel Boatlift on the Miami Labor Market." Industrial and Labor Relations review, Vol. 43, No. 2, pp. 245-57. Carre, Jean-Jacques, Paul Dubois and Edmond Malinvaud. 1972. La Croissance Francaise. Paris: Editions de Seuil. Centre d'Etude ded Revenus et des Couts. 1989. Les Francais et leurs Revenus. Paris: La Documentation Francaise. Commissariat General du Plan. 1966. Ve Plan 1966-70: Rapport General de la Commission de la Main-d'Oeuvre. Paris: La Documentation Francaise. Desplanques, Guy. 1975. Les Migrations Intercensitaires de 1962 a 1968. "Collections de l'INSEE, serie D No. 39, pp. 3-89. The Economist. 1990. "Bagehot." July 7, p. 60. Filer, Randall. 1992. "The Impact of Immigrant Arrivals on Migratory Patterns of Native Workers." In Richard Freeman and George Borjas, eds., The Economic Effects of Immigration in Source and Receiving Countries. Chicago: University of Chicago Press. INSEE (Institut National de la Statistique et des Etudes Economiques). 1966. Recensement General de la Population de 1962 - Resultats du Depouillement Exhaustif. Paris: Imprimerie Nationale. _____. 1971. Recensement General de la Population de 1968 - Resultats de Sondage au 1/4. Paris: Imprimerie Nationale. _____. Unpublished data from 1968 census. LaLonde, Robert, and Robert Topel. 1991. "Labor Market Adjustments to Increased Immigration.' In John Abowd and Richard Freeman, eds., Immigration, Trade and Labor. Chicago: University of Chicago Press, pp. 167-99. Ministere du Travail. 1959-67. Statisques du Travail et de la Securite Sociale, various issues. Muet, Pierre-Alain, and Patrick Bolton. 1970. "Evolution de l'Emploi dans les Regions d'apres les Recensements de 1954, 1962 et 1968." Collections de l'INSEE, serie R No. 4, pp. 3-101. Muguet, Jean, and Paul Carrere. 1972. "Analyse Statitisque Globale de l'Evolution des Regions Francaises entre 1954, 1962 et 1968." Annales de l'INSEE, No. 34, pp. 4-29. Newey, Whitney. 1985. "Generalized Method of Moments Specification Testing." Journal of Econometrics, Vol. 29, No. 3, pp. 229-56. Padieu, R., S. Volkoff, and M. Calviac. 1971. "Les Salaires dans l'Industrie, le Commerce et les Services en 1967 et 1968." Collections de l'INSEE, serie M No. 8, pp. 3-124. Puig, Jean-Pierre. 1981. "La Migration Regionale de la Population Active." Annales de l'INSEE, No.44, pp. 41-73. Tapinos, Georges. 1975. "L'Immigration Etrangere en France." Travaux et Documents, No. 71, pp. 1-143.
|Printer friendly Cite/link Email Feedback|
|Date:||Apr 1, 1992|
|Previous Article:||The economic progress of black women, 1940-1980: occupational distribution and relative wages.|
|Next Article:||The effect of hospital bargaining unit structure on industrial relations outcomes.|