Determinants of labour migration in an enlarged European Community

(1)

EUROPEAN UNIVERSITY INSTITUTE, FLORENCE DEPARTMENT OF ECONOMICS

320 EUR

E LI I W O R K I N G P A P E R N o . 8 6/ 230 DETERMINANTS OF LABOUR MIGRATION IN AN ENLARGED EUROPEAN

by

F Z & \ \

Rosemarie Feithen

* This paper was w'ritten during my one-year stay as Jean Monnet Fellow attached to the Economics Department of the EUI.

(2)

reproduced in any form without permission of the author.

(C) Rosemarie Feithen. Printed in Italy in August 1986

European University Institute Badia Fiesolana - 50016 San Domenico (Fi) -

Italy © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(3)

During the sixties and the early seventies, labour migration in the member states of the European Community had reached a very high level. After the immigration stop in 1973 by all countries of the Community, however, migration decreased considerably. Even between the EC member states migration has slowed down despite the free movement of labour. Also in the past, free movement of labour within the European Community proved to have relatively little effect on the inter-Community migration.

In the years before the immigration stop, more than one third of the migrant workers from non-EC countries who took up employment in the European Community were of Greek, Portuguese, and Spanish origin. Therefore, it can be expected that due to the enlargement of the EC freedom of movement will increase much in its importance.

The greatest part of the migrant workers concentrated in the agglomeration areas in the centre of the European Community. The high concentration of foreign labour partly caused considerable structural changes of the national and the regional labour markets. It is assumed that after the introduction of the regulation of free movement of labour in the new member states Greece, Portugal and Spain, migration from these countries will increase considerably.

(4)

For the labour market policy and for the regional policy of the member states as well as the EC it is important to know which factors influence labour migration in the Community.

In the following migration model the determinants of international labour migration will be analysed. The model is a multiple regression model which will be tested with data for the different countries of the European Community. The data of the member states of the European Community are appropriate for such a test, since labour migration within the European Community is without any restrictions due to the regulation of free movement of labour. The model should mainly serve the purpose of estimating potential migration of labour from the new member states, Greece, Portugal and Spain.

In the first part of the paper different theoretical approaches to the explanation of labour migration will be presented as well as different migration models will be examined. The second part relates to the hypothesis of the determining factors of international labour migration and specifies the model of migration. In the third part the results of the empirical tests of the migration model are presented and a method to achieve more accurate parameter estimates will be suggested. Finally, in the fourth part, the conclusions of the econometric results are presented.

(5)

1. Migration theories

Before the model is presented, the different migration theories and migration models are considered.

There is a multitude of theoretical approaches to explanations of migration, but there is no general theory which encompasses all relevant aspects of migration. The various disciplines dealing with migration research generally start from different explanatory approaches. Most of the migration models are confined to a monocausal explanatory approach.

In the literature dealing with migration research much criticism can be found because of the lack of universality of the theoretical statements, the ad-hoc character of numerous explanatory models, and also the frequently unsatisfactory theoretical foundation of the migration models (Albrecht, 1972, 10; Hoffmann-Nowotny, 1970, 82f.; Langenheder, 1968, 151; Szell, 1972, 22). For this reason, many of the migration models have only a very limited explanatory and forecasting value. On the other hand, it is very difficult to explain in a generally valid model migration processes which are the result of complex human behaviour in various situations of decision making.

(6)

Economic theories

Most of the explanatory approaches to migration stem from the field of economics. Economic theory considers income differentials and employment opportunities as the central economic determinants of migratory movements. According to the "income differential" hypothesis (Greenwood, 1975, 418f .) the production factor labour migrates to where the factor price is the highest. The "job vacancies" hypothesis (Greenwood, 1975, 418f.) understands labour migration mainly as a reaction to the availability of jobs.

The approaches of economics assume in general individuals who act rationally and who decide on their migration according to the principle of maximizing utility. Frequently, in these models it is asumed that the migrants have complete information. In general it is also assumed that a person planning migration makes a cost-benefit- analysis and that he decides to migrate if the benefit calculated in monetary units exceeds the cost of migration

(Hoffmann-Nowotny, 1970).

Besides the economic aproaches to migration there are several models with a sociological and psychological theoretical emphasis. The majority of the migration models with a sociological approach are based on extensions of sociological theories (Hoffmann-Nowotny, 1970; Saunders, 1976) . © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(7)

Some of the migration models coming from sociology stress the psychological and social-psychological aspects of migration (Albrecht, 1972, 20, 143ff.). Other models start from an explanation which is based on the psychological decision theory, where the behaviour of the individual migrant is considered (Beshers, 1967; Grubel and Scott, 1967) .

As can be understood from the sociological literature, the sociological and psychological approach is considered to be of great importance in explaining migration. Many of those approaches start from very plausible explanations. Unfortunately, most of the relevant models cannot be tested empirically, since some of the variables cannot be quantified (Grubel and Scott, 1967) or since the measurement would be extremely difficult (Beshers, 1967) or even impossible to establish (Hoffmann-Nowotny, 1970; Albrecht, 1972). Furthermore, some of these approaches, for example the one based on decision theory, are not able to explain migration flows.

Micro- and macroanalytical approach

The various migration models can also be differentiated according to their micro- or macroanalytical approach, that is an explanation of migration as a form of individual behaviour or as a function of the particular economic or

(8)

social system (Vanberg, 1971). Most of the migration models are macroanalytical and try to explain migration flows on the basis of aggregate data. In the model which is presented in this paper, migration flows will be analysed. The analysis of migration flows implies a macroanalytical approach. But an analysis with aggregate data cannot verify whether the variables contained in the model actually influenced the migration (Langenheder, 1968). The correlations which can be proved might as well be based on pseudo correlations (Birg, 1979, 99f.).

In a study on problems of regional mobility, Mackensen/ Vanberg/Kraemer state that in many of the macroanalytical models the individual causes are neglected or not sufficiently considered. That is how they explain the failure of numerous migration models (Mackensen/Vanberg/ Kraemer, 1975, 8 ). For a causal explanation of migration the analysis of motives of migrations on a microanalytical level is a necessary condition.

As stated in the previous paragraph, the psychological approaches which examine the migration behaviour of the individual cannot explain migration flows. If the factors are to be determined that actually influence the size and direction of the migration flows there remains the possibility of introducing results of investigations on motives of migration into the macroanalytical model. In

(9)

the sociological literature on migration research there is general agreement that the explanation of migrations requires an interview of the migrating person in order to determine the causal factors of his migration decision

(Albrecht, 1972; Szell, 1972; Vanberg, 1971).

In the following model the attempt is therefore made to integrate microanalytical factors of explanation into the macroanalytical model. The determining factors which are theoretically explained in the function will be empirically rooted by including the results of different interviews on migration motives. The use of macroanalytically based and operationalised variables is a necessary condition for the hypothesis that there is a causal relationship among the relations specified in the model.

Migration models

The model which is most discussed in the literature on migration research is the gravity model. The gravity hypothesis often serves as the starting point of a model and is an essential part of most of the migration models. The approach of the gravity theory is based on the assumption that the migration movements between two regions depend on the population size of the region of origin and the region of destination as well as of the distance between the centres of population. The gravity approach in

(10)

its simple form is purely descriptive. The gravity theory which is based on the "laws of migration" developed by Ravenstein (1889) has been developed further in numerous versions. One of the main models is the model by Zipf (1946). Also Stouffer and Somermeijer start from a gravity approach. Stouffer adds in his model the factor of "intervening opportunities" (1940) and Somermeijer uses the distance as social distance as well as additional factors of atttraction (Termote, 1972). In the last mentioned models there is a clear transition from a descriptive to an explanatory model.

Most of the migration models were developed by an expansion of the gravity model by including additional variables into the model. In some of those models the gravity approach is theoretically explained (Birg, 1982). In their migration model Lowry and Rogers link the gravity approach with the economic "push-pull" hypothesis (Rogers, 1968) and Birg integrates the gravity theoretical approach into a behavioural theoretical approach (1979, 99f.). The push- pull model is a frequently used model that explains migration with the interplay of repulsive factors in the region of origin and attracting factors in the region of destination. The application of behavioural theory to migration is based on the assumption that the migrants will choose their region of destination on the basis of a comparison of the characteristics of the region of origin

(11)

and the potential region of destination (Birg, 1979, 98ff .) .

2. Hypotheses on labour migration and specification of the model

The following migration model is formulated as a macroanalytical function since migration flows will be investigated. But since the causal factors of migration are to be determined, a microanalytical explanatory approach is used when selecting the different variables, that is the operationalization of the migration determinants is done with the results of inquiries on the motives of migration. The model is constructed as a cross- section regression model. It includes the socioeconomic factors which are supposed to determine the migration between the various member states of the European Community. These factors determine essentially the direction of migration.

It is an empirical fact that the volume of migration flows fluctuates considerably according to the different periods of observation. It can be shown that these fluctuations depend to a very high degree on macroeconomic factors.

The labour migration can therefore be defined as

(12)

(1 ) a function of the different socioeconomic factors in the country of origin and in the country of destination, which are considered by the migrant and

(2 ) a function of cyclical economic factors, determining the demand of foreign labour.

As will be shown later these two groups of factors influencing migration require a separate analysis with cross-section and time series data.

The factors which are relevant for the individual migration decision have been determined on the basis of general theoretical conceptions about the migration behaviour, on empirical results of migration research and by the use of inquiries on the motives of migration.

As a result, the following factors influencing migration decisions can be deduced:

- income possibilities - employment possibilities

- improvement of professional social status - information possibilities

- factors of distance

- willingness to migrate depending on personal factors.

The economic factors are determined on the basis of economic theories and empirical observations. They are to explain the part of the volume of migration that is

(13)

dependent on the economic growth and on the situation of the labour market.

The factors of migration defined above will be operationalized with different variables. For this purpose the results of the inquiries on the motives of migration will be used.

These inquiries on the motives of migration should allow a choice of variables for the model which come very close to reality and should therefore enable more reliable statements on the relations between migrations and influencing variables. The motives indicated by the migrants themselves will also allow for a greater precision of the variables, since they inform about the kind of subjective valuation of the various factors undertaken by the migrant himself. The influence of the variables established empirically in this way will then be quantified by multiple regression analysis.

Since in the framework of this study interviewing the migrants was not feasible, different results from former inquiries on the motives of migration of European workers were used such as collected from various interviews of workers from Greece, Spain and the Federal Republic of Germany who migrated into one of the countries of the European Community in order to take up employment. They also come from a representative inquiry of the population

(14)

of all member states made on behalf of the Commission of the European Community.

A precondition for the use of the migration motives named in the inquiries is that they can be operationalized and quantified. The factors of influence defined on the microanalytical level must also be suited for a representation by the available data of official statistics. The explanatory power of the model therefore depends on the quality of the methods and the results of inquiry as well as on the available data of the official statistics.

The operationalization of the various factors of influence results in the following function:

(2.1) M . . = f(W._, U., U., AG., L., L., F W .., GNP.,

l ] D ± l 1 i 1 1 i l i

G N P ., D .., L G ., V ., V .)

1 il l i ]

i = country of origin j = country of destination

M^_. = migration flow between the countries of the European Community

W_._^ = Wj/W^ = averaqe wage in industry per hour U^, U_. = unemployment rate

AG^ = employment in agriculture

(15)

L^, Lj = labour force

FW^_. = foreign workers from i employed in j GNP^, GNPj = growth rate of the gross national product

D^_. = distance between i and j LG^ = labour force growth rate V ± , Vj = job vacancies

With the assumptions on the endogenous variable and the exogenous variables defined above, the migration model can now be elaborated.

The labour migration flows within one year between the different member states of the European Community are defined as the endogenous or the dependent variable M ^ . The exogenous or independent variables are all the factors defined to influence labour migration. The relations between the dependent variable and the independent variables will be tested by multiple regression analysis. The function is defined as a multiplicative relation of the variables. Since it has been assumed in the explanatory model that the migrants compare the different socioeconomic characteristic of the potential countries of destination, the number of comparisons corresponds to the product of the exogenous variables included in the equation. When the error term is added, the model is defined as a stochastic model, e contains the random, non systematic influence on the endogenous variable.

(16)

The migration function has the following form with the variables defined above:

(2 ,2 ) M. . 1 ] FW. . 11 b 7 G N P , e, . i]

In order to be able to estimate the parameters of the model in a multiple linear cross-section equation, the nonlinear function is transformed into a linear function by taking the logarithms of the variables. Due to the linear transformation of the equation the method of ordinary least squares can now be applied.

3. Empirical results

The migration model is estimated for

(17)

(1) the member states of the European Community with data of 1969 (equation 3.1.1.) and

(2) the member states of the EC with data of 1980 (equation 3.1.2).

In the following regression function only those variables are included which attained a level of significance of at least 95 per cent in the equation estimated with the data of 1980. (3.1) In M. .(t) = In b„ + b. In W .. (t) + b„ In L. (t) 1 3 0 1 3 1 2 1 + b, In L.(t) + b. In F W . .(t) 3 3 4 1 3 + b c In D . .(t) + b , In G N P .(t) 5 1 3 6 3 + b_ In U.(t) + b„ In U.(t) 7 1 ° j + In (t)

Hypothesis on the parameters:

b0' b 1' b 2 ' b 3 ' b 4 ' b6 ' b 7 0; b 5' b 8 < 0

The hypothesis on the parameters is as follows:

All parameters are greater than 0 except for the parameters b and b which are smaller than 0 .

5 8 © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(18)

The estimates of the two equations are shown in table 1 and 2

.

The results of equation (3.1.1) show that the variables unemployment rate in the country of origin and in the country of destination have no statistically significant parameter estimates or the wrong sign. In equation (3.1.2) the parameters are significantly different from zero. Both equations are now estimated without the variable unemployment rate.

There are several examples of empirical migration research where the variable unemployment rate, contrary to the theoretical assumptions, did not contribute to the explanation of labour migration. In various interregional migration models by Birg as well as in the modified Lowry- model and in the Lowry-Rogers model the parameters of the variable unemployment rate were highly significant but they did not have the expected sign (Birg, 1979, 100f.; 1975, 72f.; Rogers, 1969). These results could not be explained by the authors. The results seem to contradict theoretical assumptions as well as empirical observations.

It has been shown by many examples of interregional and international migration flows that there is a close relation between the volume of migration and the economic

(19)

Table 1

Migration between EC member states - 1969 estimates of equation (3.1.1)

exogenous expected regression coefficients variables sign test 1 test 2 test 3

in bQ + 20.628 (10.209) -2.386* (1.474) -1 .915** (0.928) w.. Di + 1.890 1.799 1 .677 (0.620) (0.675) (0.598) L . i + 0.397 0.442 0.419 (0.109) (0.117) (0.105) L . + 0.349 0. 303 0.319 3 (0.117) (0.126) (0 .1 0 2 ) FW. . 13 + 0.674 0.555 0.531 (0.114) (0 .1 1 1 ) (0 .1 0 1 ) Dii _ _-0.408 _-0.342 _-0.371* (0.156) (0.167) (0.154) GNP . + -3.051 _ _ J (1 .341) * " u. 1 + -0.203** -0.117** (0.166) (0.176) “ U . - _-0.335* _{0 .0 2 2 **} -D _(n _{( n}_ififn R2 0.924 0.910 0.916 degrees of freedom 21 22 24 level of significance C 5 < 5 < 1 * > 10 * > 10 * > 2 in per cent * * > 20 * * > 50 **>■ 5

Standard deviation in paranthesis

(20)

Table 2

Migration between EC member states - 1980 estimates of equation (3.1.2) exogenous variables exDected sign regression co test 1 efficients test 2 In b + _{- 2 2 .579} -27.117 (5 .649) (6.144) w . . 31 + 0 .917 1 .080 (0 .333) (0.370) L± + _{0 .532} 0.570 (0 .092) (0.085) L . + _{0 .770} 0.449 D (0 134) (0.075) FW. . 13 + 0 308 0.332 (0 065) (0.072) D. . 13 - _{- 0 .346} _-0.277* (0 .119) (0.131) GNP . + 1 .964 2.607 3 (0 .600) (0.630) u. + 0 .286* . >. _ (0 .136) u . _ _{- 0 .819} _ 3 (0 .302) R2 0 .940 0.925 degrees of freedom 38 40 level of significance in per cent * < 1 < 5 < 1 * < 5

Standard deviation in paranthesis

(21)

situation measured by factors such as the growth rate of the gross national product and the unemployment rate.

The differing behaviour of the variable unemployment rate in the two equations and the implausible results of the migration models mentioned above, show clearly that the hypothesis on the influence of the unemployment rate, on which the models are based, cannot be empirically confirmed in most of the cases.

In the following the attempt is made to find an explanation for the implausible results.

If the expected influence of the variable unemployment rate on the migration flows is analysed in a detailed way, it can be seen that the assumptions on the effect of the variable unemployment rate do not hold for all countries of the European Community in the same way. Furthermore, it is assumed here, that the cross-section analysis is not an adequate method to show the relations existing between migrations and the unemployment rate.

In case of dependence of labour migration on economic factors of influence there are substantial fluctuations of the migration flows during the course of time. If the parameters of the migration determinants are estimated in a cross-section regression, unstable parameter estimates will be the result. As already mentioned above, the influence of

(22)

economic factors cannot be determined in a cross-section analysis. The equations (3.1.1) and (3.1.2) will therefore be estimated without the variable unemployment rate. However, the influence of the variable which has been excluded from the equation is still effective. For this reason the values of the parameters of the other exogenous variables in the equation will be biased. Therefore, the influence of the variable unemployment rate as well as of other economic factors will be determined exogenously and will be eliminated afterwards from the cross-section equation.

For this purpose the variance of the migration volume, depending on economic factors, will be determined in a regression analysis with time series data. The endogenous variable of the cross-section function will then be reduced by this influence and the cross-section equation will be estimated afterwards. The cross-section model which has been adjusted by the influence of economic factors should enable more stable parameter estimates. It should therefore reflect more exactly the influence of the behavioural variables and enable more reliable predictions.

According to the preceeding considerations the labour migration will be analysed in two separate steps:

(23)

(1) First the relations between the fluctuation of the migration flows and the variation of the economic variables will be determined in a time series analysis.

(2) The economic component, computed with the time series analysis is then eliminated from the cross-section equation.

The relations between the migrations and the other behavioural factors of influence can now be estimated with the adjusted cross-section regression.

The results of the time series analysis

In the following time series model the fluctuations of the migration flows which can be attributed to the influence of economic factors will be explained. The objective is to determine the economic factors of influence as exactly as possible, i.e. to provide a degree of explanation of the variance of the endogenous variable as high as possible. The more accurately the influence of the economic variables on the volume of migrations can be determined, the more accurately the parameters of the behavioural variables can be quantified in the adjusted cross-section model.

The dependence of the migration volume on the economic labour market situation was tested with the variables unemployment rate, growth rate of the gross national product, growth rate of employment and job openings. The last two variables, however, were not statistically significant. © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(24)

If the time series are computed separately for Italy and the other member states of the European Community without Italy, it can be shown that labour migration between the member states without Italy is completely independent of changing labour market conditions and of economic growth.

The variance of the migration streams from Italy into the other member states is relatively well explained by the variables unemployment rate and gross national product

(coefficient of determination 0.91).

Also for the labour migration from Greece and Spain into the countries of the European Community a relation between the fluctuation of the migration flows and the variation of the unemployment rate can be shown. The regression functions for Greece and Spain were estimated for the period from 1960 to 1973 until the time of the immigration stop in the countries of the European Community. The coefficient of determination for Greece is 0.56 and for Spain 0.57.

When the migration flows from Italy into the other countries of the European Community are analysed according to their distribution on each member state, it can be seen that between 80 and 90 per cent of the Italian migrants go to the Federal Republic of Germany. The number of Italian migrants who go to the other countries is therefore rather

(25)

small. Therefore, only the relations between the migration flows from Italy into the Federal Republic and the economic variables of these countries will be examined in a time series model.

The migration function is tested with the variables unemployment rate, growth rate of the gross national product in the country of origin and in the country of destination as well as with the variable job openings in the country of destination. The estimates show that there is no statistically significant relation of the variables unemployment rate and gross national product of the country of origin and for the variable job openings with the dependent variable. Therefore, the migration function is estimated without these variables with the following equation: (3.2) M . .(t) = b0 U.(t) 0 3 GNP_. (t) e . .(t) 13 M. . 13 U .* 3

= migration from Italy into the Federal Republic of Germany

= unemployment rate in the FRG

= growth rate of the gross national product GNP . 3 in the FRG © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(26)

The empirical result is the following: (3.3) In H.,(t) = In 11.352 - 0.700 In U .(t) 1 3 (0.082) (0.049) 3 + 0.214 In G N P .(t) + In e . .(t) (0.054) 3 1 3 R2 = 0.946 DW = 2.190

The estimates show that both variables have the expected sign. The coefficients are significant at a level of 0.0 and less than 0.1 per cent. The coefficient of

O

determination, R , is about 0.95, that means that about 95 per cent of the variance of the migration flow is explained by the variables unemployment rate and growth rate of the gross national product in the country of destination. There is no multicollinearity. The test with the Durbin-Watson statistic shows that there is no autocorrelation of the residuals.

As the time series analysis shows, only the variables unemployment rate and gross national product of the country of destination are statistically significant. That means that the fluctuation of the Italian migration flows depends only on the labour market situation of the Federal Republic of Germany and not of Italy.

(27)

The results of the time series analysis show that the unemployment rate in the country of origin does not act as a push factor as assumed in most of the migration models. This is at least true for the model tested here.

In order to correct the cross-section equation for the economic influence, this influence must be determined with the help of the time series equation. If the variables unemployment rate and gross national product are eliminated from equation (3.3), the following function is obtained:

(3.4) In M Z i_.(t) = In 1 1 .352 + In e _ (t)

where MZ . . is the adjusted migration flow from Italy into the Federal Republic of Germany, In 11.352 is the constant term and is the error term.

The graphic display of the migration functions (3.3) and (3.4) in the following diagram for the period from 1960 to 1981 shows that the fluctuations of the migration flows are considerably reduced after elimination of the economic variables.

It is assumed that for Italian migration into the other member states of the European Community the same economic variables and the same parameter values are true as for Italian migration into the Federal Republic of Germany. The

(28)

Labour migration from Italy into the Federal Republic of Germany 1960 - 1981

3 s

7V 73 74 7a 73 M >3

-- actual migration flews --- adjusted migration flews

(29)

economic influence on Italian migration flows can then be computed for each country and can then be eliminated from the cross-section equation. The adjusted cross-section is then estimated again with the new values. The following tables 3 and 4 show the quantitative changes of the parameter estimates in the migration model of 1969 and of 1980 after elimination of the economic factors of influence.

4. Conclusions

The migration model of 1980 should also enable estimates of future migration flows between the member states of the European Community.

One problem of all socioeconomic models of explanation and prediction which has not yet been solved and which is also relevant in relation with a migration model for the countries of the European Community, is the problem of invariance. Because the relations determined with the migration function depend on human behaviour as well as on macro-economic and political factors, their range of validity is limited.

In the migration model presented here, quasi-invariant relations are assumed, as is usual with models of this kind, that means relations where only the parameter of the

(30)

Table 3

M i g r a t i o n m odel 1969

P a r a m e t e r estimates b e fore and after e l i m i n a t i o n of the e conomic influence

exogenous P a r a meter e s t i mates with estimates without

variables e conomic influence e conomic influence

cans tant term In b Q -1.915 -1 .600 W. . J i b 1 1.677 1.613 L. l b 2 0.419 0.465 L . J b 3 0.3.1 9 0.349 FW. . b 4 0.531 0.480 D. .. b 5 -0.371 -0.459 R 2 0.916 0.912 Table 4 M i g r a t i o n m odel 1980

Pa r a m e t e r estimates b e f o r e and after elimination- of the e conomic influence

1

exogenous Pa r a m e t e r e s t imates with estimates wit h o u t

variables eco n o m i c influence ec o n o m i c influence

cons tant term in b Q -27.117 -25.234 W. . J i b . 1 .080 1.130 L . l b 2 0.570 0.455 L . J b 3 0.449 0.382 FW. . b 4 0.332 0.474 GNP . J b 5 2.607 2.233 R 2 0.925 0.935 © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(31)

variables change, but no additional or other variables have to be considered in the migration function. According to the assumptions the estimated regression coefficients can only be constant for a limited period.

In the analysis of the migration flows between the member states of the European Community a difference was made between behavioural and macro-economic migration determinants. According to this assumption time dependent variations of the parameters of a migration function can be caused by

(1 ) changing migration behaviour due to changing attitudes or expectations of the migrants,

(2 ) changing economic conditions which result in changing demand for labour.

The separate analysis of the migration flows in a time series and a cross-section regression should enable a more precise determination of the different factors of influence.

Finally, a test is made on the effect which the seperate migration analysis with behavioural and economic factors and the elimination of the economic component from the cross-section equation had on the parameter estimates. The exogenous determination of the economic influence and its elimination from the cross-section equation should improve

(32)

the stability of the parameter estimates since the influence of economic fluctuations on the volume of migration is not reflected any more in the new estimates of the cross-section equation.

A comparison of the parameter estimates of the two migration models of 1969 and 1980 shows that after the elimination of the economic influence, the difference between the estimates has been considerably reduced (see table 5 ).

The elimination of the economic component from the cross- section equation had the effect that the estimated regression coefficients of all variables in the migration models of 1969 and 1980 were approximated to each other. It can be assumed that this means also an approximation to the true value of the coefficients. If the confidence interval of the estimated regression coefficients is computed for a probability of 95 per cent, it can be shown that the interval of confidence of the coefficients in the adjusted migration function has become smaller for all estimates. This is a further indication of an improvement of the parameter estimates.

The behavioural variables included in the adjusted cross- section function will change relatively slowly if it is assumed that individual attitudes and social behavioural

(33)

T a b l e 5 D i f f e r e n c e b e t w e e n t h e p a r a m e t e r e s t i m a t e s o f t h e m i g r a t i o n m o d e l s o f 1 9 6 9 a n d 1 9 8 0 ( 1 ) b e f o r e a n d ( 2 ) a f t e r e l i m i n a t i o n o f t h e e c o n o m i c i n f l u e n c e a j u 3 a> CO o CO X u co — CO O 4-1 Q) ai o O O i i 3 a C*-4 • • • • o 3 CM o o o o X a) •r-t ■u 3 • H i—1 *o £ <4-1 / - s e CN V) *H c—' (U H o o m CN CO 4-i a d> CO CO m 00 CO CO • M T3 on — CO | CN B B O — • • 4 • • • H O *-* c 0 — o o o CN QJ U QJ r—1 ON CO m ON o ON 3 X •— X -3- 00 m • 3 on cO -cf c o *3- 1 O • • • • 0 o o o o CU o a) 3 u 0) — o ON — x e U ON m CO ON ON 4-1 QJ <u c n o 1 • H 3 CM • • « • 5 *-1 M-4 o o o o o M-4 • H CO 3 X J — 4-4 w to CJ B *H r - 1o o o ON — r - . • H 6 QJco 0 0 ON CO O 4-1 O T3 CT\ o m c o CN X co 3 O •— • • * • • 0) o 0 — o o o o CN a 1 r—1 ON ON ON _ CU CO — — r - ' T3 ON CO co . m . CO 1 O • • • , 0 o o o o 1 a; 4-1 a) 0 —• CN co m X cO X X X X X X u cO a , CO CO 3 <u O r—1 C X 0) 3 0 0 *H O 3 • H •?—> Pd X to •H •r-j • H $ 5 <1) > 5 X 1*4 o o © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(34)

patterns will also change slowly. The relations expressed in the equation will therefore be relatively stable. Fluctuations of the migration flows, due to variations of the labour market situation and of the economic growth which vary much temporally, were explained separately with economic variables in the time series analysis. The parameters proved to be rather stable over different periods. The results allow the conclusion that the separate analysis of migrations with behavioural and with economic determinants as well as the elimination of the economic influence from the cross-section function produced more precise parameter estimates which also vary less over time.

(35)

REFERENCES

Albrecht, G.: Soziologle der geographischen Mobilitaet. Stuttgart 1972.

Beshers, J.M.: Population Processes in Social Systems. New York, London 1967.

Birg, H.; Analyse und Prognose der Bevoelkerungsentwick- lung in der Bundesrepublik Deutschland und in ihren Regionen bis zum Jahr 1990. DIW, Beitraege zur Strukturforschung, Heft 35. Berlin 1975.

Birg, H.: Zur Interdependenz der Bevoelkerungs- und

Arbeitsplatzentwicklung. Sonderheft des DIW Nr. 131. Berlin 1979.

Birg, H.: On the Interaction of Job Creation, Migration and Natural Population Increase within the Framework of a Dynamic Demo-economic Model. In: Environment and Planning, Vol. 14, No. 9, September 1982, p. 1141-1153.

Greenwood, M.J.: Research on Internal Migration in the United States: A Survey. In: The Journal of Economic Literature, Vol. 13, 1975, p. 397-433. © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(36)

Grubel, H.G. and A.D. Scott: Determinants of Migration: The Highly Skilled. In: International Migration, . V, 1967, p. 127-139.

Hoffmann-Nowotny, H.-J.: Migration. Ein Beitrag zu einer soziologischen Erklaerung. Stuttgart 1970.

Langenheder, W . : Ansatz zu einer allgemeinen Verhaltens-theorie in den Sozialwissenschaften. Dargestellt und ueberprueft an Ergebnissen empirischer Untersuchun- gen ueber Ursachen von Wanderungen. Koeln und Opladen 1968.

Mackensen, R„, M. Vanberg, K. Kraemer: Probleme regiona- ler Mobilitaet. Goettingen 1975.

Ravenstein, E.G.: The Laws of Migration. In: Journal of the Royal Statistical Society, Vol. LII, June 1889.

Rogers, A.: Matrix Analysis of Interregional Population Growth and Distribution. Berkeley and Los Angeles

1968 .

Saunders, H.W.: Human Migration and Social Equilibrium. In: J.J. Spengler and O.D. Duncan (ed.), Population Theory and Policy. Glencoe 1976,

(37)

Stouffer, S.A.: Intervening Opportunities: A Theory Relating Mobility and Distance. In: American Socio logical Review, 5, 1940.

Szell, G.: Einleitung: Regionale Mobilitaet als Forschungs gegenstand. In: Regionale Mobilitaet, hrsg. von G Szell. Muenchen 1972.

Termote, M. : Wanderungsmodell. In: Regionale Mobilitaet, hrsg. von G. Szell. Muenchen 1972

Vanberg, M . : Kritische Analyse der Wanderungsforschung in der B R D . Berlin 1971.

Zipf, G.K.: The P ^P ^/D Hypothesis on the Inter-city Movement of Persons. In: American Sociological Review, 11, 1946. © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research Repository.

(38)

(39)

No. 1: Jacques PELKMANS The European Community and the Newly Industrialized Countries

No. 3: Aldo RUSTICHINI Seasonality in Eurodollar Interest

Rates

No. 9: Manfred E. STREIT Information Processing in Futures Markets. An Essay on the Adequacy of an Abstraction

No. 10: Kumaraswamy VELUPILLAI When Workers Save and Invest: Some Kaldorian Dynamics

No. 11: Kumaraswamy VELUPILLAI A Neo-Cambridge Model of Income Distribution and Unemployment

No. 12: Guglielmo CHIODI Kumaraswamy VELUPILLAI

On Lindahl's Theory of Distribution

No. 22: Don PATINKIN Paul A. Samuelson on Monetary Theory

No. 23: Marcello DE CECCO Inflation and Structural Change in the Euro-Dollar Market

No. 24: Marcello DE CECCO The Vicious/Virtuous Circle Debate in the '20s and the '70s

No. 25: Manfred E. STREIT Modelling, Managing and Monitoring Futures Trading: Frontiers of Analytical Inquiry

No. 26: Domenico Mario NUTI Economic Crisis in Eastern Europe: Prospects and Repercussions

No. 34: Jean-Paul FITOUSSI Modern Macroeconomic Theory: An Overview

No. 35: Richard M. GOODWIN Kumaraswamy VELUPILLAI

Economic Systems and their Regulation

No. 46: Alessandra Venturini Is the Bargaining Theory Still an Effective Framework of Analysis for Strike Patterns in Europe?

No. 47: Richard M. GOODWIN Schumpeter: The Man I Knew

No. 48: Jean-Paul FITOUSSI Daniel SZPIRO

Politique de l'Emploi et Réduction de la Durée du Travail

No. 56: Bere RUSTEM

Kumaraswamy VELUPILLAI

Preferences in Policy Optimization and Optimal Economie Policy

(40)

No. 60: Jean-Paul FITOUSSI Adjusting to Competitive Depression. The Case of the Reduction in Working Time

No. 64: Marcello DE CECCO Italian Monetary Policy in the 1980s

No. 65: Gianpaolo ROSSINI Intra-Industry Trade in Two Areas: Some Aspects of Trade Within and Outside a Custom Union

No. 66: Wolfgang GEBAUER Euromarkets and Monetary Control:

The Deutschmark Case

No. 67: Gerd WEINRICH On the Theory of Effective Demand

Under Stochastic Rationing

No. 68: Saul ESTRIN Derek C. JONES

The Effects of Worker Participation upon Productivity in French

Producer Cooperatives

No. 69: Bere RUSTEM

Kumaraswamy VELUPILLAI

On the Formalization of Political Preferences: A Contribution to the Frischian Scheme

No. 72: Wolfgang GEBAUER Inflation and Interest: the Fisher Theorem Revisited

No. 75: Sheila A. CHAPMAN Eastern Hard Currency Debt 1970- 1983. An Overview

No. 90: Will BARTLETT Unemployment, Migration and Indus

trialization in Yugoslavia, 1958- 1982

No. 91: Wolfgang GEBAUER Kondratieff's Long Waves

No. 92: Elizabeth DE GHELLINCK Paul A. GEROSKI Alexis JACQUEMIN

Inter-Industry and Inter-Temporal Variations in the Effect of Trade on Industry Performance

84/103: Marcello DE CECCO The International Debt Problem in the Interwar Period

84/105: Derek C. JONES The Economic Performance of Producer

Cooperatives within Command Economies Evidence for the Case of Poland

84/111: Jean-Paul FITOUSSI Kumaraswatny VELUPILLAI

A Non-Linear Model of Fluctuations in Output in a Mixed Economy

84/113: Domenico Mario NUTI Mergers and Disequilibrium in Labour- Managed Economies © The Author(s). European University Institute. Digitised version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research

(41)

84/114: Saul ESTRIN Jan SVEJNAR

Explanations of Earnings in Yugoslavia: the Capital and Labor Schools Compared

84/116: Reinhard JOHN On the Weak Axiom of Revealed Preference

without Demand Continuity Assumptions

84/118: Pierre DEHEZ Monopolistic Equilibrium and Involuntary

Unemployment

84/119: Domenico Mario NOTI Economic and Financial Evaluation of Investment Projects: General Principles and E.C. Procedures

84/120: Marcello DE CECCO Monetary Theory and Roman History

84/121: Marcello DE CECCO International and Transnational

Financial Relations

84/122: Marcello DE CECCO Modes of Financial Development: American Banking Dynamics and World Financial Crises

84/123: Lionello PUNZO Kumaraswamy VELUPILLAI

Multisectoral Models and Joint Production

84/126: John CABLE Employee Participation and Firm Perfor

mance: a Prisoners' Dilemma Framework

84/127: Jesper JESPERSEN Financial Model Building and Financial

Multipliers of the Danish Economy

84/128: Ugo PAGANO Welfare, Productivity and Self-Managemenl

85/155: Francois DUCHENE Beyond the First C.A.P.

85/156: Domenico Mario NOTI Political and Economic Fluctuations in the Socialist System

85/157: Christophe DEISSENBERG On the Determination of Macroeconomic Policies with Robust Outcome

85/161: Domenico Mario NOTI A Critique of Orwell's Oligarchic Collectivism as an Economic System

85/162: Will BARTLETT Optimal Employment and Investment

Policies in Self-Financed Producer Cooperatives

85/169: Jean JASKOLD GABSZEWICZ Asymmetric International Trade Paolo GARELLA © The Author(s). European University Institute. version produced by the EUI Library in 2020. Available Open Access on Cadmus, European University Institute Research