Temporary Labour Migration from Other Yugoslav Republics and Autonomous Provinces to SR Croatia – An Atempt to Use Multiple Regression Analysis

Cilj je ovog članka testiranje značaja odabranih ekonomskih varijabli na opseg privremenih migracija radne snage iz drugih jugoslavenskih republika i pokrajina u gradove SR Hrvatske. Broj nekvalificiranih i polukvalificiranih privremenih radnika-migranata (Mj) je zavisna varijabla. Nezavisne su varijable: Gj — broj nekvalificiranih i polukvalificiranih radnika u građevinskom sektoru određenog centra j; Dj — nivo osobnih dohodaka nekvalificiranih i polukvalificiranih radnika u j; Nj — stopa nezaposlenosti nekvalificiranih i polukvalificiranih radnika u j; Uj — veličina urbane populacije u j. Autorica postavlja hipotezu da su Gj, Dj, Uj statistički značajni i pozitivno korelirani s Mj, dok se za Nj pretpostavlja obrnuto. Multipla regresija je odabrana kao najadekvatnija statistička tehnika za testiranje spomenutih hipoteza. Korišteni su podaci iz Popisa zaposlenih u organizacijama udruženog rada u SR Hrvatskoj. Dobiveni su slijedeći rezultati: vrlo visok koeficijent determinacije (R2 = 0.92) i multiple korelacije (R = 0.95). Međutim, Gj je ustanovljena kao jedina statistički značajna nezavisna varijabla. Uj i Gj imaju vrlo visok Pearsonov koeficijent korelacije (r = 0.99) što upućuje na slučaj multikolinearnosti. Osim toga, vrlo visoke vrijednosti koje su dobijene za Zagreb, najveći grad i najvažniji ekonomski centar SR Hrvatske s jedne strane, te također i vrlo niske vrijednosti koje su dobijene za 27 najmanje razvijenih općina u Hrvatskoj s druge strane, čini se da doprinose visokim R i R°. Tada je izvršena druga multipla regresija bez takvih slučajeva, i ustanovljeni su manji, ali još uvijek značajni R (0.72) i R2 (0.50). Međutim, pokazalo se da je samo Gj statistički značajna i pozitivno korelirana s Mj. Drugim riječima, mjere politike čiji je cilj utjecaj na veličinu građevinskog sektora općina u Hrvatskoj također bi doprinijele promjenama u intenzitetu toka migracija nekvalificiranih i polukvalificiranih radnika iz drugih jugoslavenskih republika i autonomnih pokrajina. Nakon analize rezidua regresije, čini se da bi modificirani model gravitacije bio najprimjereniji za testiranje faktora koji utječu na privremene migracije radne snage. Nezavisne bi varijable bile: broj zaposlenih u tradicionalnim ekonomskim sektorima; vrijeme putovanja između mjesta porijekla i odredišta; kontakti s ranijim migrantima (ne-ekonomska varijabla).The objective of this paper was to test the significance of selected economic variables influencing the volume of temporary labour migration from other Yugoslav republics and autonomous provinces to the municipalities of SR Croatia. The number of unskilled and semi-skilled temporary migrant workers (Mj) is the dependent variable. The independent variables were: Gj, the number of unskilled and semi-skilled workers in the construction sector of the municipality of work j; Dj, the wage levels for unskilled, semi-skilled workers in j; Nj, the unemployment rate for unskilled, semi-skilled workers in j; Uj, the size of the urban population of j. It was hypothesized that Gj, Dj, Uj are statistically significant and positively correlated with Mj, while the opposite was to hold for Nj. Multiple regression was chosen as the most adequate statistical technique to test the above hypotheses. Data were used from the Census of the Employed in the Organizations of Associated Labour (Enterprises) of SR Croatia. A very high coefficient of determination (R2 = 0.92) and multiple correlation coefficient (R = 0.95) were obtained. However, the only statistically significant independent variable was found out to be Gj. Uj and Gj had a very high Pearson\u27s correlation coefficient (r = 0.99), which suggests a case of multicolinearity. Furthermore, the extremely high values attributed to Zagreb, the largest city and the most important economic center of SR Croatia; on the one hand, as well as the extremely low values attributed to the 27 least developed Croatian municipalities, on the other hand, seemed to have contributed to the high R and R2. Another multiple regression was then performed without such cases and a smaller but still significant R (0.72) and R2 (0.50) resulted. However, only Gj was shown to be statistically significant and positively correlated with Mj. In other words, a policy measure the goal of which is to affect the affect the size of the publicly-owned construction sector of the Croatian municipalities would also contribute to a change in the intensity of the flow of unskilled and semiskilled workers from other Yugoslav republics and autonomous provinces. After analyzing the regression residuals, it seems that a modified gravitation model may be the most adequate one to test the factors which influence temporary labour migration. The independent variables would be: number of employed in the traditional sector of the economy; travel time between the area of origin and destination; contacts with early migrants (a non-economic variable)