Regional distribution of immigrants in Hungary 107 Hungarian Geographical Bulletin 59 (2) (2010) pp. 107–130. Regional distribution of immigrants in Hungary1 Géza Tóth2 and Áron Kincses3 Abstract The purpose of this study is to analyse the territorial characteristics of foreign nationals those having migrated to Hungary. It is aimed to explore the triggers of choice of usual plan of residence and the diff erences by citizenship. Interrelationship between the proportion of migrants having resided in a particular area and the access by public road to it will be investigated with the help of path analysis. Firstly comparative studies on the geographi- cal distribution of immigrants and that of former international migrants already sett led in the country are to be conducted. Secondly the spatial disparities in the distribution of immigrants at micro-regional level will be identifi ed by the potential method. Keywords: immigrants, regional distribution, path analysis, potential method Introduction Since the regime change (1990) Hungary has had an international migration surplus, i.e. the number of foreigners having migrated to Hungary exceeded that of Hungarian citizens who left the country. Foreigners have become an ever increasing demographic factor in Hungary as natural population change has had a negative trend leading to an approximately 30–40 thousand popu- lation loss annually along with a concomitant positive migration balance of 10–20 thousand. On January 1 2008, 174,697 foreign citizens were on an extend- ed stay in the country making up 1.74% of the resident population. It means that out of one hundred people almost two are foreigners. During the seven years following the turn of the millennium the ratio of foreigners increased by 61% on the national level (Table 1). 1 The publication of this study was supported by the HAS Bolyai scholarship. 2 PhD, leader-consultant, Hungarian Central Statistical Offi ce, H-1024 Budapest, Keleti Károly út 5–7.; research fellow, Károly Róbert College H-3200, Gyöngyös Mátrai u. 36. E-mail: geza.toth@ksh.hu 3 PhD student, leader-consultant, Hungarian Central Statistical Offi ce, H-1024 Budapest, Keleti Károly út 5–7. E-mail: aron.kincses@ksh.hu 108 Over seven years, foreigners increased in number by one-and-a-half times. Out of them, those who arrived from the countries of the Carpathian Basin (Austria, Slovakia, Ukraine, Romania, Serbia, Croatia and Slovenia) ac- count for the majority, surpassing by 5 per cent those, who arrived from the rest of the world. Most of the former came from Romania, Ukraine and Serbia. Beside these groups a signifi cant number of citizens of EU15 countries (mainly Germans and Austrians) live in Hungary. In the following the att ention will be focused on the citizens of the neighbouring countries (Table 2). Territorial distribution of foreigners In 2001, 17% of Hungary’s resident population lived in Budapest, 20% in coun- ty rank towns, 27% in other towns and 36% in villages. By 2008, in terms of proportion, those living in other towns increased up to 31% while the popu- lation of villages dropped down to 32% while there were no changes in the fi rst two categories. As far as foreign residents were regarded, Budapest was already strongly over-represented (35%) in 2001, which coincides with international trends, as capital cities are primary target destinations for migrants. This im- pact is showed in a more expressed way by those who arrived from outside the European continent (77% of Asians live in the capital city). Working-age peo- ple account for an even larger proportion when taking into consideration all towns, while in the villages the pensioners account for a bulk of migrants. Over the analysed seven years, on the one hand, the pull force of Budapest strongly increased among foreigners (43%), concomitant with a decrease in the proportion of county rank towns, along with constant rates of smaller towns and villages. Table 1. Summary data of Hungarian population Year (January, 1) Resident population Foreign citizens staying in Hungary Number As a percentage of total population 2001 2002 2003 2004 2005 2006 2007 2008 10,200,298 10,174,853 10,142,362 10,116,742 10,097,549 10,076,581 10,066,158 10,045,401 110,028 116,429 115,888 130,109 142,153 154,430 166,030 174,697 1.08 1.14 1.14 1.29 1.41 1.53 1.65 1.74 Source: HCSO 109 Reasons behind the territorial distribution of foreigners According to the neoclassical theory (Hamilton, B. et al. 1984; Venables, A.J. 1998) fl ows of humans on the macro level are determined by the push and pull factors of capital and labour. On micro level it is the regional diff erences in incomes that generate a motivation to move (Hatton, T.J. and Williamson, J.G. 2005). The population tends to be increasingly mobile in areas where considerable disparities of incomes can be identifi ed. Other motivating factors are the individual skills and abilities of the migrants as well as intentions to improve life circumstances (Borjas, G.J. 1996; Williamson, J.G. 2006). In the opinion of the authors, in the distribution patt ern of foreigners, beside these main economic motivation factors identifi ed by the literature an important role is played by the att raction of the metropolitan area of the capital city (Rédei, M. 2007; Papademetriou, D.G. 2006) as a signifi cant focus of migration and also that of border areas due to the neighbourhood of source countries. In consequence of a great number of citizens from the neighbour coun- tries geographic location has a great signifi cance in the case of migration aff ect- Table 2. Foreign citizens staying in Hungary by citizenship (January 1) Country 2001 2002 2003 2004 2005 2006 2007 2008 Austria France Netherlands United Kingdom Germany Italy EU–15 Croatia Poland Russia Romania Serbia Slovakia Slovenia Turkey Ukraine Other European Neighbouring countries Europe Asia America Africa Other and unknown 694 511 324 624 7,493 542 11,723 917 2,279 1,893 41,561 12,664 1,576 82 455 8,947 20,584 66,359 93,197 12,603 2,488 1,233 507 785 601 346 700 7,676 563 12,181 931 2,227 2,048 44,977 11,975 2,213 88 544 9,835 21,088 70,716 97,640 14,401 2,557 1,318 513 750 711 373 872 7,100 545 11,629 800 1,945 1,794 47,281 11,693 1,536 65 469 9,853 21,552 71,913 98,230 13,480 2,434 1,281 463 780 765 415 963 7,393 551 12,143 902 2,196 2,244 55,676 12,367 2,472 81 557 13,096 22,915 85,293 110,915 14,715 2,535 1,455 489 544 330 236 440 6,908 404 9,714 837 2,178 2,642 67,529 13,643 1,225 34 615 13,933 24,493 97,711 122,261 15,121 2,667 1,556 548 1,494 1,316 666 1,451 10,504 777 18,357 778 2,364 2,759 66,183 12,111 3,597 79 756 15,337 24,307 99,579 130,535 18,543 2,989 1,800 563 2,225 1,506 1,096 1,911 15,037 1,020 25,394 813 2,681 2,760 66,951 12,638 4,276 115 886 15,866 25,314 102,769 140,827 19,733 3,075 1,783 612 2,571 1,481 1,201 2,107 14,436 1,207 25,490 852 2,645 2,787 65,836 17,186 4,944 133 1,120 17,289 26,272 108,811 146,145 22,356 3,557 1,913 726 Total 110,028 116,429 115,888 130,109 142,153 154,430 166,030 174,697 Source: HCSO, own edition 110 ing Hungary (Rédei, M. 2007). As it is illustrated by Figure 1 in the surround- ings of Lake Balaton, Budapest and Pest County as well as in micro-regions4 on the Ukrainian, Romanian and Serbian borders, foreigners accounted for a greater proportion than elsewhere in 2001. Similar concentration of foreigners is not typical along the Croatian, Slovenian, Slovakian and Austrian borders. On the one hand this may be explained by the less numerous populations of these groups in Hungary, on the other hand, by smaller “diff erences in poten- tial” compared with those that are observed along the Serbian and Romanian border sections. The above areas (centre and hinterlands) and their surround- ings experienced an increase in the percentage of foreigners within the resident population from 2001 to 2008. In Hungary the following general characteristic were found in the territorial distribution of international migrants. The overwhelming part of foreigners live in Budapest and its surroundings, a smaller proportion of them is a resident of micro-regions near the borders as well as in the surroundings of Lake Balaton. Citizens from the EU15, in addition to Budapest and its ag- glomeration, give preference to sett le in the western part of the country (Illés, S. 2004), mainly in Győr-Moson-Sopron and Somogy counties. Romanian citizens are the most diff used in their geographical distri- bution; they live in large numbers along the Romanian border, in the capital city and in Western Hungary. The Serbs cluster in a wedge determined by the common border and Budapest. The Slovaks are concentrated in Northern Hungary and in the surroundings of Budapest, while for the Ukrainians, in addition to Budapest, those micro-regions are the most att ractive ones which are near to their source country. In short, it might be said that for those for- eign nationals who have come to Hungary from the neighbouring countries Budapest and Pest County are unambiguously att ractive destinations beside those micro-regions which are nearer to that country which corresponds to their citizenship, i.e. mainly close to the Romanian, Ukrainian and Serbian border sections. It is important to note that the foreigners show an interest to sett le down also in places where a human resource injection is needed, like Southern Transdanubia or North East Hungary. 4 The system of micro-regions (formerly att raction zones of micro-regions) covers the whole country. Micro-regions do not cross county borders. Every sett lement belongs to one micro-region, though through their relationship sett lements may be att racted by one or more central sett lements. The present system of micro-regions has contained 174 micro- regions since 25th September 2007 on the basis of Act CVII of 2007. In most cases the professionals of regional analyses use this level in their work. The system of Hungarian micro-regions fi ts the fi rst LAU level in the European regional breakdown. At the local level, two levels of Local Administrative Units (LAU) have been defi ned in the European Union. The upper LAU level (LAU level 1, formerly NUTS level 4) is defi ned for most, but not all of the countries. The second LAU level (formerly NUTS level 5) consists of about 120 000 municipalities or equivalent units in the 27 EU Member States. 111 Fig. 1 Regional distribution of the foreigners by LAU1 level, 2001–2008. Source: HCSO, own edition Proportion of foreigners per 100 residents, January 1, 2001 Proportion of foreigners per 100 residents, January 1, 2008 112 Previously, location theories observed the border regions as tradi- tionally backward areas, fi rst of all because borders hampered international trade fl ows and because they were threatened by possible military invasions (Anderson, J. and O’down, L. 1999). National borders have a negative eff ect on a regional economy, because these artifi cially cut off spatially interrelated regions and increase transaction costs. Diff erent taxes, languages, cultures (though in the case of borders in concern these last two cases are not valid) and business practices hamper the cross border trade – these are a basis for potential political and social instability at border regions – which discourage domestic and foreign producers to sett le down in these regions (Hansen, N. 1977). A change in this unfavourable image, as a result of a greater interna- tional integration (Papademetriou, D.G. 2006) – with the help of eliminating trade barriers and international borders (Van Geenhuizen, M. and Ratti, R. 2001) – represents a new perspective of growth in border regions (Contessi, S. 2001; Traistaru, I. et al. 2002) in the fi rst place because of geographical accessibility to large potential markets as occurred in 1993 in Europe with the establishment of a single market and aft er the establishment of NAFTA (Krugman, P. and Venables, A.J. 1996; Krugman, P. 1998). Applying path analysis to examine the territorial distribution of foreign population groups In the present analysis of the most populous groups of foreigners that live in Hungary (from Romania, Serbia, Slovakia, EU15 and Ukraine) the causes of their territorial distribution will be analysed. As it was seen, the literature put an emphasis of living standards and diff erences in payments as pull factors but the location of sett lements is prioritised, too, and this geographic factor will be examined in a somewhat more detail. With the help of path analysis, between 2001 and 2008, the average proportion of foreigners by micro-regions is to be examined by factors. In this analysis, in the fi rst place it was aimed to identify correlation between public road access to micro-regions and the proportion of immigrants. Zero order linear correlations of independent and dependent variables are broken down into two parts in the path models. One part is the eff ect that our independent variables directly have on a dependent variable; the other part is the eff ect that is produced by independent variables through other in- termediate variables (Duncan, O.D. 1966; Alwin, D.F. and Hauser R.M. 1975; Székhelyi, M. and Barna, I. 2008). Path analysis is a series of estimations of ordinary least squares built upon each other. In step 1 it is examined what an impact the primary variables have upon the indicators of a secondary group; there are as many regressions 113 as secondary variables. In step 2 it is examined how the primary and secondary variables jointly impact the tertiary ones. At last a regression is found, where all variables are put together. The impact of signifi cant indicators is analysed jointly with the explored paths (Németh, N. 2008). The following indicators were involved in our analysis: Accessibility For micro-regional centres, travel distance on public road from the “corresponding” border crossing in minutes (BORDER). For micro-regional centres, travel distance on public road from Budapest in minutes (BUDAPEST). Economic situation Personal cars per thousand residents as an average of 2000–2007 (CAR). Shops and stores that sell food per thousand residents as an average of 2000–2007 (SHOPS). Earning per taxpayer as an average of 2000–2007 (EARNING) Active enterprises per thousand residents as an average of 2000–2006 (ENTERPRISES). Social situation Natural increase/decrease per thousand residents, 2000–2007 (DECREASE). Migration balance per thousand residents, 2000–2007 (MIGRATION) Indicted cases per thousand residents as an average of 2001–2007 (CRIME). Ratio of those with secondary and higher qualifi cations to the resident population, %, 2001 (QUALIFICATION). Territorial distribution of migrants Ratio of immigrants from a given country to the resident population, 2000 (RATIO). – – – – – – – – – – – 114 These values are regarded as independent variables that explain the proportion of foreigners with a given citizenship, which constitute the de- pendent variable. In this way concerning the territorial distribution of migrants four groups of variables were put together as a total. Over our examinations, there were more indicators in the individual groups of variables, which were ex- cluded from our system as a result of preliminary calculations. In relation to the single indicator groups the following hypotheses were devised. Accessibility: the nearer is the given micro-region to Budapest as well as to the corresponding border section, the higher is the proportion of foreigners. Economic situation: the more signifi cant is the economic weight of a micro-region, the higher is the proportion of foreigners. Social situation: the more favourable is the demographic situation and the higher is the educational att ainment of the population as well as the lower is the rate of criminal off ences, the higher is the proportion of foreigners in the micro-region. Territorial distribution of former immigrants: the higher was the propor- tion of migrants in previous times, the higher it is going to be in the analysed period, too. According to our presumptions the primary explanatory factors (acces- sibility) infl uence diff erences in secondary factors (economic situation, social situation), which were analysed in detail in an article on this topic (Hardi, T. 2008), which in turn exert an impact on tertiary factors (territorial distribution of migrants in previous times). Another assumption is that the primary and secondary explanatory factors have an infl uence on the proportion of migrants not only in an indirect but in an independent way. (The arrows in Figure 2 are to illustrate this relationship in causality). Fig. 2. Causality relations of the groups of explanatory variables. Source: own edition 115 As a starting phase for the path analysis with a simple multivariate re- gression along with all independent variables based on micro-regional data, an att empt was made to explain the proportion of foreigners by citizenship. Our results are summarised in Table 3. Of them, on the one hand it may be pointed out that the variables involved in the analysis jointly explain with an R2 value of between 0.83 and 0.99 the proportion of the population with a proper citi- zenship in the resident population, on the other hand, signifi cant diff erences by citizenship may be found in the weight of the explanatory factors. Further it should be stated that the proportion of earlier migrants by micro-region has the most signifi cant explanatory meaning in all cases, i.e. the newly arrived foreigners are distributed in line with the existing patt ern. With the help of the path analysis, however, only with the geographic location of micro-regions (distance from the corresponding border and from Budapest) it was att empted to explain the proportion of foreigners and to show the importance of the geographic proximity. The location may have a direct and, through other variables, an indirect infl uence, which will also be quantifi ed. As we have two independent primary variables so the betas of binary linear regressions are broken down into indirect and direct parts by this procedure in an additive way. The schematic system of our path analysis is illustrated by Figure 2. As a next step the relations will be analysed among accessibility and the ratios of migrant groups to a resident population at micro-regional level, in the beginning irrespective of their indirect or direct role. Table 4 is to illustrate steepness at a “simple” binary regression; R is to measure closeness at this stochastic relationship. R2 is to show in percent- ages how the geographic location explains the dispersion of micro-regional Table 3. Regression results Dependent variable Denomination EU15 Serbia Romania Slovakia Ukraine β1 β2 β3 β4 β5 β6 β7 β8 β9 β10 β11 R2 Border Budapest Car Shops Earning Enterprises Decrease Migration Crime Kpfe Ratio – -0.212 0.034 -0.058 0.077 -0.413 -0.096 0.006 0.150 -0.016 0.215 0.838 0.830 -0.014 0.008 0.039 -0.017 0.006 -0.016 0.001 0.007 -0.003 -0.022 0.983 0.990 0.006 -0.065 0.170 -0.023 0.051 -0.138 0.038 0.044 -0.015 -0.059 0.863 0.920 -0.123 0.016 0.100 0.044 0.032 -0.182 -0.014 -0.031 0.086 0.095 0.817 0.820 -0.022 0.016 0.023 0.027 0.053 0.025 -0.001 0.004 -0.018 -0.016 0.971 0.960 116 distribution for foreigners with a given citizenship. So we can conclude that the geographic location explains in itself in 22–30% of the micro-regional vari- ances for foreigners with a given citizenship; that is why the geographic loca- tion plays a signifi cant role when the foreigners choose a place of residence in Hungary. To be fair, it has to be noted, based on Table 3, that for the foreigners plays an even greater role in an informed decision to choose a domicile. They will sett le down with a high probability in those micro- regions where their compatriots already live in greater numbers, who will help them in the proc- ess of management of migration, in the adaptation, in solving administrative problems, in looking for a job, in the issue of housing in general the process of integration. In the terminology of Table 4, the nearest corresponding border when analysing the countries of EU15 is the Austrian border, while in other cases the borders corresponding to citizenships. In a regression the steepness at these variables being negative means that when moving away from the border, the analysed group with a foreign citizenship as a rule accounts for a decreasing proportion, whereas positive regression indicates an increasing proportion. In a similar way, if those betas, which belong to the access time of Budapest are negative, then when moving away from the capital city the foreigners, on average, will account for a decreasing proportion of the resident population, however, in case of a positive steepness for an increasing proportion. As it can Table 4. Binary regression results between accessibility and migrants' proportions Coeffi cients Time to access the nearest border crossing point, 2008 Time to access Budapest, 2008 EU15 β R2 -0.509 0.141 0.221 Romania β R2 -0.193 -0.488 0.259 Serbia β R2 -0.575 0.203 0.284 Slovakia β R2 -0.516 0.076 0.236 Ukraine β R2 -0.489 0.228 0.303 Source: HCSO, own calculation 117 be seen at data on Table 4 – with the exception of those who migrated from Romania – in all cases the distance measured from border crossings is longer than the distance measured from Budapest, which is shown by the diff erence between standardized betas. That is in addition to the central character of the capital city, borders play a signifi cant role in the geography of migration. In the further part of the path analysis the beta value was broken down into direct and indirect paths. To this eff ect, in the fi rst place it was analysed that out of primary characters (accessibility) which and how infl uence the secondary ones (economic situation, social situation). This operation began with the distances measured from the border: The distance measured from the Austrian border – except indicted cases – produces a signifi cant eff ect on all analysed secondary factors (in case of Annex 1–5 non-signifi cant values are marked with grey). Signs in most cases are negative that is why there is higher development, bett er provision and school att ainment, etc. nearer to the border. There is only one positive sign for natural change (increase/decrease), which is in conjunction with the present demographic processes in Hungary. The closest correlation may be seen be- tween car ownership and the distance from the Austrian border (Annex 1). Annex 1. The role of distance from the Austrian border in explaining the tatio of immigrants from EU15 countries within total population in 2001–2008. Source: HCSO, own edition 118 Annex 2. The role of distance from the Romanian border in explaining the ratio of immigrants from Romania within total population in 2001–2008. Source: HCSO, own edition Annex 3. The role of distance from the Serbian border in explaining the ratio of immigrants from Serbia within total population in 2001–2008. Source: HCSO, own edition 119 Annex 4. The role of distance from the Slovak border in explaining the ratio of immigrants from Slovakia within total population in 2001–2008. Source: HCSO, own edition Annex 5. The role of distance from the Ukrainian border in explaining the ratio of immigrants from Ukraine within total population in 2001–2008. Source: HCSO, own edition 120 The distance measured from the Romanian border (Annex 2) is insignifi - cant in connection with the migration balance, criminal off ences and educational att ainment; concerning other secondary indicators it exerts a diff erent infl uence as we may have seen before. Against the distance measured from the Austrian border, here the signs are mainly positive that is the socio-economic situation is improving when moving away from the border; so the border zone may be characterized unambiguously as a periphery. In this respect the distance meas- ured from the border is in the closest correlation with the car density. However, natural change shows a decrease when moving away from the border. The distance measured from the Serbian border produces a signifi cant eff ect on only three secondary indicators (Annex 3). When moving away from the border there is an improvement in provision with food shops, in income per taxpayer and in natural increase. The distance measured from the Slovakian border exerts a signifi cant infl uence on car ownership, income per taxpayer, density of enterprises and natural increase (Annex 4). When moving away from this border there is an increase in car ownership as well as in enterprise density and a decrease in productivity and natural increase. At last the distance measured from the Ukrainian border exerts a sig- nifi cant infl uence on three secondary variables too (Annex 5). When moving away from the border there is an increase in car and enterprise density as well as a drop in natural change. The distance measured from the border is the clos- est for this last indicator. Distances measured from the Serbian, Slovakian and Ukrainian borders were in the closest correlation with the natural increase. Closeness among primary and secondary indicators may be analysed with the help of a determination coeffi cient, which shows how accessibility indicators explain diff erence from the average of socio-economic indicators. It may be pointed out that the inequality indicators fi rst of all explain dispersion at the migration balance, car ownership and productivity (accessibility interprets more than one third of dispersion in case of all the three). In spite of this, the weighed determination coeffi cient for criminal off ences is only 5%, the lowest for the analysed indicators. Aft er analysing how the primary and secondary explanatory factors relate each other we should focus our att ention on how these variables impact the tertiary ones. In 2000, the ratio of arrivals from EU15 countries to the resident popula- tion was directly and signifi cantly infl uenced by the distance measured from the Austrian border as well as the eff ect of this may be felt through specifi c data of food shops and the business density (Annex 1). Of these three paths the direct one is the strongest. In this case the sign is negative, i.e. considering 2000 there was also a decrease in arrivals from EU15 countries when moving away from the border. 121 In one respect, in 2000, the distance measured from the Romanian bor- der produced a direct and signifi cant eff ect on the ratio of those who came from Romania as well as its eff ect could be felt through the provision with cars and food shops and the productivity (Annex 2). Of the analyzed paths the direct impact of the distance measured from the border is the strongest and has a negative sign, i.e. there was also a decrease in the ratio of arrivals from Romania in 2000 along with an increase in distance. In 2000, there was a significant correlation between arrivals from Romania and the distance measured from the Romanian border, an impact was also felt through car and food shop provision as well as productivity (Annex 2). At the analyzed paths the distance from the border has the strongest direct eff ect with a negative sign, i.e. there was a decrease in the ratio of arrivals from Romania along with an increase in distance already in 2000. In 2000 only the distance measured from the Serbian border has a signifi - cant eff ect on the ratio of arrivals from Yugoslavia; there is no signifi cant correla- tion through the secondary factors. There is a decrease in the share of migrants along with an increase in the distance measured from the border (Annex 3). In 2000, on the one hand, the distance measured from the border had a direct eff ect on the ratio of those who came from Slovakia; on the other hand, it also had an indirect eff ect through the natural increase/decrease. Of the two indicated paths the direct one is the stronger and it has a negative sign, i.e. there is a decrease in the share of immigrants along with an increase in distance (Annex 4). At last, in 2000, there was a direct, signifi cant correlation between the distance measured from the Ukrainian border and the ratio of migrants from Ukraine to the resident population (of all border sections here is the strongest direct impact), as well as an indirect eff ect expressed through car and business density as well as natural increase/decrease (Annex 5). When observing how tertiary variables impact dependent ones it can be pointed out that this is signifi cant in all cases and shows the strongest standard- ized beta-coeffi cient. It means that based on our model, the share of migrants is mostly infl uenced by the territorial distribution of earlier migrants. The high- est standardised beta-coeffi cient can be observed with the immigrants from Serbia. Considering the model as a whole, in 2001 and 2008, there was a signifi - cant correlation between the distance from the border and the average share of immigrants from the EU15 and Serbia. It is not true at the distance measured from Budapest, which is not signifi cant in any case when considering its di- rect impacts. Of course it does not mean that there is no correlation between the distance measured from Budapest and the ratio of immigrants within the resident population. That has eff ects not in a direct way but rather through dif- ferent socio-economic factors. So this part of the path analysis is not detailed 122 separately in the present article, but due to the later results these calculations are also shown in Annex 1–5. Aft er identifying the “path strengths” in our model identifi cation start- ed as to the accessibility impact upon the territorial distribution of migrants. The question is how accessibility indicators (directly or, through other factors, indirectly) impact the ratio of immigrants by citizenship. When look at the variable for the distance measured from the Austrian border, as it can be seen in Annex 1 this primary factor has a direct impact of -0.2123. On the one hand indirect paths may go over the primary, second- ary and tertiary variables, at this time all ways have to be added together from the onset to the dependent variable, while the proper path sections have to be multiplied together, i.e. (irrespective of signifi cances): (-0,4148*- 0,1847*0,838)+(-0,2291*0,2976*0,838)+(-0,1749*0,0324*0,838)+(-0,268*0,6817*0 ,838)+(0,2714*-0,0822*0,838)+(-0,1653*0,0435*0,838)+(-0,027*0,1306*0,838)+(- 0,2125*-0,1349*0,838)=-0,15463. Furthermore through the primary and secondary variables: (-0,4148*- 0,058)+(-0,2291*0,07725)+(-0,1749*-0,413)+(-0,2682*-0,0958)+(0,2714*0,00642)+(- 0,1653*0,1496)+(-0,027*-0,0163)+(-0,2125*0,21524)=0,03599. Or through the primary and tertiary variables: -0,2126*0,838=-0,1782. So the indirect eff ects as a total: -0,15463+0,03599+-0,1782=-0,2968. Together with the direct eff ects: -0,2968+-0,2123=-0,5092. I.e. a partial steepness appearing in Table 4 is obtained. Total paths were calculated for the analyzed citizenships and for both accessibility indicators. The results are contained by Table 5. In general it can be pointed out that in all cases accessibility indicators has no direct impact but fi rst of all an indirect one described by socio-economic indicators. An analysis of the foreigners’ places of residence in Hungary by an indicator on location potential As it was seen, the att ractive target area in Hungary for a foreigner migrant is one where his/her compatriots with the same citizenship live (Sik, E. 1999). So with the help of a location potential indicator it could be visualised how the foreigners with a diff erent citizenship see the area of the country as a potential destination to sett le down. The used accessibility potential was calculated from the Hansen type gravitational model (Hansen, N. 1977). During the research, in the way that was described previously a gravi- tation analogy based model was calculated with a linear resistance factor (Tóth, G. and Kincses, Á. 2007). For accessible destinations, volumes were 123 determined based on the population with a corresponding nationality in sin- gle micro-regions. This present analysis takes into account what accessibility conditions are in a given area, i.e. accessible destinations in the area. Based on our model, the potential in point I of the space: where Bi, Bj volumes for accessible destinations dij distances between I and j mi- cro-region centres in minutes, while di is the own distance (in minutes), which can be calculated in a way that for the area of a given micro-region regarded as a circle, a radius is determined, which is considered as proportional with intra-micro-regional public road distances and the time required to cover this radius is regarded as an own distance. Table 5. The role of direct and indirect paths in explaining the share of immigrants within total population (standardised Β coeffi cients) Coeffi cients Access time for the nearest corresponding border crossing, 2008 Access time for Budapest, 2008 EU15 indirect direct total -0.297 -0.212 -0.509 0.106 0.034 0.141 R2 0.221 Romania indirect direct total -0.199 0.006 -0.193 -0.424 -0.065 -0.488 R2 0.259 Serbia indirect direct total -0.562 -0.014 -0.575 0.195 0.008 0.203 R2 0.284 Slovakia indirect direct total -0.393 -0.123 -0.516 0.060 0.016 0.076 R2 0.236 Ukraine indirect direct total -0.467 -0.022 -0.489 0.212 0.016 0.228 R2 0.303 Source: HSCO, own edition d B d B i i ij ij j iP 124 In case of EU15, Serbian, Romanian, Slovakian and Ukrainian citizens micro-regional potential values are mapped in 2001 and 2008 (Figure 3). Slovakian citizens, 2001 Slovakian citizens, 2008 125 Ukrainian citizens, 2001 Ukrainian citizens, 2008 126 Serbian citizens, 2001 Serbian citizens, 2008 127 Romanian citizens, 2008 Romanian citizens, 2001 128 Fig. 3. Results of potential models. Source: HCSO, own edition EU15 citizens, 2001 EU15 citizens, 2008 129 As it can be seen from these fi gures there are diff erences in the distribu- tion of location potential by micro-region in case of foreign citizens staying in Hungary. Channels can be identifi ed between Budapest and the source countries with the exception of those who arrived from Romania, for whom Budapest and its surroundings represent an att raction, but they can be found on the whole territory of the country. Strong potential corridors can be identifi ed for the Ukrainians trending in an east–west direction, for the Austrians in west–east, whereas for both the Serbs and the Slovaks in north–south. Summary Budapest and its gravity zone accounts for the residence of a predominant part of the foreign migrants, while a smaller proportion of them live in micro-re- gions along the border as well as in the surroundings of Lake Balaton. Budapest and Pest County are unambiguously att ractive destinations for those foreigners who arrived in Hungary from the neighbouring countries, but they also prefer micro-regions located nearer to the country relating to their citizenship, mainly in the vicinity of the Romanian, Ukrainian and the Serbian border. During the path analysis the variables involved in the analysis jointly explain in a decisive way the ratio of the population with a proper citizenship to the resident population, thus our hypotheses has fulfi lled. On the other hand, however, signifi cant diff erences by nationality can be pointed out in the weight of the explanatory variables. One of the most important fi ndings of our analysis is that for the average proportion of foreigners between 2001 and 2008, the strongest explanatory force was in all cases the ratio of those compatriot immigrants who came earlier. I.e. the new immigrants follow the existing spatial patt ern in their distribution. In case of all citizenship groups, accessibility indicators have no direct eff ect but an indirect one, which is fi rst of all described by socio-economic indicators. REFERENCES Alwin, D.F. and Hauser R.M. 1975. The Decomposition of Eff ects in Path Analysis. American Sociological Review 40 (1): 37–47. Anderson, J. and O’down, L. 1999. Borders, Border Regions and Territoriality: Contradictory Meaning, Changing Signifi cance. Regional Studies 33. (7): 593–604. Borjas, G.J. 1996. Labour economics. New York, McGraw Hill 279–231. Contessi, S. 2001. The Regional Location of FDI in Central Europe: an Empirical Analysis – Comparative Perspective. Bocconi University 33 p. 130 Duncan, O.D. 1966. Path Analysis: Sociological Examples. The American Journal of Sociology, 72. (1): 1–16 Hamilton, B. and Whalley, J. 1984. Effi ciency and distributional implications of global restrictions on labour mobility: Calculations and policy implications. Journal of Development Economics, Elsevier 14. (1): 61–75. Hansen, N. 1977. Border Regions: a Critique of Spatial Theory and a European Case Studies. Annals of Regional Science 11. 1–12. Hardi, T. 2008. A határtérség térszerkezeti jellemzői – Features of the Border Region’s Spatial Structure. Tér és Társadalom (Space and Society) 23. (3): 3–25. Hatton, T.J. and Williamson, J.G. 2005. Global Migration and the World Economy: Two Centuries of Policy and Performance. Cambridge, Mass. MIT Press, 488 p. Illés, S. 2004. Foreigners in Hungary: Migration from European Union. Working Papers on Population, Family and Welfare No. 5. HCSO Demographic Research Institute 44 p. Krugman, P. 1998. What’s new about the new economic geography? Oxford Review of Economic Policy 14. (2): 7–17. Krugman, P. and Venables, A. 1996 Integration, Specialization, and Adjustment. European Economic Review 40. 959–968. Németh, N. 2008. Fejlődési tengelyek az új hazai térszerkezetben. Az autópálya-hálózat szerepe a regionális tagoltságban (Development axes in the new spatial structure of Hungary. The role of the motorway-network in regional disparities.) PhD. Értekezés-PhD thesis, ELTE Regionális Tudományi Tanszék, Budapest 150 p. Papademetriou, D.G. 2006. New Migration Thinking for a New Century. Migration Policy Institute (MPI) 25 p. Rédei, M. 2007. Mozgásban a világ – a nemzetközi migráció földrajza. (The world in moving The geography of the international migration.) Budapest, Eötvös Kiadó 568 p. Sik, E. 1999. Migrációs potenciál a mai Magyarországon. A munkaerő migrációja és az Európai Unió. Budapest, Európai Tükör sorozat, 61. 93–118. Székhelyi, M. and Barna, I. 2008. Többváltozós elemzési technikákról társadalomkutatók számára – Túlélőkészlet az SPSS“-hez. (Multivariate analysis techniques for Social Research.) Survivor supply to the SPSS.) Budapest, Typotex 453 p. Tóth, G. and Kincses, Á. 2007. Elérhetőségi modellek. (Accessibility models.) Tér és Társadalom (Space and Society) 21. (3): 51–87. Traistaru, I., Nij kamp, P. and Resmini, L. 2002. The Emerging Economic Geography of EU Accession Countries. European Communities 456 p. Van Geenhuizen, M. and Ratti, R. 2001. Gaining Advantage from Open Borders. An active Space for Regional Development. Ashgate, Aldershot. 398 p. Venables, A.J. 1998. The assessment: trade and location. Oxford Review of Economic Policy 14. (2): 1–6. Williamson, J.G. 2006. Global Migration. Finance and Development 43. (3): 1–11.