Immigration and politics – the case of Norway Frøy Gudbrandsen, Department of Comparative Politics, University of Bergen Paper (draft) prepared for ”Nasjonal Fagkonferanse i Statsvitenskap”, Trondheim 3rd 5th of January 2007. The political influence on immigration is uncertain. Norwegian political parties disagree on what influence they actually have on immigration: In a television debate (Holmgang 31.05.2006), the Progress Party’s spokesperson claimed that the recent sharp increase in immigration was caused by the government’s liberal stand on immigration. On the contrary, the Christian Democratic Party spokesperson identified fluctuations in the number of conflicts in the world as the main cause of immigration to Norway. Not only politicians, but also scholars disagree on the political influence on immigration. There is disagreement on both the validity of the partisan influence on policy output in general, as well as on states’ capacity to control immigration. Is it governments or external determinants that decide the size of immigration? This study aims at answering the question. International migration has increased globally and has had an impact on society in many countries. It has proved to be an emotional and sensitive political issue, which makes research aiming at providing factual information particularly important. There is not only disagreement on the causes of immigration, but also on what to do with it. For some time immigration has been a highly contested political issue, in Norway as well as in most industrial countries. As will be demonstrated, political parties in Norway have had very different preferences to immigration policy. But has immigration flows changed according to the political stands of the parties in government? This paper analyses the effect of changes in the party composition of government on immigration to Norway. The first part of the paper examines the immigration preferences of Norwegian political parties. The primary source is party programmes. The second part is a panel analysis of the impact of changes in government on immigration flows. The period 2 1965-2004 is covered and 185 sending countries are included in the analysis. This model controls for not only features of the Norwegian society , such as unemployment, but also features of all sending countries. While ignored in most quantitative studies, the inclusion of both sending and receiving country variables is crucial for a valid test of the political influence on immigration. When either sending country variables or receiving country variables are omitted, the statistical model is underspecified, which causes serious bias in the results. The case of Norway may shed light on the political influence on immigration for several reasons. First, poor data quality is a major obstacle to reliable quantitative analyses of immigration. But the exceptional quality of Norwegian immigration data provides an opportunity to overcome this problem. Illegal immigration has the potential of invalidating official statistics, but although on the increase in recent years, illegal immigration has been comparatively small in Norway. Second, the country’s geographical position far away from all major emigration countries gives Norwegian governments the potential of stronger immigration control than most industrial countries. Third, the constitutional structure of Norway is favourable for partisan influence (Schmidt 2002:177). The consensus orientation of Norwegian politics, as well as the frequent coalition and minority governments, may have blurred out the differences in policy preferences. However, with a centralized state structure, few majoritarian constraints, the possibility of partisan influence in Norway is likely to be high (Scmidt 1996:170).1 Countermajoritarian constraints may be “federalism, an influential constitutional court and an autonomous central bank,” (Schmidt 1996:170). 1 3 POLITICAL INFLUENCE ON IMMIGRATION? There is scant theoretical literature on the effects of parties and governments on immigration as such, but the extensive political economy literature on partisan theory and economic electoral cycles may well be applied other policy areas. This literature argues that politics does indeed have an influence on policy outcomes (See for instance Hibbs 1992; Garrett 1998). To retain support among their core supporters, politicians must stay loyal to the ideological basis of their parties: “Partisan models therefore imply a political signal in economic policies and outcomes correlated with shifts in party control of government,” (Hibbs 1992:362). Political ideologies and policy preferences of different political parties can be expected to be reflected in policy outputs2. The ’parties matter’ literature has primarily been concerned with economic policy, but here it will be applied to immigration. Since the mid 1980s, all Norwegian parties, with the exception of the Labour Party 1993 programme, have statements about immigration policy in their party programmes. Voters in general do not rank immigration as important (Aardal and Valen 1989, 1995; Aardal 1999, 2003). Despite the relatively low interest for immigration as a political issue, it is likely that parties implement policies in accordance with their party programmes: “As an official document, it will be difficult for party members to resile from policies in the party manifesto, while party leaders can be charged with failure to implement published manifesto pledges when given the chance to do so” (Laver and Garry 2000:620) . Research on partisan influence on the economy come to diverging conclusions (See Blais et al. 1993; Imbeau et al. 2001) and provide no indications on the extent of political influence on 2 Stimson et al (1995) draw attention to the possible direct effect of public ophion on policies, as politicians are sensitive to public opinion also between elections. Lack of time series data on public attitudes to immigration policy makes such a study impossible. 4 immigration. A precondition for partisan influence, is state capacity to control, and on this point the immigration literature in just as inconclusive as the partisan literature. Political economists have argued that economic globalization has decreased state capacity to steer national economy (see for instance Moses 1994). Similar arguments are found in the migration literature. Sassen (1996; 2000) argues that there is a contradiction between an open economy and strict control of people flows. Pointing to the example of U.S. agricultural export, she argues that immigration may be one of the trade offs in the processes of economic globalization (Sassen 2000:67). Local agriculture in the import countries is highly influenced by agricultural trade and may transform surplus farmers in import countries into labour migrants in the U.S. Immigration today is driven by “a complex, deeply embedded, and transnational process,” and state capacity to control immigration is therefore increasingly constrained (Sassen 2000:73). Sassen (1996) has also argued that an international human rights regime, consisting of both informal norms and binding conventions, limits national control over immigration. Joppke (1998), and Guiraudon and Lahav (2000) disputes this: “Rather than global processes constraining domestic action, what we observe in the case of aliens rights is a legally driven process of self-limited sovereignty”, (Guiraudon and Lahav 2000:189). They find that national constitutions may constrain policy makers, but still Germany, France and the Netherlands have demonstrated ability to control migration (Guiraudon and Lahav 2000:190). Realising that international migration is an international process, governments have increased international co-operation on immigration. This may have reduced the possibility for partisan influence on immigration flows. According to Cornelius et al. (2004:15-20), there has been a significant immigration policy convergence, particularly between European countries. There 5 may be several reasons for this convergence (Meyers 2002), but the result of convergence is likely to be that political party preferences to a lesser extent influence what policy measures are introduced in each country. Implementation problems and unintended consequences have been found to be the main causes of the “gap hypothesis” (Cornelius et al. 2004:4), which describes the gap between restrictive policies and increasing immigration that has been experienced in most European countries that introduced the ‘immigration stop’ in the 1970s. Governments depend on the bureaucracy for their decisions to be effective. Many aspects of the implementation process may disturb the link between policy aims and policy output, which in this case is immigration flows. Howlett and Ramesh (2003:187-193) mention several reasons for an implementation gap. Actors who disagree with the policy, may seek to influence the implementation. These actors could be the bureaucrats themselves, or external activists who call on the bureaucrats as the last point of influence. There are commonly technical or administrative difficulties, both predicted and unpredicted, that makes implementation slow, incomplete or impossible. The situation that policies were intended to address could also change, so that policies become irrelevant or unsuitable for managing reality. Policies also commonly have unintended consequences (Kelly 2005:869). This happens both due to politicians lack of knowledge about the issue and because policies have secondary, and completely unforeseen, effects. An example of the first case is attempts that have been made to reduce immigration by boosting development in sending countries. Empirical research has repeatedly supported the development hump hypothesis, which states that development increases emigration, at least in a short in the short run (Castles and Miller 2003). Accordingly, such policies were found only to increase immigration (Martin and Taylor 2001: 6 105). This does, of course, not mean that policies are never efficient. On the contrary, several studies document policy influence on immigration (see for instance Holzer et al 2000, Brekke 2004, or Thielemann 2004). As pointed out by Hibbs (1993:363), “(a)ll governments respond to high and rising inflation with contractive policies, and to prolonged or sharp downturns with expansive ones.” Likewise, it can be assumed that all governments react to sharp increases in immigration with restrictive measures. Policy must be adjusted to fluctuations, whether immigration or inflation. It is therefore possible that changing levels of immigration, perhaps primarily determined by external factors, influence policy more than changing governments. The partisan theory suggests that politics could matter for immigration, but so far we can only conclude that the validity of this theory, when applied to immigration, is uncertain. PREFERENCES OF NORWEGIAN PARTIES Immigration has been a central, and sometimes a decisive, issue in recent elections in Europe. In Norway, immigration has only been the major issue once; in the 1987 local election. The election followed a dramatic rise in the number of asylum seekers that in turn put immigration high up on the media agenda. In the general election two years later, an all time high of 9%3 of the electorate considered immigration important for their choice of party. In the 1997 general election, liberalization of asylum policy was one of the central issues of the winning coalition, but this did not necessarily reflect the interest of the electorate. Only six percent considered immigration important. Immigration has, however, been more important to the supporters of certain parties. For instance, 20% of the Progress Party voters and 10% of the 3 Berfore 1989, less than 1% regarded as immigration important ( Aardal and Valen 1995:166, Aardal 1999:2021). 7 Socialist Left party voters considered it important in 1997 (Aardal 1999:24). A precondition for testing the influence of political parties on immigration is knowing their preferences. The left right scale, which as been the focus of attention in studies of economic policy, cannot be applied directly to immigration. The first part of this study is therefore an analysis of the positions of Norwegian parties on immigration. Party preferences on immigration have been described in case studies of Norwegian politics (Hagelund 2003; Tjelmeland and Brochmann 2003). This is not, however, sufficient to determine the intention of Norwegian political parties.4 Here, the primary data source for establishing the positions is party manifestos. Immigration was first mentioned in the party manifestos in 1973, but since 1985 the issue has been covered in all party manifestos. The parties’ views on three types of immigration will be studied: labour immigration, refugee/asylum and family reunion. Statements were coded as restrictive, liberal or neutral. The statements in one party manifesto were compared to statements in the other Norwegian party manifestos. Accordingly, the codes must be interpreted as liberal or restrictive relative to other Norwegian parties. All instances of ambiguity were coded neutral so that only clearly restrictive and clearly liberal statements were given these codes. No mention of immigration in a party programme may have to explanations: 1) that immigration is not seen as relevant, or 2) that there is no agreement on the issue within the party. In both cases, ‘neutral’ is considered the most appropriate code. As all three types of immigration were not always mentioned in the manifestos, newspaper Tjelmeland and Brochmann’s coverage of the preferences is rather brief. Hagelunds study is a discurse analysis of the immigration debate, which provides information a bit on the side of what needed for this study. 4 8 articles are a secondary, supplementary source5. Most newspapers’ made their articles electronic from 2004, but newspaper articles from Aftenposten are available electronically from 1983, from Dagens Næringsliv from 1988 and a few newspapers from the early 1990s. Statements from central politicians, speaking on behalf of the party, were coded. Consequently, claims about other parties and initiatives from local politicians were ignored. The preferences of the parties as expressed in newspaper articles were to a large extent consistent with the statements in the manifestos. Table 1 Preferences according to party program statements 1973 Progress Party Labour imm. 1977 R 1981 R N 1985 R R R 1989 R R 1993 R R R 1997 R R R N L Lr R r R L N Nr N R r/l L r Nl L L L l L L r l r L Refugee/asylum Family imm Labour imm. Conservative Refugee/asylum Party Family imm R Labour imm. Labour Refugee/asylum Family imm L l L N Labour imm. Liberal Party Refugee/asylum Family imm Christian Dem. Party R N Labour imm. Refugee/asylum Family imm Labour imm. R R R Centre Party Refugee/asylum Family imm Labour imm. Socialist left Refugee/asylum party L R L L L N r l L N L L L L L L L L Radical left L Refugee/asylum party L Family imm R = restrictive, L = liberal, N = neutral statements. Capital letters= from party programmes. Family imm Labour imm. N L L r L L L L L L L 2001 R R R L Nr r N r L L L L L R N N r L Lr L L L 5 The relevant articles have been traced through the electronic news archive Atekst (www.atekst.no), where 17 daily newspapers are searchable. The key words for the searches were each party name with all the following words: refugees, refugee policy, asylum, immigration, immigrant, immigration policy, foreign workers, guest worker, labour immigration, family reunion. 9 There are some common trends in the Norwegian parties’ attitudes to immigration: The parties are less concerned with family reunion than with labour migration and refugee/asylum migration, although this is the most important type of immigration to Norway. The parties also tend to be more liberal towards refugee/asylum migration than towards labour immigration. But there are also clear differences between the parties. Two parties have had stable preferences on the subject of immigration throughout the period: The Progress Party has been consistently restrictive, and the Liberal Party has, as its party name indicates, been consistently liberal. Before 1989, the Conservative Party only had statements about refugee policy in the manifestos, and these were liberal. In the 1989 manifesto, however, the party called for restrictive policies on both labour and refugee/asylum immigration, and in newspaper articles also demanded tightening of family reunion legislation. During the 1990s the Conservative Party became overtly liberal to labour immigration, and more neutral regarding refugees. The Labour Party has left out immigration from more programmes than any of the other parties. Pearlmutter (1996:377) claims that this is common for large mass parties because “they face cross-cutting cleavages that affect their core constituencies.” He points to a possible conflict between “unions who favour restrictive policies, and liberals and ethnic groups who favour expansionist policies,” and this could well be the case with the Norwegian Labour Party. In the three programmes from 1985 to 1993 the party demanded restrictive labour immigration policies, and this stand was confirmed in newspaper articles until 2005. Regarding refugee/asylum policy, the party has been shifting between liberal, restrictive and neutral, but has in general not granted this issue much attention neither in the manifestos nor in newspaper articles. The same goes for family reunion. The Christian Democratic Party supported restrictive labour immigration policies in the 1977 manifesto and had a few neutral statements during the 1980s and 1990s until a 10 liberal shift in the 2001 manifesto where it called for liberalisation of labour immigration. Since 1989, the Christian Democratic Party has been clearly liberal towards refugee/asylum immigration. Surprisingly, considering that it is a family oriented party, family reunion was only mentioned in one manifesto. The Centre Party was clearly negative to labour immigration in the manifestos from 1973 to 1981. Then there was no mention of the issue until an ambiguous statement in the 2001 manifesto, but the party repeatedly expressed their restrictive stand in newspaper articles. Regarding both refugee/asylum immigration and family reunion, the Centre Party has been liberal, and has been more concerned with protecting the right to family reunion than any other party. The Socialist Left Party has been predominantly liberal towards all types of immigration, but supported restrictive policies on labour immigration in the 1977 manifesto. Newspaper articles suggest that this view was revitalised in 2001, when several parties wanted to liberalise labour immigration. PREFERENCES OF NORWEGIAN GOVERNMENTS The partisan theory suggests that there is a connection between what parties declare in their pre election manifestos, what policies the winning party pursue in government and, finally, policy output. The preferences of governments, as derived from party programmes and newspaper articles, are presented in the table further down. The process of influence is more complicated when several winning parties form a coalition government (see Hofferbert and Klingeman 1990). Out of the 14 governments between 1965 and 2005, six governments were coalitions. In such cases several party programmes may influence government policy. The minor parties want returns from being in government, and therefore the party program of a minor party may well be determining the immigration policy of a coalition government, but generally the dominant party can be expected to have a 11 stronger influence on policy than minor parties. When the parties in a coalition government had differing preferences, the largest party was given more weigth. The bargaining result may be stated in a government declaration but in Norway such declarations have traditionally been rather brief and immigration has not been covered once. It can be expected that the party of the minister responsible for immigration affects immigration more than other parties in government. Until 2001, this was the Minister of Justice. Between 2001 and 2005 the Minister of Regions and Municipalities was the responsible minister, and since then it is the Minister of Work and Integration. The partisan influence through the responsible minsiter will be tested in the empirical analysis later. One administrative adjustment of immigration policy should also be mentioned. Until 1998 the final appeal body of immigration cases was the Justice Department, implying that the Minister of Justice had the final say in controversial immigration cases. When the Immigration Appeals Board was established in 1998, politicians were released from the burden of making decisions in individual cases6. In particular in cases with substantial media coverage, there is reason to believe that the fear of becoming unpopular among the electorate could influence the minister’s decision. For better or worse, ministers had a larger potential to influence immigration before 1988. 6 In 2004, politicians regained the rigth to instruct both the Immigration Appeals Board and the Directorate of Immigration in the interpretation and judicial assessment of legislation. 12 Table 2 Preferences of governments Year Labour immigr. Asylum/refugees Family reunion 1973 1974 1975 1976 1977 1978 1979 1980 1981 L 1982 L 1983 R L L 1984 R L L 1985 R 1986 R 1987 R 1988 R 1989 R 1990 R R r 1991 R L N 1992 R L N 1993 R L N 1994 R r l 1995 R r l 1996 R l l 1997 R l l 1998 L L L L L 1999 L L 2000 R N/l 2001 L N/L L 2003 L N/L L 2004 L N/L L 2005 L N/L L R=restrictive, N=neutral, L=liberal, empty cell= no relevant statemensts. Capital letters: from party programmes. GENERAL DETERMINANTS OF IMMIGRATION To trace possible political effects on immigration flows, it is necessary also to consider what else determines the flows. The economy in the country of emigration influences the number of emigrants, but not necessarily to the number of immigrants to Norway in particular. The development hump hypothesis suggests that countries at a middle level of development 13 produce more emigrants than countries at a high and low development level (REF). At a middle level of development, more people have the means to emigrate while a large proportion of the population has something to gain from emigrating to for instance Norway. At a high level of development, the gains from emigration may be considered lower. Applied to the case of Norway, it may be expected that immigration to Norway is low from poor and rich countries, but high from countries at a middle level. It should be stressed, however, that other factors, such as for instance distance, probably are far more important. The level of economic development of the sending country may be decisive for the total number of emigrants from a country, but this does not mean that any of them go to Norway. War and repression creates refugees and are therefore expected to influence immigration to Norway. Since 1975, the right to asylum has been one of the two most important “immigration gates” to Norway. Refugees are admitted in two ways: trough individual asylum applications or through the UN quota system. The size of the quota is determined by the Norwegian parliament. Individual asylum seekers are given a residence permit if they are considered in danger, and the number of permissions granted therefore depends on the number of asylum seekers that arrive in Norway. Demographic development may influence international migration in primarily two ways: “directly, by raising the young adult share; and indirectly, by glutting the home labour market and thus worsening employment conditions there,” (Hatton and Williamson 2003: 7). The high emigration prosperities of twenty-year olds can be explained by the fact that young people generally have weaker ties to the place where they live, as they have invested less in work place and housing. Fewer family commitments may also allow higher mobility than in later phases (See Pandit 1997: 431). Population growth in sending countries is therefore 14 expected to influence immigration to Norway. Immigration can be explained partly by economy differences between sending and receiving countries. Labour surplus and lower wages in sending countries, and labour demand and higher wages in receiving countries drive migratory flows (REF). However, for macroeconomic changes to cause changes in immigration, borders need to be relatively open. With the labour immigration stop in 19757 this primary precondition for the economy to influence immigration was also removed. That is, Norway has had a common labour marked with the other Nordic countries since the 1950s and with the EU from 1994. Immigration from these countries is therefore likely to be more influenced by economic changes. But macroeconomic changes may have influenced immigration also from countries outside the EU, as both refugees and family members of Norwegian residents could still immigrate after 1975. Refugees may, at least in theory, chose between a number of possible countries of immigration, and the economic situation in the various countries may influence choice. A country could therefore attract more asylum seekers in periods of low unemployment 8. When deciding what country a family should settle in the economic situation in both countries may matter. However, if the two countries are on very different levels of economic development, economic variations do not change the overall differences between the countries. As stated by the network theory (see for instance Arango 2000: 291), migration flows between two countries tend increase over time, independently of the causes mentioned above. 7 The only exception was highly educated personel that was very much needed in Norway, primarily in the oil industry. About a 1000 specialists were admitted a year. 8 Although wage levels and employment opportunities vary between these countries, immigrants must be very well informed to choose the country of destination according to such relatively subtle differences. 15 Formal and informal networks between migrants and their home countries developed and sustained, and the networks ease the migration process for new migrants. All forms of migration are influenced by previous flows, but family immigration is particularly dependent upon previous migration9. DATA AND METHODOLOGY The partisan influence on immigration will be tested through a quantitative analysis of panel data, using within group regression. This model is considered most appropriate, because it is expected that there are constant, country specific facors that influence immigration, that should be controlled for. Like with time differencing, within regression has the serious disadvantage that variables that are constant within groups and across time cannot be estimated. Distance and historical ties to Norway are two relevant variables that must be excluded when within regression are chosen. However, this is not problematic as long as the excluded variables are control variables. Here, none of the variables of interest are constant. The great advantage with within regression, is that all constant factors are controlled for. The problem of omitted variable bias is therefore partly removed10. The inclusion of a lagged dependent variable in a within regression model is likely to be in conflict with the assumption that the independent variables are uncorrelated with the errors. This is a particularly large problem in short panels, and is reduced as the number of time units increases. Theoretically, a solution to the problem is the use of instrument variables, but good 9 Family immigration consists of two groups: family reunion and formation of new families. The number of family reunions is by nature strongly influenced by previous immigration, as an immigrant is reunited with his/her family, while immigration through the formation of new families also depends on travel, the internet, and work and studies abroad. 10 The omission of relevant time variant variables, on the other hand, leaves the model underspecified and gives biased results. 16 instruments are difficult to find11 and IV estimates are therefore commonly inefficient. Baltagi (2005:136) sates that “even for T=30, this bias could be as much as 20% of the true value of the coefficient of interest.” Here we have 40 time units. This is very high compared to most panel analyses in the social sciences, and the bias can be expected to be much lower than in Baltagi’s example. The costs of instrument variable estimation appear to be higher than the cost of biased estimators. I therefore prefer not to use IV estimation. When using time series data autocorrelation must be tested. When using a lagged dependent variable, Wooldridge (2006:421) suggest regressing the lagged residuals, as well as the other variables, on the residuals to test for autocorrelation. The t values of the lagged residuals are low, -0,72, which means that first order autocorrelation is not present. With one year time intervals, higher order autocorrelation is unlikely. Heteroskedasticity, on the other hand, is found to be present, as the regression results with and without corrected standard errors are very different. The standard errors will therefore be corrected for panel heterskedasticity. The panel consists of 185 sending countries of immigrants to Norway. Around 87% of these countries have values on all variables for all years in the period 1965-2005. The main reason for not staying in the panel is the fact that the country did not exist throughout the whole period. It has it proved difficult to find data on states that have seized to exist, such as the USSR. It is also practically impossible trace data about geographical areas before they became independent states, such as East Timor. Some countries could not be included at all due to missing values on the independent variables. In particular, there is a lack of data on very small states. This could cause seriously biased results, as the case selection is correlated with one of the intended control variables, population size. However, as will be seen later, this 11 Instruments must be correlated with …, but at the same time uncorrelated with …. 17 variable appeared not to be significant and therefore excluded from the final analyses. An authoritarian regime type, and in particular communist regime, seems to cause missing values. Data availability tends to be correlated with economic development, but this is the case here. The dependent variable, immigration to Norway, is measured by the variable ‘First time immigration by country of birth’ from Statistics Norway. This variable has the advantage that it is comparable and reliable for the whole period 1965-2005. A similar variable, ‘Immigration by last country of settlement’ would perhaps be a more valid operationalization. If the immigrants did not live in their country of birth, factors related to their country of last settlement would be more appropriate to control for. However, the latter variable has not been subject to the same quality control as the variable to be used. The correlation between country of birth and country of last settlement is very high. The dependent variable has a mean of 78. On average, 78 people immigrate to Norway from each sending country a year. The standard deviation of 277 shows a very high dispersion. Many countries only send immigrants to Norway occasionally, while hundreds of immigrants arrive every year from countries such as Denmark, Sweden, Pakistan and the UK. The partisan theory is tested by two sets dummy variables. The first tests whether the different party compositions of the government have influenced immigration flows. The variables measure the effect of 1) a coalition of centre parties, 2) a coalition including the Conservative party, and 3) a labour party government12. The other set of dummy variables tests whether the government preferences that were established earlier lead to changes in immigration. The variables state whether the governments preferred liberal, neutral or restrictive policies. 12 The Labour party was never in a coalition government before 2005. 18 Economic development in the sending countries is measured by Purchase Power Parity from Penn World Tables, as well as a squared version of the variable to estimate a possible curve linear effect. A dummy variable indicating whether more than 1000 people died in battle that year measures war in sending countries. The data comes from the Uppsala/PRIO Armed Conflict data set. Demographic development is operationalized as annual population growth, using data from the UN. The state of the Norwegian economy is operationalized as annual GDP growth, with data from the World Development Indicators, and unemployment share of labour stock, with data from NAV. The network theory is operationalized as a lagged dependent variable. A lagged dependent variable may cause biased results when included in a within regression model. The bias decreases with increasing number of time units. Immigration has increased on a global level, due to a number of reasons, many related to globalization. Cheaper travel, internet communication, and higher telephone availability are some factors. A trend variable is included to capture the effect of such factors. Immigration from Bosnia in 1993 was extremely high, and a dummy variable is included to control for this inflow. EU membership is also controlled for through a dummy variable. PLEASE NOTE: The following sections are based on preliminary analyses (I have not yet included all control variables I intend to include). It is therefore very brief, and tables with results are not presented. I do not, however, expect the findings to change dramatically when I include the other variables. 19 ANALYSIS The first part of this study established that there is a possibility that party political preferences have an effect on immigration flows. The following hypotheses will now be tested: H1: The party composition of government influence immigration to Norway. H2: The preferences of parties in government influence immigration to Norway. H3: The preferences of the party that holds the minister responsible for immigration influence immigration to Norway. Although there are arguments in favour of a partisan influence, the parties matter literature, migration theories as well as previous empirical studies suggest that it could be difficult to reject the null hypothesis. In a regression analysis of the control variables, the lagged dependent variable, PPP and its square, unemployment in Norway, EU membership, the Bosnia crisis and the trend variable were found to have significant effects on immigration to Norway. The effects were also in accordance with theoretical expectations. GDP growth was not significant, but this may be explained by the high correlation with unemployment. More surprisingly, war was also insignificant. The R square of the regression of significant variables gives an R square of 0,76. None of the political variables were found to have significant and expected effect on immigration. “Non socialist government” had a positive effect and was significant at a 10% level, one tailed test, but with 6018 units this is considered a too liberal significance level. 20 “Restrictive government” wad a t value of 2,62, but the positive effect is not compatible with expectations. CONCLUSION Norwegian parties were found to have differing views on immigration policy, according to both party programmes and politicians statements in newspapers. However, the empirical test of the partisan theory, applied to immigration, provided no support for the theory. When in government, parties have not translated their preferences into changes in immigration. The empirical analysis covered a long time period, and both sending country and receiving country variables were controlled for. Neither the composition of government nor the government parties’ preferences change the total number of immigrants to Norway. Centre party governments have not increased immigration, and conservative governments have not successfully reduced immigration. Other factors determine the fluctuations. Many arguments against the partisan theory have been presented, such as the possible lack of state capacity, implementation problems and policy convergence. The results of the regression analyses are therefore far from surprising. The analysis of the party programmes revealed, however, that one party may favour restrictive labour immigration policy, but at the same time liberal refugee policy. It is still possible, therefore, there is a partisan influence on particular types of immigration that could not be captured when looking at total immigration, which was done here. This possibility should probably be tested before drawing final conclusions. 21 LITERATURE Arango, J. (2000). "Explaining migration: a critical view." UNESCO: 284-295. Baltagi, B. H. (2005). Econometric analysis of panel data. Chichester, John Wiley. Blais, A., D. Blake and S. Dion (1993). "Do Parties Make a Difference - Parties and the Size of Government in Liberal Democracies." American Journal of Political Science 37(1): 40-62. Brekke, J.P (2004). “The struggle for control. The impact of control policies on the arrival of asylum seekers to Scandinavia 1999-2004.” ISF Report 2004:13. Castles, S. and M. J. Miller (2003). The Age of Migration, International Population Movements in the Modern World. New York, Palgrave Macmillan. Cornelius, W. A., T. Takeyuki, P. Martin and J. F. Hollifield (2004). Controlling immigration : a global perspective. Stanford, Calif., Stanford University Press. Garrett, G. (1998). Partisan politics in the global economy. Cambridge, Cambridge University Press. Guiraudon, V. and G. Lahav (2000). "A reappraisal of the state sovereignty debate - The case of migration control." Comparative Political Studies 33(2): 163-195. Hagelund, A. (2003). The importance of being decent political discourse on immigration in Norway 1970-2002. [Oslo], Unipax : Institutt for samfunnsforskning. Hatton, T. J. and J. G. Williamson (2003). “What Fundamentals Drive World Migration?” World Institute for Development Economics Researcy, United Nations University. Hibbs, D. A. (1992). "Partisan theory after fifteen years." European Journal of Political Economy 8: 361-373. Hofferbert and Klingemann (ref. mangler) Holmgang (31.05.2006). TV2. Holzer, T., Schneider G. and Widmer T. (2000). "The Impact of Legislative Deterrence Measures on the Number of Asylum Seekers in Switzerland." Internaitonal Migration Review, 34:4, 1182-1216. Howlett, M. and M. Ramesh (2003). Studying public policy : policy cycles and policy subsystems. Don Mills, Ont., Oxford University Press. Imbeau, L. M., F. Petry and M. Lamari (2001). "Left-right party ideology and government policies: A meta-analysis." European Journal of Political Research 40(1): 1-29. Joppke, C. (1998). "Why liberal states accept unwanted immigration." World Politics 50(2): 266-+. Kelly, N. J. (2005). "Political choice, public policy, and distributional outcomes." American Journal of Political Science 49(4): 865-880. Laver, M. and J. Garry (2000). "Estimating policy positions from political texts." American Journal of Political Science 44(3): 619-634. Martin, P. L. and J. E. Taylor (2001). Managing Migration: The Role of Economic Policies. Global Migrants Global Refugees, Problems and Solutions. A. R. Zolberg and P. M. Benda. New York, Berghahn Books. Meyers, E. (2002). "The causes of convergence in Western immigration control." Review of International Studies 28(1): 123-141. Moses, J. (1994). “Abdication from National Policy Autonomy: What’s left to leave?” Politics and society 22(2), 125-148. Pandit, K. (1997). "Cohort and Period Effects in U.S. Migration: How Demographic and Economic Cycles Influence the Migration Schedule." Annals of the Association of 22 American Geographers 87(3): 439-450. Pearlmutter, T (1996). "Bringing Parties Back In: Comments on 'Modes of Immigration Politics in Liberal Democratic Societies'. The International Migration Review 30:1, 375-387. Sassen, S. (1996). Losing control? Sovereignty in an age of globalization. New York, Columbia University Press. Sassen, S. (2000). "Regulating immigration in a global age: A new policy landscape." Annals of the American Academy of Political and Social Science 570: 65-77. Stimson, J. (1995), "Opinion and Representation." The American Political Science Review. 89:1, 179-183. Schmidt, M.G. (2002). "The impact of political parties, constitutional structures and veto players on public policy" in Keman, H "Comparative Democratic Politics: A Guide to Contemporary Theory and Research, London, GBR. Sage Publicatons. Schmidt, M.G. (1996). “When parties matter: A review of the possibilities and limits of partisan influence on public policy.” European Journal of Political Research. 30. 155-183. Thielemann, E. (2004). "Does Policy Matter? On Governments' Attempts to Control Unwanted Migration." Working Paper 112, The Center for Comparative Immigration Studies, University of California, San Diego. Tjelmeland, H. and G. Brochmann (2003). Norsk Innvandringshistorie. I globaliseringens tid, 1940-2000. Norsk Innvandrinshistorie. K. Kjellstadli: 460 s. Wooldridge, J.M. (2005). ”Introductory Econometrics. A modern Approach.” Thomson South-Western. Aardal, B. (1999). Velgere i 90-årene. [Oslo], NKS-forl. Aardal, B. (2003). Velgere i villrede... : en analyse av stortingsvalget 2001. Oslo, Damm. Aardal, B. and H. Valen (1989). Velgere, partier og politisk avstand. Oslo ,, Statistisk sentralbyrå. Aardal, B. and H. Valen (1995). Konflikt og opinion. [Oslo], NKS-forl. 23