Explaining the Surge of the Populist Radical Right: A Time-Series Analysis of the Effects of Immigration and the Economy in Norway

Abstract

Populist radical right parties have become major forces in most Western democracies. Previous research has provided conflicting evidence on whether their electoral support can be explained by two structural developments: economic decline and increased immigration. Using time-series regression and almost 30 years of aggregated monthly polling data, we perform a novel test of the effects of economic decline and immigration on aggregate support for the Norwegian Progress Party. We find that the most beneficial time-periods for this party seem to be those of rising immigration and a booming economy. However, our findings also suggest that the effect of rising immigration is halted when the party holds government office. Thus, voter mobilization based on anti-immigration messages may represent a challenge for the Norwegian Progress Party and potentially other such parties going forward as they may become victims of their own success.

Introduction

Populist radical right (PRR) parties have become major political forces in a number of Western democracies in recent years. Electoral research aiming to explain this development has long debated the explanatory power of two structural developments affecting Western democracies: rising immigration and economic turmoil. These processes are assumed to induce grievances and discontent in certain groups of the electorate from which PRR parties benefit (Golder, 2016; Rydgren, 2007). Despite much scholarly attention, however, empirical studies have so far produced conflicting results. Some studies find economic decline to boost support for the populist radical right (Jackman & Volpert, 1996), other studies find the opposite result (Arzheimer & Carter, 2006; Knigge, 1998; Rooduijn & Burgoon, 2018), while some studies find no relationship at all (Norris, 2005). As for the assumed effect of rising immigration, results seem somewhat more consistent, but evidence is still far from conclusive (e.g. Lubbers et al., 2002; Lucassen and Lubbers, 2012; Rydgren, 2008; Swank and Betz, 2003). Taken together, it is unclear to what extent and how aggregate support for PRR parties is affected by rising immigration and economic developments.

In this article, we make two contributions to this research literature. First, previous research is dominated by cross-sectional research designs (e.g. Lucassen and Lubbers, 2012) or time-series cross-sectional (TSCS) analyses with limited longitudinal variation (e.g. Golder, 2003). As discussed by Van der Brug et al. (2005), such approaches may not allow for sufficient variation in the structural conditions that allegedly caused the rise of these parties, to enable an analysis of their different electoral fortunes across Europe. We put earlier conclusions to the test by using monthly time-series data, including aggregated polling data, from Norway covering almost 30 years. Our detailed longitudinal data provides us with far more variation in our determinants of interest. Moreover, by analysing within-country longitudinal variation we escape the implicit assumption in cross-sectional analyses highlighted by van Erkel and van der Meer (2016), namely that people evaluate structural conditions based on the performance of other polities at the same time point. Instead, we assume that people primarily evaluate performance longitudinally, that is, compared to historical levels within the same polity. Relative change in immigration or in economic conditions are likely to be noticed by people, both by those who are personally affected in some way, but probably also by others through media coverage. To our knowledge, this is the first study to test the explanatory power of structural demand-side determinants on aggregate support of PRR parties longitudinally over such a long period with such short time intervals between the data points.

Our second contribution is to investigate if PRR parties’ abilities to mobilize voters based on economic- and immigration-related grievances are hampered once they enter government. We examine the Norwegian case because it is an example of a European country where there has been a successful PRR party over a long period of time that has come to power not only in the legislative arena, but also in the executive, namely the Progress Party (FrP). Insights from economic voting literature suggest that any gains these parties may make from a position outside of government from rising immigration or changes in the economy are at least weakened when they become responsible for these outcomes themselves (Lewis-Beck & Stegmaier, 2013). Yet, by sustaining a populist profile, even from a position of government, it is conceivable that they could avoid the accountability mechanisms to which other parties often are subject. By assessing whether voters hold PRR parties in government accountable not only for economic performance, but also immigration, our analysis adds to previous research on electoral accountability.

We continue in the next section by examining literature on the structural determinants of the populist radical right and perspectives on political accountability. We then describe the Norwegian case, the data and the methods used. Our analysis reveals that economic developments do affect the fortunes of the Progress Party: interestingly, the party does well during times of economic growth. Rising asylum-based immigration is positively related to support for the Progress Party as long as the party forms the opposition. Once the populist radical right enters into government, an accountability mechanism appears to set in and the party no longer benefits from rising numbers of asylum seekers. Therefore, voter mobilization based on anti-immigration messages may represent a challenge for such parties going forward as they may become victims of their own success.

Structural Determinants of PRR Support

Theoretical Perspectives

Demand-side theories of electoral support often take a number of societal macro-structural processes as theoretical starting points. These processes are assumed to induce grievances and discontent in certain groups of the electorate from which specific party families benefit (Golder, 2016; Rydgren, 2007).¹

One strand of research links support of PRR parties to changes in the socioeconomic structure of European democracies, and in particular to the shift from industrial to post-industrial economies (e.g. Betz, 1994). A crucial assumption behind this argument is that this broad economic shift has benefitted people unequally. While middle-class people have benefitted from safe jobs in expanding sectors of the economy, traditional blue-collar workers have remained in labour market positions that are among the hardest hit in times of economic distress. As social democratic parties have shifted their focus away from blue-collar workers and made middle-class voters their primary electoral base, PRR parties have gradually picked up support in the lower socio-economic stratum (Gingrich & Häusermann, 2015; Oesch, 2015). Today, PRR support is typically high among young men, with lower education who are either unemployed, self-employed or a manual worker (see Golder, 2016 for a review). In short, a core constituency of the populist radical right may be disproportionally affected by an economic downturn, which may in turn heighten the economic grievances that underlie support for this party family (see e.g. Mudde & Kaltwasser, 2018 for a discussion of this argument).

One could also argue that even a fear of loss of status or of economic security could have a similar effect. It may be that an economic downturn reminds people of their economic insecurities, and thus produces an effect even among voters whose economic condition is not directly affected. This notion is supported by research showing that subjective economic insecurities shape political attitudes in systematic and significant ways (Hacker et al., 2013). A recent case study even demonstrates that expectations of status decline can be more important for PRR support than an actual experienced status decline (Im et al., 2022). In other words, not only absolute deprivation, but also fear of personal economic distress could explain a rise in support for the PRR in troublesome economic times. Relatedly, an economic downturn could also fuel political distrust and resentment (Erkel & van der Meer (2016), which in turn could drive voters to PRR parties.

Others, however, have argued that the populist radical right in Europe actually does well in good economic times, for two different reasons. First, people may be more willing to experiment with untested political alternatives when they feel economically secure (Bjørklund, 2007). Economic prosperity may therefore induce support for party families with limited governing experience, such as the populist radical right, while voters might turn to more experienced governing parties in periods of economic insecurity. Rooduijn and Burgoon (2018) find that people who are struggling financially are more likely than those who are well off to support the far right. However, this effect mainly manifests itself in good economic times. They attribute this to “risk aversion”; voting for the far right is seen as less risky when the economy is in good shape. Second, good economic times may shift the political agenda in favour of PRR parties. While the economy is likely to dominate the political agenda under conditions of economic recession, volatility, and economic underdevelopment (Singer, 2011), good economic times tend to increase the saliency of socio-cultural issues such as immigration (Van der Brug et al., 2015). As socio-cultural issues are at the heart of radical right ideology (Mudde, 2007) and tend to mobilize potential PRR voters, PRR parties may do better in good economic times.

A second but related strand of research moves beyond economic determinants to include an additional structural change that is dominant in contemporary European politics: increased immigration. Individual-level studies of the PRR make it clear that public reactions to and resistance against immigration are beneficial to the fortunes of these parties (e.g. Mughan and Paxton, 2006; Van der Brug et al., 2000), and some even see nativism as a necessary condition for radical right electoral success (Ivarsflaten, 2008). One body of research emphasizes the cultural element of anti-immigration sentiments. From this perspective, PRR parties benefit from voters’ rejections of post-materialistic values and the spread of nativist attitudes. Norris and Inglehart (2019) refer to this as the “cultural backlash thesis”. Other studies emphasize the possible economic threat that immigrants represent to native citizens. According to the “ethnic competition thesis”, voters turn to “the new radical right because they want to reduce competition over scarce resources such as the labor market, housing, welfare state benefits, or even the marriage market” (Rydgren, 2007: 250). With respect to the labour market, immigration is often assumed to reduce the relative earnings, and possibly the employment prospects, of (manual) workers with similar skills as the immigrants (Borjas, 2003).

Both the cultural and the economic argument provide us with a reason to expand on the economic perspective by including immigration as a second structural determinant of support for the populist radical right. Moreover, the ethnic competition thesis suggests that the combination of increased immigration and a declining economy would be particularly beneficial for PRR parties.

Empirical Status and Limitations in Previous Research

Despite significant theoretical attention, empirical studies assessing the relationship between these macro-structural conditions and aggregate support for the populist radical right have so far produced conflicting results (see Arzheimer, 2018; Golder, 2016; Rydgren, 2007 for reviews of the literature). Most studies assessing the effects of economic developments have focused on unemployment rates. Some find that comparatively higher levels of unemployment boost support for PRR parties (Jackman & Volpert, 1996), whereas some others find the opposite pattern to be true (Arzheimer & Carter, 2006; Knigge, 1998), while a last group of studies find that unemployment rates have no effect at all (Lubbers et al., 2002; Norris, 2005). Studies focusing specifically on economic crises have found that PRR parties often benefit from such dramatic events (Funke et al., 2016), but at the same time stress that the political consequences of economic crises vary considerably between regions and depend on the length of the economic downturn (Lindvall, 2014, 2017). Recently, a number of studies have investigated the impact of economic globalization on voting behaviour and the Brexit referendum. Some of these studies have used exposure to Chinese import to identify exogenous variation in economic hardship and find that stronger import shocks lead to an increase in support for the populist radical right (e.g. Colantone and Stanig, 2018).

Findings seem to be somewhat more consistent, although far from conclusive, for the effect of immigration (see Amengay and Stockemer (2019) for a review). While some studies find large-scale immigrant polities to demonstrate greater support for the populist radical right (Knigge, 1998; Lubbers et al., 2002; Swank & Betz, 2003), others find that the volume of immigration has no effect (Arzheimer & Carter, 2006; Lucassen & Lubbers, 2012; Norris, 2005; Rydgren, 2008). Moreover, results are hard to compare due to the use of different measures for immigration across studies.

Economic changes and immigration are sometimes seen as working together in nudging people towards PRR parties. A situation of high immigration and high unemployment is perhaps the clearest scenario of ethnic competition for scarce resources (Arzheiemer, 2018). When the economy is declining and working class jobs are taken over by immigrants, voters may vent their frustration through support for a PRR party. This narrative finds some support in the research literature. Golder (2003) shows that the combination of an economic downturn and increased immigration is particularly beneficial for the populist radical right. Strömblad and Malmberg (2016) reach a similar conclusion in their study of Swedish voters. They find that increased exposure to visible minorities leads to an increase in support for the Swedish Democrats if the district level of unemployment is high. Arzheimer (2009) on the other hand, finds the opposite pattern in a comparative study of Western Europe; and Norris (2005) finds no interaction effect at all.

Contradictory evidence, however, need not mean that the volume of immigration and economic developments are unrelated to the fortunes of the populist radical right. First, most previous studies have either relied solely on cross-sectional variation (e.g. Lucassen and Lubbers, 2012) or used time-series cross-sectional research designs with few time points and long time intervals (e.g. Golder, 2003).² As argued by Van der Brug et al. (2005: 567), these studies may have been unable to capture the true effects of sociostructural variables because “these developments are so similar in all EU countries that they cannot account for the large differences in electoral support for anti-immigrant parties that we find in these countries.“ It may therefore still be the case that “such societal developments are at the heart of rise of anti-immigrant parties.“

Second, as argued by van Erkel and van der Meer (2016), when omitting longitudinal variation from the research design, one implicitly assumes that people evaluate economic performance and rising immigration cross-sectionally; that is, based on the performance of other polities at the same time point. However, citizens may well evaluate structural conditions longitudinally—that is, “in comparison to the performance that they expect in their own country” (van Erkel & van der Meer, 2016: 178). In line with this argument, van Erkel and van der Meer (2016) find strong longitudinal effects of macroeconomic performance on political trust in Europe. To our knowledge, our study is the first to examine how support for a European PRR party is affected by within-country changes in economic performance and immigration over almost a 30 year period with monthly observations. Such detailed longitudinal data should provide the opportunity to determine whether these macro-structural developments really are at the heart of this party family.

Finally, we investigate whether the impact of the economy and immigration on PRR electoral support changes when a populist radical right party enters into government. The central finding in the economic voting literature is that voters ensure electoral accountability by rewarding or punishing sitting governments based on their economic performance (e.g. Lewis-Beck and Stegmaier, 2013). Applying this logic to PRR parties would suggest that any gains that these parties make from rising immigration or economic setbacks are weakened when the party itself is responsible for these outcomes. Yet, it is not self-evident that retrospective accountability mechanisms apply equally to PRR parties. When in government, these parties often pursue what Albertazzi and McDonnell (2005) call ‘one foot in, one foot out strategies’ aiming to balance vote- and office-seeking ambitions. In other words, rather than assuming responsibility for current policies, PRR parties often sustain their populist message from a position of government and might thereby escape normal accountability mechanisms.

Hypotheses

Taken together, the literature points to two main structural determinants behind the electoral fortunes of PRR parties. Economically oriented studies seem to agree on economic hardship as a key explanation for support of the populist radical right, but diverge with respect to the direction of the relationship. Thus, we outline two mutually exclusive hypotheses with regard to the relationship between economic conditions and support for the Progress Party:

H1a

Economic decline leads to rising support for the Progress Party.

H1b

Economic decline leads to declining support for the Progress Party.

The second main driver identified in the literature behind support for PRR parties is immigration. We expect rising immigration to benefit the populist radical right, but we acknowledge that a reasonable argument could be made for an opposite effect. The latter argument is based on the “contact hypothesis” (Allport, 1954). Interaction between immigrants and natives, as a result of actual immigration, could lead to decreased prejudice and lower support for anti-immigration parties. Thus, we also outline two mutually exclusive hypotheses with regard to the relationship between the volume of immigration and support for PRR parties:

H2a

A rise in immigration leads to rising support for the Progress Party.

H2b

A rise in immigration leads to declining support for the Progress Party.

Based on the previous discussion, we also expect that the combination of economic decline and rising immigration bodes well for the Progress Party. Specifically, we hypothesize a positive interaction effect, meaning that the positive effect of immigration is stronger when the economy is in decline and visa-versa.

H3

There is a positive interaction effect between economic decline and a rise in immigration on support for the Progress Party.

Finally, literature on economic voting tells us that the effects of economic developments on voting hinges on whether a party is in government or not (e.g. Lewis-Beck and Stegmaier, 2013). If we find evidence that a declining economy benefits the populist right (H1a), economic voting theory suggests that this effect will weaken or even be reversed when the party enters into government. A similar logic could be at play with respect to rising immigration numbers (H2a). At the same time, PRR parties could escape typical accountability mechanisms by pursuing ‘one foot in, one foot out’ strategies (Albertazzi & McDonnell, 2005). It is also unclear whether and how the impact of the economy and immigration will change when PRR parties enter into government if the support for these parties actually is driven by good economic times (H1b) and less immigration (H2b). We thus refrain from formulating specific hypotheses and instead formulate an open research question.

RQ1

How is the relationships outlined in H1a/H1b and H2a/H2b affected when the Progress Party enter into government?

The Norwegian Case

The Norwegian Progress Party (Fremskrittspartiet [Frp]) was founded in 1973 during what has been labelled the “second wave” of far-right mobilization in Western Europe. It entered parliament that same year, but did not become a major player in Norwegian politics until the late 1980s, when immigration became a political issue, and the party started to espouse anti-immigrant views. The Progress Party remained in opposition until 2013 when it joined a governing coalition for the first time with the Conservative Party (Høyre). The party remained in the cabinet (with varying coalition partners) for a little over 6 years, until January of 2020. It is fair to say that Frp is moderate when compared to other parties traditionally defined as PRR, and some have even argued that it is somewhat of a borderline case (Mudde, 2007). Yet the three core elements in right-wing populism—nativism, authoritarianism and populism—are apparent in Frp, and through its outsider position in the Norwegian party system it has historically played a role that is typical of PRR parties (Ivarsflaten, 2008; Jupskås, 2015). In the concluding section, we discuss the generalisability of our findings.

Data and Methods

To measure the support for the Progress Party we use a dataset of monthly Norwegian polling results from seven polling institutes, covering almost 30 years from January 1991 to August 2019.³ We use the average level of support (across polling institutes) for the Progress Party at each time point as our dependent variable.⁴ By pooling polls together, in all other months, we overcome an important shortcoming of single polls, that is, imprecision due to sampling error (Jackman, 2005).

As for independent variables, we rely on two different monthly measures of the state of the Norwegian economy. First, we use the monthly unemployment figures that are published by Statistics Norway. Unemployment is defined as the percentage of the workforce that is not employed, but is actively seeking employment. Unemployment levels are often found to affect election outcomes (Lewis-Beck & Stegmaier, 2013) and are a standard measure in the literature on PRR support (Arzheimer, 2018). As discussed in the theoretical section, not only might PRR parties benefit from the discontent among those who are directly affected by rising unemployment levels, but it may also create a general feeling of economic insecurity from which the party benefits.

Second, we use data on developments at the Oslo Stock Exchange. Stock rates are a real indicator of the value of Norwegian companies, and therefore of the strength of the Norwegian economy as a whole. In addition, stock market performance may be influenced by a general sense of optimism or pessimism in society with regard to economic performance. In sum, stock market prices are a good indicator of the health of the private sector economy (and in part of the public sector economy as well, since some major companies in Norway are partially owned by the state). Stock rates can fluctuate quite a bit on a short-term basis, so there certainly is enough variation. The OBX Total Return Index (OBX) on the Oslo Stock Exchange lists the 25 most liquid companies on the exchange.⁵ We used daily index rates to calculate monthly averages.⁶

As for immigration, we use monthly official figures for the registration of newly arrived asylum seekers in Norway.⁷ These are the raw numbers of people applying for asylum. Quite a few of these people will not be granted residence, but the figures are nonetheless reflective of the varying numbers of people arriving in the country as immigrants seeking shelter. Brochmann et al. (2010: 244) find that asylum seekers and refugees were much more visible than other groups of immigrants in official documents from the 1980 and 1990s, and that immigration-related political debate in this period was largely connected to questions concerning these groups. A more recent content analysis of the immigration debate in Norwegian newspapers for the period 1970–2016 also shows that asylum seekers and refugees are much more frequently discussed in the media than foreign workers (Hovden & Mjelde, 2019). Figures related to asylum seekers have continued to be reported in the Norwegian media and sometimes form the basis of a general sense of crisis with respect to immigration. That was certainly the case in Norway and in much of Europe in the autumn of 2015, when asylum seekers arrived in record numbers. Specifically, while the number of first-time asylum applicants in the European Union member states overall increased by 123% from 2014 to 2015, it grew in Norway by 179% in the same period (Eurostat, 2016). Other forms of immigration to Norway are not politically contentious in the way that asylum immigration is. We standardize our main independent variables so that regression coefficients can be directly compared.⁸

A key premise, when using measures of real world changes in the economy or in the inflow of asylum-seekers to explain voter attitudes and behaviour, is that voters are able to perceive these changes. A rise in unemployment, for instance, only directly affects a small segment of the population: those that actually lose their jobs. They do not constitute a large enough share of the electorate to have a measurable effect on the dependent variable, support for the Progress Party. If, on the other hand, a rise in unemployment is widely reported in the media, so that the public becomes aware of it, it seems much more plausible that this could affect voter attitudes and behaviour.

In order to look into this we have conducted a media search, using the Retriever database, of four Norwegian national print media; three national newspapers: VG, Aftenposten and Dagbladet, as well as the largest news wire service in Norway, NTB. The database covers the entire period for which we have data, so we are able to construct media-data on a monthly basis to match out other data. We conduct two truncated searches. The first tries to capture media coverage of arriving asylum-seekers, by searching for “asylum*” and/or “refugee*”.⁹ The number of monthly media uses of these words ranges from 142 to 1950, with a mean of 348. Clearly, these are common words in Norwegian print media.

The second search is for “unemploy*”,¹⁰ which would capture the words “unemployment”, “unemployed” and the like. These words are used between 37 and 347 times on a monthly basis; the mean is 134.

Does the varying media coverage of these topics reflect real world changes? In short, yes. The correlation between asylum numbers and media coverage (of asylum and refugees) is a Pearson’s r of 0.53. These is also a correlation the following month (0.49), or if you lag the asylum numbers two months (0.36), but the strongest correlation is on the same month.

Coverage of unemployment is also highly correlated with actual changes in unemployment the same month (0.53) or the following month (0.54) and the month after that (0.54). We have also correlated unemployment coverage with the stock market OBX index. The correlation is, as one would expect, negative the same month (− 0.46), the next month (− 0.46) and with a 2 month lag (− 0.45).

These results clearly indicate that the real changes in asylum numbers, in unemployment and in the stock market does receive immediate coverage in Norwegian media. This is a key mechanism that could make voters aware of these events shortly after they happen, which is clearly necessary for these events having any effect on voters’ support for the Progress Party. The fact that these events are covered as they happen makes it reasonable to look for short term voter effects as well as long term effects.

A key independent variable is the inclusion of the Progress Party in government. A simple and straightforward means of measuring that would be to include a dummy variable indicating which months the party has held seats in government. However, one would not expect the party to be held as strongly to account for its policies in the first month of the governing period as it would be subsequently (Narud & Valen, 2008). We therefore model accountability for government participation as something that grows with time.¹¹ The variable has the value 0 when the party is not in government and 1 for the first month in government, followed by a gradual increase, the rate of which diminishes over time.¹² The Progress Party entered into a governing coalition in October 2013 and this remains the case through the end of our time series (August 2019).

As is common in times-series studies of party support, we also model the influence of relevant events in the time period at hand. When deciding which events to include, we use Bytzek’s (2011) three criteria for inclusion: that the event under consideration is not directly produced by the dependent variable (exogeneity), that they are linked to the political world in the eyes of citizens, and that they receive high publicity. We aim for a limited use of event variables, selecting only major events that we suspect will affect our analysis (see Online Appendix Figure A1). Failing to control for events that are large outliers in our data may otherwise have distorting effects on the estimation of our models.

We also control for the months when parliamentary elections are held. We do not have polling averages for those months, but have included actual election results instead. This controls for any bias that there may be in polling averages versus actual election results. The results below indicate that the Progress Party has somewhat less support in elections than in the polls before and after the election; the difference is a little over 1% point. We do not know whether this difference is explained by errors in the polls or by an actual change in support for the Progress Party in the months when elections are held.

We analyse these data by estimating a number of autoregressive distributed lag (ADL) models. These models include lagged values of both the dependent and the independent variables. A bivariate ADL model (with two lags) may be written as follows:

$$Y_{t} = Y_{{t - 1}} + Y_{{t - 2}} + \beta _{0} X_{t} + \beta _{1} X_{{t - 1}} + \beta _{1} X_{{t - 2}} + \varepsilon _{t} .$$

Inclusion of a lagged dependent variable (Y_t−1,Y_t−2) corrects for the serial correlation that is often present in time series. Moreover, ADL models allow for exploring the way in which any changes in our determinants affect support for the populist radical right through inclusion of both contemporaneous (β₀X_t) and lagged (β₁X_t−1, β₁X_t−2) coefficients. This is important for our purposes given that the theory provides few clues with regard to the dynamics of the relationships between immigration, economic developments and support for PRR parties. Tests of any long-term relationships, or “cointegration” in error correction model (ECM) parlance, show that no long-term relationship between our determinants and our dependent variable exists.¹³ We therefore continue by analysing short-term effects. The exact lag length of our variables is chosen by the use of information criteria (Akaike information criterion (AIC) and Bayesian information criterion (BIC). Both the Ljung–Box and the Breusch–Godfrey Lagrange Multiplier tests for autocorrelation confirm that this setup filters out the autocorrelation in our series. As ordinary least square (OLS) regression is a consistent estimator of ADL models under such assumptions, we estimate our ADL models by means of OLS.

Results

We begin by plotting the monthly time series of our dependent variable, support for the Progress Party, and our three main explanatory variables for the period from January 1991 to August 2019 (Fig. 1). The largest spikes in support for the Progress Party occurred in the decade between 2000 and 2010, when support for the party rose to above 30% several times.

The arrival of asylum seekers in Norway began in earnest in the late 1980s, and the first large wave of asylum seekers came in the early 1990s, caused by the war in the former Yugoslavia. Since then, there have been several major waves of immigrants, the largest in connection with the 2015 migrant crisis.

The OBX price index has generally trended upward during the period studied, while unemployment—in addition to its short-term cyclical nature—rose sharply in the early 1990 and 2000s but was more stable in the 2010s.

Fig. 1

Support for Frp, asylum seekers, OBX price index and unemployment rates, 1991–2019

The results of our time-series analyses are displayed in Table 1. In model 1, we report an ADL model of support for the populist radical right in Norway, estimated with our three main predictor variables as well as a number of control variables.

Starting with our two economic predictors, we find that a booming stock market bodes well for the fortunes of the Progress Party. There is indeed a short-term positive effect of changes in stock prices. This suggests that there is an immediate effect of good economic news on people’s propensity to support the Progress Party. A booming stock market is in itself a sign of economic optimism, and it may also result in positive news coverage and produce a general sense of optimism about the Norwegian economy. When that happens, support for the Progress Party increases. Moreover, we can see that unemployment levels are negatively related to support for the Progress Party. This result points substantively in the same direction: the Progress Party does well in good economic times (with low unemployment).¹⁴

A change in the number of asylum seekers from one month to the next, however, is not significant in this model. Nor do we find any evidence that the combination of an economic downturn and rising immigration is what drives voters into the hands of the populist radical right in Norway. That is, we have tested the possibility of an interaction effect between changes in asylum immigration and changes in our two economic indicators, respectively. None of these interaction effects were statistically significant, suggesting that the effect of immigration is not conditioned on economic developments and vice versa.¹⁵

Table 1 ADL models of support for the progress party

To what extent do these relationships change when the Progress Party enters into a governing coalition? Are they weakened once the party becomes responsible for setting government policy, or does the party benefit equally from these factors when holding the reins of government? We investigate this question by interacting our predictor variables with the variable capturing the inclusion of the Progress Party in government. The results are listed under models 2, 3 and 4 in Table 1.

There is a significant interaction term with regard to immigration in model 2. The interaction effect is negative, and the coefficient reaches statistical significance at the 5% level. Moreover, when including the interaction term, the contemporaneous effect of immigration turns significant and increases in magnitude. Substantively, this indicates that outside of government (party in government = 0), Frp gains support from monthly changes in immigration, but that the benefit it reaps from changes in immigration gradually evaporates when the party holds government office. This contemporaneous effect probably owes much to changing discourse and varying attention given to the issue. A sudden surge in immigration, for instance, may lead to media coverage and thereby increased public attention, which may in turn affect the fortunes of the populist radical right.

For ease of interpretation, the conditioned relationship is illustrated by the marginsplot in Fig. 2. As the figure illustrates, there is a positive relationship between changes in the inflow of asylum seekers and support for Norway’s PRR party when it is not in government (government responsibility = 0). When the party is in government, and especially when it has been in government for a while, the effect is substantially weaker and not significantly different from 0.

Fig. 2

Marginal effects of monthly asylum seekers (differenced)

As for our two economic indicators, we find no statistically significant interaction-effects. This means that the positive effect of stock market prices and the negative effect of unemployment rates on support for the Progress Party are unaffected by the party’s entrance into a governing coalition in 2013. In order to see if these results hold in alternative model specifications, we model holding government office binarily and in four cuts, respectively, rather than continuously as people may have a simpler understanding of accountability than our main model presumes (see Table A2 and Table A6 in the Online Appendix). This actually produces stronger effects of government participation than the approach used in Table 1, but other than that, the results are identical.¹⁶

Discussion and Conclusion

Recent elections in established democracies have attracted considerable attention to the causes of the electoral success of PRR parties. The purpose of this paper has been to investigate the explanatory power of two much discussed, yet empirically contested, demand-side explanations of this development: economic turmoil and increased immigration. The “popular” story of economic decline and rising immigration fuelling a rise in the populist radical right finds, at best, limited support in our analyses. This should perhaps not surprise us, given that parties such as the Norwegian Progress Party has been around since long before the financial crisis in 2008, and these parties have thrived in some very prosperous countries. We find that rather than economic decline, a booming economy appears to attract voters to the Progress Party. We can speculate about the mechanism that produces this effect. It could be that economic growth leads to a shift in the political debate away from the economy toward other issues, such as immigration, upon which the populist radical right is able to mobilize voters.¹⁷ It could also be that good economic times lead to a sense of optimism and a willingness to try “untested” political alternatives. In any case, the effect appears to be almost instantaneous and short term. In other words, the effect occurs before economic changes have had the time to affect the real lives of voters. We did find that the media covers these events as they happen. It is therefore reasonable to believe that the effect is mediated by media coverage or by the public discourse in the country.

A rise in the numbers of applications for asylum in Norway has a positive effect on support for the Progress Party, when the party is in opposition. This is also likely to be a mediated effect. People probably do not experience the inflow of asylum seekers in any other way than through media coverage, until some time has passed. This result supports the widely held view that populist radical right parties mobilise voters based on their resistance to immigration. While this party family certainly attracts voters based on other concerns or grievances as well, voter reactions to increasing immigration seem to be at the heart of the populist radical right’s appeal. In that sense, our study lends credence to individual-level studies that have argued that anti-immigration mobilization is a defining characteristic of this party family (e.g. Ivarsflaten, 2008).

Contrary to the approaches of most previous studies that have relied on static research designs or limited longitudinal variation, we have used detailed time-series data including almost 30 years of monthly observations. More variation in our predictors may be an important reason for why our findings deviate from null-findings in much of the previous work on the structural determinants of the populist radical right (e.g. Norris, 2005). Another reason may be that people primarily evaluate structural developments longitudinally, rather than cross-sectionally. This explanation squares well with recent findings on political trust (van Erkel & van der Meer, 2016).

Furthermore, the Progress Party benefits less from asylum immigration when in office. This happened despite the high salience of the immigration issue during the party’s time in government. The 2015 migrant crisis happened on the Progress Party’s watch. In the 2017 parliamentary election, immigration was a top issue for voters, in part due to the efforts of Progress Party politicians (Aardal & Bergh, 2018). In these circumstances, the Progress Party still benefits less from rising immigration. This finding speaks to recent research examining the extent to which PRR parties become “mainstreamed” when they assume office. Recent studies have found that inclusion into office in some instances have incited PRR parties to moderate their radical policy positions and populist rhetoric (Akkerman et al., 2016). Our findings suggest that such adjustments to the demands of national office may represent a challenge for these parties going forward as they may become victims of their own success. Voter mobilization based on an anti-immigration message could be so effective that these parties gain positions in government. At that point, their winning electoral formula may start to fail them.

Although increased ethnic heterogeneity and economic turmoil are by no means exclusive to Norway, our findings are likely to be context-sensitive. As a point of departure for future research, we believe that three factors are likely to play a role in explaining comparative variation: the welfare state context, the party system design and political supply dynamics. First, findings by Swank and Betz (2003) indicate that the impact of economic globalization on support for the populist radical right is conditioned by national welfare structures. They find that while the volume of foreign immigration boosts support for the populist radical right, regardless of welfare state design, economic globalization does not contribute to the support for the populist radical right where national systems of social protection are comparatively comprehensive. We might therefore expect negative economic consequences of globalization to be more beneficial for PRR parties in polities with less generous welfare systems than Norway’s. Recent findings from non-social-democratic welfare regimes support this argument (Colantone & Stanig, 2018).

Second, in an open multiparty system like Norway’s, PRR parties provide dissatisfied voters with not only a voice, but also parliamentarian representation, and in some instances even the political power that comes with executive office. This may link their electoral fortunes more closely to voters’ political preferences—on issues such as immigration and the economy—and less to political distrust and general system dissatisfaction than in closed party systems where these parties often have substantial support, but less representation and hence political influence.

Finally, it may be that the impact of increased ethnic heterogeneity and economic turmoil depend on political supply dynamics. For instance, the positive effect of rising immigration that we found in Norway when the Progress party formed opposition probably depends on the party’s ownership of this issue. If PRR parties’ issue ownership on immigration should be contested, through for instance the rise of new and more radical right-wing challenger parties, they may benefit less from increased immigration.

An important question for future research is to examine these and other comparative hypotheses concerning the winning formula of the populist radical right.