Youth , Precarious Employment and Political Participation in Hungary

Young Europeans’ political responses to the economic crisis have neither been uniform nor overly promising for the future of democratic Europe. We seek to identify potential causal relationships between young peoples’ employment status and choice of political participation (i.e. both traditional and non-traditional forms of political participation, as well as emerging alternatives). Although politicians and academics highlight that young people are increasingly disengaged from conventional politics, and papers have been published about different aspects of this topic, young peoples’ perspectives and generational differences are rarely taken into account simultaneously. In this paper we characterize the consequences of the economic and employment conditions of youth on political engagement. Our paper focuses on Hungary, which has struggled with youth unemployment. The paper involves secondary data analysis of cross-national surveys, involving six datasets (2004, 2006, 2008, 2010, 2012, 2015) from the European Social Survey (ESS). Results indicate that greater involvement and responsibility in the workplace increase political participation, whereas the impact of the other labour market indicators (unemployment, work flexibility) on political participation is not straightforward.


Introduction
Since citizens' political and social disengagement is strongly related to their perception of inequality in society (Loveless, 2013: 471), and higher levels of inequality reduce citizens' support for democracy (Krieckhaus et al., 2014: 145) the 2008 economic crisis challenged social cohesion, inclusiveness, and investment in national democracies.Economic insecurity has since risen for nearly all age groups, but youth have been particularly hard hit.Young citizens1 are most likely to be faced with unemployment and unstable career prospects, and thus feel alienated and disenfranchised from society.In this context, it is important to understand how unemployment and unstable working conditions are contributing to young peoples' political participation.
Following the 2008 crisis, the use of flexible, fixed-term contracts and alternative forms of temporary employment increased, leading to an increase in earnings-related risk and job insecurity for young people at the start of their professional careers.'Generation Y' is distinct from its predecessors in terms of the precariousness of their place in a society they struggle to enter (Bauman, 2012).This cohort is likely to experience increasing cynicism about work and systemic uncertainty about the future, with little to guide their expectations of what tomorrow will bring.Although we have seen crowds of young people protesting that 'their future has been taken away', we have little evidence about the effects of these forms of employment on political participation.
Individual economic status directly impacts political engagement, with economic status positively correlated to political engagement -from political interest to voting (Verba et al., 1979;Verba et al., 1995).Young Europeans' political responses to the economic crisis have neither been uniform nor overly promising for the future of democratic Europe.
We investigate the nature and extent of youth employment conditions and opportunities in Hungary, with a view to assessing the consequences of youth unemployment and poor employment prospects on political participation and engagement, and thus on the social cohesion and democratic legitimacy of Hungarian society.

Literature review 2.1. Changing forms of political participation
Scientific discourse on the participation of young people has oscillated between two extremes.Young people are often described as the apolitical harbingers of an incipient 'crisis of democracy' (Bessant, 2004;Furlong and Cartmel, 2007).Discussions revolve around declining political interest, falling participation, and low turnout at elections in Europe.On the other hand, young people are also heralded as innovators of politics, as creators of sophisticated new forms of participation, INTERSECTIONS.EEJSP, 3 (1): 120-146.
especially online (Coleman, 2006).It may well be the case that the upcoming generation is simply more interested in inventing novel forms of political participation (Phelps, 2004;2012).The argument has been put forward that citizens today, especially younger generations, seem to prefer to participate in the extra-parliamentary realm.
The emergence of new forms of political participation presents a theoretical challenge.Stolle and Hooghe (2011) argue that it might reduce age-and gender-based inequality.Using the Political Action Survey, as well as the European Social Survey, the authors observe that gender differences with non-institutionalized participation have been substantially reduced, and in some cases even reversed, and that women tend to be more active in this regard than men.Younger people also clearly have a preference for non-institutionalized forms of participation.Based on data from the ISSP survey, Marien, Hooghe and Quintelier (2010) found that non-institutionalized forms of participation increase inequality in terms of education, but strongly reduce or even reverse gender and age inequalities.As such, both institutionalized and noninstitutionalized forms of participation have specific (dis)advantages from the perspective of preserving equal access to democratic decision-making procedures.
We expand the study of political participation to include not only traditional but also non-traditional forms of political engagement and prospective alternatives.Based on earlier findings of Oross and Szabó (2013;2017), we differentiate three participation categories.Since electoral participation ('voted in last national election') is the most important form of political participation within the Hungarian context, we distinguish it as the first form of political participation.
'Traditional' forms of political participation refer to participation in political organizations (political parties, unions), as well as forms of participation related to these organizations (such as campaigning, participation at meetings, wearing the symbols of these organizations, etc.).
'Direct' forms of political participation are those that require personal involvement but do not require long-term commitment on behalf of actors (e.g.direct forms of protest such as sit-ins, blockades, and other expressive and symbolic acts).Direct forms of political participation require few resources, are low risk and require low levels of commitment.They include the signing of statements, petitions and initiatives.
As for the link between employment status and political participation, the literature (Lorenzini and Giugni, 2012: 333-335) focuses on the capabilities and competencies useful for political participation that can be acquired in the workplace through work-related experience (Brady, Verba and Schlozman, 1995;Pateman, 1970;Schur, 2003;Sobel, 1993).In this regard, we can distinguish between the spillover model and the civic skills model (Adman, 2008).The spillover model assumes that participation in the workplace offers individuals opportunities to learn how to participate and to develop roles related to social and political participation (Pateman, 1970;Sobel, 1993).According to this model, 'participation supports participation'; that is, involvement and responsibility in the workplace impacts political participation (Sobel, 1993).Moreover, participatory mechanisms at the workplace are an opportunity to develop a sense of political efficacy (Carole Pateman cited in Adman, 2008: 118).The civic skills approach argues that people participate when they have resources (e.g.time, money, civic skills), when they have a psychological predisposition towards engagement (e.g. an interest in politics) and when they are recruited (e.g. by voluntary associations or individuals) (Brady, Verba and Schlozman, 1995: 271).Research into the political participation of unemployed youth found that this cohort were more dissatisfied with politics (Bay and Blekesaune, 2002;Bynner and Ashford, 1994).However, unemployed and employed youth participate in voting to a similar extent (Banks and Ullah, 1987;Bynner and Ashford, 1994).
Current research on the topic (Lorenzini, 2013: 182) has pointed out that in terms of the effects of unemployment on political participation, we cannot assume that unemployment represents financial, social, and psychological deprivation for all unemployed youth.It is also difficult to ascertain whether employed people are different from unemployed people, ceteris paribus (Schur, 2003).Moreover, other studies have questioned the direction of the relationship between employment and political participation, as well as the very existence of the relationship itself.Cohen and Vigoda (1999), for example, found that political participation can explain attitudes and behaviors in the workplace, reversing the relationship between the two variables.To solve this puzzle, Adman (2008) tested the effect of work on political participation using panel data.The effect found through a cross-sectional analyses does not hold when one takes into account the temporal ordering of events (first being involved in a specific workplace setting, then participating politically).In terms of new forms of political participation, employment status has only a limited impact on political participation, affecting only consumer activities (Lorenzini and Giugni, 2012).
Increasing social inequalities have mobilized young people, from Southern Europe to Northern America.Both the 'Indignados' and the 'The Occupy Wall Street' movement brought the problem of widening social inequality into razor-sharp focus (Castaneda, 2012: 10;Hickel, 2012).Whereas there has been scholarly debate about the specificities of these movements as political responses to the economic crisis, few studies have analyzed youth political participation in relation to employment status.

Labor market flexibility and employment relations
This study brings together two fields of research: political participation -mostly investigated by political scientists, and employment relations -mostly analyzed by sociologists and economists.This requires greater elaboration of the independent variables from the perspective of employment.
Employment relations are typically investigated at the individual (micro) level, and are considered important features of social and political integration.Individual employment relations are embedded into the organizational structure of the labor market.While the labor market operates at a macro level, its consequences for individuals (e.g.being employed or unemployed, working under more or less flexible employment conditions) appear at the individual level.In this way, both macro conditions and regulations create differences at the micro (individual) level.Individuals in society are affected in an unequal way by (macro-level) organizational and institutional arrangements, which make employment relations one dimension of INTERSECTIONS.EEJSP, 3 (1): 120-146.social stratification.This is not a new but a rather well established concept (Baron, 1984;Kerckhoff, 1995).
One main feature of these processes is the transformation of standard employment relations into non-standard ones, increasing the flexibility of working conditions.This transformation has been the subject of sociological and labor market studies since the 1980s (Boyer, 1988), which expanded around the turn of the century (Strath, 2000;Kalleberg, 2000).Authors claim that labor market flexibility is part of the emergence of risk society, but the danger of such risks and precarious employment, a typical outcome of labor market flexibility, differs for individuals.Consequently, flexible labor conditions increase social inequality.This process has been ongoing for a few decades and has shown how the risk of unemployment and also the risk of being precariously employed is distributed unequally among individuals, increasing polarization (Breen, 1997;Kalleberg, 2011).How social risks arise, how labor market flexibility endangers social safety, and how social inequalities consequently emerge is particularly evident in post-communist societies where full, standard employment conditions disappeared for a significant proportion of the labor force after the collapse of socialism (Cazes and Nesporova, 2003).
Although the direction of causality is frequently debated, this study claims conceptually that: 1) employment relations -measured here using the incidence of unemployment and the features of precarious employment -are distributed unequally among individuals; 2) unemployment and precarious employment weaken the social integration of individuals (not investigated here); 3) unemployment and precarious employment have negative consequences on behavior (e.g.political participation) and this latter relationship may be analyzed by defining various forms of political participation, as outlined in the paper.
There are conceptual grounds for selecting indicators for employment relations based on the literature.'Unemployment' may not need much explanation in terms of social and political (dis)integration, but see Gallie et. al (1994).Current unemployment matters most, but earlier experiences with unemployment can also have a detrimental impact.In terms of 'precarious employment', the form of any contracts is a crucial indicator, as fixed-term contracts increase employment uncertainty (Schömann et al., 1998).Standard employment refers to jobs with an eight-hour daily work load.Part-time jobs typically used to be taken by women (who could thus combine work and child care) but labor market flexibility has increased the variety of working time arrangements for all employees, with attendant consequences on social life (O'Really et al., 2000).Finally, research indicates that a low level of job autonomy is also a characteristic feature of precarious employment (Kalleberg, 2011;Letourneux, 1998)., 3 (1): 120-146.

Case selection
In this paper we focus on Hungary, which has struggled with youth unemployment, and whose situation is typical of semi-peripheral EU countries.
1.After the regime change in 1990, more than one million people lost their jobs.This caused significant, almost unsolvable problems in the labor market (Laky, 1996).By the end of the 1990s employment stabilized at a persistently low level (Gazsó and Laki, 2004;Gazsó, Laki and Pitti, 2008) and was accompanied by a longterm unemployment rate averaging six percent.Regarding their integration into the labor market, four social groups were most affected: young people, people with a low level of education, elderly employees, and Roma.
2. The global financial and economic crisis (2008)(2009) exacerbated the already fragile situation (Szabó, 2013).In terms of macro-economic performance, Hungary fared worse than the EU28 average, but Hungary is not considered to be a crisis country in the way that Italy, Spain, and Greece are.After the 2008 economic crisis, youth unemployment remained 'just below' the EU28 average in Hungary2 , which is why it can be considered a typical, semi-peripheral EU country.
3. After the elections in 2010, the new government adopted a number of measures that resulted in a substantial improvement in the employment statistics (Messing, 2012).Early retirement was restricted, rules relating to disabled pensioners were modified, and the government launched a public works program (Koltai and Kulinyi, 2013).Despite the fact that the statistics had undoubtedly improved by 2013 in comparison with the EU (among 20-64 year olds) the labor force participation rate of Hungarians was still the fifth lowest (63.2 per cent; the lowest was in Greece at 53.2 per cent)3 .The long-term (more than 1 year) unemployment rate was 50.4 percent, 3.7 percentage points worse than the 2012 figure published by the Hungarian Central Statistical Office4 .Those excluded from the labor market have huge difficulty in finding their way back, since the average duration of unemployment rose from 17.6 months to 18.0 months in 2013.It is extremely problematic that the youngest segment of potential employees (15-24 years old) are faced with unfavorable employment and unemployment conditions: their employment rate only reached 19.8 percent (and unemployment rate 27.2 percent) (KSH, 2014).
4. As for changes in participation trends among the whole population, the level of participation in non-electoral forms of political participation compared to electoral participation is still low; electoral participation is 2.5 to 3 times higher than the most preferred other form of participation (Kern and Szabó, 2011: 22).The level of nonelectoral participation is low in Hungary and the difference between young people and INTERSECTIONS.EEJSP , 3 (1): 120-146

Employment rate
Participation rate adults in this regard is high because young Hungarians take advantage of very few forms of non-electoral participation (Oross and Szabó, 2016).Unlike young people in crisis countries, young Hungarians were not mobilized by recent movements that were a political response to the economic crisis.

Research questions
We seek to identify potential causal relationships between the increase in labor market inequalities for youth and their political attitudes and choice of political participation (e.g.expanding forms of both traditional and non-traditional forms of political participation, as well as emerging alternatives).Beyond considering youth unemployment we focus on the use of flexible, fixed-term contracts and alternative forms of temporary employment because young people are typically employed in these ways (Bertolini, 2012).Since there is little evidence about the impact of unstable work conditions on political participation, we seek to reveal if young people employed in flexible work participate less, or are rather more active in new forms of political participation.
In light of the topic of this paper -the relationship between labour market inequalities and political participation -the employment situation in the country is a possible contextual effect, having an influence on activity and behaviour.Based on official national statistics, Figure 1 provides information about the employment rate, participation in the labour market, and unemployment in Hungary for the period 2004-2015.Figure 1

Hypotheses
1. 1a, Based on the civic skills approach we expect unemployed youth, all other things being equal, to be less likely to participate in elections than employed youth.
1b, Although young Hungarians have not been mobilized by recent movements, we suppose that the rise in social inequality and lack of space in institutional politics for their claims has increased the likelihood of traditional and direct participation of unemployed young people compared to their employed peers.
2 2a, On the basis of the spillover model we assume that flexible, fixed-term contracts and alternative forms of temporary employment generate less participation in the workplace and offer fewer opportunities for learning how to participate.Considering the civic skills approach we argue that young people with flexible, fixedterm contracts have less resources with which to participate.We therefore expect these young people to be less likely to participate than employed youth.
2b, Following the 2008 economic crisis, we may expect renewed interest in political issues and a revival of political activism.This kind of 're-politicization' should be especially pronounced for young people with flexible, fixed-term contracts and alternative forms of temporary employment as they have more civic skills and social contacts than unemployed youth, can organize themselves, and are likely to be affected strongly by the negative externalities of the crisis (e.g.uncertain future prospects).In particular, we expect these young citizens to be active in direct forms of political participation.

Data and methods
In this paper we employ European Social Survey data from rounds 2-7.Thus, our data cover roughly one decade from 2004-2014.Round 1 was omitted because one of the predictor variables used in the analysis (job autonomy) was not available.Data about Hungary from these six rounds were merged, leaving approximately eleven thousand observations.A process of pooled data analysis was carried out to allow the investigation of changes over time by taking the level of significance in temporal variation into account.Design weights were employed for the data, as was a process of age selection; only respondents aged 15-65 were investigated.
It is also worth mentioning that the fieldwork for the Hungarian surveys did not match perfectly the official timing of ESS data collection rounds; the Hungarian survey was delayed three times in the period between 2004 and 2014.The exact timing of collection of Hungarian data may be important as concerns interpreting and understanding the results, particularly in light of the political events in the country around the time of the survey period.These events involved elections for the national or the European parliament, and other relevant political events that may have had an impact on political activity.

Dependent variables
Turning to measurement, the main dependent variable is political participation.We conceptually distinguish between three forms of participation, as outlined above in the theoretical section; this distinction also appears when defining the variables.
Voting is a dummy variable used to indicate whether (1) or not (0) a respondent participated in the national parliamentary elections (for this variable, 18 is the lower age limit, in line with the related Hungarian legislation).The second form is based on 3 similar questions from the ESS core questionnaire: -Signed petition in last 12 months -Taken part in lawful public demonstration in last 12 months -Boycotted certain products in last 12 months 8 Different attempts at conceptualization have engendered an emerging methodological consensus according to which research that focuses on explaining political participation should seek to identify and cluster different forms of participation."Traditional" forms of political participation include participation in elections and in political organizations (political parties, unions), as well as forms of participation related to these organizations (such as campaigning, participation at meetings, wearing the symbols of these organizations, etc.)."Collective" or "direct" forms of political participation are those that require personal involvement but do not require long-term commitment on behalf of actors (e.g.direct forms of protest such as sit-ins, blockades, expressive and symbolic acts).Direct forms of political participation require few resources, are low risk, and require low levels of commitment.These include the signing of statements, petitions and initiatives.3 (1): 120-146.In principle, simple counting of these activities would have led to a scale of 0-4, and 0-3, respectively.In practice -as descriptive results will show -, both measures are very unequally distributed; the majority of respondents did not participate in any traditional or direct political activities.As prediction using such dependent variables would have been problematic, we constructed a simple typology to distinguish four categories: passive (neither traditional nor direct participation); only traditional participation; only direct participation; both forms.

Control variables: Features of work
Regarding independent variables, the main focus of the study is the role of the labour market position in determining political participation.As outlined in the conceptual section, labour market integration is expected to affect political behaviour.In this regard, unemployment is expected to have a negative impact on participation.Unemployment was measured for the respondents' current situation, as well as for the past.The indicator of current unemployment is a dummy variable taking a value of 1 if the respondent is unemployed, otherwise (0).Past unemployment is a categorical variable with three options: never unemployed in the past; unemployment of duration longer than three months; unemployment of duration exceeding one year.9 Precarious employment is another element that influences political participation negatively.Two indicators are used in this respect: limited or nonexistent work contract = 1 vs. unlimited work contract (0); part-time work: working less than 35 hours per week = 1 vs. full-time work (0).
Finally, working conditions are examined using the concept of 'work autonomy'; the related scale is based on responses to the ESS core question: 'how INTERSECTIONS.EEJSP, 3 (1): 120-146.much does/did the management at your work allow you to decide how your own daily work is/was organised?' (the original 0-10 scale was condensed to a four-item one). 10Job autonomy is defined as influence on policy decisions about organisational activity.The 0-10 scale was converted to a four-item one).

Political variables
In the paper we take into account political attitudes as possible predictors of behaviour.We apply two items 11 from the ESS core questionnaire: satisfaction with the government (0-10), and trust in parliament (0-10).

Socio-economic variables
We also include demographic control variables in the multivariate models.Gender is coded as 1 for men and 0 for women.As mentioned before, we are specifically interested in examining the association between political participation and age, in particular in terms of cohort differences: namely, how the young generation participates in politics compared to older individuals.Therefore, three age cohorts were defined: 15/18-29; 30-50; 51-65.
Level of education is another control variable with three categories: graduate, secondary level, lower level of schooling.Place of residence distinguishes whether respondents live in big city, a suburb, in a smaller town, or in a rural settlement (village).
Finally, the respondent's financial situation is also included in the analysis; we use subjective household income from the ESS core questionnaire transformed into a dummy variable where one category combines the options 'living comfortably on present income' and 'coping on present income' (1) while the other category combines the options 'finding it difficult on present income' and 'finding it very difficult on present income' (0).
For the full list of variables used in the analysis, see Table 8 in the Appendix.
In the course of the analysis we applied bivariate and multivariate techniques.Bivariate relationships examined include changes in political activities over time, and differences by age group.The exact method of multivariate analysis was decide by the form of the dependent variables.In the case of voting, we used binary logistic regression to predict the impact of explanatory variables on the probability of voting at general parliamentary elections.For these, we present the unstandardized regression coefficients (B), the odds ratios (Exp(B)) and the level of significance.For the combined typology of traditional and direct political participation, we predicted the 10 Job-related indicators (nature of work contract, working hours, autonomy) are obviously lacking here; respondents actively employed (i.e.'currently unemployed') were not asked about these features in the survey.
11 First, we tested 4 political attitude variables (satisfaction with democracy, satisfaction with government, trust in parliament, trust in politicians), but because of high correlation values (r>0.7)we decided to work with fewer model variables.Following an anova test we selected only two variables (satisfaction with government, and trust in parliament) INTERSECTIONS.EEJSP, 3 (1): 120-146.activity of respondents using multinomial logistic regression.Passivity (neither traditional nor direct participation) was the reference status, and statistical coefficients refer to the effect of the independent variables on either traditional or direct activity, or on a combination of these (i.e.both).Similarly, we present the unstandardized regression coefficients (B), the odds ratios (Exp(B)) and the level of significance related to the three options for activity, contrasted with 'passive' status in terms of political participation.

Results
To control for changing trends in political participation, we first checked whether claims about young peoples' low level of electoral participation holds in our sample (see Figure 4.).From a comparative perspective, electoral participation is not low in Hungary (76 per cent claimed to have voted in the last national election).Although the level of participation is not critically low, young Hungarians report to a lower level of electoral participation than older cohorts.
In the course of the analysis we tested our hypotheses using two multivariate models.In the first one we used binary logistic regression to investigate how the independent variables influence electoral participation; a dummy dependent indicator.The model contains all of the variables described in the methodology section, but we present only the significant ones (p<0.05) in  As we expected on the basis of the first hypothesis, current unemployment has a negative impact on electoral participation since unemployed people are less likely to participate in elections.However, precarious employment (limited or non-existent work contract; part-time work -i.e.working fewer than 35 hours per week) did not influence electoral participation.The same also holds for previous unemployment.
As for the role of labour market position in determining electoral participation, we found that working conditions, namely job autonomy, have the greatest positive impact.It seems that people who have greater autonomy in organizing their everyday work are more likely to participate in elections than others.
As for the control variables, ESS Round 3 (2006) and ESS Round 4 (2009) had a positive influence on electoral participation compared to the earlier period of data collection.Important events occurred from a political and economic perspective in both years (see Table 1 and Figure 1) but in particular in 2009, when the economic crisis hit Hungary, employment dropped, and unemployment increased.This was also INTERSECTIONS.EEJSP, 3 (1): 120-146.
the year when Prime Minister Ferenc Gyurcsány, still an important figure in Hungarian politics on the left, resigned.In fact, 2009 was the last year of the eightyear-long incumbency of the socialist government, a distinct period preceding the change of government in 2010.These political and economic events might explain why the reported electoral participation of Hungarian citizens was higher compared to 2015.
From the demographic control variables, age affected electoral participation, as postulated in our hypothesis.Compared to the 51-65 year-old cohort, 15-29 year-olds were less active in terms of electoral participation.Similarly negative effects on electoral participation were identifiable in the case of members of the middle-aged cohort, although the magnitude of this estimate is less than that of the youngest cohort.The basic indication is that participation in national parliamentary elections increases with age.
Last but not least, electoral participation was mostly strongly influenced by level of education.More highly educated individuals were more likely to vote (graduates were 2.25 times more likely to participate in elections than non-graduates).We also assume that level of education has an indirect effect on work autonomy, since the higher the level of education of a citizen, the greater the likelihood that they have freedom to organize their everyday working routine.We return to this point in the discussion.The model also shows that the more positively the respondent subjectively evaluated their income situation, the greater the probability of voting.
In order to expand on the study of political participation, we investigated not only electoral participation but also traditional and direct forms of political engagement (see Figure 5.) Source: ESS Round 2-7.Pearson Chi-Square: 71,593; sig:0,000; Cramer's V: 0,062 We identified a very high proportion of passive citizens (81 per cent) within the whole population, finding the highest fraction of passive citizens among young people (87 per cent).It transpires that this group is the least active cohort in terms of all forms of political participation, but especially in traditional forms of political participation.
For the second multivariate model we used multinomial logistic regression to investigate the influence of independent variables on traditional and direct political participation.In this model, the dependent variable is a nominal one with four categories (0 refers to passive citizens, whereas activity in traditional, direct and both forms of political participation are also distinguished).The model examines the impact of the independent variables on each category of political activities separately, while passive respondents (the majority of the sample) are the reference category.Results are presented in Table 3. Similarly to the previous model, only the significant coefficients (p<0.05) are shown in the table, but the other independent variables are controlled.Another issue to take into account when examining the findings in Table 3 is that most of the predictor variables are also categorical.Consequently, effects for passive respondents in general, and the concrete reference category for the specific independent variables overlap.
As Table 3 indicates, traditional and direct forms of political participation are affected by more issues than electoral participation is.This is a general finding: regarding the role of labour market position we find four labour market indicators that affect political participation.
Unemployment has a significant impact, but seems to have different effects on other forms of participation than electoral participation.The analysis revealed an interesting correlation between previous unemployment and political participation.Not having a job for a period of between three months and one year increased the likelihood of both traditional and direct participation.Current unemployment, however, was not a significant predictor of political participation.These results are not in line with assumptions, and we return to them in the discussion.
Similarly to the case of electoral participation, work autonomy has a positive impact on both traditional and cumulative political participation.
Furthermore, job characteristics also play a role, according to this model.Respondents who have limited working contracts are more likely to take part in traditional forms of political activity (Exp(B)=1.423),and those who work fewer than 35 hours per week are also more likely to participate in direct forms of political activity.  1 and Figure 1, the estimates reveal the obviously positive impact (i.e.greater probability) of political participation in the specified years.The magnitude of the effect is particularly large for 2009 in the case of direct activity (Exp(B)=1.919).The most recent year (2012), however, shows a reduction in traditional and direct form of political action -i.e. the probability of such activities is significantly lower than in the year 2015.We elaborate on this further in the discussion part of the paper.
From the control variables, there is a positive effect for men in contrast to women.Age affects traditional participation (in line with our hypothesis), since being in the 15-29 year-old cohort lessens the likelihood of being involved in traditional political activities.However, the model does not show the positive impact of age on direct political participation (in terms of the youngest cohort; 15-29 years old), in contrast to popular claims.We return to this issue in the discussion.
Level of education has a positive impact on direct forms of political participation: graduates are more likely (Exp(B)= 4.164) to participate.This demonstrates the cultural component of direct participation; namely, that direct democratic participation has to be learned, as happens among educated citizens.
The effects of regional variation are quite conventional.Traditional forms of political activity are more typical of people who live in villages than those who live in cities, while the same people participate significantly less in direct action.Finally, more positive subjective appraisal of income is negatively correlated to traditional and direct political participation.

Discussion
Our analysis investigated the relationship between labour market inequalities and political participation.In particular, we focused on the role of unemployment and precarious employment.Unemployment was measured using indicators that referred INTERSECTIONS.EEJSP, 3 (1): 120-146.
to the current situation and to past experience; work flexibility was approached using two indicators: 'bad' (=non-permanent) work contract, and part-time job.For political participation, we distinguished between voting, traditional forms of participation, and direct forms of participation.Based on the literature, we tested concrete hypotheses about how labour market inequalities could influence political participation.
One further goal of the analysis was to elaborate differences according to age cohort to enable us to discuss the political activity of young individuals (below 29 years of age) in Hungary.Finally, given that we analysed only one country, we intended to link the findings about Hungary to the broader political and economic context of political participation -our data cover about 10 years from 2005-2015.This is not a long period, but involved quite significant changes.On the one hand, the economic crisis hit Hungary during these years -this is an obvious example of a wider phenomenon.On the other hand, from the perspective of Hungarian politics, this 10year period involved the fall of the socialist government and the rise of a more populist, centre-right government.Some final additional information about the factors that influence political participation in Hungary were derived from the control variables -in particular, education.

Labour market situation
We begin with the 'easy part' of this discussion: job autonomy.This indicator has a strong positive impact on basically all forms of political participation.This result is in line with the theory and findings of earlier studies; namely, that more involvement and responsibility in the workplace increase political participation.Indeed, it also seems to hold true for Hungary that the mechanisms in effect in the workplace increase political efficacy.It is even more important that this result remains valid even when education is controlled for.Obviously, more educated respondents have more job autonomy, but more autonomy at the workplace also positively affects political participation.
Turning to the results the reasons for which are more complex to interpret, it is apparent that the emerging picture of the impact of the other labour market indicators (unemployment, work flexibility) on political participation is not straightforward.We believe that the seemingly contradictory results stem -at least partly -from the fact that the models include important control variables such as age and education.Thus, it makes sense to investigate the bivariate relationship between these work indicators and political participation.These results are presented in Table 4. Political activity increases with age, and this observation probably holds for voting.However, for other political activities (particularly direct ones -as mentioned above) labour market variables can be used to characterise the age cohorts.Although further observations could have been made using the dataset, we limit the analysis to these cohort differences only.
The research did not attempt a comparative chronological analysis, but assumed that changes over time were stable.However, there is evidence for the influence of the political and economic context on political participation.This becomes obvious when we look at the positive impact of the dummy year 2009.The negative impact of year 2012 (in contrast to 2015) is also understandable.Table 1 refers to the so-called Milla and Békemenet demonstrations in Hungary that occurred during this year.At first sight, the significant negative effect on participation during this year appears to be strange.However, only the Milla demonstrations were real grassroots events directed against the government (the Békemenet -'Peace March'demonstrations were organised by the government).We suggest that many respondents discounted these events as forms of political participation, although we have no evidence for this proposition.'Cheering' for the government may not be considered political participation in the same way that protest action that expresses disagreement is.For many people, demonstration implies protest.However, this issue requires further -perhaps qualitative -investigation.
Finally, we underline the very strong impact of education.From the perspective of political socialisation, this is not surprising.Additionally, one should not forget that the young cohort includes a greater proportion of more educated respondents.There has been a huge expansion of education in Hungary (particularly at the tertiary level) which means that the pool of graduates and the cohort of younger respondents (below 29 years of age) overlap.This is one more argument to consider when looking at the results of the multivariate analysis.
This paper is intended to fill a gap in the literature.Studies about the determinants of political participation that go beyond the usual political factors such as being interested in politics, and being (dis)satisfied with current political situation, are largely lacking in the post-communist context.As a first step we examined the Hungarian population in this regard.Although we did not attempt a comparative analysis, ESS data would definitely permit the expansion of this analysis to include other post-communist countries -one suggestion for further research.
also shows that in Hungary participation in the labour market was steady from 2004-2008 (employment rate of around 58 percent; unemployment rate around eight per cent).The 2008 economic crisis had an immediate and negative impact (employment rate 61 per cent; unemployment 10-13 per cent) that lasted until 2012/2013.Since 2013, due to the public employment program introduced by the government in 2012, the trend has been towards a rise in employment and a decrease in unemployment among all age groups.

Figure 2 .
Figure 2. Voted at last general elections (%).Source: ESS Round 2-7.Beyond voting, we differentiate between traditional political participation and direct political participation. 8The first form of participation is based on four questions from the ESS core questionnaire, coded as dummy variables: -Contacted politician or government official in last 12 months -Worked in political party or action group in last 12 months -Worked in another organisation or association in last 12 months -Worn or displayed campaign badge/sticker in last 12 months

Table 1 .
Table 1 provides a summary of the time period covered by the Hungarian data from 2005-2015.Timing of Hungarian data collection and important political events INTERSECTIONS.EEJSP, 3 (1): 120-146.
Table 2 below.The estimates in the table control for (but do not display) the other independent variables.

Table 3
The model shows quite substantial variation in time:ESS Round 3 (2006 Fall), ESS Round 4 (2009 Spring) and ESS Round 6 (2012 Fall) appear to differ significantly from the reference period (ESS Round 7, 2005 Spring) with respect to their influence on political participation.Whereas in earlier years turbulent changes in Hungarian economics and politics occurred, as shown in Table

Table 5 .
Results of hypotheses.