Table 7 Control variables

Age

Individual well-being tends to decrease with age though the relationship is often argued to be U-shaped rather than linear, with a minimum around age 50 (though this varies somewhat across countries, Bruni and Porta 2005).

Gender

Women tend to report higher life satisfaction than men though the difference is often small, and women’s day-to-day emotions tend to fluctuate a lot more than men’s.

Marital status, children

People with a partner report, on average, higher satisfaction scores than those without. Children have also been shown to increase well-being. (Given our focus on first-generation immigrants, who got married and had children many years ago, we would not expect these variables to be affected by current life satisfaction.)

Education

The existing evidence seems somewhat ambiguous on the relationship between subjective well-being and education (Powdthavee 2010): while most studies find that people with more years of education report higher satisfaction scores than those with fewer years of education (e.g. Helliwell 2003; Bruni and Porta 2005; Stutzer 2004; Graham and Pettinato 2002), there is some evidence that people who have completed at least a university degree report lower levels of job satisfaction and higher levels of mental distress compared to those from a lower educational background (Clark 2003), holding health and income constant. These findings may be explained by the fact that in addition to increasing income, education may also raise aspirations, resulting in a potentially ambiguous overall effect. As we are looking at adults who have completed their education a long time ago, this variable should not be affected by current life satisfaction.

Employment

Empirical findings stress the harm done by unemployment, affecting income as well as status/social expectations. Having a job includes many aspects that provide flow experiences and satisfy intrinsic needs, like being in the company of workmates, applying expertise and experiencing autonomy. Accordingly, being unemployed is repeatedly found to have large negative effects on people’s subjective well-being, with little habituation. We also include a measure of whether the respondent works in the occupation they were trained for and a variable for the degree of job (in)security, which may be of particular relevance to immigrants. To deal with possible reverse causality from life satisfaction to employment, we examine the robustness of our results to using lagged values as proxies.

Owns house/apartment

The effect of income on subjective well-being has been shown to be positive but non-linear, both at the micro and at the macro level (Easterlin 1974, 2001). Stutzer (2004) found that the positive effect of higher income can be offset by rising income aspirations. As we are worried that income may be endogenous (and may be measured with error), we do not include it in our preferred specification and use whether the respondent owns a house/apartment as a proxy. (Results are very similar if we use income instead.)

Parental characteristics

We control for parents’ education in levels and include dummy variables for whether the respondent grew up in a large/medium/small city or in the countryside.

Health

It is widely accepted that an adverse change in health reduces life satisfaction. Furthermore, the literature on mental and physical health reports great inequalities in this field among ethnic groups (Vega and Rumbaut 1991; Rumbaut 1994).

Years since migration

If the number of years in the destination country improves the economic position of immigrants, we would expect this to also increase their life satisfaction. However, this variable may also capture expectations, regret or comparisons with the home country. This may be of particular interest in our sample of guest workers, who originally arrived as temporary migrants.

Identity

To avoid the endogeneity of identity (if respondents are unhappy in Germany, they may be less likely to feel German), we use lagged identity measures as a proxy. We believe this may be reasonable given considerable variation in subjective well-being over time as responses are influenced by the mood of the day as well as random events. We use identity from 2003 as a proxy for identity in 2010 (unfortunately, the identity question was only asked in these years); life satisfaction in 2003 predicts only around 16 % of the variation in satisfaction in 2010, whereas the identity variables (especially majority identity) are more correlated over time.

German language skills

Language plays a central role in the integration of immigrants in the new labour market (Chiswick 1998; 2002) but is also important for social contacts with the host population. Given our focus on first-generation immigrants who arrived over 30 years ago, we believe that current life satisfaction should not affect German language skills.

Discrimination

Although discrimination is often put forward as a possible explanation for the lower life satisfaction of ethnic or racial minorities, general life satisfaction is likely to affect subjective perceptions of discrimination making the identification of a causal effect difficult. As instruments are hard to find, we do not include discrimination in our preferred specification but examine results both with and without this variable.

Ethnic composition of neighbourhood

Residential location is often portrayed as a key element of immigrant integration. In a study among adolescents from immigrant families, those living in ethnically homogeneous neighbourhoods reported a higher level of satisfaction with their lives than those living in heterogeneous neighbourhoods (Neto 2001), contradicting the assumption that immigrants who are in social contact with local natives and live in heterogeneous neighbourhoods should be more socially integrated and thus more satisfied. To avoid the endogeneity of the location decision, we examine robustness by restricting the sample to those who have not moved recently. We also examine results separately for those who want to/do not want to move.

German citizenship

Having German citizenship may affect the respondent’s economic opportunities as well as subjective perceptions of security or uncertainty and may also carry a more ‘symbolic’ value on the perception of immigrants in the host country. Variation due to changes in the German citizenship law is also explored.