New evidence on cross-country differences in job satisfaction using anchoring vignettes

Abstract

This paper presents results on cross-country comparison of job satisfaction across seven EU countries taking into account that people in different countries may perceive subjective questions differently. We apply a chopit model approach where the threshold parameters in an ordered probit model are re-scaled through anchoring vignettes. Compared to a traditional ordered probit model, which yields the familiar result that Denmark and Finland are ranked in the very top, the country ranking is altered when the chopit model is applied. In this case, the Scandinavian countries are ranked somewhat lower while workers from the Netherlands are found to have the highest level of job satisfaction. These results suggest that cultural differences in the way people perceive subjective questions about satisfaction make simple cross-country comparison misleading.

Introduction

Within the economics profession, a series of studies on subjective well-being, life satisfaction and job satisfaction have proliferated in recent years, (see Frey and Stutzer, 2002a, Frey and Stutzer, 2002b, van Praag and Ferrer-i-Carbonell, 2004, Clark et al., 2006, for recent, large scale surveys). In some studies, job satisfaction or overall life satisfaction has been regressed, using some kind of ordered logit or probit approach, on a number of variables using cross-country data, and country dummies have then been used to capture country-specific effects.

Interestingly, cross-country studies of happiness with life or satisfaction with work find surprisingly consistent results in the ranking of nations. For instance, Denmark and Ireland are persistently ranked in the very top with respect to well-being (Eurobarometer, 2002, Inglehart and Klingemann, 2000, Easterlin, 2001) and job satisfaction (Clark, 2005, Ahn and Garcia, 2004, Blanchflower and Oswald, 1999, Kaiser, 2002).¹

The validity of these satisfaction studies is important for a number of reasons. First, many private companies periodically evaluate their employees' job satisfaction. This is of great concern to managers who seek to abstain employees from quitting (Clark, 2001) and because it is generally believed that satisfied workers are more productive than dissatisfied workers (Pfeffer and Langton, 1993, Koys, 2001, Patterson et al., 2004).

Second, satisfaction studies often attract a lot of attention from the news media, which in itself may increase the probability that politicians will react upon the results of these studies.

Third, the demographic evolution towards ageing populations in OECD countries puts strain on public finances since, ceteris paribus, a lower share of the population will be working and pay taxes to finance government spending while, possibly, the pressure on public health care and pension systems may increase. For this reason there is a clear objective formulated in the so-called Barcelona and Stockholm targets of the European Union to increase the age in which people exit the labor market. In recent years, the European Commission has consequently devoted a lot of attention to the quality of jobs (European Commission, 2001, European Commission, 2002, European Commission, 2003), and indicators of quality have been chosen based on their ability to predict self-reported job satisfaction (EU, 2002, p. 81).

Hence, satisfaction studies, including cross-country studies, are highly important for development of public policies as well as for human resource management. For instance, if it truly is the case that job satisfaction is higher in Denmark and Ireland than in other countries then one could argue that working life should be arranged in a way that resembles the labour market in these countries.

However, the problem with studies that have been done up to this date is that it is unclear whether the results are due to the fact that it actually is nicer to work in Ireland or Denmark or whether the Irish and the Danes perceive questions of job satisfaction in different ways compared to people in other countries. More generally, one of the inherent problems with using subjective answers to questions on individual well-being is that individuals, at least from different cultures, may perceive these questions in different ways, and that there may therefore by systematic differences across groups of sub-populations. As answers to questions on individual well-being most often are reported on a discrete scale, this means, in a technical sense, that the response category cut points for the different evaluations of self-assessed happiness or job satisfaction will differ between groups.

Layard (2003) and Veenhoven (2000) both discuss the possibility of cultural differences in reporting satisfaction and happiness but note that, to date, “There is no evidence of this…” (Layard, 2003, p. 19), i.e. of cultural differences in reporting happiness, and hence conclude that “For the present it appears that self-reports of overall happiness can be meaningfully compared across nations” (Veenhoven, 2000, p. 271).

In this paper, we seek to provide evidence of whether or not there are any cultural differences in reports of satisfaction. Our approach to remedy this potential problem takes as its starting point the correction of response category cut points by the means of anchoring vignettes. This methodology has earlier been applied in other social sciences, e.g. in political science (King et al., 2004) and in medicine (Salomon et al., 2004).

The idea behind this method is that respondents, in addition to stating how satisfied they are with their own life or job, are asked to evaluate, on the same scale, how good or bad a set of hypothetical jobs or life situations are. This information is then used to rescale individuals' real evaluation of their own situation, using a joint, so-called chopit model (compound hierarchical ordinal probit, see Rabe-Hesketh and Skrondal, 2002) for the job satisfaction question and the vignettes. Thus, if one particular group, say the Danes, systematically gives higher valuations of hypothetical jobs compared to other nationalities that rank the exact same hypothetical jobs, this would indicate that differences in job satisfaction are due to cultural differences in evaluating such subjective questions and not due to better organization of workplaces or better work environment in Denmark.

Hence, in this paper we re-examine cross-country differences in job satisfaction using anchoring vignettes. We contribute to the satisfaction literature by examining to which extent cross-country differences in job satisfaction adhere to cultural differences in responding to subjective questions. This is done using data from a purpose-built questionnaire, applied to seven EU countries, and asking respondents to rank their own job as well as hypothetical jobs. The analysis extends the standard ordered probit model along two dimensions. First, and most importantly, by using vignettes we are able to re-scale respondents' answers in order to make them comparable across countries. Second, the thresholds are allowed to depend on covariates.

Our main finding is that, while the standard ordered probit model applied to our data yields the traditional result that Scandinavian countries are ranked in the top, this ranking is altered when anchoring is applied. In this case the Scandinavian countries are ranked somewhat lower while workers from the Netherlands are found to have the highest level of job satisfaction. These results provide the first evidence of cultural differences in reporting of satisfaction and indicate that simple cross-country comparison may be misleading.