There is considerable evidence that people compare themselves to others and enjoy having more than others in some domains (Veblen  1970; Hirsch 1976; Solnick and Hemenway 1998; 2005). Individuals derive utility not only from their absolute level of consumption (as conventionally assumed) but also from status related to their relative position in comparison to others of their reference group. While several theoretical advances have been achieved, the empirical investigation of positional concerns remains relatively scarce, especially among societies. While few empirical studies have been conducted in the United States (Solnick and Hemenway 1998; 2005), Sweden (Carlsson, Johansson-Stenman and Marinsson 2007) and China (Solnick, Hong and Hemenway 2007), empirical evidence remains scarce. Our contribution adds new empirical evidence from France by exploring the following hypotheses inspired from the related literature (Solnick and Hemenway 1998; 2005):
* H1: Position matters much more for some attributes than it does for others. For instance income is frequently described as more positional than leisure (Frank 1985; Frank and Sunstein 2001). People become more positional on attributes for which they enjoy higher absolute levels. For example, people are more positional on income if they already have a higher absolute level of income (Van Kempen 2003).
* H2: Position matters much more for socially visible goods than for "hidden" goods. This hypothesis is supported by the early contribution of Veblen ( 1970) about conspicuous consumption and conspicuous leisure. (1) Recent empirical evidence can be found in Chao and Schor (1998) and van Kempen (2007). Indeed, people are more likely to exhibit positional concerns for goods that convey status and serve as social markers in their reference group.
* H3: Private goods are more positional than public goods. For example, we wonder whether respondents want their countries to be ahead in humanitarian spending, regardless of the absolute amount devoted to this cause compared to private goods. Interestingly, the empirical evidence reported by Solnick and Hemenway (2005) partially contradicts the prediction of Galbraith (1958), where he was arguing that private goods are more positional than public ones.
* H4: Goods are more positional than bads. Compared to goods, people seem to care more about the absolute amount of bads rather than the relative amount (Solnick and Hemenway 1998).
* H5: People are more positional when choosing for their children than for themselves (Solnick and Hemenway 1998).
H6: Public bads are more positional than private bads. People are supposed to be more positional about air pollution than individual health for instance (Solnick and Hemenway 2005).
Moreover, given that our questionnaire shares some questions with other similar surveys administered in other countries (Solnick and Hemenway 1998, 2005 [USA]; Solnick, Hong and Hemenway 2007 [China]), we compare with caution whether some of our results in France differ significantly from those obtained in the United States and China. We hypothesize that positional concerns are less important in France, when compared to the United States because the French society is well-known for strong egalitarian values inherited from the French Revolution. We leave open whether positional concerns are stronger in France when compared to China. The remainder of the paper is as follows. The next section exposes the methods used. This is followed by a discussion of the results. And the final section provides some policy implications and conclusions.
In line with the previous contributions of Solnick and Hemenway (1998; 2005), we used a hypothetical survey (2) consisting of 26 hypothetical questions in the same format described below (Appendix 1). Following the empirical strategy of Solnick and Hemenway (2005) and to reduce the time of filling out the questionnaire, we divided the survey into two separate surveys of 13 different questions. To account for the status quo effect, we created two versions of each questionnaire, which differed only in the order in which the responses were described (Solnick and Hemenway 1998). Each question presents two states of the world. For each state, a question reports how much the respondent, or the respondent's child or country has of a certain item and how much others have. In one state of the world, the "positional state," the respondent (or the respondent's child or country) has more than others. In the other, both parties have more than the positional state but the respondent now has less than others. (In the case of bads, in the positional state the respondent has less than others, and in the absolute state everyone has less than in the positional state and the respondent has more than others). The respondent has to choose which hypothetical situation he prefers. If he has no opinion, he can select the two answers (Solnick and Hemenway 1998). Two examples are provided below:
Assume that the higher the IQ is the more clever the individual is.
 A--Your IQ is 110 ; others average 90.
 B--Your IQ is 130 ; others average 150.
Assume that air quality can be measured on a scale from 1 (unpolluted) to 10 (very polluted).
 A--The country of your child has 6 ; other countries average 8.
 B--The country of your child has 4 ; other countries average 2.
The first example is a private good; the second is a public bad. In both examples, A is the positional state. The domains included in the surveys include private goods and bads, public goods and bads and child-related items. All questions had the same general trade-offs because a person's choice between the two states of the world depends in part on the size of the trade-off between absolute and relative position. Questions were divided into the two surveys, so that no subject was asked about all of the items. For example, each subject answered only one of our two income questions. All respondents were asked about their age, gender, matrimonial situation, number of children, number of brothers and sisters, education, professional status, sport practice, religious practice, monthly income, political orientation and the major studied in order to test whether these socio-demographic variables are related to positional choices. In January 2007, the survey instrument was administered to about 370 faculty, students and staff...