Article Preview
Top1. Introduction
This paper investigates social preferences with data from web-based survey in Japan. A unique feature in our empirical work is our behavioral assumption that people make comparisons in terms of both consumption and saving, rather than that they do so merely in terms of income, as assumed in Clark and Oswald (1996).1 Our behavioral assumption implies that we allow our subjects value their own consumption and saving differently. This, of course, is a sharp contrast with the textbook assumption that consumption and saving are treated equally by consumers, with a mere difference that saving is used in future thus utility from it will be discounted.
We, in this paper, suggest that there can be certain difference in consumption and saving (other than timing) in households’ minds through a psychological pathway. A similar argument is first casted by Richard Thaler, who suggests the mental accounting hypothesis whereby households distinguish money depending on its sources in Thaler (1999).
To understand our motivation more easily, we make use of the following example. In each scenario, own consumption, own saving, reference (others’) consumption, and reference (others’) saving are defined. In one situation the subject’s consumption is given as 3 million yen and the saving as 3 million yen2, while the reference person's consumption is given as 4 million yen and the savings as 4 million yen. In the alternative situation the participant's consumption is 2 million yen and the savings are 2 million yen, but the reference person's consumption and saving are both 1 million yen. When we innocuously assume that the standard budget constraint holds and that they only care about income, just as they do in the literature of income-comparisons, their choices depend on relative strengths of the two motives: relative income or absolute levels of income. In our experiment, however, since we split income into consumption and saving, the choices made by subjects could be more complicated in deciding which situation they would prefer. The difficulty of choices would arise when subjects treat consumption and saving differently.
One virtue of our study is that we can easily disentangle these mixed and complicated motivations of subjects by resorting to an experimental paradigm of discrete choice questions. From data of choices over hypothetical scenarios such as the one given above, we can estimate utility function with effects of social preference via the random utility model framework.3 We then are able to measure the relative importance of their own consumption, own saving, reference consumption, and reference saving.
Our results are as follows. First, we can see that the total marginal utility from own income (the sum of consumption and saving) is positive as generally confirmed in almost all literature. Turning our interests into consumption and saving, we have a positive impact on utility from own saving, as the economics theory presumes. However, our result on own consumption turns extreme, and we can see that own consumption provides negative marginal utilities. Hence, a puzzling result arises.
Next, the effects from reference groups are also estimated. Social preferences in the form of income comparisons to others are estimated to be negative, as is the case of previous studies in the literature (Clark et al. 2008). The effects of reference income on utility is half as large as those from own income levels in the absolute term. In addition, results appear in the opposite way of own situation when we look at consumption and saving separately: an increase in others’ consumption does not decrease own utility while an increase in others' saving decreases own utility.
While consumption is certainly a central notion for economics, as it is one of the essential factors when households are maximizing their utility, our experimental data provide puzzling results on own and reference consumption, whereas savings seem to act as standard economic theory presumes. Given the puzzling results, we suggest that experimenters and experimental designs have to be more careful in wording of economics technical terms and in making economic interpretations out of data collected from subjects who are not familiar with economics.