**1. Introduction**

There have been increasingly urgen<sup>t</sup> calls to transform the current material-intensive economy to a circular economy [1,2]. There are many potential ways to achieve this transformation but reusing material goods as much as possible in the forms of eco-design, direct reuse, or recycling, is a prominent, simple way to do this, both on the demand and supply sides [3]. This study explores how attachment to goods is related to life satisfaction, an oft-cited aspect of subjective well-being (SWB). There exists vast literature on attachment theory, including psychological studies that focus on interpersonal relationships, that is, the intimate emotional bonds between people, such as those between parents and children (e.g., [4]). Research on attachment has been expanded to "places" or "neighborhoods" in sociology and human geography [5] (p. 144) and empirical studies revealed that many people (from 40% to 65%) demonstrated attachment to their neighborhoods [6] (p. 274). According to [7], place attachment can be categorized as either "social attachment", which includes institutional ties, social activity, and local intimates; or "affective attachment" in which the satisfaction is with the neighborhood itself [5] (p. 145). Following this categorization, we use the term attachment to goods in the latter sense—namely, we focus in this paper on material possession attachment [8]. In particular, we hypothesize the following four relationships between longer use of goods and SWB.

First, the tendency to use the same good for longer can increase e ffective income, if other things remain equal. This excess resource may be put into the further purchasing of either consumable or durable goods. In a conventional worldview, this enables us to reach higher utility, which we aim to establish in the theoretical model in the next section.

Second, in a well-known experiment, Kahneman et al. [9] compared people's willingness to accept giving up a mug that they were previously awarded. The surprising result showed that the willingness to accept was higher than the market price, implying that some subjective value was added to the owner of the mug since it was given to them. The authors' interpretation was that once you own a material good, you might feel attached to it after keeping it for a certain period of time, so you would avoid parting with it even if an equivalent or superior good could be purchased. They called this the endowment e ffect [10], which was demonstrated in a laboratory experiment [11–13] (p. 194). This e ffect, more generally called loss aversion, provides another rationale to assume that utility depends on whether consumers continue to use the same commodity. This has been partly examined in the theme of product attachment [14,15]. Recent research suggests that attachment to possessions influences disposition decisions at various stages [16,17], and previous studies have shown that strong emotional attachment to a product discourages consumers from replacing or discarding it [14,17] (p. 215).

Third, in the modern world, where there is increasing awareness of the importance of sustainability and a tendency toward a circular economy, people may be happier when they decide to keep using the same product that is still usable, even if they can a fford to repurchase a new one. They find reusing to be prosocial, whether it is due to pure environmentalism or the "warm glow" e ffect [18]. In the warm glow theory, an agen<sup>t</sup> is assumed to prefer one alternative but aspires to choose another for ethical reasons and receives psychological satisfaction (i.e., a warm glow) from acting in accordance with their aspirations [19] (p. 502). Empirical studies have revealed that the warm glow e ffect can be found in the context of environmental protection [20,21]. A related concept is the positive feelings associated with empathy that can reinforce prosocial behavior [22] (p. 60). From this perspective, people may regard throwing products away as immoral because it harms the environment or others. Note that this prosocial e ffect does not necessarily involve attachment, endowment, or loss aversion e ffects.

Fourth, there could be some latent factors that a ffect both material possession and SWB. Due to these factors, those who tend to use the same good may also tend to focus on long-term relationships, for example. An in-depth investigation of these variables is beyond the scope of the current study, but it is useful here to point to the spirit of mottainai, which is a Japanese adjective frequently used to describe the feeling when something still usable is wasted [23]. The waste from food-related activities such as grocery shopping, cooking, eating, handling surplus food, and so forth, could be reduced by following the mottainai spirit of Japanese culture [24]. Such cultural factors can a ffect both attachment to goods and SWB. Interestingly, this can be said about material goods, money, and even talent and opportunities.

To date, it has not been clear how the transformation into consumption based on a circular economy would a ffect SWB. In contrast, the vast literature has focused on the relationship between income and SWB (for the comprehensive review on this theme, see [25]). Previous studies revealed that income positively correlates to SWB within countries (e.g., [26–29]). It has been argued that income ceases to augmen<sup>t</sup> SWB once individual income level rises above a certain threshold [25]. This has been called the "Easterlin paradox" since it has been confirmed that the reported happiness does not rise in proportion to income increase in those countries such as the United States, Japan, and European countries [28,30–33]. Other studies include the perspective of aspiration for better lives, career achievement, and so forth, which plays an important role in explaining the relatively lower levels of SWB (e.g., [34]).

In contrast to the previous studies that question the role of income in raising SWB, there has also been growing research that provides evidence that income matters to augmenting SWB [25,35,36]. For example, using panel data to control for individual fixed e ffects, it has been revealed that the income of the reference group (together with their own income) significantly a ffects SWB [37]. Argyle [31]

argues that there exists no satiation point of income, as opposed to the previous studies that explored the threshold of income, in line with the Easterlin paradox. This research shows that relative income matters and strongly a ffects SWB [25,38]. Recent studies revealed that relative income and consumption (such as home-ownership) significantly contribute to increasing SWB [39,40].

Donati [41] suggests that the e ffect of material well-being on SWB is not linear but rather diminishes with higher levels of material well-being. This is in line with the economic concept of diminishing marginal utility. Empirical studies also show diminishing marginal e ffects. Gokdemir [42] shows that, in Turkey, only the consumption of durable goods is correlated with life satisfaction. DeLeir and Kalil [43] shows that out of nine consumption categories, only one, leisure, is positively correlated with SWB, using U.S. data. Zhang and Xiong [44] employs 77 consumption categories and 13 SWB indicators to investigate the relationship between consumption and SWB in Japan and shows the particularly strong correlation between relational consumption and SWB. Dumludag [45] shows that the relationship between life satisfaction and each consumption category varies in accordance with the development stage. Relatedly, Pandelaere [46] reviews studies on experiential versus material consumption and suggests that even though many studies find an advantage for experiential consumption, this e ffect does not occur for materialists, which implies that materialists do not benefit more from material than from experiential consumption owing to unrealistic expectations. The above-mentioned literature implies that material well-being cannot be easily increased and that it depends on the contents of consumption or the development stage. However, previous studies do not fully clarify the e ffect of "individuals' attitudes toward material goods" on material well-being. In this study, we thus investigate the relationship between material well-being and "attachment to material goods" in the context of a rural developing country.

The rest of the paper will proceed as follows. In the next section, we theoretically explore the effects of conventional substitution and income. In Section 3, we use a theoretically reduced form that incorporates four e ffects and empirically test our hypothesis using survey data from rural Vietnam. Finally, we provide a discussion and conclusions in Section 4.

### **2. Materials and Methods**

### *2.1. Conceptual Model*

To see how longer use of goods can a ffect life satisfaction, an oft-cited aspect of SWB, it is useful to set up a simple model of two goods and two time periods. In doing so, we compared the consumers' utility levels in two distinct contexts. Imagine two kinds of consumer goods. Good 1 is a general, perishable good that needs to be continually repurchased, such as food. Good 2 is still a consumer good, but durable to a certain extent, so that some people purchase it every time period, while others continue to use what they bought in the previous period. Examples include clothing, personal computers, and smartphones.

In the benchmark model, the consumer purchases Goods 1 and 2 in both periods, t and t + 1. In the extended model, the consumer purchases both goods in period t, but in the subsequent period, purchases only Good 1 and continues to use Good 2 bought in the previous period. Our strategy was simply to compare the two indirect utilities achieved in both models, other things being equal.

Before diving into the details, we came up with at least four relationships between the use of goods and SWB, as described in the Introduction. Formally, life satisfaction (LS), an oft-cited aspect of SWB, may be composed of material and non-material LS (e.g., [47]):

$$L\mathcal{S} = ML + \text{NML}\_{-}$$

*ML* consists of traditional economic incentives, whereas *NML* includes a wide variety of non-economic incentives, such as endowment and attachment e ffects, prosocial/pro-environmental behavior, and the spirit of *mottainai*. In the remainder of this conceptual section, we focus on the LS-equivalent of money-derived (intertemporal) utility, *ML*.

First, in the benchmark model, the representative consumer was assumed to maximize their (intertemporal) utility, *ML*, with regard to Goods 1 and 2 in both periods. Second, in the extended model, we wanted to see what would happen if the consumer continued to use Good 2 in the second period. As detailed in Appendix A, the comparison of the benchmark and extended model led us to our central finding:

**Proposition.** *In the given framework, indirect utility is strictly higher in the extended model than in the benchmark model.*

This proposition tells us that, in a basic model that strips away behavioral features, the first channel, having fewer, durable goods, is positively correlated with SWB, as mentioned in the Introduction. In particular, not having to purchase Good 2 induces both income and substitution effects. The income effect means that the money that could have been used to purchase Good 2 in the subsequent period is now freed up to enhance effective income, which increases consumption of both Goods 1 and 2. The substitution effect from/to Goods 1 and 2 depends on their relative prices as well as the discounting and interest rate.

## *2.2. Empirical Strategy*

In the previous subsection, we demonstrated that, other conditions being equal, consumers who use the same material goods for longer may report higher intertemporal SWB. In the following, we empirically test this expectation by measuring the hypothesized effects all together, leaving separate identification of the aforementioned four effects to future research. Our empirical model is as follows:

$$LS\_i = \gamma\_1 + \mathcal{g}(C\_{1i}) + \sum\_j \beta\_j Y\_{ij} + \varepsilon\_1 \tag{1}$$

Here, *i* represents the individual, and *LSi* represents life satisfaction. Material consumption (*C*1) is on the right-hand side of Equation (1), which is assumed to determine *LS* on the left-hand side. γ1 is a constant, 1 is the uncorrelated error term, and *Yij* stands for other control variables with β*j* as their coefficients. As control variables, we used variables that have been used in prior studies of SWB (i.e., age, gender, level of subjective health, education, marriage, having children, and number of family members). We also included residual consumption (i.e., consumption other than material consumption), (*total*\_*consumptioni* − (*<sup>C</sup>*1*i*)) as a control variable. We intended to decompose the effect of total consumption into material consumption (*C*1) and non-material consumption (residual consumption), which is in line with our theoretical model in the previous subsection.

This study followed [48]'s model, which utilizes nonparametric functions for consumption, so as to clearly and visually explore its different functional forms. As for the other explanatory variables, we used parametric functions. Thus, we applied semiparametric regression. We used generalized additive models (GAMs; [49]), in which the linear predictor depends not on a weighted sum of explanatory variables, as in linear regressions, but on unknown smooth functions, *g*. As such, GAMs enabled us to identify non-linear, locally diverse relationships between consumption and SWB.
