Introduction

Transitioning to a circular economy has received increasing attention as a key component of a sustainable society. In a circular economy, material and resource values are maintained for as long as possible through circular activities (e.g., remanufacturing and product lifetime extension) to achieve an efficient use of resources (European Commission, 2015). As a circular economy aims to create new business opportunities and strengthen economic competitiveness, the possibilities of economic growth through a circular economy have been discussed from an economic perspective (Miller, 2023). Lin (2020) discussed economic growth in a circular economy by comparing it with a linear economy. George et al. (2015) proposed a theoretical model of economic growth in a circular economy based on macroeconomics. The proposed model theoretically shows that pollution can be reduced by increasing the recycling rate or self-renew rate of the natural environment. Furthermore, the circular economy perspective has received attention from practitioners. For example, Philips (2023) identified circular economy as a key material topic and reports a circular revenue percentage that shows the ratio of revenue obtained by circular practices, such as access-based business model remanufactured products, to total revenue.

Extending the product lifetime is a major strategy for achieving a circular economy. To extend product lifetimes, the New Circular Economy Action Plan of the European Commission (2020) proposed concrete measures for sustainability principles, including improving product durability, reusability, upgradability, and reparability; enabling remanufacturing; restricting single use; and countering premature obsolescence. The Circular Economy Vision 2020 has been proposed for Japan, incorporating a design of products and services that prolong product use as a business model with high circularity (Ministry of Economy, Trade and Industry, 2020). The measures proposed in the above schemes focus on mitigating the limitations to product lifetime extensions from the perspective of physical or functional product lifetimes. Following these strategies, concepts and systems to comprehensively understand the strategies of product lifetime extension (Bocken et al., 2016; den Hollander et al., 2017), product design methods for increasing product circularity (Vimal et al., 2022), and business models suitable for product lifetime extension (Lewandowski, 2016; Milios, 2021) have been developed.

However, product lifetime extension is not restricted by the designed physical durability of the product. Consumers are the ultimate decision-makers regarding product replacement or disposal. Therefore, it is necessary to understand the drivers of consumer replacement behaviour. For example, Cooper (2004) categorised product obsolescence into two types based on the reasons for obsolescence. First is “absolute” obsolescence, which is product obsolescence owing to physical product malfunction. Second is “relative” obsolescence, which refers to obsolescence for reasons other than physical product failure, such as economic, technological, or psychological reasons. Referring to Cooper’s (2004) framework of reasons for product obsolescence, Yamamoto and Murakami (2021) investigated the relationship between product obsolescence and reasons for obsolescence in Japan. They showed that the average duration of use for relative obsolescence was significantly shorter than for absolute obsolescence for microwaves, whereas there were no statistically significant differences between the average duration of absolute and relative obsolescence for digital cameras and personal computers. Okada (2001) showed that consumers’ evaluations of past use experiences of their old products, expressed as “mental book value” in this study, can affect product replacement decisions. To comprehensively understand the influencing factors, van den Berge et al. (2021) reviewed previous studies on factors. They summarised that consumers’ replacement decisions are conducted based on various values related to consumers’ psychology, owned products, and new products. They argued that further empirical studies are needed about the effectiveness of the factors on product replacement decisions.

In addition, previous studies focused on the effects of specific factors on product replacement decisions. Especially, the effect of energy cost information has been widely investigated. For example, Carroll et al. (2016) analysed the effects of providing five-year energy cost on sales of dryers in Ireland. They found that it increased the sales of less energy-efficient dryers; however, it did not significantly increase the sales of high energy-efficient dryers. Del Mar Solà et al. (2021) analysed whether providing monetary information about energy savings promotes the purchase of energy-efficient appliances in Spain for three appliances (washing machine, fridge, and dishwasher). They showed that there were differences in the effectiveness of the provision on the purchase of energy-efficient appliances based on the appliance. Developing the frameworks in Carroll et al. (2016) and Del Mar Solà et al. (2021), Denny (2022) conducted a randomised controlled trial to analyse the effects of long-term energy cost labelling for four appliances (washing machine, tumble dryer, fridge freezer, and dishwasher) in Ireland. The results showed that the labelling did not have significant effects on the purchase of high energy-efficient products.

It can be considered that the duration of the actual product lifetime is determined both by the designed physical product lifetime and the consumer’s expected product lifetime. In particular, as the physical and functional limitations of product durability are mitigated through the implementation of product design policies in the transition to a circular economy, more attention is required to extend the actual product lifetime from the consumer perspective. Consumer expectations regarding product lifetime, that is, the expected product lifetime, are useful indicators. The expected product lifetime describes the period for which the consumers expect to use a product. Previous studies have investigated the expected product lifetimes of various products in different countries. To the best of the authors’ knowledge, Cooper (2004) was the first to conduct this study. The author investigated the expected lifetimes of various products in the United Kingdom (UK) through face-to-face interviews. Cox et al. (2013) examined the expected product lifetime in the UK, utilising a group discussion approach. The expected product lifetime has been investigated in several countries. Wieser et al. (2015) investigated the expected product lifetimes in Austria for a variety of products, including home appliances, clothes, furniture, and electronics. Hennies and Stamminger (2016) studied the expected lifetimes of five products (washing machine, laptop, kettle, TV, and hand mixer) in Germany. They found that expensive products or products whose owners were highly satisfied with the product lifetimes had longer actual product lifetimes. Oguchi et al. (2016) surveyed the expected product lifetimes in Japan for vacuum cleaners, mobile phones, digital audio players, and digital cameras. They proposed three types of expected product lifetimes and compared the differences in their lengths for the four products. Echegaray (2016) investigated the expected product lifetimes of 10 electronic devices (including microwaves, computers, and washing machines) in Brazil and found that the gaps between the expected and experienced product lifetimes in Brazil were relatively larger than Cooper (2004). In a study on the expected product lifetime of information and communication technology products, Woidasky and Cetinkaya (2021) analysed the expected lifetime of laptops utilised in universities. Most previous studies have shown that actual product lifetimes do not meet consumers’ expected product lifetimes.

In the context of product lifetime extension policies in a circular economy, the product design policies proposed by the European Commission (2020) or the Japan Ministry of Economy, Trade and Industry (2020) could enhance the expected product lifetime. The proposed policies advocate removing physical or functional restrictions to product lifetime extension; therefore, the consumers would assume that their products could be used for longer than before. Furthermore, if the designed physical product lifetime is improved adequately, the practical potential of the actual product lifetime extension can be determined based on whether and to what extent the enhancement of the expected product lifetime could lead to an extension of the actual product lifetime. Therefore, it is beneficial to clarify the effects of enhancing the expected product lifetime on the actual product lifetime. Moreover, if it could be determined that enhancing the expected product lifetime does extend the actual product lifetime, policies could be formulated to persuade consumers to use products for longer than before (e.g., by providing information on product durability improvements). This can broaden the approaches to policymaking for product lifetime extension.

Research on the effect of the expected product lifetime on the actual product lifetime is scarce. To the best of the authors’ knowledge, only Nishijima and Oguchi (2023) proposed a framework for analysing the effects of the expected product lifetime on the actual product lifetime and applied the framework to air conditioners in Japan. Nishijima and Oguchi (2023) showed that an increase in the expected product lifetime could contribute to extending the actual product lifetime. Employing scenario analyses, the authors quantitatively demonstrated the extent to which such an increase could extend the actual product lifetime. However, the study focused on a single product (air conditioners). The present study considers it important to apply the analytical framework by Nishijima and Oguchi (2023) to other durable products to investigate the applicability of the framework and explore if similar results can be obtained for other durable goods. Understanding whether there are differences in the effects of the expected product lifetime among products can assist in better policymaking for product lifetime extension, for example, whether policies of increasing the expected product lifetime can be similarly implemented or should be considered by individual products.

In view of the above research background, this study investigated whether an increase in the expected product lifetime could extend the actual duration of product use for refrigerators, as refrigerators (and air conditioners) are widely used in Japanese households. Recent possession rates for refrigerators and air conditioners in Japanese households are 98.1% and 86.4%, respectively (Statistics Bureau of Japan, 2016). Moreover, the average product lifetimes of these products are similar and malfunction is the primary reason for their replacement (Cabinet Office of Japan, 2021). These similarities are suitable for the investigation. The study conducted scenario analyses to show whether and to what extent an increase in the expected product lifetime affects the actual product lifetime of refrigerators. The results were compared with the results of Nishijima and Oguchi (2023). Furthermore, effective policymaking for product lifetime extension is discussed for refrigerators and air conditioners in terms of the expected product lifetime.

Methods

Dynamic Discrete Choice Model of Product Replacement Decisions for Refrigerators

Following Nishijima and Oguchi (2023), this study employed a dynamic discrete choice model (DDCM) to analyse the difference in the effects of the expected product lifetime on the actual product lifetime between refrigerators and air conditioners. The DDCM is an econometric model to quantitatively analyse consumer decisions in a “forward-looking” manner by considering product replacement decisions related to durable goods (Rapson, 2014; Rust, 1987; Schiraldi, 2011). This study assumed that consumer i who owns a refrigerator decides in year t whether the refrigerator should be kept or replaced with a newly bought refrigerator in year t. The replacement decision is expressed as a binary variable \({a}_{i,t}\). The expression \({a}_{i,t}=0\) indicates that consumer i retains the refrigerator in year t, whereas \({a}_{i,t}=1\) indicates that consumer i replaces the refrigerator in year t. In making such product replacement decisions, it is assumed that consumer i considers the following factors: the annual electricity consumption of the current refrigerator, \(e_{i,t}^\text{old}\); the annual electricity consumption of a new refrigerator bought in year t, \(e_{i,t}^\text{new}\); the purchase price of a new refrigerator bought in year t, \(p_{i,t}^\text{ref}\); and the expected “remaining” product lifetime of the current refrigerator in year t, \({\text{ERL}}_{i,t}\). Here, the expected remaining product lifetime expresses the number of years that a consumer considers the current refrigerator could continue to be used from the time the decision is made. The utility functions for retaining and replacing the refrigerators in the DDCM, \({u}_{i,t}\left({a}_{i,t}=0\right)\) and \({u}_{i,t}\left({a}_{i,t}=1\right)\), respectively, are described as follows:

$$u_{i,t}\left(a_{i,t}=0\right)=\delta_0+\delta_1\left\{\text{l}\text{o}\text{g}\left(e_{i,t}^\text{old}\right)-\text{l}\text{o}\text{g}\left(e_{i,t}^\text{new}\right)\right\}+\delta_2{\text{ERL}}_{i,t}+\epsilon_{i,t}$$
(1)

and

$$u_{i,t}\left(a_{i,t}=1\right)=\delta_3\text{l}\text{o}\text{g}\left(p_{i,t}^\text{ref}\right)+\epsilon_{i,t}$$
(2)

where \({\delta }_{0}\), \({\delta }_{1}\), \({\delta }_{2}\), and \({\delta }_{3}\) represent the DDCM parameters and \({\epsilon }_{i,t}\) represents an unobserved error term.

Based on the assumptions used in the DDCMs in previous studies (Rapson, 2014; Rust, 1987), the probabilities of choosing to retain or replace (i.e., \(P\left({a}_{i,t}=0\right)\) and \(P\left({a}_{i,t}=1\right)\), respectively) are expressed as a standard logit probability, as shown in Eqs. 3 and 4.

$$P\left(a_{i,t}=1\right)=\frac{\text{e}\text{x}\text{p}\left\{u_{i,t}\left(a_{i,t}=1\right)+\beta{\text{EV}}_{i,t}\left(e_{i,t}^\text{new},p_{i,t}^\text{ref}\left|a_{i,t}=1\right.\right)\right\}}{\sum_{{\widetilde a}_{i,t}}\left[\text{e}\text{x}\text{p}\left\{u_{i,t}\left({\widetilde a}_{i,t}\right)+\beta{\text{EV}}_{i,t}\left(e_{i,t}^\text{new},p_{i,t}^\text{ref}\left|{\widetilde a}_{i,t}\right.\right)\right\}\right]}$$
(3)
$$P\left({a}_{i,t}=0\right)=1-P\left({a}_{i,t}=1\right)$$
(4)

Here, \(\beta\) (\(0\le \beta <1\)) is a discount factor and \({\text{EV}}_{i,t}\) is the expected value function in the DDCM. Following previous studies (Nishijima & Oguchi, 2023; Rapson, 2014; Rust, 1987), the discount factor value was set to 0.9. Based on the above probabilities, the parameters of the model using the maximum likelihood estimation with the product replacement behaviour data of the refrigerators were obtained. The methods used to collect such data are described in the “Materials” section. A bootstrap method was applied to determine the statistical significance of the obtained parameters. Based on previous studies of dynamic discrete choice model using the bootstrap method (Gordon, 2009; Rapson, 2014), the number of bootstrap samples was set to 200. To avoid obtaining local optimised parameters, MATLAB (MathWorks, Inc., USA) was used, with the global optimization toolbox to estimate the parameters and the bootstrap method. The details of the mathematical structure of the DDCM used in this study and the estimation procedure of the DDCM parameters are presented in ESM Appendices A and B.

Estimated Effect of Change in Consumer’s Expected Product Lifetime on Actual Duration of Product Use of Refrigerators

A scenario analysis was conducted to clarify the effects of the expected product lifetime on the actual product lifetime of refrigerators based on the DDCM parameters obtained. Utilising the replacement probabilities in Eqs. 3 and 4, the probability that a refrigerator owned by consumer \(i\) is replaced in year t, \({\stackrel{\sim}{P}}_{i,t}\), was calculated as follows:

$${\stackrel{\sim}{P}}_{i,t}=\prod _{s=j}^{t-1}\left\{P\left({a}_{i,s}=0\right)\right\}\times P\left({a}_{i,t}=1\right).$$
(5)

where \(j\) \(\left(j\le t\right)\) represents the year in which a consumer \(i\) purchases a new refrigerator. This replacement probability implies that consumer \(i\) chooses to retain the refrigerator from year j (year of purchase) to year \(t-1\) and chooses to replace it in year t. Therefore, the replacement probability of a refrigerator is calculated as the joint probability of choosing to retain or replace the refrigerator, as described in Eq. 5. In this study, the replacement probability was set to 0 when \(t=j\) (i.e., the probability that a refrigerator purchased in year j is replaced the same year). If the maximum number of years for which refrigerators can be used physically is described as \(Y_\text{max}\), the replacement probability when the number of years of the duration of product use (i.e., \(t-j\)) reaches \(Y_\text{max}\) is defined as \({\stackrel{\sim}{P}}_{i,t}=1-\sum _{s=j}^{t-1}{\stackrel{\sim}{P}}_{i,s}\).

If the expected product lifetime is extended by \(\Delta{\text{EL}}\) years, the change in the expected remaining product lifetime is reflected as follows:

$${\text{ERL}}_{i,t}=\left\{\begin{array}{c}\left({\text{EL}}+\Delta{\text{EL}}\right)-\left(t-j\right)\quad\text{if}\left({\text{EL}}+\Delta{\text{EL}}\right)-\left(t-j\right)\geq1\\0.5\qquad\qquad\qquad\qquad\quad\;\text{if}\left({\text{EL}}+\Delta{\text{EL}}\right)-\left(t-j\right)<1\end{array}\right.$$
(6)

where \(\Delta EL\) represents the consumer’s expected product lifetime, which corresponds to the consumer’s expected remaining product lifetime in the year of the new purchase (i.e., year \(j\)). A value of \(t-j\) indicates the number of years (duration) of product use for the currently owned refrigerator in the product replacement decision year. Setting the value of the expected remaining product lifetime when the expected product lifetime is extended (as above), the replacement probabilities of a refrigerator can be calculated, if the consumer’s expected product lifetime is extended by \(\Delta{\text{EL}}\) years, using Eq. 5. Notably, when the \(\Delta{\text{EL}}\) value is set to zero, the replacement probabilities can be obtained when the expected product lifetime is unchanged. In this study, the case \(\Delta{\text{EL}}=0\) is called the baseline case. In scenario analyses about the effects of the expected product lifetime on actual product lifetime, subscript \(i\) was substituted with subscripts \(c\) and \(j\), where subscript c represents a capacity class of refrigerators. This substitution enabled the calculation of the replacement probabilities corresponding with production years and capacity classes of refrigerators for the scenario analyses. Based on the listed capacity class of refrigerators of the production year 2019, provided by Shinkyu-san (Ministry of the Environment of Japan, n.d.), eight capacity classes were considered that correspond to subscript c (≤ 200, 201–250, 251–300, 301–350, 351–400, 401–450, 451–500, and 501–550 L). The data and values for calculating the replacement probabilities are explained in the following section.

Materials

To estimate the DDCM parameters, the Japanese survey company INTAGE Inc. conducted a web-based questionnaire survey to assess the status of refrigerator replacement in Japan in 2019. The questionnaire survey was conducted in November 2019 and February 2020. The survey respondents could take the survey only once. The detailed content of the questionnaire is provided in ESM Appendix C. A screening survey was conducted to select the appropriate respondents, that is, respondents owning a refrigerator and having the capacity to purchase a replacement refrigerator. These selected respondents answered the main survey. The respondents who had replaced their refrigerators in 2019 were requested to provide certain information, including the model code of their new refrigerator, the purchase price of the new refrigerator, the number of years the old refrigerator was used, whether they changed to a different capacity refrigerator compared with their old refrigerator, the anticipated remaining product lifetime of the old refrigerator in the replacement year, and the brand of the old refrigerator. The respondents who did not replace their refrigerators in 2019 provided certain information, including the model code of their refrigerator, the number of years the refrigerator has been used, the expected remaining product lifetime of the refrigerator, whether they had a specific refrigerator in mind for replacement, the model code and assumed purchase price of the specific refrigerator, and whether they would change the capacity of the new refrigerator (if they did not have a specific refrigerator in mind).

From the model codes of the existing refrigerators, the annual electricity consumption and the capacity of these refrigerators were determined. However, no information was available about the electricity consumption of the old refrigerators that had been replaced in 2019 or the annual electricity consumption and purchase price of the new refrigerators for the respondents who did not replace their refrigerators in 2019. Therefore, these values were assumed from the questions on the number of years of using the old refrigerators, whether the respondent changed the capacity of the new refrigerators (for the respondents who replaced their refrigerators in 2019), and whether they intended to change the capacity of the new refrigerator (for respondents who had not replaced their refrigerators in 2019). From this information, the production year and the capacity of the old refrigerators (for respondents who had replaced their refrigerators) or the capacity of new refrigerators (for respondents who had not replaced their refrigerators) were assumed. This information was sufficient for setting the values of the annual electricity consumption and purchase price. The value of t (year of product replacement decision) in this study was fixed in 2019 and the respondents were asked to describe the factors relevant to their product replacement decisions in the questionnaire survey. Based on the ratio of the respondents who replaced their refrigerators to all the respondents in the survey, the ratios of the respondents who replaced and those who did not replace their refrigerators were determined in the main survey.

Initially, 3,300 respondents were obtained. From them, respondents who were unable to provide the correct model code of their current refrigerators, respondents who answered “I do not know” or “I do not remember” to any question, respondents whose current refrigerator was manufactured in 2019, and respondents who had used their refrigerator for “1 year or less” (although they had not replaced their refrigerator in 2019) were excluded. Subsequently, 611 respondents were excluded and 2,689 samples were utilised in the analyses. A summary of statistics of samples utilised in this study and a discussion of the representativeness of the sample are presented in ESM Appendix D.

The objectivity of the data of the production year, capacity class, and annual electricity consumption of refrigerators owned by the respondents was relatively assured because these data were obtained from the model code. However, the expected product lifetime was based on the beliefs of the respondents; therefore, the subjectivity of the data could not be avoided. To obtain objective data about the expected product lifetime, a methodology should be developed to measure consumers’ expected product lifetime.

To set the values for the annual electricity consumption, a web application from Shinkyu-san (Ministry of the Environment of Japan, n.d.) was used that provides information on the catalogue-based annual electricity consumption of specific home appliances. Although these values do not sufficiently reflect the actual electricity consumption, they become useful proxies as the information is publicly released and can be easily confirmed by consumers through the energy-efficiency label attached to the products. Depending on the production year, the annual electricity consumption values were measured based on the Japan Industrial Standards (JIS) 2006, which correspond to the old standards, and the JIS 2015 standard, which corresponds to the new standards. Some refrigerator manufacturers (HITACHI, Panasonic, SHARP, and TOSHIBA) provide data on the annual electricity consumption measured by both the standards for specific refrigerators. Using these data, the conversion rates of the annual electricity consumption from JIS 2006 to JIS 2015 were calculated by capacity class. The values converted to JIS 2015 were finally used.

To set the values of the purchase price of the new refrigerators, Kakaku.com (Kakaku.com Inc., n.d.), a Japanese website for researching the prices of various products, including refrigerators, was used. The product prices of the new refrigerators for each capacity class were referred to at the midpoint of the first survey period (i.e., 15 November 2019).

To analyse the effects of the expected product lifetime on the actual product lifetime, the values of the expected product lifetime \(\text{EL}\) in Eq. 6 were set, utilising the questionnaire survey. The median of the total expected product lifetime of the refrigerators, calculated by adding the number of years of product use to the expected remaining product lifetime, was 15 years. Therefore, the \(\text{EL}\) value of the refrigerators in Eq. 6 was set to 15 years as the baseline case. Furthermore, since the maximum duration of product use indicated by the questionnaire survey was 31 years, \(Y_\text{max}\) was set to 31. A detailed summary of the expected product lifetimes is presented in Tables 9 and 10 in ESM Appendix D.

Results

Product Lifetime Extension Effect of the Expected Product Lifetime for Refrigerators

Table 1 presents the estimation results of the DDCM parameters for refrigerators in Japan. Models 1 and 2 have different settings for the discount factor, namely \(\beta =0\) and \(\beta =0.9\), respectively. The signs of the parameters are all considered reasonable as the sign of the parameter of the expected remaining product lifetime \({\delta }_{2}\) is positive and the parameter is statistically significant. This means that an increase in the expected remaining product lifetime \({\text{ERL}}_{i,t}\) increases the utility of retaining the products, following Eq. 1. Since the utility of retaining is included in the denominator of the probability of replacing, and not in the numerator, as in Eq. 3, an increase in the expected remaining product lifetime decreases the probability of choosing to replace, \(P\left({a}_{i,t}=1\right)\). Following Eq. 4, this indicates that an extension in the number of years of the expected remaining product lifetime \({\text{ERL}}_{i,t}\) increases the probabilities of choosing to retain the product, \(P\left({a}_{i,t}=0\right)\). The DDCMs were estimated with different parameter settings to check the statistical robustness of the obtained models (the results are presented in ESM Appendix E). Furthermore, although the degrees of statistical significance of each parameter of the DDCMS are slightly different, the above-mentioned results of the parameters of refrigerators in this study have nearly the same trend as the air conditioners in Nishijima and Oguchi (2023). The trends can be recognised as reasonable. Considering the similar characteristics of both products, obtaining similar trends indicates that the DDCMs in this study have applicability to model the product replacement decisions.

Table 1 Base estimation results of dynamic discrete choice model parameters
Full size table

Figure 1 shows the effect of the expected product lifetime on the actual duration of product use of refrigerators using the parameters in Model 1, the log-likelihood of which was higher than in Model 2. The dotted line shows the probability of product replacement for the baseline case in which the expected product lifetime does not change (i.e., \(\Delta{\text{EL}}=0\)). The three solid lines show the probabilities of product replacement in cases where the expected product lifetime of the refrigerators is extended by 1, 2, and 3 years (i.e., \(\Delta{\text{EL}}=1,2,3\)). The replacement probabilities shown in Figure 1 are weighted by the ratios of the refrigerator size classes in the sample. The average duration of product use in the baseline case is 12.3 years. The duration of product use in cases where the expected product lifetime increases by 1, 2, and 3 years is 12.82, 13.35, and 13.89 years, respectively. The differences in the average duration of product use of refrigerators in the baseline case and each of the expected product lifetime extensions are 0.52 years (\(\Delta{\text{EL}}=1\)), 1.05 years (\(\Delta{\text{EL}}=2\)), and 1.59 years (\(\Delta{\text{EL}}=3\)). These results imply that increasing the consumer’s expected product lifetime could contribute to extending the actual product lifetime of the refrigerators. This result is consistent with Nishijima and Oguchi (2023).

Fig. 1
figure 1

Replacement probability distribution of refrigerators estimated by dynamic discrete choice models

Full size image

Comparing Product Lifetime Extension Effects on Expected Product Lifetime

The results of the duration of product use extension effects on the expected product lifetime for refrigerators were compared with air conditioners, as previously estimated by Nishijima and Oguchi (2023). Table 2 presents the extension effects for refrigerators estimated in this study and for air conditioners estimated by Nishijima and Oguchi (2023). The parameter settings of the DDCM and the calculation methods for the replacement probabilities for air conditioners used by Nishijima and Oguchi (2023) are similar to those employed in this study. As shown in the previous section, the product lifetime extension effects of the expected product lifetime for refrigerators are 0.52 (\(\Delta{\text{EL}}=1\)), 1.05 (\(\Delta{\text{EL}}=2\)), and 1.59 (\(\Delta{\text{EL}}=3\)) years, respectively. This means that the degree of the product lifetime extension effects on the expected product lifetime is approximately half the duration of the increase in the expected product lifetime. In contrast, the product lifetime duration extension effects for air conditioners are 0.9 (\(\Delta{\text{EL}}=1\)), 1.81 (\(\Delta{\text{EL}}=2\)), and 2.73 (\(\Delta{\text{EL}}=3\)) years. This means that the degree of the product lifetime extension effects on the expected product lifetime is almost equivalent to the increase in the expected product lifetime. A comparison of the estimated values for refrigerators and air conditioners showed that the product lifetime extension effect on refrigerators is less than that on air conditioners. As discussed in “Introduction,” the results of both studies are significant as the extension effects differ although the product characteristics are similar.

Table 2 Comparison of the product lifetime extension effects on the expected product lifetime (unit: years)
Full size table

Discussion

Discussion

As shown in the “Results” section, the average product lifetime of refrigerators calculated by the DDCM estimated in this study is 12.3 years. However, the average product lifetime of the replaced refrigerators in the sample is 11.78 years. A gap between the two values of average product lifetime is not too large, which proves that the DDCM of the refrigerator estimated in this study is valid for showing consumers’ replacement decisions to a certain extent. However, since there is a gap between the two values, the DDCM in this study can be improved to increase the validity. For example, the DDCM in this study did not include the subjective or psychological factors, such as product appearance and consumers’ evaluation of product functions. Certain previous studies have analysed the relationships between these factors and product replacement (Hou et al., 2020; Miller et al., 2019; Sheth et al., 1991). However, it is beyond the scope of the present study due to data limitations and will be considered in future research.

This study’s results showed that enhancing the expected product lifetime of refrigerators could contribute to prolonging the actual product lifetime. Previous studies have contended that the expected product lifetime could be affected by past experiences of product use (Cox et al., 2013; Jensen et al., 2021; Wieser et al., 2015). This implies that the product design affects the consumer’s expected product lifetime. Accordingly, product design improvements for product lifetime extension, as described in the New Circular Economy Action Plan in Europe (European Commission, 2020) and the Circular Economy Vision 2020 in Japan (Ministry of Economy, Trade and Industry, 2020), are important to extend the actual product lifetime of refrigerators, from the manufacturer’s perspective of product design and the consumer’s perspective of the expected product lifetime. Moreover, the results imply that persuading the consumers to extend the product’s use is an effective measure for achieving product lifetime extension. As indicated in the New Circular Economy Action Plan (European Commission, 2020), making information available to consumers on product lifetime and reparability is one of the sustainable product policy frameworks. A labelling system of product durability is a representative policy, which has been discussed in previous studies (Cox et al., 2013; European Economic and Social Committee, 2016; Jacobs & Hörisch, 2021). The results of the present study indicate that such a labelling system, that is, providing information on absolute product durability and improvements to product durability, urges product lifetime extension by increasing the consumer’s expected product lifetime.

Table 10 in ESM Appendix D shows that the expected product lifetime of old refrigerators, aged 1–5 years, in the replaced sample, was much shorter than the unreplaced samples. This implies that if people replace their refrigerators within a short duration of use, they expect shorter product lifetimes for the next product. This study indicates that extending the expected product lifetime can extend the actual product lifetimes of refrigerators; hence, decreasing malfunction in the early stage of refrigerator use may be important to avoid short expected product lifetime and achieve product lifetime extension. However, this study’s data could not be used for directly analysing the difference in the expected product lifetime of owned products’ use and different product use durations. To conduct a deeper discussion, data for the expected product lifetime should be obtained for analysis.

However, the degree of the effect on product lifetime extension based on an increased expected product lifetime for refrigerators is weaker than for air conditioners. This indicates that product design policies in a circular economy do not have a large effect on the extension of the lifetime of refrigerators. The causes for this difference in the extension effects based on the estimated parameters were further considered. Table 3 presents the parameter results for refrigerators, estimated in this study, and for air conditioners, estimated by Nishijima and Oguchi (2023), with \(\beta =0\). Since the variable units of the DDCM are the same for both the products, the magnitude of each factor in the DDCM could be compared directly by referring to the parameter values.

Table 3 Estimated parameters for refrigerators and air conditioners
Full size table

Regarding the parameters of annual electricity consumption \({\delta }_{1}\), the absolute value of the parameter for refrigerators (−1.926) is larger than for air conditioners (−0.791). Conversely, regarding the estimated parameters of product price \({\delta }_{3}\), the absolute value for air conditioners (−0.686) is larger than for refrigerators (−0.267). This implies that electricity consumption has a relatively larger effect on replacement decisions for refrigerators than the product price, whereas product price has a relatively larger effect on replacement decisions for air conditioners than electricity consumption. This difference could be derived from a difference in consumer understanding of the electricity consumption patterns of both products. Refrigerators are continuously operated; however, air conditioners are operated seasonally. This difference in product use situations could lead consumers to believe that the electricity consumption of refrigerators is higher than of air conditioners. It has been reported that the electricity consumption of refrigerators is higher than of air conditioners in terms of electricity consumption in Japanese households (Agency for Natural Resources and Energy, 2011; Ministry of the Environment of Japan, 2015). Therefore, consumers attach more importance to electricity consumption during refrigerator replacement than during air conditioner replacement.

Moreover, comparing the data on annual electricity consumption provided by Shinkhyu-san (Ministry of the Environment of Japan, n.d.), the mean annual improvement rates of electricity consumption over years (2015–2019) varied from 0 to 0.96% for air conditioners with nine cooling capacities (2.2, 2.5, 2.8, 3.6, 4, 5.6, 6.3, 7.1, and 8 kW), whereas the improvement rates varied from 3.2 to 7.2% for refrigerators with eight capacity classes (≤ 200, 201–250, 251–300, 301–350, 351–400, 401–450, 451–500, and 501–550 L). These values show that, compared with refrigerators, a drastic reduction in the annual electricity consumption of air conditioners can be expected through product replacement. This is an incentive for postponing the payment for product replacement. Therefore, the sensitivity of the product price of air conditioners is higher than that of refrigerators. Summarising the above findings, the reduction potential of electricity consumption by product replacement of refrigerators is higher than of air conditioners and the factor of extending product utility is less attractive for refrigerators than for air conditioners. This is one of the main reasons that the product lifetime extension by enhancing the expected product lifetime of refrigerators is lower than that of air conditioners.

Considering the efficient achievement of product lifetime extension, air conditioners have a relatively high priority for implementing product design policies to enable product lifetime extension; however, the relative importance of such policies for refrigerators is lower. Although this does not mean that the product design policies for product lifetime extension in a circular economy do not contribute to the actual product lifetime extension of refrigerators, additional policies or product design strategies are required to obtain the same degree of product lifetime extension effect for refrigerators as air conditioners. For example, a previous study has shown that consumer attachment to or engagement with products affects product replacement or disposal behaviour (Hou et al., 2020; Okada, 2001). Therefore, policymakers and product designers should consider methods to enhance consumers’ psychological attachment to their refrigerators to reinforce the product lifetime extension effect. Enabling customers to customise the appearance or offering policies that support consumers to care for their product could enhance their attachment, thereby enhancing the product lifetime extension. Clarifying the differences in the product lifetime extension effects by increasing the expected product lifetime of such products is beneficial for policymakers and product designers to plan strategies for product lifetime extension.

Limitations and Future Work

Data, such as the expected remaining product lifetime and the product use duration of old refrigerators that were replaced, were obtained through self-reported answers. Hence, the data may include bias, which must be considered while interpreting the study results. Especially for the expected remaining product lifetime, a way of obtaining more objective data should be developed.

Since the research objective is to quantitatively show the effects of an increase in the expected product lifetime for refrigerators, this study assumed a relationship in which changes in the expected product lifetime (or expected remaining product lifetime) affect decisions of retaining or replacing refrigerators. However, a more detailed relationship, including exogenous factors, should be studied. Such studies will help estimate the casual effect of the expected product lifetime and conduct insightful policy discussions for product lifetime extension.

Furthermore, the effects can differ between products due to social and demographic characteristics, such as income, age, and household size. For example, low-income people may not be able to afford to replace their products, which may affect their replacement decisions and the expected product lifetime. However, owing to the heavy computational burden of estimating the DDCMs and the limited data obtained from the survey, the expected differences could not be analysed in this study. Such analyses would provide concrete targets for implementing policies to increase the expected product lifetime and achieve efficient actual product lifetime extension. In addition, consumers’ replacement decisions in this study are limited to 2019. For obtaining longitudinal differences, analysing the effects for different years would provide suggestive implications about the effects of the expected product lifetime.

This study used a relatively simple form of the DDCM to simplify the comparison of the product lifetime extension effects between the two products and decrease the computational burden. As more sophisticated and econometrically robust estimation methodologies of the DDCM have been proposed (Abbring & Daljord, 2020; An et al., 2021; Berry & Compiani, 2020), the DDCM should be estimated using these approaches to obtain robust results.

The product lifetime extension effects of the expected product lifetime were analysed for certain products (i.e., air conditioners and refrigerators) in one country (i.e., Japan). Although the ways of setting the DDCMs should be discussed, it is important to analyse the effects of the expected product lifetime of other products in multiple countries for a comprehensive understanding. For example, the Information and Communication Technology (ICT) products or relatively small electronics, such as laptops and smartphones, would be beneficial to analyse the effects. Moreover, analysing the differences in the effects of the expected product lifetime on reasons for product replacement is important to consider effective policymaking for product lifetime extension. Although this could not be analysed due to data limitations, the authors intend to study it in future.

It is crucial to demonstrate the extent to which the product lifetime extension based on the expected product lifetime could contribute to a reduction in environmental burdens, such as resource consumption or greenhouse gas emissions. As the structure of the product lifetime model expressed by the DDCM in this study differs from that used in previous studies, expressed by engineering product lifetime distributions, such as the Weibull distribution, this study’s model cannot be applied directly to the environmental analytical frameworks used in related studies (Budzinski et al., 2020; Chahoud et al., 2021; Kagawa et al., 2011; Nishijima, 2016). Although several previous studies have analysed the product flow and environmental effects along with product lifetime changes by policy implementations using the DDCM (Nakamoto & Kagawa, 2018; Nishijima et al., 2020), their focus was on the changes in a part of the product lifetime and failed to address the changes in the entire product lifetime. To integrate the product lifetime model in this study with such environmental analytical frameworks, future studies should focus on substantial improvements to these models.

Conclusion

This study estimated the effects of the expected product lifetime on the actual product lifetime of refrigerators in Japan and compared the results with those for air conditioners. This study showed that the effects of the expected product lifetime differ based on products; hence, implementing uniform policies to enhance consumers’ product lifetime expectations for all durable products cannot achieve sufficient product lifetime extension in a circular economy. Rather, differences in the extent of the effect should be considered. Understanding the differences in the effects of the expected product lifetime on the actual product lifetime among products is beneficial for formulating effective policies for product lifetime extension. Following the results of this study, focusing on mitigating the physical limitations of product longevity is effective for actual product lifetime extension of air conditioners. However, for refrigerators, additional methods to achieve such mitigation are required to obtain the same degree of product lifetime extension. Hence, this study contributes to a framework for clarifying these differences. For the application of the study’s findings to other products, various strategies for product lifetime extension should be considered from wider perspectives based on the differences in the effects of the expected product lifetime. Such considerations would lead to the development of business models and reinforce economic competitiveness in a circular economy. Moreover, the effects of the expected product lifetime on the actual product lifetime differ based on products and countries. The methodology provided in this study could be applied to other countries for devising product lifetime extension strategies from a global perspective.