1. Introduction
Recently, environmental pollution has become a serious problem in our society. Many countries seek to prevent environmental pollution, since issues such as global warming and air and water pollution have become evident. The common feature of policies addressing these problems is that they seek to change individual behavior. The encouragement and activation of pro-environmental consumption is a good example. To encourage sustainable consumption, we have to identify the determinants of pro-environmental consumption and analyze factors that disrupt this consumption. Many countries develop policies related to encouraging pro-environmental consumption, but the effectiveness of these policies remains unclear. The following is an example of a Korean pro-environmental consumption policy; the Korean government has formulated and enforced various environmental policies intended to promote pro-environmental consumption in compliance with the 2011 Act on Encouragement of Purchase of Green Products. The total value of goods purchased by 883 government agencies and public organizations was 2.2 trillion Korean Won in 2014, and this purchase quantity has increased every year since then [
1]. However, the performance of the revitalization of green product consumption was insufficient amongst other economic agents, excluding public organizations. For example, the Green Card Point Scheme (Since 2011, green card holders have received some points for purchasing certified green products and for leading a low-carbon environment-friendly life) that was adopted as part of the environmental policy for consumers to promote eco-friendly consumption was far from successful, with Green Card application rates of no greater than 10.7% to 24.4% [
2].
According to the Third Green Product Purchase Promotion Plan, the government’s budget allocation to promote environmentally friendly consumption has increased [
3]. This increased budget includes allocations for incentive policies, educational programs, and public relations activities to boost pro-environmental consumption. However, there is a paucity of quantitative data and analyses concerning the effectiveness of educational programs and the changing public awareness of pro-environmental consumption and environmental consciousness. A range of incentives, educational programs, and public relation campaigns intended to promote environment-friendly consumption are not well received by the general public in reality [
4]. Research on consciousness regarding pro-environmental consumption has been conducted occasionally [
5], but there are few environmental indexes that diagnose the status of pro-environmental consumption. Previous indexes relevant to environmental policy are included in the Green Competitiveness Index [
6], but this index is insufficient for diagnosing the status of pro-environmental consumption in Korea. Looking more closely, this index focuses on green growth through a regional green strategy. It is classified into three categories; the basis of green growth, the promotion of green growth, and the performance of green growth. Among these, one of the four elements that evaluate promotion of green growth is the measuring of green living. However, this measurement is only evaluated through bicycle use, resource recycling, energy conservation, and traffic volume reduction. Thus, it is actually difficult to measure pro-environmental consumption through this indicator [
7].
Meanwhile, the 9th OECD (Organisation for Economic Co-operation and Development) Working Party on Integrating Environmental and Economic Policies has discussed how to measure local residents’ interest in the environment by using search queries based upon big data, such as Google Trends, and how to apply these measurement results to the development of environmental policy. Portal users normally express the intention or purpose of their search through the search keywords used [
8]. Based on the characteristics of big data queries, recent studies measured pro-environmental attitude and used it to analyze the impact of the introduction of pro-environmental policies [
9]. This is noteworthy in that queries based on big data are used to measure pro-environmental consciousness and to estimate policy implications from the analysis results without using the questionnaire survey, which is a primary conventional method for measuring pro-environmental attitude.
Therefore, the purpose of this study was to investigate the determinants and inhibitory factors of pro-environmental consumption using the pro-environmental consumption index that was suggested in this study. In other words, this study identifies factors that impact the pro-environmental consumption index based upon the analysis of data from 13 developed countries and data from the World Bank. We chose 13 counties from the six continents; Asia, South America, North America, Europe, Australia and Africa. We considered Argentina, Brazil, Mexico, South Africa, the United States, Canada, Australia, the United Kingdom, Hungary, Spain, China, Japan, and Korea. The proposed pro-environmental consumption index based on big data queries reflects consumers’ collective consciousness and can be used to understand the present status of pro-environmental consumption. Since the analysis of both the determinants and the inhibitory factors of pro-environmental consumption should be preceded by establishing pro-environmental consumption policies [
10], the results of this study could provide policy implications for how to improve pro-environmental consumption.
This paper has the following structure.
Section 2 examines the definition, motivation, measurement, and determinants of pro-environmental consumption based upon a review of the literature.
Section 3 explains the methodology of the proposed pro-environmental consumption indicator using big data queries.
Section 4 analyzes the correlation between the proposed pro-environmental consumption index, derived from queries based on big data and indexes based upon conventional questionnaire surveys, to test the reliability of the proposed index and uses the index to comparatively analyze the trends and statuses of 13 countries over the past 12 years from 2004 to 2015. Moreover, the factors impacting pro-environmental consumption indices are determined with a regression analysis to analyze the determinants of pro-environmental consumption. Finally, we present our conclusions and discuss the policy implications and limitations of the present work.
2. Literature Review
Pro-environmental consumption behaviors refer to “consumers purchasing/using/disposing of products and services, considering their impacts on others, society, and the environment” [
11,
12,
13]. Pro-environmental consumption, which previously used to imply environmental protection, is viewed as encompassing consumption geared towards sustainable development from a broader perspective [
14].
Pro-environmental consumption is divided into the purchase, use, and disposal steps [
12,
15,
16,
17]. The purchase stage refers to “searching the environment-related information to assess the quality of products based on their impacts on the environment, as well as purchasing the pro-environmental products, e.g., energy- and resource-saving products and those causing less environmental pollution and waste” [
16]. The ‘use’ stage refers to ‘saving energy or resources considering the impacts on others or the environment before one’s own convenience’. The ‘disposal’ stage refers to ‘pro-environmental consumption behavior oriented toward conservation, involving active engagement in recycling or reuse of resources’ [
16].
To measure pro-environmental consumption, previous studies used questionnaire surveys. Particularly, pro-environmental behaviors were measured as the practical pro-environmental lifestyle encompassing the purchase, use, and disposal stages [
11,
12,
13], the acceptance of pro-environmental products [
18], and the willingness to pay the extra cost of pro-environmental products [
19]. Specifically, to measure pro-environmental consumption, previous studies first developed questionnaire items concerning purchase, use, and disposal, and based indexes upon the results. The questionnaire items concerned experience purchasing pro-environmental products in the purchase stage, the extent of energy savings in the use stage, and recycling and reuse in the disposal stage.
Extensive research on diverse indexes for measuring pro-environmental consumption has been conducted around the world. Since 2011, the Green Life Index has been developed based on a biannual survey [
20]. This survey is conducted among approximately 19,000 household members aged 20 and older in 9700 sample households across various countries [
20]. The Green Life Index is composed of a green practice index and a green performance index. The practice index consists of green community, green family, and green traffic categories.
Table 1 gives detailed information on the green living indicators.
In addition, the Korea Consumer Agency has conducted evaluation research on green consumer capability and developed an index of evaluation measuring a green consumer’s capability and level of green consumption, targeting the housewife consumer. Hwang and Lee [
11] suggested an index that consisted of inner competence, external competence, and practical competence, and they calculated the respective weights of these categories as 32.3%, 29.3% and 38.4%, based on a survey [
11].
Also, there exists the Greendex, an international indicator of pro-environmental consumption.
Figure 1 visualized Greendex indicators. National Geographic and Globescan research developed this index to include 32 indicators representing the fields of housing, transportation, food, and consumer goods and have conducted a survey every two years since 2008 in 18 countries, including South Korea, in order to measure and monitor environmental consumption. Greendex conducted an online survey consisting of 64 response variables related to lifestyle and behavior, including energy use, transportation, and food consumption. They collected questionnaire responses from 1000 consumers in each country and transformed the results of the survey into scores of 0 to 99 [
21].
There has also been a survey of American consumers. Mediamark Research & Intelligence [
22] has synthetically analyzed various items about consumer shopping behavior and media user behavior by conducting a survey every year since 2003 in order to understand consumers and provide marketing solutions. They measured consumers’ green consumption status and attitude about the environment by researching which pro-environmental products were used within the last 6 months.
Although it is not a pro-environmental consumption indicator, the National Environmental Scorecard prepared by the League of Conservation Voters [
23] compares the degree of policy drafts for pro-environmental purchasing. The twenty experts of the League of Conservation Voters have examined the degree of pro-environmental policy drafted in each American state since 1970 and have calculated annual scores from 0% to 100%, according to the ratio of drafted pro-environmental policies to all pro-environmental policies. The pro-environmental policy score is calculated based upon 12 items, including air, clean energy, climate change, and so on. The League of Conservation Voters’ indicators have been used as pro-environmental policy indicators for USA states in many papers [
23]. Lyon and Yin (2010) used the League of Conservation Voters’ indicators to measure pro-environmental behavior [
24].
Recently, some research works have included the use of big data queries to measure pro-environmental indicators. Lee, Kim and Lee [
9] measured pro-environmental behavior using portal queries and analyzed how the environmental attitudes of local residents affect the introduction of pro-environmental policy. Similar to Lee, Kim and Lee [
9], herein we measured pro-environmental attitudes and anti-environmental attitudes by using big data queries on the search terms ‘recyclables’ and ‘disposables’. Then, we verified the indicators through reliability testing and conducted regression analysis to find out the determinants for pro-environmental consumption. The presently proposed method of measuring pro-environmental consumption by means of search queries based on big data has a few strengths over the previous questionnaire survey method. First, the use of big data saves a great deal of cost in comparison to the use of questionnaire surveys. Second, the proposed method precludes the distorted social desirability or response biases that are characteristic of questionnaire surveys.
4. Results and Discussion
4.1. Results of the Pro-Environmental Index
4.1.1. Reliability Verification of the Pro-Environmental Consumption Index Using Big Data
Greendex provides data every two years from 2008. Compared with Greendex, it is possible to expand the pro-environmental consumption index by collecting open data about search frequency for ‘recyclables’ and ‘disposables’. Prior to using the pro-environmental consumption index proposed by the amount of internet searches, we conducted a reliability verification between Greendex and the pro-environmental consumption index.
Reliability testing was conducted by using the collected data. At the 99.9% level of significance, the correlation between Greendex and the proposed index had a
p-value of 0.0038 and a correlation coefficient of 0.3716. This verified the positive correlation between the two indexes.
Figure 2 visualizes the relation.
Figure 2 indicates the similarity between the pro-environmental consumption index, based on big data, and the Greendex consumption index. In conclusion, we could verify that the pro-environmental indicator that we proposed is positively correlated with Greendex.
4.1.2. Analysis of Pro-Environmental Consumption Status Using the Proposed Index
To determine the status of pro-environmental consumption, the proposed index was used to comparatively analyze the trends of pro-environmental consumption by country, including Korea, and by period of time. We analyzed the present conditions of 13 countries for 12 years from 2004 to 2015. During the analysis period from 2004 to 2015, the average pro-environmental index was highest for Argentina and Brazil, both exceeding 15. This seems to be related to the pro-environmental boom of Argentina. South Korea’s score is just at the average level of about 3.13, compared to the 12 other countries.
Figure 3 shows the changes of the pro-environmental indicator over the analysis period.
Pro-environmental consumption showed a steady increase from 2004 to 2012, but tended to decrease continuously thereafter.
4.2. Regression Analysis
To explore which determinants affect pro-environmental consumption level, we conducted regression analysis. This study used four models to examine the effect of the variables consistently. Model 1 is a base model based on control variables. Health expenditure, number of people aged 65 and above, and pre-primary education were used as control variables in Model 1, based on previous research. Models 2, 3, and 4 were constructed by adding and changing independent variables, such as low GDP countries’ GDP, high GDP countries’ GDP, and past orientation, based on Model 1. Assumptions for each variable were discussed in 3.2.2.
Table 6 summarizes these results. The
F-values were statistically significant for all models.
Surprisingly, the health expenditure variable was meaningfully negatively correlated with pro-environmental consumption for all models. Based on Model 4, when per capita health expenditure increases by USD 1000, the pro-environmental indicator decreases by 2.1 points. With increasing healthcare spending, the disposable budget for pro-environmental products, which are less affordable than ordinary products, is likely to decrease.
The number of people aged 65 and above variable was statistically significantly negatively correlated with the pro-environmental consumption index at a significance level of 0.1% for all models. In Model 4, a 1% growth in the population aged 65 and above corresponds to a decline in the pro-environmental consumption index of 0.847. Given that the maximum value of the pro-environmental consumption index is 100, the size of the population aged 65 and older seems to exert substantial effects on the pro-environmental consumption index. It cannot be ruled out that the older population may in general be less able to afford pro-environmental products, owing to their tighter budgets in comparison to other age groups.
The preprimary education variable was statistically significantly positively correlated with the pro-environmental consumption index at the 0.1% level of significance for all models. Based on Model 4, an increase of 1 year in preprimary education corresponds to an increase of 4.3 in the pro-environmental indicator. This variable is relevant to early formal educational programs, suggesting the importance of early education in pro-environmental consumption. This finding is consistent with many previous reports regarding the positive relationship between education and pro-environmental behaviors [
41,
42,
43], a tendency that is found for elementary education as well.
The interaction variable low GDP countries’ GDP was analyzed in Models 2 and 4, and the results were statistically insignificant. In contrast, the high GDP countries’ GDP interaction variable had statistically significant positive effects on the pro-environmental consumption index at the significance level of 0.1%. This finding indicates that higher GDP levels have positive effects on pro-environmental consumption. For the advanced countries with a per capita GDP of USD 30,000 or more, an increase in GDP of USD 10,000 corresponded to an increase in the pro-environmental consumption index of 1.9.
Lastly, the past orientation level had statistically significant negative effects on the pro-environmental consumption index. In Model 4, the pro-environmental consumption index fell by 3.329 for an increase in the past orientation of 1. As mentioned by Joireman, Van Lange and Van Vugt [
4], pro-environmental behavior parallels concern about future generations. A past orientation on a national level could have adverse effects on pro-environmental consumption. Lee, Lee and Choi [
50] highlighted a statistically significant relationship between past orientation and suicide rates, suggesting that people’s perception or cognition regarding time points is an important factor impacting their behavior; such behaviors affected by this orientation could include pro-environmental consumption as well as suicide.
This study proposed a pro-environmental consumption indicator that could quantify the present condition of pro-environmental consumption based upon big data queries. The reliability of the proposed indicator was verified by analyzing the correlation between the proposed index and an existing pro-environmental consumption index drawn from survey data. We used the indicator to quantify the present condition of pro-environmental consumption in 13 countries during the 12 years from 2004 to 2015. Moreover, based on the proposed indicator, we identified determinants to suggest directions for plans to activate pro-environmental consumption.
5. Conclusions
The results of this study are as follows. First, the proposed pro-environmental consumption index based on big data seems to represent pro-environmental consumption well. At the 99.9% level of significance, the proposed index was positively correlated with the existing Greendex index, with a p-value of 0.0038 and a correlation coefficient of 0.3716. Second, we analyzed the present condition of pro-environmental consumption for 13 countries using the proposed index. The pro-environmental consumption indexes were highest for Argentina and Brazil, both exceeding 15, and the index for Korea was just average among the countries studied, at about 3.13. Third, regression analysis identified variables that affected the index meaningfully. Especially health expenditure, the proportion of population aged 65 and above, and past orientation were all negatively correlated with the pro-environmental consumption index, and pre-primary education and GDP in higher GDP countries were positively correlated with the pro-environmental consumption indicators in Models 3 and 4.
By suggesting a pro-environmental index based upon search query data that could represent consumers’ collective consciousness, we proposed a useful method to analyze the status of pro-environmental consumption compared with existing questionnaires. This paper has several limitations. First, the sample size is too small to take into account autocorrelation over time and the nesting of the index in each country. Therefore, we checked the basic assumptions of the OLS before analyzing determinants via OLS. We conducted a Durbin-Watson test to verify auto correlation. The test result shows that the error terms had independency because the error terms caused little auto correlation. Second, a weak correlation of 0.3716 was found for reliability verification between Greendex and the proposed index because we considered only the purchasing step of pro-environmental consumption and examined the determinants of this step. In future research, the factors of pro-environmental consumption related to the purchase, use, and disposal stages should be analyzed. Third, although researchers have difficulty collecting individual level data because of cost, compared to collecting country level data, Franzen and Meyer [
27] insisted that differences between individuals are larger than countries. In addition, despite their strength in representing the consumers’ collective consciousness, queries based on big data cannot directly indicate the consciousness of individual consumers. In other words, although big data queries can provide information about overall orientation, they do not present clear evidence at the individual level regarding whether individual search keywords imply positive or negative attitudes. Hence, further study is needed to explore a methodology for representing pro-environmental indexes for different subgroups of consumers from a macroscopic perspective. Furthermore, more specific strategies conducive to promoting pro-environmental consumption are needed.