A dataset on energy efficiency grade of white goods in mainland China at regional and household levels

To improve energy-saving management, the energy efficiency grade (EEG) was introduced by the Chinese government in the 2000s and mainly implemented for white goods (WGs) in early stages. However, due to the lack of actual statistics, how effective the promotion of high EEG WGs has been in China is still not clear. The China Energy Efficiency Grade (CEEG) of WGs dataset described here comprises (i) EEG-related data on 5 kinds of WGs at the regional (national, provincial) and household levels in China and (ii) predictions of future average EEG trends. By web crawling, retrieving and processing in SQL, the average EEG data weighted by sales in 30 provinces in mainland China from 2012 to 2019 are provided. Household WG survey data, including household information and average EEG, were collected by distributing questionnaires to 1327 households in Beijing, China. The CEEG dataset will facilitate the advancement of research on household energy consumption, household appliance consumer choice, and the assessment of energy efficiency-related policies.


Figure S1
An example of RP information manual retrieval. Figure S2 The market share of WG in the studied period.

Table S1
Details on obtaining EEG information by web crawling and manual retrieval. Table S2 A review of existing surveys related to the EEG, household energy consumption, and energy-saving attitudes conducted in China. Table S3 Descriptive statistics of the basic information in the household WG survey. Table S4 The results of REPM OLS method. Table S5 The results of HEPM Probit method. Text S1 Mathematical descriptions of the performance evaluation indicators. Text S2 Explanations of three categories of HA. Text S3 Explanations of the regional data. Text S4 Explanations of the household data.

Data cleaning
Figure S1 An example of RP information manual retrieval.       Text S1 Mathematical descriptions of the performance evaluation indicators.  (1) and (2), TP means the actual category is 1, and the identification category is 1; FN means the actual category is 1, and the identification category is 0; FP means the actual category is 0, and the identification category is 1; TN means the actual category is 0, and the identification category is 0.

Machine learning model establishment
Additional explanations on regional data Figure S2 The market share of WG in the studied period. Text S2 Explanations of three categories of WGs.
To facilitate discussion, the studied appliances are divided into 3 categories based on market performance (Figure S2), the relationship with socioeconomic variables, and other factors: the phase-out category (IWMs, ACs), phase-in category (DWMs, VACs, DWs), and compete-with-gas category (EWHs). For the phase-out category, ACs were completely withdrawn from the market in 2021 due to the revision of EEGrelated standards, and their market share declined during the studied period; IWMs might be gradually replaced by DWMs due to their disadvantages, such as more severe damage to clothes and higher water consumption. For the phase-in category, DWMs and VACs replace the original market shares of IWMs and ACs, respectively. For the compete-with-gas category, EWHs have gas water heaters that perform similar functions but use gas to compete with them.
Text S3 Explanations of the regional data.
From 2012 to 2019, the market share of WGs in the phase-out and compete-with-gas categories decreased. Specifically, the market share of IWMs decreased from 78% to 45%, that of ACs decreased from 59% to 23%, and that of EWHs decreased from 69% to 54%. In contrast, the market share of WGs in the phase-in category increased. Specifically, the market share of DWMs increased from 22% to 55%, and that of VACs increased from 41% to 77%.
As shown in Fig. 4a in the main text, the energy efficiency of IWMs, DWMs, EWHs, and VACs in the 2012-2019 period showed upward trends (i.e., the EEG values decreased). This result indicates that WGs purchased in China were more energy efficient. However, the opposite trend was observed for ACs. Additionally, a significant shock to the average EEG of ACs was observed in 2014. This finding might be because of the end of energy-efficient household appliance subsidy policies in 2013, according to All View Cloud (https://www.avcmr.com/article/detail?id=228166870681780224). Thus, consumers tended to buy cheaper but less energy-efficient ACs. At the provincial level, the trends of the upper pivot, median, and lower pivot of the box line plot (Fig. 4b to Fig. 4f in the main text) of the provincial average EEG distribution over time were all similar to the trends of the national average EEG.

Text S4 Explanations of the household data
According to the survey data, the existing WGs in 79% of households were grades 1-2 (i.e., high EEG), and when purchasing WGs, 87% of households considered the EEG to be a vital factor. That is, in most of the surveyed households, behaviors and preferences with regard to purchasing and using WGs tended toward energy-saving WGs. Additionally, Fig. 4g and Fig. 4f show that from 2019 to 2021, the percentage of high average EEG households increased from 76% to 85%, and the percentage of households considering the EEG when purchasing WGs grew from 81% to 90%. Therefore, residents' attitudes toward acquiring energy-efficient WGs are becoming more positive, and households are gradually becoming energy-efficient in the process of gradually updating their WGs. This indicates that more households could have a higher average EEG in the future.