key: cord-0761990-8jvnkbeu authors: Toosty, Nishat Tasnim; Hagishima, Aya; Bari, Wasimul; Zaki, Sheikh Ahmad title: Behavioural changes in air-conditioner use owing to the COVID-19 Movement Control Order in Malaysia date: 2022-01-06 journal: Sustain Prod Consum DOI: 10.1016/j.spc.2022.01.001 sha: 9866b39cc844e40f8bf02e6770a8c5369c978119 doc_id: 761990 cord_uid: 8jvnkbeu Remote work (working from home) became a norm rather than an exception for the global workforce during the COVID-19 pandemic, influencing every facet of life in both positive and negative ways. The stringent action of the Malaysian government in enacting the Movement Control Order (MCO) motivated the investigation of its impact on the energy consumption behaviour of working people regarding air-conditioner (AC) use. To this end, this study conducted a cross-sectional survey through an online platform. An ordinal logistic regression model (ORL) was used to analyse the collected data of 1873 respondents to determine the factors influencing the ordinal variable of interest, AC-usage behaviour during remote work. Next, the variable with unordered categories, the MCO-induced change in AC-usage behaviour, was analysed using a multinomial regression model (MLT) to identify the potential determinants. Finally, a reason analysis unveiled aspects behind the transition in AC use during remote work. This study identified stopping AC use during remote work despite using it at the office before the MCO period as the most significant change in AC-usage behaviour due to MCO. This change was frequently adopted by people with medium-level incomes and high electricity bills. By contrast, participants unfamiliar with their electricity bill were most likely to start AC use during remote work, although they did not use it before the MCO. Participants working remotely in the communal spaces of their houses preferred to stop using ACs during MCO compared to private room users. Furthermore, age group and ethnicity significantly influenced AC-usage behaviour in remote work and changes in such demeanours. These findings recommend policy interventions to expedite limited AC use for a sustainable energy sector, even during future climatic emergencies. The emergence of the novel coronavirus (SARS-CoV-2) has marked the onset of a new era of uncertainty and immense suffering, leaving the entire world in turmoil. The outbreak of COVID-19 has impacted every corner of the world. The first case presented with pneumonia of unknown cause in Wuhan on 31 December 2019 (1). This was soon followed by a surge in infections and increasing death tolls in China and many countries worldwide, leading to a global pandemic (2) . To combat the pandemic, countries worldwide imposed restrictions on the movement of their citizens, launching a new lifestyle generally known as "lockdown". These restrictions have substantially affected almost every sphere of life, including livelihoods, economic activities, health 3 facilities, and social systems, has been substantially affected (3) by these restrictions. Organisations have been compelled to design working and economic activities while conforming to the restrictions. Thus, lockdowns have led to a shift in the perspective of remote working by confining all official, production, and service-related operations as well as educational activities to the home environment. This new lifestyle plays the role of a social science experiment with a remarkable scope for energy scientists. Life under lockdown has entailed a notable shift in energy use habits, which has attracted the attention of researchers worldwide. Disruption of the wellestablished daily routine while adapting to the newly restricted life could redistribute the peak demand period and moderate the peak-hour consumption, offering more flexibility in using home appliances (4, 5) . A study on New York City's energy consumption reported a significant increase in domestic energy consumption. An increase in individual energy demand during the usual working hours on weekdays has also been reported here. This study discussed the possibility of additional cooling load during the summer days, which may worsen the economic burden on people due to increasing electricity bills (6) . European countries with stringent preventive measures, such as Spain, Italy, Belgium, and the UK, recorded an electricity consumption profile that was observed during the weekends in 2019 preceding the pandemic period (7) . Consumers in the UK adopted a smoother usage pattern during morning peak hours by postponing their early-morning activities to until later (8) . By contrast, with no lockdown during the pandemic period, Sweden illustrated an increase in electricity consumption compared to the pre-pandemic days (7) . After the enactment of social distancing in Brazil, the overall electricity loads fell, and the highest reduction was observed in the South subsystem (9) . Despite the pandemic-induced drop in total electricity consumption, the lockdown 4 substantially exacerbated the prevailing energy poverty worldwide by increasing domestic energy demand amidst the financial crisis (10) . Overall, the energy consumption behaviour of people worldwide is being restructured throughout the pandemic and accompanying restrictions. This situation necessitates a subtle analysis of this issue, especially when climate change has already made it essential to attain sustainability in energy, social ecology, and economy (11) . Many studies have reported a positive association between climate change and energy consumption. In Shanghai, China, changes in daily temperature significantly increased electricity consumption, and this association was found to be more prominent during warm days than during cold days (12) . Climate change is projected to increase the additional peak load in the USA during the summer (13) . The trend of energy demand for heating and cooling has increased since 1990 due to climate change. It is projected to continue to increase until 2030 (14) . The change in the climate of Melbourne, Australia, which generally has a temperate and oceanic climate, was estimated to lead to an increase in peak demand for electricity from 2000 to 2100. This study considered the ownership of air conditioners (ACs) during the early phase of the 21 st century (15) . Most of these studies observed a more pronounced climate-energy association for the warm period, and ACs were identified as the main factor catalysing this relationship. Therefore, the change in energy consumption behaviour due to the newly restricted lifestyle can be prominent on hotter days. Malaysia is characterised by hot and humid weather throughout the year; thus, AC is a commonly used cooling appliance. To ensure the thermal comfort of the occupants, 53 % of the electricity consumption in buildings in Malaysia, from both commercial and residential sectors, can be attributed to ACs (16) . According to the Laos Census 2015, 41.6 % of the households in Malaysia contain ACs, while the capital of Malaysia, Kuala Lumpur, reports 45.6 % individual AC 5 ownership (17) . The AC-usage pattern of Malaysian people during the pandemic can play a salient role in attaining energy sustainability, as they had to maintain a different lifestyle, embracing strict restrictions. The Malaysian government implemented the "Movement Control Order" (MCO) on 16 March 2020 to prohibit mass movements and all types of gatherings to curb the spread of coronavirus. With the MCO, all business organisations were restricted, except those engaged in manufacturing and supplying food, personal care, and pharmaceutical products (18) . The control order was extended to 3 May 2020 after which various reforms were initiated, curtailing the degree of restriction (19) . The Energy and Natural Resources Minister of Malaysia reported that the MCO promoted domestic electricity use by 23 %, although the overall electricity use decreased by 33 % during the MCO period (20) . To withstand the MCO-time economic burden, the national energy provider of Malaysia, Tenaga Nasional Berhad (TNB), announced tiered rebates on electricity bills through the Bantuan Prihatin Elektrik (BPE) package (21) . The government of Malaysia also implemented several economic stimulus packages, along with discounts on electricity bills, for the residents of Peninsular Malaysia. Thus, the behaviour of Malaysian consumers regarding a typical cooling appliance (AC) during the MCO period, in response to the abovementioned packages, could be highly informative in the field of building environmental engineering. Furthermore, any significant change in AC-usage behaviour owing to MCO implementation may highlight the AC-use behaviour with a focus on achieving energy sustainability. To this end, this study conducted a questionnaire survey to investigate the behavioural changes in work-time AC use between office work (pre-MCO) and remote work (during MCO). The collected data were analysed to detect the potential factors related to these changes. Through this analysis, we examined how this new evolving practice can be targeted to establish a more sustainable and energy-efficient lifestyle. 6 The concept of remote work or telework is not merely an innovation of the pandemic, as this work arrangement has already been established and gradually embraced over the past several years. However, the outbreak has helped facilitate its acceptance among workers (22) . Among the European Union countries, the pre-pandemic teleworking modalities were mostly occupied by knowledge-intensive services, educational activities, finance-related work, information communication technology professionals, high-paid employees, and business managers. (23) . The teleworking space has been expanded to low-and mid-level clerical employees to comply with COVID-19 restrictions (24) . Thus, the pandemic has changed global work cultures, particularly in countries with high transmission rates. A new era of teleworking with all its benefits and pitfalls has emerged through the outbreak. It may even be sustained even after the pandemic days (25) . Therefore, it is important to work on the challenges associated with the teleworking space to achieve the highest productivity and ensure work-life balance. In Malaysia, 67 % of the active companies adopted remote work. Among the working people, multinational corporate officials, staff with salaries more than MYR 6000 (USD 1442.13), and the inhabitants of Kuala Lumpur were more likely to embrace teleworking during the COVID-19 pandemic (26) . However, this trend has already started to gain popularity during the pre-pandemic period (27) . Most Malaysian people are still struggling to adapt to this new working environment. At the same time, companies are worried about the effects of telework adoption on productivity. A comfortable virtual working environment can ensure higher productivity and the welfare of companies (28) . Vischer (29) emphasised the impact of physiological comfort on worker performance. Ambient environmental conditions, such as lighting, acoustics, ventilation, and thermal comfort, were highlighted in this case. Poor indoor environment quality (in terms of 7 lighting, air quality, and noise insulation) was also found to deteriorate the adequacy of the teleworking space (30) . For Malaysia, indoor thermal comfort has been an imperative research topic because of its geographical location. The year-long exposure of buildings to high temperatures (with average daily temperature ranging from 21 °C to 32 °C) and humidity-as well as high-intensity solar radiation-is the usual climatic condition of Malaysia (16, 31) . The continuous increase in AC installations and energy consumption (16, 31, 32) has motivated researchers to seek a reasonable balance between indoor thermal comfort and environmental impact. The abundant use of AC has led to several studies encouraging energy conservation behaviour to develop sustainable energy for future generations. Ting et al. (33) demonstrated the urgent need to save energy as a preventive measure against energy sustainability threats in Malaysia and other countries throughout the world. Jalalkamali and Abbas (34) suggested a significant reduction in energy use by raising awareness among occupants regarding energy sustainability. Some psychological aspects, such as attitudes, norms, behaviour control, habits, motivation, and knowledge about energy, were successfully identified as determinants of occupants' energy-saving behaviour (35) . Therefore, it would be interesting to determine whether the energy consumption pattern of people with the wider application of remote work can encourage them to adopt an energy-efficient lifestyle. Several studies have already projected energy transitions due to the enforcement of movement restrictions in various countries worldwide during the pandemic. For example, Qarnain et al. (36) determined that social distancing and home quarantine were two of the most influential factors for intensive energy conservation behaviour during the pandemic in India. Jiang et al. explored the impact of the pandemic on energy consumption and demand, highlighting the environmental 8 impacts and lessons regarding energy recovery (37) . In China, the impact of lockdown on domestic energy consumption was determined along with predicting the post-lockdown consumption pattern (38) . A substantial increase in energy consumption due to household cooking, entertainment, heating and cooling, and lighting was identified. However, only the cooking-related consumption pattern was predicted to shift during the post-lockdown period. A study focusing on the Lagos metropolis of Nigeria compared the overall electricity demand of various sectors, such as residential, commercial, and industrial, before and during the partial lockdown (3) . No significant change in the share of overall electricity demand among these sectors was observed in this study. However, the enforcement of total lockdown depicted a different scenario, with a sharp increase in residential electricity demand and a declining demand within the industrial sector. Most of these studies focused on the overall energy consumption or electricity use in a building, rather than on any specific equipment. To project the perceived change in a definite way, the current study analysed the impact of the MCO on energy consumption behaviour through the most common cooling appliance in Malaysia (AC). This study attempted to determine the behavioural changes in AC use due to COVID-19 MCO enforcement in Malaysia. Although regional variation is evident in the climate, this country has no significant seasonal changes over the year. Being located near the equator, Malaysia has an equatorial climate, hot and humid, with a year-long summer (except in high-altitude areas) and rain. However, high temperatures persist throughout Malaysia. The average temperature of plain areas ranges between 25-32 ºC, with 75 % humidity throughout the year, and it is sufficient to 9 make one wet without rain. The daily average temperature and relative humidity (RH) of Malaysia from July 2019 to June 2020 are shown in Fig. 1 . A cross-sectional survey, spanning from 12 May to 19 May 2020, was conducted using an online platform by a Malaysian online research company. Although the survey period coincided with the hottest month (May) in Malaysia, it enquired about AC-usage behaviour during the MCO period. Thus, it concentrated on hot summer days, which are common in the usual climate of Malaysia. The data were collected by targeting people who visited the associated online publisher websites (OPWs), such as online news portals, sports, and lifestyle websites. People browsing these high traffic websites during the survey period noticed a single question from a series of questions in the survey in the form of pop-up advertisements. On the visitors' subsequent visits to various hightraffic websites, they were shown the consecutive single question of the survey. Repeated and regular visits to consume content over various OPWs made it possible for visitors who passed the screening process to complete the questionnaire. As this study was purported to unveil new ACusage behaviour in remote work, the screening question enquired whether the respondents worked remotely during the MCO period. Only the visitors of the associated OPWs who adopted MCOdays remote work were eligible to participate in the survey. It was a completely anonymous survey where eligible visitors participated based on their willingness and were free to answer or skip any survey questions. All procedures performed in this work regarding data collection and analyses were in accordance with the ethical standards of Kyushu University. The questionnaire (attached as supplementary information in the S1 file) of the survey comprised 12 questions. It enquired about demographic characteristics, such as age, gender, and ethnicity, followed by some social status-related questions, such as monthly household income and residential house type. Each question provided a specified number of options from which the respondents had to choose their answers. However, the second part of the questionnaire concentrated on the most important enquiry, which was how frequently a respondent used ACs while working remotely during the MCO period. One possible answer to this question was "AC is not installed in-house". The respondents who chose this option (10.99 % of the respondents) were excluded from the analyses. Additional questions regarding remote working rooms, reasons for both using and not using an AC during the MCO period, monthly household electricity bill, and previous AC-usage behaviour at the office were also incorporated into the questionnaire. This study analysed data extracted from 5076 participants who answered the most important question for this work: how frequently they used ACs while working remotely during the MCO period. The analysis procedure was divided into three sections: AC-usage behaviour during remote work, changes in AC-usage behaviour due to MCO, and reasons for such changes. 12 The first section analysed AC-usage behaviour during the MCO period, extracting information on the frequency of AC use while working remotely. This variable was the response variable of this section and named as "AC-usage behaviour during MCO". It was a categorical variable with four categories (not used, rarely used, sometimes used, and always used). The explanatory variables for this section were divided into four groups: previous behaviour, dwelling conditions, economic conditions, and demographic characteristics. In the second set of analyses, changes in AC-usage behaviour due to MCO implementation were analysed. This section used a new response variable that combined AC-usage behaviour during the MCO period and AC usage at the office before the MCO period. New AC-usage behaviour during the MCO period was recategorized by combining the categories "not used" and "rarely used" into a new category "not used". Another new category, "used", accumulated the old categories "sometimes used" and "always used". This recategorized AC-usage behaviour during the MCO period and AC usage at the office before the MCO period were combined, resulting in two possible behavioural changes. Those who used ACs at the office before the MCO period but did not use it during the MCO period were classified as "stopped AC use during MCO". The behaviour of not using ACs at the office before the MCO period but using it during the MCO period indicated the second type of change, that is, "started AC use during MCO". The remaining two combinations revealed no change in AC-usage behaviour. They were merged to form the "no change" category. Thus, the variable change in AC use, with three categories (no change, stopped AC use during MCO, and started AC use during MCO), was constructed to project behavioural changes in AC use. Among the set of covariates, monthly household income, monthly household electricity bill, and age group were categorised based on numerical values. The respondents with monthly household income below MYR 13 5000 (USD 1195.74) and between MYR 5000-10,000 (USD 1195.74-2391.49) were categorised into lowand medium-income groups, respectively. The category of high income consisted of participants whose monthly household income exceeded MYR 10,000 (USD 2391.49). This categorisation was based mainly on income classification by households in Malaysia. Some respondents preferred not to reveal their income levels. These respondents constituted the "not exposed" category of this covariate. In the case of monthly household electricity bills, the categories along with the numerical range of values are low (< MYR 101 or USD 24.14), medium (MYR 101 to MYR 400 or USD 24.12 -USD 96.17), high (MYR 401 and above or ≥USD 96.41), and unknown. The unknown group encompassed respondents unfamiliar with their monthly household electricity bills. People aged between 18 and 30 years comprised the young age group, while the middle-aged category included participants aged > 30 years and not more than 50 years. Respondents aged more than 50 years were assigned to the "older" group. To provide a summary of all the variables analysed in this study, the descriptive statistics of the respective categories of each variable, as well as the 95 % confidence intervals of their percentages, are presented in Table 1 . The statistical analyses for all three sections of this study were conducted using three major processes. The analyses started by extracting the descriptive statistics of the selected variables, followed by bivariate analyses. The first two sections (AC-usage behaviour during remote work and changes in AC use due to MCO) used bivariate analyses to identify the factors significantly associated with the outcomes of interest. The bivariate analyses relied on bar charts to illustrate unadjusted associations between the dependent variable and selected covariates and the chi-square test of independence to check the statistical significance of these associations. Since the unadjusted association between two variables (one covariate and the dependent variable in the present study) was investigated by bivariate analyses, further analysis was required to justify the associations found to be statistically significant in the bivariate analysis. Therefore, all the significant covariates of the bivariate analyses were incorporated into regression models to examine the fitted associations between the responses and the corresponding potential factors. The AC-usage behaviour during remote work was analysed using an ordinal logistic regression model (ORL) (42) (43) (44) (45) (46) . The multinomial regression model (MLT) investigated the changes in ACusage behaviour at work owing to MCO implementation (47) . The last section of the study explored the various reasons behind the transition in AC-usage behaviour from the pre-MCO period. An ORL relies on the cumulative logit link function, which preserves order in the outcome variable. The regression coefficients of the covariates can be easily converted into odds ratios (OR) (OR), as these coefficients are independent of category (48) . This provides a convenient method for understanding and interpreting the effects. Suppose that an ordered categorical variable involves (> 2) categories, and the response where is the threshold term for the ℎ category, and the OR of the covariate can be computed by taking the exponent of the corresponding coefficient estimates, exp( ) The basic concept of the MLT is that it is a generalised version of the binary logistic regression model (49) , which uses a logit link function to analyse a binary dependent variable. In MLT, the multicategory (more than two) nominal outcomes are analysed, conducting binary logistic regression models for each category of the response variable involving an arbitrary reference category (50) . Suppose that a response variable has (> 2) nominal (unordered) categories. For such a response variable, the MLT consists of ( − 1 ) non-overlapping logit models. Let the ℎ category of the response be the reference category and = ( 1 , 2 , ⋯ , , ⋯ , )′ be the where is a constant and = ( 1 , 2 , ⋯ , , ⋯ , ) ′ is the ( × 1) vector of the regression coefficients associated with the ℎ category of the response. The estimated regression coefficient of the model can be interpreted easily using relative risk ratio (RRR). The RRR for the covariate can be estimated using the following formula: where ̂ denotes the RRR estimate and ̂ is the estimated regression coefficient. R software was used to analyse data using these models. The parameters of the ORL were estimated using the "polr" function of the R (version 4.1.1) package "MASS" while the "multinom" function from the "nnet" package was used for the MLT. 19 The response variable for this section was AC-usage behaviour during the MCO period, consisting of four categories: not used, rarely used, sometimes used, and always used. Four types of covariates (previous behaviour, dwelling conditions, economic conditions, and demographic characteristics) were selected to detect the potential factors of AC-usage behaviour in remote work. Fig. 2 displays the results of the bivariate analysis of AC-usage behaviour during remote work. using ACs during the MCO period was evident from those who did not use ACs at the office before the MCO period. Previous AC users tended to use ACs while remote working during the MCO period more than previous non-users. The statistical significance of these differences was evident from the non-overlapping error bars. High-rise building dwellers were found to show a higher tendency not to use ACs and a lower tendency always to use ACs while working remotely, compared to the residents of other types of buildings, and these differences were statistically significant. The significant association between the working room and AC-usage behaviour during the MCO period is mostly captured by those who did not use ACs during remote work. This category contained a significantly higher percentage of respondents who worked remotely in the communal spaces of their houses during the MCO period than those who worked in private rooms. Respondents from the low-income group tended to not use ACs during remote work than highincome respondents and those who preferred not to reveal their monthly household income. The low-income group contributed to a significantly lower percentage of always using ACs during the MCO period compared to the respondents from the high-income and unexposed groups. The figure illustrates that the respondents with low electricity bills showed a higher reluctance to use ACs 21 and a lower tendency to always use ACs during the MCO period than those from all the other categories (medium, high, and unknown). A significantly higher percentage of always using AC during the MCO period is shown for respondents with high electricity bills, compared to the low and medium categories. Both the middle-aged and older respondents revealed significantly higher reluctance to AC use during the MCO period than younger participants, although no significant difference is evident for the "always used" behaviour among these three age groups. Males were found to have a significantly higher percentage of rarely using ACs during the MCO period than their female counterparts. A significantly higher tendency towards rare AC use was projected for Malay people than for other ethnicities during this period. The bivariate analysis was followed by a regression analysis using a suitable ORL-able to handle ordinal response variables with > 2categories (42)-to examine the fitted associations between the selected covariates and the response variable. Since the response variable, AC-usage behaviour during the MCO period, is a categorical variable with four levels-and because a sense of order is inherent among these levels-it was analysed using the ORL. Covariates were included in the regression model, which sequentially accounted for the extent of their importance. Previous behaviour-related covariates were identified as the most important because of the substantial association between habitual behaviours, thermal comfort actions, and energy consumption (51) . Hence, AC usage at the office before the MCO period was analysed first. Subsequently, the wellestablished relationship between indoor thermal comfort and building spaces and characteristics (52) included as the variance inflation factor for each covariate was observed to be below 5, indicating no multicollinearity among these covariates (53) . The results of these models are listed in Table 2 , showing the ORs and p values to indicate the statistical significance of the corresponding effect. × 100 % = 46 % higher odds of not using ACs during the MCO period than those from other types of houses. High-rise buildings are usually well equipped with sufficient ventilation, enabling fresh air to pass through rooms with higher outdoor velocity and making the rooms cooler than those of low-rise buildings (55) . Consequently, the inhabitants of high-rise buildings can more easily achieve the desired level of thermal comfort without using ACs during remote work. Furthermore, the security and privacy of apartments allow female workers to work without wearing hijabs, which may also have contributed to the lower AC use of high-rise building residents during the MCO period. Another influential factor was the working room at home during the MCO period because of its significant association with AC-usage behaviour in remote work. The OR for the private room users is 0.25 (< 1), which indicates less likelihood of not using ACs while working remotely in the private rooms of the houses compared to those who used communal spaces for remote work. The higher reluctance in AC use among the users of communal spaces is attributable to the openness and comfort of these places of a house, compared to bedrooms or study rooms. Consequently, a higher percentage of AC installations was exhibited in bedrooms than in communal spaces, such as living areas or dining rooms by Kubota et al. (39) . In general, rooms with larger floor areas and better natural ventilation can potentially increase indoor wind speed, resulting in the mitigation of the thermal discomfort of occupants. Thus, people can work comfortably without using ACs in the communal spaces of their houses. Moreover, AC use in these more spacious communal spaces requires a large cooling capacity which results in a higher electricity bill. Hence, communal space users were more reluctant to use ACs while working remotely during the MCO period. Both covariates related to economic conditions appeared to be significant factors for new behaviour. The ORs for the high-income group and those who did not expose their income were 0.51 and 0.57, respectively, with p values less than 0.001. Since both the ORs are significant and less than 1, the respondents from these two groups were less likely not to use ACs while working remotely than those with a medium-income level. No such significant association was observed between the low-and medium-income groups, which may be surprising. This is because the low- To analyse changes in AC usage at work that were due to implementation of the MCO, a new response variable was created by combining two variables: AC-usage behaviour during the MCO period and AC-usage at the office before the MCO period. In the first of these, the responses on AC-usage behaviour during remote work (from 5076 participants) were recategorized. The 27 categories "not used" and "rarely used" were combined to become "not used", and "sometimes used" and "always used" were combined into another new category, "used". When this recategorized AC-usage behaviour during the MCO period was combined with AC usage at the office before the MCO period (including the answers of 2772 respondents), only two of the four possible combinations represented behavioural changes: • used AC at the office before the MCO period, but did not use it during the MCO period, and • did not use AC at the office before the MCO period but used it during the MCO period. These two types of changes were renamed "stopped AC use during MCO" and "started AC use Bivariate analyses triggered significant associations between the outcome variable and covariates. The plausibility of these associations was required to be analysed, accounting for all the covariates. To this end, an MLT was used. The total number of observations included in this analysis was 1873 after discarding the missing values of the considered variables. The applicability of the MLT for analysing unordered outcome variables with more than two categories (47) motivated us to adopt this regression model in this section. This is because the response variable of interest is a categorical variable with three levels that cannot be ordered. In this model, respondents who 30 reported no change in their AC-usage behaviour in remote work during the MCO period were used as the reference category. This means that two specific models were obtained for those who stopped and started using ACs during the MCO period, and the results of both models were comparable to the no-change category. Table 3 shows the results of these models, presenting the RRRs and p values, to check the statistical significance of the effects. The cessation of AC use during the MCO period, using it at the office before the MCO implementation but not during remote work, encountered significant associations with all the covariates, except residential house type and gender. By contrast, the model for starting AC use during the MCO period (using ACs while working remotely but not at the office before the MCO period) confirmed the statistical significance of all covariates, except for residential house type and age group. In the case of working rooms at home during the MCO period, the RRRs for stopping and starting AC use during this period were 0.287 and 0.527, respectively, which were statistically significant. Since these RRRs are less than 1, the changes relative to not changing people install ACs in private rooms more often than in communal spaces, such as living areas or dining rooms (39) . Therefore, people working remotely in communal spaces, accustomed to using ACs at the office, could easily curb AC use during the MCO period. This finding was also supported by the results of the bivariate analysis, with 49.28 % of common room users ceasing to use ACs during the MCO period. This number was significantly higher than that for users of private rooms (22.34 %). However, according to the regression model, people who did not use ACs at the office before the MCO period were more likely to start AC use it during this period, relative to the no-change group, when working remotely in communal spaces. The bivariate analysis results showed that both the common room and private room users had similar percentages for the start of AC use during remote work, which were 5.61 % and 6.11 %, respectively. By contrast, the no-change category exhibited a substantial difference in the percentage values. (56) and discounts on electricity bills under economic stimulus packages and BPE (21) , the lowincome group received higher support from the government than did those with medium incomes. Consequently, the catastrophic situation during the MCO period did not considerably dissuade the low-income group from using ACs and working comfortably in remote work. At the same time, on the grounds of these subsidies, low-income people who were not used to AC consumption at the office before MCO could not rationally start using it during the MCO period. The sections presented earlier analysed different characteristics of people and depicted specific population groups who had a greater or lesser inclination to curb AC use during the MCO period. However, it is also necessary to envisage peoples' perspectives scrutinising the factors that aroused the MCO-day transition in AC-usage behaviour. Therefore, this section illustrates why Malaysian people increased or decreased AC use during remote work. The participants' thermal sensitivity 35 as well as their outlook and attitude towards governmental aid or discounts are also discussed, which can be decisive for policy makers and program managers. Table 4 displays the varying reasons for using and not using ACs while working remotely during the MCO period in Malaysia. in buildings has already been promoted, and energy conservation behaviour has been encouraged in previous studies (16, 33) . Another imperative research field for reducing the cooling load of buildings is to analyse the envelope design through passive strategies, which is also popular in Malaysia (64) (65) (66) . The building envelope was found to be the most important design parameter for achieving thermal comfort and energy efficiency in urban high-rise buildings in Malaysia (64). However, significant transformation is difficult to achieve in energy and electricity consumption behaviour without policy implications (67, 68) . A more challenging situation appears when a behavioural change pertains to a permanent usage pattern (3) . To this end, the current study analysed the potential factors behind these MCO-driven changes to AC-usage behaviour and changes in AC use, with a view to informing policy makers in designing and implementing their programs. In this way, the residents can be habituated to the rational use of AC. The findings of this work highlighted various aspects of the AC-usage behaviour of Malaysian necessitates heat adaptation to combat any upcoming temperature rise. One of the key heat resilience strategies is to achieve energy efficiency in daily life and in buildings, which will reduce the cooling demand and increase the reliability of the power grid, especially during heat waves. By contrast, energy efficiency plays a pivotal role in reducing greenhouse gas emissions with lower energy use. Therefore, policies must ensure an energy-efficient lifestyle and reasonable buildings 40 to provide a better indoor thermal environment. The findings of this study revealed some important characteristics of Malaysian people who practised limited AC use during this pandemic. Based on the analyses, some recommendations have also been made specifying the population groups that need to be focused on promoting curbed AC use. Some policy implications have also been discussed. These policy implications are paramount to attaining an energy conservation lifestyle by limiting AC use, essential for future extreme climate events. One of the limitations of this study was that it did not analyse home composition (orientation of the building), surface area of the house, number of rooms, presence or absence of any energyefficient elements, number of occupants sharing the space, tele-students sharing device or workspace, telework space flexibility in caregiving responsibilities for the youngest, difference in electricity bills due to MCO enforcement, socioeconomic status, and job profiles of the teleworkers. These factors may affect the AC-usage pattern while working remotely during the MCO period and should be the focus of future studies. Second, the analysis procedure, comprising only the urban people of Malaysia who worked remotely from home during the MCO period and could afford AC use at home, did not scrutinise issues such as hidden energy poverty. This is because most of the analysed communities had stable energy access in terms of both energy supply facilities and financial conditions. Thus, there is probably no need to consider this energy poverty concept in this study. However, further application of these analytical approaches in other geographic regions, not only in Malaysia but also throughout the world, should account for energy poverty, if necessary. Next, this study did not analyse meteorological conditions, such as daily This study projected whether mass enforcement of movement restrictions could change the thermal adaptation behaviour of workers utilising AC usage and determine the characteristics of workers who adopted such changes. This work has also discussed how these behavioural changes can be harnessed to develop a new pattern of AC use instigating energy conservation behaviour. However, both the government and policymakers need to re-evaluate their strategies and programmes to modify the energy sector to attain the goal of energy sustainability. The findings of this study will facilitate policy implications by identifying the characteristics of people to be emphasised and different means of capturing the controlled use of AC. Thus, this study contributes to ensuring a sustainable future for the energy sector in Malaysia. Government stimulus packages and electricity discounts removed the financial gap between low-and medium-income groups, and thereby erased 42 any difference in changes of their AC-usage behaviours. In light of this finding, policymakers can design strategies to attract people towards energy-saving lifestyles through special rewards or small incentives. In this way, the practice of energy conservation behaviour may evolve as a new habit to prevail even after the pandemic. Furthermore, the importance of the service enabling easy Japanese citizens' behavioral changes and preparedness against COVID-19: An online survey during the early phase of 43 the pandemic COVID-19) Events as They Happen, World Heal Organ, 2020 Energy transition in a lockdown: An analysis of the impact of COVID-19 on changes in electricity demand in Lagos Nigeria Four energy-saving lessons from the first lockdown which may help us through the winter. The Conversation How has behaviour changed under the COVID-19 lockdown? New data suggest COVID-19 is shifting the burden of energy costs to households The impact of different COVID-19 containment measures on electricity consumption in Europe Changes in domestic energy and water usage during the UK covid-19 lockdown using high-resolution temporal data Trend analyses of electricity load changes in Brazil due to COVID-19 shutdowns. Electr Power Syst Res Emergency measures to protect energy consumers during the Covid-19 pandemic: A global review and critical analysis A New Conception of Planning in the Era of Climate Change Climate change and residential electricity consumption in the Yangtze River Delta, China Impacts of rising air temperatures on electric transmission ampacity and peak electricity load in the United States Large uncertainties in trends of energy demand for heating and cooling under climate change Climate change impact on energy demand in building-urban-atmosphere simulations through the 21st century Building energy for sustainable development in Malaysia: A review. Renew Sustain Energy Rev Consumer Appliances in Malaysia. Euromonitor International Movement control as an effective measure against Covid-19 spread in Malaysia: an overview Regulations And SOPs Mondaq Connecting Knowledge & People. Available from Residential electricity usage soar 23pc during MCO, says Energy Ministry Available from Teleworking during the COVID-19 pandemic and beyond: A Practical Guide Telework in the EU before and after the COVID-19: where we were, where we head to Science for Policy Briefs Teleworkability and the COVID-19 crisis: a new digital divide? Laws of Attraction. 67% of Malaysian Companies Required Staff to Work from Home Enhancing the implementation of telecommuting (work from home) in Malaysia How to Provide A Comfortable Work Environment for Your Employees Working From Home Designing the work environment for worker health and productivity Working from home: Is our housing ready? An investigation of thermal comfort and adaptive behaviors in naturally ventilated residential buildings in tropical climates: A pilot study Thermal Comfort Assessment of an Office Building in Tropical Climate Condition Promoting Energy Conservation Behaviour : A Plausible Solution to Energy Sustainability Threats Sustainable Aspects of Electricity Consumption in Klang Valley The psychological determinants of energy saving behavior Analyzing energy consumption 48 factors during coronavirus (COVID-19) pandemic outbreak: a case study of residential society. Energy Sources, Part A Recover Util Environ Eff Impacts of COVID-19 on energy demand and consumption: Challenges, lessons and emerging opportunities COVID-19 and household energy implications: what are the main impacts on energy use? Heliyon Energy Consumption and Air-Conditioning Usage in Residential Buildings of Malaysia Department of Statistics Malaysia the Quality of Living Space -Is Larger House a Better Living Quality? Outlook Regression Models for Ordinal Data Regression models for ordinal responses: A review of 49 methods and applications A review of ordinal regression models applied on health-related quality of life assessments Ordinal logistic regression models: Application in quality of life studies Does Community Facility Play a Vital Role on Nutrition Status of Under-Five Children in Bangladesh ? Multinomial Logistic Regression: Usage and Application in Risk Analysis An Introduction to Categorical Analysis An application on multinomial logistic regression model Modeling multinomial logistic regression on characteristics of smokers after the smoke-free campaign in the area of Melaka Domestic energy consumption-What role do comfort, habit,and knowledge about the heating system play? Energy Build Factors Significant to Thermal Comfort Within Residential 50 Neighborhoods of Ibadan Metropolis and Preferences in Adult Residents' Use of Spaces Detecting Multicollinearity in Regression Analysis Habitual behaviors or patterns of practice? Explaining and changing repetitive climate-relevant actions Building Height Effects on Indoor Air Temperature and Velocity in High Rise Residential Buildings in Tropical Climate Kriteria Kelayakan Bantuan Prihatin Nasional A quarter of Malaysian workers fear job loss -survey [Internet]. The Edge Markets Regional differences in age related decrements of the cutaneous vascular and sweating responses to passive heating The Covid-19 recession: Rough times for young Malaysians International Labour Organization The Malaysian Fifth Fuel Policy: Re-strategising the Malaysian Renewable Energy Initiatives The potential and status of renewable energy development in Malaysia Development of renewable energy in Malaysia strategic initiatives for carbon reduction in the power generation sector The effect of building envelope on the thermal comfort and energy saving for high-rise buildings in hot-humid climate Energy-efficient envelope design for high-rise residential buildings in Malaysia Evaluation on Cooling Energy Load with varied Envelope Design for High-Rise Residential Buildings in Malaysia Provision of electricity to African households: The importance of democracy and institutional quality Policy making and energy infrastructure change: A Nigerian case study of energy governance in the electricity sector The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.