Skip to main content

Open Access 22.05.2024

Covid-19 lockdown, gender and income dynamics in household energy consumption: evidence from Japan

verfasst von: Shigeru Matsumoto, Viet-Ngu Hoang, Clevo Wilson

Erschienen in: Empirical Economics

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config
loading …

Abstract

Residential electricity consumption and time spent at home by household members increased while household income decreased during the COVID-19 pandemic restrictions. Using survey data of Japanese households purchasing electricity from the Tokyo Electric Power Company Holdings, before and during the pandemic, we examine the various dynamics at play involving income, increased time spent at home by both partners and the role of genders in energy consumption. Results show a positive relationship between changes in electricity consumption and changes in household income, suggesting that households reduced their electricity usage following a decrease in income. Interestingly, the results also show that consumption changes are positively correlated to changes in hours spent at home by working husbands but negatively correlated to changes in the hours spent at home by working wives.
Hinweise

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

1 Introduction

Covid19 has had a profound economic, social and political impact on all spheres of life of urban residents, the implications of which have been the subject of many publications (see, for example, Tisdell 2020) including the impact on energy demand and consumption (Jiang et al. 2021). The International Energy Agency’s (IEA) (2020) reports that the shock caused by the Covid-19 pandemic to the global demand in 2020 has been the largest in the past 70 years. Global demand experienced a decline of 6% between 2019 and 2020. With respect to strategies used to restrict the spread of the virus, IEA (2020) also reported that energy consumption fell by 25% in countries with full lockdowns and by 18% in those with partial lockdowns. Such large and sudden variations are historically unparalleled even during major economic crises (Benatia 2020).
While the COVID-19 pandemic has reduced the total energy consumption as a whole, energy consumption shifted from industry and commercial sectors to the residential sector. This is due to people needing to spend more time at home and work from home. Sabau (2020) analysed US energy consumption in the first three months following a lockdown and reported that total energy consumption decreased by 4% from the corresponding period in the previous year, although energy consumption in the residential sector increased by 8%. Abdeen et al. (2021) reported that the average household daily electricity consumption in Ottawa, Canada increased by around 12%. They show that the highest peak loads during the pandemic were 15–20% more than peaks that occurred before Covid. Other studies also confirm that electricity consumption has increased in residential buildings and has put pressure on the grid system, not only in the USA (Agdas and Barooah 2020; Khalil and Fatmi 2022) but also in other countries across the world (Raman and Peng 2021).
Similarly, the Jyukankyo Research Institute Inc (2020) confirmed that energy consumption in Japan between April and June 2020 increased by 3.2% per household from the same period of the previous year. The Institute’s report found that under a stay-at-home order, people not only increased their home time but also changed their daily routines. For example, people slept a little longer than before and ate dinner earlier with their families. In addition, they had more time to use home appliances and other electricity-powered devices. Because of changes in lifestyles and energy-consuming behaviours, one can argue that electricity consumption would change. Although it is confirmed that such time reallocation influences electricity demand patterns at the national or regional level (Narajewski and Ziel 2020), it has not been properly investigated how household energy consumption has changed with time allocation flexibility.
As much as lockdowns are disruptive to households in numerous ways, they have also provided a rare opportunity to observe household behaviour relating to energy consumption and its dynamics to answer several important questions related to household energy consumption. For example, how does energy consumption vary when both genders of inhabitants spend mandated time at home? Equally, what are the impacts on energy consumption due to changes in income? Answering these types of questions requires rich data. Until now collecting such data was prohibitively expensive in some cases and having both genders staying at home at the same time for longer periods during a study was a relatively rare phenomenon. In this paper, we use a unique natural experiment set of data collected from the Tokyo metropolitan area over the period 2018 to 2020. This dataset reports both the difference in household electricity consumption before and during the stay-at-home request and the difference in income and home hours of both genders. Furthermore, we also examine the impact of changes in income on electricity consumption during the stay-at-home order while controlling for the effects of weather and its monthly variation over the period in which this analysis is conducted. Tests are conducted to validate the robustness of the findings.
Our empirical results confirm that changes in household income were positively correlated with changes in electricity consumption, suggesting that households would consume less (more) electricity when their income decreases (increases). Importantly, our results show that electricity consumption increased when the husband spent more time at home during a stay-at-home order but decreased when the wife stayed more at home. A comparison of the coefficients further implies that the marginal energy-saving effect of wives was stronger than for husbands. These results are robust in comparison to the two base years of 2018 and 2019 and when we control for the effects of the number of days during which the wife and husband stayed at home or the shares of the spouses’ income. There are a few possible explanations for this result. First, electricity consumption could decrease when the wife spent more time at home because the female could be exhibiting more energy-saving behaviours than males at home in Asian communities (see for example, Permana, et al. 2015). Second, a husband and wife could engage in different household tasks at different intensities. According to a survey of the hours spent by Japanese couples, full-time working women in their 30 s and 40 s spent 2 h and 53 min on housework on average while husbands spent only 59 min (Japan Broadcasting Corporation, 2020). It is quite possible that the difference in household work undertaken led to the result that the relaxation of the wife’s time constraints was more likely to have induced household energy savings than the relaxation of the husband’s time constraints.
Nevertheless, insights into differences in energy consumption behaviour with respect to gender have important policy implications which can be potentially used by decision-makers to bring about behavioural changes in energy use by households.
The following section undertakes a discussion of the relevant literature. Section 3 provides a background to the situation that prevailed in the Tokyo metropolitan area before and during the study period used to analyse the data. Section 4 discusses the survey data, first describing the survey procedure and then the changes in household electricity consumption from 2019 to 2020. This section also discusses the changes in stay-at-home hours by residents followed by data on changes in household income. Section 5 of the paper discusses the empirical model which is followed by the empirical results in Sect. 6. In this section, we discuss the results of the basic models followed by two robustness checks conducted to demonstrate the validity of the results. The results are also discussed in this section. Section 7 concludes with a brief discussion of the relevant policy implications.

2 Literature review

Several studies investigate the impact of Covid-19 on aggregate energy consumption. These studies show that while total energy consumption has decreased, residential electricity consumption has increased even after controlling for the effects of weather (see, for example, Buechler et al. 2020; Jiang et al. 2021). The main purpose of this literature review is to point out that the couple's time allocation can lead to differences in energy consumption and could also influence factors surrounding energy use, including the issue of greenhouse gas (GHG) emissions.
The ways in which men and women spend time at home is significantly different (Hamermesh 2019), and therefore it is expected that the genders consume energy in a different manner. For instance, it is well known that men and women feel comfortable at different room temperatures. Women generally have a lower metabolic rate than men and hence they prefer relatively warm rooms and use less air conditioning (Kingma and van Marken Lichtenbelt 2015). Furthermore, men consume more energy for space cooling than women in summer while women consume more energy for space heating than men in winter. Biological differences between men and women can lead to differences in energy consumption patterns.
Men and women engage in different activities at home, which leads to possible gender disparities in energy consumption. In addition, there is a discrepancy in household work hours between couples, not only in Japan (Japan Broadcasting Corporation 2020) but also in other OECD and developing countries (Álvarez and Miles-Touya 2019; Ruppanner 2010). Therefore, it is perhaps not surprising to see a difference in GHG emission intensity between couples. In a study conducted by Druckman et al. (2012), on average daily GHG emissions of men and women in Britain, women’s consumption was slightly higher than for a man. CO2 emissions are around 22 kgs for an average woman compared to around 20 kgs for an average man. It is argued here that this is not surprising given that on average, women do engage in more unpaid household work than men, who spend more time at paid work.1 If people use appliances to reduce housework time, they may decrease appliance use as their home time increases. Since women do more household work, they may reduce energy consumption for appliance use more than men when home time increases.
Men and women may have different abilities to perform energy-saving activities. Carlsson-Kanyama and Lindén (2007) conducted interviews with 30 households in Sweden to compare energy saving behaviors between men and women. They reported that an energy rate system that varies with time makes women wash clothes and dishes at night and at weekends. They also argued that when the price of electricity increased, women refrained from using clothes’ driers resulting in more time spent completing their laundry task. Their result suggests that women intend to save electricity through time reallocation more frequently than men. Belaid and Garcia (2016) analyzed microdata from French the PHEBUS survey to identify the main factors influencing residential energy-saving behaviours. After taking into account all other factors, they found that males are more likely to have a negative influence on energy-saving behaviours.
Men and women may not utilize energy-saving opportunities equally. Tjørring et al. (2018) carried out a study based on responses to requests to Dutch households to adjust the timing of their electricity consumption to reduce GHGs. They indicated that there was a significantly greater response from women than men when responding to text messages. Based on extensive interviews they concluded that an important reason for this is because of the gender difference in the responsibility for household chores undertaken. Specifically, they argued that it would be more effective to send text messages to women since they are often more responsible for the laundry, which is relatively flexible in time.
Even when men and women carry out the same household work, they may not use the same amount of energy. Grunewald and Diakonova (2020) attempted to compare the energy efficiency between men and women by collecting activity diaries and power usage data at the same time. Their results reveal that women in Britain completed household work using less energy than men, i.e., women were more energy efficient than men in terms of household work. In addition, they reported that men and women performed the same activities at different times and combined multiple activities differently.
A more recent study was conducted in Nepal measuring energy consumption at the household level and behavioral change during a shortage of fuel in 2015 due to an embargo (Acharya and Adhikari 2021). The blockage had an impact on household fuel choices and the changes in energy consumption were split between various socio-economic and demographic factors of the population. In the case of the urban population, it was found that a household whose head was a better educated female was on average more likely to change to cleaner sources of fuel, such as electricity.
The above-mentioned literature regarding the influence of gender on energy consumption indicates that females are in general are more efficient users of resources (all things considered), respond better to messaging in relation to reducing GHG emissions, and are more likely to show their willingness to change to cleaner sources of energy such as electricity. However, no study has investigated how men and women increase or decrease their energy consumption as they spend more time at home. We aim to answer this question in this study, by using information about changes in electricity consumption and home time before and during the Covid-19 pandemic.

3 Covid preventative measures in force during the sampling period

Table 1 summarizes the major prevention measures against Covid introduced by the national government and the prefectural governments in the Tokyo area. On 15 January 2020, the first infected person was confirmed in Japan. The governor of Tokyo requested residents to refrain from going out on March 26 and the national government declared the first state of emergency in major city regions on April 7. The government subsequently expanded its coverage nationwide on April 16. Although strict prevention measures—including the closure of elementary and junior high schools—were introduced under this first state of emergency, they were completely lifted on May 25. Moreover, COVID-19 prevention measures introduced in Japan were less strict compared to those introduced in other developed countries. For instance, a majority of developed countries imposed penalties for those breaking Covid restrictions while Japan has thus far not implemented similar measures (Wah 2021).
Table 1
Major prevention measures introduced in the Tokyo area
2020
  
1
15
Confirmation of the first infected person in Japan
3
26
The prefectural governors in Tokyo area requested citizens to refrain from going out
4
7
Declared the first state of emergency in major city regions
16
Expanded the target area of the state of emergency to all prefectures
5
14
Lifted the state of emergency in 39 of the 47 prefectures
21
Lifted the state of emergency in 4 prefectures
 
25
Lifted the state of emergency in all regions
6
19
Eased self-restraint of movement across prefectures
7
22
Launched “Go-to-travel campaign” in the area excluding Tokyo
30
Requested restaurants to provide alcoholic beverage to shorten their business hours (from August 3 to 31)
10
1
Included Tokyo area for “Go-to-travel” campaign
 
Launched “Go-to-eat” campaign
12
14
Stopped “Go-to-travel” campaign in all prefectures
26
Suspension of new entry of foreigners
2021
  
1
8
Declared the second state of emergency in the Tokyo region
13
Expanded the target area of the state of emergency to the major city regions
2
8
Lifted the state of emergency in the Tochigi prefecture
26
Lifted the state of emergency in prefectures excluding the Tokyo region
3
21
Lifted the state of emergency in all regions
The Covid situation greatly improved after the first state of emergency was lifted. The government subsequently eased movement restrictions on June 19 and launched the "Go to travel" campaign on July 22 in which the government subsidized a part of the travel expenses with the aim of rescuing the travel-related industry. However, Tokyo was initially excluded from the campaign area since the prevalence of the Coronavirus in Tokyo was much higher than in other areas and there was a concern that travellers from Tokyo could spread the virus to other areas. On October 1, people in Tokyo were finally able to use the "Go to Travel Campaign." At the same time, another campaign—“Go to eat”—was introduced with the aim of supporting restaurants. However, a third wave of the virus occurred several weeks later. On January 8, 2021, the national government declared the second state of emergency in the Tokyo area.
The sampling period of this survey covers the summer of 2020, which covers the period when the second corona wave came and then passed (see Fig. 1). The prevention measures introduced in the second wave were much less strict than those introduced in the first wave. On the other hand, the number of infected people remained much smaller than that in the third wave. During the sampling period, the government, while trying to protect people from the Covid-19 pandemic, also attempted to mitigate the economic damage simultaneously. Therefore, the government was criticized for stepping on the accelerator and the brake at the same time.

4 Data

4.1 Survey procedure

We asked Nippon Research Center (NRC) to conduct a series of online surveys in three successive years 2018, 2019, and 2020. We focused on married couples and collected their socioeconomic and housing data. In each year, we asked the households to access TEPCO’s website and download their electricity consumption data for August, September, and October and then upload the data to the designated web page. Unlike a typical survey that asks households to record the amount of electricity consumption and payment, we used the precise billing data for this survey. By doing so we were able to avoid recording error problems.
For the year 2018, we collected data from 341 households. Together with electricity consumption data, we also collected information on socioeconomic characteristics. We further asked households about their job status and homestay.
The purpose of this study is to evaluate the effect of time spent at home and resulting changes in household income on household electricity consumption. For this evaluation, in the 2020 survey (the Covid-19-year survey), we asked households how much home hours and household income changed from the year 2019. After removing households lacking necessary information, 78 households remained for the 2019–2020 comparison.

4.2 Change in household electricity consumption

Figure 2 shows the change in the monthly electricity consumption between 2019 and 2020. The straight line rising to the right is the 45-degree line, and thus the points above or below this line indicate the cases in which monthly electricity consumption increased or decreased, respectively. While households have increased their electricity consumption on average, it is important to note that the summer of 2020 was hotter than the summer of 2019; hence, the demand for cooling through the usage of air conditioning was higher in 2020 than in 2019.

4.3 Change in home hours

As noted above, to reduce the risk of corona infection, the government urged people to stay at home unless going out was essential. However, compliance was low with outside activities reduced only modestly.
In the 2020 survey, we asked households to record the change in home hours from the previous year. Table 2 presents the changes. 57.2% of husbands and 76.3% of wives answered that home hours remained about the same. On the other hand, 3.9% of husbands and 6.4% of wives answered that their home hours decreased while 38.9% of husbands and 15.2% of wives answered that home hours increased. Therefore, the table shows that, on average, both husbands and wives spent more time at home in the Covid year.
Table 2
Change in home hours
Question: Did the amount of time you (and/or your spouse) spent at home on weekdays in August (September, October) change from the same month last year?
  
Husband
Wife
Median
Answers
Case
Share
Case
Share
 − 55
Decreased by more than 50 h per week
3
0.7%
3
0.7%
 − 45
Decreased by 40–49 h per week
3
0.7%
2
0.5%
 − 35
Decreased by 30–39 h per week
0
0.0%
6
1.4%
 − 25
Decreased by 20–29 h per week
2
0.5%
1
0.2%
 − 15
Decreased by 10–19 h per week
2
0.5%
3
0.7%
 − 5
Decreased by 1–9 h per week
7
1.6%
13
3.0%
0
About the same
249
57.2%
332
76.3%
5
Increased by 1–9 h per week
46
10.6%
40
9.2%
15
Increased by 10–19 h per week
32
7.4%
15
3.4%
25
Increased by 20–29 h per week
23
5.3%
1
0.2%
35
Increased by 30–39 h per week
15
3.4%
3
0.7%
45
Increased by 40–49 h per week
17
3.9%
1
0.2%
55
Increased by more than 50 h per week
36
8.3%
6
1.4%
No record
0
9
2.1%
Total
435
435
100.0%
100%

4.4 Change in household income

Previous studies have estimated that the short-run income elasticity of residential electricity demand ranges from − 0.450 to 1.265 (Zhu et al. 2018) with a mean of 0.239. Due to COVID-19, some households suffered significant financial losses and thus such households were expected to reduce their electricity consumption. To evaluate Covid’s impact on household electricity consumption we asked households about their income changes in the 2020 survey.
If households consider their electricity bill as part of their necessary living expenses, then they will pay it from their monthly salary. If so, any change in bonuses will have less impact on electricity usage than the change in monthly salary. To examine whether different types of income affect energy consumption differently, we asked households about the change in three types of incomes: annual income, monthly income, and summer bonuses. Table 3 summarizes the respondents’ answers. A majority of respondents said their household income remained the same: 68.7% for annual income, 75.6% for monthly income, and 75.3% for summer bonuses. Although some households answered that their incomes had increased (4.7%), a much greater number (26%) of households said their incomes had decreased.
Table 3
Change in household income
Has there been any change in annual income (monthly income, summer bonuses) including tax and public assistance compared to 2019?
Median
 
Annual income
Monthly salary
Summer bonus
 − 55
Decreased by more than 50%
3
0.7%
3
0.7%
12
2.7%
 − 45
Decreased by 40–49%
0
0.0%
1
0.2%
6
1.3%
 − 35
Decreased by 30–39%
12
2.7%
12
2.7%
9
2.0%
 − 25
Decreased by 20–29%
15
3.3%
9
2.0%
12
2.7%
 − 15
Decreased by 10–19%
42
9.3%
12
2.7%
27
6.0%
 − 1
Decreased by 1–9%
45
10.0%
46
10.2%
12
2.7%
0
About the same
309
68.7%
340
75.6%
339
75.3%
5
Increased by 1–9%
12
2.7%
9
2.0%
6
1.3%
15
Increased by 10–19%
6
1.3%
12
2.7%
0
0.0%
25
Increased by 20–29%
0
0.0%
0
0.0%
0
0.0%
35
Increased by 30–39%
0
0.0%
0
0.0%
0
0.0%
45
Increased by 40–49%
3
0.7%
3
0.7%
3
0.7%
55
Increased by more than 50%
0
0.0%
0
0.0%
3
0.7%
 
I do not want to answer
3
0.7%
3
0.7%
21
4.7%
Total
 
450
100.0%
450
100.0%
450
100.0%

5 Empirical model

We denote \({E}_{i,m,y}\) to be household \(i\)’s electricity consumption in month \(m\) in year \(y\). We use the logarithm transformation to approximate the percentage change in electricity consumption from the base line year to the Covid year in the specific month, \( \Delta E_{{i,m}} = \ln \left( {E_{{i,m,y = {\text{COVID}}}} /E_{{i,m,y = {\text{Base}}}} } \right) \). We define \(\Delta {H}_{i,s,m}\) as the change in home hours of spouse \(s\) and \(\Delta {I}_{i,k}\) as the percentage change in \(k\)th type of income.
Even if the billing month of the electricity bill is the same, households may use electricity on different days. Undoubtedly, weather conditions also affect at-home activities, hence energy consumption. To control for the change in weather conditions, we include the change in monthly cooling days, \(\Delta CDD\) (see the detailed calculations in Appendix A). Because our sampling period is from August to October, we expect that the increase in CDDs produces an increase in electricity consumption.
Taking all together, our empirical model is given as follows:
$$ \Delta E_{{i,m}} = \alpha + \user2{\beta }_{s} \Delta \user2{H}_{{i,s,m}} + \beta _{k} \Delta I_{{i,k}} + \beta _{C} \Delta {\text{CDD}} + \user2{\beta }_{m} \user2{D}_{m} + \varepsilon _{{i,m}} $$
where \({{\varvec{D}}}_{m}\) are the survey month dummies and \({\varepsilon }_{i,m}\) is an error term which is assumed to be normally distributed. In this specification, households that did not change home hours and maintain income were treated as the control group and those that were forced to change home hours or lost income became the treatment group. It is noted that treatment levels vary across households.

6 Empirical results

6.1 Basic model

Table 4 presents the estimation results of the basic OLS models. Model 1 incorporates the change in annual household income while Model 2 incorporates the change in both monthly income and summer bonus.
Table 4
Basic model
 
Base year = 2019 (N = 234)
 
Model 1
Model 2
 
Coef
Std. Err
Coef
Std. Err
Change in home hours
Wife
− 0.0021*
0.0011
− 0.0024**
0.0012
Husband
0.0017**
0.0007
0.0016**
0.0007
Change in household income
Annual income
0.0026**
0.0011
  
Monthly income
  
0.0026**
0.0012
Summer bonus
  
− 0.0011
0.0010
Change in cooling degree days
Cool. degree days
0.0765***
0.0245
0.0767***
0.0259
Survey month dummies
September
0.2143***
0.0229
0.2130***
0.0227
October
0.0213
0.0298
0.0217
0.0301
Constant
− 0.0155
0.0171
− 0.0210
0.0173
Adjusted R2
0.3311
 
0.3342
 
*, **, *** indicate statistical significance at 10%, 5%, and 1% levels, respectively. Robust standard errors are used
With respect to the impact of changes in home hours, we find that electricity consumption increased in households whose husbands stayed at home longer than the previous year but decreased in the households whose wives stayed at home longer. The sign of the coefficient remained the same even when only the home hours change of one spouse was considered. The results suggest that under the Covid-19 pandemic, husbands would have acted in a manner to increase their electricity consumption at home while wives could have acted to reduce electricity consumption. The comparison of the coefficients further implies that the marginal energy saving effect by wives was stronger than that of the husbands.
The annual income variable became positive and statistically significant at the 5% level in Model 1. Therefore, it was confirmed that households that lost income during the Covid pandemic period reduced their electricity consumption. The estimated income elasticity of electricity demand is only 0.0026. Although monthly income became positive and statistically significant at the 5% level in Model 2, summer bonuses did not become statistically significant. That is, household electricity consumption responded to the change in monthly salary but not to the change in bonuses. This is consistent with the fact that many people consider electricity bills to be a necessary expense in their daily lives.
As expected, the coefficient for CDDs became positive and statistically significant in all models. The results suggest that people use more electricity for space cooling when temperatures are high. According to the estimation results, a one percent increase in the monthly CDD increases electricity consumption by 0.08 to 0.10 percent. Since the monthly CDD in August is about 100, it increases by 30% if the daily temperature increases by 1 degree. This temperature increase leads to an electricity consumption increase of 2.4 to 3%.

6.2 Robustness check 1: household heterogeneity

The impact of COVID-19 on electricity consumption can vary between households with different socioeconomic characteristics. If the analysis is conducted without taking into account of households' heterogeneity, the effects of time and income on household electricity consumption can be masked. To examine whether household heterogeneity affects the estimation results (Table 4), we conducted the Breusch–Pagan test. Specifically, we analyzed the relationship between the square error from the basic model and the socioeconomics characteristics of households listed in Table 5.
Table 5
Socioeconomic characteristics of household
Variable
Unit
Mean
Std. dev
Min
Max
Household included in the comparison between 2019 and 2020 (N = 231*)
Number of persons
persons
2.90
0.85
2
5
Household income
10,000 yen
752.60
379.12
100
1750
Wife’s age
years old
51.47
8.88
30
73
Husband’s age
years old
54.17
9.09
31
69
Wife’s education year
years
14.36
1.54
12
18
Husband's education year
years
15.62
1.81
12
21
The age of one wife is not reported
Although the number of persons in the household is a key determinant of power consumption, household income, spouses’ age, and spouses’ education year, only wife’s age were statistically significant variables at the 10% level. Based on this test result, we conclude that socioeconomic characteristics of households do not seriously affect the conclusions relating to the effect of time and income changes on household electricity consumption.

6.3 Robustness check 2: time allocation prior to Covid

The basic model does not take into account the time allocation prior to the COVID-19 pandemic. However, those who spent much time at home prior to the pandemic may not have drastically changed their activities at home after the COVID-19 pandemic. In particular, the impact of the increase in home time on electricity usage may differ between active and retired households. To take into account of heterogeneity of time allocation prior to the Covid pandemic, we estimate the following multi-level mixed effects model:
$$ \Delta E_{{i,m}} = \alpha + \user2{\beta }_{s} \Delta \user2{H}_{{i,s,m}} + \beta _{k} \Delta I_{{i,k}} + \beta _{C} \Delta {\text{CDD}} + \user2{\beta }_{m} \user2{D}_{m} + {}_{\varvec{\mathbf\gamma}\varvec{s}} {\text{JOB}}_{s} + \user2{Z}_{i} \user2{u}_{i} + \in _{{i,m}} $$
In the survey, we asked the subjects whether they and their spouses had a regular full-time job or not. \({{\varvec{J}}{\varvec{O}}{\varvec{B}}}_{s}\) represents the full-time job participation dummy, which takes a value of 1 if spouse \(s\) had a full-time job.
In the survey, we also asked how many days they and their spouses would normally spend at home on weekdays. The responses included four answers: 0 day, 1–2 days, 3–4 days, and all days. According to the responses, we classified households into 4 groups: Groups 1, 2, 3, and 4, respectively. \({{\varvec{Z}}}_{i}\) is the covariate matrix for random effect \({{\varvec{u}}}_{i}\), which can vary between these four groups. The last term \({\epsilon }_{i,m}\) is the error term and is assumed to be multivariate normal with mean 0 and variance matrix \({\sigma }_{\epsilon }^{2}{\varvec{R}}\).
Below, we estimated two mixed effect models; the first model is that controlled for wives’ number of days at home while the second model is that controlled for husbands’ number of days at home. The reason we do not include couples’ number of days at home simultaneously is that there is a very strong correlation with the couples’ home time.2 In Table 6, Models 1-W and 1-H incorporate the change in annual household income while Model 2-W and 2-H incorporate the change in both monthly income and summer bonus.
Full-time job dummies were positive in all models and wives’ dummies were statistically significant. This result means that electricity consumption increased more in households whose wives had a full-time job. During the COVID-19 pandemic, it is evident that women also brought their work home and thus used more electricity at home.
The number of days that wives and husbands stayed at home was included in the model to account for variations in time allocation between households before COVID-19. However, the results show that they are not strongly associated with the variation in electricity consumption.
Even after controlling for socioeconomic characteristics and time allocation prior to Covid-19, the coefficient of the change in home hours remained the same. Thus, the way in which husbands increased their home hours was such that it increased electricity consumption. Similarly, wives increased their home hours in a way which also saved electricity consumption.

6.4 Robustness check 3: share of spouses’ income

Differences in household income composition may affect changes in electricity consumption. For example, retired couples may change their electricity consumption differently from double-income couples. To examine whether the household income composition affects the household electricity consumption, we estimated the model in which the full-time job dummy in Table 6 was replaced with the wife’s and husband’s income shares. The results of this model are shown in Table 7. The signs for the variable remained the same. However, we observe a larger value for the coefficients.
Table 6
Time allocation prior to Covid: controlled for job type
 
Base year = 2019 (N = 234)
 
Model 1-W
Wife’s stay
Model 1-H
Husband’s stay
Model 2-W
Wife’s stay
Model 2-H
Husband’s stay
 
Coef
Std. err
Coef
Std. err
Coef
Std. err
Coef
Std. err
Home hours change
Wife
− 0.0018*
0.0010
− 0.0018*
0.0010
− 0.0020**
0.0010
− 0.0020**
0.0010
Husband
0.0015**
0.0006
0.0015
0.0006
0.0014**
0.0006
0.0014**
0.0006
Household income change
Annual income
0.0024*
0.0011
0.0024*
0.0011
    
Monthly income
    
0.0025**
0.0012
0.0025**
0.0012
Summer bonus
    
− 0.0008
0.0007
− 0.0008
0.0007
Change in cooling degree days
Cool. degree days
0.0832***
0.0294
0.0832***
0.0294
0.0831***
0.0294
0.0831***
0.0294
Survey month dummies
September
0.2133***
0.0244
0.2133**
0.0244
0.2122***
0.0243
0.2122**
0.0243
October
0.0182
0.0269
0.0182
0.0269
0.0189
0.0269
0.0189
0.0269
Full-time job dummies
Wife
0.0807***
0.0309
0.0807***
0.0309
0.0817***
0.0308
0.0817**
0.0308
Husband
0.0485
0.0315
0.0485
0.0315
0.0424
0.0314
0.0424
0.0314
Constant
− 0.-654**
0.0321
-0.0654**
0.0321
-0.0648**
0.0321
-0.0648**
0.0321
Random-effect parameter
Constant
3.39E-16
1.46E-12
0.0000
0.0000
9.63E-24
2.00E-22
6.08E-22
1.08E-17
Variance
0.0222***
0.0021
0.0222**
0.0021
0.0222**
0.0020
0.0222**
0.0020
Wald chi 2
129.7***
129.7***
130.6***
130.6***
*, **, *** indicate statistical significance at 10%, 5%, and 1% levels, respectively. Robust standard errors are used
Table 7
Time allocation prior to Covid: controlled for income share
 
Base year = 2019 (N = 234)
 
Model 1-W
Wife’s stay
Model 1-H
Husband’s stay
Model 2-W
Wife’s stay
Model 2-H
Husband’s stay
 
Coef
Std. err
Coef
Std. err
Coef
Std. err
Coef
Std. err
Home hours change
Wife
− 0.0020*
0.0010
− 0.0020*
0.0010
− 0.0023**
0.0010
− 0.0023**
0.0010
Husband
0.0017***
0.0006
0.0017***
0.0006
0.0016***
0.0006
0.0016
0.0006
Household income change
Annual income
0.0024**
0.0011
0.0024**
0.0011
    
Monthly income
    
0.0025**
0.0012
0.0025**
0.0012
Summer bonus
    
− 0.0010
0.0007
− 0.0010
0.0007
Change in cooling degree days
Cool. degree days
0.0843***
0.0296
0.0843***
0.0296
0.0844***
0.0295
0.0844***
0.0295
Survey month dummies
September
0.2130***
0.0244
0.2130***
0.0244
0.2118***
0.0243
0.2118**
0.0243
October
0.0180
0.0270
0.0180
0.0270
0.0185
0.0269
0.0185
0.0269
Income shares
Wife
0.1688***
0.0594
0.1704***
0.0592
0.1725***
0.0590
0.1725***
0.0590
Husband
0.0408
0.0381
0.0415
0.0381
0.0351
0.0377
0.0351
0.0377
Constant
− 0.0656*
0.0366
− 0.0667*
0.0364
− 0.0672*
0.0363
− 0.0672*
0.0363
Random-effect parameter
Constant
2.19.E-05
2.98.E-04
5.62E-12
1.40E-10
1.08E-17
4.83E-14
2.16E-16
9.03E-13
Variance
0.0223***
0.0021
0.0223***
0.0021
0.0222***
0.0020
0.0222***
0.0020
Wald chi 2
128.1***
128.3***
130.65***
130.65**
*, **, *** indicate statistical significance at 10%, 5%, and 1% levels, respectively. Robust standard errors are used

6.5 Robustness check 4: comparison between 2018 and 2020

We have investigated the effects of changes in annual income and time spent at home caused by COVID-1 on household electricity consumption by comparing electricity consumption in 2019 and 2020. Therefore, we assumed that households consumed electricity under normal conditions in the year 2019 and selected it as the base year. To examine whether the above-mentioned results are influenced by the choice of the base year, we report the estimation results in which the year 2018 is used as the base year. For this analysis, we use the data of 121 households.
Table 8 presents the estimation results in which the year 2018 is used as the base year. The structure of the estimation models is the same as that in Table 6. Some variables have lost their statistical explanatory power. No statistically significant results can be observed for the income variable. In addition, the husband's home hours become statistically significant only at the 10% level. The results of the other variables are similar to those presented in Table 6.
Table 8
Base year change
 
Base year = 2018 (N = 363)
 
Model 1-W
Wife’s stay
Model 1-H
Husband’s stay
Model 2-W
Wife’s stay
Model 2-H
Husband’s stay
 
Coef
Std. err
Coef
Std. err
Coef
Std. err
Coef
Std. err
Home hours change
  
Wife
− 0.0026**
0.0011
− 0.0025**
0.0011
− 0.0028**
0.0011
− 0.0028**
0.0011
Husband
0.0011*
0.0006
0.0011*
0.0006
0.0010*
0.0006
0.0010*
0.0006
Household income change
Annual income
0.0006
0.0012
0.0006
0.0012
    
Monthly income
    
0.0014
0.0013
0.0014
0.0013
Summer bonus
    
− 0.0015*
0.0009
− 0.0015*
0.0009
Change in cooling degree days
Cool. degree days
0.1016***
0.0141
0.1001***
0.0141
0.1029***
0.0141
0.1029***
0.0141
Survey month dummies
September
0.2072***
0.0292
0.2084***
0.0293
0.2055
0.0292
0.2055
0.0292
October
0.2156***
0.0277
0.2150***
0.0278
0.2161
0.0276
0.2161
0.0276
Full-time job dummies
Wife
0.0787**
0.0352
0.0776**
0.0329
0.0721**
0.0331
0.0721**
0.0331
Husband
0.0268
0.0303
0.0281
0.0303
0.0215
0.0303
0.0215
0.0303
Constant
− 0.0678**
0.0339
− 0.0698**
0.0320
− 0.0649**
0.0320
− 0.0649
0.0320
Random-effect parameter
Constant
0.0003
0.0007
0.0000
0.000
1.25E-20
2.53E-19
2.99E-24
6.97E-23
Variance
0.0440***
0.0033
0.0443***
0.003
0.0439
0.0033
0.0439
0.0033
Wald chi 2
181.55***
180.68***
185.75***
185.75***
*, **, *** indicate statistical significance at 10%, 5%, and 1% levels, respectively

7 Conclusion and policy implications

There are a few gaps in the literature regarding the impact of COVID-19 on increased residential electricity consumption: the impact of increased home time on energy consumption, the impact of changes in income on energy consumption, and the difference in electricity consumption between the two genders. This study used a unique data set collected from Tokyo metropolitan household residents two years before the occurrence of the Covid-19 pandemic and during the pandemic in 2020 to shed more light into the three research gaps identified above.
Our empirical results confirm that changes in household incomes have a positive impact on residential electricity consumption. This suggests that when household incomes were impacted during the Covid19 pandemic then this income effect would result in a reduction in household electricity consumption as well. Furthermore, due to the COVID-19 restrictions, household electricity consumption increased when either or both household partners spent more time at home.
Interestingly, the results also show that households where the working wife spent more time at home experienced lower energy consumption, which is opposite to households where the working husband spent more time at home. Note that our results are robust across different model specifications in which the effects of weather conditions and household characteristics have been controlled for. This result suggests that the marginal energy-saving effect was stronger for working wives than for working husbands. In several model specifications, the magnitude of the energy-saving effect of the wife was nearly as much as the effects caused by the changes in the household incomes. To our knowledge, this finding has not been reported in the published literature before. More importantly, this finding highlights the importance of the gender factor on energy efficiency, not only in Japan but also in other countries with similar household characteristics.
Particularly, this result indicates that females play a key role in reducing energy consumption. This is consistent with the general findings from other studies that show females, when all other factors are taken into account, consume less or about the same amount of electricity than men and generate lower CO2 emissions. This study also supports previous research that women are more willing than men to respond to changes involving behaviour in energy consumption and are more willing to switch to cleaner sources of fuel.
The results have several potential policy implications. Most importantly, female households are more able and likely to respond to changes that result in disruptions to day-to-day routines than men. It is, therefore, important to design more focused and gender-targeted policies which can lead to more responsive energy consumption. Given the fact that women still do more housework than men and face stricter time constraints, policies to ease women's time constraints could be effective for energy conservation. The findings also indicate that general non-gender targeted messaging to households is likely to be less effective. Therefore, incorporating knowledge about the influence of gender on the practices and dynamics of energy consumption can be taken as an important driver of policy uptake by households and a determinant of the cost-effectiveness of such policies.

Acknowledgements

An earlier version of this paper was presented at the annual conference of Society for Environmental Economics and Policy Studies (Japan) in 2021. We appreciate comments from Kentaka Aruga and other participants.

Declarations

Conflict of interest

The authors declare that having no known competing financial interests or personal relationships could have appeared to influence the work reported in this paper.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creativecommons.​org/​licenses/​by/​4.​0/​.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Anhänge

Appendix A: Definition of CDD

The period of electricity usage varies between households. For example, one household may use electricity from July 4th to August 3rd while another household may use electricity from July 25th to August 24th. Such a difference in the electricity usage period will affect the change in electricity consumption when sampling the two periods.
If the average temperature of day \(d\) is given by \({\overline{t} }_{d}\), then daily cooling degree days can be defined as follows:
$${CDD}_{d}=\left\{\begin{array}{ll}{\overline{t} }_{d}-24&\quad if \, {\overline{t} }_{d}>24\\ 0& else\end{array}\right.$$
Monthly cooling degree days are obtained by totaling \({CDD}_{d}\) over the survey month, \({CDD}_{m}={\sum }_{d}{CDD}_{d}\). Finally, the percentage change in monthly cooling degree days becomes \(\Delta CDD={\text{ln}}\left(\left({CDD}_{m,t+1}+1\right)/\left({CDD}_{m,t}+1\right)\right)\). Since \({CDD}_{m}\) can take a value of 0 in cold regions, we added 1 inside the bracket.
Fußnoten
1
It is argued by Druckman et al. (2012), that on balance, the average CO2 consumption is almost identical for both genders.
 
2
We conducted a chi-square test of independence and rejected the null hypothesis of independence at the 1% level.
 
Literatur
Zurück zum Zitat Abdeen A, Kharvari F, O’Brien W, Gunay B (2021) The impact of COVID-19 on households’ hourly electricity consumption in Canada. Energy Build 250:111280CrossRef Abdeen A, Kharvari F, O’Brien W, Gunay B (2021) The impact of COVID-19 on households’ hourly electricity consumption in Canada. Energy Build 250:111280CrossRef
Zurück zum Zitat Acharya B, Adhikari S (2021) Household energy consumption and adaptation behavior during crisis: evidence from Indian economic blockade on Nepal. Energy Policy 148:111998CrossRef Acharya B, Adhikari S (2021) Household energy consumption and adaptation behavior during crisis: evidence from Indian economic blockade on Nepal. Energy Policy 148:111998CrossRef
Zurück zum Zitat Agdas D, Barooah P (2020) Impact of the Covid-19 pandemic on the U.S. electricity demand and supply: an early view from data. IEEE Access 8:151523–151534CrossRef Agdas D, Barooah P (2020) Impact of the Covid-19 pandemic on the U.S. electricity demand and supply: an early view from data. IEEE Access 8:151523–151534CrossRef
Zurück zum Zitat Álvarez B, Miles-Touya D (2019) Gender imbalance in housework allocation: A question of time? Rev Econ Househ 17(4):1257–1287CrossRef Álvarez B, Miles-Touya D (2019) Gender imbalance in housework allocation: A question of time? Rev Econ Househ 17(4):1257–1287CrossRef
Zurück zum Zitat Belaid F, Garcia T (2016) Understanding the spectrum of residential energy-saving behaviours: French evidence using disaggregated data. Energy Econ 57:204–214CrossRef Belaid F, Garcia T (2016) Understanding the spectrum of residential energy-saving behaviours: French evidence using disaggregated data. Energy Econ 57:204–214CrossRef
Zurück zum Zitat Benatia D (2020) Electricity markets under lockdown: Insights from New York. IAEE Energy Forum Special Covid-19 Edition 2020, 55–58. Benatia D (2020) Electricity markets under lockdown: Insights from New York. IAEE Energy Forum Special Covid-19 Edition 2020, 55–58.
Zurück zum Zitat Buechler E, Powell S, Sun T, Zanocco C, Astier N, Bolorinos J, Flora J, Boudet H and Rajagopal R (2020) Power and the Pandemic: exploring Global Changes in Electricity Demand During Covid-19. arXiv preprint arXiv:2008.06988. Buechler E, Powell S, Sun T, Zanocco C, Astier N, Bolorinos J, Flora J, Boudet H and Rajagopal R (2020) Power and the Pandemic: exploring Global Changes in Electricity Demand During Covid-19. arXiv preprint arXiv:​2008.​06988.
Zurück zum Zitat Carlsson-Kanyama A, Lindén AL (2007) Energy efficiency in residences—challenges for women and men in the North. Energy Policy 35:2163–2172CrossRef Carlsson-Kanyama A, Lindén AL (2007) Energy efficiency in residences—challenges for women and men in the North. Energy Policy 35:2163–2172CrossRef
Zurück zum Zitat Druckman A, Buck I, Haywarda B, Jackson T (2012) Time, gender and carbon: a study of the carbon implications of British adults’ use of time. Ecol Econ 84:153–163CrossRef Druckman A, Buck I, Haywarda B, Jackson T (2012) Time, gender and carbon: a study of the carbon implications of British adults’ use of time. Ecol Econ 84:153–163CrossRef
Zurück zum Zitat Grunewald P, Diakonova M (2020) Societal differences, activities, and performance: examining the role of gender in electricity demand in the United Kingdom. Energy Res Soc Sci 69:101719CrossRef Grunewald P, Diakonova M (2020) Societal differences, activities, and performance: examining the role of gender in electricity demand in the United Kingdom. Energy Res Soc Sci 69:101719CrossRef
Zurück zum Zitat Hamermesh DS (2019) Spending time: the most valuable resource. Oxford University Press, USA Hamermesh DS (2019) Spending time: the most valuable resource. Oxford University Press, USA
Zurück zum Zitat International Energy Agency. (2020) International Energy Agency. 2020. Global Energy Review. International Energy Agency. (2020) International Energy Agency. 2020. Global Energy Review.
Zurück zum Zitat Japan Broadcasting Corporation (NHK). (2020) NHK Data Book 2015 National Time Use Survey. Japan Broadcasting Corporation (NHK). (2020) NHK Data Book 2015 National Time Use Survey.
Zurück zum Zitat Jiang P, Fan YV, Klemes JJ (2021) Impacts of Covid-19 on Energy Demand and Consumption: Challenges. Lessons and Emerging Opportunities Applied Energy 285:116441 Jiang P, Fan YV, Klemes JJ (2021) Impacts of Covid-19 on Energy Demand and Consumption: Challenges. Lessons and Emerging Opportunities Applied Energy 285:116441
Zurück zum Zitat Jyukankyo Research Institute Inc. (2020) The analysis of the impact of the spread of the new coronavirus (COVID-19) on household energy consumption and its prevention (second report) in Japanese. August 17, 2020. http://www.jyuri.co.jp/3375/. Accessed May 24, 2021. Jyukankyo Research Institute Inc. (2020) The analysis of the impact of the spread of the new coronavirus (COVID-19) on household energy consumption and its prevention (second report) in Japanese. August 17, 2020. http://​www.​jyuri.​co.​jp/​3375/​. Accessed May 24, 2021.
Zurück zum Zitat Khalil MA, Fatmi MR (2022) How residential energy consumption has changed due to COVID-19 pandemic? An agent-based model. Sustain Cities Soc 81:103832CrossRef Khalil MA, Fatmi MR (2022) How residential energy consumption has changed due to COVID-19 pandemic? An agent-based model. Sustain Cities Soc 81:103832CrossRef
Zurück zum Zitat Kingma B, van Marken Lichtenbelt W (2015) Energy consumption in buildings and female thermal demand. Nat Clim Chang 5:1054–1056CrossRef Kingma B, van Marken Lichtenbelt W (2015) Energy consumption in buildings and female thermal demand. Nat Clim Chang 5:1054–1056CrossRef
Zurück zum Zitat Narajewski, M. and Ziel, F. (2020) Changes in electricity demand pattern in Europe due to COVID-19 shutdowns. IAEE Energy Forum Special Covid-19 Edition 2020, 44–47. Narajewski, M. and Ziel, F. (2020) Changes in electricity demand pattern in Europe due to COVID-19 shutdowns. IAEE Energy Forum Special Covid-19 Edition 2020, 44–47.
Zurück zum Zitat Raman G, Peng JCH (2021) Electricity consumption of Singaporean households reveals proactive community response to COVID-19 progression. Proc Natl Acad Sci 118(34):e2026596118CrossRef Raman G, Peng JCH (2021) Electricity consumption of Singaporean households reveals proactive community response to COVID-19 progression. Proc Natl Acad Sci 118(34):e2026596118CrossRef
Zurück zum Zitat Ruppanner LE (2010) Cross-national reports of housework: an investigation of the gender empowerment measure. Soc Sci Res 39(6):963–975CrossRef Ruppanner LE (2010) Cross-national reports of housework: an investigation of the gender empowerment measure. Soc Sci Res 39(6):963–975CrossRef
Zurück zum Zitat Tisdell CA (2020) Economic, social and political issues raised by the COVID-19 pandemic. Econ Anal Policy 68:17–28CrossRef Tisdell CA (2020) Economic, social and political issues raised by the COVID-19 pandemic. Econ Anal Policy 68:17–28CrossRef
Zurück zum Zitat Tjørring L, Jensen CL, Hansen LG, Andersen LM (2018) Increasing the flexibility of electricity consumption in private households: Does gender matter? Energy Policy 118:9–18CrossRef Tjørring L, Jensen CL, Hansen LG, Andersen LM (2018) Increasing the flexibility of electricity consumption in private households: Does gender matter? Energy Policy 118:9–18CrossRef
Zurück zum Zitat Zhu Z, Li L, Zhou K, Zhang Z, Yang S (2018) A meta-analysis on the price elasticity and income elasticity of residential electricity demand. J Cleaner Prod 201:169–177 Zhu Z, Li L, Zhou K, Zhang Z, Yang S (2018) A meta-analysis on the price elasticity and income elasticity of residential electricity demand. J Cleaner Prod 201:169–177
Metadaten
Titel
Covid-19 lockdown, gender and income dynamics in household energy consumption: evidence from Japan
verfasst von
Shigeru Matsumoto
Viet-Ngu Hoang
Clevo Wilson
Publikationsdatum
22.05.2024
Verlag
Springer Berlin Heidelberg
Erschienen in
Empirical Economics
Print ISSN: 0377-7332
Elektronische ISSN: 1435-8921
DOI
https://doi.org/10.1007/s00181-024-02593-0

Premium Partner