Data and Methods

Visualization Website for the Japan-UK Joint Project “The Future of Unpaid Work:AI’s Potential to Transform Domestic Work in Japan and UK”

About the data

Japan: The Japanese Survey on Time Use and Leisure Activities (STULA) 2016, the Statistics Bureau of Japan.

The UK: The UK Time Use Survey (UKTUS) 2014-15, Gershuny and Sullivan (2017)

Reference

Gershuny, J., Sullivan, O. (2017). United Kingdom Time Use Survey, 2014-2015. Centre for Time Use Research, IOE, University College London. [data collection]. UK Data Service. SN: 8128, http://doi.org/10.5255/UKDA-SN-8128-1

＊Access to the UKTUS 2014-2015 is obtained through the UK Data Archive https://doi.org/10.5255/UKDA-SN-8128-1. Permission to use the data on STULA 2016 was obtained through the Statistics Bureau of Japan based on the Statistics Act (Act No. 53 of 2007), Article 33 http://www.stat.go.jp/english/data/shakai/index.html.

Terminology and Methods

Unpaid Work Time Production

Average time spent on carrying out unpaid domestic work per week by country, sex, and age group.

Unpaid Work Time Consumption

Average unpaid domestic work time consumed per week by country, sex, and age group. It is estimated using the National Time Transfer Accounts (NTTA) method. The details of the method are referred to the NTTA methodology　(https://www.countingwomenswork.org/about/methodology).

Net Unpaid Work Consumption Time

Net time consumption equals time spent on consumption minus time spent on production. When the net unpaid work consumption time is positive, the per capita consumption of unpaid work exceeds the per capita production in a given country, sex and age group. Thus, this group spends more time on consuming unpaid work than on producing it and ends up with an unpaid work deficit. This deficit then needs to be filled by members of other age and sex groups within the country carrying out housework and care work that then gets “transferred” to fulfill this groups’ needs. The opposite is true when the net time consumption of unpaid work is negative. Those with a surplus of unpaid work production transfer unpaid work time to those who are in a deficit of unpaid work. To give an example, Japanese male individuals aged 35-39 have a positive unpaid work time consumption, reflecting the reality that they benefit from housework and care work done by others (often Japanese women of a similar age) and typically do little unpaid work themselves.

Experts’ Prediction of Automation in Time Use of Unpaid Work Activities in 5 years / 10 years

Expert prediction data is obtained from a Delphi survey inquiring about the probability of automation in five and ten years, as well as the anticipated market prices of the automation services and products related to 17 types of unpaid work activities*. The study involved 65 Japanese and UK experts engaged in AI-related research and business. The survey was conducted online from September 2020 to June 2021. For details on the survey methodology, please refer to Lehdonvirta et al. (2023).

*Note: The 17 types of unpaid work activities include 1) cooking, 2) washing dishes, 3) cleaning, 4) making clothes, 5) laundry, 6) ironing, 7) gardening, 8) pet care, 9) home and car maintenance, 10) grocery shopping, 11) other shopping, 12) use of public and commercial services, and household management, 13) physical childcare and supervision of infants, 14) teaching children, 15) interacting with children, 16) accompanying and transporting children, and 17) caring for family members (excluding children). The analysis and simulation combined grocery shopping and other shopping (with the automation probability averaged), presenting 16 types of unpaid work activities.

Reference

Lehdonvirta V, Shi LP, Hertog E, Nagase N, Ohta Y (2023) The future(s) of unpaid work: How susceptible do experts from different backgrounds think the domestic sphere is to automation? PLoS ONE 18(2): e0281282. https://doi.org/10.1371/journal.pone.0281282

Individual Preferences for Automating Unpaid Work

Data on individual preferences for automating unpaid work was obtained from surveys conducted in Japan and the UK. The percentage of respondents who answered ‘yes’ to the following questions regarding 17 types of unpaid work activities* is displayed by country, sex and age group:

‘If there were (automation) technology, would you consider using it in your household? Yes = 1, No = 2’

Respondents who answered yes to the above question were then asked the following questions; ‘Do you think you would use this technology in your household at ¥¥ yen/ ££ pound (the minimum price estimated by experts)? Yes = 1, No = 2’

The survey sample sizes were 24,727 people in Japan and 11,695 people in the UK.

* The 17 unpaid work activities here matched the activities list used in the Delphi survey

** The reference websites for the Japanese and UK household surveys will be uploaded shortly.

Automation Score of Unpaid Work Activities

The automation probabilities of unpaid work activities were originally obtained from our study. The probabilities are derived by multiplying the automation probabilities of unpaid work activities obtained from Japanese and English experts (by 17 types of unpaid work activities and predictions for 5 and 10 years ahead) with the percentage of individuals who intend to use those unpaid work automation services in households obtained from separate general population surveys in Japan and the UK (broken down by country, sex and age group). These figures represent the predicted automation probabilities of unpaid work activities in households.

The study also presents the automation probabilities when applying automation scores for market occupations corresponding to each type of unpaid work activity. To derive automation scores based on relevant market occupations we assigned occupations closely related to each unpaid work activity, such as assigning a chef for cooking, a maid or janitor for cleaning, and a launderer, dry cleaner and presser for laundry. Having done that, we utilized the probability of future automation for a relevant occupation for each of our 17 unpaid domestic work activities. On the website, the occupation automation probabilities can be selected from three sources: Frey and Osborne (F&O) (2017), Nomura Research Institute (NRI) (2015), and The UK Office of National Statistics (ONS) (2019). Please refer to the respective references for details on each set of the automation probabilities of occupation. Additionally, for details on the methodology linking occupation automation probabilities to unpaid work activities for the Japanese and the UK data, please refer to Hertog et al. (2023).

References

Frey C and Osborne M (2017) The Future of Employment: How Susceptible Are Jobs to Computerisation? Technological Forecasting and Social Change 114: 254–80. https://www.sciencedirect.com/science/article/abs/pii/S0040162516302244

Hertog, E., S. Fukuda, R. Matsukura, N. Nagase, and V. Lehdonvirta (2023) The future of unpaid work: Estimating the effects of automation on time spent on housework and care work in Japan and the UK, Technological Forecasting and Social Change 191: Article 122443. https://www.sciencedirect.com/science/article/pii/S0040162523001282

Nomura Research Institute (2015) Nihon no rōdō jinkō no 49-pāsento ga jinkō chinō ya robotto-tō de daitai kanō ni (49% of Japan’s working population can be replaced by artificial intelligence and robots). Retrieved from https://www.nri.com/-/media/Corporate/jp/Files/PDF/news/newsrelease/cc/2015/151202_1.pdf.

The UK Office of National Statistics (2019) The Probability of Automation in England: 2011 and 2017. Retrieved from https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/articles/theprobabilityofautomationinengland/2011and2017

Trends in Population Aggregated Net Unpaid Work Time Consumption: What Would Happen if No Additional Unpaid Work Automation Takes Place?

These estimates are obtained by multiplying the per capita net consumption of unpaid work by country, sex and age (5-year age group) with the corresponding population by country, sex and age (5-year age group). These estimates represent the total net consumption of weekly unpaid work hours at the population level, expressed in person-hours (people × hours).

Assuming that the per capita net consumption of unpaid work remains constant and as recorded around 2015 in the respective UK and Japanese time use surveys, this illustrates how the net consumption of unpaid labor hours (the population-aggregated values) would change in the future if only the population structure changes.

Similar to per capita net consumption of unpaid work hours, if this value is positive, it indicates that the consumption of unpaid work hours exceeds production (consumption > production), meaning there is a shortage of unpaid work hours. Conversely, if this value is negative, it indicates that the production of unpaid labor hours exceeds consumption (consumption < production), signifying a surplus of unpaid work hours.

Trends in Population Aggregated Net Unpaid Work Time Consumption: What Would Happen if Additional Unpaid Work Automation Takes Place?

These estimates are calculated by reducing the estimated net unpaid work time consumption using predicted unpaid work automation probabilities. These estimates illustrate the period changes in the net consumption of unpaid work hours by country, sex and age under the assumption of the potential advancement of unpaid work automation. For the predicted automation probabilities, we present results with our original predicted automation score (for 10 years ahead, see “Population Aggregated Net Unpaid Work Time Consumption”) as well as those by applying three sources of occupation automation probabilities from Frey and Osborne (F&O) (2017), Nomura Research Institute (NRI) (2015), and The UK Office of National Statistics (ONS) (2019). Please refer to “Automation Score of Unpaid Work Activities” (above) for further details about these automation probabilities.

Net Consumption of Unpaid Work Hours in Young Ages, Middle Ages and Old Ages

The values here are a sum of “Population Aggregated Net Unpaid Work Time Consumption” by age and country for the young age deficit, the middle age surplus and the old age deficit. The young and old age deficits are defined as a sum of the net time consumption deficit (the positive net time consumption) under age 30 and over 70, respectively. In contrast, the middle age surplus is a sum of the net time consumption surplus (the negative net time consumption) observed between the two deficits. The figures illustrate the extent of the shortage of unpaid work in the young and old ages and the surplus of unpaid work hours in the middle ages in Japan and the UK, depicting how these amounts may change in the future. Additionally, the graphs show how the automation of unpaid work changes the shortage and surplus of unpaid work hours at each life stage.

The Amount of Unpaid Work Hour Shortage

The values are obtained by summing “Population Aggregated Net Unpaid Work Time Consumption” over sex and age in each country. These values represent to what extent there will be a shortage of unpaid labor work hours at the population levels in both Japan and the UK in the future. Additionally, the graphs demonstrate to what extent the automation of unpaid work will alleviate the shortage of unpaid work hours in the future.

Potential Impacts of Automation of Unpaid Work on Employment Rate: An Application of the F&O Automation Score

These estimates refer to the predicted employment rates in the case where the occupational automation scores from Frey and Osborne (F&O) (2017) are applied to unpaid work, assuming that all the time freed from unpaid work transitions to paid labor. For detailed estimation methodology, please refer to Hertog et al. (2023).