- The document analyzes how framing mild information about working from home impacts people's stated preferences in an online survey conducted in Poland during the COVID-19 pandemic.
- Providing information stating there were no productivity differences or that working from home lowered life satisfaction and career prospects reduced people's willingness to work from home by 5-10% compared to a control group.
- The results suggest stated preferences on surveys about working arrangements may be influenced by the information provided and should be interpreted cautiously, as prevalence of working from home policies may be shaped by dominant narratives.
1. Will work from home eventually work?
Will work from home eventually work?
Revisiting survey evidence with framing
Joanna Tyrowicz and Lucas van der Velde
Warsaw School of Economics
FAME|GRAPE
International Centre for Economic Analysis
November 2021
2. Will work from home eventually work?
Introduction
Recent pandemic and lock-down →↑ WFH possibilities (Bloom et al. 2021)
Increased demand for remote workers (Mueller-Langera and Gomez-Herrera 2021)
Mixed evidence on worker-side
40% of workers declare to quit jobs if not allowed to WFH (Barrero et al. 2021)
but ... long run effects are less clear (Bloom et al. 2014, Mas and Pallais 2017, Angelici and Profeta 2020)
We ask reliability of self-reported preferences
Mild framing moves stated preferences (a lot)
→ ↓ 5% willingness to WFH ↑ 2 pp. higher stated productivity
3. Will work from home eventually work?
Literature on WFH
WFH as a mixed blessing
↑ productivity, ↑↓ life satisf., ↓ promotions
(Bloom et al. 2014, Angelici and Profeta 2020)
Self-selection (Mas and Pallais 2017)
Covid changed the landscape a lot: prevalence ↑ and changed selectivity patterns
Not all jobs can be done remotely (Dingel and Neiman 2020)
Role of worker heterogeneity in response to WFH
(Schifano et al. 2021, Kurowska 2020) and (Kitagawa et al. 2021, Felstead and Reuschke 2020, Etheridge et al.
2020, Morikawa 2020)
Caution about on-line surveys during the pandemic
(Adams-Prassl et al. 2020, Baert et al. 2020, Belot et al. 2020)
Sample sizes between 5-20 k.
Non-probability sampling
Low stakes in reporting preferences
4. Will work from home eventually work?
Our contribution
Explore susceptibility of stated preferences to mild treatment
Framing in an online survey (Tversky and Kahneman 1985, Bayer and Ovodenko 2019)
Comparable to other surveys in Covid literature
Survey conducted in Poland 1-year into pandemic
Being exposed to info on WFH changes
Willingness to WFH at least once a week
Wage equivalent change
And even perceptions of past productivity
5. Will work from home eventually work?
The survey: D+
Conducted in PL since May 2020
Seven waves with ∼ 2 months difference
∼ 5000 participants in each wave
Participants recruited in social media + re-invitations
half participants in each wave are new
Resulting sample is not representative
More educated, more urban and more female
Reweight observations using CBPS to match Polish LFS data details
All results were computed using weights
6. Will work from home eventually work?
The experiment
Subjects were randomly provided with additional info on WFH
Before the pandemic, for cases where WFH was an option, research has causally
demonstrated that ...
1 ... there were no differences in productivity between workers working from the office and
those working from home
2 ... working from home on average lowers life satisfaction and increases feeling of loneliness
3 ... workers working from home had a much smaller chances for promotion than colleagues
working on site
plus a control group (no message)
Messages correspond to findings in Bloom et al. (2014) .
Outcome questions follow the literature (Barrero et al. 2021)
7. Will work from home eventually work?
Context: PL during COVID
Relatively mild restrictions
Short stay-at-home order period in April-May 2020
WFH recommended for non-essential activities
Education institutions closed for (most of) academic year 2020-2021
Two waves of contagions
1 September-November 2020
2 March-April 2021
8. Will work from home eventually work?
Context: PL during COVID
Similar changes observed in other activities See
9. Will work from home eventually work?
The outcome variables (I) : Willingness to WFH
Full-sample t:Control t:Satisfaction t:Productivity t:Career
How often would you like to work remotely? (shares)
Not at all 0.36 0.42 0.33 0.31 0.40
Once or twice a month 0.08 0.09 0.10 0.06 0.07
Once a week 0.10 0.10 0.12 0.07 0.10
Twice a week 0.12 0.09 0.12 0.12 0.16
Thrice a week 0.12 0.12 0.11 0.13 0.10
Four or more days a week 0.22 0.18 0.22 0.31 0.18
Willingness to WFH 0.56 0.49 0.57 0.63 0.54
Observations 3720 952 943 921 904
Willingness to WFH modeled after Angelici and Profeta (2020)
10. Will work from home eventually work?
The outcome variables (II): Compensational wage change
Full-sample t:Control t:Satisfaction t:Productivity t:Career
Attitudes towards WFH (shares)
Positive 0.52 0.46 0.55 0.58 0.47
Neutral 0.29 0.33 0.27 0.22 0.32
Negative 0.20 0.21 0.18 0.19 0.21
Changes in wages (in %)
Wage cut willing to accept 1.57 1.56 1.62 1.94 1.03
Wage increase required to accept 10.44 11.25 11.44 8.48 10.47
Compensational wage change 1.23 1.61 1.13 0.50 1.70
Observations 3714 952 941 917 904
11. Will work from home eventually work?
The outcome variables (III): Self-reported productivity change
Full-sample t:Control t:Satisfaction t:Productivity t:Career
Productivity change (self-reported)
Above 35% 0.16 0.15 0.15 0.23 0.11
15% to 35% 0.10 0.08 0.09 0.09 0.13
5% to 15% 0.15 0.12 0.17 0.10 0.23
-5% to 5% 0.45 0.49 0.45 0.48 0.37
-15% to -5% 0.10 0.12 0.11 0.07 0.12
-35% to -15% 0.02 0.03 0.03 0.02 0.02
Below -35% 0.01 0.01 0.01 0.01 0.02
Productivity change (cont.) 8.61 7.34 7.85 11.82 7.18
Observations 1703 447 435 404 417
12. Will work from home eventually work?
Results
Our regressions take the following form
outcomei = α + βTi + γXi + δJobi + ηWFHi + i
β is treatment effect
Xi stands for ind. characteristics
gender, age, education, marital status, presence of children under 5 in hh., and urban status
Jobi stands for job characteristics
industry, ownership of the firm, firm size, full-time worker
WFHi stands for (self-reported) experience while WFH
share of tasks that can (and cannot) be done remotely, frequency of WFH in past year, positive attitude
towards WFH
13. Will work from home eventually work?
Results
outcomei = α + βTi + γXi + δJobi + ηWFHi + i
Want to WFH Comp. wage change (%) Productivity (past)
(1) (2) (1) (2) (1) (2)
t:Productivity 0.067*** 0.030** -0.41*** 0.33 2.91** 2.91**
(0.000) (0.046) (0.004) (0.113) (0.021) (0.043)
t:Satisfaction 0.026*** -0.023*** 0.039 1.05*** -0.59 0.34
(0.001) (0.004) (0.303) (0.001) (0.310) (0.613)
t:Career 0.039*** 0.011 0.18*** 0.48* 0.053 0.45
(0.001) (0.196) (0.004) (0.088) (0.631) (0.378)
mean outcome 0.56 0.53 1.23 1.58 8.61 6.53
R-squared 0.53 0.67 0.15 0.19 0.080 0.16
# observations 3714 2292 3714 2292 1700 1390
Notes : Columns (1) control only for demographic characteristics, columns (2) add controls for job characteristics
and WFH experience. P-values in parentheses, SE cluster at treatment level
14. Will work from home eventually work?
Conclusions
Mild framing has a large effect on stated preferences
Between 5% to 10% of mean outcome
Effects mediated by experience of WFH
Survey results (and international comparisons) should be taken with a grain of salt.
Prevalence of WFH will be shaped by dominant narratives.
15. Will work from home eventually work?
References I
Adams-Prassl, A., Boneva, T., Golin, M. and Rauh, C.: 2020, Work that can be done from home: Evidence on variation within
and across occupations and industries, IZA Discussion Paper 13500, Institute of Labor Economics (IZA).
Angelici, M. and Profeta, P.: 2020, Smart-working: work flexibility without constraints, CESifo Working Paper 8165, CESifo.
Baert, S., Lippens, L., Moens, E., Sterkens, P. and Weytjens, J.: 2020, The Covid-19 crisis and telework: A research survey on
experiences, expectations and hopes, IZA Discussion Paper 13229, Institute of Labor Economics (IZA).
Barrero, J. M., Bloom, N. and Davis, S. J.: 2021, Why working from home will stick, NBER Working Paper 28731, National
Bureau of Economic Research.
Bayer, P. and Ovodenko, A.: 2019, Many voices in the room: a national survey experiment on how framing changes views
toward fracking in the United States, Energy Research Social Science 56, 101213.
Belot, M., van den Broek-Altenburg, E., Choi, S., Jamison, J. C., Papageorge, N. W. and Tripodi, E.: 2020, Six-country survey
on Covid-19, Covid Economics p. 206.
Bloom, N., Davis, S. J. and Zhestkova, Y.: 2021, COVID-19 shifted patent applications toward technologies that support
working from home, AEA Papers and Proceedings 111, 263–266.
Bloom, N., Liang, J., Roberts, J. and Ying, Z. J.: 2014, Does working from home work? evidence from a chinese experiment,
Quarterly Journal of Economics 130(1), 165–218.
Dingel, J. I. and Neiman, B.: 2020, How many jobs can be done at home?, Journal of Public Economics 189, 104235.
Etheridge, B., Tang, L. and Wang, Y.: 2020, Worker productivity during lockdown and working from home: Evidence from
self-reports, Covid economics vetted and real-time papers 52, 118–151.
Felstead, A. and Reuschke, D.: 2020, Homeworking in the UK: before and during the 2020 lockdown, Wiserd report, WISERD.
16. Will work from home eventually work?
References II
Imai, K. and Ratkovic, M.: 2013, Covariate balancing propensity score, Journal of the Royal Statistical Society: Series B
(Statistical Methodology) 76(1), 243–263.
Kitagawa, R., Kuroda, S., Okudaira, H. and Owan, H.: 2021, Working from home: its effects on productivity and mental health,
Covid economics vetted and real-time papers 74, 142–171.
Kurowska, A.: 2020, Gendered effects of home-based work on parents’ capability to balance work with non-work: Two countries
with different models of division of labour compared, Social Indicators Research 151(2), 405–425.
Mas, A. and Pallais, A.: 2017, Valuing alternative work arrangements, American Economic Review 107(12), 3722–3759.
Morikawa, M.: 2020, Productivity of working from home during the covid-19 pandemic: Evidence from an employee survey,
RIETI Discussion Paper 20-E-073, RIETI.
Mueller-Langera, F. and Gomez-Herrera, E.: 2021, Mobility restrictions and remote work: Empirical evidence on demand and
supply on a european online labour market, Covid economics vetted and real-time papers 63, 1–34.
Schifano, S., Clark, A. E., Greiff, S., Voegele, C. and D'Ambrosio, C.: 2021, Well-being and working from home during
Covid-19, Information Technology People forthcoming.
Smyk, M., Tyrowicz, J. and Van der Velde, L.: 2021, A cautionary note on the reliability of the online survey data: the case of
wage indicator, Sociological Methods Research 50(1), 429–464.
Tversky, A. and Kahneman, D.: 1985, The framing of decisions and the psychology of choice, Behavioral decision making,
Springer, pp. 25–41.
17. Will work from home eventually work?
Covariate Balancing Propensity Scores
Proposed by Imai and Ratkovic (2013)
Uses FOC of probit models to derive weights for the treated sample
Ensures a perfect balance of characteristics
Applied to online survey in Smyk et al. (2021)
Reference population: BAEL 2018 and 2019
Matching variables
gender, age(10 year bins), household size, urban status (3 levels), region, labor market status (3
levels) , industry of employment (5 levels)
back
18. Will work from home eventually work?
Context: PL during COVID
back