Exploring how worthwhile the things that you do in life are during COVID-19 and links to well-being and working from home

Abstract

The COVID-19 pandemic has had a significant impact on the way we work and live, with working from home becoming more than the occasional desire but a regular feature of work and life. While an increasing number of research studies have promoted the virtues of what is often described as the positive unintended consequences of the pandemic, there are also downsides, especially during periods of imposed restrictions on the ability to get out and about, that have broadly been described as impacting mental health and life’s worth. In this paper we use data collected in New South Wales during September 2020 and June 2021, seven and 16 months after the pandemic began, to obtain an understanding of the extent to which the pandemic has impacted on how worthwhile things done in life are for workers. We investigate whether there is a systematic behavioural link with working from home, reduced commuting linked to distance to work, and various socio-economic characteristics. The evidence suggests that the opportunity to have reduced commuting activity linked to working from home and increased perceived work-related productivity have contributed in a positive way to improving the worth status of life, offsetting some of the negative consequences of the pandemic.

1. Introduction

The COVID-19 pandemic, as forced onto society, has resulted in a large number of changes in the way we live and work. From a negative perspective, Dam et al. (2020) list a range of negative outcomes of the SARS-CoV-2 and COVID-19 pandemic such as duration of the lockdown, loss of freedom, boredom and frustration, social separation, inadequate lifestyle supplies, financial crisis, increased family issues and domestic violence, anxiety related to local news as well as rising case numbers, stigma, fear of contracting the disease, uncertainty about life, fear of losing closed ones and fear of death (thanatophobia). Despite the horrors of the virus, a number of positive outcomes have resulted which have delivered support from both employers and employees for more flexible working arrangements, including the greater incidence of working for home. Recently in Australia, we have seen that moderated behaviour has reverberated into periods of no lockdown, where people are self-regulating activity where the risk of transmitting and catching the virus has been deemed too risky.

Despite the overarching negatives of COVID-19, there have been some unintended positive consequences. There is a large literature that shows the benefits of reduced commuting which is one of the outcomes of the new paradigm of extensive working from home. From a health perspective, there have been a number of studies that have explored the impact of commuting. For example, Haefner et al. (2001) found that German commuters with a one-way trip time of 45 min or more had significantly higher levels of pain, dizziness, exhaustion, and severe sleep deprivation than the sample group of non-commuters with an equivalent one-way trip time. Similarly, Stutzer and Frey (2007) found that people with long journeys to and from work are systematically worse off and report significantly lower life satisfaction, which is not explained by a compensation or benefit at the level of the household. Further examination of the German Socio-Economic Panel data from 2007 to 2013 provided contrasting results, revealing that longer commutes were only related to lower satisfaction with particular life domains, especially family life and leisure time, and that time spent on housework, childcare as well as physical and leisure activities mediated the association between commuting and well-being (Lorenz, 2018). Ingenfeld et al. (2019) found a non-linear relationship between the negative effects of commuting and the trip distance, suggesting that the negative effects of commuting are almost completely due to individuals who commute more than 80 km daily-one-way.

In the USA, Hoehner et al. (2012) found those with commuting trips exceeding a distance of 20 miles had on average a significantly higher level of cholesterol, heart disease, stroke, blood sugar, and kidney disease (limited though due to the cross-sectional nature of the study and an inability to disentangle commuting time from correlations with sedentary behaviour). Kahneman and Krueger (2006) find that morning commuting is particularly unpleasant, with the degree determined by whether the person commutes alone or with another person and suggest that interventions that reduce the amount of time spent commuting alone, such as congestion taxes and carpool subsidies, could possibly have a beneficial effect on individuals’ emotional states. One UK study found that every extra minute of commute time reduces job satisfaction, reduces leisure time satisfaction, increases strain, and reduces mental health, whereas working from home, walking to work and shorter commute times increase job satisfaction; and that shorter commute times make it more likely that an employee will stay with their job (Chatterjee et al., 2017). In an Australian context, it has been found that commuting distance has a direct impact on increased absenteeism after controlling for indirect effects (Ma and Ye, 2019, Hensher et al., 2021a, Hensher et al., 2021c, Hensher et al., 2021b).

The savings in the time spent commuting have arisen because of the shelter at home, or work from home orders that have forced many organisations to adopt remote working at a much higher rate than many thought possible. Also a positive, Australian studies have found that perceived productivity1 whilst working from home has remained largely unchanged compared to the regular work environment; although there are some occupation classes with increased productivity (Beck and Hensher, 2022, Beck and Hensher, 2020a, Beck and Hensher, 2020b, Beck and Hensher, 2020c). In the USA, studies have similarly shown that firms have not experienced productivity loses due to working from home (Bartik et al., 2020). Barrero et al. (2021) suggest a strong correlation between perceived productivity and actual gains in labour productivity: “…we estimate that the COVID-induced shift to WFH will boost true productivity in the post-pandemic economy by 3.6 (0.2) percent on an equal-weighted basis and by 4.6 (0.2) percent on an earnings-weighted basis. These gains are quite large, providing more support for the view that the COVID-induced shift to WFH will yield important benefits in the form of more output per unit of time devoted to working for pay, inclusive of commuting time”. We also know from a number of other research studies that there is a noticeable gain in productivity, reaffirmed by our research. The Productivity Commission in Australia says that “workers may be more productive at home because they have better control over their time and enjoy better work–life balance. Firms will be able to tap into a larger pool of (more productive) labour. And while not strictly a productivity impact, workers have been shown to work longer hours when working from home during the pandemic.2

Tahlyan et al. (2022) using a U.S. representative sample of 318 working adults link telework satisfaction with experienced and perceived benefits and barriers related to telework and provide a causal structure to investigate telework satisfaction. They present an ordered probit model without latent variables to understand the systematic heterogeneity in telework satisfaction (on a 5-poit scale) across various socio-demographic groups. The results suggest younger and older aged. Individuals experienced/perceived lower benefits and higher barriers to teleworking. While age was also found to be important, the survey appears to have occurred during lockdown when children were at home and not able to go to school, which was not a constraint in our study. We recognise however that having children of school age at home is a very stressful experience which might outweigh the potential benefits of reduced commuting.

Given that remote working (flexible working, working from home, working near home, or whatever term you wish to use) will be a significantly greater proportion of the work mix moving forward, it is not surprising that research has explored the interplay between work and well-being during the COVID-19 pandemic. From a prevention perspective, Alipour et al. (2021) found that working from home reduced SARS-CoV-2 infections and helped to protect firms from COVID-19 distress. With respect to indicators of negative working from home experiences, Xiao et al. (2021) find that decreased overall physical and mental well-being after working from home were associated with physical exercise, food intake, communication with co-workers, children at home, distractions while working, adjusted work hours, workstation set-up and satisfaction with workspace indoor environmental factors. Negatively perceived factors of working from home include lack of distinction between work and home life, poor eating habits, loss of self-discipline, absence of an IT department, longer working hours and frequent video calls (Statista, 2020). In four waves of data collection in France, Italy, Germany, Spain, and Sweden, covering the period May–November 2020, Schifano et al. (2021) find that well-being is lowest among those who are not working at all, but is also lower for those who are working from home, in particular among those who are older, better-educated, and/or those with young children and those with more crowded housing. Möhring et al. (2021) find an overall decline in work satisfaction which is most pronounced for mothers and those without children who have to switch to short-time work. In contrast, fathers' well-being is less affected negatively and their family satisfaction even increased after changing to short-time work.

On the other hand, Okuyan and Begen (2022) find that working from home, while having challenges, also has favourable aspects if organisations can develop and maintain a good level of productivity and encourage staff to achieve the right level of work and life balance. In terms of promoting productivity in the working from home environment, Hashim et al. (2020) highlights the importance of adequate ICT equipment and support. Interestingly, Butler and Jaffe (2021) found that simply reflecting on the benefits of being with family, more flexibility and on support of colleagues could provide greater levels of work satisfaction through the day. In a longitudinal study during the pandemic, Russo et al. (2021) find that social contacts predicted positively and could moderate stress which predicted an individual’s well-being negatively. They also found their sample of software engineers adapted to working from home over time and were able to do so productively irrespective of predictor variables they attempted to use, highlighting the widespread success of the measure (although boredom and distraction did negatively impact productivity). For those employees that experienced lower work productivity, increased stress and poorer mental health, Toniolo-Barrios and Pitt (2021) recommended mindfulness techniques to help reduce negative outcomes and provide recommendations as to how to implement them. In spatial modelling of place use and well-being in Sweden, Samuelsson et al. (2021) found that easy access to natural settings supported well-being, irrespective of the population density of visited places, and highlighted how equitable access to natural settings can increase urban resilience towards pandemics. De Vos (2020) posits that walking and cycling can be important ways to maintain satisfactory levels of health and well-being.

One overarching concern that may exist, is how worthwhile individuals think the things that they do in life are, and how that might have changed in the face of the pandemic and resulting shock to work and activity. This is often raised in the broader literature, including the grey literature, with limited evidence on whether it is a serious issue adding to the sense of social alienation and well-being that was not there in the pre-COVID-19 period. If we can identify a link between this and what are identified as positive lifestyle outcomes of the pandemic such as reduced commuting, especially for longer distance commutes, increased flexibility in working arrangements and generally higher levels of productivity in working from home, then we should be able to establish whether individuals think that the things they do in life at present have changed for the better or worse in terms of worthwhileness, which is a useful proxy for some features of well-being.

In this paper, we draw on data obtained from the third wave undertaken in September 2020 and the fourth wave in June 2021 as part of an ongoing study of working from home and its implications for travel for residents in New South Wales (NSW) (see Fig. 1 ).3 This is two cross-sectional surveys. During this survey period, NSW had experienced an extended run of low COVID-19 case numbers, almost exclusively limited to the hotel quarantine system, and thus life had largely returned to pre-COVID-19 conditions, albeit with workers having a far greater opportunity to continue to work from home to some extent, at levels much higher than before the pandemic.

Fig. 1.

Overview of data collection time points.

The paper is organised as follows. We begin with a descriptive overview of the data used to identify a number of soft variables pertaining to happiness, anxiety and life’s worth which were collected during the 3rd and 4th waves of an ongoing survey. This is followed by a brief overview of the properties of the ordered logit choice model as an appropriate analytical setting when the dependent variable is an ordered rating scale. We then present the evidence from the estimation of the ordered logit model, focussing on the direct elasticity estimates associated with each explanatory variable at each level on the 11-point rating scale for the dependent variable on life’s worth. These elasticity estimates provide the behaviourally rich evidence on the role of statistically significant influences on life’s worth that are associated with increased propensity to work from home, work productivity improvements, reduced commuting activity as well as feelings of anxiety at the time of the surveys and some socioeconomic influences. We conclude with comments on the positive offsetting effects that have contributed to mitigating the negative impacts of influences that reduce the worth of life.

2. The descriptive setting

As part of an ongoing longitudinal survey designed to investigate changes in travel and working from home since the beginning the COVID-19 pandemic in March 2020, as the survey developed over time we began asking a short series of well-being questions identical to those used in the UK Office of National Statistics Annual Population Survey (ONS, 2020), as part of their quarterly estimates of life satisfaction (in part so that subsequent comparisons in ongoing research between Australia and the UK could be made, especially given the very different experiences with COVID-19). The four questions used asked respondents to indicate on a 0–10 likert scale: (i) how satisfied they are with life nowadays, (ii) how worthwhile they think things done in life are, (iii) how happy they felt yesterday, and (iv) how anxious they felt yesterday. These four well-being questions are reported on a scale from 0 representing ‘not at all’ to 10 representing ‘completely’. Full details of the survey are provided in Beck and Hensher (2020a, 2020b), with an overview given herein. A particular focus for data collection has been New South Wales, given the overarching objective of the project funding data collection, with a view to understanding how commuting behaviour might change during the pandemic as well as in a living with a COVID-19 world. As such, there is also a greater focus on workers, their experiences with working from home, and how it has affected commuting mode and time choices.

In this paper we focus on the interplay between working from home, distance to the main office and perceived productivity when WFH compared to the regular office, and the quality-of-life measures. The sample size, date, location, and key socioeconomic characteristics are summarised in Table 1 . All surveys were conducted online using the Pure Profile customer panel with two waves of data defined as a repeated cross section. The data were cleaned using widely accepted methods (extreme outliers, speed of completion, non-sensical responses). The resulting socio-demographic profile of the sample lines up well with the Australian Bureaus of Statistics (ABS) census in 2016,4 the latest release year (the 2021 census is yet unreleased). With the exception of the number of days working from home (an average of 2.38 in Wave 3 and 1.51 in Wave 4), the socio-demographic profiles are similar and not statistically different between the two cross section waves. Hence, in the formal modelling we test for difference due to working from home and permit generic parameters across waves for all other explanatory variables.

Table 1.

Overview of Survey Samples.

Variable	Units	Wave 3	Wave 4
Sample size		333	448
Days working from home	Number of days	2.38 (2.17)	1.51 (1.99)
Distance to work	kms	21.78 (29.3)	19.84 (20.23)
Household size	number	2.61 (1.22)	3.0 (1.29)
Number of cars	number	1.78 (0.88)	2.03 (0.92)
Number of children	number	1.60 (1.01)	1.70 (0.97)
Male	1.0	0.423	0.451
Age	years	39.41 (12.28)	41.03 (14.09)
Personal income	$ per annum/1000	89.16 (58.60)	85.29 (54.19)
Occupation:
Occupation professional and managerial	1,0	0.555	0.500
Occupation sales	1,0	0.069	0.103
Occupation clerical and administration	1,0	0.225	0.194
Occupation community and personal services	1,0	0.051	0.080
Occupation technology	1,0	0.048	0.045
Occupation machine operators	1,0	0.009	0.033
Occupation labourers	1,0	0.021	0.047
Satisfied with life nowadays	Likert scale (0–10)	6.34 (1.87)	6.78 (1.97)
Worthwhile are things you do	Likert scale (0–10)	6.80 (1.89)	7.01 (2.05)
Happiness yesterday	Likert scale (0–10)	6.53 (2.04)	6.90 (2.16)
Anxious level yesterday	Likert scale (0–10)	4.80 (2.67)	4.68 (2.91)
Productivity is more than pre-COVID-19	1.0	0.33	0.29

Note: standard deviation in brackets.

The sample average for each of the four states from both waves is presented in Fig. 2 (with the 95 % confidence interval shown by the error bars). There is a noticeable increase in overall satisfaction with life in Wave 4, along with the level of happiness felt, indicating that as people moved away from the initial peak of COVID-19 and life returned to some degree of normality, the mood of people improved; however, their overall level of anxiousness and perhaps more interestingly their perspective on how worthwhile the things they did in life were, remained somewhat unchanged between the two waves, potentially indicating that despite being less satisfied and happy, the sample still found worth in the activities and work being completed.

Fig. 2.

Sample Average Scores for Each Wave on Each Scale Item.

Table 2 shows the partial Spearman’s correlation matrix (appropriate for a Likert scale) for each scale item within each wave of data collection. The correlation patterns are very similar: satisfaction with life, how worthwhile people find the things they do in life to be, and happiness all exhibit strong positive correlations, whereas they are negatively correlated with anxiousness. Perhaps interesting to note is that the weak correlations exhibited with anxiousness in Wave 3 all but disappear in Wave 4.

Table 2.

Partial correlations between the four items.

Wave 3	SatLife	SatWorth	SatHappy
SatLife	1
SatWorth	0.75	1
SatHappy	0.74	0.73	1
SatAnx	−0.23	−0.16	−0.32
Wave 4	SatLife	SatWorth	SatHappy
SatLife	1
SatWorth	0.79	1
SatHappy	0.74	0.74	1
SatAnx	−0.05	−0.07	−0.16

Given the high partial correlation amongst the first three questions across both waves, we have chosen to focus on how worthwhile individuals think things done in life currently are, which we see as the most interesting question, although we investigate in the ordered choice model the relationship this variable has with anxiety. For this variable, given we also found no statistically significant difference between the responses obtained in the two time periods (see Fig. 2), we pool the Wave 3 and Wave 4 sample for this statement. Fig. 3 which shows the full distribution of responses for this statement in the pooled sample, as we as for each wave of data collection.

Fig. 3.

Distribution of the “how worthwhile are the things you do in life” statement.

In the distribution presented in Fig. 3, we see a right-skewed distribution with rating scores of 7 and 8 dominating. This already hints at evidence that satisfaction with life, in particular as people moved away from the initial peak of COVID-19 infections, was returning to some greater degree of positive ‘normality’ and was robust for those people who were still working during this period. However what interests us is investigating whether there is a systematic and statistically significant relationship between what the accumulating literature is suggesting are positive well-being outcomes of the pandemic, centred on increased working from home, and its association with productivity, reduced commuting, and the distance to the regular office. We implement an ordered logit choice model on the 11-point scale to investigate the presence or otherwise of such a relationship.

3. The ordered logit model

The response variable of interest is a rating scale of 11 levels which can be represented by an ordered choice model of the logit form. The ordered logit model explicitly recognises the ordinality of life satisfaction and avoids arbitrary assumptions about scale. It does this by defining points on the observed scale as thresholds that recognise, in preference space, that the numerical levels of the dependent variable are not strictly linear (Zavoina and McElvey, 1975, Greene and Hensher, 2010). Formally, let Y_i* denote an unobserved (or latent) continuous variable (-∞ < Y_i* < +∞), defined in utility space and µ_0, µ₁,…, µ_j-1, µ_J denote the threshold utility points in the distribution of Y_i*, where µ₁= -∞ and µ_J = +∞. Define Y_i to be an ordinal (observed) variable for life worth at present such that Y_i = j iff µ_j-1 < Y_i* < µ_j; j = 0,1,2,…,J response levels. Since Y_i* is not observed, the mean and variance are unknown. We assume that Y_i* has a mean of zero and a variance of one. To make the model operational, we need to define a relationship between Y_i* and Y_i. The standard ordered choice model is based on a latent regression model given as equation (1).

$${\mathrm{Y}}_i^{\ast }=\beta \text{'}{\mathrm{x}}_i+{\mathrm{\epsilon }}_i,{\mathrm{\epsilon }}_i\sim F({\mathrm{\epsilon }}_i|\theta ),\mathrm{E}({\mathrm{\epsilon }}_i)=0,\text{Var}({\mathrm{\epsilon }}_i)=1$$ (1)

where θ collects the mean and threshold parameters. The observation mechanism results from a complete censoring of the latent dependent variable as follows:

$$\begin{array}{c}{\text{Y}}_i=0\text{if}{\text{Y}}_i^{\ast }\le {\mathrm{\mu }}_0,\\ =1\text{if}{\mathrm{\mu }}_0<{\text{Y}}_i^{\ast }\le {\mathrm{\mu }}_1,\\ =2\text{if}{\mathrm{\mu }}_1<{\text{Y}}_i^{\ast }\le {\mathrm{\mu }}_2,\\ \begin{array}{c}...\\ =J\text{if}{\text{Y}}_i^{\ast }>{\mathrm{\mu }}_{J-1}.\end{array}\end{array}$$ (2)

The probabilities which enter the log likelihood function are given by equations (3).

$$\text{Prob}\left[y_i=j|x_i\right]=Prob\left(Y_i^{\ast }isin{j}^{\mathit{th}}range\right)=F({\mu }_j-\beta \prime x_i)-F({\mu }_{j-1}-\beta \prime x_i)>0,j=0,1,...J$$ (3)

This formulation ensures that all the thresholds are positive, preserves the ordering of the thresholds and incorporates the necessary normalisations. The model is fully consistent, in that the probabilities are all positive and sum to one by construction.

4. Ordered choice model results

Many ordered logit models were estimated to identify candidate influences on the probability of choosing a level on the rating scale associated with the worthwhile of things done in life. These included all the variables in Table 1. The descriptive profile of the statistically significant variables is included in Table 1 with the final ordered logit model in Table 3 where we have six statistically significant influences and the threshold parameters.5 We investigated preference heterogeneity associated with the explanatory variables and threshold parameters between each Wave but did not find any evidence to support this situation and a dummy variable for Wave 3 versus Wave 4 was found to be appropriate. While one might have considered a hybrid logit model form, typically used for unordered alternatives and not ordered alternatives, we are of the view that the link between the single ordered well-being variable and the socioeconomic variables is valid and behaviourally meaningful. What we did do, however, but it did not produce anything of statistical significance, is we interacted ‘How anxious did you feel yesterday’ with a number of socioeconomic variables, suggesting that these effects as a separate variable did a very good job in capturing and influence across the full socioeconomic spectrum of the data. This is behaviourally plausible because people of all segments of society feel anxiousness. The disturbance variance was not found to be heteroscedastic.

Table 3.

The Ordered Logit Model Results.

Dependent variable: Worthwhile of things done in life	Parameter estimate (t-value)
Constant	4.192 (13.9)
Wave 3 dummy variable	−0.2272 (-1.62)
Socio-economic:
Age	0.0141 (2.92)
Household size	0.1365 (2.93)
Other influences:
Distance of workplace from home	−0.0007 (-2.64)
Number of days WFH where productivity working from home is much more than at the office	0.3867 (2.15)
How anxious did you feel yesterday? (0–10 scale)	−0.0794 (3.32)
Threshold parameters:
µ1	0.9004 (4.76)
µ2	1.5212 (9.37)
µ3	1.9843 (14.3)
µ4	2.4592 (20.9)
µ5	3.3608 (37.9)
µ6	4.1221 (54.5)
µ7	5.0757 (68.6)
µ8	6.2279 (68.6)
µ9	7.5215 (52.1)
Log-likelihood at convergence	−1545.94
McFadden pseudo-R-squared	0.13
Adjusted pseudo-R-squared	0.12

In developing the final ordered logit model, we initially had anticipated including the number of days working for home or the proportion of days working from home as a separate explanatory variable.6 What we found to be statistically significant was an interaction between the number of days working from home and the level of productivity gain that was positive. We also found that including the distance of the commuting trip represented a statistically significant influence on the worthwhile of things done in life, which not only suggest a positive benefit from working from home as reduced commuting trips and the duration of each trip.

It is important to note that since we are interested in how working from home might impact on how worthwhile people find the things they are doing in life during the pandemic, this model is restricted to those who are employed. We acknowledge that being in employment or not is likely to have a significant impact on this statement, particularly if a person was made unemployed by the pandemic. We did explore the number of days worked as an explanatory variable (as a possible alternative to the number of days working from home, including distinguishing between zero and non-zero), but it was not statistically significant (r = -0.011, sig. = 0.460). Additionally, we also examined changes in the volume of work between Wave 3 and Wave 4 and find that there is no difference in the average response to life’s worth if a person lost one or more days of work or not (6.8 vs 7.0; t = 0.911), nor is there any correlation between the change in the number of days worked in total (r = -0.036, sig. = 0.318).

Age and household size are positive and statistically significant influences on the probability of higher rankings of the worthwhile of things done in life. This indicates that older workers tend, ceteris paribus, to have a greater probability during COVID-19 of greater well-being than younger individual and this is also the case when a worker belongs to a household with more members. The latter reflects the support mechanism available and the sense of reduce isolation under restrictions to regular behaviour outside the home. The Wave-specific dummy variable (Wave 3) suggests a statistically non-significant impact at the 95 % confidence level of the mean of the unobserved influences between the two waves.

The likert scale variable ‘how anxious they felt yesterday’ was statistically significant and negative indicating that the lower the level of anxiety the greater the probability of the worthwhile of things in life. However, given a possibility that the explanatory variable defining anxiety might be endogenous, we tested it through a control function (Guevara, 2015). The control function used in this paper is defined as the residuals from an ordered logit model for anxiety. The control function correction is an auxiliary variable to correct for the endogeneity (Guevara, 2015), so it is important to include the exogenous variables that will be considered in the ordered logit model for ‘worthwhile of things done in lifew otherwise the control function correction may confound the effect of the exogeneous variable with that of the residual. In the Anxiety ordered logit model the residuals are the observed minus expected values. We first identified any explanatory variables that influence the anxiety variable. The only variable was age with a negative parameter (i.e., anxiety decreases as one ages since rank = 0 is not at all). We then interacted anxiety with age and found no statistical influence on the well-being dependent variable but as a control function we fed the residuals from the anxiety ordered logit model into the ordered logit model for’ Worthwhile of things done in life’ and they were not statistically significant, which is an appropriate test of the presence or otherwise of endogeneity.

Given the non-linear nature of the model, the estimated parameters are not appropriate sources of behavioural interpretation, and instead we calculate elasticities as a way of establishing the relationship between a change in the level of an explanatory variables and the probability of being on a particular level of worthwhileness in life scale.

In Table 4 and Fig. 4 , we summarise the direct elasticities associated with the relationship between the level of a statistically significant influence on life and the probability of how worthwhile a respondent thinks the things that they do in life are. The mean direct elasticity varies as we move from the low rating of 0 to the high rating of 10, and indicates how a one percentage change in the level of a specific variable impacts on the probability of obtaining a particular rating level in terms of how worthwhile a respondent thinks the things that they do in life are? The elasticities associated with dummy variables recognise a 100 % change. The change of sign from negative to positive signifies that at the higher levels of the rating scale that a 1 percentage change in the level of the variable of interest results in a positive increase in the percentage change in the probability of being on a particular level of worthwhileness in life scale.

Table 4.

Direct Elasticities.

Dependent variable: Worthwhile of things done in life scale	0	1	2	3	4	5	6	7	8	9	10
Age	−0.562	−0.55	−0.534	−0.512	−0.481	−0.402	−0.257	−0.04	0.224	0.428	0.534
Household size	−0.409	−0.402	−0.388	−0.373	−0.35	−0.292	−0.187	−0.029	0.163	0.312	0.388
Distance of workplace from home	0.0142	0.0139	0.0135	0.0129	0.0122	0.0101	0.0065	0.001	−0.0057	−0.0108	−0.0135
# Days WFH when productivity working from home is much more than at the office	−0.341	−0.335	−0.327	−0.316	−0.301	−0.26	−0.181	−0.049	0.138	0.308	0.407
How anxious did you feel yesterday?	0.369	0.362	0.351	0.337	0.316	0.264	0.168	0.027	−0.147	−0.282	−0.351

Fig. 4.

Distribution of direct elasticities across the 11-point scale for all explanatory variables.

The direct elasticities associated with household size and age of respondent are the most inelastic of all the explanatory variables, although the range of most of the explanatory variables is quite substantial. We find, ceteris paribus, that with a 1 % increase in age, that the percentage change in the probability of the worthwhile of things done in life decreases by 0.562 % for not all (level = 0) through to 0.04 (level; 7) and then start to increase by 0.224 % for level 8 up to 0.534 % for level 10 (completely). The household size influence is similar to age and slightly less inelastic.

Three variables of particular interest (expanded in Fig. 5 ) are the distance between the home and workplace location (a mean of 20.4 km), the incidence on working from home being much more productive than in the regular office location used pre-COVID-19 (a mean of 30 %), and how anxious a respondent was during the pandemic period (a mean of 4.75 on the 0–11 scale).

Fig. 5.

Distribution of direct elasticities across the 11-point scale for three key explanatory variables.

The sign range from negative to positive, as we move through the 11-point scale, suggests that as the distance of a commute between home and work increases, as a percentage, it results in a respondent having a greater probability of feeling that things done in life are not being worthwhile at all or to a lesser degree, and the opposite probability lowering of a response on the worth of things done in life being completely or greatly worthwhile. This aligns well with evidence in Hensher et al., 2021a, Hensher et al., 2022a, Hensher et al., 2022b, Hensher et al., 2021b on the influence that the longer commuting distance has on the increase in the probability of working from home. In contrast, as productivity because of working from home increases, ceteris paribus, the feeling that the worth of things done in life not being worthwhile at all or to a lesser degree (within the scale range of 0 to 8) decreases and the feeling that the worth of things done in life being completely or greatly worthwhile (in the 9 to 10 scale range) increases. Again, this aligns well with expectations that the work-leisure balance improves when there is greater flexibility in when and where to work (Hensher et al., 2021a, Hensher et al., 2021c, Hensher et al., 2021b). The elasticity associated with the feeling of anxiety is of interest, suggesting that when you are much more anxious, compared to not being so at all, the probability of the feeling that the worth of things done in life being completely worthwhile reduce as might be expected.

5. Conclusions and implications

The evidence in this paper suggests that during the COVID-19 pandemic, the opportunity to work from home, which has resulted in increased work-related productivity and an improved balance between time spent commuting, and time spent on non-commuting activities, has contributed in a positive way to improving the worth status of life, offsetting some of the negative consequences of the pandemic. This suggests that had the pandemic been associated with a requirement to continue commuting at the same level as pre-COVID-19, and hence a reduced ability to work from home and gain greater flexibility in work hours, the consequences on an individual’s mental state may have been much worse, putting aside the greater risk of being exposed to the virus. Importantly, there was no statistically significant relationship between anxiety and the extent of working from home, but we did find a positive relationship between working from home and the worth status of life in the presence of positive work-related productivity gains (partial correlation of 0.78).

The evidence in this paper suggests that some amount of working from home (average of 2.38 days per week in Wave 3 and 1.51 days per week in Wave 4) results in a perceived improvement in productivity and also contributes in a positive way to increasing the probability of the worthwhile of things done in life. There is, in the main, always an option to go to the office except when there is a government mandated lockdown, which was not the case during Waves 3 and 4 data collection. We also know from analysis of the same data with a different focus that the time ‘saved’ from commuting when working from home is converted to a mix of work and leisure, adding a work-related productivity benefit but also a gain if lifestyle through increased leisure (both in and out of the home. Hensher et al. (2022a) found that of the average 60.5 min per week (standard deviation of 107 min) saved from commuting, 47.5 per cent (standard deviation of 36.1 per cent) is allocated to leisure, 29.1 per cent (standard deviation of 32.7 per cent) to paid work and 23.4 per cent (standard deviation of 28.2 per cent) to unpaid work.

We have examined potential endogeneity between the number of days working from home in concert with the perceived productivity of working, and the level of anxiety felt by respondents. Our findings are such that we are confident that this is not a significant bias in the data under analysis. However, we do recognise that for some respondents it may be the case that working from home is an unproductive experience, and that lack of productivity may result in changes to the amount of anxiety felt. It should be noted that the period under analysis (September 2020 to June 2021) was a point in the pandemic in Australia where relative normality was returning to society, and workplaces were encouraging people to return to the office but predominantly still offering workers unprecedented levels of flexibility as to when they WFH and how often. In this context it is important that we plainly state that no one should consider working from home being an all or nothing approach, rather it is the mix of working from home that matters, and in a flexible world without COVID restrictions or previous workplace constraints, an individual will find an equilibrium that works best for them, and this is why we see the persistent levels of working from home over this eight month time period when community based transmission of COVID-19 was virtually nil and thus health restrictions were minimal. That said, we do acknowledge that for some individuals in this mix, there are some who may not have the choice to return to the office, who for whatever reason are not productive while working from home, and that lack of productivity in work may be increasing their anxiousness and thus lowering their worthwhile status. This is not an unimportant point, and highlights the other end of the working from home continuing: that workers should not only have increased freedoms to work from home, but the right also to come to work is equally important because the workplace can often represent an important source of social interaction and inclusion, opportunity advancement and mentoring, and at the extreme can even be a place of refuge and respite for those experiencing domestic violence.

What does this paper add to the literature? Importantly we promote a view that there have been unintended positive consequences of COVID-19 that have cushioned the severity of the pandemic to some degree, and that this should be recognised as an immediate benefit. More importantly, however, is the potential for longer term gain in well-being and lifestyle that may not have been offered up if life had continued along the journey associated with the pre-COVID-19 state of travel, commuting and associated pressures on the work-home balance. Given that it is likely that working from home will continue to feature as a greater proportion of where work is completed, it is crucial to develop and implement best practices for working from home to maintain a good level of productivity, achieve the right level of work and life balance and maintain a good level for physical and mental health.

In particular, we have shown that productivity in work can lead to a greater sense of worth and thus positive well-being outcomes. To work from home well, robust ICT infrastructure and digital literacy is important particularly as the world is expected to focus even more on digitisation and technology after the pandemic (Gasser et al., 2020). As argued by Carnevale and Hatak (2020), organisations must remain alert and adaptive to unforeseen events and find new solutions to challenges arising across many areas of their operations. They may need to find ways in which to give employees greater job autonomy and associated self-responsibility which positively impacts working from home (Shin, 2004, Stiglbauer and Kovacs, 2018), especially given that individual work personality can change (Dweck, 2008, Tasselli et al., 2018).

While we would have preferred that the virus had not taken hold, we must look forward to using this extreme event experience to obtain positive benefits to individuals, households, and society more broadly. This position must recognise that mental health and well-being, including social exclusion has not gone away (see Stanley et al., 2021) and that it remains a high priority for governments as well as for business more generally; however, let us recognise that some good has come out of the pandemic to provide some directions to better support well-being that were not on offer before COVID-19. The policy implication is very clear; namely, to continue to ensure that people can work from home successfully, and know they are contributing while doing so. Meaningful work provides meaning to life.

In ongoing data collection during the subsequent period(s) of Omicron and other variations of the virus, we will investigate the extent to which the very positive evidence in the first two years of COVID-19 (2020–2021) is seen to continue. 2022 is showing strong signs of new actions to live with COVID-19, despite Omicron tending to produce a new kind of lockdown, which we refer to as ‘voluntary lockdown’; but what seems clear thus far is that the new levels of working from home (and working near home in satellite offices) appear set to continue as a result of this positive outcome to date.

CRediT authorship contribution statement

David A. Hensher: Conceptualization, Methodology, Writing – review & editing. Matthew J. Beck: Conceptualization.

Declaration of Competing Interest

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.