Examining the persistence of telecommuting after the COVID-19 pandemic

Figures & data

Figure 1. The spatial distribution of study sample.

Figure 2. Telecommuting frequency changes during and after the pandemic.

Figure 3. Latent constructs and outcomes descriptive analysis.

Table 1. Summary statistics of the significant explanatory variables in the final model (Sample size = 583 observations).

Explanatory variables	Description	Mean	Std. Dev.
Socio-demographic & household characteristics:
Gender
Female	1: If the employee is female; 0: Otherwise	0.568	0.496
Age
Generation Z	1: If the employee is between 9 and 24 years old; 0: Otherwise	0.015	0.123
Generation Y (Millennials)	1: If the employee is between 25 and 40 years old; 0: Otherwise	0.326	0.469
Telecommuting experience
New telecommuter	1: If the employee did not telecommute before the pandemic and they are experiencing telecommuting during the pandemic; 0: Otherwise	0.374	0.484
Race
White_race	1: If the person is White/Caucasian; 0: Otherwise	0.842	0.365
Education:
Bachelor’s degree or graduate school_education	1: If the person has a bachelor’s degree; 0: Otherwise	0.413	0.493
Household structure
No. of children	Number of children in the household	0.635	1.001
Income
High income	1: If the annual household’s income is more than 125 K; 0: Otherwise	0.333	0.472
Job category
Clerical or administrative support_job	1: If the person’s job is in the clerical or administrative support job category; 0: Otherwise	0.075	0.264
Professional, managerial, or technical_job	1: If the person’s job is in professional, managerial, or technical or administrative support job category; 0: Otherwise	0.453	0.498
Commute features
Pre pandemic_ commute mode private vehicle	1: If the person commuted by their private vehicle before the pandemic; 0: Otherwise	0.674	0.469
Pre pandemic_ commute time > 30 min	1: If the person commuted more than 30 minutes before the pandemic; 0: Otherwise	0.292	0.455

Figure 4. The generalized structural equation modeling structure.

Observed determinants and latent variables are presented with white rectangles and yellow ovals, respectively. Blue rectangles demonstrate outcome variables.

Table 2. Estimation results of the generalized structural equation model.

Variables	Estimated value	p-value
Latent variable: Productive Telecommuting
Measurement variables:
No commuting to work_W2
coefficient	1.000	Constrained
constant	.273***	0.000
Feeling sad/depressed_W2
coefficient	−3.891**	0.027
constant	.0518***	0.010
More distractions at home_W2
coefficient	−5.130**	0.027
constant	0.056**	0.018
Need equipment/technology not available_W2
coefficient	−1.053**	0.043
constant	0.023**	0.017
Lack of comfortable workspace_W2
coefficient	−2.581**	0.029
constant	0.012	0.363
Child, elderly, or dependent care_W2
coefficient	−2.725**	0.028
constant	0.012	0.402
Exogenous variables:
Female	−.0152*	0.064
Professional, managerial, or technical_job category	−.0131*	0.079
Clerical or administrative support_job category	−0.017*	0.158
Latent variable: Risk Perception _Wave 1
Measurement variables:
Shopping at grocery stores
coefficient	0.845***	0.000
constant	3.081***	0.000
Riding public transit
coefficient	1.000	Constrained
constant	4.074***	0.000
Taking a taxi/ride-hailing service
coefficient	1.025***	0.000
constant	3.509***	0.000
Exogenous variables:
Female	0.119*	0.066
White_race	−0.178**	0.043
No. of children	0.083***	0.009
Latent variable: Risk Perception _Wave 2
Latent variables:
Risk Perception _Wave 1	0.854***	0.000
Measurement variables:
Shopping at grocery stores
coefficient	0.741***	0.000
constant	2.886***	0.000
Riding public transit
coefficient	1.000	Constrained
constant	3.842***	0.000
Taking a taxi/ride-hailing service
coefficient	0.846***	0.000
constant	3.257***	0.000
Exogenous variables:
Generation Z	−0.448*	0.084
Outcome variable: Telecommuting frequency_wave 1	Estimated value	p-value
Latent variables:
Risk Perception _Wave 1	0.254^♣	0.105
Exogenous variables:
High income	0.561**	0.013
Bachelor’s degree_education	0.352*	0.080
Generation Y	0.317^♣	0.138
Pre pandemic_commute time>30 min	0.358^♣	0.120
Thresholds:
$C_{11}$	−3.188***	0.000
$C_{12}$	−2.18***	0.000
$C_{13}$	−0.638***	0.000
Outcome variable: Telecommuting frequency_wave 2	Estimated value	p-value
Latent variables:
Productive Telecommuting	4.663^♣	0.133
Risk Perception _Wave 2	0.221*	0.094
Exogenous variables:
Pre pandemic_commute mode private vehicle	−0.445*	0.051
Endogenous variables:
Telecommuting frequency_wave 1	1.624***	0.000
Thresholds:
$C_{21}$	−0.424	0.32
$C_{22}$	0.648^♣	0.107
$C_{23}$	2.812***	0.000
Outcome variable: Telecommuting frequency_expectation	Estimated value	p-value
Latent variables:
Productive Telecommuting	5.5*	0.086
Exogenous variables:
New telecommuter	−1.845***	0.000
Endogenous variables:
Telecommuting frequency_wave 2	0.722***	0.000
Thresholds:
$C_{31}$	−0.725**	0.031
$C_{32}$	0.344	0.302
$C_{33}$	2.356***	0.000
Number of observations	583
Log likelihood at convergence	−5912.58
Pseudo R-squared	0.2

Note: ***, **, *, and ^♣ denote 99%, 95%, 90%, and 85% confidence level, respectively.

Figure 5. Marginal effects of ‘New telecommuter’ on telecommuting frequency expectation in the post-pandemic era.