The Future of Work
What would it take to bring people back into the office?
Haokai Du, Sravya Jahagirdhar, Malyadri Reddy Mule, Michael Salceda
December 14, 2022
Abstract
With the prevalence of remote work, the problem our team wanted to determine is what would it take to bring people back into the office. To analyze this, we created “packages” of incentives, from commuting credits and vacation days to salary increases. These incentives may entice more people to come into the office rather than working remotely. However, to make the decision more challenging, we will also test some non-compliance penalties, in particular, salary cuts. If employees don’t want to work in the office, then perhaps getting a salary cut instead of incentives will push them to consider working in the office. Based on the results of our survey, it seems that the most effective incentives for in-office work will be salary increases and monthly attendance bonuses, but should people not want to come in, either salary cuts or reducing the number of required in-person days from a full work week may be effective as well.
Introduction
Post-pandemic, many of us are eagerly waiting to get back to life as we want it, but there are a section of people who want to work remotely. A survey conducted by McKinsey & Company (Dua et al. 2022) found that, in the US, around 58% of employed respondents had an option to work remotely at least one day a week (Figure 1).
Figure 1: Graphic from McKinsey showing the proportion of workers eligible for remote work.
In addition, of those 58%, 87% take that option to work remotely (Figure 2).
Figure 2: Graphic from McKinsey showing the proportion of remote-work eligible workers that actually work remotely.
From a global survey, people from 8 out of 10 countries are considering working from home because of the benefits like no traveling and more productivity from distractions at the office (Brower 2021). The same Forbes article also stated that people from all 10 out of 10 countries surveyed are concerned the most about isolation. This raises the question: how do we bring them back to the office?
There are a few expectations from employees. What they enjoyed at home shouldn’t go unnoticed. Leaders or bosses should take necessary requirements to engage and motivate their employees to bring them back on track (Capossela 2022). Through the use of incentives (e.g., salary increases, extra vacation days, etc.) to create an incentive “package” for working in-person as well as a penalty (a salary cut in our case), we are attempting to create a situation in which respondents have to make a trade-off between staying at home and working in-person. Our survey and subsequent analysis aims to figure out what employers can do to entice employees back to the office so employees don’t feel like they’re missing out too much on what they had with remote work.
Survey Design
Our target population was anyone who: (1) currently works from home or has worked from home within the past two years, (2) has a choice or had a choice within the past two years to work from home or in-person at the office, and (3) would at least consider the idea of coming into the office. We wanted to target these particular respondents because we wanted to see what would make a respondent come into the office if they are working from home. In order to ensure respondents actually care about this topic, we also had to only select respondents that had a choice between in-person and remote work. Workers that did not have a choice and had to work remote or in-person only were not relevant to our survey.
Critical information that we collected from respondents are their industry, salary, and work benefits information. Demographically, we have the standard battery of questions (race, gender, and education). The educational information we provided to respondents covered the meaning of the attributes, particularly what each incentive entails. Full details on the educational information can be found in the Appendix.
We selected six attributes for the survey that would help people assess the available solutions and offered each respondent eight choice questions. Every attribute has a distinct quality and advantage that influence the employees’ decisions to come to work each day. The attributes we defined are as follows:
- Salary Increase: how much of a yearly salary increase a respondent will receive expressed as a percentage; ranges in value from 0 to 10
- Required In-Person Days: how many days per week a respondent is required to be in the office; ranges in value from 1 to 5
- Monthly In-Person Attendance Bonus: how much of a bonus (expressed as a percentage) a respondent will receive per month if they accept the in-office policy and if they come into the office the required amount of days (specified by the “Required In-Person Days” attribute); values can be 0, 1, or 2
- Monthly Commute Credit: how much credit in USD a respondent will receive per month to use towards commuting to the office (e.g., using rideshares like Lyft or Uber, taking public transportation, gas expenses, etc.); values can be 0, 50, 100, or 200
- Extra Yearly Vacation Days: how many extra vacation days per year a respondent will receive if they accept the in-office policy; values can be 0, 5, 10, or 15
- Salary Cut Penalty: how much of a yearly salary decrease a respondent will receive if they do NOT accept the in-office policy presented expressed as a percentage; ranges in value from 0 to 10
We only offered two options, “Accept” and “Reject” (our “no choice” option), because our survey is mostly composed of respondents who want to select between working from home or an office. The penalty of not selecting the offered options were included as the “no choice” option. See Figure 3 for an example of what our choice question looks like.
Figure 3: A sample conjoint question presented to respondents.
Data Analysis
To get to our valid sample, we had to filter many survey responses down. In total, we received 486 survey responses. More than 50% were screened out which left us with 193 responses. After that, we removed all of the survey replies that contained incomplete information and dropped responses where the response time was too fast as this potentially indicated that respondents just selected a random option without reading/considering the question fully. What was considered “too fast” was anyone falling below the 5th percentile in terms of total time spent taking the survey. For our data, this was approximately 2.55 minutes so anyone who spent less than 2.55 minutes was filtered out. After this cleaning, we found that 178 of the responses were left which represents about 36.63% of our total responses. In total, this would give us a data set of 2,848 observations.
Figure 4: Demographic distribution of our valid respondents.
From our demographic distributions (Figure 4, Table 1, Table 2), our respondents come mostly from the tech industry and the “Other” industry. We had over 50% of our respondents between the ages of 31-49 and having at least a Bachelor’s degree. Interestingly, we had an almost equal distribution of incomes. From our critical questions regarding vacation time and commute time, the majority of our respondents were given at least 7 days of vacation time a year and most of them also had a commute time to work between 15-29 minutes.
|
|
|
|
|
|
|
The utility model we estimated is as follows: \[ \tilde{u}_j = \beta_1x_1^{Salary \ Increase} + \beta_2x_2^{Req. \ In-Person \ Days} + \beta_3x_3^{Attendance \ Bonus} + \beta_4x_4^{Commute \ Credit} + \beta_5x_5^{Extra \ Vacation \ Days} + \beta_6x_6^{Salary \ Cut} \] All the variables can take on continuous values between our defined range. We did have the “No Choice”/“Reject” option, but we did not add that as a variable to estimate because it automatically interacted with salary cuts, i.e., the only time salary cut could be non-zero was when the presented in-office incentive package was rejected. Using logitr to model, we got the following coefficients and standard errors found in Table 3.
| Coefficient | Est. Value | Std. Error |
|---|---|---|
| \(\beta_1\) | 0.1677565 | 0.0197891 |
| \(\beta_2\) | -0.4143918 | 0.0413080 |
| \(\beta_3\) | 0.2561625 | 0.0769710 |
| \(\beta_4\) | 0.0027352 | 0.0008604 |
| \(\beta_5\) | 0.0422536 | 0.0112736 |
| \(\beta_6\) | -0.1673890 | 0.0201141 |
Results
Since our in-office policies did not have a “price” component, we were unable to calculate a willingness-to-pay (WTP). While some of our attributes like salary increases and salary cuts have real dollar values, they were expressed as percentages in our survey and our data so we could not use those variables as a “price” to calculate WTP. Instead, we calculated the change in utility over all the levels in each of our attributes (Figure 5).
Figure 5: A series of plots showing the change in remote work utility for each attribute as they increase in value.
From the above series of plots, we can determine that the variables that have the biggest effect on utility are salary increases, required in-person days, and salary cuts. They have quite steep slopes compared to the other attributes. One key thing to note is that attendance bonuses also play a big role (as seen in Table 3), but due to the comparative scale of the y-axis and the fact that we only varied monthly attendance bonuses from 0% to 2% in our survey, the effect on utility is not immediately obvious without looking at the coefficient itself.
It is apparent that the “Required In-Person Days” attribute is the largest in magnitude - this implies that respondents really do not want to come in. So then what would a company have to offer if they wanted a full-week in-person policy? To answer this, we did a market simulation with five scenarios outlined in Table 4.
| Accept/Reject In-Person Policy | Salary Increase (%) | Req. In-Person Days | Salary Cut (%) |
|---|---|---|---|
| Market Scenario 1 | |||
| Reject | 0 | 0 | 0 |
| Accept | 0 | 5 | 0 |
| Market Scenario 2 | |||
| Reject | 0 | 0 | 5 |
| Accept | 5 | 5 | 0 |
| Market Scenario 3 | |||
| Reject | 0 | 0 | 10 |
| Accept | 10 | 5 | 0 |
| Market Scenario 4 | |||
| Reject | 0 | 0 | 0 |
| Accept | 10 | 5 | 0 |
| Market Scenario 5 | |||
| Reject | 0 | 0 | 10 |
| Accept | 0 | 5 | 0 |
To make our simulations easier and more comparable, we just varied the salary increase and salary cut variables, keeping everything else the same. Table 4 shows the relevant variables for our market simulations - the other variables were all held at 0. The required number of in-person days was kept static at 5 so we could see how high the barrier was for the salary increase and salary cut variables to overcome a full work week of in-person, and which variable was more convincing: an increase or a cut.
Our first market scenario compares “status quo” - full work week with no incentives versus staying at home with no penalty. Our second and third market scenarios compare alternatives where the salary cut penalty and the salary increase incentives are the same level - 5% and 10% respectively. Finally, our fourth and fifth market scenarios compare alternatives where the salary cut and salary increase attributes are at their opposites, i.e., salary cut is a minimum of 0% for rejection while salary increase is at its max of 10% for acceptance and vice versa.
We are able to determine the “baseline” effect from our first market scenario - what would people be doing now without any intervention? Our second and third market scenarios are attempting to simulate at what salary difference between at-home and in-person salaries would motivate going into the office. Our final two scenarios are simulating whether either salary increases or cuts at their max values are good enough to overcome the large effect of a 5 day/week in-person policy. Figure 6 below shows our results and Table 5 shows them in tabular form.
Figure 6: A series of plots showing the predicted market share for each market scenario. Alternative 1 is the “Reject” option and Alternative 2 is the “Accept” option.
| Accept/Reject In-Person Policy | Predicted Share (%) | 95% Confidence Interval |
|---|---|---|
| Market Scenario 1 | ||
| Reject | 88.81 | [84.07, 92.16] |
| Accept | 11.19 | [7.84, 15.93] |
| Market Scenario 2 | ||
| Reject | 59.78 | [51.13, 67.97] |
| Accept | 40.22 | [32.03, 48.87] |
| Market Scenario 3 | ||
| Reject | 21.76 | [14.51, 30.99] |
| Accept | 78.24 | [69.00, 85.49] |
| Market Scenario 4 | ||
| Reject | 59.73 | [48.22, 69.92] |
| Accept | 40.27 | [30.08, 51.78] |
| Market Scenario 5 | ||
| Reject | 59.82 | [47.97, 69.97] |
| Accept | 40.18 | [30.02, 52.03] |
It is apparent that, at baseline, people do not want to come in. From our first market scenario, we see that the majority of people would most likely want to stay home - the rejection alternative has 88% of the market share. As we look at market scenarios 2 and 3, we see that people would still be willing to take a 5% salary cut over a 5% salary increase rather than go in full time; as soon as we go to 10% though, working in person is more palatable as scenario 3 shows 78% of people would accept the policy. This seems to imply that there needs to be more than a 10% difference in at-home and in-person pay for someone to be willing to come in. Finally, our fourth and fifth market scenarios show that the effect of a salary increase and salary cut are essentially equivalent - a 10% cut or bump shows almost no change in market share for each alternative across scenarios. This is supported by the coefficients for salary cut and salary increase as they are equal until the ten-thousandths decimal place.
Expanding on our market scenario analysis, we performed a one-way sensitivity analysis on our attributes (Figure 7) and confirmed our previous findings that salary cut and salary increase are both equal in terms of affecting market share.
Figure 7: Plots of our sensitivity analyses on market share for all our attributes.
What is interesting is that, in order for a 5-day/week policy to achieve at least 50% market share, either the salary cut or salary increase needs to be around 12.5%. Since each sensitivity analysis was conducted with a baseline of a 0% salary cut/increase, no other incentives, and a 5-day in-person policy, this shows us that there needs to be at least a 12.5% difference between at-home and in-office pay for someone to consider coming in full time. We can also see from our tornado plot in Figure 8 that none of the other incentives seem to play a large enough role to push market share over 50%.
Figure 8: Tornado plot of our attributes. The red dashed line is at 50% market share.
This begs the question: what about the monthly attendance bonus? That attribute had the largest positive coefficient of all the variables so why is that not reflected in our sensitivity analysis? The reason for this is, again, due to scale. The monthly attendance bonus is essentially another form of a salary increase - just with an additional caveat, namely that one has to not miss a single day of in-person work days for each month, consistently through the year. Our sensitivity analysis only varied values from 0% to 5%, but we know from the salary increase/cut plots that there needs to be at least a 12.5% difference in at-home and in-person salaries for there to be a 50% market share or more, hence the discrepancy.
Final Recommendations and Conclusions
Based on our survey results, people’s desire to work from an office still seems to be far lower than their readiness to work from home. Our results indicate that there needs to be at least a 12.5% difference between a person’s at-home salary and in-office salary. This can take the form of either salary cuts or salary increases. If there is a hard monetary value that can be attributed to particular benefits outside of salary changes, that should be used in the calculation of how much to offer a person in a particular policy. As an example, a company policy of a 5% salary increase and a 2% monthly attendance bonus with a 5.5% salary cut penalty for staying at home would convince someone to come into the office. If a company only wants to do salary cuts, a policy with at least a 12.5% salary cut penalty will need to be implemented. Both of these policies would bring people into the office according to our results since this produces a 12.5% salary difference between working at home and working at the office.
As evidenced from our analysis, other incentives such as extra vacation days and commute credit - in other words, benefits that do not directly result in a higher take-home pay - seem to not have a major effect on peoples’ attitudes towards working in the office. Therefore, companies need to do some more research on other incentives that would be more enticing for their employees. For a more immediate effect, however, direct cuts or increases on pay will be enough to convince people to work in the office - as long as they are large enough.
Limitations
For future studies and research, it would be good to address some limitations we found in our study. First, we only covered a limited number of incentives. Respondents discussed additional benefits that would convince them to come into the office such as increased health benefits, more in-office socialization options, and the possibility to have a flexible schedule. Another survey conducted could include these additional benefits which may change the coefficients of our attributes. Another limitation we found in our study is the distribution of respondents, particularly with respect to the industry. As evidenced in Figure 4 and Table 1, the majority of our respondents came from the “Other” industry - this indicates respondents are part of an industry that is not entirely covered by the ones mentioned in the industry demographic question. A future study could delve into what is included in this “Other” industry category to better understand where these respondents are coming from. One final limitation was our inability to create a willingness-to-pay model. Because we did not explicitly have a price/cost-related variable, we could not estimate a model to identify a person’s WTP for a particular policy.
Should these limitations be addressed in the future, we believe that a more complete picture of the work-from-home and in-office discussion can be formed, and companies can form policies that benefit both the company and their employees.
Appendix
The following is a copy of our survey. We only provide one example of the conjoint questions since the values were randomized for each respondent.
Welcome to our survey!
Thank you in advance for participating. This survey is being done as part of a research effort by The George Washington University. In this survey, we will be asking about remote work and in-person work policies. More specifically, we are interested in designing policies to bring people back into the office.
NOTE: This survey may not work in Firefox. It may also not work in “Incognito” mode. We recommend using another browser such as Google Chrome, Microsoft Edge, or Safari if you find the survey does not work or not using “Incognito” mode.
Consent Form
This survey is being conducted by students at The George Washington University. All responses will be used solely for academic purposes. Your responses will be confidential, and we will not collect or share any identifying data such as your name or address.
The whole survey will take approximately 10 to 15 minutes to complete. Your participation is voluntary, and you may stop the survey at any time.
If you would like to participate, please answer the following questions:
- I am 18 years old or older.
- Yes
- No
- I have read and understood the above information.
- Yes
- No
Thanks!
Before we go on, we just need a little more information.
- Do you currently work from home, or have you worked from home previously within the past two years?
- Yes
- No
- If you currently work from home, do you currently have a choice between working from home and in-person at the office? If you worked from home previously within the past two years, did you have a choice between working from home and in-person at the office?
- Yes
- No
- If you currently work from home, are you willing to at least consider the idea of coming into the office? If you worked from home previously within the past two years, would you have been willing to at least consider the idea of coming into the office?
- Yes
- No
- I am a robot.
- Yes
- No
Great work!
Now that you’ve shared a bit about yourself, we’d like you to consider a scenario in which you can choose between a new in-office work package or rejecting the package and continuing to work from home.
Let’s learn about these attributes.
In-Person Work Package Attributes
Each in-person work package will have some attributes to convince you to come into the office. Here’s a description of those attributes:
Salary Increase
The “Salary Increase” attribute refers to how much of a yearly salary increase you will receive if you decide to accept the in-office policy. The increase is shown as a percentage increase over your current yearly salary. For example, if the salary increase is 3% and your current yearly salary is $100,000, your new yearly salary will be: $100,000 + (3% of $100,000) = $103,000.
Required In-Person Days
This value refers to how many days per week you are required to be in the office if you decide to accept the in-office policy. For example, if the “Required In-Person Days” value is 3, you are required to come into the office 3 times a week.
Monthly In-Person Attendance Bonus
The “Monthly In-Person Attendance Bonus” refers to how much of a bonus you will receive per month if you accept the in-office policy and if you come into the office the required amount of days (specified by the “Required In-Person Days” attribute). This value is a percentage of your monthly salary. For example, if the “Required In-Person Days” value is 3 and the “Monthly In-Person Attendance Bonus” is 1%, you will receive a monthly bonus of 1% of your monthly salary if you had worked in the office 3 times a week that month.
Monthly Commute Credit
This value refers to how much credit you will receive per month to use towards commuting to the office (e.g., using rideshares like Lyft or Uber, taking public transportation, gas expenses, etc.) if you accept the in-office policy. For example, if the “Monthly Commute Credit” value is $100, you would receive $100 per month to use for taking the train, taking Uber/Lyft, paying for gas, etc.
Extra Yearly Vacation Days
This attribute value refers to how many extra vacation days per year you will receive if you accept the in-office policy. For example, if the “Extra Yearly Vacation Days” value is 5, you would get 5 extra vacation days to use during the year.
Salary Cut Penalty
The “Salary Cut Penalty” refers to how much of a yearly salary decrease you will receive if you do NOT accept the in-office policy presented. This value is expressed as a percentage decrease over your base salary. For example, if the salary cut is 3% and your current yearly salary is $100,000, your new yearly salary will be: $100,000 - (3% of $100,000) = $97,000.
We’ll now begin the choice tasks. On the next few pages we will show you an in-person work policy package and the rejection option.
For example, if this was the in-person work package, which would you choose?
In-Person Package
| Salary Increase: | 5% |
| Required In-Person Days: | 1 Days/Week |
| Monthly Attendance Bonus: | 2% |
| Monthly Commute Credit: | $200 |
| Extra Yearly Vacation Days: | 15 Days/Year |
Reject In-Person Package
If you choose this option, you will get a 50% salary cut.
Awesome!
We will now show you 8 sets of choice questions starting on the next page.
(1 of 8) Would you accept the following package to go back into the office? If not, you will get the following non-acceptance penalty: 3% salary cut.
In-Person Package
| Salary Increase: | 5% |
| Required In-Person Days: | 2 Days/Week |
| Monthly Attendance Bonus: | 0% |
| Monthly Commute Credit: | $0 |
| Extra Yearly Vacation Days: | 15 Days/Year |
Reject In-Person Package
If you choose this option, you will get a 3% salary cut.
Almost there!
- What industry are you currently working in/did you work in? (Choose one.)
- Tech
- Healthcare
- Government/Public Services
- Education
- Construction
- Retail
- Agriculture
- Transport
- Utilities (Gas, Electric, Water)
- Other/Not Shown
- On average, how long does/would it take for you to get to work (one way)?
- Less than 15 minutes
- 15 - 29 minutes
- 30 - 59 minutes
- 60 or more minutes
- How many vacation days does your current company offer per year?
- Unlimited vacation days/year
- Less than 7 vacation days/year
- 7 to 14 vacation days/year
- Over 14 vacation days/year
- No vacation days/year
- Of the vacation days given by your company, on average, how many do you use each year?
- Less than half
- At least half but not all
- All
- Not applicable
The following questions are optional but will help us in our study.
- What benefits and personal demands would you want from your organization so that you would be more interested in working from the office?
- [TEXT BOX]
- Do you have any circumstances that currently force you/has forced you to work from home instead of in the office (e.g., new baby, other dependents, disabilities, etc.)? If you do, please explain below.
- [TEXT BOX]
Nice job!
We’re almost done! We’d just like to ask just a few more questions about you which we will only use for analyzing our survey data.
- In what year were you born?
- [DROPDOWN FROM 1920 - 2004, “Prefer not to say”]
- What is your current gender identity?
- Male
- Female
- Trans Man
- Trans Female
- Genderqueer/Gender Non-Conforming
- Prefer not to say
- Different identity (please state)
- I identify my race as (select all that apply):
- Asian
- African American or Black
- White (Not of Hispanic or Latino Origin)
- Hispanic or Latino
- American Indian or Alaska Native
- Native Hawaiian or Pacific Islander
- Prefer not to say
- Different race (please state)
- What is the highest degree or level of school you have completed? If currently enrolled, please use the highest degree received.
- Less than a high school diploma
- High school degree or equivalent (e.g., GED)
- Some college or university, no college degree
- Trade/technical/vocational training, no degree awarded
- Associate’s degree (e.g., AA, AS)
- Bachelor’s degree (e.g., BA, BS)
- Graduate or Professional Degree (e.g., PhD, MD, JD, MS)
- Prefer not to say
- What is your annual salary before taxes and other deductions from pay?
- Less than $10,000
- $10,000 - $14,999
- $15,000 - $24,999
- $25,000 - $34,999
- $35,000 - $49,999
- $50,000 - $74,999
- $75,000 - $99,999
- $100,000 - $149,999
- $150,000 - $199,999
- $200,000 or more
- Prefer not to say
Please let us know if you have any other thoughts or feedback on this survey. Your feedback will help us make future improvements! Thank you! :)
- [TEXT BOX]