Policy Research Working Paper 10630 The Valuation of Flexible Work Arrangements Insights from a Discrete Choice Experiment in Malaysia Yashodhan Ghorpade Alyssa Jasmin Amanina Abdur Rahman Social Protection and Jobs Global Practice December 2023 Policy Research Working Paper 10630 Abstract The changing nature of work, accelerated by the experience associated earnings. The findings show substantial but not of the COVID-19 pandemic, has resulted in several funda- overwhelming preference for greater flexibility, especially mental shifts in the terms and conditions of work. Along among freelancers, and a clear trade-off between measures with the clear trend of increased nonstandard employment, of flexibility and income. The findings also show consid- including through the gig economy and platform work, erable variation in the preference for flexibility, much of this poses critical questions for policies and practices of which is not explained by worker demographics and other the organization of work arrangements, and about who observable characteristics but is consistent with other mea- may bear the costs of emerging arrangements. This paper sures of the importance attached to flexibility and earning explores whether workers in freelancing and standard work income. The analysis outlines pathways through which arrangements in Malaysia view a trade-off between flexibil- offering even a modicum of flexibility can enhance workers’ ity and income and are willing to forgo a share of earnings utility without necessarily increasing costs for employers, for greater flexibility. The paper deploys a novel discrete provides evidence of considerable preference heterogeneity, choice experiment in which respondents are asked to choose and warns against imposing uniform approaches to (in) their preferred job from two hypothetical job descriptions flexible work arrangements. with randomly assigned attributes, namely, flexibility and This paper is a product of the Social Protection and Jobs Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at yghorpade@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The Valuation of Flexible Work Arrangements: Insights from a Discrete Choice Experiment in Malaysia Yashodhan Ghorpade,1* Alyssa Jasmin* and Amanina Abdur Rahman*2 Keywords: Flexibility, Flexible work arrangements, Gig work, JEL Codes: J30, J32, M52 The rise of flexible work arrangements, spurred globally since the COVID-19 pandemic, is a fundamental shift in the functioning of labor markets. Combined with the rise in non-standard forms of employment, including those in the ‘gig’ economy, different forms of flexibility in work (viz. the location, hours, and the duration of work) simultaneously represent an opportunity to improve work-life balance among workers as well as a threat of either reducing worker productivity or entrenching irregular work practices that can harm workers’ wellbeing. Even before the COVID-19 pandemic, the growing prevalence of alternative work arrangements, in the context of increasing deregulation and liberalization of labor markets (particularly in Europe) spurred critical debates on their implications for worker protection and welfare, legal compliance, as well as for productivity. The evidence on these assertions has also been mixed. For example, Angelici and Profeta (2020), Barerro et al. (2021) and Choudhury et al (2021) found that more flexible work conditions enhance 1 Corresponding author: yghorpade@worldbank.org 2*World Bank. The authors would like to thank En Luqman Ahmed (Government of Malaysia) and Mohd Redzuan Affandi and Muhammad Farhan Hizami (Malaysia Digital Economy Corporation) for their support for this study. The authors are grateful to Yasser el Gammal, Matthew Dornan, Piotr Lewandowski and Matteo Morgandi for helpful comments on earlier versions of this paper. Any remaining errors are our own. productivity at work. On the other hand, Schneider and Harknett (2019) found that unpredictable work schedules were associated with poor sleep, psychological distress and unhappiness among workers, to a greater extent than low wages. The COVID-19 pandemic forced an exogenous adoption of alternate work arrangements, especially working from home rather than an office or dedicated worksite, for many (though not all) types of workers, in most countries (Garrote-Sanchez et al., 2021). This monumental shift demonstrated the possibilities as well as the limitations of certain alternate work arrangements to a larger number of employers and workers than those who had already adopted such arrangements before the pandemic. As several of these arrangements are predicted to continue well beyond the pandemic (Barerra et al, 2021), it is important to assess whether workers themselves value greater flexibility in work arrangements or instead prefer more standard work arrangements, as the exigencies imposed by the pandemic abate. This in turn is critical to inform regulation by the state and management practices by firms relating to working arrangements. This debate is particularly relevant in the context of rising non-standard terms of employment, in particular through the gig economy and digital platform-based work opportunities. The growth of employment in the gig economy, characterized in large part by more autonomy and flexibility, as well as lesser certainty and stability for workers also underlines the need to study differences in the valuation of flexible work arrangements among gig workers, and those in standard employment. While the rise of gig employment in the West has often come at the cost of standard employment (Berger et al, 2019), consequently raising concerns of diminishing worker protection, in many developing countries employment in the gig economy may not necessarily represent an increase in precarious employment, given the preponderance of other forms of informal employment (Ghorpade et al., 2023 for Malaysia, for example). In such settings, the valuation of flexible work arrangements (typically offered to a greater extent in the gig economy compared to in standard employment) may be starkly different from those among workers in more advanced economies. While recent studies have examined the valuation of alternative work arrangements in countries such as the United States (Mas and Pallais, 2020; Chen et al, 2019; Wiswall and Zafar, 2018), United Kingdom (Berger et al, 2019), Italy (Angelici and Profeta, 2020) and Denmark (Eriksson and Kristensen, 2014), and have found mixed evidence, little is known about such valuations in developing countries, and distinctly between gig and standard workers. This paper seeks to measure workers’ valuations of flexible work arrangements in Malaysia, a growing upper-middle income economy that is simultaneously experiencing a sharp increase in gig employment and declining informal employment over time (Ghorpade et al., 2023). We compare such valuations among digital platform workers and workers in standard employment, attempting also to explain the common and distinct correlates of the preference for flexible work arrangements between these two groups. We overcome the endogeneity between work characteristics viz. flexibility and income, typically present in observational data,3 by deploying a novel discrete choice experiment in which respondents are asked to choose one of two hypothetical job descriptions in which terms of flexibility and the associated level of earnings are varied randomly. This paper focuses on three inter-linked research questions: 3 Especially because several work characteristics may be unobservable. 2 • Do workers experience a trade-off between flexible work arrangements (with the options to work from home, or choose one’s work hours) and income? • Does this vary by workers’ current status in employment as freelancing or in standard employment, and by gender? • Which demographic and economic characteristics, as well as workers’ preferred and existing work Conditions, are correlated with the preference for flexible work arrangements? We find considerable variation in the preference for more flexible work arrangements with respect to both the location of work (home or a fixed office) and the choice of hours of work. Overall, gig economy freelancers have a stronger preference for flexible work arrangements compared to those in standard employment, even when this implies lower earnings, suggesting a selection into gig work. Yet, the share of those in standard employment who would prefer more flexible work arrangements in exchange for a part of their income is considerable, suggesting that firms may be able to design compensation mechanisms informed by the mix of wages and flexible work arrangements suitable for workers. We also find that much of the variation in the preference for flexibility is not explained by worker demographics and other observable characteristics but is consistent with measures of workers’ subjective valuation of flexibility and income in their work. We conclude that offering even a modicum of flexibility can enhance workers’ utility without necessarily increasing employers’ costs and provide evidence against imposing uniform approaches to (in)flexible work arrangements. The remainder of this paper is organized as follows: Section 2 describes the data collected for the survey and experiment used in this paper. Section 3 discusses the methodology including the identification strategy and the experimental design. Section 4 presents results including (a) descriptive statistics, (b) T- test results for assessing the randomized allocation of attributes in job descriptions, (c) econometric analysis, and (d) correlates and likely mechanisms of the preference for flexible work arrangements. Section 5 concludes with a discussion of results and their implications for policy. 1. Data Our sample consists of 1,338 respondents of which 1,038 were freelancers and gig economy workers (subsequently referred to collectively as ‘freelancers’), and 300 were workers in standard employment. The data was collected through an online survey administered by a private survey firm, Ipsos. We partnered with the Malaysia Digital Economy Corporation (MDEC) to tap into their network of digital freelancers via their Global Online Work Force (GLOW)4 members (resulting in 738 respondents), as well as an online panel of 300 gig workers and 300 workers in standard employment. The data collection period lasted between February and May 2022. The online survey questionnaire was divided into four parts: (i) respondents’ demographic background, which asks details on age, gender, schooling and marital status; (ii) employment characteristics, which asks details on respondents’ current employment such as their status in employment, occupation, duration of employment, weekly pay, and other occupations if they are working more than one job; (iii) an experimental section with sets of hypothetical job descriptions with randomly assigned attributes for the respondent to choose from; and lastly (iv) preferred work characteristics, which solicits respondents’ 4GLOW is a training program offered by MDEC for individuals interested in becoming a digital freelancer. During the peak of COVID-19, MDEC created a new program called GLOW Penjana, a highly targeted program to help individuals whose livelihoods were affected by the COVID-19 pandemic and movement restrictions. 3 valuations of alternative work characteristics. The online survey took about 15 minutes to complete on average, and could be done on a mobile phone or a laptop/ desktop computer. The full survey questionnaire including the experimental module are appended in Annex 1. 2. Methodology We attempt to understand better the reasons underlying heterogeneity in workers’ preferences for flexibility and draw lessons for management practices and wider employment policies. We deploy a vignette-based discrete choice experiment to ascertain whether workers in non-standard employment are willing to forgo defined percentages of their current incomes to have greater flexibility in the choice of the location of work (home v/s office) and the hours of work (fixed v/s determined by the worker/ employer). We describe this vignette experiment in Figure 1 and Figure 2 below, that describe the characteristics of Job A, and four versions of Job B, each of which vary only in terms of flexibility (of the location of work in Figure 1 and hours of work in Figure 2), shown in row 3, and the associated income from such a job, shown in row 7. While each respondent sees the same description of Job A, s/he is shown one randomly assigned job description from B1-B4, and is asked to choose their preferred job between the two. Figure 1: Range of Hypothetical Jobs offered to respondent to choose from: Location of Work (Job A versus one out of Jobs B1 – B4) Standard Most Flexible More Flexible than standard Job A Job B1 Job B2 Job B3 Job B4 Hours worked 40 hours 40 hours 40 hours 40 hours 40 hours per week Monday to Friday, 9 Monday to Friday, 9 AM Monday to Friday, 9 AM Monday to Friday, 9 AM Monday to Friday, 9 AM Work Hours AM to 5 PM to 5 PM to 5 PM to 5 PM to 5 PM Location of WFH 2d of your choice/ WFH 2d of your choice/ Fixed Office WFH 5d/ week WFH 5d/ week Work week week Regular pension after Regular pension after Regular pension after Regular pension after Regular pension after Retirement retirement (age 60) retirement (age 60) retirement (age 60) retirement (age 60) based retirement (age 60) based Pension based on years of based on years of service based on years of service on years of service in this on years of service in this service in this job in this job in this job job job Included in Health Included in Included in Included in compensation Included in compensation compensation Insurance compensation package compensation package package package package Unemployment No unemployment No unemployment No unemployment No unemployment No unemployment Insurance insurance insurance insurance insurance insurance Monthly Take- Equal to your current 10% less than your 20% less than your 10% less than your 20% less than your Home Salary take-home salary current take-home salary current take-home salary current take-home salary current take-home salary Note: The respondent is shown Job A, and one of Jobs B1, B2, B3 and B4 (randomly assigned), and asked to choose their preferred job between the two. 4 Figure 2: Range of Hypothetical Jobs offered to respondent to choose from: Hours of Work (Job A versus one out of Jobs B1 – B4) Standard More Flexible Less Flexible Job A Job B1 Job B2 Job B3 Job B4 Hours worked 40 hours 40 hours 40 hours 40 hours 40 hours per week 8 hours / day from 8 hours / day from 8 hours / day from 8 hours /day from Monday to Friday, 9 Monday to Friday, client/ Monday to Friday, client/ Work Hours Monday to Friday, you Monday to Friday, you AM to 5 PM employer decides work employer decides work decide the work hours decide the work hours hours hours Location of Fixed Office Fixed Office Fixed Office Fixed Office Fixed Office Work Regular pension after Regular pension after Regular pension after Regular pension after Regular pension after Retirement retirement (age 60) retirement (age 60) retirement (age 60) retirement (age 60) based retirement (age 60) based Pension based on years of based on years of service based on years of service on years of service in this on years of service in this service in this job in this job in this job job job Included in Health Included in Included in Included in compensation Included in compensation compensation Insurance compensation package compensation package package package package Unemployment No unemployment No unemployment No unemployment No unemployment No unemployment Insurance insurance insurance insurance insurance insurance Monthly Take- Equal to your current 10% less than your 20% less than your 10% more than your 20% more than your Home Salary take-home salary current take-home salary current take-home salary current take-home salary current take-home salary Note: The respondent is shown Job A, and one of Jobs B1, B2, B3 and B4 (randomly assigned), and asked to choose their preferred job between the two. Identification We seek to examine the effect of the combination of the attributes of the social insurance package offered, and earnings on the likelihood of selecting the baseline job. Causal identification is determined by the randomized allocation of the description of job B (from B1 to B4) to respondents, i.e. the attributes of the social insurance package and the earnings offered in job B are orthogonal to respondents’ observable and unobservable characteristics. We estimate a Linear Probability Model (LPM) as depicted by the following equation () = + Equation 1 5 Where: Prob (A) is the likelihood of the respondent i preferring Job A, the baseline scenario B is the randomly allocated hypothetical job description j from 1 to 4 to individual i is the error term We further analyze the correlates of the preference for a more flexible job description by estimating a Linear Probability Model with the following specification: () = + 1 + 2 + 3 + 4 + Equation 2 Where: Prob (FLEX) is the likelihood of the respondent i preferring the more flexible job option between the two choices presented, implying also a certain percentage cut in earnings compared to the less flexible option B is again the randomly allocated hypothetical job description j from 1 to 4 to individual i J is the matrix of self-reported characteristics of individual i’s current job P is the matrix of individual i's valuations of alternate job characteristics X is the matrix of individual (demographic and economic) characteristics is the error term 3. Results a. Descriptive Statistics Table 1 below presents the mean and standard deviation of key demographic characteristics of the sample, separately for freelancers and those in standard employment. For most variables we dint observe any stark differences in the two groups, with a few exceptions; freelancers tend to have slightly higher self-assessed risk appetite and are less likely to prefer receiving a delayed but higher payment compared to an immediate but lower one. They are also more likely to be single, and not to have children. Table 1 Descriptive Statistics Freelancers Standard Employment Mean Std Dev Mean Std Dev Age 35.577 10.655 36.183 9.734 Female 0.453 0.498 0.500 0.501 Single = 1 0.487 0.500 0.373 0.484 Has children = 1 0.382 0.486 0.560 0.497 6 Household size 4.122 1.907 4.090 1.698 Risk Appetite (1 - 10) 6.625 2.242 5.753 2.685 Time Preference Index (1-10) 0.499 0.500 0.377 0.485 Education Level Secondary Education 0.180 0.385 0.190 0.393 Postsecondary Education 0.250 0.433 0.213 0.410 Bachelor's degree 0.422 0.494 0.430 0.496 Professional Education 0.047 0.212 0.027 0.161 Master's Education 0.101 0.302 0.140 0.348 Ethnicity Bumiputera 0.750 0.433 0.673 0.470 Chinese 0.167 0.373 0.150 0.358 Indian 0.063 0.242 0.157 0.364 N 1,038 300 We now examine which attributes of work are deemed to be “very important” for freelancers and standard employees, and the extent to which their work adequately provides for these attributes. Figure 3 shows that among the different characteristics of work, the opportunity to earn good money, and to acquire and make use of skills are the most important characteristics valued by freelancers and standard employees alike. There do not appear to be marked differences in the share of freelancers and those in standard employment who deem each of the characteristics of work as ‘very important’. However, as Figure 4 shows, fewer freelancers report having adequate opportunities to earn good money, interact with colleagues, or work in international teams, compared to standard employees. However they do report greater flexibility in their current work arrangements. Figure 3. Work Characteristics Deemed "Very Important" 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Building Earning good Flexibility Interaction Making Use of Opportunities Working in Professional money with my Skills for acquiring International Profile Colleagues new skills Teams Freelancers Std employment 7 Figure 4. Current Job Adequately Provides Opportunities for… 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Building Earning good Flexibility Interaction Making Use of Opportunities Working in Professional money with my Skills for acquiring International Profile Colleagues new skills Teams Freelancers Std employment b. T-tests for Verifying Random Allocation of Attributes in Job Description, the Specific Version of Job B Assigned We now examine the balance in the sample of the four randomized job descriptions for each of our choices of interest, that of the location and the hours of work. Table 2 and Table 3 below show the sub-group means of key demographic and economic characteristics of the sample across the version of Job B assigned, the ‘treatment’. We use an ANOVA to conduct a joint test of significance with the null hypothesis that the sub-group means across treatment arms are not statistically different from each-other, i.e. the job description assigned is not systematically correlated with individual characteristics. We report the results for both, the location of work and the choice of determining work hours, in Table 2 and Table 3, respectively. ̅ B 2 = ̅ B 1 = Table 2: F test of joint significance of treatment arm means (H0: ̅ B 4 ): Location of Work ̅ B 3 = WFH everyday WFH 2 days a week 10% lower 20% lower 10% lower 20% lower earnings earnings earnings earnings Variable Job B1 Job B2 Job B3 Job B4 F-stat P-value Male 0.47 0.56 0.55 0.57 2.42* 0.064 Has Children 0.40 0.40 0.47 0.41 1.61 0.186 Single/ Unmarried 0.48 0.45 0.44 0.48 0.65 0.583 Age 35.22 35.84 35.82 35.96 0.34 0.799 Chinese 0.18 0.15 0.15 0.18 0.68 0.564 Indian 0.06 0.08 0.10 0.10 1.70 0.166 Bumiputera 0.75 0.75 0.73 0.70 0.74 0.527 8 EDU: up to secondary 0.20 0.16 0.22 0.16 2.12* 0.096 EDU: Post- secondary 0.23 0.21 0.25 0.28 1.70 0.165 EDU: Bachelor’s 0.41 0.48 0.40 0.41 1.90 0.128 EDU: Professional 0.05 0.06 0.03 0.03 2.80** 0.039 EDU: Masters+ 0.11 0.10 0.10 0.13 0.92 0.431 Low Income 0.29 0.26 0.26 0.25 0.44 0.722 Med Income 0.30 0.30 0.36 0.31 1.13 0.335 High Income 0.41 0.44 0.39 0.45 1.02 0.384 In Standard Employment 0.21 0.24 0.23 0.23 0.28 0.843 Time Preference Index 0.47 0.49 0.46 0.46 0.30 0.823 Risk Appetite (0- 10) 6.30 6.54 6.47 6.40 0.59 0.621 N 333 336 337 332 Table 3. F-test of joint significance of treatment arm means (H0: B1 = B2 = B3 = B4): Hours of Work Worker Decides Hours Client/ Employer decides hours 10% lower 20% lower 10% higher 20% higher earnings earnings earnings earnings Variable Job B1 Job B2 Job B3 Job B4 F-stat P-value Male 0.54 0.52 0.53 0.55 0.29 0.832 Has Children 0.41 0.41 0.44 0.42 0.26 0.851 Single/ Unmarried 0.44 0.46 0.47 0.48 0.28 0.841 Age 35.67 36.24 35.55 35.39 0.42 0.741 Ethnicity: Chinese 0.15 0.19 0.16 0.15 0.93 0.423 Ethnicity: Indian 0.09 0.10 0.05 0.10 2.54* 0.055 Ethnicity: Bumiputera 0.73 0.70 0.78 0.72 2.06 0.104 EDU: up to secondary 0.15 0.19 0.20 0.20 1.31 0.270 EDU: Post- secondary 0.26 0.22 0.25 0.23 0.64 0.589 EDU: Bachelor’s 0.44 0.43 0.42 0.41 0.26 0.855 EDU: Professional 0.04 0.04 0.04 0.05 0.61 0.610 EDU: Masters+ 0.12 0.11 0.10 0.11 0.36 0.781 Low Income 0.25 0.24 0.27 0.29 0.75 0.524 Med Income 0.32 0.34 0.30 0.31 0.48 0.698 High Income 0.42 0.42 0.43 0.40 0.19 0.902 In Standard Employment 0.22 0.23 0.22 0.22 0.03 0.992 Time Preference Index 0.50 0.49 0.44 0.45 1.25 0.290 Risk Appetite (0- 10) 6.47 6.44 6.47 6.34 0.21 0.889 9 N 334 335 335 334 On the whole, for both sets of experiments, we find that the randomized allocation of hypothetical job descriptions to respondents is balanced, with only very few variables showing a moderate level of statistical significance between group means. Later we show that our results are robust to including these variables as controls (Annexes 1 and 2). c. Econometric Analysis We sequentially examine the results for the choice of the location, and hours of work. First, we regress the dummy variable indicating a preference for working from an office (Job A) on dummies indicating the version of Job B (from B1-B4) presented to the respondent. We then present results for a similar regression for the choice of hours of work. Table 4 Likelihood of Choosing to work from an Office 5 days a week given alternate options for location of work/ earnings (total and by status in employment and gender) TOTAL Freelancing Standard Female Male Employment WFH 5d/week, 10% 0.381*** 0.348*** 0.507*** 0.400*** 0.361*** lower earnings [0.0267] [0.0294] [0.0606] [0.0372] [0.0383] WFH 5d/week, 20% 0.473*** 0.428*** 0.620*** 0.490*** 0.460*** lower earnings [0.0273] [0.0309] [0.0550] [0.0411] [0.0365] WFH every 2d/week, 10% 0.383*** 0.314*** 0.618*** 0.418*** 0.353*** lower earnings [0.0265] [0.0288] [0.0561] [0.0400] [0.0353] WFH 2d/week, 20% 0.464*** 0.426*** 0.592*** 0.441*** 0.481*** lower earnings [0.0274] [0.0310] [0.0568] [0.0417] [0.0364] Mean 0.425 0.379 0.587 0.435 0.416 N 1338 1038 300 620 718 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 Dep Var = 1 if Job A preferred to Job B, else 0. Job A implies working from an office 5 days/week and earnings equal to respondents’ current income We see that overall, about 40 percent of respondents choose to work from an office, while around 60 percent choose to work from home. Those in standard employment are much more likely to choose office, while a majority of freelancers prefer to work from home for two or five days a week. Yet, it is striking that within each of these employment types, there is considerable heterogeneity in the preference for the location of work; over 40 percent of those in standard employment would prefer to work from home for at least 2 days a week even when that implies a 10 to 20 percent wage cut. Similarly, just under 40 percent of freelancers would prefer to work from an office for five days a week if that is accompanied by higher earnings. Among both freelancers and standard workers, the share of those opting to work from an office for five days a week does not seem to vary by the two proposed options of days worked from home; whether two 10 days a week, or all five. However, respondents, especially freelancers, are sensitive to the earnings reduction associated with the option to work from home; we observe a 10 percentage points increase in the preference to work from an office five days a week when respondents face a 20 rather than 10 percent decrease in earnings for such flexibility. Among freelancers, 10% more are willing to work from an office five days a week when presented with a 20% rather than 10% wage cut. The preference is similar among those in standard employment when they are faced with a 20% rather than 10% wage cut for working from home 5 days a week. However, if offered the chance to work from home for 2 days a week, roughly the same share of standard workers (~60%) choose to work from an office regardless of the level of foregone wages this implies (between 10% and 20%). We find no significant differences by gender, which could potentially indicate selection such that only those women who can be more flexible in the choice of location of work are currently employed. Table 5 below examines the preference for working a fixed work schedule (from 9 AM to 5 PM from Monday to Friday) when offered alternate options of the workhours chosen5 either by the worker (Jobs B1 and B2) or by their client/ employer (B3 and B4). Options B1 and B2 imply greater autonomy for a worker, and therefore include a 10 and 20 percent decrease in earnings, respectively. As options B3 and B4 involve a loss of autonomy for the worker, they include a 10 and 20 percent increase in earnings. Table 5 Likelihood of Choosing Job A given alternate options for hours of work/ earnings (total and by status in employment and gender) Total Freelancing Standard Female Male Employment B1: Worker decides hours, 10% 0.541*** 0.511*** 0.652*** 0.566*** 0.513*** lower earnings [0.0274] [0.0308] [0.0577] [0.0376] [0.0399] B2: Worker chooses hours, 0.497*** 0.479*** 0.557*** 0.490*** 0.503*** 20% lower earnings [0.0273] [0.0312] [0.0563] [0.0411] [0.0367] B3: Client/ employer chooses 0.472*** 0.414*** 0.671*** 0.490*** 0.457*** hours, 10% higher earnings [0.0272] [0.0305] [0.0543] [0.0405] [0.0368] B4: Client/ employer chooses 0.533*** 0.480*** 0.711*** 0.545*** 0.524*** hours, 20% higher earnings [0.0274] [0.0313] [0.0524] [0.0418] [0.0364] Mean 0.510 0.471 0.647 0.524 0.499 N 1338 1038 300 620 718 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 Dep Var = 1 if Job A preferred to Job B, else 0. Job A implies a fixed schedule of 9 AM to 5 PM 5 days/week and earnings equal to respondents’ current income We find a higher preference for fixed work hours (9 AM – 5 PM) among those in standard employment, compared to freelancers when both groups are offered the possibility of choosing their own work hours but forgoing 10%-20% of their earnings. However, a majority of those in standard employment (~70%) still tend to prefer a regular routine compared to a scenario where the employer/ client fully determines work 5 For the same total number of hours worked in a day (8) and week (40). 11 hours even if that implies a 10%-20% increase in earnings. In contrast only 40%-50% of freelancers prefer a fixed work hours schedule (9-5) to one determined entirely by their employers/ clients if that is accompanied by an increase in earnings by 10%-20%. This suggests that those in standard employment have a higher aversion to being dictated work hours by their employers. Further, given the considerable share of both freelancers and standard workers who opt to work fixed work hours, it appears that for many workers, the regularity of work hours may sometimes matter more than autonomy. We see substantial heterogeneity in the preferences of the hours of work. Not all workers values flexibility entirely, as a majority of standard employees and a substantial share of freelancers prefers a set schedule compared to alternatives that imply full control of workhours6 either by themselves, or by their employer / client. Regularity is therefore valued by large shares of both groups of workers over both, more (lesser) own (client/employer) control over work hours decisions. d. Understanding the Preference for Flexible Work Arrangements: Correlates and Mechanisms Underlying Respondents’ Choices Location of Work Is the preference for flexibility associated with the attributes of a job that workers themselves deem to be very important to them, and whether their current job has these attributes? Among the various offerings of a job, we assess the opportunity offered to earn good money, make good use of the workers’ skills, acquire new skills, flexibility, interaction with colleagues, building a professional profile/ looking good on a CV, and the opportunity to work in international teams. We assess the associations of these factors – both in terms of whether they are desired strongly by a worker, and whether their current job provides these adequately, along with other characteristics of the workers themselves, with the likelihood of preferring more flexible work arrangements, sequentially for the choice of the location and the hours of work. Table 6 Likelihood of Choosing to work from an Office 5 days a week (total and by status in employment and gender) (1) (2) (3) (4) (5) Total Freelancing Standard Female Male Employment Standard Employment 0.115*** - - 0.115* 0.117** [0.040] [0.060] [0.054] Age -0.003 -0.006 0.029* 0.013 -0.014 [0.007] [0.008] [0.017] [0.008] [0.010] Age-squared 0.000 0.000 -0.000* -0.000** 0.000 [0.000] [0.000] [0.000] [0.000] [0.000] Female 0.025 0.048 -0.016 - - [0.037] [0.041] [0.091] Single 0.091** 0.049 0.271*** 0.095* 0.083 [0.039] [0.045] [0.089] [0.057] [0.055] 6 Excluding late night shifts; 8-hour workday between 9 AM and 10 PM. 12 Has children 0.064 0.047 0.133 0.052 0.058 [0.046] [0.053] [0.097] [0.055] [0.053] Female*Has children -0.017 -0.072 0.096 - - [0.054] [0.061] [0.121] HH Size 0.001 0.001 -0.006 -0.010 0.011 [0.007] [0.008] [0.018] [0.011] [0.011] GLOW Member -0.142*** -0.140*** - -0.188*** -0.094** [0.035] [0.036] [0.053] [0.048] Risk Appetite (0-10) -0.000 0.002 -0.006 0.001 0.001 [0.006] [0.007] [0.012] [0.009] [0.008] Time Preference Index -0.015 -0.004 -0.075 -0.005 -0.035 [0.027] [0.030] [0.060] [0.040] [0.037] Deemed as ‘Very Important’ in a job: Earning Good Money 0.056* 0.067* 0.041 0.015 0.105** [0.032] [0.036] [0.071] [0.048] [0.044] Flexibility -0.131*** -0.145*** -0.037 -0.090* -0.165*** [0.031] [0.034] [0.076] [0.047] [0.042] Interaction with colleagues 0.012 0.020 -0.009 -0.036 0.061 [0.034] [0.039] [0.075] [0.052] [0.047] Making Use of my Skills -0.032 -0.005 -0.113 0.050 -0.109** [0.036] [0.041] [0.084] [0.055] [0.050] Opportunities for acquiring new skills 0.085** 0.079* 0.080 0.075 0.086* [0.038] [0.042] [0.084] [0.060] [0.050] Building professional profile 0.005 -0.018 0.062 -0.013 0.024 [0.034] [0.039] [0.077] [0.054] [0.046] Working in International Teams 0.003 0.034 -0.118 0.038 -0.028 [0.036] [0.041] [0.078] [0.056] [0.048] Current Job Adequately Provides Opportunities for … Earning Good Money -0.050 -0.073 -0.003 -0.035 -0.050 [0.039] [0.047] [0.077] [0.058] [0.056] Flexibility -0.077** -0.104*** 0.090 -0.061 -0.086** [0.031] [0.034] [0.078] [0.047] [0.042] Making Good Use of my Skills 0.025 0.032 -0.027 -0.039 0.071 [0.040] [0.047] [0.080] [0.059] [0.056] Acquiring New Skills 0.044 0.014 0.138* 0.058 0.034 [0.039] [0.044] [0.081] [0.059] [0.053] Building my Professional Profile 0.055 0.061 -0.000 0.104* 0.021 [0.041] [0.047] [0.084] [0.061] [0.058] Working in international teams 0.030 0.063 -0.051 0.032 0.024 [0.041] [0.046] [0.087] [0.063] [0.055] 13 Interacting with colleagues -0.030 -0.025 -0.022 -0.078 -0.000 [0.036] [0.041] [0.071] [0.054] [0.048] Treatment Arms /Description of Job B (ref: Job B1 WFH 5d/week, 10% earnings cut) B2: WFH 5d/week, 20% earnings cut 0.087** 0.081* 0.057 0.082 0.101** [0.037] [0.041] [0.085] [0.055] [0.051] B3: WFH 2d/week, 10% earnings cut -0.002 -0.032 0.088 -0.010 0.007 [0.037] [0.042] [0.084] [0.055] [0.051] B4: WFH 2d/week, 20% earnings cut 0.085** 0.089** 0.068 0.046 0.129** [0.038] [0.042] [0.090] [0.056] [0.053] Constant 0.439*** 0.516*** 0.026 0.366* 0.607** [0.167] [0.182] [0.404] [0.188] [0.236] Education Controls Yes Yes Yes Yes Yes Ethnicity Controls Yes Yes Yes Yes Yes Income Level Controls Yes Yes Yes Yes Yes Mean 0.425 0.379 0.587 0.435 0.416 N 1338 1038 300 620 718 R-squared 0.089 0.078 0.142 0.110 0.108 As we see in Table 6 above, overall, there is considerable heterogeneity in preferences for flexibility. Even after controlling for a range of individual demographic, educational, and economic characteristics, preferences of job characteristics, and attributes of their current jobs, the unexplained variation in the preference for working from home is very high as the value of the R-squared of the estimation is only ~ 9 percent. However, the share of explained variation for freelancers is about half that of those in standard employment, further underlining that this is an extremely heterogeneous group with no single or dominant preference for the location of work. Overall, workers in standard employment have a lower willingness to pay for the option to work from home. Single persons, especially those in standard employment, are more likely to prefer working from an office. Surprisingly, other demographic characteristics do not appear to be associated with a preference to work from home. The association of flexibility with (i) the desired attributes of a job, and (ii) the extent to which workers’ current jobs provide them adequately shows some interesting results. Those who state that they value flexibility at work a lot, and also those who assess the level of flexibility enjoyed by them in their current work as adequate, especially among freelancers, are also more likely to prefer not to work from an office. However, those who attach him important to earning well, and to acquire new skills (the latter especially among men), prefer to work from an office. These results suggest a clear tension between attaching importance to earning money and professional development on the one hand, and flexibility with respect to the choice of working from home. Finally, we see that those in treatment arms which imply a higher wage cut (20%) for being able to work from home are 10% more likely to prefer working from an office, compared to the comparison category (presented with the option to work from home every day for 10% lower earnings). Interestingly, and somewhat in contrast to other studies on the preference for the location of work (Lewandowski et al. 2022, Aksoy et al., 2022), we do not see a difference in the preference for working from home between full (five days per week) and ‘hybrid’ options (2 days per week) at the same level of earnings reduction (i.e. 14 between treatment arms offering jobs B1 and B3 or B2 and B4). While this is indeed somewhat surprising given the salience of hybrid work in the post-pandemic milieu globally, note that no respondent is offered a choice between fully home-based and partially homebased work (i.e., jobs B1/2 and B3/4) for us to directly measure preferences between these two arrangements. Instead, we compare the preferences for each of these arrangements with respect to job A which offers no option to work from home. With a relatively small sample, we may not be able to rely on transitivity in preferences among different groups of respondents, that is, inferring the preference (if any) of B1/2 over B3/4, or vice versa, by comparing the preferences directly between B1/2 and A and B3/4 and A, respectively. Work Hours We now examine the preference for flexibility in work hours. As depicted in Table 7, job descriptions B1 and B2 offer the worker greater flexibility in the choice of work hours compared to job description A, while B3 and B4 offer lesser flexibility (as these are determined entirely by the employer / client). Therefore, when the respondent is assigned to treatment groups indicated by B1 or B2, job B is deemed as the more flexible option, whereas if s/he belongs to groups B3 or B4, the choice of job A is considered more flexible. In other words, in terms of flexibility: B3/ B4(Client / employer sets work hours) > A(Fixed schedule) > B1/ B2(Workers chooses hours) Table 7 Likelihood of Choosing More Flexible Work Hours (total and by status in employment and gender) (1) (2) (3) (4) (5) Total Freelancing Standard Female Male Employment Standard Employment -0.044 -- -- -0.200*** 0.095* [0.040] [0.058] [0.057] Age -0.004 -0.003 0.004 -0.006 -0.006 [0.007] [0.008] [0.018] [0.010] [0.011] Age-squared 0.000 0.000 -0.000 0.000 0.000 [0.000] [0.000] [0.000] [0.000] [0.000] Female 0.000 0.031 -0.090 - - [0.037] [0.041] [0.091] Single 0.008 0.012 -0.013 0.046 -0.018 [0.041] [0.047] [0.088] [0.061] [0.058] Has children 0.012 0.026 -0.131 0.059 -0.018 [0.048] [0.055] [0.100] [0.057] [0.056] Female*Has children 0.019 0.064 0.012 - - [0.056] [0.064] [0.121] HH Size -0.001 -0.003 0.019 -0.009 0.006 [0.008] [0.009] [0.016] [0.011] [0.011] GLOW Member -0.017 -0.003 - -0.120** 0.058 [0.036] [0.037] [0.055] [0.049] Risk Appetite (0-10) -0.009 -0.009 -0.007 0.001 -0.016* [0.006] [0.007] [0.011] [0.009] [0.009] Time Preference Index -0.020 -0.030 0.014 -0.003 -0.029 [0.027] [0.031] [0.060] [0.041] [0.038] Deemed as Very Important’ feature of job: 15 Earning Good Money -0.129*** -0.124*** -0.135** -0.112** -0.151*** [0.033] [0.037] [0.068] [0.049] [0.046] Flexibility 0.116*** 0.118*** 0.084 0.092** 0.125*** [0.031] [0.035] [0.077] [0.046] [0.043] Interaction with colleagues 0.028 0.036 0.015 0.006 0.057 [0.035] [0.040] [0.074] [0.051] [0.049] Making Use of my Skills 0.012 -0.009 0.058 -0.025 0.060 [0.037] [0.041] [0.084] [0.054] [0.052] Opportunities for acquiring new skills -0.035 -0.035 -0.045 0.005 -0.075 [0.039] [0.044] [0.085] [0.061] [0.053] Building professional profile -0.046 -0.049 -0.024 -0.007 -0.080* [0.036] [0.041] [0.080] [0.056] [0.048] Working in International Teams 0.032 0.016 0.147** 0.030 0.033 [0.036] [0.042] [0.072] [0.052] [0.051] Current Job Adequately Provides Opportunities for … Earning Good Money 0.071* 0.062 0.078 0.140** -0.002 [0.041] [0.048] [0.073] [0.060] [0.058] Flexibility 0.022 0.017 0.021 0.044 0.014 [0.032] [0.036] [0.083] [0.047] [0.045] Making Good Use of my Skills -0.047 -0.027 -0.096 -0.047 -0.038 [0.040] [0.046] [0.076] [0.057] [0.057] Acquiring New Skills 0.089** 0.113** -0.023 0.077 0.105* [0.040] [0.045] [0.083] [0.059] [0.055] Building my Professional Profile -0.080** -0.094** -0.013 -0.088 -0.067 [0.040] [0.046] [0.086] [0.055] [0.059] Working in international teams -0.032 -0.043 -0.070 -0.059 -0.041 [0.042] [0.047] [0.088] [0.060] [0.057] Interacting with colleagues 0.019 0.007 0.100 -0.023 0.064 [0.036] [0.043] [0.069] [0.054] [0.050] Treatment Arms description of Job B (ref: Job B1 implies worker decides work hours, 10% earnings cut) B2: Worker decides hours, 20% earnings 0.008 0.030 -0.050 0.046 -0.023 cut [0.038] [0.043] [0.081] [0.055] [0.053] B3: Client/ Employer decides hours, 10% 0.076** 0.006 0.335*** 0.130** 0.017 earnings increase [0.038] [0.043] [0.080] [0.057] [0.053] B4: Client/ Employer decides hours, 20% -0.036 -0.096** 0.199** -0.009 -0.061 earnings increase [0.038] [0.044] [0.083] [0.058] [0.052] Constant 0.636*** 0.587*** 0.346 0.575** 0.690*** [0.167] [0.183] [0.425] [0.224] [0.242] Education Controls Yes Yes Yes Yes Yes Ethnicity Controls Yes Yes Yes Yes Yes Income Level Controls Yes Yes Yes Yes Yes Mean 0.462 0.466 0.447 0.453 0.469 N 1338 1038 300 620 718 R-squared 0.056 0.073 0.186 0.094 0.070 16 Flexible work described as the choice of job B when presented B1 or B2 (implying worker chooses work hours) and the choice of Job A when presented B3 or B4 (implying client employer choose workhours); job A has a fixed work schedule of 9 AM to 5 PM Monday – Friday. As with the analysis of the correlates of the preference for working from home, here too we find that there is considerable heterogeneity in the preference for flexible work hours, as the R-squared for the estimation is less than 6 percent for the combined sample. Again, we find that the share of explained variation for freelancers is less than half that of those in standard employment, further underlining that this is an extremely heterogeneous group with no single type of preference for the hours of work. Demographic characteristics also do not appear to influence the preference for the location of work. After controlling for all characteristics, among women those in standard employment have a lower preference for flexible work hours, reflecting potentially the selection of women into standard employment in Malaysia where women’s labor force participation remains relatively low compared to other Southeast Asian economies (Schmillen et al, 2019). Among all groups, respondents who attach high importance to flexibility are much more likely to choose more flexible work hours, and those who attach high importance to earning well are less likely to do so, which suggests an internally consistent logic underlining trade-offs between income and flexibility. Those whose current work offers opportunities for acquiring new skills, especially among freelancers, value flexible working hours more. However, those whose current work offers opportunities for building a professional profile, or makes their CV look good, are more likely to prefer less flexible work hours that are either set by the employer/ client or are routine 9-5 arrangements. This offers some evidence that work that is viewed as more prestigious may entice workers to sacrifice some flexibility in favor of professional development. Interestingly, other characteristics of respondents’ current jobs, or attributes that they may value such as close interaction with colleagues or working in international teams does not seem to affect the preference for the hours of work. 4. Discussion and Conclusions This paper has attempted to understand the valuations of and trade-offs between flexibility and income among workers in standard employment and digital freelancers, in the context of Malaysia, a fast growing East Asian economy that is experiencing a sharp rise of the gig economy and digital freelancing. Analyzing the results of a unique survey experiment in which respondents are asked to choose between jobs characterized by standard working conditions, and more flexible alternatives, we draw important lessons for management practices and employment policies globally. First, we find considerable heterogeneity in preferences for flexibility with respect to the location and hours of work among freelancers as well as standard employees. While most freelancers tend to prefer working from home for two to five days per week, a sizable share (around 40 percent) of those in standard employment would also like not to have to work from an office all five days of the week, even if that came at a cost of 10 to 20 percent of their income. Equally, around 40 percent of current freelancers would like to work from a fixed office if that translates into higher earnings. 17 We find no clear or overwhelming preference for flexibility among our combined sample, as well as for subgroups of freelancers / standard workers, or men / women. Furthermore, demographic and economic characteristics explain only a very small share (6%-9%) of the total variation in the preference for flexible work arrangements, and more so for freelancers (7%-8%) compared to those in standard employment (14%- 19%). The relatively low value of the R-squared across specifications is similar to that found by Aksoy et al. (2022) who find a value of no more than 12 percent in the estimation of the preference for working from home, after controlling for a range of characteristic in their study across 27 countries. The considerable unobserved variation in the preference for flexible work arrangements therefore means that no single policy directive or framework can sufficiently accommodate some workers’ preferences without causing a loss of utility for others. Our findings indicate a trade-off between flexibility and income-related considerations for workers. We find strong evidence that the revealed preference for working from home, or for choosing more flexible work hours, is correlated positively with a higher valuation of flexibility by workers, and negatively with the importance attached to earning well. This means that firms may be able to offer greater flexibility to some workers while reducing their wage bill,7 without a commensurate decrease in workers’ utility. The fact that unlike other studies in this area (Lewandowski et al., 2022; Aksoy et al., 2022) we find no significant difference in the preference for working from an office regardless of whether workers are allowed to work two or all five days from home could suggest that offering even a moderate degree of flexibility can be utility-enhancing for workers and could be a way to balance workers’ preferences for flexibility with employers’ needs for observing and supervising staff, facilitating in-person interaction and learning, and fostering closer teamwork. While firms and workers may be best placed to undertake such negotiations directly, based on their specific requirements and preferences viz. work hours and location, the regulatory environment should in turn allow for such flexible arrangements and not mandate all-or- nothing arrangements for the choice of work hours or location. The definition of a ‘place of work’ in labor codes across countries may, for example, be limiting for allowing such hybrid work arrangements. Interestingly, we also find that most standard workers and a sizable share of freelancers prefer a fixed work schedule of working 9 AM to 5 PM on all five weekdays, over alternatives where they themselves or their employers/ clients set work hours. This indicates that regularity of work hours (represented by conventional working hours in standard employment) may be highly valued by large numbers of both groups of workers and that emerging forms of employment may only partially change the established norms of preferred work hours. We also find some evidence of a perceived trade-off between flexible work hours and professional development, especially among freelancers. Workers who are in jobs that they view as helping build their professional profile (looking ‘good’ on their CVs), are more hesitant to opt for flexible hours of work. This may potentially be because employers/ clients who offer such work may be (perceived as) less encouraging of greater flexibility of work hours, or because of workers’ own perceptions of flexible work hours as something that may compromise their standing in the eyes of their clients/ employers. Future research may be able to explore this in greater detail. While our study focuses on uncovering workers’ preferences with respect to flexibility, we do not observe employers’ and clients’ preferences for workers’ flexibility. This was particularly challenging to attempt in 7 And potentially office space-related costs. 18 the case of gig workers who typically may not have a single employer or client, and hence our focus on workers’ preferences alone. However, employers’ or clients’ preferences are also critical to understand the likely equilibrium levels of workers’ flexibility and emoluments. This is also important because employers’ and clients’ preferences for workers’ flexibility will be shaped by their assessments and/ or perceptions of worker productivity under alternate modes of flexibility. Assessments of the productivity implications of flexible work arrangements have received renewed attention in recent months, especially as the effects of the pandemic and associated containment measures have abated across several countries. Aksoy et al. (2023) report that, across countries, workers had a favorable view of their own productivity while working from home during the pandemic. Most studies on more objective measures of productivity, however, indicate lower productivity. Emmanuel and Harrington (2023) find declines in worker productivity and quality in a call center operated by a fortune 500 company in the US, attributable to remote work. These also resulted in lower promotion rates of remote workers. Gibbs et al. (2023) find that the switch to home-based work in an Indian IT company during the pandemic simultaneously resulted in longer hours of work and lower output, leading to a decline in productivity between 8 and 19 percent. Similarly, Atkin et al. (2023) find in their study in India that data entry operators randomly assigned to work from home were 18 percent less productive than those working from the office, and that those who prefer working from home were in fact less productive at home than at the office. These studies indicate why employers may be less willing than workers to value flexible work arrangements, especially related to the location of work. Few studies have been able to assess workers’ and employers’ preferences for flexible work arrangements jointly. Lewandowski et al. (2022) find that there is a considerable mismatch between what share of earnings workers in Poland are willing to forgo for the flexibility to work from home, and what their employers would like to reduce their earnings by to compensate for the perceived productivity losses associated with offering workers such options. Aksoy et al. (2022) also find a gap between the number of days of home-based work preferred by employers and employees across countries, with employees consistently preferring more days of home-based work. Our paper speaks to these results by underlining the potential trade-offs that especially gig workers are willing to make between flexibility and earnings, and providing evidence that reduced earnings and greater flexibility may be compatible for workers and employers/ clients; though to what extent would depend also on the specific considerations of worker productivity in a given sector, industry or firm. As flexible work arrangements become more commonplace in a post-pandemic milieu in Malaysia and beyond, employers and clients may potentially gain by (re)assessing the extent to which greater flexibility may affect productivity. Where productivity is harmed by flexible work arrangements, there is scope for workers and employers or clients to negotiate to arrive at equilibrium levels of flexibility and productivity. Where employers and clients find little evidence of reduced productivity, signaling greater approval of flexible arrangements can be utility-enhancing for clients, employers and workers alike. While measuring productivity in some forms of gig work can be complicated because of its heterogeneous nature, future research should also attempt to study the productivity effects of flexible work arrangements among gig workers in developing countries. A deeper understanding of the valuations of flexibility and income as has been attempted in this paper can inform the right mix of management practices for firms and employers, and policies and regulations for governing institutions to better harness the opportunities offered by the changing nature of work. 19 References Aksoy, C. G., J. Maria Barrero, N. Bloom, S. J. Davis, M. Dolls and P. Zarate (2022) "Working from Home Around the World," EconPol Forum, CESifo, vol. 23(06), pages 38-41, October. Angelici, Marta and Profeta, Paola (2020) Smart-Working: Work Flexibility Without Constraints. CESifo Working Paper No. 8165, Available at SSRN: https://ssrn.com/abstract=3556304 or http://dx.doi.org/10.2139/ssrn.3556304 Atkin, David, A. Schoar, and S. Shinde (2023) Working from Home, Worker Sorting and Development NBER Working Paper No. 31515 July 2023. https://www.nber.org/system/files/working_papers/w31515/w31515.pdf Berger, Thor, Carl Benedikt Frey, Guy Levin, and Santosh Rao Danda (2019) Uber happy? Work and well- being in the ‘Gig Economy’, Economic Policy, Volume 34, Issue 99, July 2019, Pages 429–477 https://doi.org/10.1093/epolic/eiz007 Bloom, Nicholas, Ruobing Han, and James Liang (2022) How Hybrid Working From Home Works Out? NBER Working Paper No. 30292 July 2022. DOI 10.3386/w30292 Choudhury, P(Raj), C. Foroughi, B. Larson (2012) Work-from-anywhere: the productivity effects of geographic flexibility. Strategic Management Journal, 42 (4) (2021), pp. 655-683, 10.1002/smj.3251 Emanuel, N. and E. Harrington (2023) Working Remotely? Selection, Treatment, and the Market for Remote Work. Federal Reserve Bank of New York Staff Reports, No. 1061 May 2023. https://www.newyorkfed.org/medialibrary/media/research/staff_reports/sr1061.pdf?sc_lang=en Eriksson, Tor, and Nicolai Kristensen (2014) “Wages or Fringes? Some Evidence on Trade-Offs and Sorting.” Journal of Labor Economics 32, no. 4: 899–928. https://doi.org/10.1086/676662 . Ghorpade, Y., A. Abdur Rahman, A. Jasmin, S. Yi (2023). Informal Employment in Malaysia. World Bank: forthcoming. Gibbs, M., Mengel, F., & Siemroth, C. (2023). Work from home and productivity: Evidence from personnel and analytics data on information technology professionals. Journal of Political Economy Microeconomics, 1(1), 7-41. Keith Chen, M. Keith, Peter E. Rossi, Judith A. Chevalier, and Emily Oehlsen (2019) The Value of Flexible Work: Evidence from Uber Drivers. Journal of Political Economy, 2019, vol. 127, no. 6 , 2735-2794 Garrote, Sanchez Daniel, Gomez, Parra Nicolas, Ozden, Caglar, Rijkers, Bob, Viollaz, Mariana, and Winkler, Hernan. (2021) Who on Earth Can Work from Home? World Bank Research Observer 2021 36:1, 67-100. Lewandowski, P.,Lipowska, K.,Smoter, M. (2022). Mismatch in preferences for working from home – evidence from discrete choice experiments with workers and employers : Evidence from Poland, IBS Working Paper WP 05/2022, https://ibs.org.pl/en/publications/mismatch-in-preferences-for-working- from-home-evidence-from-discrete-choice-experiments/ 20 Mas A, and A. Pallais (2020) Alternative Work Arrangements. Annual Review of Economics. 2020; 12 (1): 631-658. Schmillen,Achim Daniel; Tan,Mei Ling; Abdur Rahman,Amanina Binti; Lnu,Shahrul Natasha Binti Halid; Weimann Sandig,Nina (2019) Breaking Barriers : Toward Better Economic Opportunities for Women in Malaysia. The Malaysia Development Experience Series Washington, D.C. : World Bank Group. http://documents.worldbank.org/curated/en/734461569247873555/Breaking-Barriers-Toward- Better-Economic-Opportunities-for-Women-in-Malaysia Schneider, D., & Harknett, K. (2019). Consequences of Routine Work-Schedule Instability for Worker Health and Well-Being. American Sociological Review, 84(1), 82 - 114. https://doi.org/10.1177/0003122418823184 Wiswall, M. and B. Zafar (2018) Preference for the Workplace, Investment in Human Capital, and Gender, The Quarterly Journal of Economics, Volume 133, Issue 1, February 2018, Pages 457– 507, https://doi.org/10.1093/qje/qjx035 21 Annex 1. Full regression: Correlates of choice of Working from Home (v/s working from a fixed office for 5days/ week) (1) (2) (3) (4) (5) Total Freelancing Standard Female Male Employment Standard Employment 0.115*** - - 0.115* 0.117** [0.040] [0.060] [0.054] Age -0.003 -0.006 0.029* 0.013 -0.014 [0.007] [0.008] [0.017] [0.008] [0.010] Age-squared 0.000 0.000 -0.000* -0.000** 0.000 [0.000] [0.000] [0.000] [0.000] [0.000] Female 0.025 0.048 -0.016 - - [0.037] [0.041] [0.091] Single 0.091** 0.049 0.271*** 0.095* 0.083 [0.039] [0.045] [0.089] [0.057] [0.055] Has children 0.064 0.047 0.133 0.052 0.058 [0.046] [0.053] [0.097] [0.055] [0.053] Female*Has children -0.017 -0.072 0.096 - - [0.054] [0.061] [0.121] HH Size 0.001 0.001 -0.006 -0.010 0.011 [0.007] [0.008] [0.018] [0.011] [0.011] GLOW Member -0.142*** -0.140*** - -0.188*** -0.094** [0.035] [0.036] [0.053] [0.048] Risk Appetite (0-10) -0.000 0.002 -0.006 0.001 0.001 [0.006] [0.007] [0.012] [0.009] [0.008] Time Preference Index -0.015 -0.004 -0.075 -0.005 -0.035 [0.027] [0.030] [0.060] [0.040] [0.037] Edu: Up to Secondary 0.041 0.057 0.026 0.000 -0.021 [0.071] [0.064] [0.113] [.] [0.093] Edu: Postsecondary 0.001 0.004 0.053 -0.060 -0.058 [0.068] [0.057] [0.108] [0.071] [0.091] Edu: Bachelors -0.012 -0.012 0.064 -0.067 -0.082 [0.066] [0.053] [0.099] [0.065] [0.090] Edu: Professional 0.000 -0.032 0.297* -0.189 0.000 [.] [0.083] [0.175] [0.119] [.] Edu: Masters -0.010 0.000 0.000 -0.073 -0.069 [0.075] [.] [.] [0.080] [0.107] Chinese -0.027 0.012 -0.163* -0.049 -0.010 [0.039] [0.044] [0.088] [0.058] [0.056] Indian -0.012 0.011 -0.036 0.059 -0.072 22 [0.050] [0.065] [0.083] [0.071] [0.072] Med Income -0.016 -0.018 -0.192 -0.027 -0.015 [0.036] [0.037] [0.131] [0.050] [0.052] High Income 0.002 0.030 -0.249* -0.037 0.028 [0.037] [0.039] [0.127] [0.056] [0.052] Deemed as Very Important’ feature of job: Earning Good Money 0.056* 0.067* 0.041 0.015 0.105** [0.032] [0.036] [0.071] [0.048] [0.044] Flexibility -0.131*** -0.145*** -0.037 -0.090* -0.165*** [0.031] [0.034] [0.076] [0.047] [0.042] Interaction with colleagues 0.012 0.020 -0.009 -0.036 0.061 [0.034] [0.039] [0.075] [0.052] [0.047] Making Use of my Skills -0.032 -0.005 -0.113 0.050 -0.109** [0.036] [0.041] [0.084] [0.055] [0.050] Opportunities for acquiring new skills 0.085** 0.079* 0.080 0.075 0.086* [0.038] [0.042] [0.084] [0.060] [0.050] Building professional profile 0.005 -0.018 0.062 -0.013 0.024 [0.034] [0.039] [0.077] [0.054] [0.046] Working in International Teams 0.003 0.034 -0.118 0.038 -0.028 [0.036] [0.041] [0.078] [0.056] [0.048] Current Job Adequately Provides Opportunities for … Earning Good Money -0.050 -0.073 -0.003 -0.035 -0.050 [0.039] [0.047] [0.077] [0.058] [0.056] Flexibility -0.077** -0.104*** 0.090 -0.061 -0.086** [0.031] [0.034] [0.078] [0.047] [0.042] Making Good Use of my Skills 0.025 0.032 -0.027 -0.039 0.071 [0.040] [0.047] [0.080] [0.059] [0.056] Acquiring New Skills 0.044 0.014 0.138* 0.058 0.034 [0.039] [0.044] [0.081] [0.059] [0.053] Building my Professional Profile 0.055 0.061 -0.000 0.104* 0.021 [0.041] [0.047] [0.084] [0.061] [0.058] Working in international teams 0.030 0.063 -0.051 0.032 0.024 [0.041] [0.046] [0.087] [0.063] [0.055] Interacting with colleagues -0.030 -0.025 -0.022 -0.078 -0.000 [0.036] [0.041] [0.071] [0.054] [0.048] Treatment Arms: WFH 5d/week, 20% earnings cut 0.087** 0.081* 0.057 0.082 0.101** [0.037] [0.041] [0.085] [0.055] [0.051] WFH 2d/week, 10% earnings cut -0.002 -0.032 0.088 -0.010 0.007 [0.037] [0.042] [0.084] [0.055] [0.051] 23 WFH 2d/week, 20% earnings cut 0.085** 0.089** 0.068 0.046 0.129** [0.038] [0.042] [0.090] [0.056] [0.053] Constant 0.439*** 0.516*** 0.026 0.366* 0.607** [0.167] [0.182] [0.404] [0.188] [0.236] Mean 0.425 0.379 0.587 0.435 0.416 N 1338 1038 300 620 718 R-squared 0.089 0.078 0.142 0.110 0.108 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 Heteroskedasticity-robust standard errors Dep var = choice of Job allowing working from home (choosing i.e., Job B1/ B2 or B3/B4 allowing 5 or 2 days of working from home per week, respectively, compared to Job A implying working from a fixed office 5d/ week) 24 Annex 2: Full regression: Correlates of choosing More Flexible Hours of Work (1) (2) (3) (4) (5) Total Freelancing Standard Female Male Employment Standard Employment -0.044 - - -0.200*** 0.095* [0.040] [0.058] [0.057] Age -0.004 -0.003 0.004 -0.006 -0.006 [0.007] [0.008] [0.018] [0.010] [0.011] Age-squared 0.000 0.000 -0.000 0.000 0.000 [0.000] [0.000] [0.000] [0.000] [0.000] Female 0.000 0.031 -0.090 - - [0.037] [0.041] [0.091] Single 0.008 0.012 -0.013 0.046 -0.018 [0.041] [0.047] [0.088] [0.061] [0.058] Has children 0.012 0.026 -0.131 0.059 -0.018 [0.048] [0.055] [0.100] [0.057] [0.056] Female*Has children 0.019 0.064 0.012 0.000 0.000 [0.056] [0.064] [0.121] [.] [.] HH Size -0.001 -0.003 0.019 -0.009 0.006 [0.008] [0.009] [0.016] [0.011] [0.011] GLOW Member -0.017 -0.003 - -0.120** 0.058 [0.036] [0.037] [0.055] [0.049] Risk Appetite (0-10) -0.009 -0.009 -0.007 0.001 -0.016* [0.006] [0.007] [0.011] [0.009] [0.009] Time Preference Index -0.020 -0.030 0.014 -0.003 -0.029 [0.027] [0.031] [0.060] [0.041] [0.038] Edu: Up to Secondary -0.042 0.026 0.024 0.000 0.022 [0.071] [0.066] [0.104] [.] [0.092] Edu: Postsecondary -0.075 -0.037 0.104 -0.016 -0.031 [0.070] [0.061] [0.099] [0.070] [0.090] Edu: Bachelors -0.091 -0.059 0.080 -0.051 -0.048 [0.068] [0.056] [0.088] [0.065] [0.088] Edu: Professional 0.000 0.033 0.247 0.088 0.000 [.] [0.085] [0.177] [0.119] [.] Edu: Masters -0.064 0.000 0.000 -0.025 -0.043 [0.077] [.] [.] [0.080] [0.105] Chinese 0.042 0.094** -0.144* 0.050 0.045 [0.039] [0.045] [0.084] [0.055] [0.058] Indian -0.019 0.036 -0.098 0.071 -0.104 [0.051] [0.063] [0.084] [0.074] [0.069] Med Income -0.006 -0.002 -0.147 0.008 -0.028 [0.037] [0.039] [0.143] [0.052] [0.056] High Income -0.013 -0.002 -0.126 -0.029 -0.018 25 [0.039] [0.041] [0.136] [0.059] [0.054] Deemed as ‘Very Important’ feature of job: Earning Good Money 0.056* 0.067* 0.041 0.015 0.105** [0.032] [0.036] [0.071] [0.048] [0.044] Flexibility -0.131*** -0.145*** -0.037 -0.090* -0.165*** [0.031] [0.034] [0.076] [0.047] [0.042] Interaction with colleagues 0.012 0.020 -0.009 -0.036 0.061 [0.034] [0.039] [0.075] [0.052] [0.047] Making Use of my Skills -0.032 -0.005 -0.113 0.050 -0.109** [0.036] [0.041] [0.084] [0.055] [0.050] Opportunities for acquiring new skills 0.085** 0.079* 0.080 0.075 0.086* [0.038] [0.042] [0.084] [0.060] [0.050] Building professional profile 0.005 -0.018 0.062 -0.013 0.024 [0.034] [0.039] [0.077] [0.054] [0.046] Working in International Teams 0.003 0.034 -0.118 0.038 -0.028 [0.036] [0.041] [0.078] [0.056] [0.048] Current Job Adequately Provides Opportunities for … Earning Good Money -0.129*** -0.124*** -0.135** -0.112** -0.151*** [0.033] [0.037] [0.068] [0.049] [0.046] Flexibility 0.116*** 0.118*** 0.084 0.092** 0.125*** [0.031] [0.035] [0.077] [0.046] [0.043] Making Good Use of my Skills 0.028 0.036 0.015 0.006 0.057 [0.035] [0.040] [0.074] [0.051] [0.049] Acquiring New Skills 0.012 -0.009 0.058 -0.025 0.060 [0.037] [0.041] [0.084] [0.054] [0.052] Building my Professional Profile -0.035 -0.035 -0.045 0.005 -0.075 [0.039] [0.044] [0.085] [0.061] [0.053] Working in international teams -0.046 -0.049 -0.024 -0.007 -0.080* [0.036] [0.041] [0.080] [0.056] [0.048] Interacting with colleagues 0.032 0.016 0.147** 0.030 0.033 [0.036] [0.042] [0.072] [0.052] [0.051] Treatment Arms: Worker decides hours, 20% earnings cut 0.071* 0.062 0.078 0.140** -0.002 [0.041] [0.048] [0.073] [0.060] [0.058] Client/ Employer decides hours, 10% 0.022 0.017 0.021 0.044 0.014 earnings increase [0.032] [0.036] [0.083] [0.047] [0.045] Client/ Employer decides hours, 20% -0.047 -0.027 -0.096 -0.047 -0.038 earnings increase [0.040] [0.046] [0.076] [0.057] [0.057] Constant 0.089** 0.113** -0.023 0.077 0.105* [0.040] [0.045] [0.083] [0.059] [0.055] Mean 0.462 0.466 0.447 0.453 0.469 N 1338 1038 300 620 718 R-squared 0.056 0.073 0.186 0.094 0.070 Standard errors in brackets; * p < 0.10, ** p < 0.05, *** p < 0.01; Heteroskedasticity-robust standard errors Dep var = choice of Job with More flexible work hours ( i.e. Job B in Treatment Arms 1/2, and Job A in Treatment Arms 3/4) 26 Annex 3. Regression Results with Survey Weights We re-ran the analysis after assigning weights to observations such that the weighted sample represents the full working population of Malaysia. Weights were assigned based on adjustment factors required for the distribution of the sample with respect to age group, gender and educational qualification characteristics to follow the same pattern as the working population of Malaysia, based on the 2019 Household Income Survey (HIS). We now present the regression coefficients in tables 4 – 7 estimated using these survey weights. Annex 3. Table 8 - Likelihood of Choosing to work from an Office 5 days a week given alternate options for location of work/ earnings (total and by status in employment and gender) with survey weights TOTAL Freelancing Standard Female Male Employment WFH 5d/week, 10% 0.421*** 0.387*** 0.550*** 0.471*** 0.393*** lower earnings [0.0437] [0.0476] [0.0999] [0.0684] [0.0562] WFH 5d/week, 20% 0.518*** 0.446*** 0.725*** 0.550*** 0.499*** lower earnings [0.0457] [0.0517] [0.0720] [0.0729] [0.0581] WFH every 2d/week, 0.404*** 0.353*** 0.575*** 0.409*** 0.401*** 10% lower earnings [0.0420] [0.0475] [0.0918] [0.0694] [0.0529] WFH 2d/week, 20% 0.511*** 0.510*** 0.516*** 0.538*** 0.494*** lower earnings [0.0448] [0.0511] [0.0945] [0.0724] [0.0569] Mean 0.461 0.420 0.597 0.490 0.443 N 1338 1038 300 620 718 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 Annex 3. Table 9 Likelihood of Choosing Job A given alternate options for hours of work/ earnings (total and by status in employment and gender) with survey weights Total Freelancing Standard Female Male Employment B1: Worker decides 0.568*** 0.558*** 0.605*** 0.624*** 0.537*** hours, 10% lower earnings [0.0435] [0.0483] [0.101] [0.0641] [0.0574] B2: Worker chooses 0.564*** 0.556*** 0.587*** 0.458*** 0.629*** hours, 20% lower earnings [0.0451] [0.0512] [0.0939] [0.0755] [0.0538] B3: Client/ employer 0.503*** 0.436*** 0.730*** 0.501*** 0.505*** chooses hours, 10% higher earnings [0.0425] [0.0481] [0.0749] [0.0706] [0.0533] 27 B4: Client/ employer 0.590*** 0.547*** 0.749*** 0.610*** 0.578*** chooses hours, 20% higher earnings [0.0437] [0.0507] [0.0782] [0.0709] [0.0556] Mean 0.555 0.522 0.666 0.547 0.559 N 1338 1038 300 620 718 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 Annex 3. Table 10 Likelihood of Choosing to work from an Office 5 days a week (total and by status in employment and gender) with survey weights (1) (2) (3) (4) (5) Total Freelancing Standard Female Male Employment Standard Employment 0.118** - - 0.021 0.149** [0.057] [0.081] [0.071] Age -0.008 -0.017* 0.030 0.003 -0.026* [0.010] [0.010] [0.025] [0.014] [0.014] Age-squared 0.000 0.000 -0.000 -0.000 0.000* [0.000] [0.000] [0.000] [0.000] [0.000] Female 0.051 0.054 0.107 - - [0.058] [0.064] [0.116] Single 0.091 -0.004 0.326*** 0.069 0.067 [0.058] [0.066] [0.100] [0.095] [0.073] Has children 0.172*** 0.142** 0.223* 0.108 0.138* [0.065] [0.072] [0.130] [0.079] [0.073] Female*Has children -0.050 -0.061 -0.106 0.000 0.000 [0.082] [0.091] [0.152] [.] [.] HH Size -0.019* -0.023** 0.005 -0.037** -0.005 [0.011] [0.012] [0.021] [0.016] [0.014] GLOW Member -0.123** -0.110** 0.000 -0.245*** -0.035 [0.053] [0.053] [.] [0.079] [0.066] Risk Appetite (0-10) -0.003 0.001 -0.013 -0.016 0.010 [0.009] [0.010] [0.017] [0.011] [0.011] Time Preference Index -0.001 0.019 -0.124 0.021 -0.034 [0.042] [0.046] [0.083] [0.059] [0.052] Deemed as ‘Very Important’ in a job: Earning Good Money 0.041 0.047 -0.030 0.051 0.037 [0.048] [0.053] [0.097] [0.082] [0.061] Flexibility -0.204*** -0.226*** -0.105 -0.115* -0.272*** [0.046] [0.050] [0.093] [0.066] [0.059] Interaction with 0.007 -0.020 0.079 -0.136* 0.122** colleagues [0.053] [0.059] [0.096] [0.078] [0.061] 28 Making Use of my 0.008 0.037 0.023 -0.029 0.008 Skills [0.058] [0.060] [0.116] [0.097] [0.070] Opportunities for -0.015 -0.033 0.032 0.056 -0.019 acquiring new skills [0.062] [0.066] [0.116] [0.101] [0.072] Building 0.066 0.060 0.063 0.116 0.055 professional profile [0.053] [0.057] [0.102] [0.086] [0.059] Working in 0.040 0.105* -0.075 0.140* -0.053 International Teams [0.055] [0.061] [0.098] [0.079] [0.065] Current Job Adequately Provides Opportunities for … Earning Good -0.045 -0.053 0.004 0.041 -0.071 Money [0.062] [0.070] [0.112] [0.076] [0.074] Flexibility -0.112** -0.131*** 0.031 -0.040 -0.141** [0.046] [0.049] [0.106] [0.069] [0.056] Making Good Use 0.031 0.008 -0.010 -0.175** 0.127 of my Skills [0.063] [0.074] [0.100] [0.080] [0.078] Acquiring New 0.059 0.036 0.269*** 0.096 0.014 Skills [0.058] [0.066] [0.094] [0.086] [0.071] Building my 0.104 0.152** -0.216* 0.161 0.089 Professional Profile [0.068] [0.077] [0.110] [0.100] [0.086] Working in 0.021 0.045 -0.060 -0.003 -0.009 international teams [0.067] [0.077] [0.123] [0.095] [0.087] Interacting with -0.046 -0.005 -0.118 -0.086 -0.009 colleagues [0.054] [0.062] [0.088] [0.089] [0.066] Treatment Arms /Description of Job B (ref: Job B1 WFH 5d/week, 10% earnings cut) B2: WFH 5d/week, 0.065 0.045 0.059 -0.005 0.103 20% earnings cut [0.056] [0.062] [0.119] [0.079] [0.069] B3: WFH 2d/week, -0.049 -0.060 -0.059 -0.117 -0.017 10% earnings cut [0.056] [0.061] [0.110] [0.087] [0.070] B4: WFH 2d/week, 0.066 0.100 -0.135 0.018 0.107 20% earnings cut [0.059] [0.063] [0.117] [0.091] [0.072] Constant 0.660*** 0.948*** 0.209 0.773** 0.956*** [0.238] [0.249] [0.565] [0.342] [0.339] Education Controls Yes Yes Yes Yes Yes Ethnicity Controls Yes Yes Yes Yes Yes 29 Income Level Yes Yes Yes Yes Yes Controls Mean 0.461 0.420 0.597 0.490 0.443 N 1338 1038 300 620 718 R-squared 0.131 0.143 0.282 0.201 0.167 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 Annex 3. Table 11 Likelihood of Choosing More Flexible Work Hours (total and by status in employment and gender) with survey weights (1) (2) (3) (4) (5) Total Freelancing Standard Female Male Employment Standard Employment -0.093 - - -0.213** -0.046 [0.057] [0.085] [0.074] Age -0.012 -0.007 -0.003 0.001 -0.026* [0.009] [0.010] [0.023] [0.012] [0.015] Age-squared 0.000 0.000 0.000 0.000 0.000 [0.000] [0.000] [0.000] [0.000] [0.000] Female 0.008 0.051 -0.153 - - [0.057] [0.059] [0.098] Single -0.065 -0.031 -0.058 0.023 -0.086 [0.060] [0.071] [0.087] [0.080] [0.079] Has children -0.003 0.028 -0.033 0.074 0.003 [0.065] [0.076] [0.108] [0.075] [0.074] Female*Has children 0.039 0.031 0.121 0.000 0.000 [0.082] [0.091] [0.144] [.] [.] HH Size 0.003 -0.008 0.039* 0.006 0.004 [0.012] [0.013] [0.020] [0.015] [0.016] GLOW Member -0.025 -0.003 0.000 -0.230*** 0.068 [0.053] [0.053] [.] [0.080] [0.064] Risk Appetite (0-10) -0.005 -0.009 -0.018 0.011 -0.015 [0.009] [0.011] [0.012] [0.012] [0.012] Time Preference Index -0.056 -0.056 -0.034 0.084 -0.137** [0.042] [0.048] [0.071] [0.058] [0.054] Deemed as ‘Very Important’ in a job: Earning Good Money - -0.112** -0.223** -0.182*** -0.107* 0.127*** [0.048] [0.052] [0.087] [0.065] [0.061] Flexibility 0.107** 0.085 0.235*** 0.149** 0.094 [0.046] [0.052] [0.090] [0.066] [0.059] Interaction with colleagues 0.010 0.016 -0.059 -0.024 0.068 [0.055] [0.063] [0.095] [0.086] [0.069] Making use of my Skills 0.000 -0.029 0.262** -0.144** 0.073 30 [0.054] [0.058] [0.110] [0.068] [0.069] Opportunities for acquiring new -0.012 -0.028 -0.015 0.070 -0.066 skills [0.058] [0.064] [0.103] [0.079] [0.075] Building professional profile -0.098* -0.076 -0.157* -0.060 -0.114* [0.054] [0.059] [0.095] [0.089] [0.065] Working in International Teams 0.047 -0.005 0.174* 0.104 -0.003 [0.057] [0.066] [0.092] [0.081] [0.072] Current Job Adequately Provides Opportunities for … Earning Good Money 0.070 0.070 0.020 0.110 0.040 [0.060] [0.069] [0.093] [0.083] [0.083] Flexibility -0.037 -0.083 0.184* 0.069 -0.062 [0.048] [0.051] [0.102] [0.077] [0.060] Making Good Use of my Skills -0.012 0.007 -0.152 -0.007 -0.026 [0.060] [0.068] [0.103] [0.072] [0.083] Acquiring New Skills 0.076 0.103* 0.024 0.065 0.082 [0.055] [0.062] [0.104] [0.069] [0.073] Building my Professional -0.096 -0.063 -0.197 -0.192*** -0.077 Profile [0.060] [0.066] [0.121] [0.072] [0.081] Working in international teams -0.020 -0.053 -0.040 -0.106 0.020 [0.063] [0.074] [0.115] [0.098] [0.084] Interacting with colleagues 0.078 0.060 0.212** 0.112 0.090 [0.054] [0.061] [0.086] [0.081] [0.069] Treatment Arms description of Job B (ref: Job B implies worker decides work hours, 10% earnings cut) Worker decides hours, 20% 0.088 0.149** -0.077 0.109 0.076 earnings cut [0.059] [0.067] [0.091] [0.079] [0.077] Client/ Employer decides hours, 0.229*** 0.172*** 0.427*** 0.235*** 0.224*** 10% earnings increase [0.056] [0.063] [0.098] [0.074] [0.074] Client/ Employer decides hours, 0.054 0.017 0.200** 0.076 0.051 20% earnings increase [0.057] [0.065] [0.096] [0.080] [0.074] Constant 0.835*** 0.745*** 0.565 0.370 1.183*** [0.245] [0.267] [0.515] [0.289] [0.359] Education Controls Yes Yes Yes Yes Yes Ethnicity Controls Yes Yes Yes Yes Yes Income Level Controls Yes Yes Yes Yes Yes Mean 0.461 0.420 0.597 0.490 0.443 N 1338 1038 300 620 718 R-squared 0.131 0.143 0.282 0.201 0.167 Standard errors in brackets * p < 0.10, ** p < 0.05, *** p < 0.01 31