Gender Inequality in Research Productivity During the COVID-19 Pandemic
GGender Inequality in Research Productivity Duringthe COVID-19 Pandemic
Ruomeng Cui
Goizueta Business School, Emory University, [email protected]
Hao Ding
Goizueta Business School, Emory University, [email protected]
Feng Zhu
Harvard Business School, Harvard University, [email protected]
We study the disproportionate impact of the lockdown as a result of the COVID-19 outbreak on female andmale academics’ research productivity in social science. The lockdown has caused substantial disruptionsto academic activities, requiring people to work from home. How this disruption affects productivity andthe related gender equity is an important operations and societal question. We collect data from the largestopen-access preprint repository for social science on 41,858 research preprints in 18 disciplines producedby 76,832 authors across 25 countries over a span of two years. We use a difference-in-differences approachleveraging the exogenous pandemic shock. Our results indicate that, in the 10 weeks after the lockdown inthe United States, although the total research productivity increased by 35%, female academics’ productivitydropped by 13.9% relative to that of male academics. We also show that several disciplines drive such genderinequality. Finally, we find that this intensified productivity gap is more pronounced for academics in top-ranked universities, and the effect exists in six other countries. Our work points out the fairness issue inproductivity caused by the lockdown, a finding that universities will find helpful when evaluating facultyproductivity. It also helps organizations realize the potential unintended consequences that can arise fromtelecommuting.
Key words : Gender inequality, research productivity, telecommuting, COVID-19
1. Introduction
The Coronavirus 2019 (COVID-19) pandemic has significantly changed the way people live andwork. The pandemic has led to unprecedented societal, scientific, and economic changes. Peoplehave been forced to work from home through telecommuting, potentially affecting their productiv-ity. In this research, we study how this pandemic shock affected academics’ research productivityusing data from the largest open-access repositories for social science in the world—the SocialScience Research Network (SSRN). We provide evidence that female researchers’ productivity https://en.wikipedia.org/wiki/Social_Science_Research_Network , accessed June 2020. a r X i v : . [ c s . D L ] J u l Cui, Ding, Zhu:
Gender Inequality in Research Productivity dropped significantly relative to that of male researchers as a result of the lockdown in the UnitedStates.In response to the pandemic, the US and many other countries have mandated their citizensto stay at home. As a result of the lockdown, many people have had to perform both work andhousehold duties at home. Most countries have also closed their schools and daycare centers, whichhas massively increased childcare needs. Given that the childcare provided by grandparents andfriends is limited due to the social distancing protocol, most families must care for their childrenthemselves. In addition, restaurants have been either closed or do not allow dine-ins, which hasincreased the need for food preparation at home. Given that women, on average, are burdened withdisproportionately more childcare, domestic labor, and household responsibilities (Bianchi et al.2012), they are likely to be more affected than men during the lockdown.The lockdown has also disrupted how academics carry out their activities. Many countries haveclosed their universities, so faculties have to conduct research and teaching at home. Conductingscientific research generally requires a quiet and interruption-free environment because concentra-tion is critical for creative thinking. The unequal distribution of domestic duties means that femaleacademics are likely to be disproportionately affected compared with their male colleagues.Anecdotal evidence provides mixed support (Dolan and Lawless 2020). A recent survey involving4,500 principal investigators reported significant and heterogeneous declines in the amount of timethey are spending on research (Myers et al. 2020). Several journal editors have noticed that,while there has been a 20–30% increase in submissions as a result of the pandemic, most of thisincrease can be attributed to male academics (Beck 2020). Amano-Patio et al. (2020) find that aparticularly large number of senior male economists, instead of mid-career economists, have beenexploring research questions arising from the COVID-19 shock. Others have seen no change or havebeen receiving comparatively more submissions from women since the lockdown (Kitchener 2020).However, there is a dearth of systematic evidence on whether and to what extent the shock hasaffected gender inequality in academia.It is an important operations and societal question to understand the change in productivityand the related gender equity caused by the reorganization of work and care at home. In thispaper, we use a large dataset on female and male academics’ production of new research papers tosystematically study whether COVID-19 and the subsequent lockdown have had a disproportionateeffect on female academics’ productivity. We also identify the disciplines, universities, and countriesin which this inequality is intensified.We collect data on all research papers uploaded to SSRN in 18 disciplines from December 2018to May 2019 and from December 2019 to May 2020. We extract information on paper titles, authornames, author affiliations, and author addresses. We use this information to identify the authors’ ui, Ding, Zhu:
Gender Inequality in Research Productivity countries and institutions. We also use their names and their faculty pages to identify their gender.The final dataset includes 41,858 papers written by 76,832 authors from 25 countries. Our mainanalysis focuses on academics in the US, and we then perform the same analysis for other countries.We employ a difference-in-differences (DID) approach to estimate the effect. We compute thenumber of papers produced by female and male academics in each week. We then compare thevariations in the women and men’s research productivity gap before and after the start of thelockdown and show that the gap increased after the start of the lockdown. We also show thatfemale and male authors’ preprint volumes followed the parallel time trend before the lockdown,and we find no significant changes in the research productivity gap in 2019 during the same timeof the year. Taken together, these results suggest that the intensified disparity has primarily beendriven by the pandemic shock.We find that during the 10 weeks since the lockdown began, female academics’ research produc-tivity has dropped by 13.9% compared to that of male academics in the US. The effect persists whenwe vary the time window since the pandemic outbreak in the analysis. Our findings lend empiricalcredence to the argument that when female and male academics face a reorganization of care andwork time, women become significantly less productive by producing less number of papers, butthe quality of their uploaded papers, measured by the download and view rates, does not change.In addition, we find that the effect is more pronounced in top-ranked research universities, andthat this effect exists in six other countries.While gender inequality has been long documented for academics in terms of tenure evaluation(Antecol et al. 2018), coauthoring choices (Sarsons 2017), and number of citations received (Ghiasiet al. 2015), the COVID-19 pandemic brings this issue to the forefront. Our study is among thefirst to rigorously quantify such inequality in research productivity as a result of the pandemic,and our results highlight that this disruption has exacerbated gender inequality in the academicworld. There are concerns that, because all academics participate together in open competitionsfor promotions and positions, these short-term changes in productivity will affect their long-termcareer outcomes (Minello 2020). Thus, institutions should take this inequality into considerationwhen evaluating faculty members.Our paper contributes to the literature on productivity, a central topic in operations manage-ment. Previous studies have examined key determinants of workers’ productivity, such as peereffects (Song et al. 2018, Tan and Netessine 2019), workers’ workloads (Tan and Netessine 2014),incentive schemes (Chen et al. 2019), and unfairness in aligning workers’ compensations with pro-ductivity (Pierce et al. 2020). In particular, multitasking has been shown to reduce productivity forworkers who perform complex tasks because of their limited cognitive capacities (KC 2014, Brayet al. 2016, KC 2020). In our context, when working from home, academics have an increasing Cui, Ding, Zhu:
Gender Inequality in Research Productivity need to allocate their cognitive capacity across house and work tasks, making those who have morehousework-related distractions to struggle due to multitasking. The unequal distribution of homeresponsibilities means that women are more likely to deal with multitasking at home during thelockdown. We contribute to the literature by showing that female academics are more negativelyaffected in research productivity, highlighting fairness as an important factor to consider whenmeasuring productivity.Our work also sheds light on the fairness issues that could arise from telecommuting, an opera-tions choice faced by companies. Since working from home can provide a flexible work schedule foremployees and reduce office-related costs for companies, an increasing number of companies arechoosing this operating model. Between 2005 and 2015, the number of US employees who choseto telecommute increased by 115% (Abrams 2019). By 2019, about 16% of the total workforcein the US was working remotely full time or part time (U.S. Bureau of Labor Statistics 2019).After the pandemic hit, for example, Twitter and Facebook announced that their employees areallowed to work from home permanently (McLean 2020), and JP Morgan plans to expand itstelecommuting program (Kelly 2020). Despite the growing popularity of telecommuting, scholarsand practitioners still lack a comprehensive understanding of the managerial and societal impactsof this operations choice (Nicklin et al. 2016). We contribute to the literature by pointing out theproductivity inequality phenomenon caused by the lockdown and telecommuting, which might leadto a longer-term unemployment risk for women, an unintended consequence that companies andsociety should take into account when making their operation choices or implementing performanceevaluation policies.
2. Literature Review and Theory Development
Our work is closely related to three streams of literature: (1) productivity, (2) social operations,and (3) telecommuting.Our work studies productivity, a central topic in operations management. When working fromhome during the lockdown, researchers may need to allocate their cognitive capacity across home-related and work-related tasks. They have to deal with more distractions and multitasking. Proirstudies have shown mixed effects of multitasking on workers’ productivity, such as an increasedservice speed with a lower service quality for restaurant waiters (Tan and Netessine 2014) or aslower processing speed for bank associates (Staats and Gino 2012) as a result of multitasking.Multitasking could lead to a more profound productivity loss for jobs that require a higher levelof cognitive capacity. For example, in the judiciary system, reducing the level of multitasking hasbeen shown to help judges focus on the most recent cases, reduce the switching costs between ui, Ding, Zhu:
Gender Inequality in Research Productivity cases, and increase the case completion rate (Bray et al. 2016). In healthcare contexts, excessivemultitasking and frequent interruptions in the work flow have been shown to hinder the productivityand effectiveness of nurses’ care (Tucker and Spear 2006) and physicians’ discharging (KC 2014),processing (Berry Jaeker and Tucker 2017), and medication delivery activities (Batt and Terwiesch2017). In our research context, multitasking is one of the challenges that women are more likely todeal with at home during the lockdown. Therefore, we contribute to the productivity literature bypointing out an equal outcome in research productivity in such disruptions.This paper sheds light on a key social issue—fairness and equity—in research productivity,adding to the growing literature on operations’ social impacts. Several recent influential papers byTang and Zhou (2012), Lee and Tang (2018), and Dai et al. (2020) encourage OM researchers towork on socially responsible topics that are important to corporations and society at large. Thegrowing literature has examined the social impact of operations strategies, such as the use of reviewinformation to reduce racial discrimination arising in the sharing economy (Cui et al. 2020a, Mejiaand Parker 2020) and the gender inequality driven by specific compensation schemes (Pierce et al.2020). The literature on gender bias has shown evidence that female researchers or students tend tobe discredited when they are evaluated alongside equally competent male candidates (Moss-Racusinet al. 2012, Sarsons 2017), that women are more likely to be assigned tasks that are undesirable(Chan and Anteby 2016), more service-oriented (Guarino and Borden 2017), and with less futurepromotion opportunities (Babcock et al. 2017), and that women are often responsible for morehousework and childcare (Schiebinger and Gilmartin 2010, Misra et al. 2012). In our context, whenworking from home, the unequal distribution of housework means that women are more likely todeal with non-work-related tasks during the lockdown and experience a decline in productivity.A recent survey involving 4,500 principal investigators shows that female scientists self-reporteda sharper reduction in research time during the COVID-19 lockdown, primarily due to childcareneeds (Myers et al. 2020). We contribute to the literature by providing direct evidence that thelockdown affects productivity and exacerbates gender inequity in the workplace, potentially leadingto a long-term career risk for women, an unintended consequence that companies should considerwhen designing their operations models and performance evaluation policies.Our work is also related to the emerging literature on organizations’ telecommuting choices.We demonstrate an unexpected fairness and social issue that stems from the challenge created bythis digital operating model. Transitioning from traditional in-office work to telecommuting mightaffect workers’ behavior and productivity though the team work effect and peer effect. For example,Dutcher and Saral (2012) observe that workers do not exert free-riding behavior when a teamis comprised of in-office workers and telecommuters, and Bloom et al. (2015) demonstrates thattelecommuting is able to improve productivity when it is performed in a quiet environment. Our Cui, Ding, Zhu:
Gender Inequality in Research Productivity work adds to this stream of literature by showing that this digital operating model may lead to anunintended consequence in gender inequality, which adds to a more comprehensive understandingof telecommuting not yet developed in the literature.
The outbreak of COVID-19 has caused substantial disruptions to academic activities. In responseto the pandemic, most countries have closed their schools and daycare centers and mandated thattheir citizens be quarantined at home. As a result, researchers from more than 1,100 collegesand universities had to carry out both work and household duties at home (National Conferenceof State Legislatures 2020). Conducting scientific work often requires hours of interruption-freeenvironment. When working from home, researchers are likely to experience more distractions fromhousework, resulting in excessive multitasking across research, housework, and childcare tasks.Multitasking means that workers have to allocate their limited cognitive capacity across multipletasks. This hinders productivity due to the setup cost associated with switching between tasksand the difficulty of focusing on and organizing relevant information (KC 2014). Substantial evi-dence from neurology and psychology research indicates that multitasking and interruptions areubiquitous to human brains (Mark et al. 2008). Multitasking has been shown to induce stress andfrustration (Mark et al. 2008), negatively influence the retention of information (Clapp et al. 2011),make people more easily distracted (Levitin 2014), and exhaust their cognitive capacity (Janssenet al. 2015). Researchers who experience more distractions from housework might struggle morefrom multitasking.Women are on average, disproportionately burdened with childcare and household responsibili-ties (Bianchi et al. 2012). They are shown to spend almost twice as much time as men on houseworkand childcare in the US (Bianchi et al. 2012). Moreover, there are 8.5 million more single moth-ers than single fathers in the US (Alon et al. 2020). Even in the gender-egalitarian countries ofnorthern Europe, women are responsible for almost two-thirds of the unpaid work (The EuropeanCommission 2016). Among heterosexual couples with female breadwinners, women still do most ofthe care work (Chesley and Flood 2017). The same pattern exists in academia (Schiebinger andGilmartin 2010, Andersen et al. 2020). Women professors spend more time doing housework andcarework than men professors across various ranks; for example, 34.1 hours versus 27.6 hours perweek for lecturers, 29.6 hours versus 25.1 hours per week for assistant professors, and 37.7 hoursversus 24.5 hours per week for associate professors (Misra et al. 2012). The unequal distribution ofdomestic duties means that female researchers are more likely to be disrupted when telecommut-ing from home, and in turn they are more likely to experience multitasking during the lockdown.Taking these factors together, we thus hypothesize that female researchers are more likely to bedisproportionately affected in their productivity compared with male researchers. ui, Ding, Zhu:
Gender Inequality in Research Productivity
3. Data and Summary Statistics
We collect data from SSRN, a repository of preprints with the objective to rapidly disseminatescholarly research in social science. We gather data on all social science preprints submitted fromDecember 2018 to May 2019 and from December 2019 to May 2020. We extract information onpaper titles, author names, author affiliations, and author addresses. We use the authors’ addressesto identify their countries. The COVID-19 outbreak began at different time points in differentcountries, so we collect each country’s lockdown start date of lockdown from news sources and theUnited Nations’ report. We drop authors without addresses or with addresses in more than onecountry because we cannot determine when these authors were affected by the lockdown. We alsodrop countries without a sufficient number of authors in our data set. The final data set consist ofa total of 41,858 papers in 18 disciplines produced by 76,832 authors from 25 countries.To identify the authors’ genders, we first use a database called
Genderize , which predicts thegenders based on their first names and provides a confidence level. About 78% of the authors’genders were identified with over 80% confidence levels. For the remaining authors, we use AmazonMechanical Turk to manually search for their professional webpages based on names and affiliationsand then infer their genders from their profile photos. Our dataset contains a total of 21,733 femaleacademics and 55,099 male academics.We aggregate the number of new preprints at the weekly level. We then count the number ofpapers uploaded by each author in each week. To measure the effective productivity for preprintswith multiple authors, when a preprint has n authors, each author gets a publication count of 1 /n . Finally, we aggregate the effective number of papers to the gender level: in each week, we countthe total number of papers produced by male and female authors separately in each social sciencediscipline.Figure 1 plots the time trend of preprints in aggregation from December 3, 2019 to May 19,2020 in the US. The vertical line represents the week of March 11, 2020, which is the start ofthe implementation of the nationwide lockdown measures in the US. We can observe that maleacademics, on average, have submitted more preprints than female academics, and that femaleand male academics’ research productivity evolved in parallel before the lockdown. After the lock-down started, however, male academics significantly boosted their productivity, whereas femaleacademics’ productivity did not change much, indicating an increased productivity gap. https://en.unesco.org/covid19/educationresponse , accessed June 2020 Available at https://genderize.io/ , accessed June 2020 Note that in many social science disciplines, author names are listed in alphabetical order. Most universities were closed in the week of March 11, 2020. Source: https://gist.github.com/jessejanderson/09155afe313914498a32baa477584fae?from=singlemessage&isappinstalled=0 , accessed June 2020.
Cui, Ding, Zhu:
Gender Inequality in Research Productivity
This graph plots the time trend of the number of preprints for female academics and male academics. The vertical linerepresents the start of the lockdown due to COVID-19 in the US.
To ensure that our results are not driven by seasonality, we plot the time trend of preprintsduring the same time window in 2019 in Appendix Figure A.1. We observe a similar pattern beforethe week of March 11, 2019, but there is no significant change in the productivity gap after thatweek.We use the authors’ affiliations to identify their universities. To ascertain whether the produc-tivity gap is intensified or weakened across top-ranked and lower-ranked research universities, wecollect social science research rankings from three sources: QS University Ranking, Times HigherEducation, and Academic Ranking of World University. We then use these data to rank USuniversities.Table 1 reports the summary statistics for the weekly number of preprints by gender and disci-pline as well as split sample statistics prior to or after the lockdown from December 3, 2019 to May19, 2020, spanning 24 weeks. This sample includes 9,943 preprints produced by 15,494 authors inthe US and 21,065 preprints produced by 37,997 authors across all countries. The average numberof submissions per week is 444.6 in the US and 877.7 across all 25 countries. Notably, while thetotal research productivity in the US was boosted by 35% after the lockdown, male authors seemto be the main contributors to this increase. Available at , accessed June 2020. Available at , accessed June 2020. Available at , accessed June 2020. ui, Ding, Zhu:
Gender Inequality in Research Productivity About 78% of the preprints fall under multiple disciplines. Note that when computing the totalpreprints, we count the paper only once when aggregating across disciplines to avoid multiplecounting. When computing the number of preprints in each discipline, we separately count all ofthe papers classified under each one. We observe substantial variations across disciplines. Among18 disciplines, Political Science, Economics, and Law received the most submissions, whereas Geog-raphy, Criminal Justice and Education received the fewest submissions. While there was a largeincrease in productivity in several disciplines, such as Economics, Political Science, Finance, HealthEconomics, and Sustainability, after the COVID-19 outbreak, other disciplines showed no obvi-ous increase. A few disciplines, such as Anthropology, Cognitive, and Information Systems, evenexperienced a decline.
Table 1 Summary Statistics
All observations Before Lockdown After LockdownLevel Weekly no. of preprints Mean Std. dev Max Min Total Mean Std. dev Mean Std. devAllDisciplines(US only) All 444 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4. Empirical Results
In this section, we identify the effect of the COVID-19 outbreak on research productivity. We firstelaborate our identification methodology that leverages the exogenous pandemic shock by using Authors self-classify their own preprints into disciplines when they upload their papers. SSRN reviews and approvesthese classifications. Cui, Ding, Zhu:
Gender Inequality in Research Productivity a DID regression. We then report the estimation results of gender inequality in the US, acrossuniversities, and across countries.
Our identification exploits the lockdown as a result of the COVID-19 outbreak as an exogenousshock that has caused substantial disruptions to academic activities, requiring academics to conductresearch, teach, and carry out household duties at home. The validity of our approach depends onthe assumption that the shock is exogenous with respect to the researchers’ anticipated responses.If a particular gender group of researchers anticipated and strategically prepared for the shock byaccelerating the completion of their current research papers, among others, this could confoundthe treatment effect. In reality, this possibility is unlikely because of the rapid development ofthe situation. COVID-19 was regarded as a low risk and not a threat to the US in late January(Moreno 2020), and no significant actions had been taken other than travel warnings issued forfour countries until late February (Franck 2020). It quickly turned into a global pandemic afterthe declaration of the World Health Organization on March 11, 2020, followed by the nationwideshelter-in-place orders within a week. We adopt the DID methodology, a common approach used to evaluate people’s or organizations’responses to natural shocks (Seamans and Zhu 2013, Calvo et al. 2019, Cui et al. 2020b). Weperform the DID analysis using outcome variables on two levels: the total number of preprintsaggregated across all disciplines and the number of preprints in each discipline.We first compare the productivity gap between female and male researchers prior to and afterthe pandemic outbreak using the following model specification with aggregate-level data: log ( P reprints gt ) = c + F emale g + βF emale g × Lockdown t + γ t + (cid:15) gt , (1)where g denotes the gender, t denotes the week, log ( P reprints gt ) represents the logged numberof preprints uploaded for gender g during week t , γ t is the time fixed effect, and (cid:15) t is the errorterm. The time fixed effect γ t includes a set of weekly time dummies that control for time trends.The dummy variable F emale g equals 1 if gender g is a female, and 0 otherwise. The dummyvariable Lockdown t equals 1 if week t occurs after the lockdown measure was adopted (i.e., theweek of March 11, 2020), and 0 otherwise. Its main effect is absorbed by the time fixed effects.The coefficient β estimates the effect of the lockdown on female academics’ research productivityrelative to male academics productivity. Source available at , accessed June 2020. ui, Ding, Zhu:
Gender Inequality in Research Productivity We also use discipline-level panel data to estimate the effect with the following DID specification: log ( P reprints igt ) = c + F emale g + βF emale g × Lockdown t + γ t + δ i + (cid:15) igt , (2)where i denotes each discipline, δ i is the discipline fixed effect that captures the time-invariantcharacteristics of discipline i , log ( P reprints igt ) represents the logged number of preprints uploadedfor discipline i for gender g during week t , and (cid:15) igt is the error term. As before, we include the timefixed effect γ t . Table 2 reports the estimated effect of the pandemic shock on research productivity at the aggre-gate level using Equation (1). Table 3 reports the estimated effect at the discipline level usingEquation (2). In each analysis, we use 14 weeks before the lockdown as the pre-treatment periodand 6 to 10 weeks after the lockdown as the post-treatment periods. The analyses yield consistentresults. First, in line with our summary statistics, the results show that fewer preprints are pro-duced by female academics than by male academics in general. Second, since the lockdown began,there has been a significant reduction in female academics’ productivity relative to that of theirmale colleagues, indicating an exacerbated productivity gap in gender. The coefficient of the inter-acted term in Column (1) of Table 2 suggests a reduction of 17.9% in females’ productivity overthe six-week period after the lockdown relative to the males’, and the coefficient of the interactedterm in Column (5) suggests an average reduction of 13.9%. Table 2 Impact of Lockdown on Gender Inequality
Dependent variable: No. of preprints (in logarithm) in aggregation6 weeks 7 weeks 8 weeks 9 weeks 10 weeksVariables (1) (2) (3) (4) (5)
F emale − − . − . − . − . F emale × Lockdown − − . − . − . − . R .
982 0 .
982 0 .
982 0 . ∗ p < . ∗∗ p < . ∗∗∗ p < . We then repeat the analysis as in Table 2 for each discipline separately. Table 4 reports the coef-ficients of the interacted term,
F emale g × Lockdown t , for each discipline. We find that the genderdifferences significantly intensified in several disciplines, namely, Criminal, Economics, Finance,Health Economics, Political Science, and Sustainability. Because the outcome variable is logged, the percentage change in the outcome variable is computed as e coefficient − Cui, Ding, Zhu:
Gender Inequality in Research Productivity
Table 3 Impact of Lockdown on Gender Inequality at the Discipline Level
Dependent variable: No. of preprints (in logarithm) by discipline6 weeks 7 weeks 8 weeks 9 weeks 10 weeksVariables (1) (2) (3) (4) (5)
F emale − − . − . − . − . F emale × Lockdown − − . − . − . − . R .
836 0 .
839 0 .
841 0 . ∗ p < . ∗∗ p < . ∗∗∗ p < . Table 4 Impact of Lockdown on Gender Inequality in Each Discipline
Dependent variable: No. of preprints (in logarithm) by discipline6 weeks 7 weeks 8 weeks 9 weeks 10 weeksDiscipline (1) (2) (3) (4) (5)Accounting − . − . − . − . − . − .
015 0 .
049 0 .
123 0 .
112 0 . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − .
010 0 . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . .
033 0 .
042 0 .
070 0 .
070 0 . .
081 0 .
088 0 .
097 0 .
140 0 . − . − . − . − . − . .
069 0 .
169 0 .
157 0 .
148 0 . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . ∗ p < . ∗∗ p < . ∗∗∗ p < . Table 5 replicates the DID analysis using Equation (2) for a subset of academics based on therankings of their affiliated universities. Due to our focus on social science, we use the 2020 QSWorld University Ranking for social sciences and management as the main analysis. We separatelyanalyze academics in universities ranked in the top 10, 20,..., and 100. The results show that theCOVID-19 effect is more pronounced in top-tier universities and that this effect in general decreasesand becomes less significant as we include more lower-ranked universities. We find similar resultswhen using the two other rankings, as shown in Appendix Table A.1. It is possible that some authors are affiliated with more than one academic institutions. We use the highest rankedinstitution as their affiliation in such cases. ui, Ding, Zhu:
Gender Inequality in Research Productivity Table 5 Impact of Lockdown on Gender Inequality by University Ranking
Dependent variable: No. of preprints (in logarithm) by disciplineUniversities 6 weeks 7 weeks 8 weeks 9 weeks 10 weeksby QS Ranking (1) (2) (3) (4) (5)Top 10 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . ∗ p < . ∗∗ p < . ∗∗∗ p < . Finally, we examine how the estimated gender inequality varies across countries by replicatingthe analysis for academics in each country. Figure 2 illustrates the impact on the productivitygap graphically by plotting the estimates of the interacted term with 90% and 95% confidenceintervals, where a negative value represents a drop in female academics’ research productivityrelative to that of male academics. We can observe that most countries—21 out of 25 countries—have experienced a decline in female researchers’ productivity. In addition to the US, six countrieshave shown statistically significant declines: Japan, China, Australia, Italy, the Netherlands, andthe United Kingdom. Note that because SSRN is a repository primarily used by US researchers,SSRN’s preprints for other countries might be limited in number, which might weaken our abilityto detect changes.In short, we find that the lockdown has adversely affected female researchers’ productivity rela-tive to that of male researchers. We also find a large heterogeneity of such gender inequality acrossdisciplines, universities, and countries.
5. Robustness Checks
In this section, we report the results of several robustness tests. Specifically, we check the paralleltrends assumption and conduct falsification tests to ensure that our estimated effects are notidiosyncratic. In addition, we test the change in research quality by measuring the download andabstract view rates.
The key identification assumption for the DID estimation is the parallel trends assumption: beforethe COVID-19 shock, female and male researchers’ productivity would follow the same time trend.In Appendix Figure A.1, which presents the time trends of preprints in 2019, the visual inspection Cui, Ding, Zhu:
Gender Inequality in Research Productivity
Figure 2 Impact of Lockdown on Gender Inequality across Countries
This graph plots the estimates of the interacted term with 90% and 95% confidence intervals in each country. The negativevalues represent female academics’ research productivity drop relative to that of male academics across countries. shows the two gender groups’ parallel evolution before the shock. We then test this assumption byperforming a similar analysis to Seamans and Zhu (2013) and Calvo et al. (2019), where we expandEquation (1) to estimate the treatment effect week by week before the shock. Specifically, we replace
Lockdown t in Equations (1) with the dummy variable T ime tτ , where τ ∈ {− , − , ..., − , − , } and T ime tτ = 1 if τ = t and 0 otherwise, indicating the relative τ th week of the outbreak, log ( P aper it ) = c + F emale i + − (cid:88) τ = − T ime tτ + − (cid:88) τ = − β τ F emale i × T ime tτ + (cid:15) it . (3)The benchmark group is the week of the pandemic outbreak. The coefficients β − to β − identifyany week-by-week pre-treatment difference between the female and male researchers, which weexpect to be insignificant. We then repeat the same analysis with our discipline-level data. ui, Ding, Zhu: Gender Inequality in Research Productivity Appendix Table A.2 presents the estimation results. The test results show no pre-treatmentdifferences in the research productivity trends between female and male academics, which supportsthe parallel trends assumption.
To show that our estimate effects are not an artifact of seasonality, we test whether such a declinein female productivity also existed in 2019. Appendix Table A.3 reports the summary statistics in2019. We repeat the same analysis specified in Equation (1) for the same time window in 2019. Ifour results simply capture seasonality, we would be able to find significant effects in 2019. AppendixTable A.4 reports the falsification test results. The placebo-treated average treatment effects areinsignificant, implying that women’s productivity did not decline significantly in the previous year.
One might question whether the difference in productivity is because male researchers increasedthe volume of their production at the expense of quality since the lockdown began. If this is true,the quality difference between male and female researchers’ preprints should have increased sincethe lockdown. We test this possibility using data on how many times the abstract has been viewedand the preprint has been downloaded for each preprint, the two primary quality indicators used bySSRN to rank preprints. Appendix Table A.5 reports the summary statistics of these two variables.We compare the average number of abstract views and downloads between preprints from maleand female researchers prior to and after the pandemic outbreak using the same specification inEquation (1) at the aggregation level: log ( Abstract V iews gt ) or log ( Downloads gt ) = c + F emale g + βF emale g × Lockdown t + γ t + (cid:15) gt . (4)Appendix Tables A.6 and A.7 report the estimation results. The average treatment effects areinsignificant, suggesting that after the lockdown, female and male researchers’ research quality didnot change significantly, suggesting that our findings are unlikely to be driven by the shifts inresearch quality.
6. Conclusions
Our paper adds to the long-standing literature on gender equality, an important topic in socialscience. For example, the literature has shown evidence of fairness in parental leaves (Lundquistet al. 2012), inequality in tenure evaluation (Sarsons 2017, Antecol et al. 2018), recognition (Ghiasiet al. 2015), and compensation (Pierce et al. 2020). Researchers have therefore investigated businessinnovations to help empower women (Plambeck and Ramdas 2020). The COVID-19 crisis brings along-existing issue to the forefront—the inequities faced by women who often contribute more in Cui, Ding, Zhu:
Gender Inequality in Research Productivity childcare and housework. We contribute to the literature by providing direct tests of the impactof the pandemic shock on gender inequality in academia.We show that, since the lockdown began, women have produced 13.9%–17.9% fewer researchpapers than men in the US. We also find that the effect exists in several disciplines and amongtop-ranked universities. Finally, we find that the increase in productivity inequality is significantin seven countries.Our findings indicate that, if the lockdown is kept in place for too long, female academics incertain disciplines at top-ranked universities are likely to be significantly disadvantaged, a fairnessissue that may expose women to a higher unemployment or career risk in the future. We hopeour findings could increase the awareness of this issue. Actions could be taken to balance domesticresponsibilities among spouses and set up an expectation of a fair allocation of efforts in housework.Universities need to take this potential gender inequality into account as they implement policiessuch as tenure clock extensions to the faculty in response to the pandemic. Our findings also indicatethat telecommuting may have unintended consequences on gender inequality. As the COVID-19outbreak accelerates the trend toward telecommuting, institutions and firms should take genderequality into consideration when implementing telecommuting policies. We hope that this workcould serve as a stepping stone to stimulate more research on the synergy between operations andsocial issues.Our study has a few limitations. First, our study focuses on social science disciplines, and thusthe findings may not be generalizable to other disciplines. Second, we have limited informationabout the researchers in our dataset. Future research could collect additional information, such astheir parental status, to directly test the mechanism underlying the observed empirical patterns. ui, Ding, Zhu:
Gender Inequality in Research Productivity References
Abrams, Zara. 2019. The future of remote work
Alon, Titan M, Matthias Doepke, Jane Olmstead-Rumsey, Michele Tertilt. 2020. The impact of COVID-19on gender equality. Tech. rep., National Bureau of Economic Research.Amano-Patio, Noriko, Elisa Faraglia, Zeina Hasna. 2020. Who is doing new research in the time of COVID-19? Not the female economists https://voxeu.org/article/who-doing-new-research-time-covid-19-not-female-economists .Andersen, Jens Peter, Mathias Wullum Nielsen, Nicole L Simone, Resa E Lewiss, Reshma Jagsi. 2020. Meta-research: Is COVID-19 amplifying the authorship gender gap in the medical literature? arXiv preprintarXiv:2005.06303 .Antecol, Heather, Kelly Bedard, Jenna Stearns. 2018. Equal but inequitable: Who benefits from gender-neutral tenure clock stopping policies?
American Economic Review (9) 2420–2441.Babcock, Linda, Maria P Recalde, Lise Vesterlund, Laurie Weingart. 2017. Gender differences in acceptingand receiving requests for tasks with low promotability.
American Economic Review (3) 714–47.Batt, Robert J, Christian Terwiesch. 2017. Early task initiation and other load-adaptive mechanisms in theemergency department.
Management Science (11) 3531–3551.Beck, Dani. 2020. The COVID-19 pandemic and the research lab .Berry Jaeker, Jillian A, Anita L Tucker. 2017. Past the point of speeding up: The negative effects of workloadsaturation on efficiency and patient severity. Management Science (4) 1042–1062.Bianchi, Suzanne M, Liana C Sayer, Melissa A Milkie, John P Robinson. 2012. Housework: Who did, doesor will do it, and how much does it matter? Social forces (1) 55–63.Bloom, Nicholas, James Liang, John Roberts, Zhichun Jenny Ying. 2015. Does working from home work?evidence from a chinese experiment. The Quarterly Journal of Economics (1) 165–218.Bray, Robert L, Decio Coviello, Andrea Ichino, Nicola Persico. 2016. Multitasking, multiarmed bandits, andthe italian judiciary.
Manufacturing & Service Operations Management (4) 545–558.Calvo, Eduard, Ruomeng Cui, Laura Wagner. 2019. Disclosing product availability in online retail. Forth-coming at Manufacturing & Service Operations Management.Chan, Curtis K, Michel Anteby. 2016. Task segregation as a mechanism for within-job inequality: Women andmen of the transportation security administration. Administrative Science Quarterly (2) 184–216.Chen, Xirong, Zheng Li, Liu Ming, Weiming Zhu. 2019. The incentive game under target effects in ridesharing:A structural econometric analysis. Available at SSRN 3479675 .8 Cui, Ding, Zhu:
Gender Inequality in Research Productivity
Chesley, Noelle, Sarah Flood. 2017. Signs of change? at-home and breadwinner parents’ housework andchild-care time.
Journal of Marriage and Family (2) 511–534.Clapp, Wesley C, Michael T Rubens, Jasdeep Sabharwal, Adam Gazzaley. 2011. Deficit in switching betweenfunctional brain networks underlies the impact of multitasking on working memory in older adults. Proceedings of the National Academy of Sciences (17) 7212–7217.Cui, Ruomeng, Jun Li, Dennis J Zhang. 2020a. Reducing discrimination with reviews in the sharing economy:Evidence from field experiments on airbnb.
Management Science (3) 1071–1094.Cui, Ruomeng, Tianshu Sun, Zhikun Lu, Joseph Golden. 2020b. Sooner or later? promising delivery speedin online retail. Working Paper .Dai, Yue, Tianjun Feng, Christopher S Tang, Xiaole Wu, Fuqiang Zhang. 2020. Twenty years in the making:The evolution of the journal of manufacturing & service operations management.
Manufacturing &Service Operations Management (1) 1–10.Dolan, Kathleen, Jennifer Lawless. 2020. It takes a submission: Gendered patterns in the pagesof ajps https://ajps.org/2020/04/20/it-takes-a-submission-gendered-patterns-in-the-pages-of-ajps/ .Dutcher, E Glenn, Krista Jabs Saral. 2012. Does team telecommuting affect productivity? an experiment.Tech. rep., Working Papers in Economics and Statistics.Franck, Thomas. 2020. Us expands iran travel restrictions over coronavirus, raises advisory for southkorea and italy .Ghiasi, Gita, Vincent Larivi`ere, Cassidy R Sugimoto. 2015. On the compliance of women engineers with agendered scientific system. PloS one (12) e0145931.Guarino, Cassandra M, Victor MH Borden. 2017. Faculty service loads and gender: Are women taking careof the academic family? Research in Higher Education (6) 672–694.Janssen, Christian P, Sandy JJ Gould, Simon YW Li, Duncan P Brumby, Anna L Cox. 2015. Integratingknowledge of multitasking and interruptions across different perspectives and research methods.KC, Diwas Singh. 2014. Does multitasking improve performance? evidence from the emergency department. Manufacturing & Service Operations Management (2) 168–183.KC, Diwas Singh. 2020. Heuristic thinking in patient care. Management Science (6) 2545–2563.Kelly, Jack. 2020. The massive work-from-home covid-19 test was a great success and will be the newnorm .Kitchener, Caroline. 2020. Women academics seem to be submitting fewer papers during coronavirus. neverseen anything like it, says one editor. ui, Ding, Zhu: Gender Inequality in Research Productivity ubmitting-fewer-papers-during-coronavirus-never-seen-anything-like-it-says-one-editor .Lee, Hau L, Christopher S Tang. 2018. Socially and environmentally responsible value chain innovations:New operations management research opportunities. Management Science (3) 983–996.Levitin, Daniel J. 2014. The organized mind: Thinking straight in the age of information overload . Penguin.Lundquist, Jennifer H, Joya Misra, KerryAnn OMeara. 2012. Parental leave usage by fathers and mothersat an American university.
Fathering (3) 337–363.Mark, Gloria, Daniela Gudith, Ulrich Klocke. 2008. The cost of interrupted work: more speed and stress. Proceedings of the SIGCHI conference on Human Factors in Computing Systems . 107–110.McLean, Rob. 2020. These companies plan to make working from home the new normal. as in forever .Mejia, Jorge, Chris Parker. 2020. When transparency fails: Bias and financial incentives in ridesharingplatforms.
Management Science .Minello, Alessandra. 2020. The pandemic and the female academic .Misra, Joya, Jennifer Hickes Lundquist, Abby Templer. 2012. Gender, work time, and care responsibilitiesamong faculty.
Sociological Forum , vol. 27. Wiley Online Library, 300–323.Moreno, Edward J. 2020. Government health agency official: Coronavirus ’isn’t something the americanpublic need to worry about’ https://thehill.com/homenews/sunday-talk-shows/479939-government-health-agency-official-corona-virus-isnt-something-the .Moss-Racusin, Corinne A, John F Dovidio, Victoria L Brescoll, Mark J Graham, Jo Handelsman. 2012.Science facultys subtle gender biases favor male students.
Proceedings of the national academy ofsciences (41) 16474–16479.Myers, Kyle R, Wei Yang Tham, Yian Yin, Nina Cohodes, Jerry G Thursby, Marie C Thursby, Peter ESchiffer, Joseph T Walsh, Karim R Lakhani, Dashun Wang. 2020. Quantifying the immediate effectsof the covid-19 pandemic on scientists. arXiv preprint arXiv:2005.11358 .National Conference of State Legislatures. 2020. Higher education responses to coronavirus (covid-19) .Nicklin, Jessica M, Christopher P Cerasoli, Katie L Dydyn. 2016. Telecommuting: What? why? when? andhow?
The impact of ICT on work . Springer, 41–70.Pierce, Lamar, Laura W Wang, Dennis J Zhang. 2020. Peer bargaining and productivity in teams: Genderand the inequitable division of pay.
Manufacturing & Service Operations Management .Plambeck, Erica, Kamalini Ramdas. 2020. Alleviating poverty by empowering women through business modelinnovation: M&som insights and opportunities.
Manufacturing & Service Operations Management .0 Cui, Ding, Zhu:
Gender Inequality in Research Productivity
Sarsons, Heather. 2017. Recognition for group work: Gender differences in academia.
American EconomicReview (5) 141–45.Schiebinger, Londa, Shannon K Gilmartin. 2010. Housework is an academic issue.
Academe (1) 39–44.Seamans, Robert, Feng Zhu. 2013. Responses to entry in multi-sided markets: The impact of craigslist onlocal newspapers. Management Science (2) 476–493.Song, Hummy, Anita L Tucker, Karen L Murrell, David R Vinson. 2018. Closing the productivity gap:Improving worker productivity through public relative performance feedback and validation of bestpractices. Management Science (6) 2628–2649.Staats, Bradley R, Francesca Gino. 2012. Specialization and variety in repetitive tasks: Evidence from ajapanese bank. Management science (6) 1141–1159.Tan, Tom, Serguei Netessine. 2014. When does the devil make work? An empirical study of the impact ofworkload on worker productivity. Management Science (6) 1574–1593.Tan, Tom, Serguei Netessine. 2019. When you work with a superman, will you also fly? An empirical studyof the impact of coworkers on performance. Forthcoming at Management Science.Tang, Christopher S, Sean Zhou. 2012. Research advances in environmentally and socially sustainableoperations. European Journal of Operational Research (3) 585–594.The European Commission. 2016. Women and unpaid work: recognise, reduce, redistribute! https://ec.europa.eu/social/main.jsp?catId=89&furtherNews=yes&newsId=2492&langId=en .Tucker, Anita L, Steven J Spear. 2006. Operational failures and interruptions in hospital nursing.
Healthservices research (3p1) 643–662.U.S. Bureau of Labor Statistics. 2019. Employed persons working at home, workplace, and time spentworking at each location by full- and part-time status and sex, jobholding status, and educationalattainment, 2019 annual averages . ui, Ding, Zhu: Gender Inequality in Research Productivity i Appendix
This graph plots the time trend of the number of preprints for female academics and male academics. The vertical linerepresents the placebo lockdown week (the week of March 11) in 2019. i Cui, Ding, Zhu:
Gender Inequality in Research Productivity
Table A.1 Robustness to Different University Rankings
Dependent variable: No. of preprints (in logarithm) by disciplineUniversities 6 weeks 7 weeks 8 weeks 9 weeks 10 weeksby Times ranking (1) (2) (3) (4) (5)Top 10 − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . − . ∗ p < . ∗∗ p < . ∗∗∗ p < . ui, Ding, Zhu: Gender Inequality in Research Productivity iii
Table A.2 Parallel Trends Test
No. of preprints (in logarithm) in aggregation No. of preprints (in logarithm) by disciplineVariables (1) (2)
F emale × T ime − − . − . . . F emale × T ime − − .
013 0 . . . F emale × T ime − − . − . . . F emale × T ime − .
060 0 . . . F emale × T ime − − . − . . . F emale × T ime − − . − . . . F emale × T ime − − . − . . . F emale × T ime − − . − . . . F emale × T ime − − . − . . . F emale × T ime − .
355 0 . . . F emale × T ime − .
130 0 . . . F emale × T ime − . − . . . F emale × T ime − .
069 0 . . . F emale × T ime − .
092 0 . . . R .
894 0 . × Time, in Equation (3). The coefficientsfor 6, 7, 8, 9 and 10 weeks since the lockdown are presented in columns (1)–(5), respectively. Note that weomit reporting estimates of other variables for brevity. Time fixed effects at the weekly level are included in allregressions. Significance at ∗ p < . ∗∗ p < . ∗∗∗ p < . v Cui, Ding, Zhu:
Gender Inequality in Research Productivity
Table A.3 Summary Statistics for December 2018 - May 2019
All observations Before March 2019 After March 2019Level Weekly no. of preprints Mean Std. dev Max Min Total Mean Std. dev Mean Std. devAllDisciplines(US only) All 401 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Table A.4 Falsification Test
Dependent variable: No. of preprints (in logarithm) in aggregation6 weeks 7 weeks 8 weeks 9 weeks 10 weeks(1) (2) (3) (4) (5)
F emale × Lockdown .
042 0 .
061 0 .
088 0 .
080 0 . R .
980 0 .
980 0 .
979 0 .
980 0 . F emale × Lockdown .
092 0 .
094 0 . .
085 0 . R .
877 0 .
877 0 .
871 0 .
873 0 . × Lockdown, in Equa-tion (1). The coefficients for 6, 7, 8, 9 and 10 weeks since the lockdown are presented in columns(1)–(5), respectively. Note that we omit reporting estimates of other variables for brevity. Timefixed effects at the weekly level are included in all regressions. Significance at ∗ p < . ∗∗ p < . ∗∗∗ p < . ui, Ding, Zhu: Gender Inequality in Research Productivity v Table A.5 Summary Statistics for Downloads and Abstract Views
All observations Before Lockdown After LockdownLevel Groups Mean Std. dev Min Max Mean Std. dev Mean Std. devNo. ofdownloads perpreprint All 40 . . . . . . . . . . . . . . . . . . . . . . . . . . .
67 226 . . . . . . . . . . . . . . . . . . . . . Table A.6 Impact of Lockdown on Abstract Views
Dependent variable: No. of Abstract Views (in logarithm) in aggregation6 weeks 7 weeks 8 weeks 9 weeks 10 weeksVariables (1) (2) (3) (4) (5)
F emale -0.054 − . − . − . − . F emale × Lockdown .
088 0 .
074 0 .
067 0 . R .
913 0 .
935 0 .
948 0 . ∗ p < . ∗∗ p < . ∗∗∗ p < . Table A.7 Impact of Lockdown on Downloads
Dependent variable: No. of Downloads (in logarithm) in aggregation6 weeks 7 weeks 8 weeks 9 weeks 10 weeksVariables (1) (2) (3) (4) (5)
F emale -0.044 − . − . − . − . F emale × Lockdown -0.027 − . − . − . − . R .
866 0 .
891 0 .
910 0 . ∗ p < . ∗∗ p < . ∗∗∗ p < ..