Skill-Biased Technical Change, Again? Online Gig Platforms and Local Employment

May combine two claims.

The growth of online gig platforms creates challenges for traditional incumbent industries because these platforms have better matching algorithms, flexible schedules, and lower entry barriers.

“The rise of these gig platforms has brought challenges to traditional, incumbent industries 1 due to their superior matching algorithms (Cramer and Krueger 2016), flexible work schedules (Hall and Krueger 2018), and lower entry barriers (Schwellnus et al. 2019).”

• cites Cramer and Krueger 2016 sourceSupported
• cites Hall and Krueger 2018 sourceNot checked — no full text of the source
• cites Schwellnus et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Agrawal et al. (2019) suggest that AI may replace occupations involving prediction tasks while complementing occupations requiring decision-related tasks.

“For example, Agrawal et al. (2019) suggest that AI may substitute occupations involving prediction tasks (e.g., forecasting), while complementing occupations requiring decision-related tasks (e.g., strategic planning).”

• cites Agrawal et al. 2019 sourcePartly supported

  The Conclusion says AI can substitute capital for labor in prediction tasks, and the passage on individual workers says it may diminish occupations whose core skill is prediction. It also says prediction and decision tasks are complementary, but the Conclusion cautions that AI may favor either labor or capital in decision tasks. Thus, complementing decision-related occupations is a possibility, not a general outcome established by the paper. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

SBTC theory holds that technological progress complements skilled workers by raising their relative productivity and replaces unskilled workers by automating their tasks.

“The A canonical theory in this discourse is SBTC (Acemoglu 2002, Autor et al. 2003), which posits that technological progress complements skilled workers by boosting their relative productivity and substitutes unskilled workers by automating their tasks.”

• cites Acemoglu 2002 sourceSupported
• cites Autor et al. 2003 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The widespread availability and affordability of computers has increased demand for high-skilled workers who can use these tools, a pattern known as capital-skill complementarity.

“The ubiquity and affordability of computers have also spurred the demand for high-skilled workers capable of effectively leveraging these tools, a phenomenon known as capital-skill complementarity (e.g., Krusell et al. 2000, Acemoglu 2002).”

• cites Krusell et al. 2000Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Acemoglu 2002 sourcePartly supported

  Section 1.2 cites evidence that the rapid spread of computers increased demand for skills, and Section 5.1 discusses computers as high-tech capital associated with demand for educated workers. But the copy does not establish affordability as a driver; Section 4.2 also cautions that some skill upgrading in computerized industries may reflect an existing trend rather than a direct effect of computers. No page numbers are available in the supplied passages. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

May combine two claims.

Online gig platforms lower entry barriers for workers and thereby promote self-employment opportunities.

“Third, online gig platforms lower barriers for workers to enter the job market, thus promoting selfemployment opportunities (Vallas and Schor 2020).”

• cites Vallas and Schor 2020 sourceSupported

May combine two claims.

Online gig platforms can automatically match clients and workers, streamline transactions, and manage reviews, tasks that overlap with middle-skilled workers' duties.

“Online gig platforms can automatically match clients and housekeeping workers, streamline transactions, and manage reviews (Vallas and Schor 2020), tasks that heavily overlap with the duties of incumbent middle-skilled workers.”

• cites Vallas and Schor 2020 sourcePartly supported

  In “ADVENT OF THE PLATFORM REVOLUTION” (page unavailable), the paper describes gig workers engaging with customers through platforms and describes digitally mediated transactions. In “PLATFORMS AND THE SHARING ECONOMY: AN AGENDA FOR RESEARCH” (page unavailable), it discusses customer reviews and ratings. These passages support the general platform functions, but do not establish that platforms automatically match housekeeping workers with clients or streamline their transactions. The statement about middle-skilled workers follows the citation and is not assessed as attributed to this source. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

Online gig platforms reduce search costs for housekeeping services, which may increase demand for incumbent workers.

“First, gig platforms reduce search costs for housekeeping services (e.g., Goldfarb and Tucker 2019), potentially boosting the demand for incumbent workers (Schwellnus et al. 2019).”

• cites Goldfarb and Tucker 2019 sourceSupported
• cites Schwellnus et al. 2019Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Incumbent workers may move to related occupations in other sectors that require similar skill levels.

“Another option for incumbent workers is to move to related occupations in other sectors that require similar skill levels (Moscarini and Vella 2008).”

• cites Moscarini and Vella 2008 sourcePartly supported

  The Introduction (page unavailable) discusses occupational mobility and gives an example of a worker doing the same tasks in different industries. It does not establish that workers move to related occupations in other sectors requiring similar skill levels; its cross-industry example involves the same occupation. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

General-purpose technologies typically create new occupations or move workers into other occupations.

“In terms of labor market outcomes, GPTs typically create new occupations (e.g., computer-related) or relocate the workforce to other occupations (e.g., workers in routine roles shifting to more analytic occupations) (Brynjolfsson et al. 2018).”

• cites Brynjolfsson et al. 2018 sourceCould not verify

  The Introduction (p. 3) says AI may bring fundamental changes to work, and §1.1.2 (p. 5) mentions possible new roles for radiologists and pathologists. Neither passage says general-purpose technologies typically create new occupations or move workers from routine to analytic occupations; searches for those occupational shifts found no supporting passage in our copy. (LLM, conf=0.72)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Online gig platforms can provide a stable income source for workers because of lower entry barriers and flexible schedules.

“Prior studies have demonstrated that online gig platforms can serve as a stable income source for workers due to their lower entry barriers and flexible work schedules (e.g., Schor et al. 2020).”

• cites Schor et al. 2020 sourcePartly supported

  The paper describes flexible schedules and says delivery apps can be joined without physical assets or education (A new analytic approach: platform dependence; Variation across the platforms: a vertical hierarchy; page numbers unavailable). But it does not establish that these features provide workers with stable income: it describes platform earnings as precarious, especially for workers who depend on them (Describing our respondents; Variation across the platforms: a vertical hierarchy). (LLM, conf=0.87)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

Technology displaces and reinstates labor.

“This observation is an important empirical support for the recent theoretical work in labor economics, which posits that “technology displaces and reinstates labor” (Acemoglu and Restrepo 2019).”

• cites Acemoglu and Restrepo 2019 sourceSupported

May combine two claims.

Standard DiD estimates can be problematic when earlier adopters are compared with later adopters and when treatment effects are dynamic.

“Studies suggest that the estimations may be problematic when comparing the earlier adopters with the late adopters and when there exists a dynamic treatment effect (Goodman-Bacon 2021).”

• cites Goodman-Bacon 2021 sourceSupported

Classified as probably not a substantial claim.

Recent studies recognize gig platforms as catalysts for entrepreneurial ventures.

“This perspective is supported by several recent studies that recognized gig platforms as catalysts for entrepreneurial ventures (e.g., Barrios et al. 2020, Vallas andSchor 2020).”

• cites Barrios et al. 2020 sourceNot checked — no full text of the source
• cites Vallas and Schor 2020 sourcePartly supported

  In “Incubators of Entrepreneurialism” (page unavailable), Vallas and Schor describe the view that platforms lower barriers and enable people to earn from their assets. They acknowledge some truth in that view but argue that platforms are unlikely to foster the envisioned crowd-based capitalism. The claim omits this substantial qualification. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

May combine two claims.

Wald test F-statistic equals 38.59 (p < 0.000) for the sum of coefficients.

“The F-statistic equals 38.59 (p < 0.000), suggesting the TaskRabbit effect on the supply of middle-skilled workers is negative and significant, supporting H2a (i).”

May combine two claims.

TaskRabbit entry is positively and significantly associated with incumbent middle-skilled workers becoming self-employed incorporated (0.027, p<0.05), but not self-employed unincorporated (p>0.1).

“Table 6 shows that TaskRabbit entry is positively and statistically significantly associated with incumbent middle-skilled workers becoming self-employed incorporated (0.027, p < 0.05) instead of self-employed unincorporated (p > 0.1).”

May combine two claims.

The study's empirical findings indicate decreased demand for middle-skilled workers and unchanged demand for low-skilled workers in traditional housekeeping businesses.

“Our empirical findings indicate a decreased demand for middle-skilled workers, with the demand for low-skilled workers unchanged in traditional housekeeping businesses.”

May combine two claims.

TaskRabbit entry does not significantly affect wages of managers, despite a decline in their employment.

“Interestingly, TaskRabbit entry does not significantly affect the wages of managers,foot_22 despite a decline in their employment.”

May combine two claims.

The study shows minimal impact of online gig platforms on the wages of incumbent housekeeping workers, although some subsamples show a negative wage effect on middle-skilled workers.

“Fourth, we show minimal impact of online gig platforms on the wages of incumbent housekeeping workers, though some subsamples show a negative wage effect on middle-skilled workers (Tables D4 and D18, Appendix D).”

May combine two claims.

No significant effects of TaskRabbit are observed on average annual wages of workers in skill-related occupations in other industries or self-employed workers in the housekeeping industry.

“Furthermore, we do not observe significant effects of TaskRabbit on the average annual wage of workers in skill-related occupations in other industries and self-employed workers in the housekeeping industry.”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

Probably not a substantial claim: may not pass S3 (load-bearing).

The mean number of workers in treated MSA units being treated is 15,379.

“We further calculate the mean number of workers in treated MSA units 32 being treated, which is 15,379.”

May combine two claims.

Handy has similar size, service areas, and expansion history to TaskRabbit.

“According to our analyses of the competitors of TaskRabbit (Table D12), Handy is one platform that has similar size, service areas, and expansion history as TaskRabbit.”

May combine two claims.

GSC estimates show a negative and statistically significant TaskRabbit effect on the number of housekeeping managers (β = -0.146, p < 0.01) and no significant effect on low-skilled workers (p = 0.48 > 0.1), supporting H2a.

“The effect of TaskRabbit entry on the number of housekeeping managers is negative and statistically significant (𝛽 = -0.146,p < 0.01), while effect of TaskRabbit entry on the number of low-skilled workers is not significant (p = 0.48 > 0.1), supporting our hypotheses (H2a).”

May combine two claims.

TaskRabbit entry does not significantly affect the number of workers in unemployment and skill-related occupations.

“While TaskRabbit entry does not significantly affect the number of workers in unemployment and skill-related occupations (Columns 1 and 2)”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

May combine two claims.

The mean number of middle-skilled workers in treated units post-treatment is 2,133, and the potential average without treatment is 2,844.

“the mean number of middle-skilled workers from treated units in the post-treatment period, which is 2,133 and transfers to the potential average number had each of these units not been treated 2,844”

The quote only partly states this claim.

Including Thumbtack search intensity in the model yields consistent results.

“We included the search intensities for”

May combine two claims.

The hourly wage coefficient for middle-skilled housekeeping workers is positive but not statistically significant.

“Although the coefficient for hourly wages is positive, it is not statistically significant.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The majority of TaskRabbit city entries occurred during summertime, with a few later in the year.

“We found the entry times for the majority of the cities are during the summertime, and a few of them were entered at the later time of the year.”

May combine two claims.

TaskRabbit entry is associated with higher counts of housekeeping establishments overall and in the 1-4, 5-9, 10-19, and 20-49 employee size categories, while estimated effects for larger size categories are not statistically significant.

“TaskRabbit | 0.056*** 0.079*** 0.043* 0.047** 0.039* | -0.014 | -0.024 | -0.093 | -0.076”

May combine two claims.

Pre-treatment trends in middle-skilled employment are close between treated and untreated locations, and middle-skilled employment declines significantly after TaskRabbit entry.

“As seen, pre-treatment trends in middle-skilled employment of treated and untreated locations are remarkably close to each other, while middle-skilled employment declines significantly after the TaskRabbit Entry.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Low-skilled workers such as janitors and cleaners, whose jobs involve intensive manual tasks and interpersonal interactions, are less vulnerable to automation.

“In contrast, low-skilled workers (e.g., janitors, cleaners), whose roles entail intensive manual tasks and interpersonal interactions, are less vulnerable to automation.”

May combine two claims.

Dixon et al. (2021) show that robots can replace managerial roles involving supervision and monitoring while creating new work opportunities for both high- and low-skilled workers, supporting polarized SBTC theory.

“Dixon et al. (2021) demonstrate that, while robots can replace managerial roles involving supervision and monitoring, they can simultaneously create new work opportunities for both high-and low-skilled workers, supporting the polarized version of SBTC theory.”

• cites Dixon et al. 2021 sourceNot checked — no full text of the source

May combine two claims.

Online gig platforms reduce search costs and improve matching efficiency by connecting workers with clients according to service requirements and preferences.

“Specifically, these platforms excel in connecting workers with clients based on service requirements and preferences, thus reducing search costs and enhancing matching efficiency (Schwellnus et al. 2019).”

• cites Schwellnus et al. 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The spread of workplace computers and technologies has automated routine-intensive jobs, leading to substitution of low-skilled workers.

“For example, the widespread adoption of workplace computers and technologies has led to the automation of routine-intensity jobs (e.g., cashiers, calculators), resulting in the substitution of low-skilled workers.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Flexibility is the main reason workers are drawn to online gig platforms.

“Hall and Krueger (2018) suggest that flexibility is the primary reason that workers are attracted to online gig platforms.”

• cites Hall and Krueger 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Prior research finds that technology complements low-skilled workers doing manual tasks while replacing middle-skilled workers doing routine, easily automated cognitive tasks.

“Prior research has explored technology's role in complementing low-skilled workers performing manual tasks (e.g., truck driving) while substituting middle-skilled workers engaged in routine and readily automated cognitive tasks (e.g., record keeping) (Autor and Dorn 2013).”

• cites Autor and Dorn 2013 sourceNot checked — no full text of the source

May combine two claims.

Online gig platforms may reduce the number of incumbent workers while stimulating local labor markets by moving those wage workers into self-employment.

“While online gig platforms may decrease the number of incumbent workers, they can stimulate local labor markets by redistributing these wage workers to self-employment, a novel avenue for encouraging new entrepreneurial endeavors.”

Classified as probably not a substantial claim. May combine two claims.

Skill-Biased Technical Change theory holds that technology increases relative demand for high-skilled workers by raising their productivity while replacing low-skilled workers.

“Notably, Skill-Biased Technical Change (SBTC) theory suggests that technology raises the relative demand for high-skilled workers by boosting their productivity while substituting low-skilled workers (Card andDiNardo 2002, Acemoglu andAutor 2011).”

• cites Card andDiNardo 2002Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Acemoglu andAutor 2011Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Li et al. (2021) show that online gig platforms provide flexible work for low-skilled and unemployed workers, reducing overall unemployment.

“For example, Li et al. (2021) show that online gig platforms offer flexible work opportunities for low-skilled and unemployed workers, thereby reducing overall unemployment.”

• cites Li et al. 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Online gig platforms should not be viewed as simply skill-biased; their entry produces a labor redistribution effect that moves incumbent middle-skilled workers into self-employment.

“Integrating the main employment effects of online gig platforms with the observed labor movement, our study shows that online gig platforms may not naively be viewed as skill-biased; instead, their market entry exerts a labor redistribution effect that shifts incumbent middle-skilled workers to self-employment.”

May combine two claims.

TaskRabbit is one of the earliest and largest online gig platforms that matches freelancers to local domestic housekeeping tasks.

“Second, TaskRabbit, founded in 2008, stands as one of the earliest and largest gig platforms enabling clients to outsource small offline domestic tasks (e.g., cleaning) to local independent workers (Isaac 2015).”

• cites Isaac 2015Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Burtch et al. (2018) find a negative relationship between Uber entry and local entrepreneurial activity, implying that online gig platforms provide viable work for unemployed or underemployed people.

“Connecting the proliferation of online gig platforms to job opportunities, Burtch et al. (2018) identify a negative relationship between Uber entry and local entrepreneurial activity, implying that online gig platforms offer viable work opportunities for the unemployed or underemployed.”

• cites Burtch et al. 2018 sourceNot checked — no full text of the source

May combine two claims.

The observed labor movements support the Self-employment Effect but not the Unemployment Effect or the Relocation Effect.

“These results corroborate the Self-employment Effect, but not the other two labor movements.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Autor and Dorn (2013) show that growth in high-skilled and low-skilled jobs often comes at the expense of middle-skilled workers whose tasks are codifiable, easily automated, and vulnerable to technological displacement.

“Autor and Dorn (2013) elucidate the polarized distribution of employment among skilled occupations; they show that the growth of high-skilled and low-skilled jobs often occurs at the expense of middle-skilled workers whose tasks are codifiable, easily automated, and thus susceptible to technological displacement.”

• cites Autor and Dorn 2013 sourceNot checked — no full text of the source

Classified as probably not a substantial claim. May combine two claims.

After TaskRabbit entry, middle-skilled workers could face unemployment if redundant, while low-skilled workers, whose roles do not overlap with platform functions, might see stable or increased demand.

“Post the entry of TaskRabbit, incumbent middle-skilled workers could face unemployment if deemed redundant, whereas low-skilled workers, whose roles do not overlap with the gig platform functionalities, might see stable or increased demand for their services.”

• AssumptionRests on an assumption: Online gig platforms cannot readily substitute for low-skilled workers performing in-person manual housekeeping tasks.

Classified as probably not a substantial claim.

We included MSA-specific linear and quadratic time trends to capture time-varying unobserved heterogeneity that may correlate with housekeeping demand changes.

“We also included MSA-specific linear (𝜆 !𝑡) and quadratic time trends (𝛿 !𝑡 & ) to allow for a unique trajectory of potential socio-economic and regulatory patterns within each MSA over the sample period, further capturing the time-varying unobserved heterogeneity that may correlate with housekeeping demand changes within individual MSAs.”

Classified as probably not a substantial claim.

Online gig platforms may improve operational efficiency by making middle-skilled workers obsolete when their tasks, such as scheduling and supervising, can be replaced by technology.

“Online gig platforms might enhance operational efficiency by rendering incumbent middle-skilled workers obsolete when their tasks are replaceable by technology (Autor et al. 2003), such as scheduling and supervising services.”

• cites Autor et al. 2003 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

If online gig platforms can manage low-skilled workers more effectively, demand for middle-skilled workers would decline.

“As a result, the demand for middle-skilled workers would decline if online gig platforms could more effectively manage low-skilled workers (e.g., Violante 2008).”

• cites Violante 2008 sourceNot checked — no full text of the source

May combine two claims.

Online gig platforms provide work flexibility and autonomy, letting workers choose when and where they work.

“A key advantage of online gig platforms lies in their provision of work flexibility and autonomy (Vallas andSchor 2020, Anderson et al. 2021).”

• cites Vallas andSchor 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Anderson et al. 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Standard errors were clustered at both MSA and year levels.

“We clustered standard errors at both the levels of MSA and year.22”

Classified as probably not a substantial claim. May combine two claims.

Online gig platforms like TaskRabbit streamline matching of labor supply with local service demand, potentially displacing middle-skilled managers or supervisors in traditional housekeeping companies whose functions overlap with the platforms.

“Online gig platforms, such as TaskRabbit, streamline the process of matching labor supply with local service demand, potentially displacing middle-skilled managers or supervisors in traditional housekeeping companies that share major functional overlaps with these online gig platforms.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The IPUMS dataset has been widely used in labor economics and IS literature.

“This dataset has widely been used in labor economics (e.g., Kerr and Lincoln 2010) and IS literature (e.g., Burtch et al. 2018).”

• cites Kerr and Lincoln 2010Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Burtch et al. 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

Consequently, displaced middle-skilled workers may opt for freelancer work to secure reliable income.

“Consequently, these workers may opt for freelancer work to secure a reliable income.”

Probably not a substantial claim: may not pass S3 (load-bearing).

After ride-sharing was introduced in New York, demand for taxi drivers rose because clients could more easily find a taxi.

“Nandakumar (2020) notes that following the introduction of ride-sharing services in New York, the demand for taxi drivers increased as it became easier for clients to find a taxi.”

• cites Nandakumar 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

MSAs were used as geographic units because TaskRabbit typically entered a broad region rather than a single city, and labor usually moves within such a region.

“We used MSAs as the geographic units for analysis because TaskRabbit typically entered a broad region (e.g., LA metro), instead of a single city, and labor usually moves within such a region (i.e., MSA) that consists of a city and surrounding communities linked by social and economic factors.”

Probably not a substantial claim: may not pass S4 (based).

Online gig platforms could counteract the surge by replacing a large portion of middle-skilled workers' jobs.

“Yet, online gig platforms could counteract this surge by replacing a large portion of middle-skilled workers' jobs.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The CBP dataset provides annual subnational economic data by industry and MSA and has been widely used in IS research.

“The CBP dataset provides annual subnational economic data by industry and MSA, and it has widely been used in extant IS research (e.g., Kim and Hann 2019).”

• cites Kim and Hann 2019 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Following existing studies, we included covariates that potentially influence TaskRabbit entry and local housekeeping employment.

“Following existing studies on local employment and online gig platforms (e.g., Berger et al. 2018), we included several groups of covariates that potentially influence TaskRabbit entry and local housekeeping employment.”

• cites Berger et al. 2018 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

TaskRabbit entry might shift middle-skilled workers, rather than low-skilled ones, to unemployment.

“Thus, it is possible that TaskRabbit's entry might shift middle-skilled workers, rather than low-skilled ones, to unemployment.”

Probably not a substantial claim: may not pass S4 (based).

A surge in low-skilled workers may lead to gradually growing demand for middle-skilled workers to oversee and facilitate matching.

“With a substantial surge in low-skilled workers, a gradually growing demand for middle-skilled workers may follow to oversee and facilitate the matching of low-skilled workers.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

As gig platforms enhance operational efficiencies, demand for middle-skilled workers in traditional roles may wane, potentially displacing them.

“First, as gig platforms enhance operational efficiencies for companies, the demand for middle-skilled workers in traditional roles may wane, potentially displacing them from incumbent employment.”

Classified as probably not a substantial claim. May combine two claims.

Online gig platforms may be more biased against, and may substitute for, middle-skilled workers than low-skilled workers.

“Hence, online gig platforms may exhibit a greater bias against, and even substitute, incumbent middle-skilled workers compared to low-skilled workers.”

• AssumptionRests on an assumption: Online gig platforms cannot readily substitute for low-skilled workers performing in-person manual housekeeping tasks.

May combine two claims.

TaskRabbit effect on supply of middle-skilled workers is negative and significant, supporting H2a(i).

“The F-statistic equals 38.59 (p < 0.000), suggesting the TaskRabbit effect on the supply of middle-skilled workers is negative and significant, supporting H2a (i).”

Classified as probably not a substantial claim.

Prior research suggests workers are more likely to shift to related occupations with similar job skills, minimizing retraining needs.

“Prior research suggests that workers are more likely to shift to related occupations that share similar job skills, minimizing the need for extensive retraining (Robinson 2018, Cheng andPark 2020).”

• cites Robinson 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Cheng andPark 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Studies indicate that automation can make middle-skilled managerial jobs obsolete because technology often performs these functions more cost-effectively.

“Studies indicate that automation can render such jobs obsolete (Ford 2015, Casey 2018) as technology often performs these functions more cost-effectively.”

• cites Ford 2015Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Casey 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Dixon et al. (2021) suggest that robot adoption increases demand for low-skilled workers who perform residual tasks not yet automated.

“For instance, Dixon et al. (2021) suggest the adoption of robots leads to an increased demand for low-skilled workers capable of performing residual tasks that robots have not yet automated.”

• cites Dixon et al. (2021) sourceNot checked — no full text of the source

May combine two claims.

Although rising industry demand can increase overall supply of housekeeping workers, middle-skilled workers are arguably less affected than low-skilled workers.

“Consequently, while rising industry demand can augment the overall supply of housekeeping workers, middle-skilled workers are arguably less affected than low-skilled workers.”

May combine two claims.

Barrios et al. (2020) illustrate that ride-sharing platforms have spurred increased business registrations and improved loan accessibility by providing a stable income source for self-employed drivers.

“For instance, Barrios et al. (2020) illustrate how ride-sharing platforms have spurred increased business registrations and improved loan accessibility by providing a stable income source for self-employed drivers.”

• cites Barrios et al. (2020) sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

We employed a staggered Difference-in-Differences framework with two-way fixed effects, following several IS studies.

“We employed a staggered Difference-in-Differences (DiD) framework with two-way fixed effects to estimate the impact of TaskRabbit entry on local incumbent housekeeping employment (H1), following several IS studies on the impact of online gig platforms (e.g., Burtch et al. 2018, Brynjolfsson et al. 2019).”

• cites Burtch et al. 2018 sourceNot checked — no full text of the source
• cites Brynjolfsson et al. 2019 sourceNot checked — no full text of the source

May combine two claims.

The results support the hypothesis that TaskRabbit entry negatively affects the number of incumbent housekeeping workers, but not the hypothesis about wages.

“The results support H1a (i) but not H1a (ii).”

Probably not a substantial claim: may not pass S1 (asserted).

With platforms like TaskRabbit, middle-skilled housekeeping workers may leave their roles and enter other skill-related non-housekeeping occupations requiring similar management and communication skills.

“In our context, with the emergence of platforms like TaskRabbit, middle-skilled housekeeping workers may leave their current roles and enter other skill-related (but non-housekeeping) occupations, such as those that require similar levels of management and communication skills (Tables C1 and C2 in Appendix C).”

May combine two claims.

Online gig platforms may not simply show bias against certain low-skilled occupations; they may redistribute workers across employment modes.

“Against this backdrop, we argue that online gig platforms may not simply exhibit bias against certain (low-skilled) occupations, but they may redistribute workers across employment modes.”

Classified as probably not a substantial claim.

Employment growth for middle-skilled workers may be slower because their job functions overlap with online gig platforms.

“Second, the increase in middle-skilled worker employment might be slower due to their job functions overlapping with those of online gig platforms.”

May combine two claims.

Prior studies on online gig platforms' labor effects have presented competing theories and relied mostly on anecdotal evidence.

“Extant studies have debated the role of these platforms on local labor markets, often presenting competing theories and relying mostly on anecdotal evidence (Fraiberger and Sundararajan 2017, Schor 2017, Schwellnus et al. 2019).”

• cites Fraiberger and Sundararajan 2017Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Schor 2017Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Schwellnus et al. 2019 sourceNot checked — no full text of the source

May combine two claims.

The empirical exploration challenges a substitution explanation often posited by critics of online gig platforms and supports a labor redistribution explanation in which the rise of online gig platforms shifts middle-skilled workers in traditional businesses to self-employment.

“Our empirical exploration challenges a substitution explanation often posited by critics of online gig platforms, but rather it supports a labor redistribution explanation-the rise of online gig platforms shifts middle-skilled workers in traditional businesses to self-employment.”

Classified as probably not a substantial claim.

To address the non-randomness of TaskRabbit entry, the main analysis included a comprehensive set of covariates to account for potential confounders.

“To address the non-randomness of TaskRabbit's entry, we included a comprehensive set of covariates in our main analysis to account for potential confounders.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The analysis is limited to two major gig platforms, TaskRabbit and Handy.

“We acknowledge the existence of other gig platforms in similar markets, but our analysis is limited to two major ones (TaskRabbit and Handy).”

May combine two claims.

Online gig platforms not only streamline business operations by matching and supervising, but also give middle-skilled workers in traditional companies opportunities to start their own ventures and participate in the gig economy.

“For practitioners, our findings imply that online gig platforms not only streamline business operations by matching and supervising, but also provide middle-skilled workers in traditional companies with opportunities to start their own ventures and participate in the gig economy.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The study uses only one online gig platform, TaskRabbit, as its empirical focus.

“Second, we acknowledge that we use only one online gig platform, TaskRabbit, as our empirical focus to illustrate the employment effects of gig platforms on incumbent housekeeping workers, although other platforms exist.”

May combine two claims.

Extant studies suggest self-employed incorporated workers are more likely entrepreneurs starting businesses, while self-employed unincorporated workers are more like freelancers or independent contractors.

“Extant studies suggest that self-employed incorporated workers are more likely to be entrepreneurs who start their own businesses, while self-employed unincorporated workers are more like freelancers or independent contractors who work on their own (Rubinstein and Levine 2020).”

• cites Rubinstein and Levine 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

The study explores only one subgroup of service occupations: housekeeping occupations.

“Third, while online gig platforms can affect a broader range of service occupations, our study only explores one subgroup of them-housekeeping occupations.”

May combine two claims.

Further in-depth analysis reveals that online gig platforms do not completely replace jobs of incumbent middle-skilled housekeeping workers but instead redistribute these workers to self-employment, stimulating local entrepreneurial endeavors.

“Our further in-depth analysis reveals that online gig platforms do not completely replace jobs of incumbent middle-skilled housekeeping workers, but rather they redistribute these workers to self-employment, stimulating local entrepreneurial endeavors.”

May combine two claims.

Online gig platforms do not necessarily replace middle-skilled workers but instead create opportunities for them, driving them toward local self-employment activities.

“Upon further exploration, our findings reveal that online gig platforms do not necessarily replace middleskilled workers but instead create job opportunities for them, driving them toward local self-employment activities.”

May combine two claims.

Past housekeeping employment and wage trends did not predict TaskRabbit entry, which suggests reverse causality is not a major concern.

“The results show that past employment and wage trends do not predict TaskRabbit entry (see Table D2, Appendix D), suggesting that reverse causality may not be a serious concern in our study.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Due to data limitations, the study cannot track how many self-employed workers use TaskRabbit for their businesses.

“Due to data limitations, we cannot track how many self-employed workers use TaskRabbit for their businesses.”

May combine two claims. The quote only partly states this claim.

Across 1,000 random placebo assignments, the mean placebo treatment effect was not different from zero and not significantly different from the DiD estimates, indicating the observed TaskRabbit effects are not spurious.

“found the mean of placebo treatment effects does not deviate from zero, let alone significantly different from our DiD estimates in Table 4, indicating that the observed effects of TaskRabbit entry were not spurious.”

May combine two claims.

Online gig platforms have substantially reshaped U.S. labor markets by lowering search costs and giving workers flexibility and autonomy.

“Online gig platforms have substantially reshaped the U.S. labor markets by reducing search costs (Chen andHorton 2016, Goldfarb andTucker 2019) and granting workers flexibility and autonomy (e.g., Burtch et al. 2018).”

• cites Chen andHorton 2016Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Goldfarb andTucker 2019Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Burtch et al. 2018 sourceNot checked — no full text of the source

May combine two claims.

This corroborates its role in improving operational efficiency and reducing reliance on these workers.

“Additional analysis shows that TaskRabbit's entry significantly reduces working hours for middle-skilled workers (Table E2, Appendix E), further corroborating its role in improving operational efficiency and reducing reliance on these workers.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The observed effects may apply to other service occupations, such as moving services, that share similar tasks with housekeeping occupations.

“We argue that the observed effects may apply to other service occupations (e.g., moving services) that share similar tasks with housekeeping occupations.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The study could not account for every platform because of data constraints.

“Nevertheless, we could not account for every platform due to data constraints.”

May combine two claims.

Low-skilled workers may not transition because online gig platforms cannot directly replace them and they have little incentive to become self-employed.

“Our additional findings suggest that low-skilled workers may not transition because online gig platforms cannot directly replace low-skilled workers, and these workers have little incentive to move to self-employment.”

May combine two claims.

Technology substitutes routine tasks but complements complex manual tasks that are not easily automated.

“This aligns with recent developments in STBC theory, particularly the polarized employment in service occupations (Autor and Dorn 2013), which suggests that technology substitutes routine tasks but complements complex manual tasks that are not easily automated.”

• cites Autor and Dorn 2013 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The findings are specific to new businesses in the housekeeping industry and may not apply to other industries.

“Finally, our findings are specific to new businesses in the housekeeping industry and may not apply to other industries (e.g., new technology).”

May combine two claims.

Initial evidence suggests gig platforms are skill-biased by substituting incumbent middle-skilled workers whose routine tasks heavily overlap with services offered by gig platforms, while having little impact on low-skilled workers who perform manual, labor-intensive tasks.

“Our initial evidence suggests that gig platforms are “skill-biased,” by substituting incumbent middle-skilled workers whose routine tasks heavily overlap with the services offered by gig platforms, while having little impact on low-skilled workers who perform manual, labor-intensive tasks.”

May combine two claims.

Online gig platforms may facilitate the redistribution of middle-skilled labor toward self-employment, thereby fostering local entrepreneurial activities.

“Instead, they may facilitate the redistribution of the middle-skilled labor force toward self-employment, thereby fostering local entrepreneurial activities.”

May combine two claims.

Subsample robustness checks using coarsened exact matching, treated-only MSAs, top 50 MSAs, and excluding post-2017 samples all produced results consistent with the baseline DiD estimates.

“In each of the above cases, results remained consistent with our baseline DiD estimates.”

May combine two claims.

Aggregate-level results are highly consistent with individual-worker data and further corroborate the self-employment effect (H3c).

“The results from the aggregate level analysis are highly consistent with those using the individual worker data, further corroborating the hypothesized self-employment effect (H3c).”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

Probably not a substantial claim: may not pass S3 (load-bearing).

County Business Patterns data have been widely used in prior studies to examine or control local establishment counts by industry.

“County Business Patterns have been widely used in prior studies to examine or control the number of local establishments by industry.”

• cites Greenwood and Gopal (2015) sourceNot checked — no full text of the source
• cites Kim and Hann (2019) sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Skill percentile is measured by the U.S. mean occupational wage in 1980, following Autor and Dorn (2013).

“PerAutor and Dorn (2013), the skill percentile is measured by the U.S. mean occupational wage in 1980.”

• cites Autor and Dorn (2013) sourceNot checked — no full text of the source

May combine two claims. The quote only partly states this claim.

The negative wage effect in the building and dwelling industry subsample supports the idea that demand for incumbent middle-skilled workers declined.

“These results bolster the notion of a diminished demand for incumbent middle-skilled workers.”

May combine two claims.

In the subsample of only treated MSAs, the significant negative effects of TaskRabbit remain, supporting the baseline DiD estimate.

“The significant negative effects of TaskRabbit remain, further supporting the baseline DiD estimate.”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

May combine two claims.

Handy entry has similar negative effects on middle-skilled workers, corroborating the generalizability of the main TaskRabbit findings.

“We found similar negative effects of the entry of Handy on middle-skilled workers, corroborating the generalizability of our main findings from TaskRabbit.”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Routine-intensive occupations tend to fall in the middle of the skill distribution, while manual-intensive occupations tend to be at the bottom.

“Routine-intensive occupations (e.g., first-line managers) tend to fall in the middle of the distribution (thereby being classified as the 'middleskilled'), while manual-intensive occupations (e.g., cleaners) tend to be at the bottom of the distribution (thereby 'low-skilled').”

• cites Autor and Dorn (2013) sourceNot checked — no full text of the source

May combine two claims.

The synthetic control model constructs a weighted combination of untreated MSAs that resembles treated MSAs in pre-treatment covariates and outcomes, satisfying the parallel trend assumption.

“The synthetic control model constructs a weighted combination of untreated MSAs (i.e., synthetic controls) that closely resembles the covariates and outcome of the treated MSAs in the pre-treatment periods, which naturally satisfies the parallel trend assumption.”

• cites Abadie andGardeazabal 2003Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Abadie et al. 2010 sourceNot checked — no full text of the source
• cites Bai 2009Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Bacon decomposition shows negative TaskRabbit effects on housekeeping worker numbers for all three groups, suggesting the main negative effects are not driven by the timing-varying group.

“As seen, the effects of TaskRabbit on the number of housekeeping workers are negative for all three groups, suggesting that main negative effects we observed are not driven by the timing-varying group.”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

The quote only partly states this claim.

The pre-treatment trend results support the validity of the baseline DiD estimates.

“These results support the validity of the baseline DiD estimates.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Adjusting the treatment variable for TaskRabbit entry months in October, November, or December by recoding treatment to 1 for the next calendar year yields consistent results.

“We also adjusted the measurement for the treatment variable based on the specific TaskRabbit entry month of a year (i.e., if the entry month is October, November, or December, we recoded the treatment equal to 1 for the next calendar year).The results are consistent and are available in online Appendix D.”

May combine two claims.

Past employment numbers and average wages do not statistically significantly predict TaskRabbit entry, which precludes the reverse causality concern.

“As seen, the number and average wage of workers in the past years do not statistically significantly predict TaskRabbit entry, precluding the concern of reverse causality.”

May combine two claims.

The permutation test's pseudo effects are centered around zero, and the t-test fails to reject the null that the average random-treatment effect differs from zero.

“Figure D1 shows the distribution of the pseudo effects, which are centered around zero, and the t-test fails to reject the null hypothesis that the average estimated effect based on random treatments is statistically different from zero.”

May combine two claims.

The heterogeneity-robust DiD model still shows significant effects for middle-skilled workers and no effect for low-skilled workers.

“As seen in Table D10, we still observe significant effects for middle-skilled workers while no effect for low-skilled workers.”

• AssumptionThe authors state this as an assumption: The difference-in-differences estimates are valid only if treated and untreated MSAs would have followed parallel trends absent TaskRabbit entry.

May combine two claims.

Thumbtack's role in matching is less direct than TaskRabbit's, which may limit the strength of its impact.

“These could suggest a less direct role in matching compared to TaskRabbit, which may limit the strength of its impact.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A Pew Research survey indicates that about 70% of gig platform workers keep their current jobs and use online gig platforms as side jobs.

“For instance, a Pew Research survey indicates that around 70% of gig platform workers maintain their current jobs and use online gig platforms as side jobs.”

• cites Pew Research surveyNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

TaskRabbit use improves operational efficiency in housekeeping firms, reducing the need for middle-skilled workers to spend hours on tasks that the platform also handles.

“These results support the notion that the use of TaskRabbit enhances operational efficiency in housekeeping companies, consequently reducing the need for middle-skilled workers spending their working hours on functions (e.g., matching and supervising) overlapped with those facilitated by the online gig platform.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The company-size results complement the proposed mechanism.

“which further complement our proposed mechanism.”

Probably not a substantial claim: may not pass S3 (load-bearing).

CEM has been widely used to address selection bias in IS studies.

“CEM has widely been used to address selection bias in IS studies (e.g., Bapna et al. 2016).”

• cites Bapna et al. 2016 sourceNot checked — no full text of the source