Published Work
Sarah Bana, Kelly Bedard, Maya Rossin-Slater, Jenna Stearns. 2022. “Unequal use of social insurance benefits: The role of employers.” Journal of Econometrics, https://doi.org/10.1016/j.jeconom.2022.02.008
Published PDF
Bana, Sarah H., Bedard, Kelly and Rossin‐Slater, Maya. 2020. “The Impacts of Paid Family Leave Benefits: Regression Kink Evidence from California Administrative Data.” Journal of Policy Analysis and Management, doi:10.1002/pam.22242
Published PDF
Bana, Sarah H., Benzell, Seth G. and Solares, Rodrigo Razo. 2020. “Ranking How National Economies Adapt to Remote Work.” Sloan Management Review.
Media Mentions: BBC Business Daily—Homeworking’s Winners and Losers, Valor Economico—Brasil é o quinto país com maior dificuldade para o home office
Working Papers
work2vec: Using Language Models to Understand Wage Premia
Stanford HAI article featuring the research
Does the text content of a job posting predict the salary offered for the role? There is ample evidence that even within an occupation, a job's skills and tasks affect the job's salary. Capturing this fine-grained information from postings can provide real-time insights on prices of various job characteristics. Using a new dataset from Greenwich.HR with salary information linked to posting data from Burning Glass Technologies, I apply natural language processing (NLP) techniques to build a model that predicts salaries from job posting text. This follows the rich tradition in the economics literature of estimating wage premia for various job characteristics by applying hedonic regression. My model explains 83 percent of the variation in salaries, 19 percent (13 percentage points) over a model with occupation by location fixed effects. Using an attribution method called integrated gradients, I decompose these elements into locations, job titles, experience levels, education levels, skills, activities, and firm names. This decomposition demonstrates the relative contribution of each of these factors to earnings.
Investment in Cybersecurity Talent: Evidence from Firms’ Responses to Data Breaches
Available at SSRN
with Erik Brynjolfsson, Wang Jin, Sebastian Steffen, and Xiupeng Wang
Do firms react to data breaches by investing in cybersecurity talent? We assemble a unique dataset on firm responses from the last decade, combining data breach information with detailed firm-level hiring data from online job postings. Using a difference-in-differences design, we find that firms indeed increase their hiring for cybersecurity workers. While this effect is statistically significant, the economic magnitude is small, which is consistent with firms' lack of incentives to improve their cybersecurity infrastructure. Further, we collect data from the MIT MediaCloud and Google Trends to measure media and public attention following breach events. We find that firms with greater media and search attention after a breach are three times as likely to post a cybersecurity job. With an increase in both the value of data as well as the number of cyberattacks, our research provides important insight into how media coverage and public attention can provide proper incentives for firms to make substantive IT investments to safeguard their customer data.
Connecting Higher Education to Career Skills
with Hung Kim Chau, Morgan Frank, and Baptiste Bouvier
[Draft available upon request]
Slides for ASSA 2023
Colleges and universities provide essential skills for students entering the workforce. However, little systematic research has been done on what is being taught, where, and to whom. This paper uses a novel dataset of 10 million course syllabi to answer this question. Syllabi provide detailed information on skills taught to students though they have not been the subject of broader analyses because of the challenge in collecting and analyzing such data. Using natural language processing techniques, we apply word embeddings to represent each detailed work activity (DWA), the atomic unit of labor, in a syllabus. Then, we compute similarities between DWAs and syllabi. This allows us to differentiate fields of study and universities based on the content of their courses, and ask a multitude of questions regarding differences over time and across universities. In particular, we find that DWAs that are taught together across all fields of study improve predictions of the topics that will become relevant to a particular field in the future. We find that, overall, syllabi reassuringly mirror pre-existing constructs of university similarity (e.g. the Carnegie classification, and self-reported peers), though there is substantial variation in what is being taught at universities. Combining skills data with earnings outcomes from the Post-Secondary Employment Outcomes (PSEO), we look at differences in skills within and across fields of study and how those skills affect earnings after graduation. Finally, we show that skills associated with high earnings are not taught equally: syllabi data provide one mechanism for universities reinforcing inequality.
Identifying Vulnerable Displaced Workers: The Effect of State-Level Occupation Conditions
Which attribute of a worker's job — their industry or their occupation — plays a larger role in determining future labor market outcomes? Understanding the dominant attribute and their relative weights allows policymakers and researchers to more accurately measure potential exposure to labor market shocks, and to target the relevant populations with interventions. Yet limited government measurement of short-term occupation level employment has inhibited such a comparison. In this paper, I derive a measure of short-term occupation conditions in a worker's state using a shift-share approach. This measure facilitates a comparison between vulnerability to industry conditions and vulnerability to occupation conditions. I estimate the effect of these conditions on displaced workers' labor market outcomes. While both state-level industry and occupation conditions appear to affect displaced workers' labor market outcomes, variation in occupation conditions completely explains the relationship between industry conditions and subsequent outcomes. This implies that the dominant worker attribute is their occupation, and suggests that large negative shocks to occupation-level employment have major labor market consequences for those workers.
Media Coverage: Wall Street Journal - Real Time Economics Blog
work2vec: Learning the Latent Structure of the Labor Market
[New results forthcoming!]
with Erik Brynjolfsson, Daniel Rock, and Sebastian Steffen
Slides ASSA 2023
Job postings provide unique insights about the demand for skills, tasks, and occupations. Using the full text of data from millions of online job postings, we train and evaluate a natural language processing (NLP) model with over 100 million parameters to classify job postings' occupation labels and salaries. To derive additional insights from the model, we develop a method of injecting deliberately constructed text snippets reflecting occupational content into postings. We apply this text injection technique to understand the returns to several information technology skills including machine learning itself. We further extract measurements of the topology of the labor market, building a “jobspace” using the relationships learned in the text structure. Our measurements of the jobspace imply expansion of the types of work available in the U.S. labor market from 2010 to 2019. We also demonstrate that this technique can be used to construct indices of occupational technology exposure with an application to remote work. Moreover, our analysis shows that data-driven hierarchical taxonomies can be constructed from job postings to augment existing occupational taxonomies like the SOC (Standard Occupational Classification) system. Exploring further the model structure, we find that between 2010 and 2019, occupations have become increasingly distinct from each other in their language, suggesting a rise in specialization of tasks in the economy. This trend is strongest for managerial, computer science, and sales occupations.
Works in Progress
Estimating the Cost of Advance Notice for Firms Conducting Mass Layoffs
with Jacob Morris
We estimate the cost of advance notice to firms conducting mass layoffs in Ontario, Canada. Using a quasi-regression discontinuity design where firms can manipulate advance notice lengths by laying off fewer workers, we find that advance notice has a significant cost: firms are willing to strategically lay off fewer workers in order to decrease the notice they give to the remaining workers. 8 percent of layoff events bunch below a threshold to avoid giving an additional four weeks of notice.
worldSML
with Erik Brynjolfsson, Tom Mitchell, Daniel Rock, Morgan Frank, and Iyad Rahwan
Using occupational distributions of fifty countries, we describe country-level suitability for machine learning.