Communicating Corporate Culture in Labor Markets: Evidence from Job Postings

1. Introduction

Corporate culture is a critical attribute when choosing a job.¹ A longstanding literature in economics suggests that nonwage attributes, such as culture, can influence job searches (e.g., Mas and Pallais 2017, Maestas et al. 2023). Job seekers have increasingly emphasized culture in recent years, with some evidence suggesting that they now consider culture to be as important as salary when evaluating job prospects (Estrada 2020). Despite its importance, culture remains difficult to observe, resulting in significant information frictions and search costs for job seekers (Guiso et al. 2008, 2015b; Grennan 2019; Graham et al. 2022). We examine whether corporate job vacancy postings emphasizing information about core cultural values alleviate such frictions and improve hiring outcomes.

We argue that job postings with culture information help companies better identify and attract job seekers with similar cultural preferences. This prediction builds on two assumptions. First, based on prior research, we assume that employees value culture (Sheridan 1992, Glassdoor 2018). Second, job postings can provide job seekers with information that is more relevant, accessible, and salient than information provided elsewhere, such as corporate websites and job review sites.² Culture information from these sources may be too general and may not apply to the position of interest, which resides within a business unit or location with its own subculture.³ In contrast, job postings contain relevant culture information for specific positions. Moreover, job seekers relying on external information sources bear additional search costs, as they need to collect and assimilate information from multiple sources while evaluating different job prospects.⁴

Our prediction is not without tension. For instance, job seekers may instead focus primarily on advancement opportunities, specific job requirements (e.g., skillsets), or salary, which firms have been under increasing regulatory scrutiny to disclose (Liu 2022). This focus can render culture information less important in the job search (deHaan et al. 2023). In addition, job seekers may view noncompany sources, such as Glassdoor or informal networks, as more informative, reducing the role of job postings as a primary source. Ultimately, the extent to which the level of culture information in job postings is associated with hiring outcomes warrants an empirical investigation.

We focus on five cultural values that are most frequently present in corporate websites and official value statements (Guiso et al. 2015b, Sull et al. 2020): integrity, teamwork, innovation, respect, and quality. We adopt a semisupervised machine learning approach from Li et al. (2021) to measure the extent to which job postings discuss these values. Specifically, we train a neural network model using approximately 12 million job postings from 2010–2020 to learn the meanings of words and phrases in the context of job postings and use the model’s output to construct a dictionary containing the words most relevant to each cultural value. We then follow prior studies (e.g., Li et al. 2021, Graham et al. 2022) and calculate a composite score (disclosed culture) that measures the degree to which job postings broadly emphasize culture (across multiple dimensions), or the salience and intensity of culture language in the job posting.

In practice, culture varies within a company and across teams, and hiring managers write job postings to reflect the needs and cultural fit for their teams (Castilla and Rho 2023). To provide context on this, we begin by analyzing sources of variation in disclosed culture. We find that, although firm fixed effects explain a sizeable portion of the variation in our culture measure (31.7%), the majority remains unaccounted for. These statistics support evidence from practice regarding job postings varying within firms. When including firm and job characteristics as determinants, we find that culture is more likely to be emphasized in roles requiring greater experience or managerial or interpersonal skills, and in periods of high job demand. By contrast, culture information is less prevalent in technical roles, those for which employers prioritize skills, and when firms rely more on compensation as a recruitment tool. Our subsequent analyses control for these characteristics and include a comprehensive set of fixed effects to help isolate the effects of culture information.

Our main analyses examine the relation between disclosed culture and employee inflows. We obtain detailed data on worker inflows from Revelio Labs. Our baseline results show a strong positive relation between disclosed culture and subsequent worker inflows at the firm-Metropolitan Statistical Area (MSA)-quarter level. The results are economically meaningful: a one within-fixed-effect standard deviation increase in disclosed culture is associated with a 9.7% increase in employee inflow, relative to the sample median. In additional analyses, we corroborate our main findings via two alternative hiring outcome measures. We show that disclosed culture shortens vacancy durations, suggesting quicker identification of suitable candidates. At the same time, more salient culture disclosures are associated with lower subsequent outflows, consistent with new hires having better cultural alignment with the company.

Next, we explore the mechanisms underlying this effect. We begin by conducting two cross-sectional analyses to examine predictable variation in the effect of disclosed culture on inflows. The first set of analyses examines how much job seekers value culture as reflected in job features and labor market conditions. We find that disclosed culture is more strongly associated with employee inflows in creative roles but that its effect weakens for entry-level positions. These results suggest that culture information is more effective when job seekers place greater emphasis on cultural fit but plays a limited role when they focus primarily on career opportunities. In addition, we find the effect of culture information weakens when the unemployment rate is high and when firms have fewer openings, consistent with culture being particularly valuable when job seekers face less competition in the labor market.

Our second set of analyses examines the role of information asymmetry, as our predictions assume that job seekers have limited information about culture. We thus expect the effect of disclosed culture to weaken when job seekers have access to more and better alternative sources of information about culture. Consistent with this argument, we find that our results are less pronounced among larger facilities (with presumably better information environments) and for firms with more high-quality reviews available on job review sites. Overall, our cross-sectional results suggest that disclosed culture matters more when job seekers value cultural fit and less when they have access to alternative sources of culture information.

Next, we assess the cultural fit channel, which suggests that culture information helps job seekers sort into jobs that better fit their preferences. Assessing cultural fit poses an empirical challenge as we cannot observe job seekers’ preferences. To circumvent this challenge, we perform two sets of analyses. First, we examine scenarios that may heighten job seekers’ preferences for certain cultural values. The first test uses the Black Lives Matter (BLM) movement as a shock that increases the weight that job seekers place on the respect dimension of culture (Li et al. 2021). The second test exploits time-varying patterns of corporate misconduct in the industry as settings in which job seekers will emphasize the integrity dimension of culture. Consistent with expectations, we find that job postings containing respect-related information have a stronger effect on worker inflows following BLM and those containing integrity-related information have a similar effect when industries have experienced more significant violations in recent periods.

Second, we further assess the cultural fit channel by examining job candidates’ reviews of the interview process on Glassdoor. If culture information facilitates cultural fit, we expect firms providing this information to have higher offer rates to candidates and candidates to have higher acceptance rates, as job seekers apply to better-fitting positions. We find results consistent with these predictions, indicating that culture information fosters cultural fit and, consequently, better matches between firms and job seekers.

Finally, we conduct a battery of robustness tests that help strengthen our claim that culture information influences worker inflows. For example, we include additional controls such as a firm-level culture measure based on earnings calls (Li et al. 2021) and the education and experience of incoming workers, use alternative scalers, and ascertain the robustness of our findings across different subsamples. Our results persist across all settings.

Our study contributes to the literature across several dimensions. First, we extend the academic literature on corporate culture (e.g., Guiso et al. 2015a, b; Grennan 2019; Pacelli 2019; Dikolli et al. 2020; Li et al. 2021, 2025; Graham et al. 2022) by providing large-scale evidence on culture information in job postings. We show that, unlike culture information in conference calls, corporate websites, or regulatory violations, culture information in job postings varies within firms, likely reflecting the subcultures of business units. Our finding that this information influences hiring outcomes extends recent research on job postings as informative disclosures with labor and capital market implications (Sran 2025). As such, we also contribute to the nascent literature on labor and accounting (Barrios 2022, Choi et al. 2023, Hann et al. 2024, Ham et al. 2025) and complement the longstanding literature on the role of nonwage attributes in labor market sorting (e.g., Sullivan and To 2014, Mas and Pallais 2017, Hall and Mueller 2018, Maestas et al. 2023). Relatedly, our finding that job seekers’ cultural preferences shape their responses to job postings also contributes to the behavioral economics of accounting (Hanlon et al. 2022), reinforcing the growing evidence that individual preferences and work environment characteristics are important determinants of economic outcomes.

Second, we extend studies examining the value of information that results from job searches (e.g., Hales et al. 2018, Huang et al. 2020, Dube and Zhu 2021, Campbell and Shang 2022). These studies primarily focus on information from employee review platforms such as Glassdoor and Indeed, and how that information affects firm behavior and relates to capital market outcomes. We extend this literature by examining how job postings alleviate information frictions related to job searches. In doing so, we also build on the nascent line of accounting research examining employee job search costs (Choi et al. 2023, deHaan et al. 2023). Our results are consistent with culture information in job postings reducing information asymmetry regarding firm core values and job search costs for employees.

More broadly, our focus on culture disclosures is relevant in light of recent Securities and Exchange Commission (SEC) initiatives requiring firms to provide more detailed descriptions of their human capital. Early evidence from the human capital disclosure mandate suggests that firms recognize the importance of corporate culture, as nearly 20% of them discuss this issue (e.g., Batish et al. 2021). Our paper complements recent studies on human capital disclosure (e.g., Bourveau et al. 2022, Demers et al. 2022, Mayew and Zhang 2022) by exploring the disclosure of human capital factors in unconventional disclosure platforms such as job postings.

2. Prior Literature and Hypothesis Development

2.1. Prior Literature

Our objective is to show how culture information in job postings influences firms’ hiring outcomes. Broadly speaking, prior studies define corporate culture as an intangible asset that represents the shared assumptions, values, and beliefs that inform employee behavior (Kreps 1990, Schein 1990, O’Reilly and Chatman 1996). Corporate culture is difficult to measure or describe, particularly for firm outsiders.

Prior studies have established a variety of frameworks and methodologies for examining and measuring culture. One line of research infers corporate culture based on the behavior or attributes of executives or key individuals in the firm (e.g., Biggerstaff et al. 2015, Davidson et al. 2015, Liu 2016, Pan et al. 2017, Merkley et al. 2020). Other studies attempt to infer a firm’s culture based on associations or patterns that cannot otherwise be explained by common observables (e.g., Cartwright and Cooper 1993, Cronqvist et al. 2009, Fahlenbrach et al. 2012, Pacelli 2019). Recent work has further attempted to measure specific dimensions of corporate culture, going beyond a firm-level composite measure (e.g., Guiso et al. 2015b, Li et al. 2021). Our study builds on this literature and adopts a multidimensional view of corporate culture, following Guiso et al. (2015b) and Li et al. (2021).

Research supports the notion that corporate culture matters for both firms and job seekers in the context of job searches. From the firm’s perspective, extensive research emphasizes the benefits of aligning employee values with organizational values. Akerlof and Kranton (2000, 2005) argue that when employees’ identities align with the organization, they are more likely to act in line with organizational objectives. Van den Steen (2005) shows that the alignment of beliefs improves coordination in organizations. Ghoshsamaddar et al. (2021) use a regulatory shock that affects employee mobility to show that a strong culture helps companies retain employees. Jeffers and Lee (2019) develop a measure of corporate culture using coworker connectivity on LinkedIn and show that firms with stronger cultures depend less on explicit contracts to retain human capital. Recent studies show that employee selection can help achieve value alignment (Chatman 1991, Kristof 1996, Campbell 2012, Abernethy et al. 2015, Swaney 2017, Liu et al. 2019, Deller and Sandino 2020). Cai (2023) provides further evidence that formal culture-fit systems during employee selection can ease the challenge of embedding values and improve the diffusion of culture within organizations. Consistent with these insights, practitioner evidence indicates recruiting benefits for companies that promote culture in job postings (Johnson 2020).

From the job seeker’s perspective, culture matters in job choice. Prior research examines how individuals sort into jobs based on their preferences and skills (e.g., Roy 1951). A growing literature shows how nonwage attributes can influence job searches, suggesting an important role for culture (Akerlof and Kranton 2005, Bénabou and Tirole 2011, Sullivan and To 2014, Mas and Pallais 2017, Hedblom et al. 2019, Ashraf et al. 2020, Choi et al. 2023, Maestas et al. 2023). Experiments further show that applicants often choose organizations with more supportive cultures, even at lower pay levels (Catanzaro et al. 2010). Relatedly, symbolic attributes such as values and image strongly affect employer attractiveness (Lievens and Highhouse 2003). Gadgil and Sockin (2020) use a sample of corporate scandals and show that negative reputation shocks decrease worker sentiment, largely because of diminished perceptions of management and culture. To our knowledge, limited research examines how job seekers respond to firms’ direct communication of culture.⁵ Instead, prior studies focus almost exclusively on the usefulness of alternative sources of information about culture. In particular, the focus has largely been on job review sites, such as Glassdoor and Indeed, where employees rate employers. In general, these studies suggest that job review sites provide useful information to employees and market participants (Hales et al. 2018, Chemmanur et al. 2019, Green et al. 2019, Huang et al. 2020, Dube and Zhu 2021, Campbell and Shang 2022).

Yet, growing evidence suggests there are limitations to the information provided by job review sites, given that they rely on voluntary data from employees (Karabarbounis and Pinto 2018, Gadgil and Sockin 2020, Marinescu et al. 2021) and are often too general to capture variation within firms. In contrast, job postings offer more precise information because they are written for specific roles and locations. This perspective suggests the presence of subcultures within organizations and is consistent with prior research in economics and organizational behavior that highlights variation in values across business units, functional roles, and geographic locations (e.g., Sackmann 1992, Hofstede 1998, Boisnier and Chatman 2003).

2.2. Hypothesis Development

Jobs bundle pecuniary and nonpecuniary attributes. The literature suggests that nonwage attributes can influence job searches and labor market sorting (e.g., Sullivan and To 2014, Mas and Pallais 2017, Hall and Mueller 2018, Maestas et al. 2023). In parallel, theories of person-organization fit argue that individuals self-select into firms whose values they share (e.g., Chatman 1991, Kristof 1996). Research on employer branding similarly shows that firms’ communicated identities meaningfully influence applicant quality and decisions (Cable and Turban 2001, Collins and Han 2004). In recent years, job seekers have increasingly emphasized culture (Estrada 2020). Yet, despite its importance, culture remains difficult to observe, thus generating information frictions for job seekers (Guiso et al. 2008, 2015b; Grennan 2019; Graham et al. 2022).

The information economics literature suggests that credible, informative disclosure can reduce mismatch risk and improve match efficiency (e.g., Stigler 1961). Job postings are a natural vehicle for such disclosure because they occur early in job searches and are less costly for applicants to process than other sources such as corporate websites and job review sites. A job posting provides hiring managers with an opportunity to communicate the benefits and requirements associated with a position, which is particularly valuable, given job seekers’ limited time and attention. Estimates suggest that the typical job seeker applies to 21 to 80 positions and spends only a few minutes reviewing each posting (Morgan 2020, Kolmar 2023).

In addition, external sources have at least two limitations. First, job review sites often require job seekers to input a review before they can observe reviews through a “give-to-get” model (e.g., Dube and Zhu 2021), which increases information acquisition costs. Second, external sources may contain information that is less relevant for a given position. For example, job review sites may contain descriptions of jobs at different locations or roles within a firm. Corporate websites are also “too vague” or contain information that is more relevant for other stakeholders, such as customers or investors (e.g., Boulland et al. 2026). These information sources may not be able to communicate distinct subcultures within a firm (e.g., Jones 1983, Sackmann 1992, Hofstede 1998). Therefore, job postings can help reduce the information frictions for job seekers to learn the culture associated with a position.

We expect that companies providing culture information can better attract workers and fill positions more efficiently. As noted by Wickman (2012), a company’s ability to hire strong employees is in part determined by how well it incorporates its core values into hiring. Culture information allows job seekers to sort into companies and positions that are a better cultural fit. Such sorting can be useful as the average corporate job opening receives 250 applications (Kolmar 2023). Culture information can help limit the pool of candidates to only those who believe they are a good fit for the position and lead to a higher acceptance rate, resulting in better hiring outcomes.

We note that our prediction is not without tension and draws on prior work in accounting and labor that examines how job seekers respond to firm disclosures (e.g., Choi et al. 2023, deHaan et al. 2023). These studies show that workers respond to disclosures when they value the underlying attribute and when they believe that the disclosure is credible and informative. A competing view is that job seekers may place little value on culture, with salary dominating their decisions. Indeed, a recent study suggests that 67% of job seekers cite salary as a top factor (Glassdoor 2018). Moreover, job seekers may learn about culture from alternative sources, reducing the incremental value of job postings. It is thus not clear whether and to what extent culture information will impact hiring outcomes.

3. Data, Variable Measurement, and Univariate Statistics

3.1. Measuring Culture Information in Job Postings

We adopt a semisupervised machine learning approach from Li et al. (2021) to measure the extent to which firms describe their core values in job postings. We focus on the five most common core values that firms highlight on their corporate websites (Guiso et al. 2015b). These values are integrity, teamwork, innovation, respect, and quality.

We obtain job posting data from Lightcast (formerly known as Burning Glass Technologies) for the period January 1, 2010, to December 31, 2020. Lightcast sources job postings from more than 65,000 websites, including company career sites, national and local job boards, and job posting aggregators.⁶ We retain job postings with more than 100 words, yielding an initial sample of about 25 million job postings from 4,911 firms.

The key challenge in measuring corporate culture is that firms discuss their values in different ways and that their language evolves. To overcome this challenge, we identify seed words that reliably define each cultural value and use a machine learning approach (word embedding) to expand these seed words to high-quality dictionaries that are specifically applicable to job postings (Mikolov et al. 2013, Li et al. 2021).⁷

Specifically, we use the culture seed words from Li et al. (2021) as a starting point and add “ethically,” “fair,” “commitment,” “diverse,” and “inclusive,” and remove “create” and “pride” based on how culture is described in the job posting context. (A full list of our seed words is provided in Table A.1, panel A, in Appendix A.) We train the Word2Vec word embedding model based on parsed job posting text to represent each word as a 300-dimensional vector that captures its semantics.⁸ To construct the expanded dictionaries, we first take the average of the vectors of seed words under the same culture category (Integrity, Teamwork, Innovation, Respect, and Quality) to represent that cultural value. For each cultural value, we then use the 500 words (phrases) with the closest association with that value, that is, the highest cosine similarity in vectors. Finally, we remove duplicate words and words that do not fit from the dictionaries. Table A.1, panel B, lists the most representative words in our expanded dictionary for each cultural attribute.⁹

After obtaining the expanded dictionaries, we measure each cultural value in a job posting by the weighted count of the number of words associated with that value divided by the total number of words in the posting. We calculate the weight using the term frequency-inverse document frequency (tf-idf) to account for the importance and prevalence of a word both within the posting and across the corpus. Table B.1 in Appendix B provides two examples of job postings from the same firm (Amazon) during the same period (2019 Q3) for different positions and locations, which exhibit different levels of disclosed culture. In the empirical analyses, we aggregate the culture variables to the unit of observation by taking the average of the values for all job postings in that unit. Most analyses are at the firm-MSA-quarter level, which proxy for a firm’s local hiring unit (such as a facility or office), to align with the notion of subculture. Empirically, this allows us to explore within-firm variation in disclosed culture and hiring outcomes across geographic locations and organizational functions.

We construct a composite score that covers all five cultural dimensions, Job Culture, following Li et al. (2021) and Graham et al. (2022). Defined as the sum of the five cultural dimensions, the score captures the salience and intensity with which firms describe their core values in job postings. To ease interpretation, we standardize it to have a mean of zero and a standard deviation of one. Details on the procedure of constructing the expanded dictionaries and measuring culture variables are in Appendix C.

3.2. Employee Inflow

In our analyses, we focus on employee inflows as our primary measure of successful hiring outcomes. Inflows provide a direct and observable indicator of whether culture information in job postings translates into successful hires. We obtain employee inflow data from Revelio Labs. The data provider estimates employee inflows and outflows from a variety of sources, including online professional profiles, resumes, and government data (Fadhel et al. 2022).¹⁰ The data cover more than 1.1 billion profiles, 4.5 million companies, and 5,200 cities and states. We define Inflow as the number of employees joining the firm-MSA in a quarter scaled by its beginning-of-quarter total number of employees.

3.3. Other Variables

Our analyses include a wide set of controls for job posting and firm characteristics. Job posting controls include the firm-MSA-quarter average of job posting features: requirements on education (Education) and experience (Experience), job complexity as reflected in the number of skill requirements (Skill Count), the number of words in the job posting (Length), recruitment demand measured with the number of job postings released (Hiring Intensity), job benefits including insurance (Insurance), other compensation (Other Comp), and work flexibility (Flexibility), and the percentage of job postings with nonmissing salary information (Salary Disclosed). We also conduct analyses that control for advertised mean salary information when available (Advertised Mean Salary) or expected salary based on similar job postings (Advertised Mean Salary (Expanded)). To mitigate the concern that both culture-related language in job postings and hiring outcomes are driven by jobs’ underlying tasks and responsibilities, we further control for the share of job postings across Standard Occupational Classification (SOC) categories. Nevertheless, fully separating culture disclosure from job content disclosure remains empirically challenging, and we acknowledge this as a caveat of our study.

In all analyses, we aggregate job posting controls to the unit of observation (e.g., firm-MSA-quarter) by taking the average of the values for all job postings in that unit. Firm controls include size (Size), profitability (return on assets (ROA)), leverage (Leverage), market-to-book ratio (MTB), and the percentage of intangible assets (IntanRatio). In subsequent analyses, we consider alternative outcome variables, such as hiring duration (Hiring Duration), employee outflows (Outflow), unscaled inflows and inflows scaled by postings, and Glassdoor measures related to the interview process, including the share of interview reviews mentioning culture (Interview Culture), the offer rate (Offer Rate), and the offer acceptance rate (Accept Rate). Variable definitions can be found in Table D.1 in Appendix D.

3.4. What Determines Disclosed Culture?

Figure 1 illustrates the time series trends in the five cultural dimensions (Integrity, Teamwork, Innovation, Respect, and Quality) in our sample period. Firms show a notable rise in the extent to which they highlight culture in job postings beginning around 2013–2014, with the trend continuing upward through the end of the sample period (2020). To assess sources of variation in disclosed culture, we employ variance decomposition and determinants analyses at the job posting level.

First, we estimate the following regressions to explore how much disclosed culture varies across firms, period, location, and occupation:

where j indexes a job posting, f indexes firm, n indexes month, i indexes industry, s indexes occupation, and c indexes county.¹¹ We report the estimates of Equation (1) in Table 1. Each row presents the adjusted R² from a regression of Job Culture on the corresponding fixed effects. The table reveals several trends. With respect to independent fixed effects, firm fixed effects explain nearly one-third of the variation (approximately 31.7%) in disclosed culture, consistent with studies documenting distinct firm cultures (Guiso et al. 2015b, Li et al. 2021). In contrast, location and time appear less meaningful. When we examine interactive fixed effects, we find that much of the variation in Job Culture remains unexplained. Even firm-occupation fixed effects, which essentially capture a job function within a firm, account for slightly more than half (53.1%) of the variation in Job Culture. The use of culture language in job postings also varies substantially within a firm at a given time, as firm-month fixed effects only explain 42.0% of the variation in Job Culture. Overall, this analysis suggests that culture information in job postings is far from boilerplate. Instead, it appears to be more granular than the culture information provided through traditional external sources, which are often at the firm level. This pattern supports the notion that subcultures exist within organizations (e.g., Sackmann 1992, Hofstede 1998, Boisnier and Chatman 2003). It also aligns with practical considerations, as job postings often reflect the preferences and priorities of hiring managers within specific teams or subunits.

Table 1. Variance Decomposition

Next, we examine firm and job characteristics that can explain disclosed culture. Culture information may be more important for firms seeking to differentiate themselves or appeal to candidates for whom cultural fit matters more, for example, in roles with high interpersonal demands. Firms may also highlight culture information when they face pressure in attracting talent. On the other hand, firms may hesitate to emphasize culture in roles where technical skills matter more to avoid narrowing the candidate pool. We include these additional considerations in Equation (1) and estimate the following model:

where Factors captures firm characteristics, job types, and posting features.

Table 2 reports the regression results, with models progressively adding groups of variables or fixed effects. Column (1) includes firm-month and county-month fixed effects along with firm characteristics and job role indicators. The results highlight meaningful heterogeneity across roles: Managerial Roles, Creative Roles, and Sales Roles exhibit higher culture scores, whereas Blue-Collar Roles and Engineer Roles show lower scores. This pattern suggests that disclosed culture is more salient in people-facing or soft-skill-intensive roles. We also find a positive association between disclosed culture and Hiring Intensity, consistent with the notion that firms emphasize culture when recruiting pressure is high during growth phases.

Table 2. Determinants Analysis

In column (2), we add firm-occupation fixed effects to account for variation across positions, and in column (3), we include job requirements from postings: Experience, Education, and Skill Count. In column (4), we add job benefits, including Salary Disclosed, Insurance, Other Comp, and Flexibility.¹² The model with all fixed effects, time-varying firm characteristics, and job requirements only explains less than two-thirds of the variation in Job Culture, consistent with the variance decomposition showing that Job Culture captures the subculture of the unit. In addition, positions requiring more experience are more likely to emphasize culture, possibly because these roles involve people management or other managerial responsibilities. Postings demanding more specific skills (as captured by skill count) contain less culture information, suggesting that firms filling technical roles prioritize skill requirements over cultural attributes. Finally, postings that disclose salary information are significantly less likely to emphasize culture.

Overall, the findings show that firms emphasize culture in job postings strategically, focusing on roles requiring greater experience, managerial, or interpersonal skills, whereas technical roles prioritize skills and may downplay culture to prevent narrowing the applicant pool. Culture becomes more salient in tight labor markets but less so when firms rely on compensation as a recruitment tool.¹³

4. Disclosed Culture and Hiring Outcomes

We focus on employee inflows, which most directly measure hiring outcomes. Our main analyses assess whether disclosed culture in a job posting is associated with increased employee inflows. We estimate the following regression at the firm-MSA-quarter level:

where f indexes firm, m indexes MSA, and t indexes calendar quarter.¹⁴ Controls include job-, firm-, and firm-MSA-quarter-level characteristics discussed in Section 3.3. Fixed effects account for various types of unobserved heterogeneity, such as time-invariant firm-MSA (i.e., facility) characteristics.¹⁵ Standard errors are clustered at the firm, MSA, and quarter levels, to account for within-group correlation.¹⁶ If disclosed culture improves inflows, we expect β₁ to be positive. Table 3 reports the summary statistics for the sample. The main variable of interest, Inflow, has mean (median) levels of 4.7% (1.4%).

Table 3. Summary Statistics

Table 4, panel A, provides the results from estimating Equation (3). We consider several empirical specifications. In column (1), we first estimate a model with only firm control variables. The coefficient on Job Culture is positive and significant at the 1% level, suggesting that disclosed culture is positively associated with subsequent employee inflow. In column (2), we include the full set of controls and fixed effects and observe that the coefficient remains positive and significant at the 1% level. In terms of economic magnitude, a one within-fixed-effect standard deviation increase in disclosed culture is associated with an approximately 0.14-percentage-point increase in worker inflows, representing a 9.7% increase relative to the sample median.¹⁷

Table 4. Job Posting Culture Information and Employee Inflows

Columns (3) and (4) extend the specification dynamically. Using inflows measured in quarters t and t+2, we find the coefficients on Job Culture to be positive and significant, indicating that disclosed culture is positively associated with inflows beginning in the contemporaneous period and persisting through the second quarter after job postings, though with smaller magnitudes.¹⁸

Next, we conduct a series of robustness and additional tests to ascertain the relation between disclosed culture and hiring outcomes. We begin by examining whether the effect of disclosed culture persists after accounting for expected salary, arguably the most important consideration for job seekers. We use the advertised salary present in job postings to measure expected salary. We do not include salary in our main analyses because the vast majority of job postings (around 92% of job postings in our sample) do not disclose salary information.¹⁹ In addition, some companies only provide a salary range that is wide and uninformative (e.g., Brown 2023, Stahle 2023). In column (1) of Table 4, panel B, we reestimate Equation (3) while controlling for the natural log of the mean salaries (Advertised Mean Salary) disclosed in job postings within a firm-MSA-quarter. As expected, the sample in these analyses is substantially smaller than that in our baseline tests (308,075 observations compared with 1,303,803 observations). Job Culture continues to load positively and significantly at the 1% level. The coefficient on the advertised salary measure is positive but insignificant, possibly because of the lack of power in this analysis.

To further explore this issue, we construct an expanded expected salary sample using average advertised salaries from similar job postings.²⁰ This approach expands the sample to 532,525 observations (column (2)). We continue to find that Job Culture loads positively and significantly. We document a positive and significant coefficient on the advertised salary variable (Advertised Mean Salary (Expanded)), suggesting that salary information is also important in attracting workers. The results presented in columns (1) and (2) indicate that job seekers view culture information to be distinct from salary information.

Next, we use job posting duration as an alternative measure of hiring outcomes. We expect more culture information to lead to a quicker identification of suitable candidates and thus a shorter job posting duration. To test our prediction, we reestimate Equation (3) and replace the dependent variable with Hiring Duration, the natural log of the days the job posting remains open and active for applicants (Chen and Li 2023).²¹ As shown in column (3) of panel B, the coefficient on Job Culture is negative and significant, suggesting that disclosed culture helps firms fill vacancies more quickly.

Finally, we extend the analysis by exploring the relation between Job Culture and subsequent employee outflows. Minimal culture disclosure may lead job seekers to make overly optimistic assumptions about cultural fit, resulting in unmet expectations and higher outflows (Sheridan 1992). To test this, we reestimate Equation (3) using average employee outflow over the subsequent four quarters (Outflow) as the dependent variable, capturing potential departures that arise after employees have had sufficient time to recognize a poor cultural fit, search for alternative opportunities, and ultimately leave the organization.²² We control for Dissatisfaction, an indicator variable of whether a facility’s Glassdoor culture rating falls in the bottom 30% in a quarter. This control helps account for broader workplace conditions that drive departures, ensuring that the estimated effect of Job Culture captures disclosure-related matching rather than dissatisfaction-driven turnover. As reported in column (4), the coefficient on Job Culture is negative and significant at the 10% level, consistent with the notion that employees in high Job Culture firms have better culturally aligned new hires.²³

Overall, the evidence in this section indicates that salient disclosed culture in job postings facilitates better job seeker-firm matching, improving hiring outcomes.

5. Mechanism Tests

Having established a robust relation between disclosed culture and employee inflow, we next assess the mechanisms for our findings. First, we conduct cross-sectional analyses to examine predictable variation in the effect of disclosed culture on inflows. Next, we provide evidence on the cultural fit channel using the variation in the salience of certain dimensions of culture and tests about the interview process.

5.1. Cross-Sectional Analyses

We conduct two sets of cross-sectional analyses to shed light on the mechanisms underlying the observed effect. Our proposed mechanism is that job seekers respond to disclosed culture because it reduces frictions in understanding firm culture and facilitates better sorting, which implies two conditions. First, job seekers value culture. Second, they perceive job postings to be informative about culture.

5.1.1. The Role of Job Seekers’ Valuation of Culture.

We begin by examining whether job seekers’ responses to culture information vary with how much they value culture, which we measure using job features and labor market conditions. First, cultural fit is likely to matter more to job seekers interested in creative roles because supportive environments that encourage psychological safety, risk-taking, and collaboration enhance employees’ creativity and innovativeness (e.g., Woodman et al. 1993, Harvey 2014). By contrast, job seekers of entry-level positions often prioritize career opportunities, which may diminish the importance of culture information. To test these predictions, we use occupations in Arts, Design, Entertainment, Sports, and Media (SOC code 27) to proxy for creative roles (Creative Roles) and postings with no education or experience requirements to proxy for entry-level jobs (No Education Requirement and No Experience Requirement).

We reestimate Equation (3) augmented individually with these variables and their interactions with Job Culture. Column (1) of Table 5 reports the results for creative roles, in which we use firm-MSA-quarter-occupation-level analysis to exploit variation in job types within facilities. We find the coefficient on Creative Roles × Job Culture to be positive and significant at the 5% level, suggesting that disclosed culture has a stronger effect on attracting employees for creative roles. Columns (2) and (3) report the results based on educational and experience requirements, respectively. The coefficients on No Education Requirement × Job Culture and No Experience Requirement × Job Culture are negative and significant at the 10% and 1% level, respectively, consistent with the notion that disclosed culture has a weaker effect on inflows for entry-level jobs.

Table 5. Disclosed Culture, Employee Inflows, and Job Features

Second, labor market conditions can shape how much job seekers consider culture information. When competition for jobs is high, job seekers may prioritize securing employment over culture. Therefore, we expect their responses to culture information to weaken in more competitive labor markets. We measure the supply of labor using local unemployment rate (Unemployment) and the demand of labor using changes in firms’ number of job postings (Δ Hiring Intensity). Columns (4) and (5) report the results. Consistent with our prediction, we observe a negative coefficient on Unemployment × Job Culture (significant at the 5% level) and a positive coefficient on Δ Hiring Intensity × Job Culture (significant at the 1% level). Thus, disclosed culture affects inflows more when job seekers face less labor competition.

5.1.2. The Role of Information Asymmetry.

Next, we explore the role of information asymmetry that a job seeker faces with respect to learning about a position’s culture. As discussed in Section 2, job seekers may learn about culture from alternative sources, which reduces their reliance on job postings. We therefore expect the effect of job postings on inflows to weaken when these alternative sources are more prevalent and helpful. To test this conjecture, we reestimate Equation (3) and interact Job Culture with three proxies for information asymmetry. First, we use facility (i.e., firm-MSA) size, assuming that larger facilities are more visible locally, reducing information asymmetry. Second, we consider the number of current employee reviews on Glassdoor and, third, the helpfulness of those reviews, as rated by Glassdoor users.

Table 6 presents the results. Column (1) reports the results based on facility size. Large Facility is an indicator variable that equals one if the total number of employees of a facility is in the top 30% of firm-MSAs in that quarter, and zero otherwise. We find the coefficient on Large Facility × Job Culture to be negative and significant at the 1% level, indicating that the effect of disclosed culture on inflows weakens when the facility is larger and thus culture information is more easily accessible. Column (2) reports the results based on the number of Glassdoor reviews. More Reviews is an indicator variable that equals one if the count of current employee reviews on Glassdoor over the past year is in the top 30% of firm-MSAs in that quarter, and zero otherwise. The coefficient on More Reviews × Job Culture is negative and significant at the 1% level, suggesting that the effect of disclosed culture on inflows weakens when more reviews are available. Column (3) reports the results based on the helpfulness of Glassdoor reviews. High Helpfulness is an indicator variable that equals one if the average helpfulness of reviews from current employees is in the top 30% of the firm-MSAs in that quarter, and zero otherwise. The coefficient on High Helpfulness × Job Culture is negative and significant at the 1% level, suggesting that the effect of disclosed culture on inflows weakens when Glassdoor reviews are more helpful.

Table 6. Disclosed Culture, Employee Inflows, and Information Asymmetry

Taken together, our results show that disclosed culture has a stronger effect on inflows when job seekers place a higher value on cultural fit and a weaker effect when alternative information sources about culture are more prevalent and helpful. These findings are consistent with job seekers responding to disclosed culture driving inflows.

5.2. Culture-Fit Channel

Next, we examine the cultural fit channel using two additional empirical designs. The first one delves into situations when certain cultural dimensions have a higher salience for job seekers. The second test explores the hiring process using job applicants’ reviews of the interview process on Glassdoor.

5.2.1. Individual Cultural Dimension Analyses.

Because we lack direct measures of job seekers’ preferences, we use two settings where external conditions may heighten the importance of specific cultural dimensions. These settings allow us to infer job seekers’ preferences based on changes in how individual cultural dimensions affect inflows. First, we use the BLM movement, which began at the end of May 2020, as a plausibly exogenous shock that increased job seekers’ preference for the respect cultural value, which reflects diversity, equity, and inclusion (Li et al. 2021). The protests were one of the largest social movements in U.S. history (Buchanan et al. 2020). We expect culture information related to “respect” to more strongly attract employees after BLM. To test this conjecture, we define BLM as an indicator that equals one for the months between June 2020 and November 2020 and zero for the months between December 2019 and May 2020 and estimate the following regression using a monthly-level sample during this year:

where f indexes firm, m indexes MSA, and n indexes month.²⁴ The independent variable Respect is the standardized culture score for the Respect value, and we expect β₁ to be positive. Table 7 presents the results. In column (1), the coefficient on BLM × Respect is positive and significant at the 5% level, suggesting that respect-related culture information more strongly affects inflows after BLM.²⁵ In column (2), we conduct a placebo analysis using the standardized sum of the scores of the other four cultural values, that is, Non-Respect Culture, as the independent variable. We do not find a similar effect.

Table 7. Job Posting Culture and Employee Inflows: Individual Cultural Dimension Analyses

Next, we use industry-level violations as a setting that is associated with increased preferences of job seekers for integrity. Regulatory violations and corporate misconduct heighten ethical concerns, potentially making integrity-related culture more important for job seekers. We thus expect culture information related to integrity to be more attractive to job seekers in industries experiencing higher levels of recent corporate misconduct. To test this conjecture, we estimate the following regression:

where i indexes industry, f indexes firm, m indexes MSA, and t indexes quarter. Integrity is the standardized culture score for the Integrity value. Violation is defined as the number of significant violations in each industry-quarter, where a violation is considered significant if its penalty falls within the top quartile of the cross-sectional distribution. We expect β₁ to be positive. In column (3), the coefficient on Violation × Integrity is positive and significant at the 5% level, suggesting that integrity-related culture information leads to greater inflows when industries experience more misconduct. A placebo test using Non-Integrity Culture (standardized sum of the scores of the other four cultural values) shows no similar effect (column (4)).

In sum, findings from these two analyses support the cultural fit channel, showing predictable variation in how disclosed culture affects inflows that aligns with changes in job seekers’ preferences.

5.2.2. Interview Analyses.

Next, we assess the cultural fit channel using the context of job interviews. Interview outcomes provide evidence about the hiring process itself and shed light on the intermediate steps in recruiting, which complements our main analyses based on hiring outcomes. If disclosed culture facilitates cultural fit, firms should be more likely to extend offers to interviewed candidates and candidates who receive offers should be more likely to accept when culture information is more salient in job postings. To test these predictions, we estimate the following model at the firm-quarter level:

where f indexes firm, t indexes quarter, and i indexes industry. We use a firm-quarter-level analysis because interview data do not contain location. We define three Interview Outcome variables: (1) Interview Culture is the percentage of interviews that explicitly mention “culture” or “cultural,” (2) Offer Rate is the share of interviews that result in a job offer, and (3) Accept Rate is the percentage of offer receivers who accept the offers.

We report the results in Table 8. Panel A presents the summary statistics of the sample. Culture is mentioned in 2.3% of interview reviews, 42.4% of interviews yield offers, and 86.3% of offers are accepted. Panel B presents the regression results. In column (1), using Interview Culture as the dependent variable, we find that the coefficient on Job Culture is positive and significant at the 1% level. This result indicates that firms highlighting culture are more likely to emphasize culture during interviews, further validating that firms discuss culture in job postings to seek employees who will fit with their cultures.

Table 8. Job Posting Culture and Interviews

We then test whether disclosed culture in job postings is associated with a higher likelihood that firms extend offers to job seekers (Offer Rate) and their acceptance of offers (Accept Rate). Consistent with the notion that more disclosed culture helps attract more suitable candidates into the interview pool, we find that Job Culture is positively associated with both likelihoods (columns (2) and (3)).

Overall, our interview analyses suggest that culture information facilitates better cultural fit and, consequently, better matches between firms and job seekers.

6. Additional Analyses

Our results thus far indicate a strong and positive association between disclosed culture and employee inflows and that this effect is likely driven by disclosed culture facilitating job seeker-firm matching. Next, we conduct several additional analyses to assess alternative explanations and ascertain the robustness of results. First, to help strengthen our argument that culture information in job postings rather than that in other sources drives our effects, we control for ECall Culture, the earnings-call corporate culture measure developed by Li et al. (2021), Website Culture, the discussion of culture on firms’ “About Us” pages, and Glassdoor Culture, the Glassdoor culture ratings from current employees.²⁶ In columns (1)–(3) of Table 9, panel A, we find that after controlling for ECall Culture, Website Culture, and Glassdoor Culture, Job Culture is still positively associated with inflow, suggesting that job postings contain culture information that is distinct and informative to job seekers.

Table 9. Job Posting Culture Information and Employee Inflows: Robustness Tests

One possible concern is that firms communicating culture in job postings may be lowering hiring standards to meet staffing needs, driving our results. Untabulated analyses show no relation between disclosed culture and the education and experience of new hires, inconsistent with this explanation. To further mitigate this concern, we control for the education and experience of the inflow workforce in Equation (3). In column (4), we continue to find that the coefficient on Job Culture is positive and significant.

Next, to ensure that the denominator used to scale employee inflows does not drive our results, we use an unscaled measure of inflows or inflows scaled by the number of job postings in the previous quarter, which approximately measures labor demand, as the dependent variable.²⁷ We report the results in columns (5) and (6). The coefficients on Job Culture remain positive and statistically significant.

We also explore whether culture acts as a motivator or a hygiene factor for job seekers, drawing on Herzberg’s two-factor theory. If culture is only a hygiene factor, we would expect that its presence, regardless of emphasis, would reduce dissatisfaction and that increasing the amount of culture information would have little effect. To examine this, we group firm-MSA-quarters into Bottom Culture (bottom 30%) and Top Culture (top 30%) within each quarter and reestimate Equation (3) including these indicators. The results are reported in column (7). We find that the coefficient on Bottom Culture is negative and that on Top Culture is positive, implying that the degree of cultural emphasis matters, beyond whether culture is present. That is, culture functions as a motivator for job seekers who value culture alignment rather than serving as merely a baseline hygiene factor.

We use job postings and employee inflow sourced primarily through online platforms, which likely overrepresents online job postings and white-collar jobs. In panel B, we consider subsamples to ensure the generalizability of our findings. First, we examine whether technology firms drive our results, because they may be more likely to advertise culture. We reestimate Equation (3) excluding facilities of firms headquartered in California where many technology firms reside or facilities of firms with a two-digit Standard Industrial Classification (SIC) code of 35 or 73. The results are reported in columns (1) and (2), and our main inferences are unchanged.

Second, we consider whether job specificity or job type drives our findings. In column (3), we retain only observations specifying low skill requirements (in the bottom 30% in the average number of skills of the firm-MSAs in that quarter). In column (4), we follow Baruffaldi et al. (2020) and control for the prevalence of AI-related keywords in job postings for a subsample of technology firms. In column (5), we exclude firm-MSA-quarter observations with job postings that demand skills typically associated with leadership positions, such as project management and people management. Our results remain robust in all tests, suggesting that our findings are not caused by skill-related information in job postings or confined to managerial roles.

Third, we test the robustness of our results across different firm sizes. Because Revelio estimates employee inflows and outflows based on data sources such as online professional profiles and resumes, these measures may be less accurate for small firms because of their limited publicly available information. We reestimate Equation (3) for large and small firms (top and bottom 30% in terms of total assets at the beginning of the quarter, respectively) separately and report the results in columns (6) and (7). The coefficients on Job Culture remain positive and significant for both samples. Nevertheless, despite our effort to ensure the generalizability of our findings, we acknowledge that our results may not extend to labor market populations that are underrepresented in the data (such as those with only offline job postings).

Finally, one central prediction in our study is that making culture information salient in job postings helps attract job seekers and ultimately helps fill employment vacancies. Ideally, we would measure the effect of a job posting on the firm’s ability to fill the corresponding job. However, because this information is unobservable, we rely on firm-MSA-quarter-level aggregation for our main tests. To alleviate concerns about measurement errors, we reexamine Equation (3) for a subsample of observations that only contain one SOC code. As shown in column (8), our inferences remain unchanged, as the coefficient on Job Culture remains positive and significant at the 1% level. In an untabulated test, we repeat the analysis at the firm-MSA-quarter-occupation level while including quarter-occupation fixed effects. We find that the coefficient on Job Culture remains positive and significant, further demonstrating the robustness of our results. Note that these two additional analyses further mitigate the concern that both culture-related language in job postings and hiring outcomes are driven by jobs’ underlying tasks and responsibilities.

In sum, although we cannot rule out all alternative explanations, the collective evidence from our empirical analyses based on different outcome variables and empirical designs suggests that disclosed culture likely improves hiring outcomes through facilitating job seeker-firm matching.

7. Conclusion

Corporate culture represents one of the critical factors that job seekers consider. We examine the labor market effects of emphasizing corporate culture information in job postings and find that highlighting core values in a job posting attracts job seekers, especially when cultural fit matters more to those job seekers and information asymmetries are more pronounced. Our findings are relevant in light of recent SEC requirements emphasizing the disclosure of material information about human capital, such as information about culture. In addition, our study has practical implications. Greater transparency about culture can help job seekers make better decisions about potential employers. Despite these benefits, managers should be mindful that emphasizing cultural fit too strongly may inadvertently discourage diversity, suggesting a need to balance culture alignment with inclusion goals.

We view our paper as opening several fruitful directions. First, researchers could develop empirical designs that more cleanly separate the cultural framing of a job from its underlying tasks and responsibilities. Second, although our analyses focus on job seekers’ responses, future research can examine how managers decide which aspects of culture to disclose and whether these disclosures lead to better matches in terms of performance outcomes.