Do job postings predict company revenue growth? What official studies show

Whether job postings predict company revenue growth has two answers. For a whole economy, central banks and statistics offices have studied job advertisements against employment, unemployment and surveyed vacancies, some with correlation figures and forecast tests, and a 2013 Reserve Bank of Australia study finds changes in vacancies and advertisements positively correlated with employment growth one and two quarters ahead. For the revenue of one company we found no official study that tests the number of its postings, and no coefficient to quote. This post sets out what the official research measured, each result with its sample and in its own terms, and then gives you a test to run on dated postings for the companies you follow.

Is there a study of the correlation between job postings and company revenue growth?

We found no study by a central bank, a statistics office or a securities regulator that reports a correlation between the number of a company's job postings and its later revenue growth. The five official studies below measure a neighbouring relation: advertisements and vacancies against employment growth, the vacancy rate and unemployment, for a country or a currency area.

Three things hold across the five. The outcome is employment, a vacancy rate or unemployment for an economy, never the revenue of a company. The lead is short: a quarter or two in the 2013 study, the first year or so in the 2025 one. And the relation moves: in the 2013 table the two-quarter coefficient for the official vacancy series is 0.48 over the whole sample and 0.27 over its last three years. Its authors conclude that their models explained around half the variance of next-quarter employment growth in sample and forecast only slightly better than a simple autoregressive model.

Job postings data as a leading economic indicator: research on hiring demand, the labour market and company growth

Official research treats job postings data as a leading indicator of employment over short horizons. Reserve Bank of Australia studies of 2013 and 2025 found changes in advertisements and vacancies positively correlated with employment growth up to two quarters ahead, and an advertisements series that improved unemployment forecasts over the first year or so. Company growth needs its own test.

A posting records an intention to hire

The official definitions say what is being counted. In the United States a job opening in the Job Openings and Labor Turnover Survey meets three conditions: a specific position exists with work available for it, the job could start within 30 days, and the employer is actively recruiting from outside the establishment. Posting a notice on the internet is one form of active recruiting on the survey's list, and openings are counted on the last business day of the month. Eurostat's definition of a job vacancy turns on the same thing: a paid post for which the employer is taking active steps to find a candidate and which the employer intends to fill, at once or within a set period.

Both definitions describe an intention and an effort, and the hire is a separate count. The Bureau of Labor Statistics divides the hires of a month by the openings left at its end, and for 2017 that ratio ran from 0.46 in federal government to 2.0 in arts, entertainment and recreation. The same number of open roles stands for different numbers of hires from one industry to the next.

An online advertisement is one more step from the hire. An advert can hold several jobs and may stay online after the post is filled, and the 2013 Reserve Bank of Australia article warns that an advertisement placed with several sources can be counted more than once. In the Job Postings dataset each posting a company publishes on its careers page is one record with the dates it was first and last observed, as why we read job postings from the company's own board explains.

From the labour market to one company

A national series averages over many employers, so the habits of any one of them wash out. One company's postings keep those habits: a role left open all year, a posting that replaces a leaver and adds no capacity. The chain from a posting to revenue is also longer than the chain to employment: the role has to be filled, the hire has to become productive and the product has to sell, each with its own delay.

Job posting data as a signal of layoffs and of revenue growth: the research

We found no official research that scores job posting data as a signal of layoffs or of revenue growth for one company. Layoffs leave dated official records: in the United States, written notice from larger employers at least 60 calendar days ahead of a mass layoff, and a current report from a listed company. Postings can be tested against both.

The Worker Adjustment and Retraining Notification Act requires employers with 100 or more employees to give at least 60 calendar days' advance written notice of plant closings and mass layoffs that affect 50 or more employees at a single site of employment. A company listed in the United States reports specified events in a current report on Form 8-K, often within four business days of the event. One item of the form is named Costs Associated with Exit or Disposal Activities, and staff guidance from the Securities and Exchange Commission says a company terminating employees as part of a plan to exit an activity need not file the report until it has informed the employees affected.

Both records follow the decision, and whether postings show a cutback earlier is the thing to test. Hiring Activity gives the open, new and closed postings of each company for every closed day, and Company News classifies company announcements and regulatory disclosures into 13 event types, one of which is layoffs and restructuring. Take those dated events as the outcome and count two things for the companies you follow: how often a fall in open postings came first, and how often a fall was followed by no event at all. Detecting restructuring from postings and announcements works through the pattern and its false alarms.

How investors use job postings data to forecast company growth

Investors who use job postings data to forecast company growth read it as a dated record of where a company intends to add people, set between two reporting dates. The count of open roles, their split by job function and seniority, and the countries they appear in each feed a forecast that the next report confirms or corrects.

A listed company in the United States files annual and quarterly reports, and its revenue arrives with them, while a careers page changes daily. The data supports four readings.

• Level against the company's own past. Open roles against the same company's trailing average, or per employee using Employee Headcount, so that companies of different sizes compare.
• Mix. Sales roles add capacity to sell, and engineering roles add cost first. Job function and seniority labels separate them, and Sales Team Metrics follows the sales organisation week by week, as in tracking sales team build-out.
• Geography. A first role in a country where the company had none is a dated sign of a market being opened.
• Direction against peers. A company rising against its industry differs from one rising with it, and Market Series gives the hiring measures by industry, country and size band on same-store cohorts.

Using hiring data in equity research between earnings builds the series step by step.

Job postings as alternative data: hiring signals, firm performance and stock selection

We found no figure from a central bank, a statistics office or a securities regulator for how the number of job postings, used as alternative data, relates to firm performance or to stock returns. A hiring signal used in stock selection therefore rests on your own test, and the test is only as sound as the dates in the data.

An academic paper on job ads and stock returns reports a result for its own companies, period and source of postings: read those three before the coefficient. Like any alternative data, postings fail a back-test in a few known ways, and each can be checked on a sample before a model is fitted.

In Fokals hiring data every posting carries the date it was first observed, daily rows are written once after the day closes and are never revised, and a posting already published at the first observation carries a baseline flag and stays out of the count of new postings. Why we write each table once sets out the rules. A test of this kind measures a relation in past data for your list, and nothing in this post is investment advice.

A worked example: one company, eight quarters

Acme Robotics is an invented company, and every figure in this example is illustrative.

Assumptions:

• Open postings are read from Hiring Activity on the last day of each quarter, and the quarter before the first ended with 50.
• The signal for a quarter is the percentage change in open postings since the previous quarter end.
• The outcome is the revenue growth of the following quarter over the quarter before it, from your fundamentals data.
• Eight pairs are compared by rank, with Spearman's coefficient, so that no single quarter dominates.

The squared differences sum to 38, and with eight pairs the coefficient is:

1 - (6 * 38) / (8 * (8 * 8 - 1)) = 1 - 228 / 504 = 0.55

A coefficient of 0.55 reads as support for the signal, and eight pairs cannot carry it. Eight outcomes can be ordered in 40,320 ways, and 3,447 of them give a coefficient at least as high against the same eight changes in postings: 8.5 per cent, or about one unrelated series in twelve. Two more runs come before any conclusion: the same calculation against revenue growth of the same quarter and of the quarter before. If the coefficient is as high when revenue comes first, postings at Acme Robotics follow revenue as much as they lead it.

A test to run on the companies you follow

The example becomes a test when it runs across companies and over more quarters, in six steps.

1. Fix the list as of each date. Take the companies you followed on each test date, those since acquired or delisted included, and join them to the postings on ISIN or share-class FIGI, the keys set out in the identifiers that make company data joinable.
2. Build the signal from dated rows. Open postings at each quarter end and the change over the quarter, as in the query below.
3. Scale it. Divide the change by the company's own trailing average, and subtract the same measure for its industry.
4. Pair it with an outcome dated when it was published. Revenue growth for the next quarter, from your fundamentals data, keyed by the day the report came out.
5. Measure across companies, quarter by quarter. Rank the companies by signal and by outcome at each quarter end, compute the rank coefficient, and read its average and its spread over all quarters beside the same statistic on shuffled outcomes.
6. Run the controls. Repeat with revenue of the same quarter and of the quarter before, split the signal by job function, and score the rule on a later period than the one it was chosen on.

The signal of step 2, as a query on Hiring Activity:

with ranked as (
  select company_id, isin, day, open_postings,
         date_trunc('quarter', day) as quarter_start,
         row_number() over (
           partition by company_id, date_trunc('quarter', day)
           order by day desc
         ) as from_end
  from company_hiring_daily
  where isin is not null
    and day < date_trunc('quarter', current_date)
),
quarter_end as (
  select company_id, isin, quarter_start, open_postings,
         lag(open_postings) over (
           partition by company_id order by quarter_start
         ) as open_before
  from ranked
  where from_end = 1
)
select company_id, isin, quarter_start, open_postings, open_before,
       (open_postings - open_before) * 1.0 / nullif(open_before, 0) as change_in_quarter
from quarter_end
where open_before is not null
order by company_id, quarter_start;

The same frame answers the layoff question: keep the signal and replace the outcome with the dated events of Company News. The hiring dataset page describes the datasets, and the data dictionary defines every field in the query.

Frequently asked questions

The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.

What this page says about the products it names was checked against their public documentation on 7 October 2026. Product and company names are trademarks of their owners. Fokals is not affiliated with them or endorsed by them.