Tracking entry-level and senior hiring as an economic indicator

Tracking entry-level against senior hiring gives you an economic indicator that is written every week: the share of the roles companies open that is addressed to people starting out, beside the share that asks for experience. This guide defines the measure, sets it beside what official statistics publish about new entrants, gives two queries, on the weekly market series and on postings, and works an example on an invented industry. It ends with the tests that stop a change of labels or of company mix from passing as a change in hiring.

What does tracking entry-level against senior hiring show as an economic indicator?

It shows which levels of experience employers ask for in the roles they open. The measure is the share of postings opened in a window at entry level, beside the share at senior level, for one industry. Once labels, company mix and calendar are ruled out, a lasting shift is a change in what employers ask for.

Three measures carry the reading, and all three come from one count: the postings opened in a window.

Set the bands before the first run and keep them. On the Fokals seniority scale of nine levels, this guide takes intern and entry as the entry band and senior and lead or principal as the senior band, and reports mid and the four management levels apart. A people manager and a senior specialist are different roles, and one merged band hides which of them moved.

The reason to watch the mix comes from research on what employers ask for. The authors of a 2014 working paper of the Federal Reserve Bank of Boston studied a large database of online job postings from 2007 to 2012 and found that employers' skill requirements, a bachelor's degree and experience among them, rose when the supply of relevant job seekers was larger. A 2016 paper by the same authors found that the requirements fell as the labour market improved from 2010 to 2014. Both findings are the authors' own, on requirements stated in postings, for those years. They make the mix worth measuring, and they leave open what a shift means in any other period.

A posting records an intention to hire. The measure counts roles advertised, at the level the employer advertises them.

Is the seniority of job postings a labour market indicator? Entry-level and senior hiring data as an economic signal

The mix of entry-level and senior roles in job postings is a labour market signal of demand: it shows which levels of experience employers ask for. Official statistics report the other side: they count people looking for a first job and describe the requirements of jobs already held. Compare their direction, never their level.

Four publications from official bodies report on entry into work in the United States. Each was read on 7 October 2026.

Three differences decide how a postings reading sits beside these.

• The unit. The household survey counts people. The requirements survey describes jobs that workers hold, and its handbook gives its objective as measuring job requirements, not the characteristics of workers. A posting is a role an employer advertises and has yet to fill.
• The meaning of entry. A new entrant is an unemployed person looking for a first job, with no previous work experience. In the occupation assignments, entry is the typical path into an occupation, and the page notes that multiple paths are not captured. A seniority level places one advertised role on a provider's scale.
• The calendar. The count of new entrants is monthly, the graduates' rates are quarterly and the requirements estimates carry a reference year. A postings series is written weekly.

The Bureau's survey of job openings publishes its estimates by industry, region and state and, at the total private level, by establishment size class. On the page that sets this out we found no split by the level of the position. A level on each advertised role is what labelled postings add.

Entry-level hiring data sources: job postings with a seniority level

A data source for entry-level hiring needs every posting dated and labelled with a seniority level under a named version. Fokals delivers this in three datasets: Job Postings, with the level on each role; Hiring Activity, with open postings by seniority for each company and day; and Market Series, with weekly seniority shares by industry.

The first two belong to the hiring product, refreshed daily, with every role collected as the company publishes it. Market Series is written weekly, as of each Sunday, and each as-of date is written once and never revised.

Two fields describe experience in different ways. The seniority level is a label, each posting carries the version of its labels, and a posting keeps its version until it closes. The minimum years of experience is the employer's own figure. Each is a check on the other.

Reading the mix for one industry in the weekly series

The first query reads Market Series (market_series) for one industry and adds the levels of each band. Like the queries after it, it is written for PostgreSQL, and every date, industry and band is a parameter.

with bands as (
  select
    as_of,
    window_days,
    sum(case when subject in (:entry_subjects)  then count    end) as entry_postings,
    sum(case when subject in (:senior_subjects) then count    end) as senior_postings,
    sum(case when subject in (:entry_subjects)  then previous end) as entry_previous,
    sum(case when subject in (:senior_subjects) then previous end) as senior_previous,
    sum(case when subject in (:entry_subjects)  then rate     end) as entry_share,
    sum(case when subject in (:senior_subjects) then rate     end) as senior_share
  from market_series
  where family = 'seniority_share'
    and dimension_kind = 'industry'
    and dimension = :industry
    and window_days in (30, 180)
    and as_of between :first_as_of and :last_as_of
  group by as_of, window_days
)
select
  as_of, window_days, entry_postings, senior_postings, entry_share, senior_share,
  1.0 * entry_postings / nullif(senior_postings, 0) as entry_to_senior,
  1.0 * entry_previous / nullif(senior_previous, 0) as entry_to_senior_previous
from bands
order by window_days, as_of;

In a row of the seniority_share family, count is the number of postings at one level opened in the window, rate is their share of the postings opened, from 0 to 1, and previous is the same count over the window just before. A metric with a subject is written family:subject, as in seniority_share:senior, and the metric catalogue of the API lists the subjects of each family. The last two columns set the entry-to-senior ratio beside the ratio of the window before.

The share is taken over postings, so it follows the employers behind them. When one employer opens many more roles than usual, the industry's mix leans towards that employer's mix. The next query measures how much.

Holding the companies and the labels fixed in Job Postings

The second query reads Job Postings (job_postings) for a list of companies you fix in advance and pass as :company_ids: the companies you follow in the industry, by your own classification. It compares a base window, from :base_start to :split, with a current window, from :split to :current_end.

with opened as (
  select
    company_id,
    case when first_seen_at < :split then 'base' else 'current' end as win,
    labels::jsonb ->> 'seniority' as seniority,
    experience_years_min
  from job_postings
  where company_id = any(:company_ids)
    and label_version = :label_version
    and found_on_first_read = false
    and first_seen_at >= :base_start
    and first_seen_at <  :current_end
),
per_company as (
  select
    company_id,
    count(*) filter (where win = 'base')    as n0,
    count(*) filter (where win = 'current') as n1,
    count(*) filter (where win = 'base'    and seniority in ('intern', 'entry')) as e0,
    count(*) filter (where win = 'current' and seniority in ('intern', 'entry')) as e1
  from opened
  group by company_id
),
matched as (
  select *, sum(n0) over () as total0
  from per_company
  where n0 >= :min_postings
    and n1 >= :min_postings
)
select
  count(*)                         as companies,
  1.0 * sum(e0) / sum(n0)          as entry_share_base,
  1.0 * sum(e1) / sum(n1)          as entry_share_current,
  sum(1.0 * n0 / total0 * e1 / n1) as entry_share_fixed_mix
from matched;

A posting enters by first_seen_at, the time it was first observed. The filter on found_on_first_read leaves out postings already published when a company's postings were first observed, for which first_seen_at is not the opening date. The filter on label_version keeps both windows under one version. Run the count once without it: rows that the filter removes mark a version boundary inside your windows.

The last column weights each company's current entry share by its share of base-window postings. Fixed mix minus base is the change inside companies. Current minus fixed mix is the effect of companies changing weight. An official index holds the mix fixed in the same way: the Employment Cost Index of the Bureau of Labor Statistics uses fixed employment weights, so that it measures the change in labour costs independent of employment shifts among occupations and industries.

To hold the job function mix fixed instead, group by labels::jsonb ->> 'job_function' in place of the company. For the two cross-checks, end the same query with this select.

select
  win,
  count(*)                                             as opened,
  count(*) filter (where coalesce(seniority, '') = '') as unlabelled,
  count(experience_years_min)                          as stating_years,
  avg(experience_years_min)                            as mean_years_asked
from opened
group by win;

An empty level means no level was set for that posting, and the entry share above counts those postings in its denominator. The mean of experience_years_min covers the postings that state a minimum.

A worked example on an invented industry

The industry, the companies and every figure below are invented, Acme Robotics included, and the example is illustrative.

Assumptions:

• The list is four companies of one industry: Acme Robotics and companies B, C and D.
• The base and current windows are 30 days each and do not overlap.
• Every posting in both windows carries the same label version, and none has an empty level.
• The entry band is intern and entry, and the minimum is 20 postings per company and window.

1. Headline. 140 / 500 = 28.0% in the base window and 150 / 700 = 21.4% in the current one: a fall of 6.6 points.
2. Base weights. Acme Robotics 100 / 500 = 0.20, B 0.40, C 0.20, D 0.20.
3. Fixed mix. 0.20 × 10.0% + 0.40 × 25.0% + 0.20 × 40.0% + 0.20 × 30.0% = 2.0% + 10.0% + 8.0% + 6.0% = 26.0%.
4. Inside companies. 26.0% minus 28.0% is a fall of 2.0 points, all of it at company B: 0.40 × 5.0 points.
5. Company mix. 21.4% minus 26.0% is a fall of 4.6 points. Acme Robotics opened three times as many roles at its usual entry share of 10.0%, and its weight went from 0.20 to 300 / 700 = 0.43.
6. Counts. Entry-band postings rose from 140 to 150.

The headline describes an industry turning away from entry-level roles. The working shows something narrower: 4.6 of the 6.6 points are one employer opening more of the roles it always opens, 2.0 points are a shift at one company, and more entry-level roles were opened than before. What company B's 2.0 points mean stays open. Ten postings in 200 can be chance, a change in the functions it recruits for, or a graduate intake that fell in the base window. A second window, and the decomposition by job function, narrow it down.

What a shift can mean, and what else could explain it

A fall in the entry share beside a rise in the senior share is what you would see if employers asked for more experience in the roles they open. It is also what you would see in each case below. Run the test for each before a shift goes into a note or a model.

A shift that survives every test describes what the employers on your list advertised. Why they did so is a further question. Employers may ask for more experience because they are cautious about the months ahead, or because more job seekers are available, which is what the authors of the Boston papers found for their period. Postings establish what was advertised. The reason needs other evidence. The data dictionary defines every column used here, and sector hiring trends for macro and thematic research covers the other weekly hiring series.

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.