Methodology

How the data works

VacancyRatio uses a single computed metric to compare graduate supply with job vacancy demand. This page explains what the numbers mean, how they are collected, how forecasts are generated, how AI is used, and the important limits of this approach.

Demo data

All figures shown are illustrative demo data. Realistic in magnitude but not sourced from live APIs. When connected to a live provider (e.g. Adzuna), a Live · API badge will replace Demo data throughout the product.

Data collection

Vacancy Data

Sourced from job posting APIs and official labour surveys, tagged by field, location, skills, and date. Deduped by title, company, and city.

field + location keynational-level

Graduate Data

Completions per field per year from national education bodies. Excludes enrolments, dropouts, and late-year graduates. Always national — city vacancy is estimated.

completions / yrUG + PG

Field Mapping

Graduates follow ISCED categories. Vacancies (ISCO, SOC, Adzuna) are mapped to 8 harmonised field slugs. Boundary assignments are approximate.

8 field slugsISCED aligned

Deduplication

Same vacancy from multiple providers is collapsed using a compound key of normalised title, company, and city. Only one record counted. Demo data is not deduped.

title · company · city

The Formula

Both metrics are derived at query time — never stored. Ensures auditability and consistency across views.

// Primary metric

vacancyRatio = relevantVacancies / graduates

// Normalised

vacanciesPer100Graduates = vacancyRatio × 100

Above 1.0 → more vacancies than graduates

Below 1.0 → more graduates than vacancies

Supply-demand signal, not individual job probability.

Interpretation Thresholds

Ratios are mapped to five named signals. These thresholds are editorial judgements, not regulatory definitions.

Signal	Range	Meaning
Critical Shortage	> 2.00	Vacancies are more than double the graduate supply. Employers are actively competing for talent.
Strong Demand	1.00 – 2.00	More vacancies than graduates. The market actively absorbs new entrants.
Balanced	0.75 – 1.00	Supply and demand broadly aligned. Competitive but conditions are fair.
Surplus	0.50 – 0.75	More graduates than vacancies. Differentiation matters.
Significant Surplus	< 0.50	Graduates substantially outnumber available roles. Specialisation and strategic targeting are important.

How Forecasts Are Generated

Historical trend extraction. The Compound Annual Growth Rate (CAGR) is computed separately for vacancy demand and graduate supply, using the available historical years.

Projection. Each component is independently extrapolated forward using the observed CAGR. The projected vacancy ratio is then derived from the two projected series.

Outlook labelling. If the projected ratio improves by more than 8% relative to its starting value, the outlook is labelled Growing. If it declines by more than 8%, it is labelled Declining. Otherwise it is Stable.

Confidence is capped at Medium when demo data is in use. Projected data is always visually distinguished — dashed lines, lighter colours, and explicit "Projected" labels.

How AI Is Used

When an ANTHROPIC_API_KEY is set, the "What This Means" market summary is generated by Claude (claude-haiku). The model receives structured, pre-computed data — vacancy ratio, signal label, CAGR figures, outlook label, top skills — and writes a 2–3 sentence summary grounded only in those numbers.

AI does not generate

Vacancy counts, graduate counts, ratios, or signals

Forecasts or skill frequencies — those are all computed from data

Statistics, companies, salaries, or trends not present in the supplied data

When no API key is set, a static template-based summary is shown instead. AI-generated summaries are labelled with an AI badge.

Limitations of Geographic Comparisons

The location comparison feature shows estimated vacancy concentrations across cities. These distributions are illustrative — they are not drawn from city-level vacancy surveys. They represent plausible shares based on known economic geography, but they should be treated as indicative, not authoritative.

When a city is selected in the Explore view, the graduate count remains national (since city-level graduate data is not available) while the vacancy count is estimated at city level. The resulting ratio is therefore not directly comparable to a national ratio and is labelled accordingly.

Planned Graduate Data Sources

Intended sources for graduate data when live integration is enabled.

🇮🇳 India

Graduates: AICTE Annual Report; UGC Annual Report; Indian Nursing Council

Vacancies: Ministry of Labour & Employment Vacancy Survey; NASSCOM Jobs Reports; RBI Employment Survey

🇺🇸 United States

Graduates: National Center for Education Statistics (NCES); HRSA

Vacancies: Bureau of Labor Statistics JOLTS

🇬🇧 United Kingdom

Graduates: Higher Education Statistics Agency (HESA); Nursing & Midwifery Council (NMC)

Vacancies: ONS Vacancy Survey; NHS Digital

🇩🇪 Germany

Graduates: Destatis (Federal Statistical Office)

Vacancies: Bundesagentur für Arbeit

🇦🇺 Australia

Graduates: Department of Education, Skills and Employment (DESE); AHPRA

Vacancies: Jobs and Skills Australia

Vacancy Data Providers

The provider architecture allows live APIs to be connected without changing any UI or calculation code. Only lib/providers/ needs to be updated.

Adzuna

Job listings APISupported

US, GB, AU, DE, IN and many others. Real-time job postings, salary data, category-level aggregates.

BLS JOLTS

Official labour surveyPlanned

US monthly job openings by industry and occupation. Highly authoritative but not real-time.

UNESCO UIS API

Graduate dataPlanned

Global graduation completions, 200+ countries, ISCED classification.

Eurostat API

Graduate + vacancy dataPlanned

EU graduation and employment data, harmonised across member states.

OECD Stats API

Cross-country employmentPlanned

Employment and labour market data across OECD members.

Limitations

Graduation figures count completions, not enrolments. Part-time and distance learners may be under-represented.

Vacancy counts reflect posted roles at a point in time, which may differ from actual hiring volumes or filled positions.

Field mappings between graduation classifications (ISCED) and vacancy classifications (ISCO, SOC, Adzuna categories) are approximate. Many graduates work in adjacent fields.

The ratio does not account for geographic distribution within a country — national averages can mask significant regional imbalances.

City-level vacancy distributions are illustrative estimates, not survey data.

Skills data is derived from job posting text analysis. Postings are not a perfect census of employer demand.

Forecasts use simple CAGR extrapolation and assume current trends continue — they do not model policy changes, economic shocks, or structural shifts.

Do not use for financial, career, or institutional planning without primary source verification.