YouthPOWER defines and measures youth opportunity at the district level. To do it, we bring together ~180 indicators across 27 government databases, and turn them into a single comparable score for every district. Young people are taken as those aged 15 to 29, with some measures using a wider window up to 35, or the 15-to-21 school-age band where a measure concerns being in education.
Education affects work. Work affects migration. Infrastructure affects enterprise. Public institutions affect participation. Skilling matters when it connects to real labour-market pathways. A composite captures these interactions better than any single indicator — youth opportunity is local, multi-dimensional, and measurable.
Counts are first expressed as rates (per population, workforce, or youth population) so districts compare regardless of size. Every indicator is then scored 0–100 against a fixed standard — the distance from a worst-acceptable level to a policy target where one exists, or position within the observed range where no target is citable — not by ranking districts against each other. In every pillar, scores blend the inputs in place (40%) with the outcomes young people actually see (60%), minus a gap penalty where inputs run ahead of results. The five pillars combine into one 0–100 score. Full detail in Score Calculation below.
Each measure is checked against the official published figure at the finest level it is published — the district where available, otherwise by adding our district values up to the state and national level and comparing against the published state or national figure. State and national rates are computed as a single ratio of summed numerators to summed denominators (for survey-based measures, the survey's own state estimate), not as an average of district rates.
The full technical annex (every scored measure with its target, basis, and issuing ministry and year, plus the scoring formulas, the aggregation weights, and the robustness and uncertainty results) is available as a separate document. Open the technical annex →
Only foundation and outcome indicators count toward the score. Context indicators are shown for interpretation. This keeps the score focused on the strength of local systems and the results they produce.
Every measure comes from official government data; not every one is scored. A measure is scored only when it (1) helps answer the pillar's central question, not merely its context; (2) is available for the district itself, or imputed to it by an explicit method, for enough districts; (3) is clean rather than confounded — measures attached to the wrong place (credit concentrated in the metros), measures that mislead in the Indian context (a low open-unemployment rate usually reflects distress self-employment, not strength), measures whose direction is ambiguous (counts of enterprises or schools can signal either depth or a surfeit of tiny units), and political representation, which cannot fairly be reduced to a score, are shown where they inform but kept out of it; (4) adds information rather than echoing another measure (correlated cuts collapse to one; composites are scored with their parts shown beneath); and (5) varies across districts. Measures shown but not scored are displayed as evidence, so local knowledge can read them in context.
Indicators are scored across the five pillars, grouped into foundation and outcome. Context indicators (population, geography) — and anaemia and men's low-BMI, which are shown on the health card for context — are displayed alongside but not scored.
Indicators come in different units — percentages, counts, ratios, rupees, seats. Each scored indicator is normalised to 0–100 against a fixed standard, not by ranking against other districts. Where a policy target exists (for example near-universal access, or a maximum acceptable level like early marriage or personal-loan-heavy credit), the score is the distance from a worst-acceptable level to that target — so a district at the target scores 100 regardless of where others sit. Where no target is citable, the score is the value's position within the observed range, after trimming the most extreme one percent at each end so a single outlier district does not set the scale. Where higher is better, higher values score higher; where lower is better (unemployment, NEET, early marriage, ITI vacancy), lower values score higher. After this step, higher always means better.
Within each pillar the scored indicators are grouped into foundation and outcome, then blended — with the outcome side weighted more heavily, and a gap penalty where the inputs run ahead of the results:
Pillar Score = 40% Foundation score + 60% Outcome score − Gap Penalty (20% of the foundation-minus-outcome gap, capped at 10 points)
The outcome side carries more weight because access on paper is not enough; a district should score highly only when young people see results.
If a district appears to have the right conditions, but young people are not seeing results, the score is pulled down.
The five pillar scores combine into one 0–100 YouthPOWER score, weighted toward the conditions that most directly shape young people's economic pathways:
YouthPOWER Score = 0.25 × Opportunity + 0.25 × Work + 0.20 × Participation & Agency + 0.15 × Education + 0.15 × Skilling
A state or national score is not a simple average of its districts. It combines a population-weighted mean (half the weight) with the median (three tenths) and the average of the weakest quarter of districts (two tenths), so a state is credited for lifting its floor, not only its average. State-level indicator values are population-weighted aggregates of their districts, shown with the spread of district values.
A lower score means young people face stronger constraints from their local conditions: weaker institutions, thinner markets, poorer access, lower public capacity, or worse outcomes.
Read it with the pillar scores and underlying indicators. Two districts can reach the same overall score for different reasons — one strong in education but weak in workforce outcomes, another with enterprise infrastructure but poor skilling access. The score is not only a ranking; it is a way to identify the specific systems expanding or limiting youth opportunity.
Where a pillar has too few measured indicators in a district to score it reliably, that pillar is marked not rated rather than shown as an artificially low number, and the overall score is formed from the pillars that can be measured. This mainly affects a small number of very small or newly created districts.
Not every card counts toward the score. On a scorecard, cards that feed the index carry a terracotta top-rule; cards shown for context carry a thin grey rule and are not scored. A few labour measures, open unemployment and the employment rate among them, are shown as context because they read counter-intuitively in the poorest agrarian districts, where subsistence and unpaid family work absorb people; the real labour-market signal lives in participation, formality and job structure.
Read differences in scores, not fine rank positions: a gap of more than about a point is a real difference, while a smaller gap means two districts are effectively level. We also support district stakeholders, elected, administrative, and civic, in acting on the score, including with tools built for their specific needs.
Because the survey-based measures carry sampling error, we measured how much a district's score can move from sampling alone. A district's overall score sits within about plus or minus 0.9 points, so two districts are a real difference only when their scores differ by more than roughly a point; closer than that, read them as level, wherever they happen to fall in a list. The precision varies by area: Education and Readiness (Skilling) rest on administrative records and carry no sampling error; Opportunity is tight (about a point); Work is wider (about 2.6 points); Participation, entirely survey-based, is widest (about 5 points), so Participation comparisons need a larger gap to be real.
We also tested the weights: moving any pillar's weight, or setting all five equal, leaves the comparison between districts essentially unchanged, so the picture does not depend on one particular choice of weights. And 87 percent of scored district figures are measured directly in the district; the rest are mostly newly-formed districts that inherit their parent's rate where their own data is too thin, with a small share estimated or pooled across a metropolitan area.
Finally, we tested the score against NITI Aayog's National Multidimensional Poverty Index. Across about seven hundred districts the two are strongly and inversely correlated (a rank correlation of about −0.74): districts that score higher for young people are districts where multidimensional poverty is lower.
Each figure is checked against the official published number for the survey or administrative release it comes from. Two measures are shown on a slightly different basis, for the reasons below.
What each indicator is compared against
Two comparisons sit beside a place's own figures, neither of which affects the score. The typical state is the middle value across all states. Comparable states is the middle value across the other states in the same region. Both are worked out separately for each measure, so the state sitting in the middle changes from one measure to the next, and neither figure describes any single state. The regions are the regional councils, with three changes: Sikkim is placed with the north eastern states, of which it is a North Eastern Council member; Uttarakhand sits with the northern hill states; and Rajasthan with the large central states it resembles in size and income. Union Territories are included where they have an elected legislature and are otherwise left out. District comparisons work the same way: the typical district is the middle value across all districts, and comparable districts is the middle value across the districts of the comparable states.
Population figures are projections
With no fresh enumeration since 2011, the population counts used as denominators and as the youth-population base are projected forward from the last count using official projections rather than a new headcount. States that have seen many district splits and mergers — Rajasthan most of all — are the hardest to project cleanly.
A few large metropolitan areas are pooled for shared facilities
For a few large metropolitan areas whose districts form one continuous built-up city — Delhi, Mumbai, Hyderabad, Chennai, Kolkata and Ahmedabad — certain facility-access measures (train-service frequency, college enrolment, training-institute places, libraries, bank branches, and road network) are calculated across the whole metropolitan area and shown for every district within it, because residents cross district lines to use these facilities. Most other measures are shown at the individual district level; where a newly-formed district's own data is too thin, it takes its parent district's rate.
Some employment and wage figures are modelled where local samples are thin
Where a district's local sample is small, or for newly-formed districts, some employment and wage figures are modelled from the wider area rather than measured in that district; they are best-available estimates and should be read as indicative.
Some measures are shown for context and suppressed where the sample is too small
A handful of indicators, such as anaemia and men's nutrition, are shown for context rather than scored. Men's figures on the agency, gender-norms and marriage measures are likewise shown for context only and are not scored, because the men's survey sample is small in many districts. Where the underlying survey sample for a district is too small to support a figure, the district carries its region's instead, and where neither can support one the measure is left unavailable, so a few unusually small samples do not produce misleading extremes.
Recently bifurcated districts borrow from their parent
Districts carved out after 2011 that a source does not yet report on their own take their parent district's rates directly, since a rate is unchanged by splitting a district. Absolute numbers, such as school counts, are instead shown as an estimated split of the parent's total.
A few districts carry no value for some indicators
This happens where the source itself has a gap, not because the service is absent, for example rail frequency in some Kerala districts.
Some measures are published only at the state level
Higher-education enrolment is one. Each district is shown at its state value, because a district-level figure is not published.
Delhi is often reported as a single unit
Banking and credit are reported for the Delhi National Capital Territory as a whole, so the territory-wide figure is shown for each of its constituent districts. Library provision is treated the same way: the libraries recorded for the capital are spread across its districts by population.
Credit is measured by place of utilisation; deposits by where they are held
Credit is measured by place of utilisation. Deposits are measured by where they are held. Some head-office bias may remain in both. Banking figures are as on March 2026; branch counts as on March 2026.
Business registration is a live, growing database
The registered-enterprise counts are a point-in-time snapshot of a register that keeps growing as more firms sign up, so the totals rise over time.
Unincorporated-enterprise figures are indicative at district level
The count of unincorporated enterprises, and the workers and value added that go with it, comes from a sampled survey rather than a full census. District figures are estimated from the survey and are best read for relative comparison between districts rather than as exact local totals.
How formal-employment figures are measured
Formal-employment coverage is measured across each metropolitan area rather than by single district, so it reflects where work sits rather than where head offices register; even so, the largest cities can read high. Where a district's figure cannot be made reliable it is left unscored rather than shown at a misleading level. State and national figures are the sum of the district counts, and a state's leading industries are the sectors that top the most of its districts, not a ranking by total employment.
Apprenticeship places are counted where training happens, then measured across the metropolitan region
Apprentices are recorded at the establishment where they train, not where they live. An industrial hub that draws young trainees from a wide area would otherwise read far above its own youth population. To correct this, apprenticeship places in each large metropolitan region are measured across the whole region and applied to every district within it, the same way formal-employment coverage is handled, so the figure reflects where the work sits rather than where a single plant happens to be. A few standalone industrial districts that sit outside any such region still read high for the same reason; there the figure should be taken as a sign of concentrated local industry, not an unusually large share of resident youth.
MUDRA loan figures are spread evenly and annualised
Where MUDRA lending is reported without a district breakdown, it is spread evenly across the state's districts. For most states (Bihar aside) the figure is a seven-year cumulative total from April 2016 to January 2023, which we annualise by dividing by seven and express relative to the labour force.
Connectivity is a composite measure
It combines proximity to a railway station, how frequently trains actually run, and the reach of the rural-road network, because no single one of these captures how connected a district really is. The station and frequency components draw on open and modelled data rather than a single official release.
Local college enrolment counts colleges located in the district
This figure tallies students enrolled in colleges and universities located in the district, set against its 18-23 population. Enrolment is credited to where the college sits, so young people who study in another city are counted there. A low value therefore reflects few colleges nearby rather than fewer students continuing into higher education. District figures are held to the official state total.
Enrolment ratios can exceed 100%
A gross enrolment ratio compares students enrolled to the local age-group population, so it can pass 100% where schools draw over-age, repeating, or out-of-area students — common in districts that act as regional schooling hubs. The figure shown is the true ratio; for scoring it is capped at 100.
College counts are aligned to the official national total
The number of colleges shown is adjusted so the national total matches the officially reported count. The underlying institution list includes some specialised teaching and off-campus centres that official reporting counts separately, so district figures are scaled proportionally to reconcile with the published total. This affects the displayed counts only and does not change any scored measure.
The single-teacher-school rate counts schools that have pupils
The share of schools run by a single teacher is measured across schools that have pupils enrolled. A small number of schools reporting no current enrolment are set aside, since a lone staff member at a school with no pupils is not a meaningful measure of teaching deployment.
Skilling-scheme figures are point-in-time readings from live databases
PMKVY and apprenticeship figures come from live dashboards. PMKVY figures are annualised over 2023-2025 (data up to December 2025): trainees are assessed and certified in the years after they enrol, so assessment and certification are counted cumulatively across years, not within a single year. Apprenticeship figures are as on 1 April 2026.
The worker-to-population ratio counts the whole community, not just working-age adults
This counts workers as a share of the whole population (all ages), so it reads lower than a conventional working-age figure — it shows how much of the entire community is in paid work, which is the lens this measure is built for. It is shown for context and is not scored.
How wages are scored, and the living-wage benchmark
The wage measures that count towards the score are the mean daily wages a district actually pays — regular salaried and casual day-wage — together with the ratio of women's to men's wages, which counts as an equity signal. A higher wage is read as more opportunity, since young people move toward better-paying work even where living costs are higher. Casual pay is shown alongside as the wage floor, which keeps a high salaried wage from carrying a district on its own. Casual daily wages are read against a living-wage benchmark of about ₹540 a day, which most casual work falls short of.
Where the ₹540 living-wage benchmark comes from
The ₹540-a-day figure is the need-based national minimum wage recommended by a 2019 government expert committee — ₹375 a day at 2018 prices, costed for a family of 3.6 — brought to current prices with inflation. It is a yardstick for reading what casual work pays, not a legal minimum.
Representation shares for small districts rest on a few seats
The share of a district's MLAs who are women, or under 45, is taken over only the assembly seats that fall within the district. Where a district holds very few seats, that share rests on a small base — a single seat moves it sharply — so in those districts it is best read as a broad signal rather than a precise figure. The count of MLAs itself follows the assembly constituencies that fall within the district; a constituency spanning two districts is counted in each.
The Members of Parliament shown chair the district development committee
The Members of Parliament shown for a district are the chair and co-chairs of its District Development Coordination and Monitoring Committee; the number shown varies from district to district according to the official committee composition.
Constituencies do not map one-to-one onto districts
Parliamentary and assembly constituencies do not map one-to-one onto districts. Where a constituency spans more than one district, it is shown in each of them.
Parliamentary-question counts run to the last completed session
Parliamentary-question figures run through the last completed session of the 18th Lok Sabha. A question tabled jointly by several members is counted under each of them.
Only centrally sponsored working-women's hostels are counted
The hostel count includes only centrally sponsored hostels. Some states — Tamil Nadu among them — also run their own state-funded hostels, which are not yet captured here.