Unemployment fell,
but employment did not rise.
The single-digit illusion: unemployment fell because the labour market emptied out
The unemployment rate is a fraction with a moving denominator. When people stop looking for work they drop out of that denominator, and the rate falls without a single job being created. To measure how much of the improvement is real, we recompute the rate holding the participation rate constant .
Official unemployment against the participation-adjusted rate
The hatched area is hidden unemployment: the difference between the official figure and what would be reported if participation had stayed at its base-year level. Change the base year with the slider below.
Choose the participation base year
The question is simple: if the share of people entering the labour market had stayed at its base-year level, what would unemployment be today? Employment counts stay exactly as published and are never touched; only the size of the labour force is recomputed.
What became of the 6.62 million who reached working age in nine years
Each square is one percent of the net addition to the population aged 15+ between 1395 and 1404.
Three rates, three different stories
Unemployment collapsed, participation declined, and the employment ratio — which is insensitive to both — barely moved.
The same pattern everywhere: urban and rural, women and men
If the fall in unemployment were real, different groups would show different patterns. Instead, all four repeat the same thing: the unemployment line falls and the participation line falls with it. Both charts show the same four groups over the same years.
DATA
Between 1395 and 1404 the population aged 15+ went from 59.5 to 66.1 million (+11.1%). Over the same period the employed rose by 2.30 million, the unemployed fell by 1.18 million, and the inactive population grew by 5.51 million.
Participation went from 43.2% to 40.6% and the employment ratio from 37.8% to 37.5% .
ANALYSIS
The adjusted rate shows two distinct phases. Between 1395 and 1398 the improvement was real : the adjusted rate also fell from 12.4% to 8.9% . From 1399 onward the two diverge; the official rate reaches 7.5% while the adjusted rate locks into a 12–14% band.
In other words, almost all of the reported improvement after 1399 is explained by the relative shrinking of the labour market, not by job creation.
CONCLUSION
The unemployment rate has become a misleading indicator for this economy. The alternative proposed here is the employment-to-population ratio , which fell 0.3 pp in nine years and tells the truer story: Iran's economy has not created jobs in proportion to its growing working-age population.
Any policy that measures its own success by the unemployment rate is exposed to this error.
The absent half: the country invests in women and then does not use them
For the first time in the available statistics, women form the majority of Iran's university students. In the very same year, women's labour force participation hit its lowest level of the decade. The two lines move in opposite directions.
The human-capital scissors
Women's share of higher-education enrolment against women's labour force participation. Both as percentages, on one axis.
In 1403–1404, 52 of every 100 university students were women. In 1404, of every 100 employed people only about 15 were women.
The gender gap in the labour market, 1404
Three key labour market indicators by gender. The length of each bar is the size of the gap.
Where do women study?
Gender composition of students by major field of study, 1403–1404. The figure beside each bar is the total enrolment in that field.
DATA
Women's share of students rose from 46.1% in 1394–1395 to 51.9% in 1403–1404. Over the same period women's participation fell from 16.3% to 13.4% and their employment ratio from 12.9% to 11.4% .
Women are the majority in art (69.5%), basic sciences (68.7%), medicine (58.8%) and humanities (57.8%), and a minority in engineering (24.8%).
ANALYSIS
The fall in women's unemployment from 20.7% to 15.0% looks positive at first glance, but it happened alongside a fall in their employment ratio — the Act 1 pattern, only sharper. The official definition does not count an educated woman who has stopped actively searching as unemployed.
The mix of fields matters too: industry — which provides 32.8% of national employment — draws on a field where only a quarter of students are women.
CONCLUSION
Iran's largest untapped human resource is 28.7 million economically inactive women aged 15+ — a population larger than everyone currently employed in the country.
If women's participation returned merely to its 1397 level (17.6%), roughly 1.4 million people would join the labour force. That is the return on an investment the country has already made in educating them.
Warning bell, spring 1405: where the decade-long trend broke
The most recent quarterly data in this dataset marks the first serious reversal. Comparing spring 1405 with spring 1404 — like-for-like quarters, which cancels out seasonality — unemployment did not merely rise; the absolute number of employed people fell.
The last five quarters: total, youth and women's unemployment
All three series jumped together in spring 1405 — the signature of a general shock rather than noise in one group.
Where did the shock come from? Sector employment change, spring 1405 against spring 1404
Sector headcounts are estimated by multiplying each sector's share by total employment in that quarter, then differenced.
Where did the new jobs come from? Nine years, three sectors
Official data publish each sector's share as a percentage , and percentages can deceive: agriculture's share fell, but did it lose jobs or merely grow more slowly? Switch to "Headcount" to see the same data in people. The answer is unambiguous.
DATA
Spring 1405: population 15+ at 66,671 thousand (+828 thousand against spring 1404), labour force 27,136 (+32), employed 24,672 (−451), unemployed 2,464 (+483).
Industry's share of employment fell from 33.0% to 31.0% ; services rose from 52.4% to 53.8% and agriculture from 14.6% to 15.1% .
ANALYSIS
Translating shares into headcounts shows industry lost about 643 thousand jobs, while services added only 110 thousand and agriculture 58 thousand. The shock originates almost entirely in industry.
The simultaneous rise in agriculture's share — a sector shrinking throughout the decade — is the classic signature of labour taking refuge in informal, low-productivity work during a contraction.
CONCLUSION
This is a turning point, not noise. For the first time in the period studied, growth in the working-age population has coincided with an absolute decline in employment.
Methodological caveat: this finding rests on a single quarter and should be checked against summer and autumn 1405. That said, the size of the move (1.8 pp on the unemployment rate) exceeds any seasonal fluctuation recorded in this dataset.
The empty cradle: the demographic engine is shutting down
So far we have talked about the people who exist. This act is about those who do not. Registered births fell 41.6% in nine years — and the decline accelerated in the last two.
The vital scissors: births and deaths on a collision course
The hatched area is natural population increase. In 1395 there were 4.1 births for every death; by 1404 that ratio had fallen to 1.9.
Marriages fell 39%, divorces did not budge
Bars are event counts; the dashed line is divorces per 100 marriages in the same year.
Households keep shrinking
Average household size from the 1335 census to 1400. The urban–rural difference has all but disappeared.
DATA
Births: 1,528,053 in 1395 to 892,278 in 1404 (compound annual growth rate −5.8%). Deaths: 369,751 to 459,984 (+2.5% per year). Natural increase: 1,158,302 down to 432,294.
Winter 1404, with 207,578 births, is the lowest-birth quarter recorded in this dataset.
ANALYSIS
Two forces are at work at once: falling births, rooted in fewer marriages and smaller households, and rising deaths, which are largely a product of population ageing rather than deteriorating health. 1399 and 1400 are the COVID exception; after that, deaths returned to their long-run trend.
The fall in rural births (−43.5%) is slightly steeper than in cities (−41.0%), meaning the rural fertility advantage is also disappearing.
CONCLUSION
Iran's natural population increase has shrunk by nearly two-thirds in nine years. This is no longer a slow trend; it accelerated in 1403 and 1404.
Because marriage leads births and marriages hit a period low in 1404, a continued near-term decline is close to certain. Pro-natalist policies, if they are having an effect, are not yet visible in the available data.
The school queue is getting shorter: a wave already on its way
School enrolment peaked at 16.97 million in 1401–1402, an echo of the 1393–1396 birth wave. That wave is over, and what follows is far smaller — something we know not from any forecast but from the number of children already born .
The peak has passed
The line is registered school enrolment by academic year. It peaks in 1401–1402, echoing the birth wave of 1393 to 1396.
And the size of the cohorts still to come
Each bar is the sum of six consecutive years of births — the population that fills the whole of primary school about seven years later. The percentage under each bar is the change from the first window.
Higher education: students came back, graduates did not
Enrolment is recovering after the sharp 1394–1399 decline, but the number of graduates is still falling.
DATA
School enrolment: 17.88 million (1380–81) → trough of 13.23 million (1391–92) → peak of 16.97 million (1401–02) → 16.86 million (1403–04).
Total births from 1395 to 1400 come to 7,808,954; from 1399 to 1404, 6,235,729.
ANALYSIS
School enrolment mirrors the birth wave with a six-to-seven-year lag. Since births have fallen continuously from 1395, enrolment enters a sustained decline from around 1405–1406: first in primary, then with a lag in secondary and higher education.
The 43% drop in graduates happened while enrolment has been rising since 1400 — pointing to a mix of weak intakes in the intervening years and, plausibly, more dropouts or longer time-to-degree.
CONCLUSION
The country is on the verge of a major reorganisation of educational infrastructure: surplus capacity in primary schools within a few years, then in secondary schools and universities.
On the other side, the flow of young workers into the economy will be markedly thinner after about 1415. Iran's demographic window — the period when the working-age share is high — is closing, while (per Act 1) it has yet to be fully used.
Official projections under strain
The "Population and Human Resources" sheet places two things side by side that are rarely compared: the population projection to 1415 and actual birth and death statistics. Put them together and a gap appears — one that has widened rapidly in the last two years.
The gap between projected growth and actual natural increase
Each bar shows how far actual natural increase (births minus deaths) ran ahead of or behind the population growth the official projection expected . Green: reality ahead of projection Red: reality behind projection
| Year | Projected population increase (thousands) | Actual natural increase (thousands) | Gap (actual minus projected) | Dominant explanation |
|---|
A live population model to 1415 — change the assumptions yourself
The teal line is the official projection and stays fixed. The crimson line is this project's model: it starts from the actual 1404 figure and moves forward on births, deaths and migration alone. Move the three sliders below to see how sensitive the result is to those assumptions.
Iran grows more urban because the countryside empties
In the official projection, the rural and non-settled population declines in every single one of the years from 1396 to 1415.
DATA
The official projection implies a population increase of 601 thousand for 1404, while actual natural increase that year was 432 thousand — a gap of 169 thousand. In 1403 the gap was 111 thousand.
The projected annual population growth rate falls from 1.12% in 1397 to 0.36% in 1415. The rural population goes from 20.75 to 18.20 million (−12.3%).
ANALYSIS
The gap has two distinct origins and should not be read as one. The 1399 and 1400 gaps (+97 and +74 thousand) come from excess COVID deaths, an event no projection anticipated. The 1403 and 1404 gaps (+111 and +169 thousand) have a different source: births falling faster than the model assumed.
The scenarios show that if the current trend continues, natural increase reaches zero somewhere between 1412 (trend continuation) and 1415 (moderated scenario), and turns negative thereafter.
CONCLUSION
This is not a criticism of the Statistical Center's model; every demographic projection rests on assumptions and must be revised when reality moves. Our finding is that that moment has arrived: the gap has doubled in two years, and the cause is identifiable.
Practical implication: planning that relies on 91.4 million people in 1415 — from subsidy and water allocation to school capacity and pension funds — should also accommodate a lower scenario of roughly 88 to 89 million.
Two hourglasses emptying at once
This dashboard's argument is one sentence: Iran faces two constraints at the same time, and each makes the other worse. On one side the demographic engine is shutting down, and the inflow of workers will be smaller in the decades ahead. On the other, right now — while the working-age population is still growing — the economy has failed to absorb it, and the official unemployment rate has concealed that failure. The demographic window is closing before it was ever really opened.
1 · Change the indicator
In an economy with falling participation, the unemployment rate is a misleading measure. employment-to-population ratio and the absolute number of employed people should be the primary yardstick for employment policy. The former fell 0.3 pp in nine years.
2 · The largest reserve is women
28.7 million women aged 15+ are outside the labour force, even though women are the majority of university students. Returning women's participation merely to its 1397 level would add around 1.4 million people to the labour force.
3 · Take spring 1405 seriously
The first quarter in which jobs fell and unemployment rose together, with a clear origin in industry (−643 thousand jobs, estimated). It needs checking against the next two quarters, but its magnitude is unprecedented.
4 · Revise the projections
The gap between the official projection and actual natural increase has more than doubled in two years. Long-term planning should cover a scenario of 88 to 89 million in 1415, not only 91.4 million.
5 · The education clock rings first
The decline in school enrolment begins around 1405–1406 and, unlike the labour market, is close to certain — the children are already born. Surplus school capacity is an opportunity for reallocation, not a crisis.
6 · Where the two crises meet
Unless productivity and participation rise, a shrinking future workforce means lower output and compounding pressure on pension funds. The time to use the young population that already exists is now.
Data, method, corrections and transparency
Every figure presented here as a derived indicator appears in this section with its formula, assumptions and source. The aim is that any reader can reproduce each calculation independently.
1 · Data sources
All figures are extracted from the "Population and Human Resources" sheet of the Macroeconomic and Social Indicators of Iran workbook. The raw file is published alongside this project and can be downloaded here . No external data has been added. The agency responsible for each table, as stated in the file itself, is listed below with a link to its official site:
- Statistical Center of Iran amar.org.ir — national population projection to 1415; average household size; population aged 15+ by activity status; participation rate; unemployment for all ages, 15–24 and 18–35; employment ratio; share of major economic sectors in employment. Labour data come from the Center's Labour Force Survey.
- National Organization for Civil Registration sabteahval.ir — registered births and deaths by sex and by urban/rural area; registered marriages and divorces, annual and quarterly.
- Ministry of Education medu.ir — school enrolment by gender and educational stage.
- Ministry of Science, Research and Technology msrt.ir — higher-education students and graduates by field of study and gender. Institutions under the Ministry of Health are included in these same tables.
The cleaned dataset this dashboard runs on can be downloaded from the buttons at the foot of this page as JSON and CSV , so that every figure here can be checked independently.
2 · Data cleaning and corrections
Before any analysis, every table with a "total = male + female" or "total = urban + rural" structure was tested automatically for consistency. The test found nine inconsistencies. Five were corrected on conclusive internal evidence; the rest were flagged and left untouched. No correction affects any national total; all of them concern internal components.
| # | Series and item | Value in file | Corrected value | Evidence | Action |
|---|---|---|---|---|---|
| 1 | Births 1395 — male, national | 766,142 | 786,142 | Urban male + rural male = 605,523 + 180,619 = 786,142. The corrected value also moves the sex ratio at birth from 103.3 (outside the natural range) to 106.0, consistent with 106.1 in 1396. | Corrected |
| 2 | Births 1396 — rural total | 344,355 | 344,365 | Rural male + rural female = 177,029 + 167,336 = 344,365; urban + rural then matches the national total exactly. | Corrected |
| 3 | Deaths, winter 1404 — urban total | 908,866 | 90,866 | One extra digit; urban male + female = 53,410 + 37,456 = 90,866. The published value is eight times the entire quarter's deaths. | Corrected |
| 4 | Deaths, winter 1404 — rural total | 23,805 | 22,805 | Rural male + female = 12,093 + 10,712 = 22,805. After this correction, 90,866 + 22,805 = 113,671, exactly the quarter's total. | Corrected |
| 5 | Deaths, summer 1404 — rural total | 25,531 | 23,531 | Rural male + female = 12,403 + 11,128 = 23,531; and 91,141 + 23,531 = 114,672, the quarter's total. | Corrected |
| 6 | Deaths 1395 — breakdown by sex (all three areas) | Male 161,774 Female 207,884 | — | The male-to-female death ratio in 1395 is 0.78, whereas from 1396 to 1404 it stays between 1.20 and 1.30. The inversion occurs simultaneously in the national, urban and rural rows — the clear signature of two transposed columns. Lacking conclusive evidence, we did not correct it. | Flagged |
| 7 | Unemployment rate, aged 18–35, 1395 | 26.8% | — | A 7.1 pp fall in a single year (to 19.7% in 1396), while the 15–24 group fell only 1.0 pp and the whole population only 0.3 pp. A sign of a definitional break or recording error rather than real change. | Excluded from trend analysis |
| 8 | Death totals by sex, 1395 to 1401 | Discrepancy of 12 to 93 | — | Male + female always falls slightly short of the total. Our reading: records of unknown sex. The discrepancy is under 0.03% and reaches zero from 1402 onward. | Left as published |
| 9 | Population projection — sum of components | Discrepancy ±1 thousand | — | In 16 rows the components sum to one thousand more or less than the total. This is rounding to the nearest thousand, not an error. | Left as published |
Some quarterly row labels in the file carried a trailing space (for example "Spring 1404 "), which was normalised before joining tables. The "pre-primary" row in the enrolment table is recorded as "000", read as "not available" and excluded from the calculations.
3 · Derived indicators and how they are computed
a) Participation-adjusted unemployment — this project's central indicator
The question: if the participation rate had stayed at its base-year level (base), with the same actual number of employed people, what would the unemployment rate be?
U*t = ( A*t − Et ) / A*t = 1 − Et / ( Pt × rbase )
Ht = A*t − At ← hidden unemployed
where P is the population aged 15+, E the employed population, A the published labour force and r the participation rate. Worked example for 1404 with a 1395 base (43.2%): A* = 66,148 × 0.432 = 28,576 thousand; U* = 1 − 24,822 / 28,576 = 13.14%; hidden unemployed = 28,576 − 26,837 = 1,739 thousand.
Limitation: this indicator does not assume everyone who left is truly unemployed. Part of the fall in participation may reflect more education, retirement, or an ageing population structure. U* should therefore be read as an upper bound on true unemployment and U as a lower bound. The analytical message is the distance between them.
b) Gender breakdown of the 15+ population and employment (analytical estimate)
The file does not publish the 15+ population by gender, but it does give men's and women's employment ratios. The breakdown can be reconstructed by solving a single-unknown equation:
For 1404 with P = 66,148, E = 24,822, rm = 63.8%, rf = 11.4%: 32,979 thousand men and 33,169 thousand women; 21,041 thousand employed men and 3,781 thousand employed women ⟹ women's share of employment 15.2%.
Independent validation: the breakdown was reconstructed using the gender-specific participation rates and Unemployment rate — which were not used in the calculation: the reconstructed labour force comes to 26,837.5 against a published 26,837 (error 0.002%) and reconstructed unemployment to 2,010 against 2,014 (error 0.19%). This agreement indicates the estimated breakdown is reliable.
c) Sector employment and the spring 1405 shock
Spring 1404: 33.0% × 25,123 = 8,291; spring 1405: 31.0% × 24,672 = 7,648 ⟹ change −643 thousand.
Rounding error: shares are published to one decimal place (±0.05 ), equivalent to roughly ±12 thousand people. The change observed in industry is 50 times that error, but smaller changes (agriculture +58 thousand) should be read cautiously.
d) Potential primary-school entry cohort
Comparing C(1395) = 7,808,954 with C(1399) = 6,235,729 ⟹ −20.1% across a four-year shift of the window.
Assumptions: near-universal enrolment, negligible child mortality, zero net migration. This calculation does not forecast the level of future enrolment — only the relative shrinkage of cohorts, which is highly insensitive to those assumptions.
e) Population scenarios and the birth–death crossover
t* = ln(B0/D0) / ln( (1+gd) / (1+gb) ) ← crossover year
gb = −5.80% is the actual compound growth rate of births over 1395–1404, and gd = +2.46% the equivalent for deaths; the death figure is confirmed by computing the post-COVID period (1398–1404: 2.56%). The moderated scenario halves the fall in births (−3.5%) to allow for the possible effect of support policies and a floor on fertility. Crossover: 1412 (trend continuation) to 1415 (moderated). Net migration is assumed zero in both scenarios.
4 · What we do not claim
- We do not claim the official statistics are wrong. The Statistical Center's definition of unemployment is the ILO international standard. Our critique concerns the adequacy of that indicator for this economy, not its correctness.
- We do not claim all 5.5 million who stayed inactive want work. Some are certainly students, elderly or retired.
- We do not claim the Statistical Center's projection is wrong. We say its assumptions have drifted from the last two years of data and warrant revision.
- The extrapolated scenarios are not forecasts; they are sensitivity tests and are highly sensitive to the assumption that the trend continues.
- The spring 1405 analysis rests on a single quarter. It is a solid finding but needs confirmation from subsequent quarters.
- No causal relationship is claimed; the marriage–birth correlation is a time-lagged association, not proof of causation.
5 · Tools and workflow
- Extraction and testing: a Python script using openpyxl read the sheet, and an automated consistency test was run over every table with a "total = components" structure. The nine inconsistencies reported in section 2 are the output of that test.
- Calculations: every derived indicator was computed in code and checked by cross-validation (section 3b). No figure is quoted from memory; all are computed from the file itself.
- Implementation: the charts are drawn as raw SVG with no external library, so the file works fully offline.
Responsibility for the accuracy of the data, calculations and inferences rests with the author, Milad Shabani. That is why every key figure in section 3 is shown with its formula and inputs, so it can be audited.
6 · Reproducibility
The cleaned dataset this dashboard runs on is embedded in this very file. You can extract it:
This page is a self-contained HTML file: no external library for the charts, no software to install, and mobile-friendly. The only external resource is the web font, which falls back to a system font offline with nothing breaking.