The Geography of a Degree

 

What 404,436 M.Phil and Ph.D holders tell us about the landscape of higher education — and about the country that produced them

 

A data-driven analysis of the Population & Housing Census 2023, covering all 136 districts of Pakistan. 

Somewhere in the enumeration books of the 2023 census, 404,436 people told an enumerator that the highest qualification they held was an M.Phil or a Ph.D. They are a very small group. Pakistan has 118.8 million people aged twenty and above; the advanced-degree holders among them work out to one person in every 294, or 341 for every 100,000 adults.

That number is worth pausing on, because it is easy to read it either too gloomily or too cheerfully. Read one way, it is a scandal: a country of 241 million has fewer people with a research degree than live in Sargodha city. Read another way, it is an achievement: two decades ago that figure would have been a fraction of what it is now, and among Pakistanis in their late twenties the density is nearly double the national average. Both readings are defensible. What neither captures is the finding that jumps out of the district-level data most forcefully — not how many, but where.

Because the striking thing about Pakistan's advanced-degree population is not its size. It is its address.

1. A national map that behaves like a city map

Spread the 404,436 across the country's 136 census districts and the distribution collapses almost immediately. Ten districts — seven per cent of all districts — account for 50.6 per cent of every M.Phil and Ph.D holder in Pakistan. Thirty-five districts account for 80 per cent. The remaining hundred-odd districts, home to tens of millions of adults, share what is left.

Figure 1. Each circle is a district, positioned at its headquarters town. Size shows the number of M.Phil/Ph.D holders; shade shows holders per 100,000 adults aged 20 and above. Source: PBS Census 2023, Table 13; author's calculations.

Lahore alone holds 62,072 — more than fifteen per cent of the national total and almost a quarter of Punjab's. The seven districts that make up Karachi hold 31,442, or 69.9 per cent of everything in Sindh. Strip Karachi out and the rest of Sindh — a province of more than 26 million adults, with a proud educational history in Hyderabad, Sukkur and Khairpur — falls to 86 holders per 100,000, a level below Balochistan's.

The concentration curve below is the standard way economists picture this kind of inequality. If advanced degrees were spread evenly, the blue line would sit on the dashed diagonal. It does not come close.

Figure 2. A concentration (Lorenz) curve for M.Phil/Ph.D holders across 136 districts. The bottom half of districts, ranked from fewest holders to most, together hold about four per cent of the national total.

Rates make the same point without the distortion of population size. Islamabad records 1,662 holders per 100,000 adults. Torghar, in the mountains of Hazara, records 23. That is a seventy-two-fold gap inside a single country and a single school system. Forty-six districts — with 14.6 per cent of Pakistan's adults living in them — record fewer than 100 per 100,000. The median district manages 171, exactly half the national average, which tells you how thoroughly a handful of metropolitan outliers pull the mean upwards.

Figure 3. Highest and lowest district densities. Note the two panels use different horizontal scales — the lowest twelve districts would be almost invisible on the left-hand scale.

Look closely at the top twelve and a second pattern appears. It is not simply a list of big cities. Abbottabad (514 per 100,000), Malakand (475), Quetta (473), Lower Dir (448) and Bahawalpur (509) are not metropolitan giants, yet they out-perform districts many times their size. What they share is a university, a cluster of degree colleges, or a concentration of public employment. The map of degrees is, to a striking degree, a map of institutions.

What exactly is being counted here? The census asks for a person's highest completed level of education and offers 'M.Phil/Ph.D' as a single tick-box. It does not separate the two, does not ask which institution awarded the degree, and does not verify anything. This matters enormously for interpretation. Pakistan's Higher Education Commission maintains a directory of recognised doctorates that lists roughly 32,640 Ph.D holders — about one-twelfth of the census figure. The overwhelming majority of the 404,436 therefore hold an M.Phil or MS, not a doctorate. Throughout this article 'advanced degree' or 'holder' means the merged census category. Where the distinction matters — as it does for any claim about research capacity — it is flagged.

2. The rural fault line

Pakistan is still a majority-rural country: 58.6 per cent of adults aged twenty and above live in rural areas. Those areas contain 30.7 per cent of the country's advanced-degree holders. Put as rates, urban Pakistan records 571 holders per 100,000 adults and rural Pakistan records 178 — a gap of 3.2 to one.

The interesting part is that this gap is not the same gap everywhere. It varies so much by province that a single national 'urban premium' is close to meaningless.

Figure 4. M.Phil/Ph.D holders per 100,000 adults aged 20+, by province and by locality. Islamabad's rural population alone out-performs every other province's urban population.

Three things stand out. First, Khyber Pakhtunkhwa's rural areas record 237 holders per 100,000 — the highest rural density of any province, above rural Punjab's 180 and more than five times rural Sindh's 43. For a province usually discussed in terms of educational deficit, that is a genuinely surprising number. Second, Sindh's rural collapse is the outlier of the dataset: at 43 per 100,000, rural Sindh has roughly one-quarter of the national rural rate, and Sindh's respectable-looking provincial average of 168 is almost entirely a Karachi artefact. Third, Khyber Pakhtunkhwa's urban rate of 732 is the highest of any province — higher than Punjab's 713 — which means KP's problem is not that its cities under-perform but that so few of its people live in them: only 16.5 per cent of its adults are urban.

There is a wrinkle worth knowing about. Six districts — Lahore and five of the seven Karachi districts — are classified as entirely urban in the 2023 census, with no rural population at all. Any rural rate calculated for them is undefined, and this workbook leaves those cells blank rather than reporting a misleading zero.

Rural and urban in the census The census does not classify people by how far they live from a city. It classifies the settlement they live in: areas inside a municipal, cantonment or corporation boundary are urban, everything else is rural. Boundaries lag reality. Fast-growing peri-urban settlements around Lahore, Karachi and Peshawar are often still counted as rural, while some sleepy district headquarters have been urban for decades. The urban–rural split is best read as a rough proxy for settlement type, not as a measure of remoteness.

3. Why here and not there?

A reasonable first guess is that this is simply urbanisation: degrees cluster where cities are. The data only partly agrees. Across the 136 districts, the correlation between a district's urban share and its density of advanced-degree holders is +0.32 — real, but modest. Urbanisation explains roughly a tenth of the variation between districts. Nine-tenths is something else.

Figure 5. Urbanisation against degree density. Karachi's districts sit at 100 per cent urban with middling densities; Abbottabad reaches 514 per 100,000 while barely one adult in five lives in a town.

Four mechanisms do most of the remaining work, and they compound one another.

Institutions create their own gravity

An advanced degree is earned in a physical building. Pakistan has roughly 262 recognised universities and degree-awarding institutions, and their distribution is heavily skewed towards provincial capitals and a handful of divisional headquarters. Where a university has stood for decades — Peshawar, Multan, Bahawalpur, Abbottabad, Jamshoro — the surrounding district shows a visible bulge of degree-holders. Where none has ever stood, the bulge is absent. This is the single most legible pattern in Figure 1.

The state is the largest employer of the highly qualified

In most low-density districts, the people holding an advanced degree are college lecturers, senior schoolteachers, health managers and civil servants. Their locations are set by posting policy, not by market demand. Because senior public posts cluster in divisional and district headquarters, degree-holders cluster there too — which is why the map of degrees so often looks like a map of administrative hierarchy.

Credential inflation in the teaching professions

Since the early 2000s an M.Phil or MS has become the effective entry qualification for a college lectureship and a promotion requirement in many school cadres. A large share of the 404,436 are therefore teachers who took a degree to move up a pay scale. This helps explain the Khyber Pakhtunkhwa rural anomaly: a province with a large public-education workforce, a strong tradition of male out-migration financing study, and a dispersed network of colleges can accumulate rural degree-holders even without cities.

People move — twice

The census records where a person lives, not where they studied. A young woman from Layyah who completes an M.Phil in Lahore and stays there is counted in Lahore. This makes the map partly a migration map. And the migration does not stop at the national border: the Bureau of Emigration recorded about 860,000 Pakistanis leaving for overseas employment in 2023, among them roughly 100,000 classified as highly skilled — doctors, engineers, IT professionals, accountants. Districts that produce well but retain badly will look poorer on this map than they deserve.

Stock versus flow — the most important distinction in this article A stock is how many people hold a degree today. A flow is how many acquire one each year. The census measures stock only. The two can point in opposite directions. A district may be producing graduates steadily and still show a thin stock because they leave. Conversely, a district may produce almost nobody and still show a decent stock because it hosts the offices that employ them. Any policy conclusion drawn from a stock map alone risks building a college where the problem was retention, or funding scholarships where the problem was that graduates never come home.

4. Two gender stories, pointing opposite ways

At the national level, the gender picture is better than most readers will expect. Women are 41.3 per cent of all advanced-degree holders. Female density is 289 per 100,000 against 389 for men, giving a Gender Parity Index of 0.74. In a country where female literacy is 52.8 per cent against male literacy of 68.0 per cent, near-parity at the very top of the education ladder is a genuine and under-reported achievement.

Then you disaggregate, and the achievement turns out to belong to a particular geography.

Figure 6. The Gender Parity Index by district — female holders per capita divided by male holders per capita. Deep red districts have almost no female advanced-degree stock at all.

Punjab reaches a parity index of 0.91, with women making up 46.3 per cent of its holders. Khyber Pakhtunkhwa records 0.31, Balochistan 0.44 and Sindh 0.65. Only six of 136 districts reach or exceed parity — Sialkot (1.27), Faisalabad (1.14), Lahore (1.11), Sheikhupura (1.08), Narowal (1.01) and Sargodha (1.00). All six lie in one contiguous belt of north-central Punjab. At the other end, Washuk, Lower Kohistan, Kolai-Palas Kohistan, Khyber, Upper Kohistan, Upper Dir and Orakzai all record a parity index at or below 0.05, meaning that for every twenty men with an advanced degree there is at most one woman.

The sharpest version of the finding is rural. Rural women record 121 holders per 100,000 against 534 for urban women — a 4.4-fold gap, wider than the overall urban–rural gap. In 64 of the 130 districts that have any rural population, fewer than fifty rural women hold an M.Phil or Ph.D; in 51 of them, fewer than twenty. These are not small percentages of a large number. They are absolute counts you could fit in a classroom.

Figure 7. Women as a share of all M.Phil/Ph.D holders, rural and urban, by province. In rural Khyber Pakhtunkhwa fewer than one in five holders is a woman.

Geography and gender do not offset one another here; they stack. The districts where women are most absent from advanced education are, with few exceptions, the same districts where advanced education is scarcest overall. A rural woman in Upper Kohistan faces both deficits at once.

The Gender Parity Index, in plain terms The index divides the female rate by the male rate. A value of 1.00 means women and men hold advanced degrees in equal proportion to their numbers. A value of 0.50 means women are half as likely as men. It is not the same as the female share of holders. A district can have a low female share simply because it has few holders of either sex; the index corrects for that by comparing like with like.

The reversal hiding inside the stock

Now split the same 404,436 people by age, and the story turns over completely.

Figure 8. Density and female share by age group. The census records no M.Phil/Ph.D holders below the age of twenty.

Among holders aged forty and above, 26 per cent are women. Among those aged 35–39, 35 per cent. At 30–34 it is 43 per cent, at 25–29 it is 53 per cent, and among holders aged 20–24 women make up 63 per cent — a parity index of 1.72. Women are not catching up in Pakistani postgraduate education; in the youngest cohort they have already overtaken men by a wide margin. World Bank figures put the female-to-male ratio in Pakistani tertiary enrolment at 0.96 in 2023, so this is not a census artefact.

Two readings compete, and the honest answer is that both are partly true. The optimistic reading is that decades of investment in girls' schooling are finally arriving at the top of the system. The sceptical reading is that young men with the same academic record are leaving university earlier for jobs or for emigration, so the female majority partly reflects male exit rather than female advance. Either way, the practical implication is the same: the map in Figure 6 is a photograph of the past. The stock is dominated by cohorts educated in the 1980s and 1990s, when female participation was far lower. If current flows hold, the gender map will look materially different by the 2030s — but only in the places that have flows at all.

The age data carries a second message. Density peaks at 601 per 100,000 in the 30–34 cohort and 593 at 25–29, then falls to 222 among those aged forty and above. That step is the imprint of the post-2002 expansion of Pakistani higher education — the HEC years, when university numbers and scholarship funding rose sharply. Whatever else can be said about that period, it is visible in the census.

5. How thin is thin? Pakistan in comparative perspective

Comparisons across countries are treacherous here, because the census category merges M.Phil with Ph.D while international statistics count doctorates only. The honest approach is to show both bounds.

Figure 9. Doctoral attainment in comparative perspective. OECD values are doctorate-holders as a share of adults aged 25–64; Pakistan's two bars are the census M.Phil+Ph.D category and the HEC PhD Directory count, both as a share of adults aged 20+.

Across the OECD, about 1.3 per cent of adults aged 25 to 64 hold a doctorate. Slovenia leads at roughly 3.6 per cent and Switzerland is close to 3 per cent; the United States sits near 2 per cent. Pakistan's merged M.Phil-and-Ph.D category reaches 0.34 per cent of adults. Its recognised doctorate count — the HEC directory's roughly 32,640 Ph.D holders — is about 0.03 per cent, some forty times below the OECD average.

Adjacent indicators tell a consistent story. Pakistan spends about 0.16 per cent of GDP on research and development, against a world average an order of magnitude higher. Its tertiary gross enrolment ratio sits at roughly 12 to 13 per cent, well below India's and far below Iran's or Turkey's. Where Pakistan does better than expected is in reported researchers per million inhabitants — 474 by the World Bank's 2023 figure, above India's most recent reported value — although that indicator is compiled differently across countries and should not be pushed too hard.

Gross Enrolment Ratio (GER) GER is the number of people enrolled at a level of education, whatever their age, divided by the population of the official age group for that level. Because it counts over-age and repeating students in the numerator, GER can exceed 100 per cent. A tertiary GER of 13 per cent means the equivalent of thirteen per cent of the 18–23 age group is enrolled in higher education — not that thirteen per cent of young people will graduate.

The comparative exercise cuts both ways. Pakistan's advanced-degree stock is thin by any international standard. But the thinness is not uniform: Islamabad's 1,662 per 100,000 — about 1.7 per cent of adults — puts a single Pakistani district within sight of OECD-average doctoral attainment, if the merged category is taken at face value. The problem is less that Pakistan cannot produce advanced degrees than that it produces them in a handful of postcodes.

The provincial picture at a glance

Province	Holders	Per 100,000	Female %	Urban %	Districts
Punjab	268,281	405	46.3%	74.3%	36
Sindh	45,006	168	37.5%	89.4%	30
Khyber Pakhtunkhwa	59,481	318	23.5%	37.9%	35
Balochistan	9,647	169	29.6%	64.5%	34
Islamabad C.T.	22,021	1,662	40.9%	55.4%	1
PAKISTAN	404,436	341	41.3%	69.3%	136

Table 1. M.Phil/Ph.D holders aged 20+, Census 2023. 'Per 100,000' is holders per 100,000 adults aged 20 and above. Figures sum exactly to the published national total.

6. What this data cannot see

A census table is a powerful instrument and a blunt one. Five limitations should govern how far any of the above is pushed.

The merged category is the most serious. Because M.Phil and Ph.D are counted together, nothing in this dataset measures doctoral research capacity — the variable that actually matters for national research strategy. A future census that splits the two, or a linkage to the HEC's degree-award register, would be a low-cost fix with a high analytical return.

Everything is self-reported. Educational attainment is recorded from a household member's answer, with no credential check. Misreporting is plausible in both directions and cannot be quantified.

Nothing here measures quality. Two districts with identical counts may have graduates from institutions with entirely different standards, supervision capacity and research output. The census counts certificates, not competence.

Discipline is invisible. A district whose holders are overwhelmingly in education and Islamic studies faces a different development problem from one whose holders are in agriculture, engineering and medicine, but Table 13 cannot tell them apart.

Finally, coverage. Azad Jammu & Kashmir and Gilgit-Baltistan fall outside these tables entirely. District boundaries are those of the 2023 census, so districts notified since — Murree, Talagang, Wazirabad, Kot Addu, Taunsa — remain inside their parent districts here.

7. What follows for policy

Six propositions follow from the data rather than from prior conviction. Each is stated with the finding that motivates it.

Geographic weighting in scholarship allocation

Finding: 35 districts hold 80 per cent of the stock. A purely merit-ranked national competition draws its applicant pool from those same districts and reproduces the map. District quotas, district-weighted merit lists, or reserved seats indexed to a district's current density are the direct lever, and they can be designed without lowering the standard applied within each district's pool.

Treat the forty-six low-density districts as a single cohort

Finding: 46 districts, with 14.6 per cent of Pakistan's adults, record fewer than 100 holders per 100,000. They share a profile — thin institutional presence, low female share, weak retention — that makes them a natural unit for a package: sub-campus or affiliated-college provision, hard-duty allowances for qualified faculty, and dedicated M.Phil seats with a return-service condition.

Concentrate female-participation spending where the deficit is, not nationally

Finding: rural female density is 121 per 100,000, and 51 districts have fewer than twenty rural women holding an advanced degree. The interventions that work are well documented — provision close to home, hostel and transport support, female faculty recruitment, stipends conditional on completion. What the data adds is a targeting rule: the marginal rupee buys far more in rural Khyber Pakhtunkhwa, rural Balochistan and rural Sindh than in Punjab, where the belt around Sialkot and Faisalabad has already reached parity without intervention.

Separate the retention problem from the production problem before spending capital

Finding: the census measures residence, not production. Before any large campus commitment, a district's stock should be read against its actual output — which requires linking this database to HEC enrolment and graduation data by institution. Building teaching capacity in a district whose graduates leave will not raise its stock.

Split M.Phil from Ph.D in the next enumeration

Finding: the HEC directory's 32,640 recognised doctorates against the census's 404,436 implies the category is dominated by master's-level qualifications. Until the two are separated, no census-based statement about Pakistan's research capacity can be trusted.

Use the geography operationally

Finding: this dataset now carries coordinates for every district. Overlaid against university locations, road networks and district development indices, the cold spots in Figure 1 stop being a chart and become a siting decision — which is the point at which analysis turns into policy.

Last word: the question the next census will answer

The 2023 census caught Pakistani higher education at an unusual moment. The stock it recorded belongs largely to men educated in the 1980s and 1990s and concentrated in a dozen districts. The flow it also recorded — visible in the youngest cohorts — is majority female and growing fastest in exactly the places that already have institutions.

Those two facts set up the question for the 2030s. If the flow continues and stays where it is, Pakistan will end the decade with a more gender-balanced advanced-degree population that is even more geographically concentrated than today's. If the flow can be pushed outward — to the forty-six districts below 100 per 100,000, and to the rural women who are almost absent from them — the same decade produces something closer to a national research base.

The census cannot decide which of those happens. It can only make the choice visible, which is what a good census is for.

 

How this analysis was done

All primary figures are computed from Table 13 of the Population & Housing Census 2023 (Population aged 5 years and above by educational attainment, sex, age group and rural/urban), published by the Pakistan Bureau of Statistics for each province and for Islamabad. 'Population aged 20 and above' is the sum of the census age groups 20–24, 25–29, 30–34, 35–39 and 40 & above; the census records no M.Phil/Ph.D holders below the age of twenty, so the 20+ base captures the entire reported stock. Densities are holders per 100,000 adults in the relevant population. District figures were reconciled against the published provincial and national totals for all sixty locality-by-sex combinations and agree exactly, including the national totals of 118,770,232 adults aged 20+ and 404,436 M.Phil/Ph.D holders. District coordinates are headquarters-town locations in WGS84.

References

1.  Pakistan Bureau of Statistics. Population & Housing Census 2023, Table 13: Population (5 years and above) by educational attainment, sex, age group and rural/urban — provincial and district tables. https://www.pbs.gov.pk/

2.  Pakistan Bureau of Statistics. Census 2023 literacy release: national literacy 60.65 per cent; male 68.0, female 52.84; urban 74.09, rural 51.56. Reported in Daily Times, 'Literacy rate reaches 60.65%; gender, urban-rural divides persist'. https://dailytimes.com.pk/1312836/

3.  Higher Education Commission of Pakistan. PhD Directory and PhDs Produced by Pakistani Universities. https://www.hec.gov.pk/english/universities/Pages/PhD-Produced-by-Pakistani-Universities.aspx

4.  Government of Pakistan, Finance Division. Pakistan Economic Survey, chapter on Education (universities, enrolment and education expenditure). https://www.finance.gov.pk/survey/chapter_25/10_Education.pdf

5.  OECD. Education at a Glance 2025 — 'To what level have adults studied?' (doctoral attainment across OECD countries). https://www.oecd.org/en/publications/2025/09/education-at-a-glance-2025_c58fc9ae/full-report/to-what-level-have-adults-studied_c1d483b0.html

6.  World Population Review. PhD Percentage by Country (Slovenia, Switzerland, United States comparative shares). https://worldpopulationreview.com/country-rankings/phd-percentage-by-country

7.  World Bank. Research and development expenditure (% of GDP) — Pakistan. https://data.worldbank.org/indicator/GB.XPD.RSDV.GD.ZS?locations=PK

8.  World Bank. Researchers in R&D (per million people). https://data.worldbank.org/indicator/SP.POP.SCIE.RD.P6

9.  World Bank. School enrolment, tertiary (% gross) and ratio of female to male tertiary enrolment. https://data.worldbank.org/indicator/SE.TER.ENRR

10.  Bureau of Emigration & Overseas Employment, Government of Pakistan, 2023 emigration statistics; summarised in 'Pakistan's economy suffers as brain drain reaches record levels in 2023', WION. https://www.wionews.com/business-economy/pakistans-economy-suffers-as-brain-drain-reaches-record-levels-619158

11.  Pakistan Institute of Development Economics (PIDE). Situation of Brain Drain in Pakistan, with a focus on the Healthcare Sector. https://pide.org.pk/research/situation-of-brain-drain-in-pakistan-with-a-focus-on-the-healthcare-sector/