Every February, Pakistan runs a lottery. The ticket is paid for not in rupees but in the best years of young graduates’ working lives. This year, 43,189 people sat the first stage of the Central Superior Services examination. Perhaps 250 will become officers.
We call this merit. A better name is rent dissipation. My rough central estimate is that the CSS queue burns about PKR 11 billion a year, or roughly PKR 44 million for every officer allocated: more than fifty years of an ordinary graduate’s earnings to select one bureaucrat. The real question is not why so few pass. It is why so many queue, and what the exam measures once they do.
The odds
A candidate who sits the first paper has about a 0.6 percent chance of becoming an officer.
FPSC press release, CSS-2026 written result (31 Aug 2026); allocation estimated from the CSS 2025 final result
The preliminary MCQ test, meant to weed out the unserious, passes 85 percent of those who sit it. It screens almost no one; it simply adds a fee, an exam, and a coaching market to the front of the queue. The written stage passed 2.74 percent this year, and that rate barely moves whether 13,000 or 18,000 sit. The cut-off falls where thousands are bunched together, and subjectively marked essay and précis papers decide most failures. Near the line, rank is substantially luck.
That noise lengthens the queue. If results tracked ability tightly, weak candidates would read their scores and leave. A noisy exam tells everyone they are close, so trying again is rational, cycle after cycle between the ages of 21 and 30, each taking about 15 months from paper to allocation. How many attempts, at what age, after how many years of preparation? FPSC does not publish it.
The bill
Suppose half of the 17,000-odd written-exam candidates prepare full-time: about 8,500 person-years withdrawn from work each cycle. Add some 2,500 more from those who prepare but never reach the written stage.
Value each year at a modest graduate wage of PKR 70,000 a month plus academy fees and big-city rent, about PKR 1 million in all, and divide by roughly 250 officers. Every input is open to argument; the conclusion is not.
Author’s estimate from FPSC 2026 counts, a graduate shadow wage of PKR 70k/month, and published academy fees
Even the low case spends twenty years of a graduate’s earnings per officer.
What the queue says about power
Anne Krueger and Gordon Tullock showed that competitors for a rent keep spending until the prize is dissipated. Run that logic backwards and the queue becomes a price signal. A candidate who spends PKR 2.5 million for a cumulative chance of perhaps 4% must value the post at over PKR 60 million.
Author’s estimate; BPS-17 initial basic pay from the federal BPS-2026 notification
The pay-slip cannot explain that. BPS-17 starting basic pay under the 2026 scales is PKR 54,140 a month, below the graduate wage used here. The value lies elsewhere: tenure no private job offers; official houses, cars, and staff; plots reserved for officers; discretion over postings, permits, land records and policing; post-retirement seats on boards; status in the family and the marriage market; and, for some, returns that appear on no pay-slip at all. Almost every item is a transfer from the rest of society to the holder of the office.
A country that pays more for power than for production will get more people queuing for power.
The costs off the spreadsheet
Misallocated talent. Murphy, Shleifer and Vishny showed that talent flows to wherever returns are highest, and that the rules of the game decide whether those returns come from adding value or from capturing it. Where rent-seeking pays better than producing, the brightest become rent-seekers, and growth suffers twice: from the output they no longer create, and from the rents they extract from those who do. CSS is such a rule. It draws the ambitious top of every cohort with a prize that is not a chance to build anything but a licence to allocate, permit and control.
The losers pay too. Preparation builds exam-specific capital, curated narratives and optionals chosen for scoring, that private employers barely value. Candidates leave at 29 with a CV gap and a lasting earnings scar. For many women, “preparation” becomes a socially acceptable exit from a labour market whose female participation is already among the world’s lowest. And an academy industry has grown whose product is exam technique and whose business model depends on candidates believing they are close.
So what are we measuring?
The design descends from Macaulay’s 1854 report on the Indian Civil Service: select generalist gentlemen by essay and general knowledge, then give them tenure for life.
Much of the left column measures not ability but preparation capital: who can afford years without income. In 2025, 23 of the top 30 came from Punjab. Then a single noisy signal at 25 fixes a 35-year career, with no lateral entry and no performance-based exit. That permanence is precisely why the signal is worth so much.
The last gate: a room with no rubric
After twelve written papers, the final word goes to a conversation. The viva voce carries 300 of 1,500 marks; below 100, a candidate is out. A board judges “moral character” and “suitability”, and the psychological assessment is folded, unscored, into the same 300. Among tightly bunched written-stage survivors, a 200-mark spread can decide who gets the Administrative Service or Police and who gets Audit. There is no published rubric, no recording, no independent second scorer, no meaningful appeal and no published distribution of marks. Years of costly testing in the name of objectivity end with the most valuable decision handed to the least measurable stage.
What to do
The fix is not a harder exam or a softer one. It is a smaller prize and an honest draw.
Shrink the rent. Lateral entry, specialist contract cadres, performance-linked pay and exit; monetise perks and end officer plot schemes. When the prize nears the market wage, the queue shortens by itself.
Make the lottery honest. Set a demanding competence threshold, then select at random among those who clear it. Another year of cramming then buys nothing.
Test what the job needs. Quantitative and case-based policy problems, double-blind essay marking, and a structured, recorded viva with a published rubric.
Publish the data. Attempts per candidate, age at success, viva marks and the score distribution near the cut-off.
Until then, Pakistan will keep running its most expensive lottery, and keep calling the result merit.
Copyright Business Recorder, 2026
The writer served as the Deputy Chairman of the Planning Commission. X: @nadeemhaque; YouTube: @SiaLytics and Substack: Aid, Policy and Growth