For much of its history, Pakistan has measured its economy through estimates, surveys, and guesswork. The informal sector, where most people earn their livelihoods, remained a statistical blind spot. That changed in 2023, when the Pakistan Bureau of Statistics (PBS) embedded an economic census within the country’s first fully digital population and housing census. The result was not only the most comprehensive economic mapping in the country’s history, but also South Asia’s largest digital enumeration exercise: 7.14 million establishments identified, 25.3 million jobs recorded, and nearly 40 million structures geo-tagged.
The significance lies as much in the method as in the numbers. Pakistan moved from clipboards to tablets, from paper registers to real-time dashboards, from fragmented surveys to geo-tagged databases. Field teams equipped with digital devices logged each establishment with coordinates, sector codes, and workforce size. Machine learning and natural language processing were deployed to clean the messy raw data, which was riddled with spelling errors, Roman Urdu, and vague descriptions. At one stage, over half of all establishments were filed under the dreaded “Others”.The PBS response was to build an algorithmic dictionary of more than 8,000 business types, refined through AI and human verification in Lahore, Karachi, and Faisalabad. Classification accuracy rose to around 85 per cent, a first for Pakistan’s bureaucracy.The exercise saved over seven billion rupees compared with running separate surveys. For a country where state machinery is more often seen as lethargic than nimble, it was a striking proof of concept: that digital infrastructure, properly deployed, can compress time, cut costs, and improve coverage at scale.
The portrait of the economy that emerged is sobering. Services dominate, accounting for nearly 58 per cent of establishments, with retail trade alone making up 2.8 million shops across the country. Manufacturing and production, often celebrated in political speeches, account for barely a quarter. The social sector — schools, hospitals, madrassas, mosques — represents 14 per cent. Within this category sits one of the most arresting findings: more than 600,000 mosques, employing over two million people, a workforce larger than the entire factory sector. Cattle farming, with 1.1 million units, also surpasses formal manufacturing.
Employment data reinforce the image of a fragmented service economy. Services employ 45 per cent of the workforce, the social sector 30 per cent, and production just 22 per cent. But the structure is fragile. Only 7,000 establishments qualify as large firms with more than 250 employees; just 35,000 fall into the medium category. The vast majority are micro and small businesses, typically employing fewer than five people. They sustain livelihoods but leave little room for productivity gains, tax compliance, or global competitiveness
The census also brings into view the household economy, which has long been invisible. Around 28.5 per cent of households reported some form of income-generating activity. Tailoring, tuition services, poultry farming, embroidery, beauty parlours, food preparation, and freelance digital work emerged as common occupations. For women and youth, these are often the first points of entry into economic life. Yet here, too, the digital promise faltered. Nearly 40 per cent of household activities ended up in the miscellaneous category, offering little insight to policymakers. Entire swathes of unstructured and mobile activity — street vendors, food carts, seasonal kiosks — were excluded altogether. In a country where the informal economy is the rule rather than the exception, such gaps matter.
The digital backbone of the census deserves both credit and critique. On one hand, Pakistan demonstrated that it could orchestrate a technologically complex, nation-wide data exercise: real-time dashboards tracked enumerators, geo-tagging placed every establishment on the map, and AI tools standardized chaos into categories. On the other, technology exposed the fragility of the system. Algorithms struggled with Roman Urdu and inconsistent spellings. Field verification revealed that even after AI corrections, misclassification lingered. And while machine learning accelerated processing, it could not compensate for the absence of crucial variables. No financial data was collected, leaving policymakers blind to revenues, profits, or investment levels. No gender-disaggregated data was recorded, masking women’s contribution. Inconsistencies across classification tables — schools double-counted in one, hospitals missing in another — undermine reliability.
Yet even in its imperfections, the census signals a turning point. For the first time, Pakistan has not relied on sample surveys or indirect estimates, but on digital counting, establishment by establishment, structure by structure. The implications extend far beyond statistics. A live, geo-tagged business register can serve as the foundation for tax reforms, investment planning, and labour policies. It can illuminate where services cluster, where industry is thin, and where household enterprises thrive. It can also expose inequities in access to infrastructure, finance, or markets. If the same digital public infrastructure were extended to health, education, or social protection, Pakistan could finally bridge its chronic gap between data and decision-making.
The international context is telling. India has conducted seven economic censuses since 1977, with its latest fully digitised. Bangladesh completed its fourth in 2024 with tablet-based enumeration. Japan has long treated its economic census as a routine pillar of governance. Pakistan, by contrast, has managed just one attempt before — an abortive effort two decades ago. Its 2023 census, then, is not just a statistical exercise but an institutional leap, showing that the state can, when pushed, modernise how it knows itself. The risk is that this achievement becomes a one-off, commemorated rather than repeated. Without institutionalisation, the Business Register will quickly go stale, and the hard-won digital apparatus will atrophy.
The census is, in the end, a mirror. It reflects an economy less industrial than imagined, more dependent on services, faith-based institutions, and household labour than policymakers admit. It reveals resilience in informality, fragility in enterprise structure, and gaps that digital tools can highlight but not yet close. Most importantly, it shows that Pakistan is capable of governing by evidence when it chooses to. Whether it continues to count — regularly, digitally, and comprehensively — will decide if this moment marks a statistical footnote or a foundation for economic reform.
Read more: Pakistan Bureau of Statistics – Economic Census Report (http://www.pbs.gov.pk/)
Follow the PakBanker Whatsapp Channel for updated across Pakistan’s banking ecosystem.