Pakistan's manufacturing sector is routinely described as being in decline. Shrinking shares of GDP, stagnant export baskets, and low industrial investment are cited as evidence of deindustrialisation. But a competing explanation deserves serious scrutiny: that the decline is at least partly a measurement artefact.
This raises an important question: Is it a genuine structural decline or statistical invisibility? Such statistical invisibility can occur through the systematic exclusion of informal, unregistered, or subcontracted enterprises.
The Census of Manufacturing Industries (CMI) and the Quantum Index of Manufacturing (QIM) are the two primary industrial data sources for Pakistan. However, these sources have structural limitations that exclude informal, small-scale, and unregistered producers. If these producers represent a substantial share of actual manufacturing activity, then the official picture of deindustrialisation is incomplete. The evidence from Gujranwala suggests that both are occurring simultaneously, but the policy response is focused entirely on the first, while the second goes unaddressed.
Gujranwala is the third-largest industrial city in Pakistan after Karachi and Faisalabad. The city is home to over 20,000 industrial units, more than 500,000 workers, and contributes roughly USD 2.5 billion in annual exports, as reported by the Gujranwala Business Centre.
The city is also unique for its diversity in manufacturing products such as ceiling fans, motor pumps, ceramic sanitary ware, steel pipes, auto parts, and light engineering goods. Approximately 7% of the national GDP and PKR 40 billion annually in tax revenue comes from this city.
Considering the base design of the CMI and the QIM, Gujranwala’s large industrial base is invisible to official statistics. The CMI only collects data from establishments registered under the Factories Act 1934.
This means the CMI covers only those industrial units with ten or more employees. Gujranwala has 13,000 cottage industries and thousands of subcontracted component producers, many of which operate below the ten-worker threshold and are therefore excluded.
According to the Pakistan Economic Census, there are 696,558 manufacturing establishments in Pakistan
The QIM still relies on weights from the outdated CMI 2015–16. This measurement scale assigns marginal importance to several products, such as light engineering and metalwork. Moreover, the ceramic sector employs 36,000 people in Gujranwala alone but contributes only a statistical rounding error.
The problem is not limited to Gujranwala; surgical instruments in Sialkot, textile SMEs in Faisalabad, and pharmaceutical supply chains in Lahore go through the same data collection and measurement errors.
Multiple structural deficiencies are causing the measurement failure. First, the CMI is a decade out of date. The last census was conducted in 2015–16, which was also delayed due to administrative barriers and a lack of coordination between the Pakistan Bureau of Statistics (PBS) and relevant provincial ministries.
The QIM weights, Small Scale Manufacturing (SSM) estimates, and national manufacturing accounts are based on this outdated baseline. A decade of structural development and industrial restructuring is invisible in official numbers.
Second is the coverage threshold of at least ten employees. This excludes the majority of industrial units. According to the Pakistan Economic Census, there are 696,558 manufacturing establishments in Pakistan.
The CMI covers only a fraction of these units due to structural limitations. Small and Medium Enterprises Development Authority (SMEDA) estimates that there are 5.24 million SMEs in Pakistan, contributing approximately 40% of GDP. Yet their statistical foundations rely on surveys and imputations rather than direct measurement.
Third, there is no integration between government data systems. Tax records from the Federal Board of Revenue (FBR), utility connections from DISCOs and Sui Northern Gas Pipelines Limited (SNGPL), and customs export data exist in separate institutional silos. Gujranwala’s USD 2.5 billion in exports cannot be reconciled with the CMI production data because many producers are not registered.
To address these statistical blind-spot problems, Pakistan urgently needs a new industrial census alongside cluster-based surveys for major manufacturing hubs such as Karachi, Faisalabad, Gujranwala, Sialkot, and Gujrat. In this new census, the threshold should be lowered to five employees, with a dedicated small-units module included.
Furthermore, the FBR, PBS, and utility providers should establish a data-sharing protocol by the end of FY2026, creating a real-time industrial unit framework. Since informal manufacturers’ exports already flow through formal customs channels, linking records would provide a strong analytical understanding of the current coverage gap.
At a time when evidence-based decision-making has become the driving force, Pakistan’s industrial data systems remain outdated and fragmented. Before declaring the industrial sector a lost cause, it is important that policymakers first ensure they are measuring the sector accurately.