What fieldwork reveals about Pakistan’s uneven state capacity

A woman in Lahore checks her phone for a Benazir Income Support Programme (BISP) payment alert. The system correctly identifies her as the beneficiary, but the handset belongs to her husband, who sees the message first. Meanwhile, at Karachi port, a trader navigates the Pakistan Single Window (PSW). The portal is slick; but while it might streamline documentation, the discretion in risk assessment has simply moved from paper to screen. ^[1] 

These moments, observed during my summer placement in Pakistan with the International Growth Centre, point to a simple idea: capacity is remarkably uneven - often sophisticated and primitive within the same transaction. This pattern continued to emerge over several weeks spent speaking with bureaucrats, local academics, and policymakers across Punjab. 

Pakistan has built impressive infrastructure for digital identification, payments, and service delivery. The problems - the human interfaces, discretionary decisions, and household dynamics that determine whether these systems actually deliver - emerge at the last mile.

Observing state capacity through transactions

State capacity is revealed through routine transactions and mundane moments where citizens encounter the state, such as checking a balance, clearing goods, registering for benefits, or filing complaints. These interactions reveal where systems function and where they break down. At this level, capacity isn't binary but rather lumpy across different stages of the same workflow.

Understanding that unevenness, transaction by transaction, is the first step toward addressing it. Why do certain transactions work while others fail? What distinguishes successful from failing interfaces? How can friction points become learning opportunities? These questions matter for policy, and as Pakistan expands digital systems and social protection, understanding these patterns could improve interface design, not just infrastructure.

Fieldwork revealed patterns that warrant systematic research

1. Poverty registries disagree on how to classify households: Punjab's Socio-Economic Registry (PSER) and the federal National Socio-Economic Registry (NSER) sometimes classify the same household differently. This isn't just administrative noise; it signals that both systems weight information differently – the federal system uses standardised indicators, while provincial systems incorporate local knowledge. These disagreements highlight the blind spots in each approach.
   
   While research has previously focused on centralised proxies versus local knowledge, Pakistan's dual registry system provides an unusual opportunity to study this systematically without presuming which registry is ‘right’. There is scope to investigate what covariates explain registry divergence, and what this reveals about optimal targeting design.
    
2. Digital transfers take place through shared phones: BISP is intended to transfer money to women, but men often control the phones. This dynamic appeared repeatedly during fieldwork: the alert arrives, the husband sees it first, and he manages the withdrawal. This isn't outright capture, but it isn’t empowerment either.
   
   These observations echo research on intra-household bargaining and digital payments, and a growing literature which shows that payment and device design can affect effective control and outcomes. The scale of the BISP makes this particularly important, since it reaches millions of families in Pakistan. If digital financial inclusion does not automatically translate to female empowerment, small, low-friction interface choices could matter more than we assume – neutral SMS wording, optional voice messages in local languages, and ‘quiet’ balance displays at agents could help improve outcomes.
    

3. Digitisation doesn’t remove discretion: With the PSW digitising customs, clearance times are dropping, but risk assessment and physical examinations remain discretionary. Once the system flags consignments, officers decide whether to inspect them. As a result, human experience continues to guide the identification of suspicious shipments. This isn't necessarily a problem, since experienced officers might spot patterns that algorithms miss. However, it suggests that digitisation relocates rather than eliminates discretion.
   
   Evidence from ports and bureaucratic settings indicates that, depending on incentives and monitoring, such discretion can either help (through tacit expertise) or harm (if officers engage in rent-seeking behaviour) the process. 

   Pakistan's ongoing digitisation efforts raise the question: when discretion moves from paper to digital interfaces, how do information sequencing and time limits affect bureaucratic behaviour and clearance variance?
    

4. Complaints can be used as diagnostic data: Punjab's Ombudsman offices function reasonably well. When citizens file complaints, offices investigate, and departments respond. But the Ombudsman sees patterns and persistent problem types without aggregating lessons. Each complaint contains information about service delivery failures that isn't being systematically analysed.
   
   Best practices followed by international organisations highlight how grievance systems can be treated as operating inputs, not just mailboxes. Punjab's Ombudsman system provides an opportunity to examine how complaint data can identify capacity gaps and predict where service delivery is likely to fail. Publishing fix rates and time-to-resolution for repeat patterns could turn reactive complaint handling into proactive capacity diagnostics, and help target fixes before problems escalate.

What can researchers learn from internal variation?

Pakistan has built the digital infrastructure, but small details determine whether systems deliver. Phone-sharing patterns and household dynamics affect whether women control transfers, what officers see on screens shapes customs decisions, and how complaints get routed determines redress. 

Every point at which processes break down contains data about system failures. A systematic analysis of these friction points could inform capacity improvements faster than sweeping reforms. 

Pakistan's uneven state capacity is typically framed as a problem. For researchers, it can be an opportunity. The coexistence of sophisticated systems and basic failures creates natural experiments in governance. Variations – between provinces within Pakistan, between digital and manual systems, and between successful and failing programmes – are an opportunity to design credible identification strategies and generate valuable insights without requiring cross-country comparison.

Explore IGC's work in Pakistan

[1] The Benazir Income Support Programme is a federal social protection initiative in Pakistan. Launched in 2008 to reduce poverty and empower women, it provides quarterly cash assistance to eligible low-income families. Eligibility is determined through the National Socio-Economic Registry, and payments are disbursed nationwide through partner banks and mobile services.

The Pakistan Single Window is an Integrated digital platform that allows parties involved in trade to lodge standardised information and documents to fulfil all import, export, and transit-related regulatory requirements. It aims to reduce the time and cost of doing business by digitalising Pakistan’s cross-border trade and eliminating paper-based processes.