Can Digital Work Graduate into Export Firms?

Executive Summary. Bangladesh has proved that people and businesses can sell digital services abroad, but current national indicators cannot show how often cross-border work becomes a durable export firm. ICT services accounted for 18.6% of service exports in 2011 and 9.51% in 2024, a decline of 9.09 percentage points. That share does not measure export dollars and can fall when other services grow. Bangladesh Bank provides a level-side anchor: telecommunications, computer, and information service credits were BDT 74,914 million in FY2023-24, against debits of BDT 17,046 million, leaving a BDT 57,868 million surplus. The category represented 10.7% of total service credits. Bangladesh also had 527 secure internet servers per million people in 2024, compared with 5,734 in Vietnam. But export shares, a broad balance-of-payments category, and server density do not count exporter firms, recurring clients, workers, or graduation. The policy opportunity is to build that missing firm-level evidence, then lower the verified barriers to contracts, certification, finance, and scale.

A successful activity lost export weight

The familiar digital story starts with talent. Young Bangladeshis learned to code, design, manage data, and provide remote services. Online platforms reduced the need for a foreign sales office. Payment channels, training programs, and connectivity made cross-border work possible. Those achievements are real.

The export composition, however, did not consolidate around ICT services. Their share of service exports reached 18.6% in 2011 and stood at 9.51% in 2024, 9.09 percentage points lower.

Source: World Bank World Development Indicators, ICT service exports as a share of service exports, 2010-2024.

A share is a demanding measure. It can decline even if ICT exports rise, provided other service exports rise faster. It is also affected by classification and balance-of-payments reporting. The chart therefore does not support the claim that Bangladesh lost half its digital export revenue. It supports a more precise claim: digital services did not become a progressively larger pillar of the recorded service-export basket.

That distinction separates an activity measure from a firm measure. Export shares show the recorded basket. They do not reveal whether sellers repeatedly win larger contracts, train workers, manage quality, protect data, finance receivables, or build reputations beyond platform profiles. Bangladesh's early digital success lowered the entry barrier to cross-border work. The available series leave the second transition unmeasured.

The policy challenge is therefore not to manufacture another headline about freelancers. It is to measure how often successful individuals become teams, how often teams become export firms, and which obstacles are associated with each transition. Incorporation, accounting, sales, security controls, contracts, management, and working capital are plausible fixed costs to test, not findings established by the current indicators. Buyers also require different assurances as contracts become larger: a platform rating may support a small task, while a bank, hospital, or multinational buyer may require an entity that can sign, audit, insure, and remedy failure.

The surplus is real but still narrow

Bangladesh Bank recorded BDT 74,914 million in credits for telecommunications, computer, and information services in FY2023-24. Debits were BDT 17,046 million, producing a BDT 57,868 million surplus.

Source: Bangladesh Bank, Balance of Payments 2023-2024.

This is the strongest counterweight to a pessimistic reading. The category earns more foreign service income than it pays. It is not an imagined sector or a promise waiting to start. Yet the same year, total service credits were BDT 699,822 million, making the digital category 10.7% of receipts.

The Bangladesh Bank category is broader than software exports. It combines telecommunications, computer, and information services. The WDI share uses an internationally standardized ICT service grouping, but the two should not be treated as identical series. Their role here is complementary. The national account shows a positive cross-border balance. The international series shows that ICT's share of the service-export basket is below its earlier peak.

That combination establishes a recorded foothold but cannot determine the structure behind it. A surplus may come from independent workers, small firms, larger companies, telecommunications providers, or a changing mix. The possibility is to learn which paths compound into recurring business contracts and which remain isolated tasks.

Recurring contracts change the economics. A platform assignment ends when the task is delivered. A managed service, software subscription, outsourced process, or long-term data contract can create more predictable revenue and support staff specialization. It also demands stronger guarantees from the seller. The present data cannot show which guarantees most often block Bangladeshi sellers, so contract, security, management, and finance constraints should be measured before programmes are designed around them.

The infrastructure gap is institutional as well as physical

Secure internet servers provide one view of the infrastructure supporting authenticated digital transactions. In 2024, Bangladesh recorded 527 servers per million people. India recorded 1,212, Indonesia 2,939, Thailand 3,150, and Vietnam 5,734. Pakistan recorded 110.

Source: World Bank World Development Indicators, secure internet servers per million people, 2024.

Vietnam's density was 10.9 times Bangladesh's. The indicator should not be romanticized. Cloud services allow a Bangladeshi firm to use infrastructure physically located abroad. Server counts depend on hosting patterns, reporting, population, and the architecture of modern platforms. A low count does not prove that a particular exporter is insecure.

It does reveal the thin domestic footprint of authenticated hosting relative to several relevant economies. That footprint matters for the surrounding ecosystem: data centers, operations staff, security practices, local enterprise demand, disaster recovery, and the everyday expectation that firms transact through secure systems. The comparison is not an argument for forcing data to stay inside Bangladesh. Localization without reliability can raise costs and weaken resilience. It is evidence that the industrial substrate remains limited.

The larger institutional gap includes contract enforcement, data protection, cross-border payment documentation, tax treatment of exported services, professional liability, and certification. A buyer purchasing a recurring service asks where data sits, who can access it, what happens after a breach, which law governs the contract, and whether the vendor can compensate a failure. Training another developer does not answer those questions.

The missing product is evidence about export firms

The indicators available here count export composition, a broad service balance, and secure servers. They do not measure graduation. The missing metric is whether revenue becomes an organization that can employ others, retain clients, and sell a more complex service next year.

Graduation requires working capital because foreign buyers often pay after delivery. It requires sales capability because large contracts do not arrive through the same channels as small tasks. It requires standard legal documents and credible dispute resolution. It requires managers who can price a project, control scope, and protect margin. These are firm capabilities, not coding skills.

The state does not need to build software companies. It can reduce the fixed cost of becoming one. Standard export contracts, clear tax guidance, interoperable payment records, affordable security certification, and credit against verified receivables are public or coordinated inputs. They help firms compete without prescribing what they build.

The distinction from the existing internet-access narrative is important. Household connectivity asks whether a person can reach the network. This essay asks whether an enterprise can make a durable promise across that network. The first enables participation. The second creates an industry.

What would change this conclusion

The graduation question becomes answerable when comparable business data connect the recorded BDT 57,868 million surplus to sellers, repeat clients, employment, and firm survival. A larger ICT share than the current 9.51% would show export composition changing, but it still would not reveal firm graduation by itself.

Three moves would measure graduation and then test what helps it.

1. Create a verified digital-export firm register built from transactions, not claims. Link consented payment records, export declarations, tax identity, employment, and repeat clients in aggregated statistics. Owner: Bangladesh Bank, National Board of Revenue, and Bangladesh Bureau of Statistics. Success signal: the country can explain who produces the current BDT 74,914 million in credits and how many sellers become durable firms.
2. Provide a shared trust stack for exporters. Offer standard contracts, security-assessment support, data-governance templates, and recognized certification through competing accredited providers. Owner: ICT Division and relevant standards bodies. Success signal: small firms can qualify for larger recurring contracts without owning the full compliance apparatus.
3. Finance verified foreign receivables. Allow lenders to underwrite confirmed contracts and payment histories while retaining credit risk. Owner: Bangladesh Bank and commercial lenders. Success signal: firms can hire and deliver before the client pays, and ICT's export share moves above 9.51% without relying on reclassification.

The counterargument

The strongest objection is that the metrics understate a decentralized digital economy. Platform earnings may be misclassified, routed through payment channels that statistics do not fully capture, or earned by workers who sensibly remain independent. Secure servers are increasingly located in global clouds, so domestic density may have little relationship to export capability.

That objection is substantial. It is why the argument does not infer the number of freelancers, software firms, or export dollars from either chart. Measurement gaps could make the recorded sector smaller than the real one. Independence can also be an efficient choice for a professional who does not want employees.

But measurement opacity is itself an institutional weakness. A sector cannot receive well-designed finance, negotiate tax treatment, or diagnose firm graduation if its core transactions are invisible. Global cloud use also strengthens the case for contract, security, and continuity capability, even if it weakens the case for local server construction.

Bangladesh has already demonstrated digital talent and a positive service balance. The unrealized possibility is to see the organizational path clearly, then help repeat work become a trusted vendor relationship and the next contract become larger than the last.

---

Data sources: World Bank WDI ICT export share and secure-server density, plus Bangladesh Bank Balance of Payments, retrieved on the publication date. The two export classifications are related but not identical, and the server indicator is treated as an ecosystem proxy rather than a firm count.

Sources

1. World Bank WDI, ICT service exports, BX.GSR.CCIS.ZS: https://api.worldbank.org/v2/country/BGD/indicator/BX.GSR.CCIS.ZS?format=json&per_page=100
2. World Bank WDI, Secure internet servers, IT.NET.SECR.P6: https://api.worldbank.org/v2/country/BGD;IND;PAK;IDN;THA;VNM/indicator/IT.NET.SECR.P6?format=json&per_page=1000
3. Bangladesh Bank, Balance of Payments 2023-2024: https://www.bb.org.bd/pub/annual/bop/bop_2024.pdf