Having worked in the VAT management system for nearly three decades since 1995, I have had the opportunity to observe numerous policy initiatives and administrative reforms. Despite these efforts, Bangladesh continues to have one of the lowest tax-GDP ratios in the world, a reality that clearly fails to justify many of the measures undertaken over the years. The fundamental reason for this outcome is that reforms were not implemented with proper prioritisation. Consequently, VAT has not been collected in line with the true potential of the economy.
To understand why VAT collection remains below the potential, it is necessary to identify the core problem accurately. The most critical weakness in the VAT management system is not the number of registered entities but the absence of a reliable mechanism for recording sales data. Proper recording of sales data should, therefore, be treated as the highest priority. If sales data are accurately captured, the scope for VAT evasion is virtually eliminated. At the same time, reliable sales data facilitate income tax collection and the preparation of accurate financial accounts. If VAT, income tax, and accounting systems were aligned through proper sales data recording, the impact on the economy would be transformative.
At present, substantial time and manpower are devoted to VAT registration, maintenance of purchase and sales registers, submission of VAT returns, and related compliance activities. However, these processes have not been implemented flawlessly, and the expected outcomes have not been achieved. In contrast, even if one function-namely, proper maintenance of sales records-was carried out effectively, VAT collection would have increased dramatically. Compliance should never be treated as an end in itself; it is meaningful only insofar as it ensures correct VAT collection.
This point becomes clearer when we consider two hypothetical scenarios. In the first scenario, there is no VAT registration, no purchase and sales registers, and no VAT return submission, yet all sales data of all entities are preserved in a central server in real time. In such a system, the correct amount of VAT could be easily assessed and collected solely on the basis of sales data. In the second scenario, all entities are registered, registers are maintained, and VAT returns are submitted, but widespread VAT evasion continues and the correct amount of VAT is not collected. From the perspective of revenue mobilisation, the first scenario is far superior.
A closer examination of the VAT landscape confirms this idea. The real problem in Bangladesh's VAT system is not non-registration. Almost all manufacturers, importers, exporters, large traders, and large service providers are already registered. The entities that remain outside the VAT net are mainly medium and small traders and service providers with relatively low turnover, whose customers are largely members of the general public. Evidence of VAT evasion, however, is frequently visible among registered entities, for example at restaurant counters where sales are often underreported.
In terms of revenue contribution, only about 5 per cent of total VAT is collected at the trading stage, while approximately 50 per cent is collected at the manufacturing stage and about 40 per cent at the service stage. Discrepancies between turnover declared in VAT returns and figures reported in audited financial statements are commonly observed even among large entities. This clearly indicates that the preservation of accurate sales records, rather than the expansion of registration among small traders, should be the primary focus of VAT administration.
Current data further reinforce this argument. About 58 per cent of total VAT is paid by only 109 entities registered under the Large Taxpayers Unit (LTU), VAT Commissionerate. On this basis, it can reasonably be concluded that the top 1,000 entities contribute around 90 per cent of total VAT. As of November 2025, there are approximately 645,000 VAT registrations, meaning that the remaining 644,000 entities together contribute only 10 per cent of total VAT. Of these, 100,000 entities account for just 1.55 per cent of VAT revenue. Even if another 100,000 entities were brought under the VAT net, the increase would be negligible.
By contrast, the economy has the potential for nearly a 200 per cent increase in VAT collection. This potential cannot be realised simply by registering more small shopkeepers and service providers. Instead, if the sales data of the top 25,000 entities were accurately recorded, VAT collection could realistically be doubled.
Despite this reality, special registration campaigns continue to be emphasised. Under such campaigns, VAT registration is often forcibly granted based on minimal information such as name, address, and nature of business. These entities are later required to submit registration forms, trade licences, TINs, rental agreements, and bank certificates to regularise their status. Only after scrutiny and the issuance of an ID and password in the e-VAT system can VAT returns be submitted online. Experience shows that most forcibly registered entities never appear for regularisation. Even among those that do, many fail to submit VAT returns. Those who submit returns often file zero returns or pay negligible amounts of VAT, making little difference to overall VAT collection.
Over time, the number of VAT returns submitted becomes disproportionately low compared to the number of registered entities, necessitating deregistration drives. From registration to deregistration, the administrative workload of VAT offices increases significantly, yet the number of entities paying substantial VAT does not rise. This is because most newly registered entities are small or medium traders and service providers with limited VAT liabilities. Organic growth in VAT registration is therefore sufficient. What is urgently needed is a targeted effort to ensure proper recording of sales data, which would increase the number of entities contributing meaningful amounts of VAT.
Historical experience supports this conclusion. When VAT was introduced in Bangladesh in 1991, a large number of registrations were issued through special campaigns. Eventually, it was found that while there were around 700,000 registered entities, only about 35,000 VAT returns were submitted. Investigations revealed that most non-filers were non-existent entities. Consequently, extensive deregistration activities were undertaken during 1997-98. This experience clearly demonstrates that VAT collection cannot be significantly increased merely by expanding registration numbers; accurate sales data preservation is far more effective.
A similar pattern appears to exist in income tax administration. Currently, there are about 13 million TIN holders in Bangladesh. A large number do not file returns. Among those who do file, many submit zero returns, and among those who pay tax, many pay only the minimum amount. This raises an important question: has the number of taxpayers paying substantial income tax increased in proportion to the rise in TIN registrations?
Recognising the importance of sales data, efforts to record it began in 2004 with the introduction of Electronic Cash Registers (ECRs), followed by VAT software and Electronic Fiscal Devices (EFDs). However, none of these initiatives succeeded. More recently, an API-based, machine-less approach has been introduced, but it too has yet to demonstrate sufficient promise. Over the past 21 years, no visible progress has been achieved in the systematic recording of sales data. This underscores the need to place sales data recording at the very top of the priority list in VAT management.
There are differing opinions regarding the appropriate information technology model for recording sales data. In my view, it is sufficient to capture a limited set of invoice information-such as the names of buyer and seller, item details, price, and VAT amount-in a central server on a real-time basis. Such a model would require less financial investment, less manpower, and less technological complexity.
In an era of rapid technological advancement, recording all sales data in real time is not a difficult task. It is far less complex than the systems operated by global technology companies or even those used in Bangladesh's telecom and banking sectors. This can be achieved using domestic financing, local technology, and existing human resources; what is required is a clear policy decision and strong initiative. During the 1980s, the proposal to introduce a National Identity Card was widely considered impossible, yet it was later successfully implemented and now supports numerous administrative functions. Proper sales data recording can similarly support multiple objectives, including the accurate collection of VAT and income tax.
Recently, 131,000 VAT registrations were issued under a special campaign. An objective assessment of how many of these newly registered entities submit VAT returns and how much VAT they actually pay would provide valuable guidance for future policy decisions. VAT collection cannot be significantly increased by registering small shopkeepers and service providers alone. The number of entities paying substantial VAT does not rise through such measures, as most new registrations involve entities with minimal contribution to total VAT revenue.
In conclusion, the core problem in Bangladesh's VAT management system is not insufficient number of VAT registrations, but the widespread concealment of sales by registered entities. Capturing sales records on a real-time basis would enable the collection of the correct amount of VAT, strengthen income tax administration, and improve the quality of financial reporting. Therefore, ensuring proper sales data recording must be given the highest importance and the topmost priority in the VAT management system of Bangladesh.
Dr. Md. Abdur Rouf is a VAT Specialist who is Chairman of International VAT Training Institute. roufvat@gmail.com
© 2026 - All Rights with The Financial Express
