Tax digitalisation should not only be linked to digital service tax, digital permanent establishments, the taxation of crypto assets, or even OECD pillars one and two discussions. In fact, the first consequence of widespread tax digitalisation is the imposition of new policy standards that allow the adaptation of legislation to the possibility of automation and machine reading instruments (ensuring the creation and validation of legal metadata).
Moreover, the design of a data strategy at international, national, or company level may lead us to a scenario where we can use specific modelling tools and select the most useful technologies for each tax problem (under EU or OECD guidelines, for example).
This relationship between tax and technology is not a novelty. However, the recent increase in data availability has allowed for a paradigm shift, considering that for the usage of technology, proper management of data is first required. In fact, what we need is qualitative, raw, and structured data. Data mining and quality issues will follow as part of a data awareness culture.
This data-centric rationale reflects the pre-eminent role and data monopoly that the tax authorities have these days.
Take the example of Portugal, where we have more than 60 ancillary obligations to be filed periodically, not covering tax returns themselves (and adding up to automatic exchange of information, fulfilment of the Common Reporting Standard requirements, and financial information exchange). Until this massive data lake is made public, the unbalance and unfair relationship between the tax authorities and taxpayers will stay untouchable.
On the one hand, the tax community tends to think about digitalisation, robot process automation, artificial intelligence (AI), and machine learning, among others, as if these soundbites would be equivalent. On the other hand, we tend to believe in digitalisation as a science fiction topic that will, sooner rather than later, replace all our human work. After all, if the Deep Blue chess computer beat Garry Kasparov in 1997, what can a computer do in 2022?
In fact, machines (or bots, as they are now named) cannot do a lot. Set your brilliant tax minds to rest that we will not soon be replaced – but we do need to adapt.
What machines are already experts at is the performance – countless times – of several binary tasks (Boolean-valued functions). Thus, it seems advisable to request them to act as supplementary tools in research, HR, filing returns, compliance, etc. Only then, after evolving to a development phase of data mining, may we arrive at a level of legal maturity that will face severe limitations towards what science fiction commands in our imagination.
Nevertheless, a degree of opacity still associated with some AI models is one of the biggest obstacles to its growth. For instance, requiring visibility and auditability of the algorithms ruling tax IT software may become a new international standard.
Adding to the above, ‘merely’ data-driven solutions may work fine as well, provided we ensure explainable design thinking (using decision trees) and transparent solutions as the modern preservation of taxpayers’ rights. This is not the future but the present: the need for translators between IT and tax is creating a new role for tax advisers.
The future is already here
The second big boost in technology democratisation is the low code/no code basic idea, even having features available based on Microsoft Power Apps, with interfaces for commonly used software, and avoiding overengineering the systems (with affordable IT solutions). This allows, for instance, dashboarding from country-by-country reporting to profit and loss clustering, litigation screening, and so many others.
The combination of low code/no code tools with specialised sectorial tools as well as enterprise resource planning (ERP) system integration leads tech-driven companies to a different level of real-time controlling and proactive tax strategy and vision.
There are already quite a few tax/legal start-ups that are booming, such as Blue J Legal, Do Not Pay, Jurimetría, Codex Stanford, E-clear, Taxdoo, WTS Global, Summitto, and Luminance. They are already out there pushing taxation to the boundaries of the current technical limitations.
All these examples – from judicial analytics to decision tree implementation, to machine learning and some AI components governing transfer pricing matters – teach us that in each tax problem there is an opportunity to model a process, improve it, and automate it.
Although it is true that Tim Berners-Lee’s initial idea of the Semantic Web, or Web 3.0, did not flourish, easy communication among different profiles and the agile use of the JSON-LD language (as an example) have allowed significant developments by the biggest players in the world (Google, for example) that will sooner or later extend to tax-related domains (while in Portugal, contact with the tax authorities is mainly covered by XML format files).
Furthermore, there are several use cases from a public sector perspective; take VAT and customs matters as examples. The tax authorities are effectively using machine learning technology for anomaly detection through mirror analysis (cross checking import declarations with export declarations) or real-time processing of VAT inputs to speed up refunds or pre-filled-in VAT returns. Not to mention the chatbots introduced across public authorities to respond to basic queries from taxpayers.
It all boils up to data governance and data awareness as an international standard. Imagine a brave new world where policy options need to be sustained by data, contributing to tax transparency as well as measurement of the economic impact of the options. Thus, technology is able to serve public policy options and tax collection, and is available to tax professionals and suitable to be adapted and enhanced by market needs.
And so, it is the case, my fellow tax professionals: ask not what technology can do for you, ask what you can do for technology!
