The Future of Data Management for Tax

Technology tools these days can do amazing things. They can help process and manipulate data or perform calculations that previously took hours or days to prepare; they can help visualize previously hidden insights; and increasingly, they can do so adaptively and intelligently. So why is technology not the immediate answer?

Very simply, because technology itself cannot fix poor data. However, it can help identify it. Technology cannot step in where you don’t have the data you need to begin with. And perhaps most of all, technology cannot require you to use it — you need to make that leap!

For many people, newly embarking on the technology journey, expectations often lie at either of two extremes. There are those who believe that at the click of a button, a perfectly completed and accurate compliance filing will be produced.

And at the other extreme, there are people who are underwhelmed at the technology, seemingly disappointed that their tools may not be as simple as a Google search.

It’s important to recognize that at this stage of the evolution of technology, it’s about making incremental gains which, progressively add up over time to transform the Tax function. Furthermore, it’s also critical to ensure that the technology works in harmony with people and processes.

The immediate answer to your data problems is contained in Proposition 5.

This proposition is merely acknowledging that the data management landscape is getting increasingly complex, and every Head of Tax should be able to navigate their way around it. In a recent compliance discussion for a multinational client, a group of tax professionals sat around a table being asked to provide answers to questions such as:

1. Where is your compliance data stored — on premises or in the cloud?
2. Is that compliance data encrypted both at rest and in transit?
3. What are the tools you can make available to transmit the data to you?
4. What data cleansing tests do you carry out? Do you provide the results back to the company to help correct errors at source?
5. Who will have access to the data, and where will it be located?
6. Do you have tools to anonymize personal data?
7. Do you provide variable access controls, so the company can control the levels of access granted within the organization?

The era of cloud storage, increased data privacy laws (especially GDPR) and cyber security incidents have made these issues become mainstream concerns for the Head of Tax. Proficiency in these areas means being able to ask the right questions, manage risk and to work with subject matter experts.

Of all the propositions stated, this is perhaps the single most important given the extent of organizational change which needs to take place. For many years, Tax functions were the passive recipients of data within organizations. The ‘owners’ of the data were typically the Finance teams. New ERP systems could be implemented, without any active representation from Tax.

Not surprisingly, this led to challenges in securing the right data needed by the Tax function and exacerbated the use of individual spreadsheeting solutions. Data could not be reconciled from multiple systems, and so began the era of the tax professional every month producing detailed computations for reconciliation purposes.
This needs to change.

In December 2020, the OECD released its ‘Tax Administration 3.0’² blueprint, and it contemplates a new era of tax administration where tax processes are moved increasingly into business systems, where taxation is moved closer to taxable events, and where tax is embedded into daily lives and businesses as though they are ‘natural systems’. Put simply, the enabler for all of this is the ability to harness trusted tax data.

As we will explore in Propositions 7 and 8, accountability for the data is growing increasingly important because the data is what will effectively be audited.

²Source: OECD (2020), Tax Administration 3.0: The Digital Transformation of Tax Administration, OECD, Paris.

As we learned from Proposition 5, the Tax function needs to take a more active role in demanding the data they need, and to become active users of it.

In those organizations which are well advanced in their transformation journey, a common feature is the appointment of a Chief Data Officer for Tax, or someone with equivalent responsibilities. These roles will grow in the years to come as technology and digitalization take hold across more and more disciplines.

Broadly speaking, the role of a Chief Data Officer for Tax is to:

1. represent the Tax function in any internal organizational discussions around ERP system changes or upgrades, and manage continued access and retention of data in legacy systems;
2. identify and work with key stakeholders to obtain the data needed by the Tax function;
3. ensure that the data which is collected at source can be trusted;
4. work with IT and other professionals to bring together data from multiple source systems;
5. ensure that data is stored for appropriate retention periods required by regulations;
6. be able to retrieve data effectively and efficiently in response to tax authority queries.

Interestingly, in a September 2021 KPMG benchmarking survey³ of transaction tax teams around the world, 44 percent of those teams were comprised of non-tax specialists. This reflects the continued evolution of skillsets being required to manage Tax functions. The appointment of a Chief Data Officer for Tax represents a clear next step in this evolution.

³Source: 2021 KPMG Global VAT/GST Benchmarking Survey, KPMG International.

Remember the days when tax authorities would randomly select an audit target, serve notice of an intended tax audit, provide you with a month or so to get your information together, and then meet with you for a gentle fireside chat to talk about how to resolve the issues. We believe the days of this happening are numbered.

To highlight the flaws in this approach with an analogy, imagine you are on a flight with 250 people landing in a foreign country. There are two customs officials designated at the immigration counters to check the paperwork of those landing, carry out searches and inspect for illegal contraband or undeclared goods.

If you are tasked with running this operation in a way which produces the highest benefit for the lowest investment in resources, would you use those scarce resources to carry out a cursory check of the baggage of a random sample of people in the hope of success? Or would you use analytics to identify higher risk travelers, such as those without an obvious reason for travel, those who pay cash for their ticket at the airport or those traveling from high-risk locations. The approach you choose can make the difference between a 1 percent success rate on a random sample, or a 30 percent success rate based on a high-risk group. Of course, you would likely choose the latter.

Tax audits are no different to the airport analogy. Tax officials will use data points they already have to risk profile higher-risk transactions or events and analyze that data to determine their audit targets. Furthermore, with the advent of real- time reporting and other similar initiatives, the tax authority may even know the answer to their query before they even contact you.

In short, the tax audit has already begun before you are even notified.

In the global arms race being carried out around tax technology, we believe the indisputable winners will be the tax authorities. They have both the most to gain and the most to lose. Their advantage lies in the economies of scale in any technology tools they invest in, plus they typically have the right to require others to use their tools.

For several years now, KPMG member firms have been alerting clients to the progress being made by tax authorities in their data and analytics tools, and in the capabilities of their people to carry out analytics. The tipping point is fast approaching. What is contributing to this tipping point moment is that tax authorities are increasingly requiring taxpayers to provide transactional data to be transmitted to them, through measures such as real-time reporting or electronic invoicing. Even measures such as DAC 7 out of the European Union is designed to tap the data of digital platform intermediaries in a bid to manage the compliance obligations of (higher-risk) participants who sell their goods or services through those platforms.

Historically, tax authorities were limited by the (limited) data they requested from taxpayers. However, today, the authorities increasingly have your data, and they are progressively improving their ability to analyze it.

Do you know what stories are sitting in your data?

Of all the propositions, this is perhaps the most confronting to tax compliance professionals. Will the current cohort of tax compliance professionals go the same way as the door-to-door encyclopedia salesperson, the VHS or DVD store operator, or even the milkman who once delivered your milk?

We think the closest analogy to the change impacting tax compliance is the comparable changes witnessed to traditional journalism over the past 10 years. The onset of digital newspapers initially heralded a period of considerable uncertainty, with many pundits predicting its demise. However, what has emerged is a number of new business models — there’s a host of new ‘journalists’ in the media market, including podcasters, bloggers and even a whole new generation of ‘influencers’, and the channels for communication with the public have broadened considerably too.

The point here is simple — we anticipate that tax compliance as it is known it today will die. It will likely cease being an ‘after-the-fact’ event and will evolve into a real-time or even a pre-emptive activity, but as it dies it will likely spawn a new range of skills required to support these new activities. As a result, data science experts, systems and process designers, and controllers will rise and become the new leaders in the profession.

An interesting by-product of this change is the expectation that all taxes will be interconnected to a greater degree — for example, the discipline involved in bookkeeping will more closely tie to indirect tax, which in turn will tie to statutory accounting and corporate income tax. The connection that each of them will have is to data management and, in particular, the repurposing of data or outputs from one function to another.

If we think back to Proposition 10 for a moment, if data management is to replace traditional tax compliance, then why would any taxes continue being an ‘after-the- fact’ obligation. Why would those taxes not become real time?

Take the example of a VAT or GST. In a typical supply chain involving a manufacturer, wholesaler, retailer and end consumer, there are three different sets of output tax paid on the ‘value added’ at each stage of the supply chain. There are also three different parties claiming input tax for the costs they incur in their stage of the supply chain. That’s an awful lot of money flows taking place through the tax system solely to collect the VAT or GST on the sale to the ultimate consumer. Whereas historically the rationale for this was to minimize fraud, nowadays the multi-stage credit offset nature of a VAT or GST arguably enables fraud. There should be a better way.

We believe the better way is that each of these obligations will be determined, imposed and collected on a real-time basis. After all, why would the tax authorities wish to take on fraud or credit risks when there are tools or solutions which can bring about timely and efficient collections. It’s similar to one of the reasons why governments often impose withholding taxes — they do so in many cases because it’s more effective to collect the tax on a real-time (or near real-time) basis.

Along similar lines, transfer pricing these days is inherently much more real-time data-driven than in the past — concepts of operational transfer pricing are driving what previously occurred by way of after-the-fact adjustments into real-time transactional level outcomes.

What flows from all of this is that if the determination, imposition and collection of taxes happen on a real-time basis, errors in the data used to calculate those taxes will also become a lot less tolerable. Penalties will be imposed based on those data errors with minimal leeway given for correction.

To end on a high note, our last proposition looks at the question of whether highly complex decision-making can be automated through the use of data techniques. Our unequivocal view here is yes.

Let’s explore this further through the lens of a commonly occurring decision, such as whether 10,000 separate items of expenditure are deductible for corporate tax purposes or not.

1. In making these decisions, taxpayers will often need to demonstrate they have incurred the relevant expenditure. Optical character recognition technology can be deployed to read the relevant contracts and invoices, and analytics carried out to determine if this condition is met.
2. It may be necessary to consider whether the relevant expenditure is for a private or non-business purpose. Plainly, the categorization or identification of certain expense codes or supplier types used in a system can assist in this decision-making. Higher-risk areas such as hotel expenditure, gifts, entertainment and other expenses can be evaluated through analytics tools.
3. There is often a need to determine if the relevant expenditure is to be capitalized or is immediately deductible. Even for an area like this which may typically draw on years of training and experience, machine learning can be deployed in support. For example, in exercising their judgment, tax professionals will typically do so through a framework built around the evaluation of certain factors or indicia. Once those factors or indicia have been identified, it is possible to apply learning tools to copy or mimic the decision-making of those tax professionals.

Finally, even if automation cannot be deployed to solve 100 percent of the decisions made every day, the reality is that the use of independent thought and judgment typically only occupies a fraction of the functions performed all day. Put simply, it is fiction to believe that there are professionals who simply sit at their desk all day making decisions which require their core subject matter expertise. Many of the processes and tasks which surround them but occupy considerable parts of their day can be automated.