Technology is enabling businesses to perform tasks more efficiently and provide services to customers in ways they never thought possible. Yet these novel processes and products present considerable challenges to multistate businesses trying to determine their overall state income tax liability. States can be incredibly reactive to economic changes, and little guidance exists indicating how businesses should treat activities enabled by technology – especially when it comes to identifying how much tax they owe different states through the apportionment process. This situation creates tremendous uncertainty, as well as the need to consider how state-by-state reactions to economic shifts may affect apportionment.

To understand how apportionment determines corporate state income tax liability, it’s helpful to take a step back and consider the overall corporate state income tax calculation. In most states, corporate income tax is based on a measure of federal taxable income. This base is modified by several state-specific additions and subtractions. In most cases, the modified base is apportioned to compute the tax liability in states where the business is active.

Apportionment is derived from a formula that varies significantly from state to state and can also be industry-dependent. Historically, businesses apportioned income by calculating three separate ratios or factors – payroll, property, and sales – comparing their in-state activity to their activity everywhere else. More recently, and for a host of reasons, states have modified their apportionment formulas, often skewing toward one that heavily weights the sales factor. Many use just a single sales factor. The rules governing the calculation of in-state activity, known as sourcing, are extremely state-specific.

The impact of automation

Companies’ reliance on automation for core business activities will proliferate in the coming years across many industries. The shift from dependence on employees who perform manual, repetitive tasks to robotic support has the potential to uproot a business’ apportionment calculation – changing both the payroll and property factors. Companies don’t have to pay robots, of course, which decreases the effect of the payroll factor. One way businesses may react is by concentrating their employee footprint in just a few states. If those states base apportionment on the payroll factor, overall apportionment to those states may rise, increasing the company’s tax liability.

Let’s turn to the property factor. Robots may supplant employees, yet they don’t draw a salary. What’s more, states would be hard-pressed to characterize payments for the automated processes robots perform as “payroll.” Instead, the property factor comprises the value of these items. While the amount of such property is likely to increase nationwide over time as automation takes hold, the apportionment implications are not uniform. Companies with greater exposure to states that use the property factor for apportionment will feel the effects most profoundly.

Once far-off but increasingly realistic advancements like deliveries by drones or driverless cars further complicate the scenario. Businesses would need to account for robots’ movement across state lines, and this mobility could lead to fluctuations in property factor values. Savvy businesses would be well-served to model the effect of relocating their employee base from states that base apportionment on the payroll factor to states that employ the sales factor.

Sales is no simpler

Businesses are not only using technology to make their own operations more efficient – they’re also selling solutions based on artificial intelligence, blockchain, and robotic process automation to customers. But sales factor considerations that may create dramatic state income tax consequences accompany these potentially lucrative revenue streams. The type of product being sold is a key determinant of how revenue is reflected in the sales factor. If it is considered tangible personal property, the sale is generally sourced to the customer’s location. But if a product is considered a service or an intangible, more complex and inconsistent sourcing rules apply. In some cases, the location where the service offering was developed or performed controls for sales factor apportionment; in others, states may require market-based sourcing measures.

The first question companies must consider is the type of product being sold. But even this seemingly simple starting point has gray areas. Take computer software. Even though it has existed for some time, states do not take a consistent approach to its classification under the sales factor. Some treat canned software as tangible personal property, while customized software is classified as a service or intangible. Some states reflect the entire amount received for the sale of software (including ancillary services and products) as one type of product; others take a bifurcated approach. These distinct approaches result in remarkable inconsistency, and in some cases will result in two states – or no states – staking a claim to the sale.

Without dismissing this complexity, assume all states treated the total revenue received by a business for an AI software product as a service. The second question for companies to grapple with remains: how should they apportion the income? This one requires a close examination of sourcing methods and an unavoidable state-by-state analysis. Determining where a business has developed or performed the activities that ultimately delivered the solution to a customer can be painstaking, though the location of the business’ payroll and property may offer clues.

Moreover, while market-based sourcing sounds simple enough – just look to where the customer was billed – in practice, the analysis is far from straightforward. Certain states, for example, eschew billing address as a guide and require companies to identify where a customer recognized the benefit of the service. This can be all but impossible for a company to pinpoint. Is it the customer’s headquarters, which paid for the service and decides where it’s used? The location where it’s actually used? Or perhaps the customer is deploying the software for its own customers’ benefit, rendering the location potentially unknowable?

Embrace uncertainty for modeling, proactivity

It won’t happen overnight, but states will eventually modernize their corporate income tax laws to align with the fast-changing technology landscape. They’ve already started to address how to source sales of infrastructure-, platform-, or software-as-a-service, just a few years after these products emerged. But until then, expect an ongoing lack of standardization in states’ approach to these issues, as each is driven by a different set of considerations to optimize revenues generated from corporate income tax. Some states might deem the replacement of human workers with robots a negative development they don’t want to encourage. In response, they could institute special tax regimes to discourage employee-replacing automation as an incentive for states to preserve their revenue bases and foster growth.

Businesses have a unique opportunity to turn uncertainty to their advantage as states contend with how technology reshapes apportionment. First, they should model how increasing automation, or the development of a significant new revenue stream, may alter their presence in a state. If guidance is lacking or inconclusive, the business could adopt an apportionment approach for its corporate income tax return filings that it deems reasonable, easy to determine, or financially prudent. Or, as an alternative, the business could proactively advocate for a reasonable approach by providing a roadmap of its proposed sourcing method in private letter ruling requests to states prior to filing its return. In any event, it seems apt for companies to apply innovative approaches to the sourcing intricacies of innovative activities that increasingly define the business landscape.

Jamie Yesnowitz

Jamie Yesnowitz

Jamie Yesnowitz is a principal and SALT National Tax Office leader at Grant Thornton LLP.