Can a cost-of-performance state end up sourcing service receipts to the customer's location?

Two undefined terms carry the result

The model cost-of-performance rule has two working parts, and the statutes define neither of them. Utah's former statute is typical: a sale other than tangible personal property was in the state if "the income-producing activity is performed in this state," or if it was performed both in and outside the state and "a greater proportion of the income-producing activity is performed in this state than in any other state, based on costs of performance" (Utah Code section 59-7-319(1)(a) and (b), as quoted in Appeal No. 09-0578). Oregon's statute used the same words before 2018, and so did Pennsylvania's subparagraph 17 for 2011.

The definitions come from regulations, most of them drawn from the same model. Oregon's rule provided that "income producing activity" "applies to each separate item of income and means the transactions and activity directly engaged in by the taxpayer in the regular course of its trade or business for the ultimate purpose of obtaining gains or profit" (OAR 150-314.665(4)(2), as quoted in AT&T Corp., 357 Or 691). Utah's rule defined the activity as "the transactions and activity directly engaged in by the taxpayer in the regular course of its trade or business," and costs of performance as "direct costs determined in a manner consistent with generally accepted accounting principles" (as quoted in Appeal No. 09-0578). Massachusetts described the activity as "a transaction, procedure, or operation" that "results in a separately identifiable item of income" (830 CMR 63.38.1(9)(d)(2), as quoted in TIR 13-12).

Those two definitions are the levers. The first decides the scope of the inquiry: whether the activity is the taxpayer's business as a whole, a product line, a contract, or a single sale to a single customer. The second decides what is counted once the activity is fixed: whether headquarters, product development, management and other shared functions are direct costs of performing a given sale, or indirect costs left out of the comparison. When the activity is the whole business and every cost counts, the receipt goes to the provider's home state, which is the result most taxpayers associate with cost of performance. When the activity is the individual sale and only the costs of delivering that sale count, the comparison often lands where the customer is, because that is where the field staff, local facilities and local vendors who deliver the service sit.

Narrowing the activity to the in-state sale

The Oregon Supreme Court resolved the scope question for its statute in AT&T Corp. v. Department of Revenue, 357 Or 691 (2015). The taxpayer sought refunds on the theory that its income-producing activity was the operation of its long-distance network as a whole, measured by a cost study of the network. The Court described that position as "network-based" and rejected it: "The correct understanding, however, is transaction-based; it examines individual sales to customers." It read the rule's phrase "each separate item of income" to mean an individual exchange between buyer and seller, whether billed per minute or by flat-rate subscription, and held that the analysis "generally begins with each item of income." Because the taxpayer's cost study did not identify the correct income-producing activities or calculate the costs of performance for them, "AT&T thus failed to meet its burden of proof," and the refund denial was affirmed. The Court expressly did not reach the parties' other questions, including how particular carrier charges are counted.

The Idaho Supreme Court reached a similar result for an internet access provider in Cable One, Inc. v. Idaho State Tax Commission (Idaho 2014). The taxpayer argued that its activity was providing internet access across all the states it served. The Court instead defined the activity as providing internet service to Idaho customers, then asked where the activities that produced that income took place. It counted the in-state local office and employee costs and the in-state network capacity purchased from other carriers as Idaho activity, treated a back office in another state as out-of-state activity, found that the greater proportion of direct costs was incurred in Idaho, and sustained the Commission's treatment of the Idaho customers' receipts as Idaho sales.

The Utah Tax Commission applied the same reasoning to education in Appeal No. 09-0578, an initial hearing order covering tax years 2003 through 2006. A for-profit school headquartered in another state operated one campus in Utah. It argued for an "operational" approach under which its overall operations were the income-producing activity and most costs sat at headquarters. The Division argued that the activity was the sale of educational services to individual students. The Commission agreed with the Division and sourced to Utah the tuition the school received "in exchange for providing educational services to its Utah students," finding that "[t]he services are consumed by Utah customers at a location in Utah" and that the direct costs were "primarily, if not exclusively, incurred in Utah." Administrative and curriculum costs incurred elsewhere were indirect, and the school had not shown otherwise under generally accepted accounting principles.

The common thread is that none of these decisions abandoned cost of performance. Each applied it to a narrower activity and a narrower cost pool, and the comparison then favored the state where the customer was served.

The network reading still wins some cases

The transaction-by-transaction reading is not universal. Massachusetts litigated the same question against the same telecommunications taxpayer and lost. The Appellate Tax Board found that the relevant activity was "the operation and maintenance of a global network," and the Appeals Court affirmed, noting the Appellate Tax Board's finding that "it would be nearly impossible to use a transaction-based analysis" (Commissioner of Revenue v. AT&T Corp., 82 Mass. App. Ct. 1106 (2012), a Rule 1:28 decision, as described in TIR 13-12). The Commissioner's response, in TIR 13-12 (August 20, 2013), was to treat the decision as one fact pattern and to state that a transactional approach will often be required under the regulation, depending on what activity creates the customer's obligation to pay.

The two AT&T decisions, read side by side, show what decides the outcome. Both courts applied the same model language. Oregon put the burden on the taxpayer to prove its network theory with a cost study built for that theory and found the study inadequate; Massachusetts accepted findings that a sale-by-sale analysis could not practically be performed. A taxpayer that wants the network reading needs a record showing that its revenue is produced by an integrated operation and that transaction-level costing is impracticable. A taxpayer that wants the transactional reading needs contract-level or customer-level cost records showing where each sale was delivered.

Reading the activity as the customer's receipt of the service

Pennsylvania took the second route, which does not depend on where costs fall. In Synthes USA HQ, Inc. v. Commonwealth, No. 11 MAP 2021 (Pa. Feb. 22, 2023), the taxpayer, headquartered in Pennsylvania, sold research and development and management services to affiliated customers that received them in Pennsylvania, in other states and abroad, and incurred the greater proportion of its costs in Pennsylvania. For the 2011 tax year, subparagraph 17 of the corporate net income tax statute used the model cost-of-performance language. The Department of Revenue read "income-producing activity" as the customer's receipt of the benefit of the service, at the customer's location, and had long applied that reading; the Office of Attorney General, defending the case, argued for the cost-of-performance reading, which would have assigned the receipts to Pennsylvania. The Court held that neither the statute nor the regulations define the key terms, adopted the Department's benefit-received reading over the Attorney General's, and affirmed a remand for calculation of the taxpayer's refund.

The posture is the reverse of the usual audit. The taxpayer, not the state, wanted the customer-location result, because its customers were mostly outside Pennsylvania and its costs were inside. The reading cuts both ways: on the same reasoning, an out-of-state provider whose Pennsylvania customers received its services in Pennsylvania in those years had those receipts sourced to Pennsylvania. Pennsylvania has since legislated the point. Effective January 1, 2014, sales of services are in Pennsylvania "if the service is delivered to a location in this State" (72 P.S. section 7401(3)2.(a)(16.1)(C)).

An example of the scope question

A building-systems inspection company is headquartered in State A, where its engineering, scheduling software, management and billing staff work. It also keeps a field office in State B, staffed by inspectors who perform every inspection for State B customers and who rent local equipment and hire local laboratory services. Both states use the model cost-of-performance rule. State B customers pay the company $5 million a year. Of the total costs associated with those contracts, $1.8 million are headquarters costs allocated from State A and $1.4 million are State B inspector wages, travel, equipment rental and laboratory fees.

On a network or whole-business reading, the activity is the company's inspection business, the shared headquarters costs dominate, and the $5 million is assigned to State A. On a transaction reading of the kind Oregon, Idaho and Utah applied, the activity is each inspection sold to a State B customer, the direct costs of performing it are the State B inspectors and vendors, the headquarters costs are indirect, and the $5 million is assigned to State B. If State A applies the network reading on its return review and State B applies the transaction reading on audit, both states claim the same $5 million, and the double count arises without any state adopting market sourcing.

Where the customer-location result is refused

Virginia shows the limit of the argument when the taxpayer is the one asking for it. Va. Code section 58.1-416(A) still sources a sale other than tangible personal property to Virginia if the income-producing activity is performed there, or if a greater proportion of it is performed in Virginia than in any other state, based on costs of performance. In The Corporate Executive Board Co. v. Virginia Department of Taxation, Record No. 171627 (Va. Feb. 7, 2019), a research and advisory firm headquartered in Arlington sold its services mostly to customers outside Virginia, while the employees who developed its online content worked in Arlington and the content was stored on servers there. Under cost of performance, nearly all of its receipts were assigned to Virginia, while market-sourcing states taxed the same receipts by customer location. The company challenged the assessments under the Commerce and Due Process Clauses and, in the alternative, sought an alternative method under Va. Code section 58.1-421 that would have sourced its sales to customer billing addresses. The Court rejected both. It held that "the existence of duplicative taxation does not, by itself, violate the Constitution," found that the double taxation arose from other states' changes rather than from Virginia's method, and noted that the General Assembly had kept cost-of-performance sourcing for services for decades and had not changed it despite repeated studies and bills.

Virginia's own statute shows the legislative route instead. Section 58.1-416(B) sources service receipts of a debt buyer to Virginia when the debt is collected from a Virginia resident or a business with its commercial domicile in Virginia, and service receipts of a qualifying property information and analytics firm to Virginia when the customer receives the benefit of the services there, regardless of where the provider performs the work. Customer-location sourcing exists in Virginia, but only where the General Assembly has written it in, industry by industry.

Why the older readings still matter

Most of the states in these decisions have since replaced the statute. Pennsylvania and Massachusetts adopted market rules for years beginning in 2014; Massachusetts now places a sale in the Commonwealth if "the corporation's market for the sale is in the commonwealth" (TIR 13-12). Oregon adopted market-based sourcing for services and intangibles effective January 1, 2018 (Oregon Department of Revenue, 2018 Oregon New Law Update). Utah now sources a service receipt to Utah if the purchaser "receives a greater benefit of the service in this state than in any other single" state or foreign country (Utah Code section 59-7-319(3)(a)), a market rule that keeps the all-or-nothing structure of the old test. Idaho moved to market sourcing in House Bill 563 (2022 Idaho Sess. Laws ch. 52, retroactive to January 1, 2022): a service is now an Idaho sale "if and to the extent the service is delivered to a location in this state" (Idaho Code section 63-3027(13)(c)).

Three groups of taxpayers still live under the older readings. The first is anyone with open years under a cost-of-performance statute, on audit or in a refund claim, where the scope of the activity and the cost pool remain the questions. The second is the Idaho communications company. Section 63-3027(15) lets a "communications company," defined in section 63-3027(1)(e) by reference to telecommunications, signal transmission and broadcast businesses that meet ownership and gross-sales tests, elect to source sales other than tangible personal property by the income-producing activity and costs of performance. A company that makes that election in order to keep its receipts at its home state takes the statute with the Cable One reading of it: the activity is the service provided to Idaho customers, and in-state network and office costs count. The third is the taxpayer in a state that has kept cost of performance, where the Oregon, Idaho, Utah and Pennsylvania decisions are persuasive authority for an auditor and Virginia's decision is persuasive authority against a taxpayer seeking relief from the statute.

Practice notes

The first question in a cost-of-performance state is how that state, for the year at issue, identifies the income-producing activity: by the business as a whole, by contract, or by sale, and whether its agency or courts have read the term as the customer's receipt of the service. The second is which costs it treats as direct costs of performance, and whether allocated headquarters, development and management costs are in or out of the comparison. A provider whose field delivery costs sit in its customers' states is exposed to transaction-level sourcing in those states even where the statute reads as cost of performance, and a provider that relies on a whole-business reading is in the strongest position with a record that shows an integrated operation and the impracticability of costing each sale. Cost records kept by contract and by delivery location answer either argument. Where the provider's home state and its customers' states read the same statute in opposite ways, the double count is a modeling item for each open year, and a request for an alternative method is weighed against the Virginia court's view that duplicative taxation alone does not show distortion. An Idaho communications company deciding whether to elect cost of performance tests the election against the Cable One reading before relying on it. G&G State Tax Group provides multistate sales factor sourcing reviews, refund claims and audit representation on service receipts alongside its state and local tax practice.