How We Know the 2022 Shock Changed Firm Behavior

The Stanford study of the U.S. R&D capitalization rule asks a causal question, not merely a descriptive one: did firms exposed to the delayed deduction experience a different change in taxes and investment than otherwise comparable firms that had not yet been exposed? This chapter reconstructs the study’s answer in plain language. The aim is to show what the fiscal-year comparison, its design checks, and its sample can support—and why the evidence is informative without being a randomized experiment or a forecast for every company.

The distinction matters because a simple before-and-after comparison would be difficult to interpret. R&D spending and cash taxes can change for many reasons: the economy, a product cycle, a pandemic recovery, financing conditions, or an industry shock. If December year-end companies happened to cut R&D in 2022, that fact alone would not tell us whether the tax rule caused the cut. The study’s design creates a more disciplined comparison by using the date on which the policy first reached different fiscal-year calendars.

A calendar creates a comparison group

For a December fiscal-year-end company, the fiscal year ending December 31, 2022 was the first fiscal year affected by the capitalization requirement. A company whose fiscal year ended September 30, October 31, or November 30, 2022 was not yet subject to the rule for that fiscal year. Its first affected fiscal year came later. The Stanford researchers therefore compare the December group with the September–November group during the same broad period.

This is a difference-in-differences design. The name describes the logic rather than a special kind of tax calculation. First, measure how the outcome changes for the firms exposed in 2022. Second, measure how it changes for firms not yet exposed in that same fiscal year. The estimated policy effect is the first change minus the second change.

The comparison is useful if the two groups would otherwise have moved in similar ways, or if the analysis can account for important differences between them. That condition cannot be proven directly. It can be examined through the behavior of the groups before the policy took effect and through controls that make the comparison more credible.

The calendar is doing more than sorting companies by month. It supplies a source of staggered exposure inside the same country, under the same federal policy, during the same macroeconomic period. The firms are not separated into “tax-friendly” and “tax-hostile” jurisdictions. They face the same rule, but the rule reaches their fiscal-year accounts at different times. That reduces some sources of comparison noise that would arise from comparing firms across countries or across very different tax systems.

The treatment group is narrower than “all December companies.” The researchers searched financial-statement disclosures and retained December year-end U.S. firms that specifically reported a deferred tax asset related to the R&D capitalization change. The deferred tax asset is evidence that the firm recognized a difference between book and tax treatment associated with future deductions. It also provides a concrete way to identify firms that disclosed being affected, rather than assuming that every firm with an R&D line faced the same magnitude of change.

What the design controls—and what it cannot control

The study includes firm fixed effects.