Intangible Capital (Brynjolfsson)

In the AI Story

The measurement crisis operates at multiple levels. At the firm level, companies cannot manage what they cannot measure — a firm unable to see its intangible capital formation cannot identify which investments produce the most value or allocate resources toward them. At the national level, productivity statistics designed for industrial output systematically undervalue intangible investments, leading to policy decisions calibrated to a partial and misleading picture of the economy. At the investor level, the gap between book value and market value grows ever harder to explain without invoking intangibles explicitly, creating information asymmetries that distort capital allocation.

Brynjolfsson's proposal of GDP-B illustrates the magnitude of what standard metrics miss. Through choice experiments, he estimated that the median American would require over $17,000 per year to give up search engines alone — yet GDP values search engines at their market price of zero. The gap between measured value and actual value suggested GDP was systematically understating the economy's output by significant margins, even before AI entered the picture.

The AI transition amplifies the measurement failure along every dimension. Output quality changes that AI enables — more maintainable code, higher-quality decisions, expanded creative range — show up only weakly in standard metrics. The organizational learning occurring as millions of workers develop new skills is invisible. The institutional adaptation being built as organizations redesign processes around AI capabilities cannot be captured by frameworks designed to measure physical throughput. The economy is investing heavily in intangible assets whose returns will materialize over years and decades — and the investment is invisible to the metrics that guide policy.

The policy consequence is underinvestment. If national statistics show slow growth during the AI dip, policymakers will be reluctant to make the large-scale investments in education, infrastructure, and institutional adaptation that the transition requires. They will point to modest measured returns as evidence the technology has not proven its value. They will not understand that the value is accumulating in intangible form, building beneath the surface like geological pressure before an earthquake.

Origin

Brynjolfsson's work on intangibles built on the broader literature in intangibles economics — particularly Charles Hulten and Leonard Nakamura's work on intangible investment measurement — but integrated it with his firm-level empirical research on IT and productivity. His 2002 Brookings paper with Lorin Hitt and Shinkyu Yang, Intangible Assets: Computers and Organizational Capital, was a landmark in quantifying the magnitude of organizational capital formation that accompanied IT investment.

The framework connects to Jonathan Haskel and Stian Westlake's Capitalism Without Capital (2017), which documented the rising share of intangible investment in advanced economies, and to the broader reform movement in national accounting that seeks to make intangibles visible in productivity and GDP statistics.

Key Ideas

Intangibles dominate modern firm value. The book-to-market gap represents capital that accounting systems cannot record but markets clearly recognize.

AI requires intangible complements. The technology's value is realized only in combination with organizational, human, and institutional assets.

Measurement blindness produces strategic blindness. What cannot be measured cannot be managed, priced, or funded at appropriate scale.

National statistics systematically undervalue the economy. The GDP-B thought experiment suggests the gap runs to trillions of dollars annually.

Underinvestment feedback loop. Invisible gains → modest measured returns → reduced political will for further investment → persistent paradox.

Debates & Critiques

Measurement economists debate the best methods for quantifying intangibles at firm and national levels, with competing approaches producing different estimates. A more fundamental debate concerns whether some intangibles are measurable in principle — the argument from tacit knowledge researchers that certain forms of organizational competence resist formalization and therefore measurement altogether. Brynjolfsson accepts the limit but argues that better measurement, even if imperfect, is better than none.