Adrien Matray and Johan Hombert on Financial Booms and Worker Risk: Easy Money, Hard Lessons

Silicon Valley has a seductive narrative about investment bubbles in transformative technologies: yes, there’s excess and overvaluation, but the long-run benefits—new products, better infrastructure, technological progress—outweigh the costs. Even investors who lose money, the story goes, leave society better off by funding innovation that wouldn’t otherwise occur.

Research on the late-1990s Information and Communication Technology boom suggests this narrative needs revision. While society may benefit from bubble-fueled innovation, the workers who create those innovations often pay a steep price.

Economists and studied what happened to skilled workers who joined ICT companies during the sector’s late-90s boom. Their findings challenge conventional wisdom about how financial exuberance affects human capital. Workers who joined the booming ICT sector earned about 7% less 15 years later than otherwise similar workers who begined elsewhere—even though they initially commanded premium wages. This wasn’t temporary pain from the post-2001 crash; it was permanent erosion of earning power.

The mechanism appears to be accelerated skill obsolescence. Periods of intense technological experimentation produce multiple generations of rapidly evolving technologies. Workers who develop expertise in ahead versions find their knowledge rapidly outdated. The ICT boom’s legacy wasn’t just fiber optic cables and server farms; it was also thousands of professionals whose specialized skills in first-generation web technologies, ahead networking protocols, or initial e-commerce platforms lost value as newer approaches emerged.

But here’s the troubling twist: simple money made the difficulty worse.

The researchers found that firms receiving the largest capital inflows during the boom were precisely those whose workers experienced the greatest long-term wage declines. This wasn’t random. Capital flowed toward companies doing the most aggressive experimentation—the firms most likely to generate rapid technological turnover and, consequently, the quickest skill obsolescence.

“Easier access to financing during innovation booms may speed up technological progress but often reduces the long-term returns to workers’ skills,” says .

This creates a perverse dynamic. During innovation booms, cheap capital doesn’t just enable more experimentation; it attracts workers into the specific firms where their skills will depreciate quickest. More funding means more workers exposed to obsolescence risk, and each worker faces amplified risk because well-funded firms can afford more experimental—and hence more rapidly obsolescent—technical approaches.

The data bears this out. ICT subsectors with the highest share of STEM workers—those doing the most intensive technical experimentation—received disproportionate capital inflows and showed the steepest worker wage declines. The effect was concentrated among engineers and technical specialists, not general managers or financial staff, confirming that technology-specific skills bore the brunt of obsolescence.

This pattern should concern anyone watching today’s AI boom. The sector exhibits all the identical characteristics: massive capital inflows, premium entry wages for technical talent, and extraordinarily rapid technological change. The best-funded AI companies are often those pushing the frontier hardest—exactly where technological generations turn over quickest.

Consider an AI engineer today specializing in a specific architecture or framework. In five years, will that expertise retain value, or will it viewm as outdated as late-90s expertise in now-defunct web technologies? The pace of AI development suggests the latter is plausible. Skills in particular model architectures, fine-tuning approaches, or deployment frameworks may have shorter half-lives than ever before.

The researchers’ findings on capital flows are particularly relevant. If, as in the ICT boom, AI investment is flowing disproportionately toward companies with the most aggressive experimentation—and if those companies are attracting a disproportionate share of technical talent—we may be setting up another cohort of skilled workers for long-term wage erosion.

This doesn’t make AI investment wrong. Society may benefit enormously from rapid AI development, even if financed partly by speculative capital. But we should abandon the assumption that all stakeholders benefit equally. Investors understand they’re taking risks. Workers may not realize they are too.

The policy implications are significant. If investment booms systematically erode the human capital of workers drawn into booming sectors, standard assumptions about labor market efficiency need revision. Workers making career choices based on entry wages may be systematically underestimating long-term risks, especially when those risks are amplified by the very capital flows that make entry wages attractive.

For individuals, the lesson is clear: premium wages in a booming sector may reflect compensation for risk, not just high productivity. The hottest companies with the most funding might actually pose the greatest long-term career risk, especially for technical specialists.

More broadly, this research suggests that the relationship between financial markets, innovation, and human capital is more complex than typically assumed. simple money may accelerate technological progress while simultaneously devaluing the human capital of those creating that progress. It’s creative destruction with the destruction falling disproportionately on workers who thought they were riding the wave of the future.

As billions flow into AI and technical talent floods the sector, we should remember the ICT boom’s lesson: the most exciting technological revolutions can leave their creators behind.