A new study by economists at Goldman Sachs has discovered something that should come as no surprise to most workers: losing your job to new technology can seriously mess up your life in ways that will cause ripples of hardship for a long time.

It’s easy enough to follow the logic here. Losing your job in a capitalist system comes with notoriously crappy financial consequences, leading to both short- and long-term impacts while you scramble to restore your income and put bread on the table.

But in studying labor market impacts of previous technological upheavals, like the computer rush of the 1980s, the analysts found workers displaced by new tech can have a harder time finding a new job and recouping lost income than those let go for other reasons. That bodes poorly for AI job loss, the economists argue, which could be even more intense depending on who you ask.

Even if the displaced worker finds a job, those short term impacts have a way of echoing for years, a term the authors call “scarring.” In previous real-world scenarios, technological displacement led to delayed homeownership, lower lifetime income, and even a lower chance of marriage.

And when we say years, we mean it. In the decade following tech-driven layoffs, the economists found, displaced workers experienced earnings growth at a rate nearly 10 percent slower than their peers. In other words, on top of frustrating your immediate financial situation, technological upheaval similar to the kind AI represents can have a permanent impact on how much companies are willing to pay peons for certain skills.

“Overall, these patterns suggest that AI-driven displacement could impose lasting costs on affected workers, with substantially larger effects when job losses coincide with a recession,” economists and report authors Pierfrancesco Mei and Jessica Rindels wrote, per the WSJ.

There are some strong caveats here. The term “scarring,” for example, implies that this is some natural process, rather than a social one. Policy choices like mandated severance, automation taxes, work placement programs, and democratic ownership over workplace decisions can drastically change these outcomes for workers.

That the threat of AI automation comes at a time when we don’t have many of those systems in place may or may not be a coincidence. But the status quo doesn’t have to remain in place. Technology doesn’t need to lead to mass layoffs or poverty. These are policy decisions — and policy, luckily for us, is never set in stone.

More on AI automation: AI Expert Says It’s Time to Stop Freaking Out About AI Taking Our Jobs