That’s according to a working paper circulated by the National Bureau of Economic Research (NBER) and written by economists Dave Atkin and Antoinette Schoar at MIT and Sumit Shinde at UCLA. The researchers observed data entry workers in Chennai, India, who were working either in an office or at home, and found that those randomly assigned to working from home were 18% less productive than office workers. Most of that productivity difference was clear from the first day of work, they wrote; the rest was a result of slowed learning for the remote group over the course of two months. 

A more striking observation: Workers who preferred working from home were substantially (27%) less productive at home than when they were at the office. That productivity gap was only 13% for workers who prefer being in the office. Atkin, Schoar, and Shinde chalk that discrepancy up to those who are less likely to prefer in-person work owing to demands at home, such as childcare or eldercare, which can gobble up time throughout the workday.

This data only adds fuel to the fire for bosses who are unwavering in their stance that being physically together is vital for productivity, collaboration, and teamwork. Elon Musk has said workers are “pretending to work”; Morgan Stanley CEO James Gorman has worried that the price of remote work is productivity; Mark Zuckerberg suggested that new hires fare better with three days of in-person work alongside colleagues each week; and Salesforce CEO Marc Benioff said that new hires “do better if they’re in the office.” Meanwhile, remote workers largely insist they’re more productive in their own environments; they certainly work more hours.

But the latest data points in favor of bosses. “Workers who most need to be around the house during the workday are exactly the workers who are most distracted working from home,” Atkin explained to Fortune. The extent of that distraction can vary based on the differences in amenities between one’s home and office, commuting challenges, house size, or in each country’s access to childcare. 

Workers for whom it isn’t much of a problem to come to the office several days a week (or even say they’re more productive there) are, naturally, more likely to show up. That’s a “sorting device” that doesn’t even require a boss’s intervention, the researchers note, and stands to increase the productivity advantage of working from the office versus from home in the real world—or “nonexperimental settings.” 

The findings don’t mean an office return is the solution

Now, the study is solely based on workers in Chennai, with radically different management and cultural norms. In developing countries, homes tend to be smaller, more cramped and hot, and have noise pollution, Atkin offered, all of which would make homes poor environments for productivity. Self-selection is an important determinant of productivity, he added. 

But recent studies of IT and call center workers in the U.S. have come to similar conclusions, Atkin said. “This gives us some faith that the results are not unique to [India].” The data also follows recent research from remote work guru Nick Bloom, who found that remote work is associated with 10% to 20% lower productivity. 

Atkin, Schoar, and Shinde purposefully chose to observe jobs for which output was easily quantifiable, and that didn’t require teamwork. Those are typically the best candidates for remote work—collaborative jobs are likeliest to suffer if not in the same place.

“We think our results are most applicable to low-skilled job environments and those hiring new workers,” Atkin said, citing earlier research showing productivity is minimally impacted by moving longtime office workers to remote or hybrid plans.

Despite the well-known benefits of in-person work, many high-performing workers nonetheless opt to work from home—as do workers who would stand to gain the most, productivity-wise, from being in the office. But it’s not entirely black-and-white, the researchers write. “To know whether such choices are optimal from the worker’s perspective, we need to better understand their preferences and know more about the nature of the constraints under which they are making their decisions.”

All told, the current state of disagreement between bosses and workers on where to work results in a “significant loss” in workforce productivity, they write. One major potential solution: a legitimate social safety net. Policies that address constraints like childcare and eldercare stand to have “substantial effects on aggregate productivity.” 

That could certainly be a big takeaway for the U.S., where, unlike countries such as Sweden, Switzerland, or Spain, a social safety net is nonexistent. Just one in four private-sector workers has paid family leave, per the U.S. Bureau of Labor Statistics, and the U.S. is the only first-world country without paid maternity leave. Public spending for childcare in Norway and in Iceland is nearly $30,000; in the U.S., it’s $500. 

These statistics leave little doubt that greater investments in social support—rather than a mandated office return—would free up workers to do their jobs more effectively. Women, the ones disproportionately saddled with caregiving responsibilities, probably could have told you as much.