When I teach behavioral science to executives at the London School of Economics, I am always bemused by how most students complain about overconfident CEOs, with Bezos, Musk, and Zuckerberg among the names most mentioned. I will probe them on why overconfidence is such a problem for them–and we will get to a place where they understand that overconfidence causes CEOs to underestimate the costs and risks associated with the projects that they want to undertake and at the same time overestimate their benefits.

My students are correct. However, they are still missing several key points. In periods of growth, the errors made by overconfident CEOs are often buffered by the upswing of the economic cycle. In contrast, during upswings, underconfident CEOs lose out because they stay hamstrung with risk aversion. Without overconfident CEOs, we would see far fewer innovative moon shots being taken that bring about technologies to advance our capabilities in health, protect the environment, work more effectively, and… well, go to the moon.

Meanwhile, underconfident CEOs procrastinate and watch their firms’ growth stagnate as they run the same old business models. Success is endogenous to beliefs. As soon as someone with the enthusiasm of Elon Musk gets behind a project, the odds of it succeeding are amplified. Why? People are more likely to follow a CEO who is excited and can articulate where they want to go in an inspiring narrative. In contrast, employees are unlikely to be motivated to work harder by a CEO who lacks irrational exuberance about their own idea.

Things get difficult for the overconfident CEO in periods of economic decline when they are publicly listed. Raising additional capital for new ideas is constrained as investors look for safe options to safeguard their finances, and debt becomes more expensive. For many big technology companies, it is actually possible to weather the storm by dipping into the large cash reserves on their balance sheet and retaining their employees through this negative cycle. Meanwhile, small-to-medium technology firms have no choice but to consider cutting jobs and monetizing assets to avoid financial distress. However, regardless of holding large cash reserves, big tech firms are still planning large numbers of layoffs. So, what gives? Why aren’t they protecting their employees?

Shareholders expect cost-cutting during periods of economic decline. CEOs cater to this psychological expectation in an attempt to preserve their companies’ share prices.

However, top technology companies also seem to be trying to take advantage of the market uncertainty to reorganize their employee mix to become more allocatively efficient over the medium term–and moving to hire more “creators.”

During the pandemic, salaries in top tech companies hit record levels in the pursuit of so-called top talent. The problem? The narrative that genius is available on the labor market for a high price causes a ratcheting of salaries. This is called the Lake Woebegon effect–which occurs when firms distort pay upward because they want to signal to the outside market that they are employing THE top talent. Trouble is, genius is hard to screen for in search and hiring. When some of that new talent arrived, their performance was mediocre and their biggest contribution was an excessive wage bill.

The Lake Woebegon effect is a bigger problem for big tech. When companies are small, it is easier to make sure that everyone employed by the firm is adding value. However, as companies scale, there is a risk of bloat. David Rolfe Graeber describes in his theory of “bullshit jobs” a number of types of potentially pointless jobs, including box tickers (those who create the appearance that something useful is being done) and taskmasters (those who create extra work for those who do not need it). Think of too many people employed in governance, middle management, or human resources. It is only excessive numbers of folk in these roles that create issues for the firm. At excessive numbers, the box tickers and taskmasters tip into “busywork” and create more unnecessary work for the “creators.” Creators add value by creating new products or services that are in demand, generating direct income for the core business.

Data suggests that companies are laying off up to 50% of their human resources teams, with recruiters,  marketing, and sales professionals also highly likely to be laid off. Overall, it is safe to conclude that creators are safer than those in “bullshit jobs.”

Since the Global Financial Crisis, slow growth has been evident in the U.S. and other major developed economies. One culprit for this could be the rise in “bullshit jobs.” To the extent that these roles can be identified within technology companies, it is unsurprising their ambitious CEOs wish to cull them to pave the way for future innovation and roles for creators.

However, the evidence does suggest that following culls, firm performance also declines. This echoes the importance of treating employees fairly during layoffs and paying the severance that is deserved rather than messing employees around like Twitter. It also underlines the importance of giving certainty to those employees who remain, to avoid leaving employees confused like Slack did, which will undoubtedly impact their productivity.

The problem with big tech is not that they have pulled the lever on mass layoffs–it’s how they are executing this decision.

Grace Lordan, P.h.D, is an associate professor of behavioral science and the founding director of The Inclusion Initiative at the London School of Economics.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

More must-read commentary published by Fortune:

• A soft landing is playing out—but optimism needs to be for the right reasons
• People are much less likely to trust the medical system if they are from an ethnic minority, have disabilities, or identify as LGBTQ+, according to a first-of-its-kind study by Sanofi
• The U.S. has thwarted Putin’s energy blackmail. Europe says ‘Tanks a lot!’
• Once the ‘intellectual blood banks’ of the rich and powerful, can speechwriters be replaced with ChatGPT?