When Takeovers Meet GitHub Users: Productivity Takes a Hit

When a firm becomes a takeover target, most attention goes to shareholders, synergies, and the deal price. Much less attention goes to the people who actually work at the firm. In our working paper, we ask a simple question: what happens to the productivity of knowledge workers at the target firm once a takeover is announced?

The answer is stark. Using detailed data from GitHub, the world’s largest code-hosting platform, we find that contributions by GitHub users at target firms drop by 14% after a takeover announcement. This decline is not a blip: it persists for months, and code quality seems to deteriorate at the same time.

Measuring productivity in the age of GitHub

Knowledge-worker productivity is famously hard to measure. For programmers, software engineers, data scientists, and related roles, GitHub leaves a rich digital footprint of daily work. We use the weekly number of actions on GitHub — commits, pull requests, issues, etc. — to create a proxy of how much output individual GitHub users generate.

To estimate the effect of takeover announcements on output, we rely on a stacked event-study design. Different takeovers are announced at different points in time, which allows us to compare workers at firms that have just become targets to a control group of workers at firms that will be acquired only later. This setup lets us trace weekly productivity from 12 weeks before to 12 weeks after the announcement, while keeping attention on a window in which most deals have not yet been completed and formal restructuring has barely started.

Quantity drops, bugs rise

Output is not the only margin that moves. For a subset of public repositories, we construct a simple “bug rate” measure: the number of issues labelled as bugs divided by the number of commits, that is, saved snapshots of code changes. After takeover announcements, the bug rate increases in affected repositories. 

Looking separately at contributions to private and public repositories, we find that most of the decline comes from private repositories, which likely capture core, firm-internal development work. This pattern underscores that the productivity costs show up where firms actually build and maintain products, not only in public-facing side projects.

Using the approach of Holub and Thies to translate GitHub commits into dollars, the observed decline corresponds to $6,000 of lost output per worker over a six-month period after the announcement. For a typical target firm that employs many GitHub users, these losses add up quickly.

Stress, anxiety, and the fear of layoffs

Why do GitHub users reduce effort once a takeover is announced? Several explanations   are plausible. Some may be distracted; others may become anxious and find it harder to concentrate. Interestingly, we don’t find any evidence that workers try to signal effort — whether genuine or “pretend” effort — to protect their jobs.

Our evidence points to stress and anxiety tied to the fear of layoffs. First, we show that the decline in productivity is larger for within-industry takeovers, where the acquirer and target operate in the same sector. In such deals, the overlap in tasks and skills is greater, and cost-cutting via layoffs is more likely.

Second, we exploit U.S. state-level variation in the Inevitable Disclosure Doctrine (IDD), a legal regime that makes it easier for firms to argue that departing employees will inevitably reveal trade secrets at a competitor. Prior work shows that firms in IDD states are more likely to be acquired because acquirers want access to their skilled workforce. In these states, target employees should be less worried about losing their jobs after a takeover, and that is exactly what we find: takeover announcements in IDD states do not, on average, reduce worker productivity.

Third, we look at partial acquisitions, which typically do not involve a full change of control and are less likely to trigger large restructuring plans. Here, the results are weaker. This again fits the idea that it is the perceived layoff risk, not just any corporate transaction, that drives the productivity decline.

Taken together, these tests support stress and anxiety as the dominant mechanism, while also being compatible with some distraction effects.

Why boards and bidders should care

From a corporate governance perspective, our findings add a missing piece to the cost–benefit analysis of takeovers. Acquirers and boards are used to thinking about premiums, synergies, and integration costs. They are much less likely to factor in that announcing a deal may quietly reduce the output and code quality of the very knowledge workers they plan to retain.

If the main channel is distraction, takeovers temporarily nudge GitHub users away from their core tasks, and the firm suffers a short-run dip in output. If stress and anxiety dominate, announcements produce both an output loss and a hidden tax on workers’ mental health. In either case, these costs are real and material for modern firms whose value depends crucially on the productivity of their engineers and data scientists.

The policy implication is straightforward: evaluations of mergers and acquisitions should move beyond capital and product markets and systematically incorporate the effects on knowledge workers. For firms, the message is even more direct. When planning and announcing takeovers, boards and acquirers should invest in clear communication and credible employment guarantees, not only to be fair to employees but also to protect the value of the deal. In short, ignoring how takeover announcements affect the productivity and mental health of knowledge workers is not just unfair to employees; it is a costly mistake for acquirers.

__________

Beate Thies is an Assistant Professor in the Department of Economics at the University of Vienna. 

Andras Danis is an Associate Professor at CEU, Department of Economics.

Evgeny Gushchin is a PhD student in Economics at Central European University.

This blog is based on a paper presented at the Tenth Annual Mergers and Acquisitions Research Centre (MARC) Conference, held in London and hosted by Bayes Business School in collaboration with ECGI. Visit the event page to explore more conference-related blogs.

The ECGI does not, consistent with its constitutional purpose, have a view or opinion. If you wish to respond to this article, you can submit a blog article or 'letter to the editor' by clicking here.