Agency theories predict that older firms make value-destroying acquisitions to benefit managers. Neoclassical theories predict instead that such firms make wealthincreasing acquisitions to exploit underutilized assets. Using IPO cohorts, we establish that, while younger firms make more related and diversifying acquisitions than mature firms, the acquisition rate follows a U-shape over firms' life cycle. Consistent with neoclassical theories, we show that acquiring firms have better performance and growth opportunities and create wealth through acquisitions of nonpublic firms throughout their life. Consistent with agency theories, older firms experience negative stock price reactions for acquisitions of public firms. EXISTING THEORIES ON THE ROLE OF ACQUISITIONS and diversification over the life cycle of firms offer two very different views. Agency theories predict that firms make wealth-destroying acquisitions and diversify when they are mature because their cash flow outstrips their internal growth opportunities and management becomes more entrenched, so that it pursues growth at the expense of shareholders (e.g., Mueller (1972) and Jensen (1986Jensen ( , 1993). Neoclassical theories predict, instead, that firms pursue acquisitions in an effort to make the best use of their valuable scarce assets (see Maksimovic and Phillips (2013) for a review). As a result, better-performing firms and firms with better growth opportunities create value through acquisitions, including diversifying acquisitions. To the extent that firms that go public are better-performing firms, we would expect them to acquire those assets that they can make better use of after going public and having easier access to external finance. As newly public firms exploit their growth opportunities, their acquisition rate should fall. Eventually, however, firms with valuable scarce assets may acquire new assets to keep making optimal use of these scarce assets, so that these theories * Aslı M. Arıkan is with Kent State University. René M. Stulz is with Ohio State University, NBER, and ECGI. We thank participants at seminars at