Entrants often need to make considerable sunk investments with highly uncertain returns. The option to exit if returns are low reduces investment risks and stimulates innovation. In this paper, Michele Bisceglia (University of Toulouse) and Jorge Padilla, Joe Perkins, and Salvatore Piccolo of Compass Lexecon examine the interaction between exit policy and up-front investment by entrants, finding an inverted U-shaped relationship between innovation and exit value. Consumer welfare is shaped by a trade-off between encouraging firms to stay in the market through higher exit barriers and stimulating investment through a more permissive exit policy. As future returns become more uncertain, consumer welfare maximization requires lower exit barriers. These insights are applied to optimal merger policy and bankruptcy law.
Industrial economists have traditionally centered their research agenda on competition, innovation, and market structure around the concept of market entry. Less emphasis has been devoted to industrial exit, even though barriers to exit have been debated in policy circles on both sides of the Atlantic for many years (e.g., Ezekiel, 1992; Frank, 1988). Market entry and exit are two sides of the same coin. Both contribute to ensuring the benefits of competition and innovation. The threat of entry disciplines market power by incentivizing incumbents to keep prices low and offer what consumers demand. Market exit, instead, is a crucial instrument to sanction unprofitable business ideas, thereby spurring the industry life cycle through creative destruction (Schumpeter, 1942).
While the policy debate to date has primarily focused on barriers to entry and their impact on competition, effective competition, and thus a well-grounded industrial policy, is also shaped by exit. Barriers to exit weaken the selection mechanism that relocates demand and resources across and within industries when market conditions change. They may also affect competition, innovation, and ultimately economic growth.
In this paper, the authors consider how the exit policy can stimulate or deter demand-enhancing (e.g., quality-improving) investments when investments are sunk and demand is uncertain. In our baseline framework, there are two firms. One is an incumbent with committed investments, while the other is a challenger, who must decide whether or not to undertake an investment before learning of its demand. Then, upon privately observing the demand realization, the challenger decides whether to leave the market and enjoy an exit value (which, as we explain below, is affected by the regulatory regime in place) or remain active and compete with the incumbent. Finally, firms in the marketplace compete for customers by setting quantities (Bayes-Cournot competition) based on committed investment levels.
This framework is inspired by the growing policy and strategic influence of many infrastructure sectors, such as the wireless communication sector, where entrants must sink significant investments with long and uncertain payback periods. Risk aversion, capital market failures or regulatory uncertainty can mean that private investment decisions result in infrastructure provision below the socially optimal level. In response to such concerns, investment in some
infrastructure sectors is supported by explicit subsidies. However, such interventions, which can influence exit costs, may be costly to consumers and governments, and can also limit the benefits of rivalry between firms to win customers.
Within the framework outlined above, the authors characterize the challenger’s exit strategy and study how the exit value affects the market outcome. They show that exit has a selection effect with strategic implications. The challenger chooses to stay in the market and thus competes with the incumbent only when it observes favorable demand states. Hence, when exit becomes more likely, as implied by a more attractive exit value, the incumbent is less aggressive, since it expects the challenger’s output to be higher, and quantities are strategic substitutes.
Building on these insights, the authors then examine the challenger’s incentive to invest, for a given exit value. They find that the return of the investment is positive and determined by two intuitive forces. First, when the challenger invests, it exits less often compared to the case of no investment. Second, conditional on remaining in the market, the challenger has a larger market share when it invests since the investment spurs demand. More interestingly, the relationship between the challenger’s exit value and its incentive to invest is non-monotone. In particular, they find
that this relationship is inverted U-shaped — i.e., an increase in the exit value tends to stimulate investment when the exit value is not too large, reducing the incentive to invest when it is large enough. Hence, they find that the investment return is maximized for an intermediate exit value, which in turn is increasing with demand volatility, market size, and the degree of substitutability between the challenger’s and the incumbent’s products. When the exit value is large, further increases in that value reduce the investment return since they make exit more likely and the return on investment only materializes when the challenger remains in the market. In contrast, when the exit value is small, an increase in that value increases the set of circumstances (i.e., demand realizations) for which the investment pays off and results in a larger market share due to the selection effect discussed above.
The authors also consider the potential implications for consumer welfare. They show first that the investment benefits consumers and identify an under-investment problem. Next, they characterize the optimal exit policy that a regulator whose objective is to maximize consumer surplus would choose. This exercise hinges on the idea that regulators can influence the challenger’s exit option by designing policies that, for example, impact its labor-related exit costs (e.g., costs related to employees’ contractual rights such as staff redundancy costs and insurance benefits), bankruptcy rules (e.g., filing and litigation fees), environmental regulations (e.g., remediation costs), etc. The authors’ main finding is that for intermediate values of the investment costs — i.e., when the exit value affects the challenger’s incentives to invest — the regulator always sets a positive exit value determined by the challenger’s (binding) incentive compatibility constraint. That is, in this range of parameters, the regulator’s first-order concern is always to solve the underinvestment problem. The intuition is as follows: for a given exit value, by increasing consumers’ willingness to pay, the investment creates value, and when the firm is indifferent between making the investment or not, this value is fully appropriated by consumers, who must therefore be better off with the investment than without it. The range of parameters in which this happens expands with demand uncertainty.
The authors then examine instances in which the exit value depends on whether the challenger has invested or not. They develop two applications where this can happen (e.g., as in Schary, 1991). First, they allow for the possibility of a merger to monopoly, in which the merger profitability depends on whether the investment takes place or not — i.e., the challenger’s exit value coincides with the takeover price offered by the incumbent upon observing its investment decision. Second, they consider the case in which the challenger is able to recoup a fraction of its investment cost
when it exits, which is determined by the prevailing bankruptcy law.
When the exit value is determined by the takeover price, the regulator faces the following trade-off. On the one hand, a restrictive merger policy, which prohibits the challenger’s acquisition by the incumbent, strengthens competition in the product market, which benefits consumers since it induces lower prices and greater product variety. On the other hand, a lenient merger policy, which allows any merger to occur, fosters investment incentives, which in turn increases the products’ quality and/or quantity available to consumers. The authors show that, even though, absent horizontal differentiation between products, any acquisition in equilibrium is a killer merger (i.e., the incumbent always shuts down the acquired firm’s product), a lenient merger policy maximizes consumer surplus in industries which require relatively more expensive investments. Interestingly, the regulator can improve over a fully lenient merger policy by adopting a transaction price-contingent policy.
Similarly, consumers benefit from efficient bankruptcy rules that minimize rent dissipation and enable the challenger to recoup a relatively large share of the ex-ante investment when it decides to quit, thereby enhancing its incentive to invest. In this light, the authors’ approach can help to understand the wide disparities in the pace of infrastructure investment between countries with similar income levels, due to differences in merger policy and bankruptcy law.
Their results remain qualitatively unchanged when considering multiple incumbents: in this case, the optimal policy still requires a positive exit value to secure the investment by the challenger when the market is sufficiently concentrated or demand is sufficiently volatile. They also consider the possibility of leapfrogging — i.e., a technology such that, upon investment, the challenger’s product features higher quality than the incumbent’s one. In this case, consumers are more likely to benefit from a policy mandating a positive exit value the larger the impact of the investment on the challenger’s quality. Their findings remain true qualitatively also when considering a technology in which the investment itself has an uncertain, rather than deterministic, impact on demand, and allowing for a continuum of investment levels.
The paper is organized as follows. After the authors clarify their contribution to several strands of existing literature, in Section 2 they describe the baseline two-firm model and analyze the effects of exit policy on investment and consumer welfare. The exit value is endogenised in Section 3. In Section 4, they present some extensions and robustness checks of the main results. Section 5 concludes. Proofs of the main results are in the Appendix. Additional proofs are contained in the online Appendix.
 Jorge Padilla, Joe Perkins, and Salvatore Piccolo are economists at Compass Lexecon. The authors’ views do not necessarily represent the views of Compass Lexecon or its clients. The opinions in this paper are the authors’ sole responsibility.
 Indeed, “most guidelines link exit barriers to entry barriers, as exit costs can deter entry if firms can anticipate
them before entering”(OECD, 2019).
 For instance, Gruber (2019) concludes that meeting the European Commission’s aims in the Digital Agenda
for Europe and the European Gigabit Society would require about €384 billion of investment in wireless and
broadband technology between 2019 and 2025. However, such investments are made when there is uncertainty
both about the scale of consumer demand for new technologies and about the proportion of value that will be
captured by infrastructure providers. He then identifies an investment gap of around €254 billion.
 For instance, in the energy sector, network firms typically own regulated asset bases that receive guaranteed
returns, while low-carbon generators often receive public support through top-up payments or guaranteed prices:
see, e.g., Roques and Finon (2017).
 If the investment cost is sufficiently low, so that there is no under-investment, the regulator optimally sets
the exit value at zero. The reason is simple: conditional on the investment taking place, consumers always benefit
from rivalry, which is achieved by minimizing the challenger’s incentives to leave the industry. By contrast, when
the investment cost is so high that the challenger never invests irrespective of its exit value, the regulator sets a
positive exit value when products are sufficiently homogeneous and demand volatility is low. This is because the
loss from monopoly becomes a second-order concern when the challenger’s products are inferior, as it happens
when it does not invest and horizontal differentiation is negligible.
 Of course, a lenient merger policy is effective in stimulating investments if and only if antitrust authorities can
commit in advance to their policy, as otherwise they would have incentives to implement a strict merger policy
after the challenger sunk its investment.
 For instance, GSMA (2020) expects 48% of mobile connections in North America to be 5G in 2025, 47%
in Greater China, and 34% in Europe. Similarly, it expects almost $300 billion of mobile network operator
investment in their networks in the US between 2020 and 2025, compared to around $170 billion in Europe.