Traditional arguments in favour of multilateral agreements are based on the theoretical premise that the free market equilibrium is efficient (Irwin 1996). This proposition has a long tradition and was originally based on the assumption that markets are perfectly competitive so that firms do not have market power and their relative prices perfectly map their relative marginal costs of production, which is a key requirement for market allocative efficiency.

However, as the relevance of monopolistic power was recognised, trade economists moved away from perfect competition, mostly embracing (for analytical convenience) monopolistic competition, whereby firms derive their market power from product differentiation even in the absence of any strategic interaction (Costinot and Rodriguez Clare 2014). With some surprise, they still found that free trade is the best multilateral trade policy as the free market equilibrium continues to be efficient, and there is, therefore, no room for improving the allocation of resources across firms and countries through policy intervention (Melitz and Redding 2015). This surprising result holds under the (most widely used) assumption that firms’ markups are constant across firms so that relative prices still map relative marginal costs. Accordingly, the original insight that free trade allows governments to attain (constrained) efficiency irrespective of technological or geographical differences across countries is not fundamentally altered by firms having market power.

The question is begging: Does the view that multilateral free trade is efficient holds in a world in which market power and markups are changing asymmetrically between large and small firms? Our new paper (Nocco et al. 2019) argues that the answer is “yes and no”. Yes, because in the very same world peddling global prosperity requires even stronger multilateral cooperation. No, because in such world free trade, though still better than autarchy, is not necessarily efficient.

Free trade is not efficient when market power varies across firms

With monopolistic competition, free trade is inefficient if one is willing to accept that larger, more productive firms can charge higher markups than smaller, less productive ones. This is a side effect of Marshall’s Second Law of Demand (MSLD), according to which demand becomes more inelastic with consumption, a property that is both theoretically and empirically appealing (Mrazova and Neary 2013, Mayer et al. 2016). When MSLD holds, more productive firms do not fully transmit their cost advantage to prices so that the ratio of their prices to the prices of less productive firms is larger than the ratio of their respective marginal costs (Melitz and Ottaviano 2008). Allocative efficiency is therefore violated – in their quest for monopoly rents, more productive firms are happy to leave a larger market share than optimal to less productive firms. As a result, a larger amount of productive resources than optimal is ‘trapped’ in low productivity uses.

This misallocation of resources is stronger when the intensity of competition is weaker. This is a direct implication of state-of-the-art measures of competition intensity. Take, for instance, the ‘Boone indicator’, defined as the elasticity of profits to marginal cost in a given market (Boone 2008). Under MSLD, the Boone indicator is negative as long as lower cost firms are more profitable. By raising the profitability of lower cost firms relative to higher cost ones, tougher competition decreases the Boone indicator (i.e. increases its absolute value), leading to a more efficient allocation of resources between higher and lower cost firms. This implies that the inefficiency of free trade is more pronounced for countries with less competitive home markets. These are countries with smaller domestic market size, worse state of technology in terms of higher innovation and production costs, and worse geographical situation in terms of closer proximity to other trade partners. We call these countries ‘disadvantaged’.

If differences in market size, technology and geography are large (small), welfare inequality between disadvantaged and advantaged countries is inefficiently large (small) under free trade. In particular, we show that the free market is less selective than optimal so that the number of producers in each country and the number of locally sold goods are too large. This applies to exports as well, in particular in disadvantaged countries, as they are the least efficient. Moreover, the free market offers a product mix with fewer low-cost varieties than optimal. This inefficiency of free trade is magnified by excessive levels of consumption of goods produced in far locations compared to those of goods produced nearby or locally. As disadvantaged countries are more distant from their trade partners, this inefficiency is more pronounced in their case.

Stronger multilateral cooperation is needed when market power varies across firms

These findings imply that multilateral cooperation can improve on the free market outcome. However, efficient multilateral policies are quite complex in that firm- and location-specific tools are needed to increase (decrease) the sales of low (high) cost firms to all countries, but especially to disadvantaged ones, and reduce firm entry in all countries, but especially in disadvantaged ones. In particular, two types of policy tools are required. Multilaterally agreed per-unit subsidies (taxes) or tariffs are needed to promote (curtail) the sales of low (high) cost firms, especially those located in the periphery of the trade network. Simultaneously, multilaterally agreed lump-sum taxes on profits are needed to discourage excessive market entry when subsidies are introduced, especially in disadvantaged countries. Interestingly, multilateral policy intervention can improve on free trade even when those refined instruments are not available as is usually the case, for example, when a country can implement only a per-unit trade subsidy common to all firms selling in its market and a lump-sum profit tax common to all its producers.

In any case, binding multilateral agreements are badly needed as countries have the usual incentive to deviate unilaterally from the efficient multilateral policies to improve their terms of trade (Bagwell et al. 2016). On the one hand, each country has an incentive to introduce a lump-sum profit tax in order to reduce the number of its producers and allow them to export at monopoly price. This tax is higher for advantaged countries. On the other hand, each country has also an incentive to use per-unit subsidies in order to induce marginal-cost pricing for imports and domestic sales. The average subsidy is higher for disadvantaged countries.

Concluding remarks

That free trade is the efficient multilateral trade policy is a hallmark of international economics. This conclusion might not hold in a world in which market power is asymmetric between large and small firms. In this world there are, however, multilateral policies that, accounting for the geography, market size and technology of the trade partners, can increase global welfare with respect to the free trade outcome. These policies promote the sales of low cost firms and trim the sales of high cost firms, in particular in disadvantaged countries. To follow up on Paul Krugman’s famous quip (Krugman 1987), while free trade is an idea that has irretrievably lost its innocence as its status has shifted from optimum to reasonable rule of thumb, the same cannot be said for multilateralism.

Authors’ note: The views expressed here are those of the authors and do not represent in any manner the European Commission.