Governments' use of `output' policies such as export subsidies to give domestic exporters a strategic advantage is limited in practice by their political cost: trading partners may retaliate and highly visible transfers entail high political costs domestically. `Industrial' policies, such as taxes or subsidies on investment or adjustment, do not in general violate the rules of GATT and encounter less domestic opposition.

In Discussion Paper No. 450, Research Fellow Larry Karp and Jeffrey Perloff use a simple multi-period model in which a home firm and a foreign firm each export their entire output to a world market, and all the dynamic effects stem from adjustment costs. Each firm takes its rival's policies in the current period as given and chooses its current rates of output and investment. At the beginning of each period, the home government chooses an export subsidy and an `industrial policy' (modelled as a measure of the extent to which the government drives a wedge between the social adjustment cost and that faced by the firm). The foreign government is passive and does not retaliate.

Karp and Perloff assume further that the current decisions of governments and firms are determined only by the value of capital inherited from the previous period and the current decisions of other agents. Thus, if the government chooses a non-equilibrium industrial policy in the current period, firms will change in the next period, but the government's deviation has no further effect, and in particular firms' expectations of future government behaviour will not change.

Introducing a dynamic framework does not affect the conclusion of previous static analyses that output policies can be effective, but it does significantly affect the efficacy of industrial policies. Investment decisions depend not only on current but also on future government policies, which are likely to vary with the capital stock and therefore depend on past levels of investment. Previous analyses using static or two- period models have found that governments have strategic incentives to intervene in imperfectly competitive international markets through the use of subsidies or taxes on exports or production. This remains qualitatively correct in a dynamic model, but the benefits from industrial policy are smaller, and perhaps negligible. Even as the period of commitment becomes arbitrarily small, however, industrial policies can still be used for non-strategic ends, such as offsetting any distortions caused by the use of output policy.

Karp and Perloff's results suggest therefore that the GATT negotiators may have been correct albeit inadvertently to focus on limiting the use of output policy, since industrial policy is probably less of a threat to free trade motivated by the need to offset such output policies. Their results depend on the assumption that governments' ability to commit is limited. This will apply to industries undergoing rapid change for which future conditions are very uncertain, such as the high-technology industries for which popular support for industrial policy has been strongest. This analysis suggests that industries that replace capital at wider intervals, such as traditional manufacturing, are better candidates for the use of industrial policy. Karp and Perloff note in conclusion that industrial and output policies are even less likely to be strategically useful if their simplifying assumptions are dropped, for example to allow foreign governments to intervene or firms in other countries to enter the market.

Why Industrial Policies Fail: Limited Commitment
Larry S Karp and Jeffrey M Perloff

Discussion Paper No. 450, August 1990 (IT)