Testing creative destruction in an opening economy

GAICIL Insights

Open the borders and the strongest gain most.¹ That is the uncomfortable, useful finding of Testing creative destruction in an opening economy, by Philippe Aghion, Johannes Fedderke, Peter Howitt and Nicola Viegi. Trade liberalization is often expected to raise productivity, and its channels are debated. The authors analyze the relationship using a framework of technological innovation and creative destruction applied to sector-level South African manufacturing data.

The framework separates direct and indirect effects. Liberalization can raise productivity directly by exposing firms to competition and foreign technology. Indirect effects work through a differential impact depending on distance from the international technological frontier. Theory predicts sectors closer to the frontier should benefit more. Openness is a ladder whose rungs are spaced by capability.

The results confirm it. Liberalization's productivity growth impact is positive, directly and indirectly. Trade protection has a direct negative impact on productivity growth. Sectors closer to the frontier benefit most significantly, and the greatest positive impact falls on sectors close to the frontier that faced low product market competition before liberalization. Findings are robust to controlling for potential endogeneity of trade protection. The biggest winners were capable sectors that had been sheltered.

Pricing power gets a twist. Liberalization enhances the importance of pricing power for productivity growth, and failing to control for reduced trade protection biases the estimated impact of pricing power downward. Studies ignoring the trade regime may misread how market power relates to growth.

The application to AI is direct. Artificial intelligence diffusion runs through the same channels. AI is a frontier technology, and trade and openness affect how fast it spreads. Openness benefits sectors near the frontier most. For economies adopting AI, already advanced sectors may gain most from foreign technology and competition, while sectors far from the frontier may need complementary investments to benefit. Openness alone will not lift the laggards.

Sheltered sectors are the opportunity. Liberalization benefited sectors that previously had low competition, as firms under new pressure invested in improvement. AI-intensive services and digital markets shielded from competition might gain similarly from openness. Protection that feels safe may be the very thing holding a capable sector back.

Protection's negative direct effect is a warning. Restrictions on trade in AI-related goods, services and technologies may reduce productivity growth. Security concerns may motivate some restrictions, and their productivity costs deserve careful weighing. Openness to foreign technology and competition supports innovation and productivity. Every AI trade barrier has a productivity invoice.

The South African setting matters. An economy undergoing liberalization provides evidence relevant to developing and emerging economies, many now considering how to engage with AI. Openness combined with investments moving sectors closer to the frontier can enhance productivity.

Distance to frontier is a practical lens. Sectors and countries at different distances need different policies. Those near the frontier gain from openness and competition that spur innovation. Those far behind may need skills and infrastructure to build absorptive capacity before openness yields large gains. Sequencing matters: capability first for some, competition first for others.

Sector-level evidence helps target. Knowing which sectors stand to gain most from openness lets governments sequence reforms and pair them with support where gains are likely smaller.

Policy follows. The research supports trade liberalization as a productivity tool, particularly for sectors near the frontier with limited prior competition. Combining openness with policies helping lagging sectors catch up broadens the gains.

Stepping back, trade liberalization raises productivity through direct and indirect channels, with the greatest gains in sectors near the frontier that faced little prior competition. As AI reshapes global trade in technology and services, openness will remain an important productivity driver, and distance to the frontier will shape how much each sector benefits.

Something for emerging economies weighing AI protectionism to sit with: shielding a capable domestic sector from foreign AI competition may be protecting it from its best chance to improve. The sectors to protect, if any, are the distant ones, and even they need investment above all.