The effect of migration policy on growth, structural change, and regional inequality in China

GAICIL Insights

The greatest growth story of the century may be, at heart, a story about permission to move.¹ The effect of migration policy on growth, structural change, and regional inequality in China, by Tongtong Hao, Ruiqi Sun, Trevor Tombe and Xiaodong Zhu, revisits an astonishing period. Between 2000 and 2015, China's aggregate income quadrupled, provincial income inequality fell by a third and the share of employment in agriculture fell by half. The authors identify worker migration as central. By 2015, almost 300 million workers lived and worked outside their area or sector of hukou registration.

The method combines rich individual-level data on worker location and occupation with a spatial general equilibrium model of China's economy. Internal migration costs fell by 45% over the fifteen years, with the cost of moving from agricultural rural areas to non-agricultural urban ones falling even more. These reductions contributed substantially to growth and account for the majority of the reallocation of workers out of agriculture and of the drop in regional inequality.

The comparison with rival explanations is the revealing part. Changes in trade costs, capital market distortions, the average cost of capital and productivity each contribute meaningfully to growth. Migration policy, however, is central to structural change and regional convergence. The authors also document that the growth slowdown between 2010 and 2015 is associated with smaller reductions in inter-provincial migration costs and a larger role for capital accumulation and credit expansion. When people stopped moving as freely, growth leaned on credit.

The policy conclusion is direct: further reductions in inter-provincial rural-urban migration costs could have large benefits for growth, structural change and regional income equality. Letting workers move to where they are most productive is one of the most powerful levers policymakers have, and one of the cheapest.

Now map this onto artificial intelligence. AI is expected to drive a new wave of structural change, shifting labor demand across sectors, occupations and regions. China's experience shows the gains from structural change depend heavily on workers' ability to reallocate. When barriers fall, workers move toward higher productivity activities, growth accelerates and regional gaps narrow. When barriers persist, growth relies more on capital accumulation and benefits spread less evenly. The AI dividend will be collected by economies that let labor follow it.

Mobility in the AI economy takes several forms. Geographic mobility matters as new activities cluster in particular places. Occupational mobility, the ability to move between tasks and sectors, may matter even more as AI changes the content of jobs. Policies that lower transition costs, through portable benefits, accessible training and flexible rules on residence and work, could help economies capture AI's gains as reduced migration costs helped China capture industrialization's. Every economy has its own hukou, in licensing rules, non-portable benefits and zoning; few call it by that name.

AI could also cut mobility costs directly. Digital job matching, remote work and AI tools that help workers acquire new skills lower the frictions of moving between locations and occupations. The paper's quantitative framework offers a way to measure such effects, estimating reallocation costs and their consequences for growth and inequality.

The slowdown finding is a warning for AI strategy. Investment in AI capital alone may not deliver broad-based growth if labor cannot reallocate to complement it. Capital deepening combined with labor mobility produced China's most dynamic period. Economies pouring money into data centers should spend equal attention on the institutions that let workers move.

For policy, the research supports reforms that reduce barriers to labor reallocation. Registration systems, licensing requirements and other rules tying workers to places or sectors impose costs that compound during rapid technological change. Removing them can promote growth and reduce regional inequality simultaneously.

Distilled to one claim: structural transformation runs through people. China's remarkable growth was driven in large part by policy changes that let workers move. As AI reshapes economies, the freedom and ability of workers to reallocate will again determine how large the gains are and how widely they are shared.

For AI industrial policy, the sting is this: the highest-return AI investment a government makes may involve no chips at all. Cutting the cost of a worker changing city or occupation could outperform another subsidy for compute.