Capital Flows to Developing Countries: The Allocation Puzzle

The development-accounting literature holds that most cross-country income differences reflect TFP differences, which in a textbook neoclassical growth model has a sharp implication for capital: a country whose productivity is converging on the frontier should invest more and borrow more from abroad, both to build the capital stock and to smooth consumption forward. This paper tests that implication on 68 developing countries over 1980-2000 – a period chosen because financial openness rose sharply and because two decades is long enough to look past crises and world cycles – and finds the cross-country correlation runs the wrong way. Korea, with average TFP growth of 4.1 percent a year and an average investment rate of 34 percent, received almost no net capital inflows; Madagascar, whose TFP fell 1.5 percent a year with an investment rate barely reaching 3 percent, received 7 percent of GDP in inflows every year. Fitted across the sample the slope of net inflows on productivity growth is -0.72 (p = 0.1 percent), and it survives controlling for initial capital scarcity, initial debt and population growth. This is what the authors name the allocation puzzle, and they are careful to separate it from the Lucas puzzle about the small overall level of flows: since developing-country productivity on average did not catch up at all (average catch-up -0.10), the small aggregate level is not especially surprising, and is consistent with Lucas’s own explanation. The puzzle is about the cross-section – Asia caught up (0.19) yet borrowed only 11 percent of initial output, while Latin America (-0.24) and Africa (-0.17) fell behind yet received 37 and 39 percent. Two decompositions narrow the target. Inserting a capital wedge calibrated to each country’s investment rate and a saving wedge calibrated to its observed flows, the investment wedge alone cannot account for the pattern; matching the data requires a saving wedge strongly negatively correlated with catch-up, so that catching-up countries behave as if subsidising saving and falling-behind countries as if taxing it. Hence “the allocation puzzle is a saving puzzle.” Splitting flows into public and private following Aguiar and Amador (2011), private inflows are positively related to catch-up (slope 0.29, p = 4 percent) while public flows are strongly negatively related (slope -0.79, p < 1 percent), with reserve accumulation doing much of the work: open developing countries whose catch-up rose 10 percentage points accumulated reserves worth about 30 percent of initial output. Financial openness does not attenuate the puzzle but strengthens it, because public outflows respond to growth more strongly than private inflows do in open economies. The paper is explicit that it offers a diagnosis rather than a solution: the wedges “should not [be interpreted] as an explanation,” and the closing survey of candidate mechanisms is “meant… to provide a tentative road map,” with “no attempt… made to discriminate empirically between these explanations.”

Olivier Jeanne | Macro Paper Warehouse

Capital Flows to Developing Countries: The Allocation Puzzle