Public debt, output growth, and the external sector in developing countries

We discuss two critical elements of the equation describing the debt sustainability dynamics: long-run output growth, which affects the automatic debt dynamics, and possible solvency/liquidity feedback effects on output growth emanating from external public debt. In this context, we first investigate the balance-of-payments-constrained growth (BPCG) model as a possible benchmark for long-run output growth. We contribute to the BPCG literature in several ways. First, we examine whether the BPCG model is relevant for low-income countries, a group of countries heavily under-represented in this strand of post-Keynesian literature. Second, we contribute to the existing debate by using modelling approaches that control for the presence of cross-sectional dependence in the macro panel data, such as the co-integration test proposed by Westerlund (2007), and the Common Correlated Effects (CCE) mean group estimator proposed by Pesaran (2006). Specifically, we test the BPCG model on a sample of 19 low-income/lower middle-income countries and compare the results with a control group of 14 upper-middle/high income countries. The sample covers the period from 1979 to 2014. All countries in our sample have experienced long-term current account deficits or near-deficits, and should therefore provide for a strong test in favour of the BPCG model. We estimate import and export functions for both country groups on an aggregated basis and on an individual country level, and test empirically the near-identity critique most recently put forward by Blecker (2016a) and Razmi (2015).We find evidence of co-integration between real imports and real domestic income, but could not establish a statistically significant co-integrating vector for real imports, domestic income and two proxies for relative prices. The estimated aggregate income elasticity of import demand is robust and statistically significant for both country groups, and exhibits unit elasticity in both country samples. The size of the income elasticity of import demand for high income/upper middle-income countries is smaller than in other studies (e.g. Cimoli et al., 2010; Lanzafame, 2014), which find these elasticities to be on the order of 2. This difference in size may be related to the sample of countries, which in this chapter could all plausibly be seen as balance of payments constrained, and the accounting for the presence of cross-sectional dependence, which other work does not control for.We also found the weak form of the BPCG model to have good predictive power on the aggregate level for low-income/lower-middle-income countries. However, we obtain the same predictive power when empirically testing the near-tautology critique by using the ratio of the long-run growth rates of real imports and real domestic income in place of the estimated income elasticity for imports. This result seems to give credence to the near-identity critique. Finally, we could not establish a statistically significant co-integrating relationship between real exports and real foreign income. This result challenges the strong form of the BPCG model.We further present evidence of debt overhang and liquidity constraints in a sample of 61 developing countries during the period 1979 to 2013. The approach taken here differs from other work in this strand of literature in two key respects. First, the impact of the external public debt stock and external debt service on real growth is conditioned on the country’s institutional and policy performance, which we divide into weak and medium/strong performers as determined by the World Bank’s Country Policy and Institution Assessment (CPIA) index. Second, poolability is tested to identify country sub-samples with relevant coefficient homogeneity. We provide evidence of a statistically significant concave relationship between the stock of external public debt to GDP and real output growth for countries assessed as medium and strong institutional performers. This debt threshold ranges between 30 and 40 percent of GDP for 26 poolable countries. Weak institutional performers do not face a debt overhang, but instead appear to be liquidity constrained. The latter takes the form of a linear negative, contemporaneous relation between the debt service-to-revenue ratio and real output growth. The estimated liquidity elasticity ranges between -0.7 and -0.8 for 23 poolable countries. We cannot find evidence of liquidity constraints in medium and strong institutional country performers.