Development And The Public Debt-laffer Curve

Once more, the issue of government debt and its impact on economic growth has dominated the forefront of politicians , policymakers and academics agendas following the record-breaking expansion in global government debt, amidst the COVID-19 pandemic. In 2020, global public debt stocks grew by over $12 trillion (~20% increase in debt-to-GDP) as governments used aggressive fiscal policy measures to contain the virus and mitigate the economic costs to businesses, households and financial markets. Notwithstanding, pre-COVID debt trajectories displayed a similar unpleasant picture with average global debt-to-GDP increasing from 43% in 2007 to around 60% in 2019 (Figure 1b). This has come as the 2007/8 financial crisis has dragged on economic growth, increased private sector credit constraints and amplified investor uncertainty, slowing the global recovery. Considering these public debt and growth trends and the inconclusive academic and policy debate on the implications of debt, this paper sets out to uncover the true cost of high public debt for economic growth.

There are several positive and negative channels through which the accumulation of government debt can affect long-run economic growth. Early studies focused on the private sector crowding-out hypothesis whereby government debt accumulation lowers domestic investment, depleting capital stocks and slowing economic growth. Other channels suggest that high levels of government debt can harm economic growth through increased long-term interest rates, volatile inflation, greater investor uncertainty of future tax rates and less inward investment. Conversely, debt proponents argue that these consequences are not experienced at low levels of government debt. Instead, public debt accumulation via productive capital investment can enhance infrastructure, improve human capital and develop financial markets, leading to faster growth in total factor productivity (TFP) and output (Delong and Summers, 2012). Another strain of research proposes the debt-Laffer curve where low levels of government debt accumulation accelerate growth due to benefits of higher TFP and capital accumulation. Although as debt levels grow too high, growth declines as these benefits are dwarfed by the consequences of crowding-out, capital flight, investor uncertainty and increased inflation (Checherita-Westphal and Rother, 2012 Claessens, 1990).

This paper employs a dataset of 174 developed and developing countries over the 40-year period 1980-2019 to provide empirical evidence regarding the relationship between government debt and long-run average per capita GDP growth. Specifically, this investigation will attempt to determine whether the impact of public debt on long-run economic growth is (i) non-linear and (ii) influenced by a country s development stage. In doing so, regression estimates attempt to uncover the maximum level of public debt that can be sustained without debt being a burden on economic growth and investigate whether this threshold varies across countries at different stages of development.

Econometrically, this analysis utilises three regression specifications. First, the pooled ordinary-least-squares (POLS) estimator is used to provide motivation and preliminary cross-country estimates. Second, the two-way error fixed effects (TW-FE) estimator is used to remove the possible endogeneity caused by cross-country and cross-time heterogeneity. In light of the well-known reverse causality issue in long-run growth regressions (Kumar and Woo, 2010) as well as the possible feedback effects of current and past economic shocks the TW-FE estimator may still be subject to bias[1]. Consequently, the third and main specification employed is the two-stage least-squares (2SLS) TW-FE estimator, where first order lagged public debt values are used as instruments. The latter two specifications use five-year average data to ensure that the effects of business cycles do not confound estimates. The robustness of these results is tested by producing estimates using ten-year average periods as well as removing outlier countries[2]. In order to control for the fact that a countries development stage may influence the relationship between public debt and economic growth an interaction term between public debt and the development indicator is included. Growth controls are inspired by the findings of (Sala-i-Martin et al., 2004) and the countries development level is defined by the United Nations (UN) World Economic Situation and Prospects report (WESP) (2021).

The main results indicate that: (i) there is an inverted u-shaped relationship between government debt-to-GDP and long-run average per capita GDP growth and (ii) the threshold for which government debt begins to harm long-run economic growth is 114% for developed economies and 96% for developing economies. Thus, if the average government debt-to-GDP ratio increased by 5 percentage points (pp) to 62% and 60% in the developed and developing sample, average per capita GDP growth is estimated to increase by 0.68pp and 0.57pp, respectively. These inferences support the debt-Laffer curve (Claessens, 1990) and the debt-overhang (Krugman, 1988) hypotheses. The results also imply that for any given debt-to-GDP ratio, further accumulation has a more positive growth-effect in developed nations than in developing counterparts. This is consistent with the notion that developed nations are more tolerant to holding large government debt (Cordella et al., 2010). Beyond the main results, supplementary analysis scrutinises King and Levine (1993). In doing so, this paper finds that by not including government debt in their regressions, estimates are likely to suffer from omitted variables bias. Moreover, this reveals how important financial development is in assisting government debt accumulation and therefore economic growth.

Furthermore, this investigation contributes to the existing body of literature in four ways. First, in order to highlight the importance of development in the public debt-growth nexus, this study establishes two separate relationships between government debt and long-run growth based on development stage. Second, this study represents one of the largest datasets in both the cross-sectional and time-series dimension. Third, this paper emphasises the need for instrumental variables (IV) to mitigate bias caused by reverse causality. Lastly, to avoid the common issue of small-sample bias in generalised method-of-moments (GMM) estimators (Arellano and Bond, 1991), this analysis employs the less frequently used 2SLS TW-FE estimator.

[1] A detailed discussion of the econometric concerns can be found in Section 4.4.

[2] Outliers are listed in the footnotes of Figure A1 in the appendix.