Major current account imbalances have largely corrected in most countries in the aftermath of the 2008 Global Crisis, but stock imbalances persist (Lane and Milesi-Ferretti 2014).
Accumulating deficits have led the US net international investment position (NIIP) to reach an unprecedented -40% of GDP. This contrasts with NIIPs of +64% for surplus economies such as Japan. The euro area has a broadly balanced NIIP – but this masks disparities within. In the post-crisis period, the NIIPs of periphery countries such as Spain had reached close to or beyond 100% of GDP, but have eased since then. In contrast, the surplus of Germany, for example, persists, further raising its NIIP beyond 60% of GDP. A first question is whether current NIIP levels are prudent from the viewpoint of external stability risks. And as NIIPs may have evolved out of sync with their main drivers over the past, a related question is whether current NIIP levels are justified in terms of economic fundamentals, and therefore likely to persist further.
But what is an appropriate NIIP level for a given country? The NIIP reflects the aggregate net financial assets of all agents in an economy. Theory thus suggests that the NIIP should respect the intertemporal budget constraint of the economy, but says little about how much the NIIP should deviate from zero for a particular country, in a particular moment in time. This contrasts to the case of current accounts, where a rich literature aims at estimating benchmarks for the assessment of external positions.1 The analogous literature on NIIP assessment is indeed very scarce. Notable exceptions are the few papers that aim to derive NIIP thresholds, which imply a high probability of currency crises or sudden stops. Catão and Milesi-Ferretti (2014) estimate the probability of external crisis from a panel probit model, with NIIP being among the significant explanatory variables for crises. Their model estimates suggest a threshold for the NIIP in the proximity of -50%.2
To our knowledge, the only NIIP benchmarks estimated in the existing literature are derived from a large sample of countries and provide a common point of reference for the NIIP of very diverse countries. This contrasts with the undisputed fact that the riskiness of a given NIIP stock depends on a number of country characteristics that are not accounted for by one-size-fits-all benchmarks. Country-specific benchmarks therefore seem needed for a proper assessment. The availability of country-specific NIIP benchmarks would also help the analysis of current account sustainability, as they would make it possible to obtain NIIP targets based on well-defined criteria for the computation of NIIP-stabilising current accounts.
In Turrini and Zeugner (2019), we propose a framework to fill this gap by estimating two sets of country-specific NIIP benchmarks:
- Prudential NIIP thresholds represent a country-specific negative NIIP level beyond which a country runs a significant external stability risk.
- NIIP norms represent the NIIP level that can be explained by fundamentals such as demographics, income per capita, resources, or financial structure.
Prudential NIIP thresholds
Prudential NIIP thresholds aim to estimate the NIIP level beyond which there is high risk of an external crisis, following the approach and definition by Catão and Milesi-Ferretti (2014). We update their dataset focusing on the starts of ‘external crises’, defined as significant use of IMF resources or as a sovereign default. We use a signalling approach à la Kaminsky et al. (1998) to test whether various macroeconomic indicators can signal such crises when they surpass a certain level. The threshold is selected to maximise signal power, i.e. to minimise the risk of missing crises or triggering false alerts.
Compared to other indicators of external financial assets, the NIIP performs well in terms of the signalling power of its common threshold equal to -25% of GDP (the only exception being an indicator that only includes the non-equity component of the NIIP°). However, the signal power of the NIIP is further increased if the variable is interacted with other variables that are normally associated with external crises. We experiment interactions with alternative structural slow-moving, country-specific characteristics that proxy borrowing constraints and debt tolerance. Income per capita turns out to deliver a comparatively strong performance as a choice for such an interaction term, as it embeds information not only on economic development but also on institutions and other structural characteristics that are normally associated with higher default probability on a given amount of foreign debt.3 Moreover, the interaction of the NIIP with per-capita income allows us to derive straightforward NIIP thresholds that are country-specific.
NIIP norms are estimated as the stock equivalent of current account norms. As NIIP series are hardly stationary in the panel, the estimation of an empirical model for the NIIP would require panel co-integration analysis (e.g. Lane and Milesi-Ferretti 2002). To overcome the short-sample limitations of panel co-integration tests and to obtain NIIP norms that can safely be interpreted as values determined only on the basis of fundamental drivers that can be considered as broadly exogenous, we follow a different route – namely, exploiting the fact that annual changes of NIIPs roughly correspond to the current account balances.
The specification of the current account norms follows Phillips et al. (2013). Regressors, whenever appropriate, are expressed as differences with respect to the world average. Current accounts are thus determined by what happens domestically in comparison to the rest of the world. This transformation induces stationarity of explanatory variables. It also provides a straightforward interpretation for the policy variables so transformed, which can be seen as deviations from a common norm corresponding to world averages. The regressors are divided into controls and fundamentals, with the latter being the only ones used to construct current account norms. Norms from this framework thus represent the current account that would typically be obtained for a country with certain characteristics, if its policy were set to the world average.
We demonstrate that the estimation of current account norms, with opportune adaptations in the definition of fundamentals, provides a good approximation for the benchmark NIIP in differences. The NIIP norm therefore amounts to the cumulation of these current account norms.
The main patterns and trends of NIIP benchmarks
We estimate NIIP benchmarks for 65 advanced and emerging economies. The median for country-specific NIIP norms over the 1995-2016 period is about -17% of GDP, while the median prudential threshold is about -44%.
The estimated benchmarks display a number of patterns.
First, the two benchmarks are found to be negatively correlated across countries (Figure 1), as the same factors that underpin scope for external borrowing – notably relatively low per-capita income – are also correlated with lower tolerance for high foreign debt (e.g. Lane and Milesi-Ferretti 2002).
Figure 1 NIIP norms versus NIIP prudential thresholds.
Note: 65 advanced and emerging economies, country-specific averages over available time periods
Second, the median NIIP and the median NIIP norm appear to comove over time (Figure 2), although norms do not display a comparable increase in their cross-country dispersion over recent decades as the one observed for the actual NIIP (Figure 3), suggesting that the recent growing external stock imbalances across the world are not fully justified by fundamentals.
Figure 2 NIIP, NIIP norm, NIIP prudential thresholds: Evolution of median value across 65 advanced and emerging economies
Figure 3 NIIP, NIIP norm, NIIP prudential thresholds: Evolution of standard deviation across 65 advanced and emerging economies
Third, NIIP gaps are fairly persistent over time, and are generally positively correlated across countries (Figure 4). The same country that has a relatively large gap with respect to norm tends to also have a large gap with respect to the prudential threshold. Only in few cases, observed mainly in emerging economies, are NIIPs that are above norm at the same time below the prudential threshold.
Figure 4 NIIP gaps with respect to NIIP norm versus NIIP gaps with respect to prudential thresholds
Note: 65 advanced and emerging economies, country-specific averages over available time periods
Fourth, NIIP gaps display a negative, although weak, relationship with subsequent medium-term changes in the NIIP/GDP ratio. Most importantly, NIIP gaps are a better predictor of subsequent adjustment in the NIIP than the NIIP level itself, which confirms the importance of country-specific as opposed to one-size-fits-all benchmarks.
NIIP benchmarks are among the tools regularly employed to assess external positions in EU surveillance, notably in the context of the Macroeconomic Imbalances Procedure. The evidence points to a quite differentiated picture across the EU. While a number of countries display positive NIIPs above norm, in some countries the NIIP is largely negative, below what is justified by fundamentals and in some cases below prudential thresholds (Figure 5). The evolution of NIIPs across the EU and the euro area is regularly monitored, and the NIIP benchmarks provide an anchor for such an assessment.
Figure 5 NIIPs and benchmarks across the EU, 2017
Source: European Commission (2019)
Catão, L.A.V., and G.M. Milesi-Ferretti (2014), "External liabilities and crises", Journal of International Economics, 94, 18-32.
Ca’ Zorzi, M., A. Chudik, and A. Dieppe (2012), "Thousands of models, one story: Current account imbalances in the global economy", Journal of International Money and Finance, 31, 1319-1338.
Chinn, M.D., and E.S. Prasad (2003), "Medium-term determinants of current accounts in industrial and developing countries: an empirical exploration", Journal of International Economics, 59, 47-76.
Debelle, G., and H. Faruqee (1996), "What determines the current account? A cross-sectional and panel approach", IMF Working Paper No. 96/58.
European Commission (2019), Alert Mechanism Report, Report from the Commission to the European Parliament, the Council, the European Central Bank, and the Economic and Social Committee, Brussels, COM(2018) 758 final.
Kaminsky, G., S. Lizondo and C. Reinhart (1998), "Leading Indicators of Currency Crisis", IMF Staff Papers 45, 1-48.
Lane P.R., and G.M. Milesi-Ferretti (2002). "Long-term capital movements,", in NBER Macroeconomics Annual 2001, Volume 16, 73-136.
Lane, P.R., and G.M. Milesi-Ferretti (2014), "Global imbalances and external adjustment after the crisis, IMF" Working Papers No. 14/151.
Lee, J., J. D. Ostry, G.M. Milesi-Ferretti, L.A. Ricci and A. Prati (2008), "Exchange rate assessments: CGER methodologies", International Monetary Fund, Washington D.C.
Phillips, S., L. Catão, L. Ricci, R. Bems, M. Das, J. Di Giovanni, D.F. Unsal, M. Castillo, J. Lee, J. Rodriguez and M. Vargas (2013), "The external balance assessment (EBA) methodology", IMF Working Papers No. 13/272.
Turrini A., and S. Zeugner (2019), “Benchmarks for Net International Investment positions”, European Commission European Economy Discussion Paper no. 97 (forthcoming Journal of International Money and Finance)
Zorell, N. (2017), "Large net foreign liabilities of euro-area countries", ECB Occasional Paper No. 198.
 See, for example, Debelle and Faruqee (1996), Chinn and Prasad (2003), Lee et al. (2008), Ca' Zorzi et al. (2009) and Phillips et al. (2013) on the estimation of current account norms in line with fundamentals. The main alternative approach, regularly used by policy institutions, is to assess current account positions in terms of their implications for the evolution of the NIIP. To this purpose, current account benchmarks are derived from the requirement for the NIIP stock to stabilise or to reach a given target by a given date (e.g. Lee et al. 2008).
 Zorell (2017) performs an analogous approach and obtains similar results.
 Consistently, available evidence from probit regressions (e.g. Catão and Milesi-Ferretti 2014) indicates that the probability of external crisis is significantly higher in countries with lower per-capita income.