Fernando Eguren-Martin, Cian O’Neill, Andrej Sokol and Lukas von dem Berge
While planes were grounded, capital flew out of emerging market economies in response to the acceleration in the spread of the virus in the early stages of the Covid-19 pandemic. Was this capital flight predictable once you account for the sudden deterioration in the global financial environment? In this post we present a model that helps to think about how financial conditions and international capital flows are linked. We then apply this methodology to events observed between March and May 2020, and find that the model predicted a large increase in the likelihood of capital flight. However, the scale of outflows was abnormally large even once the sharp tightening in financial conditions is accounted for.
A series of recent papers have advanced the understanding and monitoring of ‘tail risks’ (the probability of observing extreme outcomes) across a large number of variables, including GDP growth, inflation and exchange rates. This research agenda, motivated by the growing importance, in policy and market settings alike, of going ‘beyond the mean’ and studying risks around macroeconomic outcomes, has emphasised the need to model the tails of distributions explicitly, as they are likely to display dynamics and drivers of their own.
In this post (and underlying Staff Working Paper), we apply these ideas to the study of capital flows, which are particularly prone to extreme outcomes (typically called ‘sudden stops’ and ‘bonanzas’). The paper builds on previous work, and is also closely related to ongoing work at other institutions, most notably the International Monetary Fund.
To study the forces driving capital flows, the usual approach is to distinguish between ‘pull’ (domestic) and ‘push’ (international) factors, as they have been shown to have differing impacts. We follow this approach by looking at the information contained in domestic and global financial conditions, which we measure as the common component across a wide range of asset prices, first within a country, and then at the global level. The focus on financial conditions as our main source of information is motivated by the fact that asset prices are forward-looking variables affected by expectations about the outlook and risk considerations (as are capital flows), and that they are easy to measure and observe in a timely fashion.
Unlike earlier approaches, our framework provides a characterisation of the entire distribution of capital flows into a country as a function of domestic and global financial conditions. We estimate these relationships using a panel data set of emerging market economies over the 1996 Q1-2018 Q4 period. Chart 1 shows three distributions of non-resident portfolio flows for the mean country in our sample. The first one (in blue) is the distribution that we would observe if local and global financial conditions were at their historical averages (the mean level in our sample). This can be thought of as the probability distribution of capital flows in ‘normal times’. The orange line shows a distribution conditional on global financial conditions being two standard deviations tighter than their historical mean (while country-specific conditions remain at average levels). In a situation of global stress (a ‘push’ shock) the distribution not only shifts to the left (outflows become more likely), but the left tail also becomes fatter (a sudden stop becomes even more likely). The same is true when local financial conditions tighten (a ‘pull’ shock – yellow line), albeit less markedly.
Chart 1: Distribution of non-resident portfolio flow to EMs as a function of local and global financial conditions
One of the advantages of our approach is that it allows us to extract timely signals from asset prices in order to update the likelihood of different capital flow outcomes, because these are typically reported with a significant publication lag. The outbreak of Covid-19 at the beginning of 2020 provides a prime example of the usefulness of our framework for providing an early warning of tail risks. Chart 2 shows the probability distribution of non-resident portfolio flows in a monthly version of our model that conditions on financial conditions prevailing in March 2020. Given the tightening in both global and local EM financial conditions, the distribution shifts to the left and the left tail fattens (ie particularly large outflows become more likely). Now we can place the realised (IIF-estimated) observation of non-resident portfolio flows during the March-May period into the distribution we would have predicted back in March. The star in Chart 2 does that, and it shows that outflows were extremely large even given the prevailing conditions at the time; that is, the realisation is still very much in the left tail of the distribution which factors in the tight financial conditions observed in March. Nevertheless, the conditional distribution assigned a probability of about 12% to this outcome or worse, while the distribution based on average financial conditions would have basically treated it as a zero probability event. For context, an analogous calculation for the global financial crisis puts outflows during October-December 2008 (or worse) as a 30% probability event, conditioning on financial conditions as of October. This suggests outflows during Covid-19 were larger than those seen during the global financial crisis, not only in absolute terms but also once the deterioration in the financial environment is considered.
Chart 2: Predicted and realised non-resident portfolio flows to EMs during Covid-19
In sum, we present a framework for characterising the entire distribution of capital flows to emerging market economies. Our framework not only measures the probability of extreme events as a function of prevailing market conditions, but does so in a timely fashion, before capital flows data or estimates are released. The model shows that capital outflows observed during the height of the Covid-19 pandemic exceeded those expected on the back of the deterioration in financial conditions.
Fernando Eguren-Martin works in the Bank’s Global Analysis Division, Cian O’Neill works in the Bank’s Stress Testing Strategy Division, Andrej Sokol works at the European Central Bank and Lukas von dem Berge works in the Bank’s International Surveillance Division.
If you want to get in touch, please email us at firstname.lastname@example.org or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.