VoxEU Column Development Global economy International Finance

Capital flows at risk: Taming the ebbs and flows

Capital flows to emerging markets have continued to be highly volatile since the Global Crisis. This column uses a new framework to show that country characteristics and policy responses matter for risks to future capital flows. It finds that good institutions support stable capital flows over the medium horizon, and while foreign exchange interventions seem to help mitigate downside risks to inflows caused by worsening global conditions, a tightening of capital flow measures in response to an adverse global shock is found to be counterproductive.

Capital flows to emerging markets have continued to be highly volatile since the Global Crisis, exposing these markets to recurring surges and reversals and posing challenges to their policymakers. While a sizeable body of research has focused on understanding push and pull factors driving capital flows (see Calvo et al. 1993 and the subsequent literature, as surveyed in Koepke 2019), few have examined in detail the role of policy frameworks and policy actions. Moreover, nearly all of the existing work focuses only on the effect of policies on average capital flows, and studies have so far neglected the impact of policies on the whole probability distribution of future flows.

In a recent paper, we provide a new framework for how policymakers can monitor and manage the ebbs and flows of foreign capital (Gelos et al. 2019). We focus on what current global and domestic macroeconomic and financial conditions can tell us about future capital flows. This new approach allows policymakers to quantify capital flows risks – especially capital flight and surges – and evaluate policy tools to mitigate them, thus building the foundation of a risk management framework for capital flows. Finally, this framework also allows to evaluate the inter-temporal policy trade-offs in smoothing capital flows.

To understand the intuition for our framework, consider a stylised probability density of future capital flows to a given emerging market (Figure 1). The black dashed line in the graph represents the initial state, where the mass of the density is relatively far to the right, indicating positive inflows in most future states of the world and only a small probability of outflows (represented by the small dashed area in black). The dotted vertical line shows the median predicted flows of (2% of GDP in this example). The red density represents a subsequent state where the outlook for capital flows has deteriorated (say, due to an adverse external shock). The median falls to 1.5% of GDP and the probability of capital outflows is substantially higher, reflected in a larger dashed area in red.

Figure 1 Monitoring capital flows: Shift in predicted capital flows density after a shock

Policy actions may affect the expected post-shock distribution of capital flows. The red density function in Figure 2 shows the same post-shock distribution of future capital inflows as in Figure 1. Suppose next that in response to an adverse global shock, the central bank takes some mitigating actions – for example, by intervening in the foreign exchange market. In our stylised example, such a policy action not only increases the expected median inflows conditional on a negative global shock (blue density function) from 1.5% to 1.8% of GDP, but it also reduces the tail risks associated with the global shock (the left tail of the post-shock flows distribution becomes thinner and the probability of net capital outflows declines, as shown by the blue dashed area).

Figure 2 Managing capital flows: Domestic policies and resilience to global shocks

Source: IMF staff calculations.

In this framework, risks to capital flows can be quantified in two ways. First, we can calculate the probability that capital flows will fall below a certain threshold (e.g. zero, as highlighted in Figures 1 and 2). Second, we can estimate the amount of outflows that would be reached or exceeded with a given probability, which we call ‘capital flows at risk’ (CaR). The financial risk management literature and Adrian et al. (2019) on ‘growth at risk’ quantify the latter using the 5th percentile of the distribution.

We use a quantile regression approach to estimate the entire probability distribution of future portfolio flows as a function of current global financial conditions, the domestic macroeconomic environment, and domestic structural characteristics, as well as current policy responses. We use estimates for a range of quantiles to construct an empirical distribution of predicted average capital flows during a specified period in the future. We then fit the empirical distribution to a skewed-t probability distribution (Azzalini and Capitanio 2003).

Our results show that country characteristics and policy responses matter for risks to future capital flows. For example, deeper domestic financial markets increase the likelihood of a rebound in capital flows after an adverse shock to global financial conditions. As Figure 3 shows, the median short-term flows (defined as flows 1–2 quarters ahead) after the global shock are only 0.3% of GDP in countries with shallow financial markets (red density function), compared to 1.8% of GDP in countries with more developed financial markets (blue density function). The 95th percentile value (corresponding to capital flows ‘surges’) increases with the depth of financial markets too (from 4% to 8.75% of GDP). 

Figure 3 Financial market development and a shock to global financial conditions

Notes: Red solid line shows the distribution of future portfolio inflows after a one standard-deviation increase in the BBB yield when the financial market depth variable is set at the value equal to 20th percentile in the sample. Blue dashed line shows the distribution of future portfolio inflows when the financial market depth variable is set at 80th percentile in the sample. 

We find that good institutions support stable capital flows over the medium horizon. Countries with a higher score in the Rule of Law index and more transparent central banks face fewer large inflows and outflows in response to global shocks in the medium term, defined here as 5–8 quarters ahead (there is no discernible effect in the short term, however). We also find that some policy frameworks involve intertemporal trade-offs. For example, more flexible exchange rate regimes are linked to higher risks of both large inflows and outflows in the immediate aftermath of an adverse global shock. In the medium term, however, more flexible exchange rate regimes seem to support a larger rebound of flows. 

Looking at policies, foreign exchange interventions seem to help mitigate downside risks to portfolio inflows caused by worsening global conditions. By contrast, a tightening of capital flow measures in response to an adverse global shock is found to be counterproductive (i.e. it exacerbates the risk of large outflows of capital). This may well be because capital flow measures were not sufficiently comprehensive, leading to leakages. Finally, we find little evidence for the effectiveness of monetary and macroprudential policies in shielding countries from risks caused by global shocks, although the latter seem to reduce somewhat the likelihood of capital flow surges in the medium term.

Overall, these results highlight the complexity of policymaking, and potentially shed light on why different countries choose different policy actions – for example, if the time discount factor is much higher for some policymakers than for others, they may choose a particular course of action even if their respective economies have similar characteristics and face the same global shocks.

The capital flows at risk methodology provides a promising framework for further research. In particular, further work could examine the role of fiscal policies and the differential effects of structural characteristics, policies, and global variables on different types of capital flows, such as bank lending and foreign direct investment. The effects of combining different policies could also be explored. 

Authors’ note: The views expressed herein are those of the authors and should not be attributed to the IMF, its Executive Board, or its management.


Adrian, T, N Boyarchenko, and D Giannone (2019), “Vulnerable Growth”, American Economic Review 109(4): 1263-89

Azzalini, A and A Capitanio (2003), “Distributions Generated by Perturbation of Symmetry with Emphasis on a Multivariate Skew t Distribution,” Journal of the Royal Statistical Society, Series B, 65: 367-389.

Calvo, G A, L Leiderman, and C Reinhart (1993), “Capital Inflows and Real Exchange Rate Appreciation in Latin America: The Role of External Factors,” IMF Staff Papers: 108-151.

Gelos, G, L Gornicka, R Koepke, R Sahay, and S Sgherri (2019), “Capital Flows at Risk: Taming the Ebbs and Flows”, IMF Working 19/279.

Koepke, R (2019), “What Drives Capital Flows to Emerging Markets? A Survey of the Empirical Literature”, Journal of Economic Surveys 33(2): 516–540.

1,581 Reads