Today, we are pleased to present a guest contribution written by Jon Frost (Bank for International Settlements), Hiro Ito (Portland State) and René van Stralen (De Nederlandsche Bank). The views expressed represent those of the authors, and do not necessarily represent those of De Nederlandsche Bank, or any other institutions the authors are affiliated with.
Volatile capital inflows can be a headache for countries’ policymakers. While foreign capital is often needed to support economic growth and investment for countries, it is notoriously “fickle” – surging in when the economy is doing well, only to “suddenly stop” or “fly out” when markets take a turn for the worse. The Covid-19 pandemic has only underscored how quickly foreign capital may stop. In normal times, policymakers thus face a choice: should they try to restrict inflows, or to address currency mismatches?
There is a large debate distinguishing these two types of policies to address the risks of inflows. The first type is capital controls, which explicitly distinguish between investors based on residence, e.g. limiting domestic investments by foreign investors. A second type is macroprudential policies, which set certain limits on activities but without distinguishing by residence. Some of these policies put limits on foreign currency mismatches, thus addressing one key type of risk from foreign inflows.
Our new paper assesses these policies with aggregate, country-level data, at annual frequency, for 83 countries over 2000–17. We find that the latter approach – macroprudential policies – may be more effective in addressing risks from volatile inflows. Indeed, we find that the volume of capital inflows is lower, and both surges in inflows and banking crises are less frequent when authorities have used macroprudential policies, especially those that target foreign currency mismatches. We do not find that the imposition of capital controls leads to the same outcomes. Overall, the results indicate that macroprudential policies – especially those that target foreign currency – may be more effective at responding to volatile capital inflows than capital controls that discriminate on the basis of residency.
Our data come from various sources, including the IMF International Financial Statistics and the recent literature. The data on macroprudential policies come from Cerutti, Claessens and Laeven (2017), and data on capital controls come from Fernández et al (2016). Banking and currency crises come from Laeven and Valencia (2018), while capital inflow surges and other episodes comes from Forbes and Warnock (2012). (In each case, we use the latest available updates of the data sets.)
To compare countries, we use a propensity score matching methodology – a two-step process in which we first estimate the probability that a country will activate a certain policy measure, and in a second step compare outcomes in countries with similar probabilities in a treatment and control group. To illustrate how this works, we are essentially comparing countries such as Indonesia in 2013, which instituted an FX-based MaP, with Jamaica in 2013, which had a similar propensity score but did not take a measure. The difference between the two is called the average treatment effect, and can be interpreted similarly to the coefficient in an ordinary least squares regression.
There are, of course, limitations to this approach. Given the large sample and high level of aggregation, we have to compare countries that may be quite different. While we exclude financial centers from the sample, we still have countries at very different levels of economic and financial development. Predicting which countries may take policy actions is difficult, and the results can change given different choices of variables, and different data sets on policy action. (We show some alternative approaches in our robustness checks.) Despite the large sample, there are relatively few observations of banking and currency crises; thus, these results must be interpreted with some caution. Finally, due to annual frequency, we cannot identify shorter-term effects of policies, which may be quite relevant.
Still, we think that this study offers new and compelling evidence, and an illustration of how the propensity score matching method can be used in assessing policy effectiveness. There are a number of promising avenues for further research with the macroprudential and capital control data. As the experience with these policies grows, we hope that research findings can inform more effective policy making for the benefit of the world’s economies.
References
Cerutti, Eugenio, Stijn Claessens and Luc Laeven (2017), “The Use and Effectiveness of Macroprudential Policies,” Journal of Financial Stability 28: 203-224.
Fernández, Andrés, Michael W. Klein, Alessandro Rebucci, Martin Schindler and Martín Uribe (2016), “Capital Control Measures: A New Dataset,” IMF Economic Review 64(3): 548-574.
Forbes, Kristin J. and Francis E. Warnock (2012), “Capital flow waves: surges, stops, flight, and retrenchment,” Journal of International Economics 88(2): 235-251.
Laeven, Luc and Fabián Valencia (2018), “Systemic Banking Crises Revisited,” IMF Working Papers WP/18/206, Washington, D.C.: International Monetary Fund.
This post written by Jon Frost, Hiro Ito and René van Stralen.
The propensity score matching seems like a great way to make better comparisons and analogies on country policy. It’s good to remember that FX traders and people who want their capital to be safe, come up with very creative ways to move their money, such as Chinese citizens use of bitcoin and crypto-currency type instruments to “mutate” or “reshuffle” their home currency into something they view as more trustworthy or at least to diversify their currency holdings. I learned from the paper, but SOME of it went over my head. I admit to being “slow to the take” sometimes. Maybe it was expounded on in the paper and I missed it, can you tell me what the green color and dash lines in Table 10 represent?? Thanks.