Dampening Global Financial Shocks: Can Macroprudential Regulation Help (More than Capital Controls)?

Katharina Bergant; Francesco Grigoli; Niels-Jakob H Hansen; Damiano Sandri

doi:10.5089/9781513547763.001.A001

Jump to Content Jump to Main Navigation

Dampening Global Financial Shocks: Can Macroprudential Regulation Help (More than Capital Controls)?

Author: Katharina Bergant¹, Mr. Francesco Grigoli, Mr. Niels-Jakob H Hansen, and Mr. Damiano Sandri

¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: kbergant@imf.org, fgrigoli@imf.org, nhansen@imf.org, dsandri@imf.org

Publication Date:: 26 Jun 2020

eISBN:: 9781513547763

Language:: English

Keywords:: WP; capital control; real GDP; net capital; output gap; ln VIX; trend GDP; GDP growth; monetary policy shock; policy rate; Capital controls; Central bank policy rate; Emerging and frontier financial markets; Capital outflows; Capital flows; Global; outflow shock; US monetary policy shock; capital control measure; capital outflow shock; stringent capital control; capital requirements

Download PDF (1.6 MB)

We show that macroprudential regulation can considerably dampen the impact of global financial shocks on emerging markets. More specifically, a tighter level of regulation reduces the sensitivity of GDP growth to VIX movements and capital flow shocks. A broad set of macroprudential tools contribute to this result, including measures targeting bank capital and liquidity, foreign currency mismatches, and risky forms of credit. We also find that tighter macroprudential regulation allows monetary policy to respond more countercyclically to global financial shocks. This could be an important channel through which macroprudential regulation enhances macroeconomic stability. These findings on the benefits of macroprudential regulation are particularly notable since we do not find evidence that stricter capital controls provide similar gains.

Abstract

1 Introduction

A growing literature documents that fluctuations in global financial markets can severely affect financial and macroeconomic conditions in emerging markets.¹ When global financial conditions are buoyant, emerging markets tend to enjoy higher economic growth supported by abundant foreign capital inflows. Conversely, a tightening in global financial conditions—for example a sudden spike in global risk aversion as triggered by the COVID-19 pandemic—can considerably depress economic activity in emerging markets.

According to conventional macroeconomic theory, emerging markets should be able to offset the impact of global financial shocks by relying on exchange rate flexibility. Empirical evidence shows that a flexible exchange rate tends to soften the impact of foreign financial shocks on the domestic economy (Obstfeld et al., 2005). However, it falls short of providing full insulation. Global financial shocks can destabilize even emerging markets that have a flexible exchange rate, as documented in Rey (2015) and Rey (2016) among others.² The vulnerability of emerging markets to global financial shocks leads to recurrent calls for policymakers to deploy additional policy tools. The discussion often centers on the role of capital controls and foreign exchange intervention as these tools directly target international financial transactions. However, there is growing awareness that macroprudential policies can also play an important role in stabilizing credit markets as discussed later in the literature review.

Against this background, the paper examines whether macroprudential regulation—which involves a broad range of measures to buttress financial stability—can dampen the macroeconomic impact of global financial shocks on emerging markets. The hypothesis underpinning the analysis is that by reinforcing balance sheets, preventing excessive risk taking, and limiting foreign currency exposures, macroprudential regulation strengthens the resilience of the domestic financial sector and thus enhances macroeconomic stability.

Ostry et al. (2012) provide early evidence in favor of this hypothesis, showing that countries with stronger macroprudential regulation were more resilient during the global financial crisis. Similarly, Neanidis (2019) finds that stronger bank supervision reduces the negative impact of volatile capital flows on economic growth.³

This paper examines the dampening effects of macroprudential regulation more systematically, using a broad set of global financial shocks and analyzing the experience of 38 emerging markets between 2000 and 2016. During this time, emerging markets were exposed to highly volatile global financial conditions, driven by large swings in US policy rates, global risk aversion—proxied by the VIX—and capital inflows, as illustrated in panels 1 and 2 of Figure 1.⁴ Global financial volatility significantly affected emerging markets. Panels 3 in Figure 1 show that GDP in emerging markets grew rapidly during the buoyant years before the global financial crisis and contracted sharply during it.

Meanwhile, emerging markets have gradually tightened macroprudential regulation. Panel 1 of Figure 2 shows the average number of macroprudential tightening actions per country in emerging markets using the IMF’s integrated Macroprudential Policy (iMaPP) database compiled by Alam et al. (2019) .⁵ By cumulating the tightening actions (net of the loosening ones) for each country since 1990, we construct an approximate measure of the stringency of macroprudential regulation. Panel 2 of Figure 2 shows that macroprudential regulation has been considerably tightened over the years, especially since 2005. The global financial crisis led to a temporary loosening but emerging markets returned to tighten regulation during the subsequent recovery. Panel 2 also illustrates a substantial dispersion in the level of macroprudential regulation across countries.

By exploiting both time-series and cross-country variation in macroprudential regulation, we first show that a more stringent level of regulation significantly reduces the sensitivity of GDP growth in emerging markets to global financial shocks. These results are robust to a broad set of endogeneity tests, alleviating concerns about reverse causality and omitted variables.

We also find that the dampening effects of macroprudential regulation display decreasing marginal returns. At more stringent levels of macroprudential regulation, further macroprudential tightening becomes less effective in strengthening resilience. This is consistent with concerns about circumvention, whereby excessive macroprudential regulation may push financial activities outside the regulatory perimeter and increase cross-border borrowing.⁶

The dampening effects of macroprudential regulation are not driven by a particular set of tools. A broad range of macroprudential measures contribute to enhancing macroeconomic resilience to global financial shocks. These include macroprudential tools that boost bank capital and liquidity, limit foreign exchange exposures, and avert overly risky forms of credit. However, the dampening properties of these tools are heterogeneous and depend on the type of global financial shock hitting the economy. We also document that macroprudential regulation leads to symmetric dampening effects against positive and negative global financial shocks. A higher level of macroprudential regulation supports GDP growth when global financial shocks are adverse but it lowers economic activity when global financial conditions are favorable.

We also analyze if macroprudential regulation may entail negative spillovers. If a country shields itself using macroprudential regulation against global financial volatility, other countries may end up getting more exposed. However, we do not find evidence of such negative spillovers. Rather, we find that a higher level of macroprudential regulation in one country tends to enhance macroeconomic stability in other countries. This suggests that macroprudential regulation reduces the propagation of global financial shocks, possibly because enhanced resilience in a given country leads to more stable cross-border trade and financial flows.

After documenting the dampening properties of macroprudential regulation on economic activity, we investigate if macroprudential regulation allows for a more countercyclical response of monetary policy to global financial shocks. This could be an important channel through which macroprudential regulation may enhance macroeconomic stability. Central banks in emerging markets are generally reluctant to cut policy rates when global financial conditions tighten even after controlling for expected inflation (Obstfeld et al., 2005; Aizenman et al., 2016; Aizenman et al., 2017; Han and Wei, 2018; Cavallino and Sandri, 2020; Bhattarai et al., 2020). This is likely because they fear that a sharp exchange rate depreciation or large capital outflows may jeopardize financial stability. By strengthening balance sheets and limiting risk taking, macroprudential regulation should alleviate these concerns and help monetary policy to focus more squarely on macroeconomic stabilization. We find that macroprudential regulation indeed allows for a more countercyclical monetary policy response. At low levels of macroprudential regulation central banks tend to respond procyclically, by increasing rates when global financial conditions tighten. But at more stringent levels of regulation, the monetary policy response becomes countercyclical, involving a decline in policy rates when global financial conditions tighten.

Finally, to put in perspective the benefits from macroprudential regulation uncovered by the analysis, we ask how they compare to those from capital controls. The existing literature generally considers capital controls as the key policy tool beyond exchange rate flexibility to protect emerging markets from global financial shocks. This view originates from the Mundell-Fleming trilemma whereby restrictions on capital flows are expected to provide countries with greater control over domestic financial and macroeconomic conditions, thus strengthening resilience to external shocks. Following the same empirical approach used to analyze macroprudential regulation, we ask whether more stringent capital controls can also dampen global financial shocks and support a more countercyclical response of monetary policy. Using a broad set of capital control indicators provided by Chinn and Ito (2008), Fernández et al. (2015), Quinn and Toyoda (2008), and Pasricha et al. (2018), we do not find evidence that tighter capital controls provide similar benefits to those of tighter macroprudential regulation. These findings are in line with those presented in Forbes et al. (2015) and Frost et al. (2020). Using propensity score matching procedures, these studies find that macroprudential tools, especially FX-based measures, are more effective in influencing macroeconomic outcomes than capital controls.

The results of the analysis are subject to two main caveats. First, as later discussed in more detail, indexes of macroprudential regulation are still subject to significant measurement limitations. Therefore, our findings will need to be re-evaluated as better measures become available. Second, although the results are robust to a broad range of tests that exploit both time-series and cross-section identification and control for various variables, some endogeneity concerns may linger. In particular, we cannot rule out that the results could be biased by a variable that has been omitted from the empirical specifications, beyond the various ones considered in the robustness tests. To create a bias, this variable should comove over time and across countries with macroprudential regulation and it should affect macroeconomic resilience. Measurement limitations and omitted variable concerns also apply to the analysis of capital controls.

Literature review Our results relate to a growing literature on the effectiveness of macroprudential policies in controlling domestic credit and house prices. Various papers use country-level panel data as in our analysis. Cerutti et al. (2017) find that borrower-based and financial macroprudential tools affect credit growth in emerging markets. Similarly, Fendoğlu (2017) and Akinci and Olmstead-Rumsey (2018) document that macroprudential policies can curb credit and house price growth. Several other papers confirm that macroprudential policies can stabilize real house prices (Crowe et al., 2013; Cerutti et al., 2015; Kuttner and Shim, 2016; Wong et al., 2011). Looking at credit cycles, Dell’Ariccia et al. (2012) show that macroprudential policies can reduce the frequency of credit booms and decrease the severity of subsequent busts. Focusing on bank balance sheets, Claessens et al. (2013) find that macroprudential measures based on credit limits are effective in controlling leverage.

Micro-level studies confirm the effectiveness of macroprudential policies in affecting credit growth. Using data from Spain, Saurina (2009) and Jiménez et al. (2017) find that dynamic provisioning can mitigate credit booms and busts. Aiyar et al. (2016) and Dassatti et al. (2019) use bank-level data for Uruguay and the UK, respectively, and show that reserve and capital requirements have significant effects on credit. Igan and Kang (2011) and Tillmann (2015) use sectoral data from Korea to show that limits to loan-to-value and debt-to-income ratios moderate credit growth.

In addition to financial variables, a few papers have analyzed the effects of macroprudential regulation on GDP growth. Some papers find that a tightening in macroprudential policies leads to a temporary decline in GDP (Kim and Mehrotra, 2018; Eickmeier et al., 2018; Richter et al., 2019). Other papers focus on longer-term effects, finding that macroprudential policies tend to boost average economic growth (Boar et al., 2017; Agénor et al., 2018; Neanidis, 2019).

Our paper differs from the literature because we do not analyze the impact of changes in macroprudential regulation on domestic conditions. We instead examine if the level of macroprudential regulation—and later the stringency of capital controls—affects the transmission of global financial shocks to the domestic economy. Besides providing a different perspective on the effects of macroprudential regulation, this research question is subject to much less severe endogeneity concerns. While changes in regulation are likely triggered by economic developments that complicate identification, the level of regulation when global financial shocks materialize is largely pre-determined being the outcome of easing and tightening decision over the previous years. Furthermore, the robustness of the results can be validated using cross-sectional identification by examining whether countries that have on average tighter macroprudential regulation over a certain period tend to be less destabilized by global financial shocks. This approach neglects any time-series variation in macroprudential regulation, thus further addressing concerns about reverse causality.

The rest of paper is structured as follows. Section 2 analyzes if macroprudential regulation can dampen the effects of global financial shocks in emerging markets. Section 3 considers whether macroprudential regulation allows for a more countercyclical monetary policy response. Section 4 examines if capital controls are as effective as macroprudential regulation in dealing with global financial shocks. Section 5 concludes by summarizing the key insights of the paper and discussing avenues for future research.

2 Dampening effects of macroprudential regulation

Macroprudential regulation involves a broad set of policy tools that aim to contain the build up of systemic vulnerabilities and strengthen the resilience of the financial sector. These include measures to increase bank capital and liquidity, reduce leverage in the household and corporate sectors, and prevent currency mismatches. The hypothesis motivating the analysis in this paper is that, by buttressing financial sector stability, macroprudential regulation should also enhance macroeconomic resilience to global financial shocks. Does the empirical evidence support this logic?

To address this question, we estimate panel regressions of GDP growth in emerging markets over a vector of global financial shocks and their interactions with the stringency of macroprudential regulation.⁷ The coefficient estimates on the interaction terms indicate if the impact of global financial shocks on emerging markets’ GDP is affected by the level of macroprudential regulation.

More precisely, the analysis is based on the following panel regression:

\begin{matrix} Y_{i, t} = α_{i} + β S_{i, t} + γ (S_{i, t} . M \Pr u_{i, t}) + δ (S_{i, t} . M \Pr u_{i, t}^{2}) + ζ M \Pr u_{i, t} + θ M \Pr u_{i, t}^{2} + κ C_{i, t} + ε_{i, t} & (1) \end{matrix}

where Y_i,t denotes quarterly real GDP growth for country i at time t and α_i is a country fixed effect for country i. The variable S_i,t denotes the vector of global financial shocks and MPru_i,t is the level of macroprudential regulation. To allow for non-linear effects, the specification includes interaction terms of the shocks with the squared level of macroprudential regulation. Note that the coefficients on the interactions between the shocks and macroprudential regulation are estimated by exploiting both time-series and crosscountry variation in the data. They indeed capture both if the impact of the shocks becomes less severe in a given country if the level of macroprudential regulation tightens over time; and if the impact of the shocks is less severe in countries that have a tighter level of macroprudential regulation.

The regression specification (1) includes a vector of control variables C_i,t that, following Obstfeld et al. (2019), features lagged GDP growth, lagged log of real GDP per capita, institutional quality, and a linear trend. Furthermore, the vector C_i,t includes the lagged output gap to control for growth dynamics over the business cycle and commodity terms of trade since several emerging markets are large importers or exporters of commodities.⁸ We exclude time fixed effects since they would prevent us from estimating the impact of common global financial shocks on GDP and thus assess the quantitative importance of the dampening effects of macroprudential regulation. We will include time fixed effects in the robustness section, showing that the results are unchanged.

The vector S_i,t includes three types of global financial shocks. First, we consider shocks to US monetary policy to capture changes in the international risk-free rate. We use the shocks identified by Iacoviello and Navarro (2019) that have been shown to have tangible effects on emerging markets.⁹ Second, we include the Chicago Board Options Exchange’s Volatility Index (VIX), which is commonly used in the literature as a proxy for shocks to international risk premia. Finally, the vector S_i,t includes net capital inflows to capture shocks to the supply of foreign capital.¹⁰ Since net capital flows are also affected by domestic pull factors, we isolate the variation due to global push factors by instrumenting net capital inflows to country i with gross inflows to the other emerging markets following Blanchard et al. (2017). The idea is that the total amount of foreign capital flowing to emerging markets, except the country in question, captures the general appetite of international investors for emerging markets’ assets.

Most papers in the literature analyze only one of these three shocks. Our empirical framework includes all of them to capture all major sources of global financial shocks and understand which of them is more detrimental to emerging markets. Furthermore, the analysis can examine if macroprudential regulation enhances macroeconomic resilience against a broad set of shocks or only particular types. We will report the regression results throughout the paper by inverting the sign of net capital inflows, thus considering the effects of net capital outflows. In this way, all three shocks (to US rates, VIX, and net outflows) are expected to have a negative impact on emerging markets.

We construct the index of macroprudential regulation MPru_i,t by cumulating the macroprudential tightening actions net of the loosening actions reported in the iMaPP database from 1990 (the first year in the iMaPP database) to 2016 (the last year at the time of this work).¹¹ A similar approach has been used by Akinci and Olmstead-Rumsey (2018) to analyze how the stance in macroprudential regulation affects credit growth. This index is subject to two measurement drawbacks which reflect the limitations of existing macroprudential data. First, countries may have had a different level of macroprudential regulation in 1990, thus confounding cross-country rankings. Second, the dummy-type policy action indicators of the iMaPP database record when macroprudential regulation was tightened and loosened but not the intensity of those changes.¹²

These measurement problems could affect the accuracy of our regression estimates but are unlikely to drive our results on the dampening effects of macroprudential regulation. As discussed for example in Akinci and Olmstead-Rumsey (2018) and Forbes (2018), measurement imprecision should bias the analysis against finding significant effects associated with macroprudential regulation rather than generate spurious evidence about their benefits. To mitigate concerns about measurement problems, we will show that the results are robust to using both time-series and cross-sectional variation in the data.

The estimation approach uses a two-stage least squares procedure, where net capital inflows to country i (as well as their interaction with macroprudential regulation and with macroprudential regulation squared) are instrumented with gross capital inflows to all other emerging markets (and their interactions). Given the complex correlation structure of the error term involving dependence across economies, autocorrelation, and heteroscedasticity, we apply the Driscoll and Kraay (1998) correction to the standard errors to make statistical inference.¹³

2.1 Baseline results

Table 1 reports the regression results. We start in column (1) by analyzing the impact of global financial shocks on emerging markets’ GDP growth without controlling for the level of macroprudential regulation. An increase in the VIX and an outflow of capital have negative and highly statistically significant effects on economic growth. US monetary shocks have a detrimental effects on growth if the regression does not control for other global financial shocks, as shown in column (2). But they lose significance when controlling for the VIX and capital flows. Interestingly, this suggests that changes in US monetary policy affect emerging markets through changes in risk premia proxied by the VIX and the effects on the supply of foreign capital, rather than through changes in risk-free rates. Similar arguments are developed in Kalemli-Özcan (2019), who shows that US monetary policy transmits globally through changes in risk premia. Regarding the control variables, the coefficient on the lag of the output gap turns out to be negative and significant, suggesting that deviations from potential growth tend to be reduced over the following quarter. The instrumentation approach for net capital flows appears reliable since the F-statistic is well above the conventional threshold.¹⁴

^{Table 1:}

Dampening Effects of Macroprudential Regulation on GDP Growth

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. MPru is divided by 10 to ease the visualization of the coefficients. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

^{Table 1:}

Dampening Effects of Macroprudential Regulation on GDP Growth

	All EMs	All EMs	All EMs	No fixed ER	All EMs
	(1)	(2)	(3)	(4)	(5)
Lag dependent variable	0.071 (0.057)	0.117 (0.088)	0.072 (0.050)	0.012 (0.054)	0.072 (0.048)
Lag output gap	-0.378*** (0.035)	-0.313*** (0.045)	-0.369*** (0.035)	-0.342*** (0.033)	-0.368*** (0.035)
Lag ln real GDP per capita	-0.880* (0.505)	-0.144 (0.440)	-1.689** (0.707)	-1.111* (0.630)	-1.713** (0.749)
Institutional quality	-0.281 (0.644)	-0.484 (0.455)	-0.132 (0.813)	-0.536 (0.938)	0.078 (0.930)
Linear trend	0.003 (0.005)	-0.002 (0.005)	0.005 (0.007)	0.002 (0.006)	0.002 (0.007)
Commodity terms of trade	0.050 (0.035)	0.056 (0.038)	0.037 (0.050)	0.083 (0.052)	0.035 (0.058)
US monetary policy shock	-0.158 (0.168)	-0.449** (0.189)	-0.008 (0.311)	-0.077 (0.318)	0.017 (0.357)
Ln VIX	-0.712*** (0.182)		-1.556*** (0.312)	-1.571*** (0.287)	-1.762*** (0.487)
Net outflows	-0.186*** (0.046)		-0.393*** (0.113)	-0.378*** (0.122)	-0.624*** (0.183)
Mpru			-1.380*** (0.401)	-1.480*** (0.404)	-1.514 (1.234)
US monetary policy shock * Mpru			-0.100 (0.109)	-0.074 (0.118)	-0.164 (0.221)
Ln VIX * Mpru			0.631*** (0.143)	0.609*** (0.148)	0.997** (0.419)
Net outflows * Mpru			0.108*** (0.036)	0.095** (0.043)	0.319*** (0.111)
MPru2					0.237 (0.209)
US monetary policy shock * MPru2					0.037 (0.037)
Ln VIX * MPru2					-0.129* (0.073)
Net outflows * MPru2					-0.040** (0.016)
Observations	2,260	2,260	2,260	1,658	2,260
Countries	38	38	38	32	38
F-statistic	73.1		33.1	24.4	18.6

^{Table 1:}

Dampening Effects of Macroprudential Regulation on GDP Growth

	All EMs	All EMs	All EMs	No fixed ER	All EMs
	(1)	(2)	(3)	(4)	(5)
Lag dependent variable	0.071 (0.057)	0.117 (0.088)	0.072 (0.050)	0.012 (0.054)	0.072 (0.048)
Lag output gap	-0.378*** (0.035)	-0.313*** (0.045)	-0.369*** (0.035)	-0.342*** (0.033)	-0.368*** (0.035)
Lag ln real GDP per capita	-0.880* (0.505)	-0.144 (0.440)	-1.689** (0.707)	-1.111* (0.630)	-1.713** (0.749)
Institutional quality	-0.281 (0.644)	-0.484 (0.455)	-0.132 (0.813)	-0.536 (0.938)	0.078 (0.930)
Linear trend	0.003 (0.005)	-0.002 (0.005)	0.005 (0.007)	0.002 (0.006)	0.002 (0.007)
Commodity terms of trade	0.050 (0.035)	0.056 (0.038)	0.037 (0.050)	0.083 (0.052)	0.035 (0.058)
US monetary policy shock	-0.158 (0.168)	-0.449** (0.189)	-0.008 (0.311)	-0.077 (0.318)	0.017 (0.357)
Ln VIX	-0.712*** (0.182)		-1.556*** (0.312)	-1.571*** (0.287)	-1.762*** (0.487)
Net outflows	-0.186*** (0.046)		-0.393*** (0.113)	-0.378*** (0.122)	-0.624*** (0.183)
Mpru			-1.380*** (0.401)	-1.480*** (0.404)	-1.514 (1.234)
US monetary policy shock * Mpru			-0.100 (0.109)	-0.074 (0.118)	-0.164 (0.221)
Ln VIX * Mpru			0.631*** (0.143)	0.609*** (0.148)	0.997** (0.419)
Net outflows * Mpru			0.108*** (0.036)	0.095** (0.043)	0.319*** (0.111)
MPru2					0.237 (0.209)
US monetary policy shock * MPru2					0.037 (0.037)
Ln VIX * MPru2					-0.129* (0.073)
Net outflows * MPru2					-0.040** (0.016)
Observations	2,260	2,260	2,260	1,658	2,260
Countries	38	38	38	32	38
F-statistic	73.1		33.1	24.4	18.6

The level of macroprudential regulation and its interactions with global financial shocks enter the regression specification in column (3). The coefficients on the VIX and net outflows remain significant. Importantly, the coefficients on the interaction terms between the shocks and macroprudential regulation are negative and highly statistically significant. This implies that a more stringent level of macroprudential regulation dampens the effects of global financial shocks on GDP growth. Column (4) shows that the results are robust to excluding periods during which countries had a fixed exchange rate.

Column (5) extends the regression specification to include the squared level of macroprudential regulation and its interactions with the shocks. The results corroborate the contractionary effects of VIX increases and capital outflows as well as the dampening effects of macroprudential regulation. Moreover, the interaction terms of the VIX and capital flows with the squared level of regulation are negative and statistically significant. The dampening effects of macroprudential regulation are thus subject to decreasing marginal returns, thereby weakening as the level of regulation tightens.

To better understand the dampening effects of macroprudential regulation, it is helpful to analyze the derivative of GDP growth with respect to a given global financial shock j. Based on equation (1), the derivative is equal to:

\begin{matrix} \frac{\partial Y_{i, t}}{\partial S_{j, t}} = β + γ_{j} M \Pr u_{i, t} + δ_{j} M \Pr u_{i, t}^{2} & (2) \end{matrix}

which is a nonlinear function of the level of macroprudential regulation. The first two panels in Figure 3 illustrate the derivatives with respect to the VIX and capital flow shocks. More specifically, the charts show the impact of these shocks on GDP in emerging markets as a function of the stringency of macroprudential regulation on the horizontal axis. The third panel in Figure 3 shows the distribution of macroprudential regulation in emerging markets in the regression sample (between 2000–2016) and at the end of 2016.

At the lowest level of macroprudential regulation in the sample, an increase in the VIX or an outflow of capital considerably reduce economic growth in emerging markets. For example, a doubling of the VIX—an increase similar to the one occurred during the global financial crisis—leads to a decline in quarterly GDP growth by 1.8 percentage points. Similar effects are triggered by a net outflow worth three percent of GDP. Macroprudential regulation can considerably dampen these effects. When faced with the same shocks, a country with a median level of macroprudential regulation in the sample would experience a GDP decline of only 0.5 percentage points. In fact, if the level of macroprudential regulation is sufficiently tight, the VIX and net capital outflows no longer have statistically significant effects on emerging markets’ GDP.

The first two charts in Figure 3 illustrate also that the dampening effects of macroprudential regulation face decreasing marginal returns. A tightening in macroprudential regulation becomes progressively less effective in strengthening resilience to global financial shocks. These non-linearities are possibly consistent with problems of circumvention. As the stringency of regulation increases, domestic borrowers have stronger incentives to seek credit in the unregulated shadow financial market or from international lenders.¹⁵ These forms of credit are likely to be more sensitive to global financial conditions and thus could weaken the dampening effects of macroprudential regulation.

2.2 Robustness

A possible concern about the analysis is that the level of macroprudential regulation may respond to changes in GDP growth, in which case the results would be confounded by reverse causality. In this regard, it is reassuring that the level of macroprudential regulation is quite persistent and much less volatile than quarterly fluctuations in GDP growth. In fact, since the level of macroprudential regulation is obtained by cumulating all past tightening and loosening macroprudential actions, it is largely predetermined to the realization of global financial shocks and their GDP responses. Besides, macroprudential policies are generally not adjusted in reference to growth developments as documented in Richter et al. (2019) for the case of loan-to-value ratios.¹⁶

Nonetheless, in Table 2 we perform various robustness tests for reverse causality. As policymakers might be more prone to change macroprudential regulation in bad times, the first test excludes all observations with negative GDP growth (column 1). We then check if the results are robust to lagging the level of macropru-dential regulation by one quarter (column 2) and one year (column 3). Finally, to rule out reverse causality concerns, we replace the time-varying levels of macroprudential regulation with time-invariant country averages between 2000—2016 period (column 4). In this specification, the dampening effects of macroprudential regulation are identified by exclusively relying on cross-country heterogeneity in the stringency of macro-prudential regulation. In other words, we ask if countries that on average have a more stringent level of macroprudential regulation are less affected by global financial shocks.

^{Table 2:}

Robustness for Reverse Causality: Dampening Effects on GDP Growth

^{Table 2:}

Robustness for Reverse Causality: Dampening Effects on GDP Growth

	Excluding negative GDP growth	MPru = one-quarter lag of MPru	MPru = One-year lag of MPru	MPru = country average of MPru
	(1)	(2)	(3)	(4)
Lag dependent variable	-0.054 (0.039)	0.076 (0.051)	0.076 (0.054)	0.079 (0.052)
Lag output gap	-0.180*** (0.024)	-0.367*** (0.035)	-0.353*** (0.034)	-0.368*** (0.035)
Lag ln real GDP per capita	-0.791* (0.474)	-1.608** (0.736)	-1.833** (0.822)	-1.675** (0.754)
Institutional quality	-0.650 (0.501)	-0.197 (0.813)	-0.541 (0.811)	-0.343 (0.766)
Linear trend	-0.002 (0.005)	0.005 (0.007)	0.010 (0.008)	0.007 (0.007)
Commodity terms of trade	0.079* (0.048)	0.039 (0.050)	0.041 (0.055)	0.052 (0.039)
US monetary policy shock	0.207 (0.166)	-0.013 (0.298)	-0.008 (0.265)	-0.046 (0.314)
Ln VIX	-0.597*** (0.221)	-1.504*** (0.305)	-1.498*** (0.298)	-1.356*** (0.402)
Net outflows	-0.258*** (0.071)	-0.396*** (0.111)	-0.424*** (0.108)	-0.444*** (0.151)
MPru	-0.639** (0.323)	-1.286*** (0.375)	-1.534*** (0.388)
US monetary policy shock * MPru	-0.120* (0.068)	-0.101 (0.100)	-0.133 (0.090)	-0.070 (0.116)
Ln VIX * MPru	0.259** (0.120)	0.598*** (0.133)	0.683*** (0.137)	0.444** (0.208)
Net outflows * MPru	0.067*** (0.024)	0.111*** (0.035)	0.125*** (0.035)	0.149** (0.062)
Observations	1,846	2,235	2,153	2,260
Countries	38	38	38	38
F -statistic	35.1	32.9	32.6	29.7

^{Table 2:}

Robustness for Reverse Causality: Dampening Effects on GDP Growth

	Excluding negative GDP growth	MPru = one-quarter lag of MPru	MPru = One-year lag of MPru	MPru = country average of MPru
	(1)	(2)	(3)	(4)
Lag dependent variable	-0.054 (0.039)	0.076 (0.051)	0.076 (0.054)	0.079 (0.052)
Lag output gap	-0.180*** (0.024)	-0.367*** (0.035)	-0.353*** (0.034)	-0.368*** (0.035)
Lag ln real GDP per capita	-0.791* (0.474)	-1.608** (0.736)	-1.833** (0.822)	-1.675** (0.754)
Institutional quality	-0.650 (0.501)	-0.197 (0.813)	-0.541 (0.811)	-0.343 (0.766)
Linear trend	-0.002 (0.005)	0.005 (0.007)	0.010 (0.008)	0.007 (0.007)
Commodity terms of trade	0.079* (0.048)	0.039 (0.050)	0.041 (0.055)	0.052 (0.039)
US monetary policy shock	0.207 (0.166)	-0.013 (0.298)	-0.008 (0.265)	-0.046 (0.314)
Ln VIX	-0.597*** (0.221)	-1.504*** (0.305)	-1.498*** (0.298)	-1.356*** (0.402)
Net outflows	-0.258*** (0.071)	-0.396*** (0.111)	-0.424*** (0.108)	-0.444*** (0.151)
MPru	-0.639** (0.323)	-1.286*** (0.375)	-1.534*** (0.388)
US monetary policy shock * MPru	-0.120* (0.068)	-0.101 (0.100)	-0.133 (0.090)	-0.070 (0.116)
Ln VIX * MPru	0.259** (0.120)	0.598*** (0.133)	0.683*** (0.137)	0.444** (0.208)
Net outflows * MPru	0.067*** (0.024)	0.111*** (0.035)	0.125*** (0.035)	0.149** (0.062)
Observations	1,846	2,235	2,153	2,260
Countries	38	38	38	38
F -statistic	35.1	32.9	32.6	29.7

Across all these specifications, the coefficients on the interaction terms between macroprudential regulation and the VIX or net outflows remain positive and highly statistically significant. Therefore, the dampening effects of macroprudential regulation appear robust to reverse causality.

A second concern about the analysis is that the results could be affected by omitted variable bias. The dampening effects attributed to macroprudential regulation could be driven by country characteristics or policy actions that are correlated with macroprudential regulation and have been omitted from the analysis. To address these concerns, the regression specification is augmented with interaction terms between the global financial shocks and various factors that may affect resilience.

These factors include country structural characteristics such as institutional quality and financial devel-opment;¹⁷ fiscal variables such as gross public debt in percent of GDP, gross public debt in foreign currency in percent of total public debt, and the cyclically-adjusted fiscal balance in percent of GDP; and monetary policy variables such as the monetary policy rate and the anchoring of inflation expectations (Bems et al., 2018). We also control for the exchange rate regime, distinguishing between fixed and floating exchange rates (Ilzetzki et al., 2019); for the stringency of capital controls (Fernández et al., 2016); and for the stock of official reserves in percent of GDP. Table 3 shows the dampening effects of macroprudential regulation are robust to all the omitted variable tests.¹⁸

^{Table 3:}

Robustness Tests for Omitted Variables, Dampening Effects on GDP Growth

^{Table 3:}

Robustness Tests for Omitted Variables, Dampening Effects on GDP Growth

	X = institutional quality	X = financial development	X = gross public debt	X = gross public debt in foreign currency	X = cyclically adjusted balance	X = monetary policy rate	X = inflation expectation anchoring	X = fixed ER regime	X = capital flow measures	X = official reserves	Including time fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)
Lag dependent variable	0.071 (0.050)	0.064 (0.050)	0.059 (0.051)	0.090* (0.050)	0.103** (0.049)	0.085 (0.060)	0.221*** (0.071)	0.071 (0.049)	0.120** (0.047)	0.072 (0.052)	0.058 (0.059)
Lag output gap	-0.370*** (0.035)	-0.371*** (0.035)	-0.373*** (0.037)	-0.350*** (0.034)	-0.316*** (0.024)	-0.326*** (0.037)	-0.252*** (0.053)	-0.372*** (0.036)	-0.325*** (0.029)	-0.373*** (0.038)	-0.401*** (0.034)
Lag In real GDP per capita	-1.758** (0.774)	-1.405* (0.805)	-1.584** (0.720)	-0.794 (0.723)	-0.764 (0.587)	-2.705** (1.086)	-5.289 (3.605)	-1.509** (0.673)	-1.164* (0.639)	-1.437* (0.749)	-0.660 (0.467)
Institutional quality	0.433 (1.232)	-0.077 (0.887)	-0.284 (0.718)	-0.112 (0.840)	0.097 (0.582)	-0.449 (0.659)	0.432 (1.001)	-0.182 (0.797)	-0.564 (0.810)	-0.395 (0.772)	-0.844* (0.446)
Linear trend	0.006 (0.007)	0.008 (0.008)	0.007 (0.007)	0.003 (0.007)	0.003 (0.006)	0.006 (0.010)	0.029 (0.025)	0.004 (0.007)	0.002 (0.006)	-0.000 (0.007)
Commodity terms of trade	0.035 (0.048)	0.034 (0.058)	0.007 (0.051)	0.056 (0.051)	0.041 (0.029)	0.016 (0.036)	-0.172 (0.190)	0.034 (0.053)	0.028 (0.047)	0.071 (0.045)	0.047* (0.026)
US monetary policy shock	0.038 (0.323)	0.359 (0.376)	-0.295 (0.407)	-0.030 (0.148)	-0.338 (0.247)	0.328 (0.377)	-0.021 (0.389)	-0.017 (0.293)	-0.042 (0.361)	0.009 (0.254)
Ln VIX	-1.623*** (0.326)	-2.169*** (0.488)	-1.265* (0.649)	-1.372*** (0.418)	-1.294*** (0.299)	-1.999*** (0.656)	-1.555*** (0.484)	-1.540*** (0.282)	-1.996*** (0.324)	-1.592*** (0.343)
Net outflows	-0.412*** (0.120)	-0.410*** (0.109)	-0.552*** (0.160)	-0.491*** (0.141)	-0.251*** (0.091)	-0.302** (0.126)	-0.358** (0.157)	-0.419*** (0.112)	-0.321*** (0.094)	-0.525*** (0.132)	-0.108*** (0.032)
MPru	-1.412*** (0.418)	-1.285*** (0.435)	-1.439*** (0.464)	-1.428*** (0.355)	-1.402*** (0.405)	-1.399*** (0.467)	-1.679** (0.808)	-1.348*** (0.412)	-1.357*** (0.343)	-1.128*** (0.435)	-1.090*** (0.402)
US monetary policy shock * MPru	-0.108 (0.115)	-0.105 (0.107)	-0.111 (0.129)	-0.069 (0.075)	0.012 (0.081)	-0.125 (0.123)	-0.109 (0.145)	-0.087 (0.104)	-0.058 (0.098)	-0.011 (0.098)	-0.003 (0.073)
Ln VIX * MPru	0.644*** (0.152)	0.627*** (0.149)	0.620*** (0.171)	0.641*** (0.131)	0.559*** (0.148)	0.652*** (0.172)	0.816** (0.341)	0.615*** (0.148)	0.575*** (0.121)	0.522*** (0.150)	0.449*** (0.156)
Net outflows * MPru	0.117*** (0.042)	0.124*** (0.043)	0.131*** (0.042)	0.101*** (0.032)	0.065** (0.029)	0.094** (0.040)	0.097* (0.051)	0.102*** (0.035)	0.091*** (0.031)	0.064** (0.028)	0.025* (0.014)
US monetary policy shock * X	0.207 (0.128)	-0.999** (0.455)	0.007** (0.003)	-0.164 (0.444)	-0.008 (0.017)	-0.053* (0.031)	-0.118 (0.482)	-0.002 (0.104)	-0.117 (0.212)	-0.008 (0.005)
Ln VIX * X	-0.242 (0.272)	1.937 (1.193)	-0.005 (0.010)	-0.476 (0.589)	0.005 (0.032)	0.074 (0.081)	-2.203 (1.631)	-0.026 (0.280)	1.223** (0.492)	0.007 (0.011)
Net outflows * X	0.043 (0.081)	0.080 (0.270)	-0.003 (0.002)	-0.203 (0.146)	0.001 (0.009)	0.008 (0.006)	-0.475 (0.373)	-0.093* (0.054)	0.081 (0.076)	-0.009*** (0.003)
X		-8.999* (4.719)	0.025 (0.033)	2.978 (1.870)	0.131 (0.102)	-0.309 (0.251)	7.720 (5.596)	0.519 (0.877)	-4.315** (1.874)	0.020 (0.034)
Observations	2,260	2,260	2,194	1,949	1,965	1,796	1,200	2,260	1,925	2,207	2,260
Countries	38	38	38	35	35	34	20	38	31	38	38
F-statistic	5.8	10.5	2.7	9.8	29.9	19.8	0.7	24.3	18.2	10.8	23.4

^{Table 3:}

Robustness Tests for Omitted Variables, Dampening Effects on GDP Growth

	X = institutional quality	X = financial development	X = gross public debt	X = gross public debt in foreign currency	X = cyclically adjusted balance	X = monetary policy rate	X = inflation expectation anchoring	X = fixed ER regime	X = capital flow measures	X = official reserves	Including time fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)
Lag dependent variable	0.071 (0.050)	0.064 (0.050)	0.059 (0.051)	0.090* (0.050)	0.103** (0.049)	0.085 (0.060)	0.221*** (0.071)	0.071 (0.049)	0.120** (0.047)	0.072 (0.052)	0.058 (0.059)
Lag output gap	-0.370*** (0.035)	-0.371*** (0.035)	-0.373*** (0.037)	-0.350*** (0.034)	-0.316*** (0.024)	-0.326*** (0.037)	-0.252*** (0.053)	-0.372*** (0.036)	-0.325*** (0.029)	-0.373*** (0.038)	-0.401*** (0.034)
Lag In real GDP per capita	-1.758** (0.774)	-1.405* (0.805)	-1.584** (0.720)	-0.794 (0.723)	-0.764 (0.587)	-2.705** (1.086)	-5.289 (3.605)	-1.509** (0.673)	-1.164* (0.639)	-1.437* (0.749)	-0.660 (0.467)
Institutional quality	0.433 (1.232)	-0.077 (0.887)	-0.284 (0.718)	-0.112 (0.840)	0.097 (0.582)	-0.449 (0.659)	0.432 (1.001)	-0.182 (0.797)	-0.564 (0.810)	-0.395 (0.772)	-0.844* (0.446)
Linear trend	0.006 (0.007)	0.008 (0.008)	0.007 (0.007)	0.003 (0.007)	0.003 (0.006)	0.006 (0.010)	0.029 (0.025)	0.004 (0.007)	0.002 (0.006)	-0.000 (0.007)
Commodity terms of trade	0.035 (0.048)	0.034 (0.058)	0.007 (0.051)	0.056 (0.051)	0.041 (0.029)	0.016 (0.036)	-0.172 (0.190)	0.034 (0.053)	0.028 (0.047)	0.071 (0.045)	0.047* (0.026)
US monetary policy shock	0.038 (0.323)	0.359 (0.376)	-0.295 (0.407)	-0.030 (0.148)	-0.338 (0.247)	0.328 (0.377)	-0.021 (0.389)	-0.017 (0.293)	-0.042 (0.361)	0.009 (0.254)
Ln VIX	-1.623*** (0.326)	-2.169*** (0.488)	-1.265* (0.649)	-1.372*** (0.418)	-1.294*** (0.299)	-1.999*** (0.656)	-1.555*** (0.484)	-1.540*** (0.282)	-1.996*** (0.324)	-1.592*** (0.343)
Net outflows	-0.412*** (0.120)	-0.410*** (0.109)	-0.552*** (0.160)	-0.491*** (0.141)	-0.251*** (0.091)	-0.302** (0.126)	-0.358** (0.157)	-0.419*** (0.112)	-0.321*** (0.094)	-0.525*** (0.132)	-0.108*** (0.032)
MPru	-1.412*** (0.418)	-1.285*** (0.435)	-1.439*** (0.464)	-1.428*** (0.355)	-1.402*** (0.405)	-1.399*** (0.467)	-1.679** (0.808)	-1.348*** (0.412)	-1.357*** (0.343)	-1.128*** (0.435)	-1.090*** (0.402)
US monetary policy shock * MPru	-0.108 (0.115)	-0.105 (0.107)	-0.111 (0.129)	-0.069 (0.075)	0.012 (0.081)	-0.125 (0.123)	-0.109 (0.145)	-0.087 (0.104)	-0.058 (0.098)	-0.011 (0.098)	-0.003 (0.073)
Ln VIX * MPru	0.644*** (0.152)	0.627*** (0.149)	0.620*** (0.171)	0.641*** (0.131)	0.559*** (0.148)	0.652*** (0.172)	0.816** (0.341)	0.615*** (0.148)	0.575*** (0.121)	0.522*** (0.150)	0.449*** (0.156)
Net outflows * MPru	0.117*** (0.042)	0.124*** (0.043)	0.131*** (0.042)	0.101*** (0.032)	0.065** (0.029)	0.094** (0.040)	0.097* (0.051)	0.102*** (0.035)	0.091*** (0.031)	0.064** (0.028)	0.025* (0.014)
US monetary policy shock * X	0.207 (0.128)	-0.999** (0.455)	0.007** (0.003)	-0.164 (0.444)	-0.008 (0.017)	-0.053* (0.031)	-0.118 (0.482)	-0.002 (0.104)	-0.117 (0.212)	-0.008 (0.005)
Ln VIX * X	-0.242 (0.272)	1.937 (1.193)	-0.005 (0.010)	-0.476 (0.589)	0.005 (0.032)	0.074 (0.081)	-2.203 (1.631)	-0.026 (0.280)	1.223** (0.492)	0.007 (0.011)
Net outflows * X	0.043 (0.081)	0.080 (0.270)	-0.003 (0.002)	-0.203 (0.146)	0.001 (0.009)	0.008 (0.006)	-0.475 (0.373)	-0.093* (0.054)	0.081 (0.076)	-0.009*** (0.003)
X		-8.999* (4.719)	0.025 (0.033)	2.978 (1.870)	0.131 (0.102)	-0.309 (0.251)	7.720 (5.596)	0.519 (0.877)	-4.315** (1.874)	0.020 (0.034)
Observations	2,260	2,260	2,194	1,949	1,965	1,796	1,200	2,260	1,925	2,207	2,260
Countries	38	38	38	35	35	34	20	38	31	38	38
F-statistic	5.8	10.5	2.7	9.8	29.9	19.8	0.7	24.3	18.2	10.8	23.4

Column (11) in Table 3 shows that the results are also robust to the inclusion of time fixed effects, which capture all factors common to the countries in our sample. In this specification, we can no longer estimate the impact of US monetary shocks and the VIX since these shocks are also common to all countries. However, we can still estimate the interaction terms of the global financial shocks with the level of macroprudential regulation. The interactions with the VIX and capital outflows remain positive and statistically significant, thus confirming the dampening effects of macroprudential regulation. Finally, we also verified that the results are robust to excluding any country from the sample and to dropping the period of the global financial crisis.¹⁹

2.3 Symmetric dampening effects

The previous analysis established that macroprudential regulation reduces the sensitivity of GDP growth in emerging markets to VIX and capital outflow shocks. Are these dampening effects at play against both positive and negative shocks? To address this question, we extend the regression specification to include dummies that differentiate between an increase or decrease in the global financial shocks:

\begin{matrix} Y_{i, t} & = & α_{i} + β^{+} D_{i, t}^{+} s_{i, t}^{n} + β^{-} D_{i, t}^{-} s_{i, t}^{n} + γ^{+} D_{i, t}^{+} s_{i, t}^{n} M \Pr u_{i, t} + γ^{-} D_{i, t}^{-} s_{i, t}^{n} M \Pr u_{i, t} \\ + & ς^{+} D_{i, t}^{+} M \Pr u_{i, t} + ς^{-} D_{i, t}^{-} M \Pr u_{i, t} + β^{^{n}} S_{i, t}^{^{n}} + γ^{^{n}} S_{i, t}^{^{n}} M \Pr u_{i, t} + κ C_{i, t} + ε_{i, t} & (3) \end{matrix}

where $s_{i, t}^{n} ϵ S_{i, t}$ is a specific global shock, $D_{i, t}^{+} (D_{i, t}^{-})$ is a dummy variable that takes value one when shock $s_{i, t}^{n}$ is positive (negative), and zero otherwise. The coefficients γ⁺ and γ^- measure the dampening effects of macroprudential regulation under a positive and negative shock realization, respectively. The vector $S_{i, t}^{n}$ includes the global financial shocks other than $s_{i, t}^{n}$ .

Table 4 shows that the γ⁺ and γ^- coefficients for both the VIX (column 1) and capital outflows (column 2) are statistically significant and positive. Furthermore, the coefficients are quite similar in size, and a Wald test confirms that they are not statistically different. Therefore, macroprudential regulation entails symmetric dampening effects against both positive and negative global financial shocks. This implies that, while a tighter level of macroprudential regulation supports economic growth in case of negative financial shocks, it also lowers economic activity when global financial shocks are positive.

^{Table 4:}

Symmetric Dampening Effects of Macroprudential Regulation on GDP Growth

^{Table 4:}

Symmetric Dampening Effects of Macroprudential Regulation on GDP Growth

	Asymmetric dampening against ln VIX	Asymmetric dampening effects against net outflows
	(1)	(2)
Lag dependent variable	0.070 (0.047)	0.100* (0.051)
Lag output gap	-0.396*** (0.036)	-0.388*** (0.037)
Lag ln real GDP per capita	-1.676** (0.738)	-1.968** (0.776)
Institutional quality	-0.075 (0.854)	-0.048 (0.931)
Linear trend	0.005 (0.007)	0.008 (0.007)
Commodity terms of trade	0.035 (0.054)	0.033 (0.077)
US monetary policy shock	0.069 (0.251)	0.082 (0.314)
Ln VIX		-1.555*** (0.329)
Net outflows	-0.420*** (0.106)
US monetary policy shock * MPru	-0.117 (0.101)	-0.132 (0.122)
Ln VIX * MPru		0.634*** (0.158)
Net outflows * MPru	0.114*** (0.035)
Mpru * D+	-1.101** (0.491)	-1.324*** (0.466)
Mpru * D-	-1.721*** (0.606)	-1.346*** (0.436)
Ln VIX * D+	-1.080** (0.475)
Ln VIX * D-	-2.139*** (0.461)
Ln VIX * D+ * MPru	0.535*** (0.164)
Ln VIX * D- * MPru	0.751*** (0.225)
Net outflows * D+		-0.456*** (0.142)
Net outflows * D-		-0.478*** (0.163)
Net outflows * D+ * MPru		0.126** (0.051)
Net outflows * D- * MPru		0.132** (0.055)
Observations	2,260	2,260
Countries	38	38
F -statistic	27.3	6.9
Wald test (p-value)	0.421	0.943

^{Table 4:}

Symmetric Dampening Effects of Macroprudential Regulation on GDP Growth

	Asymmetric dampening against ln VIX	Asymmetric dampening effects against net outflows
	(1)	(2)
Lag dependent variable	0.070 (0.047)	0.100* (0.051)
Lag output gap	-0.396*** (0.036)	-0.388*** (0.037)
Lag ln real GDP per capita	-1.676** (0.738)	-1.968** (0.776)
Institutional quality	-0.075 (0.854)	-0.048 (0.931)
Linear trend	0.005 (0.007)	0.008 (0.007)
Commodity terms of trade	0.035 (0.054)	0.033 (0.077)
US monetary policy shock	0.069 (0.251)	0.082 (0.314)
Ln VIX		-1.555*** (0.329)
Net outflows	-0.420*** (0.106)
US monetary policy shock * MPru	-0.117 (0.101)	-0.132 (0.122)
Ln VIX * MPru		0.634*** (0.158)
Net outflows * MPru	0.114*** (0.035)
Mpru * D+	-1.101** (0.491)	-1.324*** (0.466)
Mpru * D-	-1.721*** (0.606)	-1.346*** (0.436)
Ln VIX * D+	-1.080** (0.475)
Ln VIX * D-	-2.139*** (0.461)
Ln VIX * D+ * MPru	0.535*** (0.164)
Ln VIX * D- * MPru	0.751*** (0.225)
Net outflows * D+		-0.456*** (0.142)
Net outflows * D-		-0.478*** (0.163)
Net outflows * D+ * MPru		0.126** (0.051)
Net outflows * D- * MPru		0.132** (0.055)
Observations	2,260	2,260
Countries	38	38
F -statistic	27.3	6.9
Wald test (p-value)	0.421	0.943

Figure 4 provides an illustration of the dampening effects of macroprudential regulation against positive and negative shocks.²⁰ It shows the GDP growth rate differential between a country with high and low level of macroprudential regulation over the sample period 2000–2016. These levels are based on the 75th and 25th percentiles of the distribution of macroprudential regulation in the sample of analysis.

Higher regulation would have delivered significantly stronger economic growth in the early 2000s and during the global financial crisis, when global financial conditions were adverse. For example, higher macroprudential regulation would have increased quarterly GDP growth by about 0.6 percent between the fourth quarter of 2008 and the second quarter of 2009. However, macroprudential regulation would have lowered economic growth considerably in the years before the global financial crises, when global financial conditions were buoyant. By doing so, macroprudential regulation reduces the amplitude of economic fluctuations by sustaining growth in the face of adverse shocks while lowering economic activity when global financial conditions are supportive.

Maintaining a high level of macroprudential regulation to dampen negative financial shocks is thus not costless since it implies foregoing growth opportunities when global financial conditions are favorable. At the same time, it would be improper for policymarkers to tighten regulation only when global financial conditions become adverse. Constraining excessive risk-taking and credit provision when financial conditions are loose— thus limiting economic activity—is a key channel through which macroprudential regulation ensures greater resilience at times of financial distress. These considerations call for further analysis on how to optimally adjust macroprudential regulation to maximize the dampening effects against negative shocks without unduly constraining economic activity when financial conditions are supportive.

2.4 Categories of macroprudential measures

Up to this point, we have used an overall index of macroprudential regulation that combines a broad range of individual measures recorded in the iMaPP database. We now examine if the dampening effects of macroprudential regulation are driven by specific measures. To investigate this issue, we replicate the analysis using more disaggregated categories of macroprudential regulation, including measures targeted at bank capital and liquidity, credit demand (such as loan-to-value ratios), credit supply (such as limits on credit growth), and foreign currency exposure.²¹

Table 5 analyzes the dampening effects of each macroprudential category. Figure 5 displays the results, reporting the coefficient estimates on the interaction terms of each macroprudential category with the VIX (panel 1) and net capital outflows (panel 2). Positive and statistically significant coefficients denote dampening effects. Measures targeted at credit demand, FX exposure, and liquidity dampen the impact on GDP growth arising from the VIX. Macroprudential measures targeted at bank capital, credit demand, and credit supply strengthen resilience against net capital outflows.

^{Table 5:}

Dampening Effects of Macroprudential Categories on GDP Growth

^{Table 5:}

Dampening Effects of Macroprudential Categories on GDP Growth

	Mpru = capital	Mpru = credit demand	Mpru = credit supply	Mpru = FX exposure	Mpru = liquidity
	(1)	(2)	(3)	(4)	(5)
Lag dependent variable	0.061 (0.054)	0.067 (0.053)	0.067 (0.055)	0.066 (0.055)	0.071 (0.054)
Lag output gap	-0.374*** (0.039)	-0.381*** (0.036)	-0.374*** (0.035)	-0.383*** (0.036)	-0.372*** (0.035)
Lag ln real GDP per capita	-2.012** (0.972)	-0.773 (0.594)	-0.936 (0.584)	-0.682 (0.531)	-1.193* (0.627)
Institutional quality	1.456 (1.439)	0.039 (0.733)	-0.207 (0.688)	-0.417 (0.661)	-0.300 (0.677)
Linear trend	0.009 (0.010)	-0.001 (0.006)	0.002 (0.006)	-0.000 (0.005)	0.003 (0.006)
Commodity terms of trade	0.058 (0.079)	0.049 (0.047)	0.046 (0.046)	0.049 (0.038)	0.052 (0.037)
US monetary policy shock	-0.059 (0.194)	-0.155 (0.168)	-0.146 (0.172)	-0.162 (0.174)	-0.160 (0.173)
Ln VIX	-0.524 (0.435)	-0.741*** (0.193)	-0.672*** (0.197)	-0.777*** (0.179)	-0.766*** (0.196)
Net outflows	-0.495*** (0.177)	-0.247*** (0.064)	-0.223*** (0.059)	-0.199*** (0.049)	-0.205*** (0.053)
Mpru	8.040 (12.012)	-3.021 (2.376)	0.886 (5.846)	-10.270*** (2.420)	-2.044** (0.887)
US monetary policy shock * Mpru	-1.106 (1.190)	-0.245 (0.422)	-0.349 (0.855)	0.169 (0.646)	0.067 (0.136)
Ln VIX * Mpru	-1.119 (4.752)	1.663* (0.866)	0.806 (1.918)	3.324*** (0.756)	0.846*** (0.248)
Net outflows * Mpru	2.390** (1.129)	0.638*** (0.223)	0.709** (0.300)	-0.115 (0.128)	0.062 (0.072)
MPru²	-19.297 (19.623)	2.019 (2.647)	-2.833 (6.464)	10.829*** (2.698)	0.315 (0.434)
US monetary policy shock * MPru²	0.501 (2.061)	0.390 (0.356)	0.245 (0.858)	0.017 (0.743)	-0.054 (0.078)
Ln VIX * MPru²	5.784 (7.565)	-1.179 (1.005)	0.073 (2.201)	-3.068*** (0.844)	-0.127 (0.146)
Net outflows * MPru²	-3.214** (1.624)	-0.426*** (0.163)	-0.886 (5.846)	0.318** (0.145)	0.015 (0.042)
Observations	2,260	2,260	2,260	2,260	2,260
Countries	38	38	38	38	38
F -statistic	4.0	14.2	12.7	22.9	12.4

^{Table 5:}

Dampening Effects of Macroprudential Categories on GDP Growth

	Mpru = capital	Mpru = credit demand	Mpru = credit supply	Mpru = FX exposure	Mpru = liquidity
	(1)	(2)	(3)	(4)	(5)
Lag dependent variable	0.061 (0.054)	0.067 (0.053)	0.067 (0.055)	0.066 (0.055)	0.071 (0.054)
Lag output gap	-0.374*** (0.039)	-0.381*** (0.036)	-0.374*** (0.035)	-0.383*** (0.036)	-0.372*** (0.035)
Lag ln real GDP per capita	-2.012** (0.972)	-0.773 (0.594)	-0.936 (0.584)	-0.682 (0.531)	-1.193* (0.627)
Institutional quality	1.456 (1.439)	0.039 (0.733)	-0.207 (0.688)	-0.417 (0.661)	-0.300 (0.677)
Linear trend	0.009 (0.010)	-0.001 (0.006)	0.002 (0.006)	-0.000 (0.005)	0.003 (0.006)
Commodity terms of trade	0.058 (0.079)	0.049 (0.047)	0.046 (0.046)	0.049 (0.038)	0.052 (0.037)
US monetary policy shock	-0.059 (0.194)	-0.155 (0.168)	-0.146 (0.172)	-0.162 (0.174)	-0.160 (0.173)
Ln VIX	-0.524 (0.435)	-0.741*** (0.193)	-0.672*** (0.197)	-0.777*** (0.179)	-0.766*** (0.196)
Net outflows	-0.495*** (0.177)	-0.247*** (0.064)	-0.223*** (0.059)	-0.199*** (0.049)	-0.205*** (0.053)
Mpru	8.040 (12.012)	-3.021 (2.376)	0.886 (5.846)	-10.270*** (2.420)	-2.044** (0.887)
US monetary policy shock * Mpru	-1.106 (1.190)	-0.245 (0.422)	-0.349 (0.855)	0.169 (0.646)	0.067 (0.136)
Ln VIX * Mpru	-1.119 (4.752)	1.663* (0.866)	0.806 (1.918)	3.324*** (0.756)	0.846*** (0.248)
Net outflows * Mpru	2.390** (1.129)	0.638*** (0.223)	0.709** (0.300)	-0.115 (0.128)	0.062 (0.072)
MPru²	-19.297 (19.623)	2.019 (2.647)	-2.833 (6.464)	10.829*** (2.698)	0.315 (0.434)
US monetary policy shock * MPru²	0.501 (2.061)	0.390 (0.356)	0.245 (0.858)	0.017 (0.743)	-0.054 (0.078)
Ln VIX * MPru²	5.784 (7.565)	-1.179 (1.005)	0.073 (2.201)	-3.068*** (0.844)	-0.127 (0.146)
Net outflows * MPru²	-3.214** (1.624)	-0.426*** (0.163)	-0.886 (5.846)	0.318** (0.145)	0.015 (0.042)
Observations	2,260	2,260	2,260	2,260	2,260
Countries	38	38	38	38	38
F -statistic	4.0	14.2	12.7	22.9	12.4

These findings indicate that the dampening effects of macroprudential regulation are not driven by a narrow set of measures. In particular, they are not limited to measures targeted at foreign currency exposures that could operate more similarly to capital controls by constraining borrowing in foreign currency. Macroprudential regulation that ensures adequate bank capital and liquidity and prevents excessive risk-taking in credit provision also plays a crucial role in fostering resilience to global financial shocks. This suggests that countries that want to enhance the reliance against both VIX and capital flow shocks should adopt a well-rounded macroprudential framework rather than narrowly focusing on few specific measures.

2.5 Cross-country spillovers

This section examines if macroprudential regulation generates cross-country spillovers. If a country protects itself from swings in global financial conditions through tight macroprudential regulation, other countries could be exposed to greater volatility.²² For example, measures that curb risk taking in a given country could lead to the relocation of risky financial activities to other countries (Houston et al., 2012; Jiménez et al., 2017; McCann and O’Toole, 2019), thus making them more susceptible to global financial shocks. However, macroprudential regulation may also entail positive cross-country spillovers. If a country becomes more resilient to global financial shocks owing to macroprudential regulation, other countries may enjoy greater stability through less volatile trade and financial flows with that country.

To assess possible cross-country spillovers, we extend the specification in equation (1) as follows:

\begin{matrix} Y_{i, t} = α_{i} + β S_{i, t} + γ (S_{i, t} M \Pr u_{i, t}) + \bar{γ} (S_{i, t} {\bar{M \Pr u}}_{i, t}) + \bar{ζ} {\bar{M \Pr u}}_{i, t} + κ C_{i, t} + ε_{i, t} & (4) \end{matrix}

Besides interacting the global financial shocks with the level of macroprudential regulation in a given country, the regression includes an interaction term of the shock vector with the average level of regulation ${\bar{M \Pr u}}_{i, t}$ in emerging markets other than country i. This new interaction term captures if the sensitivity of GDP growth in a given country to global financial shocks is affected by the level of macroprudential regulation in other countries. The average level of regulation ${\bar{M \Pr u}}_{i, t}$ is computed by weighting countries according to the average size of gross capital inflows that they received during the sample period. The idea is that countries that are more integrated financially are likely to generate larger spillovers.

Table 6 reports the regression results considering alternative ways to select the countries to be included in the average ${\bar{M \Pr u}}_{i, t}$ . In column (1), we start by computing ${\bar{M \Pr u}}_{i, t}$ using the average level of regulation in all other emerging markets besides the country i. We then differentiate countries across several characteristics to capture that spillovers are more likely to occur across similar economies, following Giordani et al. (2017).

^{Table 6:}

Regressions of Real GDP Growth, Spillovers

^{Table 6:}

Regressions of Real GDP Growth, Spillovers

	Others’ MPru = EMs’ average	Others’ MPru = average across EMs in the same region	Others’ MPru = average across EMs in the same income class (time varying)	Others’ MPru = average across EMs in the same income class (time invariant)	Others’ MPru = average across EMs in the same risk class (time varying)	Others’ MPru = average across EMs in the same risk class (time invariant)
	(1)	(2)	(3)	(4)	(5)	(6)
Lag dependent variable	0.024 (0.055)	0.070 (0.050)	0.071 (0.048)	0.055 (0.048)	0.067 (0.055)	0.052 (0.047)
Lag output gap	-0.382*** (0.046)	-0.381*** (0.037)	-0.381*** (0.039)	-0.360*** (0.034)	-0.339*** (0.031)	-0.359*** (0.032)
Lag ln real GDP per capita	-1.448 (1.235)	-1.579** (0.723)	-0.992 (0.779)	-1.763** (0.728)	-3.431*** (1.038)	-1.568** (0.771)
Institutional quality	2.967 (1.911)	0.416 (0.865)	0.950 (1.083)	0.394 (0.884)	2.327* (1.203)	0.865 (0.883)
Linear trend	-0.089* (0.047)	0.008 (0.008)	-0.003 (0.008)	0.011 (0.009)	-0.011 (0.008)	-0.009 (0.009)
Commodity terms of trade	0.089 (0.097)	0.063 (0.051)	0.024 (0.063)	0.022 (0.050)	0.063 (0.063)	0.067 (0.050)
US monetary policy shock	0.137 (0.699)	0.028 (0.349)	-0.173 (0.475)	-0.253 (0.448)	0.008 (0.387)	-0.058 (0.331)
Ln VIX	-1.469* (0.794)	-1.748*** (0.472)	-1.236** (0.579)	-1.498*** (0.471)	-1.817*** (0.484)	-1.669*** (0.458)
Net outflows	-1.367*** (0.479)	-0.600*** (0.153)	-0.683*** (0.226)	-0.535*** (0.168)	-0.828*** (0.225)	-0.626*** (0.152)
MPru	-2.864*** (0.932)	-1.582*** (0.459)	-1.855*** (0.556)	-1.429*** (0.456)	-1.479*** (0.463)	-1.630*** (0.453)
US monetary policy shock * MPru	-0.166 (0.145)	-0.105 (0.120)	-0.097 (0.136)	-0.088 (0.125)	-0.109 (0.116)	-0.084 (0.100)
Ln VIX * MPru	1.207*** (0.374)	0.709*** (0.168)	0.731*** (0.197)	0.635*** (0.165)	0.747*** (0.178)	0.692*** (0.165)
Net outflows * MPru	0.148*** (0.054)	0.120*** (0.040)	0.138*** (0.053)	0.121*** (0.044)	0.145*** (0.049)	0.105*** (0.034)
US monetary policy shock * others’ MPru	0.005 (0.195)	-0.025 (0.070)	0.059 (0.075)	0.077 (0.061)	-0.021 (0.102)	0.004 (0.069)
Ln VIX * others’ MPru	-0.515 (0.435)	0.031 (0.131)	-0.149 (0.189)	-0.004 (0.139)	0.130 (0.144)	0.034 (0.104)
Net outflows * others’ MPru	0.418** (0.175)	0.120*** (0.031)	0.113** (0.044)	0.068** (0.031)	0.170*** (0.050)	0.116*** (0.032)
Others’ MPru	4.296* (2.388)	-0.027 (0.367)	0.863 (0.682)	0.148 (0.433)	0.486 (0.480)	0.469 (0.336)
Observations	2,260	2,192	2,260	2,260	2,108	2,260
Countries	38	37	38	38	35	38
F -statistic	4.1	19.6	9.4	6.4	10.8	20.7
Wald test (p-value)	0.063	0.993	0.384	0.078	0.511	0.739

^{Table 6:}

Regressions of Real GDP Growth, Spillovers

	Others’ MPru = EMs’ average	Others’ MPru = average across EMs in the same region	Others’ MPru = average across EMs in the same income class (time varying)	Others’ MPru = average across EMs in the same income class (time invariant)	Others’ MPru = average across EMs in the same risk class (time varying)	Others’ MPru = average across EMs in the same risk class (time invariant)
	(1)	(2)	(3)	(4)	(5)	(6)
Lag dependent variable	0.024 (0.055)	0.070 (0.050)	0.071 (0.048)	0.055 (0.048)	0.067 (0.055)	0.052 (0.047)
Lag output gap	-0.382*** (0.046)	-0.381*** (0.037)	-0.381*** (0.039)	-0.360*** (0.034)	-0.339*** (0.031)	-0.359*** (0.032)
Lag ln real GDP per capita	-1.448 (1.235)	-1.579** (0.723)	-0.992 (0.779)	-1.763** (0.728)	-3.431*** (1.038)	-1.568** (0.771)
Institutional quality	2.967 (1.911)	0.416 (0.865)	0.950 (1.083)	0.394 (0.884)	2.327* (1.203)	0.865 (0.883)
Linear trend	-0.089* (0.047)	0.008 (0.008)	-0.003 (0.008)	0.011 (0.009)	-0.011 (0.008)	-0.009 (0.009)
Commodity terms of trade	0.089 (0.097)	0.063 (0.051)	0.024 (0.063)	0.022 (0.050)	0.063 (0.063)	0.067 (0.050)
US monetary policy shock	0.137 (0.699)	0.028 (0.349)	-0.173 (0.475)	-0.253 (0.448)	0.008 (0.387)	-0.058 (0.331)
Ln VIX	-1.469* (0.794)	-1.748*** (0.472)	-1.236** (0.579)	-1.498*** (0.471)	-1.817*** (0.484)	-1.669*** (0.458)
Net outflows	-1.367*** (0.479)	-0.600*** (0.153)	-0.683*** (0.226)	-0.535*** (0.168)	-0.828*** (0.225)	-0.626*** (0.152)
MPru	-2.864*** (0.932)	-1.582*** (0.459)	-1.855*** (0.556)	-1.429*** (0.456)	-1.479*** (0.463)	-1.630*** (0.453)
US monetary policy shock * MPru	-0.166 (0.145)	-0.105 (0.120)	-0.097 (0.136)	-0.088 (0.125)	-0.109 (0.116)	-0.084 (0.100)
Ln VIX * MPru	1.207*** (0.374)	0.709*** (0.168)	0.731*** (0.197)	0.635*** (0.165)	0.747*** (0.178)	0.692*** (0.165)
Net outflows * MPru	0.148*** (0.054)	0.120*** (0.040)	0.138*** (0.053)	0.121*** (0.044)	0.145*** (0.049)	0.105*** (0.034)
US monetary policy shock * others’ MPru	0.005 (0.195)	-0.025 (0.070)	0.059 (0.075)	0.077 (0.061)	-0.021 (0.102)	0.004 (0.069)
Ln VIX * others’ MPru	-0.515 (0.435)	0.031 (0.131)	-0.149 (0.189)	-0.004 (0.139)	0.130 (0.144)	0.034 (0.104)
Net outflows * others’ MPru	0.418** (0.175)	0.120*** (0.031)	0.113** (0.044)	0.068** (0.031)	0.170*** (0.050)	0.116*** (0.032)
Others’ MPru	4.296* (2.388)	-0.027 (0.367)	0.863 (0.682)	0.148 (0.433)	0.486 (0.480)	0.469 (0.336)
Observations	2,260	2,192	2,260	2,260	2,108	2,260
Countries	38	37	38	38	35	38
F -statistic	4.1	19.6	9.4	6.4	10.8	20.7
Wald test (p-value)	0.063	0.993	0.384	0.078	0.511	0.739

In column (2), ${\bar{M \Pr u}}_{i, t}$ is computed using only countries within the same geographical region. Columns (3) and (4) differentiate countries based on whether their GDP per capita is above or below the median of the country sample. Column (3) does so by considering the GDP levels in each quarter, so that countries move across groups over time. Column (4) considers instead that average levels of GDP between the 2002/2016 sample period, in which case the group assignment is time invariant. Following the same approach used for GDP, columns (5) and (6) differentiate countries according to their risk class. We use a composite risk index provided by the Political Risk Service that broadly captures the economic, financial, institutional, and political risks of a given country. This index has been used in Giordani et al. (2017) to analyze spillovers from capital flow management measures.

The analysis does not find evidence of negative spillovers. On the contrary, spillovers tend to be positive vis-à-vis net outflows. The coefficient on the interaction between net outflows and the average level of macroprudential regulation in other emerging markets is positive and significant regardless of how the average ${\bar{M \Pr u}}_{i, t}$ is computed. This implies that a country becomes more resilient to capital flow shocks if other emerging markets have a higher level of macroprudential regulation.

Regarding the magnitudes, the positive spillover effects associated with capital flow shocks appear sizeable. Across columns (2) to (6) the coefficients on the interaction of net capital outflows with ${\bar{M \Pr u}}_{i, t}$ are similar to those on the interaction with each country’s level of macroprudential regulation. In other words, countries experience similar dampening effects from their own level of macroprudential regulation than from the average level in other similar countries. However, it is important to recognize that the coefficients on the interaction of net outflows with capture the effect of a one-unit increase in the average level of macroprudential regulation in all other emerging markets within the same group. This is a larger macroprudential tightening than a one-unit increase of an economy’s own level of macroprudential regulation.

3 Macroprudential regulation and monetary policy

According to the Mundell-Fleming trilemma, countries that are open to capital flows can retain monetary independence if they have a flexible exchange rate (Fleming, 1962; Mundell, 1963). Monetary independence can be broadly defined as the ability to set interest rates and stabilize domestic macroeconomic conditions independently of swings in global monetary and financial conditions. In line with the trilemma, there is evidence that policy rates in countries with flexible exchange rates are less responsive to US monetary policy and the VIX than in countries with fixed exchange rates (Obstfeld, 2015).

However, even in emerging markets with flexible exchange rates, several central banks tend to increase policy rates in response to a US monetary tightening or a spike in the VIX, even after controlling for inflation (Obstfeld et al., 2005; Aizenman et al., 2016; Aizenman et al., 2017; Han and Wei, 2018; Cavallino and Sandri, 2020; Bhattarai et al., 2020). This is possibly to limit fluctuations in exchange rates and capital flows that may undermine financial stability. In these situations, monetary policy appears to operate procyclically, exacerbating the negative effects of tighter global financial conditions on domestic economic growth.

Against this backdrop, we ask whether macroprudential regulation—by mitigating financial stability concerns—can allow for a more countercyclical monetary policy response to global financial shocks. Some evidence in this favor is provided by Aizenman et al. (2017) who find that a tighter stance of macroprudential policy reduces the sensitivity of policy rates in developing countries to monetary policy conducted in center economies (the US, Japan, and the Euro Area).²³ We examine this issue further by using a panel regression framework and by considering a broader set of global financial shocks.

The analysis is based on the following panel specification:

\begin{matrix} I_{i, t} = α_{i} + β S_{i, t} + γ (S_{i, t} . M \Pr u_{i, t}) + ζ M \Pr u_{i, t} + κ C_{i, t} + ε_{i, t} & (5) \end{matrix}

where the dependent variable I_i,t is the policy rate in a given emerging market i at time t. The vector of global shocks S_i,t includes the US policy rate, the VIX, and net capital outflows which we instrument as in the previous sections. Note that we use the US policy rate rather than the US policy rate shocks as in Section 2. This is in line with the literature on the trilemma and with the fact that policy rates in emerging markets react to changes in actual US policy rates rather than to their unexpected components (as will be shown in Table 7). The coefficients γ capture if the response of domestic monetary policy to global financial shocks is affected by the level of macroprudential regulation. Specifically negative coefficients would indicate that monetary policy responds more countercyclically to a tightening in global financial conditions when macroprudential regulation is tighter.

^{Table 7:}

Regressions of Domestic Policy Rates, Baseline Results

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 7:}

Regressions of Domestic Policy Rates, Baseline Results

	(1)	(2)	(3)	(4)	(5)
US policy rate	0.597*** (0.068)	0.359*** (0.034)	0.295*** (0.037)		0.366*** (0.066)
Ln VIX	1.396*** (0.482)	0.587** (0.283)	0.599** (0.302)	0.732** (0.304)	2.098*** (0.451)
Net outflows	0.404*** (0.118)	0.347*** (0.079)	0.337*** (0.084)	0.169 (0.123)	0.418*** (0.127)
Expected inflation, next 12 months		1.235*** (0.075)	1.224*** (0.077)	1.251*** (0.079)	1.227*** (0.075)
Output gap		0.250*** (0.082)	0.148** (0.066)	0.151** (0.070)	0.208*** (0.071)
Real credit growth			0.050*** (0.014)	0.047*** (0.016)	0.050*** (0.014)
Commodity terms of trade			-0.085** (0.042)	-0.113*** (0.038)	-0.102** (0.044)
US policy shock				-0.061 (0.157)
MPru					0.220*** (0.053)
US policy rate * MPru					-0.013*** (0.004)
Ln VIX * MPru					-0.116*** (0.020)
Net outflows * MPru					-0.004 (0.004)
Observations	1360	1262	1250	1250	1250
Country FE	Yes	Yes	Yes	Yes	Yes
F-statistic	73.98	56.42	49.70	26.32	21.98

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 7:}

Regressions of Domestic Policy Rates, Baseline Results

	(1)	(2)	(3)	(4)	(5)
US policy rate	0.597*** (0.068)	0.359*** (0.034)	0.295*** (0.037)		0.366*** (0.066)
Ln VIX	1.396*** (0.482)	0.587** (0.283)	0.599** (0.302)	0.732** (0.304)	2.098*** (0.451)
Net outflows	0.404*** (0.118)	0.347*** (0.079)	0.337*** (0.084)	0.169 (0.123)	0.418*** (0.127)
Expected inflation, next 12 months		1.235*** (0.075)	1.224*** (0.077)	1.251*** (0.079)	1.227*** (0.075)
Output gap		0.250*** (0.082)	0.148** (0.066)	0.151** (0.070)	0.208*** (0.071)
Real credit growth			0.050*** (0.014)	0.047*** (0.016)	0.050*** (0.014)
Commodity terms of trade			-0.085** (0.042)	-0.113*** (0.038)	-0.102** (0.044)
US policy shock				-0.061 (0.157)
MPru					0.220*** (0.053)
US policy rate * MPru					-0.013*** (0.004)
Ln VIX * MPru					-0.116*** (0.020)
Net outflows * MPru					-0.004 (0.004)
Observations	1360	1262	1250	1250	1250
Country FE	Yes	Yes	Yes	Yes	Yes
F-statistic	73.98	56.42	49.70	26.32	21.98

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

The regression specification also includes control variables from an augmented Taylor rule, such as expected inflation over the next 12 months, the output gap, real credit growth, and commodity terms of trade. The inclusion of expected inflation is important to show that monetary policy is not simply responding to the effects of global financial shocks on domestic inflation. We only consider periods in which countries had a flexible exchange rate and thus retain some degree of monetary autonomy²⁴

3.1 Baseline results

Table 7 reports the baseline results. We start in column (1) by including only global financial shocks. The results show that policy rates in emerging markets tend to respond pro-cyclically to a tightening in global financial conditions. Specifically, policy rates increase in response to rises in the US policy rate and the VIX, possibly to contain the exchange rate depreciation. Policy rates also increase in response to net capital outflow shocks, likely to attract more foreign investment. These results are broadly consistent with those in the literature (Obstfeld et al., 2005; Aizenman et al., 2016; Aizenman et al., 2017; Han and Wei, 2018; Cavallino and Sandri, 2020; Bhattarai et al., 2020).

Column (2) shows that the monetary policy response remains procyclical when controlling for expected inflation and the output gap. The output gap and expected inflation have the expected signs, with central banks reacting to a higher output gap or expected inflation by tightening monetary policy. These results are robust to controlling for real credit growth and commodity terms of trade in column (3). For all the specifications, the F-tests point to instrument validity. Column (4) shows that there is no correlation between emerging markets’ policy rates and US monetary policy shocks. This is consistent with an uncovered interest parity logic: if emerging markets aim to stabilize exchange rates and capital flows, they should adjust monetary policy in line with actual US policy rates rather than with the unexpected components.

Column (5) reports the full regression specification, including macroprudential regulation and its interactions with global financial shocks. The estimates show that the interactions with the US policy rate and the VIX are negative and statistically significant. As illustrated in Figure 6, this implies that macroprudential regulation enables monetary policy to respond more countercyclically to global financial shocks. Panel 1 and 2 show that at low levels of macroprudential regulation, emerging markets tighten monetary policy in response to a hike in US monetary policy or an increase in the VIX. A more stringent level of macroprudential regulation dampens this procyclical response. In fact, a sufficiently high level of macroprudential regulation allows central banks in emerging markets to react countercyclically by lowering policy rates when US policy rates and the VIX increase.

Macroprudential regulation can thus support a more countercyclical monetary policy response to US monetary policy and the VIX. However, the results in column (5) of Table 7 also show that macroprudential regulation does not have tangible effects on the monetary policy response to capital outflow shocks. Indeed, capital outflows appear to trigger a monetary tightening in emerging markets independently of the level of macroprudential regulation. This suggests that even in countries with tight macroprudential regulation, central banks continue to face important policy trade-offs in responding to fluctuations in capital flows.

3.2 Robustness

In this section we check if the results on the interplay between macroprudential regulation and monetary policy are robust to a broad set of endogeneity tests. There are two primary concerns. First, the results could be affected by reverse causality if countries move macroprudential regulation in response to policy rate changes. For example, this could happen if countries raise macroprudential regulation when policy rates decline to mitigate financial stability concerns from monetary easing. Second, there could be an omitted variable bias if macroprudential regulation correlates with country characteristics and policy tools that affect monetary policy and are not included in the regression.

Table 8 reports the robustness tests to address the reverse causality concern. In columns (1) and (2), we lag the level of macroprudential regulation by one quarter and one year, respectively. In column (3), we use the average level of macroprudential for each country between 2000–2016. In this case, the estimation relies solely on the cross-country variation in regulation, investigating if monetary policy in countries with tighter regulation respond more countercyclically.

^{Table 8:}

Regressions of Domestic Policy Rates: Robustness to Reverse Causality

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 8:}

Regressions of Domestic Policy Rates: Robustness to Reverse Causality

	MPru = one-quarter lag of MPru	MPru = one-year lag of MPru	MPru = country average of MPru
	(1)	(2)	(3)
US policy rate	0.366*** (0.065)	0.409*** (0.067)	0.351*** (0.063)
Ln VIX	2.120*** (0.452)	2.175*** (0.480)	2.407*** (0.402)
Net outflows	0.406*** (0.126)	0.391*** (0.130)	0.566*** (0.182)
Expected inflation, next 12 months	1.207*** (0.075)	1.186*** (0.080)	1.238*** (0.079)
Output gap	0.204*** (0.071)	0.160** (0.071)	0.157** (0.069)
Real credit growth	0.052*** (0.014)	0.052*** (0.015)	0.052*** (0.015)
Commodity terms of trade	-0.102** (0.045)	-0.091** (0.046)	-0.084** (0.042)
MPru	0.217*** (0.055)	0.242*** (0.061)
US policy rate * MPru	-0.014*** (0.004)	-0.018*** (0.004)	-0.005 (0.003)
Ln VIX * MPru	-0.116*** (0.021)	-0.129*** (0.025)	-0.124*** (0.021)
Net outflows * MPru	-0.003 (0.004)	-0.001 (0.004)	-0.014** (0.007)
Observations	1241	1211	1250
Countries	25	25	25
F -statistic	22.97	22.15	20.53

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 8:}

Regressions of Domestic Policy Rates: Robustness to Reverse Causality

	MPru = one-quarter lag of MPru	MPru = one-year lag of MPru	MPru = country average of MPru
	(1)	(2)	(3)
US policy rate	0.366*** (0.065)	0.409*** (0.067)	0.351*** (0.063)
Ln VIX	2.120*** (0.452)	2.175*** (0.480)	2.407*** (0.402)
Net outflows	0.406*** (0.126)	0.391*** (0.130)	0.566*** (0.182)
Expected inflation, next 12 months	1.207*** (0.075)	1.186*** (0.080)	1.238*** (0.079)
Output gap	0.204*** (0.071)	0.160** (0.071)	0.157** (0.069)
Real credit growth	0.052*** (0.014)	0.052*** (0.015)	0.052*** (0.015)
Commodity terms of trade	-0.102** (0.045)	-0.091** (0.046)	-0.084** (0.042)
MPru	0.217*** (0.055)	0.242*** (0.061)
US policy rate * MPru	-0.014*** (0.004)	-0.018*** (0.004)	-0.005 (0.003)
Ln VIX * MPru	-0.116*** (0.021)	-0.129*** (0.025)	-0.124*** (0.021)
Net outflows * MPru	-0.003 (0.004)	-0.001 (0.004)	-0.014** (0.007)
Observations	1241	1211	1250
Countries	25	25	25
F -statistic	22.97	22.15	20.53

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

Across all these specifications, macroprudential regulation continues to support a more countercyclical response of monetary policy to global financial conditions. The only difference from the baseline results emerges when using the average levels of macroprudential regulation. In this case, regulation supports a more countercyclical response to capital flow shocks rather than to changes in US monetary policy.

Table 9 reports the robustness tests to alleviate concerns about omitted variable bias. In columns (1) to (8), we augment the baseline specification with additional structural and policy variables both in levels and interacted with macroprudential regulation. These include institutional quality (column 1), financial development (column 2), gross public debt (column 3), public debt in foreign currency (column 4), the cyclical adjusted balance (column 5), the anchoring of inflation expectations (column 6), capital controls (column 7), and the level of official reserves in percent of trend GDP (column 8). As in the baseline regression, the coefficients on the interaction of macroprudential regulation and the VIX remain negative and significant in all tests. The interaction coefficients with the US policy rate are also negative and significant, except for the specification augmented with official reserves. In that case, the interaction remains negative but loses statistical significance at the 10 percent level. Finally, column (9) shows that the interaction coefficients with the US policy rates and the VIX remain negative and significant even if we include time fixed effects.

^{Table 9:}

Regressions of Domestic Policy Rates: Robustness to Omitted Variables

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 9:}

Regressions of Domestic Policy Rates: Robustness to Omitted Variables

	X = institutional quality	X = financial development	X = gross public debt	X = government FX debt	X = cyclical adjusted balance	X = inflation expectation anchoring	X = capital flow measures	X = official reserves	Including time fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
US policy rate	0.347*** (0.073)	0.053 (0.136)	-0.061 (0.128)	0.583*** (0.090)	0.566*** (0.079)	0.235*** (0.070)	0.315*** (0.089)	0.324*** (0.097)	2.899*** (0.428)
Ln VIX	1.941*** (0.486)	2.757*** (0.699)	0.187 (0.963)	1.260** (0.540)	1.960*** (0.376)	1.884*** (0.401)	1.490*** (0.434)	1.637*** (0.435)	0.643** (0.312)
Net outflows	0.415*** (0.123)	0.178 (0.168)	0.347* (0.187)	0.591*** (0.142)	0.331*** (0.100)	0.146 (0.185)	0.221*** (0.081)	0.591*** (0.174)	0.076 (0.047)
Expected inflation, next 12 months	1.218*** (0.077)	1.197*** (0.084)	1.230*** (0.072)	1.209*** (0.085)	1.182*** (0.070)	1.150*** (0.097)	1.205*** (0.078)	1.143*** (0.077)	1.161*** (0.075)
Output gap	0.202*** (0.070)	0.189*** (0.070)	0.194*** (0.066)	0.212*** (0.072)	0.227*** (0.068)	0.170* (0.087)	0.175*** (0.067)	0.162*** (0.058)	0.120* (0.069)
Real credit growth	0.048*** (0.013)	0.047*** (0.012)	0.047*** (0.013)	0.035*** (0.011)	0.049*** (0.013)	0.031 (0.020)	0.042*** (0.011)	0.050*** (0.013)	0.034*** (0.009)
Commodity terms of trade	-0.104** (0.047)	-0.075 (0.046)	-0.148*** (0.048)	-0.114** (0.058)	-0.119*** (0.045)	-0.043 (0.035)	-0.082* (0.042)	-0.081** (0.040)	-0.038 (0.038)
MPru	0.201*** (0.055)	0.228*** (0.042)	0.172*** (0.062)	0.166*** (0.050)	0.148*** (0.055)	0.160 (0.101)	0.190*** (0.046)	0.164*** (0.053)	0.084** (0.040)
US policy rate * MPru	-0.012*** (0.004)	-0.012*** (0.003)	-0.008* (0.004)	-0.017*** (0.004)	-0.021*** (0.004)	-0.008** (0.004)	-0.013*** (0.003)	-0.005 (0.004)	-0.016*** (0.002)
Ln VIX * MPru	-0.111*** (0.021)	-0.107*** (0.016)	-0.097*** (0.024)	-0.099*** (0.019)	-0.088*** (0.020)	-0.074** (0.037)	-0.104*** (0.017)	-0.091*** (0.020)	-0.064*** (0.015)
Net outflows * MPru	-0.004 (0.004) (2.117)	-0.002 (0.004) (6.952)	-0.002 (0.003) (0.040)	-0.005 (0.003) (3.402)	-0.008* (0.005) (0.289)	0.001 (0.004) (4.867)	-0.004 (0.003) (3.336)	-0.000 (0.003) (0.069)	0.001 (0.002)
US policy rate * X	0.039 (0.058)	0.670** (0.310)	0.009*** (0.002)	-0.407** (0.189)	-0.010 (0.009)	-0.101 (0.196)	0.067 (0.102)	-0.008* (0.005)
Ln VIX * X	0.568 (0.441)	-1.851 (1.609)	0.037** (0.016)	2.290** (1.126)	0.068 (0.087)	-1.066 (1.428)	0.478 (0.966)	0.009 (0.021)
Net outflows * X	-0.053 (0.178)	0.362 (0.392)	0.001 (0.003)	-0.525** (0.235)	-0.045** (0.020)	0.167 (0.190)	0.329** (0.137)	-0.016*** (0.005)
X	0.942	-0.803	-0.144***	-6.287*	-0.716**	2.055	-0.121	-0.127*
Observations	1250	1250	1239	1157	1212	976	1250	1236	1250
Countries	25	25	25	24	24	18	25	25	25
F-statistic	3.952	6.900	11.87	12.45	5.799	2.749	20.18	22.77	15.46

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 9:}

Regressions of Domestic Policy Rates: Robustness to Omitted Variables

	X = institutional quality	X = financial development	X = gross public debt	X = government FX debt	X = cyclical adjusted balance	X = inflation expectation anchoring	X = capital flow measures	X = official reserves	Including time fixed effects
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
US policy rate	0.347*** (0.073)	0.053 (0.136)	-0.061 (0.128)	0.583*** (0.090)	0.566*** (0.079)	0.235*** (0.070)	0.315*** (0.089)	0.324*** (0.097)	2.899*** (0.428)
Ln VIX	1.941*** (0.486)	2.757*** (0.699)	0.187 (0.963)	1.260** (0.540)	1.960*** (0.376)	1.884*** (0.401)	1.490*** (0.434)	1.637*** (0.435)	0.643** (0.312)
Net outflows	0.415*** (0.123)	0.178 (0.168)	0.347* (0.187)	0.591*** (0.142)	0.331*** (0.100)	0.146 (0.185)	0.221*** (0.081)	0.591*** (0.174)	0.076 (0.047)
Expected inflation, next 12 months	1.218*** (0.077)	1.197*** (0.084)	1.230*** (0.072)	1.209*** (0.085)	1.182*** (0.070)	1.150*** (0.097)	1.205*** (0.078)	1.143*** (0.077)	1.161*** (0.075)
Output gap	0.202*** (0.070)	0.189*** (0.070)	0.194*** (0.066)	0.212*** (0.072)	0.227*** (0.068)	0.170* (0.087)	0.175*** (0.067)	0.162*** (0.058)	0.120* (0.069)
Real credit growth	0.048*** (0.013)	0.047*** (0.012)	0.047*** (0.013)	0.035*** (0.011)	0.049*** (0.013)	0.031 (0.020)	0.042*** (0.011)	0.050*** (0.013)	0.034*** (0.009)
Commodity terms of trade	-0.104** (0.047)	-0.075 (0.046)	-0.148*** (0.048)	-0.114** (0.058)	-0.119*** (0.045)	-0.043 (0.035)	-0.082* (0.042)	-0.081** (0.040)	-0.038 (0.038)
MPru	0.201*** (0.055)	0.228*** (0.042)	0.172*** (0.062)	0.166*** (0.050)	0.148*** (0.055)	0.160 (0.101)	0.190*** (0.046)	0.164*** (0.053)	0.084** (0.040)
US policy rate * MPru	-0.012*** (0.004)	-0.012*** (0.003)	-0.008* (0.004)	-0.017*** (0.004)	-0.021*** (0.004)	-0.008** (0.004)	-0.013*** (0.003)	-0.005 (0.004)	-0.016*** (0.002)
Ln VIX * MPru	-0.111*** (0.021)	-0.107*** (0.016)	-0.097*** (0.024)	-0.099*** (0.019)	-0.088*** (0.020)	-0.074** (0.037)	-0.104*** (0.017)	-0.091*** (0.020)	-0.064*** (0.015)
Net outflows * MPru	-0.004 (0.004) (2.117)	-0.002 (0.004) (6.952)	-0.002 (0.003) (0.040)	-0.005 (0.003) (3.402)	-0.008* (0.005) (0.289)	0.001 (0.004) (4.867)	-0.004 (0.003) (3.336)	-0.000 (0.003) (0.069)	0.001 (0.002)
US policy rate * X	0.039 (0.058)	0.670** (0.310)	0.009*** (0.002)	-0.407** (0.189)	-0.010 (0.009)	-0.101 (0.196)	0.067 (0.102)	-0.008* (0.005)
Ln VIX * X	0.568 (0.441)	-1.851 (1.609)	0.037** (0.016)	2.290** (1.126)	0.068 (0.087)	-1.066 (1.428)	0.478 (0.966)	0.009 (0.021)
Net outflows * X	-0.053 (0.178)	0.362 (0.392)	0.001 (0.003)	-0.525** (0.235)	-0.045** (0.020)	0.167 (0.190)	0.329** (0.137)	-0.016*** (0.005)
X	0.942	-0.803	-0.144***	-6.287*	-0.716**	2.055	-0.121	-0.127*
Observations	1250	1250	1239	1157	1212	976	1250	1236	1250
Countries	25	25	25	24	24	18	25	25	25
F-statistic	3.952	6.900	11.87	12.45	5.799	2.749	20.18	22.77	15.46

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

4 How does macroprudential regulation compare with capital controls?

The previous analysis has shown that macroprudential regulation can significantly dampen the impact of global financial shocks on economic activity in emerging markers (Section 2). Furthermore, it allows monetary policy to respond more countercyclically to global financial shocks (Section 3). To put these benefits in perspective, in this section we ask if capital controls can provide similar gains.

The empirical and theoretical literature has placed great emphasis on the role of capital controls in dealing with global financial conditions.²⁵ Since these tools directly reduce capital flows, they are expected to limit the sensitivity of emerging markets to global financial shocks. Nonetheless, there are also concerns about their effectiveness. For example, capital controls may “leak”, as investors try to circumvent restriction by channeling funds through alternative ways that can prove more volatile. Furthermore, countries that maintain capital controls may signal their willingness to impose new restrictions during a crisis. This could unnerve foreign investors and lead to higher volatility.

To assess the possible benefits from capital controls, we replicate the analysis in the previous sections using four different indicators of capital controls from Chinn and Ito (2008), Fernández et al. (2015), Quinn and Toyoda (2008), and Pasricha et al. (2018).²⁶ The goal is to examine if, for at least one of these indicators, there is systemic evidence that more stringent capital control restrictions provide similar or greater benefits to those from a tighter level of macroprudential regulation.²⁷

We start by analyzing if capital controls can dampen the effects of global financial shocks on GDP in emerging markets. To do so, we re-estimate equation (1) using the indexes of capital controls instead of macroprudential regulation. Table 10 reports the results, using specifications with and without quadratic terms for capital controls. We denote the stringency of capital controls with CC.

^{Table 10:}

Dampening Effects of Capital Controls on GDP Growth

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

^{Table 10:}

Dampening Effects of Capital Controls on GDP Growth

	Chinn and Ito (2008)		Fernández et al. (2015)		Quinn and Toyoda (2008)		Pasricha et al. (2018)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Lag dependent variable	0.072 (0.053)	0.072 (0.054)	0.126** (0.052)	0.134*** (0.051)	0.178*** (0.057)	0.194*** (0.054)	0.113 (0.088)	0.109 (0.086)
Lag output gap	-0.365*** (0.034)	-0.362*** (0.034)	-0.327*** (0.028)	-0.337*** (0.032)	-0.304*** (0.025)	-0.312*** (0.025)	-0.300*** (0.031)	-0.302*** (0.031)
Lag ln real GDP per capita	-0.430 (0.490)	-0.428 (0.482)	-0.721 (0.509)	-0.544 (0.509)	-0.519 (0.559)	-0.113 (0.485)	0.211 (0.584)	0.175 (0.767)
Institutional quality	-0.369 (0.666)	-0.474 (0.700)	-0.323 (0.727)	-0.127 (0.803)	0.382 (0.468)	-0.064 (0.483)	-0.313 (1.267)	-0.131 (1.262)
Linear trend	-0.002 (0.005)	-0.002 (0.005)	0.000 (0.005)	0.001 (0.005)	-0.002 (0.004)	-0.008* (0.004)	-0.026** (0.011)	-0.027** (0.012)
Commodity terms of trade	0.045 (0.036)	0.040 (0.037)	0.037 (0.038)	0.041 (0.047)	-0.063 (0.055)	-0.053 (0.054)	-0.121 (0.124)	-0.111 (0.119)
US monetary policy shock	-0.185 (0.189)	-0.235 (0.144)	-0.147 (0.205)	-0.206 (0.212)	-0.106 (0.196)	-0.168 (0.190)	-0.025 (0.212)	0.033 (0.205)
Ln VIX	-0.789*** (0.247)	-0.760*** (0.226)	-1.033*** (0.212)	-0.882*** (0.311)	-0.881*** (0.238)	-0.870*** (0.210)	-0.720*** (0.246)	-0.596** (0.272)
Net outflows	-0.104*** (0.035)	-0.087*** (0.030)	-0.132*** (0.036)	-0.093*** (0.028)	-0.077** (0.037)	-0.101** (0.040)	-0.364*** (0.119)	-0.371*** (0.117)
CC	-2.190 (1.721)	-2.904 (4.137)	-3.201* (1.712)	-9.395 (7.393)	-2.913 (2.315)	2.233 (3.683)	0.039 (0.105)	0.281 (0.198)
US monetary policy shock * CC	0.031 (0.149)	0.520 (0.735)	-0.081 (0.165)	0.363 (0.660)	-0.275 (0.244)	0.492 (0.467)	-0.012 (0.008)	-0.035** (0.018)
Ln VIX * CC	0.232 (0.540)	0.204 (1.127)	0.844* (0.445)	1.301 (1.833)	0.982 (0.703)	-0.942 (1.217)	0.007 (0.035)	-0.070 (0.063)
Net outflows * CC	-0.221** (0.094)	-0.427 (0.309)	-0.127 (0.093)	-0.792 (0.547)	-0.300* (0.153)	0.285 (0.268)	0.012* (0.006)	0.016 (0.014)
CC²		0.889 (4.602)		7.075 (7.223)		-8.961 (7.902)		-0.010 (0.007)
US monetary policy shock * CC²		-0.546 (0.852)		-0.490 (0.695)		-1.194 (0.785)		0.001 (0.000)
Ln VIX * CC²		-0.011 (1.312)		-0.766 (1.832)		3.539 (2.619)		0.003 (0.002)
Net outflows * CC²		0.253 (0.407)		0.790 (0.614)		-0.957 (0.644)		-0.000 (0.000)
Observations	2,220	2,220	1,925	1,925	1,856	1,856	918	918
Countries	37	37	31	31	30	30	16	16
F -statistic	32.6	10.6	22.7	10.5	13.4	3.2	12.6	2.1

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

^{Table 10:}

Dampening Effects of Capital Controls on GDP Growth

	Chinn and Ito (2008)		Fernández et al. (2015)		Quinn and Toyoda (2008)		Pasricha et al. (2018)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Lag dependent variable	0.072 (0.053)	0.072 (0.054)	0.126** (0.052)	0.134*** (0.051)	0.178*** (0.057)	0.194*** (0.054)	0.113 (0.088)	0.109 (0.086)
Lag output gap	-0.365*** (0.034)	-0.362*** (0.034)	-0.327*** (0.028)	-0.337*** (0.032)	-0.304*** (0.025)	-0.312*** (0.025)	-0.300*** (0.031)	-0.302*** (0.031)
Lag ln real GDP per capita	-0.430 (0.490)	-0.428 (0.482)	-0.721 (0.509)	-0.544 (0.509)	-0.519 (0.559)	-0.113 (0.485)	0.211 (0.584)	0.175 (0.767)
Institutional quality	-0.369 (0.666)	-0.474 (0.700)	-0.323 (0.727)	-0.127 (0.803)	0.382 (0.468)	-0.064 (0.483)	-0.313 (1.267)	-0.131 (1.262)
Linear trend	-0.002 (0.005)	-0.002 (0.005)	0.000 (0.005)	0.001 (0.005)	-0.002 (0.004)	-0.008* (0.004)	-0.026** (0.011)	-0.027** (0.012)
Commodity terms of trade	0.045 (0.036)	0.040 (0.037)	0.037 (0.038)	0.041 (0.047)	-0.063 (0.055)	-0.053 (0.054)	-0.121 (0.124)	-0.111 (0.119)
US monetary policy shock	-0.185 (0.189)	-0.235 (0.144)	-0.147 (0.205)	-0.206 (0.212)	-0.106 (0.196)	-0.168 (0.190)	-0.025 (0.212)	0.033 (0.205)
Ln VIX	-0.789*** (0.247)	-0.760*** (0.226)	-1.033*** (0.212)	-0.882*** (0.311)	-0.881*** (0.238)	-0.870*** (0.210)	-0.720*** (0.246)	-0.596** (0.272)
Net outflows	-0.104*** (0.035)	-0.087*** (0.030)	-0.132*** (0.036)	-0.093*** (0.028)	-0.077** (0.037)	-0.101** (0.040)	-0.364*** (0.119)	-0.371*** (0.117)
CC	-2.190 (1.721)	-2.904 (4.137)	-3.201* (1.712)	-9.395 (7.393)	-2.913 (2.315)	2.233 (3.683)	0.039 (0.105)	0.281 (0.198)
US monetary policy shock * CC	0.031 (0.149)	0.520 (0.735)	-0.081 (0.165)	0.363 (0.660)	-0.275 (0.244)	0.492 (0.467)	-0.012 (0.008)	-0.035** (0.018)
Ln VIX * CC	0.232 (0.540)	0.204 (1.127)	0.844* (0.445)	1.301 (1.833)	0.982 (0.703)	-0.942 (1.217)	0.007 (0.035)	-0.070 (0.063)
Net outflows * CC	-0.221** (0.094)	-0.427 (0.309)	-0.127 (0.093)	-0.792 (0.547)	-0.300* (0.153)	0.285 (0.268)	0.012* (0.006)	0.016 (0.014)
CC²		0.889 (4.602)		7.075 (7.223)		-8.961 (7.902)		-0.010 (0.007)
US monetary policy shock * CC²		-0.546 (0.852)		-0.490 (0.695)		-1.194 (0.785)		0.001 (0.000)
Ln VIX * CC²		-0.011 (1.312)		-0.766 (1.832)		3.539 (2.619)		0.003 (0.002)
Net outflows * CC²		0.253 (0.407)		0.790 (0.614)		-0.957 (0.644)		-0.000 (0.000)
Observations	2,220	2,220	1,925	1,925	1,856	1,856	918	918
Countries	37	37	31	31	30	30	16	16
F -statistic	32.6	10.6	22.7	10.5	13.4	3.2	12.6	2.1

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

Across all specifications from column (1) to (8), we confirm that an increase in the VIX or a capital outflow have negative effects on GDP growth in emerging markets. A more stringent level of capital controls would dampen these effects if the coefficient estimates on the interaction terms between capital controls and the shocks are positive. We do not find any systematic evidence that this is the case. Most coefficients are insignificant. Some are even negative and statistically significant, thus suggesting the capital controls may actually exacerbate the transmission of global financial shocks. The interaction coefficients are positive and weakly statistically significant only in column (3) against the VIX and in column (7) against net capital outflows. Overall, the results suggest that the stringency of capital controls does not affect the impact of global financial shocks on economic activity in emerging markets. This is in sharp contrast with the results on macroprudential regulation.²⁸

We also explore if capital controls have clearer dampening effects against global financial shocks when we distinguish between controls on capital inflows and controls on capital outflows. This distinction is possible only using the indexes provided by Fernández et al. (2015) and Pasricha et al. (2018). The results are reported in Table 11 and, again, do not show any systemic pattern. Most coefficients on the interaction terms between capital controls and the shocks are insignificant. There are three cases where the coefficients are positive and significant, but there is an equal number of negative coefficients.²⁹

^{Table 11:}

Dampening Effects of Inflow and Outflow Capital Controls on GDP Growth

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

^{Table 11:}

Dampening Effects of Inflow and Outflow Capital Controls on GDP Growth

	Fernández et al. (2015)				Pasricha et al. (2018)
	CC on inflows		CC on outflows		CC on inflows		CC on outflows
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Lag dependent variable	0.124** (0.053)	0.125** (0.051)	0.127** (0.051)	0.125** (0.055)	0.112 (0.091)	0.105 (0.093)	0.092 (0.096)	0.095 (0.093)
Lag output gap	-0.326*** (0.028)	-0.327*** (0.029)	-0.329*** (0.028)	-0.327*** (0.032)	-0.299*** (0.034)	-0.299*** (0.032)	-0.288*** (0.039)	-0.300*** (0.037)
Lag ln real GDP per capita	-0.713 (0.503)	-0.685 (0.525)	-0.691 (0.510)	-0.587 (0.520)	0.254 (0.636)	0.344 (0.664)	-0.115 (0.872)	0.644 (0.812)
Institutional quality	-0.221 (0.739)	-0.385 (0.743)	-0.393 (0.727)	0.513 (0.869)	-0.351 (1.306)	-0.543 (1.333)	0.257 (1.363)	0.033 (1.501)
Linear trend	0.001 (0.005)	0.001 (0.005)	0.000 (0.005)	0.003 (0.006)	-0.022** (0.010)	-0.027** (0.012)	-0.033** (0.013)	-0.034** (0.015)
Commodity terms of trade	0.034 (0.038)	0.033 (0.041)	0.039 (0.038)	0.071 (0.058)	-0.151 (0.128)	-0.109 (0.129)	-0.146 (0.132)	-0.155 (0.135)
US monetary policy shock	-0.178 (0.204)	-0.152 (0.181)	-0.125 (0.202)	-0.244 (0.220)	-0.050 (0.213)	-0.003 (0.217)	0.049 (0.244)	0.129 (0.253)
Ln VIX	-0.992*** (0.185)	-0.895*** (0.287)	-1.039*** (0.225)	-0.872*** (0.278)	-0.705*** (0.237)	-0.641** (0.251)	-0.344 (0.370)	-0.368 (0.374)
Net outflows	-0.127*** (0.039)	-0.124*** (0.037)	-0.140*** (0.035)	-0.103*** (0.028)	-0.344*** (0.121)	-0.367*** (0.130)	-0.537*** (0.187)	-0.513*** (0.189)
CC	-3.267* (1.686)	-2.548 (6.667)	-2.775* (1.548)	-15.964* (8.493)	0.100 (0.139)	-0.379 (0.315)	-0.703* (0.391)	-0.633 (0.591)
US monetary policy shock * CC	-0.030 (0.154)	-0.359 (0.717)	-0.105 (0.160)	0.774 (0.706)	0.018 (0.012)	0.082*** (0.027)	0.043 (0.036)	0.076 (0.058)
Ln VIX * CC	0.852* (0.435)	0.110 (1.911)	0.762* (0.419)	3.029 (1.923)	-0.064 (0.050)	0.042 (0.102)	0.157 (0.106)	0.230 (0.183)
Net outflows * CC	-0.152 (0.101)	-0.123 (0.473)	-0.104 (0.084)	-1.194** (0.594)	-0.012 (0.008)	-0.034** (0.016)	-0.078** (0.038)	-0.050 (0.050)
CC²		-0.997 (6.966)		14.307* (8.099)		-0.009 (0.016)		-0.037 (0.059)
US monetary policy shock * CC²		0.421 (0.791)		-0.946 (0.764)		0.003*** (0.001)		0.002 (0.006)
Ln VIX * CC²		0 828 (2.035)		-2.841 (1.979)		0.001 (0.006)		0.020 (0.018)
Net outflows * CC²		-0.058 (0.526)		-1.236* (0.631)		-0.000 (0.001)		0.001 (0.005)
Observations	1,925	1,925	1,925	1,925	918	918	918	918
Countries	31	31	31	31	16	16	16	16
F -statistic	30.2	18.5	19.3	6.5	9.1	6.2	3.7	2.5

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

^{Table 11:}

Dampening Effects of Inflow and Outflow Capital Controls on GDP Growth

	Fernández et al. (2015)				Pasricha et al. (2018)
	CC on inflows		CC on outflows		CC on inflows		CC on outflows
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Lag dependent variable	0.124** (0.053)	0.125** (0.051)	0.127** (0.051)	0.125** (0.055)	0.112 (0.091)	0.105 (0.093)	0.092 (0.096)	0.095 (0.093)
Lag output gap	-0.326*** (0.028)	-0.327*** (0.029)	-0.329*** (0.028)	-0.327*** (0.032)	-0.299*** (0.034)	-0.299*** (0.032)	-0.288*** (0.039)	-0.300*** (0.037)
Lag ln real GDP per capita	-0.713 (0.503)	-0.685 (0.525)	-0.691 (0.510)	-0.587 (0.520)	0.254 (0.636)	0.344 (0.664)	-0.115 (0.872)	0.644 (0.812)
Institutional quality	-0.221 (0.739)	-0.385 (0.743)	-0.393 (0.727)	0.513 (0.869)	-0.351 (1.306)	-0.543 (1.333)	0.257 (1.363)	0.033 (1.501)
Linear trend	0.001 (0.005)	0.001 (0.005)	0.000 (0.005)	0.003 (0.006)	-0.022** (0.010)	-0.027** (0.012)	-0.033** (0.013)	-0.034** (0.015)
Commodity terms of trade	0.034 (0.038)	0.033 (0.041)	0.039 (0.038)	0.071 (0.058)	-0.151 (0.128)	-0.109 (0.129)	-0.146 (0.132)	-0.155 (0.135)
US monetary policy shock	-0.178 (0.204)	-0.152 (0.181)	-0.125 (0.202)	-0.244 (0.220)	-0.050 (0.213)	-0.003 (0.217)	0.049 (0.244)	0.129 (0.253)
Ln VIX	-0.992*** (0.185)	-0.895*** (0.287)	-1.039*** (0.225)	-0.872*** (0.278)	-0.705*** (0.237)	-0.641** (0.251)	-0.344 (0.370)	-0.368 (0.374)
Net outflows	-0.127*** (0.039)	-0.124*** (0.037)	-0.140*** (0.035)	-0.103*** (0.028)	-0.344*** (0.121)	-0.367*** (0.130)	-0.537*** (0.187)	-0.513*** (0.189)
CC	-3.267* (1.686)	-2.548 (6.667)	-2.775* (1.548)	-15.964* (8.493)	0.100 (0.139)	-0.379 (0.315)	-0.703* (0.391)	-0.633 (0.591)
US monetary policy shock * CC	-0.030 (0.154)	-0.359 (0.717)	-0.105 (0.160)	0.774 (0.706)	0.018 (0.012)	0.082*** (0.027)	0.043 (0.036)	0.076 (0.058)
Ln VIX * CC	0.852* (0.435)	0.110 (1.911)	0.762* (0.419)	3.029 (1.923)	-0.064 (0.050)	0.042 (0.102)	0.157 (0.106)	0.230 (0.183)
Net outflows * CC	-0.152 (0.101)	-0.123 (0.473)	-0.104 (0.084)	-1.194** (0.594)	-0.012 (0.008)	-0.034** (0.016)	-0.078** (0.038)	-0.050 (0.050)
CC²		-0.997 (6.966)		14.307* (8.099)		-0.009 (0.016)		-0.037 (0.059)
US monetary policy shock * CC²		0.421 (0.791)		-0.946 (0.764)		0.003*** (0.001)		0.002 (0.006)
Ln VIX * CC²		0 828 (2.035)		-2.841 (1.979)		0.001 (0.006)		0.020 (0.018)
Net outflows * CC²		-0.058 (0.526)		-1.236* (0.631)		-0.000 (0.001)		0.001 (0.005)
Observations	1,925	1,925	1,925	1,925	918	918	918	918
Countries	31	31	31	31	16	16	16	16
F -statistic	30.2	18.5	19.3	6.5	9.1	6.2	3.7	2.5

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The estimations are based on a sample of EM from 2000Q1 to 2016Q4. All specifications include country fixed effects. Driscoll-Kraay standard errors are reported in parentheses. ***, **, and * indicate statistical significance at 1, 5, and 10 percent, respectively.

We now turn to the question of whether capital controls can enable countries to conduct a more countercyclical monetary policy in response to global financial shocks. We re-estimate equation (5) by replacing the index of macroprudential regulation with capital control measures. If stricter capital controls allow for a more countercyclical monetary policy response, the coefficients on the interactions between capital controls and the shocks should be negative and statistically significant.

Table 12 reports the regression results. We do not find evidence that more stringent capital controls support a more countercyclical monetary policy response. Most coefficients on the interaction terms in columns (1) to (4) are insignificant. Only the interaction with the US policy rate in column (4) is negative and weakly significant. There are also several instances in columns (1) and (2) where the coefficients are positive and strongly significant, thus suggesting that tighter capital controls are associated with a more procyclical response of monetary policy. Table 13 reports similar findings when we differentiate capital controls on inflows and outflows.

^{Table 12:}

Regressions of Domestic Policy Rates: Capital Controls

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 12:}

Regressions of Domestic Policy Rates: Capital Controls

	Chinn and Ito (2008)	Fernández et al. (2015)	Quinn and Toyoda (2008)	Pasricha et al. (2018)
	(1)	(2)	(3)	(4)
US policy rate	0.153*** (0.050)	0.110* (0.057)	0.258*** (0.051)	0.336*** (0.049)
Ln VIX	0.798*** (0.266)	-0.127 (0.381)	0.212 (0.363)	0.718* (0.370)
Net outflows	0.056 (0.072)	0.094* (0.055)	0.185** (0.075)	0.249** (0.109)
Expected inflation, next 12 months	1.077*** (0.082)	1.207*** (0.079)	1.207*** (0.082)	1.167*** (0.102)
Output gap	0.115** (0.057)	0.106* (0.059)	0.147** (0.069)	0.121 (0.091)
Real credit growth	0.031*** (0.011)	0.044*** (0.011)	0.040*** (0.014)	0.055*** (0.018)
Commodity terms of trade	-0.043 (0.037)	-0.062 (0.039)	-0.019 (0.043)	-0.040 (0.059)
CC	7.983*** (2.515)	-1.504 (3.278)	4.462 (5.664)	0.062 (0.146)
CC * US policy rate	0.279*** (0.095)	0.274*** (0.083)	-0.152 (0.181)	-0.019* (0.011)
CC * Ln VIX	-0.702 (0.710)	0.887 (0.974)	0.635 (1.747)	-0.042 (0.044)
CC * Net outflows	0.404*** (0.145)	0.415*** (0.136)	0.420 (0.341)	-0.063 (0.068)
Observations	1250	1250	1171	806
Countries	25	25	23	15
F-statistic	15.07	27.80	8.975	2.818

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 12:}

Regressions of Domestic Policy Rates: Capital Controls

	Chinn and Ito (2008)	Fernández et al. (2015)	Quinn and Toyoda (2008)	Pasricha et al. (2018)
	(1)	(2)	(3)	(4)
US policy rate	0.153*** (0.050)	0.110* (0.057)	0.258*** (0.051)	0.336*** (0.049)
Ln VIX	0.798*** (0.266)	-0.127 (0.381)	0.212 (0.363)	0.718* (0.370)
Net outflows	0.056 (0.072)	0.094* (0.055)	0.185** (0.075)	0.249** (0.109)
Expected inflation, next 12 months	1.077*** (0.082)	1.207*** (0.079)	1.207*** (0.082)	1.167*** (0.102)
Output gap	0.115** (0.057)	0.106* (0.059)	0.147** (0.069)	0.121 (0.091)
Real credit growth	0.031*** (0.011)	0.044*** (0.011)	0.040*** (0.014)	0.055*** (0.018)
Commodity terms of trade	-0.043 (0.037)	-0.062 (0.039)	-0.019 (0.043)	-0.040 (0.059)
CC	7.983*** (2.515)	-1.504 (3.278)	4.462 (5.664)	0.062 (0.146)
CC * US policy rate	0.279*** (0.095)	0.274*** (0.083)	-0.152 (0.181)	-0.019* (0.011)
CC * Ln VIX	-0.702 (0.710)	0.887 (0.974)	0.635 (1.747)	-0.042 (0.044)
CC * Net outflows	0.404*** (0.145)	0.415*** (0.136)	0.420 (0.341)	-0.063 (0.068)
Observations	1250	1250	1171	806
Countries	25	25	23	15
F-statistic	15.07	27.80	8.975	2.818

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 13:}

Regressions of Domestic Policy Rates: Capital Controls on Inflows and Outflows

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 13:}

Regressions of Domestic Policy Rates: Capital Controls on Inflows and Outflows

	Fernández et al. (2015)		Pasricha et al. (2018)
	CC on inflows	CC on outflows	CC on inflows	CC on outflows
	(1)	(2)	(3)	(4)
US policy rate	0.163*** (0.047)	0.101 (0.064)	0.336*** (0.044)	0.288*** (0.063)
Ln VIX	0.183 (0.309)	-0.242 (0.384)	0.788** (0.353)	0.574 (0.451)
Net outflows	0.130** (0.058)	0.095* (0.056)	0.193** (0.097)	0.323** (0.157)
Expected inflation, next 12 months	1.205*** (0.078)	1.213*** (0.079)	1.205*** (0.097)	1.174*** (0.099)
Output gap	0.109* (0.060)	0.110* (0.060)	0.108 (0.093)	0.127 (0.087)
Real credit growth	0.045*** (0.012)	0.044*** (0.011)	0.052*** (0.016)	0.063*** (0.020)
Commodity terms of trade	-0.069* (0.038)	-0.062 (0.039)	-0.029 (0.054)	-0.030 (0.052)
CC	0.024 (3.011)	-2.201 (2.912)	-0.746** (0.336)	0.359 (0.396)
CC * US policy rate	0.219*** (0.076)	0.258*** (0.085)	0.034** (0.017)	0.035*** (0.012)
CC * Ln VIX	0.421 (0.880)	0.985 (0.896)	0.248** (0.101)	-0.047 (0.110)
CC * Net outflows	0.398*** (0.147)	0.384*** (0.122)	-0.058** (0.028)	0.016 (0.030)
Observations	1250	1250	806	806
Country FE	Yes	Yes	Yes	Yes
F-statistic	26.95	30	8.712	5.566

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

^{Table 13:}

Regressions of Domestic Policy Rates: Capital Controls on Inflows and Outflows

	Fernández et al. (2015)		Pasricha et al. (2018)
	CC on inflows	CC on outflows	CC on inflows	CC on outflows
	(1)	(2)	(3)	(4)
US policy rate	0.163*** (0.047)	0.101 (0.064)	0.336*** (0.044)	0.288*** (0.063)
Ln VIX	0.183 (0.309)	-0.242 (0.384)	0.788** (0.353)	0.574 (0.451)
Net outflows	0.130** (0.058)	0.095* (0.056)	0.193** (0.097)	0.323** (0.157)
Expected inflation, next 12 months	1.205*** (0.078)	1.213*** (0.079)	1.205*** (0.097)	1.174*** (0.099)
Output gap	0.109* (0.060)	0.110* (0.060)	0.108 (0.093)	0.127 (0.087)
Real credit growth	0.045*** (0.012)	0.044*** (0.011)	0.052*** (0.016)	0.063*** (0.020)
Commodity terms of trade	-0.069* (0.038)	-0.062 (0.039)	-0.029 (0.054)	-0.030 (0.052)
CC	0.024 (3.011)	-2.201 (2.912)	-0.746** (0.336)	0.359 (0.396)
CC * US policy rate	0.219*** (0.076)	0.258*** (0.085)	0.034** (0.017)	0.035*** (0.012)
CC * Ln VIX	0.421 (0.880)	0.985 (0.896)	0.248** (0.101)	-0.047 (0.110)
CC * Net outflows	0.398*** (0.147)	0.384*** (0.122)	-0.058** (0.028)	0.016 (0.030)
Observations	1250	1250	806	806
Country FE	Yes	Yes	Yes	Yes
F-statistic	26.95	30	8.712	5.566

Source: Authors’ calculations.Notes: Net inflows (in percent of trend GDP) for each country are instrumented using gross inflows to other EMs (in percent of trend GDP). Results are presented in terms of net outflows. The US policy rate is the effective federal funds rate except during the zero lower bound period where the implied policy rate from Wu and Xia (2015) is used. The estimations are done using fixed effects on a panel of EMs, excluding countries with pegged and freely falling exchange rates (Ilzetzki et al., 2019), during 2000Q1 to 2016Q4. Driscoll-Kraay standard errors are reported in parenthesis. ***,**, and * indicate statistical significance on 1, 5, and 10, percent level, respectively.

The lack of systematic effects of capital controls on monetary policy could appear in contradiction with earlier contributions to the literature. Shambaugh (2004) finds that capital controls reduce the comovement of domestic policy rates with the interest rate in a foreign base country using data between 1973 and 2000. Our analysis considers the period post 2000 when international capital markets have become considerably more integrated, possibly allowing investors to more easily circumvent capital controls. Han and Wei (2018) also document that capital controls reduce the comovement between policy rates in emerging and advanced economies and the US policy rate. However, their results apply especially to countries with fixed exchange rates.³⁰ Similarly, Aizenman et al. (2016) find that capital controls reduce the sensitivity of policy rates in emerging markets to monetary policy in major economies, but do not allow the effect to differ across exchange rate regimes. Our regression analysis considers instead only countries that have flexible exchange rates and that should therefore already enjoy a significant degree of monetary independence. In this case, capital controls do not seem to provide additional benefits for monetary policy. Furthermore, our analysis features a broader set of global financial shocks relative to the literature, including not only US monetary policy but also movements in the VIX and shocks to capital inflows.

Overall, the results suggest that a tighter level of capital controls does not help emerging markets to dampen global financial shocks nor it supports a more countercyclical monetary policy response. Against this background, the strength of the results on the benefits of macroprudential regulation appears even more notable.

The fact that the regression analysis did not find significant effects associated capital controls should not be used to dismiss possible gains from these tools. Restrictions on capital account transactions could be helpful in case of very large global shocks. Furthermore, in line with the approach used to examine macroprudential regulation, we have only tested if a more stringent level of capital controls affects the transmission of global financial shocks and the monetary policy response. The gains from capital controls could instead materialize only by optimally adjusting capital controls in response to shocks. This is indeed the perspective developed in the recent theoretical models whereby capital controls should lean against systemic risks associated with domestic and external developments.

5 Conclusions

The key result of the paper is that macroprudential regulation can significantly dampen the macroeconomic impacts of global financial shocks on emerging markets. More specifically, a tighter level of regulation reduces the sensitivity of GDP growth in emerging markets to fluctuations in risk premia and changes in the supply of international capital flows. The dampening effects are symmetric, reducing the effects on GDP from both positive and negative global financial shocks. Therefore, while more stringent regulation leads to a lower decline in GDP when global financial conditions tighten, it also constrains economic activity when global financial conditions are buoyant.

The dampening effects of macroprudential regulation are not driven by a narrow set of instruments. A broad range of macroprudential measures targeting liquidity, capital, foreign exchange exposures, and risky forms of credit contribute to enhancing macroeconomic resilience to global financial shocks. The effects of specific measures are heterogeneous and depend on the specific type of financial shock hitting the economy.

The analysis does not find evidence of negative cross-country spillovers from macroprudential regulation. On the contrary, a higher level of macroprudential regulation in one country appears to also strengthen resilience to capital flow shocks in other countries. This is possibly because, if a country uses macroprudential measures to enhance macroeconomic stability, other countries may benefit through more stable trade and financial linkages.

One possible channel through which macroprudential regulation strengthens macroeconomic resilience is by allowing monetary policy to respond more countercyclically to global financial shocks. The empirical evidence suggests that at low levels of macroprudential regulation, central banks in emerging markets tend to increase policy rates when global financial conditions tighten. This is likely because of financial stability concerns arising from movements in exchange rates and capital outflows. However, at higher levels of macroprudential regulation, central banks tend to lower policy rates when global financial conditions tighten, thus cushioning the impact of adverse financial shocks on domestic economic growth.

The benefits of macroprudential regulation uncovered in the analysis are particularly notable since we do not find evidence that capital controls provide similar gains. Despite using various measures of capital controls, the impact of global financial shocks on emerging markets’ GDP is not systematically affected by the stringency of capital flow restrictions. Furthermore, the stringency of capital controls does not support a more countercyclical response of monetary policy to global financial conditions.

Therefore, the findings of the analysis suggests that a sound macroprudential regulatory framework may go a long way in helping emerging markets to strengthen resilience against global financial shocks. To maximize the benefits of macroprudential regulation, policymakers should consider using a broad range of measures rather than focusing on a narrow set of tools. Imposing capital controls to limit cross-border financial transactions does not appear to be a valid substitute to adopting a solid macroprudential framework.

Appendix A. Data

Table A.1 lists the data sources used in the analysis. The country coverage is reported in Table A.2. The sample consists of 38 EMs, based on the April 2020 World Economic Outlook classification, from the first quarter of 2000 to the last quarter of 2016. The criteria used for the country and period selection are: (i) a population larger than one million, (ii) at least 10 years of GDP data, (iii) at least 5 years of data for net capital inflows, (iv) data on macroprudential regulation from the iMaPP databases of Alam et al. (2019).

^{Table A.1:}

Data Sources

Source: IMF staff compilation.

Topics

Countries

Series

About

Resources

Contributor Notes

Abstract

1 Introduction

2 Dampening effects of macroprudential regulation

2.1 Baseline results

2.2 Robustness

2.3 Symmetric dampening effects

2.4 Categories of macroprudential measures

2.5 Cross-country spillovers

3 Macroprudential regulation and monetary policy

3.1 Baseline results

3.2 Robustness

4 How does macroprudential regulation compare with capital controls?

5 Conclusions

Appendix A. Data