Post-financial crisis monetary policy of emerging markets : an analysis of the first significant divergence in 40 years of emerging countries´ monetary policy when facing economic crises

P

OST

-F

INANCIAL

C

RISIS

M

ONETARY

P

OLICY OF

E

MERGING

M

ARKETS

An analysis of the first significant divergence in 40 years of emerging countries´ monetary policy when facing economic crises.

Antonio Verdú Hernández

(11832320)

University of Amsterdam

Faculty of Economics and Business Supervisor: Dhr. prof. dr. S.J.G. van Wijnbergen

MSc Economics

Statement of Originality

This document is written by Student Antonio Verdú Hernández who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

A

BSTRACT

This thesis investigates the potential influence of the Federal Reserve and European Central Bank on the type of monetary policy conducted by the emerging countries of their same continent during crises, putting special focus on the recession of 2014-2015. The study consists of the analysis, through a Probit model, of the reasons that could have motivated opposite monetary policy strategies between American and European emerging markets to face the shock of 2014. A potential explanation for this event is the different monetary policy carried out by each large central bank of the continent, ECB and FED. This would imply that emerging economies, at least in the last recession, have been more dependent on the larger countries´ influence than on their own macroeconomic fundamentals to design their monetary policies.

Key words: Countercyclical and procyclical monetary policy, emerging economies, reference rate.

T

ABLE OF CONTENTS

1. I

... 5

2. L

... 7

3. F

... 10

4. F

,

2014 ... 12

5. T

M

... 18

5.1 S

... 18

5.2 T

... 20

5.3 P

... 21

5.3.1 A

EME

... 21

5.3.2 A

:

1983-2001

2002-2016 ... 24

5.3.3 A

:

E

A

... 27

6. C

... 30

7. R

... 32

5

1.

I

NTRODUCTION

Emerging market economies (EMEs) have been an indispensable component of global growth since 1980. These economies have been characterized by elevated levels of growth and volatility of their GDP in contrast with advanced economies. Since 1990, EMEs have been propelling the global growth up to the point that if a crisis would hit them severely, worldwide GDP growth could fall to levels close to zero. According to the agency Reuters, most of this growth in 2018 and following years will be driven by emerging markets. In fact, 33 percent of the real GDP global growth in 2018 is expected to be performed by China, and 45 percent by other EMEs.

Undoubtedly, emerging markets play a key role for the world economy, and the way they face and recover from crises should be in the common interest for the rest of countries. From a monetary point of view, EMEs in general have been combating downturns very differently over time, without any defined type of monetary policy. In the financial crisis of 2007-2008 EMEs loosened their reference rate of monetary policy to levels historically low to cushion that shock. This countercyclical response supposed an important departure from previous crises, in which most of the EMEs tightened policy rates to avoid large capital flights and currency depreciations.

During the last recessive episode, 2014-2015, the reaction regarding monetary policy of all the EMEs in general did not continue performing countercyclical policies, as they did in the financial crisis. This response was very diverse and without a clear trend for the different EMEs all over the world. In contrast, if this set of countries are separated by continent, it is possible to find a correlation in monetary policy for European and American EMEs. The majority of the emerging countries in Europe conducted countercyclical policies for the period 2014-2015, while the American ones mostly did exactly the opposite, increase rates.

EMEs in general had been carrying out procyclical monetary policies until 2007-2008, when they implemented countercyclical ones. This suggests that this kind of economies have been conducting a similar monetary policy strategy since 1970, firstly procyclical and already in 2008 countercyclical. Nonetheless, in 2014, it was the first time in four decades in which their reaction regarding policy rates in crises differed.

6

According to Brahima Coulibaly (2012), the type of monetary policy implemented by every emerging country depends just on its macroeconomic fundamentals and institutions, so if an EME possesses fundamentals and financial institutions of a certain standard, the EME will be able to conduct countercyclical policies, like advanced countries.

The aim of this study is to determine if this divergence in monetary policy for the period 2014-2015 among the American and European EMEs has been motivated by their own fundamentals or if there has been another external factor affecting this decision.

The external factor considered for the study will be the monetary policy conducted by the large economy´s central bank of every continent, FED in the case of America and ECB for Europe. Therefore, trying address if the shift back to procyclical policies of some EMEs have been a result of the erosion of certain macro fundamentals due to the financial crisis or, to some extent, due to the potential influence of large central banks on the EMEs´ monetary policy. If so, is this a new phenomenom or large central banks have always influenced EMEs´ policy rate decision?

The study will be carried out by analyzing the behavior regarding policy rates of 31 EMEs during crises. For the model, after having excluded African and Asian economies for its lack of a common monetary policy during 2014-2015, the model will study the cases of America and Europe with a Probit model, being the binary dependent variable countercyclical or procyclical monetary policy {0,1}. For this purpose, the analysis will consist of 18 European and American EMEs from 1975 until 2017, examining 143 years of crisis (annual GDP falls) and their monetary policy reactions to these ones.

7

2.

L

ITERATURE REVIEW

It may seem a normal policy prescription to lower interest rates during crises in order to foster the economic activity, close the negative output gap and decrease unemployment levels due to its cushioning effects, or at least this is what the Keynesian models and many other theories illustrate. In contrast, during last four decades it has been possible to see an important number of economies carrying out policies that, a priori, might seem not very logical for its potential adverse effects. These economies are mainly a substantial portion of the emerging markets that, in fact, were doing exactly the opposite for downturns, putting in practice procyclical1 _{monetary policies.}

There are several studies that analyze the cyclicality of fiscal policies, showing that procyclical fiscal policies seem to be the norm for developing countries, as shown in Talvi and Végh (2000). In contrast, there is not much research done about the cyclicality of monetary policy in emerging economies. This is due to the difficulty to find a common policy among them, since these policies have strongly depended on the exchange rate regime during pasts decades.

Although there are differences among emerging countries, especially between those from different continents, a significant percentage of these countries have been conducting procyclical monetary policies during lasts decades, as shown in the IMF paper “Monetary Policy in Emerging Markets: Taming the Cycle” (2013) which observes

the monetary policy of 49 EMEs for the lasts 40 years.

Procyclical monetary policies can pose a problem to the health of an economy in terms of amplifying upswings and deepening recessions. With respect to crises, procyclical policies make credit more expensive, reduce inflation, lead to less creation of business and increase unemployment, therefore deepening more in the downturn.

One could wonder why the central banks of these economies decided, and some of them still decide, to implement policies that may seem not the best option to recover from a crisis. Indeed, these procyclial policies were conditioned and motivated by the own

1 _{For the purpose of this thesis, countercyclical monetary policy will be defined as a positive correlation} between movements in output and interest rates by the central bank, while it will be procyclical when the correlation is negative, that it, interest rates rising when output is declining.

8

characteristics of these economies, such as the inflation level, amount of liabilities maintained in foreign currency, credibility of the government and financial integration among other factors. These features of the EMEs especially from the 1970 until 2000 were much weaker and volatile than those of the advanced economies. Therefore, easing monetary policy could have been perceived as a turn to an overly loose monetary regime which could have led to risks of excessive inflation, debt monetizing or disproportionate currency depreciation. Events that would have worsened the credibility of private investors and institutions, increasing their risk premium and making finance more expensive.

The literature available points out the following factors as the most important ones determining the EMEs´ monetary policy during crises:

• Credibility of the monetary policy. According to the paper of Calderon et al. (2003) “Counter-Cyclical Monetary Policy Interventions in an Economy”, one of the key elements for EMEs to carry out countercyclical policies is a high degree of confidence of creditors on that the objectives marked by the central banks will be pursued and accomplished. Nevertheless, the lack of credibility along with expansionary policies can be perceived as a regime with potential risks of high inflation and currency depreciation, what is an unfavorable environment for private investments, therefore leading to possible capital flights.

• Percentage of debt maintained in foreign currency. In a situation in which a country strongly depends on foreign funding denominated in non-national currency, it disincentivizes policymakers to decrease interest rates when facing a crisis, since lowering rates and then depreciating the currency (unless the other counterpart is implementing same policies and in a similar or bigger proportion) will increase the real value of the debt of national private and public agents. • Economic and financial integration. In a study of Yakhin (2008) “Financial

Integration and Cyclicality of Monetary Policy in Small Open Economies”, he found

a strong relationship between countries with an advanced financial integration and the implementation of countercyclical monetary policies. While those that were closer to a regime based on autarky were carrying out procyclical policies. Hence, highlighting the importance for EMEs of having commercial and financial openness with other economies.

9

• Inflation level. Several studies and policymakers agree that economies with low inflation levels, generally between one and four percent, have more ease to achieve expansionary monetary policies, since these ones would have lower risks to increase prices to elevated levels. This factor is crucial since a low and stable level of inflation is, generally, the main objective of a central bank. Alesina and Summers (1993) showed that lower inflation economies and more independence of central banks are highly correlated.

• Inflation target. The goal of inflation targeting is to provide the economy with a nominal anchor by committing to a certain rate. As stated by Bernanke et al. (1999)2 _{“inflation targeting is perceived to be able to anchor inflation}

expectations more rapidly and durably than other strategies”. Thus, Brahima Coulibaly (2012), analyzing the EMEs´ monetary policy behavior from 1970 to 2010, argues that the adoption of inflation targeting facilitated their ability to implement countercyclical policies.

In this way, countries with a substantial part of its total debt denominated in national currency, good degree of credibility in the accomplishment of the scheduled objectives by their policymakers, inflation levels around one and four percent along with a strong commitment to keep them low and stable, and are well integrated financial and economically with the rest of the world, should not have problems to implement countercyclical tools during recessions. Otherwise, if an important number of these variables would behave in the opposite way, countercyclical policies could pose risks to the economy.

These more volatile fundamentals in comparison to advanced economies have constrained EMEs to perform policies more independently, being limited by the state of their own macroeconomic fundamentals.

In contrast, during the financial crisis of 2007 EMEs did a remarkable turn regarding monetary policy, a shift to countercyclical policies. Brahima Coulibaly in his paper “Monetary Policy in Emerging Market Economies: What Lessons from the Global Financial

Crisis?” analyzed the monetary policy of 188 advanced and emerging countries from

1970 to 2009 to understand the factors that motivated such shift, concluding that it was

10

mainly due to the improvement of certain macroecomic fundamentals, adoption of inflation targeting and financial institutions.

3.

F

INANCIAL CRISIS AND CHANGE IN THE

MONETARY POLICY STANCE

As seen in the previous section, EMEs generally have been conducting procyclical monetary policies in crises since the seventies, mainly due to the lack of soundness in their fundamentals, financial and economic integration and potential losses in private investors´ confidence.

Throughout the financial crisis of 2008 the vast majority of EMEs loosened their monetary policies to cushion the global shock. This was, for the first time in the last four decades, a significant departure from preceding crises in which the monetary policy strategy was based on increases on interest rates to defend the value of the currency and avoid capital flights.

Figure 1: Number of EMEs conducting procyclical monetary policies during crises (out of a sample of 31)

17 24 3 0 5 10 15 20 25 30 1988-1992 1998-2002 2007-2010

11

Having analyzed the monetary policy of a sample of 31 EMEs3 _{from different continents}

during the period 2007-2009 and considering countercyclical policy when the reference rate of monetary policy and GDP decrease at the same time from the beginning of 2007 until the end of 2009, only three of them did not implement countercyclical measures (Venezuela, Croatia and Argentina). What means that more than 90% of them made a significant shift in the way they were facing crises regarding monetary policy rates. In order to give some perspective to the significance of this shift, Brahima Coulibaly found that during recessive periods in the seventies, EMEs lowered rates in only 30 percent of the occasions. While this percentage for the financial crisis increased to over 80 percent.

Brahima examines the causes of such an important change in the implementation of monetary policy. To this aim, he analyzes how the variables that determine or can determine the type of monetary policy performed during a crisis have evolved over time, as well as to what extent these affect the decisions of the central banks.

After having selected a group of relevant variables4 _{for monetary policy, he explored}

how many of them had been significant enough to have enabled this change of policy. Finding that stronger macroeconomic fundamentals, reforms in their financial markets and the adoption of inflation targeting had been the most principal factors to use monetary policy as a macroeconomic stabilizer. What suggests that the shift from pro to countercyclical policies in 2007-2009 can be explained by the positive evolution of these factors, that indeed is, by the evolution of EMEs towards more advanced and resilient economies.

This turn in the EMEs’ monetary policy supposed a crucial progression regarding monetary instruments towards more developed countries. Thus, it could be suggested that now EMEs can conduct policies independently and use the monetary policy as a macroeconomic stabilization tool. What would let them recover more rapidly from recessive periods and therefore have a more stable output path.

3 _{Turkey, Croatia, Bulgaria, Albania, Russia, Hungary, Poland, Macedonia and Romania for Europe; Chile,}

Colombia, Peru, Argentina, Uruguay, Mexico, Brazil, Venezuela, Ecuador and Costa Rica for America; South Africa, Egypt, Tunisia and Lebanon for Africa; India, China, Thailand, Malaysia, Indonesia, Philippines and Kazakhstan for Asia. Set of countries qualified as EME for the IMF.

4 _{Being these variables: Financial openness, inflation level, credibility, financial development, foreign exchange}

12

This brings into question whether the new method of monetary policy, generally speaking, will be permanent or this was just an isolated event during the financial crisis? Furthermore, has this shift been just caused by the positive evolution in the variables previously explained? Could the EMEs return to conduct procyclical monetary policies even though these key variables have kept improving? Most of the EMEs suffered another recession in 2014-2016, but the policymakers’ decision was slightly different from the previous one.

4.

F

IRST DIVERGENCE IN MONETARY POLICY

,

2014

In 2014 another downturn took place, especially for EMEs. During the period 2014-2016 most of the emerging countries experimented important outflows, falls in equity, currency and GDP fostered by the China´s growth slowdown, FED’s tapering programme and the lower price of certain commodities, mainly oil.

Out of the same sample of 31 EMEs, 25 suffered GDP falls in that period. A ratio that increases without considering India and China, that just experienced slowdowns.

The literature has highlighted the impact of non-domestic factors, such as the global financing environment and the fluctuations in commodity prices, as the main originating sources of this recessive period. Among these external factors, sluggish

external demand has grown at levels below the pasts post-crisis years after 2009; drops in commodity prices during 2014 until 2016, especially oil, have led to lower revenues to

certain EMEs that particularly have commodity-driven earnings and therefore deteriorating their fiscal positions; Political instability and geopolitical tensions have also played a role in the reduction of global trade for economies like Russia, Brazil and Argentina among others; Global funding conditions have eroded in comparison with the post-crisis recovery, especially due to the anticipation of the tapering period implemented by the FED with beginning in 2015, an event that appreciated the Dollar Index by around 20 percent already in 20145_{; Last and probably the most important, the}

unwinding of the stock market and depreciation on the Renminbi in China in 2015 spread

5 _{United Nations in the paper “Development Issues #12: Emerging Economies and the Monetary Tightening} Path in the United States”.

13

out volatility in the financial markets and widened credit spreads along a big portion of the EMEs.

Overall, although there are certain risks of contagion among EMEs, the feedback loops of tighter financial and funding conditions with the real economy might undermine their potential growth.

Regarding monetary policy in this new shock. The study will consider countercyclical policy when the reference rate of monetary policy and GDP decrease simultaneously from the last month of 2014 until the end of 2015. Then, having analyzed the sample of 31 EMEs, the proportion of countries implementing countercyclical policies is 25

percent, while the percentage of them implementing procyclical policies rose to 42 percent. Then, being the 33 percent remaining economies that, either did not change

interest rates during that period or did not suffer any fall in GDP.

These results show a change in the monetary policy strategy in comparison with the previous episode, in which around 90 percent of the EMEs conducted countercyclical monetary policies.

This outcome is not significant enough to show any clear monetary policy stance that could represent a common trend for all the EMEs, as in pasts recessions. However, separating these economies in their respective continents it is possible to distinguish a common tendency.

Taking a closer look to the dominant type of monetary policy in the different continents, it can be seen that there is certain correlation between the policies of countries that are geographically close. For Africa and Asia, the evidence is very ambiguous due to an important percentage of the EMEs of both continents did not have GDP falls. For instance, in the case of Asia; China, India and Philippines did not experience GDP drops, while Kazakshtan and Malaysia pursued procyclical strategies, and Thailand and Indonesia countercyclical ones to recover from the shock.

On the other hand, there is more conclusive evidence about the monetary policy stance for European and American EMEs. In the case of Europe, seven out of nine emerging countries continued conducting countercyclical policies, while only Russia, with a procyclical policy, and Macedonia, maintaining its rates constant, did not follow the

14

common trend. For the ten American EMEs of the sample, seven implemented procyclical policies and only one continued with a countercyclical one, being the two remaining with rates unchanged and without falls in GDP.

This data suggests that European and American EMEs are characterized by a trend during the last crisis episodes regarding monetary policy, unlike Africa and Asia. Nonetheless, the sign of their policies has been the opposite for the downturn of 2014. While Europe has had 77 percent of its EMEs conducting countercyclical policies, 70 percent of the American ones have pursued exactly the opposite strategy.

The Figure 2 depicts the mean of procyclical policies (PCMP) during recessions over time, being the bars closer to 0 representing a larger usage of countercyclical policies, and those ones closer to 1 procyclical policies. The left-hand side graph shows the monetary policy behavior of American EMEs, while the right one represents European EMEs. Therefore, suggesting a divergence regarding monetary policy in the recession of 2014-2015.

Figure 2: Mean of procyclical policies conducted by American and European EMEs

As shown by Brahima Coulibaly in his paper previously mentioned, EMEs all over the world had been mostly pursuing procyclical policies to face crises until the financial crisis of 2008. In contrast, American EMEs in 2014-2015 have come back to a way of doing monetary policy in recessions that seemed to be obsolete. This leads to a discussion of what have been the differences between the EMEs of these two continents that have not allowed American economies to continue cushioning shocks with

15

accommodative policies, unlike the European economies which maintained the preexisting interest rates levels.

According to Brahima, the EMEs could start conducting countercyclical monetary policies as a result of, mainly, the improvement on its macroeconomic fundamentals, financial institutions and adoption of inflation targeting. So, has this shift back to procyclical policies been a result of the erosion of certain macro fundamentals due to the financial crisis or there might be other factors that could explain it?

Figure 3 portrays some of the most determining macro fundamentals6 _{for the EMEs}

monetary measures, as shown by the paper of Brahima Coulibaly, such as the ratio debt to GDP, Inflation (CPI), external debt to GDP and the financial openness (represented by the sum of exports and imports of goods and services measured as a share of GDP). As can be noted, Europe, that still implements countercyclical policies, have had lower inflation and a larger degree of financial openness, which might justify the divergence in monetary policies among the two continents. However, European EMEs have performed worse than the American ones in other crucial variables for the determination of this type of measures like external debt and debt to GDP. For this reason, it is not possible to firmly confirm that macroeconomic fundamentals have been the main cause of this divergence.

6 _{Macroeconomic fundamentals depicted as the yearly average of every variable of the 18 American and} European EMEs.

16

Figure 3: Fundamentals evolution of Americans and Europeans EMEs

The large economies´ central banks of each of the two continents, Federal Reserve and European Central Bank, have been conducting non-conventional monetary policies since the beginning of the financial crisis, lowering interest rates to historical minimums. The FED started its tapering program and announcements of raises in interest rates in 2014, implementing the first of several increases in rates in 2015. In contrast, the ECB has not implemented any measure related with tapering or raises in rates yet, in fact, they keep promoting that the asset purchase program could be extended if necessary. This divergence in monetary policy of the two central banks could have played a role in determination of the EMEs´ monetary policy. At the same time that the FED was raising rates most of the American EMEs did the same, while the European ones maintained their rates low and unchanged, as the ECB did.

The IMF paper “To Hike or Not to Hike? Monetary Policy in Latin America During Fed Liftoff” (2015) study the way EMEs depend on the US monetary policy. They address the question: are emerging countries´ monetary policies, Latin American ones especially, moving in tandem with US rates or they just follow their business cycle? They decomposed the movements on interest rates in two components: the first one reflects changes in rates to stabilize output and inflation, while the second one is everything else, specifically to what extent these ones are due to shifts in US rates. They found out that, after controlling for domestic conditions, there have been significant spillovers for

17

EMEs, especially Latin American countries. Therefore, being more likely to follow FED´s lead.

Furthermore, the raise of rates of the FED and ECB could spark risk scenarios to EMEs that, for instance, took on excessive short-term debt in dollars/euros when the rates were lower and both currencies cheaper, or potential capital flights to US bonds that now have a higher yield in the case of American EMEs. Likewise, it could create a vicious circle in which capital flights depreciate more their currency and appreciate the US dollar, leading to a higher external debt, what would lead to even further outflows from EMEs to safer assets. As Figure 4 depicts, in 2014-2015, higher US yields along with the recession produced outflows from American EMEs, driving down the value of their currencies7 _{vis a vis with the US dollar, what could have stimulated larger outflows.}

Figure 4: Relation US yields-American EMEs´ outflows-EER

All in all, we could reformulate the question at issue given that these two large central banks could potentially have caused, or substantially influenced, the divergence in monetary policy between European and American EMEs for the crisis of 2014-2015. Then, adding to Brahima Coulibaly´s analysis about the variables that determine the EMEs´ monetary policy, the possible influence of large central banks geographically close to them, that is, the FED and ECB.

7 _{Value of American EMEs´ currencies measured as an average of the effective exchange rate of Brazil, Chile,} Mexico, Peru, Argentina, Venezuela and Colombia.

0 20 40 60 80 100 120 140 160 180 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 2 2010 2011 2012 2013 2014 2015 2016 FDI outflows from American EMEs US 5 years bond EER Latin America

18

5.

T

HE

M

ODEL

In the present section the model will be introduced. It will attempt to address the following research question: Has the stance of monetary policy in EMEs been influenced

by whether the large central bank of its respective continent, ECB or FED, was tightening or loosing rates during recessions?

An analysis of the factors that have affected European and American EMEs´ monetary policy over the last 40 years.

5.1 S

To analyze it, the study will make use of a Probit model for panel data. The model will consist in a binary dependent variable. Being this one either countercyclical or procyclical monetary policy {0,1}. Considering procyclical {1} when the reference rate of monetary policy increases and GDP decreases from December of one particular year until December of the next one, and countercyclical {0} when interest rates and GDP decrease simultaneously, for the same period of time.

As independent variables, it collected most of the variables that Brahima Coulibaly used in his paper previously mentioned. He defines the following variables as the ones that determine more significantly the EMEs’ monetary policy. These are the followings:

• External debt to GDP (EXTD): ratio between the debt a country owes in foreign currency and its nominal GDP.

• Inflation (INF): measured by the CPI.

• Foreign exchange reserves (FOREXR): reserve assets held by a central bank in foreign currencies.

• Current account (CA): net trade in goods and services of a country divided by the nominal GDP.

• Debt to GDP (DEBT): ratio debt to GDP.

• Short term debt to external debt (STEXTD): ratio between the debt with maturity of one year or less and interest in arrears on long-term debt, and the external debt.

19

• Trade volume to GDP (TRADE): sum of exports and imports of goods and services measured as a share of GDP. This indicator measures the financial openness of the economy.

• Short term debt to total reserves (STTR): ratio between debt having an original maturity of one year or less and interest in arrears on long-term debt, and central bank´s total reserves including gold.

• Inflation target (IT): Binary variable that is {0} when the central bank of a certain country has not targeted any kind of inflation level and becomes {1} when this one has been adopted.

Furthermore, to capture the possible influence of each large central bank, European Central Bank and Federal Reserve, on the EMEs’ monetary policy of their own continent, a dummy variable (MPLE) will be added to represent the type of monetary policy implemented by the two large central banks during the years that the EMEs have suffered drops in their GDP. This will be done by giving the value {1} when they were increasing interest rates from December of one year to the next one and {0} when they were decreasing8 _{during EMEs crises.}

Additionally, there might be few cases in which an economy does not alter their interest rates during the year of the drop in GDP, therefore being more complicated to identify the type of monetary policy conducted. For this, the direction of the last change in interest rates, either it was upwards or downwards, will be considered to determine the type of policy. In such a way that, if an economy has not varied its rates during a downturn but it raised them one year before the drop in GDP, the monetary policy during that downturn will be considered procyclical, because of the previous rise in rates, and vice versa.

The model´s regression looks as follows:

𝑃𝐶𝑀𝑃𝑖 = 𝛽0+ 𝛽1𝐼𝑁𝐹𝑖 + 𝛽2𝑇𝑅𝐴𝐷𝐸𝑖 + 𝛽3𝐹𝑂𝑅𝐸𝑋𝑅𝑖 + 𝛽4𝐶𝐴𝑖 + 𝛽5𝐷𝐸𝐵𝑇𝑖 + 𝛽6𝑆𝑇𝐸𝑋𝑇𝐷𝑖

+ 𝛽₇𝑆𝑇𝑇𝑅_𝑖 + 𝛽₈𝐼𝑇_𝑖+ 𝛽₉𝐸𝑋𝑇𝐷_𝑖 + 𝛽₁₀𝑀𝑃𝐿𝐸_𝑖+ 𝑢

8 _{Due to the Eurozone was created in 1999, the interest rate for the binary variable MPLE (Monetary Policy} Large Economy) has been determined by the short-term interest rates of seven countries that joined the Euro. These ones are: The Netherlands, Spain, France, Italy, Finland, Germany and Belgium. So, if there are at least four countries out of the simple of seven tightening rates, the dummy variable becomes {1}, and vice versa when they are lowering rates, that is, the dummy will become {0}.

20

5.2 T

HE DATA

The present model consists of 18 emerging countries, 10 American and 8 European, having collected 143 years of recession (yearly GDP falls) from 1983 to 2016. It makes use of common macroeconomic variables taken from economic and financial databases, such as the World Bank, DataStream, Bank of International Settlements (BIS) and IMF data.

More specifically, external debt, inflation and foreign exchange reserves have been collected from Datastream; trade volume, short term debt, both to total reserves and to total external debt, and current account from the World Bank; debt to GDP and inflation from the IMF; interest rates from the webpage of every central bank and the BIS; finally, inflation target data also collected from the central banks´ webpages.

Every variable except inflation, large economy´s monetary policy and inflation target are measured in US dollars. All the data is yearly, and the observation collected is the last one of every year.

The kind of monetary policy conducted by these ones, pro or countercyclical, have been done by analyzing the reference rate of monetary policy and the yearly GDP in the way argued in the first paragraph of the section Specification of the model.

Moreover, due to the following explanatory variables´ coefficients in the Probit model are scaled9 _{and do not represent the exact impact on the dependent variable. Then, it is}

not possible to interpret them as the result of an increase of one percentage point of the independent variable on the dependent one. Instead, the effect of every variable on the type of monetary policy conducted will be interpreted by the Odds Ratio. This will be done by calculating every variable´s average and adding to this one the 20% of its own value in order to get the top quartiles of every variable, hence computing the odds of

9 _{Scaled means that, in the case another model with the same data, such as Logit or a linear regression, would} have been used instead the current Probit, the sign of the coefficients would have been the same, but its values would have been different since they are scaled differently.

21

having procyclical monetary policy during a crisis when a certain variable has performed above or below its average plus/minus 20%.

5.3

P

ROBIT MODEL

This section will present the results obtained from the model. All the countries selected for the sample share common characteristics for the fact of being emerging economies. Nonetheless, there is wide variety amongst countries of different continents and periods of time of the sample. Because of this, the study of the factors that determine the EMEs´ monetary policy will be divided in three parts: a first analysis of all the EMEs jointly; a second one dividing the sample in two periods: 1983-2001 and 2002-2016; and a third one separating the EMEs by continent. The lasts two will attempt to test if there are significant changes in the explanatory variables between the two different periods of time and among EMEs of different continents.

5.3.1 ANALYSIS OF ALL THE EMES JOINTLY

The Table 1 show the results found for the regression previously showed with clustered standard errors to correct for heteroskedasticity and autocorrelation of the time series. It explores 143 years of crisis of 18 EMEs from 1983 to 2016. Positive signs of the coefficient favor the implementation of procyclical monetary policies and negative signs countercyclical ones.

The variables which favor the implementation of countercyclical monetary policies as they increase, with a negative coefficient are; current account, financial openness (trade

volume to GDP), short-term debt to total reserves and the adoption of any kind of

inflation target by the central bank. In contrast, variables backing procyclical policies, with positive coefficient, are external debt to GDP, inflation, government debt to GDP, short-term debt to external debt, foreign exchange reserves and the large economie´s monetary policy of each continent.

Most of the results have the expected sign and effect, except foreign exchange reserves and short-term debt to total reserves, that are non-significant. Besides, inflation, external debt, current account, debt to GDP, short-term to total external debt are not statistically significant either.

22

On the other hand, financial openness, the adoption of inflation targeting and large economies´ monetary policy are significant for significance levels of 5%. Therefore, being these three variables the ones that have influenced the most the EMEs monetary policy decisions during crises since 1983.

Table 1: Probit model of all the EMEs jointly

(1) (2) (3) (4)

Procyclical Monetary Policy PCMP PCMP PCMP PCMP

(PCMP) External 0.00751 0.00765 0.00784 0.00815 debt (0.76) (0.82) (0.77) (0.85) Inflation 0.00364 0.00363 0.00374 0.00372 (0.90) (0.91) (0.89) (0.90) Foreign exchange 0.000973 - 0.00203 - reserves (0.04) (0.08) Current -0.0195 -0.0196 - - account (-0.64) (-0.63) Debt to GDP 0.00860 0.00849 0.00897 0.00876 (1.39) (1.28) (1.50) (1.39) Short-term debt 0.0126 0.0126 0.0150 0.0152 to external debt (0.70) (0.70) (0.83) (0.82) Financial -0.0193** -0.0192** -0.0197** -0.0196** openness (-2.26) (-2.27) (-2.36) (-2.37) Short-term debt -0.00507 -0.00511 -0.00514 -0.00523 to total reserves (-1.12) (-1.20) (-1.11) (-1.21) Inflation -0.576** -0.572** -0.536* -0.526* targeting (-2.04) (-2.06) (-1.90) (-1.86) Monetary Policy 0.825** 0.825** 0.827* 0.827* of large economies (1.97) (1.97) (1.96) (1.95) (MPLE) cut1 _cons -0.356 -0.364 -0.306 -0.323 (-0.62) (-0.69) (-0.53) (-0.62) sigma2_u

_cons 7.61e-39 2.49e-34 5.37e-34 1.87e-34 (0.05) (0.51) (0.51) (0.17)

N 102 102 102 102 t statistics in parentheses

*p<0.1, ** p<0.05, *** p<0.01

The variables removed have been the least significant of the sample.

This outcome suggests that stronger macroeconomic fundamentals, financial openness and the adoption of inflation targeting increase the EMEs´ probabilities of implementing

23

countercyclical policies. Regarding the significant variables, a country that possesses a degree of financial openness 20% higher than the average of all the EMEs, that is at least 60.85% trade to GDP, have lower odds of performing procyclical monetary policy in a downturn, as represented in Table 4 (Annex 3). The inflation target makes countries that adopted it 61% more likely to lower interest rates to cushion shocks. Nevertheless, the variable that represents the monetary policy of the large central bank of the continent has the larger coefficient and with an odds ratio of 3.65, suggesting that every time that an EME suffers a recession along with increases of interest rates by the FED/ECB, that EME will be 3.65 times more likely to conduct procyclical policies than if any of these two large central banks would have left its rates unchanged.

With respect to the research question, the main objective of this study is to test the ECB and FED´s influence on the EMEs´ monetary policy of their same continent, doing it by testing the significance and coefficient of that variable (MPLE). Then, if it is significant and its coefficient one of the largest among the other significant variables, the model will confirm the substantial impact of both large central banks on the EMEs geographically close to them.

Indeed, the variable of interest is statistically significant and has the largest coefficient. Therefore, confirming the hypothesis that American and European EMEs´ monetary policies have been very influenced by large central banks from 1983 to 2016.

Then, how is it? How can the FED or ECB potentially influence the policies of other countries and specifically emerging economies? This can be explained mainly through

two channels of transmission: currency depreciations and capital flights. When the FED or ECB increase rates during a recession that affects certain EMEs, the Euro or Dollar will be appreciated10_{, then increasing the real value of the private debt denominated in}

foreign currency and harming private investors´ confidence. Likewise, capital flights are likely to happen when these large central banks are increasing rates due to a better yield of government securities like treasuries.

10 _{The Euro or Dollar will be appreciated with the increase in rates unless the counterpart is highly demanded} during that period, an event that is not very likely to happen when an emerging economy is in recession.

24

These two mechanisms pose a strong incentive for the EMEs´ policymakers for not conducting countercyclical policies even if they might have cushioning effects beneficial to recover from the downturn.

Finally, these results attribute substantial importance to the influence of the FED and ECB on the EMEs of their same continent but, is it also representative for European and American EMEs separately? Has this influence been so significant for the whole period of the sample?

5.3.2 ANALYSIS BY PERIOD:1983-2001 VS 2002-2016

It should be taken into consideration that the relevance of the independent variables determining the monetary policy decisions could have changed in these two periods since the tendency of these policies during crises changed significantly from 2002 on. In order to test the possible variation of the variables´ coefficient and significance over time, the sample will be divided in two periods, the first one from 1983 to 2001 and the second one from 2002 to 2016, as represented in the Table 2 (Annex 1).

• 1983-2001

Beginning with the analysis of the first period, financial openness and government

debt are the variables that seem to determine more significantly the EMEs´

monetary policy, being significant for p<0.1. These two variables affected the monetary policy decision in diverse ways. Financial openness, with a negative sign, increases the probabilities of a certain EME lowering rates during crises, while government debt poses an incentive for the opposite. In fact, countries in the top quartiles of government debt, with 20% more debt than the average, will be 1.95 times more likely to implement procyclical policies.

Throughout this period, the monetary policy of large central banks did not have a substantial impact on EMEs´ policies, since it has a large p value. Indicating that just the evolution of certain macroeconomic fundamentals and degree of openness with the rest of the world determined the policymakers´ decision making.

• 2002-2016

As seen in the section Financial Crisis and Change in Monetary Policy, the new millennium and especially the crisis of 2008 supposed a crucial departure from

25

previous recession episodes regarding monetary policy. Most of the EMEs carried out countercyclical policies to combat the financial crisis when, generally, this type of economies was conducting procyclical ones to defend the value of their currencies and preserve the confidence of investors.

The independent variable inflation targeting adoption (IT) is eliminated for the analysis of this period, due to most of the countries that adopted any kind of inflation target did it in the late 90s. The ones that did not do it, in general, did not put in practice this measure. Therefore, being not possible to capture the change of the binary variable.

The significant variables are external debt, financial openness and the monetary

policy of large central banks. External debt increases substantially its significance

with respect to the first period, indicating that EMEs got so indebted in foreign currency that it became essential to determine the behavior of EMEs´ policies. Financial openness enhances its coefficient and significance during this time lapse, being the countries with the highest degree of financial openness of the sample 76% more likely to implement countercyclical policies from 2002 until 2016.

Finally, the variable large economies´ monetary policy becomes the most relevant. It presents the largest coefficient and lowest p value. FED and ECB´s policies are extremely crucial for this period. This is demonstrated through the results. Every time these two large central banks have been increasing/decreasing rates during EMEs recessions, the EMEs were 28.9 times more likely to conduct procyclical/countercyclical policies, that is, to follow their same behavior regarding monetary policy even though they are in distinct parts of the cycle. This signifies that EMEs, more recently, have been strongly influenced by larger economies, up to the point that the FED and ECB policies have determined to a much greater extent their policies in comparison with their own macroeconomic fundamentals.

Both periods have shared characteristics regarding the determinants of EMEs´ monetary policy, like the degree of financial openness. Nonetheless, other variables that were not significant at all until 2001 gained much importance in the following years. They are external debt to GDP and the large economies´ monetary policy, being the latter the most significant one.

26 What could explain this change? The financial crisis of 2008 affected severely most of the

advanced and developing economies, making large central banks such as the FED or ECB undertake extraordinary monetary measures and decreasing their interest rates to historically low levels, as well as diminishing the value of their currencies11_{. The fact}

that this was a global crisis and generally was faced with massive accommodative monetary policies, broke the incentive of EMEs for raising interest rates to protect the value of their currencies and to avoid capital flights, due to the advanced economies were decreasing rates even more aggressively than EMEs.

These events made emerging countries get increasingly indebted in a foreign currency such as the US dollar, since a more depreciated dollar posed an opportunity to get cheaper credit. After some years, from 2014 on, tapering and raises in interest rates from advanced economies, specifically the US, started to dominate the world economic scene.

This new panorama of hikes in rates brought back threats for EMEs related with capital flights seeking safer and higher yields in advanced economies (in comparison with previous years), currency depreciations and increase of external debt in real terms. Therefore, increasing their dependence with larger economies during this period between 2002 and nowadays. In contrast, since the ECB has not started yet to perform rises in rates, European EMEs still are anchored in the zero-lower bound.

As can be seen in Figure 5, procyclical monetary policies have lessened in the second period, corresponding these ones mainly to American economies in the recession of 2014-2015. Also, there was a substantial increase in the correlation between procyclical policies and the large economies´ monetary policy for the same period. Up until 2001 this correlation was very close to zero, while from that point on it rose to a positive moderate level (0.622). This higher correlation in comparison with the first period can be explained by the larger exposure of EMEs to capital outflows. Hence, when advanced economies have been characterized by very low rates, EMEs could benefit from inflows of investors looking for better yields and carry out countercyclical policies. On the other hand, when the things reversed and more resilient economies began their announcements of increases of interest rates, they suffered important outflows which

11 _{The value of the nominal effective exchange rate of the US Dollar and Euro decreased during the financial} crisis.

27

they had to defend by implementing raising rates, in the case of American EMEs and the FED. Then, following the same pattern than richer economies regarding monetary policy and, therefore, leading to a lower percentage of procyclical policies.

Figure 5: Mean PCMP and correlation PCMP-MPLE over time

5.3.3 ANALYSIS BY CONTINENT:EUROPE VS AMERICA

So far, the EMEs´ analysis have focused on the study of all of them jointly. Likewise, the economies of both continents might interpret the evolution and state of the variables selected in the model differently for the elaboration of their monetary policies. Table 3

(Annex 2) depicts two analysis, one for each continent, being possible to appreciate

differences in the factors that determine the policymakers´ decisions. • Europe

Europe, for the period 1983-2016, has had two main components influencing significantly its monetary policy decisions: inflation and, again, financial openness. Inflation has a positive coefficient, meaning that increases of this one encourages the implementation of procyclical policies. More specifically, European countries that have had inflation levels below 4%, which is the maximum inflation percentage set as a target (by The Central Bank of the Russian Federation), have been 81% more likely to lower interest rates during economic shocks. In the case of financial openness, as in the other analysis, it has a negative coefficient, indicating a negative correlation with procyclical monetary policies.

28

Additionally, the previous variables are significant for a p value<0.05. Instead, the monetary policy of the large central bank, ECB in this case, is significant for a p

value≤0.1. Therefore not being as determining as other variables, but having played

an important role in the decision making of monetary policy, especially since the beginning of the 2000s.

• America

For the case of American EMEs, four out of the ten variables selected appear to be significant for a p value<0.05. These ones are: external debt to GDP, current account,

inflation targeting and the monetary policy of the large central bank of the continent,

which is the FED in this case.

The sign of the coefficient for every of them is the expected, as increases in the current account and the adoption of an inflation target by the central bank favor the election of countercyclical measures. In the other hand, external debt and the monetary policy of the large central bank have positive signs. In the case of the FED policies, when this one raises rates during a crisis of an EME, emerging economies will be 2.51 time more likely to increase rates.

Comparing the EMEs of both continents, it is possible to notice some differences between them. For instance, the two variables that have determined most significantly the policymaker´s decision of American EMEs from the 80s until nowadays have been inflation targeting and the monetary policy of the large central banks. Variables that are not macroeconomic fundamentals per se. However, for European EMEs the macro fundamentals have been more crucial to define their monetary policy strategy, giving less importance to factors which are extremely significant for other emerging markets, like inflation targeting and the policies of the nearest large central bank.

The difference regarding the dependence on the large central bank might be explained by the type of policies implemented by the FED and ECB. According to Figure 6, the FED has been conducting on average more increases in rates (represented by MPLE) during EMEs´ crises than the ECB, what could have motivated further procyclical measures (represented by PCMP) by American EMEs for motives related with the preservation of investor´s confidence and reduction of capital flights, as discussed before.

29

Figure 6: Relation procyclical policies and ECB/FED policies

Nonetheless, this difference between both continents, especially regarding the influence of larger economies on their policies, is reduced if the period of time of the analysis shortens from 2002 to 2016. Period in which, as seen in the previous section, the variable MPLE gains importance. This suggests that, even if during the whole period the effect of the ECB policies have not had enough significance, it has influenced European EMEs monetary policies as well.

Ultimately, despite American and European EMEs have had some different factors affecting their policies since the 80s, they have been implementing very similar monetary measures during crises until 2014, year in which their policies for that downturn diverged very notably. What could have been motivated by the differences between the FED and ECB policies in that same period of time.

30

6.

C

ONCLUSION

The principal aim of this study is to test if the monetary policy stance of American and European EMEs during crises has been significantly influenced by whether the FED or ECB were tightening or loosening their reference rate. The empirical analysis is based on three Probit models, in a way that if the variable that represents the large economies´ monetary policy (MPLE) is significant and has one of the largest coefficient along the different regressions, the hypothesis of the FED and ECB affecting the EMEs´ behavior will be confirmed.

According to the results, the variable “MPLE” plays a very important role in most of the occasions. It determines crucially the American and European EMEs´ monetary policy decisions jointly, and more substantially after 2002. Nonetheless, before 2002, certain macroeconomic fundamentals and the degree of financial openness had a larger weight determining the type of policies implemented by the EMEs.

The financial crisis of 2008 supposed an unique event in which the majority of the EMEs all over the world reversed their trend of procyclical monetary policies to recover from shocks, conducting countercyclical ones. An important departure from past crisis episodes that still kept most of the American and European EMEs following the same pattern facing downturns. It was during the recession of 2014-2015 when both set of economies diverged notably regarding monetary responses to the crisis for the first time.

More specifically, applying the odds ratio to the model´s outcome, after 2002 both set of EMEs were 28.9 times more likely to implement raises in interest rates during a shock if the respective large central bank was increasing them too, and lower them if this one was decreasing them. Suggesting that the EMEs´ policymakers have had a strong incentive to follow the behavior of larger economies to shape their policies, rather than the evolution of their own fundamentals. An incentive mainly motivated by fears of capital flights and large currency depreciations.

In conclusion, considering the results of the model, this divergence can be potentially explained by the different orientations of monetary policy of the FED and ECB, being the

31

first in a process of tightening and the latter still accommodative. Hence, being the state of their macroeconomic fundamentals rather secondary.

32

7.

R

EFERENCES

Alberto Alesina and Lawrence H. Summers. (1993). Central Bank Independence and Macroeconomic Performance: Some Comparative Evidence. Vol. 25, pages 151-162. Angela Monaghan. (2014). Fragile economies under pressure as recovery prompts capital flight. The Guardian, website:

https://www.theguardian.com/business/2014/feb/02/emerging-markets-brazil-indonesia-south-africa.

Ben S. Bernanke & Mark Gertler. (1999). Monetary policy and asset price volatility.

Federal Reserve Bank of Kansas City, Issue Q IV, pages 17-51.

Brahima Coulibaly. (2012). Monetary Policy in Emerging Market Economies: What Lessons from the Global Financial Crisis?. International Finance Discussion Papers, Board

of Governors of the Federal Reserve System.

Caixa Bank research. (March 2017). Diverging monetary policy in the emerging economies. Caixa Bank, monthly report, website:

http://www.caixabankresearch.com/en/diverging-monetary-policy-emerging-economies.

Calderon et al. (2003). Counter-Cyclical Monetary Policy Interventions in an Economy. A

Paper Presentation to the Committee of Central Bank Governors in SADC.

Donal McGettigan, Kenji Moriyama, J. Noah Ndela Ntsama, Francois Painchaud, Haonan Qu, and Chad Steinberg. (2013). Monetary Policy in Emerging Markets: Taming the Cycle. IMF Working Paper.

European Central Bank. (2016). The slowdown in emerging market economies and its implications for the global economy. ECB Economic Bulletin, Issue 3, article 1.

Fructuoso Borrallo, Ignacio Hernando y Javier Vallés. (2016). The Effects of US Unconventional Monetary Policies in Latin America. Banco de España.

Jose De Gregorio. (2010). Monetary Policy and Financial Stability: An Emerging Markets Perspective. International Finance (commentary), pages 141–156.

33

United Nations: Development Policy and Analysis Division. (2017). Emerging economies and the monetary tightening path in the United States. United Nations: Development

Issues, No. 12.

Yasin Mimir and Enes Sunel. (2018). A recipe for monetary policy in emerging market economies. VOX, CEPR Policy Portal, website:

34 Sample: 1983-2001 Sample: 2002-2016

Period

8.

A

NNEX

Annex 1: Results of the model dividing the sample by period (pertaining to section 5.3.2). Table 2: Probit model by period

(1) (2) (3) (4) (5) (6) PCMP PCMP PCMP PCMP PCMP PCMP External -0.0107 -0.0108 -0.00960 0.0703* 0.0711* 0.0655* debt (-0.78) (-0.82) (-0.80) (1.98) (1.96) (2.25) Inflation 0.00179 0.00180 0.00173 -0.00750 - - (0.85) (0.88) (0.94) (-0.14) Foreign 0.00811 0.00939 - -0.0103 -0.0149 - exchange reserves (0.14) (0.19) (-0.15) (-0.26) Current -0.00364 - - 0.0755 0.0638 0.0658 account (-0.05) (0.60) (1.13) (1.27) Debt to GDP 0.0188* 0.0189* 0.0183* -0.0120 -0.0129 -0.0113 (1.98) (2.13) (2.04) (-0.84) (-0.99) (-0.83) Short-term debt 0.0268 0.0277 0.0295 0.0545 0.0616 0.0480 to external debt (0.61) (0.76) (0.81) (0.58) (1.01) (1.09) Financial -0.0184 -0.0187 -0.0184 -0.0605* -0.0602* -0.0587* Openness (-1.82) (-1.89) (-1.92) (-2.32) (-2.36) (-2.28) Short-term debt -0.00435 -0.00439 -0.00462 -0.00689 -0.0104 -0.00564 to total reserves (-0.99) (-1.03) (-1.17) (-0.20) (-0.61) (-0.71) Inflation -0.582 -0.585 -0.547 - - - targeting (-1.01) (-1.02) (-1.05) Monetary policy 0.0723 0.0621 0.0607 2.413*** 2.400*** 2.378*** of large economies (0.14) (0.13) (0.13) (3.77) (4.12) (3.96) cut1 _cons -0.824 -0.823 -0.858 0.355 0.320 0.466 (-0.95) (-0.96) (-1.02) (0.38) (0.35) (0.50) sigma2_u

_cons 2.88e-34 1.00e-34 2.54e-34 1.90e-33 4.64e-34 9.27e-34 (0.23) (0.11) (0.38) (0.16) (0.26) (0.47)

N 54 54 54 48 48 48 t statistics in parentheses

* p<0.05, ** p<0.01, *** p<0.001

35

Europe America

Annex 2: Results of the model dividing the sample by continent (pertaining to section 5.3.3).

Table 3: Probit model by continent

Continent (1) (2) (3) (4) pcmp pcmp pcmp pcmp External -0.0160 -0.0166 0.0417* 0.0416* debt (-0.72) (-0.72) (2.17) (2.14) Inflation 0.00382** 0.00369* 0.000142 - (2.96) (2.46) (0.22) Foreign exchange 0.0459 0.0493 0.0728 0.0713 reserves (1.07) (1.33) (1.76) (1.79) Current 0.0193 - -0.117* -0.116* account (0.38) (-2.01) (-2.00) Debt to GDP 0.0149 0.0161 -0.0148 -0.0149 (0.92) (0.86) (-1.66) (-1.67) Short-term debt 0.0375 0.0362 0.0292 0.0288 to external debt (1.65) (1.61) (0.91) (0.87) Financial -0.0803*** -0.0797*** 0.00943 0.0104 openness (-3.53) (-3.62) (0.55) (0.57) Short-term debt 0.0145 0.0147 -0.00259 -0.00231 to total reserves (1.40) (1.38) (-0.77) (-0.58) Inflation -0.380 -0.419 -1.209** -1.225** targeting (-0.46) (-0.51) (-2.86) (-2.85) Monetary policy 2.358 2.354 0.784* 0.781* of large economies (1.64) (1.66) (2.56) (2.51) cut1 _cons -0.595 -0.577 -0.623*** -0.618** (-1.27) (-1.19) (-3.32) (-3.23) sigma2_u

_cons 1.22e-33 2.61e-34 4.35e-33 8.85e-36 (0.55) (0.22) (0.74) (0.61)

N 41 41 61 61 t statistics in parentheses

* p<0.05, ** p<0.01, *** p<0.001

36

Odds by period Odds by continent

Procyclical Monetary Policy

Annex 3: Odds of PCMP and the most significant variables in the three Probit models (pertaining to section 5.3).

Table 4: Odds ratio from the Probit model

All EMEs jointly 1983-2001 2002-2016 Europe America

(1) (2) (3) (4) (5) PCMP PCMP PCMP PCMP PCMP External 1.18 1.29 2.12 3.67 0.41 debt LOW (0.56 - 2.5) (0.38 - 4.77) (0.68 - 6.71) (0.98 - 13.62) (0.14 - 1.18) External 0.81 0.81 0.4 0.35 1.53 debt HIGH (0.38 - 1.73) (0.26 – 2.52) (0.10 - 1.37) (0.08 - 1.38) (0.54 - 4.54) Inflation LOW 0.33* 0.43 0.25* 0.19* 0.53 (0.14 - 0.78) (0.12 - 1.43) (0.06 - 0.87) (0.03 - 0.88) (0.16 - 1.81) Inflation HIGH 3.24* 1.01 4.37* 5.21* 2.22 (1.37 - 7.81) (0.15 - 5.07) (1.27 - 16.18) (1.13 - 32.16) (0.67 - 7.16) Debt to GDP LOW 0.67 0.64 0.78 1.51 0.37 (0.31 - 1.45) (0.20 - 2.08) (0.24 - 2.50) (0.42 - 5.32) (0.13 - 1.09) Debt to GDP HIGH 1.64 1.95 0.61 1.75 1.62 (0.74 - 3.68) (0.59 - 7.06) (0.17 - 2.08) (0.48 - 6.24) (0.53 - 5.29) Financial 1.92 1.18 2.4 1.75 2.55 openness LOW (0.88 - 4.29) (0.38 3.86) (0.70 - 8.38) (0.48 - 6.24) (0.70 - 11.53) Financial 0.49 0.64 0.24* 0.98 0.47 openness HIGH (0.21 - 1.13) (0.18 - 2.32) (0.05 - 0.94) (0.25 - 3.65) (0.16 - 1.38) Inflation targeting 0.39* 0.22 1.25 0.14* 0.5 (0.18 - 0.84) (0.04 - 1.09) (0.36 - 4.56) (0.02 - 0.679) (0.18 - 1.39) Monetary policy 3.65* 0.83 28.9* 4.2 2.51 of large economies (1.56 - 8.90) (0.26 - 2.65) (5.13 - 279.62) (0.59 - 48.05) (0.89 - 7.39)

HIGH: it denotes the mean of the variable plus its 20% to indicate what are the odds, in a certain sample, of a variable performing in its top

quartiles of conducting procyclical or countercyclical monetary policy.

LOW: it denotes the mean of the variable minus its 20% to indicate what are the odds, in a certain sample, of a variable performing in its

lower quartiles of conducting procyclical or countercyclical monetary policy.

If the coefficient is below 1 it indicates that, that specific variable in that state (high or low) is more prone to determine countercyclical policies during crises. In the other hand, coefficients above 1 denote higher probabilities to implement procyclical policies.

The brackets show the confidence interval. If every of the parts of the confidence interval is either below 1 or above 1, the coefficient will be significant, indicating a strong relationship between a certain variable in a certain state and procyclical/countercyclical monetary policies. If the interval includes 1 inside, this one will not be significant. The significant coefficients are characterized by a star (*).