Monetary policy independence of the Hungarian Central Bank (HCB) considering global and regional spillovers

(1)

Monetary policy independence of the

Hungarian Central Bank (HCB)

considering global and regional

spillovers

Géza Kovács-Dobák S3743772 kovacsdgeza@gmail.com

Supervisor: dr. Andreas C. Steiner

(2)

2

1. Abstract

The question of the thesis is whether the regional and global spillovers affect the monetary policy rather than the trilemma of international economics, suggested by previous papers and whether the Hungarian Central Bank is independent from these. We analyse this by identifying through the literature what constitute as independence, and what factors influences it. We build our model in 4 steps: (1) Taylor rule factors (2) domestic factors (3) regional spillovers (4) global influences. We find that none of these factors affect the independence of the HCB which would imply that it is independent, but further research is needed to better understand this field.

Three key words: international monetary economics; global financial cycle; monetary independence

(3)

3

2. Content:

3. Introduction

The research area of this thesis is international monetary policy, but in this case through one specific monetary policy setting, and how other international factors, in some cases, other monetary decisions affect the conduct of the originally researched institution.

This paper was born as a result of recent years developments on the field of monetary policy dependence and financial systems. By this we mean, that as the 2008 crisis hit the US and globe, and later on the 2012 crisis hit Europe and the globe as well central banks were puzzled to cope with the challenges the global financial- and European sovereign crisis has brought upon them. Following the 2008-2012 period, the central banks have adopted a new approach which lead us to the world of zero-to-lower-bound policies, forward guidance and so on, without touching the interest rates.

The paper written by Helen Rey and Silvia Miranda Agrippino have highlighted a great flaw in how we understand our economies and the impossible trilemma a country faces when it sets their determining economic policy approaches to the aspect of the economy that a country can control on the international field. It has pointed out a bigger force, a global financial cycle plays a bigger role on the field of international economics between nations, where independent monetary policy setting is not a pursuable option.

Following the ideas of this paper, we started to think on whether this applies to all countries, and what does the literature says about independent policy setting of countries. With me, who has a Hungarian heritage and background, I have decided to test this out on the Hungarian Central Bank and try to measure the independence of their monetary policy setting. But in order to measure it, we first need to understand how monetary policy operates, and what are the factors that influence monetary policy from the official to the unofficial factors.

(5)

5

4. Literature Review

Hungary, as not being part of the Eurozone has its benefits and costs. Having an independent Central Bank not bound by the European Central Bank’s decisions these days are starting to be revaluated and appreciated. However, the question lies how independent can a Central Bank be from global and regional spillovers? This paper is aiming to explore the role of the HCB in its monetary policies and how like ECB or FED policies and events might influence or effect the Hungarian economy which to the HCB might need to react. Similarly, global financial cycles can also influence and effect the HCB’s policies which should also be controlled for.

In order to talk about monetary policy, first we need to look at how scholars define monetary policy, and for later purposes to identify our variables, we need to define also the monetary policy instruments a central bank can utilize. Friedman (Friedman, 2000) defines monetary policy as “…one of the two principal means (the other being fiscal policy) by which government authorities in a market economy regularly influence the pace and direction of overall economic activity, importantly including not only the level of aggregate output and employment but also the general rate at which prices rise or fall.”

A central bank, however, can not only use monetary policies to regulate the financial market but also macro- and microprudential measures. Furthermore, monetary policy also utilizes many instruments, which we need to define clearly (Friedman 2000):

• Open-market operations • Reserve requirements • Interest rate settings • Central bank lending • Macro target settings

However, most importantly, we need to look at what the Hungarian Central Banks defines as their legally accepted framework. According to Act CXXXIX of 2013 the HCB is permitted to use the following monetary policy instruments:

“Article 18

As instruments of its monetary policy, the MNB shall:

a) accept deposits in relation to its account management activity, and provide credit against adequate collateral, subject to the restrictions defined in Article 146;

b) buy and sell securities as well as acting as intermediary of securities in the spot and derivative markets within the framework of open market operations and repurchase agreements;

c) issue securities;

(6)

6 f) regulate minimum reserves…” (MNB2, 2015)

Out of these instruments we will focus on the parts that the central bank has exogenous influence over – namely the interest rates and the minimum reserves.

Monetary policy independence is a highly contentious topic, especially in an ever-globalizing economy, but it is most-often subject of the Mundell-Fleming trilemma (Mundell-Fleming, 1963). The generally accepted theory on macroeconomic policy trilemma is that there are 3 elements of it:

• Independent monetary policy • Fixed exchange rate

• Capital controls

Out of these 3, only two can prevail by sacrificing the third element. However, this theory as the original study’s date shows as well, is outdated, and additional understanding of international macroeconomics has been added in recent years what this literature review aims to showcase. The post-2008 zero-to-lower-bound (ZTLB) monetary policy has introduced non-conventional monetary policy instruments and further transmission channels. In a lecture at the University of Groningen, the president of the Dutch Central Bank, Klaas Knot (Knot, 2019) mentioned that besides interest rate channel, portfolio rebalancing channel, exchange rate channel and signalling channel also operated as transmission channels by central banks.

Recent study from the ECB has voiced that it’s standard policies might affect non-euro countries, namely Hungary as well (Falagiarda et al. 2015). The same paper also recognizes that the Federal Reserve of the United States of America’s policies might also affect the policies of the HCB. This finding corresponds with the findings regularly cited study on “The Global Financial Cycle and Monetary Policy Independence” by IMF Economist Hélène Rey (Rey 2015). She argues that global factors, especially the centre country’s monetary policy is quint-essential in determining monetary policy. In case of EU countries, this can be expanded to the role of the European institutions as well. Monetary policy effectiveness is often doubted, especially considering financial globalisation. Georgiadis and Mehl argues financial globalisation has an ambiguous effect on monetary policy effectiveness (Georgiadis and Mehl, 2016). Their finding implies that financial globalisation has modified the transmission of monetary policy by strengthening the importance of the exchange rate channel. Therefore, the role of central banks is in a revaluation process in recent years, just as much their macroeconomic policy choices. But most importantly, what Georgiadis and Mehl argues is that “domestic financial conditions are increasingly determined by developments in the rest of the world.”

(7)

7

isolated manner, but global and regional factors should be accounted for, especially risks factors reaching over boarders.

Independence of a central bank and it’s monetary policy is also defined by the interdependence of actors of the globalized financial market, which was famously characterized by Bernanke (2007) saying that “With globalized financial markets comes increased financial interdependence. One statistical indicator of that interdependence is that the correlations between long-term interest rates in the United States and those in other industrial countries are high and appear to have risen significantly in the last few years.” However, he goes on saying, that globalization has not reduced the ability of the FED to influence domestic conditions, yet globalization has added an extra dimension to monetary policy.

To measure the independence of monetary policy and its instruments is an even harder task. A commonly referred paper by Cukierman, Webb, and Neyapti is mentioned often, as a basis for inspiration and which has one of the most diversified methodological sources in the field (Cukierman et al., 1992). The indicator that the authors have constructed is one of the most widely used indicator, as it is the first to argue that a gap exists between de juro and de facto independence. Their main finding is that an inflation-based index of overall central bank independence contributes significantly to explaining cross-country variations in the rate of inflation They measure how central bank independence work, use them to rank central banks by their degree of independence, and explores the relation between their independence and inflation outcomes. Their finding, however, focuses also mostly on legal problems. An even defining research paper exploring the defining factor of GDP growth and inflation over interest rate was written by John B. Taylor in 1993 (Taylor, 1993). He talks about the relationship between the nominal target interest rate of the FED and the target / realized inflation, plus the deviation of real GDP growth. He argues, that the main factors that should be taken into account of interest rate setting policies are inflation and GDP growth: “According to this research, good policy rules typically call for changes in the federal funds rate in response to changes in the

price level or changes in real income” (Taylor, 1993). However, due to them being from

1992-3 the findings of the paper might not be as relevant as they were in their time due to new findings on the field, especially in light of the Rey paper from 2015. Nonetheless, as a base for our computation we are going to test for the presence of the Taylor in case of the Hungarian Central Bank’s monetary policy setting. Therefore, two of our factors will be:

- Inflation - GDP growth

(8)

8

the conduct of monetary policy and what they use in their empirical analyses are the following instruments:

• Government spending • Exchange rate

• Monetary financing of public debt • GDP Growth

Other papers, such as the paper from Hausmann, Panizza, Stein notes that the exchange rate regime of the country can also be deterministic on the monetary policy (Hausmann et al., 2001). Although their finding is relevant, however, since we are only measuring one country’s monetary policy the exchange rate regime is not distinctive, since in order it to have an effect there should be a control (group of) country with differing exchange rate regimes.

All these theories have been rarely applied specifically to Hungary in order to check for monetary independence in light of regional and global spillovers. A few exceptions, as mentioned before, is one of the ECB’s own assessment of non-conventional policies on CEE countries. Another great paper written by Jesus Crespo-Cuaresma and.Cezary Wójcik in 2004 (Crespo-Cuaresma, Wójcik; 2004) examines Hungary with Poland and the Czech Republic in regard of exchange rate regimes and how they correlate through time. However, their research is quite limited from that respect, that they only check for one variant of monetary policy instrument (comparison of interest rates) and also, they do not account with many regional or global spillover effects. This consequently means that while their finding that there is a weak linkage with monetary policy independence in light of global events is quite limited.

But other than that, we cannot really find any literature on Hungarian monetary policy independence in light of regional and global spillovers, which leads us to conclude that the concerning field is relatively under-researched. This issue may originate from the fact that the determinative paper by Rey on global factors playing a role in the independence of central banks is relatively new (Rey, 2015).

When talking about the corresponding factors what could influence the monetary policy decision of the HCB, we refer to this same Rey paper. We define these factors as:

• VIX index – Risk aversion • ECB monetary policy decisions • FED monetary policy decisions • EU Banking Leverage

• US Banking Leverage • Global Risk Aversion • Global Market Volatility

• Global credit inflows and outflows

(9)

9

potential drivers of the global financial cycle are the US monetary policy and the general risk aversion keenness of market actors.

Figure 1 The figure represents the mechanism of the global financial cycle and its transmission channels. Source: ECB

(10)

10

Figure 2 This was the end result of the ECB's Global Stock Market Factor index which was conducted in 63 countries on their returns on stocks in correlation to global financial cycle.

What is interesting to note on this figure is that Hungary stands quite high on the spectrum with an estimated variance over 0.5 that puts Hungary on the 21st_{place out of the 63 countries. This}

does not seem to be a significant number, but if we consider that the eurozone countries and the US are by nature are more prone to the global financial cycle Hungary does stand out from this figure. Especially considering their neighbouring CEE countries mentioned in previous ECB paper like the Czech Republic, Slovakia or Poland ranking way under them. In this regard Poland that relatively survived the financial crisis without a recession due to its global risk aversion and low capital openness is not surprising (Kovács-Dobák, 2017), but for Slovakia being part of the eurozone it is.

(11)

11

1. Does the Hungarian Central Bank in the price stability mechanism follow the Taylor Rule in their monetary policy?

2. Do the domestic economic factors have an impact on the conduct of monetary policy?

3. Do regional, mainly European level factors influence the independence of the Hungarian Central Bank’s monetary policy?

4. Do global spillovers influence the independence of the Hungarian Central Bank’s monetary policy?

5. Methodology and data:

In order to assess the independence of the monetary policy instruments of the Hungarian Central Bank we need identify the parameters through which we want to measure the independence. The review of the existing literature provides a good benchmark on the deemed dependent and independent variables on the field in order to perform our regression analyses.

First, we need to check for the measured dependent variables which are the 3 most common instruments a central bank uses to change the course of the monetary policy:

- Interest rates

- Reserve requirements - Open Market Operations

However, due to Open Market Operations are mostly a series of variables, the ability to measure goes beyond the scope of this master thesis. Therefore, we restrict our variables to only the first two. In this analysis we will try to observe how the change in interest rates and the reserve requirements are influenced by certain factors of the (world) economy.

Regarding for data on these variables, the Hungarian Central Bank provides a long-lasting series of data from 1990 for interest rates and 1994 for reserve requirements until present times. However, we will have to restrict our time period of the analyse between 1999-2018, because of the founding circumstances of the Eurozone and the euro (as being connected with 3 of the independent variables). We will use quarterly data for each of the variables, as for datasets like GDP Growth and inflation finding reliable monthly dataset is unlikely. We will try to regress these datasets with our two dependent variables in order to find which factors influence them. We will apply the same independent variables for the two dependent variables, but separately and build our model and results around them. First the independent variables implicated by the Taylor rule will be tested for, then later on we will introduce all of the internal factors. After that, we will introduce our regional and global factors as well. With this gradual introduction of set of variables to our regression mode we can test for the robustness of our dataset, whether each variable changes value after further variables have been added to our model.

(12)

12

website: “The most important objective of overall economic policy is a stable economic growth that is sustainable over the long term. Central banks can support this goal best by keeping inflation low and stable through conducting a predictable and credible monetary policy […] Price stability is defined as low, but not zero, inflation, explained by downward nominal rigidities, the risk of deflation, the need for positive nominal interest rates and statistical measurement errors in the consumer price index. Based on these considerations, price stability is set at around 2%–3% by the majority of central banks of countries with inflation targeting regimes, while the European Central Bank (ECB) defines it as ‘below, but close to 2%’.” (MNB 4, 2019) . For that specific reason, we will dedicate a separate control analyses for inflation which is defined to be the influencing factor in monetary policy instruments.

However, we needed to make a distinction between the variables that are ought to influence the monetary policy and what for they are deemed to change the course, and what are variables that are not in the scope of the central bank to pursue those corresponding economic results.

Therefore, we need to make separate branches of independent variables and test for them gradually. There is the most simple regression analyses of the Taylor rule, to check whether data confirms the theory in case of the Hungarian economy. There are internal factors that the central bank should aim to tender with its policies in order to improve the domestic economy; we call these “internal factors”. Those that are external factors, and which are not in the interest or the control of the central bank and can be divided into two further groups as well: regional factors and global factors.

The internal factors consist of the following variables, out of which the bold points are used for the Taylor rule test:

- GDP Growth - Inflation

- Exchange rate of the USD - Exchange rate of the Euro

- Change in government budget balance - Monetary financing of public debt

We define these variables to be worthy of measuring for two reasons, that is either it influences the domestic conduct of the central bank or the central bank actually is mandated to improve those variables through its monetary policy instruments (MNB 2, 2019). It does not require a lot of explanation that as most central banks follow a price stability principle the HCB needs to pay attention to the inflation of the Hungarian economy.

(13)

13

the Forint. The Hungarian Central Bank has a ready, day-to-day basis dataset on the exchange rate changes for both currencies that we are able to utilize for our analysis. Previous IMF papers and the existing literature (Segalotto et al., 2009) draws the attention on the fact that the change in government budget balance and monetary financing of public debt may also influence the course of the central bank. This is due to the fact in case of looming budget deficit, the central bank can decide to help out the government and buy their treasury bills in order to revitalize the public financing, however, it does come with a price. If the government of the corresponding central bank does not perform well enough, then the central bank needs to intervene. For both indicators, the Central Statistical Agency of Hungary provides a ready dataset up from 2005. We can assess the monetary policy independence through this regression, by view the internal dependence of the monetary policy instruments. By that, we can claim and draw relation between the degree that the internal factors influence the monetary policy instruments.

We need to define what we mean by regional and global factors as well. Under regional factors we understand the European Union’s economic influence, more importantly from the perspective of a central bank, the European Central Bank and the banks in the eurozone. Under global factors we mainly look at global capital flows and global financial market forces. Furthermore, we also

The external factors which consist of the regional factors in bold and the global factors consist of the following variables:

- ECB interest rate decisions

o Main refinancing operations’ interest rate o Marginal lending operations’ interest rate - EU Banking Leverage

- FED effective interest rate decisions - VIX Index

- US Banking Leverage - Global Market Volatility - Global Risk Aversion

- Global cross-border credit to all sectors

We chose the following indicators based on the Rey paper mentioned in the literature review, that identifies the VIX index, the global flow of capital, global market actions and the US-EU leverage on the global financial cycle. The points in bold refers to the fact that we have available data on them until 2018 Q4. The points with regular font refer to the limitation of the dataset. This is explained by the source of the data, as we could only find the original dataset of the Rey paper which had data until 2012 Q4 from 1999 Q1.

(14)

14

lending operations’ interest rate. As mentioned, the dataset also comes from the ECB’s Statistical Data Warehouse.

Since the USA has measures and indicators of nearly every aspect of the economy, it is easy to access target effective interest rates of the FED from the FRED site. For the sake of the time period we use data from 1999 Q1 until 2018 Q4, although data is available for longer periods.

Figure 3 The change in the interest rates of the HCB, ECB, FED between 1999-2018. Source: HCB; ECB; FED.

The VIX index, highlighted by Helen Rey as one of the main indicators of the global financial cycle, the FRED once again provides a long series of data, so we once again chose dataset from 1999 Q1 until 2018 Q4.

As in case of the global factors, the US banking leverage, the global market volatility, the global risk aversion and the global cross-border credit to all sector are provided from Agrippino’s private website who write the global financial cycle paper with Rey, that also came with the corresponding dataset (Agrippino, 2019). Therefore, we got access to complex factors that we can use for our analysis, however, the number of observations is limited, as the paper was published in 2015, and they only use data till the end of 2012.

6. Analysis:

Dataset analyses:

(15)

15

Table 1 This tables shows the summary statistics of the variables before transformation. Source is self-collected and created.

(1) (2) (3) (4) (5)

VARIABLES N mean sd min max

resreq 80 4.088 3.127 1 12 i 80 6.709 4.017 0.900 16 inf 80 4.642 3.090 -1.059 10.78 gdpgr 80 2.480 2.810 -7.557 5.304 exEUR 80 275.6 25.88 236.1 324.1 exUSD 80 232.9 37.09 157.2 303.6 govdef 80 -4.523 2.302 -9.273 -1.647 mfpd 80 196.6 112.2 39.18 453.4 vix 80 19.94 7.874 10.31 58.60 ecb_ml_i 80 2.534 1.772 0.250 5.750 ecb_mro_i 80 -0.157 0.288 -0.750 0.500 fed_i 80 1.917 2.041 0.0700 6.530 glob_crosscred 57 18,765 7,351 8,266 30,690 EU_banklev 57 1.366 0.0902 1.123 1.522 US_banklev 57 0.746 0.0686 0.620 0.840 glo_mkt_vol 56 137.1 46.98 37.01 248.6

After extracting the summary statistics of our dataset, we can see that in most of the variables we have a steady 80 observations, however, in case of the EU & US banking leverage, global cross-border capital flows and the global market volatility, our sample is smaller, 56-57. With a mostly 80 observations dataset we can yield substantial results, but even with 56, considering that these are quarterly data and that this dataset contains 19 years of economic activity.

In most cases of our variables we don’t see any outstanding results, as standard deviations and means are not differing so much from our minimum and maximum values. However, in case of our GDP growth, monetary financing of public debt, the VIX index, global cross-border capital flows and global market volatility we can see outliers that can be explained with a common reason. This factor is the global financial crisis, that caused recession in economic growth to even 7,557% monetary financing of public debt have skyrocketed during this period. Subsequently, global indexes like the VIX, market volatility and the cross-border capital flows also reached peak values as the crisis resulted in higher volatility in all markets. The case of inflation is different though, as although it reaches at its peak a 10,78% rate, significantly outpacing the standard deviation it was not as a result of the crisis, but due to the post communistic inflation rate, that was dropping since 1990, but in 1999 was still high.

(16)

16

corrgram for each of the variables to check for 12 lag-period for both dependent variables the independent variables. We find that through the 12 periods the autocorrelation for most of the variables has a downward shifting trend of autocorrelation.

Table 2 This table aims to show as an example through the interest rate the auto correlation pattern.

-1 0 1 -1 0 1 LAG AC PAC Q Prob>Q [Autocorrelation] [Partial Autocor]

1 0.9301 0.9562 71.829 0 |--- |--- 2 0.8529 -0.1999 133.01 0 |--- -| 3 0.7656 -0.1634 182.95 0 |--- -| 4 0.6883 0.1646 223.84 0 |--- |- 5 0.6325 0.097 258.84 0 |--- | 6 0.5903 0.0972 289.73 0 |---- | 7 0.5563 0.1727 317.54 0 |---- |- 8 0.5282 0.111 342.96 0 |---- | 9 0.5008 -0.0046 366.14 0 |---- | 10 0.4729 0.0622 387.1 0 |--- | 11 0.4423 0.07 405.7 0 |--- | 12 0.4129 -0.1193 422.14 0 |--- |

We can observe correlation in case of (but not limited to) the interest rate, which would suggest that the previous value determines the present, but it is easy to understand that this is only because interest rates are following each other, especially in case of a time series analysis. However, partial autocorrelation could be an issue we need to pay attention to in this case. However, in none of the cases of the variables do we observe partial autocorrelation. In some cases, there isn’t even autocorrelation as for the observations have such an irregular pattern that from there is no connection between the previous value and the present value of the variables. Since we do not utilize a Vector Autoregression (VAR) model we cannot rely on tests to identify how many periods we should lag our variables so that the dependent variables could already come into effect of the change of the independent variables. Therefore, we rely on economic intuition, as there is a 3-months period between each observation. We can assume with confidence that an event that my need a reaction from the HCB would be coming at least 3 months after the event, especially considering that the monetary policy setting committee meets twice every month (from those only one of the is interest rate setting meeting). From this we can conclude that we can rely on our economic intuition that lagging our variables with only one period is sufficient enough.

We should also check for unit root trends, being a time-series analysis. We conduct simple Dicky-Fuller tests on each of the lagged variables to check for additional trends in the observations of the variables. In most cases the dfuller test yields a non-rejection of the H0, as the t-statistic is smaller than the critical values, meaning that unit root trends exists in most of

(17)

17

the variables. There are only two exceptions to this, namely the lagged ECB main refinancing operations interest rate and the lagged global market volatility index where the t-statistic passes the critical value requirement for this sample size.

Table 3 The Dicky-Fuller test of the lagged variable for the global market volatility. We can see that the t-statistic passes the H0 hypothesis with -4,245, as for an observation size of this, -2,926

is the 5% critical value. This means that the variable does not contain unit root trends.

Dickey-Fuller test for unit root Number of obs = 55 --- Interpolated Dickey-Fuller ---

Test 1% Critical

5%

Critical 10% Critical

Statistic Value Value Value

Z(t) -4.245 -3.573 -2.926 -2.598

MacKinnon approximate p-value for Z(t) = 0.0006

Table 4 The Dicky-Fuller test of the lagged variable for the ECB’s main refinancing operations interest rate from the Ray paper. We can see that the tstatistic passes the H0 hypothesis with

-3,453, as for an observation size of this, -2,908 is the 5% critical value. This means that the variable does not contain unit root trends.

Dickey-Fuller test for unit root

Number of obs = 78 --- Interpolated Dickey-Fuller --- Test 1% Critical 5% Critical 10% Critical

Z(t) -3.453 -3.541 -2.908 -2.589

We can easily correct for the rest of the variables by taking the first differential of these variables and then reapplying the dfuller test. After generating those variables and running the dfuller tests, we find we can now reject the H0 of unit root existence, as our t-statistic is now bigger than critical value assigned to size of the observation. Figure 4 is provided as an example to illustrate the change in one variable before and after taking the differential value of the variable. For this we take inflation as an example:

Table 4 The Dicky-Fuller test of the lagged variable for inflation of Hungary. We can see that the tstatistic doesn't passes the H0 hypothesis with 1,781, as for an observation size of this, -2,907 is the 5% critical value. This means that the observations contain unit root trends which

have to be corrected for.

Dickey-Fuller test for unit root Number of obs = 79 --- Interpolated Dickey-Fuller --- Test 1% Critical 5% Critical 10% Critical

(18)

18

Table 5 T he Dicky-Fuller test of the lagged variable for inflation of Hungary. We can see that the t-statistic passes the H0 hypothesis with -5.582, as for an observation size of this, -2,908 is the 5% critical value. This means that the observation does not contain unit root trends anymore, as

a result of the differential values.

Dickey-Fuller test for unit root Number of obs = 77 --- Interpolated Dickey-Fuller ---

Test 1% Critical 5% Critical

10% Critical

Z(t) -5.582 -3.542 -2.908 -2.589

After lagging and generating the differential values of each variable where needed, we can test the variables for outliers and the normality. Our dataset we test for skewedness and kurtosis as well. Most of our variables passed the normality test, as they were normally distributed and had no significant kurtosis or outliers. We conduct sktests as well on the variables and check for outliers on the histogram of the variables that we identify from the sktest would need closer observation. In most cases, as mentioned, the variables are normally distributed, and if not, by the nature of the data it has outliers (such as in the case of the lagged variable of the ECB’s main refinancing operations rate). In other cases, we can see that the variable has significant outliers that would be easy to drop from the observations, namely in the case of the euro exchange rate (lagged, differential value).

However, the significance of those observations cannot be neglected, thus omitted. The reason is that the exchange rate volatility has been an influencing factor explicitly for the HCB before twice (2003, 2008) (Origo, 2008). Therefore, we do not drop these outliers, due to economic intuition 0 .0 2 .0 4 .0 6 .0 8 D e n si ty -20 -10 0 10 20 30 ldexEUR

(19)

19

and its contribution to the whole analysis.

Establishing the model:

We are analysing all these data through time series regression models for each of the dependent variables. For the sake of simplicity, we start off by providing only the main factors of the influencing indicators on the monetary policy, provided by the Taylor rule (Taylor; 1993). Therefore, the equation that we use as a start for the interest rates is shown by equation (1) and for the reserve requirements is shown by equation (2):

:

(1) 𝒾"#$ = β'𝑐𝑜𝑛 + β,𝑖𝑛𝑓 + β0𝐺𝐷𝑃𝑔𝑟

(2) ℝ"#$ = β'𝑐𝑜𝑛 + β,𝑖𝑛𝑓 + β0𝐺𝐷𝑃𝑔𝑟

where:

- 𝒾"#$: is the interest rate of the Hungarian Central Bank through the time period

- β': is the coefficient of the constant and con is our constant

- β,: is the coefficient of the inflation and inf is the lagged and differential inflation

observation

- β0: is the coefficient of the GDP growth and GDPgr is the lagged and differential GDP

growth observation

- ℝ"#$: is the reserve requirement to commercial banks assigned by the Hungarian

Central Bank

With this, we are trying to assess the simplest relation between the change in interest rate or the reserve requirement and the price stability with GDP growth change through the time period.

With this simple regression model, we can start our analyses and further deepen it with each variable cluster added along the way. Therefore, we will remain with one equation per dependent variable. For the next step we will add al the internal factors that influence the domestic economy which gets us to equation (3) for the interest rate and (4) for the reserve requirement:

(3) 𝒾_"#$ = β_'𝑐𝑜𝑛 + β_,𝑖𝑛𝑓 + β₀𝐺𝐷𝑃𝑔𝑟 + β₇𝑒𝑥𝐸𝑈 + β_<𝑒𝑥𝑈𝑆 + β_>𝑔𝑜𝑣 + β_@𝑚𝑜𝑛𝑓𝑖𝑛

(4) ℝ"#$ = β'𝑐𝑜𝑛 + β,𝑖𝑛𝑓 + β0𝐺𝐷𝑃𝑔𝑟 + β7𝑒𝑥𝐸𝑈 + β<𝑒𝑥𝑈𝑆 + β>𝑔𝑜𝑣 +

β_@𝑚𝑜𝑛𝑓𝑖𝑛 Whereby the new variables are:

• β7: is the coefficient of exchange rate of Euro and exEU is the lagged and differential

(20)

20

• β_<: is the coefficient of exchange rate of US dollar and exUS is the lagged and differential exchange rates observations

• β_>: is the coefficient of the government debt and gov is the lagged and differential government debt observations

• β_@: is the coefficient of the monetary financing of public debt and mfpd is the lagged and differential monetary financing of public debt observations

The third step in building our model is equation (5) and (6) which includes also the regional factors for the interest rate and reserve requirement dependent variables:

(5) 𝒾_"#$ = β_'𝑐𝑜𝑛 + β_,𝑖𝑛𝑓 + β₀𝐺𝐷𝑃𝑔𝑟 + β₇𝑒𝑥𝐸𝑈 + β_<𝑒𝑥𝑈𝑆 + β_>𝑔𝑜𝑣 + β_@𝑚𝑜𝑛𝑓𝑖𝑛 + β_B𝐸𝑀𝑅𝑂 + β_F𝐸𝑀𝐿 + β_H𝐸𝑈𝐵𝐿 (6) ℝ_"#$ = β_'𝑐𝑜𝑛 + β_,𝑖𝑛𝑓 + β₀𝐺𝐷𝑃𝑔𝑟 + β₇𝑒𝑥𝐸𝑈 + β_<𝑒𝑥𝑈𝑆 + β_>𝑔𝑜𝑣 + β_@𝑚𝑜𝑛𝑓𝑖𝑛 + β_B𝐸𝑀𝑅𝑂 + β_F𝐸𝑀𝐿 + β_H𝐸𝑈𝐵𝐿 Where by: • βB: is the coefficient of the main refinincing operations rate of the ECB and 𝐸𝑀𝑅𝑂 is the

the lagged and differential refinancing rate observations

• β_F: is the coefficient of the marginal lending rate of the ECB and 𝐸𝑀𝐿 is the the lagged refinancing rate observations

• β_H: is the coefficient of the EU banking leverage and 𝐸𝑈𝐵𝐿 is the the lagged and differential EU banking leverage observations

With the similar analogy, we include the last cluster of independent variables, the global factors, to both the interest rate and the reserve requirement of the HCB:

(1) 𝒾_"#$ = β_'𝑐𝑜𝑛 + β_,𝑖𝑛𝑓 + β₀𝐺𝐷𝑃𝑔𝑟 + β₇𝑒𝑥𝐸𝑈 + β_<𝑒𝑥𝑈𝑆 + β_>𝑔𝑜𝑣 + β_@𝑚𝑜𝑛𝑓𝑖𝑛 + β_B𝐸𝑀𝑅𝑂 + β_F𝐸𝑀𝐿 + β_H𝐸𝑈𝐵𝐿 + β_,',𝑉𝐼𝑋 + β_,,𝐹𝐸𝐷 + β,0𝑈𝑆𝐵𝐿 + β,7𝐺𝐶𝐶 + β,<𝐺𝑀𝑉 + β,>𝐺𝑅𝐴 (2) ℝ_"#$ = β_'𝑐𝑜𝑛 + β_,𝑖𝑛𝑓 + β₀𝐺𝐷𝑃𝑔𝑟 + β₇𝑒𝑥𝐸𝑈 + β_<𝑒𝑥𝑈𝑆 + β_>𝑔𝑜𝑣 + β_@𝑚𝑜𝑛𝑓𝑖𝑛 + β_B𝐸𝑀𝑅𝑂 + β_F𝐸𝑀𝐿 + β_H𝐸𝑈𝐵𝐿 + β_,',𝑉𝐼𝑋 + β_,,𝐹𝐸𝐷 + β,0𝑈𝑆𝐵𝐿 + β,7𝐺𝐶𝐶 + β,<𝐺𝑀𝑉 + β,>𝐺𝑅𝐴

• β_,': is the coefficient of the VIX index and 𝑉𝐼𝑋 is the the lagged and differential index observations

• β,,: is the coefficient of effective target interest rate of the FED and 𝐹𝐸𝐷 is the the

lagged and differential interest rate observations

• β,0: is the coefficient of the US banking leverage and 𝑈𝑆𝐵𝐿 is the the lagged and

differential US banking leverage observations

(21)

21

• β_,<: is the coefficient of the global market volatility and 𝐺𝑀𝑉 is the the lagged

volatility rate observations

• β_,>: is the coefficient of the global risk aversion and 𝐺𝑅𝐴 is the the lagged and

differential aversion observations

With this we can estimate the factors that might influence the monetary policy of the Hungarian Central Bank and also assess which factors influence more the decisions, so that we can a draw a conclusion on whether the HCB is independent from regional and global factors.

Regression analyses:

With the following equations we can finally start our regression models to check for connections between the dependent and the independent variables. We test and showcase each cluster of factors one after another. With this we aim to test each variables robustness, whether they remain significant throughout the regressions. The model yields the following table:

Table 6 This table shows the regression results for the interest rate of the HCB in relations to our independent variables. Each column adds a cluster of variables to our results.

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

VARIABLES Taylor Factors Internal Factors Regional Factors Global Factors

Interest rate

Inflation 0.0740 0.140 0.0813 0.127

(0.471) (0.236) (0.595) (0.545)

GDP Growth -0.0870 -0.0826 -0.0875 -0.139

(0.266) (0.323) (0.415) (0.382)

Euro exchange rate 0.0262 0.0295 0.0603*

(0.117) (0.172) (0.0723)

Dollar exchange rate -0.00560 -0.00950 -0.0212

(22)

22

ECB’s marginal lending rate 0.125 0.537

(0.773) (0.425)

ECB’s main refinancing operations rate 0.947* 0.944 (0.0645) (0.167) EU Banking leverage 5.652 9.343 (0.218) (0.159) VIX Index -0.00134 (0.970) FED Effective target interest

rate

-0.541 (0.238) Global cross-border capital

flows

-0.000366 (0.244)

US Banking leverage 15.01

(0.206)

Global Market Volatility 0.00168

(0.751)

Global Risk Aversion -0.0166

(0.262) Constant -0.179* -0.197** -0.0486 -0.135 (0.0681) (0.0495) (0.760) (0.865) Observations 78 78 56 47 R-squared 0.023 0.064 0.206 0.297 pval in parentheses *** p<0.01, ** p<0.05, * p<0.1

(23)

23

interest rate. However, this significance disappears over the next regression, where we also account for the global factors. In this last stance of regression, although our refinancing rate seize to be significant, the exchange rate of the EUR/HUF become significant with 0,0628 p-value on a 10% significance level. This means that at our final regression if the euro exchange rate changes by 1 Forint we can expect 0.0512 percentage point change in the interest rate.

What is although interesting is that our R-squared values when we don’t include our regional and global factors are much lower (0,023 and 0,064), however, as we add the regional factors there is a significant rise in the R-squared value. Same applies to the global factors. Besides this, we can see that our constant variable at the first two models are significant. This means that the yet omitted variables contain significant variables. As we add the regional and global factors, the significance of the constant disappears with a huge shift from a starting p-value of 0,0495 to a 0,76. From this, and our squared p-values we can deduct that the regional and global factors contributed to the understanding of the factors that influence the interest rate changes. Our computation got better and resulted more precise results (although not significant).

What is the most interesting result of our regression model does not confirm the existence of the Taylor rule in case of the HCB, as the inflation and the GDP growth does not yield a significant result in any of the stages of our regressions. Not even close, although it is important to mention that Taylor has used the difference of the expected values of inflation and interest rates in his computation.

From a robustness standpoint we can deduct that our dataset is not so robust as the variables that do happen to be significant are not significant throughout all the 4 models. Also, those variables that are present at least 3 times in our models we can see a highly changing p-value which just further suggest that our dataset is not so robust.

We progress ahead with our analyses by presenting the regression results for the reserve requirement in relations to our independet variables which can be found in table 3.:

Table 7 This table shows the regression results for the reserve requirement of the HCB in relations to our independent variables. Each column adds a cluster of variables to our results.

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

Reserve requirement

Inflation -0.000251 0.0251 0.0382 -0.0589

(0.997) (0.762) (0.724) (0.659)

(24)

24

(0.894) (0.853) (0.752) (0.670)

Euro exchange rate 0.0214* 0.0242 0.0384*

(0.0710) (0.116) (0.0726)

Dollar exchange rate -0.0137* -0.0195* -0.0200

(0.0739) (0.0739) (0.151) Government deficit 0.0117 0.0419 0.106 (0.884) (0.667) (0.357) Monetary financing of public debt -0.000605 -0.00119 -0.00239 (0.708) (0.641) (0.463)

ECB’s marginal lending rate -0.219 -0.720*

(0.478) (0.0987)

ECB’s main refinancing operations rate -0.463 -0.139 (0.197) (0.745) EU Banking leverage 2.226 1.917 (0.491) (0.645) VIX Index -0.0267 (0.242) FED Effective target interest

rate

flows

5.07e-05 (0.798)

(0.913)

Global Market Volatility -0.00762**

(0.0301)

Global Risk Aversion 0.00580

(0.535)

Constant -0.141** -0.153** -0.300** 0.662

(25)

25

Observations 78 78 56 47

R-squared 0.000 0.060 0.175 0.430

pval in parentheses *** p<0.01, ** p<0.05, * p<0.1

What we can frst see from this table is that most of our independent variables are not significant, especially at last regression results with a R-squared value of 0,43, so quite close to the 0,5 statistical significance. The inflation and the GDP growth isn’t significant as well throughout the regressions, however, it is important to note that the Taylor rule is not applicable in case of the reserve requirements. Nonetheless, this leads us to believe, that if the HCB does follow through with its monetary policy mandate to control inflation, that means that they are using other transmissin channels than the interest rate or the reserve requirements.

During our internal factor regression we can see that our exchange rates (USD & EUR) are both significant. The dollar and the euro has a significance on a 10% significance level with the euro having a 0,071, and the dollar having a 0,0739 p-value. This means that if the euro exchange rate changes by 1 Forint then the reserve requirements is expected to change by 0,0214 percentage point and in case of dollar this is a -0,0137-percentage point change in reserve requirements. This can be interpreted that in case of a change in exchange rates, the HCB may assign a higher reserve requirement.

In the case of regional factors, the euro will be dropped as a significant variable, however the dollar remains one on a 10% significance level and a p-value of 0,0739. However, this significance disappears when we add the global factors. However, the euro makes a comeback in the last phase with a 10% significance level. What does turn out to be significant in the last phase when the global factors are added is the global market volatility with a on a 10% significance level. Our regression model tells us that by every unit change in the global market volatility there is a -0,00762 percentage point change in the reserve requirements. This can be interpreted that in case global market volatility rises, signalling a more volatile global conditions of the traded goods, than the monetary policy is loosened, and if there is a drop in the volatility of the global conditions of the traded goods, then the central bank may assign higher reserve requirements to build capital buffers against the future global market volatility drops. Besides the market volatility, the regional factor of ECB’s marginal lending rate becomes significant after adding the global factors on a 10% significance level, however, this is still not enough to make the regional effect a reasonable addition our model, as it is a standalone variable in that cluster.

However, what is most worrying in our analyses is that our constant remained significant in all phases except the last, meaning that there have been many variables omitted in our analyses. However, our R-squared did get better, meaning that our variables did fit better on the regression line, especially after adding the global factors.

(26)

26

Although, it is more robust, as we can at least see the US dollar to be significant for two consecutive periods. Also, those variables that are present at least 3 times in our models we can see a highly changing p-value which just further suggest that our dataset is not so robust.

To further analyse the robustness of our model we decided to separate our dataset at the start of the sovereign crisis period of Europe. We decide to test for our dataset’s robustness due to the ZTLB monetary policies that many economies have adopted after the crisis. Due to these policies our initial regression analyses might not reflect adequate results.

We separate our dataset at 2012 Q1 and we present our finding as pre-crisis and post-crisis tables. First, we present our pre-post-crisis regressions for the interest rate and the reserve requirement:

Table 8 This table shows the regression results for the interest rate of the HCB in relations to our independent variables. Each column adds a cluster of variables to our results for the time period before the crisis (2012Q1).

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

Interest rate

Inflation 0.0628 0.104 0.0592 0.127

(0.681) (0.554) (0.728) (0.545)

GDP Growth -0.104 -0.0879 -0.0521 -0.139

(0.370) (0.501) (0.686) (0.382)

Euro exchange rate 0.0252 0.0322 0.0603*

(0.307) (0.188) (0.0723)

Dollar exchange rate -0.00477 -0.0115 -0.0212

(0.780) (0.489) (0.328)

Government deficit -0.0246 -0.107 -0.131

(0.874) (0.483) (0.469)

Monetary financing of public debt

-0.00170 0.00255 0.00432

(0.638) (0.511) (0.400)

ECB’s marginal lending rate 0.0925 0.537

(0.843) (0.425)

ECB’s main refinancing operations rate

(27)

27 (0.0672) (0.167) EU Banking leverage 5.952 9.343 (0.261) (0.159) VIX Index -0.00134 (0.970) FED Effective target interest

rate

flows

-0.000366 (0.244)

(0.206)

Global Market Volatility 0.00168

(0.751)

Global Risk Aversion -0.0166

(0.262) Constant -0.162 -0.192 -0.0418 -0.135 (0.277) (0.218) (0.807) (0.865) Observations 51 51 51 47 R-squared 0.018 0.055 0.201 0.297 pval in parentheses *** p<0.01, ** p<0.05, * p<0.1

Table 9 This table shows the regression results for the reserve requirement of the HCB in

relations to our independent variables. Each column adds a cluster of variables to our results for the time period before the crisis (2012Q1).

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

VARIABLES Taylor Factors Internal Factors Regional Factors Global

Factors

(28)

28

Inflation -0.00924 0.0196 0.0347 -0.0589

(0.932) (0.869) (0.773) (0.659)

GDP Growth 0.00399 0.0428 0.0456 0.0431

(0.961) (0.627) (0.616) (0.670)

Euro exchange rate 0.0358** 0.0278 0.0384*

(0.0361) (0.109) (0.0726)

Dollar exchange rate -0.0253** -0.0216* -0.0200

(0.0327) (0.0698) (0.151) Government deficit -0.0118 0.0226 0.106 (0.911) (0.834) (0.357) Monetary financing of public debt 0.000892 -0.000813 -0.00239 (0.715) (0.766) (0.463) ECB’s marginal lending rate -0.243 -0.720* (0.463) (0.0987) ECB’s main refinancing operations rate -0.403 -0.139 (0.294) (0.745) EU Banking leverage 3.000 1.917 (0.420) (0.645) VIX Index -0.0267 (0.242) FED Effective target

(29)

29 Volatility

(0.0301)

Global Risk Aversion 0.00580

(0.535) Constant -0.197* -0.225** -0.321** 0.662 (0.0636) (0.0359) (0.0107) (0.198) Observations 51 51 51 47 R-squared 0.000 0.115 0.187 0.430 pval in parentheses *** p<0.01, ** p<0.05, * p<0.1

In case of the pre-crisis value, we can see strong robustness to our original model as there most of our value are the same or close to the same. Also, our significant variables remained significant, and there were no further variables that became significant; therefore we can see that our model goes well with the pre-crisis period.

For the second period we present the post-crisis regressions in the appendix, due to its results. As we suspected, the post-crisis period with the ZTLB monetary policies combined with the low level of observation available has led us to omit most of our variables. In both cases of the interest rate and the reserve requirement, we can see that as a result of omitted variables our model won’t even have anything to regress for the global factors phase, but there is already a significant variable, however, due to the many wrong results, this can be neglected. Even though this goes against economic intuition to drop some of the variables, as economically they seem to be related to the variable, and also previous papers were able to find similar results with each of the variables, it is reasonable to assume that they didn’t use the same exact amount of variables and method. Therefore, we decide to drop one of the variables, one-by-one, starting with the variables that are the least significant. By testing the optimal amount and right variables, we come to the conclusion that we drop the inflation, GDP growth, VIX, index, government debt and the global market volatility variables in case of the interest rate. From this, the latter was is easier to explain, as to begin with we saw no reason to keep them, as it explains a similar to the VIX index, and government debt would probably be under the influence of the interest rate, so reverse causality might be the case for them. As for the inflation and GDP growth, once again reverse causality can be argued, but considering the importance of these two variables, it goes against economic intuition (even if it helps the econometric understanding). As for the VIX index it is hard to argue why, however, at the end of the 4th_{phase, after adding the global factors, the VIX index is the least significant with 0,97} p-value, so it is the first to be dropped.

(30)

30

not easy to argue for any of them why they are being dropped, as all three could be influencing the reserve requirement by economic intuition. However, this do seem to suggest, that the US economy’s role is not as important from the perspective on Hungary, as one would think considering that we drop two US variables, and a global capital one.

Takin into note these changes in our interest rate regression model, we come to the following regression table results:

Table 10 This table shows the results of our regressions if we omit certain variables in some cases of the interest rate and the reserve requirement. By this we yield more significant variables. Variables in bold represent the dependent variables

(1) (2)

VARIABLES d_i d_resreq

Reserve requirements

Inflation -0.0679

(0.576)

GDP Growth 0.0471

(0.594)

Euro exchange rate 0.0603** 0.0396**

(0.0425) (0.0237)

Dollar exchange rate -0.0243 -0.0218*

(0.218) (0.0561)

Government deficit 0.114

(0.283) Monetary financing of public debt 0.00456 -0.00228

(0.307) (0.452)

VIX Index -0.0248

(0.226)

ECB’s marginal lending rate 0.433 -0.723**

(0.467) (0.0401)

ECB’s main refinancing operations rate

1.017* -0.111

(0.0933) (0.768)

EU Banking leverage 10.50* 2.358

(0.0797) (0.512)

Global market volatility -0.00811***

(0.00317)

Global risk aversion -0.0159 0.00465

(0.205) (0.520)

Interest rate

FED Effective target interest rate -0.490 (0.208) Global cross-border credit flows -0.000437*

(0.0866)

(0.331)

(31)

31 (0.655) (0.0473) Observations 47 47 R-squared 0.268 0.429 pval in parentheses *** p<0.01, ** p<0.05, * p<0.1

As we can see in the case of the interest rate the new significant variables are the EU banking leverage, the ECB’s main refinancing operations rate and the global cross-border credit flows on a 10% significance level. This would suggest that two out of the three regional variables would be significant, which would imply that our regional factors actually have an influence on the interest rate setting of the HCB. This would be in line with the ECB’s own research paper by Falagiarda, McQuade and Tirpák from 2015 which also said that non-conventional monetary insturments would influence the non-euro economies. Furthermore, the global cross-border credit flow leads us to believe if there would be a sudden decrease in global capital flows in Hungary, that could lead to the interest rate to be raised, so that it could attract capital for the economy. As for our R-squared value, it does not change that drastically, as it is 0,268, and before that it was 0,295.

In the case of the reserve requirement, our previously significant VIX index turned insignificant, yet four new variables turned significant with their significance level in paranthesis: euro exchange rate (5%), dollar exchange rate (10%), the ECB’s marginal lending rate (5%) and the global market volatility (1%), however, with the latter, the effect of the global market volatility is quite negligable. As for the dollar and euro exchange rate making a comeback for being significant is no surprise as reserves are ought to be built up against exchange rate fluctuations. More welcoming news is that we were able to increase our R-squared value quite a bit, to 0,429 which is quite close to the statistically more acceptable range of 0,5.

7. Conclusions:

Generally we were looking for the independence of the Hungarian Central Bank, and its monetary policy decisions. Our hypothesis question was (1) weather the Hungarian Central Bank follow the Taylor rule monetary policy, (2) or do domestic economic factor determine the monetary policy (3) or whether regional / European level factors influence the decisions of the HCB (4) or whether global spillovers that affect the HCB’s monetary decisions the most?

We tried to approach this with a time-series regression model where we identified the relevant variables through the literature. After constructing our model and equations, we have ran 6 regressions on the interest rate’s and the reserve requirement’s effects on the monetary policies. Firstly we just included in our model the Taylor rule factors, than the domestic; the regional; than the global factors finally.

(32)

32

stability. This is further strengthened by the fact that even after omitting some of the least significant variables (or in case of interest rate, the two variable being the omitted) we see no change in their insignificance.

As for, whether domestic economic factors may determine the monetary policy, we were able to see that most of the factors do not have an affect on the interest rate and the reserve requirements. However in case of the euro exchange rate, we do see that as we add the global factors to our model, it becomes significant. However, this is not a surpirse considering the fact that the HCB has changed the interest rate in 2003 and 2008 by 300 base points explicitly due to exchange rate fluctuations (Origo, 2008). This is also shown in the reserve requirements results, as both exchange rates are significant influencers of the reserve requirements till we add the global factors. Although, at the end the dollar and the euro doesn’t turn out to be significant, we cannot neglect their results as an important influencing factor.

As for the regional / European factors influence, we can observe that when we add the regional factors, the ECB’s main refinancing operations rate become significant, but when we add the global factors, this disappears. Nonetheless, its importance cannot be downplayed, which might be explained by the fact that during the crisis the ECB heavly invested in banks by expanding their balance sheets through refinancing operations (e.g.: LTRO) which due to the similar bank present in Hungary as in the eurozone countries, might have also affected the operations of bank here, what for the HCB could have reacted to. However, these are just speculations, to give a definite answer why the main refinancing operations were significant influencing factors, we would need a separate analyses for that. However, more interestingly when we omit some of our variables, out of the three regional variable two becomes significant reiterating an ECB papers finding, that policy setting in the ECB may influence non-eurozone EU member states’ monetary policy. However, due to the lack of economic grounding of this computation we cannot conclude, especially since the reserve requirement is not significant, that the regional factors influence the

As for the global factors, for both of our regression models there is only one factor that remains significant by the end of our 4th_{phase. In case of the interest rate, this is the euro}

exchange rate, and for the reserve requirements this is the VIX index. We can just suspect that the VIX index significance can be connected to capital buffer building by banks, assigned by the central banks, but as mentioned before the euro’s significance can be connected to the two HCB decision as a result of exchange rate fluctuation.

In both of the internal measures we were unable to confirm the Taylor rule approach of the HCB in practice, since inflation and GDP growth had a no significant effect on the reserve requiremet, nor on the interest rates. So to answer the (1), the Hungarian Central Bank does not follow the Taylor rule.

(33)

33

factors influenced the reserve requirements. So we can reject that the domestic factors would influence the monetary policy decisions.

In case of hypothesis question (3), we once again have to reject the notion that regional factors would influence the HCB’s monetary policy decisions (thus make it less independent). The main refinancing operations rate might be significant in case of the interest rate, however, most variables, in the two measured dependent variables are not. In case of the omitted variables, the significance of these variables in the 4th_{phase increases in the interest rate}

regression, but it is not a solid and definite finding. Therefore we have to acknowledge that the HCB is independent from the regional spillovers as well, but we also acknowledge that to a certain degree it may influence it.

As for answering (4) hypothesis question, we have to reject our notion, that global spillovers would affect the monetary policy decision making of the HCB, as just only in the case of the reserve requirement is the VIX index significant, however, no other variable is. This leads us to conclude that the HCB is independent from the global spillovers.

This paper has many aspects that could be followed upon, further specified or expaneded yet we believe that this field is realtievly underresearched. One aspect that could be improved is for example the dataset from Rey’s global financial cycle paper, that only extends till 2012 Q4 to be expanded till 2018 Q4, in line with other variables. Furthermore a recent paper by Goczek and Mycielska suggest to measure the monetary policy independence through the short term bonds of EURIBOR and BUBOR to be included in as variables, as they did with the Polish short-term bonds (Goczek et Mycielska, 2017). But most importantly, what was also reaffirmed in the first regression, that as we added more variables, so did our results come clearer, and our constant less significant. By this we can conclude that additional variables helped to understand what influences our dependent variables, however, more could have been found, that remains to be identified.

(34)

34

8. Appendix:

Autocorrelation:

(35)

(36)

36

(37)

37

(38)

38

(39)

39

(40)

40

(41)

41

(42)

42 Dfuller tests:

Table 19 Dfuller test of the ECB's main refinancing operations rate. We can see that it passes the critical value to pass the test.

Table 20 The Dfuller test of global market volatility. We can see that it passes the critical value so it passes the test, meaning there is no unit root trend in it.

(43)

43

Table 22 GDP growth not passing the Dicky-Fuller test as it does not pass the critical value.

Table 23 Euro exchange rate not passing the Dicky-Fuller test as it does not pass the critical value.

(44)

44

Table 25 Government deficit not passing the Dicky-Fuller test as it does not pass the critical value.

Table 26 Monetary financing of public debt not passing the Dicky-Fuller test as it does not pass the critical value.

(45)

45

Table 28 ECB's marginal lending rate not passing the Dicky-Fuller test as it does not pass the critical value.

Table 29 FED effective interest rate not passing the Dicky-Fuller test as it does not pass the critical value.

(46)

46

Table 31 EU Banking leverage not passing the Dicky-Fuller test as it does not pass the critical value.

Table 32 US banking leverage not passing the Dicky-Fuller test as it does not pass the critical value.

(47)

47

Normality tests of the variables through histograms:

Table 34 Histogram of the ECB's main refinancing operations rate.

(48)

(49)

(50)

(51)

(52)

(53)

(54)

(55)

(56)

56

(57)

(58)

(59)

(60)

60

Summary statistics after variable transformation:

(1) (2) (3) (4) (5)

VARIABLES N mean sd min max

ldexEUR 78 0.929 8.135 -15.36 32.22 lecb_mro_i 79 -0.158 0.290 -0.750 0.500 lglo_mkt_vol 56 137.1 46.98 37.01 248.6 d_i 79 -0.191 0.842 -2.500 3 d_resreq 79 -0.139 0.593 -4 0 ldinf 78 -0.0773 0.947 -2.451 2.269 ldgdpgr 78 0.0336 1.246 -4.642 3.854 ldexUSD 78 0.696 12.19 -20.49 42.16 ldgovdef 78 0.0373 0.931 -4.776 4.253 ldmfpd 78 -5.273 45.97 -219.5 237.1 ldvix 78 -0.185 5.898 -13.60 33.52 ldecb_ml_i 78 -0.0545 0.380 -2.250 1 ldfed_i 78 -0.0367 0.447 -1.650 0.680 ldglob_crosscred 56 279.9 873.5 -1,874 2,271 ldEU_banklev 56 0.00307 0.0303 -0.0664 0.0972 ldUS_banklev 56 -0.00268 0.0183 -0.0907 0.0386 ldglob_rk_av 47 2.353 17.85 -26.48 75.83

Post crisis regressions of the interest rate:

(1) (2) (3)

VARIABLES Taylor Factors Internal Factors Regional Factors

(61)

61 ldgovdef -0.00796 0.0717 (0.936) ldmfpd 0.000808 0.00433 (0.462) o.ldecb_ml_i - o.lecb_mro_i - o.ldEU_banklev - ldinf 0.180*** 0.213*** (0.00582) (0.00264) Constant -0.219*** -0.220*** -0.585 (6.14e-05) (0.000204) Observations 28 28 6 R-squared 0.304 0.431 1.000 pval in parentheses *** p<0.01, ** p<0.05, * p<0.1

Post crisis regressions of the reserve requirements:

(1) (2) (3)

VARIABLES Taylor Factors Internal Factors Regional Factors

(62)

62 ldexUSD 0.000609 0 (0.891) ldgovdef 0.0306 0 (0.697) ldmfpd -0.00207** 0 (0.0243) o.ldecb_ml_i - o.lecb_mro_i - o.ldEU_banklev - ldinf 0.0188 -0.00396 (0.704) (0.937) Constant -0.0380 -0.0495 0 (0.319) (0.215) Observations 28 28 6 R-squared 0.009 0.248 pval in parentheses *** p<0.01, ** p<0.05, * p<0.1 Stata Commands:

import excel "F:\Géza PC- 2017.03.24\Dokumentumok\MSc Szakdoga\adatok\data-for-all-variables_quarterly.xlsx", sheet("Alapkamat") cellrange(A1:R81) firstrow clear

. gen n = _n

(63)

63

summarize

outreg2 using summary.doc, replace sum(log)

. gen linf =L1.inf

. gen lgdpgr = L1.gdpgr

. gen lexEUR = L1.exEUR

. gen lexUSD = L1.exUSD

. gen lgovdef = L1.govdef

. gen lmfpd = L1.mfpd

. gen lvix = L1.vix

. gen lecb_ml_i = L1.ecb_ml_i

. gen lecb_mro_i = L1.ecb_mro_i

. gen lfed_i = L1.fed_i

gen lglob_crosscred = L1.glob_crosscred

. gen lEU_banklev = L1.EU_banklev

Monetary policy independence of the Hungarian Central Bank (HCB) considering global and regional spillovers