1
An Application of the Taylor Rule: Monetary Policy in Italy and
the Netherlands 1985-2014
Bachelor Thesis Economics
Laura Wollny
10390510
2
Table of Contents
1. Introduction ... 3
2. Literature Review... 4
2.1 Euro zone and the monetary policy of the ECB ... 4
2.2 The Taylor Rule ... 7
3. Data and Methodology ... 10
3.1 The Model ... 10
3.2 Methodology ... 11
3.3 Sample Selection and Data Collection ... 13
3.4 Data Description ... 14
3.5 Hypothesis ... 16
4. Results ... 17
4.1 Results for Italy ... 17
4.2 Results for the Netherlands ... 18
5. Analysis ... 19
5.1 Analysis of the Results 1985-1998... 19
5.2 Analysis of the Results 1999-2014... 21
5.3 Limitations ... 24
6. Conclusion ... 25
Bibliography ... 27
3 1. Introduction
It can be called an experiment when 11 out of 15 European Union (EU) member states decided to deepen their economic integration further by creating a single European currency in 1999. The National Central Banks (CB) handed over their sovereignty to the newly created European Central Bank (ECB) which sets the monetary policy for the entire Euro zone ever since (Faust, Rogers, & Wright, 2001, p. 1). Meanwhile, the world has experienced a worldwide financial crisis that ultimately led to a sovereign debt crisis in several European countries (Nechio, 2011, p. 4). As a consequence of the newly established interconnections, all member states are affected and have to cooperate to better the economic situation (Ali, 2012, p. 428). Nevertheless, tensions are rising between the national governments. Cohen even speaks of a crisis of the European Union (2012, pp. 693-694).
The main objective of the ECB is to ensure price stability which is defined by an inflation rate below but close to 2% over the medium term (Sauer & Sturm, 2007, p. 377). The prime instrument for that are the short-term interest rates set by the ECB (ECB, 2011, p. 93). With the turmoil on the financial markets and the European debt crisis evolving, the interest rate setting decisions of the ECB have been in the focus lately (Moons & Van Poeck, 2008; Nechio, 2011). Currently, the interest rates are at historically low levels (Nechio, 2011, p. 1). Furthermore, the economic conditions of the Euro zone members differ widely, from relatively stable states e.g. Germany to states which heavily rely on outside support e.g. Portugal, Ireland, Italy, Greece and Spain, commonly known as PIIGS (Ali, 2012, p. 425). However, the ECB takes the data of the entire Euro zone and not the single states into account
for their decisions (ECB, 2011, p. 69).
This brings us to the central question of this paper: How well do the Euro zone interest rates fit its individual member states? Phrased differently, does monetary policy deviate more or less from the Taylor Rule since the introduction of the Euro zone?
To answer this question, I will compare the Taylor Rule recommendations with the actual interest rates for the periods before and after the creation of the Euro zone. The Taylor Rule is a simple monetary policy rule for CBs which relates the target interest rates to the inflation level, the production and an equilibrium interest rate (Taylor, 1993, p. 202). Taken that the Taylor Rule is the policy the ECB should follow, it is possible to determine if the introduction of a single European monetary policy suits the individual member states. Several studies (Belke & Klose, 2010; Belke & Polleit, 2007; Fourçans & Vranceanu, 2004; Moons & Van Poeck, 2008; Nechio, 2011) investigate the ECB monetary policy after 1999. Moons and Van Poeck (2008) as well as Nechio (2011) compare the fit of recommended interest rates for
4
different Euro zone member states according to their economic conditions with the ECB’s common interest rate. However, Belke and Polleit (2007) are reluctant to study the European strategy before 1999 because they find it to be inconsistent in terms of the monetary policy. In this paper the time period of 1985 until 2014 is considered, thus, the paper adds to the
discussion by extending the time period. In this paper the focus will lie on two countries: Italy and the Netherlands. Both countries are founding members of the Euro zone. But while Italy experienced high public debt and belongs to the group of Euro zone countries that struggle with the European debt crisis (Ali, 2012, p. 425), the Netherlands form a group with stable economies such as Austria and Germany (Cohen, 2012, p. 690)
The paper will be organized as follows. The next section will give an overview of the central literature in the field. This will extend the discussion on the Euro zone, monetary policy and the Taylor Rule. After that the data usage and methodology will be explained in detail. The subsequent sections contain the results and the analysis of them. The paper concludes with a short summary of findings.
2. Literature Review
2.1 Euro zone and the monetary policy of the ECB
Today, the Euro zone consists of 19 member states. The latest accession was Lithuania in the beginning of 2015 (ECB, 2015). Faust et al. call the integration process an “experiment without precedent” (2001, p. 1). And indeed, applying Appleyard, Field and Cobb’s levels of economic integration, the Euro zone is the only example of a monetary union which is defined by a single currency and a common monetary policy under a common CB (2009, p. 394). As a consequence of the high level of integration, the ECB states that the Euro zone can be
compared to the large economies of Japan or the United States (ECB, 2011, p. 29). DeGrauwe (1997) and Gros and Thygesen (1998) emphasize that the ECB has a higher level of
independence than the Fed, because of its cross-border combination and collective decision making process (as cited in Faust et al., 2001, p.2).
The monetary policy of the ECB is decided by the governing council which includes the six executive-board members and representatives of all member National CBs (ECB, 2011, p. 19). According to the ECB, the objective of price stability avoids high costs of inflation or deflation and secures economic growth (ECB, 2011, p. 56). The ECB targets price stability by changing short-term interest rates which in turn influence the money market rates. Through a sophisticated process of agents’ actions and their expectations, the so-called monetary transmission mechanism sets the real economic conditions. However, according to
5
the theory of the long run neutrality of money, prices are affected but output and employment are not permanently affected (ECB, 2011, pp. 62-63). Therefore, the ECB focuses on price stability which according to the theory can be influenced and not on output or employment levels.
In order to join the Euro area, a candidate state has to fulfill a number of conditions, known as the convergence criteria. This is because the country loses is possibility to conduct an independent monetary policy, such as exchange rate and interest rate adjustments, in reaction to changing economic conditions. However, large economic differences in the individual member states exist (Nechio, 2011, p. 1).
While the ECB (2011, p. 52) states that inflation and output growth variances are reasonable in relation to other large currency areas, Fendel and Frenkel (2009, p. 1293) find that the inflation differentials are noteworthy. They explain that the term inflation differential describes the difference between annual national inflation rates and the Euro zone average (2009, p. 1294). Nonetheless, ECB (2011, p. 52) and Fendel and Frenkel (2009, p. 1294) agree that the inflation differentials have decreased considerably due to the convergence criteria. For the two countries considered in this paper, the inflation in 2014 was 0.2% in Italy and 1% in the Netherlands compared to 5.2% and 1.9% respectively in 1995 (OECD). The overall Euro zone inflation was at 0.4% in 2014 (Eurostat) .Thus, currently being below the targeted 2%. Already before the creation of the Euro zone, DeGrauwe wondered how low inflation states would be convinced to be united with high inflation states. He predicted that the ECB will have to focus on price stability even more than the German CB, the
Bundesbank, which firmly focused on low inflation. (1997, as cited in Faust et al., 2011, pp. 1-2). In reality, the Harmonized Index of Consumer Prices (HICP) used by the ECB is a weighted average of inflation values of all member states (Fourçans & Vranceanu, 2004, p. 584). Fendel and Frenkel show that inflation differentials influence the ECB’s decisions. The ECB considers inflation differentials to be addressed by the overall objective of price stability (2009, pp. 1293-1296). In contrast to that, they conclude that with large differences in
inflation the ECB was cautious to fight an total inflation gap due to the risk of deflation in the states with currently low inflation (2009, pp. 1301-1302).
Cohen states that the differences between the founding members of the Euro zone were apparent. He names not only inflation differentials but also the development level and the asymmetric reactions to economic shocks. Furthermore, he mentions that although Italy and Belgium did not fulfill the public debt convergence criteria, they were admitted to the Euro zone (2012, p. 691). Additionally, Björkstén and Syrjänen (1999, p. 9) recognize
6
differences in six variables such as the GDP growth rates, the debt-to-GDP ratio and the fiscal balance. In 2010, the real GDP growth rates were at -4.5% for Greece and in contrast to that, at +3.7% for Germany (Ali, 2012, p. 428). In addition, Björkstén and Syrjänen find that the differing variables are correlated (1999, p. 15) Moreover, investors perceive risk of the single Euro zone states differently. This can be seen from the large interest rate spreads compared to the secure German bonds (ECB, 2011, p. 128). The diverse economic conditions across the Euro zone member states are reinforced by the current crisis, according to Ali. He says that the crisis reveals different problems in the member states such as Spain’s housing bubble and Greece’s social unrest due to poor public finances, but generally it is described by high levels of public debt in the Euro zone states (2012, p. 427). The ECB reacted to the severe
consequences for the money market by decreasing the interest rates to secure liquidity and strengthen investor confidence. In addition, non-standard measures such as quantitative easing, which is an asset purchase program to stimulate the economy despite the low current interest rates, were introduced (ECB, 2015).
With the crisis the divergence of Euro zone members has become problematic, according to Cohen. He says that the divergence of the Euro zone states is given by the vast economic differences of the member states e.g. their differences in balance of payments accounts (2012, pp. 691-692). The question emerges whether the ECB’s common interest rates are the correct rates for the individual states in reaction to the changing economic
conditions.Nechio (2011, p. 3) estimates the optimal interest rates for several Euro area states
and finds that the recommendations differ with the various levels of inflation and
unemployment. Moons and Van Poeck (2008, pp. 195-196) agree with Nechio. They find large gaps for the optimal and actual interest rates for some states e.g. the Netherlands , in particular states with high inflation compared to the average. Nechio (2011, p. 3) mentions that states that are struggling with a sovereign debt crisis would need lower rates to service their debt obligations. Further, inappropriate interest rates can destabilize the economy and deepen the economic recession (Moons & Van Poeck, 2008, p. 196). Cohen (2012, p. 692) and Ali (2012, pp. 428-429) see the problem of the Euro zone crisis in the numerous policies. As noted before, the Euro zone has one monetary policy but numerous fiscal policies (ECB, 2011, pp. 14-15). Ali and Cohen think that the solutions to the crisis in the Euro zone are an increased coordination of fiscal policy, structural reforms and a tighter regulation of the financial sector. Ali (2012, p. 427) emphasizes the fact that there are nevertheless countries as new candidates looking to join the Euro zone.
7 2.2 The Taylor Rule
In 1993, Taylor developed a monetary policy rule that can be used as a broad benchmark by policymakers. According to him, a monetary policy rule is preferable to discretion because it avoid the time inconsistency problem (1993, pp. 198-199). The time inconsistency problem refers to the revocability of discretionary monetary policy and the problems that it brings in terms of investor confidence in the CB (Kydland & Prescott, 1977, p.474). In his paper, Taylor (1993, p. 200) assumes that the instrument of CBs is the adjustment of interest rates which, according to Borio, is used by most CBs in developed countries (1997, as cited in Fendel & Frenkel, 2009, p. 1296). Furthermore, Clarida, Galí and Gertler attribute the success of monetary policy to inflation targeting (1998, p. 1034). In practice, this means that the CB reacts to changes in economic variables such as the production level, the money growth, the exchange rate or the inflation rate and adjusts the interest rates accordingly (Taylor, 1993, p. 200). The rule which Taylor found describes the nominal target interest rate as a function of the real equilibrium interest rate, inflation and output (1993, p. 202). The formula and its implications will be discussed in more detail in the the data and methodology section.
Taylor finds that the benchmark developed by him fits reasonably well for the Federal Reserve (Fed), the CB of the United States, in the period of 1987-1992 (1993, p. 202). Other scholars (Belke & Polleit, 2007; Clarida et al., 1998; Judd & Rudenbusch, 1998) have extended Taylor’s research and investigated the fit of the rule in different periods for the Fed. Their results correspond to the Taylor’s opinion that his rule captures the most important elements of monetary policy for the Fed. Clarida et al. (1998, p. 1050) confirm Taylor’s findings. They show that the Fed’s monetary policy between 1979 and 1994 can be described by the Taylor Rule. Belke and Polleit (2007, p. 2203) also find that the Taylor Rule fits the Fed in the period 1987-2005 relatively well. However, they include money growth and the dollar-euro exchange rate as variables to improve the fit. In contrast to that, Clarida et al. (1998, pp. 1049-1050) find that money growth was only significant in a short period of time in which non-borrowed reserves were targeted. Moreover, Judd and Rudenbusch (1998, p. 4) estimate the fit with regard to three different Fed chairmanships. From 1970 until 1998 the monetary policy can be described by a form of the Taylor Rule. But the estimated parameters and the fit deviate with the different chairmen (Judd & Rudenbusch, 1998, p. 6). Besides, the actual interest rates have been proven to follow the Taylor Rule in the United Kingdom since 1992 (Davradakis & Taylor, M., 2006, p. 11), in Japan from 1979 to 1994 (Clarida et al., 1998, p. 1050) and in the seven largest Latin American countries (Moura & de Carvalho, 2010, p. 403).
8
Furthermore, the Taylor Rule is widely used to describe the ECB’s monetary policy. Before the introduction of the Euro, Clarida et al. (1998, p. 1034) advised the future ECB to follow an inflation targeting practice as followed successfully by the Bundesbank prior to 1999. The German CB’s monetary policy was seen as a benchmark for the other European countries (Clarida et al., 1998, p. 1042). However, this statement is refuted by Faust et al.. In their study, the Taylor Rule is used to approximate the behavior of the ECB on the basis of the Bundesbank monetary policy. They find that the Bundesbank’s reaction function follows the Euro zone interest rates only until the introduction of the Euro. Afterwards, the estimated function predicts higher interest rate than the ones observed (2001, p. 9). Peersman and Smets find empirical evidence that the simple Taylor Rule stabilizes inflation and output for a weighted average of the five states; Germany, France, the Netherlands, Austria and Belgium. They recommend the Taylor Rule for the future use of the ECB (1999, p. 106).
Yet there is some critique on the application of the Taylor Rule. First of all, Faust et al. (2001, p. 6) and Peersman and Smets (1999, p. 86) state that CBs do not explicitly describe their behavior with the Taylor Rule. Peersman and Smets expound that the Taylor Rule is limiting in a sense that too little variables are used and the market mechanisms is simplified too much for policymakers to commit to such a straightforward rule (1999, pp. 86-87). Kozicki questions whether the Taylor Rule can be used if the decision making process is not captured by it. She explains that the estimations for the Taylor Rule do not fit the actual
interest rates in a time of successful monetary policy (1999, p. 17).Moreover, she criticizes
that the Taylor Rule produces significant deviations when small variations exist in the
variables, i.e. if different measures are used for the variables or other coefficients are assigned (1999,p. 24). This view is contrary to Peersman and Smets (1999, p. 88) who find that
variations in the estimated parameters are not significant. Kozicki concludes that the Taylor Rule cannot be followed by the policymakers (1999, 24).
Nevertheless, Peersman and Smets find that the Taylor Rule has two major
advantages. First, it is very straightforward in its implications. Secondly, it does not rely on a forecasting model. They think that its use as a standard and a communication instrument for monetary policy is justified (Peersman & Smets, 1999, p. 87). Despite her criticism, Kozicki also finds that the Taylor Rule depicts valuable characteristics of monetary policy (1999, p. 6). Faust et al. (2001, p. 6) and Judd and Rudenbusch (1998, p. 3) come to a similar
conclusion. They describe the Taylor Rule as a straightforward empirical mechanism for monetary policy that helps to make forecasts for future actions of the CB and helps evaluating monetary policy and its alternative measures. These findings are in line with Taylor (1993)
9
himself. He acknowledges that a monetary policy rule cannot be taken as a rigid rule which has to be followed blindly. Therefore, the name Taylor Rule can be misleading. He thinks that inheriting the general characteristics of his rule is valuable (1993, pp. 208-209).
Finally, Svenson finds evidence that the Taylor Rule is the best backward-looking reaction function a CB can use to target inflation (1999, as cited in Sauer & Sturm, 2007, p. 378).
Previous studies (Fourçans & Vranceanu, 2004; Sauer & Sturm, 2007) have estimated the monetary policy rule of the ECB. The ECB’s monetary policy is based on two factors:
monetary and economic analysis. Monetary analysis is the assessment of inflation as a monetary phenomenon in relation to money growth. Economic analysis includes
consideration of the overall price stability (ECB, 2011, p. 10). Sauer and Sturm imply that these factors justify the use of the Taylor Rule for the ECB (2007, pp. 377-378). They evaluate the best fitting Taylor Rule for the ECB and find it to be a forward-looking specification (2007, p. 393). The issue of forward-looking Taylor Rule is introduced by
Clarida et al.. A forward-looking approach means that expected values for inflation and output are used which according to them is what policymakers do (1998, pp. 1037-1038). This
approach is followed by e.g. Fendel and Frenkel (2009), Sauer and Sturm (2007) or Peersman and Smets (1999). Besides, Fourçans and Vranceanu also estimate the Taylor Rule for the ECB and find it applicable to the ECB. Yet, in their study, adding the dollar-euro exchange rate as an additional variable in the formula improves the accuracy (2004, p. 590). Belke and Polleit add to the previous results by investigating the fit of the Taylor Rule. They find that an adjusted form with variables such as money growth and exchange rate applies for the ECB in 1999-2005. However, Belke and Polleit also mention that the Taylor Rule fits the Fed’s monetary policy better than the ECB’s (2007, pp. 2207-2208). In addition, other questions have been investigated by applying the Taylor Rule. For example, Belke and Klose
approximate the differences in the monetary policy of Fed and EBC before and during the crisis with the Taylor Rule. Their finding is that the ECB in contrast to the Fed focused primarily on inflation fighting and neglected the importance of the output gap (2010, p.19). Furthermore, they highlight the importance of the inclusion of additional variables during exceptional circumstances such as the current financial crisis (2010, p. 6). Another instance of the various applications of the Taylor Rule is the study of Fendel and Frenkel. They use the Taylor Rule to examine the importance of inflation differentials in the decision making of the ECB (2009, p. 1297). While the previously mentioned studies lay the foundation for the application of the Taylor rule, the following two papers are closely related to the research in this paper. As mentioned before, Moons and Van Poeck (2008) and Nechio (2011) investigate
10
differences in the fit of single Euro zone states with the common monetary policy with the help of the Taylor Rule. Nechio finds that the monetary policy of the ECB fits the Euro zone since 2005 as a whole but not necessarily to the peripheral states (2011, p. 4). She defines the peripheral states as the states being the European states that have been severely hit by the sovereign debt crisis, namely Greece, Ireland, Portugal and Spain (Nechio, 2011, p. 3). Moons and Van Poeck share this view. According to them the precision for the individual states is low and they do not expect a convergence effect in the future (2008, p. 198).
Similarities and Differences between these papers and this study will be discussed in the next section.
3. Data and Methodology
Having explained the theoretical background on the ECB’s monetary policy and shown examples of the Taylor Rule in the literature review, the following section extends the discussion on the application of the Taylor Rule. As in previous studies, the Taylor Rule will be used as the empirical method to compare the fit of the recommended interest rates of the individual states, namely Italy and the Netherlands, with the actual monetary policy.
Furthermore, this section explains the set-up of the empirical research.
3.1 The Model
Mathematically, the Taylor Rule is described by the following:
it = πt + r* + α (πt – π*) + β (yt)
Where it is the nominal short-term interest rate which the ECB can influence, πt the current
inflation and the r* the real equilibrium interest rate. The third term describes the inflation
gap, the difference between actual and target inflation. Yt is a measure for the output gap
which is specified as the difference between actual and potential output as a percentage of potential output. Alpha and Beta are coefficients which Taylor both assigned to be 0.5 (Judd & Rudenbusch, 1998, p. 5).
It follows that if the inflation is above its target the Taylor Rule recommends to increase the interest rates. The next variable, the real equilibrium interest rate, takes a number of factors into account such as the inflation rate but also the import price inflation (Belke & Klose, 2010, p. 27). Most studies (Kozicki, 1999; Moons & Van Poeck, 2008; Sauer & Sturm, 2007) including Taylor (1993), in which the Taylor Rule is not empirically estimated, assume it to be at 2%. Another important part of the model is the coefficients. There are said to be two
11
conditions with respect to the coefficients of inflation and output gap. The so-called Taylor Principle implies that the coefficient on inflation needs to be larger than one to stabilize the economy. An inflation coefficient smaller than one does not influence the real rate sufficiently and would enhance inflation further (Fourçans & Vranceanu, 2004, p. 583). In contrast to that, the coefficient on the output gap is only supposed to be positive to counteract negative
developments (Sauer & Sturm, 2007, p. 380). Furthermore, a negative output gap, meaning actual output being below its potential level, implies a decrease of the interest rate. The
formula also indicates that inflation and output gap can contradict each other. This means they need to be balanced. Kozicki notes that the inflation gap can also stand for the long-run goal of price stability and the output gap for short run adjustment to cyclical conditions.
Alternatively, a positive output gap can be interpreted as future increases of inflation. Therefore, the inclusion of the output gap is a way to address these concerns for the future development of the inflation level (1999, p. 7). This theory on the output gap is related to macroeconomic models such as the IS and the Phillips curve and the sophisticated monetary transmission mechanism. The adjustments of the interest rate ultimately influence real economic conditions (Kozicki, 1999, p. 6). A discussion of this aspect is behind the scope of this paper. It will be assumed that the Taylor rule provides the optimal answer to economic conditions to reach the objective of price stability.
3.2 Methodology
The Taylor Rule is used to analyze the differences between the pre-Euro zone and the Euro zone monetary policy fit for Italy and the Netherlands. Eventually it will show whether the Taylor Rule deviates more or less from the actual interest rates since the introduction of a single monetary policy for the Netherlands and Italy. From the discussion above we can conclude that the Taylor Rule is the appropriate form for the analysis. Even if the ECB
explicitly focuses on price stability and not on full employment an alternative interpretation of the output gap can justify the application of the Taylor Rule for the ECB. According to Sauer and Sturm (2007, p. 378) the output gap as a forecaster of inflation allows the use of it in the Taylor Rule for the ECB. Furthermore, the ECB does not describe their monetary policy with the Taylor Rule, but previous researchers (Judd & Rudenbusch, 1998; Peersman & Smets, 1999; Taylor, 1993) agree that the Taylor Rule is a important tool to evaluate monetary policy. Moreover, similar questions to the one considered in this paper, have been investigated with the Taylor Rule (Moons & Van Poeck, 2008; Nechio, 2011).
12
However, numerous specifications of the Taylor Rule exists. One commonly made change in the functional form is to include additional variables, as mentioned before, researchers have come to different results. Another aspect is the time frame which is being used. Clarida et al. (1998, p. 1037) introduced a forward-looking Taylor Rule. Faust et al. (2001), Fendel and Frenkel (2009) and Sauer and Sturm (2007) follow this approach. Nevertheless, there is some dissent on this issue. While the researchers emphasize the advantages of a forward-looking form, they use different moments of when the data is reported. There is ex ante and ex post measurement, ex post is reported later and can be reviewed before, while ex ante is the information available before the decision. Faust et al. (2001, p. 11) , Belke and Klose (2010, p. 9) and Sauer and Sturm (2007, p. 376) prefer ex ante data because it is said to produce more precise results and be more realistic with respect to the data policymakers have at hand. Nevertheless, Belke and Polleit (2007, p. 2198) use ex post data in the forward-looking form of the Taylor Rule. In addition, Judd and Rudenbusch (1998, p. 6) state that they find results similar to the study of Clarida et al. (1998), but they use the backward-looking form. Another reason not to use a forward-looking model is pointed out by Peersman and Smets. They remark that an important advantage of the Taylor Rule is that it does not require forecasting model (1999, p. 86). Finally, the issue of interest rate smoothing is disregarded in the model. Interest smoothing describes the gradual fashion in which CBs tend to change their interest rates. Reasons for the smoothing of interest rates can be an ongoing learning process or the concern to disrupt the bond and equity market with sudden changes (Faust et al., 2001, pp. 5-6). The topic is disregarded in this paper since the long term structural trend is investigated and not the best fit of the actual monetary policy with the Taylor Rule. Several researchers (Belke & Polleit, 2007; Fendel & Frenkel, 2009; Judd and Rudenbusch, 1998; Sauer & Sturm, 2007) smooth the interest rates. On the other hand, Mishkin adds to the discussion by mentioning that the interest rate smoothing changes during times of crisis (as cited in Belke & Klose, 2010, p. 10).
Previous studies do not reach an overall consensus on which functional form to use and which additional variables to add. Moreover, it is not established how interest rate smoothing behaves during times of the recent crisis which is nevertheless part of the sample. Another concern is that the studies do not solely focus on the Euro zone. Thus, the papers on the functional form of the Taylor Rule are not comprehensive. It is not practical to use one of these papers as a standard for my paper. Therefore, I would have to conduct my own
econometric estimation of the Taylor Rule. However, the sophisticated estimation methods and the data handling are beyond the scope of my paper. Thus, focusing on the original
13
research question of this paper it is sufficient to apply the original form of the Taylor Rule. Furthermore, Moons and Van Poeck (2008, p. 194) and Nechio (2011, p. 1) use a simple form of the Taylor Rule for their research to investigate the fit of Euro zone economic conditions with the monetary policy of the ECB. Their research is closely related to my own research. While both studies are focused on Euro zone member states, Moons and Van Poeck (2008, p. 193) extended their study towards Euro zone candidates. However, their research is aimed at the time period after the establishment of the Euro zone, this paper is going to study two single Euro zone member states and their fit of the actual monetary policy with the Taylor Rule before and after the introduction of the Euro zone.
The real equilibrium interest rate of 2% is derived from previous papers. Also given is the ECB’s inflation target of 2%. Using the simple functional form of Taylor and the same coefficients, the formula looks like this:
it = πt + 2% + 0.5 (πt – 2%) + 0.5 (yt)
which can be rewritten as:
it= 1+ 1.5πt + 0.5yt
Assuming this form the Taylor principle is fulfilled. Differences in inflation levels and output gaps will lead to different Taylor Rule recommendations. However, as mentioned before the ECB determines the interest rates with respect to the economic conditions in all Euro zone member states.
The set-up of the research is as follows. First, the interest rates recommended by the Taylor Rule are calculated. Then, the difference between actual interest rate and the Taylor Rule recommendation is computed. By diving the sample into two subsamples, namely before and after the creation of the Euro zone, the weighted average of the differences can be
compared. Furthermore, two robustness checks are applied.
3.3 Sample Selection and Data Collection
The empirical research covers a time frame from 1985 to 2014. This is because the data for the output gap is available until 1985. The research will follow Taylor’s (1993) approach and report the data quarterly with an annual percentage change.
For the Taylor Rule recommendation data on the inflation rate and the output gap is required. The data on inflation is obtained from the OECD. More specifically, it is data on consumer prices of all items and is measured as a annual percentage change and reported
14
annual percentage change. To match the inflation measurement, I approximate the quarterly output gap by dividing the annual value by four and calculating the annual percentage change for every quarter by adding the previous four quarter figures. Otherwise, the sample would contain the same data points for four periods. This is not an econometric method of an approximation, however, it is sufficient for this paper. Other measures of the output gap are beyond the scope of this paper. Furthermore, Peersman and Smets find that estimation errors of the output gap do not significantly alter the performance of the Taylor Rule (1999, p. 106). Due to the data on the output gap the values for the first three quarters of 1985 cannot be reported. Thus, the sample size consists of a total of 117 observations, 53 quarters before the creation of the Euro zone and 64 quarters after 1999.
For the observed interest rates, I take the last reported value of the respective quarter. Assuming, that the policy is a reaction to the current economic conditions. Therefore, I apply Taylor’s backward-looking form. For Italy the interest rate is the official discount rate
obtained from Italy’s CB, the Banque d’Italia. They report the values in percent on a monthly basis on the last day of the month. The Dutch interest rate used is the fixed advance rate which is reported irregularly over the years by the Dutch CB, De Nederlandsche Bank. Furthermore, for the Euro zone, starting in 1999, I follow the studies of Fendel and Frenkel (2009), Moons and Van Poeck (2008) and Sauer and Sturm (2007) which take the Euro Overnight Index Average (EONIA) to compare the Taylor Rule with the actual interest rates. It is obtained from Eurostat. The EONIA fluctuates around the main refinancing rate and between a band of the marginal lending and deposit rate, the other key interest rates of the ECB (ECB, 2011, p. 100). An advantages of the EONIA is that it is reported on a regular basis.
3.4 Data Description
In this section the data used for the empirical analysis will be shown in its context. The data shown is the same as described in the previous section.
15 Figure 1. Inflation rates for Italy and the Netherlands
The inflation level for the Netherlands stayed below the Italy’s inflation level until 1997, since then the graph show that the inflation rates of the two countries are converging. It can also be seen that the inflation fluctuates around the 2% target rate but shows large
deviations as well. The inflation rate of the Netherlands is currently below the ECB’s target rate for inflation of 2%. Nevertheless, Italy has even experienced deflation with -0.1% in the third quarter of 2014. Therefore, in the sample the inflation gap for Italy has been positive in the beginning in contrast to the Netherlands. This is turn means that the inflation has been above the targeted 2% for Italy and below for the Netherlands. Since 1997 the inflation fluctuates closer around the targeted 2% but has decreased recently for both countries.
Figure 2. Annual output gaps for Italy and the Netherlands -2 0 2 4 6 8 10 Q1 -1985 Q1 -1987 Q1 -1989 Q1 -1991 Q1 -1993 Q1 -1995 Q1 -1997 Q1 -1999 Q1 -2001 Q1 -2003 Q1 -2005 Q1 -2007 Q1 -2009 Q1 -2011 Q1 -2013 inflation Italy inflation Netherlands -8 -6 -4 -2 0 2 4 6 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015
Output gap Italy Output gap Netherlands
16
The annual output gaps of Italy and the Netherlands exhibit a co-movement. The largest gap between the two countries output gaps was in 2004 with a difference of 3%. The output gap has been constantly negative since 2009 for Italy and since 2012 for the
Netherlands. However, Italy’s output gap is below the one of Netherlands, being at -5.9% and -3.2% respectively. The negative output gap indicated that the two countries produce output below their actual potential. Both output gaps are expected to rise in 2015 and 2016.
Figure 3. Actual interest rates for Italy and the Netherlands, since 1999 ECB rates
The interest rates differ largely for Italy and the Netherlands. The Italian interest rates were well above the Dutch rates. In 1985 Q3 the difference between the interest rates of the two countries was as large as 9.5%. While the Italian interest rates show an overall downward movement, the Dutch rates are less consistent. From 1999 onwards both countries use the ECB interest rates. The ECB interest rates connect to the previous interest rates of the single Euro zone member states smoothly. The ECB interest rates are approximately between 2% and 4% until the end of 2008. Since 2009 the interest rates have decreased substantially and even reached negative values.
3.5 Hypothesis
The research question of this paper is whether the implementation of a single monetary policy in the Euro zone has changed the fit of the actual interest rates with the Taylor Rule in Italy and the Netherlands. Looking at the discussion, I expect to find that the interest rates
recommended by the Taylor Rule deviate more from the actual interest rates since the creation of the Euro zone because the Euro zone states lost their possibility to conduct an independent
-2 0 2 4 6 8 10 12 14 16 18 1985 Q1 1987 Q1 19 89 Q1 1991 Q1 1993 Q1 1995 Q1 1997 Q1 1999 Q1 2001 Q1 2003 Q1 2005 Q1 2007 Q1 2009 Q1 2011 Q1 2013 Q1 interest rates IT interest rates NL interest rates EZ
17
monetary policy. Moons and Van Poeck (2008, p. 198) and Nechio (2011, p. 4) describe that the monetary policy of the ECB does not match the economic conditions of the single Euro zone states particularly well. Given that they use the Taylor Rule for a similar analysis, I am expecting to find similar results, taking into account that the economic conditions of Italy and the Netherlands differ.
Assuming that the decisions of the ECB are based on the weighted average of the Euro zone economies, deviations from the average in inflation and output gap imply a gap between recommended interest rates for the single state and actual interest rates of the ECB. The data on inflation and interest rates for Italy and the Netherlands seem to be aligned; A higher inflation level implying a higher actual interest rate (see Figure 1 and 3). Additionaly, from 1999 onwards the inflation rates of Italy and the Netherlands show convergence and the interest rates also seem to depict this moderate level of inflation differences. Moreover, Ali (2012), Belke and Klose (2010) and Cohen (2012) point out the consequences of the crisis for the Euro zone. The data on output gap for both countries already indicates a change; It
declines after 2009. The output gap for Italy decreases stronger. The model implies that a decrease of the output gap leads to a decrease in the optimal interest rate. Thus, I expect to find a difference in the fit of the Taylor Rule and the interest rates for the two countries caused by the recent crisis.
4. Results
4.1 Results for Italy
Figure 4. Results of the application of the Taylor Rule in comparison with the actual interest rates for Italy -4 -2 0 2 4 6 8 10 12 14 16 18 1985 Q1 1986 Q3 1988 Q1 1989 Q3 1991 Q1 1992 Q3 1994 Q1 1995 Q3 19 97 Q1 1998 Q3 2000 Q1 2001 Q3 2003 Q1 2004 Q3 2006 Q1 2007 Q3 2009 Q1 2010 Q3 2012 Q1 20 13 Q3 TR recommendation Actual Interest Rates
18
1985-1998 2.18
1999-2014 -1.95
Table 1. Weighted differences between the actual interest rates and the Taylor Rule recommendation for Italy
Generally, a co-movement of Taylor Rule and actual interest rates can be observed as well as an overall downward trend. Italy’s Taylor Rule recommendation has been below the actual interest rates before the end of 1998 with a short period of exception in 1995. Afterwards the trend reversed itself. The reversion of the trend can also clearly be seen from the weighted differences, the value changed by approximately 4% to a negative range. The weighted differences are calculated by summing up the differences between actual interest rates and the Taylor Rule and dividing the resulting sum by the number of quarters in the subsample. The interest rates recommended by the Taylor Rule were above the actual interest rate with an exception in 2009 and since 2013. From the weighted differences it can be seen that in absolute terms the difference was slightly higher in the first sample.
4.2 Results for the Netherlands
Figure 5. Results of the application of the Taylor Rule in comparison with the actual interest rates for the Netherlands
1985-1998 1.81
1999-2014 -2.12
Table 2. Weighted differences between the actual interest rates and the Taylor Rule recommendation for the Netherlands -2 0 2 4 6 8 10 1985 Q1 1986 Q3 1988 Q1 1989 Q3 1991 Q1 1992 Q3 1994 Q1 1995 Q3 19 97 Q1 1998 Q3 2000 Q1 2001 Q3 2003 Q1 2004 Q3 2006 Q1 2007 Q3 2009 Q1 2010 Q3 2012 Q1 20 13 Q3 TR recommendation Actual Interest Rates
19
The results for the Netherlands show a similar pattern with respect to the change from a positive to a negative weighted difference. First the Taylor Rule values are below the actual interest rates and this reverses itself in 1996. Short exceptions can be observed in 2004 and 2006. The weighted difference also changes from 1.81 to -2.12 by approximately 4%. Notably, the graph exhibits a co-movement of Taylor Rule recommendation and actual interest rates. While the actual interest rates describe a downward trend with large
fluctuations, the Taylor Rule recommendations for the Netherlands fluctuate and do not show a clear overall trend. Another difference between the results of the two states is that the axis shows a smaller range for the Netherlands. Thus, the proportions are depicted differently in the graphs. The robustness tests are featured in the Appendix.
5. Analysis
The results are analyzed to form a conclusion with respect to the research question, whether the interest rates of the single countries deviate more from the Taylor Rule recommendation since the adoption of a single monetary policy. Thus, this section will evaluate the results in detail and also show the limitations of the research.
The results suggest that the first hypothesis is not supported. Instead of deviating more after the creation of the Euro zone as expected, the deviation of actual interest rates from the Taylor Rule recommendation for both countries has not substantially changed in absolute terms. However, the weighted differences between actual and Taylor Rule interest rates show a clear pattern; the deviation changes from a positive to a negative difference. This means that first the actual interest rates are above the interest rates recommended by the Taylor Rule and that this pattern reverses itself. Notably, the two countries show a similar pattern.
Furthermore, concerning the second part of the hypothesis, the results show that the actual interest rates declined since the beginning of the crisis. In contrast to the development of the actual interest rates, the Taylor Rule recommended a strong increase in 2010. From these results follow three questions. What has caused the deviations? Why does the pattern reverse itself in 1998 for Italy and in 1996 for the Netherlands? And lastly, what role does the recent crisis play for the performance of the interest rates?
5.1 Analysis of the Results 1985-1998
In the first subsample the actual interest rates have generally been above the interest rates recommended by the Taylor Rule. This implies that the Dutch and Italian CB pursued high levels of inflation fighting. An explanation could be their orientation towards the German
20
Bundesbank which was known for its tight control of price stability. Their monetary policy was seen as a benchmark for the Euro zone before 1999 (Peersman & Smets, 1999, p. 89). Clarida et al. reinforce this argument. They proved the dependence of the Italian CB on the German monetary policy. Furthermore, they find similar results for Britain and France (1998, p. 1058). Italy and the Netherlands were members of the European Monetary System (EMS) which prepared the establishment of the Euro zone (Peersman & Smets, 1999, p. 89). As a part of the EMS an exchange rate peg was imposed on the member states (Peersman & Smets, 1999, p. 89). Clarida et al. emphasize that the Italian CB did not conduct independent
monetary policy in the 1980s and 1990s. The preparations for the European monetary union meant that Italy did not have the options of floating exchange rates nor independent capital control (1998, p. 1052). This means that these countries did not follow the inflation fighting approach of the Bundesbank out of economic considerations for the optimal interest rates for their countries but were required to follow the monetary policy of the German CB (Clarida et al., 1998, p. 1052). When East and West Germany were re-unified in 1991 the Bundesbank increased the interest rates due to inflation concerns. This lead to the EMS collapse in 1992. The CBs of Italy and the Netherlands had to choose to either follow along with high interest rates or to devalue their currency and exit of EMS (Clarida et al., 1998, p. 1060 ). The effects on Italy and the Netherlands can be seen in data. For both countries the actual interest rates peak around 1992, while the Taylor Rule simultaneously recommends lower interest rates. After 1992 the actual interest rates start to fall for both countries (Figure 4 and 5).
Overall the actual interest rate seem to follow the qualitative trend of Taylor Rule rather well. Nevertheless, there are great quantative differences in the fit between actual and Taylor Rule interest rates for Italy and the Netherlands (see Figure 4 and 5). The biggest differences are in the periods with strong decreases in inflation rate or output gap. This implies that the policymakers reply slowly to strong changes. A possible explanation is that policymakers do not know how persistent the change will be beforehand. Another explanation is the interest rate smoothing of the CB.
In addition, that the two countries were part of the EMS meant that as a part of the system the exchange rates were fixed between the member states (Clarida et al., 1998, p. 1035). Thus, the monetary policies of Italy and the Netherlands were constrained. The constraint violates Taylor’s assumption of flexible exchange rates (1993, p. 201). Hence, an interpretation of the results which are based on the Taylor Rule is problematic.
Looking at the interest rates of Italy in the first subsample 1985 to 1999 (Figure 1) the question arises whether Italy has pursued a 2% inflation target. The graph shows that inflation
21
has been well above 2% and has only converged towards the 2% target since 1999. Peersman and Smets (1999, p. 89) have noticed this issue and have excluded Italy from their research on the appropriateness of the Taylor Rule for the Euro zone. Thus, the inflation target of 2% used in this paper may not be appropriate to study Italian monetary policy before 1999.
5.2 Analysis of the Results 1999-2014
The results for the second subsample show a clear reversion of the pattern. The actual interest rates lie beneath the Taylor Rule recommendations with some exceptions. For both countries the weighted difference decreased by almost 4%, however, the absolute differences remain approximately the same. This suggests that the ECB responds less to inflation in comparison to the National CBs of Italy and the Netherlands. The Italian and the Dutch CBs are less known for inflation fighting compared to the German CB (Faust et al., 2001, p. 3). Therefore, this finding stands in contrast to DeGrauwe’s prediction that the ECB will focus on inflation even more than the Bundesbank which was known for its inflation fighting approach before the introduction of the Euro zone (1997, as cited in Faust et al., 2001, p.2). Instead, the result confirms Faust et al.’s study. They found that the reaction function derived from the
Bundesbank’s monetary policy predicts higher rates than the ones observed after the
introduction of the Euro (2001, p. 4). This means that the ECB, in terms of the model, uses a lower weight on inflation and/or a higher weight on the output gap. Therefore, contradicting Belke and Klose who found that the ECB focuses on inflation and neglects the output gap (2010, p.19). However, Belke and Polleit (2007, p. 2200), Fourçans and Vranceanu (2004, p. 592) and Sauer and Sturm (2007, p. 393) find similar results but they partly attribute an imprecision in the coefficient weights to the backward-looking form of the Taylor Rule. Here the limitations of the research must be taken into account. The interpretation is derived from the results of two of 19 Euro zone member states, thus, it is questionable if the results found here can be applied to the ECB and the Euro zone as a whole. To be able to generalize the conclusion, all Euro zone member states must be considered.
Between 1999 and 2008 the actual and Taylor Rule interest rates exhibit a
co-movement for Italy. On the other hand, the fit for the Netherlands is more volatile. The Taylor Rule recommendation spikes in 2000 and 2008 in the Netherlands. For a short time period in 2004 to 2006 the actual interest rates are very close to the interest rates recommended by the Taylor Rule for the Netherlands (see Figure 6). The result corresponds to the findings of Nechio, who found a close fit for the Euro zone with the actual interest rate in 2005 to 2006
22
(2011, p. 2). The Taylor Rule in this period seems to be driven by the fast decline of the Dutch output gap (see Figure 2).
Figure 6. Actual interest rates in comparison to the Taylor Rule recommendations for Italy and the Netherlands 1999-2014
In the last quarter of 2008 the actual and Taylor Rule interest rates decline rapidly. In this period inflation and output gap fall strongly (see Figure 6). While the actual interest rates continue to decrease, the Taylor Rule would have recommended increasing the interest rates again. Therefore, a large difference in the fit develops. This can be explained by the effects of the financial crisis. Since 2008 the ECB has tried to restore investor confidence and ensure liquidity in the financial market by keeping the interest rates low (ECB, 2011, p. 90). In addition the sovereign debt crisis in some Euro zone member states weakens investor confidence further and complicates monetary policy for the ECB (ECB, 2011, p. 125). Moreover, assuming that ECB use interest smoothing, the changes in the key interest rates would be subtle changes rather than abrupt changes accommodating the abrupt decrease in the output gap for both countries since the start of the crisis and the comparatively volatile
inflation rates. Thus, adding interest rate smoothing to the Taylor Rule specification is likely to improve the fit. However, inflation rate and output gap fall again in 2011 and thus, the fit improves (see Figure 1, 2, 3). In Italy where the output gap is substantially lower (see Figure 2), the Taylor Rule recommends negative interest rates. According to Nechio, the low interest rates can help a country to cope with their public debts (2011, p. 3).
While the Taylor Rule recommendation for Italy and the Netherlands have been close to each other before, the recommendations for the two countries diverge since 2012 again (see
-4 -2 0 2 4 6 8 10 1999 Q1 2000 Q1 2001 Q1 2002 Q1 2003 Q1 2004 Q1 2005 Q1 20 06 Q1 20 07 Q1 20 08 Q1 20 09 Q1 20 10 Q1 20 11 Q1 2012 Q1 2013 Q1 2014 Q1 ECB Rates TR Recommendation IT TR Recommendation NL
23
Figure 6). This finding is in line with the studies of Moons and Van Poeck (2008) and Nechio (2011). According to Moons and Van Poeck, the gap between optimal national interest rates and Taylor Rule interest rates for the Netherlands should be larger and should be smaller for
Italy (2008, p. 195). This view is also shared by Björkstén and Syrjänen (1999, p. 18).In
contrast to this, Nechio (2011, p. 2) shows that even though the Taylor Rule has matched the Euro zone as a whole since 2005, there are differences in the accuracy in the fit for the peripheral countries. However, she does not include Italy in the peripheral countries because she thinks the economic conditions are closer to the more stable Euro zone states whereas the peripheral countries such as Greece, Ireland and Portugal show large deviations in inflation and unemployment. According to her study, the Netherlands is in the group of core countries and hence, will match the actual interest rates of the ECB well (2011, p.4). Looking at the results in this paper, there is not a large difference between the weighted differences in absolute terms of Italy and the Netherlands after 1999 (see Table 1 and 2). Nevertheless, in the beginning of 2012 the difference between actual and Taylor Rule interest rates was -4.4% for Italy and -4.2% for the Netherlands. These values are well above the average difference over the full period after the creation of the Euro zone.
Assuming that the decisions of the ECB are based on the weighted average data of all Euro zone states, deviations in the fit of actual and Taylor Rule interest rates can be explained by deviations in inflation rate and output gap from the mean. According to Moons and Van Poeck, this implies a larger gap for the Netherlands because they have experienced relatively low inflation (2008, p. 195). The results of this paper show that the absolute weighted
difference is 0.17% larger for the Netherlands (see Tables 1 & 2). This difference persists in the robustness test (see Table 3). While the Taylor Rule recommendations have been above the ECB’s interest rates for both countries, Italy’s recommendation is now above the actual interest rates and the Netherlands’ recommendation is above (see Figure 6). The difference seems to be due the difference of the output gaps of the two countries since 2009. In addition, the weight of Netherlands in the calculation of the Euro zone average is small and therefore, not crucial for the ECB’s decision (Björkstén & Syrjänen, 1999, p. 11). On the other hand, Fendel and Frenkel suggest that the ECB was cautious to fight inflation because of the risk of deflation that it would cause the low inflation states (2009, p. 1301). Hence, this suggests that the ECB does not only consider an average of economic conditions in the Euro zone but also considers outliers.
The ECB calls their monetary policy a success. Until 2011 the medium term inflation fluctuated around its target of 2% (ECB, 2011, p. 129). Looking at the results of this paper,
24
the crisis manifests itself in the data. The historically low interest rate limits the scope of the ECB’s actions substantially, therefore, the ECB introduced non-standard measures (ECB, 2015). The Euro zone consists of 19 different countries, thus, it is inevitable that a single monetary policy does not fit all countries. A better fit can only be reached if the economic conditions of the countries converge. Nevertheless, Moons and Van Poeck observe that the
Euro zone is lacking convergence (2008, p. 196).Therefore, the solution to the imprecise fit
of actual and Taylor Rule interest rates lies in the coordination of fiscal policy. Ali (2012, p. 428) and Cohen (2012, p. 692) suggest that the European public debt crisis needs an
increased coordination of the fiscal policy of the single Euro zone states.
Finally, the absolute deviation in the fit between actual and Taylor Rule interest rates remains the same for both countries but changes its pattern from a positive to a negative difference. This means that in the first subsample the Taylor Rule interest rates are below the actual interest rates and are above the actual interest rates in the second subsample. However, it is difficult to compare the two subsamples. The reason for the deviations seem to be of a different nature. Firstly, whereas the Euro zone can be compared to large open economies such as the economy of the United States, the two countries cannot be considered large open economies before 1998. Additionally, the Euro zone has flexible exchange rates, while Italy and the Netherlands were bound in a system of fixed exchange rates before 1998.
Nevertheless, the imprecision of the fit of the actual interest rates with the interest rates recommended by the Taylor Rule is of interest for the Euro zone. Moreover, the crisis clearly manifests itself in the data. The great problems of the Euro zone need to be addresses with fiscal and monetary policy. Therefore, the difference between actual and Taylor Rule interest rates deserves attention even though the deviation has existed before the creation of the Euro zone.
5.3 Limitations
The most important limitation of this research is given by the Taylor Rule. In this paper, it was assumed that the rule captures the monetary reaction function of the ECB and provides an optimal monetary policy. Furthermore, it is based on the assumption that the ECB as well as the two individual countries use inflation targeting and short term interest rates as their instrument. Furthermore, Taylor assumes flexible exchange rates (1993, p. 201). The ECB fulfills these assumption. However, Italy’s and the Netherlands’ dependence on fixed exchange rates with other European countries in the EMS violates this assumption.
25
interpretation becomes more complicated. The Taylor Rule was chosen for this research because it is the most commonly used reaction function and is assumed to capture the most important elements of monetary policy. Fourçans and Vranceanu (2004, p. 594) remark that even if the Taylor Rule follows the monetary policies of CBs relatively well, it does not make it an optimal monetary policy. In reality, it is likely that not only inflation rate and output gap determine the interest rates of the CBs but that the underlying macro-economical models are more sophisticated.
The functional form of the Taylor Rule provides threats to the internal validity by itself. As discussed in the methodology, the Taylor Rule has been estimated for the Fed and the ECB. Whereas this paper makes use of the simple form of the Taylor Rule. Nevertheless, empirically estimated reaction functions may not depict an optimal monetary policy. Another problem is given by the changes in the parameters that come from the recent crisis. Belke and Klose find that for example the real equilibrium interest rate changes in exceptional times such as the crisis (2010, p. 5). Moreover, according to Belke and Polleit, the ECB behaved not according to their usual pattern during the crisis (2007, p. 2208). Thus, for the sake of
simplicity, it was decided to use the original form developed by Taylor (1993) himself. Even if an empirical estimation of the Taylor Rule would improve the validity, it is difficult to compare and develop the reaction functions of before and after the creation of the Euro zone.
Another aspect of the limitations derives from the availability of the data. While inflation is measured quarterly and reported frequently, data on the output gap is only available as an annual percentage change since 1985. The use of a sophisticated method to approximate the quarterly output gap is likely to increase the precision. However, potential output is always an estimate. Thus, it always conveys some uncertainty.
6. Conclusion
In this paper the question, whether the interest rates deviate more or less from the Taylor Rule since the creation of the Euro zone in Italy and the Netherlands, was investigated. For the analysis covering 30 years the Taylor Rule was used as a benchmark. The results suggest that the difference does not change in absolute terms but reverses its pattern. Before 1998 the actual interest rates have generally been above the ones suggested by the Taylor Rule and after 1999 the actual interest lie below the Taylor Rule recommendation. It is difficult to interpret the deviation pre-1999 because the economic conditions violate the assumptions of the Taylor Rule. The simplest reason for the deviation in the time period 1999-2014 is that the ECB attribute less weight on inflation and more on the output gap. Furthermore, during the
26
crisis the declining ECB interest rates temporarily contradicted the interest rates recommended by the Taylor Rule for both Italy and the Netherlands.
To be able to compare the two subsample periods, namely before and after the introduction of the Euro zone, there has to be more research on the appropriate Taylor Rule specifications for the period before 1999. For example, as mentioned before, the data on inflation contradicts the view that the Italian inflation target was at 2% before 1999. Future research should also focus on an empirical estimation of the Taylor Rule for the ECB, also considering the special circumstances of the crisis period. From the theoretical implications follow practical issues; What is the optimal monetary policy for the ECB as a whole? It is a job for future scholars to extend the research to a greater number of Euro zone member states and to find out if this optimal monetary policy is given by the Taylor Rule.
27 Bibliography
Ali, T. (2012). The impact of the sovereign debt crisis on the Eurozone countries. Social and Behavioural Sciences , 62, 424-430.
Appleyard, D., Field, A., & Cobb, S. (2009). Economic Integration. In International Economics (7th ed., pp. 392-417). New York: McGraw Hill.
Belke, A., & Klose, J. (2010). (How) Do the ECB and the Fed react to financial market uncertainty? The Taylor rule in times of the crisis. DIW Berlin Discussion Papers , 972, 1-31.
Belke, A., & Polleit, T. (2007). How the ECB and the US Fed set interest rates. 39 (17), 2197-2209. Björkstén, N., & Syrjänen, M. (1999). Divergences in the Euro Area:a cause for Concern? Bank of
Finland Discussion papers , 1-34.
Clarida, R., Galí, J., & Gertler, M. (1998). Monetary policy rules in practice. Some international evidence. European Economic Review , 42 (6), 1033-1067.
Cohen, B. (2012). The future of the euro: Let's get real. Review of International Political Economy , 19 (4), 689-700.
Davradakis, E., & Taylor, M. (2006). Interest Rate Setting and Inflation Targeting: Evidence of a Nonlinear Taylor Rule for the United Kingdom. Studies in Nonlinear Dynamics & Econometrics, 4, 1558-3708.
ECB (2015). Lithunania (2015). Retrived from
https://www.ecb.europa.eu/euro/changeover/lithuania/html/index.en.html (accessed 10.2.15).
ECB (2011). The Monetary Policy of the ECB. Retrived from https://www.ecb.europa.eu/pub/html/index.en.html ECB (2015). Press Release 22.1.15. Retrived from
http://www.ecb.europa.eu/press/pr/date/2015/html/pr150122_1.en.html. (accessed 15.2.15).
Faust, J., Rogers, J. H., & Wright, J. H. (2001). An Empirical Comparison of Bundesbank and ECB Monetary Policy Rules. FRB International Finance Discussion Paper , 705, 1-24.
Fendel, R., & Frenkel, M. (2009). Inflation differentials in the Euro area: did the ECB care? Applied Economics , 41 (10), 1293-1302.
Fourçans, A., & Vranceanu, R. (2004). The ECB interest rate rule under the Duisenberg presidency. European Journal of Political Economy , 20 (3), 579-595.
Judd, J., & Rudenbusch, G. (1998). Taylor’s Rule and the Fed: 1970–1997. FRBSF ECONOMIC REVIEW (3), 3-16.
Kozicki, S. (1999). How useful are Taylor rules for monetary policy? Economic Review-Federal Reserve Bank of Kansas City , 84, 5-34.
Kydland, F. & Prescott, E. (1997). Rules Rather than Discretion: The Inconsistency of Optimal Plans. Journal of Political Economy, 85 (3), 473-492.
Moons, C., & Van Poeck, A. (2008). Does one size fit all? A Taylor-rule based analysis of monetary policy for current and future EMU members. Applied Economics , 40 (2), 193-199. Moura, M., & de Carvalho, A. (2010). What can Taylor rules say about monetary policy in Latin
America? Journal of Macroeconomics , 32 (1), 392–404.
Nechio, F. (2011). Monetary policy when one size does nor fit all. FRBSF Economic Letter , 18, 1-5. Peersman, G., & Smets, F. (1999). The Taylor rule: A useful monetary policy benchmark for the
28
Sauer, S., & Sturm, J. (2007). Using Taylor rules to understand European Central Bank monetary policy. German Economic Review , 8 (3), 375-398.
Taylor, J. (1993). Discretion versus policy rules in practise. Carnegie-Rochester Conference Series on Public Policy. 39, 195-214. North-Holland: Elsevier Science Publishers B.V.
29 Appendix
A.1 Robustness tests
To verify that the above results are robust against deviations in the measurement, two
robustness tests are applied (R1 and R2, see Table 3). Firstly, the Taylor Rule is computed by taking the average of the inflation rates for a year and using the annual values for the output gap. This is then compared with the average interest rates over a year. The second test takes the end of the year inflation rates and again the annual output gaps. The so computed Taylor Rule recommendation is then compared with the end of the period interest rates.
Italy Weighted differences Netherlands Weighted differences
1985-1998 1999-2014 1985-1998 1999-2014
original 2.18 -1.95 original 1.81 -2.12
R1 2.01 -1.93 R1 1.78 -2.09
R2 2.13 -1.88 R2 1.79 -2.08
Table 3. Robustness test
The results imply that the original method is robust. The robustness test show only small deviations in the weighted differences and exhibit the same graph movements.
