The effect of the establishment of the ECB on inflation in the
Eurozone
Denise de Vor
Bachelor Thesis
University of Amsterdam Faculty Economics and Business BSc Economie en Bedrijfskunde Economie en Financiering
Student number: 10606394
Thesis supervisor: Gabriele Ciminelli Date: June 29th, 2016
Statement of Originality
This document is written by Student Denise de Vor who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
In this paper it is analysed whether the establishment of the European Central Bank as a highly independent central bank affected the inflation rates in the Eurozone. In the OLS regression models that are used to answer the research question, a dummy variable is added
that distinguishes whether the country is part of the Economic and Monetary Union or not. Subsequently, the sign and significance of the coefficient of this dummy variable will show if
being part of this union will give the country an extra boost in lowering the level and variability of inflation. The estimation outcomes show that a high level of central bank independence will result in a lower level and variability of inflation, provided that the right measure of central bank independence is used. In addition, being part of the EMU will not
Contents
1. Introduction ... 5
2. Literature review ... 7
2.1. Theoretical framework... 7
2.2. Empirical evidence ... 10
3. Dataset and methodology ... 14
3.1 Methodology ... 14
3.2 Hypotheses... 15
3.3 Legal central bank independence... 16
3.4 Data ... 17 4. Results ... 23 4.1 OLS regressions ... 23 4.1.1 Results 1980-1989 ... 23 4.1.2 Results 1998-2007 ... 27 4.2 Robustness check ... 30 5. Conclusion ... 34 6. References ... 36 7. Appendices ... 37
Appendix 1. Criteria list of the legal central bank independence measure of Cukierman, Webb, and Neyapti (LVAU, LVAW)... 37
Appendix 2. Criteria list of the legal central bank independence measure of Dincer and Eichengreen (CBIU, CBIW)... 40
Appendix 3. Scatterplots 1980-1989 ... 45
1. Introduction
At the 1st of June 1998, the European Central Bank was established and, right away, they
behaved as an independent central bank. Although the Federal Reserve, the central bank of the United States of America, is a highly independent central bank, the Maastricht Treaty has made the ECB the most independent central bank in the world (Mishkin, Matthews, &
Giuliodori, 2013, p. 290). The main reason to behave as an independent central bank, as stated on their website, is that it is conducive to maintaining price stability. According to the ECB, price stability is defined as a year-on-year increase in the Harmonised Index of Consumer Prices (HICP) for the euro area of below 2%. Hence, in their point of view, high central bank independence will result in a low and more stable inflation rate.
The Maastricht Treaty contains several laws that make the ECB an independent central bank. Overall, it states that the ECB will have complete autonomy in conducting the common monetary policy of the European Union (Eijffinger, & de Haan, 1996). Other elements of the Maastricht Treaty that make the ECB highly independent is that the members of the Executive Board have terms of eight years, whereas heads of the National Central Banks have terms of at least five years. In addition, it states that the Euro system determines its own budget, that the governments of the member countries are not allowed to issue instructions to the ECB, and that the ECB is prohibited from granting loans to national public-sector entities (Mishkin, Matthews, & Giuliodori, 2013).
The view of the ECB that this high central bank independency will result in lower inflation is probably based on the results of the studies that investigated the relationship between the level of central bank independence and the level of inflation. The first studies about this relationship were published in the 1980s and, up until today, it is a highly researched topic. In those articles, it is often argued that a high level of central bank
independence coupled with an explicit mandate for price stability are important institutional devices for maintaining that stability (Eijffinger, & de Haan, 1996). The evidence of the Bundesbank is one of the reasons that a large number of countries increased their level of central bank independence in order to maintain price stability. In the post-World War II period, Germany had one of the lowest inflation rates, and their central bank, the Bundesbank, was relatively autonomous (Eijffinger, & de Haan, 1996).
Conversely, in the study of Cukierman, Webb, and Neyapti (1992) some striking observations were found, resulting in the conclusion that legal central bank independence may be neither necessary nor sufficient for low inflation. For example, they found that Belgium, Japan, Morocco, and Qatar had very low inflation rates, but were also ranked in the lowest
quartile of legal central bank independence. Apparently, the high level of central bank independence which the ECB has created does not necessarily mean that there are low inflation rates. Because of this, the following research question is investigated in this paper: Did the establishment of the ECB as an independent central bank influenced the inflation rates of the Eurozone?
The research question is answered using OLS regressions of the level and variability of inflation on, among others, a central bank independence variable and a dummy variable that distinguishes EMU-countries from non-EMU-countries. The regressions are done for two different time periods. One period before the establishment of the ECB, and one period after the establishment of the ECB. The results show that it does matter which measure of central bank independence is used as a proxy of actual central bank independence. If the right measure is used, more independency will lead to a lower level and variability of inflation. Finally, being part of the EMU will, besides of having a highly independent central bank, not give a significantly extra boost in lowering the level and variability of inflation.
To examine whether the independency of the ECB has affected the inflation rates of the EMU-countries is important because, according to Rogoff (1985), greater independency of the central bank comes at the cost of higher output volatility. So, if the very high
independency of the ECB did not resulted in a significant lower level and variability of inflation, one can consider whether the benefits of having an independent central bank outweigh the costs.
The paper proceeds as follows. In the next section, the literature review is discussed. The dataset and methodology are explained in the third part. In the fourth part, the results of the regressions are presented and discussed. Finally, a conclusion is drawn in the fifth part.
2. Literature review
2.1. Theoretical framework
The theoretical background of the negative relationship between the level of central bank independence and the level of inflation was first provided by Barro and Gordon (1983). In their paper, they showed that discretionary monetary policy led to the time inconsistency problem. The time inconsistency problem means that if a central bank has a high degree of discretion in conducting monetary policy, it has a constant political pressure to boost economic growth and to reduce unemployment. The economy cannot, however, exceed its potential GDP or its natural rate of unemployment over time (Crowe, & Meade, 2007). In the long run, this discretionary monetary policy will lead to an ‘inflationary bias’, which means that that there will be persistent higher inflation than socially optimal with the same level of unemployment. One possible solution to the inflation bias, as stated by Rogoff (1985), is to delegate monetary policy to an independent central bank that is more inflation averse than the government and that could credibly commit to a certain inflation target. The reason why it is necessary for the central bank to have a credible commitment to a certain inflation level is researched by Walsh (2009). He states that inflation targeting could improve the short-run trade-off between output gap and inflation volatility (Walsh, 2009). It could do so by
anchoring the public’s beliefs about future inflation. If the central bank credibly commits to a certain inflation target, there will be greater stability of inflation expectations, which in turn will lower the volatility of both inflation and real activity. In other words, if there is a clear formal target that anchors the expectations about the central bank’s goals, it allows the central bank to reduce both inflation and output volatility (Walsh, 2009).
The theory mentioned above can be illustrated by a simple model of the inflationary bias. In this model, the society’s loss function is given by the following equation:
𝐿 = 1
2 [𝑎(𝜋 − 𝜋
∗)2+(𝑦 − 𝑘𝑦∗)2] (1)
In this equation, L stands for the loss of society, a is the coefficient that measures the aversion for inflation variability and is strictly positive, π is the real inflation, π* is the initial
equilibrium real inflation, y is total output, y* is the initial equilibrium of total output, and k is
preferred by the policy makers. For example, if k>1, a level of output greater than the market clearing level is desired by the politicians. This results in a greater loss of society.
To further rearrange the above equation, the Philips curve relationship is added, which is as follows:
𝑦 = 𝑦∗+ 𝑏(𝜋 − 𝜋𝑒) (2)
where y and y* have the same meaning as in equation (1) and b is the coefficient of the
difference between real inflation and expected inflation and measures the extent to which surprise inflation stimulates output. Substituting equation (2) in equation (1) results in the following equation:
𝐿 = 1 2 [𝑎𝜋
2+((1 − 𝑘)𝑦∗+ 𝑏(𝜋 − 𝜋𝑒))2] (3)
By taking the first derivative of equation (3) with respect to inflation, the inflation equation is determined:
𝜋 = 𝑏
𝑎 + 𝑏2 (𝑏𝜋
𝑒+ (𝑘 − 1)𝑦∗) (4)
Next, if it is assumed that the inflation expectations are equal to zero, then the following value of inflation under discretion is derived:
𝜋𝐷= 𝑏
𝑎 + 𝑏2 (𝑘 − 1)𝑦 ∗ (5)
where the superscript D is added to indicate that this is a discretionary solution, that is, it is the solution in which the central bank does not commit to keeping inflation to a particular level. Hence, the problem of time inconsistency is now clearly shown. It is assumed that the inflation expectations are zero, but as is shown by equation (5), the level of inflation is not equal to zero. In other words, the discretionary solution is not an equilibrium, because πe ≠ πD.
As a result, there is an inflation bias. The loss associated with this level of inflation, πD, and
𝐿𝐷= 1
2
𝑎(𝑘 − 1)2 (𝑎 + 𝑏2) (𝑦
∗)2 (6)
Next, it is analysed in the case of an independent central bank that could credibly commit to low inflation. In this case, k is equal to one, because the central bank fully
determines monetary policy. As a result, the society’s loss function is given by the following equation:
𝐿 = 1
2 [𝑎(𝜋 − 𝜋
∗)2+(𝑦 − 𝑦∗)2] (7)
Again, by substituting equation (2) in equation (7), the following equation is derived:
𝐿 = 1 2 [𝑎𝜋
2+(𝑏(𝜋 − 𝜋𝑒))2] (8)
Taking the first derivative of equation (8) with respect to inflation results in the following equilibrium level of inflation under commitment:
𝜋𝐼 = 𝑏 𝑎 + 𝑏2𝜋
𝑒 (9)
where the subscript I is added to indicate that this is the solution under an independent central bank with a credible commitment. If it is again assumed that the inflation expectations are equal to zero, the level of inflation will also be equal to zero. So, in this case, there is an equilibrium and the inflation bias is eliminated. In other words, if monetary policy is delegated to an independent central bank that is more inflation averse than the society, this will serve as a commitment device that allows for sustaining a lower rate of inflation. In addition, if πe = πI = 0, and, as a result, yI = y*, the loss of society under commitment, LI, will
be equal to zero. Hence, when comparing the loss of society under discretion to the loss of society under commitment, the society’s loss under discretion will be higher than the loss under an independent central bank with a credible commitment.
A more theoretical point of view of the inflationary bias is that subjecting central banks to more political pressures would impart this bias. According to Mishkin, Matthews, and Giuliodori (2013) the primary goal of politicians is to win the next elections, which may make them short-sided. Because of this, they will seek short-run solutions to problems, which
will have undesirable long-run consequences. For example, the politicians would like to increase the money supply to lower the interest rates. However, if there is too much money supply, there will be more money chasing fewer products, resulting in an increase in the inflation rate. As stated by Mishkin, Matthews, and Giuliodori (2013), the advocates of an independent central bank, therefore, believe that a politically insulated central bank will be more concerned with long-run objectives, causing a more stable inflation rate.
Besides the advantage that an independent central bank will lead to a lower and more stable inflation rate, Rogoff (1985) stated that the independency of the central bank comes at a cost. He suggests that by having an independent and inflation-averse central bank there is a loss of discretionary monetary policymaking which, as a consequence, will result in an increase in output variability. In his model, as in the model of Barro and Gordon (1983), having a conservative central banker leads to a lower inflationary bias and a lower variance of inflation. Conversely, the variance of output is an increasing function of the conservatism of the central banker (Rogoff, 1985). The reason for this is that the more inflation-averse central banker tolerates a more cyclical variability in economic activity since he is engaged in a less discretionary stabilization economic policy. Nevertheless, Rogoff (1985) concluded that the gains from the lower inflation rate possibly will exceed the losses from the higher output variability. So, on net, society is better off by appointing a conservative central banker who is inflation averse. However, according to Rogoff (1985), it is not optimal to appoint a central banker who is only concerned with low and stable inflation. Although it is possible to induce less inflationary wage bargains, it distorts the central bank’s responses to unanticipated disturbances, particularly supply shocks.
2.2. Empirical evidence
Although the model of Rogoff (1985) seems feasible, Alesina and Summers (1993) do not find that there is a relationship between the level of central bank independence and the level of output variability in the OECD countries (Alesina, & Summers, 1993). In addition, they also do not find a strong relationship between the level of central bank independence and the unemployment rate, and between the level of central bank independence and the real interest rates. As a result, their conclusion is that the level of central bank independence does not have large costs in terms of real macroeconomic performance (Alesina, & Summers, 1993). The overall effect of the level of central bank independence on output variability is, thus, still ambiguous.
The papers that researched the empirical evidence of the level of central bank
independence on macroeconomic factors such as inflation and output growth do not measure the actual independence of the central bank. Instead, they mostly measure the legal
independence. As mentioned by Cukierman (1992), actual independence, as opposed to legal independence, not only depends on legislation but also on many other factors, such as
informal arrangements with the government, the quality of the work force of the central bank, and the personalities of the key individuals in the bank (de Haan, & van ‘t Hag, 1995).
Because these other factors are impossible to quantify, most existing research has focused on measuring the legal independence of a central bank. One of the most frequently used
measures is the legal independence measure of Cukierman (1992). This measure ranges from zero (dependent ) to one (independent) and is based upon four variables: (1) variables
concerning the appointment, dismissal and term of office of the chief executive officer of the bank, (2) variables concerning the resolution of conflicts between the executive branch and the central bank, (3) the final objectives of the central bank as stated in its charter, and (4) legal restrictions on the ability of the public sector to borrow from the central bank (Cukierman, Webb, & Neyapti, 1992, p. 356-359). A more detailed description of this measure of central bank independence is discussed in the dataset and methodology section.
Another notable empirical finding is that the inverse relationship between legal central bank independence and the level of inflation, that is investigated by Cukierman, Webb, and Neyapti (1992), does not hold for developing countries. The reason why this relationship does not hold in developing countries is due to the legal-independence measure that is used. The legal-independence measure is a reasonable proxy only if there is sufficient respect for the rule of law in the country under consideration (Cukierman, Webb, Neyapti, 1992).
According to Cukierman (1994), there is reason to believe that the law is a poorer indication of actual practice in less developed countries than in industrial countries, since developing countries have less regard for the law. To still be able to measure the
independence of the central bank in developing countries, the actual frequency of change of the governor of the central bank, also known as the turnover rate (TOR), is used. In this case, a higher turnover rate indicates a lower level of independence. According to Cukierman, Webb, and Neyapti (1992), a more rapid turnover creates dependence because if the political authorities frequently get the opportunity to choose a new governor of the central bank, they will choose one who will do their will. In addition, frequent turnover might be caused by firing those who challenge the government. It should be noted, however, that a low turnover rate does not necessarily imply a high level of central bank independence because if a
governor is relatively subservient, he may stay in office for a long time (Cukierman, Webb, Neyapti, 1992).
When the relationship between the turnover rate and the level of inflation in
developing countries is investigated, a strong positive relationship appears. Hence, by using the turnover rate as a measure of actual independence in developing countries, there also is a significant negative relationship between the level of central bank independence and the level of inflation (Cukierman, Webb, Neyapti, 1992). In short, provided that the right measure of the level of central bank independence is used, the level of inflation is negatively related to the level of central bank independence in industrial as well as in developing countries.
Besides the empirical finding of Cukierman, Webb, and Neyapti (1992) that the level of central bank independence influences the inflation rate in developing countries, having a clear inflation target does also affect the level of inflation in the developing countries (Walsh, 2009). It has to be noted, however, that no central bank appears to be a strict inflation targeter. Instead, the central banks are so-called flexible inflation targeters, that is, they are concerned with both inflation and real economic stability (Walsh, 2009). In his research, Walsh (2009) found that developing countries that adopt inflation targeting faced a reduction in average inflation between 3 percent and 6 percent, depending on the method used to date inflation targeting and whether initial inflation was controlled for (Walsh, 2009). In addition, the policy of inflation targeting had a significant effect in reducing inflation volatility and lowering real economic volatility in developing countries (Walsh, 2009).
According to Walsh (2009) it is, however, more difficult to draw conclusions about the effect of inflation targeting on the level and variability of inflation in industrial countries. The reason for this is that the decline in inflation among industrialized countries over the past 20 years coincided with the adoption of inflation targeting. The decline in inflation is to a large degree due to the benign economic environment of the last two decades. During that period, all the OECD countries, both IT and non-IT countries, faced a lower and more stable inflation rates (Walsh, 2009). Because of this, no valid conclusion about the effect of inflation targeting in industrial countries can be made.
To sum up, there are a lot of studies that investigated the relationship between the level of central bank independence and macroeconomic factors. Based on the previous studies, it seems that a high level of central bank independence should lower the level and variability of inflation. In addition, a lot of research has been done on the effectiveness of inflation targeting. Here, it is also found that a clear inflation target should lower the level of inflation. Since the ECB is a highly independent central bank and has a clear inflation target
(they are aiming an inflation target of below, but close to two percent) it is worthwhile to investigate the effect of the establishment of the ECB on inflation in the Eurozone. The dataset and methodology of this research is discussed in the next section.
3. Dataset and methodology
3.1 Methodology
In this research it is analysed whether the formation of the ECB as an independent central bank had an effect on the inflation in the EMU-countries. Because the turnover rate of the governor of the central bank might be a better predictor of actual independence, or at least in developing countries, regressions are done with both the legal central bank independence measure and the turnover rate. The resulting four OLS regression are of the following form:
𝜋 = 𝛽0+ 𝛽1𝐿𝐶𝐵𝐼 + 𝛽2𝑂𝑝𝑒𝑛𝑛𝑒𝑠𝑠 + 𝛽3𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐷𝑒𝑝𝑡ℎ + 𝛽4𝐸𝑀𝑈 + 𝜀 (1)
𝜎𝜋 = 𝛽0+ 𝛽1𝐿𝐶𝐵𝐼 + 𝛽2𝑂𝑝𝑒𝑛𝑛𝑒𝑠𝑠 + 𝛽3𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐷𝑒𝑝𝑡ℎ + 𝛽4𝐸𝑀𝑈 + 𝜀 (2)
𝜋 = 𝛽0+ 𝛽1𝑇𝑂𝑅 + 𝛽2𝑂𝑝𝑒𝑛𝑛𝑒𝑠𝑠 + 𝛽3𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐷𝑒𝑝𝑡ℎ + 𝛽4𝐸𝑀𝑈 + 𝜀 (3)
𝜎𝜋 = 𝛽0+ 𝛽1𝑇𝑂𝑅 + 𝛽2𝑂𝑝𝑒𝑛𝑛𝑒𝑠𝑠 + 𝛽3𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝐷𝑒𝑝𝑡ℎ + 𝛽4𝐸𝑀𝑈 + 𝜀 (4)
where π is the level of inflation and σπ is the standard deviation of inflation. The variable LCBI stands for legal central bank independence, and TOR for the turnover rate of the central bank’s governor. The variable Openness is the share of the sum of exports and imports to GDP, FinancialDepth is the share of private credit by deposit money banks and other
financial institutions to GDP, and EMU is a dummy variable that is equal to one if the country is part of the Economic and Monetary Union, and equal to zero otherwise. Finally, ε is the error term. These four regressions are done for a time period before the ECB was established, namely the period of 1980 until 1989, and for a time period after the establishment of the ECB, from 1998 until 2007. Per time period, cross-sectional regressions are done for multiple countries, both (current) EMU- and non-EMU-countries. In the regressions, robust standard errors are used, because it cannot be assumed that the error term has the same variance across all observations.
When the data about all the variables is collected, the OLS regressions are done in Stata. To clearly show how the coefficients change when another variable is added, the regressions are done in steps. The first regression is the baseline regression of the level of inflation or the variation of inflation with the main explanatory variable, LCBI or TOR, only. Next, the dummy variable EMU is added and, subsequently, the other control variables are
added one by one in additional regressions. Finally, one regression is done with all the control variables, the dummy variable, and both TOR and LCBI.
The regressions in this research are based on the regressions done by Dincer and Eichengreen (2013), because their regressions are clear, easy to interpret, and match with the research goal of this paper. The variable EMU is added in this paper to be able to make valid conclusions about the effect of the formation of the ECB as an independent central bank on inflation.
3.2 Hypotheses
In the first two regressions that contain the legal independence measure as explanatory variable, it is expected that the coefficient β1 is significantly negative. This is in line with the
existing theories about the relationship between the level of central bank independence and the level and variation of inflation because if the coefficient β1 is significantly negative, a
higher level of central bank independence will cause a lower level and variability of inflation. In the last two regression in which the turnover rate is the measure of central bank
independence, it is expected that the coefficient β1 is significantly positive, because a higher
turnover rate indicates a lower level of independence, which in turn should increase the level and variability of inflation.
Furthermore, in the period of 1980 until 1989, the fact that the country is part of the EMU in the future should not have an effect on the level and variability of inflation, so, in this period, the coefficient β4 should not be significantly different from zero. In the period of 1998
until 2007, it is expected that the EMU coefficient has a negative sign, since this means that being part of the EMU should have an additional negative effect on the level and variability of inflation. There are, however, two possibilities of the significance of the EMU coefficient. Because the model already contains an independency variable, either LCBI or TOR, the high level of independence of the ECB should already be captured by this. If it is assumed that these independence variables are a good measure of the actual level of central bank
independence, the EMU coefficient should not be significant. Conversely, if it is assumed that these independence variables are not a good measure of the actual level of central bank
independence, the EMU coefficient could be expected to be significant. The reason for this is that the high level of independency of the ECB is not well captured by the independence variable and thus should be captured by the EMU variable. It should be noted, however, that if the EMU coefficient turns out to be significant, this could also be due to some other factor that is shared by all EMU-countries.
3.3 Legal central bank independence
Before the hypotheses described above can be tested, it has to be clear how the level of central bank independence is measured. In the past few years there has been developed a lot of measures of the level of central bank independence. In this paper, two different measures of legal central bank independence are used, because no singe independence measure was available for the two time periods. Eventually, the legal independence measure of Cukierman, Webb, and Neyapti (1992) is used in the first time period. In the second period, the legal independence measure of Dincer and Eichengreen (2013) is used. The reason why those two measures are chosen is because they have much in common and are easy to interpret.
The coding of the legal central bank independence of Cukierman, Webb, and Neyapti (1992) is based upon two principles. First, they coded only a few narrow but precise legal characteristics, and second, they used the written information from the charters. In the charter of the central bank, the legal characteristics are grouped into four clusters of issues
(Cukierman, Webb, & Neyapti, 1992, p. 356):
1. The appointment, dismissal, and term of office of the chief executive officer of the bank;
2. The policy formulation, concerning the resolution of conflicts between the executive branch and the central bank over monetary policy and the participation of the central bank in the budget process;
3. The objectives of the central bank;
4. The limitations on the ability of the central bank to lend to the public sector. Thereafter, the clusters are built up from 16 legal variables, each coded on a scale of zero, which is the lowest level of independence, to one, which is the highest level of
independence. In each cluster, the level of central bank independence is determined as follows. In the first cluster, central banks in which the legal term of office of the executive officer is longer, and in which the executive branch has little legal authority in appointing or dismissing the governor are classified as more independent. Next, in the second cluster, central banks with wider authority to formulate monetary policy and which can resist the executive branch in cases of conflict are ranked as more independent. Considering the objective of the central bank, in the third cluster, there are six possible ratings according to whether price stability is the only objective of the central bank, or if there are also other objectives stated which might conflict with price stability. In essence, the objectives variable does not reflect the general level of independence from the government. In Rogoff’s (1985)
terminology, it measures the strength of the ‘conservative bias’ of the bank’s charter, meaning that it measures the legal mandate of the bank to focus only on the objective of price stability (Cukierman, Webb, Neyapti, 1992). Finally, in the fourth cluster, a central bank that has tighter limits on its lending to the public sector is classified as more independent. These limitations consist of a number of more detailed variables, such as limitations on advances and securitized lending, and restrictions on the maturity and interest rates of the loans (Cukierman, Webb, Neyapti, 1992).
At the final step, weights are given to the variables. Cukierman, Webb, and Neyapti (1992) used the weights that they considered to be most plausible. This resulted in the
weighted average of the legal independence, called LVAW. When a component was missing, the weights of the remaining components were expanded proportionately to sum to one. The unweighted measure is called LVAU. In Appendix 1, the criteria list of the legal central bank independence measure of Cukierman, Webb, and Neyapti is described.
For the second period that is examined in this research, from 1998 until 2007, another measure of legal independence is used. Dincer and Eichengreen (2013) augmented
Cukierman, Webb, and Neyapti’s criteria by adding other aspects of central bank
independence that was emphasized in subsequent literature. The aspects they added were measures of limits on the reappointment of the CEO, measures of provisions affecting (re)appointment of other board members similar to those affecting the CEO, restrictions on government representation on the board, and intervention of the government in exchange rate policy formulation (Dincer, & Eichengreen, 2013). The unweighted average of the variables is called CBIU. Subsequently, they also give weights to the variables, resulting in a measure of legal independence called CBIW (Dincer, & Eichengreen, 2013). The criteria list for the extended legal central bank independence indices CBIU and CBIW are described in Appendix 2.
3.4 Data
In the following section, it is first described how the data is gathered. Secondly, the results of the scatterplots between the level of central bank independence and the level of inflation are presented and discussed.
The dependent variable, the inflation rate, is the annual inflation rate in consumer prices which is measured as the annual percentage change in the cost to the average consumer by acquiring a basket of goods and services that may be fixed or changed at specified
from the database of the Worldbank. Because there are large outliers in the inflation rates, the inflation variable is transformed in order to reduce heteroscedasticity of the error. As a result, there will be an improvement of the efficiency of the estimate (Cukierman, Webb, Neyapti, 1992). The inflation is transformed by dividing every year’s inflation rate by this inflation rate plus one. Next, the inflation rate of regressions (1) and (3) are calculated by taking the
geometric average over the decade of the transformed inflation rates. The variation of inflation of regressions (2) and (4) is calculated by the standard deviation of the transformed inflation rate over the decade.
As stated in section 3.2, there are two measures of legal central bank independence. One is the measure of Cukierman, Webb, and Neyapti (1992), and the other is the extended version of that measure, which is introduced by Dincer and Eichengreen (2013). For both measures, the weighted averages of legal independence are used. The weighted legal
independence measure of Cukierman, Webb, and Neyapti, LVAW, is used in the first period, and the weighted legal independence measure of Dincer and Eichengreen, CBIW, is used in the second period. For both periods, the arithmetic mean of the decade is included in the regression for every country.
The data about the explanatory variables Openness and FinancialDepth are also gathered from the database of the Worldbank. The variable Openness is the annual percentage of total trade to GDP, as it measures the openness to trade in every country. The variable FinancialDepth is the share of private credit by deposit money banks and other financial institutions to GDP, and captures the size of the financial sector relative to the economy. Finally, for both variables, the arithmetic mean is calculated for the decade.
For the first period, the turnover rates of the governors of the central banks are gathered from the article of Cukierman, Webb, and Neyapti (1992). They measured the turnover rate as the average number of changes per year in the central bank’s governor in the period of 1980 until 1989. The resulting turnover rate takes a value from zero to one. For example, if the governor is replaced every five year, the turnover rate equals 0.2. For the second period, the turnover rates were calculated using the same measure as Cukierman, Webb, and Neyapti. The information about the terms of the past governors is collected from the websites of the central banks.
In the period of 1998 until 2007, seven countries were deleted out of the sample when compared to the first period. This was due to missing data. As a result, the sample of the first period contains 47 countries, and the sample of the second period contains 40 countries.
Before the regressions are done, scatterplots between the relationship of the level of central bank independence and the level of inflation are made first. For the level of central bank independence, both the legal independence measure and the turnover rate of the
governor of the central bank are considered. The level of inflation is the geometric average of the transformed inflation rates.
The scatterplots of the first period and second period containing all countries are presented below in Figures 1 to 4. The scatterplots of the first period containing industrial and developing countries separately are shown in Figures 5 to 8 in Appendix 3. For the second period, these are shown in Figures 9 to 12 in Appendix 4.
As stated in the literature review, the relationship between the level of infla tion and the level of legal central bank independence is less strong in developing countries than in
industrial countries. Because of this, it is assumed that the scatterplot between legal
independence and inflation of the developing countries shows a less solid downward sloping line than the scatterplot of the industrial countries. Instead, if a scatterplot is made between the turnover rate of the governor of the central bank and the level of inflation for developing countries, it should show a solid upward sloping line since this positive relationship between the turnover rate and inflation is confirmed in the research of Cukierman, Webb, and Neyapti (1992).
In the first period, from 1980 until 1989, the scatterplots between the legal
independence measure and the transformed inflation result in upward sloping fitting lines. This is in contradiction with the theory that is discussed in the literature review. This result can be explained by the fact that in that period a lot of countries, both industrial and developing, suffered high inflation rates due to the high oil prices. Especially the Latin
American countries were hit hard, and they ended up in a financial crisis. Some of these Latin American countries, for example Argentina, Peru, and Nicaragua, not only faced high
inflation rates, but also had relatively independent central banks. As a result, the fitting line of the scatterplot that contains developing countries only is upward sloping. The upward sloping fitting line of the scatterplot of both the industrial countries and the developing countries can also be explained by the possibility that the legal independence measure is not a good proxy for the actual central bank independence in this period. As already mentioned in the literature review, legal independence is not a good proxy for actual independence in developing
countries due to the fact that there is less respect for the law. For the industrial countries, it could be that in the 1980s the laws of some central banks were incomplete and did leave a lot of room for interpretation. As a result, even when the law is quite explicit, the governance of
the central bank might be different in reality. Hence, the legal independence measure might not fully capture the actual central bank independence of industrial countries. So, in this period, the turnover rate might be a better measure of the level of central bank independence. As a matter of fact, the scatterplots between the turnover rate and the transformed inflation show an upward sloping line in both developing and industrial countries.
In the second period, which ranges from 1998 until 2007, the scatterplots between the legal independence measure and the transformed inflation are in line with the expectations based on the literature review. In the industrial countries, a negative relationship is shown and in the developing countries this negative relationship does not exist. Instead, it is a positive relationship, which might be caused by the fact that legal independence is not a good proxy for actual central bank independence in developing countries. Finally, the scatterplots between the turnover rate and the transformed inflation shows a positive relationship in both industrial and developing countries.
Figure 3. Scatterplot between legal CBI and inflation, 1998-2007
4. Results
4.1 OLS regressions
In the following subsections, the results of the OLS regressions of the two time periods and of the robustness check are presented and discussed. In all regressions, ‘*’ means that the value is significantly different from zero at the 10% significance level, whereas ‘**’ means that the value is significantly different from zero at both the 5% and the 10% significance level.
4.1.1 Results 1980-1989
The results of the regressions of the first time period are shown in tables A1 to A4. In Table A1, where the results of the regressions with the level of inflation as dependent variable are shown, the coefficient of LCBI is positive in all the regressions. The coefficient is, however, not significantly different from zero. In other words, the results suggest that the level of legal independence does not have a significant influence on the level of inflation. As discussed in section 3.4, this can be due to the fact that some countries, even those with relatively high legal independence, faced extremely high inflation rates in this period. Besides, legal independence may not be a good proxy for actual central bank independence in this period. The coefficient of the dummy variable EMU is significantly negative. This is also in contrast to what is expected. Apparently, the former EMU-countries already had something more that results in a lower level of inflation before the EMU was even established. A possible explanation for this result is that, during this period of high inflation rates, European countries were following the low inflation rates of Germany. As a result, the European countries faced relatively lower inflation rates than countries in the rest of the world.
Table A2 shows the results of the regressions with the variation of inflation as the dependent variable. The coefficient of LCBI is again positive, but now it is also significantly different from zero, which means that the level of legal independence has a significant positive effect on the variability of inflation. This result is in contradiction with the
expectations, but can be explained by the fact that in developing countries, central banks with a lot of formal independence were not very independent in practice, as discussed in the literature review. As a result, legal central bank independence is, at least in developing countries, not a good measure of actual central bank independence. This could result in a positive relationship between central bank independence and inflation, instead of a negative relationship.
Table A1. OLS regression of the level of inflation on the explanatory variables
Table A2. OLS regression of the variation of inflation on the explanatory variables
Coefficient (1) (2) (3) (4) (5) (6) Constant 0.1156 (0.094)* 0.1206 (0.067)* 0.1732 (0.026)** 0.2193 (0.006)** 0.2464 (0.004)** 0.1243 (0.051)* LCBI 0.1325 (0.417) 0.2040 (0.238) 0.1788 (0.303) 0.1882 (0.252) 0.1721 (0.286) 0.1295 (0.307) TOR 0.4100 (0.004)** Openness -0.0008 (0.024)** -0.0005 (0.113) -0.0007 (0.01)** FinancialDepth -0.0022 (0.001)** -0.002 (0.001)** -0.0014 (0.008)** EMU -0.1182 (0.001)** -0.1040 (0.002)** -0.0722 (0.011)** -0.0668 (0.015)** -0.0117 (0.684) R2 0.0093 0.1048 0.1643 0.2429 0.2709 0.4959 Standard error of the regression 0.16735 0.16088 0.15724 0.14965 0.1486 0.12507 Number of observations 47 47 47 47 47 47 Coefficient (1) (2) (3) (4) (5) (6) Constant 0.0279 (0.184) 0.0298 (0.112) 0.0387 (0.237) 0.0590 (0.005)** 0.0619 (0.006)** 0.0352 (0.115) LCBI 0.1046 (0.151) 0.1312 (0.068)* 0.1270 (0.083)* 0.1265 (0.065)* 0.1249 (0.074)* 0.1155 (0.074)* TOR 0.0896 (0.023)** Openness -0.0001 (0.170) -0.0001 (0.560) -0.0001 (0.388) FinancialDepth -0.0006 (0.007)** -0.0006 (0.01)** -0.0005 (0.028)** EMU -0.0440 (0.009)** -0.0416 (0.013)** -0.0304 (0.028)** -0.0299 (0.036)** -0.0178 (0.222) R2 0.0348 0.1144 0.1245 0.1871 0.1889 0.2534 Standard error of the regression 0.06738 0.06527 0.06564 0.06326 0.06393 0.06208 Number of observations 47 47 47 47 47 47
In contrast, the results of the regressions with the variable TOR as measure of central bank independence, as stated in Tables A3 and A4, are in line with the expectations. Both in the regressions with the level of inflation as dependent variable and in the regressions with the variability of inflation as dependent variable, the coefficient of TOR is significantly positive. This means that a higher turnover rate, so a lower level of central bank independence, will result in a significantly higher level and variability of inflation. In addition, the coefficient of the dummy variable EMU is not significant, which means that in this period it does not matter for the level and variability of inflation whether the country is an EMU-country in the future or not. This result is also in line with the expectations.
In addition to the fact that the results of the regressions with the variable TOR as measure of central bank independence match with the expectations, the R2 of those
regressions are higher than the R2 of the regression with variable LCBI as measure of central
bank independence. Especially in the first regressions in which either TOR or LCBI is the only explanatory variable, the difference in R2 is obvious. The R2 measures the fraction of the
variance of the dependent variable that is explained by the independent variables. So, in the regressions with the variable TOR, a greater part of the variance in the level and variability of inflation is explained by the turnover rate than by the legal independence measure. Together with the results of the sign and significance of the coefficients of LCBI, TOR, and EMU it can be concluded that the turnover rate is a better proxy for actual central bank independence in this period.
Table A3. OLS regression of the level of inflation on the explanatory variables
Table A4. OLS regression of the variation of inflation on the explanatory variables
Coefficient (1) (2) (3) (4) (5) Constant 0.0423 (0.162) 0.0602 (0.127) 0.1074 (0.006)** 0.1460 (0.014)** 0.1681 (0.002)** LCBI TOR 0.4692 (0.001)** 0.4407 (0.004)** 0.4543 (0.001)** 0.3956 (0.007)** 0.4151 (0.002)** Openness -0.0009 (0.001)** -0.0007 (0.007)** Financial Depth -0.0017 (0.007)** -0.0014 (0.012)** EMU -0.0417 (0.150) -0.0243 (0.417) -0.0133 (0.634) -0.0046 (0.863) R2 0.3455 0.3564 0.4371 0.4372 0.4873 Standard error of the regression 0.13603 0.13641 0.12905 0.12904 0.12461 Number of observations 47 47 47 47 47 Coefficient (1) (2) (3) (4) (5) Constant 0.0330 (0.001)** 0.0433 (0.004)** 0.0524 (0.001)** 0.0708 (0.002)** 0.0743 (0.001)** LCBI TOR 0.1206 (0.002)** 0.1056 (0.009)** 0.1082 (0.005)** 0.0911 (0.013)** 0.0941 (0.01)** Openness -0.0002 (0.124) -0.0001 (0.290) FinancialDepth -0.0005 (0.024)** -0.0005 (0.034)** EMU -0.0219 (0.093)* -0.0186 (0.178) -0.0129 (0.277) -0.0115 (0.352) R2 0.1371 0.1553 0.1735 0.2053 0.2125 Standard error of the regression 0.06371 0.06374 0.06378 0.06254 0.06299 Number of observations 47 47 47 47 47
4.1.2 Results 1998-2007
The results of the regressions of the period from 1998 until 2007 are shown in Tables B1 to B4. Overall, the coefficient of the variable LCBI is positive and not significantly different from zero in both the regression with the level of inflation as dependent variable and the regression with the variation of inflation as dependent variable, as shown in Tables B1 and B2. In other words, legal central bank independence does not influence the level and variability of inflation. This could, again, be due to the fact that legal independence is not a good proxy for actual independence in developing countries.
In addition, in the regression with the level of inflation as dependent variable, the coefficient of the dummy variable EMU is always negative, which is in line with the
hypothesis that being part of the EMU should have an additional negative effect on the level of inflation. The coefficient is, however, mostly not significant. As stated in the hypothesis, this could be explained by the assumption that legal independence is a good measure of the actual independency of the ECB, and that the EMU-countries do not have some extra effect that lowers their level of inflation. In contrast, if the regression with the variation of inflation is considered, the dummy variable EMU is both negative and significant. This suggests that being an EMU-country will result in a lower variation of inflatio n.
Table B1. OLS regression of the level of inflation on the explanatory variables
Table B2. OLS regression of the variation of inflation on the explanatory variables
Coefficient (1) (2) (3) (4) (5) (6) Constant 0.0505 (0.000)** 0.0215 (0.131) 0.0304 (0.073)* 0.0929 (0.000)** 0.0965 (0.000)** 0.0779 (0.001)** LCBI -0.0088 (0.614) 0.0863 (0.054)* 0.0793 (0.076)* 0.0181 (0.594) 0.0154 (0.654) 0.0069 (0.839) TOR 0.1202 (0.213) Openness -0.0001 (0.075)* -0.0000 (0.355) -0.0000 (0.479) FinancialDepth -0.0006 (0.000)** -0.0006 (0.000)** -0.0006 (0.000)** EMU -0.0689 (0.01)** -0.0639 (0.016)** -0.0260 (0.167) -0.0239 (0.208) -0.0125 (0.475) R2 0.0020 0.1833 0.1954 0.4656 0.4685 0.4991 Standard error of the regression 0.04998 0.04582 0.04611 0.03758 0.03801 0.03744 Number of observations 40 40 40 40 40 40 Coefficient (1) (2) (3) (4) (5) (6) Constant 0.0232 (0.000)** 0.0032 (0.701) 0.0062 (0.525) 0.0335 (0.003)** 0.0344 (0.005)** 0.0171 (0.177) LCBI -0.0025 (0.812) 0.0631 (0.041)** 0.0608 (0.053)* 0.0342 (0.158) 0.0336 (0.176) 0.0258 (0.272) TOR 0.1097 (0.067)* Openness -0.0000 (0.282) -0.0000 (0.569) -0.0000 (0.950) FinancialDepth -0.0003 (0.002)** -0.0003 (0.002)** -0.0003 (0.001)** EMU -0.0476 (0.008)** -0.0459 (0.011)** -0.0294 (0.033)** -0.0289 (0.04)** -0.0185 (0.148) R2 0.0005 0.2679 0.2721 0.4248 0.4253 0.5041 Standard error of the regression 0.02845 0.02467 0.02494 0.02217 0.02247 0.02118 Number of observations 40 40 40 40 40 40
The results of the regressions where the turnover rate of the governor of the central bank is the measure of central bank independence are somewhat different. The results are shown in Tables B3 and B4. In the regressions with the level of inflation as dependent variable, the coefficient of TOR is always positive, but not significant. Hence, the turnover rate of the central bank’s governor has a positive effect on inflation, but this effect is not significantly different from zero. In the regressions where the variation of inflation is the dependent variable, the coefficient of TOR is always positive, and, at least in the regression of the total model, significantly different from zero. Hence, when the control variables and the dummy variable are added to the regression, the turnover rate has a significant positive effect on the variation of inflation.
Finally, in both the regressions with the level of inflation as dependent variable and with the variation of inflation as dependent variable, the coefficient of EMU is negative in all regressions and not significant in the regressions of the total model. This could be explained by the assumption that the turnover rate is a good measure of the actual level of independence. The variable TOR captures the high independency of the ECB, and the ECB does not give its countries something more that results in a significantly lower level or variability of inflation.
Table B3. OLS regression of the level of inflation on the explanatory variables
Coefficient (1) (2) (3) (4) (5) Constant 0.0153 (0.381) 0.0326 (0.105) 0.0428 (0.039)** 0.0783 (0.000)** 0.0807 (0.000)** LCBI TOR 0.1797 (0.155) 0.1228 (0.325) 0.1132 (0.351) 0.1252 (0.189) 0.1223 (0.202) Openness -0.0001 (0.031)** -0.0000 (0.447) Financial Depth -0.0006 (0.000)** -0.0006 (0.000)** EMU -0.0255 (0.01)** -0.0233 (0.014)** -0.0098 (0.244) -0.0093 (0.272) R2 0.0824 0.1316 0.1549 0.4966 0.4986 Standard error of the regression 0.04792 0.04725 0.04725 0.03647 0.03691 Number of observations 40 40 40 40 40
Table B4. OLS regression of the variation of inflation on the explanatory variables
4.2 Robustness check
The number of countries in the sample is small because of the limited data about the legal independence variable in the two time periods. To check whether the results discussed in the previous subsections are also valid for a larger set of countries, a robustness check is done. A new sample is made, including 63 counties in the period of 1980-1989 and 56 countries in the period of 1998-2007. Because the legal independence measure is not available for this larger set of countries, only the turnover rate is used as a proxy for the actual level of central bank independence. Next, the same regressions are done as stated in section 3.1. The results of the regression of the first period are shown in Tables C1 and C2 and the results of the regressions of the second period are shown in Tables D1 and D2.
In the first period, the coefficient of the variable TOR is significantly positive in all the regressions. Hence, the expectation that the turnover rate has a positive effect on the level and variability of inflation is confirmed in this larger sample. In the original regressions of section 4.1.1, the coefficient of TOR was also significantly positive. Hence, the results are robust.
In the regressions of the total model, the coefficient of the dummy variable EMU is negative and not significant. In the original regressions, the same results occurred. As noted
Coefficient (1) (2) (3) (4) (5) Constant -0.0034 (0.763) 0.0062 (0.615) 0.0105 (0.397) 0.0283 (0.003)** 0.029 (0.003)** LCBI TOR 0.1487 (0.075)* 0.1173 (0.159) 0.1132 (0.167) 0.1184 (0.047)** 0.1176 (0.051)* Openness -0.0000 (0.053)* -0.0000 (0.607) FinancialDepth -0.0003 (0.001)** -0.0003 (0.001)** EMU -0.0141 (0.003)** -0.0131 (0.004)** -0.0064 (0.122) -0.0063 (0.135) R2 0.1744 0.2206 0.2336 0.4848 0.4853 Standard error of the regression 0.02585 0.02546 0.02559 0.02098 0.02127 Number of observations 40 40 40 40 40
earlier, the fact that the value of the coefficient is not significant suggests that the turnover rate is a good proxy for actual central bank independence. In other words, the high
independency of the ECB is fully captured by the variable TOR and the EMU-countries do not have something extra that lowers their level or variability of inflation. Since this result is carried over in the larger sample, it can be stated that the result is robust.
Table C1. OLS regression of the level of inflation on the explanatory variables
Coefficient (1) (2) (3) (4) (5) Constant 0.0491 (0.071)* 0.0613 (0.057)* 0.1227 (0.000)** 0.1329 (0.001)** 0.1667 (0.000)** TOR 0.4349 (0.001)** 0.4149 (0.003)** 0.4281 (0.000)** 0.3929 (0.002)** 0.4089 (0.000)** Openness -0.0011 (0.000)** -0.0009 (0.000)** Financial Depth -0.0017 (0.001)** -0.0013 (0.002)** EMU -0.0392 (0.095)* -0.0231 (0.322) -0.0002 (0.993) 0.0039 (0.856) R2 0.2928 0.3021 0.4205 0.3988 0.4736 Standard error of the regression 0.13198 0.13219 0.12147 0.12373 0.11676 Number of observations 63 63 63 63 63
Table C2. OLS regression of the variation of inflation on the explanatory variables
In the second period, the coefficient of the variable TOR is taking a negative, but not significant, value in the regression of the total model where the level of inflation is the dependent variable. In the original regressions of section 4.1.2, the coefficient of TOR was also positive but not significant in the regressions where the level of inflation was the dependent variable. Hence, the result that the turnover rate does not have a significant effect on the level of inflation is robust. In the regressions where the variation of inflation is the dependent variable, the coefficient of TOR takes a significant positive value in both the original regression and the regression of the robustness check. So, the result that the turnover rate has a significant positive effect carries over in the larger sample. In other words, it is robust.
The coefficient of the EMU variable is, both the regression with the level of inflation as dependent variable and in the regression with the variation of inflation as dependent variable, negative and not significant. This is in line with the original regressions in section 4.1.2. In both cases it can be assumed that the turnover rate is a good proxy for the actual level of independence and that the EMU-countries do not have a significant lower level or variability of inflation when compared to other countries in the world. Because the results carry over in the larger sample, it can be assumed that these are robust.
Coefficient (1) (2) (3) (4) (5) Constant 0.0349 (0.000)** 0.0418 (0.000)** 0.056 (0.000)** 0.0636 (0.000)** 0.0706 (0.000)** TOR 0.1278 (0.001)** 0.1166 (0.002)** 0.1196 (0.001)** 0.1098 (0.003)** 0.1132 (0.002)** Openness -0.0002 (0.006)** -0.0002 (0.033)** Financial Depth -0.0005 (0.007)** -0.0004 (0.017)** EMU -0.022 (0.026)** -0.0182 (0.074)* -0.01 (0.290) -0.0092 (0.337) R2 0.1448 0.1616 0.1980 0.2130 0.2319 Standard error of the regression 0.06063 0.06053 0.0597 0.05914 0.05893 Number of observations 63 63 63 63 63
Table D1. OLS regression of the level of inflation on the explanatory variables
Table D2. OLS regression of the variation of inflation on the explanatory variables
Coefficient (1) (2) (3) (4) (5) Constant 0.0495 (0.045)** 0.0718 (0.043)** 0.0824 (0.025)** 0.1168 (0.011)** 0.1184 (0.01)** TOR 0.0394 (0.673) -0.0316 (0.791) -0.0303 (0.801) -0.0300 (0.798) -0.0298 (0.801) Openness -0.0001 (0.046)** -0.0000 (0.668) Financial Depth -0.0007 (0.001)** -0.0007 (0.001)** EMU -0.0461 (0.057)* -0.0433 (0.076)* -0.0204 (0.265) -0.0201 (0.283) R2 0.0022 0.0437 0.0527 0.1918 0.1921 Standard error of the regression 0.08741 0.08637 0.08679 0.08017 0.08093 Number of observations 56 56 56 56 56 Coefficient (1) (2) (3) (4) (5) Constant 0.0179 (0.108) 0.0299 (0.059)* 0.0364 (0.026)** 0.0518 (0.009)** 0.0537 (0.007)** TOR 0.0546 (0.239) 0.0165 (0.773) 0.0173 (0.761) 0.0172 (0.738) 0.0175 (0.736) Openness -0.0001 (0.018)** -0.0000 (0.246) Financial Depth -0.0003 (0.000)** -0.0003 (0.001)** EMU -0.0248 (0.022)** -0.0230 (0.034)** -0.0122 (0.131) -0.0119 (0.153) R2 0.0191 0.0733 0.0887 0.2330 0.2350 Standard error of the regression 0.0407 0.03993 0.03997 0.03667 0.03698 Number of observations 56 56 56 56 56
5. Conclusion
The aim of this paper was to examine the effect of the establishment of the ECB as an independent central bank on inflation in the EMU-countries. The highly researched relationship between the level of central bank independence and inflation appears to be negative in previous studies. Because the ECB is the most independent central bank in the world, it is interesting to analyse whether this high independency gave the EMU-countries an extra boost in lowering the level and variability of inflation.
The research was done by four OLS regressions in which the dependent variable is either the level or variation of inflation. The level of central bank independence is measured by either the legal independence or the turnover rate of the governor of the central bank. Based on the findings of previous studies discussed in the literature review, the hypotheses were that the coefficient of the LCBI variable should be significantly negative in all
regressions, since this would mean that a higher level of central bank independence lowers the level and variation of inflation. By the same way, the coefficient of the TOR variable should be significantly positive. Furthermore, the coefficient of the dummy variable EMU should not have a significant effect on the level and variation of inflation in the first period. In the second period, a significant negative value of the coefficient of the variable EMU would suggest that the independence variable is not fully capturing the actual level of central bank independence. In this case, being part of the EMU might give a country an extra boost in lowering the level and variability of inflation. In contrast, if the coefficient of the EMU variable is not
significant, this means that it can be assumed that the independence variable is a good
measure of actual central bank independence. The high independency of the ECB is now fully captured by the independence variable.
The estimation outcomes of the regressions were not always in line with the hypotheses. In the first period, the coefficient of the variable LCBI is never found to be significantly negative, as was expected. The coefficient of the variable TOR, however, did have the expected significantly negative value. From this it can be concluded that the turnover rate is a better proxy of actual central bank independence than the legal independence
measure. This result is also confirmed in the scatterplots of section 3.4.
Besides the striking results of the sign and significance of the coefficient of the variable LCBI, the regressions with LCBI as independence variable resulted in a significantly negative value of the coefficient of the dummy variable EMU. In the regressions where the turnover rate was the independency variable, the coefficient of EMU had the expected nonsignificant value. The reason of these differences between the outcomes under legal
independence and the turnover rate can, again, be due to the fact that legal independence is not a good proxy for actual independence. Hence, it will lead to biased results.
In the second time period, the legal independence again did not have a significant influence on the level and variation of inflation. The turnover rate did not have a significant effect on the level of inflation, but did have a significant positive effect on the variation of inflation. In addition, the coefficient of the dummy variable EMU was mostly not significant, which means that the high independency of the ECB is fully captured by the independency variable and that there is no additional effect of being part of the EMU on inflation.
To sum up, the results show that the turnover rate is a better proxy for the level of actual central bank independence than the legal independence measure. When the level of central bank independence is measured by this turnover rate, the estimation outcomes show a negative relationship between the level of central bank independence and the level and variability of inflation. Finally, being part of the EMU will, overall, not have a significant additional effect on inflation.
The research can be improved in several ways. First of all, the sample should be extended by using a larger sample, provided that more data should be available. In this sample, only 12 EMU-countries were involved. At this moment, the Eurozone consists of 19 countries. Furthermore, it might be interesting to analyse the effect of the recent financial crisis on inflation for, on the one hand, the Eurozone, and on the other hand countries with lower central bank independence. By doing this, it can be examined whether having an independent central bank is also beneficial for the level and variability of inflation in times of crisis. Finally, it might be necessary to create a more realistic measure of central bank
independence. Using the turnover rate only is worrisome, because a high turnover rate does not necessarily mean that the central bank is more independent. Besides, the legal
independence measure does not take into account to what happens in reality. A combination of legislation and actual occasions might be a better measure of actual central bank
