The effectiveness of Quantitative Easing on fighting deflationary tendencies in Europe.
Bachlor Thesis
Author: Demi Vollemans
Student Number: 10313680
Supervisor: Damiaan Chen
University of Amsterdam
Specialization: Economics and Finance
Table of Contents
1. Introduction 2. Literature review
3. Empirical data and model. 4. Results
5. Conclusion 6. Bibliography 7. Appendix
Statement of Originality
This document is written by Student Demi Vollemans 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.
1. Introduction.
The European Central Bank (ECB) was founded on the 1st of July 1998, since then the European economy has suffered from an economic downturn. The most recent example of an economic downturn is the Financial Crisis in 2008. In times of economic distress central banks try to boost the economy by implementing stimulating monetary policies. In October 2008 the European Central Bank saw the spread between the 3-month EURIBOR and the overnight index swap rate deteriorate, which indicates that the economic situation in Europe was worsening. Then the ECB decided to implement Credit Easing, this means that
commercial banks were able to borrow as much as they needed at a fixed rate (Fawley, 2013). The ECB chose this policy in order to enhance the liquidity of the economy. By doing so the ECB tried to stimulate investments and consumption. To prevent the economy from
stagnating more the Bank of England and Bank of Japan decided to choose an even more rigorous monetary policy, namely Quantitative Easing (QE). This is a program where different assets are purchased by the Central Bank, which basically means that the money supply is increased.
The Credit Easing didn’t have enough impact to boost the inflation in Europe, so on 22 January 2015 ECB announced to implement an expanded asset purchase program. This
program implies that the ECB is going to purchase assets up to an amount of 60 billion euro, on a monthly bases. The ECB will continue these purchases until at least September 2016. Last December the ECB announced that the QE program will be extended till March 2017, because the program has not produced the desired effects yet. In England and Japan, where the Central banks implemented QE earlier, they are doubtful about the effects of the expanded asset purchasing program.
Up to now a significant amount of research on the monetary; Quantitative Easing. Most empirical research has been done on the effect of QE on the interest rate and economic growth. The overall results of the research were ambiguous (Fawley, 2013). Empirical
analysis by Kimura et al. (2003) tried to contribute to the ongoing discussion about the effects of this monetary policy at zero interest. They examined the Quantitative monetary Easing in Japan which adopted the program in March 2001, and tried to examine the effects of the program. In Japan, the year on year the money base grew by 20 percent and the interest rates are nearly zero. A Bayesian-VAR research is used where overtime coefficients can change since they assumed the money demand is non-linear. The most evident effects were that when QE implemented it provided extremely favorable monetary circumstances, and thereby it
succeeded in maintaining financial market stability. The abundant provision of liquidity has significantly calmed down the investors’ concerns about liquidity financing, this might have played a key role in fighting a deflationary spiral. Currently the Bank of Japan (BoJ) is still increasing the money supply but it fails to boost inflation.
Since QE is introduced by the ECB the deflationary tendencies have also not been subsided, and little research has been done on the effects of QE on inflation that is why this paper can contribute to the ongoing discussion about the effects of QE on short-term inflation expectations, by answering the following question: “ Is money supply a strong instrument to fight deflationary/low inflation expectations, when used as Quantitative Easing?”. This is an important question because Central Banks are implementing monetary policies to stimulate their economy. When it turns out that the stimulating effect of increasing money supply is no longer a strong tool to change the inflation expectations, the ECB should consider changing their policy.
A stable economy consists of stable inflation- and interest rates, this is important for economic decision making. According to the Fischer equitation the inflation expectations are also an important determent for the real interest rate. Because inflation expectations influence the real interest rate it also has an effect on the economy as a whole. The lower inflation expectations are, the higher the real interest rate is, the more difficult it will be for the ECB to prevent further deflationary tendencies. So it is important to find out whether inflation
expectations are still stirred by implementing Quantitative Easing, and if they are not, this monetary policy can be considered to be ineffective.
To investigate if the inflation expectations in the European economy are stirred in times of QE, Ordinary Least Squares(OLS) estimations will be conducted. The all regressions are based on the Taylor rule, the equitation is rewritten to a function where the inflation expectations are the dependent variable. The second regression will be an elaboration of the first regression, with variables who might explain the inflation expectations like crude oil prices, money supply and a dummy variable for times when QE is implemented. The data and numbers of the variables and inflation expectations will be retrieved from DataStream.
Firstly, this paper presents a literature review, which elaborates on the policy of the ECB, and the effects of Quantitative Easing on expectations according to macro-economic theories. Secondly, the empirical model will be introduced and the reason why particular variables are chosen will be explained. Thirdly, the results of the regression will be shown. Finally, the results will be analyzed and a conclusion is drawn.
2. Literature review.
This literature review is set up in the following way. The objectives of monetary policy of the ECB and how they achieve this in normal times will be explained. The next part elaborates on macro-economic theories on the effect of increasing money supply, and the liquidity trap will be discussed. This will be followed up by explanations according research on the situation in Japan, where the Bank of Japan failed to increase inflation. Thereafter the economic situation of Europe and what important variables could possibly play a role in affecting the inflation expectations will be discussed.
The primary objective of the ECB is to guarantee price stability. The quantitative definition of price stability according to the ECB is defined as: “a year-on-year increase in the Harmonized Index of Consumer Prices (HICP) for the euro area of below 2%”. Price stability has to be maintained over the medium term" (European Central Bank, 2015). Through maintaining stable inflation the ECB wants to minimize unemployment, create a stable increase in economic welfare and contribute to financial stability. By making
statements the Central Bank tries to increase its transparency, in order to make sure agents are well-informed and can create expectations, so they can allocate their resources well (European Central Bank, 2015). Their reality is different, since the 4th quarter of 2011 the rate of
inflation in Europe has only been declining, nevertheless the monetary tools the ECB uses to fight low inflation.
Central banks use the unconventional tools such as quantitative easing to help a stagnating economy when traditional monetary tools are not effective anymore. Quantitative easing simply increases money supply and should be a stimulant for the economy when the key interest rates converge to their lower bound. Quantitative Easing increases liquidity and low interest rates create an incentive for households and firms to invest (European Central Bank, 2015). As stated in the introduction, the actual effects of QE on the economy are ambiguous. There are several theories that discuss why this policy can be ineffective and possibly disadvantageous for the economy, when it end up in a trap. The theory discussed is Keynes’s traditional “liquidity trap”, introduced in the paper of Hicks in 1937. A liquidity trap is a situation where increasing money supply does not have an effect on the interest rate anymore. The liquidity trap occurs when an increasing money supply only affects nominal interest rates. Hypothetical, according to the theory developed by Kregel, central banks will not set nominal interest rates below zero. When the nominal interest rates hit the zero bound, the real interest can only be reduced when the inflation is increasing. If prices are not
responding or increasing, the effect will be that people prefer, or become indifferent to holding cash instead of bonds, so they can become perfect substitutes (Kregel, 2000). When increased money supply by the Central Bank doesn’t result in a lower real interest rate, this indicates that the policy is ineffective. In situations like this the economy is said to be in a liquidity trap (Kregel, 2000, p. 1). The first theories of the liquidity trap were designed after this phenomenon occurred during The Great Depression in the 1930’s. Short-term Interest rates were close to the zero bound, but unless the unconventional monetary policy the economy was not responding in a positive way.
Nowadays there are more sophisticated views on the “liquidity trap”, often referred to as “the modern view of the liquidity trap” which takes inflation expectations into account. Since the 1990’s more data has been available which enables researchers to do better research on this phenomenon. Modern research sheds new light on the recovery of the economy after The Great Depression. Evidence was found by Eggertsson (2008), who suggests that the modern analysis of the liquidity trap indicates that monetary policy was far from being ineffective during the Depression. Eggertsson claims that the economy recovered due to people’s expectations. He/she states that the expectations about the future money supply and inflation were the key factors in determining the aggregate demand (Eggertsson, 2008).
Since the asset prices collapsed in the early 1990’s the Japanese economy has been in a spiral of low economic growth. The Bank of Japan (BoJ) managed to keep the overnight call rate to the zero bound until August 2000.The moment the economy showed some recovery the rate was increased to 0.25 percent (Kimura, 2003). However in March 2001, when the
economy stagnated again, the BoJ announced its new monetary policy, namely Quantitative Easing. The BoJ increased the number of current accounts for commercial banks. This should result in a decrease to zero in the already low overnight call rate (Spiegel, 2006). The growth rate of the Monetary Base in the first year of QE was about 30%, the overnight call rate dropped to 0.001 percent. An important element of this policy was that the BoJ’s stated it would continue this QE regime until the deflationary tendencies will have subsided. In October 2013 the BoJ expanded the program by increasing the invested money to the record amount of $102 billion a month but still the market was not showing an increase in inflation. Therefore, it can be said that Japan is in a trap. (Krugman, 1998)
Kregel states that a liquidity trap is a kind of a credibility problem, if the monetary policy is not effective, and is in a liquidity trap, it must be so that the Central Bank fails to make the public believe the policy is going to be sustained (2000). People with “rational expectations” fail to believe that the money supply will constantly be increased by the Central
Bank even when the rates become higher at the long-term. With new higher interest rates in the long run, and therefore also fail to increase inflation constantly (Kregel, 2000). The Fischer relation is used to determine the real interest rate (𝑖), and it depends on expectations of inflation (𝜋𝑒) and the nominal interest rate (𝑟).
𝑖 = 𝑟 + 𝜋𝑒
The problem in Japan is deflation, the money growth rate is increasing, but the prices are declining. The only explanation for this phenomenon is that it is expected that the QE program will be reversed at some point, which results in a negative money growth rate, and the stimulating effect on inflation weakens. Since the real interest rate cannot be negative, and the people expect deflation (−𝜋𝑒), the economy gets stuck in a situation where the nominal interest rate is higher than the real interest rate (Kregel, 2000, p. 5). The credibility problem is not about the BoJ failing to create inflation, but it is about the controlling of the yield curve (the long term rates). Even if Japanese investors would believe that the central bank is able to set money growth rates and its short term interest rates, they would not believe that the bank is in control of the shape of the yield curve, which they do not want to be going upward, but which it normally is. When investors believe the central bank is able to create inflation , due to the zero-bid policy there is a chance of zero that the short-term interest rate would decrease more since they cannot go below zero. Therefore, the expected change in short term rates will become positive. This means that if the short term rates rise by more than the square of themselves it is expected that the long term rates will rise more than that. This will eventually cause the expectation that the economy will fail to respond positively to the policy (Kregel, 2000, p. 6). A stationary yield curve cannot be guaranteed, even if the BoJ succeeded in implementing a credible monetary inflation policy. To stimulate the economy investors should increase expectations for the rate of return on new investments, which will not happen if the interest rate rises along with the rate of inflation (Kregel, 2000, p. 7).
Japan is the most recent and the most discussed case where QE is implemented but fails to stimulate the economy. Opponents of the policy are therefore doubtful about the effects of QE. They predict that Europe, will face the same struggle to meet the ECB’s objective to keep inflation “close but below the 2% level”. As Kregel stated, Quantitative Easing mainly works through expectations. An inflation trap is therefore only a true trap when the Central Bank fails to stir expectations (Eggertsson, 2008). People in Japan a expecting too much, the BoJ is proving a lot of liquidity to the commercial banks but they do not increase the liquidity to the their consumers (Wes Goodman, 2015). When the bank is credible enough
to create positive inflation expectations in the Eurozone, then the monetary base is not an exhausted instrument and Quantitative Easing is a good policy to stimulate the economy. Quantitative Easing is not the first expanding policy the ECB is implementing to calm or stimulate the markets, before the Financial Crisis the ECB eased money lending for central banks at a prevailing policy rate to increase liquidity. When the Financial Crisis erupted, the ECB, lowered the rates even more and enhanced credit support (Jürgen Stark, 2009). In 2015 the ECB realized they were out of tools and implemented QE. Liquidity and the balance sheet of the ECB were expanded when the QE program started. This paper will investigate whether the effects of increasing the monetary base on inflation expectations are significant.
Hoffman and Zhu (2013) investigated in their research the effects of large asset purchased by the Bank of England and the Federal Reserve (Fed) on inflation expectations. They conducted an event-study analysis and looked at the effects of the announcements on the inflation expectations. The analysis suggests that there was a statistically significant impact on the medium- and long term inflation swap expectations in particular in the United States. In the United Kingdom the effects were not significant, this could indicate that other factors were influencing the inflation expectations, such as changing the economic sentiment (Hofmann, 2013, p. 33).
In 2010 Guidolin and Neely conducted a small research about whether the large-scale asset purchasing program of the Fed changes inflation expectations on 10-year Treasury Inflation-Protected Securities (TIPS). They showed that there was a small effect on the 10-year inflation expectations, according to Neely this effect should not be neglected, although it does not necessarily correlate with the final inflation expectations. Expectations rose after the announcement of the program, so it might have limited power to raise TIPS –implied
expected inflation. Therefore, Central Banks might conclude that the Purchasing Program is helpful to fight deflation (Neely, 2008).
Other factors like the dramatic fall of the oil prices between July 2014 and January 2015 could play a role in explaining the low inflation. The logarithms of crude oil prices in the energy component of the Consumer Price Index (CPI) showed a remarkable stable relationship in the period 1995 till 2015 in the United States ( Alejandro Badel, 2015). In Europe it has already been proved that the fall of oil prices have a direct effect on inflation via the energy component of the HICP. A key variable in explaining the inflation expectation is actual inflation, this means that when actual inflation rates are low or even negative, it might lower the future inflation expectations (Clemente De Lucia, 2014).
3. Emperical Model and Data.
3.1 Description of the Data.
The all regressions will be done in the period January 2011 till January 2016. In January 2015 the ECB announced that the bank would begin in March to start buying assets, but it is
assumed that the data and market is already influenced via the Efficient Market Hypothesis, since the announcement in January. I realize the sample of the regression is small, and not satisfying The Rule of Thump (which stated that at least 100 samples are needed for a
reliable regression), but maybe we are still able to find significant results. For regressions, it is chosen to start sampling data in 2011, because when a larger time span is used, the data could still be influenced by chocks from the economic crisis. So we assume that since January 2011 the economy is resorted from significant influences in the previous period.
Monthly data is used for all variables, except for the GDP gap and inflation
expectations. Since the data for GDP is only available on quarterly basis, and the ECB Survey of Professional Forecasters (SPF) is a quarterly survey of expectations for the rates of
inflation. This is solved by to interpolate the quarterly data. This is the best way to determine the values between the known samples. The only data available for inflation expectations is a quarterly survey, these forecasters are most reliable, compared to other possible parameters for expectations like 1-year inflation linked swaps. Also to determine the missing monthly values, interpolation is used. All variables in the period February 2015 till December 2015 were given a 1 to indicate that Quantitative Easing was conducted that period. All periods before February 2015 are given a 0 indicating that Quantitative Easing was not conducted that period. All other variables are retrieved from the database Datastream. The total number of observations of the estimation are 59. A summary of the descriptive data is represented in table 1.
Table1
Variable Mean St. Deviation Minimum Maximum
Expected inflation 1.445 0.8823 0.1 2.6 Inflation 1.4 1.113 -0.6 3 GDP gap -2.761 0.8960 -3.68 -0.89 Ln(Monetary Base) 15.4925 0.10407 15.3639 15.6997 3-month EURIBOR- rate 0.475 0.5317 -0.1263 1.5976
Crude oil prices 96.71 24.7088 37.13 125.77
Dummy variable for QE
0.2 0.4034 0 1
3.2 The Model.
The main question in this paper is:
Is money supply(QE) a good instrument to positively stir inflation expectations in the Eurozone?
To answer this question, the following model is created, with inflation expectation in period t as the dependent variable. To see if the variable money growth has an significant effect on inflation expectations, 3 different regression will be conducted. The all regressions are based on The Taylor Rule. The variable desired inflation is replaced by inflation expectation, the Taylor Rule is defined below. The second and third regression are an elaboration of the first regression, with additional variables to check if these variables have a significant effect on the inflation expectations.
The Taylor Rule:
𝑖𝑡 = 𝜋𝑡𝑒+ 𝑟𝑡+ 𝛽𝜋(𝜋𝑡𝑒− 𝜋𝑡∗) + 𝛽𝑦(𝑦𝑡− 𝑦𝑡∗ ) (1) (1+𝛽𝜋) 𝜋𝑡𝑒 = 𝑖 𝑡− 𝑟𝑡+ 𝛽𝜋𝜋𝑡∗− 𝛽𝑦(𝑦𝑡− 𝑦𝑡∗ ) (2) 𝜋𝑡𝑒 = 1+𝛽𝛽𝜋 𝜋 𝜋𝑡 ∗+ 1 1+𝛽𝜋(𝑖𝑡− 𝑟𝑡) + −𝛽𝑦 1+𝛽𝜋(𝑦𝑡− 𝑦𝑡 ∗ ) (3)
The Fisher equitation is assumed, this gives the following;
𝜋𝑡𝑒 = 𝛽
0+ 𝛽1𝜋𝑡− 𝛽2(𝑦𝑡− 𝑦𝑡∗ ) + 𝜀𝑡 (4)
Regression (4) will the first model that is tested. Below are the elaborations of the model.
𝜋𝑡𝑒 = 𝛽0+ 𝛽1𝜋𝑡+ 𝛽2(𝑦𝑡− 𝑦𝑡∗ ) + 𝛽3𝐿𝑛(𝑀1𝑡) + 𝛽4(𝑂𝑖𝑙𝑡− 𝑂𝑖𝑙𝑡−1) + 𝜀𝑡 (5) 𝜋𝑡𝑒 = 𝛽0+ 𝛽1𝜋𝑡+ 𝛽2(𝑦𝑡− 𝑦𝑡∗ ) + 𝛽3𝐿𝑛(𝑀1𝑡) + 𝛽4(𝑂𝑖𝑙𝑡− 𝑂𝑖𝑙𝑡−1) + 𝛽5𝑄𝐸 + 𝜀𝑡 (6) 𝜋𝑡𝑒 = 𝛽0+ 𝛽1𝜋𝑡+ 𝛽2(𝑦𝑡− 𝑦𝑡∗ ) + 𝛽3(𝑂𝑖𝑙𝑡− 𝑂𝑖𝑙𝑡−1) + 𝛽4𝑄𝐸 + 𝜀𝑡 (7)
2. Explanation of the variables:
𝜋𝑡𝑒 = the expected inflation rate in period t 𝛽0 = the constant of the regression
𝑖𝑡 = nominal interest rate
(𝑦𝑡− 𝑦𝑡∗ ) = The natural logarithm of the Gross Domestic Product Gap, the output gap, the annual difference between target output and real output measured in quarters in period t
𝑂𝑖𝑙𝑡− 𝑂𝑖𝑙𝑡−1 = the first difference for crude oil prices by subtracting period t-1 from period t 𝐿𝑛(𝑀1𝑡) = the natural logarithm of the monetary base (M1), of the Eurozone, in period t QE = the dummy variable indicating, if Quantitative Easing is contacted.
𝜋𝑡 = the rate of inflation in period t, measured from Harmonized Index of Consumer Prices (HICP) of the ECB. HICP includes,
𝜋𝑡∗ = the desired inflation rate of the ECB in period t (constant value, “close but below 2%)
𝜀𝑡 = the error term in period t 3.3 Prediction of the regression coefficients.
In this section the relationship of the chosen independent variables with the dependent variable are predicted. The first independent variable is the inflation rate in the traditional Taylor equitation (1), there, 𝛽𝜋 and 𝛽𝑦 are estimated to be 0.5, so assumed that 𝛽𝜋 and 𝛽𝑦 are greater than 0 but smaller than 1, the constant 𝛽0 is predicted to be positive. After the Taylor-rule is rewritten, given the values of 𝛽𝜋 and 𝛽𝑦, expected is that the first dependent variable (𝛽1) is there for predicted to be positive and (𝛽2) predicted to be negative.
The intuition behind 𝛽1 is that inflation in period t, is an important variable that explains most of the inflation expectations and inflation in the next period. Miccoli and Neri (2015) did a research on “surprises” on inflation expectations, where oil prices and
unemployment were the most important variables. They found that these “surprises” did have an effect on inflation expectations. They suggested that adaptive behavior forms expectations
in the Eurozone. When events take place that might be disadvantageous for the economy, the expectations become more negative in the future. For instance, when the real inflation rates disappoint for a long time this might harm the expectations in the next period. Expected is that the coefficient will be positive, because any rate of inflation will raise the expectations in the next period, and deflation will lower the expectation in the next period.
The second explanatory variable is the output gap. When the economy is in a
downturn the real output will deviate from the “potential” output, which means the gap will be negative. A negative gap could indicate economic stagnation for investors, this would negatively influence the inflation expectations in the next period. However, a positive gap would not be good as well, this could mean the economy is overheated. In a situation like this there is too much demand and factories and workers work beyond their efficient level. This scenario is not very likely to happen in the Eurozone. In the graph in the appendix it seen that the output gap is getting smaller overtime but the numbers are positive. Therefor we expect a negative coefficient for 𝛽2. The output gap is calculated the according to the following equation: 100 * (Ln(GDP, real) - Ln(Potential output))).
The variable ln(M1) is taken into account in the second (5) and third estimation (6) of inflation expectations. Because the numbers of money supply are very high, the natural logarithm is used. Monetary Base is included in the regression to see if the Quantitative Easing program has an effect via money on expected inflation . The ECB wants this variable to be significant positive, the main goal of the ECB is to maintain price stability. If increasing the money supply has no effect on inflation expectations, then this is a serious problem, because when the ECB cannot stir the expectations of agents, the European economy might end in a liquidity trap, like the case is in Japan.
The fourth variable is crude oil prices, this coefficient should function as a correction for inflation, since it is possible that the inflation is not that low if it is corrected for low oil prices. When the low oil prices are taken into account in the inflation expectations, it is expected that this will positively influence the dependent variable. There is a strong positive correlation of 0.8019 between inflation expectations and oil prices. So there is expected that an increase in oil prices, will increase inflation expectations, so a positive coefficient for the first difference for crude oil prices is predicted. In this model the first difference for Crude oil Prices is used, the Dickey-Fuller test showed that the variable was not stationary, to solve this, there is chosen to use the first difference.
In regression (6) a dummy variable for the implementation of QE is included. It is predicted that the implementations should have a positive effect on the inflation expectations,
since the ECB wants the unconventional policy to work via the signaling channel. Since the ECB wants to give a clear signal that to the public that it seriously committed to increasing inflation and so inflation expectations. A positive coefficient is predicted
In the last regression, the variable for M1 is left out. The increase of money supply and the dummy QE are highly correlated, it is predicted that in regression (6) it is possible that QE or ln(M1) will not be significant. In order to test if both variables have an impact on inflation expectations, the variables ln(m1) and QE are also tested separately.
3.4 The Hypothesis
The hypothesis of the current research is based on economic theory, the Taylor Rule and intuition, as discussed in section 2. The first hypothesis is about testing the Taylor Rule, does QE have a positive effect on expectations via low interest rates and GDP gap. Besides that we want to test about how the inflation expectations are stirred by QE, money supply or crude oil prices.
Testing the Taylor Rule;
(1) 𝐻0: 𝛽0> 0, 𝛽1>0 and −𝛽2>0 𝐻1: 𝛽0 ≤ 0, 𝛽1≤0 and −𝛽2 ≤0
The theory is as follows; during times of Quantitative Easing, the ECB expands its asset purchasing program. The ECB buys assets and securities from commercial banks. The prices of these assets will riseand the central banks will lower its interest. This enables commercial banks to loosen their lending standards, due to higher reserves. The lower rates will stimulate investments. The GDP- gap becomes smaller when more output is generated when liquidity is enhanced and more investments are made. The wealth of households increases and more money is available for consumption, therefore the aggregate demand will increase. More demand for goods will cause the consumer prices to rise. Since we assume adaptive behavior of investors and households, a small increase in inflation will cause the expectations in the next period to be higher. The ECB announced its target inflation is “close but below 2% inflation” when inflation becomes higher it is expected that inflation expectations will rise in times of Quantitative Easing.
4. Results
To estimate the inflation expectations in period t a normal ordinary least squares regression was conducted. In the regression the function time-series is used for estimating the Newey West standard errors. The Newey West estimators deal with problems in the estimation like normality, heteroscedasticity and autocorrelation. There might occur autocorrelation due to the fact that inflation and inflation expectations are both in the regression, and their
correlation is high. It is most likely that the variance of the error term is not constant, therefore the estimation is corrected for heteroscedasticity. The regression is performed in Stata. The results of the estimations of the regressions are presented in table 2.
Table 2. Variable/regression (4) (5) (6) (7) Inflation (Newey-West st. error) 0. 8069*** (0. 036756) 0.4939 *** (0.0772) 0.4734*** (0.0809) 0. 6865*** (0. 0626) GDP-gap -0.09123 ** (0. 03898) -0.0915* (0.0457) -0.1255** (0.0604) -0.0276 (0. 0601) Ln(Money) - -3.8138*** (0.7253) -4.7383*** (1.2910) -
First Difference crude oil prices - -0.0093** (0.0044) -0.0103** (0.0045) -0.0074 (0.0049) QE - - 0.1683 (0.1943) -0.4209*** 0. 1214 Constant -0. 02794 (0. 1583) 59.5820*** (11.3626) 73.8047*** (19.9781) 0. 4875* (0.2625) F-test 349.02 233.43 186.03 186.18 R-squared 0.9112 0.9453 0.9461 0.9324
4.1 Discussion of the results.
As predicted, the regression predicted an positive value for inflation and an negative
coefficient for the output gap. In the first regression is found that a 1 percent point change in inflation will increase inflation expectations by 0. 8069 percent point. The output gap has an coefficient of -0.09123 this means, assumed that the gap is negative, that a 1% decrease in output gap will increase inflation expectations with 0.09123 %. Both explanatory variables are significant at a level at 1%, however the constant is not significant, nothing meaningful can be said about the constant.
In regression (5) and (6) the coefficient for inflation is lower, almost half its size compared to regression (4), but still significant at a level of 1%. A 1 percent point increase in the rate of inflation increases by 0.4939 percent point in (5) and 0.4734 percent point in (6). The coefficient is still positive as predicted but lower than in (4), this can be explained by the fact that in (5) and (6) more explanatory variables are added. In estimation (5), all these variables are significant, so inflation expectations are not only influenced by real inflation rate but also oil prices, money supply and GDP gap.
In the second regression (5), where more explanatory variables are included, gives as predicted a positive value for inflation and an negative sign for output gap, but output gap is nog significant at a 10% level. There is found that M1 has a significant effect on inflation expectations, a 1% increase is money supply will lower inflation expectations by 3.813%. In the estimation (6) is the coefficient also negative, and significant at a level of 1%. A 1% increase in M1 decreases inflation expectations by 4.7383%. this is not as predicted because a positive effect of increasing money on expectations is desired by the ECB. An explanation for this is the data, since the data only shows that the inflation expectations have been decreasing since November 2011. However, this negative effect is not preferred by the ECB, the variable money supply is highly significant in both regression (5) and (6), this is means that the effect of changing money supply has an effect on inflation expectations. This is a good signal for the central bank that money supply is an instrument to stir expectations, but not (yet) in the
positive direction.
The oil variable was not stationary so the first difference variable is used, this variable is significant at a level of 5% in both estimation (5) and (6). The negative coefficient means that when oil prices start to increase the effect on the inflation expectations will be negative, this is not as predicted. There was a positive relationship expected with the dependent variable. There is found that if the price of crude oil fall by 1$ in a month, the dependent
variable will decrease by 0.0093, and 0.0103 percent points, respectively to estimation (5) and (6). An explanation for this could be is that when oil prices decrease, people know that this will influence the HICP since the component energy is included in the HICP. So the effect of oil prices will only have a small (negative) effect on inflation expectations, because
apparently the market is not impressed (anymore) by decreasing oil prices.
Regression (6) differs from (5), in (6) the dummy variable QE is included to test if QE has an effect on expectations. The variable is not significant so there is no evidence found that the effect of QE on inflation expectations differs from zero. This can be explained by the fact that only 11 observations in times of QE are available, since the unconventional monetary policy is implemented in February 2015. It makes sense that there is no significant effect yet of the dummy variable. The R-squared has increased a little in regression (6) after QE was included, and the variable is not significant, so including QE into the regression is not essential in predicting inflation expectations.
Regression (7), ln(M1) is left out and only the dummy variable is included, now QE is significant at a level of 1% in times of QE the inflation expectations are 0.4209 percent points lower. This is effect is also not desired by the central bank. Again, the data shows that indeed in times of QE the inflation expectations since January 2015 has only been decreasing, so it makes sense that the coefficient is negative. However in regression (7) the variables GDP-gap and oil prices are not significantly different from zero anymore, while they were in regression (5) and (6). Taken into account that these variables did explain parts of the inflation
expectations in (5) and (6), more research should done on which variables explain inflation expectations better.
The constant in estimation (5) and (6) is, significant in both estimations at a level of 1%. Notable, that the coefficient of the constant in (5) and (6) is a lot higher than in (4) where the constant is not significant. In a hypothetical situation where all variables would be zero, the inflation expectation would be 59.5820 percent points in (5) and even 73.8047 percent point in (6), which makes no sense.
5. Conclusion.
Several policies were implemented by the ECB after the financial crisis, since November 2011 the inflation rate has only been declining. In January 2015 the European Central Bank announced it was going to implement an expanded asset purchasing program called
Quantitative Easing, in order to prevent the economy from further deflationary tendencies. This unconventional monetary policy is conducted to stimulate the economy and to achieve
the ECB’s main objection, namely maintaining price stability. The ECB decided to purchase, on a monthly bases, assets to an amount of 60 billion euro, these purchases would continue until at least September 2016. After a year of purchasing assets the ECB is still failing to meet its inflation target. QE has been implemented by other central banks some are still doubtful about the effects of QE. An important aspect of these policies is to manage the expectations. The Bank of Japan conducted QE since 2011, but are failing to create inflation, and stir the inflation expectations.
In this paper the simple Ordinary Least Squares (OLS) estimation was performed, with the dependent variable inflation expectations in period t. The dependent variable was
regressed on a inflation expectations and the GDP-gap according the Taylor-Rule. Thereafter, this regression was elaborated with more explanatory variables, namely, the logarithm of Monetary Base M1 and the first difference of oil prices. The third regression included all these variables plus a dummy variable for QE. The data for the rate of inflation expectation was retrieved from a survey done by the ECB’s Survey of Professional Forecasters (SPF). The most important finding was that the Taylor-Rule holds while explaining inflation expectations and the significance for the variable monetary base.
The hypothesis, that there is a significant positive effect of inflation on inflation expectations is found. Also a significant negative coefficient for GDP-gap is found so 𝐻0 is rejected. The Taylor-rule compiles with explaining inflation expectations, and predicts the expected values of the coefficient. The hypothesis that there is a positive effect of money supply on expectations cannot be accepted, however a significant effect is found. Even though increasing money supply has not the desired effect on inflation expectations the effect is significant and therefor can stir expectations.
There is a significant effect of a change in oil prices on the dependent variable in regressions (5) and (6), it can be concluded that due to decreasing oil prices people set
expectations higher. This is not as predicted but significant and maybe a positive sign that the market is not really affected by the dramatic fall of in oil prices, and people know that real inflation will be lower due to low oil prices but when corrected for oil prices the inflations for the other components of the HICP is not that low. The significance of the effect of oil prices on inflation expectations is in line with the conclusions of research done by Alejandro Badel (2015). He showed there is an remarkable relationship between crude oil prices and inflation. Another conclusion is that inflation expectations is not better estimated when QE is in the regression. And leaving M1 out of the regression makes oil prices and GDP-gap
an effect on expectations. This could be alarming since the ECB might expect that the impact of this unconventional policy should be large on expectation, even in it is only implemented for a year. However, using OLS requires a larger sample for the dummy variable QE, to find a significant result.
There are some weaknesses of the research, the number of observations after
Quantitative Easing was conducted are very limited. Most data is only available on a monthly basis, GDP and inflation expectations only on quarterly basis, and the dummy variable for QE had only 11 observations. It is advocated that at least 100 observations are needed to get reliable results. Another weakness is that the estimation could suffer from omitted variable bias. It is possible that there are variables omitted from the regression which might explain a part of the dependent variable. According to literature, surveys are the best indicators for the rate of inflation expectation, the survey used in this paper is conducted by the ECB’s SPF, the question is how independent the SPF is from the incentives of the ECB. So sample selection for inflation expectations could be biased, but this is only speculation and not assumed.
A request for follow-up research is needed, the amount of research done of QE on inflation exceptions is small. Literature suggests that the main component of the effectiveness of QE is managing expectations. This paper has found evidence that increasing money supply stirs inflation expectations in a negative way, research can be done how/when this
expectations will become positive. Besides more research should be done on which variables influence the market’s inflation expectation most. Especially for the policies conducted in Japan, and nowadays for the Eurozone when there is more data available. The external validity of this model should be tested, it would be interesting to investigate if this model still holds for other datasets.
6. Bibliography.
Alejandro Badel. (2015, October 8). Predicting the Impact of Oil Prices on Inflation. Opgeroepen op January 9, 2016, van The Federal Reserve Bank of St. Louis: https://www.stlouisfed.org/on-the-economy/2015/october/predicting-impact-oil-prices-inflation
Clemente De Lucia. (2014). How risky is the fall of oil prices for inflation? BNP Paribas.
Eggertsson, G. (2008). Liquidity Trap. The New Palgrave Dictionary of Economics. Second Edition. Eds. Steven N. Durlauf and Lawrence E. Blume. Palgrave Macmillan. The New Palgrave Dictionary of Economics Online. Palgrave Macmillan.
European Central Bank. (2015). European Central Bank. Opgehaald van https://www.ecb.europa.eu/mopo/intro/benefits/html/index.en.html
European Central Bank. (2015, Januari 22). European Central Bank. Opgehaald van https://www.ecb.europa.eu/press/pr/date/2015/html/pr150122_1.en.html
Fawley, B. W. (2013). Four stories of quantitative easing. Federal Reserve Bank of St. Louis Review, 51-88.
Hofmann, B. a. (2013). Central bank asset purchases and inflation. BIS Quarterly Review March.
Jürgen Stark. (2009). Monetary Policy before, during and after the financial crisis. University Tübingen.
Kimura, T. a. (2003). The effect of the increase in the monetary base on Japan’s economy at zero interest rates: an empirical analysis. Monetary Policy in a Changing
Environment, Bank for International Settlements Conference Series, 276-312. Kregel, J. A. (2000). Krugman on the liquidity trap: why inflation won't bring recovery in
Japan. Jerome Levy Economics Institute Working Paper no. 298, 8.
Krugman, P. R. (1998). It's baaack: Japan's slump and the return of the liquidity trap. Brookings Papers on Economic Activity, 137-205.
Miccoli, Marcello and Neri, Stefano et al. (2015). Inflation surprises and inflation expectations in the euro area. Italy: Bank of Italy, Economic Research and International Relations Area.
Neely, C. J. (2008). Is inflation an international phenomenon?
Spiegel, M. M. (2006). Did quantitative easing by the Bank of Japan" work"? FRBSF Economic Letter.
Wes Goodman. (2015, January 18). Bloomberg. Opgeroepen op December 2015, van Bloomberg Business: http://www.bloomberg.com/news/articles/2015-01-19/boj-showing-liquidity-trap-holds-warning-for-ecb-japan-credit
7. Appendix.
Dickey Fuller test for crude oil prices.
Augmented Dickey-Fuller test for unit root Number of obs = 58
--- Interpolated Dickey-Fuller --- MacKinnon approximate p-value for Z(t) = 0.9624
D. oil Coef. Std. Err t P>t [95% Conf.
Interval] Oil L1 0.001936 0.0399065 0.05 0.961 [-0.0780385; 0.0819105] LD 0.144854 0.1402823 1.03 0.306 [-0.1362779; 0.4259859] Cons -1.175004 4.043573 -0.29 0.772 [-9.278505; 6.928498]
The P-value >0.05 so the variable is not stationary.
Graph 1. Expected inflation 0 0,5 1 1,5 2 2,5 3 15 /1/20 11 15 /4/20 11 15 /7/20 11 15 /10/2 011 15 /1/20 12 15 /4/20 12 15 /7/20 12 15 /10/2 012 15 /1/20 13 15 /4/20 13 15 /7/20 13 15 /10/2 013 15 /1 /2 0 14 15 /4/20 14 15 /7/20 14 15 /1 0/ 2 01 4 15 /1/20 15 15 /4/20 15 15 /7/20 15 15 /10/2 015
Graph 2. The 3-month EURIBOR
Graph3 output gap
0 0,5 1 1,5 2 2,5 3 15 /1/20 11 15 /4/20 11 15 /7/20 11 15 /10/2 011 15 /1/20 12 15 /4/20 12 15 /7/20 12 15 /10/2 012 15 /1/20 13 15 /4/20 13 15 /7/20 13 15 /10/2 013 15 /1/20 14 15 /4/20 14 15 /7/20 14 15 /10/2 014 15 /1/20 15 15 /4/20 15 15 /7/20 15 15 /10/2 015
-1,0000 -0,5000 0,0000 0,5000 1,0000 1,5000 2,0000 2,5000 3,0000 3,5000 15 /1/20 11 15 /4/20 11 15 /7/20 11 15 /10/2 011 15 /1/20 12 15 /4/20 12 15 /7/20 12 15 /10/2 012 15 /1/20 13 15 /4/20 13 15 /7/20 13 15 /10/2 013 15 /1 /2 0 14 15 /4/20 14 15 /7/20 14 15 /10/2 014 15 /1/20 15 15 /4/20 15 15 /7/20 15 15 /1 0/ 2 01 5 Graph 4 Oil prices Graph 5. Inflation 0 20 40 60 80 100 120 140 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61
