The impact of the global financial crisis on the
synchronization of financial cycles in the EMU
MSc Economics Thesis
Name:
Mariëlle Dreuning
Student number:
10559035
Email:
marielledreuning@gmail.com
2
Statement of Originality
This document is written by Mariëlle Dreuning who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
This thesis researches the impact of the global financial crisis on the synchronization of financial cycles in the EMU. Financial cycles are constructed for eleven EMU countries and for every country the synchronization between its financial cycle and the financial cycle of the euro area as a whole, the reference cycle, is calculated. The results indicate that the crisis had a positive impact. There is a structural break in the time trend of synchronization at the fourth quarter of 2008. Financial cycles have been diverging since the start of the EMU. From the fourth quarter of 2008, they are diverging less.
4
List of contents
1. Introduction ... 5
2. Literature review ... 6
2.1 Financial cycle synchronization in the EMU ... 7
2.2 Explanations for divergence ... 7
2.2.1 Credit developments ... 7
2.2.2 House price developments ... 9
2.3 The impact of the global financial crisis ...10
2.3.1 Financial disintegration ...10
2.3.2 House price heterogeneity ...11
3. Methodology ...11
3.1 Constructing financial cycles ...11
3.2 Synchronization measure ...12 3.3 Regression ...13 3.4 Data ...14 4. Results ...15 4.1 Descriptive analysis ...15 4.1.1 Financial cycles ...15 4.1.2 Synchronization ...20 4.2 Regression ...21 5. Conclusion ...23 References ...24
Appendix A: Financial cycles ...27
5
1. Introduction
Before the global financial crisis of 2008-2009, finance was mainly viewed as a sideshow to macroeconomic fluctuations. As the crisis proved that this was a dangerous and incorrect presumption to make, an increasing amount of work has been devoted to capturing the interactions between the financial and the real side of the economy since then. In particular, renewed attention was given to the financial cycle, a concept that is related to the procyclicality of the financial system. Borio (2014) defines it as self-reinforcing interactions between perceptions of value and risk, attitudes towards risk and financing constraints, which translate into booms followed by busts. The financial cycle can be most parsimoniously described in terms of credit and property prices (Borio, 2014). Financial cycle estimates are measures of the time-varying dimension of systemic risk in the financial system and thus capture patterns that can have important macroeconomic consequences. Peaks in the financial cycle are closely associated with financial crises and recessions that coincide with its contraction phase are especially severe (Borio, 2012). The financial cycle provides a fairly accurate measure of the build-up of risk of such a crisis in real time, which is why it plays a crucial role in macroprudential policy. The idea there is to build up buffers during the expansion phase of the financial cycle which can be drawn upon when the booms turn into busts. The countercyclical buffer in Basel III is an example of a macroprudential tool that depends on the properties of the financial cycle.
Policymakers are increasingly reorienting their policies towards a macroprudential perspective. For the euro area, this reorientation requires knowledge on the synchronization of financial cycles within the area to be able to design the appropriate framework as it relevant for the question whether macroprudential policy should be euro area wide or should remain at the national level. Financial spillovers from one country to another can be very strong in a monetary union and national authorities might not have enough incentive to internalize these (Merler, 2015). This makes financial stability in the euro area a supra-national issue and provides a rationale for entrusting the European Central Bank (ECB), being a central authority, with stronger macroprudential powers. Whether this can be successful largely depends on the heterogeneity within the EMU. When there is more heterogeneity it becomes harder to design an appropriate policy for the whole area. So from this point of view, a higher degree of synchronization of financial cycles in the EMU would be desirable. However, enhanced synchronization may reduce the benefits of international diversification for financial intermediaries (De Nicolò and Tieman, 2006). Those benefits exist when financial markets are affected by country specific factors, which makes it likely that the correlations between asset returns from different countries are lower than correlations within countries (De Santis and Gerard, 1997). In that case, internationalizing a portfolio diversifies
6 away risk. When countries become more synchronized, less risk can be diversified away and financial intermediaries become more sensitive to common shocks. Systemic risk is higher and a large negative shock may lead to international contagion.
Little is known yet about the synchronization of financial cycles. The literature that is available shows that financial cycles in the EMU have diverged since its start, meaning that the level of synchronization has decreased. This can be explained by different developments across the EMU in both credit and house prices, two important variables for the financial cycle. The different developments in credit are rooted in the financial integration that came with the EMU. House prices developments differed due to heterogeneity in mortgage markets that led to a heterogeneous response of house prices to monetary policy. In general, research has pointed out that the global financial crisis has caused financial disintegration, which makes it possible that the crisis has also induced a change in the synchronization of financial cycles. At the same time, heterogeneity in house price developments remains although it was lower during the crisis and is decreasing in general. This leads to the main research question of this thesis: What was the impact of the global financial crisis on the synchronization of financial cycles in the EMU?
This thesis will construct financial cycles based on the credit-to-GDP ratio, credit growth and house prices. The band pass filter developed by Christiano and Fitzgerald (2003) will be applied to capture the cyclical components of the time series of these variables, after which a synthetic financial cycle is constructed by averaging the three obtained individual cycles. This is done for eleven countries in the EMU. For every country the synchronization between its financial cycle and the financial cycle of the euro area as a whole, the reference cycle, is calculated. The novelty of this research lies in the use of an approach that is familiar in business cycle literature but that is new to financial cycle literature, namely the synchronization measure proposed by Bordon et al. (2013). Contrary to the methods chosen in existing financial cycle literature, this measure takes both the sign of the relationship between the two financial cycles and the differences in amplitude into account. Regression analysis is used to assess the impact of the global financial crisis on the synchronization measure.
The rest of this thesis is structured as follows. Section 2 provides an overview of the relevant literature. Section 3 describes the methodology and the data. Section 4 gives the results of applying this methodology and section 5 concludes.
2. Literature review
First, the literature on the synchronization of financial cycles in the EMU will be reviewed. This literature is very limited, but the main conclusion is that financial cycles have diverged
7 since the start of the EMU. Therefore, the second part will give an overview of trends that provide an explanation for this divergence. Finally, it is analyzed whether there is a change in these trends since the crisis.
2.1 Financial cycle synchronization in the EMU
According to Stremmel (2015), the best-fitted financial cycle measure for Europe includes the credit-to-GDP ratio, credit growth and house prices. This is in line with the general finding that the financial cycle can be most parsimoniously described in terms of credit and property prices (Borio, 2014).
Merler (2015) divides the euro area in North, Centre and South and estimates financial cycles for those clusters based on credit growth and house prices. She shows that the correlation of those financial cycles with a euro area aggregate cycle has decreased since 1999, indicating that there has been divergence in financial cycles since the start of the EMU. The heterogeneity is confirmed by Schüler et al. (2015), who calculate the degree of concordance of financial cycles for 13 EU countries and find that there is co-movement only two-third of the time.
Based on a dispersion measure that indicates whether financial cycles diverge or converge over time, Stremmel (2015) concludes that financial cycles in Europe become more synchronized during recessions.
2.2 Explanations for divergence
The explanations for the divergence of financial cycles will be divided into two parts. The first part focusses on how financial integration led to very different developments in credit at the national level. The second parts evaluates how house price developments differed across the EMU due to heterogeneity in mortgage markets that led to a heterogeneous response of house prices to monetary policy.
2.2.1 Credit developments
The divergence of financial cycles in the euro area is deeply rooted in the financial integration that followed the implementation of the single currency (Merler, 2015). Financial integration has been happening at the global level long before the start of the EMU, but the introduction of the euro has made the process even more rapid and sizeable in the euro area. The most common measures for financial integration are based on either prices or quantities.
Price-based integration looks at discrepancies in prices or returns on assets caused by the geographic origin of those assets (Baele et al., 2004). The law of one price should hold when financial integration is complete, so financial integration represents to what extent
8 frictions in the market are absent. As figure 1 shows, interest rates in the sovereign bond, money and interbank market converged rapidly in the run up to the implementation of the euro (Merler, 2015). De Sola Perea and Van Nieuwenhuyze (2014) point out that differences in interest rates reflect different monetary policy rates, exchange rate risk and differences in macroeconomic fundamentals of countries. The implementation of the euro removed exchange rate risk and unified monetary policy, so the convergence implied that financial markets valued the fundamentals of member countries as if they were identical. The wide variations in solvency among sovereigns were neglected, which has the implicit assumption that EMU members will be bailed out if necessary. Figure 1 makes clear that the convergence also meant an overall decline in interest rates.
Quantity-based integration assesses indicators of cross-border activity like volumes of cross-border loans, which signal how easy it is for foreigners to access a regional credit market (Baele et al., 2004). The introduction of the euro was followed by a massive increase in bank lending activity within the euro area, driven by a rise in cross-border activity. The loans from banks in the euro area to their domestic borrowers doubled over ten years, while loans from those banks to residents in other euro countries almost tripled (Merler, 2015).
The convergence of the interest rates and the increase in cross-border activity imply greater financial integration (see also Lane, 2008). The integration led to very different credit developments at the national level. Taylor (2008) shows that interest rates were too low compared to the rates implied by a simple Taylor rule, especially for southern countries. This
-5% 0% 5% 10% 15% 20% 25% 30% Q 1 -1 9 8 0 Q 2-1 98 1 Q 3 -1 9 8 2 Q 4 -1 9 8 3 Q 1 -1 9 8 5 Q 2 -1 9 8 6 Q 3 -1 9 8 7 Q 4 -1 9 8 8 Q 1 -1 9 9 0 Q 2-1 99 1 Q 3 -1 9 9 2 Q 4 -1 9 9 3 Q 1 -1 9 9 5 Q 2 -1 9 9 6 Q 3 -1 9 9 7 Q 4 -1 9 9 8 Q 1 -2 0 0 0 Q 2-2 00 1 Q 3 -2 0 0 2 Q 4 -2 0 0 3 Q 1 -2 0 0 5 Q 2 -2 0 0 6 Q 3 -2 0 0 7 Q 4 -2 0 0 8 Q 1 -2 0 1 0 Q 2-2 01 1 Q 3 -2 0 1 2 Q 4 -2 0 1 3 Q 1 -2 0 1 5 Q 2 -2 0 1 6 Q 3 -2 0 1 7
Figure 1: Interest rates
Austria Belgium Finland France Germany Greece Ireland Italy Netherlands Portugal Spain
9 resulted in massive capital flows from the north to the south of the EMU as credit demand rose sharply (Gros, 2012). Banks were also actually able to supply more credit for two reasons. First of all, the greater financial integration of the EMU coincided with the strong increase in the use of securitization. Securitization provides banks with a new source of financing, which makes them increase their credit supply (Loutskina, 2011). Second, being part of the EMU meant that banks now had access to a euro area wide deposit base instead of having to rely on their core deposit base. This is reflected in the strong correlation between domestic credit growth and intra-euro area debt liabilities found by Lane and McQuade (2013). Lane and McQuade also highlight that the standard deviation of domestic credit growth rose sharply during 2003-2008 when compared to previous periods, indicating that the credit boom was certainly not uniform across countries. Therefore, the financial integration that led to different credit developments offers an explanation for the divergence in financial cycles within the EMU.
2.2.2 House price developments
Another explanation may come from looking at house price developments. The decline in interest rates accompanying the convergence process translated into a decline in mortgages rates, boosting the demand for housing and thus house prices (Hilbers et al., 2008). While at the start of the euro the most eastern member countries still had significantly higher mortgage rates than countries further west, in 2008 there is relatively little variability in the cost of mortgages left (Miles and Pillonca, 2008). However, the mortgage market itself is not very integrated. There are some spillovers from country-specific house price shocks in the euro area, but they are of relatively low magnitude (Vansteenkiste and Hiebert, 2011).
Hilbers et al. (2008) find that house prices in Europe have shown diverging trends. In almost all European countries, house prices have increased sharply, but there are large cross-country differences. Spain, Ireland and the Netherlands are examples of countries where house prices more than doubled, while house prices in Germany actually slightly declined. An explanation for these differences could lie in the finding of Carstensen et al. (2009) that there is a heterogeneous reaction of house prices to monetary policy among European countries. A source for this heterogeneity are the institutional characteristics of mortgage markets across the euro area, which show significant differences in for example the typical loan-to-value ratio, the average typical term of a mortgage or early repayment fees (IMF, 2008). The latter constrains households’ ability to refinance and therefore weakens their response to an interest rate change. The heterogeneity in mortgage markets can be linked to differences in their development. Even though the deregulation in mortgage markets, which began in the early 1980s in many advanced economies, pushed all countries toward more competitive housing financing models, the process took place at a different
10 pace and to a different extent across countries, resulting in significant cross-country differences with regard to mortgage market development (IMF, 2008). The research of Iacoviello and Minetti (2003) and later on the research of Calza et al. (2013) both show that the response of house prices to monetary policy shock is significantly stronger in those countries where the mortgage market is more developed. The heterogeneity in mortgage markets and therefore the heterogeneous response of house prices to monetary policy, leading to different developments in house prices, can help explain the divergence in financial cycles in the EMU.
2.3 The impact of the global financial crisis 2.3.1 Financial disintegration
There has been financial disintegration in Europe since the crisis, both from a price-based perspective, as interest rates were diverging, and from a quantity-based perspective, as there was a sharp reduction in cross-border exposures (Darvas et al., 2014; De Sola Perea and Van Nieuwenhuyze, 2014; Horvath, 2017). The disintegration peaked during 2011-2012 and declined after, but it remains higher now than before the crisis (Horvath, 2017).
The divergence in interest rates is a result from the fact that financing costs of banks started to differ across the EMU from 2009. The cross-country dispersion was mainly related to the negative bank-sovereign feedback loop, the quality of banks’ balance sheets and redenomination risk and was passed through to lending rates (Ruscher and Vašiček, 2016). The negative bank-sovereign feedback loop arises because banks tend to hold large amounts of sovereign debt, often with a strong home bias, and because sovereigns implicitly guarantee banks’ liabilities. If one party runs into trouble, the balance sheets of the other also deteriorate, creating a negative feedback loop. Besides financing costs, borrower risk also diverged among countries, which affects the mark-up that banks charge on loans. As mentioned before, the convergence in interest rates from the onset of the euro neglected wide variations in solvency among sovereigns. According to De Sola Perea and Van Nieuwenhuyze (2014), this implies that the pricing of risk was not efficient. The crisis caused a reappraisal of risk among investors, suggesting a structural change.
With regard to quantity-based integration, there has been a continued re-domestication of banks’ activity as cross-border bank loans decreased. The home bias in banks’ assets became stronger in many countries after the crisis, which is a sign of financial disintegration (Baele et al., 2004). For Spain and Portugal for example, the international diversification achieved between 1999 and the crisis was almost completely reversed by the end of 2013 (Darvas et al, 2014).
11
2.3.2 House price heterogeneity
The global financial crisis caused a drop in house prices across the EMU, which can be seen as a correction of earlier imbalances. The decline in house prices from 2008 was much stronger in countries that had also experienced a larger increase in house prices before the crisis than in countries where this was much more moderate (ECB, 2015). Therefore, cross-country heterogeneity in euro area house price developments remains (ECB, 2008; ECB, 2009; ECB, 2010; ECB, 2011; ECB, 2012; ECB, 2013; ECB, 2015; ECB, 2016). However, the dispersion was smaller in 2007-2008 (ECB, 2010). This is in line with the report of the IMF (2018), which shows that the synchronization of house prices advanced economies sharply increased before and during the crisis. The increase contributes to a more general trend of rising synchronization across countries. In conclusion, the heterogeneity in house price developments remains but is was lower during the crisis and is decreasing in general.
3. Methodology
This sections explains the steps that need to be taken to answer the research question and motivates the methodological choices in those steps. The first part explains how the financial cycles are constructed. After that, the measure for synchronization is defined. The third part describes the regression analysis to investigate the impact of the crisis. Finally, information on the selected dataset is provided.
3.1 Constructing financial cycles
Constructing financial cycles consists of two steps. First, individual medium-term cycles have to be isolated from the data on variables that are relevant for the financial cycle. Second, the information must be combined into a summary indicator to construct a synthetic financial cycle. The relatively new and limited literature on financial cycles has not reached consensus yet on the optimal way to do this. The majority of the papers uses either turning point analysis or frequency-based filters to perform the first step. The former identifies cycles by dating the peaks and troughs of the variables. The second approach filters a time series for its cyclical components. The two methods yield similar characteristics for the financial cycle (Galati et al., 2016), thus the choice between them does not seem critical for the outcome of this research. Frequency-based filters are favorable from an analytical viewpoint because they allow for the construction of a synchronization measure at every point in time, whereas turning point analysis only dates peaks and troughs. Within this class of filters, the Hodrick-Prescott filter and the band pass filters developed by Baxter and King (1999) and Christiano and Fitzgerald (2003) are the most commonly used. Band pass filters provide cycles that are smoother than those from the Hodrick-Prescott filter and also allow for a more
12 straightforward comparison of series (Stremmel, 2015). Christiano and Fitzgerald show that their filter is favorable to the Baxter and King filter, especially when it is used to extract frequencies lower than the business cycle. A stylized fact from the financial cycle is that it has a substantially lower frequency than the business cycle (Borio, 2014). Therefore, the band pass filter developed by Christiano and Fitzgerald will be applied to isolate medium-term cycles from the data on the credit-to-GDP ratio, credit growth and house prices. Beforehand, all variables need to be standardized to ensure comparability of units. Also, an augmented Dickey-Fuller test will be performed to check whether a time series is stationary or not, because this affects the required parameters for the filter.
For the second step, the methodology of Borio (2012) will be followed. A synthetic financial cycle measure is created by averaging the individual cycles of the credit-to-GDP ratio, credit growth and house prices for each point in time. This is possible because of the favorable characteristics of the frequency-based filter method and because the individual cycles have comparable units of measurement.
3.2 Synchronization measure
Different measures of synchronization have been chosen in the existing literature on financial cycle synchronization. Merler (2015) looks at the correlation of the financial cycles of the different clusters with the financial cycle of the euro area aggregate. Stremmel (2015) takes the one year cross-country standard deviation of the filtered time series as an indicator of cycle dispersion. These measures are taken from the literature on business cycle synchronization and while they are common approaches, I argue that they are not optimal. To measure the synchronization between two financial cycles sufficiently, both the sign of the relationship between the two synthetic financial cycle indicators and the differences in amplitude should be taken into account. The correlation coefficient does not take proper account of the differences in amplitude and the standard deviation neglects the direction of the dispersion.
Therefore, this thesis will use another approach used in business cycle literature that is new to financial cycle literature. Bordon et al. (2013) propose a measure that takes both the sign and difference in amplitude into account. This measure can be represented by the following equation:
𝜑𝑖𝑡 = 1 −
(𝜃𝑖𝑡− 𝜃𝑟𝑡)2
𝜃𝑖𝑡2+ 𝜃 𝑟𝑡2
Where 𝜑𝑖𝑡 is the measure of synchronization between the financial cycle of country i and the
13 country i at time t and 𝜃𝑟𝑡 is the synthetic financial cycle indicator of the euro area. As
explained in section 3.1, the synthetic financial cycle indicator is the average of the cyclical components of the credit-to-GDP ratio, credit growth and house prices. The euro area financial cycle functions as the reference cycle. A high, positive value of 𝜑𝑖𝑡 implies a high
level of synchronization between the financial cycle of country i and the euro area reference cycle at time t. A low, negative value of 𝜑𝑖𝑡 implies a low level of synchronization.
3.3 Regression
The impact of the global financial crisis on the synchronization of financial cycles is analyzed by investigating whether the time trend in synchronization has changed when the crisis emerged. The period under investigation starts at the beginning of the EMU, the first quarter of 1999, and ends at the third quarter of 2017. The following equation is estimated:
𝜑𝑖𝑡 = 𝛼 + 𝛽1𝑡 + 𝛽2𝐷𝑡 + εit
Where 𝜑𝑖𝑡 is the synchronization measure specified in section 3.2, t is a linear time trend, D
is a dummy variable that is 0 when t < T and 1 when t ≥ T and εit is the error term. T
represents the quarter in which there is a structural break in the time trend. A positive value of beta indicates convergence of financial cycles, whereas a negative value suggests divergence.
To test if there is a structural break, the likelihood ratio test will be used. The likelihood ratio test compares the goodness of fit of two models, a null model and an alternative model. The null model is nested in the alternative model and the likelihood ratio expresses how many times more likely the data are under the alternative model than under the null model. The likelihood ratio can then be used to calculate a p-value to decide whether or not to reject the null model. The null model here is:
𝜑𝑖𝑡 = 𝛼 + 𝛽1𝑡 + εit
The alternative model is:
𝜑𝑖𝑡 = 𝛼 + 𝛽1𝑡 + 𝛽2𝐷𝑡 + εit
Where the components are as specified above. When the test rejects the null model, it is likely that there is a structural break in the time trend at T. To improve the accuracy, the likelihood ratio test will be performed for a wide range of different values for T. The T for which the outcome is most significant will be used to see if the null model should be rejected.
14 In case a structural break around the crisis is found, another test will be performed to see whether there is a second break later on. For example, it is possible that the time trend only differs for a few years after the crisis, implying that there is a second break in the time trend after which the time trend in synchronization is similar to the trend before the first break. The method to test for a structural break as described above will be repeated. Now the null model is:
𝜑𝑖𝑡 = 𝛼 + 𝛽1𝑡 + 𝛽2𝐷1𝑡 + εit
The alternative model is:
𝜑𝑖𝑡 = 𝛼 + 𝛽1𝑡 + 𝛽2𝐷1𝑡 + 𝛽3𝐷2𝑡 + εit.
Where t is a linear time trend, 𝐷1 is a dummy variable that is 0 when t < 𝑇1 and 1 when t ≥ 𝑇1,
𝐷2 is a dummy variable that is 0 when t < 𝑇2 and 1 when t ≥ 𝑇2 and εit is the error term. 𝑇1
represents the quarter in which the first structural brake was found and is equal to T as determined above. 𝑇2 represents the quarter in which there is a second structural break in
the time trend and it is assumed that 𝑇2> 𝑇1. Again, the likelihood ratio test will be used to
see if there is a second structural break.
3.4 Data
The selected variables for the financial cycles are the credit-to-GDP ratio, credit growth and house prices because these variables yield the best-fitted financial cycle measure for Europe (Stremmel, 2015). For the credit-to-GDP ratio and credit growth, quarterly data on total credit to the private non-financial sector at market value from the Bank for International Settlements (BIS) database is used. Credit growth will be measured as the four-quarter difference in log-levels (Stremmel, 2015). Data on nominal GDP is obtained from the International Monetary Fund (IMF). The IMF series on GDP are yearly data, thus a transformation into quarterly series is necessary to meet the units of measurement of the other variables. House prices will be represented by nominal quarterly indexes from either the ECB Statistical Data Warehouse or the Organisation for Economic Cooperation and Development (OECD), depending on the availability per country. The indexes on house prices are standardized to 1 to be in the same unit of measurement as the other variables.
To be able to capture the cyclical component of the variables properly, long series of the data are required. Therefore, countries for which data is only available for recent years cannot be included in the research. Financial cycles will be constructed for eleven countries in the EMU; Austria, Belgium, Germany, Finland, France, Greece, Ireland, Italy, the
15 Netherlands, Portugal and Spain. Data on credit and GDP is available from 1980Q1 until 2017Q3 for all countries. The availability of house prices varies per country; the starting date ranges between 1980 and 2000. When applying the band pass filter, all available data will be used. A financial cycle for the euro area as a whole will also be constructed. To obtain the credit-to-GDP ratio and credit growth for the whole area, the sums of the eleven individual countries’ values for credit and GDP will be used. For house prices the GDP weighted average of the individual countries’ indexes will be taken. The weight on each index is calculated as the GDP of the country divided by the GDP of the whole euro area.
4. Results
4.1 Descriptive analysis 4.1.1 Financial cycles
This section will briefly comment on the construction of the financial cycles and will discuss a few characteristics of those cycles. The band pass filter by Christiano and Fitzgerald (2003) is used to isolate medium-term cycles from data on the credit-to-GDP ratio, credit growth and house prices for each selected euro area country and for the euro area as a whole. The filter requires different parameters when the series is stationary. Therefore, augmented Dickey-Fuller tests are performed for every variable and every country in this research to test whether the time series are stationary. Under the null hypothesis, the series has a stochastic trend. Under the alternative hypothesis, the series is stationary. Table 1 show the results of the augmented Dickey-Fuller tests. A 5% significance is used to decide which series are stationary.
Table 1: Augmented Dickey-Fuller tests
Credit-to-GDP ratio Credit growth House prices
Austria -0.741 -2.709* -1.212 Belgium -2.860 -3.906*** -2.293 Finland -2.374 -2.469 -2.807 France -1.463 -3.027** -1.938 Germany -1.545 -2.022 -1.165 Greece -1.626 -1.369 -1.678 Ireland -1.947 -3.528*** -3.117 Italy -1.818 -1.411 -2.409 The Netherlands 0.165 -1.526 -2.655 Portugal -2.058 -2.405 -2.289 Spain -2.490 -1.816 -2.267
16
Euro area -2.098 -2.005 -3.713**
*10% significance, **5% significance, ***1% significance
After applying the Christiano and Fitzgerald band pass filter, the obtained individual cycles are combined into a synthetic financial cycle by taking the average of the three series at every point in time. Graphs of the financial cycles for the eleven euro area countries and the euro area as a whole can be found in appendix A.
A first glance at the characteristics of the financial cycles can be taken by looking at the descriptive statistics as shown in table 2. Since financial cycles are synthetic, the numbers do not provide much insight when analyzed in isolation but comparing the numbers makes it possible to detect unexpected results.
Table 2: Descriptive statistics
Mean Std. dev. Min. Max.
Austria 0.0022 0.0193 -0.0346 0.0469 Belgium 0.002 0.0527 -0.1398 0.1948 Finland 0.0003 0.0564 -0.1085 0.2166 France 0.0006 0.0215 -0.0573 0.0543 Germany 0.0007 0.0253 -0.0736 0.0755 Greece -0.0014 0.0404 -0.1035 0.1345 Ireland 0.005 0.1376 -0.3535 0.5151 Italy 0.0002 0.0242 -0.0709 0.0802 The Netherlands 0.001 0.0327 -0.1026 0.0892 Portugal 0.0004 0.0456 -0.1329 0.1196 Spain -0.0001 0.0383 -0.0906 0.0836 Euro area 0.0006 0.0171 -0.0361 0.0568
Ireland stands out when looking at the standard deviations, as the number is high compared to the rest of the column. A graphical analysis suggests that the standard deviation has increased over time for Ireland, and more specifically that the standard deviation is higher since the crisis. The minimum and maximum values for Ireland are also relatively high in absolute terms compared to the rest of the numbers and these values are observed after the crisis.
According to Claessens et al. (2011) the main features that characterize financial cycles are duration, amplitude and slope. The rest of this section analyzes these features. A distinction is made between an expansion phase, a downturn and a full cycle. An expansion
17 phase is the period between a trough and a peak, a downturn is the period between a peak and a trough and a full cycle consists of an expansion phase and a downturn. To analyze the duration of the cycle, the average duration of the expansion phase and the average duration of the downturn are calculated. The average duration of the expansion phase is defined as the number of quarters from a trough to the next peak divided by the number of peaks. Similarly, the average duration of the downturn is defined as the number of quarters from a peak to the next trough divided by the number of peaks. Following Harding and Pagan (2001), the formula for the average duration of the expansion phase is:
𝐷 = ∑ 𝑆𝑡
𝑇 𝑡=1
∑𝑇−1(1 − 𝑆𝑡+1)𝑆𝑡
𝑡=1
Where 𝑆𝑡 is binary variable that equals 1 in expansions and 0 in contractions. A similar
formula for the average duration of the downturn is used. Adding up the average duration for the expansion phase and the average duration for the downturn yields the average duration of a full cycle. Table 3 provides an overview of the average duration for every country and the euro area in quarters.
Table 3: Average duration financial cycles
Average duration full cycle Average duration expansion Average duration downturn Austria 8.15 3.47 4.68 Belgium 13.57 7.45 6.12 Finland 11.85 5.22 6.63 France 12.95 7.41 5.54 Germany 15 8.61 6.39 Greece 11.37 4.37 7 Ireland 13.91 5.9 8.01 Italy 12.31 6.21 6.1 The Netherlands 11 5.38 5.62 Portugal 14.8 7.69 7.11 Spain 13.81 6.6 7.2 Euro area 12.87 6.17 6.7
The average duration of a full financial cycle ranges between 8.15 quarters and 15 quarters. This is not in line with the finding of Schuler et al. (2015) that financial cycles on average last
18 7.2 years in European countries. Austria seems to be a bit of an outlier with the average duration of 8.15 quarters for the full cycle, which is quite low compared to the other countries. For the majority of the countries the average duration of the downturn is larger than the average duration of the expansion phase, which is the opposite of the result from Schuler et al. (2015). Especially for Greece and Ireland the downturns last much longer on average than the expansions.
The next characteristic up for review is the amplitude. Again, a division is made between the average amplitude of the expansion phase and the average amplitude of the downturn and adding those two gives the average amplitude of the full cycle. The average amplitude of the expansion phase is defined as the total change during expansions divided by the number of peaks. Likewise, the average amplitude of the downturn is defined as the total change during downturns divided by the number of troughs. Following Harding and Pagan (2001), the formula for the average amplitude of the expansion phase is:
𝐴 = ∑ 𝑆𝑡
𝑇
𝑡=1 ∆𝑦𝑡
∑𝑇−1𝑡=1(1 − 𝑆𝑡+1)𝑆𝑡
Where 𝑆𝑡 is binary variable that equals 1 in expansions and 0 in contractions and ∆𝑦𝑡 = 𝑦𝑡−
𝑦𝑡−1 with 𝑦𝑡 being the value for the financial cycle at time t. Table 4 displays the average
amplitudes for the eleven selected countries and the euro area as a whole.
Table 4: Average amplitude financial cycles
Average amplitude full cycle Average amplitude expansion Average amplitude downturn Austria 0.0696 0.0338 0.0358 Belgium 0.2113 0.1159 0.0954 Finland 0.1581 0.08 0.0781 France 0.0922 0.0488 0.0434 Germany 0.1084 0.0558 0.0526 Greece 0.1469 0.0732 0.0737 Ireland 0.4883 0.2115 0.2768 Italy 0.0928 0.0548 0.038 The Netherlands 0.1351 0.0674 0.0677 Portugal 0.224 0.1186 0.1054 Spain 0.154 0.0756 0.0784
19
Euro area 0.0699 0.0369 0.033
The average amplitude of the full cycle lies between 0.0696 and 0.4883. The latter is the average amplitude of Ireland, which seems to be an outlier. The relatively high value here is consistent with the earlier finding of the relatively high standard deviation for Ireland. The averages for the expansions and the averages for the downturns lie rather close to each other, but for the majority of the countries the average amplitude of the expansion phase is a bit higher than the average amplitude of the downturn.
The third and final characteristic to look at is the slope, which measures the violence of a given cyclical phase (Claessens et al., 2011). The average slope is the ratio of the average amplitude to the average duration. This is calculated for the full cycle, for the expansion phase and for the downturn by dividing the previously calculated average amplitude by the average duration. Table 5 shows the results of these calculations.
Table 5: Average slope financial cycles
Average slope full cycle Average slope expansion Average slope downturn Austria 0.0085 0.0097 0.0076 Belgium 0.0156 0.0156 0.0156 Finland 0.0133 0.0153 0.0118 France 0.0071 0.0066 0.0078 Germany 0.0072 0.0065 0.0082 Greece 0.0129 0.0168 0.0105 Ireland 0.0351 0.0358 0.0346 Italy 0.0075 0.0088 0.0062 The Netherlands 0.0123 0.0125 0.0120 Portugal 0.0151 0.0154 0.0148 Spain 0.0112 0.0115 0.0109 Euro area 0.0054 0.0060 0.0049
Ireland stands out with regard to both the average slope of the expansions and the average slope of the downturns as the values are substantially higher than the rest. This is consistent with the previous two features, as the average amplitudes for Ireland are relatively high while the average durations for Ireland are aligned with the other countries. For the majority of the countries, the average slope of the expansion phase is larger than the average slope of the downturn. Using the aforementioned definition by Claessens et al. (2011), this means that
20 expansions are more violent than downturns. However, it should be noted that the differences are small.
4.1.2 Synchronization
This section will provide a first glance at the behavior of the synchronization of the financial cycles of the eleven selected euro area countries over time. A synchronization measure is constructed by comparing the financial cycle of a country with the euro area reference cycle. Appendix Bcontains graphs for every country with the synchronization plotted over time. For most countries, this fluctuates quite heavily. No clear pattern can be detected regarding the behavior of the synchronization measure around the crisis.
Table 6 shows what the average synchronization for each country is for the whole lifespan of the EMU (1999Q1 until 2017Q3), for the pre-crisis period (1999Q1 until 2007Q4) and for the period during and after the crisis (2008Q1 until 2017Q3). It should be noted that the division in these two time periods is not based on an analysis of the data yet. This means that the division does not represent the timing of the impact of the crisis on the synchronization of financial cycles accurately. Table 6 also displays the difference between the average of the pre-crisis period and the period during and after the crisis and whether that difference is significantly different from zero.
Table 6: Average synchronization with the euro area reference cycle
1999Q1 – 2017Q3 average 1999Q1 – 2007Q4 average (1) 2008Q1 – 2017Q4 average (2) Difference (2) - (1) Significance difference (p-value) Austria 0.1988 0.4240 0.0139 -0.4101 0.0127 Belgium 0.1973 0.1471 0.2436 0.0965 0.491 Finland 0.1203 0.0027 0.2289 0.2262 0.1074 France 0.2063 0.1860 0.2250 0.039 0.8052 Germany 0.0784 0.2762 -0.1048 -0.381 0.0152 Greece 0.1886 0.3300 0.0581 -0.2719 0.0847 Ireland 0.164 0.1816 0.1477 -0.0339 0.7401 Italy 0.3387 0.3393 0.3382 -0.0011 0.9944 The Netherlands 0.0222 -0.1168 0.1505 0.2673 0.0552 Portugal -0.0891 -0.2238 0.0352 0.259 0.0643 Spain 0.2196 0.3151 0.1313 -0.1838 0.1699
21 Using a 5% significance level, the difference is only significant for Austria and Germany, who both have a lower average during and after the crisis than before.
4.2 Regression
To determine the best-fitted value for T for the dummy variable D, which is 0 when t < T and 1 if t ≥ T, likelihood ratio tests are performed. The results of the likelihood ratio tests are depicted in figure 2. The horizontal axis has the different quarters that were chosen for T and the vertical axis contains the chi-squared obtained from the likelihood ratio test.
The quarter with the highest chi-squared is chosen and that is the fourth quarter of 2008. The chi-squared from the likelihood ratio test for 2008Q4 is 16.64 and the associated p-value is 0.000, which means that the null model is rejected at the 1% significance level. It is likely that there is a structural break in the time trend of the synchronization of financial cycles in the EMU at 2008Q4. The timing makes sense, because 2008Q4 is the quarter immediately after the collapse of Lehman Brothers, which is often seen as the start of the financial crisis.
Table 7 shows the estimates for the regression equation 𝜑𝑖𝑡 = 𝛼 + 𝛽1𝑡 + 𝛽2𝐷𝑡 + εit,
where D = 0 if t < 2008Q4 and D = 1 if t ≥ 2008Q4. 0 2 4 6 8 10 12 14 16 18 Chi -s q u a re d
22
Table 7: Regression results
Variable Coefficient Standard error P-value
t -0.0101692*** 0.002176 0.000
Dt 0.0017926*** 0.0004407 0.000
α 1.929968*** 0.381522 0.000
*10% significance, **5% significance, ***1% significance
Overall, the level of synchronization declines over time as −0.0101692𝑡 + 0.0017926𝐷𝑡 < 0. That means that financial cycles in the EMU are diverging, which is consistent with the research of Merler (2015). The positive coefficient for 𝐷𝑡 implies that the divergence has been weaker since the crisis, which suggests that the crisis had a positive impact on the synchronization of financial cycles in the EMU.
Since a structural break in the time trend in synchronization around the crisis was found, the likelihood ratio tests are also performed to test for a second structural break. The results of these likelihood ratio tests are shown in figure 3.
The quarter with the highest chi-squared is the fourth quarter of 2010. The associated p-value is 0.1318, which means that the null model is not rejected. It is not likely that there is a second structural break after the crisis in the time trend of the synchronization of financial cycles in the EMU.
0 0.5 1 1.5 2 2.5 Chi -s q u a re d
23
5. Conclusion
The literature on financial cycles is growing steadily since the crisis. However, little is known yet about the synchronization of financial cycles. This is particularly relevant for a monetary union such as the euro area. On the one hand, financial cycles that are more synchronized make it easier to conduct a one size fits all policy. On the other hand, enhanced synchronization decreases the benefits of international diversification for financial intermediaries, which implies higher risk. There has been a trend of decreasing synchronization, or divergence, of financial cycles within the EMU since its start. Two explanations are the enhanced financial integration that the EMU brought along, leading to different credit developments at the national level and the heterogeneity in mortgage markets and therefore the heterogeneous response of house prices to monetary policy, leading to different developments in house prices. There has been financial disintegration since the crisis. The heterogeneity in house price developments remains but was lower during the crisis and is decreasing in general. Therefore, this thesis has researched what the impact was of the global financial crisis on the synchronization of financial cycles in the EMU.
The results indicate that the crisis had a positive impact on the synchronization of financial cycles in the EMU. It is likely that there is a structural break in the time trend of synchronization of financial cycles in the fourth quarter of 2008. The findings confirm that financial cycles have been diverging since the start of the EMU. From the fourth quarter of 2008 financial cycles are diverging less than before, indicated by a less steep decline of the level of synchronization over time. The timing can be explained by the collapse of Lehman Brothers at the end of the third quarter of 2008, after which the crisis manifested fully. No second structural break at a later point in time was found.
The policy implications of the results are tricky. Enhanced synchronization makes it easier to design one policy that fits all EMU countries, which would favor entrusting the ECB with stronger macroprudential powers. This can be desirable because financial spillovers from one country to another can be very strong in a monetary union and national authorities might not have enough incentive to internalize these. As the chance of success of a macroprudential policy for the euro area as a whole grows with increasing synchronization, the rationale for having one strengthens. However, the finding that financial cycles are diverging less after the crisis does not imply convergence. It is not clear how much convergence or how little divergence would be necessary to make a single policy for the EMU a success. Higher synchronization also implies higher risk for financial intermediaries because the benefits of international diversification decrease. This is an important result for European supervisors and should be monitored appropriately.
24
References
Aikman, D., A. G. Haldane, and B. D. Nelson (2015). Curbing the Credit Cycle. The Economic Journal 125 (585), 1072–1109
Baele, L., Ferrando, A., Hördal, P, Krylova, E. and Monnet, C. (2004) Measuring financial integration in the Euro area. ECB Occasional Paper Series No. 14
Baxter, M. and R. G. King (1999) Measuring Business Cycles: Approximate Band-Pass Filters For Economic Time Series. The Review of Economics and Statistics 81 (4), 575– 593
Bordon, I., Reade, J. J., & Volz, U. (2013). A New Measure of Business Cycle Concordance. University of London and German Development Institute
Borio, C., Drehmann, M. and K. Tsatsaronis (2012) Characterising the financial cycle: don’t lose sight of the medium term! BIS Working Paper No. 380
Borio, C. (2014) The financial cycle and macroeconomics: What have we learnt? Journal of Banking & Finance 45, 182–198
Calza, A., Monacelli, T. and Stracca, L. (2013) Housing finance and monetary policy. Journal of the European Economic Association 11 (1), 101-122
Carstensen, K., Hülsewig, O., Wollmershäuser, T. (2009) Monetary policy transmission and house prices: European cross-country evidence. CESifo Working Paper, No. 2750 Christiano, L. J. and T. J. Fitzgerald (2003) The Band Pass Filter. International Economic
Review 44 (2), 435–465
Claessens, S., Ayhan Kose, M. and Terrones, M. (2011) Financial cycles: What? How? When? NBER International Seminar on Macroeconomics 7 (1), 303-344
Darvas, Z., De Sousa, C., Huettl, P., Merler, S. and Walsh, T. (2014) Analysis of developments in EU capital flows in the global context. Bruegel Final Report. De Nicolò, G. and Tieman, A. (2006) Economic Integration and Financial Stability: A
European Perspective. IMF
De Santis, G. and Gerard, B. (1997) International Asset Pricing and Portfolio Diversification with Time-Varying Risk. The Journal of Finance 52 (5), 1881-1912
De Sola Perea, M. and Van Nieuwenhuyze, C. (2014) Financial integration and fragmentation in the euro area. National Bank of Belgium Economic Review ECB (2008) Economic Bulletin. Issue 7/2008
ECB (2009) Economic Bulletin. Issue 6/2009 ECB (2010) Economic Bulletin. Issue 12/2010 ECB (2011) Economic Bulletin. Issue 11/2011 ECB (2012) Economic Bulletin. Issue 5/2012 ECB (2013) Economic Bulletin. Issue 11/2013
25 ECB (2015) Economic Bulletin. Issue 6/2015
ECB (2016) Economic Bulletin. Issue 7/2016
Galati, G., Hindrayanto, I., Koopman, S. J., Vlekke, M. (2016) Measuring financial cycles with a model-based filter: Empirical evidence for the United States and the euro area. DNB Working Paper No. 495
Gros, D. (2012) Macroeconomic Imbalances in the Euro Area: Symptom or cause of the crisis? Centre for European Policy Studies Policy Brief No. 266.
Harding, D. and Pagan, A. (2001) Extracting, Analysing and Using Cyclical Information. MPRA Paper 15, 1-34
Hilbers, P., Hoffmaister, A. W., Banerji, A. and Shi, H. (2008) House Price Developments in Europe: A Comparison. IMF Working Paper No. 211
Horvath, R. (2017) Financial market fragmentation and monetary transmission in the euro area: what do we know? Journal of Economic Policy Reform.
Iacoviello, M. and Minetti, R. (2003) Financial liberalization and the sensitivity of house prices to monetary policy: theory and evidence. The Manchester School 71 (1), 20-34
IMF (2008) World Economic Outlook April 2008. Housing and the Business Cycle. Washington, DC.
IMF (2018) Global Financial Stability Report April 2018: A Bumpy Road Ahead. Washington, DC.
Lane, P. (2008) EMU and Financial Integration. Institute for International Integration Studies Discussion Paper No. 272
Lane, P. and P. McQuade (2013) Domestic credit growth and international capital flows. ECB Working Paper No. 1566
Loutskina, E. (2011) The role of securitization in bank liquidity and funding management. Journal of Financial Economics 100 (3), 663-684.
Merler, S (2015) Squaring the cycle: financial cycles, capital flows and macroprudential policy in the euro area. Bruegel Working Paper
Miles, D. and Pillonca, V. (2008) Financial innovation and European housing and mortgage markets. Oxford review of Economic Policy 24, 145-175
Ruscher, E. and Vašiček, B. (2016) Revisiting the Real Interest Rate Mechanism. Quarterly Report on the Euro Area 14 (4), Institutional Paper 016. Brussels
Schüler, Y., P. Hiebert and T. Peltonen (2015) Characterising the financial cycle: a multivariate and time-varying approach. ECB Working Paper No. 1846
Stremmel, H. (2015) Capturing the financial cycle in Europe. ECB Working Paper No. 1811 Taylor, J. (2008) The Financial Crisis and the Policy Responses: An Empirical Analysis of
What Went Wrong. Congressional paper, Ottawa, November 14th
26 area countries? Evidence from a global VAR. Journal of Housing Economics 20 (4), 299-314
27
Appendix A: Financial cycles
-0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Austria -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Belgium -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Finland
28 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 France -0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Germany -0.15 -0.1 -0.05 0 0.05 0.1 0.15 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Greece
29 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 2 0 0 7 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Ireland -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Italy -0.15 -0.1 -0.05 0 0.05 0.1 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 The Netherlands
30 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Portugal -0.1 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 0.1 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Spain -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 1 9 8 1 Q 1 19 82 Q 2 1 9 8 3 Q 3 1 9 8 4 Q 4 1 9 8 6 Q 1 1 9 8 7 Q 2 1 9 8 8 Q 3 1 9 8 9 Q 4 1 9 9 1 Q 1 1 9 9 2 Q 2 1 9 9 3 Q 3 1 9 9 4 Q 4 1 9 9 6 Q 1 1 9 9 7 Q 2 1 9 9 8 Q 3 1 9 9 9 Q 4 2 0 0 1 Q 1 2 0 0 2 Q 2 2 0 0 3 Q 3 2 0 0 4 Q 4 2 0 0 6 Q 1 20 07 Q 2 2 0 0 8 Q 3 2 0 0 9 Q 4 2 0 1 1 Q 1 2 0 1 2 Q 2 2 0 1 3 Q 3 2 0 1 4 Q 4 2 0 1 6 Q 1 2 0 1 7 Q 2 Euro area
31
Appendix B: Synchronization
-1.5 -1 -0.5 0 0.5 1 1.5 2 0 0 0 Q 1 2 0 0 0 Q 4 2 0 0 1 Q 3 2 0 0 2 Q 2 2 0 0 3 Q 1 20 03 Q 4 2 0 0 4 Q 3 2 0 0 5 Q 2 2 0 0 6 Q 1 2 0 0 6 Q 4 2 0 0 7 Q 3 2 0 0 8 Q 2 2 0 0 9 Q 1 2 0 0 9 Q 4 2 0 1 0 Q 3 2 0 1 1 Q 2 2 0 1 2 Q 1 2 0 1 2 Q 4 2 0 1 3 Q 3 2 0 1 4 Q 2 2 0 1 5 Q 1 2 0 1 5 Q 4 20 16 Q 3 2 0 1 7 Q 2 Austria -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Belgium -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Finland32 -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 France -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Germany -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Greece
33 -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Ireland -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Italy -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 The Netherlands
34 -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Portugal -1.5 -1 -0.5 0 0.5 1 1.5 1 9 9 9 Q 1 1 9 9 9 Q 4 2 0 0 0 Q 3 2 0 0 1 Q 2 2 0 0 2 Q 1 2 0 0 2 Q 4 2 0 0 3 Q 3 2 0 0 4 Q 2 2 0 0 5 Q 1 2 0 0 5 Q 4 2 0 0 6 Q 3 2 0 0 7 Q 2 2 0 0 8 Q 1 2 0 0 8 Q 4 2 0 0 9 Q 3 20 10 Q 2 2 0 1 1 Q 1 2 0 1 1 Q 4 2 0 1 2 Q 3 2 0 1 3 Q 2 20 14 Q 1 2 0 1 4 Q 4 2 0 1 5 Q 3 2 0 1 6 Q 2 2 0 1 7 Q 1 Spain
