1
Risk ratings and Dutch institutional investors
How risk ratings affect the size of assets held by Dutch pension funds and insurance companies in different countries during the crisis
Yashini Sagoenie 10381228
BSc Economics and Business
Specialisation: Economics and Finance
June, 2015
2
Statement of Originality
This document is written by Student Yashini Sagoenie 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.
3
Acknowledgements
Upon completion of this thesis, I would like to thank the following people. First of all, I wish to thank D. van Dijk PhD for guiding me through this process. Despite the fact that I still need to do the introductory course in Econometrics next year, I have been able to do extensive regressions and interpret the results adequately thanks to the guidance of Mr. Van Dijk. Besides, he quickly gave critical, useful advice and comments, which has helped me writing an even better thesis.
Moreover, I am grateful to my parents for their support and advice when I had to write my thesis.
4
Abstract
This thesis investigates whether political risk rating, regulatory risk rating and trade credit risk rating influence the amount of assets held abroad by Dutch pension funds and
insurance companies and hence investigates if a flight to quality exists in the period 2008 – 2013. The dataset consists of the amount of assets held in 74 countries, both Eurozone and non-Eurozone countries. This is tested against political risk rating, regulatory risk rating and trade credit risk rating. Other control variables are consumer price index, gross domestic product, current account and foreign domestic investment. The results show that political risk rating and trade credit risk rating significantly affect the amount of assets held in a country, implying that a higher (thus safer) risk rating leads to a higher amount of assets held which can be considered to be a flight to quality. Besides, evidence for home bias is also found.
Keywords: pension funds, insurance companies, political risk, regulatory risk, trade
5
Content
1. Introduction ... 6
2. Literature review ... 8
2.1 Pension funds and insurance companies ... 8
2.2 Flight to quality and flight home ... 9
2.3 Other related literature ... 10
3. Methodology ... 12
4. Data and descriptive statistics ... 14
4.1 Data selection ... 16 4.2 Descriptive statistics ... 19 4.3 Hypotheses ... 21 5. Results ... 22 5.1 All countries ... 22 5.2 Eurozone ... 24
5.3 Heteroscedasticity and autocorrelation ... 25
5.4 Empirical results and discussion ... 26
6. Limitations and recommendations ... 27
7. Conclusion ... 28
References ... 30
6
1. Introduction
In April 2015, insurance companies and pension funds accounted for 39% and 33% of total institutional assets under management, respectively (EFAMA, 2015). Together they are responsible for more than half of the total assets under management in Europe. Yet, there hasn’t been much research on their trading behaviour during the financial crisis, whilst research on the trading behaviour of other investors has been highly popular.
Since insurance companies and pension funds hold relatively large amounts of assets, they can influence the (European) financial markets. Hence, it is interesting to look at how the crisis changed their trading behaviour and their asset allocation among different countries. Bijlsma and Vermeulen (2015) examined the trading behaviour of Dutch insurance companies during the crisis where they focused on the ‘flight home’ and the ‘flight to quality’. Their results imply that insurers invested more in northern European bonds instead of Southern and Dutch bonds. Insurers hence ‘flew to quality’ but they didn’t become more home biased. One of the main reasons why institutional investors wanted to hold more Northern-European securities than Southern-European is because of the risk that was involved when holding these securities. Greece, Spain and Italy, for example, had deteriorating government balances and fiscal deficits which also led to social and political unrest; riots and strokes were not
uncommon in these countries. Besides, the political parties became more and more polarised.
Since the before mentioned institutional investors are important for financial markets, it is especially interesting to look at the role of country risk ratings on institutional investors’ asset allocation during and after the crisis. The Netherlands play a big role on the institutional investment market. Dutch pension funds account for almost 20% of all European assets under management (PWC, 2014), which is equal to around €1,200 billion (DNB, 2015). Insurance companies’ assets are equal to 69% of GDP (Bijlsma and Vermeulen, 2015).
In this thesis will be examined whether Dutch insurance companies and pension funds experienced a flight to quality based on country risk ratings. After looking at statistical data from the DNB (Dutch Central Bank) on total amount of assets held in certain countries and risk ratings derived from DataStream (Oxford Economics) the following research question is derived:
To what extent did the Dutch institutional investors change the amount of assets held in different countries as response to changes in country risk?
7
A multiple regression model will be used in this data set to see if there is a significant relation between the country risk ratings and the value of the assets that institutional investors hold in the specific countries.
This approach differs from existing literature because this thesis pays attention to the amount of assets held by both Dutch pension funds and insurance companies in countries within Europe, but also outside of Europe. Besides, the country risk rating will be based on: political risk, regulatory risk and trade credit risk. By looking at countries outside of Europe, the effect of political risk can be more significant since the Arab Spring is an example of mainly
political instability outside of Europe. Moreover, looking at countries outside of Europe gives some more insights in the US and Asian market. The US are interesting because of the high deficit and because of the fact that the US dollar is the most important currency in the world. Asian markets on the other hand are growing faster than European markets and could lead to potential diversification benefits for institutional investors. In general this thesis will focus on the macroeconomic conditions that have an influence on the country weights of institutional investors; firm-specific circumstances are assumed to be constant.
Besides, the majority of the papers focus on the change in trading behaviour by other investors such as banks, mutual funds and hedge funds. However, these investors differ significantly from pension funds and insurance companies. Banks’ main activities are collection of deposits and providing loans, insurance companies and pension funds on the other hand provide risk pooling and risk transformation (Insurance Europe, 2014). Due to these differences in activities, their balance sheets are different. Insurance companies and pension funds try match their assets and liabilities as much as possible by making the balance sheets more stable, banks however don’t have matching assets and liabilities as one of their main objectives.
The rest of the article will be organised in the following way: in part 2 existing literature on related topics will be discussed. Part 3 will present the data and the empirical methodology including a description of the model. Part 4 will show the empirical results, while these results will be discussed in part 5. Finally, the conclusion will be provided in part 6.
8
2. Literature review
2.1 Pension funds and insurance companies
In the period of 2009 – 2013, pension funds’ assets increased on average by 8.2% per year, whereas insurance companies had an average annual growth of 4.1% (OECD, 2014).
Although both pension funds and insurance companies are types of institutional investors that provide insurance against certain events and that invest part of the premiums that they receive from participants who are saving for their retirement and/or (life) insurance, they also differ significantly.
Besides the fact that pension funds’ assets’ annual growth is higher, investors also have most long-term commitments. Pension funds namely invest premiums from the (private) pension plans. Since many workers start saving early, pension funds do not expect to pay out benefits within 30 - 50 years from now. Life insurance companies often have a longer time span as well, whereas non-life insurance companies most often have a short term horizon. Due to this different time horizons pension funds are inclined to take more risk and are less sensitive to a change in their reserves. Due to this distinct term, the companies face other risks. Longevity risks are one of the risks that pension funds face as opposed to insurers that are prone to financial distress costs. Pension funds, however, are basically immune to default (Gorter and Bikker, 2011). Despite having different time horizons, pension funds and (life) insurance companies are more similar to each other than to mutual funds or hedge funds for example. The latter two pool money to invest it. Together, insurers account for one-third of all
institutional investments, which is equal to US $ 20 trillion (Bijlsma and Vermeulen, 2015). In this thesis institutional investors comprises both pension funds and insurers.
Choosing for Dutch companies has some advantages: Dutch insurers are flexible in deciding how to allocate their assets because they are not restrained by regulations due to a principle-based approach in the Netherlands (Bijlsma and Vermeulen, 2015). Also, as stated by Euracs, the Dutch pension funds have the highest global per capita pension reserves; this makes it interesting to look at how this money is invested. Furthermore, the Dutch pension and
insurance system are relatively unique in the world: most employees participate in the pension system and the (health) insurance system is obligatory for every employee.
9
2.2 Flight to quality and flight home
A flight to quality occurs in times of economic and/or financial distress, when investors invest more in safer, more liquid assets and less in risky assets (Beber, Brandt and Kavajecz, 2009). This is also described as flight to liquidity. In their research, Beber, Brandt and Kavajecz (2009) compare the importance of liquidity versus credit quality. Their results suggest that most investors chase liquidity over credit quality in times of severe economic distress. In case of a flight home or home bias, investors prefer investing in their own country over investing in other countries (Bijlsma and Vermeulen, 2015).
This thesis builds upon a recently published working paper from the DNB on insurance companies’ trading behaviour during the crisis that was mentioned before. Bijlsma and Vermeulen (2015) found already in their dataset that the investments in equities declined during the period 2006 - 2013 from 24% to 10%, whereas the investments in government bonds grew from 48% to 58% in that same period. The share of investments in other types of bonds stayed the same. Also, the dataset shows that Dutch insurers hold about one-third of all their assets in the Netherlands hence about two-third of the investments are abroad.
After testing their hypotheses with the F-tests, Bijlsma and Vermeulen’s paper (2015) showed that insurance companies were involved in procyclical investment behaviour during the recent financial crisis. Evidence was found that there was a flight to quality right after the Lehman Brothers collapsed. After Draghi’s ‘whatever it takes to preserve the Euro’-speech in July 2012, this flight to quality disappeared. Hence the fact that the insurers invested more into safer assets worsened the crisis, since capital left the already weakened countries. They didn’t find significant results for home bias. They didn’t find proof for a flight home from
investments in non-euro countries. However, they did find significant results showing that the insurer held more assets in Northern European countries during the crisis. This flight to more Northern European countries can be explained by the fact that they have less political,
regulatory and trade credit risk. It was a public secret that most of the Southern European countries were risky countries to invest into.
The same results were found for mutual funds in Raddatz and Schmukler’s research (2011), although Gianetti and Laeven (2012) found that banks turned out to be more prone to home bias. Raddatz and Schmukler (2011) stated that investors react to a crisis in a country by taking away their money and investing it in countries that perform relatively well; a pro-cyclical reaction. They explain that fund managers behave the same way as investors do,
10
however their reaction is even stronger due to the fact that investors take the money out of the fund and valuation changes in countries (2011). The valuation changes in countries imply that they have to change the country weights in the portfolio.
2.3 Other related literature
According to Manconi, Massa and Yasuda (2010), institutional investors played a big role in the financial crisis. Though, insurance companies and pension funds were not the ones that exacerbated the crisis: the mutual funds were responsible because they have an even shorter time horizon implying that they are more sensitive to volatility in the financial markets. When the crisis started and the securitized bonds turned out to be intoxicated, the mutual funds immediately started selling these securitized bonds and buying the relatively safe corporate bonds. In contrast, insurance companies did not really change their portfolio as long as they met certain capital requirements. This could be explained by the fact that they have a longer time horizon and different purposes than the mutual funds.
On the other hand, De Haan and Kakes (2010) found that Dutch pension funds have the most contrarian investment strategies (against the market/anticyclical) especially when the markets are not stable. Life insurance companies only have contrarian strategies under certain policies in which clients carry the investment risk. Non-life insurance companies are least likely to have a contrarian strategy and tend to have a momentum strategy (more procyclical). This can be explained by the fact that the non-life insurers have a shorter time horizon than life insurers and pension funds. The contrarian strategies are also more evident for sells instead of buys, implying that the investors are more willing to take risk when it comes to future capital gains. Despite the different levels of contrarian strategies, both pension funds and (non-) life
insurance companies tend to have a contrarian strategy. Therefore, their results imply that the institutional investors invest against the market conditions. Thus in times of financial distress, institutional investors can have stabilising effects on the markets.
According to Kacperczyk and Schnabl (2009), institutional investors used to hold a significant share in the money market funds before the crisis. After the collapse of the Lehman Brothers, there was a run on the money market funds and institutional investors reduced their share in these funds by more than $172 million. This high reduction in shares can be explained by the fact that asset-backed securities turned out to be riskier than previously thought. Although this
11
run was weakened by a temporary deposit insurance introduced by the U.S. Department of Treasury, the number of commercial papers outstanding dropped by almost 30%. Since this was the case in the US, it is highly possible that the discovery of these risks is one of the reasons why there was a large capital outflow from the US.
In a research done by Ferreira and Matos (2008) on the role of institutional investors around the world, they find that institutional investors have a preference for countries that have a good disclosure standard, because this leads to more transparency. They also have a preference for stocks from countries that are relatively close to their home country, but not necessarily English-speaking. Moreover, they prefer countries that are more developed. They also examine more factors that influence both US and non-US institutional investors’
preferences and they explain that the amount invested in foreign companies is also very often driven by firm-specific circumstances in combination with macroeconomic factors. Although they claim they look at companies worldwide, they do make a distinction between US and non-US companies.
Driessen and Laeven (2007) look at companies located in mid-size and large economies to see if international diversification leads to (cost) benefits for institutional investors while looking at financial and macroeconomic variables. One of the results they find is that global
diversification led to an average increase in the Sharpe ratio of 7.8% for developed countries in case of USD returns and no short-sale restrictions. Developing countries, however, have an average increase in the Sharpe ratio of about 13.5% under the same circumstances. Reason for this is that developing countries are not yet as integrated in the world market as developed countries. Applying this information to the Netherlands would lead to an average increase of 7.8% in average excess return/volatility ratio for Dutch institutional investors. The most important conclusion that Driessen and Laeven find in their paper, is that the country risk seems to be a valid determinant for estimating the diversification benefits.
The disadvantage of their research is that it was done just before the crisis. They also use country risk measures to measure if and how the benefits from international diversification change over time. It is highly interesting to see if their findings still hold during and after crisis.
12
All these different strategies are important to analyse the trading behaviour of institutional investors. Trading behaviour and risk preferences determine when and by how much they decide to take away the money from certain countries.
3. Methodology
Since the research is on assets held by institutional investors in the time period 2008 – 2013 for 74 different countries, there are time-serial and cross-sectional relations within the data. Hence a panel data set is created to look at how the variables affected the assets held in certain countries. By doing a multiple regression, significant outcomes may occur.
The y-variable or the dependent variable in the multiple regression is the amount of assets held by the institutional investors in the 74 different countries in the period 2008 – 2013. The x-variables or the independent variables are the political risk rating (x1), the regulatory risk rating (x2) and the trade credit risk rating (x3). The control variables are consumer price index (x4), GDP per capita (x5), current account (x6), foreign direct investments (x7) and the dummy variable Eurozone (x8). See table 1 for more information. Table 2 in the next chapter provides a more elaborative description of the data.
The basic model looks as follows:
Ln(Assets)i,t = β0 + β1PRi,t + β2RRi,t + β3TCRi,t + β4CPIi,t + β5ln(GDP)i,t + β6CAi,t + β7FDIi,t + β8EURi,t + αi + εi,t
Since the dataset is panel data, both the pooled ordinary least squared (OLS) or the fixed effects model can be used to test the model. The latter model looks at the relation between the x- and y-variables within the countries. The model above is a country fixed model meaning that this model controls for characteristics that differ over countries, but are constant over time. These characteristics are assumed to still exist in later periods, hence they are called time-invariant, for example the fact that Spanish people tend to take a break (siesta) during the day. By taking out the outcome of the time-invariant variables it will be possible to reduce the omitted variable bias (Torres-Reyna, 2007). The fixed effects model is an often used model because its assumption come closest to reality. Alpha controls for the fixed effects and
13
epsilon is the error term. Pooled OLS neglects the fact that there exist country-specific differences.
The Hausman test1 outcome showed that the null hypothesis2 was rejected and hence confirmed the expectation that the fixed effects model is the appropriate model for this dataset. Despite this outcome both the pooled ordinary least squared and fixed effects method are tested to see if they lead to the same conclusions.
Table 1 Description of the different variables used in model.
Variable Description Time period Level Source
Assets Amount of assets held in, in euros, converted into natural logarithm
2008 – 2013 Country Dutch Central Bank
PR Political risk rating (categorised)
2008 – 2013 Country DataStream RR Regulatory risk
rating (categorised)
2008 – 2013 Country DataStream TCR Trade Credit risk
rating (categorised)
2008 – 2013 Country DataStream CPI Consumer Price
Index, annual percentage 2008 – 2013 Country WorldBank GDP Gross Domestic Product, in euros, converted into natural logarithm 2008 – 2013 Country WorldBank CA Current Account, in billion euros 2008 – 2013 Country WorldBank FDI Foreign Direct
Investment, in billion euros
2008 – 2013 Country WorldBank
EUR Dummy variable: country in Eurozone or not
2008 – 2013 Country European Central Bank
The table provides an overview of the variables used in the regression model. The assets are the dependent variable. Political risk rating, regulatory risk rating and trade credit risk rating are the independent variables. Inflation, Gross Domestic Product, Current Account and Foreign Direct Investment are the control variables. Eurozone is the dummy variable. The overview describes the variable, provides the time period during which data is gathered, shows at which level the data is gathered and tells where the data is gathered from.
1 Hausman: H = (β
re – βfe)’[ Var(βfe) − Var(βre)]-1(βre – βfe) (Clark & Linder, 2012) 2
14
An additional model that is tested includes the time fixed effects. The time fixed effects model controls for factors that are similar for all countries, but they differ over time. The time fixed effects looks as follows, where the lambda controls for time fixed effects:
Ln(Assets)i,t = β0 + β1PRi,t + β2RRi,t + β3TCRi,t + β4CPIi,t + β5ln(GDP)i,t + β6CAi,t + β7FDIi,t + β8EURi,t + αi + λt + εi,t
This research focuses on how the country risk affects the amount of assets held by Dutch pension funds and institutional investors in certain countries, the hypotheses that will be tested are discussed in the next chapter.
In total four regressions will be run: the first regression is the pooled OLS without control variables, the second regression is the pooled OLS with control variables, thirdly the fixed effects model with country fixed effects will be tested and the fourth regression is the fixed effects model with country- and time fixed effects. The four regressions are run in order to see if there are significant differences among them.
4. Data and descriptive statistics
In this section the data is presented and it is shown how this data is used to answer the research question. First the data selection is shown, hereafter an overview of the variables’ characteristics is given and this part will end with the hypotheses.
When looking at the dataset, it is interesting to see that overall the amount of assets held by the institutional investors was significantly lower in the period 2009 - 2011; just after the crisis started (see also figure 1).
Besides, some countries have a large decrease or increase in the amount of assets held by pension funds and insurers in the abovementioned period. Data on the PIIGS countries (Portugal, Ireland, Italy, Greece and Spain), for example, shows that pension funds and insurers had a maximum decrease in the amount of assets held respectively by 87.6%; 12.8%; 78.2%; 99.2% and 62.5%. Figure 2 shows how the size of the assets changed during the period of 2008 – 2013. One of the reasons for these immense decreases could be the high risk of default during the financial crisis where the institutional investors were exposed to.
15
The US also had a significant decrease of 43.7% which can be explained by the higher default risk, although the investors apparently thought the US were more trustworthy.
The institutional investors, on the other hand, increased the assets held in Germany by 49.8%. Germany was considered to be one of the safer countries. Figure 3 shows the change in the size of assets in some of the important countries, such as the US and Germany.
Figure 1 Total assets held in billion euros
The figure provides a graphical view of the change in amount of the total amount of assets in other countries held by Dutch pension funds and insurance companies in the period 2008 – 2013.
Figure 2 Assets in PIIGS per year
The figure shows the size of the assets held by institutional investors in the period 2008 – 2013 in the PIIGS countries; countries that were in financial problems during the financial crisis. These countries include: Greece, Ireland, Italy, Portugal and Spain.
517.938
435.561 446.123 434.512
500.809 498.964
2008 2009 2010 2011 2012 2013
Total assets held in billion euros
Total assets held
0 5 10 15 20 25 30 35 40 2008 2009 2010 2011 2012 2013 A m o u n t o f assets h e ld in b ln € Greece Ireland Italy Portugal Spain
16
Figure 3 Assets in important countries
This figure looks at the asset allocation in different countries that show an increase in the amount of assets held by institutional investors in the period 2008 – 2013. Belgium, France, Germany, United Kingdom prove to be stable. Although the US had a decrease in the amount of assets held by institutional investors, they are assumed to be a strong economy. Hence the US is also included in this graph.
4.1 Data selection
4.1.1 Dependent variable
Since this research answers the question how the three different risk components affect the size of the assets outstanding in different countries, the most important variable is the assets held by pension funds and insurers in the different countries around the world. The DNB-website offers a clear overview of all the countries in which the institutional investors invest and it is divided over different sectors. The focus lies at the pension funds and insurance companies, since that is the main focus point of this research. Since this is the case, only data on pension funds and insurance companies is used. The overview included 242 countries of which some didn’t even have assets held by institutional investors. Hence, the countries used in this research are the countries that have significant amounts of assets held by pension funds and insurers and the countries for which ratings could be found. In total, 74 countries are observed over a time period of five years (2008 – 2013).
4.1.2 Independent variables
Different credit rating agencies (Moody’s, S&P, Fitch) look at the situation in a country and rate this country based on different aspects and weights. These agencies are not the only ones
0 20 40 60 80 100 120 140 160 180 2008 2009 2010 2011 2012 2013 A m o u n t o f assets h e ld in b ln €
Assets in important countries
Belgium France Germany United Kingdom United States
17
that rate countries; the OECD, Oxford Economics and other independent institutions rate countries based on different aspects.
The data on the different risk factors is found on DataStream (Oxford Economics). The three incorporated risk ratings are: political risk, regulatory risk and trade credit risk. All three factors have a scale of 1 – 7 where 1 is implies high risk and 7 implies low risk. Political risk looks at the social stability in terms of government stability, corruption, sovereign risk, sanctions and political unrest. Iraq has the highest political risk with a risk rating of approximately 2.5. Iraq is well-known for its high social instability, corruption and high political unrest due to the war. Regulatory risk focuses on trade restrictions, fiscal policies and exchange controls. Bolivia currently has a regulatory risk of 1. Although new regulatory rules have been implemented and banking supervision has become more strict, the IMF still fears that for example minimal credit quotas can be very harmful for the financial system (AMB, 2014). Also, the recent nationalisation of three Spanish companies has led to higher regulatory risk (Fides, 2013). Trade credit risk is related to the credit risk related to defaults,
bankruptcies and insolvency of government. In the early beginning of the credit crisis, Greece had a trade credit risk of 6, however a few years later it was downgraded to a trade credit risk of 2. The increased risk of default by the Greek government and the risk of Greece leaving the Eurozone have led to higher risk.
Some countries didn’t have a risk factor, thus they were also taken out of these data. The total dataset now consists of 74 countries, these countries can be found in Appendix table 7. The risk ratings were categorised into three subgroups: category 1 consists of the factors 1 – 3, category 2 incorporates factors 4 – 5 and category 3 includes 6 – 7. Categorising the risk factors makes it easier to interpret them.
Although International Country Risk Guide is the most popular proxy for risk measure, the risk ratings provided by DataStream are chosen. The International Country Risk Guide looks at political, economic and financial risk whereas the DataStream risk ratings focus on
political, regulatory and trade credit risk. Since this research investigates the investment behaviour of institutional investors, the risk rating categories provided by DataStream were more suitable for this research, since regulatory risk incorporates the investment climate and the trade credit risk is directly link to the creditworthiness of the government. These two categories are relevant for the institutional investors’ asset allocation.
18 4.1.3 Control variables
According to Oxfordeconomics, risk warnings are amongst others based on GDP growth, CPI inflation, current account balance and government debt. By looking at some of these variables separately, it is possible to see if they also affect the amount of assets held in these countries. Although they might be correlated, it could be interesting to find for example that the risk ratings are adjusted quite late as a response to the changes in the before mentioned factors.
The Consumer Price Index (CPI) was found in the Worldbank’s database and measures the price inflation. The CPI is measured in annual percentages and is based on the Laspeyres formula. Laspeyres is a ‘base-weighted’ index since one base year quantity is adjusted for prices in different years.
Gross Domestic Product (GDP) is per capita in current USD and is found in the Worldbank’s database. It is equal to the country’s GDP divided by the midyear population. GDP is the standard measure of the value of goods and services produced by a country in a certain time period minus the value of the imports. Luxembourg and Norway have the highest GDP per capita ($106,803 and $92,836 respectively).
The current account balance (CA) is part of the balance of payments and is measured in current USD. This information was derived from the Worldbank’s database together with the foreign direct investment (FDI). The current account is the sum of the net exports of goods and services and the net primary and secondary income and indicates how healthy a country’s economy is. If the current account is negative for example, it would imply that a country has imported more than it exported and hence it has to borrow money from abroad. When a company attains a significant interest (more than 10%) in a company abroad, it is foreign direct investment. FDI can be seen as a form of globalisation.
GDP, current account and foreign direct investments are in billion euros. The original data was in dollars, however all numbers are converted into euros by taking one average exchange rate throughout the years. By taking one average exchange rate, the amounts are not affected by any exchange rate volatility. The average exchange rate is derived from taking the average exchange rate from dollar to euro for every day during the period 2008 – 2013. These
19
The last factor states whether a country was part of the Eurozone or not, this information is derived from the European Central Bank’s website.
Figure 4 Overview of the Euro area (dark blue), EU member countries with currency pegged to euro (light blue) and EU member countries with floating currency (green) (the Economist, 2015)
4.2 Descriptive statistics
Table 2 shows an overview of the summary statistics for this model. Table 6 in the appendix shows the correlation between all the variables. The correlation matrix shows that GDP and CPI have some correlation with risk ratings. In case the results show that GDP and CPI are relatively unsignificant, this can be ignored.
Table 2 shows the number of observations, the mean, standard deviation and the minimum and maximum observation. Assets and GDP have high means and high standard deviations, by taking the natural logarithm of these two, these high averages can be avoided and it will be easier to interpret the results. Current account and foreign direct investments originally also had high numbers, however, since the numbers are in billion euros the numbers seem lower and less volatile and it becomes easier to interpret the results.
In this dataset, no country has a political risk rating below 2.5, which means that political risk in this dataset is not as high as regulatory risk and trade credit risk in some countries. This is supported by higher standard deviations for these two risk rating being more volatile. Trade credit risk rating and consumer price index have fewer number of observations than the other variables, because some countries are missing. Since the number of missing variables is low, the outcomes are still reliable. CPI has a relatively high standard deviation, which can be explained by some high outliers such as the hyperinflation in Belarus.
20
Table 2 Summary statistics
Variable Observations Mean Standard deviation
Min Max
Assets 432 6.59e+09 1.90e+10 0 1.64e+11
Political risk 432 4.7918 1.0457 2.5 6.916
Regulatory risk 432 5.7847 1.6804 1 7
Trade credit risk 426 4.2873 1.7617 1 7
Consumer Price Index 418 4.6117 5.3786 -4.48 59.22 Gross Domestic Product 432 17067.56 16658.74 722.30 83239.52 Current Account 432 -1.2147 56.0131 -502.78 308.11 Foreign Direct Investment 432 -0.2749 24.2133 -168.79 133.93 Eurozone 432 0.2014 0.4015 0 1 Ln(assets) 432 19.32071 4.5523 0 25.82 Political risk (categorised) 432 2.2060 0.6143 1 3 Regulatory risk (categorised) 432 2.5671 0.6710 1 3
Trade credit risk (categorised)
426 1.9859 0.8057 1 3
Ln(GDP) 432 9.2131 1.1205 6.58 11.33
This table summarises the results of all variables included in the model including the unconverted variables (Assets, political risk rating, regulatory risk rating, trade credit risk rating and Gross Domestic Product), the dependent variable (ln(assets)), the independent variables (political risk rating categorised, regulatory risk rating categorised, trade credit risk rating categorised), the control variables (Consumer Price Index, ln(Gross Domestic Product) and Foreign Direct Investment) and the dummy variable Eurozone.
The mean and standard deviation for variables Eurozone and the categorised risk ratings have little additional value, because they are dummy variables. Eurozone can only be 0 or 1 and the risk ratings can only be 1, 2 or 3 (see also paragraph 4.1.2 and 4.1.3). Hence an unrounded average doesn’t tell that much.
Figure 5 in the Appendix shows that the residuals are non-normally distributed, however they are also not really skewed. Testing with clustering will help to solve this problem.
The graphs (Appendix figure 6, graph 7) show that Belarus had a big change in the inflation, so called hyperinflation, (from 7.74% in 2010 to 53.23% in 2011) which can be explained by a huge devaluation of the currency in 2011 as a result of gap in the balance of payment (Badkar, 2011). Moreover, the graphs in Appendix figure 7 show that all the Eurozone countries had a sharp fall in the CPI in 2009, a rise in 2010-2011 and slowly started to decrease/stabilise again in 2012. The sharp fall was caused by the crisis, the rise can be
21
explained by the actions that were taken by the governments in order to prevent further worsening of the crisis, including the announcement of three new supervisory authorities by the EU. One factor affecting the stabilisation of the inflation is Draghi’s famous speech in 2012.
4.3 Hypotheses
Based on articles that were discussed in chapter 2, where a flight to quality but no home bias was found in times of economic and financial distress in the Eurozone, the following
hypotheses are derived to see if the same results are found for countries both in- and outside the Eurozone.
Hypothesis 1.1: Higher political risk rating positively affects the amount of assets held by institutional investors in a country
Hypothesis 1.2: Higher regulatory risk rating positively affects the amount of assets held by institutional investors in a country
Hypothesis 1.3: Higher trade credit risk rating positively affects the amount of assets held by institutional investors in a country
Thus, higher risk ratings, implying relatively safer countries, should have relatively higher amounts of assets held by Dutch institutional investors. However, the Eurozone countries specifically will be tested as well for the same relationships to see if the results are in line with what the articles found before.
Hypothesis 2.1: Higher political risk rating positively affects the amount of assets held by institutional investors in a Eurozone country
Hypothesis 2.2: The regulatory risk rating positively affects the amount of assets held by institutional investors in a Eurozone country
Hypothesis 2.3: The trade credit risk rating positively affects the amount of assets held by institutional investors in a Eurozone country
22
5. Results
This paragraph consists of a part in which the results will be analysed and discussed and a part that covers for heteroscedasticity and autocorrelation.
5.1 All countries
From table 3 it becomes clear that political risk rating is significant in all four regressions, hence from this can be derived that in general the amount of assets held by Dutch pension funds and insurance companies in other countries is positively affected if the political risk rating increases by 1. Thus, if the political risk in a country becomes less, the amount of assets held by institutional investors increases by 1.588%, 1.271%, 2.584% or 1.285% respectively. This seems to be intuitive, since institutional investors tend to invest in relatively safe assets and safe countries.
Furthermore, trade credit risk rating is significant in both pooled OLS models and the model that controls for country and time fixed effects, implying that the amount of assets held by Dutch institutional investors is positively related to the trade credit risk rating. Hence, if trade credit risk rating goes up by 1, assets held by the investors increase by 2.022%, 1.683% or 1.861% respectively.
Also, EUR is significant in the pooled OLS with all control variables and the model that controls for country and time fixed effects, which implies that Dutch institutional investors invest 1.522% and 1.591% respectively more in Eurozone countries. Although not mentioned as one of the hypotheses, this result could imply a flight home if the Eurozone is considered to be the domestic market because of the same currency.
Ln(GDP) is only significant in the country fixed effects model. Since both variables are in natural logarithm, the outcome is an elasticity. Hence, a one-percent change in ln(GDP) leads to a 5.991%-change in assets. The constant term is significant in all four models and stands for the predicted value of y in case all independent variables are simultaneously equal to zero. Since some of the independent variables are not equal to zero, the constant term is not
23
Table 3 Results for pooled OLS with and without control variables and country fixed effects (excluding and including time fixed effects) for all countries
All countries Pooled OLS without control variables (1) Pooled OLS with control variables (2) Country fixed effects with control variables (3)
Country & time fixed effects (4) PR 1.588*** (3.35) 1.271** (2.55) 2.584*** (3.08) 1.285*** (2.60) RR 0.065 (0.17) -0.132 (-0.31) 0.059 (0.08) 0.051 (0.12) TCR 2.022*** (5.97) 1.683*** (4.74) -0.245 (-0.51) 1.861*** (5.22) CPI -0.049 (-1.25) 0.036 (1.08) -0.034 (-0.83) Ln(GDP) 0.218 (0.72) 5.991*** (5.79) 0.018 (0.06) CA -0.005 (-1.51) 0.002 (0.30) -0.005 (-1.39) FDI -0.005 (-0.64) 0.005 (0.40) -0.003 (-0.39) EUR 1.522** (3.03) 0.101 (0.07) 1.591** (3.19) Constant 11.650*** (14.67) 11.502*** (5.62) -41.372*** (-4.16) 13.363*** (6.32)
Fixed effects No No Yes Yes
R2 0.2866 0.3211 0.2399 0.3433 N p-value F 426 0.0000 412 0.0000 412 0.0000 412 0.0000
The table shows the results for the different regressions that are done for all countries included in this dataset. The t-statistic is given between parentheses. Pooled OLS is the normal OLS regression with and without control variables (CPI, ln(GDP), CA, FDI, EUR). Country fixed effects account for the factors that differ over country but are fixed over time. The country & time fixed effects also accounts for factors that are constant among the countries but differ over time. *** significant at 1% ** significant at 5% * significant at 10%. 0: omitted because of collinearity. Moreover, R2 is provided to show what percentage of variance can be explained by this model. Besides, N and the p-value of the F-test are included to test whether the model is significant.
Significance levels are the same for both country and time fixed effects and pooled OLS with control variables. However, as explained in the methodology, fixed effects is the appropriate model since that model takes into account both factors that differ over time but are similar across countries and the factors that differ among countries but are constant over time. Hence the results imply that in this case political risk rating, trade credit rating, Eurozone and
constant term are significant.
All tables are significant according to the performed F-test with a 5% p-value, hence the regression models have a good fit for the data.
24
5.2 Eurozone
Table 4 Results for pooled OLS with and without control variables and country fixed effects (excluding and including time fixed effects) for the Eurozone countries
Eurozone
Pooled OLS without control variables (1)
Pooled OLS with control variables (2)
Country fixed effects with control variables
(3)
Country & time fixed effects (4) PR 1.227*** (3.03) -0.940** (-2.00) 0 -0.999** (-2.12) RR 0 0 0 0 TCR 2.013*** (7.14) 1.155*** (4.39) 0.563*** (3.06) 1.376*** (4.80) CPI -0.251* (-2.71) -0.077 (-1.59) -0.374** (-2.43) Ln(GDP) 3.446*** (6.29) 6.836*** (5.09) 3.325*** (5.97) CA 0.003 (0.73) 0.001 (0.26) 0.001 (0.35) FDI 0.027** (3.09) -0.008 (-1.63) 0.030*** (3.19) EUR 0 0 0 Constant 14.043*** (13.87) -13.081** (-2.87) -48.827*** (-3.63) -11.832** (-2.51)
Fixed effects No No Yes Yes
R2 0.5166 0.7074 0.5620 0.7285 N p-value F 87 0.0000 87 0.0000 87 0.0000 87 0.0000
The table shows the results for the different regressions that are done for countries that are part of the Eurozone. The t-statistic is given between parentheses. Pooled OLS is the normal OLS regression with and without control variables (CPI, ln(GDP), CA, FDI, EUR). Country fixed effects account for the factors that differ over country but are fixed over time. The country & time fixed effects also accounts for factors that are constant among the countries but differ over time. *** significant at 1% ** significant at 5% * significant at 10%. 0: omitted because of collinearity. Moreover, R2 is provided to show what percentage of variance can be explained by this model. Besides, N and the p-value of the F-test are included to test whether the model is significant.
The Eurozone results in table 4 have more significant variables than all countries together. Again it is found that pooled OLS and fixed effects have similar outcomes. However, as explained before, the model with country and time fixed effects is the most relevant model. The table shows that both the political risk rating and the trade credit risk rating are
significant, although the political risk rating has the opposite relation. Thus, if the political risk rating goes up by 1, the amount of assets held in that country by the Dutch institutions will decrease by 0.999%. This doesn’t seem to be in line with what is expected and what was found for all countries; to find that the lower risk would have led to higher amounts of assets
25
held. This opposite relationship can be explained by the fact that none of the Eurozone countries has a political risk rating of 1. Hence, relationship between the different political risk ratings could have changed as compared to all countries.
Besides, the control variables are more significant as compared to all the countries together. If CPI, ln(GDP) or FDI increases by 1, the assets held decrease by 0.374%, increase with 3.325% or 0.030% respectively, keeping all other variables constant. Since an increase in inflation leads to higher expected inflation and thus also higher nominal interest rates, bond prices tend to decline. As a result, the value of the assets held diminishes (Tatom, 2011).
Again, all tables are significant according to the performed F-test with a 5% p-value, hence the regression models have a good fit for the data.
5.3 Heteroscedasticity and autocorrelation
These four regressions are run again but now they are also tested for heteroscedasticity and autocorrelation. Heteroscedasticity implies that the covariance of the error term of two variables at one point in time are not equal to zero. The variability of the y-variable will become wider or narrower in the future. Autocorrelation occurs when the covariance of the error term of one variable is not equal to zero for a certain time period (Van Dijk, 2015). By controlling for heteroscedasticity and autocorrelation, it is possible to achieve results that are even more reliable because fewer errors will occur.
Table 5 shows that the same variables turn out to be significant for the country and time fixed effects model, although some significance levels have changed. Coefficients are the same as without clustering, however, the t-statistic values have changed. In fact, they all have become smaller. If the t-value becomes smaller, the confidence interval has a smaller range. Hence a smaller confidence interval with significant results implies more reliable outcomes.
All tables are significant according to the performed F-test with a 5% p-value, hence the regression models have a good fit for the data.
26
Table 5 Results for clustered pooled OLS with and without control variables and clustered country fixed effects (excluding and including time fixed effects) for all countries
All countries Pooled OLS without control variables (1)
Pooled OLS with control variables (2) Country fixed effects with control variables (3)
Country & time fixed effects (4) PR 1.588** (2.45) 1.271* (1.78) 2.584 (1.22) 1.285* (1.82) RR 0.065 (0.08) -0.132 (-0.14) 0.059 (0.16) 0.051 (0.05) TCR 2.022*** (3.47) 1.683*** (2.71) -0.245 (-0.50) 1.861*** (2.88) CPI -0.049 (-1.22) 0.036 (0.49) -0.034 (-0.75) Ln(GDP) 0.218 (0.50) 5.991*** (2.85) 0.018 (0.04) CA -0.005 (-1.17) 0.002 (0.56) -0.005 (-1.06) FDI -0.005 (-0.52) 0.005 (0.99) -0.003 (-0.34) EUR 1.522** (0.62) 0.101 (0.862) 1.591** (2.51) Constant 11.650*** (5.56) 11.502*** (3.23) -41.372* (-1.91) 13.363*** (3.87)
Fixed effects No No Yes Yes
R2 0.2866 0.3211 0.2399 0.3433 N p-value F 426 0.0000 412 0.0000 412 0.1620 412 0.0000
The table shows the results for the different regressions that are done for all countries in this dataset, however corrections have been made to control for heteroscedasticity and autocorrelation. The t-statistic is given between parentheses. Pooled OLS is the normal OLS regression with and without control variables (CPI, ln(GDP), CA, FDI, EUR). Country fixed effects account for the factors that differ over country but are fixed over time. The country & time fixed effects also account for factors that are constant among the countries but differ over time. *** significant at 1% ** significant at 5% * significant at 10%. 0: omitted because of collinearity. Moreover, R2 is provided to show what percentage of variance can be explained by this model. Besides, N and the p-value of the F-test are included to test whether the model is significant.
5.4 Empirical results and discussion
This research shows that political risk rating and trade credit rating are the most important factors that positively influence the amount of assets held by Dutch pension funds and insurers in different countries. Hence, hypotheses 1.1 and 1.3 are not rejected. Though, hypothesis 1.2 can be rejected. These results imply that indeed a flight to quality is visible since the institutional investors change the amount of assets held in a certain country depending on the change in the risk rating.
When focusing on Eurozone countries, the same results are applicable despite the fact that this change is less severe. Besides, the control variables seem to have a bigger influence on the
27
amount of assets held by the institutional investors than for all countries. Hence, this research shows that political risk rating and trade credit risk rating affect the amount of assets held by Dutch institutional investors abroad and hypotheses 2.1 and 2.3 are not rejected. They have higher and more significant effects on this than regulatory risk rating. It suggests that
countries that achieve a relatively higher risk rating (which means they are safer), can lead to more capital inflow.
The results are in line with what was found in abovementioned articles that stated that there was a flight to quality in the case of institutional investors. Bijlsma and Vermeulen (2015) found this flight to quality only after Draghi’s speech, whereas this research shows a flight to quality throughout the crisis period. However, Bijlsma and Vermeulen didn’t find a flight home. If the Eurozone is considered to be the domestic market, this research shows that there was indeed a flight home.
The results don’t support Ferreira and Matos (2008) who found that institutional investors have a preference for countries with a high disclosure standard, since the regulatory risk ratings turned out to be relatively insignificant.
6. Limitations and recommendations
Despite the fact that most results were quite significant, more research can be done in order to see if the results still hold. For example, by comparing the International Country Risk Guide outcomes with these results could lead to a different conclusion. Besides, this study only focuses on the crisis period (2008 – 2013), these results might differ in non-crisis periods.
Other interesting fields for further studies are for example comparing the amount of assets held by Dutch pension funds with the amount of assets held by Dutch insurance companies or comparing the amount of assets held by Dutch institutional investors with the amount of assets held by institutional investors in other countries, such as Germany or Belgium. If the same results are found, then it can be proven that risk ratings can influence capital flights. However, it should be tested whether these risk ratings are often adjusted relatively late as a response to changes in GDP growth, CPI, current account balance and government debt. In
28
the dataset one can see that Greece’s trade credit risk rating only became low in 2012, while their problems started around 2009.
Besides, it is found that regulatory risk rating is not significant. However, the dataset shows that regulatory risk ratings didn’t change as often as political risk rating and trade credit risk rating. If risk ratings don’t change for some countries, it is hard to see if it affected the amount of assets held by institutional investors.
Simultaneous causality should also be taken into account. Despite the fact that the results were significant, hence implying that risk ratings affect the amount of assets held in a certain country. It is quite likely that the amount of assets held by Dutch institutional investors in fact affected the risk ratings. Especially when looking at the trade credit risk rating it is
noteworthy to mention that if there was indeed a flight to quality, the amount of assets held in Greece for example has decreased. In return, this decline could have increased chances of default for the Greek government.
7. Conclusion
After conducting a fixed estimates regression on the panel data set, it was found that political risk rating and trade credit risk rating are the most important factors affecting the amount of assets held by Dutch pension funds and insurance companies in other countries most. The same results were found for the Eurozone countries although these effects were smaller and more control variables were significant.
The results found in this research imply that political risk rating and trade credit risk rating are important determinants for the amount of assets held in certain countries. Hence, to answer the research question: Dutch pension funds and insurance companies change the amount of assets held in different countries as a response to changes in political and trade credit country risk during the crisis. This can indicate a flight to quality. Besides, a flight home is found when the Eurozone is considered to be the domestic market. Pension funds and insurance companies do not change the amount of assets held as a response to a change in regulatory risk.
29
However, more research has to be done to see if the same results hold in a different time period, for other countries and for other risk rating companies. Besides, simultaneous causality should be taken into consideration.
30
References
AMB Country Risk Report Bolivia. (2014, August 18). Retrieved June 17, 2015, from http://www3.ambest.com/ratings/cr/reports/Bolivia.pdf
Beber, A., Brandt, M. W., & Kavajecz, K. A. (2009). Flight-to-quality or flight-to-liquidity? Evidence from the euro-area bond market. Review of Financial Studies, 22(3), 925-957.
Badkar, M. (2011, August 31). Railroaded By Hyperinflation, Belarus Is Now Running Out Of Meat. Retrieved June 17, 2015, from http://www.businessinsider.com/bealrus-hyperinflation-running-out-of-meat-2011-8?IR=T
Bijlsma, Melle, and Robert Vermeulen. "Insurance companies' trading behaviour during the European Sovereign debt crisis: Flight home or flight to quality?." (2015).
Clark, T. S., & Linzer, D. A. (2015). Should I use fixed or random effects? Political Science Research and Methods, 3(02), 399-408.
De Haan, L., & Kakes, J. (2011). Momentum or contrarian investment strategies: evidence from Dutch institutional investors. Journal of Banking & Finance, 35(9), 2245-2251. De Nederlandsche Bank (2015). Balance sheet of pension funds [Data file]. Retrieved June
29, 2015, from https://www.statistics.dnb.nl/index.cgi?lang=uk&todo=Pen1 Driessen, J., & Laeven, L. (2007). International portfolio diversification benefits:
Cross-country evidence from a local perspective. Journal of Banking & Finance, 31(6), 1693-1712.
European Fund and Asset Management Association. (2015). Asset Management Report 2015. Retrieved from http://www.efama.org/statistics/SitePages/Asset%20Management %20Report.aspx
Ferreira, M. A., & Matos, P. (2008). The colors of investors’ money: The role of institutional investors around the world. Journal of Financial Economics, 88(3), 499-533.
Fides, R. Bolivia es un país de alto riesgo político para las inversions [Bolivia is a country with high political risk for investments]. (2013, March 21). Retrieved June 17, 2015,
31
from http://hoybolivia.com/Noticia.php?IdNoticia=78466&tit=bolivia_es_un_pais_ de_alto_riesgo_politico_para_las_inversiones
Gorter, J., & Bikker, J. A. (2011). Investment risk taking by institutional investors. Insurance Europe. (2014). Why insurers differ from banks. Retrieved from http://www.
insuranceeurope.eu/uploads/Modules/Publications/why_insurers_differ_fr om_banks.pdf
Kacperczyk, M., & Schnabl, P. (2009). When safe proved risky: commercial paper during the financial crisis of 2007-2009 (No. w15538). National Bureau of Economic Research.
Manconi, A., Massa, M., & Yasuda, A. (2012). The role of institutional investors in propagating the crisis of 2007–2008. Journal of Financial Economics, 104(3), 491-518.
Organisation for Economic Cooperation and Development. Pension Markets in Focus 2014. Retrieved from http://www.oecd.org/daf/pensions/pensionmarkets
PWC Market Research Centre. (2015). [Interactive Map March 19, 2015]. European Institutional Investors. Retrieved from http://www.pwc.lu/en/press-releases/2015/a-bird-s-eye-view-of-european-institutional-investors.jhtml
Raddatz, C., & Schmukler, S. L. (2012). On the international transmission of shocks: Micro-evidence from mutual fund portfolios. Journal of International Economics, 88(2), 357-374.
Taking Europe’s pulse. (2015, May 7). The Economist. Retrieved from http://economist.com Tatom, J. (2011). Inflation and asset prices. Networks Financial Institute Working Paper,
(2011-WP), 26.
Torres-Reyna, O. Panel Data Analysis Fixed and Random Effects. (December 2007). Retrieved June 17, 2015 from http://www.princeton.edu/~otorres/Panel101.pdf Van Dijk, D. June 19th 2015, personal communication
32
Appendix
Table 6 Correlation matrix for independent and control variables
Political risk Regulatory risk Trade credit risk Consumer Price Index Gross Domestic Product Current Account Foreign Direct Investment Eurozone Political risk 1.0000 Regulatory risk -0.2427 1.0000 Trade credit risk -0.3396 -0.0401 1.0000 Consumer Price Index 0.0732 0.1792 0.1146 1.0000 Gross Domestic Product -0.3439 -0.3306 -0.3177 0.0191 1.0000 Current Account -0.0219 0.0811 -0.0260 0.0597 -0.0149 1.0000 Foreign Direct Investment -0.0313 0.0451 -0.0379 0.0151 -0.1281 0.3371 1.0000 Eurozone 0.0733 -0.0525 0.0965 0.0833 -0.3186 -0.0566 0.0041 1.0000 _cons 0.1659 0.0520 0.2742 -0.2932 -0.8770 -0.0019 0.1807 0.3297
The matrix shows the correlation between all the independent variables and control variables used in this model. It turns out that there are no significant correlation between variables. The highest correlation is between GDP and the risk ratings.
33
Figure 5 Non-normal distribution of residuals
This figure shows the distribution of the residuals, the distributions are not normally distributed. They are also not skewed. To correct for this, clustering will be used in the regressions.
0 .0 0 0 .2 5 0 .5 0 0 .7 5 1 .0 0 N o rm a l F [(r e si d -m) /s] 0.00 0.25 0.50 0.75 1.00
34
Table 7 All countryid numbers and their country
Countryid Country Countryid Country 1 Albania 37 Israel 2 Argentina 38 Italy 3 Australia 40 Jamaica 4 Austria 41 Japan 5 Bahamas 42 Kazakhstan 6 Bahrein 43 Latvia 7 Belarus 44 Lithuania 8 Belgium 45 Luxembourg 9 Bolivia 46 Malaysia 10 Brazil 47 Mexico 11 Bulgaria 48 Morocco 12 Canada 49 New Zealand 13 Chile 50 Norway 14 China 51 Pakistan 15 Colombia 52 Panama 16 Costa Rica 53 Peru 17 Croatia 54 Philippines 18 Cyprus 55 Poland 19 Czech Republic 56 Portugal 20 Denmark 58 Romania 21 Ecuador 59 Russia 22 Egypt 60 Serbia 23 Estonia 61 Singapore 24 Finland 62 Slovakia 25 France 63 Slovenia 26 Georgia 64 South-Africa 27 Germany 65 South-Korea 28 Ghana 66 Spain 29 Greece 67 Sweden 30 Hong Kong 68 Switzerland 31 Hungary 69 Thailand 32 Iceland 70 Tunisia 33 India 71 Turkey 34 Indonesia 72 Ukraine 35 Iraq 73 United Kingdom 36 Ireland 74 United States
35
Figure6 CPI in time period 2008 – 2013 for every country in dataset
This figure shows the consumer price index per country during the period 2008 – 2013. In general, the CPI is quite stable or has some small jumps. Belarus, however, had a very volatile CPI in this period which suggests hyperinflation. This can be explained by the devaluation in 2011 as a result of a gap in the Balance of Payment.
0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74
C
PI
year
Graphs by countryid36
Figure 7 CPI in time period 2008 – 2013 for all countries that are part of the Eurozone
This figure shows the consumer price index per country during the period 2008 – 2013 for all Eurozone. In general, the CPI has the same pattern.
-5 0 5 10 -5 0 5 10 -5 0 5 10 -5 0 5 10 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 2008 2010 2012 2014 4 8 18 19 23 24 25 27 29 36 38 45 56 62 63 66 C PI year Graphs by countryid
