1
The correlation between
the credit cycle and
the economic cycle
in the European Monetary
Union
Abstract:
Credit is a crucial factor in the economy of today. This thesis analyzes the
correlation between the credit cycle and the economic cycle in the EMU. An
meta-analysis was conducted and the results were that there is a significant
correlation between the credit cycle and the economic cycle for some
countries. The correlation is stronger in South Europe than in North Europe.
Furthermore, the correlation in the EMU as a whole is positive and significant,
while for some countries a negative significant correlation is found.
Name: Jouke Jagersma
Student number: 10352767
Bachelor program: Economie & Bedrijfskunde
Specialization: Financiering & Economie
Supervisor: Egle Jakucionyte
Date: 30-01-2017
2
Statement of Originality
This document is written by Jouke Jagersma 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 source other than those mentioned in the text and it 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 Table of contents
1. Introduction ... 4
2. Literature review ... 4
2.1 What is a macroeconomic cycle? ... 4
2.2 What is a credit cycle? ... 6
2.3 Views on the role of debt and credit in relation to the real economy ... 8
2.4 Hypotheses ... 11 3. Methodology ... 11 4. Results ... 12 4.1 Data ... 12 5. Conclusion ... 17 6. References ... 19
4 1. Introduction
Since the Sumer civilization in South Iraq developed the idea of issuing debt and register borrowers and lenders around 3000 BC, the world is involved with debt. In the early days of debt system, farmers became so indebted that their own children were forced to become slaves to repay the debt (Greaber, 2014). People struggled since than with the role of debts and credits in the society. Everybody wants easy access to loans, but nobody wants to have more debt than they can repay. Central banks were founded and found their role in the economic landscape the last century. Fisher (1933) laid the foundation under economic science with his analysis on over-indebtedness.
Bernanke and Friedman wrote also great papers with a contribution on credit science. This paper tries to contribute on credit science. Recent empirical work shows that there is interaction between credit and the real economy. Cristiano et al. (2010) describes the effect of financial shocks to the real economy. There is also a debate in Europe about the
effectiveness of monetary policy, since the inflation is now very low for years. Monetary policy has effect on the credit market. This paper will focus on the correlation between the credit cycle and the real cycle in the Eurozone, to contribute to the discussion how large the dependence is on credit in the Eurozone. The main research question of this paper will be: what is the correlation of credit cycles and real cycles in the euro zone and the correlation change over time in the euro zone? First, an analysis of historic literature on credit and real cycles will be given. Second, three hypotheses are formed. As last, an overview of the found results, and accordingly the hypotheses will be discussed and analyzed. This paper tries to contribute to the empirical literature by using a very recent dataset and a discussion about the differences within the Euro zone.
2. Literature review
This literate review consists of three parts. In the first part the theory behind a macroeconomic cycle and its fluctuations will be explained. In the second part of the review, the existence and variables of the credit cycle will be discussed. In the third part, the interaction between a credit and a real business cycle will be reviewed.
2.1 What is a macroeconomic cycle?
In this paragraph the theory behind real macroeconomic cycles will be discussed. A general overview of macroeconomic cycles will be given and the factors that influence such cycle.
Modern economies have significant short-run fluctuations in aggregate output,
consumption and employment. One important result from research on economic cycles is that these economic fluctuations do not display a regular pattern (Romer, 2012). A
recession, in the Eurozone defined as two consecutive quarters with a negative GDP growth, is therefore hard to predict. Romer states that the prevailing view among economists is that the real economy is disturbed by various factors of the economy, such as consumption or government purchases. These components of the economy then generate at a more or less random interval negative shocks to the real economy. These shocks create a cycle around a
5
positive upwards path of the economy. Romer uses unemployment and GDP as the two most important variables to measure the real cycle. This cycle is mostly measured with GDP. The economic cycle switch between GDP growth and a recession, where GDP growth is negative. It is difficult to know in which part of the cycle the current economy is. There are many models that tries to exclude the trend part of economic growth to isolate the cyclical part of the economy. One of the suggestions of Romer is to find the deviation of a long term constructed trend line.
Another important finding of Romer is that since the Second World War, fluctuations in the United States are unevenly distributed over the components of real output. This means that some components, for example inventory investment, are not very meaningful to normal GDP growth. Only 0.6% of the GDP growth is explained by inventory investment, while the average share in fall in GDP in recessions relative to normal growth is 44.8%. Inventory investment has large fluctuations, because it has a large downward potential according to data research of Romer. The consumption of durable goods, business
investment and residential investment are also accountable for disproportionate shares of output fluctuation compared to a normal state of the economy. The third result of Romer’s data analysis is that output growth is distributed roughly around its mean. However, the output illustrates an asymmetry, as output is relative a long time above its usual path with short interruptions where output is relatively far below its growth path. The major
macroeconomic schools, the Keynesian and the Austrian, agree on these findings of Romer, but have different in the hypotheses concerning the shocks and propagation mechanisms on the real cycle.
Solow (1957) was one of the first economic scientists who developed an aggregate production function that can generate time series with the same complex volatility as a real economy. Solow found that the growth of output per person can mostly be attributed to technological progress, as from 1910 to 1948 technological shocks were accountable for 87,5% of the growth of output per person. According to Solow, the growth of the use of capital in the economy is not as important for the growth of the economy. King and Rebello (1993) make a large contribution to the real business cycle research. The first real business cycle theories, such as the Solow growth model, argue that technological shocks are the most important source for fluctuations in the real business cycle. They tested a real business cycle model, developed from Solow, on the economic data from 1940 to 1990. One finding is that technological development must be very volatile to produce a real business cycle. Their second finding is that the elasticity of labor, which means the responsiveness of demand to labor when there is a change in the market wage rate, is a large explanatory variable for fluctuations in the real business cycle instead of only technological factors as Solow suggests.
Galí (1999) agree with King and Rebello (1993) that technological progress as most important factor behind real business cycle fluctuations was overestimated by Solow. Galí found that there is a negative response of worked hours to a positive technology shock. This correlation combined with the Solow model, can result in a lower impact of technology shocks on the real business cycle because the role of capital increases when worked hours
6
decline. The non-technological shocks, such as monetary shocks, are the most important explanatory factor for economic shocks. The estimates of Galí (1999) point to a pattern of positive co-movement of GDP and hours associated with nontechnology shocks, with an estimated correlation of 0.97. The theory which underlines this finding is that technological shocks have a permanent impact on the real business cycle, while the nontechnology shocks have a transitory impact on the economy. Therefore, monetary shocks can be a good
explanatory variable for fluctuations in the real business cycle.
Conclusively, real macroeconomic cycles are influenced by technological and non-technological factors. To measure this cycle, GDP and unemployment is common. There is however debate between major economic schools how mechanisms, such as shocks from consumption or monetary policy react to the macroeconomic cycle and what creates conjuncture. Gali (1993) also indicates that monetary shocks are a large contribution to some fluctuations of the real business cycle.
2.2 What is a credit cycle?
In this paragraph the credit cycle will be discussed and the components that contribute to shocks of the credit cycle. Also, important papers about the changing role of credit in the modern economy will be discussed.
There is a large role of credit as variable in the economy, as Schularick and Taylor (2012) states. Their conclusion is the result of a quantitative analysis on a dataset for 12 developed countries over the years 1870-2008. They describe how we are now living in the “age of credit”, because financial innovation made credit less dependent on monetary aggregates, such as money supply. Monetary aggregates are broad categories that measure money supply. The Central Bank has a lot of control of the money supply. In line with their
conclusion, they found large growth in bank loans relative to broad money since 1950 and especially since 1970. The explanation behind this finding was that after 1700, a wave of financial liberalization arose that made access to debt instruments easier. Those new debt instruments lead to more non-monetary bank liabilities. The looser link between monetary aggregates led to an expansion of the role of credit in the macroeconomic cycle in the post-1945 era. Past credit growth was in their regression the best predictor of financial crises. Therefore is the dependence of credit on money supply less than before 1945.
The monetary policy of the FED that gave more incentive to debt finance companies with debt until it resulted in a widespread crisis (Bordo, 2008). Bordo (2008) wrote an historical perspective on the crisis of 2007-2008. The last crisis which was triggered by events in the financial sector was comparable with the Great Depression. The crunch of credit triggered both crises. A difference with the previous credit crisis is that over the last decades, there were more important changes to the financial system and they led to a greater role of credit in the macroeconomic world.
A credit cycle is a self-reinforcing interaction between risk and value of credit. In times of a high cyclical conjuncture, the supply of credit is high because risks are regarded low
(Bordo,2014). When risks of credit are assumed low, the value of credit is high. This interaction between risk and value is everlasting relationship. In a credit crisis, value
7
disappears because the risks are assumed too high for the current value and therefore values will be adjusted accordingly. The credit cycle is independent if there a no other macroeconomic variables that has influence on this interaction between risk and value. Valgreen (2014) investigated the independency of the credit cycle and underlined the view that an independent credit cycle doesn’t exist, but that the credit conditions can be seen as shocks to real cycles instead of a cycle itself. The credit conditions that are used as example are the price of lending, transaction costs and accessibility of credit. The argumentation is that still nobody has found a debt/income variable that consistently affects future cyclical swings. Credit shocks to the economic cycle are for example monetary policy and credit conditions.
Aikman, Haldane and Nelson (2013) wrote a paper where they sketched a credit cycle. They found that across industrialized countries credit cycles can be measured by variation in the ratio of bank lending to GDP. This variable is easy identifiable and regular for a large period of time. They found that for the USA real GDP growth did not differ as much as the mean growth rate of bank loans between 1880 and 2008. This can be explained by a larger amplitude of the credit cycle. A credit cycle is independent when it is not influenced by variables of the real cycle. They found however a direct link of the growth in the credit-to-GDP ratio on the chance of a banking crisis. A one percentage point increase in the growth rate of credit to GDP for one year resulted on average with an increase in the probability of a banking crisis of 0.18.
The importance of analyzing the credit cycle and its effect is illustrated by Aikman, Haldane and Nelson from the fact that credit cycles are more synchronized than at the beginning of the century. Cross-country credit cycle correlations after 1979 are twice as high than between 1945-1979. However, the absolute level of correlation is still low despite the globalization of the financial economy since 1979. The growth of correlation between credit cycles will lead to less risk differences between economies, while the national central banks have different policies. Without coordination between central banks, arbitrage
opportunities, based on the different policies but correlated risk, will exist. The correlated risk exists because the risk of country based credit cycles will influence one other. The correlation between risk should cause correlation in value as (Bordo,2014) suggest. If policy however differs, values will differ more between countries than only risk based value. Therefore investigating the credit supply is not only interesting for macro-economic policy, but also for the commercial financial institutions.
Kohlmann and Zeugner (2012) documented in their paper that there is a significant negative link between leverage and the volatility of future real activity. More leverage means that an economy is more indebted. The relationship between volatility and leverage is that when leverage is high, an economy is more indebted, so the value adjusting shock, described by Fisher(1933), is more intense in terms of GDP. However, leverage information would not have predicted the financial recession of 2008 better than other conventional macroeconomic predictors. It agrees with Schularick and Taylor (2012) that a more indebted economy will be more volatile.
8
Financial recessions are more harmful to the GDP and other macroeconomic variables according to the research of Jorda, Schularick and Taylor (2011). Even after the recession, the economy had a slower growth rate, less investment spending and less credit growth than usual, especially in the United States. They state that a credit build-up in an economic boom heighten the vulnerability of an economy. Credit can have a large imploding effect on the real economy, if credit supply tightens. This results in lower investment which results in lower business activity. The lower business activity has again effect the credit supply, so therefore credit has a high effect on the vulnerability of the real economy.
De Gregorio and Guidotti (1995) wrote a paper on the influence of financial development on economic growth. They tested if the GDP per capita growth was positively correlated with the domestic credit to private sector ratio for Latin-American countries and high-income countries. They found that the per capita real output growth is positively correlated with the amount of credit supplied.
Conclusively, there is some debate about the independency of the credit cycle. There is consensus about the fact that more indebted economy, more credit compared to GDP, leads to a more volatile real economy.
2.3 Views on the role of debt and credit in relation to the real economy
This section will discuss some papers which reviewed the role of credit and debt on the real economy. As first, one of the founders of debt research will be discussed, Fisher. Than will some recent papers concerning the interaction between credit and GDP/recessions be examined.
Fisher (1933) is the first important article about the role of the financial markets in the real economy. He states that capital items in an economy, such as homes, gold, money, credits and debits are an essential variable in a macro-economic cycle. Such capital items aren’t always in equilibrium. He declines the theorem of a general equilibrium. The general equilibrium theorem explains that the behavior of supply, demand, and prices in a whole economy with many markets who interact. These interactions has a the result according to the theory that the interaction between supply and demand will result in an overall general equlibrium. If there is investment and speculation with borrowed money, over-indebtedness is the result according to Fisher. Over-over-indebtedness, the situation where there is too much debt relative to national wealth and income, is always the beginning of a
recession according to Fisher (1933). This was his conclusion after analyzing the economic history of the United States until the start of the Great Depression.
The economy had a great debt problem during the Great Depression. A debt problem occurs when there isn’t enough investment profit to pay the interest on existing loans, which happened back then. This is problematic because investment has reached the point where it is not profitable anymore. Loans cannot be paid off, so banks will come in danger and companies will have less access to debt finance. The markets boomed before the Great Depression and the consequence was that everybody invested with borrowed money. That wasn’t a problem until the investment profit dropped under the interest rate. The result of an economy with a debt problem is distress selling, which causes lower deposits at the bank
9
and less money in circulation. The consequence of this situation is a decline in the price level. This decline affects the net worth of companies which leads to fewer employees hired and less production inputs hired. The chain reaction of Fisher brings us back at a lower point of price level because lower output results in less money in circulation. Concluding, price level can always fall lower and lower if there is no offsetting cause in a crisis. He called this: “There is then no tendency of the boat to stop tipping until it has capsized” (Fisher, 1933). The new macro-economic cycle of boom and recession will start after a universal liquidation of all the debts. This is the natural way of the economy described by Fisher. He conclude, that it can always be an option to prevent an economic crisis by deflating the price level to the point where outstanding debts are contracted by the borrower and expected by lenders. He meant with this solution that the control of the price level is essential for an economy and that it can be achieved through monetary instruments. Fisher underlined the role of the investment market and what damage a flood of credit supply can do to an economy. This argumentation of Fisher must lead to simple recession preventing policy.
There are two important views on the structure and dynamics of money, credit and the influence on the macroeconomic cycle (Freixas and Rochet, 1997). One, for simplicity called the “money view”, is strongly influenced by “A monetary history of the United States, written by Friedman and Schwartz (1963). Until 1963, the consensus between economists was that monetary policy was very ineffective. The only job of central bankers back then was to keep interest rates low when unemployment was high. The most economists didn’t think that central banking policy was the solution for lowering inflation rates.
Friedman and Swartz (1963) did a lot of case studies on monetary actions in times of recession. The first one is where they criticized the role of the FED before the Great Depression. In the beginning of 1928, the FED raised the federal funds rate to make the economy more stable. Friedman and Swartz ( 1963) called this the “Great Contraction”. The result of a higher federal funds rate is lower money supply because banks have to pay more for lending money out to each other. The banks will as a result ask a higher interest rate to consumers, so the demand for credit will be lower. Lower demand for credit will result in a lower stock of money in the circulation. This lower stock of money drops investments and therefore will it harm the real economic cycle. Friedman stated that this is the reason that the market crashed in 1929 and the following Great Depression.
Bernanke is the most famous recent adept of the credit view. He carried out this view while he was president of the Federal Reserve. Bernanke (1983) writes that there is a paradox concerning the financial crisis of 1930, namely that the financial collapse had real effects on the macroeconomic cycle beside the effects through monetary channels. The Great Depression cannot be explained with monetary actions because the Great Depression was too long and too deep to have only monetary explanations . His conclusion is that financial institutions and their supply of credit must be recognized, because they can
influence costs of transactions and thus the market allocations. These market allocations can be harmed due to less investment. His advice is that financial institutions must evolve, because well-performing institutions can offset policy mistakes or external shocks.
10
Bernanke, Gertler, Gilchrist (1993) makes a model in line with the view of Bernanke (1983). They criticize the real business cycle model of 1993, the Keynesian view and the view of Modigliani-Miller (1958) on the real economy, because they assume that conditions in credit and financial markets do not influence the real economy. They included in their model money and price stickiness. The money supply of an economy is controlled by the central bank and price stickiness is a logical assumption because the price is never completely flexible in the credit market due to contracts. Money and price stickiness influences monetary transmission channels and that influences the credit market a lot. The financial accelerator, an endogenous development of the financial market, can affect the real economy.
Christiano, Motto and Rostagno (2010) provides a macroeconomic model where they clearly reference back to Fisher and Bernanke. This model helps to study the capital and financial side of the whole economy. They included asymmetric information and agency problems in financial contracts, shocks to the market risk of capital, bank funding conditions and more attention to central bank liquidity instead of market liquidity if private credit is not a possibility anymore. These are all financial factors which affects the credit market. They found that financial risk shock in their model accounted for one third of the volatility of investment in the Eurozone. It was also tested relative to GDP, in the Eurozone they found that the financial risk shock is responsible for 35% of the fluctuations of the GDP cycle.
Samarina, Zhang and Bezemer (2016) investigates the cross-country coherence of credit cycles in the Eurozone. The credit cycles of the Eurozone countries fluctuated often in the time period between 1990-2013. A measurement of the credit cycle was total bank credit or non-financial business loans. The credit cycles showed more synchronicity because the currency risk was eliminated between the countries. They found that the lower currency risk results in more synchronization between the real business cycle and the credit cycle. In addition, they state that more trade openness causes divergence in cycles and financial deregulation contribute to more synchronization between the real business cycle and the credit cycle within countries. The harmonization of the European credit cycles has the consequence that a credit problem in Italy will harm other countries in the Eurozone more than before the introduction of the Euro. This underlines the view that credit cycles, even a foreign credit cycle can have a large impact on the domestic credit cycle.
Bruche and González-Aguado (2010) found in an empirical analysis that different phases of the business cycle and the credit cycle are evident in recovery rates. Credit downturns seem only imperfectly aligned with recessions. They stated that credit supply has influence on the macroeconomic cycle, but they are not significantly related according to their model. This is in contrast with Cristiano et al. (2010), but the difference is that Bruche and González-Aguado (2010) focuses in their research on recovery rates, not on fluctuations of GDP.
Borio and Lowe (2004) found evidence that credit is so important for the macro economy that central banks must consider credit when setting their monetary policy. They found an easy predictor of banking crises. The two main variables are that the credit volume to GDP must exceed its long-term trend at the same time when inflation-adjusted equity prices
11
exceed their long-term trend. The reasoning behind is that expansion in credit relative to GDP does not have to be harmful, if the asset prices grew at a sustainable rate. If credit and assets grow at a sustainable rate, a financial imbalance will not occur. A financial imbalance occurs when debts are larger than assets, so if risks are not priced in correctly. Where this paper also refers to Fisher (1934). The likelihood of a disruption increases greatly if the two variables exceed their long-term trend at the same time.
Conclusively, credit has influence on the economic cycle according to the reviewed literature. Fisher (1933) explains that the indebtedness of an economy has influence on the macroeconomic cycle. This is confirmed by Bernanke et al. (1993) and Cristiano et al. (2010), both papers found a significant influence of credit on the economic cycle. The interaction between risk and value of credit is under influence of economic variables.
2.4 Hypotheses
Following from the literature discussed above, three hypotheses are formed. The first hypothesis (1) is that the credit cycle is highly correlated with the economic cycle. This follows from the findings of Bernanke, Gertler, Gilchrist (1993) and Schularick and Taylor (2012). These papers state that shocks the credit market, such as bank lending, affect the economic cycle. The second hypothesis (2) is that the credit cycle is now increasingly more correlated with the economic cycle than in the past. The third hypothesis (3) is that the credit cycle and economic cycle is more correlated in north-Europe than in the south of Europe because more financial developed countries are expected to have more correlation between credit supply and GDP as Aikman, Haldane and Nelson (2013) suggests. South Europe had also a bigger recession due to governmental deficits and therefore Hall (2012) suggests that the credit market of south Europe is disturbed due to their response to the credit crisis and euro crisis.
3. Methodology
First, a selection of countries will be made from the Eurozone. The first twelve countries that adopted the euro as first is a logical starting point. From the twelve countries, two countries are left out of the data in this paper, Luxemburg and Ireland. Ireland and
Luxemburg have less data available than the other countries for the selected time period, which is 1967 until 2016. The countries involved in this paper are Greece, Germany, the Netherlands, Portugal, Austria, Italy, France, Belgium, Finland and Spain. Furthermore, the euro area as a whole is included in the analysis.
As correlation coefficient, the Pearson coefficient is used in this paper. It will be two-tailed tested against an alternative hypothesis of 0, which means no correlation between the variables used. This test for the correlation coefficient will be done after a Hodrick-Prescott filter is used to remove the cyclical component of a time series data from raw data and isolate the cyclical component for the analysis. For this Hodrick-Prescott filter, a noise-ratio of 100 will be used, which is normal for yearly data (McElroy, 2008).
To test whether the credit volumes are highly correlated with the economic cycle,
12
GDP. This ratio is also used in Basel III to advise if countercyclical actions must be taken by policymakers (Drehmann and Tsatsaronis, 2014). The variable is also a good early warning indicator for a banking crisis (Borio and Lowe, 2004). The domestic credit to private sector by banks is defined by the World Bank as financial resources provided to the private sector by other depository corporations (deposit taking corporations except central banks), such as through loans, purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment. To measure the real macroeconomic cycle, time series of the GDP adjusted for inflation will be downloaded from the World Bank too. Here the GDP is defined as the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. The GDP at market prices is all
computed in current $ to make comparison possible.
For testing the first hypothesis, if the credit cycle is highly correlated with the economic cycle, a Hodrick-Prescott filter will be used filter the trend out of the data. The correlation will be calculated between the GDP growth rate and the credit as percentage of GDP. To test the second hypothesis, that the credit cycle is more correlated with the real cycle now than before, again a Hodrick-Prescott filter will be used and for two time ranges, 1967-1997 and 2001-2015 correlations will be calculated and compared. The first time range begins with the financial deregulation and ends before the introduction of the euro. 1998-2000 is left out of this paper because the credit data is not available for all the Eurozone countries. The second time range will give the correlations during the presence of the Euro era.
For the third hypothesis, countries will be grouped. The partition of the countries will be the same as Hall (2012) use. For north-Europe, Germany, Belgium, Austria, The Netherlands, Finland and France will be grouped. For south-Europe, Greece, Italy, Spain and Portugal will be grouped. For a new credit divided by GDP variable, a weighted average is taken. The weighing factor is the GDP of the country. The test range will be the same as for separate countries, 1967-2015. To analyze what the influence of the credit crisis and the Euro crisis was, will the years after 2007 be eliminated. The intuition behind this is there was always a recession in at least one of the countries involved since the begin of 2008, according to the data of the World Bank.
4. Results
The next section will give a complete overview of the used data and the found results after analysis of the data.
4.1 Data
Table 1 gives the descriptive statistics of the credit to GDP ratio as percentage of GDP for the time range between 1967 and 2015. Only Greece has data for all selected countries. For other countries the years (1998) , 1999 and 2000 are missing. For Germany the years 1967 and 1968 are also missing. Spain and Portugal has the highest standard deviations, but Greece and the Netherlands have also high standard deviations. The three highest standard
13
deviations are all from South European countries. Remarkable is that Italy, the other South European Country, has the lowest standard deviation in credit to GDP ratio of all the analyzed countries.
Table 1
Country Obs. Mean. Std. dev. Min Max EMU 46 74.49 18.76 38.88 106.15 ESP 47 92.55 37.29 44.85 172.41 AUT 46 77.60 18.46 41.50 100.18 PRT 47 85.35 37.42 44.59 159.83 ITA 47 65.09 14.36 47.09 94.71 GRC 49 47.31 31.68 15.45 118.11 NLD 46 76.21 31.54 29.44 118.62 BEL 46 43.15 20.69 16.20 75.05 DEU 44 85.12 15.31 59.60 113.29 FRA 46 72.42 23.82 28.59 96.78 FIN 47 62.93 18.46 39.61 95.45 Graph 1 visualizes the growth of the GDP through time of the selected countries without the Hodrick-Prescott filter. The crises of 2007 and the euro crisis is easy to identify. The credit crisis is the first fall in GDP after a long rise since 2000. Graph 2 visualizes the volume of credit compared to the GDP of the selected countries, also without the Hodrick-Prescott filter. The credit volume is especially large in Spain, Portugal, Greece and the Netherlands compared to the real economy.
14
With the HP filter used on the data, the difference between the data and the trend line is revealed. Figure 3 shows for Greece, Italy, the Netherlands and Germany the difference from the trend line for the credit to GDP ratio. The high peak of Greece at the end of the credit crisis is remarkable, the credit to GDP ratio is in 2011 10% above its trend line, while it is in 2008 far under its trend. One more observation is that the volatility of the German credit cycle is low compared to other countries.
4.2 Analysis Results
In this section the results of the correlation analysis will be discussed. The data are prior to analysis transformed with a Hodrick-Prescott filter as described.
Table 2 Country Correlation 1967-2015 Correlation with Credit/GDPt-1 Correlation GDPt-1 EMU 0.257* 0.455** 0.088 ESP -0.566** -0.329** -0.208 AUT -0.374** -0.419** -0.201 PRT -0.472** -0.247* -0.123 ITA -0.372** -0.137 -0.088 GRC -0.250* -0.131 -0.142 NLD -0.142 -0.158 -0.058 Figure 3
15 BEL 0.102 0.136 -0.040 DEU 0.054 0.038 -0.023 FRA -0.061 -0.070 -0.033 FIN -0.038 0.004 -0.021 Note. **=p≤.05 *=p.≤.10.
Table 2 represents the correlations between credit divided by GDP and GDP between 1967 and 2015. There are six significant correlations with a p-value below 10% if there is no variable lagged. The significant correlations are the European Monetary Union as a whole, Spain, Portugal, Italy, Greece and Austria. The first hypothesis, that there is a correlation between the credit cycle and the economic cycle is not accepted for all countries. Spain, Portugal, Italy, Greece and Austria has a negative correlation, which means that the credit cycle moves opposite to the macroeconomic cycle for these countries. Noteworthy is that all the South European countries have a significant negative correlation between the credit cycle and the economic cycle.
In the third column, there are the correlation results if the credit/GDP cyclical movement is lagged with one year. It gives slightly different results as in the first column, except that the correlation for the European Monetary Union is now more significant with a stronger linear relationship between credit/GDP and GDP and that the correlation for Greece and Italy is not significant anymore. In the last column, the results are given if the cyclical movements of GDP are lagged with one year. All the correlations are non-significant, so there is no relationship between the GDP of the year before and the credit/GDP ratio between 1967-2015. Table 3 Country/Area Correlation 1967-1998 Correlation 2001-2015 Correlation 2008-2015 EMU 0.317* 0.204 -0.464 AUT 0.214 -0.607** -0.958** FIN 0.218 -0.673** -0.957** NLD 0.170 -0.363 -0.710* BEL 0.205 0.285 0.921** DEU -0.140 0.049 0.089 FRA 0.304* -0.467* -0.937** ESP -0.187 -0.638** -0.972** PRT -0.298* -0.613** -0.986** ITA -0.340* -0.619** -0.930** GRC -0.126 -0.213 -0.649 Note. **=p≤.05 *=p.≤.1
Table 3 represents the correlations between credit divided by GDP and the GDP for three periods, 1967-1998, 2001-2015 and 2008-2015. Only four found correlations for 1967-1998 are significant, the European monetary Union, Portugal, France and Italy. Remarkable is that if the Euro years are excluded, the correlations for Spain and Austria became non-significant. Also, that the correlation for France and Italy are significant until the introduction of the
16
euro in those countries. This implies that during the euro era, the relationship of the credit cycle is disturbed for these countries, while for Spain and Austria is disturbance takes place before the euro era.
For the second time period, the euro era, five correlations are significant with a p-value < 5%. The correlations of the Austria, Finland, Spain, Portugal and Italy between credit and GDP are significant and positive. The second hypothesis is not valid for all countries, except for the previous mentioned countries. All four South European countries have stronger relationships between the credit cycle and the GDP cycle in the Euro era than before, but Greece has in both time ranges not a significant linear relationship. The results of the correlation analysis is in line with the reviewed literature. Aikman, Haldane and Nelson (2012) conclude that a more financial deregulated system and more globalization leads to a stronger connection between credit and GDP, because credit plays now a greater role in the economy and therefore affects the macroeconomic cycle now stronger than before. For many countries however, there are still no significant results. The second selected time period is with N=14 can be too small, which influences the found correlations.
In the last column, the correlations are given for the crisis years in Europe. Belgium and has a strong positive and significant correlation between the credit cycle and the economic cycle. While for many countries, such as Austria, Finland, France, Spain, Portugal, the Netherlands and Italy a very strong negative correlation exists for the European crisis years. This implies that the credit cycle move opposite to the economic cycle, but in Belgium the correlation shows co-movement. Figure 4 and 5 illustrates the difference between the credit movement in Belgium and in Portugal. The credit cycle raised compared to its normal path, while the credit cyclical movement of Portugal came below the trend line of the credit cycle.
Figure 5 Figure 4
17 Table 4 Country/Area Correlation 1967-2015 Correlation 1967-2007 Correlation 2001-2015 North Europe 0.136 0.192 -0.188 South Europe 0.416** 0.626** 0.388 Note. **=p≤.05 *=p.≤.10
Table 4 represents the correlations between credit volume divided by GDP and GDP for the grouped countries. North Europe exists of Germany, France, the Netherlands, Finland, Austria and Belgium. South Europe consists of Greece, Spain, Portugal and Italy. The correlation between the credit cycle and the real cycle is always stronger in South Europe, which means that the third hypothesis is not accepted if the data range is set from 1967 to 2015. One possible explanation can be that Germany influences the data from North-Europe strongly, because the credit divided by GDP ratio in Germany drops from 113% in 1997 to 77% in 2015.
One remarkable result is that when the euro crisis and capital crisis from 2008 until 2015 is left out of the data, North Europe and South Europe has a stronger correlation. This means that between 2007 and 2015 the relation between credit/GDP and GDP was weaker in Europe than before 2007. An interesting suggestion for further research is what the response to the crisis years did to the correlation between credit and GDP . South Europe became more indebted while the economy was in a recession. In North Europe, the difference is smaller because they lacked a strong divergent movement of credit/GDP and GDP. In the last column, the correlations are given for the Euro era. Both correlations are not significant, but a difference in correlation is slightly visible between the two groups. This can implicate that credit policy has more effect on South Europe than in North Europe.
5. Conclusion
This paper investigates the correlation between real cycles and credit cycles in the Euro zone. First, the insights that economic theory and empirical research provides are discussed. In the found literature, some discussion exists about how large the role of credit is on the real macroeconomic cycle and what the relationship is. This paper tries to contribute to the discussion how large the role of credit is. To investigate the correlation between credit cycles and real cycles, the variables credit divided by GDP is used to measure the credit cycle. GDP is used to measure to real cycle. A Hodrick-Prescott filter is used to construct cycles from the data. The analysis shows that there is a significant correlation between credit cycles and real cycles for the EMU, Spain, Portugal and Austria between 1967 and 2015. Only the European Monetary Union has a positive correlation between the credit cycle and the economic cycle. If the GDP is one year lagged, the analysis provides only non-significant results, while if the credit/GDP is one year lagged, the EMU, Spain, Portugal and Austria still have a significant correlation.
For the time range between 1967 and 1997, significant results are found for EMU, Portugal, France and Italy. If the European crisis years are analyzed, there are many
18
significant correlations while only Belgium has a positive correlation. Most of the countries has a strong (<-0.7) negative relationship. Those countries are the Netherlands, Austria, Finland, Spain, Portugal, France and Italy. In Belgium, the credit cyclical movement
commoved with the GDP cycle upwards after the credit crisis. One remarkable result of the analysis is when the data is grouped to North Europe and South Europe. For South Europe, the correlation is higher than in North Europe, but the South European correlation rises a lot when the years after 2007 are excluded. South Europe experienced weak co-movement of the credit cycle and the economic cycle during the crisis years in Europe.
For future research, explanations for the shift in correlations between 2007 and 2015 is worth investigating. The change in correlations of North and South Europe during the crisis years in Europe is an interesting result. Yearly data is a shortcoming of this paper, when quarterly data is made available, more research can be done with the data. An analysis of the influence of quantitative easing on the correlation between real cycles and credit cycles will be also a fascinating research, because in North Europe the correlation between the credit cycle and economic cycle is weak. This can imply that credit policy has less effect on the North European economy than in South Europe.
19 6. References
Aikman, D., Haldane A. G. & Nelson B. (2013). Curbing the credit cycle. Economic Journal. Vol 125, pp. 1072-1109
Bernanke, B. S. (1983). Nonmonetary Effects of the Financial Crisis in the Propagation of the Great Depression. The American Economic Review. Vol. 73, No. 3, pp. 257-276.
Bernanke, B. S., Gertler, M., & Gilchrist, S. (1999). The financial accelerator in a quantitative business cycle framework. Handbook of macroeconomics, 1, 1341-1393.
Bordo, Michael D. (2008). “An Historical Perspective on the Crisis of 2007-2008,” NBER Working Paper Series 14569. Cambridge, Mass.: National Bureau of Economic Research, December.
Borio, C. (2014). The financial cycle and macroeconomics: What have we learnt?. Journal of Banking & Finance, 45, 182-198.
Borio, C. E., & Lowe, P. W. (2004). Securing sustainable price stability: should credit come back from the wilderness?.
Bruche, M., & Gonzalez-Aguado, C. (2010). Recovery rates, default probabilities, and the credit cycle. Journal of Banking & Finance, 34(4), 754-764.
Christiano, L. J., Motto, R., & Rostagno, M. (2010). Financial factors in economic fluctuations. De Gregorio, J. & Guidotti, P., E. (1995). Financial Development and Economic Growth. World development. Vol. 23, No.3, pp.433-448.
Drehmann, M., & Tsatsaronis, K. (2014). The credit-to-GDP gap and countercyclical capital buffers: questions and answers. BIS Quarterly Review March.
Fisher, I. (1933). The Debt-Deflation Theory of Great Depressions. Econometrics. Vol. 1, No. 4, pp 337-357
Freixas, X. & Rochet, J. (1997). Micro Economics of Banking. The MIT Press.
Friedman, M. & Schwartz, A. (1963). A Monetary History of the United States, 1867-1960. Princeton University Press.
Gali, J. (2000). The return of the Phillips curve and other recent developments in business cycle theory. Spanish Economic Review, 2(1), 1-10.
Graeber, D. (2014). Debt-updated and expanded: the first 5,000 years. Melville House Hall, P. A. (2012). The economics and politics of the euro crisis. German Politics, 21(4), 355-371.
King, R. G., & Rebelo, S. T. (1999). Resuscitating real business cycles. Handbook of
macroeconomics, 1, 927-1007.
Kohlmann, R. & Zeugner, S. (2012). Leverage as a predictor for real activity and volatility. Journal of Economic Dynamics and Control. Vol 36, No. 8, pp. 1267–1283
20 McElroy, T. (2008). Exact formulas for the Hodrick‐Prescott filter. The Econometrics
Journal, 11(1), 209-217.
Modigliani, F. & Miller, M. H., (1958). The Cost of Capital, corporation Finance and the theory of Investment. The American Economic Review. Vol. 48, No. 3, pp. 261-297.
Romer, D. (2012). Advanced Macroeconomics. Mcgraw-Hill Irwin, University of California. Samarina, A., Zhang, L., & Bezemer, D. (2016) Mortgages and Credit Cycle Divergence in Eurozone Economies.
Schularick, M. & Taylor, A. M. (2012). Credit Booms Gone Bust: Monetary Policy, Leverage Cycles and Financial Crises, 1870–2008. American Economic Review . Vol. 102, No.2., pp. 1029-1061.
Solow, R. M. (1957). Technical change and the aggregate production function. The Review of
Economics and Statistics. Vol. 39, No. 3, pp. 312-320.
