University of Groningen Faculty of Economics and Business MSc International Economics and Business
Master’s Thesis
The Influence Of Debt On Consumption and Investments:
A Post-Crisis Eurozone Perspective
Abstract
This paper empirically estimates the effect of debt growth on consumption and investments. It ventures to estimate the different effects of public and private debt growth in the years before and after the financial crisis. The empirical analysis is based on panel data from 19 Eurozone countries in the period 1990-2015. This paper conducts a dynamic panel data that is estimated using a system GMM estimator. The short-term as well as the long-term is considered. The findings suggest that there is a non-linear relation between public and private debt and consumption in the short-term. A weaker non-linear effect is found for the relationship of public debt growth and consumption in the years after the crisis. Regarding investments, a non-linear relationship with public debt growth is found in the long-term analysis.
Keywords: Private debt, Public debt, Financial Crisis, Consumption, Investments
Author: Rick Westland
Contact: R.westland@student.rug.nl Student number: S2418916
Date: 10/01/2017
2
Table of Contents
1. Introduction
...32. Trends in Debt and Economic Growth
...53. Literature Review
... 83.1 Consumption ...8
3.2 Investments ... 10
3.3 Interlinkage between Government and Private Debt...12
4. Data and Methods
...144.1 Dependent and Independent variables ... 15
4.1.2 Control Variables ... 16
4.2 Research Model...18
4.3 Robustness ... 17
4.3.1 Robustness of the panel...17
4.3.1 Robustness of the GMM Estimator...18
5. Results...
...195.1 Preliminary results ... ……….... ….. 19
5.2 Estimation results ... 20
5.2.1 The short-term effects of debt on consumption...20
5.2.1 The short-term effects of debt on Investments...23
5.2.3 The Effects of Debt on Consumption and Investment Pre- and Post-Crisis...23
5.2.4 Long-Term Effects of Debt on Consumption and Investments...24
6. Discussion and Limitations...
... 247. Conclusion...
...26References...
...283
1. Introduction
Financial crises are caused by excesses – mostly monetary - which lead to a boom and an inevitable bust (Taylor, 2009). For instance, the recent financial crisis was preceded by a global housing boom due to the build up of excessive amounts of private credit. In the first quarter of 2006 the housing market turned which resulted in a burst of the housing bubble (Acharya et al., 2009). The high levels of private credit are a major cause of the economic crisis, but what are its implications in the aftermath of the financial crisis?
The relationship between debt and economic growth has received considerable attention in international economic literature over recent decades. The majority of the literature argues that there is a negative relationship between debt and economic growth. Specifically, when government debt to GDP levels in advanced and emerging markets reach over 90% it will be associated with lower economic growth (Reinhart and Rogoff, 2010 ; Checherita-Westphal et al, 2012 ; Arcand et al, 2015). This finding could be biased, since most government debt to GDP levels only rose to levels above the threshold in the years after the crisis. The post-crisis fall in GDP could also have resulted from austerity measures. Albeit not explicitly mentioned, Checherita-Westphal (2012) do control for austerity measures by including countries1 with government debt to GDP levels above at least 90% before the crisis. Hence, negative effects on economic growth occur due to high levels of public debt.
Most of the existing empirical studies on the relation between debt and economic growth focus primarily on the role of public debt and not so much on the role of private debt. Cecchetti et al., (2011) was one of the first to assess the role of different types of debt and the effects on economic growth. The main finding has resulted in a threshold private debt to GDP level of 85%, after which private debt becomes a drag on economic growth (Cecchetti et al., 2011). In 1990 the level of bank credit to GDP was situated near 78%, in the subsequent years it has increased to a peak of 118% in 2009 (IFSD IMF, 2016). The level of bank credit to GDP in the Eurozone was on average 113,5% in the years 2007 through 2012 (IFSD IMF, 2016). Gambacorta (2014) posits that recessions associated with financial crises in economies with bank-oriented systems are three times as severe as in those with a market-oriented structure. Therefore, the findings by both Cecchetti et al (2011) and Gambacorta (2014) imply that the rapid increase of private debt in the Eurozone pre-crisis could be a possible explanation of the sluggish economic recovery post-crisis.
1
4 The sluggish economic recovery has been extensively examined by several researchers. Claessens et al (2010) found that countries with house price appreciation and higher levels of private debt before the 2007 crisis suffer from a longer and more severe recession. For instance, a 1% increase in pre-crisis growth in borrowing translates into an additional loss in cumulative real output of 0.2% on average (Feldkircher, 2014). An even more comprehensive study has been conducted by Bezemer (2015), who investigated the composition of debt. When the share of household mortgage credit increases before the crisis, recessions post-crisis appear to be more severe (Bezemer, 2015). To summarize, both public and private debt appear to have a negative or non-linear effect on economic growth after a certain threshold level. Furthermore, economic growth post-crisis suffers from the development of private debt pre-crisis.
Hence, the implications of high debt levels or growth on economic growth are explained in recent papers. Yet, a thorough analysis on the effects of the different GDP components is missing in recent literature. For policy implications it will be useful to know which transmission channels of economic growth will be most harmed and thus which type of debt should be more carefully monitored. Therefore, this study will attempt to assess the implications of high debt growth, in specifically private debt, on the growth of consumption and investments. In addition, this thesis attempts to provide an explanation for the difference in economic growth rates before and after the financial crisis. The answer to the research question is determined by investigating the following subquestions: (1) is there a non-linear relationship between growth of debt and consumption or investments (2) is there a difference between the effects of public and private debt (3) is there a difference in the effects pre- and post-crisis and (4) is there a difference in the short-term and long-term?
5 The remainder of this paper is organized as follows. Chapter 2 provides additional information on the trends in public debt, private debt and economic growth in the Eurozone. Chapter 3 covers the relevant literature on debt, consumption and investments, and establishes the hypotheses. Chapter 4 introduces and explores the data and methodology, followed by empirical results in chapter 5. Thereafter, limitations and the conclusion of this research are presented in chapter 6 and 7 respectively.
2. Trends in Debt and Economic Growth
The increasing level of European private and government debt has gained importance in recent decades, as part of economic growth theories. The rapid development of indebtedness is not to be explained by one specific cause, however Ocampo et al. (2008) argue that capital market liberalization has a major part in it. Capital market liberalization should have enhanced capital flows from industrial to developing countries and financial stability, though it also appears to coincide with changes in economic growth (Ocampo et al., 2008). Figure 1 illustrates the trend to rising level of debt, and its composition, as a percentage of GDP in the years 1999 through 2016. Each year in the graph presents the average debt level of all the Eurozone countries2. Private debt is composed of debt for household and (non-financial) corporations. Total debt is composed of both private debt and government debt3.
As illustrated in figure 1, the debt levels as a percentage of GDP have seen a significant increase over the past two decades. Where the level of total debt was equal to 190 percent of GDP in 1999, it has increased to a level of 260 percent of GDP in 2016. Specifically, the total debt is composed of 92 percentage points of government debt and 165 percentage points of private debt in 2016. Hence, government debt levels have been increasing with an average annual growth rate of 1.2 percent in the years 1999 through 2016, whereas it was 2.4 for private debt. To summarize, there has been a strong growth in Eurozone debt to GDP levels, with a most prominent role for private debt.
2
Composed of 19 countries: Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Portugal, Slovak Republic, Slovenia, and Spain. (World Economic Outlook Database IMF, 2016).
3 Gross government debt consists of all liabilities that require payment or payments of interest and/or
6 Figure 1. Average total debt in Eurozone countries, 1999-2016
As figure 2 illustrates, economic growth moves on different growth paths and trends than both public and private debt. Panel a in figure 2 shows that government debt levels were relatively stable in the years before the financial crisis, whereas panel b illustrates that the levels of private debt kept increasing. New empirical evidence confirms that financial crises tend to be associated with excessive private debt levels in both advanced and emerging market economies (IMF Fiscal monitor, 2016). The downward peak in GDP growth in the year 2007 appears to relate to the build of private debt in the previous years. Moreover, the financial crisis of 2007 appears to be related with the drastic increase in the level of government debt in the Eurozone, as can be seen from the top panel of figure 2.
Figure 2 Panel a: Average Eurozone growth rates of debt and GDP, 1995-2016
-5 -4 -3 -2 -1 0 1 2 3 4 5 0 10 20 30 40 50 60 70 80 90 100 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Year Av era ge G DP G ro w th (% ) Av era ge G o ve rn m en t De b t (% /G DP)
General government gross debt Gross domestic product, constant prices 0 50 100 150 200 250 300 1999 2001 2003 2005 2007 2009 2011 2013 2015 Av era ge d eb t (% G DP) Year
Credit to Corporations Household Debt Private debt Government Debt Total Debt
7 Figure 2 Panel b: Average Eurozone growth rates of government debt and GDP, 1999-2016
Though, public debt appears not to be at the root of the 2007 financial problems, it is still not without risks. Entering a financial crisis with a weak fiscal position exacerbates the depth and duration of the ensuing recession (IMF Fiscal Monitor, 2016). The reason is that the absence of fiscal buffers prior to the crisis significantly curtails the ability to conduct countercyclical fiscal policy. According to the above presented graphs and the debt theory, there seems to be a negative covariation between public and private debt. This covariation is illustrated in figure 3 and can be explained by the Fundamental Identity Theory put forward by Godley (1974). Figure 3.Annual average growth rate of Debt/GDP in the Eurozone, 1999-2016
The Fundamental identity is a three-sector financial balances model, consisting of the private sector, public sector and the current account balance (Lavoie, 2012). When the foreign balance does not change or is equal to zero, it will consequently lead to a balance between the
-5 -4 -3 -2 -1 0 1 2 3 4 5 0 20 40 60 80 100 120 140 160 180 1999 2001 2003 2005 2007 2009 2011 2013 2015 Year Av era ge G DP G ro w th (% ) Av era ge G o ve rn m en t De b t (% / G DP)
Private Debt Gross domestic product, constant prices Source: BIS and IMF
-0,06 -0,04 -0,02 0 0,02 0,04 0,06 0,08 0,1 0,12 0,14 0,16 1999 2001 2003 2005 2007 2009 2011 2013 2015 A ve rag e d e b t (% GDP ) Years
8 private and public sector. Hence, every increase in the private financial surplus has to be followed by a government deficit (Lavoie, 2012).
3. Literature Review
3.1 Consumption
As mentioned earlier, the financial crisis of 2007 has been caused by a burst in the housing bubble that was accompanied by huge amounts of private credit. For consumers, the major implication of the burst of the housing bubble is the tremendous drop in the value of their assets, while being left highly leveraged. In a paper on the effect of household debt and consumption in Denmark, Andersen et al (2014) give an example on the possible relation between both variables. That is, assume an economy with two families X and Y. The families are homogeneous in terms of size, age, income and their house has been bought in the same year. Family X owns a house worth 1 million Euro and has a gross debt of 0.75 million Euro, while family Y owns a house worth 0.5 million Euro and has a gross debt of 0.25 million Euro. Hence, the families have the same net wealth due to a similar absolute difference of 0.25 million Euro between assets and liabilities. However, the Loan-To-Value (LTV)-ratio in family X is 75 percent, while family Y’s LTV-ratio is 50 percent. Family X is more leveraged than family Y. Highly leveraged households are expected to respond different to changes in financial circumstances. Therefore, this section will elaborate on the effects of different debt levels on consumption growth.
The first possible transmission channel through which the levels of consumption are affected appears to be the debt-overhang theory. The theory argues, that firms have reduced incentives to undertake profitable investments when decision makers seek to maximize equity value, because part of the return from a current new investment makes existing debt more valuable (Krugman, 1998., Diamond, 2014). The debt overhang theory does also apply to consumers, where it is assumed to create a need for household deleveraging that, in turn, has been depressing the level of household spending4. In a paper on debt overhang, Dynan (2012) examined the effects of high debt levels on household consumption levels in the US after the financial crisis. Highly leveraged homeowners had larger declines in consumption spending
4
9 than other homeowners ((Dynan, 2012 , Mian and Sufi (2009)). Moreover, for the same housing wealth losses, low income households cut back on spending twice as much than rich households (Mian and Sufi, 2014). Also, retail sales dropped much more sharply in countries with higher leverage (Mian et al., 2011). The debt overhang theory relates to the findings by Reinhart and Rogoff (2010) and Checchetti et al. (2011), since both relate high levels of debt to have a negative result on respectively consumption and GDP. Hence, a non-linear relationship appears to be present after a certain threshold level of debt.
Another channel which could possibly influence the consumption levels in a highly leveraged economy is the access consumers have to credit. For most homeowners, the most important source of credit is borrowing against their home(s) (Andersen et al., 2014). By using their home equity, credit is secured on the housing stock but not invested in it. Hence, it represents additional funds available for reinvestment or to finance consumption spending (Davey, 2001). The decrease in equity values after the financial crisis could have implications for the amount consumers are able to take on as credit and therefore affect the level of consumption in an economy. An external negative shock on housing prices increases the external finance premium, which leads to a further decrease in housing demand and also spills over into consumption demand (Campbell et al., 2007). Thus, a drop in house prices could impede the consumption growth level (Campbell et al., 2007). In addition, US households with weak balance sheets adjust their housing demand more strongly in the face of income shocks (Lamont and Stein, 1999). Therefore, housing price fluctuations are interpreted as a strong role for borrowing constraints. Highly leveraged household could also be constrained in obtaining leverage, because the regulations on credit could become more strict. Financial institutions could be less willing to supply credit to households which would decrease consumption in the short term (Andersen et al., 2014)
10 on consumption growth. Moreover, the high debt levels post-crisis will disable consumers from borrowing, which will affect consumption. This leads to the following hypotheses: Hypothesis 1a: Private debt growth will increase consumption growth at lower levels of private debt, but dampen growth of consumption when private debt accumulates.
Hypothesis 1b:Growth in private debt after the financial crisis will decrease consumption more than before the crisis.
Hypothesis 1c: Debt growth will be more harmful to consumption in the long-term than in the short-term.
3.2 Investments
The major drawback of taking up private debt as mortgages is that it only increases house prices and total debt in the economy. By fuelling asset prices, households tend to feel a bit wealthier. Nevertheless, no goods and services will be created, hence it does not lead to additional income. This outcome is different when an entrepreneur obtains a loan. Debt enables an entrepreneur to invest which could result in a future flow of income which can be used for deleveraging. More importantly, the debt enables an entrepreneur to create profit and jobs. Therefore, investments are the next channel to be examined in this paper.
11 In contrast, debt can also have a negative effect on growth. The effects can be identified at two different moments in time, when debt is increasing and when debt levels are already high5. The first way in which debt might affect investment levels is by crowding out the investments (Reinhart et al., 2012). In his paper on financial cycles, Cleassens (2011) posits that higher private debt will enhance private consumption, though reduce private savings and investments. According to the debt dynamics theory6, government debt levels will increase due to an increase in the budget deficit. In addition, public savings will decrease which in turn results in declining investments. Weil (2012) argues that lower saving rates will lead to a decrease in TFP7 growth, which implies lower economic growth. Furthermore, reduced investments shrink capital stocks, which as well impedes economic growth (Mankiw, 1999). In summary, these authors argue that debt can have negative effects on investments as well, albeit these findings rely on the assumption that savings are needed for investment.
The two subsections above identified the negative and positive effects of increasing debt levels on investments.. In an economic situation of high debt (e.g. post-crisis) the effects seem to refer back to the debt overhang theory. The debt overhang theory argues that profits planned for new investments will be used to service existing liabilities, therefore firms will be reluctant to make new investments before they have deleveraged their situation (Erken et al., 2015). Thereby, high debt burdens at the corporate level restrains turnover and investment growth (Randveer et al., 2011). Specifically, firms with higher debt overhang invest less prior to the European crisis and this effect intensified during the crisis (Kalemli-Ozcan, 2015). Another channel which supports both the debt overhang theory and the crowding out theory is the expectations channel. With the trend in pre-crisis debt accumulation, investors can expect an increase in future debt levels which are larger than the repayment ability. Therefore, future taxes will be higher and investors will decrease their expectations of returns which in turn will lead to lower levels of investment post-crisis (Pattilo et al., 2002). Another important channel through which public debt accumulation can affect growth is that of long-term interest rates. Higher long-term interest rates, resulting from more debt-financed government budget deficits, can crowd-out private investment, thus dampening potential output growth. In
5 For instance, before the Financial crisis of 2007 and after the Financial crisis of 2007. 6
∆debt = (primary deficit/GNP) - (seignorage/GNP) + (real interest rate - growth rate) x d. Holding all else constant a larger primary deficit will increase government debt levels. The noninterest deficit has to be financed with new debt to the extent that this deficit exceeds the amount of money creation by the central bank. In addition, nominal interest expenditures have to be refinanced with new debt (Fischer, 1990).
7
Total Factor Productivity (TFP) is the portion of output not explained by the amount of inputs used in
12
summary, debt appears to affect investments both positively and negatively. However, the positive effects appear to be affected when debt levels are high. Therefore, the second set of hypotheses will be as follows:
Hypothesis 2a: Private debt will have non-linear effects on investments when private debt
accumulates.
Hypothesis 2b: The non-linear effect will be higher in the years after the crisis.
3.3 Interlinkages Between Government and Private Debt
As mentioned before in section 2 on trends in debt, there seems to be a negative covariation between private and government debt. The current empirical literature on debt has extensively examined the bidirectional relationship between private and government debt and its effects on economic growth. Lavoie (2012) argues that in a three sector balances model, the public and private sector move in opposite directions when the foreign balance does not change. Hence, private and public debt are interlinked with each other through a bidirectional relationship.
In a stable economic situation, private debt increases alongside asset prices which enhances consumption and investment levels. Consequently, higher fiscal revenues will be obtained and public debt will decrease (Ludwig, 2004, Lettau, 2004). The emergence of the financial crisis in 2007 reflects a different causation between private and public debt. Once again private growth decreased, however the main explanation for the sudden increase of government debt growth in 2007 (figure 3) is the costs of bailouts. By using fiscal resources to support banks, households, and nonfinancial corporations in restructuring balance sheets, government debt levels increased significantly (IMF, 2016).
13 prudential ratios and it will limit banks’ ability to borrow from capital markets (Martynova, 2015). Hence, banks’ funding is constrained and lending to the private sector reduced. Banks will have also have a motive to deleverage themselves, since an increase in the sovereign risk raises the chance of large balance sheet losses (IMF, 2016).
The strong interlinkage between private and public debt could also affect the effect both of these types of debt have on consumption and investments. In their research on growth effects of government expenditure and taxation in developed countries, Folster and Henrekson (2001), found a negative relation between public expenditure and economic growth. In addition, in a paper on the impact of government debt on private consumption in OECD countries, Berben (2005) shows that the relationship is non-linear. Furthermore, Becker (1997) posits that an increase of government bond holdings results in a decrease in private consumption. Hence, an increase in public debt negatively effects consumption growth. The post-crisis period could also have specific implications in affecting both investments and consumption. Increases in productive government expenditure, financed out of a rise in the tax rate, will be less growth-enhancing in highly indebted European economies (Adam and Bevan). Also, Corsetti et al., (2012) find that when public debt or the deficit is high, the impact responses of government spending on output are lower than in the baseline scenario8. In addition, a spending expansion during a crisis appears to be associated with more inflation and currency depreciation ( Corsetti et al., 2012). Both of these factors could also be a drag on economic growth post-crisis. At high levels of public debt, the government will impose fiscal tightening, consequently reducing both consumption and investments (IMF, 2016). As mentioned before, an increase in public debt will reduce the growth in private debt. In result, credit is rationed less efficiently and the process of debt absorption is delayed which will make the recession longer (Andrés et al., 2016). In addition, credit disruptions prompt a sharp widening of corporate credit spreads (Bassett et al., 2014). As a consequence, the process of debt absorption is delayed which will make the recession longer (Andrés et al., 2016).
In summary, public debt growth appears to have a non-linear effect on consumption and investments as well. Moreover, the interaction between private and public debt works through different transmission channels. Mostly, public debt growth appears to affect the effect of private debt on consumption and investments indirectly. This leads to the hypothesis below:
8 Baseline scenario is an economy with low public indebteness and a healthy financial system (Corsetti et al.,
14 Hypothesis 3: Public debt growth will have a non-linear effect on both consumption and investments.
Hypothesis 3b: The effect is stronger in the years after the financial crisis
4. Data And Methods
In this chapter the research model is presented and applied to test the hypotheses. Furthermore, a comprehensive account of the data collection that took place is described. Finally, the relevant descriptive statistics of the final dataset and the specifications of the regression analysis are presented.
4.1 Data
In order to examine the theory and hypotheses from section 3, the empirical research in this thesis will be based on the model constructed by Checherita-Westphal and Rother (2012). In their research, Checherita-Westphal and Rother (2012) analyzed the impact of high and growing government debt on economic growth. The constructed model in this research will firstly differ in the sense that the dependent variables will be components of economic growth, and not economic growth on aggregate. Secondly, Checherita-Westphal and Rother make use of debt levels, while this thesis uses growth levels of all the variables. Thirdly, the variable private debt is included. To capture the full impact of debt on the GDP components, two dependent variables are analyzed; Private Consumption and Investments. We investigate the effects in 19 Eurozone countries, namely, Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Portugal, Slovak Republic, Slovenia, and Spain between the years (1990-2015). Data originates primarily from Eurostat, The Bank of International Settlements (BIS), The world Bank (WB), International Monetary Fund (IMF), and the Organization for Economic Cooperation and Development (OECD).
15 public debt compared to private debt. A reasonable explanation for this finding is the significant increase in public debt levels in specific Euro countries after the crisis. The average growth rate of Consumption appears to be slightly decreasing on average. While investments follows a different growth path. The average growth rate of investments was 3,6% on average.
4.1.1 Dependent and Independent Variables
The first dependent variable, Consumption¸ can be described as household final consumption expenditure and is obtained from the World Development Index at Worldbank. The second dependent variable, Investments, is collected from the International Monetary Fund (IMF) and corresponds to total investments (as a percentage of total GDP).
The main and most important explanatory variables are Private Debt and Public Debt. Private debt is consolidated credit to the private financial sector, which includes credit to non-financial corporations and Households & Non-Profit institutions serving households (NPISHS). The data for private debt is expressed in percentage of Gross Domestic Product (GDP). The data on private debt for all Euro Countries are obtained from both the BIS Statistic Warehouse and Eurostat. From the Economic Outlook World Database, the variable public debt has been collected. Public debt corresponds to General government gross debt which includes debt liabilities in the form of SDRs, currency and deposits, debt securities, loans, insurance, pensions and standardized guarantee schemes, and other accounts payable (IMF, 2016).
4.1.2 Control Variables
16
4.2 Research Model
In order to examine the effects of debt on both consumption and investment and to test the hypotheses derived in section 3, a dynamic panel data with both country and time fixed effects is considered . In general, panel data have the advantage to analyze data in both, the cross-section as well as time-series dimension. In addition, the most important aspect of a dynamic panel is that the lagged dependent variable is used as a dependent variable. Taking into account the previous years’ levels of consumption and investments, is justified in growth models because the performance should rely on the preceding years (Cecchetti et al., 2011). The model captures both the short- and long-term effects of debt on consumption and investments. The short-term effects are measured in the baseline model. The long-term analysis has been conducted by transforming the data into 3 year non-overlapping growth rates. Cecchetti et al., (2011) argue that five-year growth rates will reduce potential effects of cyclical movements. This thesis applies 3 year non-overlapping growth rates to not harm the total number of observations too much. The following model is considered to examine the role of debt on consumption:
Consumptionit = β0 + β1Consumptionit +ln β2P_Debttit + β3G_Debtit + lnβ3P_Debt2it + β4G_Debt 2it + β5Xit + αt + λi + εit (1)
Where Consumptionit is the level of consumption in an economy at time t, P_Debttit is the total level of private debt consisting of corporate and household debt, G_Debtit is similar, though it represents public debt, P_Debt2itis the squared variable of private debt to capture any non-linear effects. Consequently, G_Debt 2itcaptures the non-linear effect of government debt. Xit is a vector including the control variables. Furthermore, αt measures the time dependent fixed effects, λi country-specific fixed effects and εit is the error term. This same baseline model holds for the investments:
Investmentsit = β0 + β1Investmentsi(t-1) +ln β2P_Debttit + β3G_Debtit + lnβ3P_Debt2it + β4G_Debt 2it + β5Xit + αt + λi + εit
(2)
17 sufficient estimation techniques. In this case, the better option to circumvent endogeneity is to consider an instrumental variables (IV) approach (Almarzoqi et al., 2015). The most suitable IV approach in this paper is the System GMM estimator outlined in Arellano and Bond (1991). Their estimation technique transforms the model into first differences and uses both lagged dependent variables and endogenous variables as instruments (Arellano and Bond (1991). The GMM estimator is preferred over 2-SLS estimators, since it corrects for heteroskedasticity and autocorrelation that may be present in the error structure by using the consistent estimator (Cecherita-Westphal & Rother, 2012). The GMM estimator presents some efficiency gains over the traditional IV/2-SLS estimator derived from the use of the optimal weighting matrix, the overidentifying restrictions of the model, and the relaxation of the independent and identical distribution (i.i.d.) assumption (Baum et al., 2007)
The consideration for GMM estimation is based on the characteristics of GMM mentioned in the paper on the introduction to system GMM by Roodman (2009). In his paper, Roodman (2009) mentions the following characteristics for the data to be useful. Firstly, the panel should have a relative small number of observations N over short time T periods. This condition does not really hold for the sample since the Eurozone consists of 19 countries and the years range from 1990 through 2015. Secondly, variables on the right of the equation do not necessarily have to be exogeneous. Beyond the dynamic aspect of the model, there are strong suspicions that some of the variables are to some extent exogenous. Thirdly, the left hand or dependent variable should be dependent on its past realizations. Hence, the lagged dependent variable is to be found on the right side of the equation, which implies a dynamic panel. Lastly, the panel should incorporate fixed individual effects to cope with unobserved heterogeneity. These unmeasured country specific effects are very likely to appear.
4.3 Robustness
4.3.1 Robustness of the Panel
18 To test for any Autocorrelation in the model, the Wooldridge test is applied. The Wooldridge test for autocorrelation unravels the autoregressive structure of the panel (Drukker, 2003). The significant outcome entails that the model suffers from serial correlation. This issue can be solved in two ways. Firstly, by Specifying cluster-robust standard errors (Cameron & Miller, 2015), as it allows for error correlation within the cross-section. Secondly, by adding a lagged dependent variable. To deal with autocorrelation both solutions are included in the model.
In addition, the Breusch-Pagan test has been conducted to analyze if any heteroskedasticity is at present. The test reveals that the errors are heteroskedastic across observations, which asks for the inclusion of robust standard errors (Breusch & Pagan, 1979).
Lastly, any presence of multicollinearity has been tested. A correlation matrix for annual growth rates and long-term growth rates has been included in Appendix 2. Variables that show a high level of correlation have been highlighted. Not many variables are highly correlated. Moreover, the Variance Inflation Factor (VIF) is used which can detect the severity of multicollinearity. All variance inflation factors, VIF, show low values. Hence, multicollinearity is not of influence in this analysis
4.3.2.Robustness of the GMM Estimators
The GMM estimation is the primary estimation method to be used in growth regressions. In addition to normal panel data assumptions, the GMM also requires its specific robustness tests. The tests aim to check the validity of instruments, hence they should not be correlated with the error term. To test this assumption a Sargan-Hansen test has been conducted. The results are reported in the regression result tables after J-test. In this test the null hypothesis assumes that the error term is not correlated with any of the instruments. In the regressions illustrated in Appendix 3, the significant value implies some indication that the error term is correlated with the instruments.
19 second order autocorrelation, which implies robust models. Hence, lagged debt variables are validated as instruments.
The last assumption to be tested is the number of instruments which could make the estimation less robust. Roodman (2009) argues that the included number of instruments enhances the size of the variance matrix, where the GMM estimator is based on. When the sample includes many instruments and is relatively small, it could be lacking information to provide an estimation for the matrix, consequently results will be inaccurate. In addition, when the number of instruments increases so does the probability that instruments are endogenous. When fitting a larger number of instruments, the number of wrongly chosen instruments will increase making endogeneity more pronounced. There has yet not been a major clarification, however Roodman (2009) acclaims two steps to deal with these problems. Firstly, the number of instruments should always be smaller than the number of observations, which holds in the sample (17<313). Furthermore, the number of instruments should be smaller than the number of estimations. Therefore, the collapsing method by Holtz-Eakin, Newey, and Rosen (1988) has been conducted resulting in an efficient number of instruments. Using the rule of thumb in Roodman (2009), we limit the number of instruments to less than the number of countries in each of our regressions by restricting the number of lags in the GMM-style instruments to 2, and by using the collapsing method of Holtz-Eakin, Newey, and Rosen (1988).
5.Results
5.1 Preliminary Results
20 Figure 4: preliminary examination in the short-term
Starting with the upper left scatter plot, we see a positive correlation between growth in private debt and consumption. The upper right figure represents an opposite causation for private debt and investments. Hence, this provides some first preliminary evidence of the literature that accumulating debt decreases investments. The lower figures both show a positive relation between public debt and consumption and investments. Both of these effects contradict with the existing literature. The relationship between these variables is thoroughly analyzed in the correlation matrix (table 1 of appendix 2). The correlation matrix shows that only public debt is significantly correlated with the dependent variables.
5.2 Estimation Results
5.2.1 The Effects of Debt on Consumption in the Short-Term
The first GDP component to be analyzed is consumption. Table 1 in appendix 3 shows the results of the regression on the dependent variable consumption. This first two columns show
21 a GLS and fixed effects estimation, whilst column 3 is the first column where the system GMM estimation is introduced. Finally, column 4 and 5 show the GMM estimation on consumption before and after the global financial crisis, which is noticeable in the reduction of observations.
The general interpretation of the results will be obtained from column 3, which shows the general short-run effect of private debt growth on consumption growth in the years 1990 through 2015. Since none of the variables have been transformed into their natural logarithm this model can be considered as a level-level regression. Hypothesis 1a and 3a both expect that a non-linear relationship between debt and consumption will occur. Consequently, the linear variable is expected to be positive, whereas the squared term is negative. Table 1shows that a 1 percentage point increase in private debt increases the growth rate of consumption with 0.025 percentage points, albeit not significant. The squared private debt variable has been incorporated to measure the non-linear effects between private debt and consumption. The significant outcome in the regression confirms that a non-linear relationship between private debt growth and consumption growth is present. Admittedly, the outcome is consistent and provides an explanation for hypothesis 1a, but the outcome is still questionable. The coefficient of the squared term is rather small which leads to high threshold debt levels. Figure 5: The non-linear relation between short-term debt and consumption growth
These nonlinear relationships are illustrated in figure 5, which provides important evidence on the debt trajectory of both types of debt. The blue line depicts a threshold level of 12.5 percent, after which the positive effect on consumption growth gradually declines. Hence, the negative effects occurs when countries experience high annual growth in private debt.
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0,45 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Gr o wt h r ate o f Co n su m p tion (% )
Growth rate of debt (%)
22 A nonlinear relation has also been found for the public debt variable. In this case, not only the squared term but also the linear variable has a significant outcome. However, with 0.04 the magnitude of the public debt variable is considerably higher than its private counterpart, while the squared term is equal to the public one. Hence, the non-linear or negative effect of debt on consumption will be pronounced at a higher debt level when considering public debt. In summary, when private debt accumulates, the negative effect on consumption is more prevalent than with public debt.
The lagged dependent variable or dynamic specification Consumption(t-1) is negative and significant. Thus, this paper is able to find evidence of the persistence of the lagged variable in which increases in the preceding growth rate will decrease the growth rate of the following year.
Three of the control variables, savings, taxes and government expenditure are highly significant and have a negative effect on consumption. This outcome is as expected since Savings is income that is not consumed, taxes will decrease disposable income and increases in public consumption will decrease private consumption. Due to insignificance, it is not possible to give any conclusions on the other control variables.
5.2.2 The Effect of Debt on Investments in the short-term
The second estimated GDP component is investments. The estimation results of the regression on investments are presented in table 2 of appendix 3. The first two columns include GLS and fixed effects estimations. Although, both estimations are technically incorrect, they provide a useful check on our results particularly in regard to the coefficient of the lagged dependent variable. The coefficient appears to be positive for every regression. However, the estimation of the lagged dependent variable is insignificant in both the GMM estimation and the GMM for the years after the crisis, hence the choice of a dynamic specification is not validated in these cases.
23 Considering the control variables, 2 of them are significantly different from zero. Government expenditure appears to be positively related to investments, where unemployment has a negative relation.
5.2.3 The Effects of Debt on Consumption and Investment Pre- and Post-Crisis
As mentioned before, public and private debt went on different growth paths in the years preceding and after the global financial crisis. Column (4) and (5) in the regression results table control for the years pre- and post-crisis respectively.To start, the only significant variable in the estimation results before 2007 is the squared private debt term. Private debt appears to have an u-shaped relation with consumption in the years before the crisis. This could mean that when private debt accumulated at a very high pace as during the global housing boom, it slightly increased consumption growth rates. According to Cecchetti et al. (2011) and Dynan (2012), debt accumulation has a negative effect on both economic growth and consumption during periods of high debt. Therefore, hypothesis 1b and 2b propose that the non-linear or negative effects will be more pronounced in the aftermath of the global financial crisis. The estimation results in column (5) are in sharp contradiction with the expectations and recent literature. There is a strong insignificant positive relation between private debt growth and consumption.
In contrast, a significant non-linear effect of public debt on consumption is found. Although, the non-linear or negative effect of debt growth is not found to be stronger compared to the full sample. Nevertheless, this finding is still of considerable importance. The inverted u-shaped relationship dives below zero after the annual debt growth level is higher than 54 percent. This number appears not to be realistic, but many countries experienced a significant increase in government debt growth after the ’07 crisis. Hence, in an attempt to save their economy, policymakers and economists will have unwillingly harmed the growth in consumption.
24
5.2.4 Long-Term Effects of Debt on Consumption and Investments
To analyze the long-term effects of debt on consumption and investments the cumulative 3 years non-overlapping growth rates have been constructed. Investigating non-overlapping growth rates enables to capture a long-run impact more adequately and avoid cyclical fluctuations, mitigate endogeneity problems and especially reverse causation, and tests the robustness of the debt turning point.
Table 1 in appendix 4 presents the results for the long-term estimations. A strong negative effect of private debt on consumption growth is present in the long-term, though insignificant. Cecherita-Westphal and Rother (2010) find a non-linear effect of public debt on economic growth in the long-term. This thesis finds a positive effect of public debt growth on consumption. The squared term is negative, but insignificant and with a very small magnitude. Thus, we are not able to give any robust conclusion on the debt turning point in the long-term.
Regarding the long-term effects on investments, non-linear effects are only found for public debt growth. Despite being highly significant, the non-linear effects are smaller compared to the short-term analysis. Hence, in the longer run there seems to be less reason to worry about government debt accumulation when considering investments.
As illustrated in the regression tables in appendix 4, the long-run analysis comes with several limitations. Firstly, The GMM approach generally works best with large N and small T (Cecchetti et al., 2011). Therefore, a 3 years overlapping growth rate instead of a 5 years growth rate is constructed. Secondly, the AR(2) p-value is equal to 0.098 in table 1. Consequently, the null hypothesis of no serial correlation can be rejected and thus included instruments are weak. Thirdly, standards errors in the estimations are relatively high indicating that coefficients are not well estimated. Fourthly, the p-value of the Sargen-Hansen J-test of over-identifying restrictions which tests for overall validity, ranges between 0.082 and 0.009 in table 1 and 2 respectively. Thus, we are able to reject the null hypothesis that all instruments as a group are ex*ogenous which is not demanded.
6. Discussion and Limitations
25 thorough analysis of these and some other limitations. There are a number of fundamental limitations that must be taken into account and could be improved upon.
Firstly, analyzing annual growth rates should once again be reconsidered because it can result in relatively large differences. E.g. Finland underwent a deep depression in the early 90’s. As a result, initially low government debt levels rose sharply leading to extremely high annual growth rates. The same accounts for government debt levels in Eurozone countries such as Ireland, Greece, Latvia, Lithuania and Italy, specifically after the 2007 financial crisis. Although this thesis attempts to provide additional evidence for the effect of these events, the resulting outliers can have a major influence on the estimation results. To cope with outliers they have been manually omitted. It would have been better to correct the data by using an algorithm developed by Billor et al. (2000). The advantage of using this algorithm is that outliers are efficiently detected and all variables are treated in the same way. Another efficient method to correct for outliers is Winsorizing (Watson, 1990). Since most of the outliers were present in the data for investments and public debt, a more efficient method could affect the estimation of the relation between these two.
Secondly, the Eurozone is investigated due to many similarities between the included countries. Furthermore, the depth and duration of the global recession has been more severe in the Eurozone. The outcomes are limited because they can’t be related to developing countries or low-debt countries. Pattilo (2004) finds that low levels of debt have a positive effect on both economic and TFP growth in developing countries. Hence, when giving any policy implications it would be good to examine whether differences in the effect of debt accumulation exists between developing and developed countries. This can be applied by including an advanced/emerging economy distinction as conducted in ‘’a global house of debt’’ by Bezemer and Zhang (2015).
Thirdly, section 3.3 introduces the interlinkage between both public and private debt. The causation between these two variables could be found by incorporating an interaction variable between public and private debt in the empirical model. Due to a lack of time this effect has not been estimated.
26 is more, the private debt variable is collected from the BIS Statistic Warehouse and Eurostat. Two different sources are used because not every source was able to supply a full dataset. When both sources differ in their way of collecting and analyzing data, bias could be present.
7. Conclusions
After the global financial crisis erupted in 2007, many economists started debating on the effects of the increased indebtedness of developed economies. One of the most important contributions to this debate comes from Cecherita-Westphal and Rother (2012) and Cecchetti et al. (2011) who argue that debt has a negative effect on economic growth after certain threshold levels.
The aim of this study is to provide further insights to the effects of debt on two components of GDP, namely consumption and investments, in 19 Eurozone countries during the years 1990 through 2015. In addition, the difference between the effects of debt growth before and after the 2007 financial crisis in the Eurozone is analyzed. In this study the method by Cecherita-Westphal and Rother (2012) was followed and slightly changed. Firstly, the dependent variable GDP growth is changed into the variable Consumption or investments. Secondly, all variables have been transformed to annual growth rates. Lastly, private debt has been added to the regression to examine the different effects in the composition of debt. The empirical approach investigates whether a non-linear relationship between debt and consumption or investments is present in the short-term, whether the effects differ in the long-term, through which channels the effect is most likely to occur and what the difference before and after the financial crisis is.
For public debt and private debt this thesis finds a non-linear relationship with consumption in the short term. Lower growth levels of debt increase the growth rate of consumption. However, increasing growth rates are associated with a negative effect on consumption growth. As expected, the non-linear relationship is more pronounced for private debt compared to government debt. For the years after the crisis a non-linear relationship is also found between public debt and consumption growth. Though, the effect is unexpected since the non-linear relationship is weaker than in the full sample. In the long-term analysis, a positive effect of public debt growth on consumption is found.
27 with higher investment growth before the financial crisis. In the long-term analysis, a linear relationship between public debt growth and investments is found. However, the non-linear effect appears to be rather small implying that negative effects on investment growth only exists when the 3 year average growth rate is relatively high.
This thesis provides import policy insights on the relation between debt growth and growth of two GDP components. The findings provide a better understanding of the different channels through which debt harms or enhances economic growth. In the short-term, private debt should be more carefully monitored to decrease the possibility of negative effects on consumption growth. In addition, policymakers and analysts in countries where government debt levels increased at high rates after the crisis, can now explain the decrease in consumption growth.
So does growth in debt explain the sluggish economic recovery in the Eurozone after the global financial crisis? According to the existing empirical literature it does. This thesis agrees upon that finding and finds that debt accumulation negatively affects consumption growth rates. Regarding future research, a more comprehensive analysis on the effects of debt on economic growth can be performed by including the additional GDP components. Furthermore, the scope of the analysis could be improved by incorporating emerging economies and by investigating the interaction effect between public and private debt.
28
REFERENCES
Acharya, V. V., & Richardson, M. (2009). Causes of the financial crisis. Critical
Review, 21(2-3), 195-210.
Adam, C. S. and D. L. Bevan (2005), “Fiscal deficits and growth in developing countries”, Journal of Public Economics, Vol. (4), pp. 571-597.
Almarzoqi, R., Naceur, M. S. B., & Kotak, A. (2015). What matters for financial
development and stability? (No. 15-173). International Monetary Fund.
Andersen, A. L., Duus, C., & Jensen, T. L. (2014). Household debt and consumption
during the financial crisis: Evidence from Danish micro data (No. 89). Danmarks
Nationalbank Working Papers.
Andrés, J., & Thomas, C. (2016). When Fiscal Consolidation Meets Private Deleveraging.
Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The review of economic studies, 58(2), 277-297.
Aschauer, D. A. (1988): “The Equilibrium Approach to Fiscal Policy,” Journal of Money, Credit and Banking, 20(1), 41–62.
Bassett, W. F., Chosak, M. B., Driscoll, J. C., & Zakrajšek, E. (2014). Changes in bank lending standards and the macroeconomy. Journal of Monetary Economics, 62, 23-40.
Baum, C. F., Schaffer, M. E., & Stillman, S. (2007). Enhanced routines for instrumental variables/GMM estimation and testing. Stata Journal, 7(4), 465-506.
Berben, R. P., & Brosens, T. (2007). The impact of government debt on private consumption in OECD countries. Economics Letters, 94(2), 220-225.
Bezemer, D., Grydaki, M., & Zhang, L. (2016). More mortgages, lower growth?. Economic Inquiry, 54(1), 652-674.
Billor, N., A. S. Hadi, and P. F. Velleman (2000). BACON: blocked adaptive computationally efficient outlier nominators. Computational Statistics & Data Analysis 34(3), 279–298.
29 Cameron, A. C., & Miller, D. L. (2015). A practitioner’s guide to cluster-robust
inference. Journal of Human Resources, 50(2), 317-372.
Campbell, J. Y., & Cocco, J. F. (2007). How do house prices affect consumption? Evidence from micro data. Journal of monetary Economics,54(3), 591-621.
Canning, D., and P. Pedroni (2004): “The Effect of Infrastructure on Long-Run Economic Growth,” mimeo, Harvard University.
Cecchetti, S. G., Mohanty, M. S., & Zampolli, F. (2011). The real effects of debt.
Checherita-Westphal, C., & Rother, P. (2012). The impact of high government debt on economic growth and its channels: An empirical investigation for the euro area. European Economic Review, 56(7), 1392-1405.
Cipriani, M., Fostel, A., & Houser, D. (2012). Leverage and asset prices: an experiment. FRB of New York Staff Report, (548).
Claessens, S., Kose, M. A., & Terrones, M. E. (2010, September). Financial cycles: what? how? when?. In NBER International Seminar on Macroeconomics 2010 (pp. 303-343). University of Chicago Press.
Comin, D. (2010). Total factor productivity. In Economic Growth (pp. 260-263). Palgrave Macmillan UK.
Davey, M. (2001). Mortgage equity withdrawal and consumption. Bank of England
Quarterly Bulletin, Spring.
Diamond, D. W. and He, Z. (2014), A Theory of Debt Maturity: The Long and Short of Debt Overhang. The Journal of Finance, 69: 719–762. doi:10.1111/jofi.12118 Drukker, D. M. (2003). Testing for serial correlation in linear panel-data models. Stata
Journal, 3(2), 168-177.
Dynan, K. (2012). Is a household debt overhang holding back consumption?. Brookings Papers on Economic Activity, 2012(1), 299-362.
Easterly, W., and S. Rebelo (1993): “Fiscal Policy and Economic Growth,” Journal of Monetary Economics, 32, 417–458.
Elmendorf, D. W., & Mankiw, N. G. (1999). Government debt. Handbook of
macroeconomics, 1, 1615-1669.
30 Feldkircher, M. (2014). The determinants of vulnerability to the global financial crisis 2008 to 2009: Credit growth and other sources of risk. Journal of International Money
and Finance, 43, 19-49.
Fischer, S., & Easterly, W. (1990). The economics of the government budget constraint. The World Bank Research Observer, 5(2), 127-142.
Fölster, S., & Henrekson, M. (2001). Growth effects of government expenditure and taxation in rich countries. European Economic Review, 45(8), 1501-1520.
Garleanu, N., & Pedersen, L. H. (2011). Margin-based asset pricing and deviations from the law of one price. Review of Financial Studies, 24(6), 1980-2022.
Godley, W. and F. Cripps (1974) ‘Demand, inflation and economic policy’, London and Cambridge Economic Bulletin, 84 ( January).
Godley, W., & Lavoie, M. (2012). Monetary economics: an integrated approach to
credit, money, income, production and wealth. Springer.
Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1988). Estimating vector autoregressions with panel data. Econometrica: Journal of the Econometric Society, 1371-1395.
IMF – International Financial Statistics Database July (2016). Series on credit.
IMF – World Economic Outlook Database Oktober (2016). Series on gross domestic product and general government gross debt.
Irmen, A., & Kuehnel, J. (2009). Productive government expenditure and economic growth. Journal of Economic Surveys, 23(4), 692-733.
Kalemli-Ozcan, S., Laeven, L., & Moreno, D. (2015). Debt Overhang in Europe: Evidence from Firm-Bank-Sovereign Linkages. University of Maryland, mimeo.
Kaminsky, G. L., Reinhart, C. M., & Végh, C. A. (2005). When it rains, it pours: procyclical capital flows and macroeconomic policies. In NBER Macroeconomics
Annual 2004, Volume 19 (pp. 11-82). MIT Press.
Krugman, P. (1988), Financing vs. forgiving a debt overhang: Some analytical issues, NBER Working Paper No. 2486.
Lamont, O and Stein, J (1999), ‘Leverage and house-price dynamics in US cities’, RAND Journal of Economics, Vol. 30(3), pages 498-514.
Lettau, M., & Ludvigson, S. (2003). Understanding trend and cycle in asset values:
Reevaluating the wealth effect on consumption (No. w9848). National Bureau of
31 Ludwig, A., & Sløk, T. (2004). The relationship between stock prices, house prices
and consumption in OECD countries. Topics in Macroeconomics, 4(1).
Mankiw, N. G., Romer, D., & Weil, D. N. (1990). A contribution to the empirics of
economic growth (No. w3541). National Bureau of Economic Research.
Martynova, N. (2015). Effect of bank capital requirements on economic growth: a survey.
Mian, A., & Sufi, A. (2015). House of debt: How they (and you) caused the Great
Recession, and how we can prevent it from happening again. University of Chicago
Press.
OECD (2016), Household spending (indicator). doi: 10.1787/b5f46047-en (Accessed on 24 September 2016)
Panizza, U., & Presbitero, A. F. (2013). Public debt and economic growth in advanced economies: A survey. Swiss Journal of Economics and Statistics,149(2), 175-204.
Pattillo, C. A., Poirson, H., & Ricci, L. A. (2002). External debt and growth (No. 2002-2069). International Monetary Fund.
Poirson, M. H., Ricci, M. L. A., & Pattillo, M. C. A. (2004). What are the channels
through which external debt affects growth? (No. 4-15). International Monetary Fund.
Popov, A. A., & Van Horen, N. (2013). The impact of sovereign debt exposure on bank lending: Evidence from the European debt crisis.
Roodman, David. "How to do xtabond2: An introduction to difference and system GMM in Stata." Stata journal 9.1 (2009): 86.
Stiglitz, J. E., & Ocampo, J. A. (2008). Capital market liberalization and
development. Oxford University Press on Demand.
Taylor, John B. The financial crisis and the policy responses: An empirical analysis of
what went wrong. No. w14631. National Bureau of Economic Research, 2009.
Watson, C. J. (1990). Multivariate distributional properties, outliers, and transformation of financial ratios. Accounting Review, 682-695.
World Bank - World Development Indicators (2016). Series on Household Final Consumption Expenditure, Population, Age Dependency Ratio, General Government Expenditure and Savings.
32 Zhang, L., & Bezemer, D. (2015). A global house of debt effect? Mortgages and
33
Appendix 1: Descriptive Statistics
Table 1: Descriptive statistics for the annual growth rates before controlling for outliers Variable N Mean SD Median Max Min
Consumption 444 -0.0024 0.0204 -0.0012 0.1053 -0.0762 Investments 459 0.0120 0.1351 0 0.4794 -0.3916 Private Debt 375 0.0259 0.0686 0.0205 0.4730 -0.2108 Public Debt 426 0.0208 0.1020 0.0046 0.6093 -0.2524 Taxes on Income 398 0.0078 0.1199 0.0031 0.8382 -0.5323 Savings 444 0.0012 0.0402 0 0.1654 -0.1453 AgeDepRatio 444 0.0073 0.0746 0.0048 0.4943 -0.2810 Unemployment 459 0.0012 0.0117 0.0021 0.0331 -0.0330 Countrysize 459 0.0005 0.0924 0 0.6364 -0.3030
Table 2: Descriptive statistics for 3 year non-overlapping growth rates
Variable N Mean SD Median Max Min
34
Appendix 2: Correlation matrix
Table 1: Correlation Matrix for Annual Growth Rates
Table 2: Correlation Matrix for the 3 Year Non-Overlapping Growth Rates Consumption Investments
Private Debt
Public
Debt Taxes Savings
AgeDep- Ratio Unem- ployment Country size Consumption 1.00 Investments 0,2269 1.00 Private Debt 0.0659 -0.0128 1.00 Public Debt 0.2576* 0.6323* -0.0643 1.00 Taxes -0.0959 -0.2625* -0.0776 -0,2022* 1.00 Savings -0.7594* -0.2726* -0.0766 -0,2980* 0,1133 1.00 AgeDepRatio 0.0351 0.0447 -0.1597* 0,1077 0,0108 -0,0433 1.00 Unemployment -0.3201* -0.4784* 0.0068 -0,4339* 0,1523* 0,4535* -0,0856 1.00 Countrysize -0.0338 0.0058 -0.0067 -0,0519 0,0095 -0,1308 -0,1461* -0,0968 1.00 Consumption Investments Private Debt Public
Debt Taxes Savings
35
Appendix 3: Empirical Results
Table 1: Regression Results for Consumption in the Short-Term
(1) (2) (3) (4) (5)
VARIABLES GLS Fixed Effects sGMM sGMM<2007 sGMM>2008 Consumption(t-1) 0.0317 0.000816 -0.294** -0.067 -0.324*** (0.0308) (0.0276) (0.114) (0.054) (0.099) Private Debt 0.0244** 0.0188** 0.025 -0.040 0.059 (0.0104) (0.00745) (0.027) (0.027) (0.051) Private Debt2 -0.000661*** -0.000491*** -0.001** 0.002** 0.002 (0.000159) (9.04e-05) (0.000) (0.001) (0.005) Public Debt 0.0284*** 0.0342** 0.040** -0.031 0.054* (0.00842) (0.0128) (0.016) (0.021) (0.028) Public Debt2 -0.000203** -0.000275 -0.001** 0.002 -0.001** (9.64e-05) (0.000164) (0.000) (0.001) (0.000) Taxes -0.0149*** -0.0171*** -0.029*** -0.006 -0.026 (0.00567) (0.00537) (0.010) (0.004) (0.015) Government Exp. -0.185*** -0.181*** -0.135** -0.262*** -0.106 (0.0189) (0.0428) (0.059) (0.032) (0.071) Age.Dep.Ratio -0.0506 0.0206 -0.030 0.134 -0.052 (0.0602) (0.0901) (0.126) (0.120) (0.236) Unemployment 0.0211** 0.0224** 0.008 0.015 0.029 (0.00899) (0.00969) (0.011) (0.023) (0.024) Countrysize -0.179** 0.232 -0.198 -0.073 -0.325 (0.0809) (0.279) (0.236) (0.255) (0.213) Savings -0.275*** -0.270*** -0.228*** -0.358*** -0.233*** (0.0124) (0.0322) (0.034) (0.038) (0.047) Constant 0.0583 -0.115 0.110 -0.002 0.042 (0.115) (0.217) (0.174) (0.290) Observations 314 314 314 170 106 R-squared 0.725 Number of Id 19 19 19 19 19 Country FE YES No. of instruments 17 17 17 AR1 p-value 0.020 0.059 0.038 AR2 p-value 0.210 0.649 0.326 J-test 0.000 0.001 0.000
36
Table 2: Regression Results for Investments in the Short-Term
(1) (2) (3) (4) (5)
VARIABLES GLS Fixed Effects sGMM sGMM<2007 sGMM>2008
Investment(t-1) 0.322*** 0.313*** 0.302 0.717*** 0.044 (0.0392) (0.0791) (0.196) (0.105) (0.117) Private Debt 0.283** 0.285** 0.381 0.223 -0.266 (0.118) (0.135) (0.230) (0.421) (0.586) Private Debt2 -0.00260 -0.00266* -0.003 -0.007 0.008 (0.00179) (0.00145) (0.003) (0.009) (0.046) Public Debt 0.343*** 0.346*** 0.273 0.459** 0.041 (0.100) (0.114) (0.204) (0.182) (0.293)
Public Debt2 3.27e-05 -1.67e-05 -0.001 0.000 0.000
(0.00109) (0.00118) (0.003) (0.010) (0.003) Taxes -0.189*** -0.191** -0.179 0.028 -0.610 (0.0626) (0.0824) (0.126) (0.114) (0.399) Government Exp. 0.813*** 0.806** 0.789* 0.099 1.165 (0.217) (0.332) (0.410) (0.439) (0.678) Age.Dep.Ratio -0.434 -0.102 0.077 0.145 -1.674 (0.680) (0.904) (0.704) (0.820) (2.068) Unemployment -0.992*** -0.994*** -0.969*** -0.303 -1.227*** (0.103) (0.176) (0.207) (0.212) (0.298) Countrysize 1.162 2.009 1.369 3.409*** 0.409 (0.908) (1.191) (0.835) (1.081) (1.838) Savings 0.363*** 0.374* 0.324 0.051 0.852* (0.136) (0.192) (0.213) (0.319) (0.410) Constant -1.089 -1.456* -0.918 -2.037* 6.679 (0.910) (0.822) (1.806) (1.050) (4.508) Observations 313 313 313 169 106 R-squared 0.642 Number of Id 19 19 19 18 19 Country FE YES No. of instruments 17 17 17 AR1 p-value 0.016 0.013 0.035 AR2 p-value 0.666 0.987 0.513 J-test 0.123 0.640 0.006
