Troubles for Credit Markets? A look at Capital Structure and Government Debt in Europe

Section 1: Introduction

Given the global shock of the covid-19 pandemic and the subsequent awakening of the liquidity issue that underlies the global economy, I thought it would be interesting to analyse the corporate debt situation in the old continent. This is an ambitious project: the goal is to understand (i) corporate structure decisions and (ii) the role played by the national government debt on the capital structure of European firms. Hence, the aim here is to have a general idea of the situation and perhaps to lay the foundation to further studies. The idea is to look at the corporate structure drivers and to test whether government debt crowds out private debt. I will skip the literature review and classic theories (contained in the full paper) and jump straight to the methodology in Section 2. The project required the collection of data on a large sample of European companies (EU-15 countries over the 2000-2017 period) and the analysis (via OLS regression) of the data. The analysis is contained in Sections 3 and 4. In section 3, I adopt the empirical specification studied by Rajan and Zingales (1995) and replicate their model using, of course, a different sample. In Section 4 I then extend the specification to include sovereign debt, finding some support for both the pecking order theory and the crowding out story. Conclusions and limitations are finally discussed in Section 5. With more time, there should be more robustness checks as well as more data.

Section 2: Methodology

2.1 Introduction

Previous international studies have established that capital structure is cross-sectionally correlated with certain factors. The extent to which firms are levered in an economy does not seem to depend solely on the share of external financing that banks account for in that economy (Rajan and Zingales, 1995). The first part of this project builds upon previous literature and uses tested methods to analyse the drivers of capital structure in Europe, in a period characterised by a series of events that work as a semi-natural experiment. In the second part, the government debt will be added as an independent variable to try to draw conclusions on its impact on the corporate sector.

2.2 Data set

The data set used, namely the EU-15 countries in a time period between 2000-2017 has not been analysed yet, to the best knowledge of the writer, in the existing literature. The choice of the sample is based on the growing importance and influence of the European market in the international economy. The European Union was established in 1993 by 12 original member states, with the adoption of the single currency happening in the year 2000. Although the union has grown to 28 members, this study considers only the EU-15 (as defined in 2013) as they are characterised by homogeneous financial characteristics and a high degree of socio-economic convergence that will contribute to the strength of the results of the analysis.

The data set includes both market-oriented and bank-oriented systems. In market-oriented markets, such as the UK, firms raise funds mainly in capital markets and banks act as the lenders only as a last resort. The literature view on the impact of a large stock market on capital structure is mixed. On one hand, large stock markets provide entrepreneurs with opportunities to diversify their portfolios. Thus, in countries with developed stock markets, there may be an incentive for firms to substitute equity for long-term debt. In this case, a developed stock market is expected to be negatively associated with long-term debt (Deesomsak, R., Paudyal, K. and Pescetto, G., 2004). On the other hand, the existence of a developed stock market might increase the firms’ ability to access long-term debt. Grossman (1976), and Grossman and Stglitz (1980) demonstrate that prices quoted in stock markets at least partially reveal information that only informed investors possess. A positive correlation between the development of stock markets and firms leverage is therefore expected, as the release of information might make, in this case, lending to a listed firm less risky.

In a bank-oriented market, such as Germany, France and Italy, banks play a leading role in allocating financial resources and providing most of the credit to the economy. Diamond (1984) demonstrates that banks or other financial intermediaries have better access to information because of their economy of scale. They might also have greater incentives to use the collected information to discipline the borrowers. For this reason, it is suggested that a developed banking sector will facilitate access to external financing, and larger financial intermediaries are expected to have a positive effect on firms’ leverage. According to Allen and Gale (1999), significant differences exist in outcomes between systems in which banks play the dominant role and those where they do not. A developed banking sector leads to an increase in the availability of short-term financing since such form of financing enables banks to use their comparative advantage in monitoring. Demirguc and Maksimovic (1999) present different empirical results by studying firms with different sizes in 30 developing and developed countries. They find that in countries with large banking sectors, small firms have less short-term debt. When it turns to large firms, they fail to find a significant relationship between the size of banking sectors and debt ratio.

The industry-based variables are obtained from Thomson Eikon DataStream. All the data is corrected and adjusted to the single currency euro, to eliminate the effect of the exchange rate and instability. As common in previous literature, outliers have been winsorised to adjust the data were necessary. The time period chosen is 2000-2017, as it includes the effects on liquidity due to financial integration, market liberalisation and the global financial recession.

2.3 Variables

Leverage

A common trait of the existing literature is to use leverage (or debt ratio) as a dependent variable. However, the decision implies further qualification of the data, namely whether to use market value or book value of leverage, and the choice between short-term debt, long-term debt and total debt.

Book value is the cost of a particular asset as recorded on the balance sheet. It is a backward-looking measure, as the price reflects the past financial activities which remain consistent as long as the firm maintains ownership. It is, therefore, helpful information to track past profits and losses. Additionally, according to generally accepted accounting principles (GAAP), tangible assets shown on balance sheet can only be valued at their respective book value. The issue with this is that firms that have ‘consumed’ their assets do not show any significant remaining value, but, nonetheless, it can be assumed that firms are exposed to competitive pressures that force them to reinvest in new equipment. This brings the assets’ values closer to market value and, above all, this is the basis on which firms make their investment decisions. The optimum leverage therefore incorporates the target value as part of a firm’s future capital structure decision. In contrast to the book value, the market value indicates the current value of an asset. This extracts the present value of the real investment options and therefore is forward-looking. Debt financing can thus distort future growth opportunities. The profit or loss lies in the difference between the market and book value if an investment has been owned for a long period of time (Zhu, 2014). Myers (1977) found that a series of rational reasons lie behind the managers' motivation to define leverage targets in terms of book values. Graham and Harvey (2001) reported that managers prefer book values when deciding on the capital structure of a firm. Rajan and Zingales (1995) use the higher level of stock leverage in France and Italy as an example to prove that the results from a market value measurement are slightly different from book value. Bowman (1980) also supported this suggestion. In light of the above, the book value of leverage will be used in this project.

The second distinction to be made is on the use of short-term debt, long-term debt or total debt. According to Bradley, Jarrel and Kim (1984), the tax credits that firms aim to maximise are not applicable to all short-term debt but apply to most current liabilities. Differences in the length of repayment will cause differences in the cost of capital. There are mainly four reasons for this. First, with respect to short-term liabilities, the use of long-term liabilities has more opportunities to influence turnover. Every turnover period follows another one. Therefore, using long-term debt, profitability is higher than that which could be achieved using short-term debt. This also indicates the demand for higher returns for long-term creditors. Second, with the repayment of long-term debt, and considering the impact of compounding interest, the cost of capital can be higher than that from using short-term debt. Third, long-term debt is subject to a greater inflationary impact, according to the formula that the nominal interest rate is equal to the total amount of the real interest rate and expected price changes; where the nominal interest rates on long-term liabilities are bound to be higher than the nominal interest rate of current liabilities. Finally, due to the longer period of bounding, long-term debt can result in instability in the business, and therefore, there will be greater credit risk of default (Zhu, 2014). Acharya et al. (2011) show that using ’net debt’ (defined as the book leverage net of cash and cash equivalents) reflects a more realistic measurement of leverage since the firm will use cash in addition to debt in case of bankruptcy. Given that the information on cash reserves is not available at the industry level, this project will use the book value of the total debt ratio as the dependant variable.

Tangibility

According to Rajan and Zingales (1995), If a large fraction of a firm’s assets is tangible, then the risk of the lender suffering the agency cost of debt is reduced as tangible assets can be used as a collateral. Both the trade-off theory and the agency theory, therefore, predict the positive relationship between tangibility and the level of debt. Similarly to the approach used by a large number of previous studies, this project uses the fixed assets divided by total assets as a measurement of tangibility.

Growth

Myers (1977) find that highly levered companies are more likely to pass up profitable investment opportunities, therefore firms expecting high growth should opt for a greater amount of equity finance. The trade-off theory strengthens these findings and predicts a negative relationship between growth and a firm’s capital structure because an increase in growth opportunities leads to higher agency costs and therefore lowers managerial discretion (Booth et al., 2001). On the other side, the pecking order theory assumes that firms issue equity when their market value is high, thus high growth firms are encouraged to issue debt. As suggested by Rajan and Zingales (1995), the ratio of the book value of assets less the book value of equity plus the market value of equity all divided by the book value of assets is used in this project as a proxy for growth opportunities.

Scale

Past literature provides mixed conclusions on the relationship between capital structure and the firm size. Larger firms tend to be more diversified and fail less often, so size may be an inverse proxy of the probability of bankruptcy, and therefore should have a positive impact on the supply of debt (Rajan and Zingales, 1995). The proxy for scale is either the logarithm of assets (Chen and Strange, 2005) or the logarithm of sales (Rajan and Zingales, 2003; Titman and Wessels, 1988; Whited, 1992). This study will use the logarithm of total sales.

Profitability

Profitability is recognised by the theory as one of the essential determinants of capital structure. The static trade-off model and the agency theory (Jensen, 1986) show a positive relationship between profitability and leverage, while the pecking order theory (Myers, 1984) predicts a negative relationship. Myers and Majluf (1984) confirm a negative relationship based on the fact that firms will prefer to finance with internal funds rather than debt. Jensen (1986), on the other side, shows a positive relationship only if the market for corporate control is effective and forces firms to commit to paying out cash by levering up. If the market is ineffective, instead managers of profitable firms tend to avoid the use of debt as a measure of control. The trend in the literature is to measure profitability by the return on assets (ROA), measured as the ratio of operating earnings before interest, tax, depreciation and amortisation (EBITDA) over total assets (Myers, 1977; Titman and Wessels, 1988; Harris and Raviv, 1990; Rajan and Zingales 1995; Wiwattanakantang, 1999; Deesomsak, Paudyal and Pescetto, 2004; Noulas and Genimakis, 2011). An alternative is to use return on equity (ROE), which shows how much net income is generated based on the shareholders’ equity (Heffman and Fu, 2010), or pre-tax profit margin (as the ration of pre-tax profits to sales) (Hall, G., Hutchinson, P. and Michaelas, N.,2004; Panno, 2003). The first method (EBITDA over total assets) will be used in this project.

2.4 Econometric Issues

The literature identifies a series of issues that typically accompany the econometric analysis. Some limitations are at the base of the quantitative analysis and cannot, therefore, be dealt with, however, the literature points out some precautions that should be taken to ensure the reliability of the results.

Gaps in the data is a common issue when dealing with large samples. Although the databases are improving, some information might have not been provided, might not exist or could get lost during the transition between different platforms. Missing data could result in a loss of explanatory power, and therefore it becomes necessary to tackle the issue. Three methods emerge from previous studies: the first is to drop an industry when key data are not available; the second is to replace the gap with the mean of aggregate values; the third solution is called ‘hot deck’, and consists to select fixed values from another observation with the same covariates. A combination of the three methods, depending on the applicability, has been used in this project.

Another typical issue that arises when the number of observations is relatively large compared to the number of years is autocorrelation. Autocorrelation could be due to a series of factors. For example, in European countries, interconnected economic activities might result in a ‘spillover’ effect leading to spatial autocorrelation. The literature provides a series of tests that can be used to confirm whether the autocorrelation problem is essential, namely the Likelihood-based testing, the Box-Piere and Ljung test or the Breusch-Godfrey test.

Outliers are also present in the dataset. According to Wooldridge (2010), outliers are observations that are numerically distant from the rest of the data, which can affect the efficiency of the model. To eliminate insignificant data, outliers have been winsorised and values of the sample above a relevant percentile have been ignored.

Finally, multicollinearity, a statistical phenomenon in which two or more predictors variables in a multiple regression model are highly correlated, will be tested using the Variance Inflation Factors (VIF) (Daoud, 2009).

2.5 Estimation Method

According to the Gauss-Markov theorem, the ordinary least square (OLS) method is the best linear unbiased estimator (BLUE) (Koop, 2008). Is the ‘best’ as it provides the smallest variance in the estimates. It is ‘unbiased’ as ensures the minimum distance between the sampling distribution and the entire observation. ‘Linearity’ refers to the relationship between the dependant and explanatory variables. The OLS improves the accuracy of the estimation by minimising the sum of all squared deviations from the line (or square residuals). In order to be BLUE, the Gauss-Markov theorem states that the OLS must satisfy five assumptions as listed below (Hayashi, 2000; Knutsen, 2008; Kennedy, 2003; Wooldridge, 2013).

a) Linearity

The parameters estimated must be linear, meaning that the dependent variable can be calculated as a linear function of a specific set of independent variables plus an error term. 

b) Random Sample of n Observations

The sample analyzed must be collected through a randomized and probabilistic mechanism.

c) Exogeneity

Exogeneity means that the regressors cannot be correlated with the error term. If this property is violated, there will be bias in the intercept, because OLS as a procedure forces the average of the error terms to be zero. Note that the assumption is about the underlying structure of the world, and the OLS procedure follows this assumption and forces the error terms into a certain structure.

e) Homoskedasticity

This assumption means the error terms associated with different observations are not related to each other, and the variance of the error term must be constant independently from the value of the regressors.

Section 3: Capital Structure in the EU-15

3.1 Introduction

The aim of this chapter is to examine the capital structure trends in the EU-15 data-set. The analysis will be performed by looking at the correlation between the debt ratio and the determinants of capital structure, building upon the pioneer work by Rajan and Zingales (1995) however applying the model to a sample limited to the EU-15 in the period between 2000-2017. The importance of studying the capital structure in Europe is initiated by the recent processes of financial integration and liberalisation and the impact on firm’s resource allocation and equity development of the capital markets, resulting in a general shift from bank-oriented to market-oriented economies. Furthermore, the financial crisis that impacted the world’s economy in 2008 proved how more reforms and regulations need to be applied to the financial markets. Therefore providing pieces of evidence from Europe will improve the existing literature and provide results that are more relevant to the present.

The test conducted by Rajan and Zingales (1995) is closely replicated in this chapter. The OLS method has been used to estimate cross-sectional models on a yearly basis and in each country of the sample to examine the relationship between the determinants of capital structure and the leverage.

3.2 Data

Industry-level data was obtained from DataStream, which provides a detailed balance sheet and income statement information of European firms. Only one source has been used to ensure that the country-level variables are consistently defined over time. The sample covers the period between 2000-2017, where the first year of the sample is determined by the adoption of the single currency in the EU-15. Since the study focuses on the time-series variation in corporate debt, all data is based on book leverage and alternately year, country and industry-level controls. Table 3.1 shows the industries considered, Table 3.2 the total observations per country and Table 3.3 the number of firms observed per industry in each country.

3.3 Summary Statistics

For the industry-level analysis, aggregates data at firm level has been used. The dependent variable is the book value of the total debt ratio, calculated as the book value of the total debt over the book value of equity. The independent variables are: tangibility, defined as the ratio of fixed assets to the book value of assets; market-to-book, the ratio of the book value of assets less the book value of equity plus the market value of equity all divided by the book value of assets; logsale is the logarithm of net sales; profitability is earning before interest tax depreciation and amortisation divided by the book value of assets. Country, year and industry have been used as control variables.

The summary statistic for each variable is represented in table 3.4.

The mean of the book value of debt to equity of the European firms was found to be among the highest in 2008. During the crisis period, firms substantially increased their debt level due to a growing necessity to raise funds to improve performances and therefore profits. It also reflects firms’ incentive of issuing debt due to tax shield advantage. Zhu (2014) found that the decline of short-term financing is consistent with its procyclical characteristics with respect to economic growth. Due to the financial constraint during crisis periods, firms prefer long-term financing instead of short-term investment. The asymmetric information hypothesis claims that short-term debt is subject to less information asymmetry and is less risky than long-term debt in terms of adverse selection (Myer and Majluf, 1984). In addition, Jensen (1986) suggests in the agency theory that, to reduce moral hazard costs, short-term financing disciplines managers. According to Zwiebel (1999), managerial control is more limited than that of long-term debt, as managers have an incentive to increase profitability and productivity due to the short maturity commitment.

3.4 Baseline Specification

The baseline specification relates the determinant of capital structure to the debt ratio at firm level. The regression estimated is:

Similarly to Rajan and Zingales (1995), all the regressors are eighteen years average (2000-2017) to reduce the noise and to account for slow adjustments. Outliers have been winsorised, and gaps in the data have been resolved by either eliminating an industry (if the data available was very little), by infilling the gap with aggregates average or by using fixed values from another observation with the same covariates. The ratio debt over equity represents the book total debt ratio, the dependant variable; the independent variables are represented by β, γ, δ, ?, ζ; αt, αc and αi are the controlling variables respectively for year, country and industry. Year-fixed effects account for worldwide events such as the financial crisis, country-fixed effects control for time-invariant country characteristics, industry-fixed effects are included to capture any industry differences. One control variable between industry and country has been applied at the same time with the years. In general, the reason to include control variables is to exclude alternative explanations while testing hypotheses with the explanatory variables.

3.5 Empirical Results

In this section the regression has been estimated, with the book value of total debt ratio as the dependant variable. Rather than analyse each country separately, this study tries to outline patterns across the EU-15 countries and discuss separately exceptions. Rajan and Zingales (1995) show similar results for European firms. Other relevant previous studies are Remmers, et al. (1974) and Stonehill, et al. (1975). Both studies focus on five countries (US, Japan, France, Norway and The Netherlands) between 1966-1972 and find that profitability and growth are generally the most important determinants of leverage. Zhu (2014) shows results for European countries between 2000-2012, highlighting a significant positive relationship between growth and leverage (as opposed to the studies described earlier) and a negative relationship between profitability and leverage.

Table 3.4 shows the coefficients for each independent variable. Tangibility and profitability show respectively a very strong positive and negative relationship with the amount of firms’ total debt. Growth opportunities (expressed as market to book) are negatively related to debt, while there is a positive relationship between leverage and size.

Tangibility

Tangibility is expressed as the ratio of fixed to total assets. The importance of tangibility lies with the fact that tangible assets are easier to collateralise and therefore firms might have easier access to long-term debt as the lender is better protected (collateral effect). A strong positive relationship is shown between tangibility and leverage, consistently with Rajan and Zingales (1995). They also argue that as firms with a closer relationship with creditors need to provide less collateral (Berger and Udell, 1994), tangibility should matter less in bank oriented country. Given that the majority of the countries analysed in this study are bank-based economies, this argument goes against the results shown in table 3.4. The reason between such discrepancies between the theoretical analysis and empirical results might be explained by the difference between the use of market and book measures. Rajan and Zingales (1995) finally argue that on a market basis, firms with a lot of fixed assets are not highly levered, thus suggesting that using market-based values the relationship between tangibility and leverage is negative.

Additionally, Zhu (2014) shows no statistically significant impact of tangibility on firms in continental Europe, but highlights that capital expenditure has a significant positive impact on short-term debt ratio in continental European firms. This is consistent with both the trade-off and pecking order theory. When capital expenditure is high, it displays confidence in future earnings and the firm is more prone to issuing debt.

Consistently with both trade-off and agency theory, a strong positive correlation is documented in Table 3.4. As this study is based on a sample where the majority of the countries have, notoriously, a bank based economy, the results suggest that in recent years one of the effects of the European financial system on the EU-15 is being a gradual change from bank to market-oriented system.

Finally, a positive relationship that is stronger than what suggested in previous studies, highlights how lenders have become, particularly with the debt crisis impacting Europe, more aware of the risks related to lending and are therefore more inclined to lend when the borrower can offer a good level of collateral.

Market to Book

Because of a higher cost of financial distress, the theory predicts a negative correlation between market to book and leverage. Fama and French (1992) also argue that the shares of firms with high leverage may be discounted at a higher rate because distress risk is priced. Table 4.4 shows a low-significance negative relationship between growth (expressed as market to book) and debt ratio, consistently with Rajan and Zingales (1995), Jensen (1986), Titman and Wessels (1988) and Fama and French (2002). The result suggests that firms with more growth opportunities tend to borrow less debt over time. Rajan and Zingales (1995) argue that the reason for the negative correlation between market to book and leverage stems from the tendency for firms to issue stock when their stock value is high relative to earnings or book value. On the other hand, Zhu (2014) shows a positive correlation in Europe between 2000-2012, suggesting that firms with substantial growth rates afford to have greater financial leverage since it can generate enough earnings to support the additional interest expenses.

The weak significance of the results and inconsistencies between different studies leaves to future research the task of evaluating closely how growth correlates to leverage. Once again, the debt crisis that hit Europe since 2010 is seen to have an impact on the capital structure decisions and firms’ access to debt.

Size (Logsale)

The positive correlation found by this study is consistent with the trade-off theory. Larger firms are more diversified and have easier access to debt, and therefore are less likely to fail. Furthermore, larger firms are more transparent and therefore less prone to the asymmetric information effect in debt financing compared to small firms (Rajan and Zingales, 1995). Consistently with Titman and Wessels (1988), smaller firms prefer short-term debt to avoid the risk of bankruptcy instead of long-term debt.

It is interesting to note that the result in Panel A, i.e. controlling for country, shows a non-significant correlation between size and leverage. After excluding multicollinearity using the VIF test (Daoud, 2009), the possibility that the p-values are not necessarily measures of strength of association or effect size has been considered; every p-value/effect size curve has a steep part somewhere, so that a small change in effect size can throw the p-value across the arbitrary significance threshold (in this case 0.1). In terms of coefficients, however, the difference between Panel A and Panel B is substantial, suggesting that the association of a predictor with an outcome is different when controlling, in this case, for the country. Assuming that the problem is not in the regression model, the different results could be attributed to significant differences in terms of size between firms when compared at country level.

Profitability

Finally, profitability is significantly negatively correlated with leverage. Consistently with Rajan and Zingales (1995), if in the short run dividends and investments are fixed and debt financing is the dominant mode of external financing, then changes in profitability will be negatively correlated with changes in leverage. The results also confirm the pecking order theory, as firms prefer internal financing over leverage when they are profitable. The dynamic trade-off theory also supports the negative relationship by arguing that firms passively accumulate profits (Kayhan and Titman, 2007). Issuing new equity is the final option because of its high costs which may arise from asymmetric information or transaction costs. In either case, the realised profitability and the available amount of earnings to be retained are important determinants of current capital structure (Myers, 1984).

3.6 Conclusions

This chapter investigates the determinants of capital structure in European firms between 2000-2017. The results provide, generally, strong evidence in support of previous literature and theory, and in particular of the trade-off theory and pecking-order theory. The transition from bank to market-oriented economies, the low cost of debt and high tax brackets in Europe have strengthened the assumptions made by the trade-off theory. Firms use leverage to take advantage of tax shield effects, though up to a certain level when bankruptcy costs reduce such benefits. However, this study has seen the increasing importance of the debt crisis that hit the EU in the nearest past, thus differentiating the situation analysed in this project from the previous literature. The pecking order approach is finally prevailing, generally, in the European firms, confirming that internal founding is preferred to debt to avoid reducing liquidity which in turn increases the risk of bankruptcy.

Section 4: Capital Structure and Government Debt

4.1 Introduction

This section adds the government debt relative to each country in the period between 2000-2017. Following the financial crisis but also the more recent debt crisis, growing attention has been given to increasing government debt levels. Yet, the impact of government debt on the corporate sector has not been explored much in the literature. The European Union, and in particular the integration on the bond market, works therefore as a quasi-natural experiment. Since the second half of the 1990s, the degree of integration in various European ?nancial markets has signi?cantly increased (ECB, 2006). The e?ect has especially been prominent in government and corporate bond markets (Pagano and Von Thadden, 2004 and ECB, 2006). Coeurdacier and Martin (2007) argue that in theory, the monetary integration can have opposing e?ects on the holdings of Euro assets by countries in the Eurozone. For instance, by reducing currency risk, integration can decrease transaction costs of trading across di?erent ?nancial markets in the Eurozone. In contrast, a single currency might increase the substitutability between assets issued by member countries which in turn decreases the Euro asset holdings of the member countries. The results of Coeurdacier and Martin (2007) suggest that the single currency decreased transaction costs for a cross border purchase of a Euro bond or equity for both Euro and non-Euro countries. Although they also ?nd evidence for the negative impact of substitution e?ect on the holdings of Euro assets, the results indicate that the positive impact of lower transaction costs dominates the substitution e?ect. More speci?cally, consistently with Lane (2006), they show that there is a “Euro bias” in bilateral bond holdings between two Euro countries.

4.2 Data

Industry-level data was collected from DataStream, as discussed in the previous sections. The new independent variable, the government debt, has been obtained from the International Monetary Fund (IMF), World Economic Outlook Database. The preferred measure is the gross debt as a percent of GDP as it is an indicator of an economy's health and a key factor for the sustainability of government finance. In accordance with the Maastricht Definition, “Debt” is defined as a specific subset of liabilities identified according to the types of financial instruments included or excluded. Debt is thus obtained as the sum of the following liability categories (as applicable): currency and deposits; securities other than shares, except financial derivatives; loans; insurance technical reserves; and other accounts payable. Changes in government debt over time reflect the impact of government deficits (OECD, 2018). The government debt variable includes both internal and external debt.

4.3 Government Debt

The use of the government debt as the sum of external and internal debt might result in distorted results in the case where an increase in the supply of government debt is absorbed by foreign investors or international ?nancial institutions such as the IMF. In the latter case, it is not expected that changes in government debt would have any impact on corporate leverage, as long as the stock of local government debt stays constant. There are two potential channels through which corporate debt can be a?ected from changes in foreign demand for government bonds. First, holding the change in total government debt constant, a larger fraction of a debt issue is absorbed by foreign investors, leaving more local funds available for corporations. Second, the increase in foreign demand for government debt can crowd out foreign investment into corporations. The latter e?ect would be more prominent in those countries where external private debt is a more important source of debt ?nancing than domestic private debt (Demirci, Huang and Sialm, 2016). According to the IMF (2015), over the period between 2003 and 2014, foreign bank lending to the private sector in emerging markets stayed below 8% of total debt, whereas domestic bank lending was above 78%. Although bond ?nancing has doubled from 8% to 16% since 2008, domestic bank lending is still the main source of debt ?nancing in emerging markets.

A study by Demirci, Huang and Sialm (2016) finds that the negative relationship between corporate leverage and government leverage is driven by domestic public debt rather than external debt. However, given that this study uses the European environment as a quasi-natural setting, it has been decided that the total gross debt as a percent of GDP (i.e. including both external and internal debt) would be a more relevant measure to compare to corporate level variables. Understanding the impact of both measures is a task for future research. 

4.4 Summary Statistics

The dependent variable is the book value of the total debt ratio, calculated as the book value of the total debt over the book value of equity. The independent variables are: tangibility, defined as the ratio of fixed assets to the book value of assets; market-to-book, the ratio of the book value of assets less the book value of equity plus the market value of equity all divided by the book value of assets; logsale is the logarithm of net sales; profitability is earning before interest tax depreciation and amortisation divided by the book value of assets; government debt to GDP is the total gross government debt as a percent of GDP. Country, year and industry have been used as control variables.

The summary statistic for each variable is represented in table 4.2.

Government debt to GDP has a mean of 65.2% and an interquartile range of 54.3% and 77.3%. Table 5.1 shows also that only four countries have a debt exceeding the GDP, namely Belgium, Greece, Italy and Portugal. The largest value is 183% for Greece in 2016, while the lowest value is 6.5% for Luxemburg in 2000.

4.5 Baseline Specification

The baseline specification relates the determinant of capital structure to the debt ratio at firm level and this time includes the government debt to GDP ratio. The regression estimated is: 

As mentioned before, similarly to Rajan and Zingales (1995), all the regressors are eighteen years average (2000-2017) to reduce the noise and to account for slow adjustments. Outliers have been winsorised, and gaps in the data have been resolved by either eliminating an industry (if the data available was very little), by infilling the gap with aggregates average or by using fixed values from another observation with the same covariates. The ratio debt over equity represent the book debt ratio, the dependant variable; the independent variables are represented by β, γ, δ, ?, ζ; αt, αc and αi are the controlling variables respectively for year, country and industry. Year-fixed effects accounts for worldwide events such as the financial crisis, country-fixed effects control for time-invariant country characteristics, industry-fixed effects are included to capture any industry differences. One control variable between industry and country has been applied at the same time with the years.

4.6 Empirical Results

Table 4.3 shows the coefficients for each independent variable. The results for the variables are, as expected, very close to what showed in the previous chapter. In particular, tangibility and profitability show respectively a very strong positive and negative relationship with the amount of firms’ total debt. Growth opportunities (expressed as market to book) are negatively related to debt, while there is a positive relationship between leverage and size. Government debt to GDP is negatively related to leverage, however, the results show low significance when controlling for the industry.

Government Debt to GDP

Similarly to previous literature (Graham, Leary and Roberts, 2014), this study finds a negative relationship between government debt and corporate leverage. Firms reduce their leverage and increase their holdings of liquid assets as a response to increases in sovereign debt. These findings are consistent with corporations increasing purchases and reducing sales of safe and liquid securities, relative to equity, in response to increases in government debt and the reduction in the price of these securities (Friedman, 1986; Krishnamurthy and Vissing-Jorgensen, 2012). Graham Leary and Roberts (2014) find an alternative interpretation of the negative correlation between government debt and corporate structure and argues that the government issues debt in bad economic times during which investment opportunities are poor and, consequently, the demand for credit is low.

Demirci, Huang and Sialm (2016) study the European countries from 1990 and 2006. Interestingly, opposed to this project they show a positive relationship between government debt and corporate leverage after the introduction of the Euro in 2000. It could be argued that the negative relationship documented in this study is the result of the high level of sovereign debt that characterised the European area after the world financial crisis. Firms have become more risk-averse and therefore might prefer to buy liquid assets rather than borrowing in a high sovereign debt environment. On the other side, during the first years after the adoption of the single market and euro currency, the integration might have weakened the crowding out e?ect through increased demand by foreign investors for government debt and/or corporate debt securities. While the former helps local investors in absorbing government debt supply and increases funds available to the corporate sector, the latter decreases ?rms’ dependence on local investors, especially on ?nancial institutions (Demirci, Huang and Sialm, 2016).

The results in Panel A show a significant negative relationship between corporate leverage and government debt. Using industry as a control variable (Panel B), instead, the coefficient becomes statistically not-significant. Firstly, multicollinearity has been excluded using the VIF factors (Daoud, 2009). The test results show a VIF factor in the range of 1, meaning that the variables are almost linearly independent. A second test is to look at the coefficients. The p-values are not necessarily measures of strength of association or effect size; every p-value/effect size curve has a steep part somewhere, so that a small change in effect size can throw the p-value across the arbitrary significance threshold (in this case 0.1). Looking at the coefficients, however, the difference between Panel A and Panel B is substantial, suggesting that the association of a predictor with an outcome is different when controlling, in this case, for industry. This effect could be justified with the Simpson’s paradox (Wikipedia, 2018), which argues that any kind of change is possible, including a change to a large, significant, value with the opposite sign. Simpson’s paradox (Wikipedia, 2018) explains how, similarly to this case, a trend appears in several different groups of data but disappears or reverses when these groups are combined.

A different interpretation of this effect could be that the industry composition is not homogeneous in the countries analysed but nevertheless correlated to the government debt. MacKay et al. (2005) find that industry factors affect not only individual firm decisions but also the joint distribution of financial characteristics within industries. Furthermore, they argue that own-financial structure depends on changes made by industry peers, highlighting the importance of industry interdependence even in a competitive environment.

4.7 Conclusions

This chapter examines the correlation between government debt and corporate leverage in listed companies in the EU-15 during the period 2000-2017. Contrarily to previous studies of European countries that found a positive relationship between sovereign debt and corporate leverage, this project shows a statistically significant negative correlation between the two variables. The comparison of these results to the previous literature highlights the influence of the debt crisis that impacted the Europe area after the financial crisis and the subsequent growing adoption of a risk-averse strategy in the corporate sector, where firms appear to prefer liquid assets to debt. The argument builds upon the pecking order theory, confirming that, in countries with high-level sovereign debt, internal funding is preferred to debt to avoid reducing liquidity which in turn increases the risk of bankruptcy.

It is interesting to note that industry, when used as a control variable, impact adversely the correlation. While some potential sources of the issue have been identified in this study, the effect should be deepened further by future literature.

Section 5: Conclusions

5.1: Conclusions

Building on the work by Rajan and Zingales (1995) and using the monetary integration and single market in the European Union as a quasi-natural experiment, this study investigates the EU-15 countries in the period between 2000-2017. The panel data is analysed with econometric techniques common to the previous literature. This project applies the quantitative methods used by previous studies (Rajan and Zingales, 1995) to provide new insights on the subject that take into account of monetary integration, single market, world crisis and debt crisis that affected Europe at the beginning of the 21st century. Additionally to the firm-level analysis, this study analyses the impact of macroeconomic variable, namely the government debt, on firms’ leverage, thus providing a macro-level understanding of capital structure decisions.

Findings on the determinants of capital structure provide strong evidence in support of the previous mainstream theories, and in particular of the trade-off theory and pecking-order theory. European firms use leverage to take advantage of tax shield effects, however up to a certain level when bankruptcy costs reduce such benefits. Nevertheless, this study documents an increasing impact of the debt crisis that hit the EU in the nearest past, thus differentiating it from the previous literature. The pecking order approach is therefore prevailing, generally, in European firms, confirming that internal founding is preferred to debt to avoid reducing liquidity which in turn increases the risk of bankruptcy.

The analysis of the correlation between government debt and firms’ leverage is controversial. Contrarily to previous work that documented a positive relationship in the European Union, this study shows that increasing levels of sovereign debt result in capital structure decision favouring equity to debt. The pecking order theory prevails, as following the world crisis and the debt crisis firms have become more risk-averse and therefore prefer internal funding to debt to reduce the risk of bankruptcy.

In terms of debt and in light of recent events, it is also observed that as the fear of rising interest rates and the increasing perception of risk for government bond is leading to a blurred risk-reward equation, investors preference might shift from government to corporate bonds. However, in a high-sovereign debt environment, investors must pay extreme caution in evaluating corporate bonds, as higher interests rate might result in a higher risk of corporate default. As large corporate shifts towards the bond market, banks, on the other side, might have to focus on serving smaller companies and individuals.

Low inflation and a highly volatile environment due to the recent political chaos support a higher overall valuation in the equity market. Furthermore, political uncertainty often results in a decline of the Euro, which is good for exporters. This projects also documents a reduction in the correlation between industries, which is another factor in support of the equity market.

The implications of this results might form the basis to future studies on new considerations that governments should take into account when increasing the public debt and, on the other side, whether or not it would be appropriate for a firm to invest into acquisitions in a certain socio-economic situation or for investors to choose between equities and fixed income in a given country. This research also links together different recent studies carried out on the impact of sovereign distress on banks’ ability to lend and capital structure decisions in Europe, thus providing a starting tool for regulators’ work, in the challenge of the post-crisis environment.

5.2 Limitations

Unfortunately, for consistency of the data I had to stop at the end of 2017. It would be interesting to extend the period as close as possible to today. Not all the current European countries are included in the sample, therefore future research could benefit from using a more comprehensive sample. The effects of historical events (world crisis, debt crisis etc.) should be studied by performing separate analysis in different time periods to isolate the impact of each event separately. Also the UK as a non-EMU (Economy and Monetary Union) and market-based economy has a unique market structure which is different from continental Europe, and therefore it would be beneficial if future studies produced a comparative analysis between EMU countries and the UK. An additional step could be differentiating between short and long-term debt, for the reasons explained in the first part of this project. Finally, more robustness checks should be performed.

Thanks for reading!