- Email: [email protected]
The effects of government borrowing on corporate financing: Evidence from Europe
ARTICLE IN PRESS
JID: FRL
[m3Gsc;September 21, 2016;13:56]
Finance Research Letters 0 0 0 (2016) 1–8
Contents lists available at ScienceDirect
Finance Research Letters journal homepage: www.elsevier.com/locate/frl
The effects of government borrowing on corporate financing: Evidence from Europe Yusuf Ayturk Istanbul University, Faculty of Political Sciences, Department of Business Administration, Gulhane Yerleskesi, Alemdar Mahallesi, Soguk Cesme Sokak, Fatih, Istanbul, Turkey
a r t i c l e
i n f o
Article history: Received 17 September 2015 Accepted 16 September 2016 Available online xxx JEL Classification: G32 G38 H63 Keywords: Capital structure Corporate debt Government borrowing Developed Europe
a b s t r a c t This study investigates the relationship between government borrowing and corporate financing decisions in 15 developed European countries for the period of 1989–2014. We find a robust negative relationship between government borrowing and corporate debt in developed European countries. However, we do not identify any significant relation between government debt and equity. The more important finding of our study is that longterm debt of large credit-worthy companies is more sensitive to government debt in comparison to that of small financially constrained companies. In addition to government borrowing, firm-specific factors and other macroeconomic control variables are statistically significant for corporate financing decisions. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Government borrowing and its consequences have been a debatable topic all over the world, especially in the Eurozone, after the global financial crisis. Actually, government debt can play an important role to cope with economic crisis in the short-term, but in the long-term it may crowd out private investment. The current Eurozone sovereign debt crisis experience has indicated that level of government debt can affect behaviour of all economic units including financing activities of corporations. At microeconomic level, determining capital structure is one of the most important decisions in corporate finance practice and theory, because capital structure can affect the firm value. The early capital structure literature focused on the firm-specific determinants of corporate financing by ignoring macroeconomic factors. Trade-off, pecking order and market timing theories are the widely accepted capital structure theories in the literature. A detailed discussion of classical capital structure theories is presented in Frank and Goyal (2008). Recently, one of the most significant topics in capital structure literature is to add macroeconomic variables as independent variables into econometric models. Little is known about the impact of government borrowing on the corporate financing structures of non-financial companies, except findings of Graham et al. (2014a, b) and Fan et al. (2012). The main purpose of this study is to investigate the impact of government borrowing on corporate financing decisions for data set of 15 developed European countries (Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and United Kingdom). Furthermore, we aim to examine sensitivity of corporate leverage to government borrowing for both financially good and bad companies. Specifically, this study seeks E-mail address: [email protected] http://dx.doi.org/10.1016/j.frl.2016.09.018 1544-6123/© 2016 Elsevier Inc. All rights reserved.
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL 2
ARTICLE IN PRESS
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
to address the following research questions: (a) Are corporate debt of non-financial firms and government debt inversely associated in developed European countries? (b) Is there a relationship between equity and government borrowing for nonfinancial firms in developed European countries? (c) Are financial leverage policies of more credit-worthy non-financial firms more sensitive to changes in government debt in comparison to policies of less credit-worthy firms in developed European countries? We use both country-level aggregate panel data and firm-level microeconomic panel data in our econometric analyses. We find a statistically significant inverse relationship between corporate debt level and government debt level. Similarly, we also identify a statistically significant inverse connection between corporate debt issuance and government debt issuance. These relationships are robust for different corporate debt measures, additional control variables and different sample periods. However, we do not identify any relationship between equity and government borrowing. The most important finding of our study is that long-term corporate debt of large credit-worthy companies is more sensitive to changes in government debt level in comparison to that of small financially constrained companies. Our findings are consistent with studies of Graham et al. (2014a and b) and Fan et al. (2012). While Graham et al. (2014a and b) have examined a long U.S. historical data set and discussed the impacts of government debt on corporate financing and investment, Fan et al. (2012) have considered international data set covering developed European countries. The remainder of our paper is structured as follows. Section 2 explains theoretical backgrounds of our motivation, reviews the literature critically and proposes our main hypotheses. Section 3 presents our sample, data set and research methodology. Section 4 reports the results of our study and presents some implications. Lastly, Section 5 concludes this study. 2. Theoretical background and hypothesis development There are three main theories of capital structure which emphasize the importance of firm-specific factors in determining corporate financial policies. These are trade-off, pecking order and market timing theories. Trade-off theory combines tax benefits of using corporate debt and bankruptcy costs (Kraus and Litzenberger, 1973) and agency costs (Jensen and Meckling, 1976) of debt financing. Pecking order theory argues that adverse selection, agency conflicts and taxes have significant effects on capital structure decisions of firms (Myers, 1984; Myers and Majluf, 1984). The most important implication of pecking order theory suggests a specific ranking rule for new fund issuance of firms such that retained earnings is preferred to debt financing and corporate debt is preferred to equity financing (Myers, 1984). Third strand in the literature is market timing theory which asserts that managers try to time new fund issuances according to time-varying costs of equity and corporate debt. A comprehensive survey conducted by Graham and Harvey (2001) for the U.S. companies presented strong evidence that managers try to time changes in interest rates by issuing new debt when they think that interest rates are low, this phenomenon is more valid for larger companies because they have more sophisticated treasury departments. Brounen et al. (2006) conducted a similar survey for detecting capital structure practices in Europe. They also presented evidence of market timing activities for European managers, but these activities are lower in comparison to their U.S. counterparts. The recent cross-country studies have been considering macroeconomic factors which have effects on corporate financing (De Jong et al. 2008; Fan et al., 2012 and Öztekin, 2015). Graham et al. (2014a and b) have shown that government borrowing have significant effects on corporate financing decisions of the U.S. firms over the last century. Based on theoretical backgrounds of Miller (1977); McDonald (1983); Taggart (1985) and Friedman (1986); Graham et al. (2014a) explain aggregate corporate debt equilibrium that is reached at intersection between imperfectly inelastic demand and supply curves. The most important implication of an imperfectly inelastic demand curve of aggregate corporate debt is to motivate a negative association between government debt and corporate debt. As long as government debt is imperfect substitute to other securities in the financial markets, changes in government debt issuance can have effects on relative returns of other securities. In the same vein, Graham et al. (2014b, p.3) claim that changes in supply of government borrowing can affect the relative returns of competing securities in such a way that supply of closer substitutes (corporate debt) is more sensitive to variations in government debt than that of poorer substitutes (equity). Depending on this reasoning, we propose the following two hypotheses: H1. There is a negative relationship between government borrowing and corporate debt for data set of developed European countries. H2. There is no significant relationship between government borrowing and equity for data set of developed European countries. Recent empirical studies of Greenwood, Hanson, and Stein (2010) and Badoer and James (2015) focus on the connection between maturity structures of government debt and corporate debt. Results of Badoer and James (2015) imply that sensitivity of long-term debt issuances of highly rated firms to supply shocks in long-term government bond issuances are higher than other securities, because they are closer substitutes. Badoer and James (2015) have also found strong evidence that increase in supply of government debt lowers only long-term corporate debt of large credit-worthy firms, not that of small financially constrained firms, in addition, they have not identified any significant relationship between short-term corporate debt issuance and government debt supply. Similarly, Graham et al. (2014b) have found that long-term debt of credit-worthy companies is more sensitive to government borrowing in comparison to that of small financially constrained companies. In line with this theoretical reasoning and recent empirical findings on the U.S. data set, we propose the following hypothesis: Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
3
Table 1 Summary statistics. Panel A. Country-level aggregate panel data
N
Mean
Stand. Dev.
Min.
Max.
Corporate leverage variables Total debt / Total assets Net debt / Total assets Long-term debt / Total assets Short-term debt Total assets Change in total debt / Total assets (t–1) Change in long-term debt / Total assets (t–1) Change in short-term debt Total assets (t–1)
390 390 390 390 375 375 375
0.29 0.19 0.21 0.08 0.03 0.02 0.01
0.07 0.09 0.07 0.03 0.05 0.04 0.02
0.12 –0.03 0.06 0.02 –0.14 –0.11 –0.06
0.54 0.47 0.41 0.18 0.38 0.30 0.16
Other firm specific variables EBIT / Total assets Net income / Total assets Market to book asset value ratio Cash / Total assets Intangible assets / Total assets
390 390 390 390 390
0.08 0.04 1.37 0.11 0.15
0.03 0.02 0.32 0.04 0.12
–0.11 –0.15 0.85 0.02 –0.08
0.18 0.10 3.40 0.24 0.57
Government debt variables Government debt / Total assets Change in government debt / Total assets (t–1)
390 375
1.56 0.10
1.37 0.21
0.24 –0.23
8.61 1.58
Macroeconomic variables Treasury bill rate Inflation rate Equity market return Real growth rate of GDP Government expenditures / Total assets Government debt / GDP
390 390 390 375 390 390
0.05 0.02 0.08 0.02 1.16 0.66
0.04 0.02 0.25 0.02 0.90 0.31
0.00 –0.04 –0.63 –0.08 0.30 0.10
0.21 0.13 0.98 0.13 6.01 1.63
Panel B. Firm-level panel data
N
Mean
Stand. Dev.
Min.
Corporate leverage variables Total debt / Total assets Net debt / Total assets Long-term debt / Total assets Short-term debt total assets Change in total debt / Total assets (t–1) Change in net debt / Total assets (t–1) Change in long-term debt / Total assets (t–1) Change in short-term debt Total assets (t–1)
79,343 78,542 79,343 79,343 69,884 69,884 69,884 69,884
0.21 0.07 0.13 0.08 0.02 0.02 0.01 0.01
0.18 0.27 0.13 0.09 0.12 0.16 0.09 0.07
0.00 –0.83 0.00 0.00 –0.32 –0.78 –0.26 –0.28
0.85 0.77 0.65 0.54 0.79 0.75 0.59 0.40
Other firm specific variables EBIT / Total assets Market to book asset value ratio Intangible assets / Total assets
78,542 78,542 78,542
0.02 1.62 0.22
0.18 1.16 0.20
–1.31 0.52 0.00
0.35 10.11 0.88
Max.
H3. Long-term debt of large credit-worthy firms is more sensitive to government debt in comparison to that of small financially constrained firms for data set of developed European countries. 3. Sample, data set and methodology We use both country-level aggregate and firm-level individual data set of 15 developed European countries. These countries are Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and United Kingdom. We exclude financial firms from our sample as in the case of most capital structure studies. Our sample of firms includes all non-financial firms listed in the national stock exchanges of countries over the period of 1989–2014. The details of all variables used in this study are reported in appendix. In our analysis, we use different samples based on macroeconomic and firm-level factors. The first group of samples consists of aggregate country-level variables which are constructed by summing firm-level values across firms at each year for each country so that we construct country-year panel data. We use sum of total assets as scale factor to calculate country-level ratios. Table 1 Panel A provides summary statistics for country-level aggregate data. Our second group of samples consists of firm-year panel data. We winsorized each firm-level variable at 1% and 99% to deal with possible outlier problem. Table 1 Panel B shows summary statistics for firm-year unbalanced panel data. Following the methodology of Graham et al. (2014b), we use three different linear corporate debt demand equations: (1) country-year panel data with aggregate values at level; (2) country-year panel data with aggregate flow values; (3) firm-year panel data with firm-specific financial data for large and small firms.
Q Cit QG = αi + β it + δ Xit + εit Ait Ait
(1)
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL 4
QG it
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
α i is country-specific time-invariant intercept. Q Cit is sum of interest bearing corporate debt for ith country at year t and
is sum of government debt for ith country at year t. Ait is sum of total assets for ith country at year t. Xit is a vector of aggregate firm-specific and macroeconomic control variables for ith country at year t and ε it is error terms. The main variables of this study are government debt and corporate debt ratios. They are scaled by sum of total assets at each year to ensure that these variables are consistent. We use stock values of government debt and corporate debt at the end of each year. We include treasury bill yield, inflation rate, equity market return, growth rate of real GDP and government expenses/total assets ratio as macroeconomic control variables based on the study of Graham et al. (2014b); earnings before interest and tax (EBIT)/total assets, intangible assets/total assets and market value of the firm/total assets as firm-specific control variables based on the findings of Frank and Goyal (2009). Our concern is on the sign and significance of government debt coefficient which is represented by beta coefficient in the model. We expect a negative relationship between government debt and corporate debt. The second equation is presented below. C
C
j Q i,tj − Qi,t−1
Ai,t−1
= αi + β
G Q Gi,t − Qi,t−1
Ai,t−1
+ δ Xit + εit
(2)
In this equation, we use changes in government debt and changes in corporate debt scaled by first lagged total assets. Control variables are the same as we have used in Eq. (1). However, all stock variables are first differenced and scaled by first lagged total assets. Similar to Eq. (1), we expect a negative relationship between change in government debt and change in corporate debt; on the other hand, we expect to find no significant relationship between government debt issuance and equity issuance. In addition to two preceding models with aggregate data, we can use firm-level data to investigate our third hypothesis. As we have discussed in preceding section, Friedman (1986) and Graham et al. (2014b) assert that corporate debt of more credit-worthy firms are closer substitute of government debt. Thus we expect that corporate debt of large credit-worthy firms is more sensitive to changes in government debt in comparison to that of small firms. Following methodology of Graham et al. (2014b), for each country and at each year, we classify all firms into 10 groups based on 10 quantiles of their total assets so that small firms are classified in the lowest quantile and large firms are classified in the highest quantile. We use firm size as a proxy for measuring credit-worthiness depending on studies of Shivdasani and Zenner (2005) and Graham et al. (2014b). We aim to compare the effects of government borrowing on small and big firms by using the model in Eq. (3) below.
yi jt =
αi + β1 g jt × I (lowi jt ) + β2 g jt × I (highi jt ) + δ1 Xi jt × I (lowi jt ) + δ2 Xi jt × I (highi jt ) + φ1Y jt × I (lowi jt ) + φ2Y jt × I (highi jt ) + εit
(3)
yijt is corporate leverage (Total Debt, Total Net Debt, Long-term Debt, Short-term Debt) of ith firm in jth country at year t. α i is firm-specific time-invariant intercept. gjt is government debt of country j at year t. I(highijt )is a dummy variable which equals to one if the firm i in country j at year t is in the highest quantile and I(lowijt ) is a dummy variable which equals to one if the firm i in country j at year t is in the lowest quantile. Xijt is a vector of firm-specific control variables of ith firm for jth country at year t. Yjt is a vector of country-specific control variables of jth country at year t. ε it is error terms. We estimate all equations by using firm fixed-effects ordinary least squares (OLS) regression analysis. In Eq. (3), difference between β 1 and β 2 coefficients represents difference between sensitivities of corporate debt to government debt for large and small non-financial firms. We conduct a simple t-test to compare β 1 and β 2 . 4. Results and discussion Table 2 reports the results obtained from estimation of Eq. (1) for alternative models. The coefficient of government debt explanatory variable is negative and statistically significant at 1% level in model 1 which implies a negative association between government borrowing and corporate leverage. To check robustness of our results, we also use alternative definitions of corporate leverage in models 2, 3 and 4. We have detected similar statistically significant negative relationship between net corporate debt and government debt. More importantly, results of models 3 and 4 indicate that both magnitude and statistical significance of government debt ratio coefficient for long-term debt model are more negative than that for short-term debt model. As suggested by Graham et al. (2014b, p.14), we also add several additional variables which can represent corporate sector investment and supply of corporate debt. These variables are change in oil price, unemployment rate, growth in money supply, change in real assets and spread between government bond and government bill. Lastly, we check the relationship between corporate debt and government debt for two different periods. Hence, we aim to check the possible effects of current crisis period with extraordinary high levels of government debt. Table 2 reports the results of estimated three models (models 5, 6 and 7), one with additional variables and two others with different periods. Estimation results of more comprehensive model (model 5) indicate that the relationship between corporate debt and government debt is still negative and significant with similar coefficient. Our results prove that negative relationship is not specific to the period after the recent financial crisis that is characterized by high level of government borrowing. Overall, our results are consistent with findings of Graham et al. (2014b) on the U.S. data. After analysis of corporate leverage at level, we also examine the same relationship by using flow data explained in Eq. (2). Table 3 presents the results of fixed-effects panel data analysis. The estimated coefficients of total debt and long-term Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
5
Table 2 Estimation results of country-level aggregate stock data. Dependent variable
Total debt / Total assets
Net debt / Total assets
Long-term debt / Total assets
Models
1
2
3
Government debt / Total assets Macroeconomic variables T-Bill return Inflation Equity market return Real growth of GDP Government Exp. / Total assets Firm characteristics EBIT / Assets Intangible assets / Total assets Market-to-book asset ratio
–0.024 –(6.26)
∗∗∗
–0.024 –(4.16)
∗∗∗
–0.030 –(7.42)
∗∗∗
Short-term debt / Total assets 4 0.006 (2.03)
∗
Total debt / Total assets Post-2007 7
–0.029∗∗∗ –(4.27)
–0.028∗∗∗ –(3.17)
–0.022∗∗∗ –(6.22)
0.832∗∗ (2.83) –0.189 –(0.83) 0.019 (0.86) –0.145 –(0.94) 0.002 (0.18)
0.186 (1.48) –0.357∗∗ –(2.31) 0.026∗ (1.83) –0.354∗∗ –(2.62) 0.024∗∗∗ (3.33)
0.329∗∗ (2.64) 0.145 (1.23) -0.002 –(0.25) 0.047 (0.56) -0.011∗ –(1.86)
0.544∗∗∗ (3.68) –0.058 –(0.27) 0.026 (1.39) –0.398∗∗ –(2.46) 0.031∗∗∗ (6.46)
0.926∗∗∗ (8.68) –0.655∗∗ –(2.60) 0.048∗∗ (2.63) –0.867∗∗∗ –(3.74) 0.029∗∗∗ (3.66)
0.078 (1.26) 0.291∗∗∗ (4.92) –0.001 –(0.32) -0.110∗ –(1.84) 0.041∗∗∗ (5.34)
–0.592∗∗∗ –(4.07) 0.178∗∗∗ (3.29) –0.032 –(1.17)
–0.565∗∗∗ –(4.00) 0.239∗∗∗ (3.63) -0.025 –(0.80)
–0.306∗∗ –(2.37) 0.196∗∗∗ (5.02) –0.044∗ –(1.83)
–0.286∗∗∗ –(4.85) –0.018 –(0.62) 0.012∗∗ (2.41)
–0.628∗∗∗ –(6.33) 0.206∗∗∗ (3.40) –0.017 –(0.69)
–0.736∗∗∗ –(6.90) 0.240∗∗∗ (3.48) –0.020 –(0.81)
–0.519∗∗∗ –(9.69) –0.058 –(1.38) –0.030∗ –(2.05)
0.075∗∗∗ (15.77) 0.387 41.2∗∗∗ 375 15
0.039∗ (1.84) 0.201 (1.26) 0.001 (0.03) 0.020 (0.79) –0.194∗∗∗ –(3.09) 0.306∗∗∗ (8.05) 0.415 270.7∗∗∗ 353 15
0.050∗ (2.10) –0.157 –(0.92) 0.023 (0.59) 0.009 (0.24) 0.546∗∗∗ (4.28) 0.337∗∗∗ (7.90) 0.495 470.5∗∗∗ 248 15
–0.019 –(1.10) 0.057 (1.04) 0.019 (1.11) –0.008 –(0.17) –0.094∗∗∗ –(6.26) 0.378∗∗∗ (14.97) 0.640 786.1∗∗∗ 105 15
Unemployment rate Money supply growth Change in real assets Treasury bond - Bill spread
R2 F statistics N Number of countries
5
Total debt / Total assets Pre-2008 6
0.514∗∗ (2.21) –0.212 –(0.94) 0.024 (1.21) –0.307∗ –(2.09) 0.014∗ (1.82)
Additional variables Change in oil price
Constant
Total debt /Total assets
0.362∗∗∗ (10.32) 0.392 24.4∗∗∗ 375 15
0.231∗∗∗ (6.12) 0.393 129.0∗∗∗ 375 15
0.287∗∗∗ (8.53) 0.494 299.5∗∗∗ 375 15
The models are estimated by country fixed-effects panel data method. Driscoll and Kraay (1998) robust standard errors are used for heteroskedasticity and autocorrelation. t-statistics are reported in parenthesis. ∗ , ∗∗ , ∗∗∗ indicate 10%, 5% and 1% significance levels respectively. Variance Inflation Factors (VIFs) are calculated to check multicollinearity problem, all VIFs are below 5.
debt models (1 and 2) are negative and statistically significant at 5% level, whereas the coefficients of short-term debt and equity models (3 and 4) are not statistically significant. From these results, we have documented a negative association between total debt issuance and government debt issuance. More importantly, we can see that the main source of this relationship is long-term debt issuance rather than short-term debt issuance. Lastly, as questioned by Friedman (1986) and Graham et al. (2014b), we examine the corporate financing decisions of different companies in response to changes in government borrowing. In order to compare sensitivities of large and small firms’ corporate debt to government debt, we estimate Eq. (3) with firm-specific fixed-effects panel data model. The results of all models for stock and flow data are reported in Table 4. For stock data (Table 4 Panel A), results of the model with long-term debt/total assets ratio indicate that the coefficient of government debt variable is −0.275 for the subsample of large firms, on the other hand, it is 0.063 for that of small firms. To compare these two results, we use a simple t-test for the difference, −0.338, and it is statistically different from zero at 1% significance level. We have not found similar results for other model with total debt. We also compare sensitivities of corporate debt issuances of large and small firms to government debt issuances as reported in Table 4 Panel B. In the model with long-term debt issuance for large firms, the estimated coefficient of government debt issuance is −0.032, whereas the corresponding coefficient of small firms is −0.006, but not statistically significant. The difference, −0.026, is statistically significant at 5% level. Overall, these results show that long-term corporate debt levels of large firms are more sensitive to government debt levels than that of small firms. This finding is also valid for debt issuance flow data. Our results are consistent with empirical findings of Friedman (1986) and Graham et al. (2014b). The main implication of our findings is that long-term corporate debt of larger firms is closer substitute to government debt for data set of developed European countries. Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL 6
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
Table 3 Estimation results of fixed–effects panel data model with aggregate flow data. Dependent variable:
Change in total debt/ Total assets 1
Change in long–term debt/ Total assets 2
Change in short–term debt/ Total assets 3
Change in Gov. Debt / Assets(t–1)
–0.059∗∗ –(2.13)
–0.060∗∗ –(2.39)
–0.009 –(1.04)
0.228∗∗ (2.47) 0.138 (0.57) 0.019 (1.05) 0.207 (1.33) –1.188∗∗∗ –(4.06)
0.105 (1.47) 0.192 (1.08) 0.003 (0.21) 0.111 (0.80) –0.875∗∗∗ –(3.32)
0.129∗∗∗ (3.62) –0.021 –(0.26) 0.008 (0.91) 0.118 (1.53) –0.369∗∗ –(2.89)
0.209 (1.59) 3.864
0.112 (1.38) 1.667
0.063 (0.93) –1.680
(0.81) –0.076∗∗∗ –(4.21) 0.008 (0.89) 0.187 5.5∗∗∗ 330 15
(0.46) –0.035∗∗ –(2.19) 0.009 (1.69) 0.137 7.6∗∗ 330 15
–(1.15) –0.028∗∗ –(2.64) –0.002 –(0.42) 0.153 5.5∗∗ 330 15
Macroeconomic variables T-Bill return Inflation Equity market return Growth of real GDP Change in Gov. Exp. / Assets (t-1) Firm characteristics Change in EBIT / assets (t–1) Change in intangible assets / Assets (t–1) Change in market-to-book asset ratio Constant R2 F statistics N Number of countries
Change in equity/ Total assets 4 0.007 (0.41) 0.140∗ (1.83) –0.260 –(0.96) 0.023 (1.42) 0.163 (1.41) –1.101∗∗ –(2.46) 0.977∗∗∗ (6.84) –0.722 –(0.25) –0.034 –(1.16) 0.017∗∗ (2.48) 0.392 14.5∗∗∗ 330 15
The models are estimated by country fixed-effects panel data method. Driscoll and Kraay (1998) robust standard errors are used for heteroskedasticity and autocorrelation. t-statistics are reported in parenthesis. ∗ , ∗∗ , ∗∗∗ indicate 10%, 5% and 1% significance levels respectively. Variance Inflation Factors (VIFs) are calculated to check multicollinearity problem, all VIFs are below 5.
Table 4 Estimation results of large and small firms’ stock data. Panel A. Stock data Total debt/Total assets Large Gov debt / Assets Control variables Constant R2 F statistics N Number of firms
Long-term debt/Total assets Small
∗∗
–0.285 –(2.07) Yes Yes 0.061 86.2∗∗∗ 12,472 2132
–0.076 –(0.71)
Difference –0.210 –(1.13)
Large ∗∗∗
–0.275 –(3.47) Yes Yes 0.062 39.0∗∗∗ 12,480 2140
Small
Difference
0.063 –(0.74)
–0.338∗∗∗ –(2.71)
Panel B. Flow data Change in total debt/ Total assets (t–1)
Gov Debt / assets Control variables Constant R2 F statistics N Number of firms
Change in long-term debt/ Total assets (t–1)
Large
Small
Difference
Large
Small
Difference
–0.043∗∗∗ –(3.64) Yes Yes 0.236 610.0∗∗∗ 12,383 2106
–0.027 –(1.26)
–0.017 –(0.89)
–0.032∗∗∗ –(3.20) Yes Yes 0.177 117.3∗∗∗ 12,406 2120
–0.006 –(0.45)
–0.026∗∗ –(2.04)
The models are estimated by fırm fixed-effects panel data method. White (1980) robust standard errors clustered by country and industry groups (45) are used for heteroskedasticity and autocorrelation. t-statistics are reported in parenthesis. ∗ , ∗∗ , ∗∗∗ indicate 10%, 5% and 1% significance levels respectively. Variance Inflation Factors (VIFs) are calculated to check multicollinearity problem, all VIFs are below 5.
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL
ARTICLE IN PRESS
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
7
5. Conclusions This study has examined the effects of government borrowing on corporate financing decisions in 15 developed European countries for the last 26 years, from 1989 to 2014. The results of our analysis show that government debt is a significant factor to determine corporate financing decisions in developed European countries. Firstly, we have detected that there is a strong negative association between corporate debt and government borrowing, whereas there is no significant relationship between equity and government borrowing. These relationships are robust for both flow and stock data. In other words, higher government borrowing levels lead to financial crowding out effect on corporate debt in developed European countries. These results should be of interest to governments and corporate managers in these countries. More importantly, we have investigated sensitivities of small and large firms’ corporate debt to government debt for data set of developed European countries. Our comparisons have revealed that long-term debt of large credit-worthy firms is more sensitive to government debt than that of small financially constrained firms. This result implies that long-term corporate debt of credit-worthy companies is closer substitute to government bonds. These companies are able to fill supply shocks in government debt of developed European countries. Acknowledgements I would like to acknowledge the financial support of Jean Monnet scholarship programme awarded by Turkish Ministry for European Union Affairs and European Union. Appendix. List and definitions of variables
Panel A. Firm-level variables Variable
Definition
Source
Units
Corporate leverage Total debt Long-term debt Short-term debt Total assets Corporate net debt issuance Corporate net long-term debt issuance Corporate net short-term debt issuance Corporate net equity issuance
Total debt / Total assets Long-term debt + short-term debt Long-term debt Short-term debt Total assets Change in total debt/Total assets (t-1) Change in long-term debt/Total assets (t-1) Change in short-term debt/Total assets (t-1) Change in book value of stockholders’ equity in excess of the change in retained earnings (Fama and French, 2005) Total debt - cash and Cash equivalents Cash and short-term investments / total assets Earnings before interest and tax / Total assets Net income / Total assets (Total liabilities + Market capitalization) / Total assets Intangible assets / Total assets Total assets - (Net Plant, property and equipment (Net) + Accounts receivables + Cash and Short term investments + Inventory)
Datastream Datastream Datastream Datastream Datastream Datastream Datastream Datastream Datastream
% Billion Billion Billion Billion % % % %
Datastream Datastream Datastream Datastream Datastream Datastream Datastream
Billion Euro % % % % % Billion Euro
1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014
Variable
Definition
Source
Units
Period
Government leverage
Government debt / Total assets of all non-financial companies Change in government debt / total assets Year-end unemployment rate Annual change in year-end industrial production index Year-end M1 money supply Annual change in year-end consumer price index Year-end local oil price Annual change in MSCI country index Year-end 10-year indicator bond yYield Year-end 3-month indicator bill yYield Nominal gross domestics product Real gross domestics pProduct based on 2010
Oxford economics
%
1989–2014
Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford
% % % Billion Euro % EU % % % EU EU
1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014
Net debt Liquid assets Profitability Net profitability Market to book asset value Intangibility Intangible assets
Period
Euro Euro Euro Euro
1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014
Panel B. Macroeconomic variables
Government net debt issuances Unemployment rate Industrial production Money supply Inflation Price of oil Equity market return Treasury bond yield Treasury bill yield Gross domestic product Real gross domestic product
economics economics economics economics economics economics economics economics economics economics economics
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL 8
ARTICLE IN PRESS
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
References Badoer, D.C., James, C.M., 2015. The determinants of long-term corporate debt issuances. J. Finance (forthcoming). Brounen, D., De Jong, A., Koedijk, K., 2006. Capital structure policies in Europe: survey evidence. J. Bank. Finance 30 (5), 1409–1442. De Jong, A., Kabir, R., Nguyen, T.T., 2008. Capital structure around the world: the roles of firm-and country-specific determinants. J. Bank. Finance 32 (9), 1954–1969. Driscoll, J.C., Kraay, A.C., 1998. Consistent covariance matrix estimation with spatially dependent panel data. Rev. Econ. Stat. 80 (4), 549–560. Fama, E.F., French, K.R., 2005. Financing decisions: who issues stock? J. Financ. Econ. 76 (3), 549–582. Fan, J.P., Titman, S., Twite, G., 2012. An international comparison of capital structure and debt maturity choices. J. Financ. Quant. Anal. 47 (01), 23–56. Frank, M.Z., Goyal, V., 2008. Trade-off and pecking order theories of debt. In: Eckbo, E. (Ed.). In: Handbook of Corporate Finance, vol. 2. Elsevier, North Holland, pp. 135–202. Frank, M.Z., Goyal, V.K., 2009. Capital structure decisions: which factors are reliably important? Financ. Manage. 38 (1), 1–37. Friedman, B.M., 1986. Implications of government deficits for interest rates, equity returns and corporate financing, (No. c7940), National Bureau of Economic Research, 67–90. Graham, J.R., Harvey, C.R., 2001. The theory and practice of corporate finance: evidence from the field. J. Financ. Econ. 60 (2), 187–243. Graham, J.R., Leary, M.T., Roberts, M.R., 2014a. A century of capital structure: the leveraging of corporate America. J. Financ. Econ. (forthcoming). Graham, J.R., Leary, M.T., Roberts, M.R., 2014b. How does government borrowing affect corporate financing and investment? (No. w20581). National Bureau of Economic Research. Greenwood, R., Hanson, S., Stein, J.C., 2010. A gap-filling theory of corporate debt maturity choice. J. Finance 65 (3), 993–1028. Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. J. Financ. Econ. 3 (4), 305–360. Kraus, A., Litzenberger, R.H., 1973. A state-preference model of optimal financial leverage. J. Finance 33, 911–922. McDonald, R.L., 1983. Government debt and private leverage: an extension of the Miller theorem. J. Publ. Econ. 22 (3), 303–325. Miller, M., 1977. Debt and taxes. J. Finance 32, 261–275. Myers, S.C., 1984. The capital structure puzzle. J. Finance 39, 575–592. Myers, S.C., Majluf, N.S., 1984. Corporate financing and investment decisions when firms have information that investors do not have. J. Financ. Econ. 13 (2), 187–221. Öztekin, Ö., 2015. Capital structure decisions around the world: which factors are reliably important? J. Financ. Quant. Anal. 50 (3), 301–323. Shivdasani, A., Zenner, M., 2005. How to choose a capital structure: navigating the debt-equity decision. J. Appl. Corp. Finance 17 (1), 26–35. Taggart, R.A., 1985. Secular patterns in the financing of US corporations. In: Corporate capital structures in the United States. University of Chicago Press, pp. 13–80. White, H.L., 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48, 817–838.
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL
[m3Gsc;September 21, 2016;13:56]
Finance Research Letters 0 0 0 (2016) 1–8
Contents lists available at ScienceDirect
Finance Research Letters journal homepage: www.elsevier.com/locate/frl
The effects of government borrowing on corporate financing: Evidence from Europe Yusuf Ayturk Istanbul University, Faculty of Political Sciences, Department of Business Administration, Gulhane Yerleskesi, Alemdar Mahallesi, Soguk Cesme Sokak, Fatih, Istanbul, Turkey
a r t i c l e
i n f o
Article history: Received 17 September 2015 Accepted 16 September 2016 Available online xxx JEL Classification: G32 G38 H63 Keywords: Capital structure Corporate debt Government borrowing Developed Europe
a b s t r a c t This study investigates the relationship between government borrowing and corporate financing decisions in 15 developed European countries for the period of 1989–2014. We find a robust negative relationship between government borrowing and corporate debt in developed European countries. However, we do not identify any significant relation between government debt and equity. The more important finding of our study is that longterm debt of large credit-worthy companies is more sensitive to government debt in comparison to that of small financially constrained companies. In addition to government borrowing, firm-specific factors and other macroeconomic control variables are statistically significant for corporate financing decisions. © 2016 Elsevier Inc. All rights reserved.
1. Introduction Government borrowing and its consequences have been a debatable topic all over the world, especially in the Eurozone, after the global financial crisis. Actually, government debt can play an important role to cope with economic crisis in the short-term, but in the long-term it may crowd out private investment. The current Eurozone sovereign debt crisis experience has indicated that level of government debt can affect behaviour of all economic units including financing activities of corporations. At microeconomic level, determining capital structure is one of the most important decisions in corporate finance practice and theory, because capital structure can affect the firm value. The early capital structure literature focused on the firm-specific determinants of corporate financing by ignoring macroeconomic factors. Trade-off, pecking order and market timing theories are the widely accepted capital structure theories in the literature. A detailed discussion of classical capital structure theories is presented in Frank and Goyal (2008). Recently, one of the most significant topics in capital structure literature is to add macroeconomic variables as independent variables into econometric models. Little is known about the impact of government borrowing on the corporate financing structures of non-financial companies, except findings of Graham et al. (2014a, b) and Fan et al. (2012). The main purpose of this study is to investigate the impact of government borrowing on corporate financing decisions for data set of 15 developed European countries (Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and United Kingdom). Furthermore, we aim to examine sensitivity of corporate leverage to government borrowing for both financially good and bad companies. Specifically, this study seeks E-mail address: [email protected] http://dx.doi.org/10.1016/j.frl.2016.09.018 1544-6123/© 2016 Elsevier Inc. All rights reserved.
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL 2
ARTICLE IN PRESS
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
to address the following research questions: (a) Are corporate debt of non-financial firms and government debt inversely associated in developed European countries? (b) Is there a relationship between equity and government borrowing for nonfinancial firms in developed European countries? (c) Are financial leverage policies of more credit-worthy non-financial firms more sensitive to changes in government debt in comparison to policies of less credit-worthy firms in developed European countries? We use both country-level aggregate panel data and firm-level microeconomic panel data in our econometric analyses. We find a statistically significant inverse relationship between corporate debt level and government debt level. Similarly, we also identify a statistically significant inverse connection between corporate debt issuance and government debt issuance. These relationships are robust for different corporate debt measures, additional control variables and different sample periods. However, we do not identify any relationship between equity and government borrowing. The most important finding of our study is that long-term corporate debt of large credit-worthy companies is more sensitive to changes in government debt level in comparison to that of small financially constrained companies. Our findings are consistent with studies of Graham et al. (2014a and b) and Fan et al. (2012). While Graham et al. (2014a and b) have examined a long U.S. historical data set and discussed the impacts of government debt on corporate financing and investment, Fan et al. (2012) have considered international data set covering developed European countries. The remainder of our paper is structured as follows. Section 2 explains theoretical backgrounds of our motivation, reviews the literature critically and proposes our main hypotheses. Section 3 presents our sample, data set and research methodology. Section 4 reports the results of our study and presents some implications. Lastly, Section 5 concludes this study. 2. Theoretical background and hypothesis development There are three main theories of capital structure which emphasize the importance of firm-specific factors in determining corporate financial policies. These are trade-off, pecking order and market timing theories. Trade-off theory combines tax benefits of using corporate debt and bankruptcy costs (Kraus and Litzenberger, 1973) and agency costs (Jensen and Meckling, 1976) of debt financing. Pecking order theory argues that adverse selection, agency conflicts and taxes have significant effects on capital structure decisions of firms (Myers, 1984; Myers and Majluf, 1984). The most important implication of pecking order theory suggests a specific ranking rule for new fund issuance of firms such that retained earnings is preferred to debt financing and corporate debt is preferred to equity financing (Myers, 1984). Third strand in the literature is market timing theory which asserts that managers try to time new fund issuances according to time-varying costs of equity and corporate debt. A comprehensive survey conducted by Graham and Harvey (2001) for the U.S. companies presented strong evidence that managers try to time changes in interest rates by issuing new debt when they think that interest rates are low, this phenomenon is more valid for larger companies because they have more sophisticated treasury departments. Brounen et al. (2006) conducted a similar survey for detecting capital structure practices in Europe. They also presented evidence of market timing activities for European managers, but these activities are lower in comparison to their U.S. counterparts. The recent cross-country studies have been considering macroeconomic factors which have effects on corporate financing (De Jong et al. 2008; Fan et al., 2012 and Öztekin, 2015). Graham et al. (2014a and b) have shown that government borrowing have significant effects on corporate financing decisions of the U.S. firms over the last century. Based on theoretical backgrounds of Miller (1977); McDonald (1983); Taggart (1985) and Friedman (1986); Graham et al. (2014a) explain aggregate corporate debt equilibrium that is reached at intersection between imperfectly inelastic demand and supply curves. The most important implication of an imperfectly inelastic demand curve of aggregate corporate debt is to motivate a negative association between government debt and corporate debt. As long as government debt is imperfect substitute to other securities in the financial markets, changes in government debt issuance can have effects on relative returns of other securities. In the same vein, Graham et al. (2014b, p.3) claim that changes in supply of government borrowing can affect the relative returns of competing securities in such a way that supply of closer substitutes (corporate debt) is more sensitive to variations in government debt than that of poorer substitutes (equity). Depending on this reasoning, we propose the following two hypotheses: H1. There is a negative relationship between government borrowing and corporate debt for data set of developed European countries. H2. There is no significant relationship between government borrowing and equity for data set of developed European countries. Recent empirical studies of Greenwood, Hanson, and Stein (2010) and Badoer and James (2015) focus on the connection between maturity structures of government debt and corporate debt. Results of Badoer and James (2015) imply that sensitivity of long-term debt issuances of highly rated firms to supply shocks in long-term government bond issuances are higher than other securities, because they are closer substitutes. Badoer and James (2015) have also found strong evidence that increase in supply of government debt lowers only long-term corporate debt of large credit-worthy firms, not that of small financially constrained firms, in addition, they have not identified any significant relationship between short-term corporate debt issuance and government debt supply. Similarly, Graham et al. (2014b) have found that long-term debt of credit-worthy companies is more sensitive to government borrowing in comparison to that of small financially constrained companies. In line with this theoretical reasoning and recent empirical findings on the U.S. data set, we propose the following hypothesis: Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
3
Table 1 Summary statistics. Panel A. Country-level aggregate panel data
N
Mean
Stand. Dev.
Min.
Max.
Corporate leverage variables Total debt / Total assets Net debt / Total assets Long-term debt / Total assets Short-term debt Total assets Change in total debt / Total assets (t–1) Change in long-term debt / Total assets (t–1) Change in short-term debt Total assets (t–1)
390 390 390 390 375 375 375
0.29 0.19 0.21 0.08 0.03 0.02 0.01
0.07 0.09 0.07 0.03 0.05 0.04 0.02
0.12 –0.03 0.06 0.02 –0.14 –0.11 –0.06
0.54 0.47 0.41 0.18 0.38 0.30 0.16
Other firm specific variables EBIT / Total assets Net income / Total assets Market to book asset value ratio Cash / Total assets Intangible assets / Total assets
390 390 390 390 390
0.08 0.04 1.37 0.11 0.15
0.03 0.02 0.32 0.04 0.12
–0.11 –0.15 0.85 0.02 –0.08
0.18 0.10 3.40 0.24 0.57
Government debt variables Government debt / Total assets Change in government debt / Total assets (t–1)
390 375
1.56 0.10
1.37 0.21
0.24 –0.23
8.61 1.58
Macroeconomic variables Treasury bill rate Inflation rate Equity market return Real growth rate of GDP Government expenditures / Total assets Government debt / GDP
390 390 390 375 390 390
0.05 0.02 0.08 0.02 1.16 0.66
0.04 0.02 0.25 0.02 0.90 0.31
0.00 –0.04 –0.63 –0.08 0.30 0.10
0.21 0.13 0.98 0.13 6.01 1.63
Panel B. Firm-level panel data
N
Mean
Stand. Dev.
Min.
Corporate leverage variables Total debt / Total assets Net debt / Total assets Long-term debt / Total assets Short-term debt total assets Change in total debt / Total assets (t–1) Change in net debt / Total assets (t–1) Change in long-term debt / Total assets (t–1) Change in short-term debt Total assets (t–1)
79,343 78,542 79,343 79,343 69,884 69,884 69,884 69,884
0.21 0.07 0.13 0.08 0.02 0.02 0.01 0.01
0.18 0.27 0.13 0.09 0.12 0.16 0.09 0.07
0.00 –0.83 0.00 0.00 –0.32 –0.78 –0.26 –0.28
0.85 0.77 0.65 0.54 0.79 0.75 0.59 0.40
Other firm specific variables EBIT / Total assets Market to book asset value ratio Intangible assets / Total assets
78,542 78,542 78,542
0.02 1.62 0.22
0.18 1.16 0.20
–1.31 0.52 0.00
0.35 10.11 0.88
Max.
H3. Long-term debt of large credit-worthy firms is more sensitive to government debt in comparison to that of small financially constrained firms for data set of developed European countries. 3. Sample, data set and methodology We use both country-level aggregate and firm-level individual data set of 15 developed European countries. These countries are Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and United Kingdom. We exclude financial firms from our sample as in the case of most capital structure studies. Our sample of firms includes all non-financial firms listed in the national stock exchanges of countries over the period of 1989–2014. The details of all variables used in this study are reported in appendix. In our analysis, we use different samples based on macroeconomic and firm-level factors. The first group of samples consists of aggregate country-level variables which are constructed by summing firm-level values across firms at each year for each country so that we construct country-year panel data. We use sum of total assets as scale factor to calculate country-level ratios. Table 1 Panel A provides summary statistics for country-level aggregate data. Our second group of samples consists of firm-year panel data. We winsorized each firm-level variable at 1% and 99% to deal with possible outlier problem. Table 1 Panel B shows summary statistics for firm-year unbalanced panel data. Following the methodology of Graham et al. (2014b), we use three different linear corporate debt demand equations: (1) country-year panel data with aggregate values at level; (2) country-year panel data with aggregate flow values; (3) firm-year panel data with firm-specific financial data for large and small firms.
Q Cit QG = αi + β it + δ Xit + εit Ait Ait
(1)
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL 4
QG it
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
α i is country-specific time-invariant intercept. Q Cit is sum of interest bearing corporate debt for ith country at year t and
is sum of government debt for ith country at year t. Ait is sum of total assets for ith country at year t. Xit is a vector of aggregate firm-specific and macroeconomic control variables for ith country at year t and ε it is error terms. The main variables of this study are government debt and corporate debt ratios. They are scaled by sum of total assets at each year to ensure that these variables are consistent. We use stock values of government debt and corporate debt at the end of each year. We include treasury bill yield, inflation rate, equity market return, growth rate of real GDP and government expenses/total assets ratio as macroeconomic control variables based on the study of Graham et al. (2014b); earnings before interest and tax (EBIT)/total assets, intangible assets/total assets and market value of the firm/total assets as firm-specific control variables based on the findings of Frank and Goyal (2009). Our concern is on the sign and significance of government debt coefficient which is represented by beta coefficient in the model. We expect a negative relationship between government debt and corporate debt. The second equation is presented below. C
C
j Q i,tj − Qi,t−1
Ai,t−1
= αi + β
G Q Gi,t − Qi,t−1
Ai,t−1
+ δ Xit + εit
(2)
In this equation, we use changes in government debt and changes in corporate debt scaled by first lagged total assets. Control variables are the same as we have used in Eq. (1). However, all stock variables are first differenced and scaled by first lagged total assets. Similar to Eq. (1), we expect a negative relationship between change in government debt and change in corporate debt; on the other hand, we expect to find no significant relationship between government debt issuance and equity issuance. In addition to two preceding models with aggregate data, we can use firm-level data to investigate our third hypothesis. As we have discussed in preceding section, Friedman (1986) and Graham et al. (2014b) assert that corporate debt of more credit-worthy firms are closer substitute of government debt. Thus we expect that corporate debt of large credit-worthy firms is more sensitive to changes in government debt in comparison to that of small firms. Following methodology of Graham et al. (2014b), for each country and at each year, we classify all firms into 10 groups based on 10 quantiles of their total assets so that small firms are classified in the lowest quantile and large firms are classified in the highest quantile. We use firm size as a proxy for measuring credit-worthiness depending on studies of Shivdasani and Zenner (2005) and Graham et al. (2014b). We aim to compare the effects of government borrowing on small and big firms by using the model in Eq. (3) below.
yi jt =
αi + β1 g jt × I (lowi jt ) + β2 g jt × I (highi jt ) + δ1 Xi jt × I (lowi jt ) + δ2 Xi jt × I (highi jt ) + φ1Y jt × I (lowi jt ) + φ2Y jt × I (highi jt ) + εit
(3)
yijt is corporate leverage (Total Debt, Total Net Debt, Long-term Debt, Short-term Debt) of ith firm in jth country at year t. α i is firm-specific time-invariant intercept. gjt is government debt of country j at year t. I(highijt )is a dummy variable which equals to one if the firm i in country j at year t is in the highest quantile and I(lowijt ) is a dummy variable which equals to one if the firm i in country j at year t is in the lowest quantile. Xijt is a vector of firm-specific control variables of ith firm for jth country at year t. Yjt is a vector of country-specific control variables of jth country at year t. ε it is error terms. We estimate all equations by using firm fixed-effects ordinary least squares (OLS) regression analysis. In Eq. (3), difference between β 1 and β 2 coefficients represents difference between sensitivities of corporate debt to government debt for large and small non-financial firms. We conduct a simple t-test to compare β 1 and β 2 . 4. Results and discussion Table 2 reports the results obtained from estimation of Eq. (1) for alternative models. The coefficient of government debt explanatory variable is negative and statistically significant at 1% level in model 1 which implies a negative association between government borrowing and corporate leverage. To check robustness of our results, we also use alternative definitions of corporate leverage in models 2, 3 and 4. We have detected similar statistically significant negative relationship between net corporate debt and government debt. More importantly, results of models 3 and 4 indicate that both magnitude and statistical significance of government debt ratio coefficient for long-term debt model are more negative than that for short-term debt model. As suggested by Graham et al. (2014b, p.14), we also add several additional variables which can represent corporate sector investment and supply of corporate debt. These variables are change in oil price, unemployment rate, growth in money supply, change in real assets and spread between government bond and government bill. Lastly, we check the relationship between corporate debt and government debt for two different periods. Hence, we aim to check the possible effects of current crisis period with extraordinary high levels of government debt. Table 2 reports the results of estimated three models (models 5, 6 and 7), one with additional variables and two others with different periods. Estimation results of more comprehensive model (model 5) indicate that the relationship between corporate debt and government debt is still negative and significant with similar coefficient. Our results prove that negative relationship is not specific to the period after the recent financial crisis that is characterized by high level of government borrowing. Overall, our results are consistent with findings of Graham et al. (2014b) on the U.S. data. After analysis of corporate leverage at level, we also examine the same relationship by using flow data explained in Eq. (2). Table 3 presents the results of fixed-effects panel data analysis. The estimated coefficients of total debt and long-term Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
5
Table 2 Estimation results of country-level aggregate stock data. Dependent variable
Total debt / Total assets
Net debt / Total assets
Long-term debt / Total assets
Models
1
2
3
Government debt / Total assets Macroeconomic variables T-Bill return Inflation Equity market return Real growth of GDP Government Exp. / Total assets Firm characteristics EBIT / Assets Intangible assets / Total assets Market-to-book asset ratio
–0.024 –(6.26)
∗∗∗
–0.024 –(4.16)
∗∗∗
–0.030 –(7.42)
∗∗∗
Short-term debt / Total assets 4 0.006 (2.03)
∗
Total debt / Total assets Post-2007 7
–0.029∗∗∗ –(4.27)
–0.028∗∗∗ –(3.17)
–0.022∗∗∗ –(6.22)
0.832∗∗ (2.83) –0.189 –(0.83) 0.019 (0.86) –0.145 –(0.94) 0.002 (0.18)
0.186 (1.48) –0.357∗∗ –(2.31) 0.026∗ (1.83) –0.354∗∗ –(2.62) 0.024∗∗∗ (3.33)
0.329∗∗ (2.64) 0.145 (1.23) -0.002 –(0.25) 0.047 (0.56) -0.011∗ –(1.86)
0.544∗∗∗ (3.68) –0.058 –(0.27) 0.026 (1.39) –0.398∗∗ –(2.46) 0.031∗∗∗ (6.46)
0.926∗∗∗ (8.68) –0.655∗∗ –(2.60) 0.048∗∗ (2.63) –0.867∗∗∗ –(3.74) 0.029∗∗∗ (3.66)
0.078 (1.26) 0.291∗∗∗ (4.92) –0.001 –(0.32) -0.110∗ –(1.84) 0.041∗∗∗ (5.34)
–0.592∗∗∗ –(4.07) 0.178∗∗∗ (3.29) –0.032 –(1.17)
–0.565∗∗∗ –(4.00) 0.239∗∗∗ (3.63) -0.025 –(0.80)
–0.306∗∗ –(2.37) 0.196∗∗∗ (5.02) –0.044∗ –(1.83)
–0.286∗∗∗ –(4.85) –0.018 –(0.62) 0.012∗∗ (2.41)
–0.628∗∗∗ –(6.33) 0.206∗∗∗ (3.40) –0.017 –(0.69)
–0.736∗∗∗ –(6.90) 0.240∗∗∗ (3.48) –0.020 –(0.81)
–0.519∗∗∗ –(9.69) –0.058 –(1.38) –0.030∗ –(2.05)
0.075∗∗∗ (15.77) 0.387 41.2∗∗∗ 375 15
0.039∗ (1.84) 0.201 (1.26) 0.001 (0.03) 0.020 (0.79) –0.194∗∗∗ –(3.09) 0.306∗∗∗ (8.05) 0.415 270.7∗∗∗ 353 15
0.050∗ (2.10) –0.157 –(0.92) 0.023 (0.59) 0.009 (0.24) 0.546∗∗∗ (4.28) 0.337∗∗∗ (7.90) 0.495 470.5∗∗∗ 248 15
–0.019 –(1.10) 0.057 (1.04) 0.019 (1.11) –0.008 –(0.17) –0.094∗∗∗ –(6.26) 0.378∗∗∗ (14.97) 0.640 786.1∗∗∗ 105 15
Unemployment rate Money supply growth Change in real assets Treasury bond - Bill spread
R2 F statistics N Number of countries
5
Total debt / Total assets Pre-2008 6
0.514∗∗ (2.21) –0.212 –(0.94) 0.024 (1.21) –0.307∗ –(2.09) 0.014∗ (1.82)
Additional variables Change in oil price
Constant
Total debt /Total assets
0.362∗∗∗ (10.32) 0.392 24.4∗∗∗ 375 15
0.231∗∗∗ (6.12) 0.393 129.0∗∗∗ 375 15
0.287∗∗∗ (8.53) 0.494 299.5∗∗∗ 375 15
The models are estimated by country fixed-effects panel data method. Driscoll and Kraay (1998) robust standard errors are used for heteroskedasticity and autocorrelation. t-statistics are reported in parenthesis. ∗ , ∗∗ , ∗∗∗ indicate 10%, 5% and 1% significance levels respectively. Variance Inflation Factors (VIFs) are calculated to check multicollinearity problem, all VIFs are below 5.
debt models (1 and 2) are negative and statistically significant at 5% level, whereas the coefficients of short-term debt and equity models (3 and 4) are not statistically significant. From these results, we have documented a negative association between total debt issuance and government debt issuance. More importantly, we can see that the main source of this relationship is long-term debt issuance rather than short-term debt issuance. Lastly, as questioned by Friedman (1986) and Graham et al. (2014b), we examine the corporate financing decisions of different companies in response to changes in government borrowing. In order to compare sensitivities of large and small firms’ corporate debt to government debt, we estimate Eq. (3) with firm-specific fixed-effects panel data model. The results of all models for stock and flow data are reported in Table 4. For stock data (Table 4 Panel A), results of the model with long-term debt/total assets ratio indicate that the coefficient of government debt variable is −0.275 for the subsample of large firms, on the other hand, it is 0.063 for that of small firms. To compare these two results, we use a simple t-test for the difference, −0.338, and it is statistically different from zero at 1% significance level. We have not found similar results for other model with total debt. We also compare sensitivities of corporate debt issuances of large and small firms to government debt issuances as reported in Table 4 Panel B. In the model with long-term debt issuance for large firms, the estimated coefficient of government debt issuance is −0.032, whereas the corresponding coefficient of small firms is −0.006, but not statistically significant. The difference, −0.026, is statistically significant at 5% level. Overall, these results show that long-term corporate debt levels of large firms are more sensitive to government debt levels than that of small firms. This finding is also valid for debt issuance flow data. Our results are consistent with empirical findings of Friedman (1986) and Graham et al. (2014b). The main implication of our findings is that long-term corporate debt of larger firms is closer substitute to government debt for data set of developed European countries. Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
ARTICLE IN PRESS
JID: FRL 6
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
Table 3 Estimation results of fixed–effects panel data model with aggregate flow data. Dependent variable:
Change in total debt/ Total assets 1
Change in long–term debt/ Total assets 2
Change in short–term debt/ Total assets 3
Change in Gov. Debt / Assets(t–1)
–0.059∗∗ –(2.13)
–0.060∗∗ –(2.39)
–0.009 –(1.04)
0.228∗∗ (2.47) 0.138 (0.57) 0.019 (1.05) 0.207 (1.33) –1.188∗∗∗ –(4.06)
0.105 (1.47) 0.192 (1.08) 0.003 (0.21) 0.111 (0.80) –0.875∗∗∗ –(3.32)
0.129∗∗∗ (3.62) –0.021 –(0.26) 0.008 (0.91) 0.118 (1.53) –0.369∗∗ –(2.89)
0.209 (1.59) 3.864
0.112 (1.38) 1.667
0.063 (0.93) –1.680
(0.81) –0.076∗∗∗ –(4.21) 0.008 (0.89) 0.187 5.5∗∗∗ 330 15
(0.46) –0.035∗∗ –(2.19) 0.009 (1.69) 0.137 7.6∗∗ 330 15
–(1.15) –0.028∗∗ –(2.64) –0.002 –(0.42) 0.153 5.5∗∗ 330 15
Macroeconomic variables T-Bill return Inflation Equity market return Growth of real GDP Change in Gov. Exp. / Assets (t-1) Firm characteristics Change in EBIT / assets (t–1) Change in intangible assets / Assets (t–1) Change in market-to-book asset ratio Constant R2 F statistics N Number of countries
Change in equity/ Total assets 4 0.007 (0.41) 0.140∗ (1.83) –0.260 –(0.96) 0.023 (1.42) 0.163 (1.41) –1.101∗∗ –(2.46) 0.977∗∗∗ (6.84) –0.722 –(0.25) –0.034 –(1.16) 0.017∗∗ (2.48) 0.392 14.5∗∗∗ 330 15
The models are estimated by country fixed-effects panel data method. Driscoll and Kraay (1998) robust standard errors are used for heteroskedasticity and autocorrelation. t-statistics are reported in parenthesis. ∗ , ∗∗ , ∗∗∗ indicate 10%, 5% and 1% significance levels respectively. Variance Inflation Factors (VIFs) are calculated to check multicollinearity problem, all VIFs are below 5.
Table 4 Estimation results of large and small firms’ stock data. Panel A. Stock data Total debt/Total assets Large Gov debt / Assets Control variables Constant R2 F statistics N Number of firms
Long-term debt/Total assets Small
∗∗
–0.285 –(2.07) Yes Yes 0.061 86.2∗∗∗ 12,472 2132
–0.076 –(0.71)
Difference –0.210 –(1.13)
Large ∗∗∗
–0.275 –(3.47) Yes Yes 0.062 39.0∗∗∗ 12,480 2140
Small
Difference
0.063 –(0.74)
–0.338∗∗∗ –(2.71)
Panel B. Flow data Change in total debt/ Total assets (t–1)
Gov Debt / assets Control variables Constant R2 F statistics N Number of firms
Change in long-term debt/ Total assets (t–1)
Large
Small
Difference
Large
Small
Difference
–0.043∗∗∗ –(3.64) Yes Yes 0.236 610.0∗∗∗ 12,383 2106
–0.027 –(1.26)
–0.017 –(0.89)
–0.032∗∗∗ –(3.20) Yes Yes 0.177 117.3∗∗∗ 12,406 2120
–0.006 –(0.45)
–0.026∗∗ –(2.04)
The models are estimated by fırm fixed-effects panel data method. White (1980) robust standard errors clustered by country and industry groups (45) are used for heteroskedasticity and autocorrelation. t-statistics are reported in parenthesis. ∗ , ∗∗ , ∗∗∗ indicate 10%, 5% and 1% significance levels respectively. Variance Inflation Factors (VIFs) are calculated to check multicollinearity problem, all VIFs are below 5.
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL
ARTICLE IN PRESS
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
7
5. Conclusions This study has examined the effects of government borrowing on corporate financing decisions in 15 developed European countries for the last 26 years, from 1989 to 2014. The results of our analysis show that government debt is a significant factor to determine corporate financing decisions in developed European countries. Firstly, we have detected that there is a strong negative association between corporate debt and government borrowing, whereas there is no significant relationship between equity and government borrowing. These relationships are robust for both flow and stock data. In other words, higher government borrowing levels lead to financial crowding out effect on corporate debt in developed European countries. These results should be of interest to governments and corporate managers in these countries. More importantly, we have investigated sensitivities of small and large firms’ corporate debt to government debt for data set of developed European countries. Our comparisons have revealed that long-term debt of large credit-worthy firms is more sensitive to government debt than that of small financially constrained firms. This result implies that long-term corporate debt of credit-worthy companies is closer substitute to government bonds. These companies are able to fill supply shocks in government debt of developed European countries. Acknowledgements I would like to acknowledge the financial support of Jean Monnet scholarship programme awarded by Turkish Ministry for European Union Affairs and European Union. Appendix. List and definitions of variables
Panel A. Firm-level variables Variable
Definition
Source
Units
Corporate leverage Total debt Long-term debt Short-term debt Total assets Corporate net debt issuance Corporate net long-term debt issuance Corporate net short-term debt issuance Corporate net equity issuance
Total debt / Total assets Long-term debt + short-term debt Long-term debt Short-term debt Total assets Change in total debt/Total assets (t-1) Change in long-term debt/Total assets (t-1) Change in short-term debt/Total assets (t-1) Change in book value of stockholders’ equity in excess of the change in retained earnings (Fama and French, 2005) Total debt - cash and Cash equivalents Cash and short-term investments / total assets Earnings before interest and tax / Total assets Net income / Total assets (Total liabilities + Market capitalization) / Total assets Intangible assets / Total assets Total assets - (Net Plant, property and equipment (Net) + Accounts receivables + Cash and Short term investments + Inventory)
Datastream Datastream Datastream Datastream Datastream Datastream Datastream Datastream Datastream
% Billion Billion Billion Billion % % % %
Datastream Datastream Datastream Datastream Datastream Datastream Datastream
Billion Euro % % % % % Billion Euro
1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014
Variable
Definition
Source
Units
Period
Government leverage
Government debt / Total assets of all non-financial companies Change in government debt / total assets Year-end unemployment rate Annual change in year-end industrial production index Year-end M1 money supply Annual change in year-end consumer price index Year-end local oil price Annual change in MSCI country index Year-end 10-year indicator bond yYield Year-end 3-month indicator bill yYield Nominal gross domestics product Real gross domestics pProduct based on 2010
Oxford economics
%
1989–2014
Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford Oxford
% % % Billion Euro % EU % % % EU EU
1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014
Net debt Liquid assets Profitability Net profitability Market to book asset value Intangibility Intangible assets
Period
Euro Euro Euro Euro
1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014 1989–2014
Panel B. Macroeconomic variables
Government net debt issuances Unemployment rate Industrial production Money supply Inflation Price of oil Equity market return Treasury bond yield Treasury bill yield Gross domestic product Real gross domestic product
economics economics economics economics economics economics economics economics economics economics economics
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018
JID: FRL 8
ARTICLE IN PRESS
[m3Gsc;September 21, 2016;13:56]
Y. Ayturk / Finance Research Letters 000 (2016) 1–8
References Badoer, D.C., James, C.M., 2015. The determinants of long-term corporate debt issuances. J. Finance (forthcoming). Brounen, D., De Jong, A., Koedijk, K., 2006. Capital structure policies in Europe: survey evidence. J. Bank. Finance 30 (5), 1409–1442. De Jong, A., Kabir, R., Nguyen, T.T., 2008. Capital structure around the world: the roles of firm-and country-specific determinants. J. Bank. Finance 32 (9), 1954–1969. Driscoll, J.C., Kraay, A.C., 1998. Consistent covariance matrix estimation with spatially dependent panel data. Rev. Econ. Stat. 80 (4), 549–560. Fama, E.F., French, K.R., 2005. Financing decisions: who issues stock? J. Financ. Econ. 76 (3), 549–582. Fan, J.P., Titman, S., Twite, G., 2012. An international comparison of capital structure and debt maturity choices. J. Financ. Quant. Anal. 47 (01), 23–56. Frank, M.Z., Goyal, V., 2008. Trade-off and pecking order theories of debt. In: Eckbo, E. (Ed.). In: Handbook of Corporate Finance, vol. 2. Elsevier, North Holland, pp. 135–202. Frank, M.Z., Goyal, V.K., 2009. Capital structure decisions: which factors are reliably important? Financ. Manage. 38 (1), 1–37. Friedman, B.M., 1986. Implications of government deficits for interest rates, equity returns and corporate financing, (No. c7940), National Bureau of Economic Research, 67–90. Graham, J.R., Harvey, C.R., 2001. The theory and practice of corporate finance: evidence from the field. J. Financ. Econ. 60 (2), 187–243. Graham, J.R., Leary, M.T., Roberts, M.R., 2014a. A century of capital structure: the leveraging of corporate America. J. Financ. Econ. (forthcoming). Graham, J.R., Leary, M.T., Roberts, M.R., 2014b. How does government borrowing affect corporate financing and investment? (No. w20581). National Bureau of Economic Research. Greenwood, R., Hanson, S., Stein, J.C., 2010. A gap-filling theory of corporate debt maturity choice. J. Finance 65 (3), 993–1028. Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. J. Financ. Econ. 3 (4), 305–360. Kraus, A., Litzenberger, R.H., 1973. A state-preference model of optimal financial leverage. J. Finance 33, 911–922. McDonald, R.L., 1983. Government debt and private leverage: an extension of the Miller theorem. J. Publ. Econ. 22 (3), 303–325. Miller, M., 1977. Debt and taxes. J. Finance 32, 261–275. Myers, S.C., 1984. The capital structure puzzle. J. Finance 39, 575–592. Myers, S.C., Majluf, N.S., 1984. Corporate financing and investment decisions when firms have information that investors do not have. J. Financ. Econ. 13 (2), 187–221. Öztekin, Ö., 2015. Capital structure decisions around the world: which factors are reliably important? J. Financ. Quant. Anal. 50 (3), 301–323. Shivdasani, A., Zenner, M., 2005. How to choose a capital structure: navigating the debt-equity decision. J. Appl. Corp. Finance 17 (1), 26–35. Taggart, R.A., 1985. Secular patterns in the financing of US corporations. In: Corporate capital structures in the United States. University of Chicago Press, pp. 13–80. White, H.L., 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48, 817–838.
Please cite this article as: Y. Ayturk, The effects of government borrowing on corporate financing: Evidence from Europe, Finance Research Letters (2016), http://dx.doi.org/10.1016/j.frl.2016.09.018