Capital structure in Spanish firms : the effect of the financial crisis

(1)

1

MASTER’S THESIS

Capital Structure in Spanish Firms:

The Effect of the Financial Crisis

Carolina López-Quiles Centeno

10604340

Supervised by Prof. Dr. S.J.G van Wijnbergen

Abstract

In this paper I empirically study the finance behavior of Spanish firms. The main focus of

this paper is to evaluate whether the financial crisis had an impact on the way Spanish firms

choose their funding sources. I test for different capital structure choice theories, and I use an

Error Correction Model to prove that Spanish firms tend to adjust to a target leverage, much

in line with trade-off theory. I find empirical evidence that after late 2008, the speed of

adjustment towards target leverage decreases. This can be explained by credit constraints and

supply-side rationing in credit markets, due to the adverse economic environment.

(2)

2

“In theory, chief financial officers spend idle moments admiring charts of the

idealised firm's optimal capital structure. […].In the real world, however, many

managers burnt their textbooks long ago and steadily increased their firms' leverage.

[…](However), the psychological scars will run deep for the private firms that bear a

disproportionate burden of overall corporate debt. Many face bankruptcy, with the

destruction of value that entails.[…] It used to be that equity, as well as lunch, was

for wimps. Not anymore.” The Economist, April 30th 2009.

Capital structure can be defined as the

composition

of a firm’s resources, divided broadly in debt and equity, the latter one being either internal or external. Firms can choose how much debt and equity to issue for their operating needs. This choice is of utmost importance, not only because it could be a determinant factor in the survival of the firm, but also because it is used as an indicator of performance by outside investors.

Historically, there have been many attempts to provide a theoretical explanation of capital structure. However, heterogeneity across firms, sectors and markets, leads to different capital structure choices, which makes it difficult to find a theoretical framework that fits the empirical findings.

The study of capital structure is of great relevance both from a microeconomic and a macroeconomic perspective. From a microeconomic point of view, it is fundamental for firms to understand the implications of different capital structure choices. Leverage allows firms to increase their operating activity today using resources for which they will only be liable tomorrow. However, largely indebted firms may face problems such as debt overhang, and ultimately bankruptcy. Equity on the other hand has the advantage of loss absorption capacity. However, issuing too much equity can be inefficient and very costly. Firms’ survival may depend on their ability to find the perfect balance between finance sources.

From a macroeconomic standpoint, understanding how firms finance themselves is fundamental for understanding the mechanics of capital markets, which in turn helps designing optimal policy and regulation.

(4)

4

Modigliani and Miller (1958) argued that capital structure is irrelevant for the firm’s value under complete markets. However, observed market imperfections, and the presence of other factors such as tax shields, managerial reputation, and bankruptcy costs have led researchers to elaborate new theories that describe finance behavior. The three most relevant theories are trade-off theory, pecking order theory and market timing theory. Evidence has been found that supports each theory individually, depending on the sample of firms, the country and the time frame under study. Since evidence is mixed, neither of the theories seems to dominate in the empirical research arena.

In this paper I attempt to test the effect on the financial crisis on the finance behavior of Spanish firms. I approach this matter by testing the three main theories of capital structure on a sample of listed Spanish firms, and I study the impact of the crisis by looking for a break in the trend after late 2008. The paper is structure as follows: Section II provides a theoretical

framework for capital structure theories, along with an overview of the previous literature on each of them. Section III describes the data and summary statistics for the analysis. Section IV

provides different methodologies to test for each theory and to evaluate break points during the crisis. Section V concludes.

II. LITERATURE REVIEW Theories of capital structure

Attempts to explain capital structure choice go back to Modigliani and Miller’s 1958 Theory of Investment, in which they argue that in complete and perfect capital markets, capital structure choice is irrelevant. If markets are perfect, there is no profit to be made by arbitrarily switching from equity financing to debt financing in the long run. If equity financing is achieved through shares and debt financing is achieved through bonds, and if the market prices of said securities are different at a point in time allowing for arbitrage, then the process of arbitrage itself will bring these prices to be equal in equilibrium1_.

Several theories of capital structure have been discussed since. The more traditional ones are

1_{If all securities are assumed to yield the same risk adjusted return, investors will want to sell the}

ones with the highest price and buy more of the low price ones, bringing the prices to an equilibrium.

(5)

5

pecking order theory and trade-off theory, although a relatively more modern one, market timing, has received special attention over the past decade.

1. Trade-Off Theory

According to this theory, firms analyze the trade-off between the costs and benefits of equity and debt and set a target debt ratio accordingly.

As explained by Myers (1984), the firm has to balance the value of interest tax shields against the costs of debt, those being costs of bankruptcy and financial distress. Furthermore, excessive debt may give way to the so-called debt overhang problem (Myers, 1977). Highly leveraged firms may find it hard to borrow and may therefore forgo investment opportunities that could very well be profitable and pay for themselves. Firms with high growth opportunities are generally more prone to set lower leverage ratios in order not to lose investment opportunities.

This theory has established that firms are supposed to substitute equity and debt until the value of the firm is maximized. Whether this hypothesis holds empirically is another thing. If trade-off theory accurately describes capital structure choice, one would expect similar firms to have similar debt ratios. However, there is a wide variation in actual debt ratios, as shown by Myers (1984). One possible explanation is adjustment costs. Moving towards the optimum may be costly and time consuming. Given that adjustment costs are generally found to be relatively small, one must conclude that the reason behind different debt ratios within similar firms must stem from the fact that firms are actually far from their optimum leverage. This is bad news for trade-off theory, because it means that firms are not paying much attention to the benefits and costs of equity and debt, but rather choosing capital structure based on other criteria.

In a more recent study of this theory, Fama and French (2012) find that, although leverage seems to move towards a target ratio, the reversion is quite slow, leaving the question of whether trade-off theory predictions are irrelevant. Furthermore, they point out that ‘there is evidence that target leverage is not a first-order consideration in financing decisions’.

2. Pecking Order Theory

Pecking order theory suggests that firms issue debt and equity in a specific order or preference. Firms prefer internal finance, so when funding is needed, they first choose to raise internal equity. If additional funding is needed, firms turn to debt issues, because they are

(6)

6

only be issued as a last resort. Pecking order theory in this sense predicts that firms will issue securities in order of preference from pure debt to pure equity, along a continuum of hybrids (for instance, convertible bonds).

However, as Myers(1984) points out, issue costs are too small for them to yield any

significant difference in capital structure from that predicted by trade-off theory. A more modern approach to pecking order suggests that the main reason why firms prefer internal to external finance is not issue costs, but rather costs of asymmetrical information and adverse selection in capital markets (de Bie and de Haan, 2007).

In addition, some point out that firms’ reliance on internal finance is a consequence of the separation of ownership and control. External finance requires the managers to be disciplined by the capital markets.

Many empirical studies, such as the ones carried out by Sunder and Myers (1999), Huang and Ritter (2009), and Lemmon and Zender (2012), have corroborated the hypotheses behind pecking order theory, while some others, such as Frank and Goyal (2003) and Fama and French (2012) have refuted them, or found several limitations in them. For instance, Frank and Goyal argue that pecking order only does well in explaining capital structure in larger firms, while it seems that small firms do not raise funds by order of preference. They find this result quite puzzling, since theory would predict that smaller firms face larger adverse selection and information asymmetry problems, and hence are expected to be greater advocates of pecking order.

3. Market timing Theory

Market timing theory is a more modern approach to capital structure theory. It suggests that firms choose their capital structure by timing their debt and equity issues to their market

performance. In other words, firms will issue equity when their stocks are overvalued (when their market-to-book ratios are high) and will issue debt when their stocks are undervalued (when the market-to-book-ratios are low).

While pecking order theory predicts that external equity is too costly and thus a last resort for firms, there is evidence that equity issues are not necessarily more expensive than debt issues, depending on risk adjustment and the market price of stock.

(7)

7

Market timing has gained relevance over the last years and has been empirically tested in a wide number of studies. For instance, Baker and Wurgler (2002) find a strong and persistent negative relationship between market-to-book ratios and leverage in US firms, that is transmitted mainly through new equity issues. This indicates not only that firms time their capital structure to the market, but also that the effects of such timing are long-lasting. Huang and Ritter (2009), and Fama and French (2012) further corroborate this hypothesis.

While the evidence of market timing behavior in the US seems rather widespread, the same cannot be said for European firms. De Bie and de Haan (2007) find that the relationship between market-to-book ratios and leverage is generally non-significant for Dutch firms. Furthermore, Bruinshoofd and de Haan (2011) find no evidence of market timing in UK firms, as well as in continental Europe. It seems that European firms behave more in line with trade-off and pecking order theory.

III. DATA

My sample consists of 82 Spanish listed non-financial firms, which is roughly 70% of all non-financial firms listed in the Madrid stock market. An alphabetical list of the firms in the sample can be found in Appendix A. Financial firms were excluded because of the obvious differences in financing methods.

In order to construct the panel for my analysis, I gathered all the balance sheets and income statements, as well as the stock market data and dividend payments for the 82 firms on a yearly basis, from 2005 to 2012.

The sample period is interesting to study because of the anomaly caused by the presence of a financial and economic crisis.

All data are downloaded from SABI2_{(Sistema de Análisis de Balances Ibéricos). This}

database is developed by Bureau van Dijk3_.

2_{http://www.bvdinfo.com/en-gb/our-products/company-information/national/sabi} 3_{http://www.bvdinfo.com/en-gb/home}

(8)

8

Table 1: Summary Statistics

4

Variable Observations Mean Std. Deviation Min Max

Book Leverage 656 0.6854177 0.2217469 0.0406116 1 Market leverage 499 0.5364143 0.2412977 0.045635 1 Market-to-Book 565 1.508284 1.267565 30375216 12.57896 Debt Issues 562 0.0496288 0.1448128 -0.8691323 0.910921 Equity Issues 566 0.0039021 0.033389 -0.1077695 0.6916288 Profitability 648 0.0952416 0.0961 -0.358514 0.5247624 Tangibility 648 0.3065859 0.2140545 0.0035401 0.9204762 Size 646 13.27225 1.932995 7.072422 17.95605

In general, the summary statistics are very similar to those shown by de Bie and de Haan (2007) for Dutch firms, with the exception of equity issues, which are notably lower in the case of Spanish firms. Having a deeper look into the balance sheets, I observed that most firms in my sample did not issue any new equity during the sample period at all, and the few firms that did, only did so once or twice. Most of the increases in book equity come from retained earnings and reserves. This can be a preliminary sign of either pecking order or trade-off theory. Acedo Ramírez et al. (2012) found evidence of trade-off theory in Spanish firms from 2000 to 2008, while Fernández, de Rojas and Zuliani (2004) found slight evidence that Spanish firms followed pecking order from 1995 to 2002. It is interesting to see which result holds for my sample period, especially in light of the financial crisis.

By looking at the correlations between the variables, it can be seen that market-to-book ratios are only weakly correlated with book leverage, although they are more strongly correlated with market leverage. Profitability is negatively correlated with both book and market leverage, which is to be expected since we have already observed that Spanish firms rely heavily on retained earnings. A word of caution regarding the correlation between profitability and leverage is needed: since the data is end-of-period data, profitability will be included in the denominator of the leverage ratio. Firm size is positively correlated with both book and market leverage, although only weakly. The correlation between tangibility and leverage is almost negligible.

(9)

9

Table 2: Correlation Matrix

Book Leverage Market Leverage Market-to-Book Profitability Tangibility Size

Book Leverage 1 Market Leverage 0.6579 1 Market-to-Book -0.1128 -0.5820 1 Profitability -0.2363 -0.5401 0.6933 1 Tangibility -0.1560 -0.0020 -0.1860 0.0513 1 Size 0.3822 0.2033 0.0065 0.1759 0.0460 1 IV. ANALYSIS

This section is devoted to testing the effect of the financial crisis on the capital structure of Spanish firms. For this, different theories of capital structure are tested on the aforementioned sample of firms. First, I will analyze the effect of market-to-book ratios on leverage. Second, I will run a joint test for the presence of either trade-off or pecking order behavior. In both analyses, a dummy for the crisis is used to test for breaks after 2008.

All regressions are run both on book leverage and market leverage. There is an ongoing debate about which of them is more accurate to judge a firm’s debt capacity, but both measures are widely used by investors, rating agencies, and firms themselves.

1. Do market-to-book ratios affect leverage?

Baker and Wurgler (2002) proved that for American firms, market-to-book ratios affect leverage and they do so mainly through equity issues. This is proof that there is market timing behavior in the US. However, the idea of market timing in the case of Spanish firms can be disregarded in light of the observation that there are little to no equity issues. It is still interesting, however, to check whether market-to-book ratios still affect leverage through any other channel. Thus, I replicate Baker and Wurgler’s (2002) method, not to test for market timing, but for other possible connections between market-to-book and leverage.

Their regression analyzes the effects of market-to-book ratios on the annual change in leverage. The change in leverage is regressed on lagged market-to-book ratio and three control variables. The three control variables are profitability, measured as EBITDA over assets, tangibility, measured as power plant and equipment over assets, and size, measured as the log of

(10)

10

sales. The reason to include these three variables is to control for the fact that larger, more profitable and more tangible firms may have easier access to debt because creditors will regard them as less risky than smaller, less profitable firms, because of their greater ability to post collateral. This may influence their capital structure choice.

The regression reads as follows:

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )

Note that this regression can be interpreted as an Error Correction Model of the form:

( )

( ) [( ) ( ) ] ( ) where ( )

is a target level of leverage and is equal to

(

)

(

)

(

)

(

)

( )

(3)

This Error Correction Model will be analyzed more formally in the next section to test for Trade-off theory, where a specific target level of leverage will be calculated. For the time being, let us analyze the impact of market-to-book ratios on firms’ leverage.

Results for this regression are shown in Table 3. A few conclusions can be drawn from this table. First, lagged market-to-book ratio is generally not significant. As expected, no evidence of market timing behavior can be deduced from these results. This result is contrary to what Baker and Wurgler (2002) found for American firms, but it goes in line with the results found by de Bie and de Haan (2007) for Dutch firms under the same regression specification.

(11)

(12)

12

Second, it is very interesting that the most significant variable across the whole sample period and also in each year individually is lagged book leverage. This result may be indicative of trade-off behavior, according to which, these firms would be reducing their debt year by year after realizing their level of debt was undesirably high.

Furthermore, profitability is overall very significant in the whole sample period, but especially after 2008. Profitability affects leverage negatively, and even more so after the crisis. This may be a sign of leverage constraints. It has been established that Spanish firms generally prefer using retained earnings instead of debt as a source of finance, but the fact that this is accentuated after the crisis, indicates that Spanish firms were forced to reduce their leverage ratios due to shrinking credit conditions.

These credit constraints seem to have a relatively smaller effect on larger firms, as indicated by a positive and significant coefficient for the size variable. Larger firms may have easier access to debt because creditors trust them more.

Credit constraints in Spain after the crisis have been a widely discussed topic in

international economic fora, and the IMF has warned about the severity of this issue on multiple occasions. The problem of credit constraints during the crisis is exacerbated by the fact that Spanish firms rely more heavily on bank loans than in other countries. The European Central Bank (2010), documented that in 2006, bank loans to Spanish firms accounted for 86% of GDP, compared to 62% un the EU, and 35% in the UK. The poor state of some Spanish banks during the crisis was critical in determining the ability of firms to raise funds.

A rather interesting result is that market-to-book seems to affect market leverage

positively, as opposed to what is expected from the theory. A possible explanation for this is that, since Spanish firms don’t issue any equity, high market-to-book ratios are not seen as an

opportunity to issue cheap equity, but rather as an indicator of solvency. When market-to-book ratios are high, creditors perceive that the firm is performing well and are more likely to give them credit because they expect them to be able to pay back. This is a sign of supply-side rationing in credit markets.

(13)

13 1.1. The effect of the crisis

In light of the interesting result that market-to-book ratios affect market leverage positively, I run the following regression adding a dummy for the crisis period, to check whether this effect is due to a break in the trend during the crisis:

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )

Table 5: The effect of the crisis

OLS regression of book and market leverage on market to book ratio, an interaction term, profitability, tangibility and size. The interaction term is computed by multiplying market-to-book ratios and a dummy variable depicting the crisis period, in order to isolate the effect of the crisis. Profitability is measured as EBITDA over total assets, tangibility is measured as fixed assets over total assets, and size is measured as the logarithm of net sales.

Change in Book Leverage Change in Market Leverage Market-to-book ratio 0,002 0,012** (0,006) (0,006)

Market-to-book ratio interacted with crisis dummy -0,000 -0,017***

(0,007) (0,005) Profitability -0,247*** -0,378*** (0,091) (0,078) Tangibility -0,002 0,007 (0,031) (0,022) Size 0,007** 0,006** (0,004) (0,003) Lagged Leverage -0,292*** -0,133*** (0,032) (0,026) Constant 0,124** 0,054 (0,051) (0,038) Number of observations 485 415 Adjusted R2 0,142 0,115 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors reported in brackets

(14)

14

Again, book leverage is mainly affected through profitability and size. The coefficient for market-to-book ratio in the regression for market leverage is, as before, positive for the entire sample period. Interestingly, the coefficient for the interaction variable with the crisis dummy is negative and statistically significant. As argued before, a positive relationship between market-to-book ratios and leverage may suggest that creditors use market values as an indicator of firm performance in order to make their credit decisions. However, the interaction term of market-to-book ratio and the crisis dummy is negative and greater in absolute value than the coefficient for market-to-book ratio alone. This means that during the crisis, the coefficient for market-to-book ratios was slightly negative. The fact that during that period, market-to-book ratios had the opposite effect on leverage as they had during normal times may indicate distress borrowing on the part of firms. With the crisis, market-to-book ratios decreased, but leverage, which was expected to decrease during normal times, actually increased.

Another interesting thing to test is whether by the end of 2012, the stress in the market had disappeared, and finance behavior was back to normal. In order to test this, I add a second interaction term with a dummy variable that takes value 0 from 2005 to 2011, and value 1 only for 2012. Results are shown in Table 6.

Table 6: Addition of 2012 Dummy

OLS regression of book and market leverage on market to book ratio, interaction terms, profitability, tangibility and size. The first interaction term is computed by multiplying market-to-book ratios and a dummy variable depicting the crisis period, in order to isolate the effect of the crisis. The second interaction term is computed by multiplying market-to-book ratios and a dummy variable depicting the year 2012.

Change in Book Leverage Change in Market Leverage Market-to-book ratio 0.002 0.012** (0.006) (0.006)

Market-to-book ratio interacted with crisis dummy -0.003 -0.015***

(0.007) (0.005)

Market-to-book ratio interacted with 2012 dummy 0.015 -0.007

(0.012) (0.009) Profitability -0.245*** -0.378*** (0.091) (0.078) Tangibility 0.000 0.006 (0.031) (0.022) Size 0.007* 0.006** (0.004) (0.003) Lagged Leverage -0.291*** -0.132*** (0.032) (0.026) Constant 0.124** 0.053 (0.051) (0.038) Number of observations 485 415 Adjusted R2 0.143 0.114 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors reported in brackets

(15)

15

The coefficient for the interaction term of market-to-book ratios with the 2012 dummy is not statistically significant, which means that the year 2012 was not different from the pre-crisis period. It seems that market behavior was back to normal.

It has been established that market-to-book ratios affect market leverage in different ways during normal times and during the crisis. It would be interesting to look more closely at the channel through which this effect is transmitted. In order to check this, I split the change in leverage in three components following Baker and Wurgler(2002). These components are equity issues, retained earnings and the residual change in leverage due to asset growth:

( ) ( )

(

) (

) [

(

)]

(5) ( ) ( ) ( ) ( ) ( ) ( )

Regressing each component separately on the variables used in the previous regression yields the results shown in Table 7.

(16)

16

As expected, market-to-book ratios don’t affect leverage through equity issues. It seems that during the crisis, market-to book ratios have a stronger negative effect on retained earnings, and a weaker but still negative effect on the residual change due to asset growth.

Interestingly, as shown in Graph 1, asset growth and market-to-book ratios seem to move together up until 2008, and then they start moving in opposite directions.

Table 7: Effects of market-to-book ratios in different components of leverage

OLS regression of the different components of leverage on market to book ratio, an interaction term, profitability, tangibility and size. The interaction term is computed by multiplying market-to-book ratios and a dummy variable depicting the crisis period, in order to isolate the effect of the crisis. Profitability is measured as EBITDA over total assets, tangibility is measured as fixed assets over total assets, and size is measured as the logarithm of net sales.

Equity Issues Retained Earnings Residual Change in Leverage Market-to-book ratio 0,001* -0,006* -0,071*** (0,001) (0,003) (0,013)

Market-to-book ratio interacted with crisis dummy 0,000 -0,010*** 0,053*** (0,001) (0,004) (0,014) Profitability -0,029*** 0,388*** -0,036 (0,011) (0,048) (0,184) Tangibility 0,002 -0,027 -0,035 (0,004) (0,017) (0,064) Size -0,001*** -0,002 0,008 (0,000) (0,002) (0,002) Constant 0,017*** 0,031 -0,047* (0,006) (0,027) (0,024) Number of observations 485 485 485 Adjusted R2 0,027 0,129 0,089 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors reported in brackets

(17)

17

One final conclusion can be derived from this analysis: market-to-book ratios mainly affect leverage through their effect on assets (the denominator of the leverage ratio) and not debt(the numerator). Since market timing theory was rejected for this dataset in light of the observation that Spanish firms generally do not issue new equity, and since the only other relevant channel through which market-to-book ratios affect leverage is through assets and not debt, it is interesting to move forward to analyzing the presence of pecking order or trade-off theory.

2. Towards solving the puzzle: Trade-off or Pecking order?

In order to formally assess the best theoretical fit to the observed capital structure choices, I use Sunder and Myer’s (1999) model.

A good test for pecking order behavior, according to them, is to check whether firms issued debt after realizing their internal cash flows were inadequate for their investment and dividend commitments.

(18)

18 They define the funds flow deficit as:

(6) where denotes dividend payments, denotes investment, denotes the net increase in

working capital, is the amount of debt due in the present period, and are operating cash flows.

The pecking order hypothesis is then tested by regressing the amount of debt issued (or paid back) on the fund flow deficit.

The regression reads:

( ) If pecking order theory holds perfectly, it is expected that =0 and =1.

It is worth noting that the regression is not an accounting identity because by construction, DEF does not include equity. This is because in a strict sense, pecking order predicts that equity will never or only very rarely be issued. Sunder and Myers argue that a broader pecking order hypothesis should accommodate for some equity issues. However, even though simplistic, this regression could appropriately describe financing behavior in the case of Spanish firms, since very few equity issues are observed.

Next, in order to check for trade-off theory, an Error Correction Model is used. This model infers that changes in the debt ratio are explained by current deviations from a target ratio. If trade-off theory holds, firms must know their target ratio, and try to steer back to it when random events push them far from it. Therefore, the regression line is:

[(

)

]

(8) where (

) denotes the target leverage.

If , this would be indicative of adjustment towards a target level. However, it is also expected that , due to the presence of adjustment costs. Steering the leverage ratio back to its target level can be costly and thus the adjustment will not be immediate and complete.

(19)

19

The actual target level of debt for each firm is unobservable, but there are a few ways of calculating an estimate.

For instance, following Baker and Wurgler’s (2002) regression used in the previous section, we can define target leverage as:

(

)

(

)

(

₎

(

₎

( )

(9) Running this regression on a first step and then using the residuals one can compute a linear prediction of the target leverage. The regression used to predict target leverage is a Tobit regression that allows to limit the value of target leverage between 0 and 1. Results for this regression can be found in Appendix C.

Table 8 reports the regression results for the funds deficit variable for pecking order and the error correction variable for trade-off theory, both run separately and jointly. As can be seen, the fund deficit variable is never significant, and it’s R-squared when analyzed individually is negligible. This is enough to reject the presence of pecking order behavior. The results for the trade-off theory test are more promising. The constant term is not significantly different from zero for book leverage, as the hypothesis predicted, although it is slightly positive in the case of market leverage. The coefficient for the error correction variable is positive and significant at the 1% level. In all cases, both for book and market leverage, and both for the individual and the joint tests, this coefficient smaller than 1.

(20)

20

Table 8: Tests for Pecking order and Trade-off Theory

OLS regressions showing the results for error correction and pecking order models. The dependent variable is the change in either book or market leverage. The independent variables are the coefficients for each model as stated in the main text.

Change in Book Leverage Change in Market Leverage

Constant 0,000 -0,001 -0,008 0,043*** 0,033*** 0,037*** (0,007) (0,006) (0,007) (0,006) (0,005) (0,006) Pecking order coefficient -0,000 -0,000 -0,000 -0,000 (0,000) (0,000) (0,000) (0,000) Error correction coefficient 0,290*** 0,257*** 0,122*** 0,133*** (0,032) (0,035) (0,027) (0,035) Number of observations 374 486 361 322 415 322 Adjusted R2 -0,003 0,141 0,128 -0,002 0,045 0,040 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors reported in brackets

(21)

21

The overall conclusion from Table 8 is that trade-off theory seems to fit the data better than pecking order theory. In addition, a coefficient for the adjustment variable lower than 1 indicates there are adjustment costs. Firms will tend towards a target level of debt, but adjustments will only be partial.

Furthermore, using this Error Correction Model, we can study the long and short term drivers of leverage. Equation (8) can be rearranged to be:

(

)

( )

(10)

Table 9: Long and Short Term Coefficients for Leverage

OLS regression of leverage on target leverage and lagged leverage. This regression shows the long-term and short-term drivers of leverage according to trade-off theory.

Book Leverage Market Leverage

Target Leverage 0.292*** 0.203*** (0.032) (0.027) Lagged Leverage 0.706*** 0.851*** (0.032) (0.027) Number of observations 485 415 Adjusted R2 0.961 0.975 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors reported in brackets

In this regression, the coefficient for lagged leverage can be interpreted as the short term driver of changes in leverage, and the target leverage coefficient can be interpreted as the long term tendency towards this target.

2.1. The effect of the crisis

Seeing as the pecking order coefficient is insignificant, I will move on to evaluate the impact of the crisis on the error correction coefficient only. As in the previous section, an interaction term with a dummy variable is added to equation (8). This dummy takes value 0 from 2005 to 2008 and value 1 from 2009 to 2012.

(22)

22 The regression now reads:

[(

)

]

[(

)

]

(11)

Table 10: Effect of the Crisis on the Error Correction Coefficient

OLS regressions showing the results for an error correction model with a crisis interaction term. The dependent variable is the change in either book or market leverage.

Change in Book Leverage

Change in Market Leverage

Error correction coefficient 0.378*** 0.234***

(0.055) (0.044)

Error correction coefficient interacted with crisis dummy -0.125* -0.134**

(0.067) (0.055) Constant -0.001 0.029*** (0.006) (0.005) Number of observations 485 415 Adjusted R2 0.153 0.079 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors reported in brackets

Table 10 shows an interesting result. The sign of the interaction term with the crisis dummy is negative, although smaller in absolute value than the error correction coefficient. This means that during the crisis, the error correction coefficient was positive but smaller than during normal times. There was still adjustment towards a target leverage, but the speed of adjustment was significantly slower. This result is coherent with Cook and Tang (2010), who find that ‘

movement toward target

leverage ratios is impeded by poor economic conditions’. Furthermore,

Drobetz et al. (2014) also document the existence of an adjustment process towards a target leverage in a sample of firms from the G-7, and find that bad macroeconomic states lead to lower speed of adjustment, especially in financially constrained firms.

This is consistent with the intuition that, given the credit constraints in Spain during the crisis, and the poor performance of the publicly owned part of the banking sector, firms found it increasingly difficult to move their leverage towards the desired target.

(23)

23 V. CONCLUSION

It is widely accepted that, contrary to Modigliani and Miller (1958), capital structure choice does matter, mainly because of market imperfections. However, optimal capital structure choice is still to a big extent a puzzle. Several attempts to theoretically describe the way firms choose their financing options have been made in the past. The three main theories of capital structure are trade-off theory, which states that firms evaluate the costs and benefits of equity and debt and set a target leverage ratio accordingly; pecking order theory, which argues that firms prefer internal to external finance, and within external, they prefer debt to equity; and market timing theory, which poses that firms time their debt or equity issues depending on the market-to-book ratios of their stock. Evidence regarding these theories is mixed. The best theoretical fit for the observed capital structures seems to vary across firm samples and time periods.

In this paper I attempted to test these three theories on a sample of Spanish firms from 2005 to 2012. This time period is particularly interesting because of the presence of the financial crisis. This paper therefore focused on studying the potential changes in finance behavior that happened in Spain after late 2008.

It appears from the analysis performed in this paper that market-to-book ratios have very little to do with book leverage, although they do affect market leverage. I found a reversal effect caused by a break during the crisis. Market-to-book ratios were proven to affect market leverage positively, due perhaps to the fact that creditors see market performance as an indicator of the firm’s solvency, and therefore give more credit to firms with higher market-to-book ratios. However, the sign of the coefficient switches from positive to negative after late 2008, when the financial crisis started to have effects in the Spanish economy. An explanation for this is that creditors no longer perceived market value to be a good indicator of solvency, and credit started shrinking.

Analyzing this effect more closely, I discovered that the channel through which market-to-book ratios affect leverage is mainly through growth in assets and not through changes in total debt. This is further prove that market-to-book ratios in the case of Spanish firms have little to no effect on the amount of debt issued.

After dropping the case for market timing and for other effects of market-to-book ratios in leverage, I moved on to testing the presence of pecking order and trade-off behavior. Using a funds deficit variable to proxy for pecking order, and an Error Correction Model for trade-off theory, I found that the best theoretical fit for Spanish firms is trade-off theory. Spanish firms set a target leverage taking into account the costs and benefits of debt, and when an event moves them away from this

(24)

24

target, they tend to steer back towards it. This adjustment, however, is not immediate and is not complete, due to the presence of adjustment costs. Trade-off theory as a best explanation for Spanish firms’ financing behavior has been proven by Acedo Ramírez et al.(2012).

Again, after 2008 the pattern seems to change. The error correction coefficient -used to

measure adjustment to target leverage- was significantly smaller during the crisis than in normal times. During the crisis, Spanish firms struggled to maintain their finances afloat, and the speed of

adjustment towards target leverage slowed down, due perhaps to the need to resort to distress borrowing to cover for the losses originated by the economic turmoil.

To sum up, it seems that capital structure theories perform quite differently in explaining observed capital structure choices depending on the country and time period under evaluation. Furthermore, it seems that significant events such as the financial crisis, tend to change the pattern followed by firms to make such choices. It can be concluded that, although theoretically powerful, these theories cannot be considered inherently applicable to all samples, and should not be

(25)

25 APPENDIX A

List of firms included in the study:

ABENGOA.SA INDUSTRIA.DE.DISEÑO.TEXTIL.SA

ABERTIS.INFRAESTRUCTURAS,.SA INVERFIATC.SA

ACCIONA,.SA LABORATORIOS.FARMACEUTICOS.ROVI.SA

ACERINOX,.SA LINGOTES ESPECIALES, SOCIEDAD ANONIMA

ACS,.ACTIVIDADES.DE.CONSTRUCCION.Y.SERVICIOS,.SA LIWE ESPAÑOLA, SA

ADOLFO DOMINGUEZ SA MARTINSA-FADESA,.SA

ADVEO.GROUP.INTERNATIONAL.SA. MEDCOM.TECH,.SA

ALMIRALL.SA MEDIASET.ESPAÑA.COMUNICACION.SA

AMADEUS.IT.HOLDING.SA MELIA.HOTELS.INTERNATIONAL.SA.

AMPER.SA MINERALES Y PRODUCTOS DERIVADOS, SA

ATRESMEDIA.CORPORACION.DE.MEDIOS.DE.COMUNICACION.SA. MIQUEL.Y.COSTAS.&.MIQUEL.SA AZKOYEN,.SA MONTEBALITO,.SA BARON.DE.LEY,.SA NATRA.SOCIEDAD.ANONIMA BEFESA.MEDIO.AMBIENTE.SOCIEDAD.ANONIMA. NEURON.BIOPHARMA.SA BODEGAS.RIOJANAS,.SA NH.HOTELES,.SA CAMPOFRIO.FOOD.GROUP,.SOCIEDAD.ANONIMA NICOLAS.CORREA,.SA CEMENTOS.PORTLAND.VALDERRIVAS,.SA OBRASCON.HUARTE.LAIN.SA CIE.AUTOMOTIVE,.SA PAPELES.Y.CARTONES.DE.EUROPA,.SA

CLINICA BAVIERA, SA PESCANOVA.SA

CODERE,.SA PRIM, SA

COMPAÑIA VINICOLA DEL NORTE DE ESPAÑA, SA PROMOTORA.DE.INFORMACIONES.SA COMPAÑIA.LEVANTINA.DE.EDIFICACION.Y.OBRAS.PUBLICAS.SA PROSEGUR.COMPAÑIA.DE.SEGURIDAD,.SA COMPAÑIA.LOGISTICA.DE.HIDROCARBUROS.CLH.SA RED.ELECTRICA.CORPORACION.SA

COMPANYIA D'AIGUES DE SABADELL, SA REPSOL SA.

CONSTRUCCIONES.Y.AUXILIAR.DE.FERROCARRILES,.SA SA,.RONSA

DURO.FELGUERA,.SA SACYR.SA.

ELECNOR.SA SNIACE,.SA

ENAGAS.SA SOCIEDAD ANONIMA DAMM

ENCE.ENERGIA.Y.CELULOSA.SA. SOLARIA.ENERGIA.Y.MEDIO.AMBIENTE,.SA

FAES.FARMA,.SA TECNICAS.REUNIDAS,.SOCIEDAD.ANONIMA

FERSA.ENERGIAS.RENOVABLES.SA TECNOCOM.TELECOMUNICACIONES.Y.ENERGIA.SA

FLUIDRA,.SA TELEFONICA, SA

FOMENTO.DE.CONSTRUCCIONES.Y.CONTRATAS.SA TUBACEX,.SA

FUNESPAÑA, SOCIEDAD ANONIMA TUBOS.REUNIDOS,.SA

GAMESA.CORPORACION.TECNOLOGICA.SOCIEDAD.ANONIMA URALITA,.SA

GAS NATURAL SDG SA URBAR INGENIEROS, SA

GENERAL.DE.ALQUILER.DE.MAQUINARIA.SA VIDRALA,.SA

GRUPO.EMPRESARIAL.SAN.JOSE,.SA VISCOFAN.SA

IBERDROLA,.SOCIEDAD.ANONIMA VOCENTO,.SOCIEDAD.ANONIMA

IBERPAPEL.GESTION,.SA ZARDOYA.OTIS,.SA

(26)

26 APPENDIX B

Variable definitions

Variable name Definition

Book Leverage Total debt/Total assets

Crisis Dummy Takes value 0 from 2005 to 2008. Takes value 1 from 2009 to 2012

Debt issues Change in book value of total debt/Total assets

Equity issues Change in subscribed capital/Total assets

Firm size Logarithm of Net Sales

Market Leverage

Book value of total debt/(Total assets-Book value of equity+Market value of Equity)

Market-to-Book ratio Total Market Capitalization/Book value of Balance Sheet Total

Profitability EBITDA/Total assets

Tangibility Physical assets/Total assets

APPENDIX C

Target leverage is calculated by regressing book and market leverage on lagged market-to-book ratio, lagged profitability, lagged tangibility, and lagged size. A Tobit regression is used to limit the values of leverage between 0 and 1.

The residuals of this regression are then used for a linear prediction of target leverage, which is used in the error correction model in Section IV.

Table 11: Estimation of target leverage

Tobit regression of book and market leverage on market-to-book ratio, profitability, tangibility and size. Leverage is limited from 0 to 1.

Book Leverage Market Leverage

Market-to-book ratio 0.006 -0.056*** (0.010) (0.010) Profitability -0.661*** -0.982*** (0.146) (0.138) Tangibility -0.095* -0.053 (0.049) (0.042) Size 0.020*** 0.039*** (0.006) (0.005) Constant 0.500*** 0.226*** (0.078) (0.070) /sigma 0.221*** 0.181*** (0.008) (0.006) Number of observations 485 431 Note: *** p<0.01, ** p<0.05, * p<0.1 Standard Errors in brackets

(27)

27 REFERENCES

Acedo Ramírez, M.A. Alútiz Hernando, A., and Ruiz Cabestre, J. (2012). “Factores determinantes de la estructura de capital de las empresas españolas” Información Comercial Española, ICE: Revista de economía, ISSN 0019-977X, Nº 868, 2012 (Ejemplar dedicado a: Liberalización del comercio interior) , pp. 155-172.

Altunbas, Y., Kara, A. and Marqués-Ibáñez, D. (2009). “Large Debt Financing. Syndicated loans versus

corporate bonds”. European Central Bank Working Papers.

Baker, M. and Wurgler, J. (2002). “Market Timing and Capital Structure”. The Journal of Finance. LVII.

No.1.

Bruinshoofd, W.A., and de Haan, L. (2012). "Market timing and corporate capital structure: a transatlantic

comparison," Applied Economics, Taylor & Francis Journals, vol. 44(28), pages 3691-3703, October.

Cook, D. and Tang, T., “Macroeconomic conditions and capital structure adjustment speed”. Journal of

Corporate Finance. Volume 16, issue 1. February 2012. pp 73-87.

De Bie, T. and de Haan, L. (2007). "Does market timing drive capital structures? A panel data study for

Dutch firms,"

De Economist, Springer, vol. 155(2), pages 183-206, June.

De Haas, R. and Peeters, M. (2006). “Towards target capital structures of firms in transition economies.

Economics of Transition”. Volume 14 (1), 133–169. The European Bank for Reconstruction and

Development.

Drobetz, W., Schilling, D. C. and Schröder, H. (2014). “Heterogeneity in the speed of adjustment across

countries and over the business cycle”.

Fama, E.F., and French, K.R. (2002). “

Testing Trade‐Off and Pecking Order Predictions About Dividends

and Debt”. Review of Financial Studies (2002) 15 (1): 1-33.

Fama, E.F., and French, K.R. (2012). “Capital Structure Choices”. Critical Finance Review, 2012,1: 59-101.

Fermández Ramos, Y., de Rojas Conde, M.C., Zuliani, G.D., (2004). “Contrastación de la Teoría del Pecking Order: El caso de las Empresas Españolas”. Universidad de Valladolid.

Frank, M. Z., and Goyal V. K., (2003). “Testing the pecking order theory of capital structure”. The Journal Of

Financial Economics.

67 (2003) 217–248

Harris, M., Raviv, A., (1991). “The theory of capital structure”. Journal of Finance 46, 297– 355.

Helwege, J., Liang, N., (1996). “Is there a pecking order? Evidence from a panel of IPO firms”. Journal of

Financial Economics 40, 429– 458.

Hovakimian, A., Opler, T., Titman, S., (2001). “The debt– equity choice”. Journal of Financial and

Quantitative Analysis, 36, pp 1-24.

Kayhan, A. and Titman, S. “Firms’ histories and their capital structures” Journal of Financial Economics,

83 (2007), pp 1-32.

Modigliani, F., Miller, M.H., (1958). “The cost of capital, corporation finance and the theory of investment”.

American Economic Review 48, 261– 297.

Myers, S.C., (1977). “Determinants of corporate borrowing”. Journal of Financial Economics 5, 147– 175.

Shyam-Sunder, L., Myers, S.C., (1999). “Testing static tradeoff against pecking order models of capital

structure”. Journal of Financial Economics 51, 219–244.

(28)

28 The sensible giants. The Economist. Issue May 2nd 2009.

Capital structure in Spanish firms : the effect of the financial crisis