How Real Estate shocks affect the corporate investment in
Europe through the collateral channel
University Of Amsterdam Amsterdam Business School MSc Business Economics Master Specialisation Finance
Author: Lorenzo Amin Student number: 10187332 Thesis supervisor: Jan Lemmen Finish date: 06-2016
PREFACE AND ACKNOWLEDGEMENTS
At the end of the study career, it is very difficult what subject to study. At the end, there are a lot that I could choose. What for me is important is that I understand a subject completely after the performance of the study and of course it must get my attention. There are a lot of people that has been supporting me through this study who I want to thank especially my supervisor, Mr. Jan Lemmen, who gave me guidelines on how to get a good structure in the whole study.
Collateral channel is yet a relatively unknown, interesting and important study. Through the 5 years of study that I gained at the University of Amsterdam, I did not once heard about this subject until 6 months ago. While the macroeconomic consequences of the collateral channel is huge. There are a few reasons why I choose to do this study. Firstly, it is a subject that few of us heard about. Second, the economic impact of the collateral channel is great. Third, the study is not performed in the Europe. Last but not least, through this study I learned a lot about the real estate, corporate investment and corporate capital structure.
NON-PLAGIARISM STATEMENT
By submitting this thesis the author declares to have written this thesis completely by himself/herself, and not to have used sources or resources other than the ones mentioned. All sources used, quotes and citations that were literally taken from publications, or that were in close accordance with the meaning of those publications, are indicated as such.
COPYRIGHT STATEMENT
The author has copyright of this thesis, but also acknowledges the intellectual copyright of contributions made by the thesis supervisor, which may include important research ideas and data. Author and thesis supervisor will have made clear agreements about issues such as confidentiality.
Electronic versions of the thesis are in principle available for inclusion in any EUR thesis database and repository, such as the Master Thesis Repository of the Erasmus University Rotterdam
ABSTRACT
In the presence of contract incompleteness, pledging collateral is important for getting funds externally. As Chaney et Al. (2012) show for the US corporations, a $1 increase in the value of a firms’ real estate assets is associated with a $0.06 increase in its investments. This collateral channel effect is statistically significant and economically important. However, research in other regions does not show the results found in the US. Therefore the external validity of the
collateral channel fails to remain. Through an empirical study, this paper studies the collateral channel for the European companies. I find that a 1 dollar increase in the real estate value of a firm leads the representing firm to increase its’ investments by 0.02 dollars. Further, firms with less capital constraints in the form of high liquidity, high profitability and high asset tangibility ratio show less evidence of collateral channel. Small firms with high growth potential and low value of real estate as part of assets also show no evidence of collateral channel. Further I
highlight the difference in the importance of collateral channel before and after the crisis of 2007 and later. We see that the collateral channel is much important after the crisis when the credit standards are tightened. What remains interesting and unexplained to study is why the effect of real estate price shocks is lower in Europe compared to the US. That Despite the ex analysis expectations that the effect should be higher for Europe because European companies are more capital constrained and prefer external financing the most.
Keywords:
TABLE OF CONTENTS
PREFACE AND ACKNOWLEDGEMENTS ... ii
TABLE OF CONTENTS ... iv
LIST OF TABLES ... v
LIST OF FIGURES ... vi
1. Introduction ... Error! Bookmark not defined. 2. Literature review ... 3
2.1 Supply side of Credit ... 5
2.2 Capital constraints ... 8
3. Data ... 11
4. Methodology ... 13
5. Results……….15
5.1 Differences in size, tangibility, profitability and liquidity………17
5.2 Visual presentation of the real estate bubble……….19
5.3 Leverage ratio………20
6. Conclusion………..21
REFERENCES ... 23
LIST OF TABLES
Table 1 Summary statistics 15
Table 2 Regression analysis 1-6 16 Table 3 Regression analyses 7-11 18
LIST OF FIGURES
Figure 1 Real estate prices 5
Figure 2 Investment rate 19 Figure 3 Leverage ratio 20
1. Introduction
In the presence of contract incompleteness, research points out that collateral pledging improves a firm’s ease to finance its projects from outside investors ex ante (Gan 2007). As Chaney et Al. (2012) show, a $1 increase in the value of a firms’ real estate assets is
associated with a $0.06 increase in its investments. Reason for this is that providing investors with the option to liquidate pledged assets ex post acts as a signal of low moral hazard
problems. According to Meyer (1990), firms’ prime financing entity is a bank, and bank loans require collateral pledging. The asset liquidation value is thus important in the determination of a firm’s debt capacity.
In their paper, Bernanke and Gertler (1989) show that in times of business downturn asset values deteriorate, reducing debt capacity of firms. This in turn causes lower investment and output, which will strengthen the downturn even further. Accordingly, this collateral channel is often the main suspect for the severity of the great depression (Bernanke 1983). At the other extreme, it is responsible for the extraordinary expansion of the Japanese economy at the end of 1980s (Cutts 1990). So the collateral value is an important factor in the economy. However, research in the Chinese market shows no evidence of this collateral channel, i.e. increase or decrease in the collateral value does not have any statistically significant effect (Wu et Al., 2013). The authors point out that the nonexistence of the collateral channel is due because companies in this country are mostly state-owned and do not face capital constraints. Facing no capital constraints, the external validity of the collateral channel established by Chaney et Al. (2012) fails to remain for China and probably in general.
Being among one of the arguments why I do this research, Campello et Al. (2010) show differences in the way European and American companies face capital constraints. Also the nonexistence of the collateral channel in China puts pressure on the collateral channel theory. Since the previous studies focused on US and Japan, it is important to test this collateral channel in the current context of increasing real estate prices in Europe. The main aim of this paper is to answer the following question; what is the effect of a $1.00 change in real estate prices on corporate investment in Europe? Through an empirical study this paper contributes to the existing literature in the following way. As far as I am informed, there is no research performed on the collateral channel in Europe nor the external validity of the collateral channel has been tested except for China.
I expect that the effect of real estate prices on investments in European countries is much stronger than that found in the US because European companies are more dependent on bank financing as external financing. The explanation for this expectation is the differences in capital constraints faced by firms and capital constraints again are explained by the state of the economy and the differences in financial systems of regions.
There are two factors that make this research different from that of Chaney et Al., the
revolutionary paper they wrote about the collateral channel. First, the time period of data used by Channey et Al. (1993-2007) represents the booming times in US, where one can assume that capital constraints were limited, confirmed by Maddaloni and Peydró (2011) among others. The low capital constraints are explained by the low interest rates, low regulatory and lending standards. In contrast, the time period that I am using in this analysis (2007-2015) consists the great recession where capital was one of the main constraints faced by firms over the world, especially in Europe as pointed by Campello et Al. (2010). The main cause was the strengthening of regulatory and lending standards. This collateral effect is even more
important during and after the crisis because the credit standards are strengthened (Khosravi (2015) & Maddaloni and Peydró (2011) and Delis and Kouretas (2011)).
Second, because of the European real estate bubble burst in 2007, the situation is similar to that of Japan from 1991 to 1993 as Gan (2007) point out where the real estate prices fell by 50%. This second argument strengthens the investigation of the exclusive effect of collateral channel on investment, controlling for other factors that contributed for less investments in this period of great recession.
Throughout this paper, in part 2, I will introduce some existing literature on the collateral channel and corporate investment. In that part, I also mention factors that determine the difficulty of bank financing. After that, in part 3, description of data that is used in this paper is performed. In the following part, part 4, I introduce the methodology that is used in this paper. Thereafter in part 5, results of the analysis are discussed. At the end a conclusion will follow.
2. Literature review
Meyer (1990) points out financial intermediaries like banks contribute in decreasing
asymmetric information and firms mostly use bank loans for their investments. Most banks require collateral to grant credit. It turns out that collateral pledging indeed decreases the cost of debt, increases credit rating and causes higher loan-to-value ratios. Benmelech et Al. (2009) argue that the ability to pledge collateral and giving the creditor the option to liquidate assets in the event of default decreases the expected losses of creditor and that in effect leads to a lower cost of debt. Another explanation for the lower cost of debt is that the moral hazard problem is reduced. That is the event that a borrower takes actions that are not in the interests of creditors because if the borrower wins, she gets all and if she loses the creditor pays the price. Collateral ensures that the borrower loses her assets in the event of default. Therefore, she will be cautious in making investment decisions (Benmelech et Al. 2009).
This paper is mostly related to the recent and emerging literature on collateral and investment. In their empirical research Chaney et Al. (2012) showed that a $1 increase in the collateral value leads the representative US Corporation to raise its investment by $0.06. In the
aggregate this sensitivity can be quantitatively important because real estate represents a large fraction of tangible assets that firms hold on their balance sheet. More specifically, 58% of the North American corporations have some real estate assets on their balance sheet and real estate assets count for 19% of the tangible assets of these firms. Meyer (1990) confirms this view that most of corporate loans are from the banking sector, and these entities require collateral pledging, so the amount of asset liquidation value is very important for corporates in getting funds for their projects.
To get to their $0.06 sensitivity, Chaney et Al. (2012) used variations in the local real estate prices as shocks to the collateral value of land-holding firms. Similar to the identification that is used in this paper, they measured the difference in collateral channel effect between land-holding and non-land-land-holding firms. Their other hypothesis is that land-land-holding firms in different regions react differently to the real estate shocks, due to differences in real estate price shocks. Main results were that land holding firms increase their investments more than non-land holding firms when the real estate prices increase. This effect is economically large in that a one standard deviation increase in real estate value explains 28% of investment’s
standard deviation (Chaney et Al. 2012). They also point out that firms in regions with greater real estate price fluctuations react stronger to these shocks.
In addition, in a similar research on the Japanese market, Gan (2007) showed by using a Difference-in-difference approach that firms with real estate holdings were more affected by the burst of real estate bubble in 1990s than firms without real estate. The author also points out that firms with greater collateral losses face more difficulties in keeping good
relationships with banks and have in general less bank credit. Nikolov (2012) confirms this view that firms have a nontangible collateral in the form of reputation. Investigating Japan was the perfect situation for Gan (2007). Corporate loans were traditionally collateralized by land and land prices dropped with 50% from 1991 to 1993. This shock is in their view a good measure for the collateral channel since the probability that the investments are hit by other factors than real estate prices is negligible. For the Japanese and the United State market, the collateral channel is apparent as empirical studies show.
In contrast to the above studies, Wu et Al. (2013) investigated the collateral channel for China. There was an enormous growth in the Chinese market since the 2000s altogether with a booming real estate market of China. Naturally, these two movements raised the question of in what way the collateral channel contributed to the growth, as was the case for the
extraordinary expansion in the Japanese economy (Cutts, 1990). They point out that the effect of the collateral channel was non-existent for China. The authors argue that the main reason for this collateral channel non-existence is that the Chinese government owns most of the big companies, the so-called state owned enterprises (SOEs). But, the collateral channel was even invisible for non-SOEs. This paper puts pressure on the external validity of the collateral channel.
Although the time period is different, but the situation in Japan investigated by Gan (2007) is to some extend similar to the situation of Europe in 2007 where the real estate bubble burst was apparent. Figure 1 shows the Commercial price index for the EU that fell dramatically since 2007. This fact makes it very interesting to investigate the effect of the real estate bubble burst on the investment rate. As was apparent for the Japanese market, the European companies with real estate must have cut their investment as a result.
Figure 1 Real Estate prices. Adapted from “commercial property market analysis: An ECB perspective”, by A. Kanitin, 2013, European Central Bank.
2.1 Supply side of Credit
The amount of bank financing both depends on the supply of and the demand for credit. Factors that affect this supply and demand of credit are not only difficult to address but are also too many to address. Think of the interest rate, bank lending standards, the economic conditions, the variations in financial products, the state of the housing market and so on. Addressing these factors to determine the supply and demand of credit is therefore too difficult and time consuming to do and so behind the scope of this study. However, I try to highlight the determinants of credit supply and demand that are important to take in
consideration while reading through this paper and making conclusions. Besides these factors, I take the growth in the great domestic product as an indicator for credit supply.
Banks in the pre-crisis period are broadly known for their soft lending standards. Financing externally was very easy in the pre-crisis period. These soft-lending standards were due to the low interest rates and the soft regulatory rules. According to Maddaloni and Peydró (2011) and Delis and Kouretas (2011) the decline in banks’ profitability due to the low interest rates is the main suspect of the high risk-taking behavior of banks. These banks increased the number of loans outstanding to maintain a certain level of profit, increasing the credit supply. On the other hand, the low regulatory capital requirements strengthened the softening of lending standards. According to these authors, this softening of lending standards
concomitantly with the soft regulatory standards is the main suspect of the last great recession.
In addition, in an attempt to extend the view of Maddaloni and Peydró (2011) and Delis and Kouretas (2011) among others, Khosravi (2015) investigates the period of low monetary policy rates on loosening of banks’ credit standards concerning enterprises, households and consumer loans. He finds that the low interest rates prior to the crisis indeed deteriorated the lending standards for all three types of loans. He contributes to the excising literature by investigating the post crisis period, where the European Central Bank injected an amount of 1 trillion euro of cheap loans into the economy. The effect of this action is unapparent in the short-run. Further research is needed as the time passes to measure the effect of this action by the European Central bank in the long run. What becomes clear from the Maddaloni and Peydró (2011) is that in the post-crisis period the credit standards are adjusted, regulatory rules are stricter so that getting loans from the banks is more difficult than ever. These findings increase the importance of collateral.
What other than the regulatory and monetary policy affect the supply of credit? Collateral value inversely affects credit levels. According to Kanitin (2013) a significant proportion of banks assets comprise of a commercial property loan. Being the direct effect, the declining commercial prices lead to deterioration of credit. The indirect effect of declining commercial prices is a decline in construction and real estate developers suggesting a decline in the supply of real estate. Banks keep assets at the Central bank to get loans from Central bank. The decrease in the commercial property prices caused a decrease in the amount of banks’ assets held at the Central Bank. This indeed further decreased the supply of credit (Kanitin, 2013). At the end, the question is whether this decline in credit supply is measurable.
Banks’ credit supply can be assessed very smartly as Becker and Ivashina (2014) showed in their study. What determines the firms to choose between bank loans or issuing bonds? According to the auteurs firms switch from bank financing to bond financing when the contraction in banks’ credit-supply is apparent. This suggests that when firms switch from bank financing to bond financing either there are tight lending standards, poor bank
performance and/or tight monetary standards (Becker & Ivashina, 2014). The authors argue that bank financing gets difficult, as the state of the economy is worse. A negative growth or a decline in the growth of the great domestic product is a sign for the declining supply of credit.
Hence, investments get lower causing less growth. As mentioned earlier, I use the growth in the GDP as an indicator for the credit supply and use it as a control variable in my analysis. Activity level and the GDP growth are very important factors in the economy, what is the role of asset prices in determining these factors. Assets prices are an enormous determinant of activity level. Think of the interest rate that determines the price of bonds. Research has focused on how monetary policy affects the asset prices through the so-called interest rate channel. Mishkin (2001) looks at the effects of the monetary policy through the collateral channel. He describes how the monetary policy affects consumption and investments through stock prices, real estate prices and exchange rates. Mishkin (2001) argues that due to the increase in money, interest rates decrease, which lowers the cost of financing housing and as a result increases their price. A higher price of housing relative to its construction costs makes it more profitable for construction companies to build more, so the housing expenditure and aggregate demand increase. Also, the increase in Real Estate prices increases household’s wealth (Case et Al., 2005). According to Case et Al. (2005) this increase in households spending will increase the aggregate demand and thus investments.
A more important effect of the monetary policy on real estate prices is through the banking credit supply. As the increase in the money supply is apparent, the real estate prices increase. Because most of the banks lending’s are backed by collateral, when the prices of real estate increases, the value of their assets increases. As a result of this increase in the banks’ capital, the supply of credit by banks will increase. This again increases the investment and aggregate demand (Mishkin, 2005). Now think of it inversely, where the real estate prices decreased after the real estate bubble burst in 2007. All the above arguments work inversely meaning that the credit supply was very low.
In the current state of the economy, with markets interacting in a great degree, the financial sector itself and the overall economy is more than ever sensitive to shocks. In a natural experiment to test whether loan supply shocks in one country affect the activity in other country. Peek and Rosengren (2000) investigated the situation of shocks in the Japanese credit supply and its effect on the real activity in the US. The dramatic decline in Japanese Real Estate prices put pressure on banks’ lending position and thus credits supply. The effect is indeed apparent in the commercial real estate economic activity in the US. Because the
Japanese banks are active internationally, the declining lending position also cut loans in the US. Again, I show how many factors can affect the supply side of the credit.
While it is important to take the supply side of credit into account, I expect that the supply-side remains fixed relatively for each firm in the sample. It is clear from the literature that, prior to the crisis, the capital supply was high and lending standards were low. The role of collateral was lower prior to the crisis as a result of the last two facts. This collateral role increases during and after the crisis as the lending standards are strengthened. This
phenomenon increases the importance of this research since the effect of collateral is more apparent or exclusive after the crisis. To grab the effect of credit supply, I use the growth in the GDP for each country as an indicator of money supply. This way, I can distinguish countries by their amount of money supply and the effect of collateral is stronger.
Furthermore, to some extent, the degree of capital constraints faced by firms can be seen as demand for external financing, which I will assess through some factors explaining capital needs in the next chapter.
2.2 Capital constraints
How capital restricted were corporations in the crisis? Campello et Al. (2010) investigated the degree of capital restrictions faced by firms around the world during the recent financial crisis. Through a survey from CFOs, they found out that the inability of financing their attractive projects from external investors, forced them to cut on investments or even bypass these. The effect was even greater for the European and Asian corporates. While the 86% of constrained firms bypassed attractive investments, unconstrained firms that bypassed their attractive investments account for only 44%, the half of the constrained ones.
Another way of collecting funds to finance their projects is the sale of assets, which was more common in Europe and Asia than in US. Overall they conclude that the European companies were more capital constrained and this lead them to cut their investments more often. This again points out the importance of capital constraints faced by firms and the ability of firms to resolve this problem by pledging collateral. Also Campello et Al. (2010) shed light on the
differences of European and US companies in terms of capital constraints and the fraction of companies that bypass their attractive investments.
But most of companies of interest in this paper are large corporations and these companies are public companies and have easy access to the capital market. By issuing equity through a seasoned equity offering they can raise money to fund their investments. One might say that capital constraint is of no importance because a company can get capital easily via issuing capital anyway to fund its profitable projects. However, there is a drawback to issuing equity. Here to shed light on the drawback of equity issuing, Myers and Majluf (1984), in their influential paper introduced the Pecking Order Theory. According to the authors, there is an order of different financing forms that companies choose when they need fund. The basic idea is that managers of a company have more information about their company than investors. So there is asymmetric information involved. Companies can finance their new projects through internal-, external financing (debt) and or equity issuing. Myers and Majluf (1984) argue that firms prefer internal- to external- and external- over equity financing. While the details of the model are out of the scope of this paper, they argue that internal financing is the cheapest. Issuing equity is the most expensive because of costs of issuing equity and the underwriters’ fee, that is 7% in the US on average (Grullon & Michaely (2004)). On the top of that, issuing equity signals that the stock is overpriced according to the authors. So looking at the costs and signalling effects and when internal financing is not possible, firms prefer external financing to equity financing.
Murray and Vidhan (2003) tested the accuracy of Pecking Order Theory in practice for the publicly traded US companies. The authors find that external financing is much greater than that predicted by Myers and Majluf (1984). This external financing sometimes exceeds the investment needs for companies. Firms with the most information asymmetric i.e. high-growth, small firms are more likely to show evidence of the pecking order theory. However the inverse is true according to the empirical evidence. Mostly large companies that were continuously listed on the stock market between 1980 and 1990 show the most accuracy in their leverage choice in line with the Pecking Order Theory. This again shows the importance of debt financing for capital restricted firms.
Besides collateral in this case as a form of decreasing moral hazard problem, intangible assets in the form of reputation to the bank also contribute. In a working paper Nikolov (2012) point out the importance of firms’ reputation as an intangible collateral while taking loans.
Companies obtain this reputation by keeping relations with the bank. It means that as a company live up to its obligations, the reputation, and so the intangible assets, increase. If the market can exclude the firms that defaulted on their loans, then it is left with firms that have good reputation and this increases the access to leverage in aggregate. According to the author, this intangible asset is very important for two reasons. First, it excludes firms that do not live up to their obligations often. Second, the amount of credit constraints faced by firms decreases this way making it easier for companies to get external financing. It would be very illogical that this exclusion of defaulting companies is not apparent in the current economy, but the authors do not make this clear.
As usual in investment research, the ex-ante capital constraint faced by firms is important. To assess this ex-ante capital constraint faced by firms, I use four factors that are important determinants. The first determinant is a firms liquidity ratio.. This is the ratio of current assets to current liabilities. As the ratio increases, firms have more cash flow or current assets to repay their current liabilities. This increases the credit worthiness of companies. Accordingly, every company is different in the point where it switches from internal to external financing. Also as highlighted by the Pecking Order theory, firms prefer internal financing the most Having a high liquidity ratio means that the company has enough internal funds to invest and the rest is distributed to the investors by dividend. This is a good ratio to rank companies from less constrained to most capital constrained firms.
Second I count for the size of the company. While Real Estate is a part of the tangible assets of firms, a larger part of companies’ assets exists of non real estate assets, which can be pledged as collateral. An example is a production factory or a machine. Assuming that firms with greater total assets have less capital constraints seems logical in this contest (Chaney et Al. 2012). The third factor is the assets tangibility of a company. As seems logical, firms with the most part of their assets that is tangible are able to pledge more collateral. The last factor that I try to take into account is the profitability of a company. As the profitability of a firm increases, the chance of default decreases and that makes it easier for firms to get funds.
3. Data
I use accounting data on European corporations active in COMPUSTAT that is collected from Wharton research data service over the period of 1987 to 2015. Most of the data selection criteria are in line with that of Chaney et Al. (2012). Because the data is only available on a global basis, I have to keep only corporations registered in Economic Monetary union (EMU) countries as of 2002. Because of their economic significance, Great Britain, Denmark and Sweden are kept in the sample despite not having euro as their home currency. The rest is dropped out of the sample under the assumption that the real estate assets are located on the same country as where a company’s headquarter is located. The last assumption is due to the fact that companies do not list their real estate holding per region or country in the
COMPUSTAT data file. As Chaney et Al. (2012), I keep only firms with non-missing total assets, firms that were active in 1993. I drop firms that are active in mining and construction-, financial-, insurance- and real estate sector.
This panel dataset contains variables that I need to measure the two important variables for this research. One is the investment rate and the second is the Real Estate value. Investment rate is measured as the ratio of capital expenditures over a firm’s property, plant and
equipment (ppent). As is common in investment literature, controlling for market-to-book gives a more accurate beta. To measure the market-to-book ratio, I need the monthly close price of each firm in the sample, which is not available in COMPUSTAT data file. I use the total asset value instead of the market to book value, which is proportionally for each firm and will also give us a reliable measure.
Next I need to measure the value of the real estate assets. Unfortunately, the value of real estate assets is at their historical costs. So I need to measure the market value. To measure the market value of the real estate assets, I first determined the value of real estate in 2007. Then, using a depreciation life of 40 years, I calculate backwards the age of the real estate assets using the proportion of depreciation over the value in 2007 times 40 as the age of the real estate assets. 2007 minus the age gives me the year of purchase.
Now this data is merged with the country level Home Price Index (HPI), which is the index factor that I use to inflate the cost price of the real estate assets to get their market value. This index is collected from the Statistical Data Warehouse, maintained by the European central bank. I only have the HPI indexes from 2000 upward with 2007 as the base year. Therefore I
miss values of the HPI index before 2000. I use the consumer price index before 2000 to address this issue because the CPI index and its’ growth is very similar to that of the HPI. This data is collected from the website Global Rates.
Furthermore firms’ cash flows ratio to past years PPE is measured. To measure the leverage ratio, I take the short- and the long-term debt over the total assets. Furthermore, I measure the long-term debt repayment and debt issuance over the PPE. The difference is a measure for the net change in long-term debt. Finally, I add the ratio of current debt change over the PPE to the net change in long-term debt over PPE to get to the total change in debt.
Furthermore, Real Estate value is measured by summing up the companies assets consisting of property plant and equipment (ppent). One problem is that the dataset captures the
accumulated depreciation and the values of real estate are noted as their book value. Since the real estate holdings are at their book value in the dataset, we use a 40 years depreciation time to value the real estate holdings backwards by measuring their age. The ratio of accumulated depreciation over the book value gives us the age of a building. After inflating the value of real estate holdings till 2007, this value fluctuates with the level of growth in the real estate prices. After that, we use this value as the base for real estate holdings. Also because the Real Estate holdings are given until 1993, I only use firms that were active in 1993. Further data selection and transformation is very much like that of Chaney et Al. (2012).
I arrange firms on the basis of their credit constraints. Firstly, firms are arranged from top to bottom by their liquidity. Measured by the ratio of current assets over current liabilities. Second, firms are arranged by their profitability. That is the ratio of operating income before depreciation minus depreciation over lagged total assets. Third, I use the lagged total assets size as the third control variable.
Finally, to address the effect of the money supply, I collect data on the growth of the GDP for each country from 2007 to 2015. This data is merged with the main data file. Furthermore in the methodology part of this paper, I use the GDP growth to as an instrument for assessing the supply side of credit.
4. Methodology
The two main variables in this paper are the investment rate and the real estate value. Similar to the methodology of Case et Al. (2001) where they investigated the effect of real estate value on consumer spending and Chaney et Al. (2012), I use two sources of identification in my empirical strategy. Within a local area, the sensitivity of investment to real estate prices across firms with and without real estate is compared. Second, the investment of land-holding firms across areas with different variations in real estate prices is compared. Through these two identifications, by using the model of Gan (2007) and Chaney et Al. (2012), I try to assess how investment rate is affected by real estate shocks in a difference-in-difference model through the following formula:
𝐼𝑁𝑉!"! =∝
! + 𝛿!+ 𝛽!"#$%&'.!𝐺! + 𝛾𝐺! + 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠!"+ 𝜀!"
Where, 𝐼𝑁𝑉!"! is the ratio of investment over lagged PPE, RE-Value
i is the value of Real estate in 2007 over the lagged PPE, measured from the original value of real estate in 1993, and Gs measures the growth in the real estate value from 2007 to year t. As Channey et Al. (2012), I control for ratio of cash flow to PPE, one year lagged market to book value of assets and the lagged leverage. αi and δt represent the time and country fixed effects that should be included to capture the shocks that are caused by for example global economy. Gc captures the overall impact of the real estate cycle on investments. Finally the β measures the extra investments that a firm does for each additional $1 that this firm actually owns, not how investments react to real estate shocks overall. This specification lets us to take distance from state-specific shocks that affect the investments of both firms with and without real estate (Chaney et Al. 2012).
Hypothesis: Within Europe, corporate investment is positively correlated with the shocks in the real estate price.
In the first regression (reg1), I regress investment on real estate value, measured as a fraction of PPE, HPI, as well as year and firm fixed effects. In the second regression (reg2), I run the same regression as in reg1 using the alternative real estate variable, which is the value of real estate over total assets. For the third regression (reg3), I repeat regression 1, but add cash flow and the market-to-book ratio as controls. For regression four (reg4), I repeat reg3, but use the
alternative real estate variable. For the fifth regression (reg5), I regress investment on an interaction variable of the state price index and a dummy variable indicating whether the firm owns any real estate in 2007 as well as year and firm fixed effects. For the sixth regression (reg6), I repeat regression 5 and add cash flow and the market-to-book ratio as controls. In good times, getting fund from a bank is not that hard. Think of the time before the financial crisis. After 2007, the financial crisis began and let the banks over the world tighten their credit conditions, the interest rate and so on. So getting fund from a bank became much harder. Indirectly, it means that the role of collateral channel is greater after 2007. To see if this is indeed true, I am dividing the time period into before and after 2007. Before crisis, the credit constraints were low and the role of collateral should be lower Chaney et Al. (2012). In regression 7 I do the same as regression 3 in but then for the period 2000 to 2007.
In regression 8, I take the small companies into account and perform regression 7 again. These are companies with log assets total up to the 75st percentile. The idea is that firms with small total assets should be more capital-constrained than firms without. Therefore, a real estate price shock should affect their investment rate more compared to the large companies. On the other hand one can argue that small firms also have relatively less real estate.
Therefore a shock in real estate prices should not affect the investment of these firms. In regression 9, I regress the investment rate on real estate value but select the sample on the basis of their liquidity. Restricting the liquidity to values lower than 3. The idea is that firms with more liquidity are also able to fund their projects internally, weakening the effect of collateral channel.
Profitability ranges from -7,72 to 34.83 in the sample. In regression 10, I restrict the sample size to firms with a profitability of 8 or lower. In the last regression, regression 11, I select companies on the basis of their asset tangibility. Companies with the most tangible assets are kept in the sample.
Last but not least I try to show how the real estate bubble is apparent through a figure. In this figure, the investments by real estate holding firms are compared to firms without real estate. What we shall see is that the investments made by real estate holding firms is much higher, reflecting the real estate bubble. For more about the methodology and data transformation, see Appendix A.
5. Results
How important is the real estate for the European corporations? To answer this question, I produced a table that contains some important variables that are mentioned in table 1 below. After removing outliers from the data to get good tails, only 23,3% of the companies show some real estate holdings. Furthermore, real estate holdings count for 24,9% of the total assets for these real estate holding companies, which is a great percentage. Fluctuations in this amount can cause serious problems for a company.
Further, from the table becomes clear that the amount of debt as part of the total assets is on average only 20% which is quit low knowing the preferences of companies to debt when we take interest tax shield into consideration. Furthermore, we see that the market to book ratio is quit high on average but the median is about 2,7 times the total assets value which is
reasonable for a publicly traded corporation. The cash-ratio is on average 86% of the total asset value. It is a bit rare that the amount of cash is so high. Even after checking for outliers, the ratio remains high. The reason is that the diversity of cash as part of total assets is great. This may be due to the fact that the need for cash in the sector that a company operates is much more than another company from another sector.
Table 2 reveals various results of the analysis. Through regression 1 to 6, the country fixed effects and year fixed effects are controlled for. In reg1 the investment rate is regressed on the ratio of real estate value and the HPI index. It turns that the real-estate-holding firms increase their investment with 0,017 when the real estate prices increase compared tot the non-real-estate-holding firms. However, the r2 is 0,083 meaning that the model is very weak. In the second regression, I regress the investment on the level of real estate as a ratio of total assets. The effect turns out to be 0,17, which is a quit high number. Again the fit of the model remains poor. One explanatory factor is the multicolinearity between the investment rate and the ratio real estate to assets total.
Adding market-to-book ratio and cash ratio to the regression that is performed in reg1, we see that the fit of the model increases. On average, real estate holding firms increase their
investments by 2% when the real estate price increases. Also, the cash ratio and the market to book ratio are statistically significant explanatory factors for the investment rate, both at a 1% significance level.
In reg4, investment is regressed on ratio real estate to assets, market-to-book value and cash ratio. Again, the cash ratio and the market-to-book ratio increase the explanatory power of the model. This time, an increase in a firms’ real estate value causes the representing firm to increase its investments with 15%.
In regression number 5, I do the same regression as in regression 3 but add a dummy variable indicating whether a firm owns real estate or not instead of the total real estate value. The results show that owning real estate increase the investments with 19% when the real estate prices increase. However, the fit of the model is very poor.
Now the question was in what extend do supply side of credit explain investments. I choose to investigate the effect of credit supply by using the growth in the GDP as an instrument
variable. Performing the regression 6 and using the real estate value as an explanatory variable, it turns that the effect of the growth in GDP is statistically not different from zero. Besides, the effect of the real estate value overall gets deteriorated. As already mentioned, most probably, the effect of credit supply remains constant across Europe due to the narrow cooperation between the European central bank and European countries.
5.1 Results of companies with differences in size, tangibility, profitability and liquidity
As argued in the introduction, the time period that a regression is performed is very important. The reason behind this is that during booming times, the credit constraints are low and when there is a recession, i.e. after 2007, the credit constraint become greater. I am dividing the time period before and after 2007. In regression 7 in table 2, we see that the same as
regression 3 in table 1 but then for the period 2000 to 2007. What seems is that the effect of real estate shocks is much smaller in the period before 2007 than the period after. It is logical because in booming times, especially the period before 2007, credit constraints were
negligible. As Delis and Kouretas (2011) point out in their paper, the interest rate was low, lending standards were very soft so getting fund from a bank was very easy.
In regression 8, I take the small companies into account and perform regression 7 again. These are companies with log assets total up to the 75st percentile. The idea is that firms with small total assets should be more capital-constrained than firms without. Therefore, a real estate price shock should affect their investment rate more compared to the large companies. As regression 8 on table 2 reveals, the effect gets weakened and is not statistically different from zero. One explanation is that the small firms have less real estate as part of total assets; therefore any shock to the real estate value has negligible effect on their creditworthiness and thus investments.
In regression 9, I regress the investment rate on real estate value but select the sample on the basis of their liquidity. Restricting the liquidity to values lower than 3. Regression 9 in table 3 shows that the real estate holding firms with low liquidity increase their investments by 0,025 when the real estate prices increase, slightly more than the overall sample. That too seems logical because firms with small liquidity are more dependent on collateral.
Profitability ranges from -7,72 to 34.83. In regression 10, I restrict the sample size to firms with a profitability of 8 or lower. Results in table 3 show that less profitable firms react less to the real estate prices. Again, it can be explained because more profitable firms are more likely to have more real estate than less profitable firms. These more profitable firms react therefore more to real estate price shocks.
In the last regression, regression 11, I select companies on the basis of their asset tangibility. Companies with the most tangible assets are kept in the sample. I keep companies with tangibility of 0,5 or higher. As expected, the effect is not significantly different from zero as table 3 shows. As firms with low tangible assets get less affected from real estate shocks, their investments too do not react much.
5.2 Visual representation of the real estate bubble
Figure 2 shows the visual representation of the sample of data that is used in this paper. Firms are divided in real-estate-holding firms, the blue line and firms without real estate, the red line. Maddaloni and Peydró (2011) mentioned the real estate bubble in their paper. Without going into details, the main idea is that the real estate prices are far more than their
fundamental prices because of speculation. This figure represents that high price of real estate and the burst of it around 2007. Before 2007, real-estate-holding firms have high investment rate. One possible explanation is that they have a great amount of real estate; they can borrow more via the collateral channel and so invest more. As the bubble bursts, the value of real estate decreases and as a result the investment rate of real estate holding firms decreases. The logical explanation is that these firms faced deterioration in their real estate and thus collateral value, meaning their ability to borrow deteriorated after the burst.
Figure 2 In this figure the investment rate of real-estate-holding firms is compared to firms without real estate. The blue line represents real estate holding firms.
5.3 Leverage ratio
In figure 3 I graphed the leverage ratio of firms. By this I made a distinction between estate-holding firms and firms without. What is clear from the figure is that on average, real-estate-holding firms’ leverage rate is much higher than that of firms without real estate. The leverage ratio of firms with real estate is almost double that of firms without real estate. This clearly shows that the collateral channel is important even here in the leverage ratio
representation. However, the figure is just a representation and no statistical proof is available to say that collateral explains exactly the difference between the capital structures of the two sorts of firms.
Figure 3 This figure represents the amount of leverage for real-estate-holding firms versus firms without real estate. The blue line represents the leverage ratio of real-estate-holding firms.
6. Conclusion
When the value of real estate increases with 1 dollar, the investments of a real estate holding firm increase by 0.02 dollar, statistically significant at a 5% level. When the period befor the crisis is compared to the period after, we see that the importance of collateral value is much lower in the period before the crisis for two reasons.
First, a period of crisis after 2007 is known for its’ low credit supply. So from this point of view, one could take the conclusion that because of low credit supply, companies’ need for external financing from a bank became greater and thus the importance of collateral value increased. Second, the lending standards were very low in the pre-crisis period as mentioned by Maddaloni and Peydró (2011).
Further we saw that the effect of real estate shocks is much lesser for companies with high liquidity, profitability and companies with low tangible assets as part of their total assets value. The collateral channel for the small companies is less apparent. The proof is that the effect of real estate price shocks has no statistically significant effect on the investments of small companies. One explanation is that these companies see deterioration in their growth opportunities and thus their investments when there is a crisis and the demand is low. Another explanation is that these companies just do not have enough real estate as part of their assets; therefore any shock to their real estate value does not have significant effect.
In the last part, we see that the real estate bubble burst is very clearly apparent just before the financial crisis began and the real estate bubble burst. Investment of real estate holding companies decreased below those firms without real estate. We also see that the collateral value leads to a higher leverage ratio for firms with real estate holdings compared to firms without. It might be logical that when collateral value decreases, firms’ financial situation deteriorates, leading to less leverage. This leverage ratio remains constant even after the bubble burst, which remains unexplained.
A more unexplained phenomenon that is ruled out of this analysis is that of the effect of real estate shocks on investments in Europe compared to US. As pointed by Chaney et Al. (2012), when the real estate value increases in the US, the representing firm increases its’ investments by 0.06 dollar. Three times of the value found in this research for the European corporations.
Ex ante, I expected that the effect should be much higher for Europe compared to US. This because European companies are more capital constrained, finance relatively more from banks and collateral gets more important in the time period that I use in this analysis Campello et Al. (2010). Why is the effect less than was expected? In my view this is a very important question that should take place in investment research. Do European banks put less value on collateral? This question remains open for further research.
REFERENCES
Becker, B. & Ivashina, V. (2014). Cyclicality of credit supply: Firm level evidence. Journal
of monetary economics 62: 76-93
Benmelech, E. & Bergman, N.K. (2009). “Collateral Pricing.” Journal of Financial
Economics 91 (3): 339–60
Bernanke, B.S. (1983). “Nonmonetary Effects of the Financial Crisis in Propagation of the Great Depression.” American Economic Review 73 (3): 257–76
Bertrand, M., Duflo, E., Sendhil, M. (2004). “How Much Should We Trust Differences-in-Differences Estimates?” Quarterly Journal of Economics 119 (1): 249–75
Chaney, T., Sraer D., Thesmar D. (2012). “The Collateral channel: how real estate shocks affect corporate investment.” American Economic Review 102(6): 2381-2409 Chevalier, E., Vath, V.L., Scotti. S. (2011). An Optimal Dividend and Investment Control
Problem under Debt Constraints. Society for Industrial and Applied Mathematics 425: 666-694
Cutts, R.L. (1990). Power from the ground up: Japans Land Bubble. Harvard Business
Review, 68: 164-72
Delis, M.D. & Kouretas, G.P. (2011). "Interest rates and bank risk-taking." Journal of Banking & Finance 35.4: 840-855.
Eisfeldt, A.L. & Rampini, A.A. (2009). “Leasing, Ability to Repossess, and Debt Capacity.”
Review of Financial Studies 22 (4): 1621–57
Gan, J. (2007). “Collateral, Debt Capacity, and Corporate Investment: Evidence from a Natural Experiment.” Journal of Financial Economics 85 (3): 709–34
Global Rates. (2016). Inflatie cijfers. http://nl.global-rates.com/economische-statistieken/inflatie/1999.aspx (Accessed 02-06-2016)
Grullon, G. & Michaely, R. (2004). The information content of share repurchase programs.
Khosravi, T. (2015). The Impact of a Low Interest Rate Environment: Empirical Evidence from the Euro Area Bank Lending Survey. (Working Paper No. 67363), Retrieved
from Munich Personal RePEc Archive website:
https://mpra.ub.uni-muenchen.de/67363/
Maddaloni, A. & Peydró, J. (2011). "Bank risk-taking, securitization, supervision, and low interest rates: Evidence from the Euro-area and the US lending standards." Review of Financial Studies 24.6: 2121-2165
Meyer, C. (1990). Financial Systems, Corporate Finance, And Economic Development. (Working Paper No. 11477), Retrieved from National Bureau of Economic Research website: http://www.nber.org/chapters/c11477
Mishkin, F.S. (2001). THE TRANSMISSION MECHANISM AND THE ROLE OF ASSET PRICES IN MONETARY POLICY. (Working paper No. 8617) Retrieved from National Bureau of Economic Research website: http://www.nber.org/papers/w8617 Murray, Z. F., & Vidhan, K.G. (2002). Testing the pecking order theory of capital structure.
Journal of Financial Economics 67: 217–248
Myers, S.C., & Majluf, N.S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13: 187-221
Nikolov, K. (2012). A MODEL OF BORROWER REPUTATION AS INTANGIBLE COLLATERAL. (Working paper No. 1490) Retrieved from European Central Bank website:
https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1490.pdf?471ae9e035d529b78b66f 275ca8dc1d1
Peek, J., & Rosengren, E. (2000). Collateral damage: Effects of the Japanse Bank crisis on real activity in the United States. American Economic Review, 30-45.
Rampini, A.A., & Viswanathan, S. (2010). “Collateral, Risk Management, and the Distribution of Debt Capacity.” Journal of Finance 65 (6): 2293–322 Statistical Data Warehouse. (2016). Residential property price indicator.
https://sdw.ecb.europa.eu/browse.do?node=2120781 (Accessed 28-05-2016)
Wu, J., Gyourko, J., Deng, Y. (2013). Is There Evidence of a Real Estate Collateral Channel Effect on Listed Firm Investment in China? (Working Paper No. 18762), Retrieved from National Bureau of Economic Research website:
http://www.nber.org/papers/w18762
Wharton Research Data Services. (2016). COMPUSTAT IQ.
APPENDIX A [Stata codes]
use "/Users/lorenzoamin/Downloads/d9ba27777984bf50.dta" * drop all non-european countries*
drop if fic== "ARE" drop if fic== "ARG" drop if fic== "AUS" drop if fic== "BFA" drop if fic== "BGD" drop if fic== "BHS" drop if fic== "BHR" drop if fic== "BLZ" drop if fic== "BMU" drop if fic== "BRA" drop if fic== "BWA" drop if fic== "CHL" drop if fic== "CHN" drop if fic== "CIV" drop if fic== "COL" drop if fic== "CUW" drop if fic== "CYM" drop if fic== "CYP" drop if fic== "ECU" drop if fic== "EGY" drop if fic== "FLK" drop if fic== "FRO" drop if fic== "GAB" drop if fic== "GGY" drop if fic== "GHA" drop if fic== "GIB" drop if fic== "HKG" drop if fic== "IDN" drop if fic== "IMN" drop if fic== "IND" drop if fic== "IND" drop if fic== "ISR" drop if fic== "JAM" drop if fic== "JEY" drop if fic== "JOR" drop if fic== "JPN" drop if fic== "KAZ" drop if fic== "KEN" drop if fic== "KOR" drop if fic== "KWT" drop if fic== "LBN" drop if fic== "LBR" drop if fic== "LKA" drop if fic== "LTU"
drop if fic== "MAR" drop if fic== "MCO" drop if fic== "MEX" drop if fic== "MHL" drop if fic== "MUS" drop if fic== "MWI" drop if fic== "MYS" drop if fic== "NAM" drop if fic== "NGA" drop if fic== "OMN" drop if fic== "PAK" drop if fic== "PAN" drop if fic== "PER" drop if fic== "PHL" drop if fic== "PNG" drop if fic== "PSE" drop if fic== "QAT" drop if fic== "RUS" drop if fic== "SAU" drop if fic== "SDN" drop if fic== "SEN" drop if fic== "SGP" drop if fic== "SRB" drop if fic== "THA" drop if fic== "TTO" drop if fic== "TUN" drop if fic== "TUR" drop if fic== "TWN" drop if fic== "TZA" drop if fic== "UGA" drop if fic== "UKR" drop if fic== "VEN" drop if fic== "VGB" drop if fic== "VNM" drop if fic== "ZAF" drop if fic== "ZBW" drop if fic== "ZWE" drop if fic== "ZMB" drop if fic== "BGR" drop if fic== "CZE" drop if fic== "EST" drop if fic== "CHE" drop if fic== "HRV" drop if fic== "HUN" drop if fic== "ISL" drop if fic== "LVA" drop if fic== "MLT" drop if fic== "NOR" drop if fic== "NZL" drop if fic== "POL"
drop if fic== "ROU" drop if fic== "SVK" drop if fic== "SVN"
**Save the only europe file
save "/Users/lorenzoamin/Desktop/Only europe.dta" **Open the only europe data
use "/Users/lorenzoamin/Desktop/Only europe.dta" **Drop duplicates
drop if gvkey == gvkey[_n-1] & fyear == fyear[_n-1]
**Restrict the sample for obervations with non-missing total assets
drop if at==. & fyear==2007
**Drop firm is not active in 2007 gen active=1 if fyear==2007
by gvkey, sort: egen active2 = total(active) drop if active2 !=1
drop active drop active2
sort gvkey (fyear)
**Drop values with gap in consecutive years destring gvkey, replace
tsset gvkey fyear
tsspell, f(L.fyear ==.)
egen length = max(_seq), by (gvkey _spell) keep if length >= 3
drop _spell _seq _end length
**Drop observation with less than 3 consecutive years duplicates tag gvkey, generate(gvkey2)
drop if gvkey2<2 drop gvkey2
sort gvkey (fyear)
**Drop financial and construction firms destring sic, replace
gen sic2 = 1 if sic >= 6000 & sic <=6799 | sic >= 1000 & sic <=1799
drop if sic2==1 drop sic2
**Replacing missing values replace dpc=0 if (dpc==.)
egen totalrealest = rowtotal(ppent), missing gen totalrealest07 = totalrealest if fyear==2007
by gvkey, sort: egen totalrealest2 = total(totalrealest07), missing
sort gvkey (fyear)
**Calculating year of purchase
gen assetage = (totaldepr/totalrealest2)*40 if fyear==2007 gen yearofpurch = 2007-assetage
replace yearofpurch=int(yearofpurch)
by gvkey, sort: egen yearofpurch2 = total(yearofpurch) **Drop values to merge the 2 datasets
drop if totalrealest2==. ren totalrealest2 re_value sort gvkey (fyear)
**Merge datasets CPI downloaded from Bureo of Labor Satistics merge m:m fyear fic using "/Users/lorenzoamin/Desktop/hpi.dta" drop _merge
sort gvkey fyear
**Create market value of real estate gen marketval= re_value*hpi
drop if marketval==.
gen ppentlag = ppent[_n-1] gen atlag = at[_n-1]
sort gvkey (fyear)
**Creating independent variables using Compustat #item8 for PPE lagged
gen state=1 if fic=="AUT" replace state=2 if fic=="BEL" replace state=3 if fic=="DEU" replace state=4 if fic=="DNK" replace state=5 if fic=="ESP" replace state=6 if fic=="FIN" replace state=7 if fic=="FRA" replace state=8 if fic=="GBR" replace state=9 if fic=="IRL" replace state=10 if fic=="ITA" replace state=11 if fic=="NLD" replace state=12 if fic=="PRT" replace state=13 if fic=="SWE" drop if marketval>900
gen retiore_value = marketval/ppentlag gen re_valueat = marketval/atlag
gen investment = capx/ppentlag gen altinvrate= capx/at
**investment rate of higher than 5 is rare. after performing under line,
** only 0,5% of observations are deleted **--->
drop if investment>20
gen market2book = atlag*2.8/at drop if at>5000
**Also less than 1% of the upper cf is deleted gen cf = (dp+ib)/ppentlag
drop if cf<-100
gen ltdebtissuance = dltis / ppentlag gen ltdebtrepay = dltr / ppentlag
gen ltdebtchange = (dltis-dltr) / ppentlag gen currentdebtchange= (dlcch)/ ppentlag
gen totaldebtchange= ltdebtchange+currentdebtchange gen totbooklev = (dlc+dltt)/at
gen shortbooklev = (dlc)/at gen longbooklev = (dltt)/at
gen totmarketlev = (dlc+dltt)/ at gen shortmarketlev = (dlc)/ at gen longmarketlev = (dltt)/ at
gen control2hetero = (oibdp - dp) / ipodate gen control3hetero = (oibdp - dp) / sic gen control1hetero = (oibdp - dp) / at gen control4hetero = (oibdp - dp) / state gen payout = (dvc + prstkc) / ib
sort gvkey (fyear)
**Checking for outliers
winsor2 retiore_value, suffix(_w) cuts (3 97) by (fyear) winsor2 re_valueat, suffix(_w) cuts (5 95) by (fyear) winsor2 investment, suffix(_w) cuts (0.5 99) by (fyear) winsor2 market2book, suffix(_w) cuts (1 97.5) by (fyear) winsor2 cf, suffix(_w) cuts (2 98) by (fyear)
winsor2 totbooklev, suffix(_w) cuts (0 98) by (fyear) winsor2 shortbooklev, suffix(_w) cuts (0 98) by (fyear) winsor2 longbooklev, suffix(_w) cuts (0 98) by (fyear) winsor2 totmarketlev, suffix(_w) cuts (0.5 97) by (fyear) winsor2 shortmarketlev, suffix(_w) cuts (0.5 97) by (fyear) winsor2 longmarketlev, suffix(_w) cuts (0.5 97) by (fyear) drop if gvkey==.
drop if fyear==. *--->
replace capx=0 if capx>2000 drop if capx==0
**Creating dummy variable
replace ppentlag=0 if ppentlag<100 replace ppentlag=. if ppentlag>800 drop if ppentlag==.
gen REowner=1 if ppentlag>0 replace REowner=0 if ppentlag==0 sort gvkey (fyear)
ren investment invrate ren cf cashratio
ren retiore_value ratioREvalue ren re_valueat ratioREvalueat **Creating table 1
tabstat invrate market2book cashratio at ratioREvalue ratioREvalueat altinvrate totmarketlev shortmarketlev longmarketlev REowner, stat(mean median sd p25 p75 n) col(stat)
**Merge GDP with the main dataset
merge m:m fyear fic using "/Users/lorenzoamin/Desktop/gdp growth.dta"
drop _merge
sort gvkey fyear drop if fyear==2006
**Create some extra variables gen re_stateindex=ppent/state gen re_year = ppent / fyear gen re_firm= ppent / gvkey ** Regression 1 to 6
sort gvkey fyear xtset gvkey fyear tab fyear, gen(YEAR) drop YEAR1
gen cluster=state*fyear *** reg1
xtreg invrate ratioREvalue hpi i.fyear, fe vce (cluster gvkey) estimates store dd1, title(reg1)
*** reg2
xtreg invrate ratioREvalueat hpi i.fyear, fe vce (cluster gvkey)
estimates store dd2, title(reg2) *** reg3
xtreg invrate ratioREvalue cashratio market2book i.fyear, fe vce(cluster gvkey)
estimates store dd3, title(reg3) *** reg4
xtreg invrate ratioREvalueat cashratio market2book i.fyear, fe vce (cluster gvkey)
estimates store dd4, title(reg4) gen dummyre=1 if ppent>0
replace dummyre=0 if ppent<10
gen interact= dummyre*re_stateindex ***reg 5
xtreg invrate dummyre cashratio market2book i.fyear, fe vce (cluster gvkey)
estimates store dd5, title(reg5) ***reg 6
xtreg invrate ratioREvalue dummyre cashratio market2book gdp i.fyear, fe vce (cluster gvkey)
estimates store dd6, title(reg6)
*** Creating table for regression 1-6
estout dd1 dd2 dd3 dd4 dd5 dd6, cells(b(fmt(%15.2g) star) t(fmt(%11.2g) par)) keep(ratioREvalue ratioREvalueat hpi cashratio market2book dummyre gdp) stats(N r2) starlevels(* 0.10 ** 0.05 *** .01) delimiter( ) end( ) label style(tex) ** Creation of variables liquidity, profitability, tangibility and size
gen liqudity = act/lct
winsor2 liqudity, suffix(_w) cuts (0.5 99) by (fyear) gen profitab = (oibdp-dp)/atlag
winsor2 profitab, suffix(_w) cuts (0.5 99) by (fyear) gen tangibility= ppent/atlag
gen logsize = log(at)
winsor2 logsize, suffix(_w) cuts (0.5 99.5) by (fyear) gen growthop = market2book_w
**Twoway line
egen mean = mean(marketval), by(fyear) sort fyear
line mean fyear
winsor2 capx, suffix(_w) cuts (0.5 99) by (fyear)
by fyear, sort: egen investrate_asset = mean(investment) if dummyre==1
by fyear, sort: egen investrate_0asset = mean(investment) if dummyre==0
graph twoway line investrate_asset investrate_0asset fyear sort gvkey fyear
by fyear, sort: egen booklev_asset = mean(totbooklev_w) if dummyre==1
dummyre==0
graph twoway line booklev_asset booklev_0asset fyear sort gvkey fyear
** creating regression 7-11 regs *** reg7
*** perform the same steps as above, but then change 2007 to 2000 and perform
*** the following regression
xtreg invrate ratioREvalue cashratio market2book i.fyear, fe vce(cluster gvkey)
estimates store dd1, title(reg7) *** reg8
*** perform the same steps as above, but drop firms with logsize>4
xtreg invrate ratioREvalue cashratio market2book i.fyear, fe vce(cluster gvkey)
estimates store dd2, title(reg8) *** reg9
*** perform the same steps as above,but drop firms with liquidity>3
xtreg invrate ratioREvalue cashratio market2book i.fyear, fe vce(cluster gvkey)
estimates store dd3, title(reg9) *** reg10
*** perform the same steps as above,but drop firms with profitability>8
xtreg invrate ratioREvalue cashratio market2book i.fyear, fe vce(cluster gvkey)
estimates store dd4, title(reg10) *** reg11
*** perform the same steps as above,but drop firms with tangibility>0.5
xtreg invrate ratioREvalue cashratio market2book i.fyear, fe vce(cluster gvkey)
