Is there a persistent impact of IPO market timing on capital structure? : evidence from the US market

Is There a Persistent Impact of IPO Market Timing on Capital

Structure? Evidence from The US Market

Master of Science in Finance Track: Corporate Finance

Master Thesis Written by Yiwen Cui 11398825 Supervisor: dr. E. Zhivotova

Statement of Originality

This document is written by Student: Yiwen Cui who declares to

take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are

original and that no sources other than those mentioned in the text

and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the

supervision of completion of the work, not for the contents.

Abstract

Whether there is a relationship between market timing attempts and change of capital structure is hotly debated and conclusions are controversial. Does marketing timing exist? Will firms act when they perceive market timing? If firms tend to perceive market timing as a good way to go IPO, will IPO during market timing change firms’ capital structure? Is a short-term effect or persistent effect? To answer these questions, I collect data from COMPUSTAT and SDC platinum and restrict my sample to only include the US companies. The sample includes data from 1990 and 2017 and uses cross-sectional data to test several hypotheses. The IPO volume in US stock market over this period shows a pattern that there are clusters where firms collectively go public or firms collectively choose not to go public. I find that firms go public in good market condition and they earn higher proceeds from IPO. Firms chose to go for market timing attempts have a lower book leverage prior to IPO compared to firms go public in the inactive market. In the IPO year, firms with market timing attempts increase their book leverage ratio compared to firms without market timing attempts and evidence shows that firms with market timing attempts have a higher capital expenditure and R&D in the IPO year and one year later. It is consistent with the equity market timing hypothesis but is not consistent with the dynamic trade-off hypothesis. Firms tend to expand their business by raising capital from equity and debt at the same time. However, there is no long-term effect of market timing attempts at changing the capital structure.

Contents

1.

Introduction ... 6

2.

Literature review ... 8

2.1. Capital Structure: Trade-off Theory and Pecking Order ... 8

2.2. Capital Structure: Marketing Timing Theory ... 10

2.3. The relationship between market timing and capital structure ... 10

3.

Data ... 12

3.1. Variables and their summary statistics ... 12

3.2. Hot and cold market ... 15

4.

Methodology ... 17

4.1. Hypothesis and Model Specification ... 18

4.2. Econometric methodology ... 19

5.

Results: the relationship between market timing and capital structure ... 20

5.1. The trend of filing Initial Public Offering in Hot Market... 20

5.2. The impact of market timing on firm’s capital structure in the short-term

... 25

5.3.

The persistent effects of market timing on firm’s capital structure ... 27

6.

Robustness Check ... 29

7.

Conclusion and Discussion ... 31

1. Introduction

In corporate finance, equity market timing is defined as the practice of issuing shares when the market values are high and repurchasing them when their market values are low. The purpose is to using market timing as a strategy to obtain benefits with possible temporary fluctuations in the cost of equity, compared to the cost of other sources of capital. This practice would allow managers to use specific market times to form the capital structure of their organisations and take advantage of the lower cost of capital at the time.

In both practical and theoretical ways, a lot of studies and articles on famous financial newspapers indicates that market timing is costly, unprofitable and unsuccessful on average. On the one hand, Chang & Lewellen (1984) mentioned in their study that there are only a few managers appear to have many market-timing skills and they are not able to outperform a passive investment strategy

collectively. On the other hand, in an anonymous survey of Graham and Harvey’s paper (2001), around two-thirds of CFOs admit to market timing. They find that the item- “the amount by which our stock is undervalued or overvalued” was seen as an important or very important consideration by managers who plan to issue or issuing equity. As a whole, market timing is considered more important than 9 out of 10 factors when making decisions to issue equity in the survey.

The issue of equity is naturally related to the capital structure because raising fund from issuing equity is a vital way of financing for many companies. Alti (2006) mentioned that if there is an impact of equity market timing on the capital structure, high persistence of market timing impacts would imply very loose leverage targets, suggesting a minimal role for traditional determinants of capital structure. According to Modigliani-Miller Theorem, in a perfect capital market where no taxes, no transaction costs, no financial distress or bankruptcy costs, equivalence in borrowing costs for both companies and investors, no asymmetric information and no effect of debt on a company’s earnings before interest and taxes, the costs of choice between debt and equity do not vary independently. So, there is no benefit from capital structure choice. But in an inefficient or segmented capital market, market timing benefits current shareholders. When a firm wants to finance its business, the firm is likely to make this decision by referring to market timing. Therefore, it is significant to figure out the relation between market timing and whether the impact persists.

about this question, especially about the lasting period of the impact. Because of those competing theories of this topic, this paper will examine the impact with a more updated dataset compared with previous studies. Moreover, according to the report written by Fresno (2018) and World Federation of Exchange (Desjardins), there are 16 exchanges that are parts of the “trillion-dollar club”, the total market capitalization of them comprises 87% of the world’s total value of equities. Among those stock exchanges, the United State hold 37.53% of the world markets with US$ 25,935 billion value.

Therefore, data from the stock market of the United State are representative of the world’s equity market. The paper will use data from the U.S capital market and is useful for managers who want to achieve their target leverage level before or during the period of IPO. Also, shareholders can have a better understanding of the relationship between market timing and capital structure.

Hence, the main question of interest is whether there is a persistent impact of equity market timing of initial public offering on the corporate capital structure. Most of the prior studied choose to use the same method to identify the market timers as Baker and Wurgler (2002). The way they identify market timers is to consider those firms have a history of raising capital when the market-to-book ratios are high. And if the impact on the capital structure still exists beyond 10 years, they will regard it as a persistent timing effect. But, this measure is subject to lots of criticism. As Alti (2006) indicates in his study, a history of simultaneous increases in the need of raising external funds and in the market-to-book ratio is possible to proxy for different characteristics of firms. For instance, the long-term growth feature that dictates low optimal leverage ratios, it may lead to a spurious relation between history and firms’ capital structure. We need to use a different measure for market timing when analyzing its long-run impact on the capital structure.

This study will use the hot-cold market measure that is similar to Alti (2006) with an updated dataset that allows comparing the result with prior studies in which the sample is from an older dataset. The measure indicated by Alti (2002) is that whether the IPO happened in a hot issue market which has high IPO volume in terms of the number of issuers, or a cold issue market. The main finding of this paper is that firms with market timing attempts tend to use the raised capital from IPO and increase book leverage to expand their business instead of using it to reduce their debts.

The remaining part of this paper is organized as follow. Section 2 provides a thorough overview of prior studies in the same research field. Section 3 includes the data sources, summary statistics of key variables and descriptive figures about the IPO volume during 1990-2017. In section 5, the results of

empirical analyses of the impact of market timing on the corporate capital structure and the robustness check are presented in detail. Section 6 will show the robustness test. And section 7 concludes and discusses the result and contributions of the study.

2. Literature review

Corporates need to raise capital to fund their operations and growth. In general, there are two sources for corporates to raise capital that is debt and equity. Capital structure is about the proportion of debt and equity which is known as the leverage ratio. One of the most significant decision corporates need to make is choosing from debt and equity to fund their operations and future growth. Debt comprises short-term debt and long-term debt. Equity mainly comprises common stock, preferred stock, and retained earnings. In the following sections, we would discuss the reasons why firms adjust their capital structure and the main factors that would induce the firm to adjust their capital structure. There are three mainstream theories introduced by previous studies explaining firms’ financing choices. They are trade-off theory, pecking order theory, and marketing timing theory.

The first famous paper about capital structure comes from Modigliani and Miller (1958). Academia summarised two core propositions from their papers: capital structure irrelevance and dividend policy irrelevance. These irrelevance propositions state that capital structure and dividend policy are

irrelevant to the firm’s value in an ideally perfect markets capital. It means that in a perfect market firms do not concern about their capital structure at all. However, Modigliani-Miller theorem has received fierce criticism about their assumptions of the perfect market. In the perfect markets they assumed, there are no taxes, no transaction costs, no bankruptcy costs, no tax shield effect, the borrowing costs are identical to every market participant and all of the market participants have symmetric information. Apparently, these assumptions about the capital market are difficult to be fulfilled in the real world.

2.1. Capital Structure: Trade-off Theory and Pecking Order

Successors of Modigliani and Miller’s theory started to release some assumptions of the perfect market and trade-off theory started to emerge. Trade-offs theory emphasizes the trade-off between

costs and benefits of debt and equity. In other words, because of the tax shield that is brought in by increasing debt issuance, firms are required to find an optimal balance between tax shield and financial distress.

Kraus and Litzenberger (1973) release assumptions that no tax and no bankruptcy cost. In their study, they find that due to taxation benefits on corporate profits and the presence of bankruptcy costs, the choice of financial leverage level has an effect on the firm’s value. Ross (1977) releases the

assumption of no hidden information. He argues that insiders have an advantage of knowing more information about the firm and they know whether their firms are undervalued or overvalued. They are able to take advantage of asymmetric information and adjust their capital structure to increase the market’s perception of value. Also, the managerial incentive compensation incentivizes insider to adjust the capital structure to increase the firm’s value. Myers (1977) releases the assumption about borrowing cost. In his theory, the firm’s assets can be valued as a call option and the value of this kind of option relies on future investment. However, the future investment decision is easy to be influenced if the firm issued too much risky debt. Because too much issuance of risky debt would force the firm to choose a suboptimal strategy. The market value of the firm would reduce if the firm follows the suboptimal investment strategies, so there is a negative relationship between corporate borrowing and firm value. Thus, trade-off theory predicts that firms benefit from moderate borrowing under this situation.

After fierce discussion and study about trade-offs theory, Myers and Majluf (1984) put forward to a new idea about the capital structure which is the origin of the pecking order theory. Pecking order theory states that when a firm is going to raise capital for funding its operation, the first choice should be raising capital through internal funds, and the second choice would be issuing debt, and finally issuing equity. They propose investors tend to be rational to interpret firms’ issuance of equity and insiders (management) have more information about the intrinsic value of the firm than investors. They assume investors are not able to precisely value firms’ existing assets and new investment opportunities. If investors found that the firm’s new investment project is with a growth opportunity and positive net present value, they would like to invest in the firm’s shares. In this case, the firm would tend to issue more equity to raise capital for funding their investment projects. Their propositions provide another reason and motivation that firms would adjust their capital structure. Also, it explains why a firm with a lot of internal funds borrows less debt and generally the main

source of external financial comes from debt. However, there is still the trade-off between debt and equity financing. When the firm has too much debt, its stock return would underperform. Billett et al. (2006) show how firms using bank loan financing is linked to their stock returns. They find that over the subsequent three years after the firm announcing bank loans, the stock return of the firm reduces.

2.2. Capital Structure: Marketing Timing Theory

After pecking order theory, scholars start to discuss the role of initial public offerings (IPOs), secondary market offerings in capital structure choice. Two anomalies had been studied by scholars that are the under-pricing phenomenon and hot market phenomenon. Market timing theory triggered heated discussions. They argue that when the management know that the firm’s stock is overvalued by investors and market, they would like to issue new equity to grab opportunities of earning abnormal returns. Although they know the mispricing would be corrected, and they may earn less return because of this, management of those firms will still obey this rule. Ibbotson et al. (1988) gather and analyses data of 8668 IPOs listing between 1960 and 1986, and they conclude that there is a 16.4% average initial return on the first trade day. In some “hot” periods, the initial returns on the first trade day are even higher because investors would be overoptimistic in these periods. Also, more firms have the tendency of issuing equity during these periods. Baker and Wurgler (2006) show that investor sentiment also plays an important role in stock returns. They categorize stocks into seven different types: small stocks, young stocks, extreme growth stocks, non-dividend-paying stocks, unprofitable stocks, distressed stocks and high volatile stocks. Interestingly, there is a negative relation between proxies for investor sentiment and the subsequent return of all types of stocks stated above. If investors have a low level of sentiment proxies at the beginning, it is more possible that these

categories of stock would earn higher subsequent returns and vice versa. It explains why firms tend to issue equity by taking advantage of this phenomena. To summarize, the market timing hypothesis states that firms are keeping for cheaper capital and in general previous capital structure play a small role in firms’ decision-making process.

2.3. The relationship between market timing and capital structure

As a pioneer in this field of study, Baker and Wurgler (2002) find that as they measure leverage by using “external finance weighted-average market-to-book ratio”, firms with low leverage level tend to raise fund when their market valuations were high, while high leverage firms would prefer to raise funds when their market valuations were low.

For proving the persistent impact of past market valuation on capital structure, they document it in three separate ways. Firstly, they use leverage regression after controlling for the current market to book ratio for getting the within-firm time series variation and explaining capital structure outcomes. It implies that the fluctuations in market valuations could cause the permanent changes in capital structure. Afterward, they test for persistence by controlling the initial capital structure level and look at the regression for the change of capital structure level from its initial level caused by the subsequent fluctuations in market valuations. The third test is to look at the effects of the lagged values of the weighted average market to book ratio on capital structure. They conclude that their results are consistent with their hypotheses that market timing has a significant and persistent effect on the corporate capital structure. However, I will use a different method for testing the existence of persistent impact of market timing on leverage level due to some criticisms of the measurement method of market timing in Baker and Wurgler’s study.

Alti (2006) shows the short-term persistence of the impact of market timing on capital structure and suggests that after going public, firms in hot-market that is characterized by high IPO volume in terms of the number of issuers would increase their leverage ratios by issuing more debt and less equity relative to those firms in cold-market. Moreover, Alti finds that the impact on the leverage of firms would entirely vanish at the end of the second year after the IPO. And this finding is contradictory to the result of Baker and Wurgler (2002). Also, Alti (2006) mentioned that the simultaneous increases in external funding needs and the market to book ratio is likely to proxy for a firm’s characteristics and dictate low optimal leverage. The simultaneous control variables are noisy proxies for firm

characteristics and there will be a spurious relation between past and capital structure obtained. Alti’s measure of market timing is relatively direct: whether the IPO takes place in a hot issue market or not. For IPOs in the cold market, it is more possible to keep their issues to a minimum due to the less favourable market conditions. In my thesis, I would use Alti’s market timing measure which is to establish the hot-cold market classification. Because by doing so, I can avoid some problems indicated in criticisms of the methodology of Baker and Wurgler. Similar to findings of Alti’s paper (2006), Hovakimian (2006) confirms the negative effect of historical market-to-book ratios on US corporate leverage, but do not corroborate its long-term persistence of the impacts.

On the contrary to findings of all the papers stated above, De Bie and De Haan (2007) do not find a significant relationship between the market timing theory (Baker et.al., 2002) and corporate leverage for the Netherlands and many other markets.

3. Data

This main sample of the study is from COMPUSTAT database by using which we could determine the IPO date and firm characteristics over the period of 1990-2017. Knowing the IPO allows us to investigate changes in leverage and other related variables before and after the IPO date. For data about IPO event, I collect the data from Securities Data Company (SDC).

The IPO year is defined as extracting the year from IPO date from COMPUSTAT dataset. This sample does not include SIC code between major group 60 and major group 67. United States Department of Labour classifies this range under Division H: Finance, Insurance, and Real Estate. The reason for excluding this division in the sample is that companies or banks from this division have many assets or liabilities that are off-balance sheet. It might cause difficulty to interpret the change in their capital structure. Furthermore, firms without complete data of total assets and firms have only one-year data will be excluded from the sample. Total assets are an important factor to scale variables and taking firm size into consideration and at least two-year data available is also crucial due to the calculation of the difference of variables. The definition of variables used in this study is described as follows and capital letters in the bracket are variable names that are defined by COMPUSTAT dataset.

3.1. Variables and their summary statistics

To summarize the sample, I create similar variables and construct a similar summary statistics table like Alti (2006). D (book debt), is defined as the sum of total liabilities (from COMPUSTAT Fundamentals Annual database, LT) and preferred stock (PSTK) minus deferred taxes showed on balance sheet (TXDB) and convertible debt (from CRSP/COMPUSTAT Merged Fundamentals Annual: DCVT). E (book equity), is total assets (AT) minus book debt we defined above. D/A (book leverage), is calculated as the book debt divided by total assets. If observations with book leverage are larger than 100%, the observations will be dropped. M/B (market-to-book ratio), is equal to the sum of book debt and market equity (the product of common share outstanding (CSHO) and stock price (PRCC_F)) divided by total assets. By producing boxplot of M/B, outliner with the high market-to-book ratio is spotted and dropped from the sample.

d/A (net debt issues), is defined as the change in D. e/A (net equity issues), is the change in book shareholders’ equity (SEQ) minus the change in retained earnings (RE). ∆RE/A is the change in

retained earnings and scaled by total assets as well. EBITDA/A is used to measure the profitability and it is earnings before interest, taxes, and depreciation (EBITDA). SIZE is defined as the nature logarithm of total revenue (REVT). Because it is impossible to calculate nature logarithm of a negative total revenue, SIZE is treated as missing value if the observation with negative total revenue. PPE/A (asset tangibility) is the net plant, property, and equipment which can be found directly in the database (PPENT). R&D/A is the research and development expense (XRD) and it is replaced by zero when the value is missing. For doing the regression analyses below, there is dummy variable RDD generated in order to take the value of one when the research and

development expenses item is missing. DIV/E denotes common dividends (DVC) divided by year-end book shareholders’ equity. CASH/A is defined as cash and short-term investments (CHE). INV/A is the capital expenditures (CAPEX). The variables ND/A, NE/A, ∆RE/A, EBITDA/A, PPE/A, R&D/A, INV/A, and CASH/A are scaled by total assets to control effect of firm size. Firm-year observations having d/A, e/A, ∆RE/A, EBITDA/A, INV/A, or DIV/E larger than 100% in absolute value are dropped. After the above data filtering criteria, the sample size reduces to 6107 observations and 1599 companies are included.

IPO year is defined based on IPO date and it is a variable (IPO year) generated as the fiscal year in which the firm goes public. IPO year+n is the nth fiscal year after the IPO. Pre-IPO is IPO date is later than the current fiscal year and IPO is IPO date within the current fiscal year. IPO+1, IPO+2, IPO+3, IPO+5, and IPO+7 are one year, two years, three years, five years and 7 years after IPO respectively. The whole sample includes only 7 observations having financial data before their IPO date and only 13 observations having financial data in the IPO year. Observations with financial data available increase significantly one year after their IPO. 1275 observations have financial data one year after their IPO. 1048, 814, 519 and 367 observations have financial data after two years, three years, five years and seven years after IPO respectively. Availability of financial data before firms’ IPO (Pre-IPO) is really low (only 7 companies in the sample with pre-IPO financial data) compared to studies Alti (2006) and Baker & Wurgler (2002). Before applying for any data selection criteria, there are only 57 companies with pre-IPO financial data. Because my study is about the US market, generally private companies are not required to publish their financial statements before going IPO. Although SEC has the requirement about IPO filing that publishes part of their previous fiscal year income statement and balance sheet, COMPUSTAT does not include these data.

Table I shows the summary statistics of firm characteristic and financing decisions of the sample. The first row shows the summary statistics for the whole sample and it is an overview of the sample.

11792 observations and 1399 companies are included in the sample. It is the sample where all observations have pre-IPO financial data. It is really important to conduct our methodologies to solve our research questions. This sample will be subset into 7 subsamples based on variables

defined above: IPO year+n. The first column of table 1 tells the story. After firms went public (IPO), the number of observations decreases over time because they are delisted or exit the COMPUSTAT database. After 7 years since their IPOs, the sample size reduces to 532. Book leverage (D/A) that is shown in the second column is an important variable for my research and it is used to measure capital structure. It decreases significantly by 10% compared to book leverage in pre-IPO. Mean of Book leverage stays at a stable level over time after IPO. Market-to-book ratio also changes

significantly after IPO. The mean market-to-book ratio in pre-IPO is smaller than 1 but after IPO the mean of its increase to more than 1.5. The variable: Change in book debt shows that on average firm capital structure changes. The proportion of debt in the capital structure reduces. When it comes to change in retained earnings and profitability, the mean of them decreases a bit after IPO. Although mean of change in retained earnings and EBITDA decreases after IPO, the mean of SIZE (nature logarithm of total revenue) grows gradually after the IPO year. The mean of cash holdings made a small jump after IPO and then it stays at around 14.50%. The mean of dividend payout ratio, capital expenditure, and the R&D expense reduces after IPO year. My result is consistent with Alti (2006) and Baker & Wurgler (2002) and variables changed over time following a similar pattern.

Table 1

Summary Statistics of Firm Characteristics and Financing Decisions

The table shows the means and the standard deviations of several firm characteristics. Except for M/B and SIZE, the rest are in percentage terms. Letter “A” standards for total assets in the following sentences and table. All variables are scaled by total assets to control the effect of company size. Book Leverage, D/A, is the ratio of book debt over total assets. Market-to-book ratio, M/B, is defined as book debt plus the market value of equity divided by total assets. Net debt issues, ND/A, is the change in book debt over total assets. Net equity issues, NE/A, is the change in book shareholders’ equity minus the change in retained earnings. ΔRE/A can be used to measure the change in retained earnings. Profitability is measured by EBITDA/A, which is earnings before interest, taxes, and depreciation over total assets. SIZE is the logarithm of total revenue. Asset tangibility, PPE/A, is defined as net plant, property, and equipment over total assets. R&D/A is the research and development expense. INV/A is capital expenditures. over total assets. DIV/E is common dividends divided by year-end book shareholders’ equity. The sample consists of IPOs between 1990 and 2017. The sample contains only company with COMPUSTAT information before their IPO year. The sample excludes financial firms, spinoffs and unit offers with SIC between 6000-6999. Firm-year observations that book leverage, net debt and net equity issues, change of retained earnings, EBITDA/A, INV/A and DIV/E is greater than 100 are dropped from the sample. Only market-to-book ratio of firm-year observations is lower than 10 will be kept in the sample.

N D/A M/B d/A e/A ∆RE/A EBIT

DA/A

SIZE PPE/A R&D/ A

INV/A DIV/E CASH /A All 11792 47.52 (22.48) 1.52 (1.09) 5.63 (20.06) 6.29 (20.92) 1.29 (12.83) 11.31 (12.22) 4.66 (2.19) 35.36 (27.24) 2.39 (6.37) 7.29 (8.58) 3.47 (9.92) 14.15 (16.88) Pre-IPO 1808 55.92 (25.31) 0.87 (0.79) 10.91 (24.4) 4.52 (27.76) 2.38 (13.09) 13.22 (14.94) 3.29 (2.22) 37.26 (28.42) 2.94 (7.97) 8.85 (10.17) 4.00 (13.18) 11.28 (16.23) IPO 1014 45.41 (21.93) 1.86 (1.25) 1.92 (28.3) 23.21 (32.22) 2.81 (16.16) 13.08 (11.31) 3.43 (1.91) 35.69 (28.02) 1.89 (5.01) 8.89 (10.47) 4.31 (13.79) 14.27 (17.57) IPO+1 1094 44.24 (22.27) 1.81 (1.26) 10.41 (20.26) 6.96 (18.58) 2.42 (13.11) 11.19 (12.05) 3.76 (1.83) 35.26 (28.22) 2.42 (6.59) 9.05 (10.37) 2.86 (9.5) 15.47 (18.98) IPO+2 1014 45.84 (22.27) 1.68 (1.22) 7.37 (19.11) 6.17 (17.76) 0.78 (12.88) 10.48 (11.83) 4.16 (1.73) 36.37 (28.21) 2.4 (5.76) 7.66 (8.31) 2.99 (9.2) 14.44 (17.95) IPO+3 900 45.78 (21.97) 1.61 (1.14) 5.56 (18.76) 6.25 (17.16) 0.16 (12.69) 9.57 (13.09) 4.45 (1.84) 36.01 (27.9) 2.49 (6.52) 6.94 (8.16) 2.86 (8.13) 14.4 (17.83) IPO+5 693 46.57 (22.05) 1.54 (1.03) 4.92 (18.06) 4.82 (15.89) 0.44 (12.81) 10.36 (11.89) 4.9 (1.89) 35.85 (26.95) 2.18 (5.65) 6.49 (7.28) 2.84 (7.84) 14.57 (16.97) IPO+7 532 46.23 (21.33) 1.58 (0.95) 3.34 (14.57) 4.45 (13.51) 1.06 (11.11) 11.15 (11.18) 5.14 (2.00) 35.34 (26.62) 2.04 (4.5) 6.41 (7.29) 2.91 (7.48) 14.21 (15.63)

3.2. Hot and cold market

Figure 1 shows the IPO volume in the North America market from 1990 to 2017. The data is from the COMPUSTAT dataset and the sample contains 10856 firms’ IPOs between 1990 and 2017. By

extracting the year from the IPO date variable in COMPUSTAT dataset of each firm we can calculate IPO volume in each year. The sample size is larger than the sample used in summary statistics. Because this sample includes all companies before applying data selection criteria set up for the previous sample. The IPO market was very active before the millennium and the volume has reached the highest point which means there were 1083 firms choose to go public in 1997 according to our sample. But after dot.com bubble, IPO volume has suddenly fallen 81.79% in 2001 compared with one year before and then it remains at a relatively low level from 2001 to 2003. However, the IPO volume reaches a small spike and the yearly average of IPO volume has reached 304 between 2004 and 2007. And shortly after the warm-up situation, the financial crisis in 2008 storms the world and the IPO volume shrinks to the level it was after-dot-com-bubble. After two years since the burst of the financial crisis, the impact of the financial crisis has gradually faded away and the IPO volume has increased to a moderate level and with small changes each year.

Figure 1

Initial Public Offering (IPO) from 1990 to 2017

To define hot and cold market, I need to make some adjustment to IPO volume to smooth out seasonal variation such as short-term fluctuations and long-term trends. I use 3-month centered moving average of the volume of IPO for each month because 3 months is a quarter in terms of financial statements. Furthermore, I further detrend this time series data by removing economic growth effect. Average GDP growth rate of the US from 1990 to 2017 is 2.42% and 0.20% per month. I multiply 0.20% by each 3-month centered moving average for each month. The hot market is defined as the 3-month centered moving average of IPO volume is above the median of the distribution of the detrended moving average IPO volume across all periods in the sample. The definition of the hot or cold market is really important for this study to identify firms’ market timing attempts. The median is 22 that is

166 384 473 701 628 740 1083 904 986 932 593 108 101 ₆₇ 272 280 268 394 111 ₈₇ 216 162 163 224 280 187 122 224 0 200 400 600 800 1000 1200 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

shown as the line in the graph. In this sample 336 months are included, 171 months (51%) are defined as “hot” month and 165 months (49%) are defined as “cold” month. There are 10787 IPOs after IPO volume detrended between 1990 and 2017. 8790 IPOs (74%) took place in the hot market and 1997 (26%) took place in the cold market. Speaking of the sample for regression analysis, among 11792 observations where 8710 observations (73%) are defined as “hot” and 12869 observations are defined as “cold” (27%). From figure 2, it shows significant the difference of IPO volume between the hot market and cold market. Prior to dot.com bubble around the beginning of 2001, there are huge

numbers of firms choosing for going public but from 2001 there are just so a few firms choosing to go public. Until 2004 the market became good and firms start to collectively go public. But after the financial crisis in 2008 market became quiet again IPO volume stay around or below the 27-year median of IPO volume. For the later regression study, I create a dummy variable to identify the hot and cold market that is equal to 1 if the moving average of IPO is above the median.

Figure 2

Detrended 3-month centered moving average of IPO volume in the North American capital market from 1990 to 2017

4. Methodology

firstly I need to figure out whether firms regard the hot market as a good timing for filing IPO. If firms tend to file their Initial Public Offerings in the hot market, whether there is a short-term or long-term impact of market timing on their capital structure. In the following sections, I will propose my hypotheses, models, and methodology to testify my hypotheses.

4.1. Hypothesis and Model Specification

Firms tend to be driven by market timing if they can benefit from filing IPO during the hot market. As our sample shows that 80% of firms filed their IPO during the hot market. The benefits can be

measured by proceeds from primary shares issued, total IPO proceeds from shares issued and the ratio of offer price over year-end book value per share. Apart from these direct benefits from IPO, there are also some indirect benefits such as deleveraging,

As stated above, this study will test three hypotheses:

Hypothesis 1: There is an impact of equity market timing on the issuance activity of companies.

Yt=k0+k1HOT+k2M/BIPO+k3EBITDA/At-1+k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7SIZEt-1 k8M/Bt-1+ k10SIC+...+ knSIC +εt (1)

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1+ k4SIZEt-1+k5PPE/At-1+k6R&D/At-1+k7RDD t-1+k8SIZEt-1 +k9D/At-1+ k10SIC+...+ knSIC εt (2)

In the regression model (1), the dependent variable Yt is defined as the total IPO proceeds divided by year-end assets, the proceeds from primary shares divided by year-end assets, the proceeds from primary shares divided by beginning- of-the-year assets, primary shares issued divided by the total outstanding shares at year-end, offer price divided by the per-share book value at year-end, and the IPO-year net debt issues divided by year-end assets respectively. This regression is designed to test whether hot market firms earn higher proceeds from IPO, so they have the motivation to go for market timing attempts by going public. In the regression model (2), the dependent variable Yt is defined as pre-IPO book leverage ratio, capital expenditure, EBITDA and dividend payout ratio. This regression is designed to test their motivations that are driven by pre-book leverage ratio, business expanding, profitability and dividend payout.

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1 +k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/At-1+ k8SIZEt-1+k9SIC+...+ knSIC +εt (3)

In this regression model (3), Yt is the change in book leverage, net equity issues, the change in cash, the change in other assets, the change in retained earnings and book leverage ratio.

Hypothesis 3: The impact of equity market timing on the corporate capital structure is persistent. Yt=k0+k1HOT+k2M/Bt-1+k3EBITDA/At-1 +k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/APRE-IPO+

k8SIC+….+ knSIC +εt (4)

The dependent variable Yt is the cumulative change in book leverage from the pre-IPO year to years

IPO+1 and to IPO+2, level of book leverage, and the regression excludes the pre-IPO book leverage. To summarize the process of hypothesis testing, I first test whether IPO volume can be treated as a good indicator for capturing firms’ market timing attempts. If so, I can use this dummy variable to test the short-term and long-term effect of market timing attempts on changing firms’ capital structure.

4.2. Econometric methodology

Based on the data structure, the sample used for later regressions is panel data where each firm has many years of financial data. However, panel data analysis is not a good fit for testing my hypotheses. For instance, if a firm went IPO in 2000 but the sample includes data from 1990 to 2017, panel data does not help to answer whether this IPO event in 2000 has a persistent effect on its capital structure. Panel data would be a better choice if I want to figure out whether there is a relationship between two variables such as the relationship between capital structure and firm value. For my research questions, cross-sectional data study is a better choice. If I subset the sample into several subsamples based on variables: IPO+n year, the data structure will become cross-sectional data. These subsamples will help to answer our research questions. These subsamples act like the snapshot of firms before or after their IPO. Ordinary Least Squares (OLS) is a good estimator for our models. To get a robust coefficient from the OLS estimator, I control industry effect based on SIC code that categorizes companies into a different industry by creating a dummy variable for every SIC into regression models.

5. Results: the relationship between market timing and capital structure

If market condition is good or firms perceive market condition is good, firms would believe that they can earn higher return from IPO by selling more shares with higher price. In the previous section, the sample shows that there are 85% companies choose to go public in hot markets. It reflects that market timing attempts of firms. In this section, I am about to dig into it to illustrate behavior of market timing attempts by firms, and costs and benefits of going public in hot markets.

In the following paragraphs, four reasons why firms would choose to go public in hot market will be tested by linear regression and group mean difference. First of all, if the firm has a high leverage ratio before IPO, the firm is likely to use IPO to reduce its leverage ratio. Second, if the firm anticipate that its business is in a rapid growing stage, the firm may consider finance its business by raising equity capital. Third, firms with low profitability is very likely to go public in the hot market because it is not easy for them to go public in cold markets. Last but not least, if the firm goes public in hot market, the firm can use the proceeds from IPO to payout its dividend. Because they guess proceeds from IPO is higher than proceeds from IPO in cold market. It ensures the firm has fund to payout its dividend by proceeds.

Table 2 shows the comparison of hot markets and cold markets firms in terms of pre-IPO book leverage, capital expenditure, EBITDA and dividend payout ratio. Panel A in Table 2 shows the comparison of the mean value of variables for hot market firms and cold market firms. Panel B shows regression result of hot market indicator on dependent variables shown in the table. Mean value of pre-IPO book leverage of hot market firms is higher than that of cold market firms and it means that hot market firms have a lower book leverage before they go public compared it to cold market firm. But the difference in the mean is not statistically significant. The coefficient of the first regression is statistically significant. It shows if firms go public in hot market, the firms’ previous book leverage is 7.63% less than firms go public in cold market. In other words, firms with low book leverage like to go public in hot market. Firms probably try to reduce their leverage ratio to their leverage ratio target because they are might overleveraged before their IPO. Therefore, this result is consistent with the first motivation stated above and it meets our expectation.

firms on average invest more capital into capital expenditure in the IPO year and the mean difference is statistically significant. It means that the second motivation stated above is plausible. The regression result shows that hot market firms have a higher capital expenditure compared to cold market firms, but the coefficient is statistically insignificant. The t-statistics of the following two columns for IPO+1 and IPO+2 are statistically insignificance and firms’ motivation to go public in hot market to raise more fund for future capital expenditure cannot be proved from these results. The third motivation about profitability is tested by regressing of hot market dummy on EBITDA and the mean value of EBITDA. The mean values show that hot market companies have a lower EBITDA compared to cold market firms except for data in IPO year. These mean differences are statistically insignificant, and the third motivation cannot be shown by reading these mean values. Linear regression results are also mixed, although the sign of coefficients is consistent with our expectation. Only coefficient of IPO+1 column is statistically significant and hot market companies has a lower EBITDA compared to cold market companies by 1.78%. The fourth motivation about dividend pay-out cannot be confirmed because t-statistics for mean difference and coefficient are statistically insignificant. Therefore, the motivation that firms tend to use proceeds from IPO to mitigate their pressure from paying out dividend before IPO.

Table 2

The Investment Regression and Profitability Regression for Hot and Cold Firms

The dependent variables for the following regressions are the debt to assets ratio for pre-IPO year, the investment divided by the assets in IPO year, one year after, two years after IPO and the EBITDA divided by assets in IPO year, one year after IPO, two years after IPO and three years after IPO respectively. The mean values of dependent variables for hot and cold firms are shown in Panel A and the t-statistics for the differences of mean values are in the parentheses. Panel B reports the coefficients of the regression model

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1+k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/At-1+k8SIZEt-1+k9SIC+….+ knSIC +εt

The intercepts are not reported in the table. Regressions in Panel B are estimated with industry-fixed effects defined by Standard Industry Codes. Values in the parentheses below coefficients are t-statistics. Market to book ration in IPO year, market to book ratio in previous year, EBITDA divided by assets in previous year, PPE divided by lagged assets, R&D expenses divided by lagged assets and the dummy variable indicating the availability of R&D expenses in COMPUSTAT dataset are included in the regression as control variables. The regression results with four different dependent variables and Hot as independent variables are shown in Columns 1-9 respectively.

INV/At EBITDA/At

t D/A

PRE-IPOt

IPO IPO+1 IPO+2 IPO IPO+1 IPO+2 IPO+3 DIV/EIPO

Panel A: Mean Values

Hot 56.70 9.59 9.02 7.49 13.15 10.87 10.14 9.52 3.74

Cold 56.55 7.64 9.10 8.00 12.95 11.78 11.17 9.70 4.55

Table 2 (Continued)

Panel B: Regression Analysis

HOT -7.63 (-1.95) 1.02 (0.86) -0.02 (-0.01) 0.14 (0.14) -0.01 (-0.01) -1.78 (-1.63) -0.66 (-0.52) -0.23 (-0.13) -1.33 (-0.34) M/BIPO — __ 0.21 (0.67) 0.19 (0.55) -0.21 (-0.73) 2.39 (5.24) 2.35 (5.38) -0.83 (-1.61) -0.13 (-0.17) 1.58 (1.37) M/B-1 — __ __ __ 1.03 (2.53) __ __ __ __ 3.75 (5.20) 3.81 (5.95) __ __ EBITDA/At-1 0.22 (1.54) 0.03 (0.48) 0.04 (0.60) 0.06 (1.02) __ __ __ __ __ __ __ __ 0.24 (1.61) SIZEt-1 -1.31 (-1.62) -0.89 (-3.16) -0.29 (-1.11) -0.87 (-3.10) 0.70 (2.67) 0.40 (1.41) 0.95 (3.05) 1.20 (2.16) 1.01 (1.79) PPE/At-1 -0.20 (-1.67) 0.22 (6.69) 0.21 (7.53) 0.21 (8.36) 0.05 (2.00) 0.05 (1.49) 0.20 (0.63) 0.07 (1.96) -0.05 (-0.51) R&D/At-1 -3.01 (-6.70) 0.002 (0.02) 0.02 (0.14) -0.09 (-1.15) 0.08 (0.38) 0.29 (1.27) 0.20 (0.78) 0.33 (0.98) 1.08 (2.51) RDDt-1 3.36 (0.69) 2.03 (1.61) 0.54 (0.46) -0.56 (-0.58) -0.15 (-0.13) 1.03 (0.82) 0.75 (0.54) -0.89 (-0.63) -8.50 (-1.46) R2 _{63.82% 47.58% 40.95% 41.83% 31.65% 21.57% 25.43% 30..61% 11.2%} N 1808 1014 1094 1014 1808 1094 1014 1808

Table 3 shows whether firms chose to go public in hot market would earn more proceeds from IPO. I collect data from SDC platinum to illustrate my conclusion. To measure the direct benefits from filing IPO, I collect variable "Proceeds Allotment included Over Sold -in this Market ($, million)", “total proceeds”, “Offer Price” and “Primary shares offered – sum of all markets (million)” to generate my variables. ProceedsT _{is total proceed of IPO from the first day to the end of fiscal year and Proceeds}P _is

total proceed of IPO in the first day of IPO. To control the effect of firms’ size on IPO proceeds, I scale them by total assets of fiscal end and beginning of current fiscal year. Offer/Book is dividing offer price by the year-end book value of per share. By these three variables I can measure the direct effect of IPO on firm’s benefits. Furthermore, I create %issued to capture the behavior of firm in hot market and cold market and whether firm chose to go public in hot market will issue more shares in the IPO date. d/At is change of book debt that can be used to measure the change of book debt in the

Panel A in table 3 shows mean value of the series of variable in hot market and cold market. Results show direct benefits from going public in hot market. All data used for Table 3 are IPO year data. From column 1 to column 4, the number of Hot is greater than Cold. it shows on average firms filed their IPO in hot market receive higher proceeds from primary shares issued and total proceeds. Mean value of Offer/book also shows hot market firm turns out with a higher valuation by market

participants. %issued shows firms’ behavior concerning about primary shares issued in the IPO date. Firms go public in hot market tend to issue more shares in the first IPO date by taking advantage of good market condition. Based on t-statistic shown in the brackets, total proceeds and proceeds from primary shares issued are statistically significant. The last column shows that firms go public in hot market reduce their book debt by less amount compared to firms go public in the cold. It is statistically significant but it contrary to my expectation.

To show the robustness of these mean values, I use boxplots to show there are no outliners in the sample. Table 4 shows outliners are excluded from the sample to make sure the mean value is not affected by outliners. Panel B shows result of linear regressions and four out of five main coefficients are statistically significant at 5% and 10% significance level. The result of first column shows that firms go public in hot market total proceeds from IPO is higher than firms go public in cold market by 5.23%. Speaking of economic significance, 5.23% of the mean value of total proceeds in hot market is equal to 1.86 million US dollar. It is a decent earning for firms. The regression result of proceeds from primary shares issued also shows a similar result that there is positive relationship between going public in hot market and proceeds from primary shares issued. However, when the proceeds from primary shares issued is scaled by total asset of previous fiscal year, it is contrary to my expectation, marketing timing effect should be significant (even though the coefficient shows a significant effect of marketing timing on proceeds, it is not statistically significant) when measured relative to total assets of previous fiscal year. Coefficient of Hot in fourth column is also statistically significant and it shows that there is a positive relationship between going public in hot market and primary shares issued in the first IPO day. When a firm goes public in hot market, the firm would like to tend to issue more shares in the IPO first day. Scaled by its year-end total shareholders’ equity, hot market firms turn out issued more than 4.26% compared to cold market firms. Coefficient of hot in fifth column is

statistically significant and it shows that if the firm goes IPO in hot market the P/B ratio is higher than firms goes IPO in cold market by 0.46. It means that a firm go public in hot market it is very likely to receive a higher valuation by market. The last column shows a result that is not what I expect. If the firm goes public in hot market, the firm’s book debt would increase by 5.8% compared to firm goes

public in cold market. Capital structure is changed after firms’ IPO, but the increase of total shareholders’ equity would be canceled out by the increase of book debt.

Table 3

Market Timing Effects on Issuance Activity

In panel A, there are mean values of hot and cold market firms for six different dependent variables that are total proceeds divided by IPO year-end total assets, IPO proceeds from issuing primary shares divided by the IPO year-end total assets, proceeds from primary shares issued divided by the asset at the beginning of year, the percentage of primary shares issued in year-end total shares outstanding, the offer price divided by the book value of each share and the net debt in IPO year divided by the asset at the end of year as shown in the table below. And the subscripts t under every variable refer to the IPO year. Panel B shows the coefficient results of regression analysis of dependent variables stated above, independent variable HOT and other control variables as the form shows below

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1+k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/At-1+ k8SIC+ … + knSIC +εt

RDDt-1 is a dummy variable which will be one if the research and development expenses are not given in the COMPUSTAT dataset. The following

regression are estimated with industry-fixed effects defined by Standard Industry Codes and the t-statistics are in the parentheses. All variables are in percentage term in regressions in Columns 1-6.

ProceedsT_/A

t ProceedsP/At ProceedsP/At-1 %Issued Offer/Book d/At

Panel A Mean Values

Hot 35.53 38.99 38.99 22.63 2.91 -2.20

Cold 31.35 34.54 38.96 22.62 2.73 -8.14

t-statistic (difference) (1.62) (1.56) (0.05) (-0.23) (0.72) (-2.26) Panel B Regression Analysis

HOT 5.23** (2.15) 5.97** (2.18) 9.06 (0.45) 4.26** (2.00) 0.46** (2.03) 5.80* (1.7) M/Bt 4.51 (5.23) 5.21 (5.39) 21.11 (1.74) -3.87 (-4.81) 0.37 (4.79) -0.07 (-0.06) EBITDA/At-1 0.13 (1.40) 0.13 (1.23) -0.57 (-1.14) -0.07 (-0.86) -0.001 (-0.06) -0.04 (-0.31) PPE/At-1 -0.007 (-0.12) -0.008 (-0.12) -0.76 (-2.34) 0.02 (0.35) 0.001 (0.13) 0.163 (1.87) R&D/At-1 0.18 (-0.64) 0.19 (0.60) 448.35 (2.93) 0.18 (0.76) -0.05 (-2.02) -0.46 (-1.14) RDDt-1 1.53 (0.41) 1.57 (0.37) 75.01 (1.46) 5.19 (1.54) -0.60 (-1.65) -10.78 (-1.96)

Table 4

Boxplots of variables about market timing effect on issuance activity

This section aims to test the hypothesis 1: There is an impact of equity market timing on the issuance activity of companies. After analysing previous book leverage ratio, capital expenditure, EBITDA, dividend pay-out ratio and IPO proceeds, IPO market volume is proper indicator for identifying firms’ market timing attempts. Table 3 shows that firms like to issue more shares in hot markets than in cold market because they anticipate earning higher proceeds from IPO in hot market. Furthermore, firms have the possibility to take advantage of IPO in hot markets to reduce their previous book leverage ratio and to raise capital for their business expansion by increasing capital expenditure. Overall, we find that there is an impact of equity market timing on the issuance activity of companies.

5.2. The impact of market timing on firm’s capital structure in the short-term

In the section, it aims to provide quantify answer to the second hypothesis: There is short-term impact of equity market timing on the corporate capital structure. To test this hypothesis, this study regresses

Table 3 (Continued) D/At-1 -0.13 (-2.61) -0.14 (-2.65) -0.86 (-2.41) 0.02 (0.49) 0.018 (3.05) -0.40 (0.72) R2 _{34.55% 33.68% 60.53% 23.22% 45.22% 21.96%} N 401 401 401 401 401 401

hot market indicator on change in book leverage in the IPO year and the result is shown in the first column of table 4. The t-statistic for mean difference is statistically insignificant and it means we cannot reject the hypothesis that the mean value of book leverage of hot market companies in IPO year is the same with that of book leverage of cold market companies. However, the coefficient of the regression is statistically significant. Firms go public in hot market and the book leverage ratio would increase by 3.89%. Our second hypothesis is also proved and there is a positive relationship between market timing attempts and change of capital structure in the short-run. Results from table 2 shows that firms probably tend to use market timing attempts to reduce their previous leverage ratio but the result from table 4 is contrary to it. Firms would not use their proceeds from IPO to reduce their leverage ratio and surprisingly they increase their book leverage. In order to figure out where does the new capital go, I decompose the change of book leverage into change in net equity, change in cash, change in other assets and change in retained earnings where other assets are total assets minus cash. From second column to fifth column are result of these decomposed components. T-statistics for change in net equity, change in cash and change in retained earnings are not statistically significant for mean difference and coefficients, except for other assets. It means that the firm goes public in hot markets, its other assets increase by 0.47% compared to cold market firms. Firms tend to invest the new capital raised by IPO into assets to expand their business rather than mainly reduce their book leverage. The result from last column confirms my speculation because the statistically insignificant coefficient proves that hot market firms does not use the new raised capital to reduce or increase book leverage. The new raised capital is used to another place.

Table 5

Short-Tern Impact of Market Timing on Capital Structure

Panel A shows the mean value for hot- and cold-market firms and the t-value of the difference between them. The subscript t of each variables denotes the IPO year. And Panel B reports the regression result of our six dependent variables, HOT and other six control variables. The regression model is described as follows:

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1 +k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/At-1+ k8SIC+ … + knSIC +εt

The regression in Panel B of following table are estimated with industry-fixed effect using the Standard Industry Codes. And the intercepts for each regression are not included in the table. The dependent variable is the change in debt to asset ratio, net equity issues, the change in cash to asset at time t, the ration of changes in other asset to asset at time t, the ratio of changes in retained earnings to assets and debt to asset ratio respectively. The

independent variable is HOT which is the dummy variable for measuring hot and cold market. Other variables are included in the regression as control variables. Values in the parentheses under coefficients are the t-statistics.

D/At- D/At-1 e/At ΔCash/At ΔOther Asset /At ΔRE/At D/At

Panel A Mean Values

Hot -10.79 22.78 6.59 14.58 2.98 46.23

Cold -12.33 23.55 7.18 10.48 2.75 45.07

t-value (difference) (1.33) (-0.38) (-0.63) (2.50) (0.21) (0.80) Panel B Regression Analysis

HOT 3.89 (2.24) -2.30 (-0.64) -2.98 (-1.71) 0.47 (0.18) -3.08 (-1.52) 2.38 (1.03) M/Bt -0.25 (-0.41) 0.97 (0.80) 2.14 (3.45) 0.52 (0.57) 0.067 (0.93) 0.13 (0.15) EBITDA/At-1 -0.44 (-5.54) 0.51 (3.20) 0.52 (6.48) 0.15 (1.29) 0.67 (7.22) -0.50 (-4.67) SIZEt-1 1.23 (2.94) -1.35 (-1.55) -0.77 (-1.84) -2.32 (-3.70) -1.12 (-2.3) 0.26 (0.48) PPE/At-1 -0.07 (-1.78) 0.26 (3.25) 0.08 (2.05) 0.005 (0.082) -0.11 (-2.53) -0.09 (-1.78) R&D/At-1 -0.02 (-0.09) -0.16 (-0.38) 0.03 (0.15) -0.10 (-0.30) -0.26 (-1.06) -0.19 (-0.67) RDDt-1 -1.40 (-0.69) -2.62 (-0.62) -3.06 (-1.50) 1.02 (0.34) -3.05 (-1.29) -0.10 (-0.04) D/At-1 -0.43 (-11.40) 0.40 (5.15) 0.09 (2.37) 0.07 (1.31) 0.017 (0.38) __ __ R2 _{60.71% 49.7% 39.34% 40.53% 18.76% 87.31%} N 961 961 961 961 961 961

5.3. The persistent effects of market timing on firm’s capital structure

The previous section proves that there is a short-run impact of market timing on capital structure. Results above show that hot market firms have a lower book leverage ratio prior to their IPOs

compared to cold market firm. And then hot market firms increase their book leverage ratio in the IPO year compared to cold market firms. In this section, the third hypothesis: The impact of equity market timing on the corporate capital structure is persistent. To test this hypothesis, I create a variable to capture the cumulative change of book leverage since their IPO. From the linear regression results, all coefficients are not statistically significant and mean differences are also not statistically significant. Because of it, we cannot statistically prove the third hypothesis and there is no relationship between market timing attempts and change of capital structure in the long run. Apart from cumulative change

of book leverage, the next group of four linear regressions. Book leverage is used as a dependent variable to test whether there is a relationship between market timing attempts and book leverage. The result shows also that all t-statistics for mean difference and coefficients suggest they are statistically insignificant. We find there is no relationship between market timing attempts and book leverage after IPO year. All mean value and coefficient of hot market dummy suggest that there is a positive

correlation between market timing attempts and book leverage and it suggests that market timing attempts is probably not aimed for reduce their book leverage. Combined results from above tables, it is plausible to guess that firms tend to use capital from IPO and leverage more to invest into their capital structure and R&D.

Table 6

Persistence of the Impact of Market Timing on Capital Structure

Panel A reports the mean value of the two dependent variables that are the cumulative change in book leverage ratio from pre-IPO year to years IPO+1, IPO+2 and the leverage level without taking the pre-IPO book leverage into account. Also, the t-statistics of the differences between the mean of two dependent variables for hot and cold firms are reported in the third row in Panel A,

Panel B shows coefficients of the regression model below

Yt=k0+k1HOT+k2M/Bt-1+k3EBITDA/At-1 +k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/APRE-IPO+ k8SIC+….+ knSIC +ε

The intercepts are not included in the table below and the regression analysis in Panel B are estimated with industry-fixed effect using the Standard Industry Codes. Values in the parentheses under coefficients are the t-statistics.

D/At- D/APRE-IPO Book Leverage D/At

t IPO+1 IPO+2 IPO+1 IPO+2 IPO+1 IPO+2 IPO+1 IPO+2

Panel A Mean Value

Hot -11.83 -10.36 __ __ 43.98 45.80 __ __

Cold -13.48 -12.04 __ __ 44.52 45.85 __ __

t-value (difference)

1.06 0.99 __ __ -0.37 -0.04 __ __

Panel B Regression Analysis

HOT 1.54 (0.86) 1.76 (0.89) 1.57 (0.88) 1.88 (0.95) 0.88 (0.45) 0.85 (0.41) 0.91 (0.47) 0.97 (0.46) M/Bt-1 0.13 (0.22) -1.69 (-2.17) __ __ __ __ 0.16 (0.26) -2.18 (-2.63) __ __ __ __ EBITDA/At-1 -0.51 (-5.74) -0.09 (-0.73) -0.51 (-5.79) -0.21 (-1.89) -0.54 (-5.6) -0.11 (-0.87) -0.54 (-5.66) -0.27 (-2.3) SIZEt-1 1.80 (3.94) 0.28 (0.52) 1.81 (3.99) 0.21 (0.39) 1.20 (2.4) 0.05 (0.08) 1.21 (2.47) -0.05 (-0.09)

Table 6 (Continued) PPE/At-1 0.02 (0.45) 0.03 (0.59) 0.02 (0.44) 0.03 (0.63) 0.02 (0.35) 0.02 (0.36) 0.02 (0.33) 0.02 (0.40) R&D/At-1 -0.24 (-1.11) 0.06 (0.21) -0.24 (-1.12) -0.03 (-0.11) -0.36 (-1.51) -0.23 (-0.72) -0.36 (-1.52) -0.36 (-1.21) RDDt-1 -1.26 (-0.57) -0.31 (-0.13) -1.28 (-0.59) -0.29 (-0.12) -0.81 (-0.34) 0.69 (0.26) -0.84 (-0.35) 0.76 (0.28) D/APRE-IPO -0.66 (-17.04) -0.68 (15.46) -0.66 (-17.1) -0.67 (-15.3) __ __ __ __ __ __ __ __ R2 _{66.38% 58.71% 53.47% 43.77% 86.1% 85.1% 87% 84.87%} N 1060 1060 1060 1060 1060 1060 1060 1060

6. Robustness Check

A factor should be discussed for robustness check. It is IPO proceeds. Whether firms with really high IPO proceeds behave differently compared to firms with relative low IPO proceeds. It is really

important to test whether the effect of market timing on capital structure is “linear” or not. In order to test this possibility, I subset the sample into subsamples based on their median. Table 6 shows that firms with IPO proceeds below median go public in hot markets, firms can earn 13.44% more proceeds from IPO compared to cold market firms and this coefficient is statistically significant. However, firms with IPO proceeds above median go public in hot markets, firms earn 19.71% less proceeds from IPO compared to cold market firms and it is statistically significant. Because IPO proceeds is closely related to valuation by market and it reflects firm size. In other words, the size of firms with high IPO proceeds is large size firm and vice versa. Small firms go public in hot market benefit more than large firms go public in hot market. The second column is regression model for testing short-run effect of market timing on capital structure and the coefficient is still statistically insignificant for two subsamples. It means that a high or a low IPO proceeds does not shape their behaviour to change their capital structure in the short-term. Third and fourth column are results for testing long-term effect of market timing on capital structure. Coefficients are not statistically significant. Due to sample size of IPO+2 the regression result is not reliable, the regression result for IPO+2 is not populated in the table. Also, the long-term effect of market timing on capital structure cannot be proved.

Table 7

Summary statistics of IPO sold proceeds on and before the first trading day (in Million US$) This table shows summary statistics of Proceeds Allotment included Over Sold -in this Market ($, million) from SDC platinum database. Based on IPO year and fiscal year. I split the whole sample into three subsamples. The table shows minimum, 25st _{quarterlies, median, mean, 75}th _{quarterlies and}

maximum.

Minimum 25th _{Quartiles Median mean 75}th _{Quartiles Maximum}

IPO 4.82 65.54 138.00 239.99 292.09 4376.00

IPO+1 3.70 101.41 187.14 298.31 363.39 3853.00

IPO+2 4.14 89.09 195.16 310.51 378.80 3003.50

Table 8

Robustness Check

This table shows linear regression results and they are estimated by OLS estimator. All linear regressions are estimated with industry-fixed effects and it is achieved by adding dummy variables (SIC code) into models. Dependent variables are IPO sold Proceeds, change of book leverage and cumulative book leverage and they are defined above to test short- and long- term effect. Independent variables are not presented in the table and independent variables are identical to previous table. For robustness check, I split the sample into two groups based on their medians that are shown in the table 7. Linear regression models are shown below in order:

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1 +k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/At-1+ k8SIC+ … + knSIC +εt

Yt=k0+k1HOT+k2M/Bt+k3EBITDA/At-1+k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/At-1+ k8SIC+ … + knSIC +εt

Yt=k0+k1HOT+k2M/Bt-1+k3EBITDA/At-1 +k4PPE/At-1+k5R&D/At-1+k6RDDt-1+k7D/APRE-IPO+ k8SIC+….+ knSIC +ε

D/At-D/At-1 D/At-D/APRE-IPO

ProceedsP_/A

t IPO IPO+1 IPO+2

IPO proceeds >Median HOT t-statistic R2 N -19.71 (-2.14) 92.5% 207 -4.94 (-1.26) 93.48% 207 -1.96 (0.82) 47.91% 101 __ __ __ __ IPO proceeds <Median HOT t-statistic R2 N 13.44 (2.11) 75.22% 208 -10.63 (-1.08) 19.38% 208 -0.02 (10) 99% 101 __ __ __ __

7. Conclusion and Discussion

For testing hypotheses and answer the research question, I collect 11800 observations for US firms from COMPUSTAT database. After applying some selection criteria and excluding outliners, I further subset the sample into subsample for the different year after IPO year. By graph, group mean

difference and linear regression, I have four main findings. Firstly, by visualizing IPO volume in the US stock market from 1990 to 2017, there is a significant difference of IPO volume in different time frames. In some time, the market is really active and IPO volume is high. But in other time, the market became inactive and IPO volume drops down significantly. It shows it is plausible to differentiate stock market into the active market and inactive market (hot and cold market).

Secondly, firms with market timing attempts (hot market firms) earn higher proceeds from IPO compared to firms without market timing attempts and hot market firms have a lower book leverage ratio prior to IPO compared to cold market firms. Because the hot market firm is very likely to raise more capital from IPO, it is plausible to believe that firms have the motivation to take advantage of market timing. The result can also be interpreted that use IPO volume to measure market timing attempts is reasonable. My first and second finding is consistent with the previous studies (Baker & Wurgler, 2002; Alti, 2006; Mahajan & Tartaroglu, 2007; De Bie & De Haan, 2007). The stock market can be differentiated between active and inactive market and firms are motivated to take advantage of the market timing to achieve their financial goals.

Also, there is a positive relationship between market timing attempts and change in capital structure in short-run. It means that firms with market timing attempts increase their book leverage ratio after their IPO. The change of capital expenditure and other assets (total assets minus cash) suggests that firms have the possibility to use raised capital from IPO and increase book leverage to invest into capital expenditure and other assets for business expansion (conclude from the significant increase of their revenue after IPO). This finding is consistent with previous studies and most of the scholars agree there is an effect of market timing attempts on capital structure.

Moreover, there is no persistent effect of market timing on capital structure. This finding is consistent with findings (Alti, 2006; Mahajan & Tartaroglu, 2007; De Bie & De Haan, 2007) but it is

inconsistent with finding (Baker & Wurgler, 2002). Baker and Wurgler state that the effect of marketing timing attempts is persistent and capital structure is shaped by the cumulative outcome of

past market timing attempts.

Overall, my finding is consistent with the equity market timing hypothesis but is not consistent with the dynamic trade-off hypothesis. Market timing hypothesis states that firms are kept looking for cheaper capital and previous capital structure plays a small role in their decision-making process. The dynamic trade-off theory states that managers are looking for an optimal balance between debt and equity. My result shows that firms tend to expand their business by raising capital from equity and debt at the same time and they are motivated to get involved in market timing attempts.

8. Reference list

Alti, A. (2006). How Persistent Is the Impact of Market Timing on Capital Structure? Journal of

Finance, 61(4), 1681-1710.

Auerbach, A. J. (1985), ‘Real determinants of corporate leverage’. In: B. M. Friedman (ed.): Corporate Capital Structures in the United States. Chicago, IL: University of Chicago Press. Baker, M. and J. Wurgler (2002), ‘Market timing and capital structure’. Journal

of Finance 57(1), 1–32.

Baker, M. and Wurgler, J. (2006). Investor sentiment and the cross-section of stock returns. The Journal of Finance, vol.61(4), pp. 1645-1680.

Baxter, N. D. and J. G. Cragg (1970), ‘Corporate choice among long- term financing instruments’. Review of Economics and Statistics 52(3), 301–324.

Benninga, S., Helmantel, M., & Sarig, O. (2005). The timing of initial public offerings.

Journal of Financial Economics, 75(1), 115-132.

doi:10.1016/j.jfineco.2003.04.002

Billett, M. & Flannery, M. & Garfinkel, J. (2006). Are bank loans special? Evidence on the post-announcement performance of bank borrowers. Journal of Financial and

Quantitative Analysis, Vol. 41(4), pp. 733-751.

Booth, L., V. Aivazian, A. Demirguc-Kunt, and V. Maksimovic (2001), ‘Capital structures in developing countries’. Journal of Finance 56(1), 87–130.

Bruinshoofd, W., & De Haan, L. (2012). Market timing and corporate capital structure: A transatlantic comparison. Applied Economics, 44(28), 3691-3703.

Chang, E. C., & Lewellen, W. G. (1984). Market Timing and Mutual Fund Investment Performance. The Journal of Business, 57(1), 57. doi:10.1086/296224

De Bie, T. and De Haan, L. (2007) Market timing and capital structure: evidence for Dutch firms,

De Economist, 155, 183–206

Desjardins, J. (2016, February 17). All of the World's Stock Exchanges by Size. Retrieved from http://www.visualcapitalist.com/all-of-the-worlds-stock-exchanges-by-size/

Faulkender, M. and M. Petersen (2006). ‘Does the source of capital affect capital structure?’Review of Financial Studies 19(1), 45–79.

Fresno, B. G. (2018, March 14). The Largest Stock Markets in the World | BBVA. Retrieved from https://www.bbva.com/en/the-largest-stock-markets-in-the-world/