Time until IPO and company performance
of American companies from 2000-2015
Amsterdam Business SchoolName Dominique Dersigni
Student number 10341668
Program Economics & Business Specialization Finance & Organization Number of ECTS 162
Supervisor Ilko Naaborg Target completion 29-06-2016
Abstract
This thesis researches the relation between the time until IPO and company performance. An empirical analysis is performed on 1686 companies in a short period of 3 months after the IPO and on 1352 companies in a long period of 24 months after the IPO. Only American companies were selected who went public in the period of 2000 until 2015. The contradictory results show that companies who push their IPO at a younger age perform better in the short run and companies who go public at an older age perform better in the long run. The short run results are different from the literature by Ritter (1991), Jovanovic & Rousseau (2001) and Chemmanur & Fulghieri (1999). However, this effect of time until IPO on company
performance is insignificant. Since the results are not significant and not entirely in line with the reviewed literature, it is hard to state economic implications. A limitation of this research can be the fact that the incorporation dates were used as founding dates.
Statement of Originality
This document is written by Dominique Dersigni who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is 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.
Table of Contents
1. Introduction ... 3
2. Literature review ... 5
2.1 Theories regarding research topic time until IPO and company performance. ... 5
2.2 Empirical findings in the literature ... 7
2.3 Summary on the literature ... 8
3. Methodology and data ... 9
3.1 Methodology ... 10
3.2 Data ... 11
4. Results and analysis ... 15
4.1 Empirical results ... 15
4.2 Robustness check ... 18
5. Conclusion and discussion ... 20
References ... 22
1. Introduction
Going public involves having the shares of a company quoted on a stock exchange (Jenkinson & Ljungqvist, 2001). There are several ways for a company to go public. The most known and the most common way is doing this via an initial public offering (IPO) of a companies’ shares to investors. The shares which are sold during the IPO can be newly created or already existing shares. The proceeds from selling the shares will go to the company if the shares are newly created. In the other case, when existed shares are sold, the proceeds from selling the shares will obviously go to the original investors. In practice, some IPOs consist entirely of new equity, with the original investors retaining their shares, some IPOs involve selling only existing equity, with no new money being raised for the company, but with the original owners selling some of their shares, and many consist of a combination of the two (Jenkinson & Ljungqvist, 2001).
Whether or not to go public is an important decision in the life cycle of a company. Theoretical literature seems to point to five common stages in the life cycle of a corporation. Theorists predicted that each stage would have complementarities among four classes of characteristics: strategy, structure, environment and decision making style. Miller and Friesen (1984) used this literature to create a typology of these stages. By using their typology and doing analyses of variance, they found empirical evidence for the prevalence of
complementarities among variables within each stage and inter-stage differences.
The five stages they described are: birth phase, growth phase, maturity phase, revival phase and the declined phase. As stated above each phase has its characteristics. The maturity phase is described as the period which is expected to happen as the sales levels stabilize, the level of innovation falls, the level of growth slows down and a more bureaucratic organization structure is established. Furthermore it is characterized as a larger and still older company with dispersed ownership.
In their findings Miller and Friesen describe each stage according to the four classes of characteristics: strategy, structure, environment and decision making. The findings for the maturity phase with respect to the situation are especially interesting. They state that the firms are indeed older and larger than those in the birth or growth phases. Besides that, they also have a substantial size advantage over most competitive firms. Moreover, their ownership is still further dispersed as the company founder retires, sells the business or goes public with the company.
Since almost all companies go through these stages, although not all in the same sequence, they pass the stage of maturity and are therefore concerned with the question how to deal with the changing composition of ownership. Thus, the issue of whether or not to go public is essential for every company in this phase.
There are several reasons why companies could make this decision to go public. One of the main reasons is to raise capital. Moreover, going public gives company founders and shareholders the option to sell their shares. In addition and as mentioned before, it allows for more dispersion of ownership. However, there are also disadvantages of going public. One particular disadvantage is the high costs which are associated with going public (Ritter & Welch, 2002).
Once a company has actually decided to go public, one of the major questions regarding public trading is: when is the optimal time for a company to file its IPO? The timing of the IPO has a big influence on the outcome and the post-IPO performance.
Therefore this thesis is taking a look at the age of the firm at the time of the IPO in relation to its performance. It especially looks for the facts whether young public firms perform better or whether older public firms perform better.
To answer the research question: ‘what is the relation between the time until American companies went public (IPO) and their performance in the period of 2000-2015?’ an analysis will be performed on 1686 American companies. These companies went public in the time period from 2000 until 2015. Data has been gathered on their incorporation date and their IPO date. With these variables the time until their IPO is calculated in months. In addition, there are company measures such as the return on assets and the basic earning power ratio. On top of that there are control variables on the crisis and industry effects. With these variables a regression is estimated. The results of the thesis should be viewed as a guideline for new founded companies.
Newly founded companies always have certain goals and objectives in the future they want to achieve. One of these goals might be going public once they reached the maturity stage. The outcome of the thesis will be defined in such a way that a company has a certain time period to strive for from the moment they incorporated. That is why this research adds relevant and useful information to the decision process of the timing of an IPO with regard to their performance.
The second part of this thesis gives an overview on the literature. It describes the main theories regarding the research topic, the empirical findings of this literature and it ends with a summary on the literature. The third part will describe the methodology. The fourth part takes
a look at the data and the descriptive statistics. The fifth part actually discusses the analysis. In the sixth and final part, a discussion and conclusion will be given.
2. Literature review
This section reviews the most important literature on IPOs, the reasons why firms choose to go public, the corporate life cycle and the timing of IPOs. Then, the empirical results from the relevant literature will be discussed. Finally a summary on the literature will be given.
2.1 Theories regarding research topic time until IPO and company performance. The first question should be: ‘why do firms go public?’ There are multiple reasons for
companies to go public. As mentioned before, the most known and common way, is going public by doing an initial public offering. Ritter and Welch (2002) give in their review several reasons for going public by an IPO. First, the desire to raise equity capital. Firms prefer to go public when they reach a certain stage in their business growth cycle and they need external equity capital to continue to grow further. Second, going public gives company founders and shareholders the option to convert some of their wealth into cash at a future date. Third, it allows for more dispersion of ownership.
Moreover Zingales (1995) points out that going public makes it easier for potential acquirers to spot potential takeover targets. Because by going public firms can facilitate their company’s acquisition for a higher value since they realize that acquirers can pressure targets on pricing concessions more than they can pressure outside investors. In this way a company can receive a higher value for the acquisition than they would get from an outright sale. Besides that, Maksimovic and Pichler (2001) state that public trading can add value to the firm as it may inspire more faith in the firm from other investors, customers, creditors and suppliers. For example, being the first in your industry to go public sometimes creates a first-mover advantage.
Finally, Pagano et al. (1998) find that going public provides another benefit; it enables companies to borrow more cheaply since the interest rate around the IPO date on their short-term credit falls and the number of banks willing to lend them rises.
A study of Jovanovic and Rousseau (2001) shows that a firm has its first innovation usually soon after founding. But then it takes years or even decades before a company goes public and lists on a stock exchange. They interpret the delay as a learning period. In this time the firm and its possible lenders learn about the firm’s optimal investment. In many companies
learning and production takes place almost simultaneously. A firm starts small, it learns a little, adds capacity, learns a little more, adds more capacity, changes its product a little, finds new suppliers, invests and hires a few more people, then learns again a little more, and the process goes on (Jovanovic & Rousseau, 2001). So, at the end of this period, an IPO leads to an influx of capital that enables the firm to put its further ideas into production. Pagano et al. (1993) supports this statement since they showed that a company’s size is significantly correlated with the probability of listing.
Chemmanur and Fulghieri (1999) recognize a similar fact that most firms start as small private companies and at some point in their growth cycle they go public. With this in mind they explain that young firms need to learn in order to gain capacity. Young firms cannot completely open up their company and its inventions in the beginning. This has to do with the fact that there are fixed costs associated with going public and proprietary
information cannot be revealed costless. So, early in its life cycle, a firm will be private but when it grows sufficiently large it can become optimal to go public (Chemmanur & Fulghieri, 1999).
Furthermore, Ritter and Welch (2002) also talk about the changing composition of IPO issuers. They state that the type of firms that went public changed over the years. The percentage of technology firms increased from about 25 percent of the IPO market in the 1980s and early 1990s to 37 percent after 1995 and an amazing 72 percent during the internet bubble, before returning to 29 percent in 2001 (Ritter & Welch, 2002).
Only 19 percent of the firms who went public had negative earnings in the 1980s. From 1995 to 1998 this gradually increased to 37 percent. Then, during the internet bubble it rose to an astonishing 79 percent. In the 1960s and the 1970s it was still unusual for an investment banker to take a firm public that did not have at least four years of positive earnings. Even in the 1980s it was still a standard to show four quarters of positive earnings before going public. Next, in the 1990s fewer and fewer companies met this condition. Still, the investment banking firm’s analyst would normally project profitability in the year after going public (Ritter & Welch, 2002). But during the period of the internet bubble it became common for companies to go public without the direct prospects of becoming profitable. This obviously changed the post-IPO performance of companies.
Nowadays it is conventional wisdom that in a period of high issuing volume the quality of firms going public deteriorates. According to Ritter (1991) this has to do with the fact that in these periods firms get overoptimistic about their prospects. Ritter (1991) also
shows evidence for this in his paper since he found that companies which went public during years of high issuing volume performed even worse than average.
2.2 Empirical findings in the literature
A lot of research has been performed on IPO pricing and allocation, and the long-run underperformance of IPOs. Ritter and Welch give a sophisticated overview in their paper of 2002 on these matters, including empirical evidence. However, just a couple of other papers have discussed the issuing activity and timing of IPOs. Most of them are literature reviews, theoretical analyses or papers including decision–tree models. Examples are papers from Zingales (1995), Chemmanur & Fulghieri (1999) and Maksimovic & Pichler (2001). Only a few of the researches on issuing activity and timing of IPOs show empirical evidence. One reason for this is that most of the formal theories on IPO timing and issuing activity are difficult to test. The problem is that most of the researches only observe and use the firms which are actually going public. Most of them do not observe how many private firms could have gone public (Ritter & Welch, 2002).
Still, there are three scientific papers with valuable empirical findings. First, Ritter (1991) wrote a paper about the time- and industry-dependence of the long run performance on IPOs. He finds in his research a strong so called monotone relation between the age of a company and aftermarket IPO performance. He also shows that the oldest firms are financial institutions which performed exceptionally good during that period and many of the youngest firms were oil and gas firms, which performed exceptionally bad. Therefore he states that the pattern of aftermarket performance is strengthened by industry effects. Besides that, he finds that the worst-performing industry in the long run had the lowest median age and the highest average initial return, while the best-performing industry in the long run has the highest median age and the lowest average initial return (Ritter, 1991).
Second, Pástor and Veronesi (2005) show evidence for the fact that IPOs come in waves. They tried to solve for the optimal time to go public and they found that private firms are attracted to capital markets especially when market conditions are favourable. These market conditions refer to a low expected market return, a high expected aggregate profitability and also a high prior uncertainty about the post-IPO average profitability in excess of market profitability. At any point in time, private firms are waiting for an
improvement in market conditions; that is, for a decline in expected market return or for an increase in expected aggregate profitability or prior uncertainty (Pástor & Veronesi, 2005). This means that if market conditions worsen, stock prices drop and IPO volume declines
because private firms choose to wait for more favourable conditions before going public. And on the other hand, when market conditions improve sufficiently, many firms choose to go public. Thus, creating a cluster of IPOs or the so called IPO wave.
Third, Lowry and Schwert (2002) argue that recent initial IPO returns of firms going public lead to the decision of other firms also going public. They observe that an increased number of companies file IPOs following periods of high underpricing. This suggests that the initial returns of recent IPOs contain information on the market's valuation of future IPOs. Lowry and Schwert find that more companies go public after periods of high initial returns because these high returns are related to positive information the companies have learned during the registration period of those offerings. Bookbuilding periods for IPOs are on average two months but they can easily last for four months. IPOs in subsequent months can therefore have overlapping registration periods. This all leads to the fact that companies’ learning processes during the registration period is causes the initial returns to be positively related to future levels of IPO activity.
2.3 Summary on the literature
Companies have several reasons to go public by an Initial Public Offering. These reasons are discussed by Ritter & Welch (2002), Zingales (1995), Maksimovic & Pichler (2001) and Pagano et al. (1998). First, the desire to raise equity capital. Second, going public gives company founders and shareholders the option to convert some of their wealth into cash at a future date. Third, it allows for more dispersion of ownership. Fourth, going public makes it easier for potential acquirers to spot potential takeover targets. Fifth, it can add value to the firm as it may inspire more faith in the firm from other investors, customers, creditors and suppliers. Finally, going public enables companies to borrow more cheaply. Jovanovic and Rousseau show that a firm has its first innovation usually soon after founding. But then it takes years or even decades before a company goes public and lists on a stock exchange. They interpret the delay as a learning period. The moment a company then decides to go public leads to an influx of capital that enables the firm to put its further ideas into production.
Chemmanur and Fulghieri (1999) say that early in its life cycle, a firm will be private but when it reaches a certain stage and grows sufficiently large it can become optimal to go public. Ritter and Welch (2002) also state that the type of firms that went public changed over the years. They explain how it became common for companies in the past decades to go public without the direct prospects of becoming profitable. This obviously changes the post-IPO performance of companies.
In addition, a lot of research has been performed on IPO pricing and allocation, and the long-run underperformance of IPOs. While just a couple of other papers have discussed the issuing activity and timing of IPOs. On top of that most of them are literature reviews, theoretical analyses or papers including decision–tree models. Only a few of the researches on issuing activity and timing of IPOs show empirical evidence. However, there are three
researches with valuable empirical findings. One is written by Pástor and Veronesi (2005). They found evidence for the fact that IPOs come in waves. Another one is written by Lowry and Schwert (2002). They argue that recent initial IPO returns of firms going public lead to the decision of other firms also going public. These statements are reinforced by the findings of Ritter (1991) as he finds a strong relation between the age of a company and aftermarket IPO performance. Besides that, he also shows results for the fact that the best-performing industry in the long run has the highest median age and the lowest average initial return.
Altogether we can interpret this literature and empirical evidence on the going-public decision by saying that firms prefer to go public when market conditions are favourable but only if they passed a specific learning period and when they reached a certain age and a certain maturity stage in their corporate life cycle. Especially the findings of Ritter (1991) suggest that companies at an older age perform better in the long run.
Additionally, with this literature it is possible to frame the hypotheses for this
research. Once again, the research question is stated as: ‘what is the relation between the time until American companies went public (IPO) and their performance in the period of 2000-2015?’. To answer this question, four hypotheses are examined. Hypotheses will test whether young companies at the time of IPO perform better in the short run. Hypothesis two will test whether old companies at the time of IPO perform better in the short run. Hypothesis three will test whether young at the time of IPO companies perform better in the long run.
Hypothesis four will test whether old companies at the time of IPO perform better in the long run.
3. Methodology and data
This section discusses the methodology and data. Section 3.1 describes the methodology and presents the model that is used in this thesis. Section 3.2 describes how the data is collected and how it is treated.
3.1 Methodology
As one can read in the literature overview, there has been done little research on the timing of IPOs and most of the researches that exist are literature reviews, theoretical analyses or papers including decision–tree models. Therefore, since not that much empirical research has been performed on the timing of IPOs, there is not a key scientific paper this thesis is based upon. Hence, this thesis performs a research with a self-designed model on the age of the firm at the time of the IPO in relation to its performance. This model is used in two time periods. First, the short run is considered, this represents the situation 3 months after the IPO date. Second, the long run is considered, this represents the situation 24 months after the IPO date. According to this information, the two econometric models are as follows.
Short run: 𝐶𝑜𝑚𝑝𝑎𝑛𝑦 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒: (𝐻𝑃𝑅 − 𝑀𝐼𝑅)3 𝑚𝑜𝑛𝑡ℎ𝑠= 𝛼 + 𝛽1𝑇𝑖𝑚𝑒 + 𝛽2𝑇𝑖𝑚𝑒2 +𝛽3𝑅𝑂𝐴 + 𝛿1𝐶𝑟𝑖𝑠𝑖𝑠 + 𝑐𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑒𝑐𝑡𝑜𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠1+ 𝜀 Long run 𝐶𝑜𝑚𝑝𝑎𝑛𝑦 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒: (𝐻𝑃𝑅 − 𝑀𝐼𝑅)24 𝑚𝑜𝑛𝑡ℎ𝑠 = 𝛼 + 𝛽1𝑇𝑖𝑚𝑒 + 𝛽2𝑇𝑖𝑚𝑒2 +𝛽3𝑅𝑂𝐴 + 𝛿1𝐶𝑟𝑖𝑠𝑖𝑠 + 𝑐𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑒𝑐𝑡𝑜𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀
In both cases the dependent variable is ‘Company Performance’. This is calculated as the Holding Period Return (HPR) minus the Market Index Return (MIR). In the case of the short run, the 3 month HPR after the IPO is taken minus the 3 months MIR after IPO. In the case of the long run, the 24 month HPR after the IPO is taken minus the 24 month MIR after IPO.
Again, in both cases the main independent variable of this research is the time until IPO, in short, ‘Time’. This could also be considered as the age of a company, since the time until IPO is calculated as the IPO date minus the Incorporation date. This explanatory variable is measured in months. Since Ritter (2001) found a strong so called monotone relation
between the age of a company and post-IPO performance, there will be accounted for age in two ways. The first way is by measuring age in a linear relation through the variable ‘Time’. The second way is by measuring age in a quadratic relation through the variable ‘Time2’. For this variable the time until IPO is taken and squared. Besides these age related variables, a 1 𝐶𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑒𝑐𝑡𝑜𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 = 𝛿 1𝐵𝑎𝑛𝑘𝑠 + 𝛿2𝐶ℎ𝑒𝑚𝑖𝑐𝑎𝑙𝑠, 𝑟𝑢𝑏𝑏𝑒𝑟, 𝑝𝑙𝑎𝑠𝑡𝑖𝑐𝑠 + 𝛿3𝐶𝑜𝑛𝑠𝑡𝑟𝑢𝑐𝑡𝑖𝑜𝑛 + 𝛿4𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛, 𝐻𝑒𝑎𝑙𝑡ℎ + 𝛿5𝐹𝑜𝑜𝑑, 𝑏𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑠, 𝑡𝑜𝑏𝑎𝑐𝑐𝑜 + 𝛿6𝐺𝑎𝑠, 𝑤𝑎𝑡𝑒𝑟, 𝑒𝑙𝑒𝑐𝑡𝑟𝑖𝑐𝑖𝑡𝑦 + 𝛿7𝐻𝑜𝑡𝑒𝑙𝑠, 𝑟𝑒𝑠𝑡𝑎𝑢𝑟𝑎𝑛𝑡𝑠 + 𝛿8𝐼𝑛𝑠𝑢𝑟𝑎𝑛𝑐𝑒 𝑐𝑜𝑚𝑝𝑎𝑛𝑖𝑒𝑠 + 𝛿9𝑀𝑎𝑐ℎ𝑖𝑛𝑒𝑟𝑦, 𝑒𝑞𝑢𝑖𝑝𝑚𝑒𝑛𝑡, 𝑓𝑢𝑟𝑛𝑖𝑡𝑢𝑟𝑒, 𝑟𝑒𝑐𝑦𝑐𝑙𝑖𝑛𝑔 + 𝛿10𝑀𝑒𝑡𝑎𝑙𝑠, 𝑚𝑒𝑡𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠 + 𝛿11𝑂𝑡ℎ𝑒𝑟 𝑠𝑒𝑟𝑣𝑖𝑐𝑒𝑠 + 𝛿12𝑃𝑜𝑠𝑡, 𝑡𝑒𝑙𝑒𝑐𝑜𝑚𝑚𝑢𝑛𝑖𝑐𝑎𝑡𝑖𝑜𝑛𝑠 + 𝛿13𝑃𝑟𝑖𝑚𝑎𝑟𝑦 𝑠𝑒𝑐𝑡𝑜𝑟 + 𝛿14𝑃𝑢𝑏𝑙𝑖𝑐 𝑎𝑑𝑚𝑖𝑛𝑖𝑠𝑡𝑟𝑎𝑡𝑖𝑜𝑛, 𝑑𝑒𝑓𝑒𝑛𝑐𝑒 + 𝛿15𝑃𝑢𝑏𝑙𝑖𝑠ℎ𝑖𝑛𝑔, 𝑝𝑟𝑖𝑛𝑡𝑖𝑛𝑔 + 𝛿16𝑇𝑒𝑥𝑡𝑖𝑙𝑒𝑠, 𝑤𝑒𝑎𝑟𝑖𝑛𝑔 𝑎𝑝𝑝𝑎𝑟𝑒𝑙, 𝑙𝑒𝑎𝑡ℎ𝑒𝑟 + 𝛿17𝑇𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡 + 𝛿18𝑊ℎ𝑜𝑙𝑒𝑠𝑎𝑙𝑒, 𝑟𝑒𝑡𝑎𝑖𝑙 𝑡𝑟𝑎𝑑𝑒 + 𝛿19𝑊𝑜𝑜𝑑, 𝑐𝑜𝑟𝑘, 𝑝𝑎𝑝𝑒𝑟
company measure variable is needed in this model. Therefore the ‘ROA’ is added, which refers to the Return on Assets. It is calculated as Net Income divided by Total Assets. Further the variable ‘Crisis’ is added, this is a dummy variable for the latest global financial crisis. The period of the Global Financial Crisis is defined from 2007-2008 (Alfaro & Chen, 2012). Especially, in the fall of 2008, world financial markets were in the midst of a credit crisis of historic breadth and depth (Campello, Graham, & Harvey, 2010). If a company existed and/or was active during the financial crisis of 2007-2008 then D1=1, if a company did not exist yet and/or was not active during the financial crisis of 2007-2008 then D1=0. This variable is added because the latest financial crisis has had a great influence on the performance of companies (Campello, Graham, & Harvey, 2010). For the robustness check of the model, which will be discussed in chapter 4, two interaction variables will be added. These interaction variables are ‘Crisistime’ and ‘Crisistime2’. They are combinations of the dummy variable ‘Crisis’ and the time-related variables.
Finally, the ‘company sector dummies’ will control for the different industry sectors in which companies are operating. Ritter (2001) also pointed out the importance of industry effects. These variables are added since it is interesting to find out if some industries outperform others. For example, it is well known that at the end of the 19th century technology firms went public younger than other firms because the technologies that they brought in were too productive to be kept out very long (Jovanovic & Rousseau, 2001). To test the statistical relationship of this econometric model, data is needed on American companies who went public from 2000 until 2015. Besides that, data is needed on their dates of incorporation and IPO, company and market returns after the IPO and data on the ROA company measure as already discussed.
3.2 Data
The sample used in this thesis consist of 1686 American companies who completed their IPO in the time period of 2000 until 2015. There has to be made a distinction between the two analyses. In the first analysis, the short-run is considered. Regarding this case, it is possible to use all 1686 companies since for every company the accounting data is available for 3 months after their IPO. In the second analysis, the long run is considered. For companies who went public in 2014 and 2015 there is no accounting data available yet for 24 months after their IPO. Therefore these companies are left out in the second part of the analysis and 1352 companies remain. Due to the fact that some companies were deleted, the dummy variable
‘sector_banks’, which is part of the company sector dummies, is left out because among the 1352 companies there were no firms active in this sector.
The first step was to get IPO data from the Zephyr Database. There were four main search criteria: 1) the companies should be American companies, 2) the companies should have done their IPO in the time period of 2000-2015, 3) the IPOs should have been
completed, not just rumoured or announced, 4) the company who filed the IPO should be the target company, not an acquirer or vendor company.
The second step was to add some variables in the Zephyr Database. The company incorporation date and the IPO date were requested. If these two variables were not available or if the complete date (day, month, year) was not available, the companies were deleted from the list. Then, the major sector of the company was requested. In addition, for the remaining companies, the ISIN numbers were retrieved.
For the third step, Datastream was used to gather the accounting variables. For every company the Net Income and the Total Assets of one year after the IPO were taken out. The accounting data one year after the IPO has been chosen because it always takes some time before events such as an IPO are reflected in the data. That being the case, this data was used to calculate the ROA. If it was not possible to calculate these company measures, the
company was deleted from the list.
The final step was also completed in Datastream. The last variables needed for this research were the companies’ monthly stock prices up to 24 months after the IPO. If the companies’ monthly stock prices were not available, the company was deleted from the list. As well as the monthly prices of the Market Index Fund for the period of 2000-2015 were retrieved. In this case, the Wilshire5000 is chosen as Market Index Fund. The Wilshire5000 is a market-capitalization-weighted index fund of all actively traded stocks in the US. Most of its stocks are traded through the NYSE and the NASDAQ. Since most of the companies that were gathered for this research are also primarily traded on the NYSE and the NASDAQ, this index fund is chosen.
With the above stated data it was possible to calculate the variable ‘Company
Performance’. With the available stock prices for the companies, the 3 month HPR and the 24 month HPR were calculated. Also, with the available Market Index Fund prices the 3 month index return and the 24 month index return were calculated. These returns were used for the dependent variable.
With the Incorporation date and the IPO date, the variable ‘Time’ was calculated by simply subtracting the Incorporation date from the IPO date. The outcome is defined in
months. If the time until IPO turned out to be negative for a company, then it was deleted from the list as it is not possible to go public before incorporation.
After all the data for the main and control variables was collected and the variables were calculated, the analyses were performed in Stata. In both scenario’s, the first step was to import the data. In the short run the 1686 companies were imported and in the long run fewer companies, 1352, were used. Then the dummy variables and the interaction variables were calculated. The final step was to perform an ordinary least squares regression with robust standard errors in both cases.
In order to discuss the results in the next chapter, the descriptive statistics and the correlation matrices have to be analyzed first.
Table I
Descriptive statistics for the short run
Variable Obs Mean Std. Dev. Min Max HPRMIR_3M 1,686 0.115144 2.205261 -0.8816972 77.33156 Time 1,686 55.85553 85.6175 -1202.301 928.0408 Time2 1,686 10445.85 56658.27 0 1445528 ROA 1,686 -0.1367142 0.3576176 -1.931496 0.5823538 Crisis 1,686 0.5937129 0.4912851 0 1 Crisistime 1,686 33.91436 77.26603 -1202.301 928.0408 Crisistime2 1,686 7116.682 52564.91 0 1445528 This table provides the descriptive statistics of variables used in the short run regressions. It says that the Holding Period Return corrected with the Mark Index Return is on average 11.5% in the period of 3 months after the IPO. Besides that, the time until American
companies went public is on average 55.9 months. Which is approximately 4,6 years counted from the date of incorporation until the IPO date. The average Return on Assets is -13,7%. The descriptive statistics for all the company sector dummy variables are listed in Appendix I.
Table II
Descriptive statistics for the long run
Variable Obs Mean Std. Dev. Min Max HPRMIR_24M 1,352 0.0549909 1.329036 -1.494839 23.70691 Time 1,352 56.0861 82.20935 -206.8705 928.0408 Time2 1,352 9899.028 48238.55 0 861259.7 ROA 1,352 -0.1202106 0.3539887 -1.931496 .5823538 Crisis 1,352 0.7352071 0.4413861 0 1 Crisistime 1,352 42.70604 76.85614 -206.8705 928.0408 Crisistime2 1,352 7726.303 43592.34 0 861259.7
This table provides the descriptive statistics of variables used in the long run regressions. It says that the Holding Period Return corrected with the Mark Index Return is on average 0.58% in the period of 24 months after the IPO. Besides that, the time until American
companies went public is on average 56.1 months. Which is approximately 4,7 years counted from the date of incorporation until the IPO date. The average Return on Assets is -11.7%. The descriptive statistics for all the company sector dummy variables are listed in Appendix II.
The correlation coefficients have been estimated and the results are shown in the matrices below. Only the most important variables are discussed. A complete overviews of the correlation coefficients are listed in the Appendix III for table III and Appendix IV for table IV.
Table III
Short run correlation matrix
HPRMIR_3M Time Time2 ROA Crisis Crisis time Crisis time2 HPRMIR_3M 1.0000 Time -0.0065 1.0000 Time2 -0.0012 0.3797 1.0000 ROA 0.0076 -0.0269 0.0396 1.0000 Crisis -0.0383 0.0179 0.0329 0.0760 1.0000 Crisistime -0.0078 0.7899 0.2422 0.0202 0.3632 1.0000 Crisistime2 -0.0021 0.2259 0.9198 0.0405 0.1120 0.2888 1.0000 There are two noticeable correlations in table III. First the high correlation of the variable Crisistime with Time and Crisistime2 with Time2. This is not strange since Crisistime and Crisistime2 are interaction variables with the time related coefficients. Besides that, these variables are less important since they are not incorporated in the main model that is used for this thesis. They are only used for the robustness check of the model. Introducing these interaction variables into the robustness model might cause multicollinearity and give high variances. However, as one can see in the next chapter, the found regression results of the main models already have quite big variances. Adding these correlated interaction variables does not make the other coefficients biased. Thus, these high correlations are not troubling and the used model for the robustness check will not be worse off.
Table IV
Long run correlation matrix
HPRMIR_24M Time Time2 ROA Crisis Crisis time Crisis time2 HPRMIR_24M 1.0000 Time 0.0227 1.0000 Time2 0.0282 0.8257 1.0000 ROA 0.1843 -0.0067 0.0466 1.0000 Crisis -0.0158 0.0400 0.0208 0.0215 1.0000 Crisistime 0.0043 0.8449 0.7052 0.0050 0.3325 1.0000 Crisistime2 0.0184 0.7267 0.8957 0.0406 0.1061 0.8080 1.0000 For this table almost the same analyses hold as for table III. Again, Crisistime with Time and Crisistime2 with Time2 have high correlations. Besides this there are three other high
correlations. Time with Time2, Crisistime2 with Time and Crisistime with Time2. Since Time2 is derived from Time, and these are still interaction variables, these high correlations are not surprising. As already explained above, these interaction variables are not used in the main model, they are only used for the robustness check. Since the results of the main model already have quite big variances, adding these correlated variables does not make the model for the robustness check worse off.
4. Results and analysis
In this section the main empirical results of the ordinary least squares regressions with robust standard errors are presented. Furthermore, answers to the hypotheses will be given. Finally, the model will be verified for robustness.
4.1 Empirical results
First the results are discussed concerning the kind of relationship between time and company performance. There are six regressions performed from which the results are shown. In the regressions one and four, the time-related variables are left out. Leaving out these variables does not explain more of the data. In regressions two and five only the variable ‘Time’ is included, which means only the linear relationship of time and company performance is considered. Since the ‘Time’ coefficient is negative in the short run and positive in the long run, there can be stated that here is a negative linear relationship in the short run and a positive linear relationship in the long run. In regression three and six both variables ‘Time’ and ‘Time2’ are included, which means the quadratic relationship of time and company performance is considered. As in both regressions the coefficient of ‘Time2’ is zero,
Table V
Regression explaining the relation between time until IPO and Company Performance Dependent Variable: HPR-MIR3 months Dependent Variable: HPR-MIR24
months (1) (2) (3) (4) (5) (6) Time -0.0002 (-0.44) -0.0002 (-0.52) 0.0002 (0.62) 0.0004 (0.53) Time2 0.0000 (0.77) 0.000 (-0.37) ROA 0.0444 (0.82) 0.0429 (0.82) 0.0417 (0.80) 0.0017 (0.01) 0.0028 (0.01) 0.0050 (0.02) Crisis -0.1794 (-1.32) -0.1787 (-1.33) -0.1794 (-1.33) -0.1238 (-1.33) -0.1255 (-1.35) -0.1259 (-1.35) sector_banks 0.3298 (2.35)** 0.3435 (2.92)** 0.3443 (2.92)** sector_chemicalsrubberplastic 0.4402 (7.60)*** 0.4453 (8.65)*** 0. 4447 (8.60)*** 0.3850 (2.20)** 0.3749 (2.19)** 0.3696 (2.18)** sector_construction 0.4122 (3.44)*** 0.4134 (3.47)*** 0.4127 (3.45)*** 0.1579 (0.45) 0.1518 (0.44) 0.1489 (0.43) sector_educationhealth 0.3433 (5.39)*** 0.3518 (6.45)*** 0.3516 (6.43)*** 0.0506 (0.46) 0.0360 (0.32) 0.0289 (0.25) sector_foodbeveragestobacco 0.3243 (3.52)*** 0.3349 (4.24)*** 0.3303 (4.12)*** 0.8533 (1.12) 0.8343 (1.08) 0.8350 (1.08) sector_gaswaterelectricity 0.2955 (3.16)** 0.3129 (4.10)*** 0.2993 (3.73)*** 0.2819 (1.37) 0.2436 (1.10) 0.2657 (1.19) sector_hotelrestaurant 0.3380 (5.04)*** 0.3485 (6.80)*** 0.3476 (6.73)*** 0.2127 (1.36) 0.1937 (1.20) 0.1866 (1.12) sector_insurance 0.5493 (3.37)*** 0.5609 (3.47)*** 0.5617 (3.46)*** 0.7017 (1.04) 0.6827 (1.02) 0.6720 (1.01) sector_machineryrecycling 0.4024 (9.34)*** 0.4097 (11.84)*** 0.4096 (11.80)*** 0.0557 (0.55) 0.0423 (0.41) 0.0354 (0.32) sector_metals 0.7628 (3.94)*** 0.7642 (3.96)*** 0.7633 (3.95)*** 0.1148 (0.45) 0.1106 (0.43) 0.1084 (0.43) sector_otherservices 0.4522 (7.32)*** 0.4570 (7.97)*** 0.4562 (7.92)*** 0.1151 (1.32) 0.1049 (1.21) 0.0997 (1.10) sector_posttelecommunication 0.3903 (5.91)*** 0.3963 (6.39)*** 0.3965 (6.39)*** -0.0324 (-0.17) -0.0445 (-0.23) -0.0516 (-0.26) sector_primarysector 0.5110 (5.51)*** 0.5149 (5.94)*** 0.5134 (5.91)*** 0.2462 (2.43)** 0.2365 (2.33)** 0.2336 (2.28)** sector_publicadmindefence 0.3132 (4.58)*** 0.3100 (4.53)*** 0.3092
(4.51)*** Omitted Omitted Omitted
sector_publishingprinting 2.2765 (1.43) 2.2842 (1.42) 2.2842 (1.42) 0.6593 (1.68) 0.6455 (1.62) 0.6380 (1.57) sector_textilesleather 0.3950 (3.17)** 0.3986 (3.29)*** 0.3988 (3.29)*** -0.1200 (-0.54) -0.1284 (-0.57) -0.1340 (-0.59) sector_transport 0.3580 (6.83)*** 0.3612 (7.61)*** 0.3608 (7.57)*** 0.3540 (2.40)** 0.3452 (2.32)** 0.3407 (2.27) sector_wholesaleretail 0.4487 (8.70)*** 0.4550 (10.63)*** 0.4531 (10.41)*** 0.6461 (2.16)** 0.6338 (2.13)** 0.6307 (2.14) sector_woodcorkpaper 0.3445 (2.18)** 0.3531 (2.30)** 0.3539 (2.30)** 1.4947 (0.95) 1.4795 (0.94) 1.4700 (0.93) Constant -0.2624 (-1.95)* -0.2589 (-1.83)* -0.2572 (-1.81)* -.0641 (-0.59) -0.0635 (-0.59) -0.0640 (-0.59) R2 0.0211 0.0212 0.0212 0.0268 0.0270 0.0223
Constant included. Absolute value of t-statistics in parentheses. * significance at 0.1, ** significance at 0.05, ***significance at 0.001
there can be concluded that no quadratic relationship concerning time until IPO could be found.
Second, the results of the short run are discussed in detail. These are the results of the third regression on the dependent variable ‘HPR-MIR3 months’. The bèta of independent
variable ‘Time’ is in a small extent negative, although not significant. However, this can be interpreted as the fact that being a slightly younger firm at the time of going public, will give better performances. The bèta of the dummy variable ‘Crisis’ is negative but not significant. It shows that companies who went public during the financial crisis or companies who were active in this period, performed indeed worse as stated in the literature by Campello, Graham and Harvey (2010). The company sector dummy variables are all significant, except for the publishing and printing sector. Still, there can be stated that companies working in publishing and printing sector and the metal-products sector outperformed the other sectors, since these two sectors have relatively the highest bètas.
Finally, the results of the long run are examined in detail. These are the results of the sixth regression on the dependent variable ‘HPR-MIR24 months’. The bèta of independent
variable ‘Time’ is, in contrast to the short run, positive but not significant. Therefore the statement above with respect to the variable time is not reinforced since in this case it can be interpreted as the fact that being an older firm at the time of going public, will give better performances. The bèta of the dummy variable ‘Crisis’ is also in the case of the long run negative. None of the company sector dummy variables are significant and the variable ‘sector_publicadmindefence’ is omitted due to collinearity. However, we can state that the companies working in the food, beverages and tobacco sector, and the wood, cork and paper sector slightly outperformed the other sectors, since all three of them have relatively higher bètas. There are also a two sectors that underperformed compared with the other sectors, namely: the post and telecommunication sector and the textiles and leather sector.
Even though the found coefficients are not significant we can state the following about the hypotheses. Hypothesis one is not rejected since the results showed that in the short run the coefficient ‘Time’ is negative and therefore companies who go public at a younger age perform slightly better. Therefore hypothesis two is rejected since older companies at the time of IPO do not perform better in the short run. Hypotheses three is rejected since the outcome presented that in the long run the coefficient ‘Time’ is also positive and therefore companies who go public at a younger age do not perform better. Hence, hypothesis four is not rejected because older companies at the time of IPO perform better in the long run. The results are not completely in line in what was expected from the literature. The expectations were that
companies who go public at an older age perform better. This expectation holds for the long run, but not for the short run. Ritter (1991) showed that the best-performing industry in the long run has the highest median age. Jovanovic and Rousseau (2001) stated that a firm
usually has its first innovation soon after founding. Then it takes years or even decades before a company goes public and lists on a stock exchange. Moreover, Chemmanur and Fulghieri (1999) recognize a similar fact that most firms start as small private companies and at some point in their growth cycle they go public. They explain that young firms need to learn in order to gain capacity. Thus, early in its life cycle, a firm will be private but when it grows sufficiently large it can become optimal to go public. With this literary background, the expectations were different from the empirical findings.
4.2 Robustness check
There are two ways in which one has to control for robustness. The first way is the robustness of standard errors. In this thesis there has been controlled for robustness of standard errors by adding the robust command in Stata before running the regression. The second way is to examine the robustness of the model. This is done by adding two interaction variables in both models. The added variables are ‘Crisistime’, which is the dummy variable ‘Crisis’ multiplied by the time until IPO, and ‘Crisistime2’, which is the dummy variable crisis multiplied by the squared time until IPO. As already mentioned in section 3, adding these highly correlated variables will not make the other coefficients biased. The variances of the main model are relatively high, and as one can see, they remain high in this robustness model as well. Therefore there can be stated that the robustness model will not be worse off.
The adjusted models are: Short run: 𝐶𝑜𝑚𝑝𝑎𝑛𝑦 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒: (𝐻𝑃𝑅 − 𝑀𝐼𝑅)3 𝑚𝑜𝑛𝑡ℎ𝑠= 𝛼 + 𝛽1𝑇𝑖𝑚𝑒 + 𝛽2𝑇𝑖𝑚𝑒2 +𝛽3𝑅𝑂𝐴 + 𝛽5𝐶𝑟𝑖𝑠𝑖𝑠𝑡𝑖𝑚𝑒 + 𝛽6𝐶𝑟𝑖𝑠𝑖𝑠𝑡𝑖𝑚𝑒2 + 𝛿1𝐶𝑟𝑖𝑠𝑖𝑠 + 𝑐𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑒𝑐𝑡𝑜𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀 Long run 𝐶𝑜𝑚𝑝𝑎𝑛𝑦 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒: (𝐻𝑃𝑅 − 𝑀𝐼𝑅)24 𝑚𝑜𝑛𝑡ℎ𝑠 = 𝛼 + 𝛽1𝑇𝑖𝑚𝑒 + 𝛽2𝑇𝑖𝑚𝑒2 +𝛽3𝑅𝑂𝐴 + 𝛽5𝐶𝑟𝑖𝑠𝑖𝑠𝑡𝑖𝑚𝑒 + 𝛽6𝐶𝑟𝑖𝑠𝑖𝑠𝑡𝑖𝑚𝑒2 + 𝛿1𝐶𝑟𝑖𝑠𝑖𝑠 + 𝑐𝑜𝑚𝑝𝑎𝑛𝑦 𝑠𝑒𝑐𝑡𝑜𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀
The regression results with these alternative models are presented in table VI below. The first regression in table VI in which the interaction variables are included shows the results of the dependent variable: ‘HPR-MIR3 months’. The main coefficient of this research, ‘Time’, is
negative which means that younger firms at the time of IPO perform better. Besides that, is the coefficient of ‘Crisis’ also negative which implies that firms who went public or firms who were active during the financial crisis performed worse. The second regression in which in table VI the interaction variables are included shows the results of the dependent variable ‘HPR-MIR24 months’. Again, the main coefficient ‘Time’ is positive and coefficient of ‘Crisis’ is
negative. If these findings are compared to the first and third regressions, one can read that the conclusions are the same. For this reason, there can be stated that the robustness of this model is verified.
Table VI
Regression explaining the relation between time until IPO and Company Performance Dependent Variable: HPR-MIR3
months
Dependent Variable: HPR-MIR24 months
(1) (2) Time -0.0028 (-1.24) .0006 (0.24) Time2 0.0000 (1.41) 0.000 (-0.27) ROA 0.0269 (0.54) .0055 (0.02) Crisis -0.3076 (-1.34) -.1194 (-0.93) Crisistime 0.0030 (1.35) -.0002 (-0.09) Crisistime2 0.0000 (-1.40) 0.0000 (0.18) sector_banks 0.4339 (3.87)*** sector_chemicalsrubberplastic 0.4313 (7.28)*** .3705 (2.09)** sector_construction 0.3643 (2.48)** 0.1515 (0.43) sector_educationhealth 0.3481 (6.49)*** 0.0298 (0.26) sector_foodbeveragestobacco 0.2850 (2.72)*** 0.8444 (1.06) sector_gaswaterelectricity 0.2185 (1.72)* 0.2591 (1.13) sector_hotelrestaurant 0.3267 (5.03)*** 0.1888 (1.14) sector_insurance 0.5787 (4.44)*** 0.6712 (1.00) sector_machineryrecycling 0.3972 (9.46)*** 0.0366 (0.33) sector_metals 0.7505 (3.85)*** 0.1090 (0.43) sector_otherservices 0.4434 (7.10)*** 0.1009 (1.07)
sector_posttelecommunication 0.3901 (5.75)*** -.0506 (-0.25) sector_primarysector 0.4824 (4.83)*** .2350 (2.29)** sector_publicadmindefence 0.3164 (4.64)*** Omitted sector_publishingprinting 2.2788 (1.42) 0.6385 (1.59) sector_textilesleather 0.4167 (3.76)*** -0.1349 (-0.59) sector_transport 0.3350 (5.11)*** 0.3421 (2.27) sector_wholesaleretail 0.4262 (7.05)*** 0.6313 (2.16) sector_woodcorkpaper 0.3500 (2.38)** 1.4700 (0.93) Constant -0.1377 (-0.60) -0.0703 (-0.49) R2 0.0223 0.0271
Constant included. Absolute value of t-statistics in parentheses. * significance at 0.1, ** significance at 0.05, ***significance at 0.001
5. Conclusion and discussion
This thesis examined the relation between the time until IPO and company performance. The results show no significance on the main coefficient. Therefore from this research and with this dataset there can be concluded that there is no relation between the time until IPO and company performance. The results that were found implied that companies who push their IPO at a younger age perform better in the short run and companies who go public at an older age perform better in the long run. However, this is not completely what was expected from the reviewed literature. Only the long run results are in line with the expectations from the literature. Jovanovic & Rousseau (2001) and Chemmanur & Fulghieri (1999) implied with their researches that if an older company at the moment of IPO performs better. Especially when it reaches a certain maturity stage in the corporate life cycle. This is reinforced by the findings of Ritter (1991) who showed that the best-performing industry in the long run has the highest median age.
Due to the fact that the results of this research are not significant and not entirely in line with the reviewed literature, it is hard to state economic implications. One limitation of this research can be the fact that the incorporation dates are used as founding dates, since for most companies the real founding dates are not available or incomplete. This could have had an effect on the coefficient of time and its significance. Another limitation might be the performance measured used to estimate company performance as none of the performance measures were significant.
A suggestion for further research is to examine a shorter time period. In this study there has been used 15 years to measure the relation between time until IPO and company performance. Since a lot can change in 15 years and other relevant factors besides the
financial crisis might have had influences, it could be better to examine a shorter time period. Another suggestion would be to perform the research with the real founding dates, though this could be very time consuming. The results however, could improve and make the research more significant.
References
Alfaro, L., & Chen, M. X. (2012). Surviving the Global Financial Crisis: Foreign Ownership and Establishment Performance . American Economic Journal: Economic Policy, 30-55.
Brau, J. C., & Fawcett, S. E. (February 2006). Initial Public Offerings: An Analysis of Theory and Practice. Journal of Finance, 399-429.
Campello, M., Graham, J. R., & Harvey, C. R. (2010). The real effects of the financial constraints: Evidence from a financial crisis. Journal of Financial Economics, 470-487.
Chemmanur, T. J., & Fulghieri, P. (1999). A Theory of the Going-public Decision . Review of Financial Studies, 249-279.
Choe, H., Masulis, R. W., & Nanda, V. (1993). Common stock offerings across the business cycle. Theory and evidence. Journal of Empirical Finance, 3-31.
Jenkinson, T., & Ljungqvist, A. (2001). Going Public. The Theory and Evidence on How Companies Raise Equity Finance . New York: Oxford University Press .
Jovanovic, B., & Rousseau, P. L. (2001). Why wait? A century of life before IPO. American Economic Review, 1-12.
Loughran, T., & Ritter, J. R. (2001). Why Has IPO Underpricing Increased Over Time? Working paper. 1-48.
Lowry, M. (2003). Why does IPO volume fluctuate so much? Journal of Financial Economics, 3-38.
Lowry, M., & Schwert, G. W. (June 2002). IPO Market Cycles: Bubbles or Sequential Learning? Journal of Finance, 1171-1199.
Ma, R. (November 1959). The Age Structure of Quoted Public Companies. Journal of the Royal Statistical Society, 158-167.
Maksimovic, V., & Pichler, P. (2001). Technological Innovation and Initial Public Offerings. Review of Financial Studies, 459-494.
Miller, D., & Friesen, P. H. (1984). A Longitudinal Study of the Corporate Life Cycle. Management Science, 1161-1183.
Pagano, Marco, Panetta, F., & Zingales, L. (1998). Why do companies go public? An empirical analysis. Journal of Finance, 27-64.
Pástor, L., & Veronesi, P. (2005). Rational IPO Waves. Journal of Finance, 1713-1747. Ritter, J. R. (1984). The 'Hot Issue' Market of 1980. The Journal of Business, 215-239.
Ritter, J. R. (1991). The Long-Run Performance of Initial Public Offerings. Journal of Finance, 3-27.
Ritter, J. R., & Welch, I. (2002). A Review of IPO Activity, Pricing and Allocations. Journal of Finance, 1795-1823.
Zingales, L. (1995). Insider Ownership and the Decision to go Public. Review of economic studies, 425-448.
Appendices
Appendix I: Complete table of descriptive statics for short run regression variables.
Variable Obs Mean Std. Dev. Min Max
HPRMIR_3M 1,686 .115144 2.205261 -.8816972 77.33156 Time 1,686 55.85553 85.6175 -1202.301 928.0408 Time2 1,686 10445.85 56658.27 0 1445528 ROA 1,686 -.1367142 .3576176 -1.931496 .5823538 Crisis 1,686 .5937129 .4912851 0 1 Crisistime 1,686 33.91436 77.26603 -1202.301 928.0408 Crisistime2 1,686 7116.682 52564.91 0 1445528 sector_banks 1,686 .0011862 .0344316 0 1 sector_chemicalsrubberplastic 1,686 .0723606 .2591609 0 1 sector_construction 1,686 .010083 .0999364 0 1 sector_educationhealth 1,686 .0189798 .1364941 0 1 sector_foodbeveragestobacco 1,686 .0136418 .1160329 0 1 sector_gaswaterelectricity 1,686 .010083 .0999364 0 1 sector_hotelsrestaurant 1,686 .0260973 .1594719 0 1 sector_insurance 1,686 .0017794 .0421574 0 1 sector_machineryrecycling 1,686 .1364176 .3433333 0 1 sector_metals 1,686 .014828 .1208999 0 1 sector_otherservices 1,686 .485172 .4999284 0 1 sector_posttelecommunications 1,686 .0136418 .1160329 0 1 sector_primarysector 1,686 .0658363 .2480693 0 1 sector_publicadmindefence 1,686 .0011862 .0344316 0 1 sector_publishingprinting 1,686 .0278766 .1646682 0 1 sector_textilesleather 1,686 .0041518 .0643199 0 1 sector_transport 1,686 .0320285 .1761279 0 1 sector_wholesaleretail 1,686 .0610913 .2395688 0 1 sector_woodcorkpaper 1,686 .0029656 .0543926 0 1
Appendix II: Complete table of descriptive statics for long run regression variables.
Variable Obs Mean Std. Dev. Min Max
HPRMIR_24M 1,352 .0549909 1.329036 -1.494839 23.70691 Time 1,352 56.0861 82.20935 -206.8705 928.0408 Time2 1,352 9899.028 48238.55 0 861259.7 ROA 1,352 -.1202106 .3539887 -1.931496 .5823538 Crisis 1,352 .7352071 .4413861 0 1 Crisistime 1,352 42.70604 76.85614 -206.8705 928.0408 Crisistime2 1,352 7726.303 43592.34 0 861259.7 sector_chemicalsrubberplastic 1,352 .0724852 .2593856 0 1 sector_construction 1,352 .0088757 .0938268 0 1 sector_educationhealth 1,352 .0177515 .1320957 0 1 sector_foodbeveragestobacco 1,352 .0147929 .1207678 0 1 sector_gaswaterelectricity 1,352 .0081361 .0898659 0 1 sector_hotelsrestaurant 1,352 .0251479 .1566322 0 1 sector_insurance 1,352 .0022189 .0470707 0 1 sector_machineryrecycling 1,352 .1442308 .3514536 0 1 sector_metals 1,352 .012574 .1114677 0 1 sector_otherservices 1,352 .4733728 .4994752 0 1 sector_posttelecommunications 1,352 .0155325 .1237037 0 1 sector_primarysector 1,352 .0673077 .2506468 0 1 sector_publicadmindefence 1,352 .0014793 .0384473 0 1 sector_publishingprinting 1,352 .0303254 .1715447 0 1 sector_textilesleather 1,352 .0051775 .071795 0 1 sector_transport 1,352 .0318047 .1755448 0 1 sector_wholesaleretail 1,352 .0650888 .2467737 0 1 sector_woodcorkpaper 1,352 .0036982 .0607229 0 1
Appendix III: Complete correlation matrix of short-run regression variables (continued on next page)
HPRMIR_3M Time Time2 ROA Crisis Crisistime Crisistime2 secto~ks sector~c sector~n sector~h sector~o sector~y
HPRMIR_3M 1.0000 Time -0.0065 1.0000 Time2 -0.0012 0.3797 1.0000 ROA 0.0076 -0.0269 0.0396 1.0000 Crisis -0.0383 0.0179 0.0329 0.0760 1.0000 Crisistime -0.0078 0.7899 0.2422 0.0202 0.3632 1.0000 Crisistime2 -0.0021 0.2259 0.9198 0.0405 0.1120 0.2888 1.0000 sector_banks -0.0007 0.0176 0.0008 0.0170 -0.0417 -0.0151 -0.0047 1.0000 sector_chemicalsrubberplastic -0.0072 -0.0070 -0.0123 -0.0610 0.0120 -0.0146 -0.0198 -0.0096 1.0000 sector_construction -0.0021 -0.0306 -0.0141 -0.0027 -0.0374 -0.0201 -0.0095 -0.0035 -0.0282 1.0000 sector_educationhealth -0.0097 0.0260 0.0025 0.0308 0.0177 0.0240 0.0040 -0.0048 -0.0388 -0.0140 1.0000 sector_foodbeveragestobacco -0.0072 0.0390 0.0548 0.0156 -0.0381 -0.0231 -0.0094 -0.0041 -0.0328 -0.0119 -0.0164 1.0000 sector_gaswaterelectricity -0.0076 0.0784 0.1480 0.0386 -0.0253 0.1048 0.1651 -0.0035 -0.0282 -0.0102 -0.0140 -0.0119 1.0000 sector_hotelsrestaurant -0.0102 0.0520 0.0214 0.0619 -0.0161 0.0369 0.0091 -0.0056 -0.0457 -0.0165 -0.0228 -0.0193 -0.0165 sector_insurance 0.0010 0.0166 0.0003 0.0171 0.0063 0.0118 0.0002 -0.0015 -0.0118 -0.0043 -0.0059 -0.0050 -0.0043 sector_machineryrecycling -0.0194 0.0503 -0.0103 -0.0598 0.0790 0.0711 -0.0013 -0.0137 -0.1110 -0.0401 -0.0553 -0.0467 -0.0401 sector_metals 0.0134 -0.0315 -0.0136 -0.0705 0.0315 -0.0249 -0.0144 -0.0042 -0.0343 -0.0124 -0.0171 -0.0144 -0.0124 sector_otherservices -0.0138 -0.0520 -0.0370 -0.0877 -0.0886 -0.0809 -0.0302 -0.0335 -0.2711 -0.0980 -0.1350 -0.1142 -0.0980 sector_posttelecommunications -0.0067 0.0036 -0.0098 0.0061 0.0348 0.0091 -0.0093 -0.0041 -0.0328 -0.0119 -0.0164 -0.0138 -0.0119 sector_primarysector 0.0031 -0.0333 0.0043 0.0863 0.0005 -0.0110 0.0064 -0.0091 -0.0741 -0.0268 -0.0369 -0.0312 -0.0268 sector_publicadmindefence -0.0038 -0.0205 -0.0063 0.0172 0.0285 -0.0130 -0.0047 -0.0012 -0.0096 -0.0035 -0.0048 -0.0041 -0.0035 sector_publishingprinting 0.1372 0.0236 -0.0023 0.0403 0.0080 0.0154 -0.0049 -0.0058 -0.0473 -0.0171 -0.0236 -0.0199 -0.0171 sector_textilesleather -0.0031 -0.0104 -0.0083 0.0483 0.0159 -0.0124 -0.0072 -0.0022 -0.0180 -0.0065 -0.0090 -0.0076 -0.0065 sector_transport -0.0115 -0.0307 -0.0161 0.0696 0.0270 -0.0054 -0.0094 -0.0063 -0.0508 -0.0184 -0.0253 -0.0214 -0.0184 sector_wholesaleretail -0.0057 0.0098 0.0321 0.0884 0.0396 0.0372 0.0374 -0.0088 -0.0712 -0.0257 -0.0355 -0.0300 -0.0257 sector_woodcorkpaper -0.0034 0.0100 -0.0038 0.0318 0.0007 0.0083 -0.0033 -0.0019 -0.0152 -0.0055 -0.0076 -0.0064 -0.0055
secto~nt sec~ance sec~ling secto~ls secto~es secto~ns secto~or sec~ence sec~ting sect~her secto~rt sector~l sect~per HPRMIR_3M Time Time2 ROA Crisis Crisistime Crisistime2 sector_banks sector_chemicalsrubberplastic sector_construction sector_educationhealth sector_foodbeveragestobacco sector_gaswaterelectricity sector_hotelsrestaurant 1.0000 sector_insurance -0.0069 1.0000 sector_machineryrecycling -0.0651 -0.0168 1.0000 sector_metals -0.0201 -0.0052 -0.0488 1.0000 sector_otherservices -0.1589 -0.0410 -0.3858 -0.1191 1.0000 sector_posttelecommunications -0.0193 -0.0050 -0.0467 -0.0144 -0.1142 1.0000 sector_primarysector -0.0435 -0.0112 -0.1055 -0.0326 -0.2577 -0.0312 1.0000 sector_publicadmindefence -0.0056 -0.0015 -0.0137 -0.0042 -0.0335 -0.0041 -0.0091 1.0000 sector_publishingprinting -0.0277 -0.0071 -0.0673 -0.0208 -0.1644 -0.0199 -0.0450 -0.0058 1.0000 sector_textilesleather -0.0106 -0.0027 -0.0257 -0.0079 -0.0627 -0.0076 -0.0171 -0.0022 -0.0109 1.0000 sector_transport -0.0298 -0.0077 -0.0723 -0.0223 -0.1766 -0.0214 -0.0483 -0.0063 -0.0308 -0.0117 1.0000 sector_wholesaleretail -0.0418 -0.0108 -0.1014 -0.0313 -0.2476 -0.0300 -0.0677 -0.0088 -0.0432 -0.0165 -0.0464 1.0000 sector_woodcorkpaper -0.0089 -0.0023 -0.0217 -0.0067 -0.0529 -0.0064 -0.0145 -0.0019 -0.0092 -0.0035 -0.0099 -0.0139 1.0000
Appendix IV: Complete correlation matrix of long-run regression variables (continued on next page)
HPRM~24M time_m~s time2 ROA crisis crisis~s crisis~2 sector~c sector~n sector~h sector~o sector~y
HPRMIR_24M 1.0000 Time 0.0175 1.0000 Time2 0.0172 0.8254 1.0000 ROA 0.0142 -0.0064 0.0468 1.0000 Crisis -0.0429 0.0406 0.0211 0.0268 1.0000 Crisistime -0.0008 0.8444 0.7049 0.0050 0.3336 1.0000 Crisistime2 0.0094 0.7263 0.8957 0.0407 0.1064 0.8077 1.0000 sector_chemicalsrubberplastic 0.0359 -0.0188 -0.0245 -0.0788 0.0191 -0.0200 -0.0224 1.0000 sector_construction -0.0024 -0.0291 -0.0133 -0.0314 -0.0326 -0.0203 -0.0102 -0.0265 1.0000 sector_educationhealth -0.0179 0.0237 0.0060 0.0755 0.0426 0.0354 0.0116 -0.0376 -0.0127 1.0000 sector_foodbeveragestobacco 0.0620 0.0491 0.0805 0.0031 -0.0653 -0.0338 -0.0123 -0.0343 -0.0116 -0.0165 1.0000 sector_gaswaterelectricity 0.0050 0.1354 0.2516 0.0300 -0.0016 0.1576 0.2829 -0.0253 -0.0086 -0.0122 -0.0111 1.0000 sector_hotelsrestaurant 0.0012 0.0665 0.0369 0.0508 -0.0214 0.0369 0.0147 -0.0449 -0.0152 -0.0216 -0.0197 -0.0145 sector_insurance 0.0178 0.0192 0.0009 0.0171 -0.0073 0.0078 -0.0004 -0.0132 -0.0045 -0.0063 -0.0058 -0.0043 sector_machineryrecycling -0.0508 0.0508 -0.0075 -0.0564 0.0746 0.0726 0.0044 -0.1148 -0.0388 -0.0552 -0.0503 -0.0372 sector_metals -0.0103 -0.0417 -0.0202 -0.0790 0.0527 -0.0308 -0.0175 -0.0315 -0.0107 -0.0152 -0.0138 -0.0102 sector_otherservices -0.0644 -0.0664 -0.0700 -0.0783 -0.0857 -0.0773 -0.0668 -0.2650 -0.0897 -0.1275 -0.1162 -0.0859 sector_posttelecommunications -0.0238 0.0058 -0.0103 -0.0026 0.0212 0.0016 -0.0130 -0.0351 -0.0119 -0.0169 -0.0154 -0.0114 sector_primarysector 0.0075 -0.0278 0.0150 0.0904 -0.0060 -0.0183 0.0155 -0.0751 -0.0254 -0.0361 -0.0329 -0.0243 sector_publicadmindefence -0.0070 -0.0240 -0.0079 0.0176 0.0231 -0.0189 -0.0068 -0.0108 -0.0036 -0.0052 -0.0047 -0.0035 sector_publishingprinting 0.0602 0.0222 -0.0053 0.0450 -0.0112 0.0097 -0.0053 -0.0494 -0.0167 -0.0238 -0.0217 -0.0160 sector_textilesleather -0.0178 -0.0123 -0.0101 0.0511 -0.0034 -0.0222 -0.0107 -0.0202 -0.0068 -0.0097 -0.0088 -0.0065 sector_transport 0.0179 -0.0264 -0.0140 0.0713 0.0419 -0.0071 -0.0091 -0.0507 -0.0172 -0.0244 -0.0222 -0.0164 sector_wholesaleretail 0.0854 0.0129 0.0434 0.0824 0.0292 0.0349 0.0583 -0.0738 -0.0250 -0.0355 -0.0323 -0.0239 sector_woodcorkpaper 0.0585 0.0123 -0.0042 0.0343 0.0089 0.0060 -0.0052 -0.0170 -0.0058 -0.0082 -0.0075 -0.0055
secto~nt sec~ance sec~ling secto~ls secto~es secto~ns secto~or sec~ence sec~ting sect~her secto~rt sector~l sect~per HPRMIR_24M Time Time2 ROA Crisis Crisistime Crisistime2 sector_chemicalsrubberplastic sector_construction sector_educationhealth sector_foodbeveragestobacco sector_gaswaterelectricity sector_hotelsrestaurant 1.0000 sector_insurance -0.0076 1.0000 sector_machineryrecycling -0.0659 -0.0194 1.0000 sector_metals -0.0181 -0.0053 -0.0463 1.0000 sector_otherservices -0.1523 -0.0447 -0.3892 -0.1070 1.0000 sector_posttelecommunications -0.0202 -0.0059 -0.0516 -0.0142 -0.1191 1.0000 sector_primarysector -0.0431 -0.0127 -0.1103 -0.0303 -0.2547 -0.0337 1.0000 sector_publicadmindefence -0.0062 -0.0018 -0.0158 -0.0043 -0.0365 -0.0048 -0.0103 1.0000 sector_publishingprinting -0.0284 -0.0083 -0.0726 -0.0200 -0.1677 -0.0222 -0.0475 -0.0068 1.0000 sector_textilesleather -0.0116 -0.0034 -0.0296 -0.0081 -0.0684 -0.0091 -0.0194 -0.0028 -0.0128 1.0000 sector_transport -0.0291 -0.0085 -0.0744 -0.0205 -0.1718 -0.0228 -0.0487 -0.0070 -0.0321 -0.0131 1.0000 sector_wholesaleretail -0.0424 -0.0124 -0.1083 -0.0298 -0.2502 -0.0331 -0.0709 -0.0102 -0.0467 -0.0190 -0.0478 1.0000 sector_woodcorkpaper -0.0098 -0.0029 -0.0250 -0.0069 -0.0578 -0.0077 -0.0164 -0.0023 -0.0108 -0.0044 -0.0110 -0.0161 1.0000
