The effect of JOBS act on US IPO information asymmetry and underpricing process

(1)

Evaluation of the Thesis by the Supervisors

The supervisor will evaluate the thesis that has been submitted on its

content and he/she will also evaluate the attitude of the student while

working on the thesis.

On each of the subjects below a score of Poor, Insufficient, Sufficient, Good

or Excellent can be obtained. The final grade will be based on these scores.

The criteria are as follows:

The thesis

1. Statement of the research proposal

2. Structure

3. Originality

4. Choice and processing of literature

5. Choice and processing of research methods

6. Quality of analysis

7. Quality of the conclusion

8. Use of language

9. Technical presentation.

The attitude of the student

1. Independence

2. Pace of work

3. Handling suggestions

4. Contact with supervisor

5. Other elements considered.

(2)

The Effect of JOBS Act on US IPO Information

Asymmetry and Underpricing Process

MSc in Business Economics, academic year 2015-2016 Universiteit van Amsterdam Hui Hu,1102396 Supervisor: Dr. T. Caskurlu Submission date: 15th

(3)

Statement of Originality

This document is written by Student Hui Hu 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.

(4)

Acknowledgement

I would like to thank our supervisor, Tolga Caskurlu for his helpful guidance in writing this thesis. I would also thank those around me who have given us a language revision and supported me through the process.

(5)

Abstract

This paper studies the effect of Jump Our Business Startups Act (JOBS Act) on Initial Public Offering (IPO) underpricing process. Information asymmetry theory for IPO underpricing is discussed in this thesis and used to explain the JOBS Act’s possible impact on IPO market. A hand-collected sample of 187 IPO firms is used to test the effect of JOBS Act. Linear regression and RD are applied in this study. Different from the previous researches, no significant difference in underpricing caused by JOBS Act is found in my study. Significant difference in underpricing is found between emerging growth companies and none emerging growth companies, however with further test, the hypothesis that the significant difference in underpricing between EGCs and None-EGCs is triggered by JOBS Act is rejected.

(7)

1.Introduction

In this thesis, I investigate the relationship between JOBS Act and IPO underpricing in US IPO market, more specifically: the difference in IPO underpricing caused by JOBS Act between emerging growth companies (EGC) and non emerging growth companies (none-EGC).

On April 5th

, 2012, the JOBS Act (Jumpstart Our Business Startups Act) was enacted, to encourage the funding of small companies which are known as emerging growth companies(EGC). Whether the JOBS Act completed its aim can be tested by different standards: the number of companies going to IPO, IPO volume, IPO proceeds, IPO cost, etc. Among all those standards, IPO proceeds is an important incentive for companies when deciding whether to go to IPO. In many cases, a large amount of proceeds are taken away from issuers’ hand via underpricing process, during which companies (or referred as issuers) choose to give up some proceeds to win a good underwriter, more investors or some other ‘benefits’. A number of papers have investigated the IPO underpricing process: information asymmetry theory is the most famous and popular theories in this field. In this thesis, I would link the effect of JOBS Act to the information asymmetry theory and illustrate why I think the JOBS Act would have an effect on IPO underpricing.

In information asymmetry theory, the first reason why IPO underpricing exists is that uninformed investors ask compensation for staying and investing in this market. According to Rock (1986), for uninformed investors in IPO market, the existence of winner’s curse and ‘unfair’ price they get make them want to quit IPO market. The uninformed investors, however, are an important part of IPO market because they provide liquidity to IPO market. Liquidity in IPO market is good for issuers to sell their equity and can provide IPO stock with

(8)

uninformed investors and to convince them back to the IPO market. As for underwriters, the reputation of successfully underwriting is quite important, they would gain a better reputation if the underwriting success rate is higher than average. The uninformed investors’ participation provides market liquidity and also enlarges IPO purchasing volume, which would help underwriter successfully underwrite. Since successful underwriting is to underwriters’ interest and underpricing would help the IPO stock win uninformed investors and in the end increase the success rate of underwriting, the underwriters have the incentive to underprice the IPO equity as well. To sum up, both issuers and underwriters are willing to underprice the equity and compensate the information asymmetry of uninformed investors and the extend of the information asymmetry determines the amount of compensation (or referred as the amount of IPO underpricing).

Apart from the information asymmetry of uninformed investors, the issuers’ uninformed extend would affect the IPO underpricing extend as well. Baron (1982) points out that the interests of issuers and underwriters are different while issuing IPO and underwriters are better informed than issuers are. For issuers, the optimal situation would be giving up as few proceeds as possible in exchange for benefits they want (investors, liquidity, price support, etc.). However, for underwriters, the situation is quite different. Acting as intermediary between investors and issuers, underwriters have two parties of customers whose interests are completely contradicted. This is because investors can get more returns or profits if the IPO is priced lower, whereas issuers would get less proceeds in that case. Both issuers and investors are underwriters’ clients, underwriters need to decide which clients’ interest to meet. According to Rock (1986), in this situation, underwriters would choose to meet investors’ interest because issuers are only one-time or two-time customers while investors are long-term customers to them. Additionally, underwriters sometimes could get ‘soft dollars’ from the investors if when they give a low price IPO to investors. Namely, the underwriters choose

(9)

to meet investors’ interest to maximize their own profit. The interest conflicts give underwriters the incentive to underprice the IPO more than needed, and the better informed the issuers are, the easier for them to avoid their equity being underpriced overly by underwriters. So the issuers’ extends of information asymmetry would affect the IPO underpricing extends as well.

JOBS Act eases the compliance requirements and disclosure obligations for emerging growth companies (EGCs) whose revenue is smaller than $1 billion in the year prior to IPO. The provisions, such as ‘instead of disclosing the audited 3-year financial statement and 5-year selected financial data in order for registration statement, EGCs can choose to disclose just 2 year audited financial statement’, would affect the extend of information asymmetry in IPO market. And in this case, JOBS Act would affect the IPO underpricing and in the end affect the proceeds of EGCs as well. Proceeds is one of the most important stuffs the issuers would consider about when deciding whether to go to IPO. JOBS Act wants to reduce the disclosure and compliance burden for EGCs to encourage their going public, whereas this act may trigger greater information asymmetry and lower the proceeds for EGCs as well. So I would like to investigate whether JOBS Act would increase the information asymmetry in IPO market and increase the underpricing extends of EGC firms.

Local linear regression, placebo experiment, RD regression and robustness test have been conducted in my study to test the whether the JOBS Act would increase the IPO underpricing extend of EGC firms. A significant difference in underpricing between EGC firms and none-EGC firms has been found in the local linear regression, and the placebo experiment confirms that the prior-to-IPO year revenue itself would not affect underpricing extend. Whereas, there is no cutoff at the threshold found in the RD test and robustness test, so the hypothesis 1 that the JOBS Act has a significant effect on IPO underpricing and proceeds for EGC firms is

(10)

rejected and then the hypothesis 2 that JOBS Act would increase the IPO information asymmetry and underpricing extend of EGC firms should be rejected as well.

2.Overview of JOBS Act (Jumpstart Our Business Startups Act)

and Information Asymmetry Theory

2.1 Provisions of JOBS Act On December 8th

, 2011, the JOBS Act was introduced in the US House of Representatives and it is on 2012 April 5th_{that this Act was signed by president and enacted as law formally.}

There are seven titles in this Act, which aims at helping the small companies to gain ‘cost-effective’ access to capital in a belief that having easier access to capital enables those small companies to create more jobs and to contribute to economic growth. This Act helps small companies get less costly access to capital in two aspects: the first is to enable companies to go public easier than before and the second is to help company raise private capital and able to stay private longer. My study would focus on the part of this Act that has effect on firms’ going public, more specifically, focus on Title 1 which contains the IPO relevant provisions EGCs applicable for.

The EGCs (Emerging Growth Companies) defined by JOBS Act are firms with revenue less than $1 billion in the most recent completed fiscal year, the firms with EGC status can choose to apply any, all or none of the provisions in JOBS Act. And the EGC status lasts till any of the following situation occurs:

a) The issuer (EGC) has an annual total gross revenue over $ 1 billion; b) It comes to the fifth anniversary of the issuer’s IPO date;

(11)

d) The issuer is recognized as a “large accelerated filer” according to Code of Federal Regulations.

Equivalently, for a company who wants to go IPO, the criteria for it can apply the JOBS Act or be an EGC is having an annual revenue smaller than $1 billion in the year prior to IPO.

The Title 1 helps EGCs get less costly access to capital mainly by easing the ‘unnecessary or overly burdensome’ regulatory requirements and the disclosure obligations, in the hope of helping EGCs reduce their cost when going public. There are four aspects that Title 1 of JOBS Act makes difference on EGCs’ IPO process and regulatory requirements.

The first is to reduce the disclosure obligations. There are two disclosure obligations: financial statement disclosure and executive compensation disclosure. For the disclosure of financial statement, instead of disclosing the audited 3-year financial statement and 5-year selected financial data in order for registration statement, EGCs can choose to disclose just 2-year audited financial statement after the enactment of JOBS Act. As for the executive compensation disclosure, firms with EGC status only need to disclose 3 named executive officers compared to 5 prior to JOBS Act. And in EGCs’ registration statement, there is no need to provide Compensation Discussion and Analysis (CD&A) section, the Summary Compensation Table is required to replace the CD&A part.

The second is to help the EGCs get a better chance to go to IPO successfully. There are also two provisions in this aspect. The first is allowing EGCs to submit their draft IPO registration statement to SEC for non-public reviewing before public filing under JOBS Act. Prior to JOBS Act, EGCs need to submit the public-reviewing registration statement directly, which is costly and risky if the EGC cannot raise money in IPO market successfully. The second is to allow the EGC issuers and underwriters to have oral or written communication with investors before going to IPO, which helps all three parties (issuers, underwriters and

(12)

investors) in IPO market get to know each other better and increase the market transparence, thus helps the EGC to gain a larger chance in successfully raising money in IPO market.

The third part focuses on lowering EGCs’ compliance cost, including future accounting standards opt-out, Public Company Accounting Standards Board(PCAOB) opt-out and executive compensation vote opt-out. Though these provisions are an important part in Title I, they affect the transparence in IPO market after EGCs’ going public not before. Thus these provisions would not affect the IPO underpricing extend which depends mostly on the information before firms’ going pubic. So in this study, I would not draw a lot attention on this part.

2.2 Information Asymmetry and underpricing process in IPO market

A number of papers have investigated the mechanism of new issue’s underpricing, and the information asymmetry theory is brought up by Rock (1986). The IPO market as a whole, can be overpriced, fairly-priced and underpriced: when the average offering price is higher than the true value of the equities in the market, the market is perceived as overpriced; when the average offering price equals the true value of equities in the market, the market is perceived as fairly-priced; when the average offering price is lower than the expected true value of equity, the market is perceived as underpriced. Rock (1986) suggests that there are both informed investors and uninformed investors in IPO market. And the informed investors are the ones whose information superior to the information of IPO issuers and uninformed investors. For the informed investors, they only do investment when there are ‘good issues’, namely the issues in which the true value of the equity is higher than its offering price. However, the uninformed investors do their investments randomly, which means that they ask for both ‘good issues’ and ‘bad issues’ due to lack of information. Because informed investors always and only request when the equity is ‘good issue’, though uninformed trader

(13)

do investment randomly, they cannot get the share of ‘good issues’ as much as they want and can always get the ‘bad issues’ in full amount. So in an overpriced or a fairly-priced IPO market, the average return the uninformed investors get is definitely not positive. Imagine a specific situation: in a fairly-priced market where 50% IPO equities are underpriced and 50% IPO equities are overpriced in a same extend, informed investors only request for the underpriced IPO. Thus there are more overpriced equities left in market than underpriced ones, even if the uninformed investors do their investment randomly in a fairly-priced market, the average return for the portfolio they get is going to be negative. To avoid this risk and loss, the uninformed investors would quit the IPO market because collecting information to become informed investors is also costly for them. Those uniformed investors, however, are quite important to IPO market and issuers by providing liquidity to market. So the issuers(firms) are willing to underprice their equity to compensate for the uninformed investors and keep them in the market.

In another word, from the perspective of investors, the underpricing is a compensation for the information collection (Chemanur 1993), this is because it’s hard and costly for investors to obtain all the information about the firms, thus investors are usually investing with asymmetric information and bearing high risk, so the IPOs are underpriced to compensate for this risk.

In this way, the underpricing in IPO is a kind of compensation for uninformed investors because of information asymmetry. Thus markets with different transparency tend to have different compensation for investors. So the extend to which equities are underpriced partly is correlated with the market transparency.

Apart from the information asymmetry of uninformed investors, there exists information asymmetry for issuers as well. Baron came up with the information asymmetry between

(14)

underwriters and issuers in 1982. He suggests that the underwriters have more opportunities to obtain market information about the investors’ interest than the issuers do before the stock going to public. For issuers, they are willing to underprice their own equity in exchange for the ‘benefit’ (investors, price support, etc.) they want (Loughran & Ritter 2002), and the optimal situation for them is to underprice their equity as less as possible. The tricky part is that issuers are less informed than underwriters are, and they would take advice from the underwriters when deciding how many proceeds they should give up. However, for the investors, the situation is different. They want the equity to be underpriced as much as possible, thus they can get more return from the equity. Because of the conflicting interests between issuers and investors, underwriters, as the intermediary between investors and issuers, have to decide which party’s interest to meet. In this case, underwriters would choose to meet investors’ interest because issuers are just one-time or two-time customers while investors would buy equity from them in a long term if the investors get good price (Rock 1986). That is to say, the underwriters would sometimes take advantage of issuers’ information asymmetry and underprice the equity more than necessary to please their investors and get benefit from them.

Underwriters can benefit from underpricing in two ways. Firstly, a cheaper IPO price is more easily accepted by market and therefore lowers the risk of unsuccessful underwriting which will do harm to the underwriter’s reputation (Carter and Manaster 1990). Secondly, underwriters can allocate underpriced shares to their favored investors in exchange for ‘soft dollars’ (Loughran and Ritter 2002). Through the ways above, underwriters can get the benefit from underpricing the stock while the right of issuers is violated. Liu & Ritter (2011) suggests that this kind of benefit which the underwriters share with their favored investors comes from the proceeds which should have been owned by issuers without underpricing process.

(15)

3. Literature Review

There have been different researches investigated the effect of JOBS Act on IPO process, and the conclusions about how the JOBS Act would affect the information asymmetry in IPO market are different.

Dambra, Field & Gustafson (2015) suggest the enacting of the JOBS Act will mitigate the information asymmetry of issuers and investors by the testing the water (TTW) provisions in Title I. TTW provision allows emerging growth companies (EGCS) to engage in oral or written communication with qualified institutional investors prior to public disclosure of the registration statement. This provision provides investors more time to evaluate, understand, and ask questions before road show (Latham & Watkins LLP 2014) and also allows the issuers to obtain and evaluate investors’ interest about the IPO offerings and get a better understanding about the market’s view on how the shares offered by the EGCs should be priced. In TTW process asymmetric information problem in both investors and issuers has been mitigated. So the amount of money on the table (Money on the table is defined as the number of shares issued multiplied by the difference between the first closing market price and the offer price. It is the dollar value of underpricing that is expected to be smaller because both issuers’ and investors’ information asymmetry have been mitigated by provisions of JOBS Act.

However according to the research from Barth, Landsman and Taylor (2014), the information asymmetry increases after the enactment of JOBS Act, and the argument of this research is mainly based on the easing disclosure provisions in Title I of JOBS Act. JOBS Act eases the disclosure obligations and regulatory requirements for the IPO firms with EGC status to help those firms gain an easier access to IPO capital raising, which would, in Barth’s opinion,

(16)

less information a firm discloses, the less information the investors get. Thus both the information asymmetry and underpricing would increase according to the information asymmetry theory in this case. Barth also investigated the the effect of the JOBS Act collectively, finding out that the JOBS Act would increase the IPO underpricing and information asymmetry on the whole. Barth came to her conclusion by comparing firms that have similar characteristics (all have revenues below the $1 billion threshold and could claim EGC status) but issue their IPO in different time. There are two groups of firms in her regression study: the treatment group (EGC firms) is the firms with EGC status (those firms have annual revenue less than $1 billion in their last completed fiscal year) and issue their IPO after the enactment of JOBS Act, which is 2012 April 5th

; the control group (None-EGC firms) is the firms that have their revenue less than $1 billion in their last completed fiscal year (below the threshold), which means they could claim their EGC status if they issue their IPO after 2012 April 5th

, however issue their IPO before the JOBS Act’s enactment (for example, a firm had a revenue less than $1 billion but issued its IPO before 2012 April 5th

). The statistical result of Barth’s research proved her hypothesis: the JOBS Act increased the information uncertainty and underpricing in IPO market.

However, the methodology Barth used may exist potential problems. She used IPO firms from different time periods to construct her sample: firms conduct IPO before April 5th

, 2012 (JOBS Act’s enactment date) are identified as Non-EGC firms (control group) and firms conducted IPO after April 5th

, 2012 are identified as EGC firms (treatment group), which in my opinion, is not perfectly comparable because the extend of IPO underpricing has strong correlation with macro economic environment. According to Loughran and Ritter (2004), IPO underpricing could change a lot over time: in 1980s, the average first-day return is 7%, and doubled during 1990-1998, and then jumped to 65% during the internet bubble. The financial crisis of 2007- 2009 provided a dramatically unstable and fluctuated market.

(17)

Though the EGC firms and NEGC firms in Barth’s sample have similar characteristics, the firms’ different macro economic environment may lead to significant difference in IPO underpricing between the treatment and control group. In this case, it is not accurate to attribute the significant difference in underpricing to JOBS Act.

So in my study, I restrict my sample data to the time period 5th

August 2012 - 31st

December 2015, a shorter time period when the US IPO market is more stable than the US IPO market in Barth’s experiment period. Because JOBS Act was enacted in 2012, the control group in my study is also different from that in Barth’s study. Control group in my study are the firms going to IPO after 5th

April 2012 but having an annual revenue larger than $1 billion in the year prior to their IPO, which means they cannot apply the JOBS Act or be an EGC even after the enactment of JOBS Act.

Though in my study, the difference in underpricing between treatment group and control group has been mitigated, there are two shortcomings. The first is that the number of the firms issuing their IPO in the time period 5th

April 2012 - 31st

December 2015 is limited, which means the number of data in my sample is limited. This shortcoming is hard to solve, because if I want to enlarge my sample, I need to choose a wider time period in my study, then the difference in underpricing caused by macro economics cannot be mitigated. The second shortcoming is the concern that whether my treatment group and control group are comparable, because firm’s revenue in the year prior to IPO for those two group of firms could be quite different. To solve this shortcoming, RD test is applied in robustness test: the firms close the $1 billion threshold are good comparisons to each other. Whether existing a cutoff in IPO underpricing at the threshold would be a crucial part to investigate.

4. Hypothesis

(18)

1. JOBS Act has a significant effect on IPO underpricing and further has a significant effect on the IPO proceeds of EGC firms.

2. JOBS Act would increase information asymmetry in market and the underpricing extend of firms with EGC status.

Notably, hypothesis 2 is based on hypothesis 1, if in any test hypothesis 1 is rejected, then the hypothesis 2 must be rejected either.

5. Methodology

5.1 Local linear regression

To test the hypothesis, local linear regression is applied to test the underpricing difference between the firms eligible to JOBS Act (EGC-eligible firms with less than $1billion revenue in the year prior to their IPO) and firms not eligible to JOBS act (EGC-ineligible firms with more than $1 billion revenue in the year prior to their IPO). Dummy variable named EGC is generated in this regression to identify whether the firm can apply the JOBS Act when going IPO, where EGC equals 1 if the firm can apply JOBS Act and equals 0 if the firm cannot. First-day return is used to capture the extent of IPO underpricing, the larger the underpricing extent the larger the first day return. First-day return is defined as the difference between first-day closing price and offering price divided by offering price.

To improve the accuracy of the model, control variables are included, the the model 1 is as follow:

!"#$% − !"# !"#$!%_!

= !!+ !!!"#!+ !!ln !""#$"! + !!!"#! + !!!"#! + !!ln !"#"$%"! + !!!&! − !" − !"#$% + !!!"#$%#&'(%)!

(19)

Model I

Where the !"#!_{equals 1 if this IPO firm has EGC status(has a annual revenue less than $1} billion in the fiscal year prior to IPO), and equals 0 if this IPO firm has None-EGC status (has a annual revenue more than $1 billion for the last completed fiscal year).

The control variables are included in this model to make this model more accurate: firm size (lnAsset), profitability (ROA), book-to-market ratio (BMR), revenue (lnRevenue), R&D fees (R&D-to-sales), and the percentage of retained earnings to net income (PctRetained).

ln !""#$"! is the the natural logarithm of total assets, and ROA is the return on asset which is generated by dividing net income with total assets, the market to book ratio is calculated by dividing book value of a firm with its market value, and ln !"#"$%"_! is the natural logarithm of revenue, and !&! is calculated by dividing firms research and development fees with firm’s sales.

!_! is the coefficient of interest which indicates the JOBS Act increase the IPO underpricing by !!. If !! is statistically significant and different from 0, then the hypothesis 1 that the JOBS Act has a significant effect on IPO underpricing process and information uncertainty holds. And if !! is not significantly different from 0, then it indicates that there is no significant effect of JOBS Act on IPO underpricing process. If !! is significant and positive, then the hypothesis that JOBS Act would increase IPO underpricing and information asymmetry holds. If !! is significant and negative, then the JOBS Act would decrease IPO underpricing and market information asymmetry.

5.2 Placebo experiment

However, there exists a concern that the IPO underpricing difference between EGC-eligible firms and EGC-ineligible firms is not caused by JOBS Act, but caused by the different

(20)

prior-to-IPO revenue itself, which means these firms have other different characteristics that may lead to different IPO underpricing. In order to address this concern about endogeneity, apart from including the control variables, a placebo experiment is conducted to confirm whether the revenue in the year prior to IPO would affect underpricing.

A random threshold $300 million is created in this experiment to imitate the $1 billion threshold created by JOBS Act. The dummy variable random-threshold is created in this model, which equals 1 if the firm has a revenue smaller than $300 million in the fiscal year prior to IPO and equals 0 if the firm has a revenue larger than $300 million in the fiscal year prior to IPO. The basic idea about this placebo test is: the two groups of firms in this experiment have different revenues in the year prior to IPO as well, if there is no significant difference in underpricing between those two groups of firms, then the threshold itself or the revenue of the year prior to IPO would not affect the underpricing extend, and in this case, we can attribute the significant difference in model I to JOBS Act.

All the control variables and data sample used in this test are the same as those used in model I, only the dummy variable has been changed, and the model is as follow:

!"#$% − !"# !"#$!%_!

= !_!+ !_!!"#$%& − !ℎ!"#ℎ!"# + !_!!"#$$%&$_!+ !_!!"#_!+ !_!!"#_! + !_!!"!"#"$%"_!+ !_!!&! − !" − !"#$% + !_!!"#$%#&'(%)_!

Model II

Similar to model I, all the control variables and data are the same, only the variable of interest is different. The coefficient of interest in this model is !!. If !! is significantly different from 0, then there exists significant difference in these two group of firms’ underpricing extend, it

(21)

means the the revenue of the year-prior-to IPO would affect firms’ underpricing extend and there exists endogeneity in model I. In this case, the result about underpricing in model I is not valid and coefficient of interest in model I cannot be interpreted as causal effect either. If !_! is not significantly different from 0, then there is no significant difference in underpricing between these two groups of firms, which also indicates the revenue of the year prior-to-IPO itself would affect underpricing process, thus the conclusion about IPO underpricing I get from model I is valid and trustworthy.

5.3 Regression discontinuity

Regression discontinuity (model III) is also used to investigate the effect of JOBS Act on IPO underpricing. The idea to apply RD regression comes is to make the firms in treatment group and control group more comparable. JOBS Act creates a $1 billion revenue threshold: the firms which generated a gross revenue larger than $1 billion in the year prior to IPO (None-EGC firms) cannot apply this act while the firms which generated a gross revenue smaller than $1 billion (EGC firms) can apply, and this divides the IPO firms in two groups: EGC-eligible firms and EGC-inEGC-eligible firms. Though there are a lot differences between those two groups of firms, the companies just below the cutoff (threshold) are good comparisons to those just above the threshold.

In order to conduct this RD test, a subsample in which firms have a similar revenue in the year prior to IPO needs to be created. In this subsample I identify treatment group and control group as eligible-A and ineligible-A respectively. eligible-A and EGC-ineligible-A are the groups of firms whose last fiscal year’s revenue close to the $1 billion threshold and should have had the similar underpricing extend if there is no JOBS Act. Thus if there is discontinuity of underpricing between EGC-eligible-A and EGC-ineligible-A firms, it is evident that JOBS Act does have effect on IPO underpricing extend. And if there is no

(22)

significant underpricing discontinuity between EGC-eligible-A and EGC-ineligible-A firms, then the underpricing difference may be caused by other characteristic instead of JOBS Act.

The model III is as follow: !"#$% − !"# !"#$!%_!

= !_!+ !_!!"# − !"#$#"%"! − !_! + !_!!"#$$%&$_! + !_!!"#_! + !_!!"#_! + !_!!"!"#"$%"_! + !_!!&! − !" − !"#$% + !_!!"#$%#&'(%)_!

Model III

All the control variables are the same as model I, only the variable of interest and sample of data are different. The firms used to test in this model are the EGC firms and none-EGC firms with revenue close to the JOBS Act threshold, namely, the EGC firms have a relatively larger prior-to-IPO revenue and the none-EGC firms with a relatively lower prior-to-IPO revenue. The specific criteria of selecting this subsample including EGC-eligible-A firms and EGC-ineligible-A firms would be discussed in data selection part. The variable of interest EGC-eligible-Ai equals 1 if the firm belongs to this subsample and is an EGC firm and equals

0 if the firm belongs to this subsample and is an none-EGC firm. The coefficient of interest in this model is !_!, if !_! is significantly different from 0, then there exists significant difference in these two group of firms’ underpricing extend, if !! is not significantly different from 0, then there is no regression discontinuity close to JOBS-threshold, which also indicates that the underpricing difference between EGC-eligible and EGC-ineligible firms might be caused by JOBS Act. So I use RD in my research to further test the firms close to the cutoff and investigate whether it is JOBS Act that triggers this underpricing difference.

(23)

6. Data and descriptive statistics

6.1 Data Collection

The data used in this thesis is made up by 4 portions: CRSP/Compustat Merged,

CRSP Daily Stock, Thomson one and a hand-collected data sample.

I firstly get the list of IPO firms which issued their IPO after 2012 April 5th (the date on which JOBS Act is enacted) from CRSP/ Compustat Merged database and download the variables including total assets, revenue, R&D expenses, market value, net income, SIC code, number of shares outstanding and retained earnings. From this sample, I exclud the firms in finance, insurance, real estate industry (with SIC code from 6000 to 6799), mining industry (with SIC code from1000 to 1499), construction industry (with SIC code from1500 to 1799) and end up with 410 firms in this sample.

The underpricing data (first-day return) for IPO firms can be obtained by two approaches. The first is to download directly from Thomson one Database and the second is to get the first day closing price from CRSP Daily Stock, and get the offering price from the registration statement in the Securities and Exchange Commission’s (SEC’s) Electronic Data Gathering and Retrieval (EDGAR) system, then divide the difference between the closing price and the offering price with offering price to get the first-day return. I try both methods and end up with the same result. The dependent variable first-day return and control variables lnAsset, lnRevnue, book to market ratio, R&D fee to sales, return on asset, retained earning to net income could be generated with the data above.

The forth portion of data collection is to hand collect the IPO firms’ revenue in the year prior to IPO. This is to identify IPO firms’ EGC status: firms with revenue of last completed fiscal year prior to IPO greater than $1 billion are identified as Non-EGC firms, while the firms

(24)

with revenue of last completed fiscal year prior to IPO smaller than $1 billion are qualified as emerging growth companies (EGC). To reflect the impact of this Act fully, I choose the firms that conducted their IPO 4 months after the enactment, namely, the firms in this sample are the ones that issue their IPO after August 5th, 2012. I hand collect the last fiscal year revenue as well as offering price data of 328 IPO firms from the 424b filings in Securities and Exchange Commission’s (SEC’s) Electronic Data Gathering and Retrieval (EDGAR) system. After excluding the the cross listing firms and firms without a completed or standard data for revenue, there are 242 firms left, with 37 None-EGC firms and 205 EGC firms in this sample. JOBS Act was enacted in 5th April, 2012 while the data in my sample is from 5th August, 2012. The reason why I exclud the first three months is that the effect of JOBS Act would be more evident and stable after a certain time period, which has already been proved in previous research. Before doing the regression, I also delete the firms with a too large or too small revenue of the year prior to IPO. And the whole sample used in model I and placebo experiment (model II) is 187 IPO firms with a revenue larger than $1million and smaller than $6 billion in the year prior to IPO.

The subsample used in RD test (model III) are the firms with a revenue larger than $300 million and smaller than $2 billion in the fiscal year prior to IPO, including 20 EGC firms and 17 none-EGC firms. The ideal situation is choosing the firms closer to the threshold to make treatment group and control group as comparable as possible. However, with limited number of data in the subsample, the criteria $300 million - $2 billion cannot be further narrowed. The advantage of this subsample is that the treatment group and control group in this subsample have already been more comparable than those in the whole sample.

(25)

6.2 Descriptive statistics

Fig.1. Initial Public Offering (IPO) first-day return for the whole sample (firms with prior-to-IPO revenue larger

than $1 million and smaller than $6 billion) after enactment of JOBS Act. The y-axis reflects the first-day return of IPOs which is also IPO underpricing extend, and the x-axis reflects the different of prior-to-IPO revenue of IPO issuers. The dashed line represents the $1 billion threshold set by JOBS Act. Collectively this scattered figure reflects the relationship between firm’s prior-to-IPO revenue and IPO underpricing extend. And whether there exists significant discontinuity close to the threshold can indicate whether JOBS Act has a significant effect on IPO underpricing process.

Fig.1. shows the IPO first-day return of firms in the whole sample with different prior-to-IPO year revenue and issuing their IPO after August 5th

, 2012. The common trend in this figure is that as the firm’s prior-to-IPO year revenue increases, the firm’s underpricing drops. It can be observed that the IPO underpricing extends of the firms with smallest prior-to-IPO year revenues are much bigger than those of the firms with largest prior-to-IPO year revenues. Notably, some firms with smallest prior-to-IPO revenue have an evident larger underpricing extends even than other EGC firms in this sample. However, for the EGC firms none-EGC firms near $1 billion threshold, the underpricing difference is not evident in this figure, so

(26)

whether there is a significant difference caused by JOBS Act remains unclear: there is no evident discontinuity of first-day return for the firms close to the threshold.

Table1

Descriptive statistics for Initial Public Offering(IPO) issuers after JOBS Act (whole sample)

This table presents the descriptive statistics for IPO issuers (including EGC-eligible firms and EGC-ineligible firms) between August 5th_{2012 and December 31}st_2015.

Characteristic Mean Median Standard

deviation

Firm characteristics

First-day return 20.55% 11.50% 0.31

Revenue for the year prior to IPO (millions) $404.31 $86.45 $687.28

Asset (millions) $488.00 $125.80 $781.91

Revenue (millions) $559.50 $125.80 $1,024.66

ROA (return on asset) -11.88% -5.26% 0.17

BMR (book to market ratio) 22.95% 18.65% 0.17

R&D-to-sales 51.74% 14.56% 0.93

Retained earnings to net income 33.45% 269.73% 12.77

Table1 presents the descriptive statistics for the whole sample IPO issuers after JOBS Act. The average first-day return (underpricing) in the time period 5th

August, 2012 – 31st

December, 2015 is 20.55% and the median of first-day return in this time period is 11.5%. Besides, the average prior-to-IPO revenue is $404.31 millions and the median is $86.45 millions, we can see that a large portion of firms in this whole sample can be categorized as EGC, and half of them have a prior-to-IPO revenue smaller than $86.45 millions which is very small comparing to the threshold set by JOBS Act.

(27)

Descriptive statistic (whole sample) comparison between EGC-eligible firms and EGC-ineligible firms, namely the descriptive statistic comparison between firms which can apply the ‘benefit’ of JOBS Act and firms which cannot in whole sample.

This table presents the mean and median of those firms’ characteristics, also indicates the difference of characteristics between those two groups of firms.

Characteristic EGC-ineligible EGC-eligible Increase in mean Increase in median

Mean Median Mean Median

Firm

characteristics

First-day

return 11.89% 6.41% 22.04% 12.92% 85.36% 101.50%

Revenue for the year prior to IPO (millions) $1,910.61 $1,884.30 $144.06 $75.44 -92.46% -96.00% Asset (millions) $2,081.81 $2,242.18 $219.16 $108.95 -89.47% -95.14% Revenue (millions) $2,545.00 $2,242.18 $224.59 $108.95 -91.18% -95.14% ROA (return on asset) 1.50% 1.03% -14.14% -9.62% -1041.76% -1033.57% BMR (book to market ratio) 17.57% 15.62% 23.87% 19.34% 35.86% 23.83% R&D-to-sales 0.77% 0.00% 60.55% 20.47% 7780.08% N/A Retained earnings to net income -501.61% 13.58% 123.70% 299.61% -124.66% 2106.12%

Table 2 presents the descriptive statistics of EGC firms and none EGC firms in the whole sample. In this comparison table, it can be observed that the first-day return for EGC firms and none-EGC firms are quite different, with an 85.36% increase in first-day return for firms with EGC status. However, for the treatment group and control group in this whole sample, other characteristics are quite different as well, the necessity of constructing a subsample with more comparable treatment group and control group can be confirmed by this table.

(28)

Fig.2. Initial Public Offering (IPO) first-day return for the subsample (firms with prior-to-IPO revenue larger

than $300 million and smaller than $2 billion) after enactment of JOBS Act. The y-axis reflects the first-day return of IPOs which is also IPO underpricing extend, the x-axis reflects the different prior-to-IPO revenue level of IPO issuers. The dashed line represents the $1 billion threshold set by JOBS Act. Collectively this scattered figure reflects the relationship between firm’s revenue and IPO underpricing extend. And whether there exists significant discontinuity close to the threshold can indicate whether JOBS Act have a significant effect on IPO underpricing process.

Fig.2. shows the IPO first-day return of firms with different prior-to-IPO year revenue and issuing their IPO after August 5th

, 2012 in the subsample. There is no common trend nor evident cutoff at the threshold that can be observed in this figure. The first-day return for both EGC firms and none-EGC firms seems to be random. However, according to the principle of RD, although the making a figure to see whether there is a cutoff is important in a RD experiment, the figure cannot be the only criteria in deciding whether there is a significant

(29)

difference caused by the treatment. Thus no further conclusion can be drawn without doing the regression test.

Table3

Descriptive statistics for Initial Public Offering(IPO) issuers after JOBS Act (subsample sample)

This table presents the descriptive statistics for IPO issuers (including eligible-A firms and EGC-ineligible-A firms) between August 5th_{2012 and December 31}st_2015.

Characteristic Mean Median Standard _deviation

First-day return 23.48% 15.57% 0.30

Revenue for the year prior to IPO (millions) 904.47 887.90 494.55

Asset (millions) 1936.72 1241.79 1734.94

Revenue (millions) 1253.28 1166.41 797.80

ROA (return on asset) 1.25% 1.33% 0.11

BMR (book to market ratio) 16.20% 17.39% 0.52

R&D-to-sales 3.16% 0.00% 0.15

Retained earnings to net income -167.12% 98.29% 27.87

According to the descriptive statistics in table 3, the average first-day return in this subsample is 2.93% larger than that in the whole sample. The mean of prior-to-IPO revenue in this subsample is $500.16 millions larger than that in the whole sample.

(30)

Table4

Descriptive statistic (subsample sample) comparison between EGC-eligible-A firms (prior-to-IPO revenue from $300 millions to $1billion) and EGC-ineligible-A firms (prior-to-IPO revenue from $1 billion to $2 billions), namely the descriptive statistic comparison between firms which can apply the ‘benefit’ of JOBS Act and firms which cannot in subsample.

This table presents the mean and median of those firms’ characteristics, also indicates the difference of characteristics between those two groups of firms.

Characteristic _{EGC-ineligible-A} _{EGC-eligible-A} Increase in mean Increase in median

_Mean _Median _Mean _Median

First-day return _17.96% _15.79% _26.96% _15.36% _50.15% _-2.74%

Revenue for the year prior to

IPO (millions) _{$1,433.52 $1,423.75} _$571.36 _$502.32 _-60.14% _-64.72%

Asset (millions) _{$2,307.67 $2,047.29 $1,666.93 $1,015.41} _-27.77% _-50.40%

Revenue (millions) _{$1,796.72 $1,542.68} _$858.05 _$680.30 _-52.24% _-55.90%

ROA (return on asset) 3.08% 0.94% -0.09% 1.48% -102.82% 57.58%

BMR (book to market ratio) _22.08% _19.49% _11.73% _14.95% _-46.88% _-23.30%

R&D-to-sales 0.66% 0.00% 4.97% 0.00% 652.51% N/A

Retained earnings to net

income -14.45% 120.34% -278.14% 61.98% 1824.83% -48.49%

Table 4 presents the descriptive statistics of A firms and the EGC-ineligible-B firms, it can be observed that except for book-to-market ratio and retained earnings to net income, all other characteristics are more comparable in this subsample. This increases my confidence that the treatment group and control group are more comparable in this RD subsample than those in the whole sample. Notably, the mean of first-day return is larger in both control group and treatment group in this RD subsample comparing with the first-day return in whole sample’s control group and treatment group, with the mean of first-day return increasing by 6.07% in control group and and 4.92% in treatment group respectively. The average first-day return of control group increasing more than that of treatment group may lead to an insignificant difference in underpricing in the RD test. This can also indicate that

(31)

the none-EGC firms with prior-to-IPO revenue larger than $2 billions experience a lower underpricing than the EGC-ineligible-A firms. If the significant difference in underpricing between control group and treatment group found in model I is caused by the smaller underpricing extend the none-EGC with prior-to-IPO revenue larger than $2 billions have, then the significant difference cannot be attributed to JOBS Act.

7. Result

7.1.1 IPO underpricing difference between eligible firms and EGC-ineligible firms

AS first step to investigate the effect of JOBS Act on IPO underpricing, I compare the IPO underpricing of firms who are eligible to JOBS Act with underpricing of the firms who are not. The IPO underpricing is scaled by first-day return. The coefficient of interest in model I is EGC, which equals to 1 if the IPO firm is EGC eligible and equal to 0 if it is not. I also control for the asset, return on asset, book to market ratio, R&D fee to sale ratio, revenue, and the retained earnings to net income. From the regression result in mode I (the result is shown in the first column of Table5), it can be observed that the IPO firms with EGC status have larger IPO underpricing extends than the firms without an EGC status do. The EGC status would increase firms’ underpricing extend by 0.183 at 95% significance level. And we can also see that although different provisions of JOBS Act have completely different effect on market transparency with some provisions that increasing the market transparency and some decreasing, its effect on IPO market transparency is negative in total according to the coefficient of EGC in model I and Information Asymmetry Theory for IPO underpricing. This is consistent with my hypothesis 2 that JOBS Act or EGC status would have negative effect on IPO underpricing. This is also in line with previous researches.

(32)

However, I can not rule out alternative causes such as other firm characteristics omitted in my model without further investigation. Thus I conduct two more regressions to determine whether the effect should be attributed to JOBS Act.

7.1.2 IPO underpricing difference in placebo experiment

To further test whether there is an effect of JOBS Act on IPO underpricing, I set up a threshold randomly as placebo experiment and the result is shown in the second column of table 5. The variable of interest is random-threshold, which equals 1 when the firm is below the random threshold ($0.3 billion) I set, and equals 0 when the firm is above the threshold. All data and control variables in this placebo test are the same as in the previous regression. As I expected, there is no significant difference between these two groups of firms. The result of this placebo experiment confirms my hypothesis: the underpricing difference between EGC firms and none-EGC firms is triggered by JOBS Act.

7.1.3 IPO underpricing in RD

To further test the effect of JOBS Act on IPO underpricing, I use RD to compare the firms just below the JOBS Act threshold ($1 billion) with the firms just above. The coefficient of interest is EGC-eligible-A in this model, which equals 1 when the firm is below the JOBS Act threshold and has an annual revenue larger than $300 million in the year prior to its IPO, while equal 0 when the firm is above the JOBS Act threshold and has an annual revenue smaller than $2 billion in the year prior to its IPO. However, no significant difference in underpricing extend is found in this RD test, which is inconsistent with my hypothesis and the result I got in previous regression.

This could because the subsample I use in this model is smaller than the whole sample I use in previous regression: in this RD test, I exclude the firms having an annual revenue larger than $2 billions in the year prior to IPO, and those firms actually have a relatively small

(33)

underpricing extend. This means, when I exclude those firms with larger revenue from my sample, there is no significant underpricing difference between EGC firms and none EGC firms. Another possible reason for this result is that the number of observations in this subsample is too small, this reason would be tested in robustness test. However, the result of this RD experiment illustrates the significant difference in IPO underpricing found in model I cannot be attributed to JOBS Act for sure and the hypothesis should be rejected.

(34)

Table5

Regression result for Initial Public Offering (IPO) issuers’ underpricing after JOBS Act

Each dependent variable measures the underpricing extend of IPO issuers (including EGC-eligible firms and EGC-ineligible firms in model I and model II, EGC-eligible-A and EGC-ineligible-A in model III) between August 5th_{2012 and December 31}st_{2015. In model I and model II, regression runs in the whole sample, which}

includes all EGC-eligible and EGC-ineligible companies issued their IPO between August 5th 2012 and December 31st_{2015 and meet sample selection criterial in data selection part. In model III, regression runs in}

the subsample, which includes EGC-eligible-A and EGC-ineligible-A companies issued their IPO between August 5th_{2012 and December 31}st_{2015 and meet sample selection criterial in data selection part. The}

coefficients are presented below, *, ** and *** indicate significance level at 10%, 5% and 1% level, respectively.

Parameter Model I Model II Model III

EGC 0.183** _[0.076] Random-threshold 0.035 [0.0863] EGC-eligible-A 0.043 [0.102] Ln(Assets) 0.007 0.002 -0.158** [0.030] [0.032] 0.066 ROA 0.117 _[0.193] 0.123 _[0.196] [-0.282] _0.483 BMR -0.647*** -0.652*** [0.094] [0.116] [0.116] 0.095 Ln(Revenue) -0.002 _[0.028] -0.022 _[0.031] [0.069] _0.105 R&D-to-Sales -0.018 -0.038 0.636*** [0.036] [0.038] [0.146] PctRetained 0.000 _[0.002] 0.000 _[0.002] -0.003 _[0.003] Cons 0.215 _[0.170] 0.483*** _[0.217] 0.849 _[0.714] R2 0.144 0.123 0.269 Number of observations 187 187 37

8. Robustness checks

In this robustness check, I want to include more observations in robustness-checking sample than that in the previous RD subsample to improve the accuracy of my test and see whether having more data in sample would make the underpricing extend of treatment group

(35)

significantly different from that of control group. Additionally, this robustness test can further confirm whether the significant difference in underpricing found in model I is due to the small underpricing extends of firms with very large prior-to-IPO revenue (larger than $2 billions), namely, whether the significant difference in underpricing found in previous regression is not triggered by JOBS Act. After doing the RD with 59 firms having prior-to-IPO revenues larger than $100 millions (comparing with $300 millions in the previous RD subsample) and smaller than $2 billion, I still find no significant difference in underpricing found. In this case the hypothesis should be rejected for sure.

9. Conclusion

In this study, I discuss the provisions of JOBS Act and relevant IPO underpricing theories and explain why there might be difference between the underpricing in EGC firms and None-EGC firms. A new type of IPO issuer, emerging growth company (None-EGC), is created by JOBS Act. JOBS Act aims at encouraging EGCs’ initial public offerings by easing the information disclosure requirement for EGCs. The question this study dealing with is whether JOBS Act increases IPO underpricing extend and thus decreases the proceeds for EGC firms when going public. To address this question I conduct a linear regression, a placebo experiment and a RD test in my study.

I find a significant increase in IPO underpricing between the firms which can apply JOBS Act and the firms which cannot do. More specifically, The EGC status would increase firms’ first-day return by 18.3%. Namely, firms which apply JOBS Act tend to have a higher underpricing when going to IPO. This is in consistent with my prediction and the study of Barth (2015): easing information disclosure increases market information asymmetry and IPO underpricing, and thus increases the IPO cost for firms. However, JOBS Act is not the

(36)

only difference between the two groups of firms in my first model, so conclusion cannot be drawn based on this result solely.

In the placebo experiment I conduct, I create a random threshold $300 million. The data sample and control variables used in this experiment are exactly the same as those used in the first regression. No significant difference in IPO underpricing is found in this test, which increases my confidence that the threshold itself would not trigger significant difference in underpricing.

To be more precise, I use regression discontinuity to investigate the underpricing difference of firms close to the $1 billion threshold required by JOBS Act. According to RD experiment principle, there should be a cutoff at the threshold if the ‘treatment’ does have effect on the dependent variable. However, in this RD test, no significant difference in underpricing is found between the EGC firms and None-EGC firms close the threshold. This finding contradicts to my hypothesis and the result of previous regression. There might be two reasons for this none significant result. The first is that in this RD test, I exclude the firms with an annual revenue smaller than $300 million in the year prior to IPO. In the chapter overview of data, it can be seen that those firms with small revenue pre to IPO are generally the ones experience the largest IPO underpricing. Including them into EGC group in model 1 would increase the underpricing extend of EGC group. Whereas excluding those firms in this RD test would decrease the IPO underpricing extend of EGC group. This might be one of reasons why there is no significant difference in underpricing in RD test. The reason why the firms with a prior to IPO revenue smaller than $300 million experience a larger extend of IPO underpricing could be investigated in the study afterwards. The second reason might be the limitation of my sample, after excluding the firms ‘far away’ from the $1 billion threshold,

(37)

there are only 37 observations in the sample, this sample might not be large enough to find out the significant difference.

Taken together, the hypothesis that JOBS Act affects IPO underpricing extend for IPO firms with EGC status and thus has a significant impact on EGC firm’s proceeds should be rejected, because no significant difference between two groups of firms in RD test and robustness test. So after all those analysis, the hypothesis that JOBS Act would increase IPO market information asymmetry and increase IPO underpricing extend for EGC firms should be rejected as well.

References

Baron, D 1982, ‘A model of the demand of investment banking advising and distribution services for new issues’, Journal of Finance, vol. 37, pp. 955-976.

Barth, ME, Landsman, WR & Taylor, DJ 2014, ‘The JOBS Act and information uncertainty in IPO firms’, Research paper, Stanford University, viewed 12 June 2016, https://corpgov.law.harvard.edu/2014/08/20/the-jobs-act-and-information-uncertainty-in-ipo-firms/

Carter, R & Steven, M 1990, ‘Initial public offerings and underwriter reputation’, Journal of Finance, vol. 45, pp. 1045-1067.

Chemmanur, TJ 1993, ‘The pricing of initial public offers: A dynamic model with information production’, Journal of Finance, vol.48, pp. 285-30.

Cohen VF & 2014, The JOBS Act Two Years Later: An Updated Look at the IPO Landscape,

Latham & Watkins LLP, Los Angeles, viewed 23 May 2016,

(38)

Dambra, M & Field, LC & Gustafson, MT 2015, ‘The JOBS Act and IPO volume: Evidence that disclosure costs affect the IPO decision’, Journal of Financial Economics, vol. 116, pp. 121–143.

Dharmapala, D & Khanna, VS, 2015 ‘The Costs and Benefits of Mandatory Securities Regulation: Evidence from Market Reactions to the JOBS Act of 2012’, Research paper, University of Chicago, viewed 12 June 2016, Social Science Research Network.

Loughran, T & Ritter, J 2002, ‘Why don’t issuers get upset about leaving money on the table in IPOs’, Review of Financial Studies, vol. 15, pp. 413–443.

Loughran, T & Ritter, J 2004, ‘Why has IPO underpricing changed over time’, Financial Management, vol. 33, no. 3, pp. 5–37.

Liu, X & Ritter, J 2011, ‘Local underwriter oligopolies and IPO underpricing’, Journal of Financial Economics, vol. 102, pp. 579-601.

Rock, K 1986, ‘Why new issues are underpriced’, Journal of financial economics, vol. 15, no. 1-2, pp.187-212.

Strand, ME & Skattum, AF 2013, ‘IPO Underpricing and family controlled firms – A study of the Norwegian market’, Master thesis, BI Norwegian Business School, viewed 1 July 2016, http://www.bi.edu/oslofiles/ccgr/student_papers/msc/ccgr_msc_052.pdf