Amsterdam Business School
Master Business Economics Finance track
T
HE CHANNELS OF PRE
-
IPO AGENCY COSTS DUE
TO PRE
-
IPO OWNERSHIP STRUCTURE AND ITS
EFFECTS ON UNDERPRICING
Master thesis by:Annisa G. Marconi 5725143 Thesis supervisor: Dr. Jens Martin Date: July 7th, 2014
Abstract
This thesis is aimed to contribute to the little research that has been done on pre-IPO relations, more specifically, the channels of pre-IPO agency costs. This is done by combining models that test the presence of pre-IPO agency problems and the cost of agency problems measured in underpricing. The results indicate that these relations are present, though not always significant. Also, it appears that the proxies for agency costs do not provide the expected results for the sample. More research in this field of study is highly wel-comed.
Table of contents
1 Introduction ... 4
2 Existing literature ... 6
2.1 Why firms go public ... 6
2.2 The separation of ownership and control ... 8
2.3 Mitigating pre-IPO agency problem ... 9
2.3.1 Monitoring ... 9
2.3.2 Take-over bids ... 10
2.4 Agency problem in IPO ... 11
2.4.1 Manifestation of agency problem in IPO ... 11
2.4.2 Mitigating agency problem in IPO ... 12
2.5 Number of shares offered in sight of dilution of control ... 12
3 Model ... 13
3.1 Channels of agency costs ... 13
3.2 Effect agency costs and insiders on underpricing ... 15
3.3 Retainment or dilution of control ... 17
3.4 Effect of retention in post-IPO managerial decisions ... 19
3.5 Effect of retention in post-IPO performance ... 20
4 Dataset ... 22
4.1 S-1 SEC filings ... 22
4.2 Post-IPO accounting and stock price data ... 23
4.3 Descriptive statistics ... 23
4.3.1 Descriptive statistics pre-IPO agency costs ... 23
4.3.2 Descriptive statistics retention and post-IPO ... 24
5 Results ... 27
5.1 Pre-IPO insider ownership as single regressor ... 27
5.2 Pre-IPO insider ownership in multivariable regressions ... 27
5.3 Efficiency ratios and underpricing ... 28
5.4 Pre-IPO ownership and the shares that go to the public ... 29
5.5 Post-IPO efficiency ratios ... 31
5.6 Post-IPO performance ... 32
6 Conclusion and limitations ... 34
Bibliography ... 36
1. Introduction
The initial public offering (IPO) of firms has been a subject for research for many years. When a company goes public, the firm becomes subject to the agency problem that is a result of the separation of ownership and control. Jensen and Meckling’s 1946 re-nowned article sheds light on this phenomenon. Much of this agency problem research was conducted on post-IPO data of public firms. This may not come as a surprise as public firms are required to disclose their data, which then are gathered and filed in da-tabases and are readily accessible in this digital age. The opposite is the case with com-pany data prior to its IPO. This has withheld the majority of scholars to investigate post-IPO theories and arguments in a ppost-IPO setting. This leaves a lot of room for new re-search and findings in the pre-IPO firm field of studies.
There are, however, only a few scholars that have found a way to collect pre-IPO data to conduct research on a firm’s pre-pre-IPO state. Ang, Cole and Lin (2000) found that the agency problem is as much present in a private firm as in a public firm. The au-thors used an anonymous survey on small and medium enterprises, the Survey of Small Business Finances from 2004, commissioned by the Federal Reserve Board, in which data such as ownership structure and operational data were surveyed and filed. This al-lowed them to test for relations between ownership characteristics and financial data. The authors argued that there are two proxies of agency costs are present at a private firm, namely the excessive spending spree by managers and the inefficient utilization of assets.
Alavi, Pham and Pham (2008) are another trio of scholars who found a pre-IPO data set they could conduct research with. Their research focused on how pre-IPO own-ership structure impacted decisions during the IPO process. They also found that some form of agency problem was present prior to an IPO, namely that non-shareholding managers merely act as an agent on behalf of the (management) shareholders.
This thesis searches to combine both researches by Alavi, Pham and Pham (2008) and Ang, Cole and Lin (2000). More specifically, the proxies argued by the for-mer trio are applied to the models of the latter trio.
Commission (SEC) requires companies to include financial and ownership data for a period of at least three years prior to the IPO. These, so called, S-1 filings are publicly available on the SEC’s website and allow for manually gathering data.
This research contributes to the scarce pre-IPO research and seeks to test the channels of existing agency problems in private companies. Also, it attempts to find whether a change in ownership structure due to an IPO affects post IPO management decisions.
The this paper is structured as follows: section 2 sets forth a literature survey of the little literature that is available, section 3 describes the model, section 4 reports and describes the data, section 5 report the result, and section 6 concludes all findings and discusses this thesis’ limitations.
2 Existing literature
Pre-IPO ownership has not been the center of research as much as post-IPO ownership, mostly due to the difficulty of obtaining the data. In this section, an overview is given about the little literature that can be found.
2.1 Why firms go public
In her literature research, Roëll (1996) has listed the most important reasons for compa-nies to go public. In descending order of claimed importance, she identified: new fi-nance, improve morale of management and employees, sell of shares of existing share-holders, and exploit mispricing.
New finance. Acquiring new finance is given as the most important reason to go public. A survey by Buckland and Davis (1989) showed that the funds are mostly used for capital investments. However, these new acquired funds are not always used to fi-nance new projects for expansion. Instead, they are often used to refifi-nance current bor-rowings. According to a survey by Ransley (1984), expansion is mostly achieved by acquisition.
In the longer term, issuing shares also affects the ability to raise equity in the future. Firstly, by issuing equity, a firm strengthens its equity base and leverage is re-duced. This mitigates agency problems, such as debt overhang. However, this can also be achieved by placing private equity and, hence, it is not a sufficient explanation on why a firm would go public (Roëll, 1996). Therefore, secondly, Roëll (1996) argues that liquidity in a company’s stock is very much preferred. In the case that a stock is not/little liquid and/or in the hands of large blockholders, the investors will require a premium as they will need to incur transaction costs if they want to sell these shares (and so will their buyers). Therefore, it will probably not be worthwhile to issue public equity and incur the costs if the shares will be allocated to a select large group of inves-tors instead of widely dispersed. In the latter case, shares will be traded more frequently and, therefore, issuing is more cost effective. In conclusion, having sufficient liquidity in the capital market and a strong equity base can be a pre-requisite to raise capital in
ful to raising equity in the future because the share price is an important signal of the company’s value. Moreover, companies have acknowledged that they have a better ne-gotiating position with banks having an enhanced status as a public company (Ransley (1984) in: Roëll (1996)). Moreover, Pagano, Panetta and Zingales (1995a) found that companies were able to raise more debt after they have issued public equity. They have attributed this to the competition between banks and other potential sources of finance.
Improve morale of management and employees. With compensation schemes in which the manager and/or employees receive company shares, firms try to motivate them to perform in the best interest of the company. This is seen as a major advantage of going public (Roëll, 1996).
Sell of shares of existing shareholders. The IPO is often used as a moment of cashing in by the existing shareholders. This is generally de-emphasized in the prospec-tuses, as it could be perceived as a sign that depresses the pricing of the shares. Pagano, Panetta and Zingales (1995b) found that approximately half of the existing shareholders do not sell at all, though they might in the future. They have also found that in the years after the IPO, divestment activities continue.
Exploit mispricing. Ibbotson and Ritter (1995) have mentioned that managers tend to time the IPO in a period in which investors see a window of opportunity for the company. Managers try to take advantage of a positive investor sentiment, in which in-vestors are willing to overpay for the shares.
As the previous are the most important reasons for (or advantages of) an IPO, the following indicate the disadvantages of an IPO. Most notably are the costs of going public and the dilution of control.
Cost of going public. Alavi, Pham and Pham (2008) have indicated two types of costs: direct costs and cost of underpricing. The direct costs include costs such as pro-motion costs and underwriter fees, i.e. costs that have to be suffered to be certain of a successful IPO. The cost of underpricing is a substantial though inevitable cost of going public to attract enough investors.
Dilution of control. Existing shareholders are conscious of the fact that they might lose control after an IPO. It can be argued that when the original own-er(s)/management’s pre-IPO ownership level is high, they might issue more shares than in the case it was low. Then it can be expected that management will choose to issue
fewer shares and will avoid attracting new large shareholders (Alavi, Pham and Pham, 2008).
Another important implication of an IPO is separation of ownership and control, which is described in the paper by Jensen and Meckling (1967). They addressed the agency problem that arises when the ownership (shareholders) and the control (manag-ers) of a firm are separated. This is discussed in the next section.
2.2 The separation of ownership and control
Jensen and Meckling (1967) compared the behavior of a manager that owns 100% of the firm with the behavior of a manager that owns less than 100% of the firm to analyze the effect of outside equity on the agency costs. They analyzed the incentives of the agent and the principle and how the relationship between these parties determines equi-librium.
An owner-manager that owns 100% of the firm will make operational decisions that will maximize its own utility. This utility is achieved by a combination of pecuni-ary and non-pecunipecuni-ary compensation. Non-pecunipecuni-ary compensation can include his physical presence at the office, where he meets his staff, observes the level of staff dis-cipline, his personal relationships with his staff, charitable contributions, and so forth. Important here is the mix that maximizes his utility, i.e. at the point where the marginal utility from spending an additional dollar is equal for each non-pecuniary reward and equal to his after-tax purchasing power.
In the case that the owner-manager owns less that 100% of the company, for in-stance 95%, the other 5% is owned by outside equity holders. The agency problem is a result of the difference in interest between the (owner-)manager and the other equity holders. In this case, the mix between pecuniary and non-pecuniary rewards that max-imizes utility for the (owner-)manager is where the marginal utility from spending an additional dollar is equal for each non-pecuniary reward and equal to an additional 95 cents of his after-tax purchasing power, and not one dollar. Thus, the (owner-) manager does not bear the full the cost of his perquisites.
Na-that are owned by the manager for 100% and those Na-that are not. This allowed the au-thors to compare the two types of company ownership. Their first measure of agency cost is measuring excessive expenses such as the manager’s consumption of perks. The second measure is an efficiency ratio, which is a proxy to the loss in revenues due to inefficient use of company assets and is calculated as the ratio of annual sales to total assets.
One of their findings is that the ratio of operating expenses to sales is higher for firms managed by outsiders than for firms managed by the owners. Measuring the effi-ciency ratio on asset utilization, the authors found that firms managed by the sharehold-ers have a higher ratio than firms managed by outside managsharehold-ers.
Non-managing shareholders can mitigate the agency problem by participating in monitoring the firm’s managers. They are incentivized to do so as this could prevent the managers from destroying firm value. Another type of monitoring is by the debt hold-ers, most commonly the banks. The authors found that external monitoring by banks leads to lower agency costs.
2.3 Mitigating pre-IPO agency problem
It is possible to mitigate the negative effects of the agency problem. The first one, men-tioned in the previous section, is monitoring by the outside share- and debtholders. An-other mechanism is the threat of a takeover.
2.3.1 Monitoring
Jensen and Meckling (1967) analyzed this measure against the agency problem in their paper too. Expending resources to monitor them can alter the possibility for managers to reap non-pecuniary benefits. Examples of monitoring are budget restrictions, auditing, incentive compensation schemes and so forth. Measures that intend to align the interest more closely to the shareholder’s interests.
Jensen and Meckling (1967) also put forth that it does not matter who does the monitoring, i.e. the outside equity holders or the owner-manager himself. It is possible for the owner-manager to expend resources to guarantee the outside equity- and
debtholders that he will limit his perquisite consumption. It is in the interest of the own-er-manager to incur these costs up until the point where the net increase in wealth, be-cause of the decrease in agency costs, is more valuable than the perquisites given up.
A potential problem with monitoring is that shareholders are not incentivized enough to actually monitor management if their share is not large enough. A sharehold-er needs to incur 100% of the monitoring costs, while his return is as small as the amount of shares he owns. If the share is large enough, he will partially internalize the benefits from his efforts to monitor (Pagano & Roëll, 1998).
To overcome this incentive problem, monitoring of management can be done by establishing a board of directors and allowing for large blockholders. There is a variety in how these measures can be implemented, not mutually exclusive, for instance, incen-tivizing the board of directors by stock ownership, having independent board members, and separated chairmen for the board of directors and the executive board. In Zajac and Westphal’s (1994) paper, they find that there are diminishing returns of behavior when increasing monitoring of management. Thus, increasing the amount of monitoring is not always better.
2.3.2 Takeover bids
Another mechanism to mitigate the agency problem is the threat of a takeover. When the shareholders own a large enough stake in the company, they are also more incentiv-ized to place a takeover bid in case management does a poor job in increasing firm val-ue. They benefit from it as they capture a part of the appreciation of the share price due to the improved management (Pagano & Roëll, 1998).
Shleifer and Vishny (1986) found that majority shareholders play an important role in takeovers. They are able to facilitate takeovers by outsiders by sharing the large gains on their own stakes with the bidder. Then it does not matter whether they are able to monitor management themselves.
Therefore, managerial owners tend to prefer issuing share publicly, as he is then able to sell little amount of shares to many potential owners, such that ownership is dis-persed. This reduces the threat of a takeover. If the managerial owners would, instead,
sell the shares privately, they can only sell to a few large shareholders (Pagano & Roeëll, 1998).
2.4 Agency problem in an IPO
This section discusses how the agency problem, i.e. the difference in interest of mana-gerial and non-manamana-gerial shareholders, can manifest itself.
2.4.1 Manifestation of agency problem in IPO
In the case that pre-IPO ownership is dispersed, the agency problem and its associated costs already exist. In particular, in the case where there are non-managing shareholders and managers holding no shares. Ang, Cole and Lin found that an owner-manager is less incentivized to consume perquisites when his level of ownership rises. This is due to his shares that rise in value as firm value increases, while his benefits from perqui-sites stay constant. In the case that an outside manager is managing the firm, then the largest/primary shareholder is the one monitoring management. Finally, agency costs should be higher when outside managers manage the firm and it should also increase with the amount of non-managing shareholders. The latter is discussed before as the shareholder’s stake should be large enough to absorb the costs of monitoring (2000).
The authors also found that pre-IPO managerial ownership influences the level of IPO costs. As mentioned before, they established two measurements of IPO costs, direct costs and underpricing. They found that higher larger pre-IPO management own-ership levels, the more effort is exerted to keep IPO costs as low as possible. They also argue that the level of underpricing is influenced by pre-IPO management ownership. Ljungqvist and Wilhem (2003) argue that managers are merely acting as the agent on behalf of the shareholders when negotiating the IPO price with the underwriters. When management ownership is high, levels of underpricing should be lower than it would be with low management-ownership. Therefore, underpricing is a result of the IPO agency problem.
It is in the best interest of the issuer to keep underpricing as low as possible, as the capital used for expansion after the IPO is reduced by underpricing (Platt, 1995).
The underwriters benefit from underpricing as they can provide their investors a lower price for the shares, which encourages future business between the underwriter and the investors (Pollock, Porac, & Wade, 2004).
2.4.2 Mitigating agency problem in IPO
While research on the existence of an agency problem in an IPO has been conducted, research on mitigating the previously discussed agency problems is very limited to not at all. Mostly, due to the difficulty of gathering data on ownership structure of private firms before they go public (Alavi, Pham, & Pham, 2006).
2.5 Number of shares offered in sight of dilution of control
Alavi, Pham and Pham (2006) found that pre-IPO ownership affects the decisions made in the IPO process. According to the authors, decisions during the process are driven by the interest of the managers. They found that the amount of shares issued is determined by the desire of the pre-IPO owners/owner-managers to keep control after the IPO. The larger their pre-IPO share is, the more they issue at the IPO, because it will be likely that they are able to retain control. In the case that their ownership is small, they are al-so like to issue a large amount of shares because they will not be able to retain control anyway. However, if pre-IPO management ownership is around 60%, it is not likely anymore to issue a large amount of shares as that would probably lead to a loss of con-trol after the IPO.
Next to the amount of shares issued, pre-IPO management ownership also influ-ences the allocation of the shares. In the case that the shares are issued to large share-holders, they might risk diminishing control or that their control is challenged by the blockholders. The authors find that in the case that insiders lose control after the IPO, the shares are allocated such that the emergence of large blockholders is prevented (Alavi, Pham, & Pham, 2006).
3 Model
A multivariable OLS regression is used to assess the explanatory power of efficiency on a firm’s underpricing at its IPO. First, the causal relationship of the level of insiders on the efficiency proxies is assessed, after which these efficiency measurements explain underpricing. All regressions are heteroskedasticity robust.
3.1 The channels of agency costs
Ang, Cole and Lin (2000) found that when management ownership is not 100% in a company, agency costs are present. According to Alavi, Pham and Pham (2006), pre-IPO management ownerships influences the level of pre-IPO costs. They distinguished be-tween direct costs, such as costs of promotion, and indirect costs, such as cost of under-pricing. This model is aimed to investigate through which channels the costs that non-management ownership brings manifest themselves and how that affects the IPO in terms of underpricing.
Following Ang, Cole and Lin (2000), the channels used in this model are two measurements for efficiency, namely the expense ratio (EXPENSE) and the asset utiliza-tion ratio (ASSETUZ). The expense ratio is measured as the ratio of operating expenses and annual sales and the asset utilization ratio is measured as annual sales divided by total assets. The relationship between the level of management owners and the expense ratio is expected to be negative. Excessive expenses due to the consumption of perks and perquisites should be reflected in in these operating expenses, which are measured by total expenses minus the costs of goods sold, interest expenses and managerial com-pensation (Ang, Cole and Lin, 2000). Therefore, the relationship between the fraction of management owners (INSIDER) and the expense ratio is expected to be negative, as his benefits will increase with the increase in firm profits, while his benefits from consump-tion of perquisites would stay the same. Table 8 in Appendix 4 reports a list of all varia-bles and their definitions.
The relationship between the level of owner-managers and the asset utilization ratio is expected to be positive, as this ratio measures how effectively management
de-ploys the firm’s assets. The higher the level of management ownership in a firm, the better the use of assets. Managers might behave in ways that may lead to lower reve-nues due to, for instance, poor investment decisions and not exerting sufficient effort, and might consume perquisites such that the firm purchases inefficient assets, e.g. fancy office space and furniture (Ang, Cole and Lin, 2000).
(1) EXPENSE = α + β1INSIDER + ε
(2) ASSETUZ =α + β1INSIDER + ε
INSIDER is constructed by calculating the fraction of shares of total shares outstanding prior to the IPO that is owned by insiders, with insiders being shareholders who are able to make day-to-day decisions, such as the CEO, the founder(s), other executive direc-tors and senior managers. Non-executive direcdirec-tors are not included (Alavi, Pham and Pham, 2006). In later models, the percentage change of insider ownership after the IPO are accounted for as well to see how that change affects the expense and asset utiliza-tion ratios after the IPO.
The outsiders of the firm, the non-managerial owners, can be categorized in large capital holders (CAPDUMMY) that has a partially controlling interest and holds at least 20% of a company’s shares, which include investors such as venture capitalists (VCDUMMY), angel investors, parent companies and joint-venture partners. It can be ex-pected that even though these large shareholders do not hold an executive position with-in the company, they are able to with-influence the decision-makwith-ing and operatwith-ing process and should therefore be controlled for. The use of a dummy variable is chosen, because it is not possible to simply aggregate the number of shares of insiders and large share-holders, as the sum would be close to one, which would probably result in a multicol-linearity problem (Alavi, Pham, & Pham, 2006). It was argued before that the higher the level of insider ownership, the lower the expense ratio and the higher the asset utiliza-tion ratio should be. Therefore, it can be argued that the relautiliza-tion between the presence of a large capital holder and venture capitalist, and the efficiency ratios should be posi-tive for the expense ratio and negaposi-tive for the asset utilization ratio.
Further, other pre-IPO variables that might affect the firm’s expenses and asset utilization and should, therefore, be controlled for is a firm’s size, as efficiency can be realized in scale economies (Alavi, Pham and Pham, 2006). This is measured by the natural logarithm of pre-IPO market capitalization and calculated by multiplying the number of pre-IPO shares with the offer price (logPRESIZE).
Also the firm’s age is controlled for, as the older the firm becomes, the better it should have established its learning curve and the more efficient it should be (logAGE) (Ang, Cole and Lin, 2000). This is the number of years from the company’s inception to the year of the IPO.
Last to be controlled for, are the company’s debt levels, the industry and year. Mellow and Parsons (1992) found that the levels of debt might influence the operating decisions, such as the decision to abandon a project. Abandoning a project may save the company on operating expenses and preserves inventory for future projects. This is an example of the deadweight costs of financial distress and is measured as the ratio of in-terest bearing debt over total assets (DEBT).Ang, Cole and Lin (2000) found that the in-dustry of the company should affect the expense and asset utilization ratios. This is treated as industry fixed effects (INDUSTRY) and also time fixed effects (YEAR), i.e. a dummy for each two-number SIC code and year.
Incorporating the variables above, the model is extended to the following:
(3) EXPENSE = α + β1INSIDER + β2CAPDUMMY + β3VCDUMMY +β4logPRESIZE
+β5logAGE + β6DEBT + γnINDUSTRYn + δnYEARn + ε
(4) ASSETUZ = α + β1INSIDER + β2CAPDUMMY + β3VCDUMMY +β4logPRESIZE
+β5logAGE + β6DEBT + γnINDUSTRYn + δnYEARn + ε
3.2 Effect agency costs and insiders on underpricing
Following Alavi, Pham and Pham (2006) and with establishing the channels of agency costs in the previous section, it can be argued that both efficiency ratios will negatively impact the IPO. This impact is measured by underpricing, which is inevitable in an IPO.
Underpricing represents the opportunity costs of the issuer because he is not able to of-fer the shares on its fair value and that can be quite costly for the issuer. Not only does the issuer receive less than the shares’ fair value, he also suffers from the cost of dilu-tion. In IPO literature, underpricing is therefore seen as a substantial cost in an IPO (Ljungqvist, 2007). As underpricing is the ratio of the difference between the first-day closing price and offer price, and offer price, MAR<0 would indicate overpricing and MAR>0 underpricing. Therefore, there should be positive relation between the expense ratio and underpricing and a negative relation between asset utilization.
Next to these efficiency ratios, the proportion of insiders should also affect the level of underpricing. Insiders should be incentivized to reduce underpricing of the shares as explained before. Ljungqvist and Wilhem (2003) argued that non-shareholder managers merely act as agents in a negotiation process with the underwriters. The un-derwriters benefit from underpricing as they can offer their investors prices below fair value. Therefore, the level of underpricing can be regarded as a result of the agency problem that existed before an IPO. Therefore, the higher the level of asset utilization, the less underpricing there should be. The larger the expense ratio, the higher the level of underpricing and the larger the share for insiders, the less underpricing there should be. Therefore, there should be a negative relation between the level of insiders and un-derpricing. Moreover, it is interesting to see how the relative change in level of insider ownership affects underpricing (INSIDECG). It can be argued that its relation to under-pricing is negative is the same as explained previously, because the higher the level of insider ownership, the lower the agency costs in terms of underpricing will be. Hence, the bigger the relative increase, the lower underpricing should be. However, it can also be argued that when the level of insider ownership is already high prior to the IPO, cur-rent insider shareholders are looking to decrease their ownership via selling their shares at time of the IPO. This argument is discussed before in section 2.1.
Underpricing (MAR) is calculated as the difference between the offer price (Poffer) and the first day closing price (Pclose) divided by the offer price (Poffer).
Further, variables that need to be controlled for as they might affect underpric-ing are the firm’s pre-IPO size (logPRESIZE), the firm’s age (logAGE), debt levels (DEBT), and its book-to-market ratio (BTM). First day market value is used for the
book-plained as a result of information asymmetry and uncertainty (Alavi, Pham and Pham, 2006; Booth and Chuba, 1996; Rock, 1986). The older and larger the firm, the more well known they are. Firms with high book-to-market ratios are generally considered as having less uncertainty and firms with debt should be more transparent due to existing relationships with banks. Their relations to underpricing are, therefore, expected to be negative. Again, industry and time fixed effects are controlled for.
(5) MAR = α + β1EXPENSE + β2INSIDER + β3INSIDECG + β4CAPDUMMY +
β5VCDUMMY + β6logPRESIZE +β7logAGE + β8DEBT + β9BTM +
γnINDUSTRYn + δnYEARn +ε
(6) MAR = α + β1ASSETUZ + β2INSIDER + β3INSIDECG + β4CAPDUMMY +
β5VCDUMMY + β6logPRESIZE +β7logAGE + β8DEBT + β9BTM +
γnINDUSTRYn + δnYEARn + ε
3.3 Retainment or dilution of control
The previous models are mostly focused on relations prior to an IPO and the proxies for the agency costs. It is, however, also interesting to analyze how many shares manage-ment-owners decide to offer, as they risk losing control after the IPO.
Alavi, Pham and Pham expected and found that companies with very high or very low levels of pre-IPO managerial shareholders offer more shares at the IPO than companies with middle levels of pre-IPO management ownership. As insider ownership is at such high levels, the owners are able to sell a large amount of shares without the threat of diluting control. The same goes for the low levels of insider ownership, the owners will lose control due to the IPO regardless of the amount of shares sold. Com-panies with middle levels of insider ownership are expected to sell a little amount of shares as they do risk losing control (assuming that this is not desired) (2008).
To measure the extent to which the shares are offered to the public, the authors divide the total amount of shares that is fully subscribed by new investors by the total amount of shares at the end of the IPO to calculate . This particular data
was not available for this research, therefore, for this research, PUBLICSHARES is calcu-lated by dividing the amount of shares that is available for trading by the public (i.e. not to be bought by existing shareholders) at the IPO by the total amount offered and sold including any overallotment.
Further, INSIDER is the level of pre-IPO insider ownership, as it is in the previous models. The variable INSIDER is expected to have a positive sign for levels of low pre-IPO ownership and high pre-pre-IPO ownership and negative for levels of pre-pre-IPO owner-ship that are somewhat in the middle. Therefore, INSIDER2 is added to test the existence of a non-monotonic relationship, i.e. to see whether the same relationship holds for all level classes of insider ownership (small, middle and large levels of insider ownership). These regressions are run separately.
Next, the presence of large capital holders (CAPDUMMY) and venture capitalists (VCDUMMY) is accounted for as they may have a significant influence in the decision making process of the IPO. CAPDUMMY is equal to 1 if a large capital holder is present and 0 otherwise, and VCDUMMY is equal to 1 if a venture capitalist is present and 0 oth-erwise. Both are expected to have a negative relation with the amount of shares availa-ble for public buys as they risk a dilution of control. Other firm characteristics are ac-counted for as they might influence the amount of capital to be raised. The firm’s pre-IPO size is accounted for, as generally larger firms (firms with larger market valuation) are probably able to raise the same amount of capital with a lesser amount of shares. Also, they are likely to have more internal capital, which reduces the requirements for external capital (logPRESIZE). The firm’s age is also accounted for. Even though young-er companies may have highyoung-er capital requirements, it can be expected that they are not able to raise such large amounts of capital due to their short existence (logAGE). The levels of interest bearing debt, calculated by dividing interest baring by total assets (DEBT) are also accounted for, as the proceeds of an IPO are often used to reverse the accumulation of debt by repaying their borrowings (Pagano, Panetta & Zingales, 1995a).
(7) PUBLICSHARES = α + β1INSIDER + β2CAPDUMMY + β3VCDUMMY +β4logPRESIZE +
(8) PUBLICSHARES = α + β1INSIDER + β2INSIDER2+β3CAPDUMMY + β4VCDUMMY +
β5logPRESIZE +β6logAGE + β7DEBT + ε
Furthermore, it is interesting to analyze to what extent the wish to keep control by the insider shareholder bears a cost. While using the same variables as the previous models, in the following model, underpricing is a result of management that tries to retain con-trol after an IPO, rather than the cost of the agency problem. It can be expected that management owners use control retention to signal higher firm quality, which was sug-gested by Leland and Pyle (1977).
(9) MAR = α + β1PUBLICSHARES + β2INSIDER + β3INSIDECG +β4CAPDUMMY +
β5VCDUMMY +β6logPRESIZE +β7logAGE + β8DEBT + β9BTM +
γnINDUSTRYn + δnYEARn + ε
3.4 Effect of retention in post-IPO managerial decisions
As it is previously argued, a higher level of insider ownership would result in a better use of assets and a more efficient policy in operational expenses, it is interesting to see whether this can be expected after the IPO as well. The question here is whether com-panies where the insider shareholders give up control after the IPO behave differently from companies where the insider shareholders retained control.
To analyze this question, models (3) and (4) are used again. However, instead of using the level of insider ownership prior to an IPO, a dummy variable is used that is equal to 1 if the managers did have control prior to the IPO and gave it up afterwards, and is equal to 0 otherwise (GIVEUPDUMMY), with giving up control defined as having less than 20% ownership after the IPO (see section 3.1). Also the presence of post-IPO large capital holders is accounted for (CAPDUMMYPOST) as well as the presence of ven-ture capitalists (VCDUMMYPOST) as they may have significant influence on managerial decisions. The percentage of insider ownership after the IPO (INSIDERPOST) and the per-centage change in insider ownership (INSIDECG) are controlled for. The company’s post-IPO size, the market value, is also accounted for as the post-IPO capital proceeds can be used
for economies of scale, which might affect the firm’s efficiency ratios (POSTSIZE). Last-ly, the company’s age and debt levels are accounted for, as with time companies are able to establish a learning curve to act more efficiently (logAGE) and the proceeds from an IPO are often used to repay debt (DEBTPOST).
Going back to the argument that a higher level of insider shareholders should lead to more efficient expense policies and use of assets, it can expected that companies where the management owners gave up control, have a higher expense ratio and a lower asset utilization ratio. Therefore, the coefficients are expected to be positive for the former and negative for the latter.
(10) EXPENSEPOST = α + β1GIVEUPDUMMY + β2INSIDERPOST +β3INSIDECG +
β4CAPDUMMYPOST + β5VCDUMMYPOST + β6logPOSTSIZE +
β7logAGE + β8DEBTPOST + γnINDUSTRYn + δnYEARn + ε
(11) ASSETUZPOST = α + β1GIVEUPDUMMY + β2INSIDERPOST +β3INSIDECG +
β4CAPDUMMYPOST + β5VCDUMMYPOST + β6logPOSTSIZE +
β7logAGE + β8DEBTPOST + γnINDUSTRYn + δnYEARn + ε
3.5 Effect of retention in post-IPO performance
Lastly, next to test the expense and asset utilization ratios after the IPO, it is also inter-esting to measure the performance after the IPO. To measure this, the return after 180 days (180DAYS) and 365 days (1YEAR) are used as the dependent variables. Models (5) and (6), in which the first day return is the dependent variable, then become:
(12) 180DAYS = α + β1EXPENSE + β2INSIDERPOST + β3INSIDECG +β4CAPDUMMY +
β5VCDUMMY +β6logPOSTSIZE +β7logAGE + β8DEBTPOST +
(13) 180DAYS = α + β1ASSETUZ + β2INSIDERPOST + β3INSIDECG +
β4CAPDUMMYPOST + β5VCDUMMYPOST +β6logPOSTSIZE +
β7logAGE + β8DEBTPOST + β9BTM + γnINDUSTRYn + δnYEARn + ε
(14) 1YEAR = α + β1EXPENSE + β2INSIDERPOST + β3INSIDECG +β4CAPDUMMY +
β5VCDUMMY +β6logPOSTSIZE +β7logAGE + β8DEBTPOST + β9BTM
+ γnINDUSTRYn + δnYEARn + ε
(15) 1YEAR = α + β1ASSETUZ + β2INSIDERPOST + β3INSIDECG +β4CAPDUMMY +
β5VCDUMMY +β6logPOSTSIZE +β7logAGE + β8DEBTPOST + β9BTM
4 Dataset
4.1 S-1 SEC filings
As mentioned before, there is a lack of information on pre-IPO ownership structure in readily available databases. Alavi, Pham and Pham (2006) used a dataset comprised of Australian firms that underwent an IPO. The Australian data is available, as Australian firms are required to submit their pre-IPO ownership structure to the Australian Securi-ties and Investment Commission (ASIC) as well as their post-IPO ownership structure to the Australian Stock Exchange (ASX).
This research focuses on IPOs in the United States of America, where the pre-IPO ownership structure is reported in a company’s S-1 filing to the SEC. These filings are available online via the EDGAR1 database. Though, these reports give a precise di-vision of the pre- and post-IPO ownership structure, there are some pitfalls that need to be kept in mind. First of all, the data needs to be gathered manually, from each firm’s prospectus. Though, the information copied from these filing were done in a very care-ful manner, it cannot be guaranteed that no mistakes were made. Secondly, there is a possibility that shareholdings are counted double. For example, in the case when one of the non-executive directors on the board represents a venture capital fund, the same lev-el of rlev-elevant ownership may be reported in the prospectus. To keep this problem to a minimum, shares held by external shareholders that are affiliated with someone who is a director or an executive officer, are excluded from the total of outsider shareholder. Thirdly, the proceeds of the IPO may be used to acquire other business owned by the owners/managers. This is an example of a related-party transaction that occurs at the same time as the IPO or is dependent on the IPO. This would result in an understate-ment of pre-IPO ownership concentration. However, there is no way to deal with this problem, which is in the scope of this research. Lastly, what has been reported in the prospectuses is still subject to change on management’s discretion and the actual num-bers might be different from the numnum-bers in the prospectuses.
4.2 Post-IPO accounting and stock price data
Post-IPO accounting data, such as balance sheet and operations data, are acquired via the CRSP-Compustat Merged (CCM) database. Gathering SEC filings into EDGAR were phased over three years and ended in 1996. Therefore, all data of US based com-panies conducting an IPO from 1996 onwards were downloaded from the CCM data-base. This resulted in 17,927 companies. A random sample resulted in the use of 157 companies for this research. Data to compute underpricing, i.e. data on offer price and price at the end of the IPO day, are gathered via Thomson ONE. Also, data on the 180-day and 365-180-day return after the IPO are gathered via Thomson One. The database did not provide the return after more than 365 days.
4.3 Descriptive statistics
4.3.1 Descriptive statistics pre-IPO agency costs
Table 1 reports the descriptive statistics of the data sample for models (1) through (6). On average, pre-IPO insider ownership is a little more than half of the outstanding shares. This decreases after an IPO to a little more than a third of the outstanding shares. This means that on average, management owners give in on the percentage of owner-ship after an IPO.
Moreover, both efficiency ratios decrease, on average, after the IPO. This can be expected in case of the expense ratio, as this would mean that a decrease of manage-ment ownership is accompanied by a more efficient policy on spending. However, it can be expected that, on average, the utilization of assets should be more efficient in case management ownership decreases, which would result in a higher ratio. This, how-ever, is not the case.
The average level of ownership by large capital holders, just as insider owner-ship, decreases after an IPO. This might give in to the notion that many existing share-holders use an IPO to sell their shares, discussed before in paragraph 2.1.
Table 1
Descriptive statistics of data for models (1) through (6)
This table reports the descriptive statistics of the dependent as well as independent variables, both pre-IPO and post-IPO. This makes comparing of pre- and post-IPO data easier. The number of firms in the sample is depicted under #IPOs.
EX-PENSE and ASSETUZ are the efficiency ratios. MAR is underpricing calculated as the first day return. INSIDER is the
propor-tion of pre-IPO insider ownership and INSIDERCG is the relative change of this proportion after the IPO. The variable AGE is the number of years from the firm’s inception until the IPO. For a more complete picture, the percentages of large capital holders (LARGECAP) and venture capitalists (VC) are included as well as the dummy variable. DEBT is the ratio of interest
bearing debt and total assets. BTM is the book to market ratio. Pre-IPO
variable #IPOs mean std. dev. min max
EXPENSE 153 5.252 27.935 0 326.6 ASSETUZ 157 1.484 2.544 0 25.631 MAR 157 0.329 0.923 -0.646 5.25 INSIDER 157 0.554 0.307 0 1 INSIDECG 155 -0.013 0.364 -1 1.976 CAPDUMMY 157 0.306 0.462 0 1 LARGECAP 48 0.413 0.198 0.190 0.981 VCDUMMY 157 0.121 0.327 0 1 VC 19 0.168 0.106 0.049 0.443 PRESIZE* 157 28,261 562,841 0 2,529,820 logPRESIZE 156 18.613 1.330 14.128 21.651 AGE** 157 22.1 28.345 1 154 DEBT 157 0.336 0.865 0 9.239 BTM 61 0.031 0.023 0.005 0.123 Post-IPO EXPENSE 140 1.996 4.554 0 46.379 ASSETUZ 141 0.839 0.764 0 3.412 INSIDER 141 0.379 0.227 0 0.934 LARGECAP 34 0.353 0.156 0.207 0.868 VC 15 0.132 0.086 0.044 0.327 * X $1,000 ** in number of years
4.3.2 Descriptive statistics retention and post-IPO
Table 2 reports the descriptive statistics for models (7) through (9). Due to missing val-ues from the dataset, the portion of shares that is available for public buys at the IPO can only be calculated for only 86 companies out of the 157. Also, as PUBLICSHARE is calculated as the amount of shares available for public of the total shares offered and sold overallotment, it is not possible to display a number that is higher than 1. This is probably a wrong entry in the database. Therefore, PUBLICSHARE is winsorized at 5%.
The issue with having only 86 companies becomes evident when analyzing the portion that is offered for public buys, when the sample is split based on ranges of pre-IPO management ownership. Table 3 reports these classes of sample splits. There are no more than 21 observations in each class, which makes a comparison between the classes not representative. Nevertheless, the mean of all ranges, except for the 20%-40% range, are between 0.7 and 0.9, which means that the average amount of shares that is
availa-ble for the public to buy are between 70 and 90 percent. This also counts for the first range of insiders who own the least percentage of shares, and as expected, they allow for large amount of shares to be sold to public, as they will not be in control anyway. These results are more or less in line with the results by Alavi, Pham and Pham (2008).
Table 2
Descriptive statistics of data for models (7) through (9)
This table reports the descriptive statistics of the dependent as well as the independent variables. The number of firms in the sample is depicted under #IPOs. PUBLICSHARE is the portion of shares available for public buys of the total amount of shares offered. MAR
is the return on the first day. INSIDER is the level of pre-IPO insider ownership relative to the total pre-IPO amount of shares out-standing. To test for the existence of a non-monotonic relation with PUBLICSHARE, INSIDER2 is added. INSIDECG is the relative change of level of insider ownership after the IPO. The variable AGE is the number of years from the firm’s inception to the IPO. Both PUBLICSHARE winsorized and not-winsorized are included. This variable is at the date of the IPO. As it can be seen, the outlier
of 2.426016 does not follow from dividing the portion of shares eligible for public buys by total shares offered and overallotment sold. A large capital holder is present in 48 firms and a venture capitalist in 19.
Pre-IPO
variable #IPOs mean std. dev. min max
PUBLICSHARE 86 0.710 0.523 0 2.426 PUBLICSHARE* 86 0.678 0.463 0 1 MAR 157 0.329 0.823 -1 5.25 INSIDER 157 0.554 0.307 0 1 INSIDER2 157 0.400 0.324 0 1 INSIDECG 155 -0.014 0.364 -1 1.976 CAPDUMMY 157 0.306 0.462 0 1 LARGECAP 48 0.413 0.198 0.190 0.981 VCDUMMY 157 0.121 0.327 0 1 VC 19 0.168 0.106 0.049 0.443 SIZE*** 157 28,261 562,841 0 2,529,820 logSIZE 156 18.613 1.330 14.12779 21.651 AGE** 157 22.1 28.345 1 154 DEBT 157 0.336 0.865 0 9.239 BTM 61 0.031 0.023 0.005359 0.123
* winsorized at 5%, above only ** in number of years
*** x $1,000
Table 3
Descriptive statistics classes of insider ownership
Descriptive statistics of the sample, divided into classes based on percentages of pre-IPO managerial ownership. Total number of firms is 86. Classes of 40%-60% and 60%-80% are both winsorized on a 5% level on the high end, due to outliers. Ang, Cole and Lin (2000) found that the proportion of total shares offered that is available for public buys is high when there was a very low level
or very high level of pre-IPO insider ownership. All classes show that the average proportion of offered shares that is available for public buys is around 70%-80%, except for the class in which pre-IPO insider ownership is between 20% and 40%. This is more or
less in line with the findings by Ang, Cole and Lin (2000).
Proportions of offered shares including overallotment that is offered to the public Sample split according to
pre-IPO ownership #IPOs Mean Std. Dev. Min Max
<20% 18 0.7 0.4701623 0 1 20%-40% 11 0.4545455 0.522233 0 1 40%-60% 21 0.8168329 0.5075859 0 2.178461 40%-60%* 21 0.7655955 0.4149093 0 1 60%-80% 19 0.7457439 0.629478 0 2.426016 60%-80%* 19 0.6835507 0.4792561 0 1.057765 80%-100% 17 0.8840881 0.323407 0 1.053019
Table 4 depicts the descriptive statistics for models (10) through (15), which test on managerial decisions and measure performance after the IPO.
Compared to the average of the pre-IPO expense and asset utilization ratios, both ratios drop after the IPO on average. This means that on average, the firms have a more efficient expense policy on the one hand, but they utilize their assets very ineffi-ciently on the other. In terms of performance, the average return after a year is higher than the average return after 180 days. On average, the change in pre-IPO ownership levels is negative, which means that on average, the level of insider ownership declines. This can also be noticed from difference between the level of pre-IPO insider share-holders, which is 55.4%, and the level of post-IPO insider shareshare-holders, which is 37.8%. In 34 firms, a large capital holder is present and in only 15 firms, a venture capi-talists is present in 15 firms, which is less than the 48 firms in which a large capital holder is present and 19 firms in which a venture capitalist is present, prior to the IPO. On average, their portions of ownership have also reduced compared to the levels prior to the IPO.
Table 4
Descriptive statistics for data models (10) through (15).
This table depicts the descriptive statistics for the dependent and independent variables. The number of firms in the sample is depicted under #IPOs. EXPENSEPOST and ASSETUZPOST are the expense and asset utilization ratios after the
IPO. 180DAYS and 1YEAR are the returns after 180 days and 365 days since the IPO, respectively. GIVEUP is a dummy
variable that is equal to 1 if insiders did have control prior to the IPO but do not have control afterwards and equal to zero otherwise. INSIDERPOST is the level of insider ownership after the IPO and INSIDERCG is the relative change of the shares owned by insiders after the IPO. LARGECAPPOST and CAPDUMMYPOST are dummy variables that are equal to 1 if
a large capital holder or venture capitalist is present after the IPO, respectively, and equal to 0 otherwise.
variable #IPOs mean std. dev. min. max.
EXPENSEPOST 139 2.003 4.569 0 46.379 ASSETUZPOST 140 0.831 0.762 0 3.412 180DAYS 139 0.093 0.290 -0.297 1.477 1YEAR 139 0.159 0.445 -0.047 2.546 GIVEUP 141 0.064 0.246 0 1 INSIDERPOST 141 0.378 0.227 0 0.934 INSIDECG 138 -0.015 0.386 -1 1.976 LARGECAPPOST 34 0.353 0.156 0.207 0.868 CAPDUMMYPOST 140 0.243 0.430 0 1 VCPOST 15 0.132 0.086 0.044 0.327 VCDUMMYPOST 140 0.107 0.310 0 1 POSTSIZE* 137 396,075 483,071 0 2,529,820 AGE** 141 17.4 22.711 1 154 DEBTPOST 141 0.219 0.300 0 2.343 BTM 60 0.032 0.024 0.005 0.123 * x $1,000 ** in number of years
5. Results
The results of the regressions are described in this section. It is divided into sub-sections per set of regressions as depicted in paragraph 3.
5.1 Pre-IPO insider ownership as single regressor
The results of models (1) and (2) are depicted in table 5. The relationship between in-sider ownership and the expense ratio, as well as the relationship between inin-sider own-ership and the asset utilization ratio are as expected, namely negatively and positively, respectively. However, this relationship is only significant for the relationship between insider ownership and the asset utilization ratio at a 5% level.
These results are in line with the expectations set forth in paragraph 4. This would indicate that firms with managers who own company shares are managed more efficiently than firms that are not. Note that neither firm fixed effects nor year fixed ef-fects are incorporated. If year and industry fixed efef-fects are incorporated, the results would not differ2.
5.2 Pre-IPO insider ownership in a multivariable regression
When controlled for the other variables, the relationship between insider ownership and each ratio does not change. However, neither of them is statistically significant. A pos-sible implication is that the agency problem due to pre-IPO ownership structure might not be as significant as Ang, Cole and Lin (2000) argued. Also, as their sample consist-ed of purely small firms, their proxy for the agency problem might not be as significant when large companies are included in the sample.
The presence of a large external capital holder is not significant, while the pres-ence of a venture capital fund is only significant at a 10% level with the expense ratio as the dependent variable. All coefficients do have the predicted sign. The influence that large capital holders and venture capitalists have on daily management decision is (in-deed) very little.
Pre-IPO firm size is not significant for neither the expense ratio nor the asset utilization ratio. Moreover, the coefficients do not have the predicted sign. It was argued that the larger the company, efficiency can be realized in economies of scale, hence it was expected that the relation between size and the expense ratio is negative and be-tween size and the asset utilization ratio positive. These results indicate that that is not the case. This could imply the larger the firm, the less efficient their expense and asset utilization policies are.
It was expected that the older a firms is, the better it should have established its learning curve and, therefore, the more efficient it should be. It is expected that the rela-tion between firm age and the expense ratio is negative and the relarela-tion between age and asset utilization positive. The results are in line with the later, though not signifi-cant, while the results are not in line with the former expectations. The results indicate that the relation between the firm age and the expense ratio are positive. Moreover, it is significant at a 5% level.
Lastly, debt levels should incentivize firms to act more efficiently. Therefore, it was expected that the relation between debt and the expense ratio is negative and posi-tive between debt levels and the asset utilization ratio. The results are in line with the former, but not with the later. However, neither coefficients are significant.
5.3 Efficiency ratios and underpricing
Both efficiency ratios are negatively related to underpricing. This was expected for the relation between the asset utilization ratio and underpricing, however, it was not for the relation between the expense ratio and underpricing.
A negative but statistically significant (at a 1% level) relation between the asset utilization ratio, might be due the use of the proceeds to finance new/existing projects and acquisitions right after the IPO. This would result in a significant increase in reve-nues and the use of assets. The negative relation between the expense ratio and under-pricing is not in line with expectations. It was expected that the expense ratio would have a positive impact on underpricing, i.e. a higher expense ratio would mean a higher level of underpricing. A possible explanation is that the expense ratio serves as an
indi-urement of excessive spending by management (Habib & Ljunqvist, 2003). This varia-ble is, however, not statistically significant.
Also, the coefficient for insider ownership is positive, which was not expected, and significant at every level. This does not support what is previously stated that insid-ers are incentivized to keep underpricing to a minimum. In fact, according to these re-sults, a higher level of insider ownership, would lead to more underpricing. A possible explanation is that insiders allow for more aggressive underpricing, because they intend to sell their existing shares. The relation between the percentage change in insider own-ership and underpricing is negative. This, on the other hand, is in line with expectations.
The results indicate that large external capital holders are negatively related to underpricing, which is also not as expected. This is not in line with the finding that large external capital holders are more interested in selling their shares than receiving a fair price (Ljungqvist & Wilhelm, 2003), nor with the findings by Alavi, Pham and Pham (2008) who found a negative relation (though it was not significant). The results for the presence of a venture capitalist are as expected and are significant at a 5% level.
In regards to pre-IPO firm size, the coefficient in both models is positive, which indicates that the larger and older the firm, the more underpricing there will be. Only the coefficient for firm age is significant. The coefficient for interest bearing debt is negative in both instances, indicating that an increase in debt levels in proportion to as-sets would mean a decrease in underpricing as well as they are more transparent due to their relationship with banks. These results are all statistically significant. The latter re-sults are in line with expectations, but the rere-sults on firm age are not.
5.4 Pre-IPO managerial ownership and the shares that go to the public
The results of models (7), (8) and (9) are reported in table 6. Firstly, models (7) and (8) are both aimed at explaining the proportion of shares available for public buys. They are both the same, except for the added squared variable of pre-IPO ownership in model (8). This quadratic variable is added to test the existence of a non-monotonic relation.
Regressing model (7) results in the pre-IPO insider ownership percentage and the presence of large capital holders to be significant at a 1% and 10% level, respective-ly. Both variables have the predicted sign as well. These variables also have the
predict-ed sign in model (8), however, the quadratic percentage of pre-IPO insider ownership is not significant. This indicates that there is no non-monotonic relation, which could have been expected due to the different retainment motives for various ranges of insider ownership. This is probably due to the lack of data on which proportion of offered shares is available for public buys. Nevertheless, the results indicate that the relation between pre-IPO insider ownership with the amount of shares offered to the public is positive. That is as expected, as manager owners that already own a large portion of the company, are able to offer a large amount of shares without risking control dilution.
Thus, there is a statistically significant relation between the levels of insider ownership and the amount of shares available for public buys. The higher the level of management ownership, the more shares will be available. As to the presence of large capital holders, the relation with the amount of shares available for public buys is nega-tive. Therefore, the firms where a large capital holder is present are like to offer 16% less shares than firms where such a capital holder is not present.
As for the other control variables, the IPO firm size does not show the pre-dicted sign. It is argued before that the larger the firm, the easier it probably is for the firm to raise the same amount of capital with fewer shares and they probably have larg-er intlarg-ernal capital. These results indicate that this is not the case. The same goes for the firm’s age. It was predicted that the older the firm is, the more shares it would be able to offer. The result here, however, indicates that the relation between firm age and the amount of shares offered that is available for public buys is negative.The level of inter-est bearing debt, on the other hand, does show the predicted sign and indicates that in-deed one of the reasons to go public is to repay existing debt. However, the results are not significant.
Further, the effect of the desire to retain control is tested in model (9), whether this will cost the manager shareholders in terms of underpricing. The proportion of shares variable does have the predicted sign, this indicates that control retention is used to signal a higher value for the firm and is statistically significant on every level. The percentage of insider ownership is also statistically significant, however it is not in line with expectations. The result indicates, as it did in models (5) and (6), that manager owners are not incentivized to keep underpricing to a minimum.
Moreover, the coefficient for book-to-market is negative and statistically signifi-cant at the highest level. This indicates that high book-to-market ratios reduce the asymmetric information problem that results in uncertainty in IPOs.
In regards to the other variables, the sign of the presence of large capital holders coefficient, indicates that there is no significant relation between large capital holders and underpricing, as was previously suggested. However, it is not statistically signifi-cant. The relation between underpricing and the presence of a venture capitalist is sig-nificant at a 5% level and shows the predicted sign. Also, the coefficient for a firm’s age is not statistically significant, though it does indicate the relation between firm age and underpricing is positive.
5.5 Post-IPO efficiency ratios
The results of models (10) and (11) are depicted in the first two columns of table 7. The coefficients for the retain-control dummy have the exact opposite sign than what was expected. This would mean that companies where the insider shareholders gave up con-trol after the IPO would have a more efficient expense policy than companies where the insider shareholders did not give up control. This is not in line with the research con-ducted by Ang, Cole and Lin (2000). A possible explanation is that when the public shareholders hold more power over the managers, the higher the firm value would be. This is one of the corporate governance arguments that are a result of the agency prob-lem due to the separation of ownership and control (La Porta, et al., 2000). Also, the relation is only significant at the 5% level with the expense ratio and not with the asset utilization ratio.
Further, for both the expense ratio and asset utilization ratio, the signs of the percentage of insider ownership are positive. This is the same as the coefficients in the pre-IPO model for the asset utilization ratio, but not the expense ratio. The sign in tion to asset utilization coincides with what was expected, however it does not in rela-tion to the expense ratio. The positive sign can be attributed to an increase of the agency problem (excessive spending by managers), as the separation between ownership and control has increased after the IPO. It is, however, not statistically significant. The
in-fluence of post-IPO insider ownership on the asset utilization ratio is significant at the highest level.
The percentage change of insider ownership after the IPO is negatively related to the post-IPO expense ratio. This means that a decrease in insider ownership leads to an increase of the expense ratio. This result is in line with the argument that the higher the level of insider ownership, the more efficient the firm should be. It is significant at a 5% level. The relation between this variable and the post-IPO asset utilization ratio is positive, which is also in line with expectations. However, it is not statistically signifi-cant.
Concerning the other variables, the presence of large capital holders has no sig-nificant influence on the expense ratio, but does have the expected sign. From these re-sults, it appears that firms where large capital holders are present after the IPO have a higher expense ratio and a less efficient utilization of assets than firms where they are not. Their influence on the latter is significant at every level.
Also, firm age has, apparently, a significant influence on the asset utilization ra-tio. It can be expected that as a company progresses over the years, it develops an in-creasing learning curve. Therefore, it should have built the knowledge to utilize its as-sets in a more efficient manner. On the other hand, it does not have a significant influ-ence on the expense policy. The coefficient does have the predicted sign. The level of interest bearing debt does not influence the efficiency ratios as expected, nor is it signif-icant.
5.6 Post-IPO performance
Post-IPO performance is proxied by the 180-day and 1-year returns. The results of models (12) through (15) are also depicted in table 7. These models follow the same reasoning as models (5) and (6) in which underpricing is the dependent variable. The difference is that in models (12) through (15) are over a time span of six months and one year after the IPO instead of one day at the IPO.
The results indicate that both efficiency ratios are positively related to the 180-day return. Which was only expected for the post-IPO asset utilization ratio. In contrast,
pected for the post-IPO expense ratio. However, the results indicate that this relation is far from significant.
The level of insider ownership is positively related to both the 180-day and 1-year returns and statistically significant at the highest level for the 180-day return. This is in line with the findings by Kim, Lee and Francis that security returns might be ex-plained by insider ownership (1988). They found that companies with high levels of in-sider ownership tend to outperform companies with low levels of inin-sider ownership. In regards to the 1-year return, not the level of insider ownership is significant, rather the percentage change in insider ownership is significant, while it is not in regards to the 180-day return. This might be due to the period of six months after the IPO where the after effects of a change in the levels of insider and outsider ownership are present. A period of one year appears to be long enough for the effects of the IPO to wear out.
The post-IPO presence of a large capital holder is only significant in model (12), while the presence of a venture capitalist is significant in models (12), (14) and (15). Also the relation between the presence of a large capital holder and the 180-day return is negative, while it is positive in relation to the 1-year return. The relation between the presence of a venture capitalist and the 180-day and 1-year return is negative in all in-stances.
Firm post-IPO size is positively related to both the 180-day return and 1-year return. This indicates that the larger the firm, the higher the return is. A possible expla-nation is the presence of economies of scale. Firm age is significantly related to the 180-day and 1-year return, however the signs differ. Firm age is negatively related to the 180-day return, indicating that the older the firm is, the lower the return will be after 180 days after the IPO. In contrast, the older the firm, the higher the return will be after one year.
The ratio of interest bearing debt is only significant in the 1-year return model. Moreover, it is positively related to the 1-year return, but negatively related to the 180-day return. This means that the higher the levels of debt, the lower the return will be after the first 180 days after the IPO. However, after a year, the relation between the levels of debt and return is positive. This might indeed confirm that firms with high lev-els of debt, are more transparent due to their relationship with banks. Lastly, the book-to-market ratio is positive and not significant in all models.
6 Conclusion and limitations
This thesis aimed to identify the channels of agency costs that already exist in private companies before they go public. It tried to contribute to the little empirical research that has been done on the relation between insider ownership structures and financial prior to an IPO.
Using efficiency ratios of the level of expense relative to revenues and of sales relative to assets as proxies for agency costs, the results indicate insider ownership does significantly affect the utilization of assets, but not the level of expenses. Controlled for other large capital holders and firm characteristics, these variables indicate the same relation, however, not significantly. When these proxies for agency costs are put into relation with the cost of underpricing (another proxy for the agency problem), the ex-pense ratio is not a significant proxy, unlike the asset utilization ratio. Moreover, the level of pre-IPO insider ownership shows a completely significant and unexpected rela-tion to underpricing. The explanarela-tion for this became clearer in models (12) through (15) as the way underpricing is calculated is essentially the first day return, just as the 180-day and 1-year returns are calculated. From here, it can be concluded that the high-er the level of insidhigh-er ownhigh-ership, the highhigh-er the return. Only, on the first day of trading, a positive return is considered underpricing. From this research, it possibly follows that underpricing might not be the best measure of the agency problem in an IPO.
The results of how many shares are made available to public buys at the IPO, are in line with previous results by Alavi, Pham and Pham (2008). The higher the level of management ownership prior to the IPO, the more shares they are willing and able to make available for public buys, as they do not risk losing control anyway. However, a non-monotonic relation between the different ranges of management ownership was expected, but was not found. This is probably due to the little amount of observations that was available. Lastly, the effects of dilution of control after an IPO on the agency costs proxies appear to be little existent. It is only significant for the effect on the ex-pense ratio.
An important limitation to this study was the difficulty of finding existing litera-ture and results, and obtaining data. While it can be seen as a contributing challenge, the
gathered carefully and to the utmost precision, a mistake-free and fully consistent da-taset cannot be guaranteed. This resulted that in several models, only so little results were deemed significant.
It is clear that this area of studies has a lot of uninvestigated territories. While most research is interested in post-IPO agency relations, these relations are present in private firms as well. This area of research would be so much better accessible and, therefore, probably popular if data would be more easily available. Requirements such as in Australia, namely submitting pre-IPO ownership structures and financial data prior to an IPO, specifically, would make data more available and precise. Also, according to these results, the proxies for pre-IPO agency costs are not as significant as desired. Re-search to what proxies would be better, would also contribute to finding agency costs in private firms.
