Underpricing of privatised IPOs: the Australian experience.

Abstract:

We investigate the price performance of initial public offerings (IPOs) of formerly state-owned companies in Australia. On average, privatised IPOs in Australia are underpriced by about 11%, which is not significantly different from the magnitude of underpricing of the privately-owned IPOs and that of privatised share offerings in other OECD countries. While it appears that the magnitude of underpricing is correlated with the size of issues and the party affiliation of the presiding government, cross-sectional analysis of our sample does not support some traditional theoretical explanations of underpricing for privatised IPOs--only the market sentiment prior to the listing is identified as a significant factor in determining the degree of underpricing. We also investigate the impact of a two-tiered pricing structure on the performance of IPOs.

Keywords:

INITIAL PUBLIC OFFERINGS; PRIVATISATION.

1. Introduction

There is evidence that government-owned firms are less efficient than comparable private-sector firms (Boardman & Vining 1989). Significant increases in profitability and operating efficiency for companies that were privatised through public share offerings have been documented (D'Souza & Megginson 1999). To reduce government deficits and improve economic efficiency, a vast transfer of state-owned assets to the private sector has occurred in the past twenty years in both developed and developing countries. Australia has been no exception to this trend.

Since the late 1980s, both Australian Commonwealth and State Governments have privatised a significant portion of the public sector as a response to fiscal pressure and as a part of continuing process of economic reform. Although most of privatisation takes place in the form of trade sale, some of the most established publicly owned businesses have been sold through public share offerings. (1) Subsequent to the large gains in share prices of several major companies such as Telstra and Commonwealth Bank of Australia, some newspaper commentators have suggested that these formerly state-owned companies have been sold too cheaply to investors at a cost to the public. They have also suggested that governments may significantly underprice such sales solely to please voters for political gains. (2) Do the critics have valid reasons to believe that political factors influence offer prices of privatised IPOs? We investigate this issue by studying the sample of Australian formerly state-owned companies which have been privatised and listed on the Australian Stock Exchange.

It has been known for decades that initial public offerings (IPOs) are underpriced on average. Therefore, we should not be surprised that privatised IPOs may also be underpriced, leading to a large initial gains for the investors. However, the samples studied in prior research mainly contain privately owned companies and the underpricing is largely attributed to the information asymmetry with respect to firms' future cash flows between large informed investors and small-uninformed investors. Should we expect that the pricing for formerly state-owned companies is more accurate and less biased? The answer is mixed. On the one hand, these formerly state-owned companies were profitable, more mature, and well established with long operating histories before the public float, and therefore, we should expect that the information asymmetry about their future cash flows to be less severe than that for privately owned companies. On the other hand, investors in these privatised IPOs also face the uncertainty about the government regulation of the firms post-IPO and even possible renationalisation, thus, some risk premium associated with government uncertainty is also expected. In addition, governments, in pushing their own agenda such as increasing shareholder ownership, may introduce rationing mechanism to allocate shares below the market-clearing price. These additional factors may affect the prices for the privatised IPOs.

In general, the evidence of initial returns on public floats of state-owned companies is consistent with that of privately-owned companies: all initial public offerings tend to be underpriced on average. Several recent international studies (Dewenter & Malatesta 1997; Huang & Levich 1998; Jones, Megginson, Nash & Netter 1999) conclude that privatised IPOs indeed are underpriced worldwide, but there does not appear to be a general tendency for privatisations to be more underpriced than privately owned company IPOs. For example, among the eight countries studied by Dewenter and Malatesta, only the samples in the UK exhibit a higher degree of underpricing for privatised IPOs, while for the samples in Canada and Malaysia the opposite is true.

Our study focuses on the Australian experience. We collect the data for the 11 privatised IPO firms in Australia in the past decade, and address two main issues. First, we wish to estimate the magnitude of the underpricing of state-owned IPOs in Australia. Is it different from that of privately owned Australian companies? Is it different from that of publicly owned companies in other OECD countries? Second, we aim to identify the factors which significantly affect the degree of underpricing. Based on prior research in the area, we test several possible factors which might affect the magnitude of underpricing of privatised IPOs: information asymmetry, government uncertainty, issue size, and market sentiment. In addition, we explore the possible impact of the two-tiered pricing system, which allows retail investors pay a fixed price for the shares while let institutional investors to bid in a range, employed in a large proportion of the firms privatised in Australia.

Our results show that on average the initial market-adjusted rates of return on the first trading day for retail investors in privatised IPOs are about 11%, which demonstrates that the degree of underpricing for privatised IPOs in Australia is not significantly larger than that of privately-owned IPOs as reported in Lee, Taylor and Walter (1996). The magnitude of the underpricing of Australian privatised IPOs is also similar to that of the OECD countries as documented in Huang and Levich (1998). To interpret this number, we notice that investors in these privatised IPOs face not only the information asymmetry problem, but also the government regulatory uncertainty. Since it is generally believed that the information asymmetry is not as severe in large, well established, state-owned companies as that in small, young, privately owned companies, the underpricing possibly includes the risk premium associated with government uncertainty.

We then perform the cross-sectional analysis to see which factors might be influential in determining the degree of underpricing. While the size of the issues and the degree of underpricing are positively correlated, and also on average, the floats under the Coalition governments (which are considered right of the centre in the political spectrum) offer higher first day trading returns for investors than otherwise, we do not find any statistical significance for several factors suggested by theories which could affect the degree of underpricing. In one regression, with explanatory variables representing information asymmetry, government uncertainty, and market sentiment respectively, we find only the impact of the variable representing market sentiment to be significant.

Despite the fact that the magnitude of the underpricing for privatised IPOs is similar to that of privately-owned IPOs in Australia, it would be naive to claim that the sole purpose of government privatisation is value maximisation. In particular, the Australian Governments have clearly stated their objective of making share ownership the Australian dream of 1990s. (3) In the past, governments have consistently structured the IPO offering to facilitate economic and political objectives. Consequently, share allocations have been tilted to favour domestic investors and employees, foreign ownership has been limited, control restrictions have been imposed on some privatised firms, and domestic investment bankers have been used as co-lead underwriters.

The remainder of the paper is organised as follows. In section 2, we review the existing literature on the theory as well as international empirical evidence of underpricing for privatised IPOs. This builds up the foundation for our empirical work. In section 3, we describe the data sample and present the results of the statistical analyses and tests. Section 4 concludes the paper.

2. Literature Review

2.1 Reasons of Underpricing for Privatised IPOs

It is well established in the literature that initial public offerings tend to be underpriced, see Ibbotson, Sindelar and Ritter (1994) for the US sample, Lee, Taylor and Walter (1996) for the Australian sample, and Loughran, Ritter and Rydqvist (1994) for international evidence. In every study, the evidence indicates that IPOs are underpriced on average, providing large initial returns to investors who are able to buy shares at the offer price.

Among many explanations of the IPO underpricing puzzle, Rock's (1986) model is perhaps the most influential. In his model, some large institutional investors are assumed to have better information about the quality of new issues than small uninformed investors. Consequently, informed large investors will attempt to buy shares only when an issue is underpriced, but the amount they can purchase is rationed. Uninformed small investors, on the other hand, do not know which issues will be underpriced or overpriced, and so will be allocated only a fraction of the most desirable new issues, while receiving all of the least desirable new issues. Uninformed investors thus face a winner's curse: if they get all of the shares they demand, it is because the informed investors do not want these shares. Faced with this adverse selection problem, uninformed investors will submit purchase orders only if IPOs are underpriced sufficiently, on average, to compensate them for the bias in the allocation of new issues.

Several alternative explanations have been proposed in the literature. For example, investment bankers and firm managers may fear litigation for misrepresentation if the new issue significantly declines in price soon after the issue. This litigation would be based on a claim that the investment banker and the firm omitted material information in the prospectus, see Thakor and Hughes (1992). It could also be true that investment bankers have a longer term and more extensive relationship with the purchasers of the new issues rather than the issuers, and will tend to favour the purchasers in exchange for implicit or explicit favours in other deals. Furthermore, investment bankers have a strong incentive to perpetuate the notion that it is an unqualified disaster if the issue is undersubscribed and deliberate underpricing may reduce the probability of such an outcome.

Many state-owned companies were in well-established industries such as transport, tele-communications, and banking, and had long operating history prior to privatisation. Therefore, it is relatively easier for investors to assess the firms' future cash flows, and the information asymmetry concerning the market value for those firms would be less severe than otherwise. However, privatisation also has some unique problems because political factors may affect the offer price and selling mechanism employed. We will term such political considerations as `government uncertainty' throughout the paper. Perotti (1995), Biais and Perotti (1999) provide several arguments in support of higher discounting for privatised IPOs because of government uncertainty.

When a government sells a state-owned firm through public float, it may pursue other political and economic objectives than simply maximising revenues from the sale. As Perotti (1995) observes, the government in this situation is both the issuer and the regulator. Such a dual role puts it in a position to affect the value of the firm not only at the initial sale, but also after the sale through government regulation and income redistribution policy. This exposes investors to an additional problem regarding a government's commitment to privatisation.

Perotti (1995) models this policy uncertainty as an asymmetric-information problem in which investors are uncertain whether a government is populist or committed to privatisation. In his model, populist governments cannot resist the political pressure that arises, after privatisation, to redistribute the firm's value. They may achieve this either by interfering in the operation of the company, changing the regulatory environment, or by reversing policy altogether. As the possibility of changes in policy is reflected in the sale price, any government prefers the announcement of a policy of restraint. In the absence of a reliable signal, however, only a policy maintained over some time can mitigate the perceived risk. Thus, a committed government has to take costly action to signal its intent. A partial sale and possibly its underpricing are signals of commitment. When uncertainty regarding a government's future policy toward interference is low, a signalling equilibrium will occur if the government chooses to privatise a firm gradually via partial sales. The initial sale of a small portion of a firm signals that the government, as a major shareholder, is willing to bear most of the redistribution costs associated with subsequent interference in it. When uncertainty regarding future policy is high, private investors are likely to perceive that a small initial sale indicates the government does not intend to relinquish control. In this case, underpricing the initial sale of large portions of the state-owned company is needed to send the market a credible signal.

Underpricing has another political effect. Biais and Perotti (1999) point out that some governments are unable to commit credibly to a market-oriented policy. Furthermore, an existing market-oriented government can be voted out in favour of a left-wing government that plans to expropriate the gains from privatisation. Biais and Perotti show that a market-oriented government's optimal privatisation strategy is to underprice shares in fixed-price offers and then ration the shares to median-class voters. The allocation of underpriced shares is an inducement for median-class voters to align their interests with those of the market-oriented government. To this end, the government should restrict the amount available to foreigners and wealthy investors. The number of shares that must be allocated to the median class and the required underpricing both increase with a country's income inequality. This is because greater income inequality implies poorer median-class voters, who require more persuasion if they are to reject a left-wing government's policies.

2.2 International Evidence of Underpricing for Privatised IPOs

Three recent comprehensive studies have documented widespread phenomenon of underpricing for privatised IPOs across many countries. For a sample of 109 privatisation IPOs in eight countries (which does not include Australian companies), Dewenter and Malatesta (1997) conclude that there does not appear to be a general tendency for privatisations to be more underpriced than privately owned company IPOs. Among the eight countries studied, only the samples in the UK exhibit a higher degree of underpricing for privatised IPOs, while for the samples in Canada and Malaysia the opposite is true.

Huang and Levich (1998), working with a large sample of 507 privatisation offerings from 39 countries over the period of 1979-1996, also present evidence of underpricing for privatised IPOs. From the cross sectional analysis conducted they find support for elements of the theories which have been rather successful in explaining the underpricing of privately-owned IPOs, but they have not found that restricting foreign ownership will increase the magnitude of the underpricing. They conclude that for the IPOs privatised by governments, the results are consistent with the notion of value maximization but not with that of political consideration.

Jones et al. (1999) build upon the work of Perotti (1995) and Biais and Perotti (1999) and focus on the political objectives and privatisation through share offerings by governments around the world. Using a sample of 630 share issue privatisations from 59 countries, they find that the magnitude of the underpricing for privatised IPOs is similar to privately-owned IPOs. They examine how governments tilt their share allocation to favour domestic investors, impose control restrictions on privatised firms, use domestic investment banker as lead underwriter, and use fixed-price rather than book building or competitive tender offers to achieve political objectives. Overall, they find that governments indeed use share privatisation for such political gains.

In a recent paper, Lee, Steiner, Taylor and Walter (2000) also contains some analysis of initial performance of the ten Australian share issue privatisations that occurred between 1989 and 1998. (4) They employ an accounting approach, which is not universally accepted, to value privatised companies and conclude that those companies were approximately priced fairly. We do not reach such a conclusion. We compare the results from our sample with the results from privately-owned IPOs and only conclude that government regulatory uncertainty seems to require some risk premium. Furthermore, we try to understand the magnitude of the underpricing and attempt to explain this more closely to the Australian features such as the two-tiered pricing mechanism.

3. Data and Empirical Testing

We begin with a description of the data set, followed by the statistical analyses and interpretation of the obtained results.

3.1 Data Set

The starting point for our sample is the report published by the Reserve Bank of Australia (see footnote 1). We add Telstra and TAB Queensland to the sample as they were privatised after the article was published. We also add Bank of Western Australia, which was listed as a trade sale in the article. Further research into the deal indicated, however, as a part of the agreement between Royal Bank of Scotland and the Western Australian government, 49% of the bank had to be offered to the public through a float. In addition, we eliminate Suncorp from the sample as Suncorp was merged with the Metway bank, which was already a publicly listed company. Table 1 lists the companies which, to date, have been privatised through public float since late 1980s. 11 companies sampled, their ticker symbols and first trading date on the Australian Stock Exchange.

Table 2 provides the descriptive statistics for the eleven privatisations. This data is collected from the companies' prospectuses. The total market value of the privatised companies is $21.10 billion; Telstra alone accounts for 67.9% of this value. Five companies restricted foreign ownership to varying degrees, and six made special offers to the current employees to guarantee an allocation up to certain limit, with the ADD employees being granted an optional stake. Restrictions on institutional shareholdings are more uniform across the companies. While ADD, SGI and TAH imposed no explicit limits whatsoever, the other companies set explicit maximum proportion institutional investors can be allocated, with the exception that TQL set a minimum of one third of the issue for institutions. The proportional ownership retained by the government is perhaps indicative of its inexperience with the process (with ADD), the size of the issue, and possibly the electoral sensitivity of the `sell-offs' of large, well established companies such as CBA, QAN and TLS. Whether the partial sales were used as a signalling tool, as suggested by Perotti (1995), is not directly testable. Notably, for all the other types of privatisations the governments retained almost no equity in the new companies.

Table 3 provides the price information for the 11 privatised companies. The price information for companies and the market are obtained from Datastream and the company prospectuses. The All-Ordinaries Accumulation Index of total returns is used as the market index. In preparing table 3, we note that the Australian institutional setting for initial public offerings has a number of differences from the US and other countries. In Australia, prices are set in the prospectuses for a majority of offerings, which are then distributed to potential small investors, while large institutional investors are often required to bid for their desired holdings within the range provided. For example, in the prospectus for TAB, institutions are instructed to bid in the range of $1.80 to $2.20 per share, while retail investors are required to pay the lesser of $2.05 per share and the final winning price of the institutional investors. In all cases, the retail price of a privatised IPO for small investors is not higher than that paid by the institutional investors.

In the US the offering price is set on the day before the listing after a lengthy book building process. Therefore, one-day raw returns are simply calculated on the basis of the offer price to the investors and the closing price on the first listing day. By contrast, the market adjusted initial return for an Australian issue is calculated by adjusting the raw return for the movement of the All Ordinaries Accumulation Index between the lodging date of the prospectus and the first date of trading, because for retail investors the price is essentially fixed at the date when the prospectus is lodged to the stock exchange. The average `delay' in listing is about 47 days for the sample in this study. Please note that the initial returns are calculated for retail investors.

It is worth noting that in Australia the public float of a state-owned company may involve multiple stages. For example, there were three subsequent sales of Commonwealth Bank of Australia and two (so far) stages for Telstra. This study, however, focuses only on the market prices and performance of the first tranch offering. The later stages of offerings are perhaps similar to seasoned offerings for privately-owned companies. Based on the information in table 3, the unadjusted first trading day mean and median returns for privatised IPOs are 11.96% and 7.32% respectively, while the market adjusted first trading day mean and median returns are 10.92% and 7.21% respectively. Since the median rates of return are lower than the mean rates of returns, the distribution of the initial return is skewed. This skewness is mainly caused by the higher initial return of Telstra's first tranch offering. For example, if we exclude Telstra from the calculation, then the average unadjusted initial rate of return would drop from 11.96% to 9.46%, while the median would only drop from 7.32% to 6.27%. It is also interesting to note that weighing the returns on the basis of the issue size, the unadjusted average returns for the 11 companies is 27.83%, while the market adjusted average return for the 11 companies is 34.14%. Although these returns are quite high, a comparison with other studies is infeasible as much of the previous research has used equally-weighted average returns.

3.2 Testing & Estimation

The first objective is to test for difference between the magnitude of underpricing for privatised IPOs and that of privately-owned IPOs. A standard t-statistic test is employed. The test statistic is given by:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where: [r.sub.p] = the average privately-owned company IPOs initial rate of return;

[sigma] = the sample standard deviation of ([r.sub.i] - [r.sub.p]); and

T = the Student's t-distribution with 10 degrees of freedom as the sample contains data for 11 companies.

Based on the previous research by Lee, Taylor and Walter (1996), [r.sub.p] = 11.86% on the market adjusted basis. This statistic is obtained by calculating the returns on all IPOs listed on the Australian Stock Exchange from January 1976 to December 1989. As we do not have access to more recent data, we compare our result with Lee, Taylor and Walter study. Substituting the data required, we obtain that T = -0.18277, implying that the magnitude of (market-adjusted) underpricing for privatised IPOs is not statistically different from that of privately-owned IPOs at all conventional confidence levels. Lee et al. (1996) also report that the unadjusted average initial return for IPOs is 16.41%. The t-statistic for the unadjusted initial returns in our sample is -1.22, which also implies that the magnitude of unadjusted underpricing for privatised IPOs is not statistically different from that of privately-owned IPOs at the conventional confidence levels.

Huang and Levich (1998) report that the initial unadjusted returns for privatised IPOs are 10.8% for OECD countries and 65.2% for non-OECD countries. Again, employing the t-test, we find that t-statistics are 0.318 and -14.61 respectively. This means that the initial returns for Australian privatised IPOs is not statistically different from that of OECD countries, but are significantly lower than that of non-OECD countries at the conventional confidence levels.

The second issue to be addressed relates to the factors which potentially influence the magnitude of initial returns for privatised IPOs. To choose the independent variables, we rely on well established theories and empirical findings on this topic. Among these, three variables are the most important: information asymmetry, government uncertainty, and market sentiment.

1. Information Asymmetry

As reviewed in section 2, in the literature for underpricing of IPOs, the most influential study is perhaps Rock's information asymmetry theory. This theory implies that the greater the uncertainty about the value of a new issue, the larger the expected underpricing. Numerous previous studies use the price volatility based on 30 trading days after the offering as the proxy for uncertainty of the issue. We believe that if a firm's unique risk is a source of such an information asymmetry, then a proper measure of this should be used. Therefore, based on the Capital Asset Pricing Model(CAPM), we use the variable [[theta].sub.i], which is defined as the difference between the total variance of the stock's return and the total variance of the market return in 30 trading days after the listing,

[[theta].sub.i] = var([r.sub.i]) - var([r.sub.m])

Note here we essentially assume that all the new issues have a beta of 1. A precise beta for a new issue is difficult, if not impossible, to estimate.

2. Government Uncertainty

Strictly speaking, government uncertainty is also a form of information asymmetry. Here, the specific part of the uncertainty is not between large informed investors and small uninformed investor--it is the investors confidence about the government's commitment of privatisation and its social policies such as income distribution and industry regulations.

According to Biais and Perotti (1999) and tested empirically by Jones et al. (1999), income distribution is important in determining the degree of underpricing as reviewed in section 2. However, we are unable to test this theory fully, because the Australian Bureau of Statistics (ABS) does not have the time series data for median household incomes or the Gini coefficient from 1988 to 1998. The ABS has collected such data on a yearly basis only from 1994 onwards. The available Gini coefficient and the median and mean household income were reported in the publication titled `Income Distribution, Australia (ABS 6523.0), 5 August, 1999'. From the reported data, there has been no significant change in income distribution in Australia from 1994 to 1998. The Gini coefficient was 0.443 in 1994-95, 0.437 in 1995-96, 0.444 in 1996-97, and 0.446 in 1997-98, respectively. Performing the statistical test based on such a small sample when the independent variable barely change will not yield useful results.

Due to lack of income distribution data, we look at potential political influence from a different angle. We notice that four out of eleven firms (BWA, SGI, TAH, TLS) were privatised under the Coalition governments which are to the right of the centre in the political spectrum. The average market adjusted initial returns for these four firms is 21.80%, which is significantly higher than that of 4.69% for the other seven firms privatised under the Labor governments. This perhaps indicates that the right of centre governments may have more difficulty in persuading the general public to accept the privatisation process and therefore offer a higher degree of underpricing for the IPOs. In the regression analysis, we employ a dummy variable to detect whether different party in power will lead to a statistically different outcome to the privatised IPOs' underpricing.

3. Market Sentiment

IPO markets also tend to be cyclical. During some periods they are `hot' and during other periods they are `cold' (see Ibbotson, Sindelar & Ritter 1994). In other words, investors' sentiment is quite important. To accommodate this notion, we use the variable [R.sub.i], the return on the All Ordinaries Accumulation Index over 30 trading days prior to the listing of stock i, to measure the market sentiment. This market sentiment theory predicts that the higher the variable [R.sub.i], the higher the initial rate of return for the IPO.

4. Other Variables

There are several other variables considered in the literature. Chowdhry and Sherman (1996) argue that IPO initial returns may be positively related to the length of time between setting the offer price and the offer date. This hypothesis is also somewhat consistent with information asymmetry theory. We include the number of days between the prospectus filing and the listing on the Australian Stock Exchange in our regression to capture the influence of time.

Beatty and Ritter (1986), and Mauer and Senbet (1992) present evidence that initial returns are negatively related to the offer size. Because of the sheer size of Telstra, using the market value directly may distort the result dramatically due to heteroscedasticity. We choose the variable [S.sub.i], the natural logarithm of the ratio of the size of issue over the quarterly GDP, to measure the size effect on the degree of underpricing for our sample. The size of the issue may also matter if the demand curve is not flat. Some privatised IPO issues such as the Commonwealth Bank, Telstra and Qantas, are large in the Australian capital market. In addition, foreign ownership are restricted for five companies in the sample. We notice that the coefficient of correlation between the relative size (the amount issued divided by the quarterly GDP) and the degree of market adjusted initial returns of the privatised IPOs is 68.68%.

Finally, we are interested in determining whether the restriction on institutional holdings has any affect on the magnitude of underpricing for privatised IPOs. Assuming that such restrictions increase the acceptability of the float to the general public and make it more politically acceptable, the initial returns for privatised IPOs should be higher for floats with such restrictions than otherwise. Variable [I.sub.i] denotes the proportion of restrictions on institutional holdings as stated in the prospectuses--however, only 7 out of 11 companies explicitly stated such restrictions in their offering documents, while Queensland TAB sets a lower limit to the institutional investors.

Because of the small sample, it is not feasible to run a single multiple variable regression. Table 4 lists several OLS regression results using various independent variables. In the first regression, reported in column 2 on table 4, we find that the higher the unique risk as measured by [[theta].sub.i], the higher the initial market adjusted returns. We also see that the underpricing under the Labor party (when Dummy = 1) is smaller. These two findings are consistent with the theories which relate the degree of underpricing with information asymmetry and political factors. However, the influence of both factors are not statistically significant.

In the second regression in which the market sentiment variable, [R.sub.i], is added, we see that [R.sub.i] significantly affects the underpricing. It suggests that if the market prior to the listing is strong, the first day trading gains will be smaller. This finding is curious because the sign of the estimated coefficient is opposite to the theoretical prediction. In the third and fourth regressions, we add more explanatory variables. The regression results are fragile in the third regression and all regression coefficients are statistically insignificant. However, in the small sample of 7 firms in the fourth regression, we test the relationship of the initial market adjusted rate of return with the institutional ownership restriction, post-listing firm specific uncertainty, and the political party in power. We find that the coefficients are all significant at 5% level or better. The degree of underpricing is positively related to the post-listing firm-specific uncertainty, and the proportion of the shares can be held by institutional institutions, but it is negatively related to the market movement prior to listing.

Judging from the regression results, we may conclude that the underpricing evidence for the Australian privatisation sample is consistent with other international evidence. Although it is surprising that most of the potential explanatory variables suggested by the traditional theories do not have much power with the exception of the fourth regression model, this result is broadly consistent with Dewenter & Malatesta (1997) who also do not find significant results for similar variables when applied to a much larger cross-country sample (N = 95) of privatisations. It is worth noting that the Dewenter and Malatesta regressions find only one variable named PRIMITIVE (which distinguishes between developed and developing capital markets) consistently significant. (5) The overall explanatory power of their regressions ([R.sup.2] = 0.494) is also comparable to the results obtained in this study. However, Huang and Levich (1998) do find that traditional theories such as information asymmetry, reputation building, and investor sentiment that are used to model the behaviour of privately owned IPOs can also be applied to privatised IPOs.

3.3 Two-Tiered Pricing System and Underpricing

As we mentioned before, in Australia, the privatised IPOs are often sold through the rather unique two-tiered pricing system, which allows retail investors pay a fixed price for the shares while let institutional investors to bid in a price range. It is typically the case that small investors place their orders first, and institutional investors submit their bids later. The final results of the bidding are announced the day after the closing of institutional bidding but one day before the share being listed on the Australian Stock Exchange. (6) In our sample, seven out of eleven companies adopted this method. They are CSL, GIO, TAB, QAN, TAH, TLS and TQL. In all cases, small retail investors are protected by an upper limit so that they won't pay higher price than institutional investors.

The two-tiered pricing structure has quite a few advantages. On the one hand, small investors can directly place their orders, knowing the exact price they are going to pay, without the need of a broker. This facilitates the government's policy goal to increase shareholder ownership in Australia. It also leads to a more equitable distribution of shares, because the distribution of shares does not depend on the relationship of the underwriters or brokers and the potential investors. (7) On the other hand, letting large institutional investors bid in a range also facilitates the movement of the final sale price closer to the market clearing price.

Since the final institutional price is generally announced one day before the trading, this price should indicate the demand for the shares, at least from institutional investors' point of view. It follows that if the clearing price is at the upper limit of the range, then the market interest for the new issue is high and therefore the first day trading price should move higher to meet some unfulfilled demand from institutions. Similarly, if the clearing price is cleared in the middle or lower part of the bidding range, then it is possible that the shares are not `hot' and the initial price increase on the first trading day would be smaller or may even negative. This sentiment right before the listing should play a role in the determination of the first day trading gains or losses of the new issue. In our sample, the range coefficient (8) is 0.667 for CSL, 0.87 for TAB, 0.667 for QAN and 1 for all remaining four companies. Simple calculation of the coefficient of correlation, which is 0.3203, reveals that there is a positive relationship between the range coefficient and the degree of underpricing. However, in the fifth regression reported in Table 4, we see that the impact of the range coefficient on the degree of underpricing is not statistically significant.

4. Conclusions

This study presents an analysis of a complete sample of 11 privatised IPOs in Australia in the past decade of economic reform. We find that on average, the privatised IPOs are underpriced with a magnitude comparable with that of privately-owned IPOs in Australia and of privatised IPOs in the OECD countries. Our results, to a certain degree, do put to rest the notion that Australian governments have sold the companies too `cheaply'. The cross sectional multivariate regression analyses do not support some of the theoretical explanations of underpricing, although several variables, such as the size and the party affiliation, as stand-alone variables have a high coefficient of correlation with the degree of underpricing. We have also argued that the Australian governments structure the offerings to further political and economic policy objectives. Because only limited data is available, we are unable to firmly establish the link between the median-voter theorem and the degree of underpricing as suggested by Biais and Perotti, but we do notice that the average market adjusted initial returns for the firms privatised under the Coalition governments are significantly higher than that of the firms privatised under the Labor governments.

Table 1
The Sample

                    Company                   Ticker   Listing Date
                                                       (dd/mm/yy)

Australian Industry Development Corporation   ADD      01/09/89
Bank of Western Australia                     BWA      01/02/96
Commonwealth Serum Laboratories               CSL      31/05/94
Commonwealth Bank                             CBA      13/09/91
Qantas                                        QAN      01/08/95
Telstra                                       TLS      19/11/97
NSW Government Insurance Office               GIO      24/07/92
TABCORP                                       TAH      16/08/94
WA State Government Insurance Office          SGI      13/04/94
NSW TAB                                       TAB      23/06/98
TAB Queensland                                TQL      16/11/99
Table 2
Descriptive Statistics for the Sample of Privatised Companies

Company   Size    Days to     Foreign   Employee    Insti-    Govern-
Ticker    ($m)    Listing *                        tutional    ment

ADD          25      51         --         5%         --       81.5%
BWA         441      47         --         --        30%        0%
CSL         299      42         --         2%        35%        0%
CBA        1311      68         --         --        44%       70%
QAN        1450      39        26.67%      --        26.67%    75%
TLS       14330      48        11.67%      --        40%       66.67%
GIO        1260      32        10%         --        35%        0%
TAH         609      44        40%         3%         --        0%
SGI         165      65         5%         2%         --        0%
TAB         945      48         --         1%        35%        0%
TQL         268      31         --         1.62%      -- **     0%

* `Days to listing' is the interval between final ASX filing and first
trading day. Foreign and institutional ownerships indicate limits
(where applicable) on ownership by these groups. Employee ownership
is the maximum proportion offered to current employees, and government
ownership is the proportion retained by the government after the
initial float; and

** TQL set the minimum allocation to institutional investors at 1/3
of the total offer.
Table 3
Privatised IPOs and Their Price Performance

                                                     [r.sub.1]
Company          [P.sub.0]   [P.sub.1]   [r.sub.1]     (adj)

ADD                 2.00        2.05        2.50%     -12.36%
BWA                 2.05        2.58       25.85%      22.65%
CBA                 5.40        6.40       18.52%      16.18%
CSL                 2.30        2.42        5.22%       5.02%
GIO                 2.40        2.52        5.00%       6.51%
QAN                 2.10        2.15        2.38%      -4.20%
SGI                 1.00        1.19       19.00%      26.46%
TAH                 2.25        2.16       -4.00%      -8.73%
TAB                 2.05        2.20        7.32%      14.49%
TLS                 1.95        2.67       36.92%      46.84%
TQL                 2.10        2.37       12.85%       7.21%

Mean                                       11.96%      10.92%
Median                                      7.32%       7.21%
Std. Deviation                             12.08%      17.13%

Note:

[P.sub.0] = the listing price in the prospectus;

[P.sub.1] = the first trading day price;

[r.sub.1] = the unadjusted return for the first trading day; and

[r.sub.1] (adj) = the market adjusted return for the first trading day.
Table 4
Regression Results

Market adjusted one-day initial returns for privatised IPOs are
regressed on the number of days between price setting and first
trading date ([D.sub.i]), after market volatility ([[theta].sub.i]),
market sentiment ([R.sub.i]), a variable indicating relative size
([s.sub.i]), a Dummy variable indicating the political party in power,
the restriction on institutional holding [I.sub.i], and the range
coefficient [c.sub.i]. The t-statistics, based on the OLS standard
errors, are given in parentheses (top). The p-values are given in
brackets [bottom].

    Regressor                   Coefficients

Constant              11.9733     18.4156     14.1822
                      (1.0543)    (2.0618)    (0.5728)
                      [0.3225]    [0.0782]    [0.5592]

Days                                           0.2601
                                              (0.6699)
                                              [0.5326]

[[theta].sub.i]        4.7378      0.2876      1.4312
                      (1.1702)    (0.0823)    (0.3197)
                      [0.2756]    [0.9367]    [0.7621]

[R.sub.i]                        -60.3207    -43.3907
                                 (-2.6409)   (-1.0797)
                                  [0.0334]    [0.3296]

[s.sub.i]                                      2.1496
                                              (0.4980)
                                              [0.6396]

Dummy                -15.3119     -9.8985     -8.4597
                     (-1.5781)   (-1.2986)   (-0.9564)
                      [0.1532]    [0.2352]    [0.3828]

[I.sub.i]

[c.sub.i]

Adjusted [R.sup.2]     0.2037      0.5441      0.4244

Sample Size           11          11          11

    Regressor             Coefficients

Constant              -19.4875      21.4672
                      (-2.7095)     (0.6596)
                       [0.07319]    [0.5566]

Days

[[theta].sub.i]         3.2653      15.5098
                       (3.5688)     (3.5430)
                       [0.0376]     [0.0383]

[R.sub.i]

[s.sub.i]

Dummy                 -25.3195     -12.1176
                     (-10.5638)    (-1.2129)
                       [0.0018]     [0.3120]

[I.sub.i]               1.3550
                       (6.7826)
                       [0.0066]

[c.sub.i]                          -36.7043
                                   (-1.0454)
                                    [0.3727]

Adjusted [R.sup.2]      0.9712       0.6738

Sample Size             7            7

(1.) See `Privatisation in Australia', Reserve Bank of Australia Bulletin, December 1997, for a detailed account of Australian privatisation process.

(2.) For example, see John Lyons: `Telstra: The float that sank billions', The Bulletin, October 26, 1999, page 67-72.

(3.) See Fred Brenchley, `Mum and dad's the word,' The Bulletin, March 16, 1999, p. 20.

(4.) We became aware of their work after submitting our paper.

(5.) This finding is also concurred by Huang and Levich (1998) where there is a significant of difference of underpricing between the OECD and non-OECD countries.

(6.) For example, in the first tranch of Telstra share offer, the offer to the public and employees was open on 15 October 1997 and closed on 31 October 1997, while institutional offer opened on 27 October 1997 and closed on 14 November 1997. The final price and basis of allocation was announced on 16 November 1997, and the shares were traded on the ASX starting on the following day, 17 November 1997.

(7.) However, privatised IPO by state governments usually favour its own residents.

(8.) If a share has the bidding range of $1.80 to $2.00 and the actual final price for institutions is cleared at $1.90, then we say the range coefficient is 0.50. In general, for a range of [[x.sub.0]; [x.sub.1]], the range coefficient is defined as

x - [x.sub.0] / [x.sub.1] - [x.sub.0]

where: x = the market clearing price for institutions.

(Date of receipt of final transcript: June 2001. Accepted by Gary Twite, Area Editor.)

References

Beatty, R.P. & Ritter, J.R. 1986, `Investment banking, reputation, and the underpricing of initial public offerings', Journal of Financial Economics, vol. 15, pp. 213-32.

Biais, B. & Perotti, E. 1999, `Machiavellian Privatization', working Paper, University of Amsterdam.

Boardman, A. & Vining, A. 1989, `Ownership and performance in competitive environments: A comparison of the performance of private, mixed, and state-owned enterprises', Journal of Law and Economics, vol. 32, pp. 1-33.

Brenchley, F. 1999, `Mum and dad's the word', The Bulletin, March 16, p. 20.

Chowdhry, B. & Sherman, A. 1996, `International differences in oversubscription and underpricing of initial public offerings', Journal of Corporate Finance, vol. 2, pp. 359-81.

Dewenter, K.L. & Malatesta, P.H. 1997, `Public offerings of state-owned and privately-owned enterprises: An international comparison', Journal of Finance, vol. 52, pp. 1659-79.

D'Souza, J. & Megginson, W.L. 1999, `The financial and operating performance of privatized firms during the 1990s', Journal of Finance, vol. 54, pp. 1397-438.

Huang, Q. & Levich, R.M. 1998, `Underpricing of new equity offerings by privatized firms: An international test', working paper, Hofstra University.

Ibbotson, R.G., Sindelar, J.L. & Ritter, J.R. 1994, `The market's problem with the pricing of initial public offerings', Journal of Applied Corporate Finance, vol. 7, pp. 66-74.

Jones, S.L., Megginson, W.L., Nash, R.C. & Netter, J.M. 1999, `Share issue privatizations as financial means to political and economic ends', Journal of Financial Economics, vol. 53, pp. 271-53.

Lee, P.J., Taylor, S.L. & Walter, T.S. 1996, `Australian IPO pricing in the short and long run', Journal of Banking and Finance, vol. 20, pp. 1189-210.

Lee, P.J., Steiner, N., Taylor, S. & Walter, T. 2000, `Australian share issue privatisations: Balancing finance and politics', working paper, University of Sydney.

Loughran, T., Ritter, R. & Rydqvist, K. 1994, `Initial public offerings: International Insights', Pacific-Basin Finance Journal, vol. 2, pp. 165-99.

Lyons, J. 1999, `Telstra: The float that sank billions', The Bulletin, October 26, pg 67-72.

Mauer, D. & Senbet, L. 1992, `The effect of the secondary market on the pricing of initial public offerings: Theory and evidence', Journal of Financial and Quantitative Analysis, vol. 27, pp. 55-79.

Perotti, E.C. 1995, `Credible privatization', American Economic Review, vol. 85, pp. 847-59.

Reserve Bank of Australia Bulletin, 1997, `Privatisation in Australia,' December, pp. 7-16.

Rock, K. 1986, `Why new issues are underpriced', Journal of Financial Economics, vol. 15, pp. 187-212.

Thakor, A. & Hughes, P. 1992, `Litigation risk, intermediation, and the underpricing of initial public offerings', Review of Financial Studies, vol. 5, pp. 437-70.

Ning Gong ([dagger]) Chander Shekhar (dagger])

([dagger]) Melbourne Business School, University of Melbourne, 200 Leicester Street, Carlton, VIC 3053. Email: [e-mail]; [e-mail]

The authors wish to thank Garry Twite (the deputy editor), an anonymous referee, and audience at the 12th Australasian Banking and Finance Conference, December 1999, in Sydney for helpful comments.