• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Are Mining-Exploration Stocks More Prone to Informed Trading Than Mining-Production Stocks?

Posted on: Friday, 20 January 2006, 06:00 CST

By Poskitt, Russell

Abstract:

This paper examines the level of informed trading in mining- exploration and production stocks listed on the Australian Stock Exchange. The probability of informed trading is higher in exploration stocks. However, this same pattern is detected in a sample of control-stocks matched by trading activity, suggesting that the difference in informed trading between production and exploration stocks reflects the impact of differences in trading activity documented in prior research. These results show that mining-exploration stocks experience the same level of informed trading as other less actively traded stocks in other industries.

Keywords:

MINING STOCKS; INFORMED TRADING.

1. Introduction

The history of the mining-exploration industry is replete with stories of insidertrading and market manipulation (e.g. Blainey 1969; Rae Committee Report 1974). Researchers have also documented that investment professionals perceive mining-exploration stocks to be highly susceptible to insider-trading (e.g. Tomasic 1991) and that directors of mining companies are prolific purchasers of company stock (e.g. Brown & Foo 1998). Moreover, despite the adoption of more stringent disclosure and insider-trading regulations in recent years, mining-exploration stocks continue to figure prominently in breaches of the continuous-disclosure regime of the Australian Stock Exchange (ASX) and in recent trading-abuse cases prosecuted by the Australian Securities and Investment Commission (ASIC).

Although mining-exploration companies appear to be over- represented in practices that are consistent with the existence of strong and potentially valuable information asymmetries, there exists no accepted explanation as to why this should be the case. This paper suggests that the incentive for informed trading derives from the high sensitivity of mining-exploration stock prices to changes in the probability of discovering an economic mineral deposit. The paper then proceeds to examine whether there is a systematic tendency for ASX-listed miningexploration stocks to be more prone to informed trading than mining-production stocks.

This paper uses the PIN model developed by Easley, Kiefer, O'Hara, and Paperman (1996) to measure the level of information- based trading in a sample of mining-exploration and production stocks and a control sample of non-mining stocks selected by matching on-market trading activity or turnover. I find that the probability of informed trading (or PIN) is significantly higher for exploration stocks than for production stocks. However, the PINs of the control-stock subsamples show the same cross-sectional variation as the mining-stock sub-samples. This finding is robust to the specification of the PIN model and the estimation interval. I conclude that the higher PINs of the exploration stocks are more a reflection of the information risk traders face in less actively traded stocks than of any inherent characteristics of the mining- exploration industry. In short, there is nothing extraordinary about the level of informed trading in mining-exploration stocks.

The remainder of this paper is structured as follows. Section 2 develops an explanation for why mining-exploration stocks might be more prone to informed trading than mining-production stocks. section 3 introduces the PIN model developed by Easley et al. (1996). The sample selection, sample partitioning and the data and its characteristics are discussed in section 4. Section 5 presents and discusses the empirical results. Section 6 summarises and concludes.

2. Theoretical Development

It is part of mining-stock folklore that insiders and their associates use their access to private information to make gains at the expense of other less well-informed investors. Blainey (1969) discusses the tricks used by mining company directors in the Victorian goldfields of the 1870s and 1880s to defraud investors.1 Although many of these practices fall into the category of market manipulation, there is little doubt that uninformed investors suffered large losses as a consequence of the severe informational disadvantage they faced. The Report of the Senate Select Committee on securities and Exchange 1974 (Rae Committee Report) provides details on the widespread insider-trading and market manipulation abuses that took place during the Australian mining boom of 1969- 1972.2

What is it about mining-exploration stocks that exposes uninformed investors to such risks? One clue lies in the nature of the mining industry. Mining activity is a three-stage process from exploration through development to production (Mackenzie & Doggett 1992). One of the key features of the mining-exploration stage is the very high level of discovery risk faced by exploration companies. Empirical studies of mineral exploration in Australia and Canada show that the probability of a 'mineral occurrence' yielding an economic mineral deposit is as low as 1% - 2% (Mackenzie & Doggett 1992). The sequential nature of mining activity also means the absence of future cash flows should an exploration company's exploration programme prove unsuccessful. Thus the only reason for a mining-exploration company's stock to have a non-zero value is the likelihood that the company's exploration programme will find an economic deposit before the company exhausts its exploration funds (Brown & Burdekin 2000).

Although several studies find that the stock prices of gold mining companies to be highly sensitive to gold price movements (e.g. Blose & Shieh 1995; Tufano 1998; Twite 2002), Brown and Burdekin (2000) suggest that gold mining companies with proven reserves (i.e. production companies) should be differentiated from gold mining companies without proven reserves (i.e. exploration companies). They argue that stock in the latter is akin to an out- of-themoney call option and that the stock price will be more sensitive to changes in the probability of finding an economic deposit than to the price of gold. There is no shortage of anecdotal evidence to support this proposition. For example, in November 2001 the stock price of Minotaur Resources jumped from A17c to A$1.65 in a single day on announcement of early drilling results and subsequent market speculation that it had discovered a rich copper deposit in South Australia.

The discussion above suggests that the rewards for private information are much higher in the mining-exploration stocks than for mining-production stocks. One would expect these strong incentives to stimulate the level of informed trading in mining- exploration stocks. This supposition is supported by Tomasic's (1991) field study on insider-trading in Australia. Although the author concluded that insider-trading occurred predominantly in smaller capitalisation stocks, he reported that the most commonly identified 'at risk' groups were mining, speculative, exploration and gold shares. Interviewees suggested a number of reasons why mining-exploration stocks were more prone to insider-trading, including: 'there are all sorts of people on the site', 'there are things like drilling reports', and 'a leak from a geologist could create insider-trading'.

Indeed the risks faced by investors in mining-exploration stocks are recognised by the ASX which imposes more stringent periodic disclosure requirements on mining companies, and on mining- exploration companies in particular. Although all listed companies must complete a half-year report and a preliminary final report for the ASX, mining-production and exploration companies must provide the ASX with a quarterly report providing details of miningproduction and development and a summary of the expenditure incurred on those activities (ASX 2002). The quarterly report must also contain a summary of exploration activities (including geophysical surveys) and a summary of expenditure incurred on exploration. In addition the ASX specifies what information should be included when reporting on mining-exploration activities and who is qualified to compile such information. Furthermore, reports of mineral resources and ore reserves prepared by mining companies must conform to the JORC Code to ensure that investors are provided with all relevant information in a clear and unambiguous fashion.3

The ASX's continuous disclosure regime also requires listed companies to keep the market informed of price-sensitive information. To comply with this requirement many exploration companies file regular reports on the progress on their exploration programmes. The ASX monitors the financial media on a daily basis to check that listed companies are complying with their continuous disclosure obligations. The ASX also monitors the market for unusual price and volume movements, a potential sign that information might not have been disclosed to the market. Recent years have also witnessed a strengthening of insider-trading regulations and a greater willingness by ASIC to refer cases to the Director of Public Prosecutions for prosecution.

Recent research suggests that mining-exploration companies figure prominently in potential breaches of the ASX's continuous disclosure regulations (e.g. Neagle & Tsykin 2001).4 Moreover mining- exploration companies continue to feature in \insider-trading and market manipulation cases brought before the courts as a result of ASIC investigations. Between July 1997 and June 2002, three of the four cases of market manipulation and two of the four cases of insider-trading referred to the Director of Public Prosecutions by ASIC involved the stocks of mining companies.5 Research on the profitability of insider-trading also shows that directors are active traders of stocks of mining companies (e.g. Brown & Foo 1998).6

This brief survey highlights that mining-exploration companies have been over-represented in practices that are consistent with the existence of strong and potentially valuable information asymmetries. However the overall level of some of this activity is relatively low and whether it is consistent with a systematically higher likelihood of informed trading in mining-exploration stocks than in miningproduction stocks is a question than can only be settled empirically. And this requires an empirical model.

3. A Model of Informed Trading

The PIN model of Easley et al. (1996) is a sequential trade model. Prior to the start of each trading day nature determines whether a private information event occurs or does not occur. Trading arises from the interaction between potentially informed and uninformed traders and a risk neutral competitive market-maker or dealer.

The PIN model, and variants thereof, has been used to examine the 'cream skimming' hypothesis (Easley, Kiefer & O'Hara 1996); the probability of informed trading in high versus low trading activity stocks (Easley et al. 1996; Brockman & Chung 2000); the impact of analyst following on the level of informed trading (Easley, O'Hara & Paperman 1998); whether informed traders prefer to trade in the stock or options market (Easley, O'Hara & Srinivas 1998); the impact of stock splits on informed trading (Easley, O'Hara & Saar 2001); the relationship between informed trading and expected returns (Easley, Hvidkjaer & O'Hara 2002), and; the influence of market structure on informed trading (Heidle & Huang 2002). One of the more important findings in this literature is that the probability of informed trading is higher for stocks that are traded less frequently (Easley et al. 1996; Brockman & Chung 2000). The basic notion is that when less frequently traded stocks are traded, it is because of traders acting on private information (Easley et al. 1996). This suggests that the experimental design in this study will need to take into account the differences in trading activity between mining-exploration and production stocks.

4. Sample Data

4.1 Sample Selection

Although this paper focuses on information-based trading in mining-exploration and production stocks, the paper employs a sample of non-mining stocks to control for the effect of trading activity on the probability of informed trading. As a first step, an initial sample of 352 mining stocks was compiled from the ordinary shares listed in the weekend edition of the Australian Financial Review dated 30th June 2001 under the sectoral headings Diversified Resources, Energy, Gold and Other Metals. Stocks were removed from this list if they were not continuously listed over the following 12- month period (21 stocks), switched their focus away from the mining industry (19 stocks), were placed in voluntary administration (1 stock) or were stock in a listed unit trust (1 stock). These filters reduced the initial sample to 310 mining stocks.

The second step involved creating a control sample of non-mining stocks. Mining stocks were retained in the mining stock sample if they could be successfully matched with a non-mining stock listed on the ASX. This trading data was provided by SIRCA (securities Industry Research Centre for Asia-Pacific). Successful matching required that, for the period 1st July 2001-30th June 2002, each pair of stocks satisfied the following criteria:

* The difference in mean daily turnover in the two stocks did not exceed 15%;7

* The difference in the number of days the two stocks traded did not exceed 15%;

* The difference in total turnover did not exceed 20%; and

* Both stock traded for at least 60 days during the sample period.

The first two criteria ensured that matched pairs of stocks not only had similar levels of total turnover, but similar levels of daily turnover and trading frequency. The last of the criteria was imposed to facilitate estimation of the PIN model. A sample comprising 275 matched pairs of mining and non-mining stocks was created using this procedure.8 The majority of the 35 mining stocks that could not be matched successfully were characterised by low mean daily turnover and infrequent trading.

Subsequently, four of the matched pairs were eliminated from the sample when selected financial information on the mining stock could not be obtained from the ASPECT financial database. This left a final sample of 271 matched pairs of mining and non-mining stocks.

4.2 Partitioning the Mining Stock Sample

The next step involved partitioning the 271 mining companies into those are engaged primarily in exploration activities and those that are largely production companies. One option was to use the three- digit industry code the ASX allocates to each stock. This three- digit code comprises a two-digit code identifying the main industry group and a third digit identifying the subgroup. Unfortunately, production and exploration are not the only subgroups employed by the ASX and to avoid further reductions in the sample size I opted to employ financial information to partition the sample.9 Specifically, I obtained information on sales, profitability and exploration expenditure for the 271 stocks from the most recent annual report published in the 12-month period 1st July 2001-30th June 2002. This financial information was obtained from the ASPECT financial database.

Sales was measured by item no. 1 : Operating revenue (sales) while operating profit was measured by item no. 100: Operating profit/(loss) after tax. Exploration expenditure presented more of a problem since mining companies differ in the way they report exploration and development expenditure. Some mining companies capitalise the expenditure on their balance sheet while others record the expenditure as a current expense in the profit and loss statement. To circumvent this problem the level of exploration expenditure was calculated by aggregating three items reported in the cash flow statement: item nos. 814 Payments for exploration and development, 815 Payments for development expenditure and 862 Exploration and development costs.

Initially, a composite measure of sales and exploration expenditure, the exploration expenditure/sales ratio, was used to partition the sample of mining stocks into exploration and production stock sub-samples. Intuitively, one would expect that companies with a low exploration expenditure/sales ratio are engaged primarily in production activities while those with a high exploration expenditure/sales ratio are principally exploration companies. However the very large numbers of companies that reported either zero sales and/or zero exploration expenditure compromised the usefulness of such a measure in this study.

The breakdown of sales and exploration expenditure presented in table 1 shows the extent of this problem. A total of 177 companies (65.3% of the sample) have a zero as either the denominator or the numerator (or both) in the calculation of the exploration/ expenditure/sales ratio.

Table 1

Sales and Exploration Expenditure of Mining Companies

As an alternative, I used an annual sales level of $2.5 million to partition the sample of mining stocks. Mining companies with annual sales of $2.5 million or more were classified as production companies while mining companies with annual sales below $2.5 million were classified as exploration companies. The resulting production and exploration stock sub-samples comprise 92 and 179 stocks respectively. Descriptive statistics on sales, operating profit and exploration expenditure for each mining company sub- sample are presented in table 2.

The data in table 2 highlight three important points. First, the Jarque Bera statistics show that many of the underlying distributions are highly non-normal, exhibiting severe skewness and kurtosis. This should not be surprising since the production stock sub-sample includes, at one extreme, a mining giant such as BHP Billiton with sales of A$28,145 million, OPAT of A$3,001 million and exploration expenditure of A$691 million and, at the other extreme, a relative minnow such as Tribune Resources NL which reports sales of A$2.7 million, OPAT of A$0.2 million and exploration expenditure of A$2.1 million.

Table 2

Descriptive Statistics for Selected Financial Data on Mining Companies

Second, companies with substantial production activities are usually profitable while those without significant production activities are generally unprofitable. For example, 50 of the 92 (54.3%) production stocks report positive OPAT while the corresponding figure for the exploration stocks is 15 out of 179 (8.4%). Median OPAT is A$0.8 million and -A$1.2 million for the production and exploration stock sub-samples respectively.

Third, the larger and more profitable companies in the production stock subsample tend to spend more on exploration activities than companies in the exploration stock sub-sample. For example, the median exploration expenditure is A$2.5 million and $0.4 million for the production and exploration stock subsamples respectively. Nonetheless, since the primary issue is whether or not a mining company has reached the production stage, sales rather than exploration expenditure was employed to differentiate between exploration and production companies.

5. Empirical Results

5.1 PIN Model Estimation

Estimation of the PIN model for a stock requires daily data on the number of buyer-initiated and seller-initiated trades. This \information was obtained from SIRCA for each of the 271 matched pairs of mining and non-mining stocks for the 12-month sample period. The parameters of the PIN model are estimated by maximising the likelihood function defined by equation (1). The parameters α and δ are restricted to [0,1] by a logit transformation and μ and ε are restricted to [0, ∞] by an exponential transformation. The likelihood function is maximised using the Newton Raphson search algorithm in the NLP (non-linear programming) procedure of SAS/OR. To minimise the chance of the algorithm locating a local rather than a global maximum, a grid search is conducted over 4,096 combinations of possible parameter values. Under this approach, the Newton Raphson algorithm commences its search procedure from the grid point yielding the highest value of the likelihood function.

5.2 Estimation Results

The PIN model estimation results are presented in table 3. Due to the nonnormality of many of the distributions of the parameter and PIN estimates the discussion uses the median as the primary measure of central tendency. The estimation results for the mining stock sub- samples are reported in columns (1) and (2). These results show that exploration stocks, relative to production stocks, have a lower frequency of private information events (α), a median of 0.1922 versus a median of 0.2443. The Wilcoxon test statistic of 4.28 indicates that this difference is significant at the 0.001 level.

The estimation results also show that mining-exploration stocks have lower arrival rates of informed (μ) and uninformed (ε) traders. The median μ are 7.57 and 15.73 for exploration and production stocks respectively whilst the respective median ε are 1.17 and 3.36 respectively. The Wilcoxon test statistics of 3.24 and 5.14 show that both of these differences are significant at the 0.01 level. However since ε relative to μ is lower for exploration stocks, the PIN is higher for these stocks, a median of 0.3904 versus a median of 0.3500.

Thus the problem with exploration stocks is not the presence of too many informed traders but the absence of sufficient uninformed traders. The Wilcoxon test statistic of 3.42 shows that this difference in PIN estimates is significant at the 0.001 level.

Table 3

PIN Model Estimates

To summarise, mining-exploration stocks have a lower frequency of private information events and enjoy a trading population with proportionately more informed traders than mining-production stocks. The first of these characteristics will reduce the PINs of exploration stocks relative to production stocks whilst the second characteristic will have the opposite effect. Overall, the second effect appears to dominate the first with the result that information-based trading is higher in exploration stocks than in production stocks. The higher PINs of exploration stocks are consistent with the supposition that mining-exploration stocks will be more prone to informed trading.

Nonetheless, the lower frequency of private information events for exploration stocks is surprising. One interpretation is that this reflects the impact of the more stringent periodic disclosure regime that exploration companies face and the regular updates that explorations companies release to the ASX on the progress of their exploration programmes. One would expect the heightened level of disclosure to mitigate information asymmetry. This explanation can be tested by examining whether this difference in α is present in the non-mining control-stock sub-samples.

Furthermore, it is possible that differences in trading activity are responsible for the higher PINs of exploration stocks since mining companies with a production focus are likely to be larger and their stocks more actively traded than the stocks of mining companies that are still at the exploration stage. Recall that PINs tend to be higher for less actively-traded stocks since, when trading does occur, the trader is more likely to be acting on private information (Easley et al. 1996). This would help explain the higher PINs of exploration stocks which appear to be driven primarily by the relative absence of uninformed traders. One way to test this explanation is to compare the PINs of the mining- exploration and production stock sub-samples with the PINs of control samples of non-mining stocks. Recall that the respective mining and non-mining stocks are matched by average daily turnover and trading frequency so this procedure should control for the impact of any difference in trading activity.

The estimation results for the control sample of non-mining stocks are presented in columns (3) and (4) of table 3. The exploration control-stocks have a lower median α than the production control-stocks but the difference in medians is not significant at the 0.05 level. This finding is consistent with the explanation that disclosure practices of mining-exploration mitigates information asymmetry by reducing the frequency of private information events in mining-exploration stocks.

The exploration control-stocks also have significantly lower median μ and ε, most notably for ε. Not surprisingly, the median PIN of the exploration controlstocks is significantly higher than the median PIN of the production control-stocks. This is an important finding since it suggests that it is differences in trading activity that are behind the higher PINs of mining- exploration stocks rather than the unique features of mining stocks such as the high sensitivity of the stock price to information concerning the outcome of exploration programmes.

Table 4

PIN Model Estimates by Quintile

Table 4

PIN Model Estimates by Quintile

I present a cross-sectional breakdown of PIN model estimation results by quintile in table 4. To do this, I partition the sample of 271 pairs of mining and non-mining stocks into five portfolios based on the average daily turnover of the mining stock during the 12-month sample period. Quintiles 1 to 4 each contain 54 pairs of stocks while quintile 5 contains 55 pairs of stocks. Quintile 1 comprises the 54 mining (and non-mining) stocks with the highest average daily turnover while quintile 5 comprises the 55 mining (and non-mining) stocks with the lowest average daily turnover. Not surprisingly, mining-production stocks account for a disproportionate share of stocks in the high turnover quintiles. For example, 42 (45.7%) of the 92 mining-production stocks are in quintile 1.

The estimation results show that the median PINs of the mining stocks rise as average daily turnover declines. The differences are quite dramatic. For example, the median PIN is 0.2694 for Quintile 1 mining stocks and 0.5447 for quintile 5 mining stocks. The same pattern is apparent in the PINs of the non-mining stocks. For example, the median PIN is 0.2740 for Quintile 1 non-mining stocks and 0.5160 for quintile 5 non-mining stocks. The source of the higher PINs for the lower turnover quintiles is the more rapid decline in ε than in μ as average daily turnover declines. There is no apparent pattern in the cross-sectional behaviour of α in either the mining or non-mining stocks.

The rise in the PINs of mining and non-mining stocks as a result of the relative scarcity of uninformed traders as turnover declines is consistent with prior research (e.g. Easley et al. 1996; Brockman & Chung 2000). The cross-sectional similarities in the PINs of the mining and non-mining control-stocks despite the variations in the exploration/production stock mix among the mining stocks across quintiles is further evidence that the risk of information-based trading in miningexploration stocks is similar to that of mining- production stocks after controlling for trading activity.

5.3 Robustness Tests

Table 5

PIN Model Estimates (6 Parameter Model)

Table 5

PIN Model Estimates (6 Parameter Model)

The impact of allowing both μ and ε to differ on the buy and sell sides of the market is apparent from the parameter estimates in table 5. The arrival rates if informed investors are higher on the buy side than on the sell side while the reverse is the case for uninformed investors. However the overall impact on PIN is minor since the modified PIN formulae attaches the same weight to arrival rates on the sell side as it does to arrival rates on the buy side.

I test the null hypothesis that the median PIN in the six- parameter model is identical to the corresponding median PIN in the four-parameter model. This hypothesis is tested for all four sub- samples. The Wilcoxon test statistics for the mining-exploration, mining-production, non-mining-exploration and non-miningproduction stock sub-samples are 0.46, 0.53, 1.43, and 0.10 respectively. These results indicate that the modifications to the PIN model do not have any significant impact on the PIN estimates.

The results reported in columns (1) and (2) of table 5 show that mining-exploration stocks have a significantly lower median α than production stocks. These stocks also have significantly lower median μ^sub b^, μ^sub s^, ε^sub b^ and ε^sub s^ than production stocks, with the differences more dramatic in the cases of ε^sub b^ and ε^sub s^. Not surprisingly, exploration stocks are also found to have a significantly higher median PIN, consistent with earlier results. Again, the impact of the relative absence of uninformed traders, on both the buy and sell sides of the market, offsets any tendency for the lower frequency of private information events to reduce PINs with the result that PINs are higher for exploration stocks than production stocks.

The results for the non-mining control-stocks are similar to those reported earlier. The results reported in columns (3) and (4) show that the exploration control-stocks have significantly lower median a. Thus the lower a of exploration stocksis found in both the mining and non-mining control-stock samples. This is not consistent with the results reported earlier and suggests that any claim that the disclosure practices of mining-exploration mitigates information asymmetry and puts downward pressure on information-based trading must be treated with caution.

Exploration control-stocks also have significantly lower median μ^sub b^, μ^sub s^, ε^sub b^ and ε^sub s^ than the production control-stocks. Again, the differences are more dramatic in the cases of ε^sub b^ and ε^sub s^ than for μ^sub b^ and μ^sub s^ and the resulting PINs are significantly higher for exploration control-stocks.

The results reported in table 6 show that the median PINs of both the mining and non-mining control-stocks rise as average daily turnover declines. For mining stocks, the median PIN rises from 0.2764 to 0.4979 between quintiles 1 and 5. For the non-mining stocks, the median PIN changes from 0.2695 to 0.4626 between quintiles 1 and 5. In both cases, this reflects a more precipitous decline in the arrival rates of uninformed investors relative to informed investors. These results are consistent with the results reported earlier and with prior research (e.g. Easley et al. 1996; Brockman & Chung 2000).11

5.3.2 PINs and Information Risk As an additional robustness test, I examine the behaviour of the market spread of mining stocks to determine if the PIN estimates developed earlier have an economically meaningful interpretation.12 Specifically, I focus on the issue of whether the PIN of a stock is an appropriate proxy for information risk.

The spread decomposition literature posits that dealer bid/ask spreads reflect three factors: order processing costs, inventory costs and adverse selection costs arising from the presence of informed traders (e.g. Stoll 1989). Empirical research shows that the market spread in an order-driven market can be decomposed into the same three components (e.g. Brockman and Chung 1999). Prior research also shows that order processing costs can be proxied by the stock price level while inventory holding costs can be proxied by stock trading activity and/or stock price volatility (e.g. Stoll 1978). Easley et al. (1996) and Easley, O'Hara and Paperman (1998) suggest that the adverse selection cost associated with a stock can be proxied by the product of the stock's PIN and the daily price range of the stock.

Table 6

PIN Model Estimates by Quintile (6 Parameter Model)

Table 6

PIN Model Estimates by Quintile (6 Parameter Model)

Prior research into market spread behaviour in a limit order market finds that spreads are negatively related to price and trading activity and positively related to stock price volatility (e.g. Aitken & Frino 1996). If PIN x ADPCH is an appropriate proxy for adverse selection costs faced by liquidity providers then spreads should be positively related to PINs (e.g. Copeland & Galai 1983).

SIRCA also provided data on four key microstructure variables for each mining stock-the average market spread per day, the average daily transaction price, daily turnover and the daily open-to-close return. The latter is used to calculate stock price volatility. Descriptive statistics on the four key microstructure variables are presented in table 7. The data reported in table 7 show that exploration stocks have higher market spreads, trade at substantially lower prices, have markedly lower daily turnover and exhibit greater price volatility on a day-to-day basis than the production stocks. The Wilcoxon test statistics reported in table 7 show that all of these differences are significant at the 0.05 level.

I estimate equation (5) using OLS. I measure PRICE and TURN in inverse terms to capture non-linearities in the relationships between these variables and SPREAD. The use of the inverse specification rather than logarithms also provides superior model fit. I expect the estimates of β^sub 1^, β^sub 2^, β^sub 3^ and β^sub 4^ to be positive.

One problem with the specification of the spread model is the high positive correlation between VOLATILITY and ADPCH. The sample correlation coefficient of 0.80 indicates that multicollinearity will be a problem if both variables are included as explanatory variables. First I estimate equation (5) excluding the PIN x ADPCH variable. The estimation results are reported in column (1) of table 8 and are consistent with prior expectations. The OLS estimates of β^sub 1^, β^sub 2^ and β^sub 3^ have the expected sign and all are significantly different from zero at the 0.05 level.

The estimation results reported in columns (2) and (3) use the Easley et al. (1996) Easley, O'Kara and Paperman (1998) measure of information risk. These results show that the estimates of β^sub 4^ are positive and significantly different from zero at the 0.05 level. This finding is robust to the specification of the PIN model. Anticipating the argument that the PIN x ADPCH variable is simply a proxy for VOLATILITY, I re-estimate equation (5) but this time replacing the PIN x ADPCH variable with PIN and re- introducing the VOLATILITY variable.

The estimation results for this version of the spread model are reported in columns (4) and (5). The results show that when the PIN variable is added the signs of the estimates of both β^sub 3^ and β^sub 4^ are positive and statistically significant. This result is robust to specification of the PIN model. These results confirm the earlier finding that stocks with greater information risk carry greater market spreads, all other things being equal.

Table 7

Descriptive Statistics for Microstructure Data

Table 8

OLS Estimation of Market Spread Model

Overall, the spread model regression results show that the risk of information-based trading, measured by either PIN or PIN x ADPCH, is priced into the market spread, consistent with the theoretical literature (e.g. Copeland & Galai 1983). This gives comfort over the issue of the meaningfulness of the PIN estimates derived from the estimation of the PIN model.

6. Summary and Conclusions

This paper examines the relative extent of information-based trading in mining-exploration and mining-production stocks listed on the ASX. I suggest that the incentive for informed trading in exploration stocks comes from the high sensitivity of the stock price to changes in the probability of finding an economic mineral deposit. I test for any systematic tendency for ASX-listed mining- exploration stocks to be more prone to informed trading than mining- production stocks using the PIN model of Easley et al. (1996).

I find that the probability of informed trading is indeed significantly higher for exploration stocks than for production stocks. To determine if differences in trading activity is responsible for this result, I compare the PINs of the mining stocks with the PINs of a control sample of non-mining stocks matched by average daily turnover and trading frequency. Exploration controls stocks are found to exhibit the same relative scarcity of uninformed traders compared to production controls stocks, with resulting higher PINs. The PINs of the control-stock sample also show the same cross-sectional variation as the PINs of the mining-stock sample. These findings are robust to alternative specifications of the PIN model and the choice of estimation interval. I conclude from this that the difference between the PINs of the exploration stocks and production stocks is due to the difference in trading activity rather than to the unique features of exploration stocks, such as the high sensitivity of the stock price to information concerning the outcome of exploration programmes.

In addition, spread-regression-model estimates show that market spreads of mining stocks are significantly positively related to the probability of informed trading in that stock, consistent with the microstructure literature. This suggests that the PIN estimates are a meaningful proxy for information risk.

This paper makes three important findings. First, the paper provides empirical evidence that the mining-exploration sector suffers from greater information-based trading than the mining- production sector. This evidence is somewhat more concrete than the anecdotal evidence that has driven market regulation in the past.

Second, and perhaps more importantly, this paper also shows that the higher levels of information risk suffered by exploration stocks more than likely reflects the fact that mining-exploration companies are less frequently traded, since less frequently traded stocks in other industries are also found to suffer from similar levels of information risk. This is consistent with Tomasic's (1991) inference that insider-trading is likely to be more prevalent in low- capitalisation stocks. This finding should be of comfort to regulators concerned over the public perception that uninformed investors face excessive information risk in mining-exploration stocks.

(Date of receipt of final transcript: June 14, 2005. Accepted by Garry Twite & Doug Foster, Area Editors.)

1. Directors used a number of methods to mislead investors about the true production potential of their mines, including salting reefs with gold fired from shotguns, erecting crushing machinery on poor reefs to persuade investors that the reef is rich and hoarding rich stone for several months and then extracting all the gold in one week. Directors were also not averse to spreading gloomy reports about a company's prospects, accompanied by bear raids on the stock, then buying the stock cheaply from the shareholders they had deceived. Geologists were also more likely to be employed by mining company promoters if they were amenable to preparing exaggerated reports. Stockbrokers were also known to employ informants in mines to obtain first news of important discoveries.

2. The Australian mining boom of 1969-\1972 had its genesis in the boom of the Poseidon NL stock and spread rapidly to other mining exploration stocks, fuelled by investor expectations and the very bullish reports of drill core intersections. Poseidon NL illustrates some of the worst practices of the period. The company's stock was subject to intense speculation over the potential of nickel in drill core samples. This speculation pushed the share price from A$0.20 to A$280 and then back to less than A$1 when the sketchiness of the prospect was made public. The company's directors and consulting geologists purchased shares ahead of the release of the first report of rich assays to the stock exchanges in late 1969. In addition, placements of stock were made to various persons and companies before the information was fully released. The consulting geologists also sold out of Poseidon NL before the unfavourable news was released to the market. No prosecutions were ever made.

3. The JORC Code of 1989 sets out minimum standards, recommendations and guidelines for public reporting of exploration results, mineral resources and ore reserves in Australia and New Zealand. The Code was drawn up by the Joint Ore Reserves Committee of the Australasian Institute of Mining and Metallurgy, the Australian Institute of Geoscientists and the Mineral Council of Australia.

4. Neagle and Tsykin (2001) report that mining companies accounted for 236 (or 25.9%) of the 911 Price Query or Please Explain notices issued by the ASX during the 1999 and 2000 calendar years. The authors find that 80.8% of companies issued a Price Query or Please Explain notice had a market capitalisation of less than $100 million and 76.5% had negative earnings. Although the authors do not provide any breakdown between mining exploration and mining production companies, one suspects that exploration companies are heavily represented since negative earnings is a prominent feature of these companies.

5. The mining company stocks involved in the market manipulation cases were Reef Mining NL, Diversified Mineral Resources NL and Diamond Rose NL. The mining company stocks involved in the insider trading cases were Mt Kersey Mining NL and Carpenter Pacific Resources NL. The defendants in the Mt Kersey Mining NL case were subsequently acquitted of the charge of insider trading.

6. Brown and Foo (1998) find that resource stocks-defined as stocks of companies in ASX industry groups 01 to 05-account for 454 (or 50.8%) of 894 directors' trades in the authors' sample. The authors find that directors' sales of resource stocks are profitable but not their purchases. Similar results are found for the non- resource stock sample.

7. Mean daily turnover is calculated over days reporting non- zero turnover.

8. The matching process was very precise, with the vast majority of matches taking place well within the allowable tolerances. For example, 230 of the 271 matched pairs (84.9%) had a difference in mean daily turnover of less than 5%.

9. For example, the Gold Industry (industry group 01) has subgroups 011 Gold Producer, 012 Gold Explorer, 013 Gold-Other Mining, 014 Gold-Oil, 015 Gold-Copper and 016 Gold-Investment.

10. Easley, Hvidkjaer and O'Hara (2002) develop a five-parameter version of the PIN model in which the arrival rates of uninformed investors differs between the buy and sell sides of the market. The six-parameter model employed in this paper an extension of this five- parameter model.

11. I also estimate the five-parameter PIN model of Easley, Hvidkjaer and O'Hara (2002). The results are not reported above but were similar to those reported for the four-parameter and six- parameter PIN models. Specifically, exploration stocks have higher PINs than production stocks but the same pattern is evident in the control-stock samples. Furthermore, the PINs of the mining and non- mining stocks exhibit the same pattern across the turnover quintiles. In addition, I estimate the four-parameter model using intraday data. The PIN model is estimated using the number of buyer- initiated and seller-initiated trades measured over 30-minute intervals. Not surprisingly, the estimates of α, μ and ε all decline. Although PINs are higher under this procedure, the patterns in the results are the same as found when using daily data.

12. In an order-driven market such as the ASX, the market spread is defined as the lowest limit sell or ask price in the limit order book less the highest limit buy or bid price in the limit order book.

References

Aitken, M. & Frino, A. 1996, 'The determinants of market bid ask spreads on the Australian stock exchange', Accounting and Finance, vol. 36, no. 1, pp. 51-63.

Australian Securities and Investment Commission, 1997/98 to 2001/ 02, Annual Reports.

Australian Stock Exchange, 2002, ASX Listing Rules.

Blainey, G. 1969, The Rush That Never Ended, Melbourne University Press, Melbourne.

Blose, L.E. & Shieh, J.C.P. 1995, 'The impact of gold price on the value of gold mining stock', Review of Financial Economics, vol. 4, no. 2, pp. 125-39.

Brockman, P. & Chung, D.Y. 1999, 'Bid-ask spread components in an order-driven environment', Journal of Financial Research, vol. 22, no. 2, pp. 227-46.

Brockman, P. & Chung, D.Y. 2000, 'Informed and uninformed trading in an electronic, order-driven environment', Financial Review, vol. 35, no. 2, pp. 125-46.

Brown, W.O. Jnr. & Burdekin, R.C.K. 2000, 'Fraud and financial markets: The 1997 collapse of the junior mining stocks', Journal of Economics and Business, vol. 52, no. 3, pp. 277-88.

Brown, P. & Foo, M.S. 1998, 'Insider-trading in Australia: evidence from directors' trades', working paper, University of Western Australia.

Copeland, T. & Galai, D. 1983, 'Information effects and bid-ask spreads', Journal of Finance, vol. 38, no. 5, pp. 1457-69.

Easley, D., Hvidkjaer, S. & O'Hara, M. 2002, 'Is information risk a determinant of asset returns', Journal of Finance, vol. 57, no. 5, pp. 2185-221.

Easley, D., Kiefer, N. & O'Hara, M. 1996, 'Cream-skimming or profit-sharing? The curious role of purchased order flow', Journal of Finance, vol. 51, no. 3, pp. 811-33.

Easley, D., Kiefer, N., O'Hara, M. & Paperman, J. 1996, 'Liquidity, information and infrequently traded stocks', Journal of Finance, vol. 51, no. 4, pp. 1405-36.

Easley, D., O'Hara, M. & Paperman, J. 1998, 'Financial analysts and information-based trade', Journal of Financial Markets, vol. I, no. 1, pp. 175-201.

Easley, D., O'Hara, M. & Saar, G. 2001, 'How stock splits affect trading: A microstructure approach', Journal of Financial and Quantitative Analysis, vol. 36, no. 1, pp. 25-51.

Easley, D., O'Hara, M. & Srinivas, P. S. 1998, Option volume and stock prices: Evidence on where informed traders trade', Journal of Finance, vol. 53, no. 2, pp. 431-66.

Heidle, H.G. & Huang, R.D. 2002, 'Information-based trading in dealer and auction markets: An analysis of exchange listings', Journal of Financial and Quantitative Analysis, vol. 37, no. 3, pp. 391-424.

Mackenzie, B. W. & Doggett, M.D. 1992, Economics of Mineral Exploration in Australia, Centre for Resource Studies (Queen's University, Kingston, Ontario) and Australian Mineral Foundation (Glenside, South Australia).

Neagle, A.-M. & Tyskin, N. 2001, ' 'Please Explain'. ASX share price queries and the Australian continuous disclosure regime', Research Report. Centre for Corporate Law and Securities Regulation, University of Melbourne.

Report of the Senate Select Committee on Securities and Exchange, 1974.

Stoll, H. 1978, 'The pricing of dealer services: an empirical study of NASDAQ stocks', Journal of Finance, vol. 33, no. 4, pp. 1153-72.

Stoll, H. 1989, 'Inferring the components of the bid-ask spread: theory and empirical tests', Journal of Finance, vol. 44, no. 1, pp. 115-34.

Tufano, P. 1998, 'The determinants of stock price exposure: financial engineering and the gold mining industry', Journal of Finance, vol. 53, no. 3, pp. 1015-52.

Tomasic, R. 1991, Casino Capitalism? Insider-trading in Australia, Australian Institute of Criminology, Canberra.

Twite, G. 2002, 'Gold prices, exchange rates, gold stocks and the gold premium', Australian Journal of Management, vol. 27, no. 2, pp. 123-40.

by

Russell Poskitt [dagger]

[dagger] Department of Accounting and Finance, University of Auckland, Private Bag 92019, Auckland. Email: r.poskitt@auckland.ac.nz

Data provided by SIRCA (Securities Industry Research Centre Asia Pacific) on behalf of ASPECT and ASX. I thank Freddy Davison for programming assistance. This paper has benefited from the comments of seminar participants at the University of Auckland, Australasian Finance and Banking Conference 2003 and Irish Accounting and Finance Association Conference 2004. All remaining errors and omissions are the author's responsibility.

Appendix

Appendix

Copyright Australian Graduate School of Management Dec 2005

Source: Australian Journal of Management

More News in this Category