Accounting Firm Rankings

Document Sample
Accounting Firm Rankings Powered By Docstoc
					The information content of Insider trading on future firm performance-
Evidence from New Zealand 1994-2005




                                           by:
                           Allan Hodgson and James McHardy
                               Amsterdam Business School
                                University of Amsterdam

                                   Contact information:
                                  a.c.hodgson@uva.nl or
                                    j.c.mchardy@uva.nl



                                  Preliminary first draft
                    Please do not quote without contacting the authors




                                        Abstract


This paper adds to the limited research compiled on insider trading (IT) trends and
profitability in New Zealand. We investigate the relationship between the firm’s future
earnings, future returns and director’s share trading, focusing on the signaling effect
directors trades have on the quality and persistence of accruals.


At an aggregate level we find that firms with income increasing accruals (POSACC) will
experience a 2.9% increase on there return on assets (ROA) if accompanied in the
preceding year by net insider buying. A similar result is found for insider selling which
suggests that diversification and liquidity based trading distorts the signaling power of
insider sales.




                                                                                            1
We find no significant association between ROA and IT in firms with income decreasing
accruals (NEGACC) only that accruals are less persistent when accompanied by net
insider buying.


At a disaggregated level we find that firms with POSACC will experience a 3.7%
increase in ROA, if insider buying comes in the form of associated party transactions.
Insider selling from directors who are direct beneficiaries delivers a similar result,
suggesting most of the liquidity and diversification effects are captured in this trade
classification. We find no significance between ROA and IT in firms with NEGACC, only
lower persistence in the firms NEGACC if the insider buying comes from associated and
beneficial parties.


Interestingly we find no significance for non beneficial transactions on either ROA or the
persistence in accruals. The latter is a relevant finding for the regulatory bodies in New
Zealand who are currently scrutinizing the disclosure requirements and relevancy of
non-beneficial transactions.


Despite the return associations with insider buying and selling in POSACC firms, there is
little evidence that a profitable trading strategy could be formed around such signals.
Portfolios formed on the direction of IT do not seem to outperform the market when
correctly adjusted for risk. Previous research in New Zealand has failed to adjust
returns for portfolio variance, size and B/M ratios and as such, has overestimated the
size of the abnormal returns from IT.


With little evidence of any consistency in the longer run abnormal returns of even the
most intimate of insiders, it is hard to understand why the regulatory bodies continue to
force companies to report non-beneficial transactions. It is also difficult to comprehend
the relevancy of the recent regulatory changes that now require companies to report all
trades undertaken by the less informed “officers” of the company.




                                                                                             2
1.     Introduction


In this paper we examine whether the presence of IT activity provides any informational
content into the persistence of earnings and of subsequent firm year returns. Specifically
we examine the relevance of IT on earnings conditioned upon the sign of there accruals
and there rankings on market capitalization (MV) and Book/Market value (B/M).
Previous research on IT has shown that insiders have been able to generate excess
returns, but the results vary between countries and the research has tended to focus on the
sign and the size of the trade.   Little research has been complied on the reasons why
insider‟s trade and whether these trades possess some form of information advantage.


The key to any research on IT is in the ability to distinguish between those trades that are
directional and those that are not. Portfolio diversification, liquidity based sales and
transactions for employee share schemes (ESS) hinder the sell side analysis while
managerial overconfidence and the tendency to hang on to losing trades hinder the buy
side. If insiders possess any information advantage then it‟s likely to come in their
assessment of the quality and persistence of accruals versus the market forecast.
Accruals consist of past allocations and future estimates which are unique to a firm.
They are often treated differently depending on which accounting methods are employed
and often subject to manipulation. Despite wide evidence of the accruals anomaly
(Sloan 1996, Barth & Hutton 2004) Pincus (2005) believes the anomaly is not a global
phenomenon, but a function of the institutional, legal and accounting structures of each
country, specifically countries which have a common law tradition, weak shareholder
protection and low share ownership concentration. With the New Zealand institutional
setting typifying the characteristics set out by Pincus, the presence of IT should help to
provide greater clarification as to the quality and/or persistence of accruals.




                                                                                             3
This paper extends on the framework laid out by Hodgson & Praag (2006) and Benish &
Vargus (2002). We analyze IT at an aggregate and disaggregated level, the latter split
into 3 classifications, (i) Beneficial trading (ii) Associated Party Trading (iii) Non
Beneficial Trading. These groupings as disclosed in the company‟s annual report and
allow for a theoretical partition between “directional” and “non-directional" trades.


The results show that future earnings as measured by the firms return on assets (ROA)
will typically be higher if accompanied in the previous year by net insider buying.
There is no significance on next years ROA for insider selling with diversification and
liquidity based trading distorting the insider selling signal. The partition into directional
and non directional trading, surprisingly, does not improve the relationship with next
years ROA relative to the simple aggregate measure.


The disaggregated measures of IT only load significantly when interacted with accruals.
ROA will be 3.7% higher for firms with income increasing (POSACC) accruals if this is
accompanied by associated party buying (IBAP). Similarly ROA will be 3.5% higher for
firms with income increasing (POSACC) accruals if this is accompanied by beneficial
selling. The insider selling result is driven by the fact that the vast majority of selling
transactions occur in larger more profitable firms. Diversification, liquidity and ESS
trades dominate these portfolios and remove any of the relevance as a signaling tool.


For firms with income increasing accruals (NEGACC) both associated party and
beneficial party buying leads to less persistence in the NEGACC component and a
subsequent rise in next years ROA. All non beneficial (directional) trading and all
insider selling variables for NEGACC firms load insignificantly on next years ROA.
Similar inconsistencies are witnesses in the return series. Positive unadjusted abnormal
returns tend to be associated with NEGACC firms who have experience associated party
and beneficial party buying in the previous year. These firms are traditionally small
firms with high B/M values and hence constitute much higher risk. Previous research
has failed to correctly measure this risk and as such there results over-state the size of
abnormal returns for IT in New Zealand.



                                                                                              4
The results show that the vast majority of IT is not a significant lead indicator on future
earnings or returns. One needs to partition heavily before any discernible associations are
found and the ability to have constructed a feasible and profitably trading strategy based
upon these signals seems most unlikely.


Surprisingly the partition into directional and non directional trading added no additional
significance over and above an aggregated measure of insider trading. This is a relevant
finding for the regulatory bodies in New Zeland who have been receiving complaints
about the administrative burden company‟s face having to report trades of “officers” of
the company and non beneficial transactions. The results suggest there is little market
relevance in the disclosure of these types of trades.


The remainder of the paper is as follows.     Section 2 outlines prior research on IT while
section 3 outlines the New Zealand institutional setting from 1994-2005. Section 4
explains the methodology employed, while section 5-6 provide and explanation on the
models employed and report on there results.


2.     Previous Research


There is a wealth of international research on IT. The vast majority of this research is
concentrated in the United States, which is a highly liquid and strongly regulated market.
All but a few highlight the profitability of insider trading.


The consensus of opinion is that IT was and continues to be, a profitable area for
company executives. The early work of De Vere & Pratt (1970), Jaffe (1974), Finnerty
(1976), showed that in the short run insiders could earn abnormal returns although the
sample periods studied were limited in that no insider transaction dates were supplied to
the SEC prior to 1965. Baesel and Stein (1979) showed that shares bought by affiliated
insiders, who included bank officers and directors, generated abnormal long-term
performance. Outside directors earned the highest abnormal returns before selling and
avoided the greatest losses by timing their sells. The authors also found evidence that


                                                                                              5
some insiders were profiting by violating securities legislation. In particular they found
an abnormal volume of insider sales prior to „bad‟ news and no decrease in insider
purchases prior to „good‟ news.

Madden & Gombola (1983), Nunn et al (1983), Seyhun (1986), showed that insider
purchases tended to be more profitable than insider sales. Jeng, Metrick and Zeckhuser
(2003) corroborated this evidence by showing that insider purchases earned abnormal
returns of more than 6% per year, while insider sales earned no significant abnormal
returns. There research also highlighted that investors or “outsiders” could earn
abnormal excess returns by implementing trading strategies based on the insider-trading
signal. This sparked debate at the time with the “efficient market” proponents.
Givoly and Palmon (1985) found that the majority of the abnormal performance of IT
was due to price changes arising from the information revealed through the trade itself.
Momentum trading generated the abnormal returns not the subsequent disclosure of
specific news about the company to which the insiders might have been privy. There
findings also suggested that there was a low incidence of IT in anticipation of impending
new disclosures. Rozeff & Zaman (1988) undertook to explain how it was possible
outsiders could earn abnormal profit when forming portfolios based upon IT signals.
They showed that the profit was driven by the size and earnings/price anomalies. After
controlling for these factors, outsider profits dropped by half and the additional
assumption of a 2% transactions cost made outsider profits zero or negative. Insider
profits were also reduced to a modest 3% per annum after assuming a 2% transactions
cost. In a related paper Rozeff & Zaman (1988) found that iinsider transactions were not
random across growth and value stocks. They found that insider buying rose when stocks
change from growth to value categories. Insider buying was also greater after low stock
returns, and lower after high stock returns. There findings are consistent with a version of
overreaction which says that prices of value stocks tend to lie below fundamental values,
and prices of growth stocks tend to lie above fundamental values.




                                                                                             6
My research also confirms that much of the abnormal return previously reported in earlier
research on IT in New Zealand is driven by size and B/M effects. Once correctly
measured, abnormal returns are difficult to distinguish and hint at being only contrarian
in nature. This is consistent with the findings of Seyhun (1992) who showed that insiders
were more likely to buy (sell) shares following periods of significant price declines
(increases) consistent with the expectation of subsequent price reversals. This is also
consistent with the results from Ferreira (1995), Lakonshok and Lee (2001), McNally &
Smith (2003), Piotroski and Roulstone (2005) who showed insiders tended to be
contrarian investors.


Ferreira (1995) showed that IT activity in bear markets was characterized by decreases in
insider sales and increases in purchases, consistent with the view that those markets are
followed by improved economic conditions. Conversely, insider sales increased and
purchases decreased in bull markets, consistent with the view that inferior market
conditions tended to follow those periods. McNally & Smith (2003) found in the month
around their trades, insiders bought on dips and sold on rallies.


Only a handful of papers have found no significant abnormal returns for insiders. Eckbo
& Smith (1998) estimated the performance of insider trades on the Oslo Stock Exchange
(OSE) during a period of lax enforcement and found zero or negative abnormal
performance by insiders. A similar result was found by Hamill, Mcllkeny and Opong
(1998) for the U.K. market but the consensus remains that IT is fruitful ground for
abnormal returns. John and Lang (1991) analyzed IT data around dividend
announcement dates and concluded that an insider‟s knowledge came from both there
access to price sensitive information and there analytical skill.


Meulbroek (1992) found that prior to the announcement of a takeover, 43% of the price
run-up in the target‟s stock occurred on IT days and Summers & Sweeney (1998) found
insider transactions related to knowledge about internal fraud. In related work Watson
and Young (1998) found that trading activity before takeover announcement was greater
by non-executive directors.



                                                                                            7
The seminal work by Sloan (1997) on the accruals anomaly has lead recent research into
the sources of IT advantage, specifically insider‟s intimate knowledge of the quality of
the firm‟s accruals. Aboody and Lev (2000) found superior abnormal returns for R&D
intensive firms relative to non R&D firms. Beneish and Vargus (2002) revealed that
income-increasing accruals appeared to be overpriced when managers were involved in
above average selling and were fairly priced when managers were involved in above
average buying. They found that investors overpriced income-increasing accruals when
firm's top executives did not trade, but that overpricing was less severe when managers
engaged in above average selling. In all they concluded that investors failed to interpret
insider-selling information correctly because it was difficult to distinguish informational
motivated selling from liquidity-motivated selling. Ke Huddart and Petroni (2003)
showed evidence that insider‟s trade on the knowledge of upcoming accounting
disclosures for up to two years, prior to the disclosure. Stock sales by insiders increased
three to nine quarters prior to a break in a string of consecutive increases in quarterly
earnings. Insider stock sales are greater for growth firms, when the prior string of
earnings increases is longer, and before a longer period of declining earnings.


Unlike the United States and to a lesser extent Europe, Australian and New Zealand
executives have historically received, far less remuneration in the form of stock or stock
options. This reduces the importance of liquidity driven selling. Brown and Foo (1997)
found that director‟s sales in Australia were profitable but not there purchases, contrary to
international findings. Director‟s sales generally occurred after abnormal stock price
rises (contrarian trading) and sales were substantially higher for resource stocks. Unlike
the U.S. results the authors found no direct relationship between IT profitability and firm
size. Anand, Brown & Watson (2002) concluded that directors‟ sales signalled an
expected deterioration in accounting performance measures, particularly in the longer
term, but there was no relationship for their purchases. The relationship for sales was
also stronger and more significant for larger trades and for smaller firms.




                                                                                              8
New Zealand research is even more scant than Australia, with the pre-emanate work
concentrated in only a handful of authors. Duncan and Etebari (1990, 1997), studied IT
in the years 1986-1993 and found evidence of excess returns for insiders on price run-ups
prior to corporate announcements. Casey and Tourani-Rad (2001) examined IT from
1993-1999 and found over a 250 day holding period that insiders earned a 15.6%
abnormal return when they purchased and generated a 11.8% loss when they sold.


Etebari, Tourani-Rad, & Gilbert (2003) revealed that New Zealand‟s two-tiered
disclosure regime (section 3) lead to superior returns for delayed disclosures relative to
immediate disclosures. There results showed that insiders earned significant profits for
both purchases and sales. Of the 2453 insider transactions from January 1995-December
2001 including 939 substantial shareholder trades they found an abnormal gain of 6.5%
for purchases and 5.9% loss for sales for the 250 trading days following the transaction.
Delayed disclosures generated an 8.9% abnormal gain for purchases and a 7.7% loss for
sales, while immediate disclosures returned a 4.1% and -4.5% abnormal returns
respectively. However size played a major factor in these results. Over the 250 days
following the transactions, small firms and non-NZSE 40 firms posted large abnormal
returns of 21.5% and 11.6%, respectively. Large firms and NZSE 10 firms both lost 1.6%
and 1.3%, respectively although these returns were statistically insignificant. From the
results, they concluded that smaller and less-researched companies accounted for a
disproportionately large share of the abnormal returns reported for delayed disclosure
purchases.


There results are not surprising as they are the result of the market model employed by
the authors to generate the expected return series (equation_ 1)                                   The market model used
assumes that the expected return for the firm is a function of its covariance with the
market (β1), the return on the market and an intercept term (α1). In theory α1 should
pick up the idiosyncratic risk of the firm i.e. the risk of the firm over and above that
implied by its beta.
1.   If over the 190 day trading window prior to the trade the firms stock has fallen in price the expected return for the stock over for
     the following 250 days would be negative!




                                                                                                                                       9
For a perfectly specified model with low residual error, α1 should capture all
idiosyncratic risk that is unique to the firm and not captured by beta e.g. B/M, size,
liquidity, momentum. If equation 1 is mis-specified, as it is for smaller firms, then all the
IT AR‟s are dubious at best.


Equation_1.                                         E( Rit ) = α1 + β1Rmt + et                      Equation_2                               ARit = Rit – E( Rit )
Rit = Average cum dividend portfolio return                                     ,   β1 =Security Beta , Rmt = Return on                        market portfolio




To illustrate, Etebari, Tourani-Rad, & Gilbert calculated individual stock betas from daily
price covariance with the NZSE40 over a 190 day window, starting 250 days prior to the
director‟s trade. For small less liquid stocks that rarely trade, the covariance with the
market can be very low and even close to zero for such a short trading window. (1)
If the market model systematically understates the beta risk of these smaller stocks then
the expected return for these firms will be lower than there risk implies and consequently
there AR‟s substantially overstated.

Figure_1 Daily price changes of Allied Work Force, Telecom NZ and NZX_all share index
January 2006-2007


                                                             25,00%                                                                                    25,00%

                                                             20,00%                                                                                    20,00%

                                                             15,00%                                                                                    15,00%
   Allied Work Force daily % changes




                                                             10,00%                                                                                    10,00%
                                                                                                       Telecom daily % changes




                                                              5,00%                                                                                     5,00%

                                                               0,00%                                                                                     0,00%
                                       -3,00%   -2,00%   -1,00%    0,00%    1,00%   2,00%   3,00%                                -3,00%   -2,00%   -1,00%    0,00%    1,00%   2,00%   3,00%
                                                             -5,00%                                                                                    -5,00%

                                                            -10,00%                                                                                   -10,00%


                                                            -15,00%                                                                                   -15,00%

                                                            -20,00%                                                                                   -20,00%

                                                            -25,00%                                                                                   -25,00%
                                                          NZX_all daily % changes                                                                   NZX_all daily % changes




                                                                                                                                                                              10
Figure_1 shows the daily price changes of a smaller capitalised stock versus a larger
capitalised stock. The market model is a poor predictor of daily price changes for the
smaller capitalised stock. The Market model implied return from a 270 day random
sample are calculated below and show that the less risky stock (TEL) demands a higher
daily return. For a 1% rise in the NZX_all the market model implies a subsequent rise of
0.4% for the small riskier stock and a 1.9% rise for the less risky stock. In fact if the
association is close to zero, which might be the case for the smallest less liquid stocks,
then the expected return for small portfolios will be understated by the entire equity risk
premium.


E(R_awf) = -1,94 + 0,44_RM     rsqr 0.4%    E(R_tel) = -0,002 + 2,14_RM rsqr 56.7%


If IT portfolios were more heavily weighted to small firms with high B/M ratios relative
to the market portfolio a simple Market Model approach could understate expected return
and hence overstate AR‟s. As shown in section 5, IT portfolios in New Zealand have
historically been skewed toward small firms with high B/M ratios and hence are riskier
than that implied by a simple market model.     It is not surprising then that Etebari,
Tourani-Rad, & Gilbert find high AR‟s for small stocks and negative returns for the
larger stocks which are better specified by the market model.


Drew, Marsden & Veeraraghavan (2005) looked at NZ firms from 1990-2002 and found
a negative relationship between firm‟s size and a stocks idiosyncratic volatility. In other
words firm size may proxy for idiosyncratic risk factors in the Fama & French asset
pricing models. Employing the Fama & French 3 factor model they found that small
firms tended to have positive slopes on SMB and large firms had diminishing positive or
negative slopes on SMB. They also found that portfolios of stocks with the highest
idiosyncratic volatility generated higher average returns than portfolios of stocks with
low idiosyncratic volatility. Firms with high idiosyncratic volatility tended to have high
betas and generated low earnings on book equity hence requiring a higher rate of return.
The use of Sharp ratios in section 5 is an attempt to capture this risk and ensure a more
equitable assessment of performance.



                                                                                             11
The authors ranked stocks based on MV and found that portfolios of the smallest stocks
generated the highest average returns compared to the average returns of the portfolios of
largest stocks.
Bryant & Eleswarapu (1997) and Vos & Pepper (1997) established similar B/M effects in
New Zealand although there sample period biases the results. Pinfold et al (2001) found
that portfolios with the highest B/M ratio produced a return 5.3% higher than the
portfolio with the lowest ratio although the authors pointed out that constructing tradable
portfolios in New Zealand based upon size and B/M was virtually impossible in reality.
The short sample period and extreme volatility of the return series meant that the results
lacked significance and the size of the market made liquidity an issue. What is clear from
all the limited work on the Fama & French approach and idiosyncratic risk is that
portfolios in New Zealand are substantially more risky than the traditional asset pricing
models would suggest.                  Results which rely solely on a simple market model to generate
AR‟s must be treated with some caution.




3.         New Zealand Regulatory Environment

Securities Amendment Act (SAA) 1988-2002
The institutional setting in New Zealand from 1994-2003 provides fertile ground for
research on IT activity. Similar to the institutional setting in Australia, the New Zealand
market despite being highly deregulated suffered from weak regulatory enforcement for
IT breaches. Coupled with a market heavily populated with small enfant companies and
a two tier disclosure regime the market should have been ripe for IT abuses. (2)
New Zealand's IT laws were built on three sources. The most important was Part 1 of
the Securities Amendment Act 1988. "Inside information" was defined as information
that was not publicly available but would be likely to affect the price of a security if it
were publicly available.


2. Given the dualist nature of the institutional setting pre and post 2004 any research into IT needs to be wary of this breakpoint. The
results and portfolios derived from the 1994-2004 samples may not be directly transferable into the current trading environment.




                                                                                                                                     12
An “Insider" was defined as a director, employee or substantial security holder, who had
inside information about the public issuer, or who had inside information about another
public issuer. The definition also covered people who were outside the company but
also received in confidence inside information about the public issuer.


Substantial Security Holding (SSH)
Under section 20(1) of the SAA every person who had a substantial security holding
(SSH) in a public issuer had to give notice to the public issuer and to the New Zealand
stock exchange (NZX) when they reached 5% or more of the voting securities of that
public issuer or body (2). In addition, under section 22(1) a SSH also had to give notice
when there holdings changed by more than 1% of the total number of issued voting
securities. Both SSH disclosures were generally reported to the NZX within 5 business
days. In contrast and despite the relatively wide and strongly worded definitions of
“insider” within the Securities Amendment Act 1988, company ddirectors only had to
disclose their security trading in the annual report, while company executives who were
non-board members were not required to disclose their trading at all. Given that NZ
firms were only required to report annually under the SAA the vast majority of investors
were oblivious to insider transactions for the majority of the financial year.


Complicating matters further for investors was a weak enforcement regime from the NZ
Securities and Exchange Commission (NZSEC). The law placed responsibility on the
party harmed by an insider, often the party on the other side of the trade, to file and prove
some form of improprietory. The difficulty in building a case coupled with the litigation
costs and the delaying tactics employed by the issuing company saw only 23 cases
brought to the NZSEC over the 14 years 1992-2005.


Securities Market Amendment Act 2002-2006
On 1st December 2002, the Securities Market Amendment Act (SMAA) 2002 was passed
and this came into effect on the 1st March 2004. Under the SMAA directors and officers
of public issuers were required to disclose, to both the NZX and the public issuer, their
relevant interests in the firm and any subsequent purchases/sales in those securities within



                                                                                            13
5 trading days of the transaction.   Under the Act, an "officer" in a listed company
included people who reported directly to the board of directors. The definition also
extended to lower tiers of responsibility and included people who managed a principal
business unit or a division of the public company.


This lead to an enormous increase in the amount of IT disclosure in 2005 and consequent
calls from industry participants for a review of the SMAA on the grounds that it had
become ineffective at improving transparency. In May 2006 Commerce Minister Lianne
Dalziel told the Securities Law Update conference in Wellington that the Government
had received feedback from share market participants that disclosure obligations were
much harder for companies to comply with than was originally anticipated. Specifically
the concern lay in relation to “officer” disclosure. Industry participants had told her the
definition of officer was too wide under the act and was acting as a disincentive for
companies listing here. A review of the SMAA is under way and will attempt to look at
the information relevancy of officer disclosures and non-beneficial transactions. The
results in this paper support the calls from the Business community that there is little
informational relevance in non beneficial trades and in fact little in the way of
informational relevance in the vast majority of insider trading. As such the regulatory
bodies could consider removing non beneficial trading from there disclosure
requirements and almost certainly consider removing “officer” disclosures.


4.     Sample and Methodology


The informational content of IT is hard to gauge as a simple sale or purchase can be a
noisy representation of the expectation of the firm‟s future performance. With a larger
percentage of executive compensation coming in the form of stock grants or stock
options in recent years, sales from insiders are often a result of cashing up for liquidity
reasons rather than a signal of deteriorating future firm performance. Extracting insider
sales for liquidity or portfolio diversification reasons is next to impossible which means
that many of the transactions are likely to distort the signaling power of insider trading.




                                                                                              14
This is also not restricted to sales alone. Insider buying also presents challenges in that
many insiders trade or a reversal of fortunes, believing firmly that there own managerial
expertise or vision will deliver higher future returns. The challenge is in extracting these
types of trades away from trades where managers may have tangible knowledge of future
profitability.


Hodgson, Praag, Cocx (2006) were able to divide director‟s trades on the Dutch market
into treasury and executive related trades and found that there was much more signaling
power in executive trading once treasury sales had been controlled. A similar approach is
employed here by sorting insider trades by there balance sheet classification, an approach
that will hopefully remove the non-informational noise alluded to in Tourani-Rad and
Gilbert (2004). We would expect to see the highest information content in beneficial and
associated party transactions and little to no informational content in non beneficial
transactions. Despite the fact there was never any legal requirement under the SMAA to
classify trades in the annual reports; most firms did so, reporting IT as either a beneficial,
non beneficial or associated party transaction. There was also never a formal definition
from either the NZX or the Securities and Exchange Commission for each of the 3
classifications but a general consensus did exist.


Associated Party (AP) interest was commonly understood to be a director sharing an
interest with a mutually beneficial third party e.g. a spouse or business partner. Some of
the associated party trading in New Zealand was the result of the establishment of family
or related party trusts, which was fuelled by gift duties and student loans means testing
back in the late 1990‟s. Non-beneficial (NB) related to those directors trading on behalf
of third party entity where the director had no obvious financial connection. Many of the
non-beneficial director‟s trades were a result of being executors to non-beneficial trusts
or trustees of their company‟s employee share schemes. Beneficial interest (B) related to
any transactions where the director was personally likely to benefit.




                                                                                             15
Market and Accounting Data
The market data was obtained via DataStream and the New Zealand Stock exchange
(NZX) while the accounting data was compiled from Bloomberg, DataStream and the
Investment Research Group Ltd (IRG).                          IRG is New Zealand's foremost provider of
information, analysis and advice about investments both locally and in Australia. It was
formed from the merger in October 2002 of Datex Services Ltd and McEwen & Co Ltd.
IT data from 1995-2003 was hand collected from individual company annual reports and
collected from the NZX post 2003.                     (3)   The entire universe of stocks listed on the NZX
over 1994-2005 was 1637 firm year observations and once stocks were removed which
had no record of IT or accounting data the sample was reduced to 1016 firm year
observations. (Table_1).                The use of average assets to scale the earnings components
and the removal of 1% of the accruals and cash flow outliers left a final sample of 891
firm year observations. Accruals were scaled by average assets along with earnings
before interest and tax (EBIT) and operating cash flow (OCF).                                  (4)   This essentially
converted EBIT into a ROA measure.                          (5)



In section 5 firms were assigned a rank of SMALL, MEDIUM, LARGE depending upon
which third there MV fell into. A similar sorting into LOW, MEDIUM, and HIGH was
done for firms based on B/M values.




3.   There were 403 trades where price was never declared in the annual report and 900 trades were a trade date was never
     specified. This was generally the result of reconciling the directors trading reports in the executive summaries with the opening
     and closing balances provided in the footnotes to the financial accounts.
4.   EBIT = Earnings before interest and tax, OCF = Operating Cash Flow, Accruals = EBIT – OCF, ACC_TA= Accruals /
     average total assets, OCF_TA= OCF / average total assets,      EBIT_TA= EBIT / average total assets
5.   It should be noted that given EBIT is the earnings metric used in this analysis, impairment, special items and changes in
     amortisation and depreciation schedules could introduce possible noise into the dependent variable. If impairment charges are
     transitory then EBIT may understate the true financial position of the firm with next year’s earnings recovering to pre
     impairment levels. Changes in amortisation and depreciation schedules may also distort EBIT.




                                                                                                                                   16
The return series could suffer from some survivorship bias. Some of the firms that were
excluded from the sample are firms that have since de-listed. The return of the market
portfolio is therefore likely to have been lower had the entire 1637 firms been included.
However including the returns of these companies without knowledge of there IT would
have severely biased the results. Additionally the reduced sample has a wide
representation of industry and size and the use of portfolio returns relative to the market
portfolio circumvents the survivorship issue somewhat.


If we were to err on the side of caution one might expect the full sample to have had a
lower average return and to have witnessed more insider buying and selling on average in
firms with deteriorating financials and returns. As such the reduced sample is likely to
overstate the significance of the positive association with insider buying on portfolio
ROA‟s and AR‟s.      It is also likely to have overstated the losses for IT portfolios formed
on insider selling. However this only reinforces the key findings in the paper that the
vast majority of insider trading looks to have little informational content.


Market and Accounting Data Summary
Of the 891 firm year observations negative accruals (NEGACC) accounted for 44% of
the firm year observations while positive accruals (POSACC) accounted for the balance.
Aggregate net insider buying (selling) was observed in 37 %( 18%) of firm years, while
no IT was observed in 45% of firm years.
What is striking about the New Zealand data is the acute differential between large and
small companies. 26% of firm year observations had market capitalisation <25mio, 71%
<200mio, while 5% of the sample was >2bio. (Table_2)


Larger firms were on average consistently more profitable than smaller firms. (Table_3)
The distribution of returns shows a significant skew to the downside for smaller firms and
this would have been more acute had the full sample been employed. (Figure_1) As will
be elaborated further, size is a clear proxy for risk in the New Zealand market.




                                                                                           17
            Table_1 Sample firms and market capitalization relative to all NZX 1995-2005


            Year               1994       1995       1996        1997         1998    1999     2000     2001    2002     2003     2004     2005   Total
            Sample_firms        49         49         56          62           65      76       81       96      106      122      128      126   1016
            NZX_total           131        137        127         126          129     123      139      133     132      144      162      154   1637
            %_mktcap          86,1%      77,0%      80,0%       81,4%        79,9%   85,3%    76,1%    84,4%   88,2%    82,4%    89,1%    87,3%   62%


            Table_2 Distribution of Market Capitalisation of New Zealand firms 1994-2005

            MV       <25m            25-50m         50-100m 100-200m 200-500m 500-750m 750m-1bio >1bio                          >2bio    >5bio
             % sample 26%              13%            13%     19%      12%       5%        2%     10%                             5%       2%
            MV       <25m            <50m           <100m <200m      <500m    <750m    <1biom    >1bio                          >2bio    >5bio
            cum_%     26%              39%            52%     71%      83%      88%       90%     10%                             5%       2%



            Table_3 Average Earnings, OCF and Accruals ranked by market capitalisation

             ROA           1995    1996        1997       1998        1999       2000       2001    2002     2003     2004       2005 average
             Small        11,3%   -0,3%        0,1%       1,8%       -5,6%       1,6%      -0,9%   -19,8%   -10,0%   -13,4%     -8,0%    -3,9%
            Medium        10,3%   14,3%       12,5%       9,8%       12,1%      11,3%      10,4%   10,1%    11,5%    13,9%       7,4%   11,3%
             Large        13,7%   12,1%       11,3%      10,1%       10,8%       8,9%       6,6%    9,4%    11,8%    11,9%      13,3%   10,9%
            F.ROA
             Small        13,4%     4,1%       -2,9%      -0,1%       -4,7%     -3,0%       0,9%   -22,5%   -15,0%     -14,5%   -18,2%    -5,7%
            Medium        12,6%    12,0%       12,1%       8,0%       11,6%     13,2%       4,5%    -5,2%     9,4%      12,5%    11,0%     9,2%
             Large        14,3%    13,3%       11,3%      10,5%       10,1%      8,8%       6,3%     7,8%    13,2%      11,5%    13,2%    10,9%


            Figure_1 EBIT_TA distribution for small, medium and large firms
              25
              20
Frequency




              15
              10
               5
               0




                   -1.5            -1              -.5                  0             .5
                                  Full sample ROA distribution small firms




                                                                                                                                   18
IT summary statistics
Similar to the results from Tourani-Rad, Gilbert & Wisniewski (2005) we find more
purchases than sales transactions due to the low level of option use as a form of executive
remuneration. Table_4 shows that 63% of the IT was in the form of associated,
beneficial or non-beneficial purchases of securities.


Table_4 - Aggregate IT transaction summary 1995-2005

Year     Firms  IBAP IBNB                IBB       ISAP       ISNB        ISB      Total
   1995      49   47    28               117         45        15          68       320
   1996      56   43    21               121         59        64          44       352
   1997      62   43    39               169         38        34          91       414
   1998      65   58    30               176         35        15          43       357
   1999      76   15    30               156         8         15         108       332
   2000      81   40    33               198         13        32         107       423
   2001      96   41    31               260         24        44         136       536
   2002     106   54    33               204         28        34         102       455
   2003     122   60    35               187         19        35          90       426
   2004     128   25    24               136         11        11          82       289
   Total    968 426    304               1724       280        299        871      3904
 % total        10,9% 7,8%              44,2%      7,2%       7,7%       22,3%
IBAP=Insider Buying Associated Party, IBNB=Insider Buying Non Beneficial, IBB=Insider Buying Beneficial
ISAP=Insider Selling Associated Party, IBNB=Insider Selling Non Beneficial, IBB=Insider Selling Beneficial



Etebari, Tourani-Rad & Gilbert (2003) examined a sample of 1254 trades by insiders
from January 1995-December 2000, 793 of which were by directors and 461 by large
block holders of the 33 companies examined over the sample. Our sample is larger with
3904 director‟s trades recorded covering nearly 130 different companies from January
1995 to December 2004. Total trades reached a peak of 536 in 2001 before falling away
to a low of 289 in 2004, as the regulatory changes mentioned in section 2 took effect.
Nearly 70% of all transactions were filed as “beneficial interest” while associated party
and non-beneficial interest accounted for the remaining 30%.




                                                                                                             19
Tourani-Rad, Gilbert & Wisniewski (2005) found a significant negative correlation
between market value (log of market cap) and the number of deals made by managing
directors i.e. as market value increases IT diminishes. This suggests that Managing
directors are less likely to trade in larger firms due in part to the larger investor scrutiny.
This is in contrast to the Hodgson & Praag results that suggested that director‟s disguise
there IT within larger more liquid stocks.


Table_5 shows that 40% of firm year trades took place in mid-tier sized companies as
ranked by market capitalisation. Small and medium sized firms with small and large
companies constituted the remaining 30% each. Table_5 also shows the average ROA
and subsequent years ROA (F.ROA) for portfolios formed on aggregate insider trading.
It is surprising to see that non beneficial (NB) transactions tend to be good lead indicators
of future earnings trends. NB buying is associated on average with the highest ROA and
F.ROA readings (9.5%, 7.9%) while NB selling is in portfolios with the lowest return. If
earnings are a driver of returns portfolios formed on NB buying and selling might be
profitable. The diversification and liquidity effects seem to be strong for insider selling
as the average ROA, F.ROA for ISAP and ISB portfolios are some of the highest.
IBAP earnings are poor relative to the market average. These initial findings suggest IT
portfolios may be less profitable than previously thought.


Table_5 ROA t and ROA t+1 for IT portfolios and IT statistics sorted by MV

Portfolio    ROA     rank F.ROA rank                 MV rank
All Stocks     6,54%      3,91%              Small   Medium     Large
POSACC        11,50%      6,20%
NEGACC         0,28%      0,58%              229       332       262
IB             6,72% 6    5,50%   5
IS             8,28% 2    5,46%   6
IBAP           4,77% 8    5,44%   7           38        60        31
IBB            7,72% 4    5,66%   4           82       125        96
IBNB           9,53% 1    7,94%   1           23        51        45
ISAP           7,70% 5    6,00%   2           22        33        22
ISB            7,85% 3    5,99%   3           45        44        48
ISNB           5,88% 7    1,64%   8           19        19        20




                                                                                              20
2. Regression Models

The regression models follow the research design in Hodgson & Praag (2005) and are
shown in Table_6.     The dependent variable is one year ahead EBIT scaled by average
total assets, which is essentially a traditional a return on assets (ROA) measure.   (21)

Model 1 starts with a simple autoregressive (AR1) process of ROA regressed against its
lag.


Model_1 establishes a benchmark for which to compare the significance of adding insider
trading, size, B/M and earnings component variables to the most naive model.
If insiders possess any information advantage then it‟s likely to come in their assessment
of the quality and persistence of accruals versus the market forecast. Earnings are
therefore decomposed in Model_2 into OCF and ACCTA, with ACCTA further
partitioned in Model_3 into income increasing (POSACC) and income decreasing
accruals (NEGACC).      If accruals are positive (negative) then the interactive dummy
variable takes the value of ACCTA or 0     (ACCTA, 0).      Like prior research we expect
to find that the ACCTA coefficient is less persistent than the OCF coefficient and that the
POSACC coefficient is more persistent than the NEGACC coefficient.


Model_4 adds two IT dummy variables to Model_3.         Insider buying (selling) takes a
value 1 if net IT per firm financial year is positive (negative) or zero otherwise. We
would expect to see the insider buying (selling) coefficient to load positively (negatively)
on future ROA. In other words the presence of insider buying (selling) is confirmation of
a higher (lower) earnings performance next year. Model 5 introduces IT variables subject
to the sign of accruals. If insider buying is present and the firms ACCTA>0 then D_IB_P
takes the value 1 or zero.




                                                                                            21
We would expect to see insider buying (selling) in firms with POSACC as confirmation
that these accruals will persist (reverse) into the future and (holding everything else
constant) result in a higher (lower) future ROA. Conversely we would expect to see
insider buying (selling) in firms with N EGACC as confirmation that these accruals will
reverse (persist) in the future and result in a higher (lower) future ROA.
We expect insider buying in NEGACC firms to be a weaker signal of improved future
earnings performance relative to POSACC as the former has a larger % of poor
performing firms relative to the POSACC portfolio. (Table_5)


Model 6 introduces the product of the insider buying dummy variables in model 2 with
the accrual vectors POSACC and NEGACC. V_IB_P (V_IS_P) takes the value of the
vector of POSACC if insider buying (selling) was present, otherwise 0.
V_IB_N (V_IS_N) takes the value of NEGACC if insider buying (selling) was present or
zero. The interactive terms provide a measure of the relative strength of the IT signal
subject to the ratio of accruals to average total assets(ACCTA) In other words do the IT
signals have a stronger influence on the firms ROA given a higher(lower) ratio of
ACCTA.


Models 7-9 replicate the earlier models 4-6 but divide the IT variables into the three
partitions, Beneficial Party (B), associated Party (AP) and Non Beneficial (NB). We
would expect to see greater significance for the finer partitions relative to the aggregate
measure specifically a significant loading of the AP and B coefficients and no
significance for the NB coefficients.




                                                                                              22
Table_7 Regression Models of EBIT_TA on earnings components and insider trading

 Model_ 1      ROA t+1 =  0 +  1_ROA t +  2_SIZE t + Vt


 Model _2      ROA t+1 =  0 +  1_SIZE t +  2_OCF_TA t +  3_ACCTA t + Vt


 Model_ 3      ROA t+1 =  0 +  1_SIZE t +  2_OCF_TA t +  3_P t +  4_N t + Vt


 Model_ 4      ROA t+1 =  0 +  1_SIZE t +  2_OCF_TA t +  3_P t +  4_N t +
                          5_D_IB t +  6_D_IS t + Vt


 Model_ 5      ROA t+1 =  0 +  1_SIZE t +  2_OCF_TA t +  3_P t +  4_N t +
                          5_D_IB_P t +  6_D_IS_P t +  7_D_IB_N t +  8_D_IS_Nt + Vt


 Model_ 6      ROA t+1 =  0 +  1_SIZE t +  2_OCF_TA t +  3_P t +  4_N t +
                             5_V_IB_P t +  6_V_IS_P t +  7_V_IB_N t +  8_V_IS_Nt + Vt


ROA = Earnings before interest and tax / average total assets, Size = Log (Market Capitalization)
OCF_TA = Operating Cash flow / average total assets, ACCTA = EBIT-OCF / average total assets
P = vector of positive accruals. Takes value of Accruals_TA if Accruals_TA > 0, 0
N = vector of negative accruals. Takes value of Accruals_TA if Accruals_TA < 0, 0
D_IB = Net Insider buying dummy variable. Takes value 1 if firm year net IT was > 0, 0
D_IS = Net Insider selling dummy variable. Takes value 1 if firm year net IT was < 0, 0
D_IB_P = dummy variable taking value 1 if there is net insider buying for firms with POSACC, 0
D_IS_P = dummy variable taking value 1 if there is net insider selling for firms with POSACC, 0
D_IB_N = dummy variable taking value 1 if there is net insider buying for firms with NEGACC, 0
D_IS_N = dummy variable taking value 1 if there is net insider selling for firms with NEGACC, 0
V_IB_P = D_IB * V_POSACC, V_IS_P = D_IS * V_POSACC




                                                                                                    23
3.        Regression Results

     Table_8 Regression Results of ROA on earnings components and aggregate insider
     trading

     ROA t+1        Model_1      sig  Model_2 sig Model_3 sig Model_4 sig Model_5                          sig   Model_6      sig
     no,obs              758               758          758          758         758                               758
     adj_rsq          33,9%             38,5%        38,9%        39,2%        39,6%                              41,6%
     constant         -0,312    0,000   -0,248 0,001 -0,285 0,000 -0,297 0,000 -0,297                      0,000  -0,281     0,000
     ROA t             0,661    0,000
     Sizet             0,017    0,000      0,013 0,001        0,014 0,000       0,014   0,000   0,014      0,000   0,014     0,000
     B/Mt             -0,011    0,253     -0,008 0,452       -0,003 0,749      -0,004   0,688   -0,005     0,659   -0,005    0,610
     OCF_TA                                0,978 0,000        1,004 0,000       0,996   0,000   0,994      0,000   1,008     0,000
     D_IB                                                                       0,029   0,064
     D_IS                                                                       0,024   0,209
     ACCTA                                 0,442 0,000
           P                                                  0,666 0,000       0,665 0,000     0,616      0,000   0,489     0,006
           N                                                  0,362 0,006       0,360 0,006     0,349      0,010   0,673     0,000
     D_IB_P                                                                                     0,028      0,063
     D_IS_P                                                                                     0,056      0,000
     D_IB_N                                                                                     0,030      0,198
     D_IS_N                                                                                     -0,022     0,547
     V_IB_P                                                                                                        0,326     0,116
     V_IS_P                                                                                                        0,640     0,001
     V_IB_N                                                                                                        -0,613    0,007
     V_IS_N                                                                                                        -0,378    0,323

     Vuong Statistics Model_ 6 versus Model_ 5,4,3

     EBIT_TA = Earnings before interest and tax / average total assets, Size = Log (Market Capitalization)
     OCF_TA = Operating Cash flow / average total assets, IB = Net Insider buying dummy variable. Takes value 1 if firm year net IT
     was > 0, 0
     D_IS = Net Insider selling dummy variable. Takes value 1 if firm year net IT was < 0, 0
     Accruals_TA = EBIT-OCF / average total assets
     V_POSACC = vector of positive accruals. Takes value of Accruals_TA if Accruals_TA > 0, 0
     V_NEGACC = vector of negative accruals. Takes value of Accruals_TA if Accruals_TA < 0, 0
     D_IB_P = dummy variable taking value 1 if there is net insider buying for firms with POSACC, 0
     D_IS_P = dummy variable taking value 1 if there is net insider selling for firms with POSACC, 0
     D_IB_N = dummy variable taking value 1 if there is net insider buying for firms with NEGACC, 0
     D_IS_N = dummy variable taking value 1 if there is net insider selling for firms with NEGACC, 0
     D_IBB = Net Beneficial (B) Insider buying dummy variable. Takes value 1 if firm year B buying was > 0, 0
     D_IBAP = Net Associated Party (AP) dummy variable. Takes value 1 if firm year AP buying was > 0, 0
     D_IBNB = Net Non Beneficial (NB) Buying dummy variable. Takes value 1 if firm year NB buying was > 0, 0
     D_ISB = Net Beneficial (B) Insider selling dummy variable. Takes value 1 if firm year B selling was > 0, 0
     D_ISAP = Net Associated Party (AP) dummy variable. Takes value 1 if firm year AP selling was > 0, 0
     D_ISNB = Net Non Beneficial (NB) Selling dummy variable. Takes value 1 if firm year NB selling was > 0, 0
     V_IB_P = D_IB * V_POSACC, V_IS_P = D_IS * V_POSACC, V_IB_NEGACC = D_IB * V_NEGACC
     V_IS_NEGACC = D_IS * V_NEGACC, V_ISB_P = D_ISB * V_POSACC, V_IBB_N = D_IBB * V_POSACC
     V_IBAP_N = D_IBAP * V_NEGACC




                                                                                                                               24
Table_9 Regression Results of ROA on earnings components and disaggregated insider
trading

ROA t+1        Model_ 7 sig Model_ 8 sig Model_9 sig
no,obs              758          758          758
adj_rsq          39.2%        39.3%        40.9%
constant         -0.302 0.000 -0.293 0.000 -0.285 0.000
ROA t
Sizet               0.014    0.000      0.014 0.000        0.014 0.000
B/Mt               -0.004    0.736     -0.004 0.698       -0.008 0.477
OCF_TA              0.997    0.000      1.007 0.000        0.991 0.000
D_IBB               0.013    0.394
D_IBAP              0.022    0.223
D_IBNB             -0.004    0.791
D_ISB               0.011    0.505
D_ISAP              0.028    0.348
D_ISNB             -0.025    0.427
P                   0.666    0.000      0.640     0.000   0.598 0.000
N                   0.359    0.006      0.344     0.010   0.603 0.000
D_IBAP_P                                0.037     0.002
D_ISB_P                                 0.036     0.001
V_ISB_P                                                    0.498 0.004
V_IBAP_N                                                  -0.553 0.003
V_IBB_N                                                   -0.452 0.070

Vuong Statistics Model_ 6   versus Model_ 5,4,3
Vuong Statistics Model_ 7   versus Model_ 4
Vuong Statistics Model_ 8   versus Model_ 5
Vuong Statistics Model_ 9   versus Model_ 6


EBIT_TA = Earnings before interest and tax / average total assets, Size = Log (Market Capitalization)
OCF_TA = Operating Cash flow / average total assets, IB = Net Insider buying dummy variable. Takes value 1 if firm year net IT
was > 0, 0
D_IS = Net Insider selling dummy variable. Takes value 1 if firm year net IT was < 0, 0
Accruals_TA = EBIT-OCF / average total assets
V_POSACC = vector of positive accruals. Takes value of Accruals_TA if Accruals_TA > 0, 0
V_NEGACC = vector of negative accruals. Takes value of Accruals_TA if Accruals_TA < 0, 0
D_IB_P = dummy variable taking value 1 if there is net insider buying for firms with POSACC, 0
D_IS_P = dummy variable taking value 1 if there is net insider selling for firms with POSACC, 0
D_IB_N = dummy variable taking value 1 if there is net insider buying for firms with NEGACC, 0
D_IS_N = dummy variable taking value 1 if there is net insider selling for firms with NEGACC, 0
D_IBB = Net Beneficial (B) Insider buying dummy variable. Takes value 1 if firm year B buying was > 0, 0
D_IBAP = Net Associated Party (AP) dummy variable. Takes value 1 if firm year AP buying was > 0, 0
D_IBNB = Net Non Beneficial (NB) Buying dummy variable. Takes value 1 if firm year NB buying was > 0, 0
D_ISB = Net Beneficial (B) Insider selling dummy variable. Takes value 1 if firm year B selling was > 0, 0
D_ISAP = Net Associated Party (AP) dummy variable. Takes value 1 if firm year AP selling was > 0, 0
D_ISNB = Net Non Beneficial (NB) Selling dummy variable. Takes value 1 if firm year NB selling was > 0, 0
V_IB_P = D_IB * V_POSACC, V_IS_P = D_IS * V_POSACC, V_IB_NEGACC = D_IB * V_NEGACC
V_IS_NEGACC = D_IS * V_NEGACC, V_ISB_P = D_ISB * V_POSACC, V_IBB_N = D_IBB * V_POSACC
V_IBAP_N = D_IBAP * V_NEGACC




                                                                                                                          25
Model_ 1 - ROA in New Zealand is mean reverting as the lagged coefficient (1) is
equal to 0.66. This is similar to the coefficients in Hodgson & Praag (0.67), Tourani-Rad
(0.71) and is within the confidence interval set down by Pincus (2004) of 0.6-0.8. The
adjrsqr is 33.9%.


Model _2 - As with prior research OCF is more persistent than accruals, or more simply
accruals mean revert much quicker than OCF. The OCF coefficient is very high at 0.98
but not dissimilar to the Tourani-Rad coefficient of 0.95. The accruals coefficient of
0.44 is lower than the Tourani-Rad coefficient of 0.54 a result of capturing a wider
sample of stocks with a lower average ROA. The adjrsqr of 38.5% and the V-stats show
this is an improved model over the simple AR1.


Model_ 3 - POSACC or income-increasing accruals are more persistent (0.67) than
NEGACC (0.36) and similar to the Hodgson & Praag coefficients of 0.65 and 0.35. The
result makes economic sense as the pressure to maximize revenues and the transitory
nature of loss making ensures NEGACC are a short run phenomenon. There is a small
but insignificant increase in adjRsqr to 38.9%.


Model_ 4 - Only the IB dummy loads significantly with a p value of 0,068. Next years
ROA will be 2.9% higher if accompanied by insider buying in the current financial year.
The IS dummy despite being positive is insignificant and in line with previous research,
while the model had a significant incremental increase in adjRsqr to 39.2%.


Model_ 5 - The IB and IS dummy variables for POSACC load significantly. Firms with
POSACC could expect to see an additional 2.8 % (5.6%) in ROA if accompanied by
insider buying (selling). (4) The IS loading makes little economic sense and is likely the
result of the liquidity and diversification effects mentioned earlier. There is a significant
incremental improvement in adjRsqr to 39.6%




                                                                                           26
Model_ 6 - NEGACC are substantially less persistent (0.673-0.613<0.362) if
accompanied by insider buying i.e. income decreasing accruals are likely to completely
reverse next year if insider buying is witnessed. The only other significant coefficient is
that POSACC are more persistent when accompanied by insider selling. The IB
coefficient for POSACC firm‟s accruals is significant at around 90% and shows that
POSACC are more persistent when accompanied by insider buying(0.489+0.326>0.616).
Both Interactive terms show that the higher the ratio of accruals/average assets the higher
the sensitivity of future ROA is to insider buying. The adjRsqr has improved
significantly to 41.6%.


Model_7 - None of the disaggregated dummy variables load significantly in the
regression unlike Model_ 2 where the insider buying coefficient was significant. This is
a surprising result given that we would have expected to see Beneficial and Associated
Party transactions loading significantly. Directors could be trying to disguise there
profitable trades by commingling transactions between classifications or might be as
simple as that the vast majority of trades are not great earnings indicators.


Model_ 8 - Only the POSACC IT dummies load significantly. Firms with income
increasing accruals on average will see a 3.6% rise on there ROA if accompanied in the
preceding year by associated party buying.                           Likewise Firms with income increasing
accruals will see a 3.5% rise on there ROA if accompanied in the preceding year by
associated party selling. A lack of significance for insider buying in NEGACC or less
profitable firms suggests many directors trades might be driven by hope and expectation
rather than knowledge of impending profitability.




4.   The full regression includes 16 possible regressors s but table_9, only reports only the significant coefficients.




                                                                                                                          27
Model_ 9 - The interactive model has NEGACC being significantly less persistent given
AP buying in the previous year (0.598-0.553< 0.362). In other words the higher the ratio
of NEGACC / average total assets the greater the sensitivity insider buying has on ROA.
Firms that exhibit above normal insider buying whilst experiencing a high NEGACC
ratio could be involved in earnings management but such an assertion requires additional
research. NEGACC is significantly less persistent if accompanied by beneficial party
buying in the previous year (0.581-0.433< 0.351). (5)                               The only significant selling
coefficient was for beneficial transactions, where POSACC were more persistent if
accompanied by insider selling. This suggests that there is little information content in the
selling transactions of insiders on ROA even after removing the noise of non-beneficial
trades. The liquidity and diversification effects seem to be dominant. When there is no
trading POSACC (NEGACC) are less (more) persistent suggesting a no trading sign is a
weak negative signal for future year ROA.


IT Portfolio construction and returns
The return series is based on annualised total returns for each individual stock i.e. annual
dividend plus capital gains/losses for the year. Portfolio returns are based on a non-
compounded $1 investment held over a single financial year. At the end of every
financial year portfolios are closed down and marked to market resulting in a series of
annualised returns. Average annual portfolio returns for the entire holding period are
calculated by taking the average annual portfolio return for each individual year from
1995-2005. Average returns and variances are therefore the results of 10 annualised firm
year observations. Portfolios constructed on insider buying (selling) would require
taking a long (short) position in the securities of firms where insider buying (selling) was
prevalent. (6)




5.   p-value is 0.07
6.   The short portfolio returns in reality should also be adjusted for holding costs.




                                                                                                                   28
                  n                                 n
Average Return = ∑ Rit              (1)        AR = ∑ Rit – Rf                (2)
                         t=1                           t=1

Abnormal returns are calculated by taking the gross portfolio return and subtracting the
1-year Treasury bill rate. Performance is measured by comparing the Sharpe ratios of
each portfolio with a portfolio of similar risk.                 (7)



Sharpe ratio = AR / σ (AR)


Portfolio Returns and Risk Rankings for IT
Portfolio performance analysis is not a simple process. The numerator (abnormal return)
and the denominator (risk) are prone to measurement error and there is no common
ground as to whether risk should be measured via the abnormal return component the
scaling factor, or both. The Sharpe ratio measures a unit of excess portfolio return per
unit of risk and as such depends on the excess return and the variance being measured
correctly. (23) Return analysis between portfolios is therefore only possible if each
portfolio of returns is compared against a portfolio of identical risk.


A simple market model approach to measuring abnormal return will under (over)
estimate the expected returns if the portfolio is more (less) heavily weighted to small,
value stocks vis-à-vis the market portfolio. (8) Failure to capture size and B/M effects
will result in risk misspecification and spurious results. Portfolios of small (large), value
(growth) stocks are likely to be riskier (less risky) and hence should be compensated for
via higher expected returns or scaled by there higher expected variance.               Similar to the
results in Pinfold et al (2001) we find no consistent excess returns for firms sorted on
size.
Table_11 shows high volatility for small firm returns and evidence that these portfolios
are significantly more risky than portfolios of large firms.



7.   Sharpe ratios cannot be used as a performance measure when returns are negative




                                                                                                    29
8.   Value stocks are stocks with high B/M values and hence are deemed to be good value and hence likely to generate a higher
     return.

There is evidence of a B/M effect for NZ stocks and if excluded from the abnormal
returns would clearly distort any inference about performance. The adjustment to returns
can come via factor loadings of SMB and HML on a market model type approach similar
to Etebari, Tourani-Rad, & Gilbert and/or by weighting returns relative to the
idiosyncratic risk of the portfolio.


Table_11 Average returns ranked on size and B/M value

               40.0%


               30.0%


               20.0%


               10.0%


                0.0%


               -10.0%


               -20.0%
                        1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

          Small_cap     16.6% 10.3% -12.9 21.0% 37.9% 31.0% 1.6% 2.1% 38.9% 22.2%
          Medium_cap 25.6% 16.1% -12.6 26.5% 13.4% -1.0% 9.6% 8.8% 28.8% 9.8%
          Large_cap     26.3% 11.0% 12.0% -11.3 32.7% 3.3% 23.4% 5.1% 24.4% 28.3%



                50.0%

                40.0%

                30.0%

                20.0%

                10.0%

                 0.0%

               -10.0%

               -20.0%

               -30.0%
                        1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

         Low_B_M        9.0% 22.5% -5.2% 32.3% 30.8% 1.2% -1.2% -2.4% 15.1% 9.0%
         Medium_B_M 16.3% 21.1% -2.9% 24.6% 12.2% 13.7% 2.2% 7.9% 44.4% 26.3%
         High_B_M       28.4% -0.4% -26.7 28.4% 12.1% 37.6% 11.5% 28.4% 37.3% 15.1%




                                                                                                                         30
The use of “risk rankings” in Table_12 is a means of identifying some of the
characteristics of each of the IT portfolios. Each firm is assigned a yearly rank from 0-
100, based upon there market capitalisation, leverage (Debt/Equity) and B/M values. The
market portfolio by construct is then expected to average 50 for each of the risk
categories (2). The risk ranking for each portfolio based on size (SR), leverage (LR) and
B/M is measured by taking the average of each yearly rank. By comparing the average
portfolio rank with that of the market portfolio rank, we can get some insight as to the
potential exposure each portfolio had to size, leverage and B/M.


Table_12 Portfolio returns and Risk Rankings for IT Portfolios

 Portfolio        Market(long) IB_all         IBAP        IBB       IBNB Market(short) IS_all             ISAP       ISB       ISNB
 Cum_return          150,1%       155,2% 174,7% 135,2% 140,5%                    -150,1%      -173,1% -142,8% -187,6% -127,8%
 average              15,0%        15,5%      17,5%      13,5%      14,1%         -15,0%       -17,3% -14,3% -18,8% -12,8%
 stdev                12,6%        19,3%      17,9%      16,8%      18,8%          12,6%       18,6%      24,3%     22,3% 21,5%
 AR_avg                8,5%         9,0%      10,9%       7,0%       7,5%         -21,6%       -23,9% -20,8% -25,3% -19,3%
 stdev_AR             12,9%        19,8%      17,8%      17,3%      18,2%          12,4%       18,5%      23,9%     22,6% 21,5%
 Sharpe_ratio          0,66         0,45       0,61       0,40       0,41           n/a          n/a       n/a       n/a         n/a
 SR                    50,4         51,7       48,7       52,2       56,3           50,4        51,3       49,8      51,0          52,3
 LR                    47,9         49,3        5,1       49,2       55,3           47,9        49,6       52,0      51,5          46,9
 BMR                   50,4         52,7       58,0       53,1       54,9           50,4        51,9       49,9      51,8          61,0
 obs                    815          325        129        303        118           815          152        77       137         58
Cum_return = Cumulative average annualised gross return 1995-2005
Average = average annualised gross return 1995-2005
Stdev = standard deviation of the average annualised returns 1995-2005
AR_avg = Portfolio average annualised return – I year risk free rate
Stdev_AR = standard deviation of AR_avg
Sharpe_ratio = AR_avg / Stdev_AR
Risk rankings assign a value from 0 (lowest) to100 (highest) for all firm year observations based upon there balance sheet ranking at
the end of every financial year. Portfolio rankings are the weighted average of each firm year observation within the portfolio.
SR = market capitalisation ranking.
LR = Leverage ratio ranking.
BMR = Book/Market ranking.
Obs = number of securities in each portfolio.




                                                                                                                                   31
At an aggregate level, abnormal returns for insider buying (IB_all, 7.8%) and insider
selling (IS_all, -23.9%) under performed the market long and short portfolios (9.1%, -
22.2%). Both portfolios had higher variances (18.3%, 18.5%) relative to the market
portfolio (14%, 13.7%) so there risk adjusted performance was inferior.
Relative to the 50.5 size rank of the market portfolio the IB_all rank of 52.4 and the
IS_all rank of 52.1 indicate a heavier skew to larger companies, which according to Fama
& French would demand a lower return.


The size factor looks to be a stronger influence than the offsetting skew that both the
IB_all (51.3) and the IS_all (51) portfolios had to value companies. These results
highlight the fact that if the risk/return characteristics are not correctly specified
erroneous conclusions can be drawn.      (9)




The same caution also needs to be exercised when analysing the performance of the
disaggregated portfolios. The average returns for the IBAP portfolio (18.2%)
outperformed the market portfolio by 2.6% and delivered a superior Sharpe ratio of 0.68
versus the market Sharpe of 0.65. However The IBAP portfolio looks to have a skew
toward small (0.495), leveraged (0.518), value stocks (0.558) and would traditionally
demand a higher expected return than that implied by the market model.


The average return for the IBB portfolio underperformed the market by a surprising 1.8%
and had an inferior Sharpe ratio of 0.42. However it is difficult to gauge the degree of
underperformance as the portfolio had a higher weighting to large stocks (52.9) which are
traditionally less risky and an offsetting higher weighting to value stocks (51.4)
Of the portfolios formed on insider selling, the ISAP portfolio outperformed the market
by an average 1.4% a year but it had a significantly higher variance (23.9%) (3). The
return performance might also be explained by the fact that the ISAP portfolio had a
higher weighting to growth stocks (48.5%) which traditionally demand a lower return.




                                                                                           32
9.   The average leverage ranking for the Market portfolio of 47.8 reflects the fact that many firms had no long-term debt hence a
     larger number of 1 observations relative to 100.

The most surprising result was for the ISB portfolio which underperformed the market
portfolio by 3.1% despite having a higher weighting to large (51.9), growth stocks (49.9)
It also experienced significantly higher variance relative to the market portfolio
confirming that diversification and liquidity rather than profitability drives insider selling.
The vast majority of Non Beneficial trading was the result of transactions surrounding
employee share schemes. As a consequence one would have expected to have seen a
heavier skew for Non Beneficial trading toward larger more mature companies. The
mature companies in NZ also tend to be more “brick and mortar” types, which are likely
to be associated with value stocks. Its not surprising then the both the ISNB and IBNB
portfolios had a heavier concentration in large value stocks.


The IBNB portfolio underperformed by 1.6% with a heavy skew toward larger
companies (57.1) and a small skew toward value firms (51.4).
The ISNB portfolio was the best performing short portfolio outperforming the market by
2.9% a year, although it experienced much higher volatility (21.5%). The return
performance was even more surprising given the ISNB portfolio had a sharp skew to
value stocks 60.4.


Fama and French size weighted portfolios
In order to try and measure the effect that size and B/M had on the true risk adjusted
performance of insider trades we sort firms based on the approach employed by Fama &
French (1996). (10) Firms are sorted by size (market capitalisation) and B/M ratios. Size
is divided into three sorts; small, medium, large while B/M is also divided into three
sorts; low, medium, high.




10. Fama and French sorted industrial firms annually by size and by B/M. For each year the size premium was constructed as the
     difference in return between small and large firms. SML is constructed as the difference in returns of an equally weighted long
     position in the 3 small firm portfolios and equally weighted short positions in the 3 big firm portfolios. SMB=1/3(SL+SM+SH)-
     1/3(BL+BM+BH). The B/M effect is calculated from the difference in returns between high B/M ratio and low B/M ratio firms.




                                                                                                                                  33
     HML is constructed as the difference in returns between an equally weighted long position in the high B/M portfolio and an
     equally weighted short position in the low B/M portfolio HML=1/3(SH+MH+LH)-1/3(SL+ML+LL)

The intersections of each of theses sort‟s results in nine portfolios; small/low,
small/medium, small/high, medium/low, medium/medium, and medium/high, large/low,
large/medium, and large/high. The average returns and abnormal returns of each of these
9 portfolios are shown in Table_13.


Table_13 Portfolio returns partitioned on size and book/market

Portfolio_size      Market         SL         SM          SH   ML              MM         MH          LL          LM     LH
average              15.6%       14,4%       19,2%      23,6% 7,3%            14,8%      16,9%      15,9%        16,2% 10,9%
US average                        12,6%       15,6%      18,4%                                       10,7%        12,5% 16,1%
stdev                 13,8%      45,0%       22,1%      28,2% 18,8%           13,3%      24,1%      18,2%        16,1% 28,8%
AR                    9,1%       10,5%       15,3%      19,7% 3,4%            10,9%      13,0%      12,1%        12,3% 7,1%
stdev_AR              14,0%      46,9%       23,9%      26,6% 22,1%           17,5%      22,0%      20,0%        16,3% 29,2%
Sharpe_ratio           0,65       0,22        0,64       0,74  0,16            0,62       0,59       0,60         0,75  0,24
Average = average annualised gross return 1995-2005
US average = annual average annualised gross return 1929-1997
AR_avg = Portfolio average annualised return – I year risk free rate
Stdev_AR = standard deviation of AR_avg
Sharpe_ratio = AR_avg / Stdev_AR



As per the vast majority of literature, returns in NZ were higher for value stocks relative
to growth and for smaller firms relative to larger firms albeit not linear.                               (11)

Normally we would expect to see a higher return from smaller stocks vis-à-vis larger
stocks given the same B/M ranking and lower returns as we moved from high to lower
B/M rankings, similar to the U.S. results provided.                          Performance for the IT portfolios
given the size and B/M effects measured by comparing the Sharpe ratios of each portfolio
relative to the appropriate Sharpe ratio for the Fama and French benchmark portfolio. e.g.
Sharpe ratios of portfolios formed on Insider buying in small/high (SH) firms will be
compared with the Sharpe ratios of the market portfolio of total small/high firms. The
results are presented in Table_14.


Despite IT portfolios delivering higher abnormal returns relative to the market portfolio,
it becomes less clear when adjusted for size and B/M. Of the 54 portfolios only 8 or
15% outperform their benchmark. There is no discernible size or B/M trend that would
lend any support to a systematic trading strategy based on IT signals. At a stretch one
might conclude that IBAP and IBB signals in mid to small sized growth stocks might be


                                                                                                                                  34
profitable as well as ISAP signals in mid tier growth stocks. However without firm
economic rationale for this it might be a case of data mining.


One could argue that associated party trades could be the result of directors trying to
disguise there trades in family trusts or associated party holdings. The skew toward
growth firms could be a function of the fact that directors better understand the growth
prospects of the firm vis-à-vis the market. If this were the case we would expect to see
some of this informational advantage surface in the accruals portion of earnings,
specifically positive accruals. Table_15 shows insider-trading portfolios formed on the
sign of accruals.




11.The only anomalies were that the LL return of 15.9% is higher than the SL return of 14.4% and the LH return of 10.9% is lower
than the returns of the LL (15.9%) and LM (16.2%) portfolios.




                                                                                                                               35
Table_14 Fama and French sorted IT Portfolio Performance 1995-2005


IBAP                   SL          SM            SH          ML            MM       MH        LL       LM       LH
àverage              15,7%        26,5%        33,8%        14,0%         11,8%    16,7%    25,2%   16,5%     9,5%
stdev                51,1%        28,5%        46,3%        39,5%         16,5%    31,9%    42,2%   18,1%     34,6%
AR_avg               9,1%         19,9%        27,3%        7,4%          5,2%     10,1%    18,6%   10,0%     2,9%
stdev_AR             51,4%        28,2%        46,0%        39,2%         16,7%    32,8%    42,5%   18,0%     35,0%
Sharpe_ratio          0,18         0,71         0,59         0,19          0,31     0,31     0,44     0,56     0,08
obs                     9            9           20           11            24       23       13       11        7
IBB                    SL          SM            SH          ML            MM       MH        LL       LM       LH
àverage              26,5%        24,3%        22,3%        16,3%         14,7%    3,7%     16,1%     6,3%    -1,6%
stdev                66,1%        32,7%        31,4%        22,4%         27,4%    29,4%    23,8%   26,9%     35,4%
AR_avg               20,0%        17,7%        15,7%        9,7%          8,1%     -2,9%    9,6%     -0,2%    -8,1%
stdev_AR             66,1%        32,7%        31,4%        22,4%         27,4%    29,4%    23,8%   26,9%     35,4%
Sharpe_ratio          0,30         0,54         0,50         0,43          0,30    -0,10     0,40    -0,01    -0,23
obs                    16           26           39           16            46       38       38       37       20
IBNB                   SL          SM            SH          ML            MM       MH        LL       LM       LH
àverage              31,7%        9,4%         45,9%        -4,1%         8,6%     4,5%     19,9%   32,7%     -2,7%
stdev                51,7%        23,8%        39,3%        34,6%         13,3%    41,4%    19,0%   51,0%     38,2%
AR_avg               25,2%        2,9%         39,4%       -10,7%         2,1%     -2,1%    13,3%   26,1%     -9,2%
stdev_AR             51,7%        23,8%        39,3%        34,6%         13,3%    41,4%    19,0%   51,0%     38,2%
Sharpe_ratio          0,49         0,12         1,00        -0,31          0,15    -0,05     0,70     0,51    -0,24
obs                     7            4           11           13            21       15       17       16       13
ISAP                   SL          SM            SH          ML            MM       MH        LL       LM       LH
àverage             -17,1%       -12,3%       -33,9%        3,9%         -17,1%   -17,1%   -28,5%    -8,7%    -7,0%
stdev                71,8%        26,4%        49,9%        38,9%         16,1%    24,7%    18,9%   35,2%     32,1%
AR_avg              -10,6%        -5,8%       -27,4%        10,5%        -10,6%   -10,6%   -21,9%    -2,1%    -0,4%
stdev_AR             71,8%        26,4%        49,9%        38,9%         16,1%    24,7%    18,9%   35,2%     32,1%
Sharpe_ratio         -0,15        -0,22        -0,55         0,27         -0,66    -0,43    -1,16    -0,06    -0,01
obs                     9            7            9            9            14        6       13        5        4
ISB                    SL          SM            SH          ML            MM       MH        LL       LM       LH
àverage             -34,3%       -17,7%       -29,2%       -15,1%        -10,0%   -20,8%   -14,3%   -15,6%    -3,9%
stdev                85,2%        40,0%        36,3%        40,2%         29,3%    27,8%    27,0%   16,5%     43,8%
AR_avg              -27,8%       -11,2%       -22,6%        -8,5%         -3,4%   -14,3%    -7,8%    -9,1%    2,6%
stdev_AR             85,2%        40,0%        36,3%        40,2%         29,3%    27,8%    27,0%   16,5%     43,8%
Sharpe_ratio         -0,33        -0,28        -0,62        -0,21         -0,12    -0,51    -0,29    -0,55     0,06
obs                    10           15           19           15            11       13       26       16        9
ISNB                   SL          SM            SH          ML            MM       MH        LL       LM       LH
àverage             -13,1%       -19,2%        3,9%        -20,2%         -8,4%   -30,2%   -20,5%   -20,1%   -16,1%
stdev                46,0%        33,3%        43,0%        24,3%         20,8%    34,2%    28,4%   14,1%     48,0%
AR_avg               -6,5%       -12,7%        10,4%       -13,6%         -1,8%   -23,7%   -14,0%   -13,6%    -9,5%
stdev_AR             46,0%        33,3%        43,0%        24,3%         20,8%    34,2%    28,4%   14,1%     48,0%
Sharpe_ratio         -0,14        -0,38         0,24        -0,56         -0,09    -0,69    -0,49    -0,96    -0,20
obs                     1            3            1            5             7        6       10        5        6

Average = average annualised gross return 1995-2005
Stdev = standard deviation of the average annualised returns 1995-2005
AR_avg = Portfolio average annualised return – I year risk free rate
Stdev_AR = standard deviation of AR_avg
Sharpe_ratio = AR_avg / Stdev_AR
Obs = number of securities in each portfolio.




                                                                                                                 36
Table_15 Portfolio returns for IT Portfolios sorted by accruals
Portfolio               Market       POSACC     NEGACC      IB_POSACC      IB_NEGACC      IS_POSACC     IS_NEGACC
Cum_return            156,3%        145,9%      168,1%       131,5%         185,6%        -202,5%        -132,7%
average                15,6%        14,6%        16,8%        13,1%         18,6%          -20,3%         -13,3%
stdev                  13,8%        14,5%        16,4%        16,7%         23,7%          19,9%           26,3%
AR                     9,1%          8,0%        10,3%        6,6%          12,0%          -13,7%          -6,7%
stdev_AR               14,0%        14,4%        16,9%        16,8%         24,2%          20,3%           26,0%
Sharpe_ratio            0,65         0,56         0,61         0,39          0,50           -0,67          -0,26
SR                     50,5%        54,1%        44,6%        54,8%         48,9%          52,5%           51,6%
LR                     47,8%        53,8%        40,0%        56,6%         41,8%          50,3%           47,7%
BMR                    50,4%        49,2%        52,0%        48,4%         55,2%          49,1%           53,5%
obs                     815           485         330          193            129             87             61
Cum_return = Cumulative average annualised gross return 1995-2005
Average = average annualised gross return 1995-2005
Stdev = standard deviation of the average annualised returns 1995-2005
AR_avg = Portfolio average annualised return – I year risk free rate
Stdev_AR = standard deviation of AR_avg
Sharpe_ratio = AR_avg / Stdev_AR
Risk rankings assign a value from 0 (lowest) to100 (highest) for all firm year observations based upon there balance sheet ranking at
the end of every financial year. Portfolio rankings are the weighted average of each firm year observation within the portfolio.
SR = market capitalisation ranking.
LR = Leverage ratio ranking.
BMR = Book/Market ranking.
Obs = number of securities in each portfolio.


Absolute returns are the highest in NEGACC firms (16.8%) and specifically in negative
accrual firms where insider buying (18.6%) and insider selling (-13.3%) were prevalent.
Both portfolios outperformed their benchmarks by 1.8% and 5.5% respectively but
neither of the portfolios Sharpe ratios outperformed their benchmarks. NEGACC
portfolios tended to be weighted more heavily to smaller, less levered value companies‟
vis-à-vis the market portfolio, while positive accrual portfolios were weighted more
heavily to larger, more levered growth stocks.




                                                                                                                                  37
At the disaggregated level the only portfolio to outperform its benchmark was the IBAP
portfolio for NEGACC which had an abnormal return of 18.9% and a Sharpe ratio of
0.78, which was higher than the benchmark Sharpe for the total negative accrual portfolio
of 0.61.       The portfolio was weighted to smaller sized (45.9%) value stocks (60.1%).
The failure to find any return superiority in the positive accruals portfolios suggests there
is little evidence that insiders are trading on some form of information asymmetry based
upon director‟s knowledge of growth prospects for the firm. The fact that most of the
returns are captured in NEGACC firms in mid to small sized value stocks points to
directors being contrarian in nature and that the market might not correctly price small
less profitable firms. Similar to Tourani_Rad (2005) the contrarian trading hypothesis
seems the most plausible and as such there is little support that a systematic trading
strategy based on IT signals would have delivered an appropriately risk adjusted
abnormal return.


Table_17 Portfolio returns for IT Portfolios sorted by Positive accruals


Accruals                POSACC        POSACC     POSACC        POSACC         POSACC         POSACC         POSACC
Portfolio                 All          IBAP         IBB          IBNB           ISAP            ISB           ISNB
Cum_return             145,9%        149,5%      120,4%       131,9%         -155,0%        -189,7%        -200,6%
average                 14,6%        14,9%        12,0%        13,2%          -15,5%         -19,0%         -20,1%
stdev                   14,5%        19,9%        17,7%        17,9%          22,8%          22,4%           30,7%
AR_avg                  8,0%          8,4%        5,5%         6,6%            -8,9%         -12,4%         -13,5%
stdev_AR                14,4%        16,8%        17,9%        17,8%          23,3%          22,8%           22,8%
Sharpe_ratio             0,56         0,50         0,31         0,37             n/a           n/a            n/a
SR                      54,1%        51,6%        54,9%        57,8%          50,7%          52,3%           54,9%
LR                      53,8%        55,8%        55,5%        56,1%            5,6%         52,5%           53,6%
BMR                     49,2%        53,2%        48,2%        45,9%          44,7%          49,4%           61,5%
obs                      485           80          185           76              45            74              36
Cum_return = Cumulative average annualised gross return 1995-2005
Average = average annualised gross return 1995-2005
Stdev = standard deviation of the average annualised returns 1995-2005
AR_avg = Portfolio average annualised return – I year risk free rate
Stdev_AR = standard deviation of AR_avg
Sharpe_ratio = AR_avg / Stdev_AR
Risk rankings assign a value from 0 (lowest) to100 (highest) for all firm year observations based upon there balance sheet ranking at
the end of every financial year. Portfolio rankings are the weighted average of each firm year observation within the portfolio.
SR = market capitalisation ranking, LR = Leverage ratio ranking, BMR = Book/Market ranking, Obs = number of securities in each
portfolio.




                                                                                                                                  38
Table_18 Portfolio returns for IT Portfolios sorted by Negative accruals


Accruals               NEGACC        NEGACC      NEGACC        NEGACC         NEGACC         NEGACC         NEGACC
Portfolio                 All          IBAP         IBB          IBNB           ISAP            ISB           ISNB
Cum_return             168,1%        255,0%      165,8%       223,3%         -110,2%        -255,9%         -88,2%
average                 16,8%        25,5%        16,6%        22,3%          -11,0%         -25,6%          -8,8%
stdev                   16,4%        44,6%        23,3%        32,5%          47,6%          38,6%           42,2%
AR_avg                  10,3%        18,9%        10,0%        15,8%           -4,5%         -19,0%          -2,3%
stdev_AR                16,9%        24,2%        26,0%        44,4%          47,6%          24,2%           38,1%
Sharpe_ratio             0,61         0,78         0,38         0,36            n/a            n/a            n/a
SR                      44,6%        45,9%        49,7%        55,7%          50,7%          51,4%           50,0%
LR                      40,0%        45,2%        41,6%        5,3%           44,2%          50,2%           37,5%
BMR                     52,0%        60,1%        56,2%        61,4%          54,1%          54,1%           58,6%
obs                      330           47          115           41              31            60              22
Cum_return = Cumulative average annualised gross return 1995-2005
Average = average annualised gross return 1995-2005
Stdev = standard deviation of the average annualised returns 1995-2005
AR_avg = Portfolio average annualised return – I year risk free rate
Stdev_AR = standard deviation of AR_avg
Sharpe_ratio = AR_avg / Stdev_AR
Risk rankings assign a value from 0 (lowest) to100 (highest) for all firm year observations based upon there balance sheet ranking at
the end of every financial year. Portfolio rankings are the weighted average of each firm year observation within the portfolio.
SR = market capitalisation ranking, LR = Leverage ratio ranking, BMR = Book/Market ranking, Obs = number of securities in each
portfolio.




                                                                                                                                  39
6. Conclusions

If insiders in New Zealand possessed any private information that they could have been
traded on profitable and without detection then it was likely to have been during the two
tier disclosure regime prior to the regulatory changes of 2003. The fact that the trading
results over this period are inconclusive casts some serious doubt on the commonly held
belief in New Zealand that all insiders trading are profitable.


We find that firms ROA will typically be higher if accompanied in the previous year by
net insider buying but this does not necessarily transpire to higher returns versus the
market when correctly adjusted for risk. There is no significance on next years ROA for
insider selling and no abnormal returns versus the market for portfolios formed on these
selling signals.


The partition into directional and non directional trading does not greatly improve the
relationship with next years ROA relative to an aggregate measure. The disaggregated
measures of IT only load significantly when interacted with accruals.      ROA will be
3.7% higher for firms with income increasing accruals if this is accompanied by
associated party buying. However portfolios formed from POSACC firms where
associated party buying is present do not outperform the market when correctly adjusted
for risk. All non directional trading and all insiders selling variables for the
disaggregated models load insignificantly on next years ROA.


For firms with income increasing accruals (NEGACC) both associated party and
beneficial party buying leads to less persistence in the NEGACC component and a
subsequent rise in next years ROA. Portfolios formed from NEGACC firms where
associated party buying is present generate higher absolute returns and higher Sharpe
ratio than the market portfolio. However these returns need to be adjusted for the
significantly higher weighting these portfolios had to small cap value stocks vis-à-vis the
market and as such there performance is ambiguous. Portfolios more heavily weighted to




                                                                                            40
small cap value stocks are riskier by composition and hence require a higher
compensatory rate of return.


Previous New Zealand research has failed to capture this anomaly and in turn has
significantly overstated IT returns. Additional research opportunities lie in a more
comprehensive analysis of the return series, specifically focusing on the exact effect size
and B/M should have had on expected return. Once the correct Fama and French factor
loadings can be specified, a more comprehensive analysis can be made as to the true risk
adjusted profitability of insider trading in New Zealand.                             (11)




11. The return data would be more comprehensively measured by considering the entire holding period return for each individual
     transaction i.e. from the actual deal date rather than financial year end. If IT returns were heavily influenced by early price
     movements as pre the Givoly and Palmon (1985) findings then the series may exclude significant returns captured in the early
     months of trading. The difficulty lies in the fact that the enlarged sample captures trading by directors where trade date and
     trade price were often not disclosed. A full and comprehensive return series would require additional disclosure information in
     order to avoid reducing the sample.




                                                                                                                                       41
References
1. Anand, A. Brown, P. & Watson, I. 2002- “Directors Trades as signals of their firms future
    financial performance: Evidence from the Australian share market “ Working Paper,
    University of Western Australia
2. Aboody, D. and B. Lev, 2000. “Information asymmetry, R&D, and insider gains.” The
    Journal of Finance, 55, (6), 2747-2766.
3. Amihud, Y. and Mendelson, Y, 1991. “Volatility, Efficiency, and trading: Evidence from the
    Japanese stock market”. The Journal of Finance, 46, 5., 1765-1789.
4. Atiase, R., 1985. “Pre-disclosure information, firm capitalization and security price behavior
    around earnings announcements”. Journal of Accounting Research, 23, 21-36.
5. Baesel, J.B and Stein, G.R., 1979 “The Value of Information: Inferences from the
    Profitability of Insider Trading“- The Journal of Financial and Quantitative Analysis,
6. Barth, M.E and Hutton, A.P 2004- “Analyst Earnings Forecast Revisions and the Pricing of
    Accruals” Review of Accounting Studies,
7. Beneish, M. D. and M. E. Vargus, 2002. “Insider trading, earnings quality and accrual
    mispricing” The Accounting Review, 77, 4, 755–791.
8. Bettis, J.C., J.L. Coles and M.L. Lemmon, 2000. “Corporate policies restricting trading by
    insiders”. Journal of Financial Economics, 57, 2, 191-220
9. Brown, P, S. Taylor and T. Walter, 1998. “The impact of statutory sanctions on the level and
    information content of voluntary corporate disclosure”, Working Paper, Department of
    Accounting, University of Sydney.
10. Brown, P., and Foo, M., 1997. “Insider Trading in Australia: Evidence from Directors‟
    Trades,” Working Paper, Department of Accounting and Finance, University of Western
    Australia
11. Bryant, P. and Eleswarapu, V. 1997, 'Cross-Sectional Determinants of New Zealand Share
    Market Returns', Accounting and Finance, vol 37, pp 181–205.
12. Casey, Q., & Tourani-Rad, A. (2001). “Are Directors Deals Dodgy? Some New Zealand
    Evidence”. Unpublished Paper, Department of Finance, Waikato Management School.
13. Dechow, P.M., (1994). "Accounting earnings and cash flows as measures of firm
    performance: The role of accounting accruals," Journal of Accounting and Economics 18:3-
    42




                                                                                                42
14. Dechow, P.M., R. G. Sloan, A. Sweeney, 1996. “Causes and consequences of earnings
    manipulation: an analysis of firms subject to enforcement actions by the SEC”,
    Contemporary Accounting Research 13, 1–36.
15. Duncan, K., & Etebari, A. 1990 “Insider Trading, Information Leakage and Pre-
    Announcement Price Movements: Evidence from the New Zealand Markets”. Pacific Basin
    Capital Markets Research, 227-241.
16. Etebari, A., & Duncan, K. (1997). “The Efficacy of Insider Trading Laws: New Zealand
    Evidence”. The New Zealand Investment Analyst, 19, 17-25.
17. Etebari, A., Tourani-Rad, A., & Gilbert, A. (2003). “Disclosure Regulations and Profitability
    of Insider Trading: Evidence from New Zealand”. Unpublished paper, Department of
    Finance, Waikato Management School.
18. Drew M, Marsden A, Veeraraghavan M, (2005) Does Idiosyncratic Volatility Matter? – New
    Zealand Evidence, Discussion paper
19. Vos, E; Pepper, B (1997). “The size and book to market effects in New Zealand”. The New
    Zealand Investment Analyst, 18, 35-45.
20. JF Pinfold, WR Wilson, Q Li 2001 -, 2001 – “Book to market and size as determinants of
    returns in small illiquid markets: The New Zealand case” Financial Services Review
21. Eckbo, B. E., and Smith D. C., (1998). “The Conditional Performance of Insider Trades”,
    Journal of Finance, Vol. 53, No. 2, 467-498.
22. Fama, E.F French, K.R 1996 “Multifactor Explanations for Asset Pricing Anomalies” The
    Journal of Finance 51 pp 55-84
23. Ferreira, Eurico J, 1995. "Insider Trading Activity, Different Market Regimens, and
    Abnormal Returns," The Financial Review, Eastern Finance Association, vol. 30(2), pages
    193-210
24. Givoly, D., and Palmon, D., (1985). “Insider Trading and the Exploitation of Inside
    Information: Some Empirical Evidence,” Journal of Business, 58 (1), 69-87.
25. Hamill, P., McIlkenny, P., and Opong, K.K., (1998). “Directors‟ Share Dealings and
    Company Financial Performance,” Working Paper.
26. Hamill, P.A., McIlkenny, P., and Watson, I. D., (1999). “Market Reaction to Directors' Share
    Dealings: Some Irish Evidence,” Irish Accounting Review, 6, 59-78.
27. Hillier, D. and A. Marshall, 1998. “The timing of directors‟ trades”, Journal of Business Law,
    454-467.
28. Hillier, D. and A.P. Marshall, 2002. “The market evaluation of information in directors‟
    trades”. Journal of Business Finance & Accounting, 29, 1-2, 77-110.



                                                                                               43
29. Hodgson & Praag 2006- “Information trading by corporate insiders based on accounting
    accruals: forecasting economic performance”, Accounting & Finance 46 (5), 819–842.
30. Jeng Metrick and Zeckhuser 2003- “Estimating the Returns to Insider Trading: A
    Performance-Evaluation Perspective”, The review of Economics and Statistics May 2003,
    Vol. 85, No. 2, Pages 453-471
31. John, K., and Lang L., (1991). “Insider Trading Around Dividend Announcements: Theory
    and Evidence,” Journal of Finance, 46(4), 1361-1389.
32. Karpoff, J.M., Lee, D., 1991. “Insider Trading before new issue announcements”, Financial
    Management, 20, pp. 18–26.
33. Ke, B., S. Huddart, K. Petroni, 2003. “What insiders know about future earnings and how
    they use it: Evidence from insider trades”? Journal of Accounting and Economics, 35, 315–
    346.
34. Hardjo Koerniadi & Alireza Tourani-Rad (2006) “Accrual or Cash Flow Anomaly? Evidence
    from the New Zealand Stock Market”, Working Paper, Auckland University of Technology
    School of Business
35. Michael S. Rozeff, Mir A. Zaman “Market Efficiency and Insider Trading: New Evidence”,
    The Journal of Business, Vol. 61, No. 1 (Jan., 1988), pp. 25-44
36. Michael S Rozeff, Mir A Zaman (1998) “Overreaction and Insider Trading: Evidence from
    Growth and Value Portfolios” The Journal of Finance 53 (2), 701–716.
37. Lakonishok, J. and I. Lee, 2001. “Are insider trades informative”? The Review of Financial
    Studies, 14, 1, 79-111
38. Lee, D.S., Mikkelson, W.H., Partch, M.M., 1992. “Managers‟ trading around stock
    repurchases”. Journal of Finance, 47, 1947–1961.
39. Madden K.P. Gombola M.H. (1983)- “Are some insiders more „Inside‟ than others, Journal
    of Portfolio Management 9, 18-22
40. Meulbroek, L.K., 1992. “An empirical analysis of illegal insider trading”. The Journal of
    Finance, 47, 5, 1661-1699.
41. McNally W.J, Smith B.F. (2003) “Do Insiders play by the rules? Canadian Public Policy–
    Analyse DE Politiques
42. O‟Connor, M., 1989, “Toward a more efficient deterrence of insider trading: The repeal of
    section 16(b)”, Fordham Law Review, 58, 309-381.
43. Palmrose Z.V., V.J. Richardson and S. Scholz, 2004. “Determinants of market reactions to
    restatement announcements”, Journal of Accounting and Economics, 37, 59–89.




                                                                                                 44
44. Pincus, M., S. Rajgopal and M. Venkatachalam (2005), “The accrual anomaly: International
    evidence”, Working paper, The University of Iowa.
45. Piotroski, J. D. and D. T. Roulstone, 2005. “Do insiders reflect both contrarian beliefs and
    superior knowledge about future cash flow realizations”? Journal of Accounting and
    Economics, 39, 55–81.
46. Piotroski, J. D. and D. T. Roulstone, 2004. “The influence of analysts, institutional investors,
    and insiders on the incorporation of market, industry, and firm-specific information into stock
    prices”. The Accounting Review, 79, 4, 1119-1151.
47. Pratt, S.P. en C.W. De Vere, 1970, “Relationship between Insider Trading and Rates of
    Return for NYSE Common Stocks”, 1960-1966, Modern Developments in Investment
    Management, 268-279.
48. Seyhun, N., 1986. “Insiders‟ profits, costs of trading, and market efficiency”. Journal of
    Financial Economics, 16, 189–212.
49. Seyhun, N., 1990. Do bidder managers knowingly pay too much for target firms? Journal of
    Business, 63, 439–464.
50. Seyhun, N., 1992. “The effectiveness of the IT sanctions”. Journal of Law and Economics,
    35, 149–182.
51. Seyhun, N., Bradley, M., 1997. “Corporate bankruptcy and insider trading”. Journal of
    Business, 70, 189–216.
52. Sivakumar, K. and G. Waymire, 1994. “IT following material news events: Evidence from
    earnings”. Financial Management, 23, 1, 23-32
53. Sloan, R.G., 1996. “Do stock prices fully reflect information in accruals and cash flows about
    future earnings”? The Accounting Review, 71, 3, 289-315
54. SL Summers, JT Sweeney 1998- “Fraudulently Misstated Financial Statements and Insider
    Trading: An Empirical Analysis” - The Accounting Review, 1998
55. Tourani-Rad, Gilbert & Wisniewski (2005) “The Impact of Regulatory Change on Insider
    Trading Profitability: Some Early Evidence from New Zealand”, Unpublished Paper,
    Department of Finance, Waikato Management School.
56. Turner, L., Dietrich, J.R., Anderson, K., Bailey, A.J., 2001. Accounting restatements.
    Working Paper, SEC.
57. Watson I., and Young A., (1998), “A Preliminary Examination of Insider Trading Around
    Takeover Announcements in Australia,” Working Paper, Department of Accounting and
    Finance, University of Western Australia.




                                                                                                   45

				
DOCUMENT INFO
Description: Accounting Firm Rankings document sample