Self Managed Superannuation Funds: Theory and Practice. By Dr Peter J Phillips University of Southern Queensland 1

Transcription

1 Phillis Self Managed Suerannuation Funds: Theory and Practice By Dr Peter J Phillis University of Southern Queensland 1 Abstract In this aer, a small reliminary samle of self managed suerannuation funds (SMSFs) is investigated. The contents of the actual SMSF ortfolios are examined and a number of microstructure features that characterise the SMSF ortfolios are described. These emirical characteristics are juxtaosed with theoretical and emirical finance to roduce an analysis of self managed suerannuation funds in theory and ractice. A theoretical economic rationale for the existence of SMSFs is develoed and microstructure features exhibited by the SMSFs such as home bias, under-diversification, high relative weightings of blue chi securities and the deloyment of infrequent ortfolio revisions (buy-and-hold strategies) are reorted and discussed vis-à-vis theoretical and emirical finance. The outcome is a first ste towards a more comlete understanding of this increasingly imortant comonent of Australia s retirement income stream. Key Words: Microstructure, Self Managed Suerannuation Funds (SMSF). JEL Codes: G11 1. Introduction The microstructure of self managed suerannuation funds (SMSFs) is an imortant area for research that remains relatively untouched. Whilst the Australian Taxation Office ublishes an aggregated summary that identifies where the trustees of SMSFs have laced their funds, this aer resents an examination of the actual ortfolio structures of a small samle of tyical SMSFs administered by a large financial services firm in South East Queensland. The contents of the SMSF ortfolios are investigated and a descrition of the comonents of a tyical self managed suerannuation fund as reflected by the information obtained from the samle is resented. A number of microstructure features that characterise SMSF ortfolio decisions, structure and erformance are uncovered and juxtaosed with theoretical and emirical finance. Modern finance theory and (Markowitz) ortfolio theory rovide the basic theoretical structure utilised in this aer. This theoretical structure vis-à-vis the emirical characteristics of SMSFs that are exhibited by the funds in the samle is the rimary focus of this aer. Hence, the aroach that shall be taken herein involves the lacement face-to-face of a descrition of the emirical characteristics of the SMSFs in the samle with the relevant ieces of modern finance theory and ortfolio theory. At times, the characteristics of the SMSFs in the samle are comared with emirical results documented and exlained by financial economists. For examle, international investments constitute a very small comonent of the SMSFs. This is in accordance with the home bias that investors have been found to exhibit and may be exlained by a variety of institutional or behavioural factors (see French and Poterba (1991, )). This study may be viewed as a first ste towards the develoment of a more comlete understanding of the microstructure of SMSFs. Most of the research contained in the existing literature focuses on other arts of the suerannuation system and deals redominantly with retail suerannuation funds. Some aers examine suerannuation choice or the investor s decision to remain with their existing or default suerannuation fund (Clark-Murhy and Gerrans (2001), Clark- Murhy, Kristofferson and Gerrans (2002) and Fry, Heaney and McKeown (2007). A few aers examine the erformance of suerannuation funds (Gallagher (2001), Drew and Stanford (2003) and Bilson, Frino and Heaney (2004; 2005)) and there have been a handful of aers that investigate the investor s selection of a suerannuation fund and suerannuation assets (Drew, Stanford and Taranenko (2001) and Drew and Stanford (2001)). The resent aer comlements these existing studies by extending the analysis that is extant in the literature to the microstructure of self managed suerannuation funds. This aer is organised as follows. In Section 2, a theoretical rationale for the existence of SMSFs is exlored. In Section 3, the data deloyed in this study are described. Whilst the samle is not large enough to ermit generalisations to the broader oulation of self managed suerannuation funds in Australia, the samle does reresent a comlete sub-oulation or subset of the global oulation of SMSFs administered by a large financial services firm located in South East Queensland. In Section 4, the equity ortions of the ortfolios are dissected and various characteristics extracted and described. This includes an analysis of the number and tye of ublicly listed Australian comanies contained in the ortfolios as well as an analysis of the inclusion of small-caitalisation issues and issues that have subsequently become delisted as a result of bankrutcy or failure of the issuing comany. In Section 5, a tyical SMSF equity and cash ortfolio is constructed and analysed. Attention is given to the erformance, risk and level of diversification of this archetyal equity and cash ortfolio. Section 6 concludes the aer. 8 Journal of Law and Financial Mangement - Volume 6, No. 1

2 Self Managed Suerannuation Funds: Theory and Practice 2. The Theoretical Rationale for SMSFs Theory and ractice converge immediately when one considers the (theoretical) economic rationale for the existence of SMSFs. The emirical fact that these suerannuation entities exist is something that must be exlored from the oint of view of modern finance theory. There are robably a large number of theoretical arguments that could be made to rovide a theoretical basis for the SMSFs that exist and are being created in Australia. One could, for examle, construct a theoretical rationale around the utility that investors may derive from the management of their own financial affairs and the greater number of investment choices that SMSFs rovide. However, recent research imlies that whilst eole are initially attracted to a large variety of choices, in the final analysis limited choices actually lead to higher levels of hainess (Iyengar and Leer, 2000). Furthermore, Benartzi and Thaler (2002) found that eole are no haier with their own ortfolios than with a ortfolio constructed by other average investors. Whilst these results are not conclusive, the resumtion that more choice makes eole better off has been weakened (Benartzi and Thaler, 2002,.1611) 2. Whilst investor choice may yet rovide the building blocks for a theoretical rationale for the existence of SMSFs, surer footing is found in the mainstream of modern finance theory. Markowitz (1952) identified the now well-known fact that covariances of asset returns lay a most imortant role in the management of ortfolio risk. If one ignores some or many covariances, the ossibility that one will achieve of an otimal result is jeoardised. However, this is exactly what occurs when an economic agent holds assets in a retail suerannuation fund whilst also holding a selection of outside assets. Under such conditions, the investment manager does not know what assets the investor has outside the managed fund and, consequently, is not aware of the covariances between in fund assets and outside assets. The investment manager does not know the comosition of the remainder of the investor s ortfolio. This roblem is comlicated by another equally imortant roblem that arises when a suerannuation fund deloys funds to more than one investment manager. This roblem of decentralised investment management also casts doubt on the likelihood that the ortfolio will be otimal, even in the absence of many beneficiaries (Share, 1981). Again, the difficulty in reaching otimality arises from the ignorance of covariances. The attainment of otimality is jeoardised when ortfolio managers cannot see the whole ortfolio. This casts doubt uon the ractice of decentralised management and the ractice of diversifying among different managers (on the basis of style and judgement) (Share, 1981). Indeed, achieving an otimal ortfolio outcome in the resence of the decentralised management roblem is only ossible under a very limited set of conditions (Share, 1981). The removal of covariance ignorance is a ossible theoretical economic rationale for the existence of SMSFs. In this aer, suort for this is rovided by showing that the in-fund assets outside assets roblem and decentralised investment management roblem can only be solved under very restrictive conditions. This task is made much simler by the fact that Share (1981) has already demonstrated the conditions under which decentralised investment management can achieve otimality. It is left to us to adat (or construct an analogy) between Share s decentralised investment management roblem and the in-fund assets outside assets roblem. An aroriately constructed analogy will ermit the utilisation of Share s notation and reasoning in the latter context. Once comleted, this will show that the two covariance ignorance roblems can be resolved only under limited conditions and an alternative feasible solution for a utility maximising investor is the initiation of a SMSF. In deriving a solution for the decentralised investment management roblem, Share (1981, ) first obtains the solution to a maximum roblem for a simle secification u t e v of the utility function, where u is the utility of the ortfolio, e is the exected return of the ortfolio, using otimum estimates of security exected returns, is the variance of the ortfolio and is the risk tolerance of the ortfolio. The solution is obtained by maximising the Largrangean function, the first order conditions of which can be exressed as follows: c... 2c 1 x1 0 e n c 1 n c 1.. nn t... 1 xn 0 0 z 1... en 0 (1) Where c ij is the covariance between the returns of securities i and j and e i is the otimum estimate of the exected return of security i. For comact exression, (1) is exressed in matrix notation as follows: DY F K t Where w F c c F k w F F c [ e 1c... e nc 0]' F k And [ e 1k... e nk 0]' Y k [ GK] t [ GF GK t { G[ w F GK t GF c c ] c k w F ]} k k w t G[ F k F ] c (2) (3) Where G is the inverse of D, [GK] is the minimum variance ortfolio, [GK ] is divergences er unit of risk tolerance, GK t GFc is the otimal assive ortfolio and is the otimal active divergences. Rewriting, Y wc[ GK t GFc ] wk [ GK t GFk ] Where [ GK t GFc ] is the otimal assive ortfolio and [ GK t GFk ] is the otimal active ortfolio. The decentralised investment management roblem can be solved if the investor allocates funds in accordance with the investment managers abilities to redict the exected return of security i, allocating w c to be invested by a assive manager who maximises the objective function c c c and w k invested by an active manager who maximises the (4) (5) u e v / t 9

3 Phillis u e v / t objective function k k k. Even though each manager acts as though his or her ortion of the fund is the whole ortfolio, the investor s objective function is maximised because [ GK t GFc ] maxmises u c and [ GK t GFk ] maximises u k. By analogy, Share s solution to the decentralised investment management roblem can be easily extended to generate a solution for the in-fund assets outside assets roblem identified reviously. If the investor manages his or her outside assets ersonally and allocates the other comonent of his or her whole ortfolio to a single fund manager, then he or she simly becomes one of the two managers to whom a weighting is rescribed. That is, the in-fund assets outside assets roblem becomes a secial case of the decentralised investment management roblem. Of course, if the investor undertakes no ersonal management of any comonent of his or her whole ortfolio, the roblem once again reverts to a decentralised management roblem described by Share. In the resence of many ossible managers (of which the investor himself or herself may be one), Share s aroach still alies but will require an estimation of the covariances among the subsets of the whole ortfolio (Share, 1981,.229). There is, therefore, a solution to the decentralised investment management roblem and the in-fund assets outside assets. The crucial oint, however, is that this solution is ossible only under limited conditions: (1) the agent s marginal rate of substitution of variance for exected return is constant; (2) unrestricted short sales are ossible; (3) it is ossible to lead each manager to maximise the aroriate objective function; (4) all managers agree about the variance and correlations of a set of securities (Share, 1981). Furthermore, a number of other issues combine to make this a best case scenario for the solution of our roblems. First, the ossibility that markets are efficient (and the imlications that would have for managers redictions) was assumed away or ignored. Second, we ignored the ossibility of many clients (a many-beneficiary roblem). And third, our agent had a well defined objective function (Share, 1981). In the absence of some of these features, the solution to the decentralised investment management roblem and the in-fund assets outside assets roblem becomes extremely roblematic and a more feasible solution to ursue for a utility maximising investor is the initiation of a SMSF. 3. Data The data considered in this investigation were obtained from a large financial services firm based in South East Queensland. The firm administers a total oulation of aroximately 130 self managed suerannuation funds. Of these, two-thirds of the self managed suerannuation funds consist redominantly of residential and commercial roerty and involve minimal direct management by the trustees and limited exosure to financial securities. The remaining third involve direct management of a ortfolio of both real assets and financial securities by the SMSF trustees. It is this segment of the total oulation of funds administered by the financial services firm that the firm ermitted detailed access to for the uroses of this research. Thus, the samle of self managed suerannuation funds consists of 41 SMSFs where the trustees directly manage a traditional mix of assets: cash, real estate, managed funds, fixed interest securities and equities. The financial services firm from which the data were sourced rovides a range of financial services to clients, including the SMSF trustees. These services include roviding financial lanning advice, administering the SMSFs and erforming accounting functions. Whilst the trustees of the SMSFs in this samle may have received financial advice or otherwise taken advantage of these services, the SMSFs are the roduct of the trustees decisions and have not been constructed for them by a rofessional financial lanner, accountant or ortfolio manager. Further, the trustees are ultimately resonsible for the management of their SMSF. Hence, this investigation generates some imortant insights into the nature of the microstructure of SMSF ortfolios and the investment choices that have been made by the trustees of these self managed suerannuation funds. The 41 ortfolios in the samle were formed between 1998 and 2004, with all additions to the equity ortion of the Investment Quantity Purchase Date Average Cost of each Security Total Dollar Cost Weighting Cash at Bank Macquarie Cash Management Trust 485, , % Westac Savings Account 2,306 2, % TOTAL 487, % Shares in Listed Australian Comanies PMP Communications Limited 1,000 30/06/ , % St George Bank 586 4/12/ , % Suncor Metway 2,520 1/11/ , % Telstra 2 Instalment Receits 22/10/1999-2,900-2,900 Telstra Cororation Limited 4,000 5/11/ , % TOTAL 41, % TOTAL FOR SMSF 570, % Table 1 - An Examle of a SMSF s ortfolio structure 10 Journal of Law and Financial Mangement - Volume 6, No. 1

4 Self Managed Suerannuation Funds: Theory and Practice ortfolios comleted by June of Whilst the samle size is too small to ermit wide-reaching generalisations regarding the total oulation of SMSFs in Australia, there is no reason to believe that there is anything atyical about the SMSFs in the samle that would result in a divergence in fundamental character from similar funds obtained from alternative data sources. Furthermore, the funds have assed through an annual audit rocess and do not exhibit any anomalous features that would lead to a non-comliance with Australian Tax Office requirements or, more broadly, the Suerannuation Industry (Suervision) Act 1993 designation being laced uon them. This ensures another level of base-line similarity with the broader oulation of self managed suerannuation funds in Australia. The financial services firm from which the data were sourced administers each of the SMSFs. This role involves overseeing the lacement of orders for securities with a brokerage firm, maintaining all of the necessary records and facilitating the auditing of the SMSFs via a third-arty auditor to ensure the continued comliance of the SMSFs with Australian Tax Office requirements and the relevant legislation. The research results resented herein are based uon accurate and comlete data concerning the ortfolios in the samle and do not need to rely on the record-keeing of individual SMSF trustees. The data itself was resented as 41 individual ortfolio summaries. The summaries list the cash at bank, the quantities of fixed interest securities, real estate and listed securities contained in the SMSF, the urchase dates, average cost and the weighting of each asset in the ortfolio. An examle of an actual ortfolio summary is resented in Table 1. The average size of the self managed suerannuation funds in this samle is aroximately $400,000. Across all the funds in the samle a total of nine different asset classes are reresented. The average dollar amounts invested in each of these nine different asset classes and the average ercentage weighting attributed to each asset class in the ortfolios is resented in Table 2. Confronted with the task of constructing a ortfolio of real and financial assets, the trustees of these SMSFs aear to direct their attention to the most familiar asset classes: cash, shares in Australian comanies and managed investments (unit trusts). Whilst this is not surrising, the concentration of investable funds into a small range of asset classes means that the efficient frontier available to these SMSF investors lies somewhat below that which could be obtained if other investment oortunities were carefully considered. The focus of the ortfolios into a narrow range of asset classes, articularly Australian cash, shares and unit trusts, may be symtomatic of a higher level of (relative) risk aversion or general deficiency of confidence among these trustees. However, the data also ermits the exloration of the extent to which the SMSFs reflect the broader tendency of investors to exhibit home biases or references for domestic equities over international investments. For examle, French and Poterba (1991) discovered that U.S. investors allocated over 90 ercent of their funds to U.S. securities. This behaviour leads U.S. investors to ignore over 50 ercent of world s foreign equity oortunities and forego the otential exected returnrisk benefits associated with overseas investments. Similar results were discovered for Jaanese investors, whilst British investors were found to hold aroximately 18 ercent of their ortfolios in foreign stocks. The very small allocation to overseas investments exhibited by the SMSF ortfolios is an indication of home bias. This is consistent with emirical results documented in numerous studies. Because Australia s stock market accounts for such a small fraction of the world s equity oortunities, the oortunities forgone by home biased Australian investors are of a greater order of magnitude. In Australia, home bias leads Australian investors to ignore more than 90 ercent of the world s foreign equity oortunities. Of course, the most imortant thing from a ortfolio management oint of view is that there are otentially considerable risk reductions available to investors who invest in overseas markets. Whilst the returns generated by the world s stock markets have become more Asset Mean Short Term Cash 80, Long Term Cash 26,625 4 ($) Mean Listed Australian Shares 159, Unlisted Australian Shares Fixed Interest 8, Unit Trusts 66, Real Estate 50, Overseas Shares 3, Derivatives TOTAL 396,995 Table 2 - Portfolio Structure: Averages across all SMSFs in the samle Notes: short term cash denotes cash held in bank savings accounts or cash management trusts. Long term cash denotes cash held in term deosits. For the SMSFs contained in this samle, real estate refers to commercial roerty, with a small number of self managed suerannuation funds holding one or two commercial roerties. Derivatives denote comany otions listed on the Australian Stock Exchange, not exchange traded otions. (%) 11

5 Phillis closely correlated, the correlation is far from erfect and could be exloited by ortfolio managers to increase the exected return of their ortfolios and decrease the total risk of their ortfolios. A variety of factors, including institutional factors and transactions costs, could lead investors to forgo the otential benefits of international diversification. However, it is also ossible that investors (mistakenly) attribute additional risks to overseas investments due to their unfamiliarity with foreign business environments (French and Poterba (1991,.225) and Tversky and Heath (1991)). Whatever the case may be, the SMSFs in this samle do aear to exhibit considerable home bias. In addition to obvious home bias, the SMSF trustees also exhibit strong references for articular asset classes and allocate only small roortions of their investable funds to others. Whilst some of the SMSFs contain long term cash investments (term deosits), real estate and overseas shares, on the whole these are not oular investment choices with 93 ercent, 83 ercent and 87 ercent of funds having zero holdings of these asset classes, resectively. The dominant asset classes are short term cash (at bank or invested in a cash management trust), listed unit trusts and listed Australian shares. None of the funds hold a zero short term cash balance and only 2.40 ercent of funds do not hold shares in listed Australian comanies. Given this dominance, the focus of the remainder of this aer is the dissection of the equity ortions (including listed unit trusts) of the SMSF ortfolios. 4. The Equity Portions of the SMSF Portfolios In this samle, short term cash investments, listed unit trusts and Australian shares account, on average, for more than three-quarters of the total value of the ortfolios. Whilst this is imortant information, a more detailed account of the equity ortion of the ortfolios reveals a number of features of the SMSFs that are not discernible from an analysis of the aggregated data. In this section, the microstructure characteristics of the equity ortions of SMSFs are reorted. This analysis reveals: (a) the Australian comanies in which the SMSF investors have invested; (b) the reresentation in the ortfolios of small caitalisation issues versus large caitalisation issues; (c) the reresentation of articular industry sectors; (4) the extent of the seculative comonent of the ortfolios, including a consideration of investments in now-delisted or bankrut business enterrises; and (5) the weighting schemes alied by the SMSF trustees to the securities in the ortfolios, the frequency of the trading exhibited by the ortfolios and the tendency to form the ortfolios over a very short eriod. 4.1 The Australian Comanies SMSF Trustees Choose for their Portfolios On average, each ortfolio in the samle contains an investment in the shares of twelve different listed Australian comanies and listed unit trusts. Whilst the average number of different equity securities held in the self managed suerannuation fund ortfolios is quite small, the 41 SMSFs in the samle invested in a total of 152 different Australian comanies listed on the Australian Stock Exchange. The five most oular 3 Australian comanies among the SMSF trustees comrising the samle are: (1) Australia and New Zealand Banking Cororation; (2) National Australia Bank; (3) Suncor Metway; (4) Telstra Cororation; (5) and Woolworths. These comanies were each resent in aroximately half of the SMSFs. The next five most oular are: (1) BHP Billiton; (2) Fosters Grou Limited; (3) Paerlinx; (4) Commonwealth Bank of Australia; and (5) Commonwealth Proerty (Office). These comanies and roerty trusts were each resent in aroximately one-quarter of the SMSFs. This information is resented in Figure 1. An interesting question arises at this oint: Do the SMSF trustees exhibit home bias at home? Home bias at home is the observed tendency of investors to exhibit a reference for comanies located within close geograhical roximity, even within their own national borders. As mentioned earlier, investors have been observed to exhibit home bias or the tendency to invest domestically rather than overseas. Coval and Moskowitz (1999) found that the tendency for investors to refer investments that are closer to home also alies to domestic ortfolios. That is, investment managers refer to invest in firms that are locally headquartered. The SMSFs in the samle are located in South East Queensland and home bias at home would result in a tendency for the SMSFs to favour firms headquartered in Queensland ANZ Commonwealth Office BHP Bluescoe CBA Fosters Grou NAB Paerlinx Figure 1 - The Number of SMSFs Containing the Most Poular Investments Suncor Telstra Woolworths 12 Journal of Law and Financial Mangement - Volume 6, No. 1

6 Self Managed Suerannuation Funds: Theory and Practice The SMSF trustees in the samle invested in a total of 152 Australian comanies. Of these, aroximately 12 ercent are Queensland comanies with headquarters in that state, with the most oular Queensland-based comanies being Suncor, ABC Learning, Flight Centre and New Hoe Cororation. This is not strong evidence for home bias at home, although some of the more obscure selections (for examle, New Hoe Cororation and Buderim Ginger), may be the roduct of home bias at home. The absence of strong evidence for home bias at home does not rule out the ossibility that the geograhical roximity of certain comanies did indeed lead some of the SMSF trustees to include these comanies in their ortfolios. Also, when considering whether the SMSF trustees exhibit home bias at home, allowance must be made for the relatively low number of Australian comanies headquartered in Queensland. Another imortant characteristic of the SMSF ortfolios is the absence of high levels of diversification. The inclusion, on average, of just twelve equity securities in the SMSF ortfolios has significant imlications for the diversification of these funds. In their classic study, Evans and Archer (1968) concluded that an investor required aroximately 15 to 20 equity securities to fully diversify his or her ortfolio and remove the effects of non-systematic risk on the total risk of the ortfolio. More recent research has determined that the number of equity securities required to achieve full diversification has increased in recent decades as individual shares have become more volatile. Cambell, Lettau, Malkiel and Xu (2001) suggest that the number of shares required is now aroximately 50. The equity ortions of the SMSF ortfolios may be exosed to considerable levels of diversifiable risk. This is one feature of SMSFs that cannot be ascertained by an examination of aggregated data. The under-diversification of the SMSFs is in accordance with the extant emirical investigations that have studied the usefulness of diversification as a guide to what one might exect of eoles ortfolios. In articular, using Federal tax return data and the Federal Reserve Board s 1962 survey of the financial characteristics of consumers, Blume and Friend (1975) found that, contrary to the assumtions that underlie certain arts of finance theory, individuals do not hold diversified ortfolios. In fact, Blume and Friend found that (American) households at the time tended to reort dividends from only one or two firms. Exlanations for this behaviour include (1) heterogenous exectations (Blume and Friend (1975,.597); (2) the inability of the investor to otimise by assessing the covariances of his or her entire ortfolio; and (3) transactions and decision costs. Most interesting, however, is the exlanation develoed by Goldman (1979). Secifically, Goldman (1979,.511) exlains how a favouritism rincile may be utilised by investors in lace of the diversification rincile: We normally give more weight to our favourite foods (than is considered normal) in the comosition of our grocery bundle, to our favourite divertissement in our entertainment bundle etc. The diversification rincile emhasises the interactions of goods; the favouritism rincile underscores the individual (in vacuo) attributes of each good. The relative imortance of diversification to favouritism is in most choice roblems deendent on circumstances e.g., wine must be chosen in coordination with entrée but erhas indeendently of the evening s entertainment. In the context of the ortfolio roblem the diversification rincile seems to dominate when the revision interval is short when return distributions are tight and symmetric. However, when the revision eriod is long (i.e. comounded return distributions are sread out and skewed) the convolution of asset returns tends to be dominated in an essential way by one asset s return. For otimality it is imerative that this dominating return be the favourite. Consequently, there is a tendency to asymtotically reduce the ortfolio weights of all nonfavourite assets The favouritism rincile may hel to exlain the underdiversification of infrequently revised ortfolios like the SMSFs in the samle. Not only are the SMSFs under-diversified and concentrated in well-known, Blue Chi comanies, but they also exhibit infrequent revisions with most ortfolios remaining stationary for a number of years (see Section 4.5 below). Deending on whether the SMSF ortfolios in the Consumer Discretionary Consumer Stales Energy Financials Proerty Trusts Health Care Industrials Information Technology Materials Telecommunication Services Utilities Delisted Figure 2 - The Reresentation in the Portfolios of Industry Sectors 13

7 Phillis samle could be assumed to concur with the class of roblems analysed by Goldman (1979), a monotonic trend towards less diversification as the length of the revision eriod increases might be consistent with the favouritism rincile and may hel to exlain why these infrequently revised SMSF ortfolios exhibit low levels of diversification. If Goldman (1979) is correct, this may still lead to an aroximately otimal ortfolio solution. The further exloration of this articular juxtaosition between finance theory and SMSF ractice is a toic that cannot be considered to fall within the scoe of the resent aer. 4.2 Industry Sector Reresentation The reresentation of industry sectors in the SMSF ortfolios also sheds imortant light uon the structures and risk rofiles of the ortfolios. Just over 26 ercent of the comanies selected for inclusion in the ortfolios are constituents of the consumer stales and consumer discretionary industry classifications. This is, erhas, an indication that the trustees invest in comanies whose names or roducts they are familiar with. Further suort for this hyothesis might be reflected by the fact that 16 ercent of the comanies selected for inclusion in the ortfolios are drawn from the financials industry classification and a further 24 ercent from the industrials and materials industry classifications. Over two-thirds of the comanies selected for inclusion in the SMSF ortfolios are derived from these five industry classifications (see Figure 2). Conversely, only a very small number of the comanies selected (in aggregate) by the SMSF trustees for inclusion in their ortfolios are constituents of the telecommunication services (1.30 ercent), roerty trusts (8.50 ercent), information technology (1.90 ercent), energy (4.60 ercent), healthcare (3.90 ercent) and utilities (1.90 ercent) industry sectors. Not surrisingly, these sectors have a relatively low reresentation among their constituents of the Blue Chi or most well-known comanies. Arguably, only Telstra Cororation (telecommunication services) fits into this category vis-à-vis the major Australian banks, insurance comanies, mining comanies and industrial comanies that constitute the sectors from which the majority of the comanies reresented across the SMSF ortfolios were drawn. 4.3 Small Cas versus Large Cas The tendency towards well-known comanies or Blue Chis is also reflected in the dominance of large caitalisation comanies in the ortfolios. Whilst the 41 ortfolios contain a total of 152 different listed Australian comanies and listed unit trusts, the dominant stock selections derive from the Australian Stock Exchange s fifty largest caitalisation comanies. This is quite evident from the discussion resented in section 4.1 above. However, further elaboration on this imortant characteristic of the SMSFs may be aroriate. To be recise, ten out of the twelve (83 ercent) most oular comanies among the SMSF trustees are constituents of the S&P/ASX 50. Furthermore, only 9 out of the 50 (18 ercent) comanies that constitute the S&P/ASX 50 are not reresented in at least one of the SMSF ortfolios. Interestingly, the stocks that do not obtain reresentation in the SMSF ortfolios are, by and large, lesser known comanies, including Asciano Grou, Centro Proerties, Goodman, Orica, Telecom NZ, and Zinifex. However, Brambles, Macquarie Bank and Santos also fail to achieve selection by the SMSF trustees. Whilst the trustees of these ortfolios did not restrict themselves entirely to Blue Chi or to 50 comanies, the Investment Quantity Purchase Date Average Cost of each Security Total Dollar Cost Weighting Cash at Bank Cash 29,606 29, % TOTAL 29, % Shares in Listed Australian Comanies ANZ /05/ % APN /03/ % BHP /03/ % Commonwealth Office 16,151 24/03/ , % Coca-Cola Amatil /03/ % Contango Microca /05/ % Hunter Hall Global /03/ % Leighton Holdings /05/ % NAB /03/ % Paerlinx /03/ % Patrick Cororation /03/ % QBE /05/ % Ronin Proerty Grou 12,844 10/05/ % SFE /03/ % Suncor /03/ % Tabcor /05/ % Telstra /03/ % Westfield Holdings /03/ % Woolworths /03/ % TOTAL 299, % TOTAL FOR SMSF 328, % Table 3 - SMSF equity and cash ortfolio structure 14 Journal of Law and Financial Mangement - Volume 6, No. 1

8 Self Managed Suerannuation Funds: Theory and Practice resence of mid-caitalisation comanies and, articularly, small-caitalisation issues is restricted to a reasonably small art of the ortfolios. This is best illustrated by resenting and examining the details of the equity ortion of another of the SMSF ortfolios (see Table 3). A close investigation of the structure of this SMSF ortfolio reveals the resence of only two or three obscure issues. Whilst it is not ossible at this stage to determine by which rocesses these comanies came to be included in the ortfolios, there is almost certainly an indeendent research or information gathering rocess being deloyed by the SMSF trustees. Imortantly, this asset selection or decision-making rocedure aears to have resulted rimarily in the selection of well-known Australian comanies and only secondarily in middle or small caitalisation issues. This has revented any reonderance of seculative issues in the SMSF ortfolios. Whilst many small caitalisation issues are reresented across the SMSFs in the samle, each SMSF has a small roortion of investable funds allocated to such securities. 4.4 Delisted (or Susended) Comanies and Bankrutcy Cases The final characteristic of the SMSF ortfolio s equity selections that shall be considered here is the reresentation in the ortfolios of delisted comanies or bankrutcy cases. Whilst 11.8 ercent of the 152 comanies originally chosen by the SMSF trustees for inclusion across their ortfolios were delisted in the eriod 2004 to 2007, these comanies were relatively unoular selections and, in each case, were usually held by just one ortfolio. Furthermore, the delisting of a articular comany was not always due to bankrutcy. Rather, on many occasions the delisting was due to a far more benign set of circumstances. However, a handful of the SMSF ortfolios were exosed the failures of Australian Magnesium and Sons of Gwalia. Fortunately, exosure to these failed comanies amounted to less than 1 ercent of the value of the affected SMSFs. Of course, the lack of largescale exosure to failed business enterrises is to be exected given the analysis resented in earlier sections that revealed the dominance of S&P/ASX 50 or Blue Chi comanies in the SMSF ortfolios. 4.5 Weighting Schedules, Trading Frequency and the Absence of Dollar-Cost Averaging There are a number of imortant miscellaneous characteristics that are revealed by thorough analysis of the microstructure of the SMSFs in the samle. Firstly, the ortfolios exhibit aroximately an equal-weighted weighting schedule. The aortionment of the investable funds among securities is undertaken so as to roughly equate, either by design or circumstance, the ercentage of total investable funds allocated to each security. The allocations are not so recise as to lead to the susicion that any significant degree of thought and calculation has been alied. Rather, the SMSF trustees have, on average, avoided large amounts of asymmetry in the weighting schedules they have alied with only one or two securities having a much smaller or much larger weighting than the other securities in the ortfolio. For examle, consider the following data concerning five ortfolios selected at random from the samle (see Table 4). The second imortant characteristic is absence of frequent trading or turnover of issues in the ortfolios. Most of the ortfolios, once formed, remained stationary for a number of years. It is therefore aroriate to suggest that the dominant strategy alied to these SMSF ortfolios has been a buy-andhold strategy. Imortantly, this aroach has robably served the SMSF trustees well and is accorded a great deal of suort by emirical finance. In fact, it has been discovered that ortfolio managers usually fail to outerform a buy-and-hold strategy. For examle, Jensen (1968) in his now classic study found that no ortfolio manager was able to consistently outerform (on a risk adjusted basis) an investment in T-bills and the market index during the eriod 1955 to More recent evidence has continued to suort this finding and reaffirm the underlying soundness of the buy-and-hold aroach. The soundness of buy-and-hold derives from a combination of factors. Firstly, there are the imlications of market efficiency. Investment managers do not aear to be able to consistently earn above equilibrium returns (unless they ossess a cometitive informational advantage). Secondly, an active trading strategy attracts higher transactions costs and management exenses. Elton, Gruber, Das and Hlavka (1993) reorted that a low turnover aroach outerformed ortfolios with high turnover. Thirdly, frequent trading or ortfolio rebalancing increases the chance that investors will miss out on being in the market on strong days. Being out of the market on just a few days over any ten year eriod has been shown Portfolio Highest Allocation to a Lowest Allocation to a Average Particular Security Particular Security Allocation % 1.66% 3.17% % 3.34% 5.16% % 1.45% 9.09% % 8.12% 8.77% % 0.29% 3.15% Table 4 - Weighting Schedule Examles Notes: the average allocations indicate the aroximately equal weighting schemes alied in these ortfolios. A few investments might attract a little more weight but usually any asymmetry in the weighting schedules leans towards underweighting of one or two securities. 15

9 Phillis to lace considerable downward ressure on the ortfolio s terminal value. For examle, whilst the S&P 500 generated a comound annual return of 18 ercent during the eriod 1982 to 1990, being out of the market for just 30 of the strongest days decreased the return to 5 ercent (Strong, 2006,. 389). The buy-and-hold strategy alied to the SMSFs is accorded strong emirical suort. The third imortant characteristic exhibited by the SMSF ortfolios is the absence of a dollar-cost-averaging aroach to building the ortfolios. Rather than steadily acquiring securities over a medium to longer term, the ortfolios tended to be constructed in a very short eriod of time sometimes in just one or two trading days. This is exemlified by the ortfolio resented in Table 3 above. The construction of a ortfolio over such a short time horizon is inadvisable. On any articular trading day, share rices may exhibit unusual movements either on the uside or downside. Whilst a ortfolio might fortuitously benefit from such movements (at least in the short term), there may be adverse effects reflected both in the volatility the ortfolio exhibits and a higher average rice aid for the selected securities. Conversely, a more gradual construction of the ortfolio through a dollar-cost-averaging aroach ensures that the investor buys more shares at lower rices and fewer shares at higher rices. Such a strategy increases the chance that the investor will ay a reasonable average rice for the securities in the ortfolio. The absence of such a strategy from the SMSFs in the samle is somewhat disconcerting. 5. A Tyical SMSF Portfolio: An Analysis The juxtaosition between finance theory and SMSF ractice becomes critical once the discussion moves to ortfolio construction and otimality. The analysis of the microstructure of the SMSFs in this samle has revealed a number of characteristics. First, cash and listed Australian equity securities (including roerty trusts) account for more than three-quarters of the total value of the ortfolios. Second, the SMSF trustees held an average of 12 equity securities in their ortfolios. Third, a large roortion of the equity investments are drawn from the consumer stales, consumer discretionary, financials, materials and industrials industry sectors. Fourth, the most oular equity investments among the SMSF trustees are Blue Chi securities drawn from the S&P/ ASX 50 with middle caitalisation issues and, articularly, small caitalisation issues reresenting smaller roortions of the total number of securities chosen by the SMSF trustees for inclusion in their ortfolios. In this section, these microstructure characteristics are utilised to build a SMSF ortfolio that is tyical of the funds in our samle. The archetye ortfolio is analysed using the tools of modern ortfolio theory. 5.1 Forming an Archetye SMSF Portfolio The SMSFs in this samle were formed during the last years of the 1990s and the early years of the 2000s. The majority of the ortfolios were formed during 2003 and Given this, the archetye SMSF constructed here is assumed to come into existence on January The total amount of investable funds is $400,000. Of this, 19 ercent or $76,000 is allocated to a cash account. An interest rate of 4.00% er annum is assumed to revail. The remainder of the investable funds are allocated to twelve equity securities (including roerty trusts). The weighting scheme alied to the archetye SMSF is an equally weighted scheme and the investment strategy deloyed is, in accordance with the aarent behaviour of the SMSFs in this samle, a buy-and-hold strategy. The equity securities were chosen for the ortfolios on the basis of the most oular securities and industry sectors identified in sections 4.1 and 4.2 (above). This archetye SMSF ortfolio exhibits most of the characteristics of the SMSFs in the samle. There is a distribution of the equity ortion of the ortfolio across twelve equity securities and a variety of different industry sectors. The ortfolio constituents were urchased on a single day rather than through a dollar-cost-averaging aroach. After allocating aroximately 20 ercent of the ortfolio to cash at bank, the weighting schedule aortions the remaining investable funds equally across the twelve equity securities. The strategy that is deloyed from the incetion date is a buyand-hold strategy with no additions or subtractions to or from the ortfolios. Whilst this archetye SMSF ortfolio aears Investment Quantity Purchase Dates Average Cost of each Security Total Dollar Cost Weighting Cash at Bank Cash 76,000 76, % TOTAL 76, % Shares and Unit Trusts AGL /01/ $27, % ANZ /01/ $27, % APN News and Media /01/04 4 $27, % BHP /01/ $27, % Commonwealth Office /01/ $27, % CSL /01/ $27, % Fosters Grou /01/ $27, % Paerlinx /01/ $27, % Suncor /01/ $27, % Telstra /01/ $27, % Wesfarmers /01/ $27, % Woolworths /01/ $27, % TOTAL 324, % TOTAL FOR SMSF 400, % Table 5 - The archetye SMSF ortfolio 16 Journal of Law and Financial Mangement - Volume 6, No. 1

10 Self Managed Suerannuation Funds: Theory and Practice to be reasonably sound, closer insection and analysis utilising the tools of modern ortfolio theory is required in order to reach informed conclusions on the soundness of this tyical SMSF ortfolio. 5.2 Analysis of the Archetye SMSF Portfolio Careful analysis of the structure of the archetye SMSF ortfolio reveals a number of imortant ieces of information regarding the diversification, exected return and risk exhibited by the ortfolio. Of most concern from a ortfolio management oint of view, is the ossibility that a large ortion of the risk of the SMSF ortfolio is attributable to firm-secific risk factors that might have been diversified away through the selection of additional securities for inclusion in the ortfolio. More aroriate levels of diversification may increase exected returns and reduce the total risk of the ortfolio. Similarly, the equal-weighted strategy that seems to have been alied, by and large, to the SMSF ortfolios in the samle and which is reflected in the archetye ortfolio, may be a subotimal weighting schedule. A higher level of exected return may be obtainable from a reallocation of the investable funds among the chosen assets. Portfolio summary statistics for the archetye SMSF ortfolio are resented below. The archetye ortfolio has erformed quite well over the eriod from Imortantly, on a risk-adjusted basis, this ortfolio has outerformed the unmanaged ASX All Ordinaries index (RVOL equals 0.17). However, the ortfolio is reasonably under-diversified with aroximately one-quarter of the fluctuations in its returns being attributable to firmsecific factors that could be diversified away. Not surrisingly, the solution of the relevant Markowitz ortfolio rogramming roblem reveals that it is ossible to increase the exected annual return of the archetye ortfolio by aroximately four ercentage oints by abandoning the equal weightings initially accorded to the securities in the ortfolio and re-weighting the ortfolio in an efficient manner. The efficient ortfolio that results exhibits a higher exected return and a lower beta coefficient (0.69). It should be noted that the archetye ortfolio incororated the best features of the SMSFs in the samle. The archetye ortfolio accurately reflects most of the features of the SMSFs in the samle. However, in constructing the archetye ortfolio the investable funds were distributed across different industry sectors to reflect the allocation of the investments on average across all funds. The result is an overstatement of the diversification exhibited by each of the SMSF ortfolios examined individually. Also, the archetye ortfolio contains the most oular investments. This tends to result in an allocation of investable funds to what may be called the bluest of Blue Chi shares. In the context of the samle of SMSFs, the archetye SMSF is a best case examle. Many of the SMSFs in the samle have, in reality, failed to emulate the erformance of the archetye ortfolio, even though the archetye ortfolio itself is far from erfect. 6. Summary and Conclusions The analysis resented in this aer revealed a number of characteristics of self managed suerannuation funds that are worth noting. First, the 41 SMSFs in the samle contain an average of twelve equity securities selected from a total of 152 different listed Australian comanies and unit trusts, onequarter of which were drawn from the consumer stales and consumer discretionary industry sectors. Second, well-known Blue Chi issues are strongly reresented across the ortfolios with most of the constituents of the S&P/ASX 50 featuring in the ortfolios. Third, the selection of the equity securities Portfolio Summary Statistic Formula Statistic Value Exected Annual Return Annual Standard Deviation Beta Percentage of Total Variance Attributable to Systematic Factors (using Share s single index model) Percentage of Total Variance Attributable to Non-Systematic Factors (using Share s single index model) Treynor s RVOL Ratio E n R w ER P i1 n n n 2 2 wi i i1 i1 j1 i j i i 21.73% 2 P wiw j ij 8.71% n i P w i i1 2 2 P M 2 P 2 ( e P 2 P ( ep ) P P M r ) r r f 74.02% 25.98% Table 6 - Portfolio Statistics for the Archetye Portfolio 17