Would you like choice of fund with that?

This week’s effort is about super choice – as will be next week’s.

It’s amazed me how much effort has been put into choice of fund and yet, particularly in the light of how little people know, how little effort has been put into trying to make those choices reasonably informed.

We have long known and for commonsensical reasons that good information is critical to economic efficiency. Friedrich Hayek argued this within the ‘Austrian’ tradition of economics in prosecuting his case in the ‘socialist calculation debates’ of the 1930s. Asymmetric information found its way into the neoclassical mainstream from the early 1960s on in the work of people like Kenneth Arrow, George Stigler, George Akerlof and Joseph Stiglitz.

Oddly however, this is often where the argument ends. The Austrian tradition has little to say (that I know of) about addressing the various ways in which markets fail in the way they handle information (prefering to focus on how much worse it is under central planning). At least within the orthodox ‘neoclassical’ tradition, market failures in the provision of information are conceded. This leads to a fairly formulaic assertion of the role of regulation to compel disclosure – for instance to consumers and to investors. But this is often where the discussion ends with little real interest in the efficacy of such regulation, and whether it can be improved – and if so how.

At the same time, the idea of ‘performance regulation’ is a buzzword in some areas of regulation like occupational health and safety and the environment. ‘Performance regulation’ is regulation that is tailored as closely as possible to achieving whatever objectives we set for it. Put another way, it targets outputs (like lower levels of industrial accidents) rather than inputs (like more machine guards). Beyond OHS and environment, economic reform has not gone very far down the path of ‘performance based regulation’.

I don’t know of any literature which asks the question ‘what would performance regulation look like in the area of information? I’ve had a go at this here in the context of a more general argument for broadening our understanding of what might constitute economic reform. (Comments appreciated)

Amidst more general comment about good choices requiring good information and some elaboration of the shortcomings of the industry that sells what we euphemistically call ‘investment advice’ this column tries to set out a strategy for ‘performance based’ disclosure regulation in the area of investment advice.

Super Choice

The advertising blitz is upon us. Employees are at long last getting the right to choose how their own hard earned superannuation is invested. But let me hazard a guess. You’ve got no idea what you’ll do with super-choice. Here’s another guess. Even if you start studying now you still won’t know what to do come 1st July.

In life as in economics, choices are only as good as the information they’re based on. And there’s always been a terrible problem in getting people good information about where to invest. It’s an intrinsically difficult problem. Markets often fail doing it and governments can be even worse.

Right now, ‘investment advice’ exists in a netherworld. Having grown organically from the sale of life-insurance, the industry is still regarded by investment funds as a ‘sales channel’ and its commercial structure is identical to the sale of fridges in a department store with each stage of the distribution chain adding its margin and using it to fund sales incentives.

At one level there’s nothing wrong with that. It’s a free country and it works for fridges. Why not for investments? The problem is that there are big rewards for sales people who can get their customers to think of them as fiduciary advisors that is, professionals who act in their clients best interests, usually in return for a fee, as a good doctor or accountant would.

The ambiguities of this situation, together with the usual gallery of rogues on the fringes of any industry make regulation inevitable. That’s no bad thing where it provides basic safeguards like some protection against fraud, fee disclosure and streamlined dispute resolution.

Beyond this, the regulation we’re pursuing will do more harm than good. Why? Firstly, beyond very simple regimes like the panel on a packet of biscuits telling you its fat and sugar content elaborate disclosure confuses and paralyses more than it empowers.

And even more importantly, trying to regulate to improve the advice of a sales oriented industry is not just wishful thinking. It legitimates the subliminal association of salespeople with fiduciary advisors. Regulation not only allows them to parade their ‘government licenced’ status but actually requires practitioners to play the role of advisor by mandating the provision of continuing risk analyses and written investment plans.

These plans generally comprise slabs of standardised text drafted before the event by investment sales executives, vetted by lawyers and disgorged from software packages upon entering of client data. The software gives the impression of independence and expertise, but is promoted as ‘sales technology’ within the industry.

So here’s how I think regulation should work. Beyond basic fee disclosure and effective dispute resolution, we should require any advisor recommending a product charging fees above some minimal level to provide some simple measure of the likelihood that this will pay off with higher net returns.

In the US where we have the best evidence, around 90 percent of investment managers underperform the market. And in Australia the ‘industry funds’ to which award super has usually been paid have generally outperformed commercial funds. Not having to remunerate investment advisors, they charge lower fees which as we’ll see add up over time.

We also need a system where advisors keep independently auditable ‘sample portfolios’ operated in ‘real time’ to avoid “advisors” selecting their best picks (and forgetting their duds) after the event. Then we could measure their performance. The investments in these portfolios would be confidential to protect advisors’ investment management secrets. But their performance in terms of absolute returns, after-tax returns and volatility would be published.

The Government or even an enterprising Opposition – might kick this off with public encouragement rather than compulsion. If leading figures called for action and initiated co-operative work on reporting standards to ensure comparability of performance measures, some industry leaders might start self reporting to highlight their own good performance. But over time I expect it would make sense to regulate to bring everyone into the fold and scale back other regulatory excesses.

And you’ve heard of the miracle of compound interest. You’ve got to decide whether you want it working for you or your ‘advisor’. If you were earning $50,000 a year, your fund was earning a 5% real rate of return and my regulation helped you avoid the 0.25% per annum trailing commission plus GST that most funds pay to advisors, your pay-out after 35 years of compulsory super would be more than $28,000 or 5% higher. If my system helped you pick a fund with an annual return just 0.5% higher again hardly a stellar result it would increase your ultimate retirement nest egg by $85,000 or over 14% on the base scenario. You’d be richer and our economy would function more efficiently and fairly.

Choice of fund is a great idea. Now let’s make it an informed choice. Oh, and there’s something else we could do with super that could make everyone even better off again including the Government. I’ll tell you next week. Until then . . .

This entry was posted in Uncategorized. Bookmark the permalink.
Notify of
Newest Most Voted
Inline Feedbacks
View all comments
2022 years ago

Choice of super fund is a great idea.

I’ll be choosing Australian Ethical Investments. I am well informed on this subject, Nick.

Cameron Riley
2022 years ago

Nicholas, the software industry uses what is called “Unit Testing”. This runs each time the project is compiled. The customer level testing, “Functional Testing” is often run each night on a server as part of the Continuous Integration process. Software is an information management process (compilation is the only contruction in it).

One of the interesting aspects of the hardcore unit testers is that they build the test first, and then the software code to match the outcome required by the test. It is like backward design, but works extremely well for having confidence in the source codes output.

Another interesting aspect is that these are automated. From testing a method, up to testing the requirements. I read through your PDF, since software developers have to deal with rigourous outcomes and a wash of information, it is possible that some of the software development technologies and procedures may translate to other areas.

Good article. I enjoyed it.

Jacques Chester
Jacques Chester
2022 years ago


I’ve often felt the same way. A lot of concepts from software engineering could be applied to law – loose coupling, refactoring, modularisation, contracted interfaces, unit testing etc etc.

Personally I think that the key to introducing a lot of these concepts might be to introduce a language which compiles to legislation-ese. Perhaps it could be a purely functional language, which has the benefit that you could automatically generate proofs of behaviour under all conditions during unit testing.

On the other hand, it’d be tremendously difficult to develop such a language; and in any case domain-specific languages have tended to be failures (COBOL for instance).

2022 years ago

Why don’t we just make legislation-ese plain English. Then we can leave the ‘compilers’ where they belong – in our heads.

Jacques Chester
Jacques Chester
2022 years ago

There’s 2 problems with “plain english”:

1. Legalese is a proven language. We use and reuse certain phrases because their meaning has been precisely determined by court cases. This is not true of regular written English. Lawyers are very conservative in this regard – if it ain’t broke, why rewrite it?

2. Following on the above, plain english is notorious for not being very clear. This is what you get in a language which is the bastard offspring of Europe (though I love it). Things as minor as apostrophes can change there entire emphasis of sentences: “We’re in town” versus “Were in town” and so on.

The general idea is that a formal language, while dull and not necessarily much more accessible than legalese, can at least be tested for common problems in advance, and in great detail. We geeks call this “testing the common case and boundary case”, and by knowing our tests in advance we can also test interactions between software, which we call “regression testing”.

2022 years ago

I hope the industry super funds do really well out of super choice.

They may not have the marketing budget of the big names but I do believe they perform better and hopefully the smell of money will attract punters despite all the hype.

2022 years ago

As to (1): I’m not sure about this – the unusual phrases lawyers use and reuse are mostly the product of centuries-old case law. There is no need for statute to reproduce the sometimes absurd vocabulary of the common law. For example, s.52 TPA: ‘A corporation shall not, in trade or commerce, engage in conduct that is misleading or deceptive or is likely to mislead or deceive’ Simple and brutally effective. There is no legal problem that cannot be solved with equally plain language.

As to (2): I don’t understand what imprecision has got to do with plain English. If there is a need to be clarify something, use more plain English. Lawyers now produce very large agreements (dealing with very complex commercial arrangements) composed of plain English. I also don’t understand what a mis-placed or omitted apostrophe has got to do with plain English.

Having said that, I’m intrigued by the notion that a specialised testing system could be set up to look at common problems in drafting legislation. God knows the drafters could use it. All these computer references are over my head though. Perhaps you ‘geeks’ could look at introducing a bit of plain English into software development.


2022 years ago

ab, fair enough. But every now and then “legalese” throws up a moment of poetry. How would you have written “with malice aforethought”, for example?

Jacques Chester
Jacques Chester
2022 years ago


Your response to my point 1) correctly pointed out that a lot of language lingers because its meaning has been established in court. While there has been a plain english movement, its scope will always be limited by what one might call backwards-compatability issues.

As for English, it is in formal terms an incredibly ambiguous language. AI researchers have been banging away at “Natural Language Processing” for years and nobody has really gotten a good handle on it yet – what we have mostly is an intricate and comprehensive collection of known problems in natural language parsing. Solutions are some way off yet.

What seems to you and me to be perfectly clear can in fact reveal several meanings when examined in a formal or strict setting. Human beings enjoy tremendously sophisticated apparatus for decoding language, even so we can fail.

To give another example, consider the sentence “I attacked the burglar with a gun”. The sentence has two distinct and quite different meanings: it might mean that the burglar who I attacked had a gun. Or it could mean I used a gun to attack the burglar.

The point of developing a formal language is that every statement in that language should have one, and only one, meaning. When designing a computer programming language this is a necessary precondition for everything else – compiling, testing, automated formal proofs, the works. If you design a language in which a given statement can have two meanings which cannot be deciphered within the context of the program, your language cannot be compiled or interpreted.

Interestingly, Perl is one language which places a very English-like emphasis on contextual understanding. Variables can be referred to by implication, rather than literally every time they are adjusted. However much of this is what’s called “syntactic sugar”: it means the same as the narrow formal constructs but makes things easier for humans. There is heated debate in the computer languages community about whether this approach is superior or not.

Turning to the question of a formal legal language, the advantages would be that it could be tested, and designed in such a way that each statement must necessarily have a single and unambiguous meaning. The ability to test that a given piece of legal-code will meet preset tests, interacts nicely with existing legislation etc etc would I think be an enormous boon to legal drafters and to the community of law in general.

Later on you would trickle down these tools into other areas. If the language includes mechanism for describing self-contained rules in the AI sense, it offers a doorway to the progressive computerisation of legal reasoning. While I imagine this would be heartily opposed by the legal profession in all respects, it might nevertheless offer one way to drive down the costs and complexities of legal reasoning, while improving its quality.

Finally, putting aside fantasies about formal languages and machine-assisted legal reasoning, the simple act of drafting legislation and other legal documents could be aided by the wholesale adoption of software engineering techniques. The job of the programmer is to turn a problem (the domain, borrowing from maths) into a representation that a computer can understand and manipulate. Consequently programmers and software engineers deal in abstractions heaped upon other abstractions. Over time the field has developed an array of techniques and methods for coping with exponential and constantly increasing complexity. You might say it’s our specialty, and the results speak for themselves.

Since law is itself about dealing with concrete matters via abstractions heaped on abstractions, the two fields have a lot in common. In terms of actually constructing law, however, we’re stuck in the software engineering state of the art circa 1960. There’s no concept of information hiding (pieces of code shouldn’t have direct access to others information), no concept of loose coupling (code shouldn’t rely on the internal working of other code in order to work – “deep linking” makes code fragile), modularisation (seperation into logical concerns allows problems to be solved in isolation), refactoring (identifying design mistakes and rearranging existing code to better represent the problem domain), unit testing (testing modules of code individually and together to preset testing standards whenever the code is altered) and so on and so forth.

Here endeth this round of ranting. Sorry for going so off-topic, Nicholas.

2022 years ago


Most of that is over my head, so I’ll just say that a lot of what you seem to want is given in the Acts Intepretation / Interpretation of Legislation Acts.


I think I’d have to go for ‘deliberately’.

Everyone else,

I hope the industry super funds kick arse too.

Nicholas Gruen
2022 years ago

Jacques, no problem at all. What’s up here may seem off topic, but its not off topic for me – as it relates to what I’m really interested in and what was suggested in my introduction rather than the column, which is the efficient design of regulation.

The other mercy is that there is much light and little heat. While a fair bit of what you write is technically over my head, I flatter myself that I may understand things of relevance to the conversation. In particular I’d be prepared to lay bets that we can’t express most regulation – or most important regulation in unambiguous commands.

Economics has been through that fantasy too.

This is what Samuelson said about it.

Let me make a confession. Back when I was 20 I could perceive the great progress that was being made in econometric methods. . . . I expected that the new econometrics would enable us to narrow down the uncertainties of our economic theories. . . . My confession is that this expectation has not worked out. . . . [I]t seems . . . that there does not accumulate a convergent body of econometric findings convergent on a testable truth.

There is something else from the world of software engineering that I think is more germane to good regulation though, and that’s the richness and immediacy of *feedback*. Our regulatory systems have not been built with this in mind, and I believe this is an important part of what ails us.

Cameron Riley
2022 years ago

Nicholas, Unit/Functional testing for regulation is an interesting aspect. The policy would be the requirements. The government/bureaucracy execution of it would be covered by unit testing. The end customers of that regulation would be covered by functional testing.

One of the aspects of software unit testing is that it is automated, repetitive and run at every possible chance. How would you automate unit/functional testing on the inputs/outputs of regulation?

Jacques, In the SSR book Scrymarch named the section on modifying government “Refactoring Government”. Which is a good description, as it is redesigning government (process) without alterings it outcome (liberal democracy).

Homer Paxton
Homer Paxton
2022 years ago

I confess that I would like to join an Industry Fund but never will because of their appalling ‘choice’ on life insurance.
Their product is absolutely hopeless.

I have yet to hear from any fund I have tried to join why they do not or cannot offer life insurance that the retail fund and everybody else offers.

For those who don’t know it is more tax effective to have life insurance with a super fund.

Jacques Chester
Jacques Chester
2022 years ago


I apologise if I went over anyone’s head. I’m not trying to win an argument using proof by obscurity. Feel free to quiz any terms I have.

I should add generally that while I don’t share the view that things can be done perfectly, it shouldn’t stop us trying. As I say, the core process of law and software are the same – dealing with concrete problems using abstract tools. Both are rooted and reason, both grapple with complexity. The difference is that one has moved much faster than the other in the question how to deal with complexity. The other has remained stuck in the first ideas it came across, its horizon hidden by high walls of detail.

2022 years ago

curiously, one way of imparting a message to potential super consumers is advertising, yet there is a growing suspicion that if they advertise, particularly on telly, one should totally avoid them, for it costs money to advertise, that’s money that could be invested

but how else is the lumpen super savers be contacted?

the only feedback here happens at the end of the cycle, when you are on the retirement scrapheap, how can that feedback into rational choice …we’re all gar-gar

and or over medium length time periods, but swapping funds will cost, …. should i stay or should i go

and what if the funds are wiped out because there is a war (with china say) and super funds are as useful as nazi or confederate war bonds

50 60 years is a bloody long time, performance/measurement is meaningless over such a period of potential histories

of course i am a dolebludger and regard the whole thing as a mouse-in-a-wheel for economic robots

its a pity we cannot have a real conversation about the diversity of ‘incentives’ in our lives

the only solution i can see coming from the market is that the better perfoming funds be made illegal or marginalised, because they cannot compete with them, socialise the losses, privatise the profits and criminalise the best performers…

and its seem the liberal government agrees

ideology and aetheneum club ingroups at their worst

and to think aetheneum clubs were once outposts of liberalism

15 years ago

sorry all. Just forgot to uncheck the ‘notify me’ box, so I’m getting emails. Just fixing that.

Don’t mind me.

I suppose I should have linked to 12 porn sites, then it wouldn’t have gotten through to bother you all.