I’m at the Australian Conference of Economists in Perth, where I’m presenting a paper on water policy. I went to a paper today on ‘home bias’ and the International Capital Asset Pricing Model. The general home bias story is that most people hold most of their assets in their own country, when standard theory suggests the optimal portfolio should be much the same for everyone, regardless of location.
Following this up, this suggests that everyone should hold assets in a given country in proportion to its vaue share in the total world market (about 2 per cent for Australia).
It struck me that there’s a paradox here. Suppose I’m considering investing in two overseas stock markets with similar characteristics (mean return, variance, covariance with the Australian market). It seems obvious that I should invest the same amount in each market. But, if you accept ICAPM this is wrong, unless the two markets have the same capitalisation. ICAPM implies that I should invest more in whichever market is larger.
The paradox can be resolved if larger markets are more stable, but casual observation doesn’t support this. Has anyone seen this issue addressed?
John Quiggin 
ICAP is wrong and specifically in the Australian case does not account for franking credits. Franking credits totally bias the portfolio towards Australia due to their tax efficiency. Even so by taking this into account the home bias is still huge.
Another reason why Australia has a higher weight is the statistical conclusion that correlations have moved higher across the global so that the case for diversification (in a more globalised world) is not as strong. For example RIO Tinto is listed in Australia and the UK. Is it an Australian company? Is it tied to Australia? Or is it really tied to global and in particular emerging market demand for its commodities.
Also ICAP assumes that markets are efficient (they are not!). For example the developed benchmark has an allocation to US equities of over 50% since 2000 which has been a losing proposition. It also had an allocation of nearly 30 (or was it 40%) to Japan at the height of their bubble so I think the global benchmark is useless as a tool to allocation your global money based on their country weightings.
Benchmarks are like to MSCI (Morgan Stanley Capital International Index) not the only way we can look at allocation across markets. Specifically the developed benchmarks exludes emerging markets like Brazil, India and China etc etc.
In the institutional market, super funds have addressed this by two ways,. First they might allocate a core (but less so these days) to a passive fund that follows the benchmark like the MSCI and then second, allocate the rest to an active manager who typically builds concentrated portfolios of stocks without regard to country or regional weightings. For example one of Australia’s (and the world for that matter) premier investors, Kerr Nielsen from Platinum runs global equities without much regard for what the allocation to country or regional markets say based on the MSCI.
On the basis of the information supplied regarding the specific model and the specific data and the estimation method, I can’t say much more than what applies to any CAPM, namely the model(s) assume complete spanning (complete markets) and this assumption is not an adequate approximation to reality.
John,
Is the following the paper in question?
http://www.crefa.ecn.ulaval.ca/cahier/0008.pdf#search=%22%22International%20Capital%20Asset%20Pricing%20Model%22%22
There is also the question of imperfect information. Even these days, it’s easier to get better investment information and understanding with a home bias. That’s in regard to equity markets.
In regard to investment in property, institutional barriers to entry can be extraodinarily high (e.g. no foreign ownership permitted) or present risk profiles (joint ventures with local partners required) that would basically prevent non-residents from ever investing in those markets.
2 tanners is right; it’s another asymmetric information effect. I’m sure I’ve read an NBER paper making this point formally (sorry, I don’t have time to chase it this morning).
The other side being ignored is transaction costs and forex margins. If I have $100K to invest, trying to build a world-straddling portfolio transaction costs are going to very quickly eliminate any benefits of such a diverse group of assets.
Throw in forex margins and charges and the return of any group of foreign assets will be minimal to negative.
For small to medium individual investors, therefore, foreign investing normally just isn’t worth it.
Even for the big boys there are still significant additional costs to investing overseas. Lack of market knowledge to correctly estimate the beta also hurts.
The theory is great – practice is another thing.
Both 2 tanners and Andrew Reynolds are correct. In the Australian market there are significantly higher returns to be made in the medium and small cap market due to inefficient information. If this wasn’t the case then boutique investmnt managers would die due to under-performance and everybody would use Index Funds.
As somebody who is familiar with CAPM and EMH and actually managing portfolios in Australia I would also add that while the theory is nice, most of it is based on historical information, where as investment decisions are about future potential returns. An example – in the period 1999 to 2002 international equity markets hit all time highs before falling while the AUD was historically very weak. Unless you are a devotee of the “time in the market – not timing the market” school, why would you buy expensive assets with a cheap currency? I didn’t and my clients’ portfolios outperformed benchmark portfolio containing the “recommended” international exposure.
I don’t think Andrew Reynolds is correct — it is extremely easy to trade some overseas assests, like shares, commodities etc., and you can get fairly reasonable exchange rates depending on the currency you want to convert and who does the conversion. The only reason that people think it is hard is that the Australian banks are not well set up. Some OS bank accounts (like HSBC) allow you to trade in almost anything (shares, gold, oil, currencies, bonds…), and the only loss you are likely to encouter is the forex loss, and this is only meaningful if you want to exchange currencies a lot (and presumably for investment accounts, you don’t, unless you want to for profitable reasons). Most of this is just with a standard bank account — you just tick a box for things you want to be able to do.
Razer is not correct in saying “that while the theory is nice, most of it is based on historical information, where as investment decisions are about future potential returns.” The CAPM is a characterisation of a solution of a model of people who are making investment decisions.
John Quiggin:
“It struck me that there’s a paradox here. Suppose I’m considering investing in two overseas stock markets with similar characteristics (mean return, variance, covariance with the Australian market). It seems obvious that I should invest the same amount in each market. But, if you accept ICAPM this is wrong, unless the two markets have the same capitalisation. ICAPM implies that I should invest more in whichever market is larger.”
First a clarifying question. What do you mean by ‘similar’?
If I interpret ‘similar’ strictly in the framework of a theoretical model of the CAPM type, then there are 3 securities (portfolios) in the ‘world economy’, each one being uniquely characterised by a set of real numbers {mean; variance} and there is an associated unique variance-covariance matrix. But, if I interpret ‘similar’ to mean ‘the same’ then there are only 2 securities (portfolios), one of which is sold under two different names (in the real world). In the latter case it would follow (under ‘standard’ CAPM assumptions; I don’t know the actual version of the ICAPM you have in mind) that investing in the 2 (foreign) securities in equal amounts is optimal at the point in time when the announced market capitalisation of the 2 (foreign = empircally differently named) securities is identical . That is, the investor would ‘commit’ at this point.
Perhaps you want to convey with the term ‘similar’ the more of less continuous price adjustments. In this case ‘similar’ could mean ‘as close as one can get in real time to a theoretical solution’.
Correction. The second last sentence should read: Perhaps you want to convey with the term ’similar’ a more or less continuous price adjustment process.
BTW PrQ, I should have welcomed you to the best part of the country – and a perfect day it is, too. Is Curtin still looking like a large set of concrete barns? It does not look like it has changed much from where I am sitting.
Ernestine – testing of CAPM and EMH is done with historical data – unless they are making itup.
Conrad – it is effing expensive to transact international trades from Australia using full-service brokers. I don’t know about internet brokers. And the paperwork to set up the accounts is a pain in the arse, too.
If you leave out the assumption of perfect information then home bias is easily explained as the rational behaviour of investers with costs to gathering information.
Any investor is going to find it easier and cheaper to get information about investments that are closer to home, particularly ones in their own country. They will have information on local market conditions, political environment, laws, corporate culture etc etc almost without making any effort to acquire it. To do the same in an overseas market, on the other hand, is much more costly.
For similar reasons an Australian is more likely to invest in markets in America or Britain rather than Germany or France. They are simply going to have more information available to them about countries that they share a language or history with.
If standard theory really does suggest that the optimal portfolio is the same for everyone then its conclusions are absurd. Everyone’s access to information is different, which suggests that everyone should have different portfolio’s. I am not an economist so feel free to correct me if I am getting my understanding wrong but does standard theory really ignore the cost of information gathering ?
Razor, yes, the testing of the theoretical model involves historical data but this is not the same thing as saying “that while the theory is nice, most of it is based on historical information, where as investment decisions are about future potential returns.â€? (in practice CAPM type models tend to be used to calculate benchmark ‘cost of capital’ for the purpose of longer term investment decisions in physical capital and to calculate bounds on trading ranges).
“If standard theory really does suggest that the optimal portfolio is the same for everyone then its conclusions are absurd. Everyone’s access to information is different, which suggests that everyone should have different portfolio’s. I am not an economist so feel free to correct me if I am getting my understanding wrong but does standard theory really ignore the cost of information gathering ? ”
To the best of my knowledge, there is no ‘standard theory’ on how ‘assets’ are priced internationally. We are talking about more than one theoretical model in the class of models known as ‘International Capital Asset Pricing Model’ (ICAPMs).
None of these models “suggest” anything. These models consist of a set of assumptions (very much akin to what people who advocate ‘free capital markets’ advocate) from which theoretical results are deduced. These are called ‘solutions’. In models such as the CAPM or the ICAMs, the solutions are in the form of equations that consitute testable hypothesis. The solutions may have other implications, which can be tested against observables. Another way of assessing these models is to check the assumptions against observables. This is a bit more tricky because it involves a bit of judgement, eg, are these assumptions adequate approximations of reality, at a particular time, at a particular place.
I don’t like the term ‘bias’ in the specific set of literature referred to in JQ’s post. In which sense could an ICAPM, which does not hold up well when tested empirically entail a bias? What or who is biased against what or whom? Surely, we don’t want to have a confusion of theoretical models with reality that might lead to a perversion of the research program in the sense that instead of rejecting a model, we end up rejecting reality! Reality, in turn is not a fixed state. I can’t see anything wrong having models which assume what some people advocate as desirable (say ‘free markets’) and to have a little look on how observations compare with the model. Some of these models have been tested many times and the overall conclusion is that they provide a sufficiently good fit to observables to be of practical use in some circumstances, but not necessarily everywhere in the world and all the time. For example, the simple CAPM was tested with US share market data and some European share market data a lot in the 1970s and 1980s. The results were very much better for the US than for Europe. Not surprising, one might say, given the additional information that share markets did not play as big a role in continental European countries, at the time, as in the US. However, I would not agree to use test results which show that the model does not ‘work well’ as evidence that reality has to be changed because reality is ‘biased’. In short, the term ‘home bias’ makes no sense to me at all. (There is probably a shorter way of expressing the point in this paragraph but I could think of it tonight).
Regarding your paradox (which applies to the theory not the empirics): Ernestine’s lengthy comment seems right.
If the two overseas markets are “identical,” then it doesn’t matter which one you invest in. The observed market sizes and portfolio shares can thus easily be consistent.
Of course, what does “similar”/”identical” mean in this context, and what’s happening inframarginally?
ICAPM is a poor way to manage/allocate money. Nice theory but completely useless in the real world.
Also making investments based on market caps is misleading. Better to make investments based on valuations compared to history eg buying low price to earnings , high dividend yields etc. Over the long term this is a sound way to manage money as opposed to country or industry weights based on market cap. The efficient school will violently disagree but research and performance studies back my conclusions.
Con,
Can you please provide a link to the research? I would be interested to have a look at instances of value investing consistently beating index tracking.
I would start with the following link from wikipedia on value investing. The seminal piece is contained in the reference section containing the paper by Fama and French.
http://en.wikipedia.org/wiki/Value_investing
Razor,
It is cheap as chips for a retail investor to trade on the US market through an online broker who is set up to do it (cheaper in fact than the same sort off service in Australia).Of course, institutional investors very often have their own brokers inhouse, which lowers their transaction costs considerably.
Full service brokers are charging more than transaction costs, they are charging for advice and insurance. The knowledge/information assymetry means that to advise on an international trade is going to take specific research which is going to cost.
still working it out , You specifically mentioned information acquisition costs. I didn’t deal with this item – my post was already long. Just a few words on this one. I am writing a bit in a vacuum because I don’t have a specific point of reference for this thread. But I can say that in the CAPM context, ‘heterogenous expectations’ about future payoff distributions can be handled without changing the structure of the theoretical solution. To the extent that information acquisition affects expectations, an element of information asymmetries is consistent with the general models.
To provide a point of reference as to the relative difficulties: Do law makers ask how much it costs for people to find out what the laws actually are? How is the meaning of some laws determined? Isn’t it via a complex and expensive process involving a hierarchy of courts?
I fully agree with you that information asymmetries are a major problem in societies and I tend to agree with you that geographic distance and language differences affect information acquisition costs (it comes out in the literature on multinational firms and in daily life)
To Ernestine and John H,
“If I interpret ‘similar’ strictly in the framework of a theoretical model of the CAPM type, then there are 3 securities (portfolios) in the ‘world economy’, each one being uniquely characterised by a set of real numbers {mean; variance} and there is an associated unique variance-covariance matrix.”
That’s exactly what I mean, with “similarl” now meaning that the mean returns of the two foreign markets are the same, and the variance-covariance is unchanged by permuting them.
If this condition is satisfied, it seems obvious that I should invest equal amounts in the two markets, regardless of their relative capitalisation.
John Q.,
Thanks for the reply. So far I have only enough information for a CAPM and not for an ICAPM. A possible solution to a numerical example of the model that is consistent with the information I have could look like {.3, .5, .2} (ie normalised security prices).
What does ‘investing equal amounts in two markets, regardless of their relative capitalisation’ mean? Which ‘market’ is bigger?
In my possibly not so humble opinion (IMPNSHO), there are fundamental conceptual problems with the ICAPM which don’t apply to the CAPM or its multi-period extension.
May I suggest you send me either a copy or a reference to the paper you have in mind by email? Alternatively, the paper I linked to above could be used as a common information set.
A question from a mug punter: Two companies have similar (but not highly correlated) prospects. Why should I invest more in the bigger one? Answers in plain English please. If you refer to CAPM, or any other model, explain how the conclusion follows from the premises.
This is a genuine question, not a rhetorical challenge. On the one hand, investing according to index weightings seems to make sense. On the other hand, I’ve never seen a killer argument for it. The kerfuffle over News Limited’s move to the US illustrates some of the issues.
a mug punter, please explain in plain English what you mean by ‘a bigger one’.
Is there any model which tells you what you should do? The CAPM and the ICAPM, which I know, do not prescribe to any individual what they should do.
JQ
Yes based on your strict assumptions of mean/variance I would allocate equally. However if we include other variables like tax (ie benefits of imputation credits to Australian investors), maybe market cap (ie prefer a liquid name over an illuiquid one all other things being equal) then the there will be biases in one stock over another.
Apologies, Ernestine Gross. I should have read your earlier posts more carefully. The basic point I’d missed is that typical models adjust the prices to clear the market. In the case I was imagining, this would make the bigger company cheaper. Whether the real world works like this is a different question.