It is very common to compare average asset mixes of pension funds across countries as a way of explaining fund performance, pension costs and investment cultures. However, assessing asset allocations of pension funds in general may be more difficult than one might expect.

To quantify pension-related assets it is necessary to identify the different sets of pension assets to avoid any overlap and/or loss of sub sets. To compare the same or similar sets of assets in various countries it is quite important to distinguish between:

q type of provision (eg state pension, private pensions, occupational pensions, savings, insurance contracts);

q type of investor (ie retail and institutional funds); and

q type of financing vehicle (eg pension funds, book reserve, insurance contracts, support funds, Pensionskassen or deposit administration funds).

Many countries, such as Austria, Japan, Spain and Germany, use more than one method of financing occupational pensions.

In Germany, for example, there is no such thing as a pension fund. Occupational pensions are financed through one or more of the following methods: book reserve (60% of occupational pension assets), Pensionskassen (20%), support funds (10%) and direct insurance contracts (10%).

There are ratios which describe a fund's exposure to equities (the equity-debt ratio - EDR) and its exposure to foreign securities (the foreign-domestic ratio - FDR). The book reserve method accrues the pension liabilities on the company's balance sheet, thus the assets are totally invested in the company (ie an EDR of 100:0).

Support funds can invest in almost any prudent asset. Sometimes support funds grant back a loan to the sponsoring company and/or use reinsurance. Therefore the exposure to equity might vary between zero and as much as 60%.

Pensionskassen, direct insurance and other insurance companies are forced to invest at least 65% in German bonds and their typical equity exposure is currently between 10% and 35%. The asset allocation for pension funds" in Germany most often quoted is that of Pensionskassen. It is important to know that these numbers are based on book values and understate the equity exposure.

Because occupational pension schemes in Germany can and do use a mixture of approaches to financing it is very difficult - and may be nonsensical - to try to assess an average asset allocation for occupational pension assets.

One way of quantifying a broader asset allocation is to use the asset allocations of Spezialfonds. Spezialfonds are used by more or less all the approaches to financing. However, Spezialfonds are also used for non-pension investments. The Spezialfonds market as at 31 July 1997 accounted for DM521bn ($288bn) of assets and had an EDR of 37:63 and a FDR of 22:78. Interestingly, non-institutional funds in Germany are invested more aggressively (EDR of 43:57 and FDR of 40:60). This is in contrast to the US, where it is the institutional investors who invest more aggressively.

Austria, with its book reserve schemes and mandatory use of "Wertpapierdeckung" (security coverage), is another good example of a country with several different financing methods. Pension schemes financed through book reserve in Austria have to hold 50% of their book reserve in government bonds. Quoted asset allocations from occupational pensions in Germany and Austria are typically ignoring equity exposure of book reserve. This demonstrates the danger of making simple country comparisons, without taking into account the different approaches to financing pensions from country to country.

Differing economic backgrounds also help to explain why pension fund asset allocations are different from country to country is due to. Inflation history and currency stability play an important role. In the UK, inflation has been high in the past (7.5% pa between 1963 and 1996) and UK gilts gave only a modest real return (1.6% pa over the same period). UK gilts even underperformed cash over this period (2.3% pa in real terms).

Against this background it seems reasonable that UK investors should look for a higher equity exposure than German investors. German inflation has been fairly low (2.8% pa between 1952 and 1996) and German bond returns have been better than German equity returns in the long run. In fact German bonds outperformed German equites between 1960 and 1996 (equities 2.45% pa and German bonds 3.74% pa in real terms). In the period between 1960 and 1980, German equities did not give any positive real return. Although some other periods show different results, it becomes easier to understand why the German investment culture tends to be equity averse.

However, in Spain, inflation has been high and equities have outperformed bonds, but equity exposure is almost unknown. This means that inflation is not the only explanation of a low EDR. The Spanish equity market is fairly small, illiquid and not fully transparent, leading to lower confidence in equities. In addition, in Spain we find also a low FDR, which is probably due to psychological effects.

The relationship between equity confidence and investors' equity return expectations is investigated in the study 'Explaining the Home Bias' by Michael Kilka and Martin Weber. According to this study the median German investor had 50% confidence in German equities and 30% in US equities (100% = full confidence, 0% = no confidence). Expectations about return were similar, ie the median German investor expected on average a return of 8.6% pa from German equities and only 4.4% pa from US equities.

Although the overall confidence in equities is higher for a US investor, the relationship between confidence and expectation of return was also biased in favour of home. The median US investor had 39% confidence in German equities but 58% in US equities and expected an average return of only 2.8% from German equities but 6.0% from US equities. This leads to the conclusion that confidence might be an important explanation for the heavy home bias of investors.

Investors may also have different objectives, based for example on their liability profile. A pension fund with a mature population and no surplus cannot afford to invest in short-term volatile securities, such as equities.

In the same way, this could explain different asset allocations between countries. By 2040 the old age dependency ratio (persons aged 65 and over as a percentage of those aged 15-64) will be nearly 50% in Germany, Italy and the Netherlands and only below 30% for Ireland, according to Eurostat figures.

In addition, there may be a relationship between expected relative pension cost and equity exposure (ie EDR). Therefore, it is useful to express pension cost as a percentage of GDP. Low EDR in Italy, Luxembourg, the Netherlands and Germany might be explained by an expected ratio of pension expenditure to GDP between 15% and 20% by 2030. In Ireland and the UK this ratio is expected to be below 10% by 2030.

Economics, investment culture and demographics may be some explanations for different asset allocations. But trying to compare different countries' asset allocations without considering other factors such as different types of pension provision, the difference between book and market values, and different investment objectives will not provide the right answer. Instead, it will be literally "comparing apples with pears".

Olaf John is a consultant, asset consulting services, with Towers Perrin in London"