PensionReforms' summary and comments
The direct ownership by households of listed shares ('stocks' in the US) has shifted dramatically over the last decades.  In the US, about 90% of listed shares were owned by individuals after World War II; that had reduced to about 30% by 2006:
"A large portion of households' ownership shares has migrated to financial intermediaries which manage private pension plans such as pension funds, mutual funds, and life insurance companies.  A similar shift of ownership shares from households to carriers of retirement assets has taken place in all countries for which long time-series of stock ownership data exist.  With this in mind, the cross-country evidence [of 20 developed countries] on the fraction of household ownership in recent years is telling.  There is not a single developed country where households own more than half of the equity market directly, with the average across countries being just 17%."

This seems not to be caused by higher costs or information shortfalls.  Those both seem to have fallen for individuals.  Financial innovation seems not to be the cause either as most of the shift took place before that became prevalent.  It seems that tax concessions for pension schemes might be the reason.  The report looks at this issue using macro data, rather than micro/household level data.

PensionReforms notes that most countries offer significant, expensive tax advantages to retirement saving schemes.  Contributions by individuals (and their employers) are usually deductible against other taxable income and the investment income earned by the scheme is usually exempt from tax.  The benefits on retirement are normally taxable (but often not the whole benefit) but retirement tax rates are usually lower than income tax during the accumulation.  As the report notes, this is the "consumption tax" approach (by contrast with the "income tax" treatment of ordinary income).  The eight countries examined in the report all use this.  PensionReforms notes that New Zealand had the consumption tax approach - from 1990, it switched to the income tax treatment.

PensionReforms notes that the shorthand way of describing the tax-advantaged, consumption tax environment relates to the three movements of money - contributions; investment income and benefits - E (contributions); E (investment income and T (benefits) - EET for short.  The most 'powerful' of these three components is the middle one - the way investment income is treated.  That's because of the very long saving period usually involved.  The report finds that this treatment seems to explain the shift of share ownership, rather than the income smoothing effect of lower tax rates in retirement.  That effect has seemingly been magnified by a combination of increasing personal tax rates and inflation (with the 'bracket creep' normally associated with inflation) and has been relatively slow to develop.  The shift in response to the influences noted has taken "...a half century to complete...".

"We have provided panel-data evidence from eight countries that changes in the fraction of household ownership is correlated with proxy variables for marginal tax rates.  Ownership in the eight sample countries follow different paths depending on features of the tax code and exposure to inflation in the 1970s and the 1980s.  As inflation takes off, the fraction of household ownership decreases fast in the United States, United Kingdom, and Sweden where marginal tax rates are high and long-term capital gains are taxed.  At the same time, the fraction of household ownership decreases slowly in Germany and Japan with tight monetary policy and no tax on long-term capital gains."

The report also finds that "...stock prices would have been much lower without the dynamic tax clientele shift that we have observed."

The report suggests a range of further possible research topics that, for example, might look at the impact of the "...tax theory of pension funds [that] may explain the growth and prevalence of inter-corporate ownership in many countries."  That may be connected to the presence of pension liabilities and the possible need to hedge those.

The report also wonders whether the same tax-influenced ownership effects apply to the ownership of real estate and bonds.  PensionReforms would be extremely surprised if it were otherwise.

The report does not address the question whether the move from direct to indirect ownership is a good thing (PensionReforms thinks the answer to that should probably be 'yes'); nor the much more important issue whether the expensive tax incentives should be deployed by the state to that end - in other words, is the distortionary effect noted in the report justified?.  PensionReforms suggests not.  Not only are the incentives distortionary and very expensive (in terms of revenue forgone) they are also regressive, costly to administer, inequitable and probably do not increase saving overall.  In other words, they seem not to work.  If a government's revenue were all based on expenditure taxes, there might be some justification for the EET "consumption tax" model.  Most countries in fact collect most tax from income so the TTE "income tax" model seems the preferable model. (File size 550 KB; 59 pp) 344