Overcoming technical objections to Colm Fagan’s AE proposals

[Brendan Burgess]

It would be useful to list out the technical objections to Colm's proposal so that they can be identified and dealt with.

Again, I would stress that these are not my opinions but a list of the objections I have seen.

What happens if we have a prolonged downturn in the stock markets like Japan?

The past is no guarantee to the future, so we don't trust the simulations.

It's too inflexible - people have no choice of whether to contribute or not, pause their contributions or increase their contributions.

Customers will be forced to buy units when the price is above the value of the underlying assets.

Customers will be forced to sell the units in retirement when the underlying value is higher than the smoothed value they are getting.

There is no need for any form of smoothing

AE does not need any form of smoothing

Pension funds should be fully invested in equities during their working life They should stay invested in equities in their retirement Lifestyling is nonsense I would in general agree with these. Some form of smoothing/risk-sharing is a good idea Sorry Brendan, butneither are a good idea...

www.askaboutmoney.com

 

[Colm Fagan]

Ok. I'll try to take the technical objections in order, starting with:

What happens if we have a prolonged downturn in the stock markets, like Japan?

I deal with that in Section 3 of the paper to the Institute and Faculty of Actuaries (pages 12 and 13). Sorry for constantly referring back to the paper to the actuaries, but it did anticipate nearly all of the technical objections.
Here is the relevant section of the paper:
The question is: how would it cope with a more prolonged downturn?
The Japanese experience since 1990 provides such a test. From its all-time high on 31 December 1989, the Japanese market fell precipitously – down 40% in 1990, down another 25% over the next two years. After three years, the total return index was just 45% of its starting level. It staged a partial recovery in the early 2000’s, but suffered a severe relapse during the Global Financial Crisis, falling more than half between 2007 and 2009. By January 2013, twenty-three years after the initial collapse, the Japanese total return index was less than 40% of its starting level. It recovered strongly in later years but was still only 82% of its January 1990 level by 1 January 2020.
Figure 5 shows the progress of Japan’s TOPIX Index with dividends reinvested in the 30 years from January 1990 to December 2019.
Figure 5

Assuming the same trapezoid pattern of cashflows as above (increasing for ten years, staying constant for ten years, then declining through zero after 30 years), the progression of smoothed and market indices (adjusted for cashflows) is as follows:
Figure 6

The fall in market values in the early years shown in figure 6 is not nearly as severe as that shown in figure 5. This is because new cashflows are being invested at lower prices, and the benefits of the improved terms are being shared across the membership. Smoothed returns are negative in some years (worst calendar year smoothed return is -0.7% in 2009), but the lowest smoothed ten-year return is a positive 19.4% (average +1.8% a year, between 2003 and 2012) and the average smoothed return over the entire 30-year period is +3.0% a year. These returns are surprisingly good against the backdrop of Japanese financial experience in that period, which included periods of negative returns, even for bank depositors.

The second technical objection is:

The past is no guarantee to the future, so we don't trust the simulations.

I presume by that is meant by that is: ".... so we don't trust the simulations based on past experience".
Once again, I refer readers to what I've written, this time my 2021 paper to the Society of Actuaries in Ireland. Here's what I wrote there (Sections 12.15 to 12.18) about seeing how the model deals with an unpredictable future:

12.15 Two thousand Monte Carlo simulations of possible future investment experience were completed for the 60 years, based on the (deterministic) cash flows in Appendix 1 and the Wilkie model of stochastic investment returns (see Appendix 4 for a high-level description of the Wilkie model). All 2,000 simulations showed the scheme staying solvent for the full 60 years, in the narrow sense that, even in the worst scenarios, fund values were positive throughout and the smoothing formula could be applied to calculate smoothed returns at all times.
12.16. Having seen no ‘failures’ at the crudely defined level of not running out of cash, we then searched for simulations where smoothed returns were negative for the final eight years or longer. Just two out of 2,000 simulations satisfied this criterion. The likelihood is that, if the projections were extended beyond year 60, both these simulations would eventually cause the scheme to run out of cash and become insolvent.
12.17. Before considering the actions required to forestall the adverse outcomes envisaged in these two simulations, it is worth considering the plausibility or otherwise of the underlying investment trajectories:

1. In one of the two simulations, the index of market values (with dividends reinvested) falls 55% in the last 11 years. The average return on cash flows over the entire 60- year period is negative, compared with an expected return of around 5% a year according to the model.
2. In the second simulation, the index of markets value falls 75% in the last 13 years, i.e., its final value is just 25% of its level 13 years previously. The average return on cash flows over the entire 60 years is equivalent to 1.9% a year.

In both simulations, the severe adverse outcomes can be explained in large part by the absence of mean reversion in the dividend component of the Wilkie model. Market collapses of such severity, and extending over such prolonged periods, are difficult if not impossible to reconcile to any plausible real-world economic scenarios.
12.18. Seeing that only two out of 2,000 random simulations of future experience show the scheme facing the possibility (not the certainty) of insolvency at some point after it has been in existence for 60 years, and that even those two simulations are a consequence of highly implausible financial trajectories, it is tempting to conclude that the scheme will remain solvent in all reasonably foreseeable future circumstances. That would be wrong. Negative cash flows will pose a new set of challenges for the smoothing approach and will require new solutions.
The more recent paper to the Institute and Faculty of Actuaries gives more detail on the "new solutions", and shows how the scheme can be set up so that it never fails, even in the most catastrophic of the above projections of possible future experience.

I'm sorry that it hasn't been possible to give more succinct answers to both "technical" questions, but a lot of effort was required to deal with the technical issues in the paper(s). It's not possible to summarise all that effort in a few sentences.

I'll deal with the "inflexible" and other arguments in a separate post.

 

[Duke of Marmalade]

Podcast of Colm deleted and posted on the main thread covering Colm's recent OP in the IT

 

[Colm Fagan]

I don't know if anyone is reading these threads, but I'll plough on regardless (for the time being anyway).
The next "technical objection" on Brendan's list is:

It's too inflexible - people have no choice of whether to contribute or not, pause their contributions or increase their contributions.

First of all, it's important to remember that auto-enrolment is inflexible by its very nature. Contributions are a fixed percentage of earnings (up to a €80k a year), i.e. starting at 1.5% from worker and employer for the first three years, increasing to 3% for the next three years, etc. My proposal exploits the inherent inflexibility to deliver more than 50% better value. That extra value couldn't be delivered if it were an ordinary DC pension. That is a very important point, sometimes forgotten. Putting it another way: the "traditional" approach treats the auto-enrolled pension scheme as if it's a "normal" DC scheme, without getting any compensation for the inflexibility. My approach gets a heck of a lot of extra benefit for the inflexibility.

Let's see what that inflexibility entails (drawing from Section 2, pages 5 to 7 of my paper to the Institute and Faculty of Actuaries)

Contributions (by employees, employers and state) are a fixed percentage of earnings (to the upper earnings limit). A worker can always stop contributing, but that means losing the employer's and state's contribution, so it's a no-brainer. AVC's will not be allowed - they must be effected through a separate private sector DC arrangement. Workers can of course leave the scheme and join another one, retaining full entitlement to the account value secured by contributions to date and will continue to earn "interest" at the same rate as active members (i.e., there will be no charge or penalty for "paid-up" accounts) but they will not be able to transfer their accrued savings to the new scheme. The accrued benefits will be paid on eventual retirement or death on the same terms as continuing contributors (as set out below).
On retirement, members must take the 25% cash option and the other 75% as a regular income. They stay in the scheme for that 75% and can withdraw between 3% and 8% of account value every year (higher percentages over age 80). They won't be allowed to change the percentage capriciously, to exploit differences between smoothed and market values, but they will be able to change it for "demographic" or "economic" reasons, e.g., reduce the percentage on becoming entitled to a state pension (if they retired before state pension age), take a reduced percentage while working part-time, then increase it on retirement from part-time work, reduce the percentage on the death of a spouse, etc. As someone who's "retired", that seems flexible to me. I would also allow the percentage to be changed, provided that it's not to exploit differences between smoothed and market value, e.g. give a year's notification of increasing from 3% to 6% a year. The bottom line is to allow lots of flexibility, so long as someone isn't trying to play the system. Then there is the option (not compulsion) to add longevity protection as per Section 5 of the paper (page 17)
It is also important to add that the member takes their pension and gratuity whenever they retire from work, irrespective of age. There are no penalties or enhancements for "early" or "late" retirement, etc. That doesn't sound like inflexibility to me.
As I said at the start, I don't know if anyone is reading these responses, so I'm inclined to leave it at that and not waste any more time on it. It does not mean I haven't answers for the other "technical objections". Far from it.

 

[Duke of Marmalade]

They don't come much more inflexible than a DB pension and yet most people would gladly switch to a (sound) DB pension than have all the bells and whistles in the world that the DC industry can dream up and will undoubtedly throw at AE, whatever guise it takes.

 

[jasdpace@gmail.]

As I said at the start, I don't know if anyone is reading these responses, so I'm inclined to leave it at that and not waste any more time on it. It does not mean I haven't answers for the other "technical objections". Far from it.

Hi Colm,

That's a pity but I understand your position. The momentum seems to be very much reduced on AAM and, frankly, in the main current thread, many of the posts didn't add a whole lot to the debate IMHO. You mentioned better quality discussions on other forums - can you advise where I can find these discussions please?

 

[Colm Fagan]

Hi @jasdpace@gmail.
The most productive forum I'm on is a small group who meet occasionally by video call to discuss burning issues face to face. The main advocate (in my case the smoothed approach to AE) makes a brief presentation at the start, then others get a chance to chip in and ask questions, I suggested something similar a while back on AAM, but there were no takers.

It occurs to me that one possible way to overcome the problem would be to organise a webinar, at which I would try to answer the questions, and where others would have the opportunity to come back to me with disagreements on what I've said, or with follow-up comments.
I'm not sufficiently technically minded to organise something of this nature, but would be happy to participate if someone else could organise it. I imaginer that it would be possible for people to retain their anonymity, if they wished.
Any takers?

That offer still stands.