« A preliminary estimate for Canadian 2011Q4 GDP growth | Main | Why I don't recognize your foreign educational credentials »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Hi Nick,

very quick response: The World Bank set out a framework for public pensions with 3 pillars. link.

Now, you can accept or reject the WB's approach, but there is a logic to it.

The WB's view in brief is that pensions try to do a number of things. The idea of the 3 pillars approach is to make sure we have as many instruments as targets.

Having OAS/GIS as the 'poverty alleviation' pillar and CPP as the 'earnings replacement' pillar makes sense to me. It also makes sense to me that they are funded differently.

Kevin: thanks for the link.

I fully understand the difference in cross-section funding. If the object is to alleviate poverty, the rich are going to pay for the poor. We have that with OAS/GIS and we don't have that with CPP. That makes sense.

But the time-series funding doesn't make sense to me. Suppose there were more poor people in odd years than in even years. Would we increase taxes/cut benefits in odd years and do the opposite in even years? That would violate tax-smoothing, as well as equity. Instead we would run deficits in odd years and surpluses in even years. And we would set up a special "snow removal" fund to handle the odd/even fluctuations and help us get the accounting right.

Nick,

I would like to add that this post ignores or assumes-away the money lost to administration costs. The government is not a perfectly efficient pass-through agency. If X is the total of all money collected for *either* CPP *or* OAS, the total money paid to benefit recipients is (X - c), where c is a positive real number.

Why is this relevant? Because it has big implications for your debt/GDP ratio. I know you don't want to talk about sustainability, but debt/GDP plays a major role in your reasoning here, so I thought I would interject that various values for "c" (i.e. various levels of "administration costs" - a polite euphemism for government waste and corruption) change the outcome of our preference between PAYGO and FF. In other words, if you can save on administration costs, you can "get away" with PAYGO (if all you're doing is comparing it to FF).

I will also add another important argument "in favor" of FF, if we must provide such excuses for government theft of resources: Rule of Law. A regime in which the rules are clearly stated and remain unchanged for all citizens is both legally "fairer" and also provides for better economic calculation (by reducing uncertainty) than a regime in which tax and benefit rates change with the seasons. As we all know, markets hate uncertainty. Even a certain loss is better than an unexpected change in future annuities.

Finally, just to be the "screaming libertarian in the room," you wrote: "Presumably there is some reason why the government is in the pension business in the first place." I think this statement grants far too much benefit of the doubt to governments. I believe that "c" is the whole reason pension plans exist. By taking some of our resources away, they create cozy, permanent jobs for Ottawa residents, where we can all put our skates on in $700,000 on-ice change rooms and pretend that the money comes from "GDP growth rates" rather than the blood, sweat, and toil of the Canadian population, who is then forced to relinquish it via the tax code.

Okay, that's it! I promise not to crowd out further discussion this time. :)

Ryan: I like your "Rule of Law" argument for FF. I see it as an application of the general "Rules vs Discretion" debate in macro. Quite apart from unfairness, PAYGO creates much greater uncertainty if people don't know in advance how the government will respond to changing demographics. "Will they cut benefits, or raise taxes??". Depends on the politics of the day.

Actually, I think we saw that effect in action, before CPP became funded. There was a lot of speculation in the 1980's 1990's that CPP would not be around in future, when it was closer to PAYGO. I don't hear that nowadays. Instead, we are all speculating about OAS. What will it be indexed to? Will it start at 65 of 67? If OAS were FF, it would be less uncertain, I think.

Nick - sometimes, however, it just is is is FF/PAYG is the answer.

OAS was started in the 1950s. If you could remember people's savings being wiped out in 1929 and never recovering, would you think to yourself "let's save for people's retirement by putting money in the stock market"? No, you'd be highly suspicious of fully funded schemes, knowing how badly they could go wrong.

Also, (and this is a variant on your intergenerational equity argument) state pensions were put in place because old people were desperately poor and unable to look after themselves in their old age. That immediate problem (desperately poor old people) was something that could never be solved with a FF scheme (unless one had a tardis and could travel back in time - and then figure out some way for people who were barely earning enough to subsist on to also accumulate sufficient wealth to fund their own retirement.)

Remember people who were 70 in the 1950s were people who were 50 in the 1930s - they could have lost everything and never had a chance to rebuild/regain their wealth.

Frances: I think there are two questions: should it be FF?; if FF, what assets should it be invested in?

If you have a debt/GDP ratio anyway, that's larger than the Fund, you can always invest in government bonds. From the secular perspective, it's just a shell-game; but from the cyclical perspective it isn't, because it means the debt/GDP ratio should be adjusted according to demographics.

(I was scared you were going to talk about burying bread in the ground, and the GE perspective!)

Kevin Milligan wrote:

Having OAS/GIS as the 'poverty alleviation' pillar and CPP as the 'earnings replacement' pillar makes sense to me. It also makes sense to me that they are funded differently.

Precisely. Poverty Amelioration means pooling and *gasp* redistribution. OAS has a presence test and low-income test (clawbacks) for benefits. CPP has a presence and earnings tests in that earnings are placed into funding. No earnings, no benefits. The two are not the same.

Further in Canada there is a constitutional wrinkle in that OAS and CPP both have specific constitutional provision to allow the Federal government to deliver them; otherwise they would be provincial jurisdictions. Of course I am banging my usual tin up that economists should pay more attention to history.

And let's ignore the fact that somebody has to pay to turn an asset into income.

I think you are correct that FF is a good principle. That would have been my default answer.

But I am having second thoughts after your comment. I think that it's reasonable to assume that there is enough variability in economic growth to generate variability in lifetime saving capacity. This mean that some generations could have less ability to save (let's imagine a big war or something like that). Then it would make sense for a goverment who took the very long view to transfer resources from one generation to the other in order to smooth the ex-ante consumption. This would argue against FF.

But such a goverment would need to have god-like forecasting powers. So it probably a bad idea as it's too easy to game.

A pension is a bet at the end of the day. You bet on the economy growing as it did in teh past and you bet that medical innovation will not accelerate. Risk can be somewhat mitigated through sharing but not eliminated. Even when fully funded those risk exists. Who should bear them? I am inclined to think that the individual should bear some of them. If medical science improves and you can still work, maybe you need to give back to society some of that gain...

How is it that life insurance companies , who work in the same game as pensions, don't seem to have these problems?
If life expectancy increases, there is more time to accumulate assets but competition will eat away those gains. If workers/beneficiaries worsens, the ability to pay would diminish as a whole. Yet, we don't hear insurances companies talking about future crisis.Do they know something we don't or are they totally blind?

In life insurance, increased longevity is to the insurer's advantage. In life annuities, it works against the company but Canadians are notoriously reticent about purchasing life annuities for themselves. They hate it and have to be dragged kicking and screaming to do it. The average age to purchase one is 80 and that is because that's when most provincial Pension Benefits Acts require it for funds discharged from a pension plan into a locked-in account.

Life Insurance and Life Annuities are a mutual hedge against each other, that's why insurers sell them. Why complain when your hedge works?

"In life insurance, increased longevity is to the insurer's advantage". Yes but , as it is updated, competition should eat away the advantage. Even though few people buy annuities on their own , some convert their RRSP into annuities instead of RRIF. So there still should be some grumbling heard from the industry.

Determinant: "Precisely. Poverty Amelioration means pooling and *gasp* redistribution."

Re-read my post, and reply to Kevin. There is an argument for redistributing from rich individuals to poor individuals. Why is there an argument for redistributing from one cohort to another cohort, based on how numerous those cohorts are? That is what FF vs PAYGO is about.

acarro: good point. If one cohort is richer than another, that is an argument for redistribution from the richer to the poorer cohort. But tax rate smoothing and benefit rate smoothing would presumably cover a lot of that.

Determinant: "Life Insurance and Life Annuities are a mutual hedge against each other, that's why insurers sell them. Why complain when your hedge works?"

Neat! Do the same companies sell both?

"Remember people who were 70 in the 1950s were people who were 50 in the 1930s - they could have lost everything and never had a chance to rebuild/regain their wealth."

That makes sense. On the other hand, the same logic would mitigate against a PAYGO system, after all, what happens to the ability of a government to fund such a program in a great depression. Maybe we should ask the Greeks. At least if a FF system, it can in theory be diversified (I.e., you're not just an unsecured creditor of the Greek government), and politically I think its probably harder to cut benefits.

Of course! Canada Life (actually Canada Life, Great-West Life and London Life are all the same company under the hood), Manulife, Standard Life and Sun Life are the players here.

By law only life insurers can sell life annuities, as BMO recently found out to its cost. Life Insurance and life annuities are both dependent on mortality risk and only life insurers are licensed to take that risk. With life annuities sicker is better for the company but a customer actually gets a better return for being sicker, this is called an Impaired Annuity. But taking up smoking on your own is pushing it.

I've been banging away steadily about life insurance, life annuities and mortality risk because life insurance companies specialize it it. I have my old Life Insurance training books to hand and there is a whole chapter devoted to life annuities, a life insurance license lets you sell both.

Jacques: I said that Life Insurance (the actual product, particularly whole life) and Life Annuities are hedges for each other. Thus no grumbling. They aren't perfectly symmetrical due to volume on one side, but deterioration in one market is generally improvement in another.

Life Insurers also engage in insurance of their own risks, an activity called reinsurance. All the big reinsurers are global and thus diversified across countries. Insurance companies have their own retention limits which you never see, they take any sized risk at retail and reinsure the excess on the reinsurance market. In this case life insurers are retail fronts, but it is all transparent and has been going on for over s century. The rules are clear, the players well-established and the business benignly boring.

Bob Smith: "At least if a FF system, .... and politically I think its probably harder to cut benefits."

That sounds right to me too. It's not obvious why. But if it's already 'earmarked" in the budget,..."flypaper effect" I think it's called. I expect that's a case where framing matters.

Nick, in a PAYGO redistribution system where benefits come from general tax revenues (OAS is paid from the Consolidated Revenue Fund, it has no special fund or levy) Workers *who pay net tax* and retirees *who pay net tax* fund the benefits of those who receive OAS.

With retirees this is even more the case due to the clawback. OAS does not take money from people without money and it also takes money from wealthy seniors. It is not US Social Security where you stop contributing once you receive benefits.

The only net beneficiaries are the elderly poor and poor workers don't pay (much) either because they don't pay net tax. Wealthy seniors certainly do pay.

Since there is no dedicated fund with an age transition from payer to beneficiary the actual inter-cohort redistribution is less than that of US Social Security.

OAS is far more a case of redistribution from wealthy to poor than transfer across generations.

Stop thinking of US Social Security. It will muddy your thinking.

Determinant: "In this case life insurers are retail fronts, but it is all transparent and has been going on for over s century. The rules are clear, the players well-established and the business benignly boring."

That's what we want!

I just meant that it wasn't exotic speculation of the kind that so recently detonated AIG. Like any market there are failures. The health and financial underwriting rules for reinsurers are clear; retail insurers incorporate them into their own standards.

you want the Government to start selling life insurance? Nick, I never knew you were such a closet Communist! Do you have a picture of Karl Marx on your night table then?

"OAS is far more a case of redistribution from wealthy to poor than transfer across generations."

Can't it be both? In all seriousness, most government spending will involve transfers from wealthy to, well, let's call them less wealthy (poor might be a stretch), by virtue of the fact that we generally raise revenue by way of progressive taxes.

On the other hand, OAS involves a transfer to a group which is (a) generally wealthier than the population as a whole (no surprise there, wealth accumulates with age) and generally less poor than the population as a whole (although in saying that
I can't forget that that's to a large extent a function of OAS/GIS. For the most part, it's a transfer to the middle class.

I think saying that life annuities are not exotic products is quite weird. They are a very complicated products. As bad as CDO if not worse... Frankly I see little difference between life annuities pools and mortgage pools.

There is a market, but that's saying nothing. There was a market for CDOs until there wasn't.

Plus aren't many insurance products back stopped by the government in many contries?
How many companies have been bankrupted by pension promises, which get dumped on the tax-payer?

Nick - my initial response was a bit snapping at things before I'd seriously thought through your arguments - sorry.

If I understand you correctly, you're not saying "if I had a tardis I'd go back and start off with a FF system." You're saying "we could reduce the debt/GDP ratio by moving from a PAYG to a (more) FF system, and this would be a good idea."

I am not aware of any time in human history when everyone in a society has been able to stop paid work at age 65 and - on the basis of their own savings - enjoy an above-poverty line standard of living. More precisely, as far as I know, at no time in history have the old have been able to live off returns to investment + *intra* generational transfers. There have always been some inter-generational transfers, or people have continued to work past 65, or both.

Moving from a PAYG system to a FF system implies not only that it is possible for people to live off their own savings when old, but that it is possible for current young workers both to fund their own retirement *and* pay for current old.

We could move the goal-posts a little by moving up the retirement age. We could cut the level of benefits received by the current generation of 65+ people. These might be good ideas, but only the first has any likelihood of happening.

But I'm still not convinced that the math works. It's policy whack-a-mole. Whack down the mole of high debt-to-GDP ratios, and up pops the mole of higher marginal tax rates on the current working-aged-population. You could say "contributions to a FF pension scheme are not taxes." True, the link between contributions paid and benefits received does ease the distortionary effects. But you're still making people save more than they want to save, and that makes earning income and thus working less attractive. The link between contributions paid in and benefits received isn't as tight as it could be - CPP benefits max out at 25 years of contributions, so there are lots of people out there paying in and getting no additional benefits for their efforts. Right now employee+employer CPP is 9.9%. How much higher than that would you be prepared to let it go?

Also, intra-generational transfers within the population of current old are difficult because the rich can move to Costa Rica.

acarraro writes:

I think saying that life annuities are not exotic products is quite weird. They are a very complicated products. As bad as CDO if not worse... Frankly I see little difference between life annuities pools and mortgage pools.

There is a market, but that's saying nothing. There was a market for CDOs until there wasn't.

Plus aren't many insurance products back stopped by the government in many contries?
How many companies have been bankrupted by pension promises, which get dumped on the tax-payer?

No. In Canada life annuities and life insurance are backstopped by Assuris, an industry-led fund designed to protect policyholders if their life insurer goes bankrupt. Your contention also argues against the evidence. Since 1945 only four life insurance companies have gone bankrupt, there have been fewer life insurance bankruptcies in Canada than their have been chartered bank bankruptcies and there have been precious few of those too. The biggest one was Confederation Life in 1993 and there was a small one just recently. Those policyholders will have their policies transferred to another, solvent company and they will suffer a 10% loss at most, some will lose not a penny, especially smaller policies. It is an excellent example of failing well.

Life Insurers engage in a business where trust is crucial and it is to their definite benefit to ensure that the entire industry's promises are honoured. That's why Assuris exists. Enlightened self-interest.

The problem with pensions, which I agree are life annuities, are not that they exist but that it was fundamentally wrong to ask widget manufacturers to be life annuity providers as well. This distinction was lost on three generations of Canadians, unfortunately.

Life Insurance and Life Annuities have existed as a business in the English-speaking world since the 1760's. Equitable Life UK was established in London in 1762 and it is still in business.

The difference is that the entire life insurance industry has robust underwriting standards for health risk and the principles of the Life Insurance Act, 1774, which state nobody may insure a risk they do not possess. Speculation in mortality risk through life insurance has been illegal for centuries.

Actually I have advocated that the principles of the the Life Insurance Act, 1774 be extended to CDO's as well as a simple and well-known remedy; the 1774 Act is only four paragraphs long. Why reinvent the wheel?

'You're saying "we could reduce the debt/GDP ratio by moving from a PAYG to a (more) FF system, and this would be a good idea."'

I'm not sure that's what Nick means. From a secular perspective, it's a wash. The "proper" way of moving from PAYGO to FF is to pay all current pensioners with newly-issued government debt, transition young workers to FF, and gradually pay off the extra debt over time. That would require a temporary _increase_ in debt/GDP, but that increase is really just accounting for existing entitlements on young workers' part.

"Presumably there is some reason why the government is in the pension business in the first place."

The same reason it runs health insurance. Insurance markets with strongly asymmetrical information typically fail.

"The government is presumably trying to help us smooth our consumption over our lifetimes, so we don't feast when young and starve when old."

No, the government is insuring you against the risk of outliving your savings.

CPP and OAS were both started as Paygo systems (CPP was moved to a partially funded model to deal with the demographic challenge faced by the boomer retirement wave). I imagine the initial decision to go with paygo was largely political, since saying "we're starting a new program and you'll get the benefits right away" has more political appeal than saying we're starting a government pension program but nobody already retired will ever benefit, only people just starting out in their careers now will get the full benefit sometime 30 years from now. Of course that's not just a political benefit, it's a large improvement in the welfare of society as a whole, so that is something to balance against your reasons for preferring a fully funded system.

Logically, it makes more sense for CPP to be fully-funded since the underlying principle of a pension is that is a reward for contributions made during one's working life. As the government collects CPP contributions, it is making a promise to future retirees that they will receive a pension in return. The OAS is just a straight transfer, and there is no connection between what was contributed and what is received so there isn't really the same logical basis for making it fully funded.

You still have the inter-generational equity/smoothing argument, but if that's your only rationale, then you're also arguing that education funding should be financed entirely be borrowing, and probably some other strange seeming results as well.

Following up on what Frances was alluding to, people in past societies were not able to retire at 65 based on what they had. Rather, in the past, people who had aged were normally taken care of by their own families in an extended family arrangement. As society has modernized, we have moved away from this extended family arrangement, to the detriment of the aged who can not fund their own retirement. From this perspective, OAS is a mechanism to replace the extended family arrangement which existed in more traditional society.

The extended family was a PAYGO system, as well as a intergeneration transfer of wealth to the aged. As OAS more or less just replaces this previous situation, having it funded as PAYGO including an intergenerational transfer of wealth seems natural.

This doesn't mean that we should keep OAS as PAYGO, but it's not clear to me that the intergenerational transfer of wealth is necessarily a bad thing -- at least in this instance.

"In a FF pension plan, the government uses the proceeds from the bonds it sells the young to invest in real assets. The government has a liability but an equal offsetting real asset. The net debt/GDP ratio is zero."

What a minute...is this funding or arbitrage?

Suppose that instead of selling bonds and buying stocks, the plan sold put options on stock indexes. The effect should be similar, and yet that doesn't seem like funding, does it?

I admit my ignorance about the Life Insurance Act, 1774. I just read something about the act on wikipedia. But it seems to me unworkable: how do they justify re-insurance? I agree that forbidding 125% mortgages is likely a good idea, but I see no solution to systematic risk...

I am sure that if you put enough regulation and barriers of entry you can make the annuity market enough of a monopoly to make sure companies fail rarely. I wonder if that is really the optimal solution though...

Frankly the idea that there is a industry body that coordinates actions of all insurance companies strikes me as pretty fishy. Surely you remember what Adam Smith said about business people gathering together...

I am pretty certain Equitable Life UK went bust not long ago. The name is the only thing that still exists...

I also thought about a different issue:
Is it possible to truly fully fund a pension system for the population as a whole?

I don't think it has ever been tried. The pension market has always applied to a small percentage of the population. I am not convinced there is enough risk capacity to handle all the various risk for the population as a whole. Or that it would be efficient to raise the cost of pension to the level where such risk would be willing taken by the market.

Maybe we should have a mechanism to make people more aware they cannot be risk-free. Maybe pension promises shouldn't be tied to fixed ages and returns and we should have a explicit system to put risk back on the individual. But doesn't that look like a pay as you go system in a certain way?

I am still inclined to think that making the effort to calculate likely cost of any investment is a good thing, but I think you should be explicit that the promises made are not written in stone...

Frances: I don't think I was clear enough. You are talking about the secular difference between FF and PAYGO, and the transition from PAYGO to FF. I think that's basically a wash, since we don't really have much of a theory of what is the "proper" long run/secular debt/GDP ratio anyway. Here's how you handle the transition: print up a load of government bonds, and give them to the pension fund. Et voila! OAS is now FF. (And the debt/GDP ratio is higher). It's just a shell game. PAYGO is observationally equivalent to FF+(a higher debt/GDP).

It's in the cyclical part that all the action is. What fluctuates when demographics fluctuate? If contributions and/or benefits fluctuate, it's PAYGO. If they don't, it's FF.

anon: yep.

Max "Suppose that instead of selling bonds and buying stocks, the plan sold put options on stock indexes. The effect should be similar, and yet that doesn't seem like funding, does it?"

Yep, in the secular part, it's a shell game. When a newborn is promised a pension, it is *as if* the government gave it a bond. A bond is just a promise, written on paper (OK, it's also transferable, but forget that in this context).

I must find a way to try to think about this and say this more clearly.

Declan: but pensions are like annuity insurance, which isn't like health insurance. Determinant says that annuity markets work OK?

"Logically, it makes more sense for CPP to be fully-funded since the underlying principle of a pension is that is a reward for contributions made during one's working life. As the government collects CPP contributions, it is making a promise to future retirees that they will receive a pension in return. The OAS is just a straight transfer, and there is no connection between what was contributed and what is received so there isn't really the same logical basis for making it fully funded."

and Kosta: "The extended family was a PAYGO system, as well as a intergeneration transfer of wealth to the aged. As OAS more or less just replaces this previous situation, having it funded as PAYGO including an intergenerational transfer of wealth seems natural."

That seems to me the sort of essentialist/framing/is is IS IS argument I wanted to reject.

Look at it in math, to eliminate the framing:

(Individual's benefits/cohort's benefits)=A(individual's contributions/cohort's contributions)

Cohort's benefits=B(cohort's contributions)

A and B are two different parameters.

When CPP was PAYGO, A=1 and B=0. Now that CPP is FF, both A=B=1.

OAS is currently A=B=0. If OAS became FF, we could keep A=0 but set B=1.

acarro: "Is it possible to truly fully fund a pension system for the population as a whole?"

Good question. You are thinking General Equilibrium. My post was Partial Equilibrium. There are two questions: are there enough safe assets so that risk can be avoided? Even if we ignore risk, are there enough assets in total? (Frances' metaphor: we can't bury bread in the ground for the boomers to eat when they are old, so what can we bury in the ground so we can pay the future young to persuade them to bake bread for the old?

Here's a better way to think about it, to eliminate the framing.

Young people pay money to the government. The government calls some of that money "contributions", and some of it "taxes". Then the government pays money to old people. The government calls some of that money "OAS" and the rest "CPP". But it doesn't matter what you call it. What matters is: the total you pay when young; the total you get when old; and (for incentive effects) the degree to which the second is linked to the first. What function should determine that linkage (that's the optimal tax question, where we trade off equality vs incentive effects)? Should that function depend on the size of your cohort relative to other cohorts (that's the FF vs PAYGO question)?

Frances: my hunch is that this PAYGO vs FF question is formally identical to our Universality question. If we could just "see" it the right way, it ought to pop right out of the Atkinson/Stiglitz(whoever) optimal tax theory solution, just like universality did.

Yep! Dammit! Assume the canonical optimal tax theory problem "scales" by population size, so you get the same answer regardless of how big the population is. Then you can use the same answer for a small country as you can for an otherwise identical country that is 10% bigger. Now, instead of countries, let the canonical model apply to cohorts. Therefore, if a unique solution exists to the canonical problem, that solution exhibits FF. QED. (Well almost!)

Assume you have 2 countries, A and B, that are identical in every way, except A is bigger than B. (If you added two B's together, you would get A).

1. You would want exactly the same tax/transfer/pension system in A and B.

2. You would not want to make transfers from A to B or from B to A.

Therefore, if cohorts differ only in size, FF is best. You treat each cohort as a country.

Nick: " You would not want to make transfers from A to B or from B to A."

Okay, now I get it, and I agree. But I'm back to it is is is is PAYG - I'm assuming (observing?) that there are currently large transfers in place from B to A, and that it is politically impossible to cut them. If we set up an OAS fund, fill it with government bonds, and then pay for OAS benefits out of that fund, B is still on the hook for A's pensions. It's a different kind of neutrality that I"m looking at - the neutrality between paying for generation A's pensions through tax revenue, and paying the interest on the bonds that pay for generation A's pensions through tax revenue. True, as you note, these aren't totally identical in that there may be stronger or weaker perceived links between benefits paid in and contributions received, and might be seen differently by bond rating agencies.

You could respond - as you did - that's only a one generational problem, and at least if we go to FF now then things will be better for generations C, D, E, etc. I'm still concerned about feasibility - i.e. any retirement system within that FF budget constraint would probably look quite different from the one we have now - though we'll probably have a very different system by the time I'm 75 or 80. Also there's the issue of desirability - if CPP/OAS had to be fully funded, would we ever choose retirement at 65 with CPP/OAS over working longer? I suspect not - that's where I'm coming from with my inevitability of raising the retirement age posts.

There is also the issue of moral hazard. Give a government (or agency, or a bunch of Bay Street types) a few billion dollars or so to manage and say "give it back to us in 30 years time with interest" - what could possibly go wrong?

"If we set up an OAS fund, fill it with government bonds, and then pay for OAS benefits out of that fund, B is still on the hook for A's pensions."

Yes, well, that's what government bonds are for. Anytime you have government bonds, future folks are on the hook for current expenditure. But AFAICT the overall budget constraints are not significantly different between FF and PAYGO, even though PAYGO may involve more redistribution among cohorts. I'm not sure that moral hazard is relevant--having a pension fund manage a bunch of government bonds is not especially difficult; arguably state pension funds shouldn't invest in equities, because the risk profile is not appropriate to a social security program. (As a separate matter, some governments may find that starting a sovereign wealth fund is more efficient than paying off debt or spending for public investments.)

Nick wrote: "That seems to me the sort of essentialist/framing/is is IS IS argument I wanted to reject."

I wasn't quite, quite, quite sure that you were categorically trying to reject the intergenerational transfer of wealth, so I left my statement a little open-ended. But now, let me expand on my line of argument.

Until recent times, societies were generally organized that children take care of their parents once those parents become aged. This is true even in societies in which survival is more challenging, such as in
Inuit society
where the elders were provided for as long as food was available.

Now modern society has in changed in many ways including people living longer (so that the elderly are more of a burden) but also in how the extended family institution has broken down such that it is much more difficult for the children to provide for their parents. This does require that the institutions we have to provide for the aged must change, but it is still unclear to me why PAYGO, or more specifically, the intergenerational transfer, must also change with these institutions.

It's not clear to me that an intergenerational transfer from young to old is necessarily a bad thing. Perhaps I am thinking about this issue from some odd moral-ethical perspective, but I think that the least the present generation can do for our forefathers (that is the generation which is now aged) is to ensure that they don't live in poverty (which is what OAS does). It is not a huge cost, and it is one that I think society can afford, especially as society is merely rewarding the previous generation for raising the current generation to its present position. Indeed, it is arguably most equitable to institutionalize this intergenerational transfer which alleviates old age poverty.

If there is some compelling economic argument to forego the intergenerational transfer, I'd be open to reassessing my position. Your transparency has merit, but it is far from compelling. Your example of a baby boom followed by a baby bust leading to very few working aged people supporting a multitude of aged is a concern. This could happen, but let me be cynical and say if it does happen that our society will follow the same pattern as the Inuit and pretty well every other traditional society and that both the young and the aged will agree that the aged will do with less.

Mind you, I was raised in a traditional Greek household, so the idea of having my parents live with me, my wife and my family until they pass is far from an anathema to me.

acarraro:

The basic point of the Life Insurance Act, 1774 is that the insuring party must have an insurable risk and cede that risk to the insurer in exchange for the premium. Reinsurance is just that, insurance of insurance. When London Life goes to Munich Re for reinsurance, it is insuring a "book" of business against known limits. It transfers that risk to Munich Re in exchange for a premium. London Life is therefore in the same shoes as I was in when I purchased a policy from them.

Both parties cannot get a risk-based gain in life insurance, only one can, the other has to have a liability to set off the "gain" from the settlement. I can't go out and take out an insurance policy on Nick's life, I have no insurable interest in him. That's what the Act is all about, ensuring that sellers of risk have legitimate reasons to sell and don't speculate.

This may be a matter of culture, life insurance as we know it was invented in England and English-speaking countries have the most experience with it, especially with life annuities.

As for the annuity market "working", the price of life annuities is a good example of the price of a life income, it is therefore very much an "information price" that tells us what it costs to provide that income given the various costs and constraints to provide that income through the established method. Many people feel this is too expensive anyway. That is a legitimate criticism and that's why we have OAS which has a different funding model than annuities do. Different model, different price. Isn't that the point of undergrad micro?

"Perhaps I am thinking about this issue from some odd moral-ethical perspective, but I think that the least the present generation can do for our forefathers (that is the generation which is now aged) is to ensure that they don't live in poverty (which is what OAS does)."

Well, that's what GIS does, which is why it's means-tested. OAS, though, is universal, albeit subject to clawback past a certain threshold (in practice, only a small percentage of the retiree population is subject to any clawback, and an even small percentage receives nothing). Fighting poverty amongst the elderly wouldn't explain why we pay $6000+ a year to seniors with an income of $69K (i.e., an income well above the average Canadian income). If the rationale for an intergenerational transfer of wealth is to protect the elderly from living in poverty, that's fine, but that wouldn't justify the existing structure of the OAS system.

Moreover, coming back to Nick's point, you could address poverty among the elderly without an inter-generational transfer of wealth. For example, if OAS was a fully-funded, universal, program, it would address poverty through an INTRA-generational transfer of wealth (i.e., yesterday's young pay both for their current retirement and those of the poorer members of their cohort). That would avoid some of the nastier demographic effects.

Hi Bob,

Thanks for the extra information on GIS. But how about considering GIS as part of OAS? This way the expanded OAS system becomes much more means-tested as a combined entity. You raise a good point about the benefits a senior making $69,000 a year receives, but this point seems to be concerned with the method of means-testing and clawbacks as well as determining what "poverty" means for the aged, rather than whether OAS should exist and be paid for through an intergenerational transfer.

Coming back to Nick's point, what I am arguing is that there doesn't seem to be a compelling reason to move OAS(or the expanded OAS+GIS) system off of PAYGO. Can there be "nastier" demographic effects? Sure, but even with the relatively "nasty" demongraphic effect that will arise as the boomers retire, the Canadian OAS system is fundamentally solvent (at least that's what I gathered from the reports over the past month). There doesn't appear to be any driving reason to change the status-quo (although I would not object to modifying/tightening the means-testing of the system).

As others alluded to in the comments, the PAYGO system effectively creates a future liability which is paid by the next generation. This liability is not particularly large, and is arguably a reasonable transfer to the older generation from the younger.

Or perhaps I am being just too comfortable in the status quo.


There's another consideration behind switching from PAYGO to FF for OAS. The PAYGO system under the Samuelson assumption real growth > nominal growth is sustainable with the first cohort entering OAS getting their benefits for free. Every other cohort, however, pays, either for themselves or for the generation before them. But again under the Samuelson assumptions of real growth > nominal growth and population growth > 0, the cost of funding OAS across the society is reasonable.

In my earlier comments I hadn't quite thought through this situation -- that is I had argued that it seemed reasonable that the younger generation pay for the older generation's OAS because the older generation raised the younger generation. Arguably this is true, but there is a more important consideration. While the older generation may not have funded their own OAS, they did fund the OAS of the generation before them. This was done with the expectation that their own OAS would be funded by others.

Let's say, as a society, we decide to switch to a FF OAS from PAYGO. How do we implement this change? Do we immediately cut off PAYGO funding for OAS for today's aged (diverting that funding to fully fund the OAS of "tomorrow's" aged)? Or do we continue to fund today's OAS out of today's working generation, and then collect yet more fees/taxes from today's working generation so that they pay for their own OAS once they become aged?

Implementation of the OAS as PAYGO allowed the first cohort in the system to receive benefits for free, but every other cohort pays for benefits (either their own, or the previous generation's). If the system switches to FF, how is the pay schedule accelerated? Will this not result in a cohort paying twice, that paying for the OAS of both the generation before and their own generation?

"Let's say, as a society, we decide to switch to a FF OAS from PAYGO. How do we implement this change?"

This was addressed in a previous post: we pay today's aged with newly issued government bonds, and future workers buy tgese bonds. The bonds may be rolled over indefinitely (which maintains the "free lunch" aspect) or paid off gradually, depending on economic growth and interest rates. In case of a short-run generational imbalance, the interest paid on government bonds may move slightly as a result of supply and demand; but speculation, portfolio choice and the like will act as a stabilizing force, and it's also easier to cushion fiscal effects by adjusting the government budget. In the long run, the only difference is in accounting.

And to follow-up on Anon's point, I think there's probably a governance advantage to borrowing to fully-fund OAS rather than proceeding on a PAYGO basis in that you put the future liability for OAS on the balance sheet rather than having a very real, but off-balance sheet, liability for future pensioners. At the end of the day, we'll have a future obligation to pay somebody (whether it's the bondholders who lent us money to shift to a FF regime or pensioners in a PAYGO regime), but by borrowing to shift to a FF regime, you make that liability explicit and force governments, politicians and voters, to deal with it or at least require that they take it into account in future planning, whereas in a PAYGO regime, everyone knows the liability is there, but out of sight, out of mind.

@anon & Bob

Thank you for the additional information. Bringing the OAS liability onto the balance sheet is probably a good idea. We can quibble about whether doing so is converting the system from PAYGO to FF. As Bob notes, the liability is still there, it's just a question of who is paying who. A true FF system would mean that the money has already been collected to pay for the future liability.

This proposal still does not address how the shift from PAYGO to FF is paid for. If we are to switch to PAYGO today, then this generation of workers would have to fund their own future OAS as well as the previous generation's present-day OAS. Anon alludes to a "short-run generational imbalance," how large is this imbalance? If it's only a minor imbalance easily handled by adjustments in interest rates, the it begs the question as to whether we should make the adjustment at all (that is, why even move from PAYGO to FF?) But if the PAYGO future liability is significant, aren't we fooling ourselves by trying to claim that we can switch to FF with only a 'minor' short-run generational imbalance?

But bringing the future liability onto the balance sheet is a good idea (as is the ability to gradually shift the liability between the present and future generations).

we are to switch to PAYGO today, then this generation of workers would have to fund their own future OAS as well as the previous generation's present-day OAS.

Not neccesarily. If you borrow to fully fund OAS, then the previous generation's OAS would be financed by bond-holders, and the repayment of the bonds could either be (a) spread-out over all future generations, or (b) if nominal GDP growth > interest rates, rolled over foreover. There's no reason why the current generation would have to take the full hit. I think Anon's right that the real short-term cost might be a hit to borrowing costs, but even then, all you're doing is shifting an off-balance sheet liability onto the balance sheet, it's not clear that that should affect the government's ability to borrow. If anything, the bond markets might even prefer the transparency provided by moving to a fully-funded regime because it reduces the uncertainty about the governments future, off balance sheet, liabilities (I suppose that cuts both ways, though, if the bond market is currency under-estimating the size of those liabilities, crystalizing them could drive up borrowing costs).

Good discussion. I'm reading but seeing no need to respond. Just remember my main point Kosta: demographic fluctuations. Not all cohorts are the same size. A bigger cohort will have bigger need for benefits, but also a bigger capacity to make contributions to fund those benefits. Which is what you would get anyway in a FF system, but not in PAYGO.

There are good arguments for changing how we account for OAS. Bringing the off-balance future liability onto the books is a worthwhile endevour. Bringing the liability onto the balance sheet would also facilitate more effective budgeting of the future expenses. By this I am alluding to Nick's last comment. If a large cohort is approaching retirement (large could mean that the cohort is larger than the cohort following it), then it probably would be prudent to start pre-funding that cohort's OAS. These are all good points.

Although I've been arguing for the maintenance of PAYGO for OAS, what I am really arguing for is the continuance of the intergenerational transfer from today's working to today's aged. The argument is based on tradition in that societies always took care of their aged, usually through an extended family arrangement, but with the breakdown of the extended family, a PAYGO OAS system seems like a natural replacement. Or more specifically, an intergenerational transfer of wealth to pay for OAS seems like a natural replacement.

I would like to argue that it also makes good economic sense for the cohort which is working to defer the funding of its retirement to the next generation so as to free up resources to either expand production in the present, or to raise the future generation. I realize that this logic is prisoner to the Samuelson's assumption that the rate of growth will exceed the real interest rate.

Another argument which supports the maintenance of the status quo is the cost of switching over to FF from PAYGO. In Bob's last comment, Bob writes that the bonds issued to pay for OAS switch would be paid off by subsequent generations. But as I read this, is this a switch from PAYGO to FF, or is this a switch from PAYGO to Pay-Sometime-Down-The-Road? It is definitely not fully funded.

I think we're arguing to separate points here. On one hand, there's how do we account for the future liability (that is off-balance sheet PAYGO versus bring it on the balance sheet and explicitly paying for the liability through some combination of pre-funding and issuing debt). The other issue is how do actually pay for the future liability? Does each generation pre-fund its retirement OAS+GIS? Or does it depend on the next generation to ensure that its "basic" living standards are met (that is an intergenerational transfer of wealth)?

For the first question, I agree with the arguments that accounting changes would be beneficial. But the second questions, I think an intergenerational transfer of wealth in this circumstance is fine.

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad