Pensions and long-term retirement saving: a macroeconomic perspective

Speaker: Andrew Bailey, Chief Executive
Location: Gleneagles
Delivered on: 16 September 2016

Key points:

• Looking at the issues on the FCA agenda, pensions and long-term retirement savings are probably top of the list in terms of their importance to our society.
• The lifetime model is a core building block of macroeconomics, and it remains an essential lens through which to assess today’s challenges around retirement income and pensions.
• If there are two big investments in the life cycle model – a home and a pension – the cost of both in real terms has risen.
• There is an argument that pension saving would be assisted by people holding more housing in their stock of pension assets, based on the real appreciation in the value of housing. I don’t subscribe to this argument.

Note: this is the text of the speech as delivered.

It is a great pleasure to be in Gleneagles today and to have the opportunity to spend the day reflecting on pensions and long-term retirement savings, which is such an important issue across many countries in the world. When I look at the issues on the agenda of the FCA with its objective from Parliament to ensure that relevant markets function well, I think that pensions and long-term retirement savings are probably top of the list in terms of their importance to our society. But, of course, only part of what is a very broad issue falls under our responsibility because there are many other aspects of economic and social policy in which the issue falls, and that demands joined up thinking and approaches.

When I look at the issues on the agenda of the FCA...I think that pensions and long-term retirement savings are probably top of the list in terms of their importance to our society.

There is in my view an important macroeconomic dimension to the pensions issue which is growing in its significance.

In addition to my role as the Chief Executive of the Financial Conduct Authority, I am also a member of the Financial Policy Committee (FPC) of the Bank of England which has the macro-prudential remit in respect of the stability of the financial system. I want to talk today about the macroeconomic aspect of pensions, something that I think requires more attention, as well as how that macro picture reflects into the challenges for the FCA. I should add that these views are my own and not those of the FPC.

Let me start with an illustration of the issue by borrowing something written by John Kay last weekend: ‘To provide yourself with 70% of your gross income for 25 years of retirement when real interest rates are zero requires setting aside 45% of that gross income every year. Should you save more to try and make up the shortfall – or, since the goal of comfortable retirement is beyond reach anyway, should you save less?’

Let’s just unpick that statement for a moment, to draw out the significance. First, it well illustrates the key assumptions in long-term planning for retirement income: the long-run course of real interest rates, and thus the rate on which to discount the return on the assets in which savings are held; the assumed duration of retirement; and the target for income in retirement. Those are very big judgements around which, as I will come to later, there is substantial uncertainty. The second big point I would draw out from the comment is the importance of this issue for what we assume about saving in the economy, and particularly the lifetime saving model. I will come to that later.

Before that, let me add another well-known but important illustration of the importance of changing macroeconomic conditions. Annual GDP growth in the advanced economies of the world is currently around 1.6% percentage points lower than in the decade prior to the global financial crisis. Put like that, 1.6% percentage points may not sound much, but that is an average annual growth of 1.2% rather than 2.8% (so, the more headline grabbing statistic is that growth in the advanced economies is 57% lower on average). There has been a major response from monetary policy, so that real interest rates and thus debt servicing costs have fallen too. To be clear, this has been a major benefit, and what I am saying today is in no sense a criticism of those actions, of which I am a strong supporter. They have supported economic activity, and have diminished the past pattern of loan defaults, failed companies on a much larger scale and people losing their homes on a much larger scale too. But, we now live in a world where we can expect debt ratios to remain high over the medium term, which is more manageable at current real interest rates but would be much more problematic in a world whenever nominal and real rates rise. In countries like the UK, this is much more of an issue for households than companies based on current debt ratios. It explains why on the FPC we have had a focus on the effects of high household indebtedness and particularly the mortgage market. The reason for this digression from pensions and retirement is not just to further illustrate the impact of very low real interest rates on long term savings but also to point out the second big impact on the lifetime saving model, namely from changes in the pattern and scale of borrowing associated with the cost of home ownership. I should also note that there are other significant factors which are affecting the ability of individuals to save for retirement.

The FPC has pointed out that an important underlying cause of these changes comes from the balance of demand for housing versus the supply of additions to the stock – i.e. new homes. There is, however a further link back to retirement saving here, namely that there are two parts to determining the overall demand for houses: the demand for houses to live in; and the demand for them to own.  The drivers of these two elements are different, and the second one is in part determined by the demand to hold housing as an asset for retirement saving. But the first one also has an important link to retirement saving because another key assumption which John Kay did not mention is whether or not people assume that they will vary their investment in housing as part of their retirement provision strategy. I will come to that later. 

Let me now turn to the lifetime saving model, or the life-cycle hypothesis. This is one of the workhorses of macroeconomics, and dates back to the early 1950s and the work of Franco Modigliani and Richard Brumberg. The core of it is that people make choices about how they spend at each point in their lives limited only by resources available over their lives and thus independent of income at each point in their lives. Decisions on spending at each point in people’s lives are thus linked to building up and running down assets to provide for retirement, among other things. An important observation on the model is that like many macroeconomic workhorses, it is founded on an apparently empirical observation, here assumed lifetime available resources, which is as a matter of fact unobservable. What we do know, however, as I observed earlier, is that available resources will change over the course of a lifetime, as will the level of uncertainty over what they will be at the point of planned retirement.

I want to add a few more pieces of the picture of the lifetime model. First, it asserts that changes in lifetime resources lead to proportionate changes in spending in all periods of life, so the link to macroeconomic policy is very clear. Second, changes in the volatility of income are likely to affect views on lifetime resources relative to assumed need. More volatile income will tend to restrain spending. Third, greater uncertainty generates a demand for greater precautionary saving. By the way, this is not a point original to Modigliani and Brumberg, as it was previously identified by Keynes. It is also worth noting here that saving is the sum of changes in assets owned by individuals and changes in their indebtedness.

The lifetime model is a core building block of macroeconomics, and it remains an essential lens through which to assess today’s challenges around retirement income and pensions. It is much more than a retirement model because it encompasses decisions taken by households on all other forms of investment, most notably as I said earlier, housing.

The lifetime model is a core building block of macroeconomics, and it remains an essential lens through which to assess today’s challenges around retirement income and pensions.

The lifetime model has been challenged over the years, but I would assess that those challenges have the effect of enriching the model rather than disqualifying it. For example, the timescales over which these calculations are happening are clearly shifting. And the traditional bright line between work and retirement is blurring, but the fundamentals of the model have held up well.

A challenge on which I want to comment briefly comes from behavioural economics. This is particularly relevant because, if I may put in an advert for the FCA here, there has been a greater take-up and development of behavioural economics at the micro-economic level of bodies such as conduct regulators, and my sense is that while behavioural macroeconomics has been around for some time, there is more to do to build the insights into macroeconomic policy. The essential insight of behavioural economics for the lifetime model is to put more emphasis on individuals’ perception and understanding of their current position rather than assume they are at all times fully discounting and recalculating their whole lifetime assumptions. To borrow the phrase of St Augustine – which is translated often as ‘make me chaste, but not yet’ - individuals heavily discount tomorrow. Behavioural economists have used the term ‘hyperbolic discounting’ to describe time inconsistent preferences in which, given similar rewards, individuals will save less today than the lifetime model would predict because they overestimate their ability and willingness to save tomorrow and at all points in the future. To borrow another phrase, does tomorrow ever come then? The answer from the models is yes-ish, but considerably later in life.

The weight given to such variances in the discount rate over time matters substantially when judging the value of additional long-term saving for retirement and thus the design of savings products. It matters more in an environment – as we are seeing in the pensions world – where the responsibility for such saving is increasingly transferred from the state and employers to individuals in society. Social security systems and defined benefit pension schemes tend to reinforce the life-cycle model of saving by prescribing saving, and doing so – through its compulsion – in a way that neutralises quite a bit of the variable discount rate argument. I offer no judgement on whether this is a good thing or a bad thing, but the shift is certainly happening, and therefore the relevance of the behavioural argument in terms of discount rate assumptions is if anything likely to grow. By the way, the so-called Lamborghini preference fits broadly under this heading of preferring today’s benefit over the longer-term value of retirement saving.  All of that said, I don’t think that the insights of behavioural economics have undermined the broad relevance of the lifetime model once we recognise that the saving behaviour of individuals can vary over time for reasons other than those the lifetime model predicts.

I want to now to come back to the issue of housing and how it fits in to the picture. The lifetime saving model is a model of every decision, not just those in retirement. Investing in a dwelling is another major lifetime decision, and it interacts – increasingly I would say – with retirement saving. Individuals borrow earlier in life to smooth the cost of investing to own a home. In some parts of Britain, the real cost of housing has risen quite markedly, which therefore increases the cost of investment. It also tends to extend the duration of borrowing to smooth that investment, and increasingly for some people that duration extends into what at least used to be regarded as retirement years. If there are two big investments in the life cycle model – a home and a pension – the cost of both in real terms has risen. While this does not in any sense invalidate the life time model, it will change behaviour within it.

If there are two big investments in the life cycle model – a home and a pension – the cost of both in real terms has risen.

Let me make two observations here. First, there is an argument that pension saving would be assisted by people holding more housing in their stock of pension assets, based on the real appreciation in the value of housing. I don’t subscribe to this argument. Why? First, because given the scale of uncertainty over long-run real returns on assets, I would not favour over-weighing to any one asset class, while recognising that a balanced investment portfolio can be exposed to property.  But, increasing the weight on housing investments could be self-defeating. In the FPC we have been concerned about increasing levels of household indebtedness. If the effect of increasing the demand for housing as an asset to own is to push up the cost of ownership, an increase in holdings of housing as pension assets will tend to increase the real cost, and thus household indebtedness.

My second observation is that an alternative approach, again best viewed within the lifetime model, is that rather than save for housing and retirement income at the same time, people would use the former to fund the latter. The distinction here with my earlier point on the weighting of property investment in a portfolio is that here the focus is on how much they invest in their own dwelling over their lifetime. They could vary this investment broadly in one of two ways: to downsize at retirement and thus release the investment; or to purchase an equity release mortgage which enables decumulation without moving. The fact that people look to borrow against home equity is not a surprise. Some argue that the costs of equity release, both up front and compounded over time, are relatively high for the individual which signals a note for caution. Others, however, can point to potential benefits, especially for those who want to remain in homes they’ve worked and paid for over their lifetime. My note of caution here would be that while the approach has an appeal in terms of the lifetime investment pattern, the accompanying financial instrument is made much more complicated by the need – again consistent with the lifetime model – to embed in it a no negative equity guarantee. This can have both prudential and conduct consequences, and the PRA has issued a discussion paper on the former.

There is an argument that pension saving would be assisted by people holding more housing in their stock of pension assets, based on the real appreciation in the value of housing.  I don’t subscribe to this argument.

To conclude, retirement saving and pensions is one of the largest issues we face. It needs to be considered broadly. There are some very big issues at stake here: the balance of who takes the risk, between the state, employers and individuals, with the balance shifting to individuals; the potential for large inter-generational shifts in income and wealth; the impact of heightened macroeconomic uncertainty on the ability to write long-term financial contracts which embed assumptions on future returns (I would add here as a financial conduct regulator that there is a contradiction between the scale of uncertainty embedded in macroeconomic views of the future and the apparent relative certainty embedded in the design of financial instruments); and finally the big issue of the appropriate balance of public policy between positive descriptions of the issue – retirement savings and pension provision – and more normative prescription from public authorities to individuals. These are the big issues.

I will end with a plug for the FCA’s work in this field. We published yesterday an update on our work on the ageing population, setting out the key areas on which we will focus over the next nine months. I encourage you all to read it.

Thank you.