The case against the Asset Meltdown Hypothesis

The impact of evolving demographics, specifically either ageing or longevity, on financial asset prices has long been debated. From an economic theory perspective, its effect is best studied through the lens of the life-cycle theory which posits three broad types of savings behaviour exhibited through an individual’s life:

  1. Younger people tend to be borrowers, looking to invest in their future well-being. They borrow to finance education or home ownership.
  2. As they mature, people increasingly become savers, paying down their debt burdens and shifting towards accumulating wealth for retirement.
  3. Finally, when they reach the retirement stage, people start to dis-save and use their accumulated wealth to finance their consumption needs, owing to the lack of employment or business income.

Following this framework, economists have developed frameworks referred to as ‘overlapping generation models’ in order to simulate the behaviour of aggregate savings in a given economy. Unsurprisingly, the theoretical results point to a significant shift in demand for assets as a population ages. First, if the relative proportion of savers to dis-savers drops, net demand for assets falls as a consequence. This decline fuels a broad-based decline in financial asset prices. Second, ageing is associated with risk averseness; leading to a decline in the demand for equities relative to bonds as portfolio allocations shift from risky to safe assets as people near retirement.

This line of thinking has led to academic circles and the investor community debating the ‘Asset Meltdown Hypothesis’ (AMH), first expressed as a pessimistic assessment of real estate prices in the US by Gregory Mankiw and David Weil in 1989[1].

The evidence is not clear-cut

Since this seminal paper, the empirical literature on the relationship between demographics and asset prices has grown. Likewise, as the baby-boomer generation has started to reach retirement age over the past decade, so the topic has only become more relevant

Overall, dependant on specifications, samples and controlling variables, the literature tends to find a positive but weak link between demographics and financial asset prices[2]. Also, results tend to be more robust in the US than in other geographies. As such, the AMH put forward in theory, only tends to be loosely validated by empirical data.

Because we are only at the beginning of the ageing transition, it would be imprudent to entirely dismiss the hypothesis based only on current data.

Life versus dynasty

A number of factors have, however, led us to doubt that the AMH will actually prove true further down the line. The primary reason is that the life cycle theory as hypothesised, does not work well in practice.  The notion fits very well on the liability side of household balance sheets: i.e. people indeed tend to borrow in the early part of their lives and pay back debt as they get older (Exhibit 1). The fit is a lot less telling on the asset side: net wealth continues rising past retirement and, in the euro area, only begins to fall after 75 years old, which contradicts the assumption of de-accumulation. And even for the latter, it looks more likely due to cohort effects dominating age effects[3].

Exhibit 1
Life cycle theory in practice

 

Source: ECB (2017), U.S. Census Bureau (2014), AXA-IM

These stylised facts have been well flagged and two convincing explanations have emerged in the literature to explain the failure of the life cycle theory. Old people tend not to liquidate their wealth because of: i) a desire to leave money behind for their loved ones, and ii) uncertainty around life expectancy and the associated costs of late-life care. A third factor suggested by the data is that the drop in income is oftentimes compensated by a drop in expenses that reduces the need to dis-save.

None of these factors look likely to fade away even as longevity expands. The first would warrant a tax reform making it extremely unattractive to leave bequests. This is an unlikely prospect considering that inheritance tax is already very high in the jurisdictions considered (40% in the US, 45% in France, 30% in Germany for example). The second would require a large expansion of social safety nets which looks difficult at a time of high public indebtedness and rapidly rising healthcare costs.

Irrational savers?

A second important aspect of the life cycle theory is the assumption that people should not only de-accumulate wealth as they approach the later stages of their lives, but they should also broadly shift their asset allocation from risky to ‘safe’ investments. This again is not really reflected in reality. In the euro area, financial assets tend to grow with age, mimicking total wealth. But the share of people holding equities also tends to grow and both US and French data suggest that the share of equities within individuals’  financial asset portfolios also tends to grow with age. (Exhibit 2).

Exhibit 2
Financial assets and equities go up with age[4]

Source: ECB (2017), U.S. Census Bureau (2014), AXA-IM

This apparently irrational behaviour can also be explained by the bequest motive that has at its heart the desire to createa dynasty, rather than make the most of only one individual life cycle. It also reflects the heavily regulated retirement and savings sector in the US and Europe that tends to channel savings towards tax-advantageous vehicles.

It is possible that future deregulation within the savings and retirement markets, as well as the gradual adoption of defined contribution systems will lead to asset allocations converging towards what the theory says is optimal. But this is likely to be very gradual.

Longevity more than ageing

A third factor mitigating the impact of demographics on asset prices in practice is that retirement is pushed out in the future as longevity expands. As a result, looking at a static ratio of savers (35-64 years-old) to elderly (65+) gives an excessively negative picture of reality (Exhibit 3). Instead, we build a dynamic ratio where retirement is assumed to be postponed by one year every five years from 2010 to 2050, reaching 73 years-old by that date. This assumption fits the reality observed since 2010 in most OECD countries. Using this dynamic ratio leads to a more positive conclusion whereby the ratio of savers to elderly remains about stable between 2005 and 2050.

Exhibit 3
70 is the new 65

Source: UN population projections, AXA IM Research

Winner takes it all

The final factor that is working against the possibility of the AMH becoming a reality in practice is the level of economic inequality.. Leaving aside the social consequences of such inequality, the demand for financial assets is less dependent on the behavioral tendencies of the average household and more reliant on the actions of high net worth households since ownership of these assets is highly concentrated at the top. In the United States, the wealthiest quintile holds over sixty percent of the financial assets (Exhibit 4).

Exhibit 4
Inequality skews financial asset ownership

Source: U.S. Census Bureau (2014), AXA-IM

This suggests that we should focus more attention on the demographic trends and preferences of the wealthiest households. As expected, the requirement to dis-save to finance consumption in old age reduces as you move up the wealth spectrum. This is evidenced by the fact that only marginal changes to the savings rates of high income households are visible as they age. Studies also indicate that labour force participation has been rising significantly in the top income quintile for people above the age of 65. In other words, rich people tend to live longer and in better health, therefore retire later than the average population. As a result, not only does inequality heavily skew the life cycle theory but it also amplifies other mitigating factors described above.

 

[1] Mankiw, G.N. and Weil, D.N. (1989) The Baby Boom, the Baby Bust and the Housing Market. Regional Science and Urban Economics.

[2] Thenuwara, W., Siriwardana, M. and Hoang, N. (2017) Demographics and Asset Markets: A Survey of the Literature. Theoretical Economics Letters.

[3] In particular, the cohort born during war time is more likely to have gotten less education and opportunities, staying poorer as a result than following cohorts throughout their life.

[4] In addition to differences between relative allocation to financial asset between the US and Euro area, the large difference in household financial assets could perhaps be explained by the scope of assets accounted for and limitations in the availability of data categorized by age group. For example, US financial assets include household assets held in 401K or IRA accounts while the Euro area covers only “voluntary pension plans”.