Demographics at a tipping point
A permanent change in the age structure of a population — its demography — can alter patterns of consumption, saving and investment. It can also affect the size of the workforce and its skillset, the rate of productivity growth, and the way in which income is distributed between labour and the owners of land and capital.
All these transitions can impact inflation, but in ways that can push in opposite directions. Despite economists’ best efforts, the implications are not always clear, either in theory or upon rigorous empirical study.
The popular consensus seems to have coalesced around the view that an ageing demographic profile is deflationary. The Japanese experience is often the first evidence cited. From the mid-1990s, Japan was the first wealthy nation to suffer a shrinking workforce. Since then its rate of inflation — and the expected rate of inflation in particular — has struggled to stay in positive territory. But, as we posit in the boxed text on page 13, there are many moving parts and we are not so sure that its lesson is relevant to the rest of the world. Moreover, there is a significant body of research setting out well-reasoned and well-evidenced arguments for the inflationary consequences of ageing.
We believe that ageing creates both inflationary and deflationary forces. Which way the balance tips may be determined by which age cohorts are growing or shrinking most rapidly. In this regard, ageing has just entered a new era. Our reading of the literature, confirmed by some modelling exercises of our own, suggests the next phase of ageing is likely to start to exert net upward pressure on inflation, but not to any alarming extent.
Saving less and spending more
A recent project called the National Transfer Accounts has confirmed what some economists have always suspected: that consumption does not fall with age. Indeed, as figure 4 shows, spending continues to increase through the average lifetime, while incomes fall (unfortunately the project is yet to publish data for the UK). The composition of an individual’s spending tends to change with age — the very old spend very large sums on healthcare, either privately or via government programmes — but the fact remains that the total amount of goods and services demanded does not decrease.
Intuitively, an inflationary impulse occurs when the proportion of the population willing and able to work starts to shrink relative to those who are no longer ‘productive’ — in other words, when more and more non-workers compete for the goods and services produced by relatively fewer workers. The UK has only just reached that stage today.
Since the 1960s, the number of non-workers — the old and the young — has shrunk relative to the number of workers. We call this the ‘dependency ratio’ because these two groups depend on the goods and services produced by those of working age. The number of producers has grown more quickly than the number of consumers. In other words, while the world has been ageing, this crucial demographic pattern has sent a profoundly deflationary impulse. But that impulse is reversing — and in some countries quite sharply. The dependency ratio troughed in the UK around 10 years ago (according to United Nations data) — from now on the proportion of pensioners is set to grow more quickly than those of working age (figure 5).
Figure 4: Spending patterns
Income and consumption by age, expressed as a percentage of average income for 30- to 49-year-olds.
Source: National Transfer Accounts and Rathbones.
A global dependency
For sure, the UK’s dependency ratio is not set to rise quite so steeply as other global regions.1 The absolute number of 15- to 64-year-olds is not set to peak until around 2072, while it has likely already peaked in Western Europe and China. That said, our economic modelling work (see appendix) confirms that the global dependency ratio better explains changes in UK inflation than the UK’s.
Globalisation has integrated global workforces: when previously isolated countries enter global supply chains and start to consume (now globalised) goods and services, the effect on richer countries is analogous to a boom in the working-age population (Goodhart & Pradhan, 2017). The decline of advanced economy dependency ratios was amplified by the integration of China and Eastern Europe into the global workforce from the early 1990s. China started that decade with just a 2% share of global manufacturing; by 2016 it had taken a 19% share. However, many of the most globally integrated emerging economies are ageing more rapidly than the UK or the US. Even with the favourable demographics of African and southern Asian economies, the global dependency ratio reached its nadir in 2012.
Now, fewer workers catering to a relatively greater demand for their wares could improve their ability to bargain for a pay rise. In theory, that will push up wages beyond what could be justified previously by their productivity, which is inflationary. The effect is magnified as older people’s spending habits shift towards services that use labour more intensively, such as healthcare and holidays. (Remember the idea of too much money chasing too few goods.)
Labour scarcity would also shift income away from capital (profits) and towards workers. The owners of capital tend to be wealthy, and the wealthy tend to save a large amount of their income. So a process that redistributes income away from the wealthy also reduces savings and increases consumption, which is likely to be inflationary too.2
In his book, The Age of Aging, the economist George Magnus points out with arresting lucidity how ill-prepared governments are for the expense of serving an aged population. And an ageing population can become a very expensive one: in California, the elderly receive three times more government spending than the young. If we assume that social safety nets and welfare states are kept in place,3 a considerable amount of income will need to transfer from workers to the old. Taxes will rise. If the transfer is progressive, funded by the wealthiest members of society, it may also be inflationary, providing the government spends the proceeds domestically. Higher taxes will reduce wealthy households’ disposable incomes and they may respond by reducing savings more than consumption.
This issue raises the question of whether government debt matters for inflation. The data suggests it has not had a significant impact since World War II (Giannitsarou & Scott, 2006). Theoretically, government debt doesn’t matter providing monetary policy remains unconstrained. Imagine an economy closed off from the rest of the world. When the government issues debt the private sector has to buy it. The government then spends (or distributes) the money it has borrowed back into the private sector. But because this transaction has changed the balance between desired savings and desired investment, interest rates will rise and that creates a new non-inflationary equilibrium. The situation becomes more complicated when we introduce open, interconnected economies and banking systems, of course. But the basic intuition should still hold — that interest rates (together with exchange rates) balance things out.
The middle-age savings spread Ageing in the UK has thus far been deflationary because dependency ratios have been falling. Middle-aged cohorts have increased particularly quickly, which has increased savings relative to consumption. This group saves the most as retirement looms (Magnus, 2009). As more people move out of middle age, savings decrease and consumption increases. An important academic study found that people’s consumption habits start to ‘turn’ deflationary in their early 30s (Juselius & Takats, 2015), which is when rates of saving start to shoot up (figure 6).
In the UK, the proportion of 30- to 64-year-olds in the population peaked in 2005, although globally it is not set to peak until 2025. Over the next decade then, this trend is likely to exert downward pressure on prices of tradeable goods and services but place upward pressure on the price of services in the UK and other advanced economies that are wholly domestically supplied.
If life expectancy continues to increase, middle-aged cohorts are likely to save more (cf. Bloom et al, 2009 for the strong link between life expectancy and saving). However, a recent report by the Institute of Health Equity at University College London found that life expectancy has slowed or even stopped rising altogether in the UK (Independent, 2017).
Figure 5: Demographic patterns
The dependency ratio for 15- to 64-year-olds suggests the proportion of pensioners is set to grow more quickly than those of working age.
Source: United Nations and Rathbones.
Ageing and efficiency
Inflationary pressures could be amplified if a shrinking workforce places upward pressure on wages, while ageing slows the rate at which companies improve their productivity — how many goods and services a given level of workers can provide. A study by ZEW, a leading economic research institute in Germany, found that industries with a higher proportion of older workers are less likely to adopt new technologies. You may remember that the NHS’s failed multi-billion pound patient record-keeping system was blamed on the inability of less-IT savvy older workers to input data (although we suspect a lack of incentives and poor implementation may also be to blame).
The economist James Feyrer has shown that the median age of those responsible for innovation in the US has been remarkably stable at around 48, whereas the median age of managers who adopt their ideas is lower at around 40. Most patent applications are also filed by the under 60s (Aksoy et al, 2016). Other studies have found evidence of a decline in labour quality in the over 50s when speed and problem solving were required, but not when verbal ability or experience matters (Skirbekk, 2003).
Taken together, these studies are worrying: ageing causes an aggregate decline in the agility of human capital, but also prohibits the adoption of new technologies that could make up for the loss. However, as ageing societies naturally demand more services that require people skills and experience, this effect may not be so significant. It also seems that the number of people in their 40s is the key for innovation and change, so we should breathe a sigh of relief that this cohort is not set to peak in the UK until 2033 and globally not until 2096.
In contrast, the number of 40-year-olds in Western Europe peaked in 2009. If this link between demography and productivity growth holds then a fall in relative productivity is likely to put downward pressure on the euro relative to the exchange rates of economies ageing less rapidly, and that could have an inflationary impact in the eurozone through rising import prices.
Two Japanese economists have found another way in which ageing can lower aggregate human capital. Older workers have a significant amount of specialised, even firm-specific, knowledge that they find difficult to transfer to new roles or new companies. When good, incumbent workers are left unemployed as a result of the natural processes of competition and creative destruction or a downturn in the business cycle, the older among them undergo a permanent loss of human capital and the economy a permanent loss of productivity (Fujita & Fujiwara, 2014).
Which way will the wind blow?
There are counterarguments to what we have just set out. The main one was first proposed in the 1930s by an economist called Alvin Hansen. As populations age, population growth also slows. Hansen said that economies only need a given stock of capital (buildings, equipment, infrastructure and land) per worker. If population growth slows, demand also falls for new houses, new offices, new machines and so on. At its most extreme, Hansen hypothesised a ‘secular stagnation’ — an enduring decline in investment spending and a chronic oversupply of savings, which pushes an economy into a semi-permanent slump and unleashes fiercely deflationary pressures.
We are not so sure that ageing will cause investment spending to slow quite so dramatically from here. More importantly, we are not so sure it will cause future investment spending to fall by more than future savings. Consider residential investment. Inertia is well documented and older people tend to keep their large family homes, necessitating more houses for a given population as that population ages. Even if older people did vacate their homes, younger generations can’t afford them today anyway so more affordable housing needs to be built. From a global perspective, as emerging markets continue to get richer, existing housing stock requires replacing as high incomes increase the demand for more floor space and second homes. This puts upward pressure on the cost of building materials (Goodhart & Erfurth, 2014).
If a scarcity of labour places upward pressure on wages, firms have an incentive to invest in labour-augmenting technology in order to mitigate some of that extra cost. Business investment has been lower than expected in most countries over the past 20 years, but this was also driven by the relative availability of labour and its relative cheapness as globalisation brought new low-cost workers into the workforce. As we have seen, the relative availability of labour has come to an end, although relative cheapness could persist if — and it may be a big if — governance, infrastructure and education can improve in Africa and the Indian subcontinent. Lastly, the impetus to limit global warming should also mandate large investment spending in clean energy production and distribution, transportation and farming.
An oft cited study by three International Monetary Fund (IMF) economists confirms that unanticipated ageing causes inflation to deviate below the prevailing trend, but it does not analyse the impact of ageing on the trend itself, which is what this report is really interested in (Yoon et al, 2014).
Figure 6: Age cohort effects on inflation
Since the late 1970s, the growing proportion of 35- to 64-year-olds has pushed down the rate of inflation in most advanced economies.
Source: Juselius & Takats and Rathbones.
Figure 7: Projected cumulative effect of ageing on UK inflation
We estimate that the aging process over the next two decades will increase the annual rate of inflation in the UK by between 0.15% and 0.23% by 2037.
Adding it all up
Our conclusion that ageing is on the cusp of becoming inflationary is supported by an increasing number of robust academic studies (cf. Goodhart & Pradhan, 2017; Aksoy et al, 2016). One of the most rigorous concluded that demographic change has caused on average a five percentage point fall in the rate of inflation across 22 advanced economies from the late 1970s, a process that was largely completed by the early 2000s (Juselius & Takats, 2015). This was attributed to the growing proportion of 35- to 64-year-olds (figure 6). The proportion of that cohort has peaked; from the age of 65, the average person carries inflationary pressures.
As a base case, we estimate that the ageing process over the next two decades will increase the annual rate of inflation in the UK by between 0.15% and 0.23% by 2037, all other things being equal (see figure 7 and appendix). We see three downside risks to this view.
First, younger workers are saving less than their predecessors (Magnus, 2009) and may not be saving enough for retirement. If this continues all the way to pensionable age, consumption spending is likely to start falling with age, potentially quite dramatically.
Second, robots. In our essay on the economic impact of automation, If the machines aren’t coming for your jobs, are they coming for your investment returns?, we argued against the ‘mass unemployment’ thesis because of the likelihood that artificial intelligence is likely to be more of a human’s complement than a substitute (‘co-bots’ rather than robots) for the foreseeable future. But if ageing limits our ability to work alongside machines, developers may redouble their efforts to produce a rival to human labour with general (non-task specific) intelligence. Initially, the increase in investment spending would be inflationary, but if they were successful, the wholesale displacement of workers would have profoundly deflationary consequences.
The third, politics. This represents perhaps the greatest downside risk to our view that ageing populations will turn inflationary. (See box ‘Will the old vote their way to deflation?’)
Ageing past the tipping point
Over the past five decades, ageing has had a profoundly deflationary effect. But what may be the crucial factor, the dependency ratio, acting on patterns of saving, investment and productivity, has reached a tipping point. From now on, a rising proportion of older people and a declining proportion of those of working age is likely to create a net inflationary impulse.
Figure 8: UK constituencies by size and demographic
The elderly tend to prefer things to stay the same and analysis of voter behaviours suggests this behavious can have a deflationary effect on an economy.
Source: Datastream and Rathbones.
Will the old vote their way to deflation?
“Demographics is not destiny”, to paraphrase George Magnus. Lawmakers can change retirement ages or incentivise and facilitate greater participation in the workforce from those of working age. That said, it is unlikely — and in some cases impossible — to legislate ageing workforces away. The IMF has estimated that to offset the impact of a declining proportion of 20- to 64-year-olds in the population to 2050, participation rates would have to rise by an average 10 percentage points across advanced economies (20 in Spain, Italy and South Korea).
In other words, if around six people in every 10 in this age group work today, seven in 10 would need to work by 2050 to offset their shrinking numbers. To put that in context, the participation rate only increased by six percentage points between 1960 and 2000, when economic and social conditions were arguably much more favourable (Magnus, 2009).
Mathematically at least, raising the retirement age could provide a more feasible offset. It would need to rise by around seven years across advanced economies to offset the decline in the proportion of 20- to 64-year-olds (and by only three years in the UK). But average retirement ages have been falling over the past two decades, and attempts to raise them in a number of European countries since the financial crisis have shown just how unpopular the policy is with voters. James Bullard, President of the Federal Reserve Bank of St Louis, believes older people vote for policies that engineer low inflation, benefiting those with savings but no wage income (Bullard et al, 2012). Although the central bank committees responsible for setting monetary policy are independent, they are appointed by elected politicians. Voter preference can more directly influence fiscal policies and we wonder whether this is why austerity — an indisputably disinflationary policy, even if you believe it necessary — has found more favour with conservative (traditionally older) voters.
Analysts at Morgan Stanley have argued that Japan failed to escape inflation in the mid-2000s because of an entrenched preference for deflation in a legislature that is biased towards the elderly — the seats with the highest proportion of elderly voters relative to youth voters require fewer votes to win. Such gerontocracies are entrenched by the habitualness of older voters, regularly turning out and voting for the same party. Our own analysis suggests that there is the foundation for
a similar bias in the UK. As the downward sloping trend line in figure 8 shows, constituencies with an older demographic contain fewer voters and therefore require fewer votes to win. They also tend to be in the regions that the ONS estimates will age the most over the next 20 years. The correlation weakens when we look at those registered to vote, however.
Is Japan a special case?
The speed at which Japan’s population is ageing is infamous. So too is its inflation problem: it’s been stuck around zero since the mid-1990s, when the 15- to 64-year-old population peaked. Many people assume that Japan’s demography led to its deflation, and that this manner of causation must be universal. We are far from convinced. Japan’s situation is unique in various ways:
The dependency ratio
Japan’s dependency ratio (non-workers relative to the working-age population) started to increase after 1992, when this ratio for the rest of the world still had much further to fall. While labour was becoming scarcer in Japan, it was becoming more abundant abroad. In a new era of globalisation, Japanese firms ‘offshored’ production, making use of this abundant labour to keep wages from rising. Japan was also next door to the epicentre of the offshoring revolution — China. Japan aged alone; the rest of the world is doing so more or less in synchrony. That’s a very important difference.
A shrinking population
Furthermore, Japan had to deal not only with a rising dependency ratio, but also a shrinking working-age population and a soon-to-be-shrinking total population all at the same time. Again, this is unique. The UK’s dependency ratio troughed in 2007 but its working-age population isn’t projected to shrink until 2072 and its total population isn’t set to shrink this century. Of course, these projections are conditioned on a reasonable degree of immigration (and Brexit may change that in the UK). Japan had legislative and cultural barriers that kept immigration extremely low, and prevented younger immigrant workers from offsetting the ageing, native workforce. Not only was Japan’s ageing exceedingly rapid, it was also poorly understood. The academic literature was nowhere near as rich, and firms and governments had little knowledge of the consequences. This situation led to an unanticipated decrease in demand — a shock that once realised contributed to a collapse in investment and spending that was already under way after a severe financial crisis (Koo, 2011; Yoon et al, 2014).
Losing market share
While household savings declined to support pensioners’ spending habits, corporate savings ballooned, which was probably a key factor (Gruber, 2015). Japanese firms also started to lose market share around the world. It’s worth considering a counterfactual: if Japanese firms had been better at exporting, would gross investment have declined by so much and would the deflationary force have been nearly so great?
Ageing and deflation
In 2014 three IMF economists released a paper entitled, Is Japan’s Population Aging Deflationary? It concluded it was, and is very frequently cited as evidence that all ageing must be deflationary. But the paper actually supports the idea that it is unlikely to be for the rest of the world. The study confirmed that aggregate demand has been well-supported by retirees spending from savings. However, Japanese savers held a large quantity of financial assets overseas, so dissaving by the elderly resulted in a repatriation of foreign savings, which in turn led to the appreciation of the yen.
Remember that Japan had offshored much of its production so its rate of inflation was much more dependent on import prices, which fall as the yen rises. The deflationary impact from currency appreciation more than offset the inflationary effects from higher demand relative to supply (Anderson et al, 2014). Clearly, if many advanced economies are repatriating foreign savings at the same time, the exchange rate effect is not going to be so great.4
1. Although that’s because current projections extrapolate past rates of immigration. Brexit may nullify this assumption, as might changing political attitudes in general. If we assume zero net migration, the dependency ratio will increase by an additional 2 percentage points over the next 20 years.
2. The wealthiest capital owners may also spend vast sums on items that don’t even show up in the most commonly referenced inflation baskets — at an extreme, yachts and old master paintings — so fewer yachts but more TVs, say, increases inflation by statistics alone.
3. An assumption in which we are fairly confident. Consider just how difficult a job Italy, Spain or France has had making minor (and essential) changes to welfare programmes in order to incentivise more people to join the workforce; or the backlash against the Conservative Party’s ‘dementia tax’ in the UK; or the difficulty the Republican Party is having rolling back Obamacare despite years of highly vocal and vitriolic criticism.
4. We include the dependency ratio in our estimation of behavioural equilibrium exchange rates, which attempt to compute the exchange rate justified by long-term economic fundamentals. The pound, which is already extremely undervalued relative to equilibrium, is unlikely to be driven lower by the UK’s relative demographic profile (Rathbones, 2016).