Determinants of the real interest rate

Remarks by Philip R. Lane, Member of the Executive Board of the ECB, at the National Treasury Management Agency

Dublin, 28 November 2019

Introduction

It is a pleasure to address the Annual Investee and Business Leaders’ Dinner organised by the National Treasury Management Agency (NTMA).^[1] I plan today to explore some of the factors determining the evolution of the real (that is, inflation-adjusted) interest rate over time.

This is obviously relevant to the NTMA in its role as manager of Ireland’s national debt and also to its other business areas (Ireland Strategic Investment Fund; State Claims Agency; NewEra; National Development Finance Agency) and related entities (National Asset Management Agency; Strategic Banking Corporation of Ireland; Home Building Finance Ireland). More generally, the real interest rate is at the core of many financial valuation models, while simultaneously acting as a fundamental macroeconomic adjustment mechanism by reconciling desired savings and desired investment patterns.

My primary focus today is on the real interest rate on sovereign bonds, which in turn is the baseline for pricing riskier bonds and equities through the addition of various risk premia. The real return on government bonds in advanced economies has undergone pronounced shifts over time. Chart 1 shows that, since the 1980s, the real return on sovereign debt has registered a steady decline towards levels that are low from a historical perspective. Looking at the 1970s, ex-post calculations of the real interest rate were also low during this period, since inflation turned out to be unexpectedly high. In contrast, the current low levels of the real interest rate take the form of low nominal yields, since inflation is itself low and stable.^[2] Understanding the root causes of the low level of real interest rates is a high priority for monetary and fiscal policymakers, financial-sector participants and the wider populations of savers and investors.

The natural rate of interest

The falling trend in yields can be interpreted as a decline in the so-called natural or neutral rate of interest (labelled as r* in academic research and policy discussions). The natural rate of interest corresponds to the level of the real short-term interest rate that defines a neutral policy stance: this corresponds to a situation in which the economy is operating at potential and inflation is at its target value, such that there is no reason for the central bank to either inject or withdraw stimulus.

In what follows I will explore the causes of this trend decline in r*.

Causes of the persistent low real yield environment

I will structure my discussion of the drivers of r* around three broad driving forces: first, the determinants of potential growth rates; second, demographics; and third, diverging developments in the returns on risky and safe financial assets.^[3] These three factors are embedded in the textbook economic growth model, which relates the equilibrium rate of interest to economic growth, population growth and the discount rate. I will primarily focus on common international trends, even if cross-country differences in these driving forces have important implications for return differentials and international capital flows.^[4],[5]

The potential growth rate

One fundamental driver of r* is the potential growth rate of the economy: in a high-growth economy, it takes a higher real interest rate to encourage the volume of saving required for the high investment levels needed to sustain a fast-growing economy. Although easier to see with the benefit of hindsight, there has been a sustained decline in the potential growth rate of advanced economies in recent decades: it is estimated to be less than half of what it was 50 years ago (Chart 2).^[6] This pattern also carries over to expectations about future growth performance. The decline in potential output growth in turn can be attributed to a decline in total factor productivity growth (Chart 3) and a corresponding decline in labour productivity growth (Chart 4).^[7]

A range of inter-related structural and technological factors have been advanced to explain the decline in the potential output growth rate. Assessing the determinants of the rate of technological innovation is beyond the scope of this speech.^[8] In addition to the pace of frontier innovation, aggregate productivity growth also depends on the rate of technology diffusion. There are signs that the diffusion rate has slowed: for instance, the gap between the labour productivity growth of firms operating at the technology frontier in a given sector and that of firms that are lagging behind the technology frontier in that sector (non-frontier firms) has increased over time (Chart 5).^[9] This pattern is even more pronounced in the services sector than in manufacturing.

This, of course, raises the question of why technological diffusion has slowed down. One candidate is the fall in business dynamism as measured by business churn: the rate at which firms exiting the market are replaced by new ones has declined measurably over the last decade (Chart 6).

Finally, shifts in the sectoral distribution of economic activity also play a role. Employment growth has been concentrated in the services sector in recent years, and productivity growth in services has been weaker than in other sectors, such as manufacturing and information technology. As a result, the growing share of services in total employment mechanically implies a drag on productivity growth in the economy as a whole (Chart 7).

Assuming that the structural shifts I have just described account for a loss in potential output growth of around 1 percent, a back of the envelope calculation suggests that these account for a decline in real equilibrium yields of a similar magnitude.

Looking ahead, it is difficult to predict whether these structural trends will be reversed. In particular, there is a wide range of views about the potential economic impact of digitalisation, automation and artificial intelligence, which depends on the success rate in converting new technologies into economically-successful business applications. In turn, aggregate investment dynamics will be heavily influenced by this discovery process.

Another open question relates to climate change and the carbon transition. On the one side, future productivity growth may be further depressed as a consequence of climate change – either directly due to natural disasters or indirectly due to required climate change mitigation measures. On the other side, the carbon transition is also a spur for investment and opens up new opportunities for innovation and productivity growth.^[10]

Demographics

Let me now turn to a second driving force for the real interest rate: demography. Over the last 50 years, demographics in advanced economies have been characterised by low fertility rates and rising life expectancy (Chart 8). As a result, people now expect to enjoy many more years in retirement than they did in the 1970s, as Chart 9 shows for the euro area. Similarly, while in the 1970s and 1980s, the ratio of the elderly (aged 65 and over) to the working-age population (aged 15-64) was around 20 percent, the European Commission projects the old age dependency ratio will rise to over 50 percent by the middle of this century (Chart 10), while the ratio of older workers relative to younger workers has trended up significantly over the last decades (Chart 11). Over the same period, the annual growth rate of the working-age population has fallen from over 1 percent to zero and is projected to turn negative (Chart 12).

But how do these demographic trends – which are without precedent – affect the equilibrium interest rate?

A number of mechanisms are at play. First, the ageing of the population can depress the demand for capital. This in turn is due to the fact that the ratio of installed capital relative to the size of the work force increases as the population ages. Second, ageing can lower productivity growth and thereby reduce investment opportunities. This materialises if the productivity of older age cohorts is lower than that of younger age cohorts. Third, in relation to savings rates, if rising life expectancy implies longer retirement periods then individuals and those responsible for publicly provided pensions face incentives to save more. Taken together, a lower level of desired investment and a higher level of desired saving imply a reduction in the natural rate of interest.

Some observers have conjectured that the negative impact of ageing on the equilibrium real rate will be reversed once the largest age cohorts retire and begin dissaving.^[11] However, the increase in the capital-labour ratio associated with a sustained phase of higher saving and a contraction in the workforce will weigh on the level of real interest rates for a considerable time.^[12]

In terms of magnitude, a wide range of estimates suggest that the downward impact on equilibrium real rates from slowing population growth and rising life expectancy over the period from 1980 to 2050 amounts to about 1 to 2 percentage points in both the United States and the euro area.^[13]

While demographics constitute a powerful inter-generational influence on aggregate savings, the intra-generational distribution of income also matters. Chart 13 shows that the share of national income earned by households in the top 1 percent income bracket has increased considerably since the 1980s, both in the euro area and, even more strongly, in the United States. Assuming that the propensity to consume out of income is lower for rich households than for those with lower income and wealth, greater inequality may be associated with an increase in savings and further downward pressure on the real interest rate.^[14]

Diverging trends in the return on capital and on safe financial assets

A further complication relates to shifts in the relation between the return on so-called safe assets (such as highly-rated sovereign bonds) and wider measures of rates of return, including equity returns and returns on higher-risk debt instruments.^[15] In related fashion, there has also been a shift in the relation between the returns on short-term instruments and those on longer-term instruments, with term premia that are much smaller now than in earlier periods (or even negative).^[16]

On the one side, we have observed that corporate bond spreads have held up and estimates of the equity risk premium have increased, in both the United States and the euro area (Chart 14). On the other, the demand for short-term money-like instruments has surged, and the yield on those instruments has dropped to unprecedented negative values.

A vibrant field of research seeks to reconcile the observed divergence in rates of return across asset categories with reference to shifts in mark-ups and risk premia.^[17] This literature also highlights the increasing value attached to safe assets that offer protection against downside risks.^[18]

The premium on safe assets is due to several inter-related factors, with the relative importance of individual factors shifting over time. First, ageing may play a role in moving portfolio preferences towards lower-risk assets, with older savers seeking safer income streams as retirement approaches. Second, at a global level, the accumulation of reserve-currency low-risk assets has been a central strategy for many emerging economies, in seeking to limit financial fragilities associated with a lack of foreign-currency liquidity.^[19] However, this force has plausibly weakened in recent years, due to the strengthening of domestic-currency financial systems in many emerging markets and the adoption of more flexible exchange-rate regimes.

Third, the safety premium also reflects the lingering effects of the global financial crisis and the euro area sovereign debt crisis (Chart 15). By revealing the harsh costs of exposure to financial disasters, such crises tend to shift portfolio preferences in the direction of assets that preserve their value during bad times. In a similar vein, in recognition of the fact that the pre-crisis financial system was not sufficiently capitalised (compared with risk exposures) and not sufficiently liquid, regulatory reforms have also fuelled demand for safe assets, since banks have a reduced risk appetite and are also required to hold more liquid assets.^[20]

In terms of the supply of safe assets, the surge in public debt in the wake of the crisis and the operation of the doom loop between the risk profiles of banking systems and sovereigns has led to a relative contraction in the share of highly-rated sovereign bonds in the euro area. The share of government bonds in the euro area with AAA-rating from at least two rating agencies collapsed by two thirds and has become more concentrated (Chart 16). In the euro area the ratio of highly-rated short-term assets relative to GDP appears low in international comparison (Chart 17). More generally, safe asset scarcity tends to be most visible in yields on assets that provide the best hedge against liquidity, interest rate, or default risk (Chart 18).^[21],[22]

Empirical Summary

The decline in potential growth rates, demographic trends and the portfolio shift towards safe assets combine to put downward pressure on the equilibrium real rate of interest. This rate is not directly observable: at any given point in time, temporary factors push the economy away from its underlying steady state so that the measured real interest rate also reflects these shocks (and the associated policy responses). However, analysis by the ESCB Working Group on Econometric Modelling indicates that, taking the impact of these driving factors into account, the equilibrium real interest rate in the euro area has been around zero or negative in recent years (Chart 19).^[23] Similar estimates are reported in the growing literature on the global equilibrium real interest rate.^[24] While Chart 19 shows that estimates of r* are inevitably not very precise, the general downward drift is quite clear.

Policy Issues

What are the implications of low equilibrium real rates for policy?

Taking monetary policy first, it is important to recognise that the primary determinants of long-term equilibrium real rates are non-monetary in nature: potential growth rates; demographics; risk preferences in portfolios. Nonetheless, the implications of the level of real interest rates for the formulation of monetary policy are profound. In particular, monetary policy ultimately has to be calibrated vis-à-vis the underlying equilibrium rate of interest. If, for example, inflation stabilisation calls for monetary stimulus due to a negative demand shock, the central bank is required to engineer an easing of financial conditions relative to the steady state. If the prevailing inflation rate falls short of the inflation aim and the equilibrium real rate is low – or even negative – this may call for a substantially negative policy rate in order to stabilise medium-term inflation.

Furthermore, if the policy rate is already close to the effective lower bound, alternative methods of providing monetary stimulus may be required to substitute for deeper rate cuts. For example, a 1 percentage point decline in steady-state inflation translates into an equivalent loss of capacity to cut the policy rate in the event of an adverse shock: this is a large magnitude, given standard calculations of the monetary easing required during downturns in order to stabilise inflation over the medium term. Options to ease monetary policy when the conventional policy space is constrained include forward guidance about the future path of policy rates; asset purchase programmes; and targeted lending operations. Since the summer of 2014, the ECB has been active in employing all of these instruments and our assessment is that these unconventional policy measures have been highly effective in easing monetary and financial conditions.^[25] At the same time, to avoid that the effective lower bound constrains monetary policy too frequently – and thus to reduce the need for unconventional policy tools – it is essential that inflation is centred at its target over the medium term, so that the limitations imposed by a low equilibrium real rate are not further compounded by an inflation trend that is also too low.

In relation to other policy dimensions, a low equilibrium real rate reflects – as I discussed earlier – low expected potential growth rates. It follows that growth-enhancing structural policies and growth-optimising public investment rates would translate into an upward shift in the equilibrium real rate. In addition, population ageing poses many policy challenges, and the policy choices that are made will have significant implications for trend savings and investment rates. In addition, in the event of negative shocks, the speed of convergence back to steady state can be accelerated if macroeconomic stabilisation is supported by counter-cyclical fiscal policies (where feasible) in addition to the stabilisation policies of central banks.

In addressing the safe-asset premium, policies that reduce negative tail risk are essential. Much has been done in the aftermath of the crisis to mitigate tail risks, especially in relation to the financial system: bank capitalisation levels have been increased, reinforced by an additional layer of loss-absorbing debt instruments; supervisory oversight has been enhanced, including through the creation of the Single Supervisory Mechanism; macroprudential policy frameworks have been initiated; macro-financial imbalances are more contained; and the European Stability Mechanism is a source of stability in the sovereign debt market. Still, it is also clear that the current system for bearing risk remains far from optimal, with the completion of banking union and the deepening of capital markets union offering much potential for risk sharing. In turn, a greater degree of risk sharing reduces the need for individual entities to self-insure through the accumulation of safe assets.

On the supply side, the sustained through-the-cycle maintenance of fiscal discipline (in combination with an active role for counter-cyclical fiscal stabilisation policies, which can help maintain the equilibrium real rate at a higher level) will over time lead to an improvement in the risk assessments of the national sovereign debt of euro area countries.^[26] In addition, the supply of safe assets could be boosted at the euro area level through a range of innovations, even if some of these proposals require a significant deepening of fiscal union at the area-wide level.^[27]