5 The Long-term global outlook for interest rates

Report made by the OECD for the commision, December 2014¹

5.1 Introduction and summary1

This report updates OECD long-term projections to 2060, previously published in the May 2014 OECD Economic Outlook (OECD, 2014a), with a particular focus on the outlook for global interest rates. The new long-term projections to 2060 extend the revised short-term projections to 2016 published in the November 2014 Economic Outlook.

The global model used to generate the long-term projections is described in Johansson et al. (2013) with a summary overview provided in the Appendix. Within this modelling framework there are three distinct forces acting on the determination of long-term interest rates: cyclical, whereby current and expected short-term policy rates exert a pull on long-term interest rates; country-specific risk premia related to the level of government and external debt; and global pressures relating to the balance between ex ante global savings and investment which influence interest rates in all countries. These forces play out in the projections of long-term interest rates, here represented by the rate of 10-year government bonds, as follows:

• The current low level of OECD long-term nominal interest rates is projected to persist over the coming two years, before rising in the baseline scenario to 2020 as output gaps progressively close and policy rates normalise; the average OECD (nominal) long-term interest rate will consequently rise from about 2.6 percent in 2016 to 4.6 in 2020.

• Relative to the previous set of long-term projections published in May 2014, this implies a more protracted period of low short- and long-term interest rates, which reflects the weaker outlook for growth, especially in the euro area.

• Between 2020 and 2030, the real OECD long-term interest rate remains stable at just over 2%, which for the median OECD country implies an interest rate-growth differential close to zero, which is similar to the differential experienced over the pre-crisis period 2000–07. Fiscal consolidation assumptions, whereby OECD countries target a gross debt-to-GDP ratio of no more than 60% play a role in restraining interest rates over this period.

• Beyond 2030, a gradual rise in long-term interest rates is projected mainly due to increasing pressure from an ex ante global savings shortage, which can be mainly traced to demographic pressures from ageing populations. Real OECD long-term interest rates rise from just over 2% in 2030 to nearly 3½ per cent in 2060 which would imply an interest-rate growth differential for the median OECD country of nearly 2 percentage points.

• The long-term projections are particularly sensitive to developments in China and India because they account for such large share of global saving. The rapid ageing of the population in China is one of the main factors driving up global interest rates after 2030.

• The interest rate projections depend on a number of key assumptions, including future levels of indebtedness. In the baseline scenario governments are assumed to target a gross debt-to-GDP ratio of 60% of GDP, which restrains interest rates both by lowering country-specific risk premia and by raising global saving. An alternative scenario, in which government debt is assumed to stabilise close to current levels would imply higher interest rates, with average OECD real interest rates up by nearly a full percentage point compared to the baseline scenario in 2030 and by over ½ a percentage point in the longer-run.

The remainder of the report is organised as follows: section 2 provides a brief summary of the long-term GDP projections; section 3 describes the interest rate projections and the underlying forces which generate them, as well as considering their sensitivity to some of the underlying assumptions. A description of the model used to generate the projections as well as the underlying assumptions is provided in the Appendix.

5.2 Long-term GDP projections

Some of the main features of the long-term projections published in the May 2014 OECD Economic Outlook include:

• The composition of global output will continue to shift towards emerging economies as well as towards Asia; the combined GDP of China and India was 33% of that of the OECD in 2010 (on a PPP basis), but is expected to rise to 73% by 2060.

• The crisis is estimated to have reduced OECD-wide potential output per capita by about 3¼ per cent, although for some European, mainly euro area, countries, the effect has been much larger. Nevertheless, an assumption underlying the long-term projections is that there has been no permanent damage to underlying trend productivity growth.

• Providing growth of the technology frontier continues at historical rates, average growth in OECD GDP per capita over the period to 2060 is projected to be similar to the 1½ per cent per annum experienced in the immediate pre-crisis period.

• With only a few major exceptions, the adverse effect of population ageing on labour utilisation in OECD countries will be largely offset by rising labour force participation. Up until 2030, this is achieved in most countries through already legislated increases in pensionable age, the positive effect of increased education and trend increases in female participation. Beyond 2030 it will require additional reforms to ensure that retirement ages are effectively indexed to life expectancy.

• For nearly all OECD countries, with the exception of Japan, the additional fiscal consolidation that would be required to reduce government debt to 60% of GDP by 2030 is less, and mostly much less, than that already expected to have been achieved between 2010 and 2015. For Japan, even a massive fiscal consolidation of nearly 11 percentage points of GDP beyond 2015 would only stabilise the debt ratio by 2030, suggesting that a complementary and more ambitious package of structural and macroeconomic policies is needed.

There have been a number of updates and changes since the long-term projections were published in May 2014 and these can have implications for the updated long-term projections:

1. Firstly, the short-term outlook for GDP has deteriorated; OECD GDP growth for both 2014 and 2015 has been revised down by about ½ percentage point between the May and November Economic Outlook projections, with much of the revision focussed on the euro area. Moreover, the short-term projections have been extended to 2016, implying a weaker recovery than the operation of the mechanical rules underlying the earlier model-based projections. A weaker short-term outlook for GDP growth implies a lower profile for policy rates and also lower long-term interest rates.

2. Historical GDP data have been substantially revised for most European countries based on the new System of National Accounts (SNA2008 or ESA2010, its European equivalent). The revisions imply upward revisions to the most recent levels of GDP, typically of between 1% and 6%, although implied revisions for recent growth rates (and hence for recent and projected potential output growth rates) are more ambiguous.

3. Recently revised estimates of Purchasing Power Parities (PPPs) for non-OECD countries from the International Comparison Program have been incorporated which substantially increase the relative level of GDP of those countries on a PPP basis; for example, they imply China’s GDP was about 20% higher and India’s about 30% higher. However, this does not directly affect the projection for interest rates, because pressures on interest rates are evaluated by aggregating global saving and investment using market exchange rates rather than PPPs.

OECD potential output growth is projected to increase by 2¼ per cent per annum, (similar to pre-crisis growth rates) to 2030, before moderating to 1¾ percentage point over the period 2030–60 (Table 1). The decline in growth rates of non-OECD countries is more marked: from averaging just under 6% per annum since 2000, they are projected to grow by 4¾ percent per annum to 2030 and only 2¾ percent per annum over the period 2030–60. The main reason for the slowdown in non-OECD growth rates is a decline in productivity growth. Although projected non-OECD productivity levels will remain well below levels in the leading OECD countries, important differences in underlying structural settings, including policies, limit the scope for catch-up. Ageing populations imply that over future decades the contribution to per capita GDP growth from labour utilisation is likely to be less than in the past. Nevertheless, an assumption that policies in OECD countries adapt to ensure that the length of working lives increase with life expectancy beyond 2030, ensures that the employment rate is not a major drag on growth in most countries.

5.3 Long-term interest rate projections

The determination of interest rates

Short-term interest rates are assumed to vary with the cycle, so that as output gaps close, short-term interest rates return to neutral levels which are determined consistently with nominal potential growth. Long-term interest rates are determined as a convolution of short-term rates plus a term premium. Additional effects on interest rates come from: country-specific premia related to the level of indebtedness; and a ‘world balancing premium’, which applies equally to all countries and ensures ex post global savings and investment remain in balance.

• Higher government debt levels are assumed to entail higher country-specific fiscal risk premia consistent with the findings of Égert (2010) and Laubach (2009): for every percentage point that the debt ratio exceeds a threshold of 75% of GDP, the fiscal risk premium applied to long-term interest rates increases by 2 basis points, with an additional increase of 2 basis points for every percentage point that the debt ratio exceeds 125%. Japan is an exception to this rule, given the high proportion of government debt which is financed domestically, so that the fiscal risk premium is computed at one quarter the rate for other OECD countries. No allowance for an additional interest rate premium is made for countries which do not have their own national currency.

• Higher levels of external indebtedness lead to higher country-specific risk premia that are reflected in market interest rates, consistent with the findings of Lane and Milesi-Ferreti (2001), Rose (2010) and Turner and Spinelli (2013). This is implemented as an increase of 2 basis points in interest rates for every percentage point increase in the ratio of net external debt to GDP. For countries that are net creditors, there is no discount placed on their domestic interest rate consistent with the findings of Turner and Spinelli (2013).

• Movements in global interest rates ensure that global saving and investment remain aligned. Hence if the ex ante sum of current account balances is more positive than previous periods, signifying an ex ante excess of global savings, then there will be downward pressure on all interest rates from a ‘global balancing premium’ which stimulates investment and so brings global savings and investment back into balance.

The forces acting on interest rates over the projection

The weak short-term outlook for OECD growth described in OECD (2014b) is reflected in a slow normalisation of policy rates, notably with no increase in short-term interest rates in the euro area before 2017. In the baseline scenario, policy rates in many OECD countries only return to neutral levels by the early 2020’s as output gaps close. Consistent with this outlook for short-term interest rates, the average2 OECD nominal long-term interest rate will rises from about 2.6 in 2016 to 4.4 per cent in 2020 (Table 2). For the median OECD country this implies an interest-rate-growth differential3 rising from negative values to close to zero by the early 2020s, which is similar to the differential experienced over the pre-crisis period 2000-07 (Figure 1, Panel A).

Over the period from the early 2020s to 2030, the level of average OECD real interest rates remains roughly stable as does the median interest-rate-growth differential in the baseline scenario, but thereafter they begin to rise gradually to 2060. This reflects a number of competing pressures on interest rates, with forces acting to reduce interest rates as follows:

• Reduced indebtedness, as government’s are assumed to gradually target a gross-government debt-to-GDP ratio of 60%, reduce country-specific interest rate risk premia both on high government debt and net external debt (Figure 2).

• A separate and distinct effect of OECD fiscal consolidation is to raise public and total saving (even though there is a partial Ricardian offset to private saving) and so put downward pressure on global interest rates (Figure 3).

• Slowing potential growth rates in non-OECD countries implies declining investment demand to maintain the same capital-output ratio which also puts downward pressure on global interest rates.

• The eventual dominant influence tending to reduce national saving in both OECD and non-OECD countries, and so raise interest rates, is population ageing, which in the case of non-OECD countries is re-enforced by the effect on precautionary saving from the wider provision of social safety nets (Figure 3).

Despite saving rates falling in nearly all countries, the global saving rate remains relatively stable to 2030 because of a compositional effect as the share of high-saving countries (especially China and India) in global output rises (Figure 4). Beyond 2030, however, the country compositional effect is out-weighed by the trend decline in savings rates in all countries so that interest rates in nearly all countries steadily rise. Real OECD long-term interest rates rise from just over 2% in 2030 to nearly 3½ per cent in 2060 (Table 2), which would imply an interest-rate growth differential for the median OECD country of about 2 percentage points (Figure 1, Panel A). This projection is, however, particularly sensitive to developments in China and India (because together they account for more than one-third of global savings) as well as to the magnitude of the demographic influence on savings.4

A second factor which delays the upwards pressure on global interest rates is the assumed fiscal consolidation across OECD countries to reduce debt-to-GDP ratios to 60%. In a variant “high debt” scenario in which government debt is assumed to stabilise close to current levels , average OECD real interest rates up are increased by nearly a full percentage point compared to the baseline scenario in 2030 and by over ½ a percentage point in the longer-run (Figure 1, Panel B).

Note: Selection of OECD Countries limited to include those for which there is historical data going back to at least 1990, namely: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Korea, Netherlands, New Zealand, Portugal, Slovakia, Spain, Sweden, Switzerland, the United Kingdom, and the United States.

Kilde: OECD.

Note: There is no debt premium for Austria, Denmark, Germany, Israel, Korea, Luxembourg, the Netherlands, Slovenia, Sweden and Switzerland for any of the three periods shown.

Kilde: OECD.

Note: The component bars identify the three most important contributions to changes in the saving rate, but do not sum to the “Total” because the smaller contributions from another factors are not explicitly included in order to make the chart more legible. The contribution of “Public Saving” nets out a Ricardian offset to private saving. The “Social safety net” contribution allows for the effect that the extension of social welfare systems (especially public health provision) will have on reducing private saving in non-OECD countries.

Kilde: OECD.

Kilde: OECD.

Bibliography

• Balassa, B. (1964), “The Purchasing Power Parity Doctrine: A Reappraisal”, Journal of Political Economy, 72.

• Égert, B. (2010), “Fiscal Policy Reaction to the Cycle in the OECD: Pro- or Counter-Cyclical?” OECD Economics Department Working Papers, No. 763.

• Frankel, J. (2006), “The Balassa-Samuelson Relationship and the Renminbi”, Harvard Working Paper, December.

• Johansson, A., Y. Guillemette, F. Murtin, D. Turner, G. Nicoletti, C. de la Maisonneuve, P. Bagnoli, G. Bousquet and F. Spinelli (2013), “Long-Term Growth Scenarios”, OECD Economics Department Working Papers, No. 1000, OECD Publishing.

• Kerdrain, C., I. Koske and I. Wanner (2010), “The Impact of Structural Policies on Saving, Investment and Current Accounts”, OECD Economics DepartmentWorking Papers, No. 815, OECD Publishing.

• Lane, P. R. and G. M. Milesi-Ferreti (2001), “Long-Term Capital Movements”, Centre for Economic Policy Research, Discussion Paper No. 2873.

• Laubach, T. (2009), “New Evidence on the Interest Rate Effects of Budget Deficits and Debt”, Journal of the European Economic Association, Vol. 7, pp.858-885.

• Loayza, N., K. Schmidt-Hebbel and L. Serven (2000), “What Drives Private Saving Across the World”, Review of Economics and Statistics, Vol. 82, pp. 165-181.

• OECD (2014a), “Growth Prospects and Fiscal Requirements over the Long Term”, OECD Economic Outlook, No. 95, May, OECD Publishing.

• OECD (2014b), OECD Economic Outlook, No. 96, November, OECD Publishing.

• Röhn, O. (2011), “New Evidence on the Private Saving Offset and Ricardian Equivalence”, OECD Economics Department Working Papers, No. 762, OECD Publishing.

• Rose, D. (2010), “The Influence of Foreign Assets and Liabilities on Real Interest Rates”, Institute of Policy Studies Working Paper 10/09.

• Samuelson, P (1964), “Theoretical Notes on Trade Problems”. Review of Economics and Statistics, 23.

• Turner, D. and F. Spinelli (2013), “The effect of government debt, external debt and their interaction on OECD interest rates”, OECD Economics Department Working Papers, No. 1103, OECD Publishing.

Appendix 1. The modelling framework and key assumptions

The global model used to extend the short-term Economic Outlook projections to 2060 is described in Johansson et al.(2013). The country coverage is all OECD countries as well as current non-OECD G20 countries (Argentina, Brazil, China, India, Indonesia, the Russian Federation, Saudi Arabia and South Africa), equivalent to 88% of world GDP in 2012 at market exchange rates. The level of detail in which OECD economies are modelled is greater than for non-OECD economies, reflecting wider data availability for OECD countries, particularly in respect of fiscal accounts.

The backbone of the model is a consistent set of long-run projections for potential output. Potential output is based on a Cobb-Douglas production function with constant returns to scale featuring physical capital, human capital (based on gradually declining returns to average years of schooling) and labour as production factors plus labour-augmenting technological progress. By projecting these trend input components, assuming a degree of convergence in total factor productivity and human capital, potential output is projected out to 2060. The degree of convergence in total factor productivity depends on the starting point, with countries farther away from the technology frontier converging faster, but it also depends on the country’s own structural conditions and policies, hence the ’conditional convergence’ nomenclature. In the long run, productivity in all countries will tend to grow at the same rate determined by the worldwide rate of technical progress, but cross-country GDP per capita gaps remain, mainly reflecting differences in technology levels, capital intensity, human capital and labour utilisation, which in turn partly depend on differences in structural conditions and policies.

Private saving rates for OECD countries are determined according to recent OECD empirical work (Kerdrain et al.,2010) which suggests that demographic effects, captured by old-age and youth dependency ratios, are important drivers of long-term trends in saving, but with additional effects from the fiscal balance, the terms of trade, productivity growth, net oil balances and the availability of credit. Total saving in OECD countries is determined as the sum of public and private saving, although there is a 40% offset of any improvement in public saving from reduced private saving due to partial Ricardian equivalence (in line with recent OECD estimates, for example Röhn, 2011). For non-OECD countries, the total saving rate is determined according to an equation, which is close to being a total economy variant of the private saving equation for the OECD, with effects from the old-age and youth dependency ratios, the terms of trade, the availability of credit, the level of public expenditure (as a proxy for public social protection) and productivity growth. As emphasised in the main paper one of the main drivers of saving rates in the long-term projection is old age dependency rates, and in this context it is worth highlighting that the effects incorporated in the model are in line with other estimates in the literature (for example, Loayza et al., 2000), but if anything are on the low side.

Movements in global interest rates ensure that global saving and investment remain aligned, whereas imbalances at the national level are reflected in current account balances. An exception is a group of major non-OECD oil exporting countries, defined to include Saudi Arabia, Russia as well as 27 smaller non-OECD countries. For these countries, no individual projections of current balances are made. Rather, the combined current account balance of all non-OECD oil exporting countries is calculated based on projections of their balance of trade in oil.

Current account imbalances are accumulated to provide a proxy for foreign asset positions – with higher levels of external indebtedness leading to higher country-specific risk premiums that are reflected in market interest rates, consistent with the findings of Lane and Milesi-Ferreti (2001) and Rose (2010). This is implemented as an increase of 2 basis points in interest rates for every percentage point increase in the ratio of net external debt to GDP. For countries that are net creditors, there is no discount placed on their domestic interest rate since no evidence for one has been found in recent OECD empirical work (Turner and Spinelli, 2013).

Fiscal closure rules acting on the primary balance ensure that the ratio of government debt to GDP is stable either at recent levels or by targeting a specific (usually lower) debt-to-GDP ratio, which in the baseline here is 60%. Debt service responds to changes in debt and market interest rates, but with lags which reflect the maturity structure of debt. Higher debt levels are assumed to entail higher country-specific fiscal risk premia consistent with the findings of Égert (2010) and Laubach (2009): for every percentage point that the debt ratio exceeds a threshold of 75% of GDP, the fiscal risk premium applied to long-term interest rates increases by 2 basis points, with an additional increase of 2 basis points for every percentage point that the debt ratio exceeds 125%. No allowance for an additional interest rate premium is made for countries which do not have their own national currency.

Assumptions in the baseline long-term scenario

The baseline scenario includes the following assumptions for the period beyond the short-term projection horizon that ends in 2014:

• Cross-country comparisons of levels of GDP and GDP per capita are made in terms of chained purchasing power parities (PPPs). PPPs are projected on the basis of differences in aggregate inflation and a relative price effect (reflecting the relative price of tradeables and non-tradeables) which is related to changes in relative living standards, through the so-called Harrod-Balassa-Samuelson effect, based on the empirical work of Frankel (2006). Nominal exchange rates adjust in line with relative aggregate inflation rates and also correct the level of the real exchange rate in line with predictions of the Harrod-Balassa-Samuelson effect. The upshot is that the GDP of low-income countries like India and China rise somewhat less over the projection, relative to high-income countries such as the United States, when measured on a chained PPP basis rather than on a fixed PPP basis.

Assumptions regarding monetary and fiscal policy are as follows:

• Policy interest rates continue to normalise as output gaps close and beyond that are directed to converge on a neutral real short-term rate, which in turn follows the potential growth rate of the economy.

• The target for inflation over the long-term is generally taken to be 2%, with the following exceptions: Australia, Poland, Iceland and Norway target 2.5%; Chile, Hungary, Mexico and Korea target 3%; Argentina, China and Russia target 4%; and Brazil, India, Indonesia and South Africa target 4.5%.

• Effects on public budgets from population ageing and continued upward pressures on health spending are not explicitly included, or, put differently, they are implicitly assumed to be alleviated through reforms of relevant spending programmes or offset by other budgetary measures.

Assumptions regarding structural policies are as follows:

• Policy influences on labour force participation differ between two stages of the projections. Recently-legislated pension reforms that involve an increase in the normal retirement age by 2030 are assumed to be implemented as planned, and the participation rates of older workers adjusted accordingly. In countries where no such reforms have been undertaken, retirement behaviour is assumed to reflect only effects coming from a rising education level. Beyond 2030, a more stylised assumption is adopted whereby the share of active life in life expectancy is assumed to remain constant, hence the legal pensionable age is implicitly assumed to be indexed to longevity.

• Structural unemployment in OECD countries is consistent with the most recent Economic Outlook projections to 2016 and thereafter is assumed to return gradually to a pre-crisis level, which is taken to be the lowest level of structural unemployment estimated over the period 2007 and 2013. Unemployment in non-OECD countries where the level is currently above the OECD average is assumed to gradually converge to the OECD average, while it remains unchanged in countries currently below the OECD average.

• The long-term trend increase in average years of schooling per worker (the basis for estimating human capital) is assumed to continue in all countries, and is calculated to have a modest positive effect on aggregate labour force participation.

• Product market and trade regulations are assumed to gradually converge towards the average regulatory stance observed in OECD countries in 2011 in countries where regulations are currently more restrictive, while for other countries, regulations remains unchanged.

• For non-OECD countries, a gradual increase in public spending on social protection is assumed, amounting on average to an increase of four percentage points of GDP to a level of provision similar to the average OECD country. It is further assumed that this is financed in a way so as to have no effect on public saving.

• Oil prices start from OECD short-term projection in the November 2014 OECD Economic Outlook, where they are $85 per barrel for Brent crude, gradually rise from there. For the period 2017 to 2031, they rise at 2% per year, essentially remaining constant in real terms. Then from 2030 on they rise at 3% per year. Given the recent volatility in oil prices, a variant scenario (not discussed in the main paper) was also considered in which oil prices fall to $60 per barrel and stay constant in real terms thereafter. Lower oil prices reduce global saving because they significantly reduce saving by net oil exporting countries. The variant scenario results in higher global interest rates, up by 0.1 percentage points by 2030 and 0.2 percentage points by 2050.