Philip R. Lane: The yield curve and monetary policy

SPEECH

The yield curve and monetary policy

Speech by Philip R. Lane, Member of the Executive Board of the ECB, Public Lecture for the Centre for Finance and the Department of Economics at University College London

London, 25 November 2019

Introduction

• In my contribution, I wish to share some thoughts about the yield curve and monetary policy.
• Over the past 200 years, gilt yields have never been as low as today.
• [2],[3] Long-term rates in the euro area have even reached negative territory.
• [4] Chart 1 UK, US, DE and euro area OIS long-term interest rates (percentages per annum) Sources: Bank of England, Bloomberg, FRED, Jord et al.
• Notes: The slope of the yield curve shown is the spread between ten-year and one-year OIS yields since 1999.
• German data back to 1972 are computed by the Bundesbank using the method of Svensson.
• German data prior to 1972 are extrapolated backwards based on historical series of short and long-term yields provided by the OECD.
• Long-term OECD yields refer to yields on outstanding listed federal securities with residual maturities of over nine to ten years traded on the secondary market.
• [5] The yield curve isa central element in the transmission of monetary policy.
• Standard and non-standard monetary policy instruments affect the whole of the term structure, which in turn is a key determinant of the financing conditions of the economy.
• [6] The financing conditions prevailing for firms and households in turn affect the level of economic activity and inflation.
• While current monetary policy is an important factor affecting the yield curve, beliefs about future monetary policy and risk premia also play a role.

Response of the yield curve to monetary policy in normal times

• Let me first revisit the benchmark case of how the yield curve reacts to the standard policy tool of variations in short-term interest rates.
• When measuring the effect of monetary policy on the yield curve, macroeconomists typically focus on monetary policy shocks: that is, surprise changes in the policy rate that occur independently of other structural shocks, such as innovations in aggregate demand or cost-push shocks.
• Since monetary policy decisions respond to these other shocks, naive calculations of the co-variation between monetary policy and the yield curve cannot distinguish the contribution of monetary policy from the impact of the underlying structural shocks.
• Accordingly, the independent role of monetary policy is revealed by focusing on the surprise element in monetary policy decisions, and there is a buoyant literature trying to identify these shocks and measure their impact on the yield curve and the economy.
• [8] A standard rate cut primarily affects the short end of the curve, while the impact peters out monotonically across the curve (Chart 3, left panel).
• Taking an average over such an expected rate path gives rise to the observed hump-shaped footprint.
• Besides the average of expected short-term rates (the expectations component), long-term bond rates typically contain term premia, which are time-varying.
• [11] Monetary policy exerts its effects on the yield curve not only through pure rate expectations but also through term premia.
• Chart 4 shows a model-based estimate of the yield curve response to a change in policy rates for the euro area.
• While Chart 4 shows that changes in rate expectations explain the bulk of the yield curve response, part of it is attributable to changes in term premia.
• Chart 4 Instantaneous impact of conventional policy surprise on expectations and term premia (basis points) Source: ECB calculations.
• Some studies have looked at specific channels by which term premia help explain certain yield curve phenomena that can hardly be reconciled by the expectations hypothesis alone.
• For instance, Hanson and Stein (2015) observe that monetary policy rate decreases tend to coincide with surprisingly strong reactions in long-term forward rates.
• Those players may try to reallocate their portfolios in response to the policy-induced short rate decline in order to keep their average portfolio yields constant.
• While this is one promising way to rationalise term premium responses to a monetary policy rate change, overall, our understanding of the interaction between monetary policy and term premia is still very limited.

Assessing the impact of non-standard monetary policy on the yield curve

The transmission of negative rates to the yield curve

• Chart 5 shows the characteristic footprint of a cut in the policy rate in negative territory on the yield curve.
• There is a clear difference: compared with standard rate cuts in positive territory, rate cuts in negative territory have had a stronger effect further out along the yield curve.
• Chart 5 Estimated effect of policy rate (deposit facility rate) cut in negative territory on the OIS curve (basis points) Sources: Based on Altavilla et al.
• This type of communication about the scope of negative rates can provide accommodation even in the absence of actual rate cuts.
• In technical terms: the presence of the lower bound induces a censored distribution for future short rates.
• [14] Communicating about a decrease in the lower bound makes rates possible that were previously perceived to be infeasible and, ceteris paribus, decreases expected future rates.
• That is, the rate cut in negative territory would lead to a more accentuated search for yield that could amplify the yield curve reaction compared with what is observed on average in positive territory.
• Through a similar mechanism negative rate policy can also be seen as complementary to our bond purchase programme (whose impact on the yield curve I will discuss in more detail later on).
• Accordingly, the presence of negative rates provides a strong incentive for banks to participate in the TLTROs, which support credit provision to the real economy at attractive conditions.

The transmission of asset purchases to the yield curve

• Let me now turn to large-scale asset purchases, including our own asset purchase programme (APP).
• There is a wide array of studies quantifying the effects of various forms of asset purchases on financial markets and macroeconomic variables.
• Extracting a quantitative easing factor from high-frequency interest rates and other financial market data confirms that news about ECB asset purchases had yield compression and flattening effects on the term structure, as shown in Chart 6.
• Chart 6 Estimated effect of quantitative easing surprise on the OIS curve (basis points) Sources: Based on Altavilla et al.
• [19] Under the signalling channel, central bank asset purchases serve as a commitment device to keep policy rates at low levels for an extended period of time.
• Under the signalling channel, the expectations component of bond yields decreases upon news about asset purchases, whereas under the portfolio rebalancing channel, the term premium component is compressed.
• Econometric modelling is required to discriminate between the two components and provide a deeper understanding of the transmission channels of asset purchases.
• An important episode studied by ECB researchers is the period before the announcement of the APP in January 2015.
• The second bar in Chart 7 also shows that the observed yield compression can be explained exclusively by the term premium.
• [22] Chart 7 Decomposed changes in ten-year Bund yields (percentage points) Sources: Based on Lemke and Werner (forthcoming).
• For instance, buying a French eight-year bond decreases the supply of that bond available in the market.
• For given demand, this drives up the price of the bond and compresses its yield.
• Securities of similar maturity are close substitutes, so their yield would also be affected, but those with very short and much longer maturities would not be greatly affected.
• Another way in which the portfolio rebalancing channel is set in motion is through duration extraction.
• In terms of asset pricing terminology, central bank bond purchases decrease the market price of duration risk (the expected excess return on a long-term bond per unit of risk).
• Similar to analysis that accompanied the asset purchases of the Federal Reserve, ECB economists have quantified the effect of our asset purchases on the yield curve.
• Chart 8 APP impact on the euro area sovereign term structure (basis points) Sources: ECB, based on Eser et al.


Chart 9 APP impact on ten-year euro area sovereign yield over time (basis points) Sources: ECB, based on Eser et al. (2019). Notes: Evolution of the APP impact on the ten-year synthetic sovereign yield (weighted average of Germany, France, Italy, Spain). The impact is derived on the basis of an arbitrage-free affine model of the term structure with a quantity factor (see Eser et al. 2019).

• For shorter-term bonds the effects are lower, such that the expectation of asset purchases is estimated to have led to a flattening of the curve.
• Moreover, by virtue of the stock effect, the yield compression was expected to be long-lasting and to halve only after five years.
• Owing to changes in macroeconomic conditions and the need to adapt the monetary policy response, several recalibrations were made to the APP.
• When the end of net purchases was announced in June 2018, the impact on ten-year sovereign bond yields was estimated to be around 90 basis points.
• The additional duration extraction associated with the renewed net purchases that started this month has contributed to a further compression of the term premium.

Summary of the yield curve effects of the ECB’s non-standard policies

Putting several pieces of analysis together, ECB economists have come up with the following summary assessment of how the various non-standard policy measures since 2014 have affected the yield curve. Chart 10 Compression of euro area sovereign yield curve due to ECB’s non-standard measures (percentage points) Sources: Based on Rostagno et al. (2019).
• As we see in Chart 10, covering the annual averages from 2014 to 2018, the APP accounted for the bulk of the estimated yield compression at the long end of the curve.
• Negative interest rates and forward guidance, by contrast, account for the lions share at the short end of the curve.
• These yield compressions have been transmitted further to an easing of financing conditions for non-financial corporations and households.

Interpreting signals from a flattening curve under central bank asset purchases

• An important and currently relevant example is the question as to what extent a flattening yield curve signals a weakening of the economic outlook.
• I will just briefly highlight how important it is to take our own action into account when interpreting the signals reflected in the yield curve.
• A recent paper examines how economists and commentators, when confronted with a flattening curve in the past, often argued that this time is different.
• First, the current flattening of the curve has mainly been induced by the long end of the curve coming down, rather than by the short end coming up.
• Hence, the inversion of the yield curve observed this time is not occurring in a context of monetary policy contraction raising the short end of the curve.
• Chart 11 Decomposition of change in the slope of the yield curve (percentage points) Sources: Bundesbank, OECD, Thomson Reuters and ECB calculations.


Chart 12 APP impact on the euro area OIS curve (percent per annum) Sources: ECB, based on Eser et al. (2019).

• Second, the flattening of the curve is to some extent driven by our own asset purchases conducted for the purpose of stimulating the economy.
• As is shown in Chart 12, one observes that the current slope would be around 70 basis points higher in the absence of the APP and thus much closer to the historical average.
• [33] What we learn from this example is that we need to be careful when reading the yield curve especially during times when our own actions have a significant influence on it.

Conclusion