Iryna Kaminska and Haroon Mumtaz
Since 2009, when policy rates reached their effective lower bound, quantitative easing (QE) has become an important instrument of central bank monetary policy. It is aimed to work via long-term yields. The literature confirms that QE helped lower long-term yields. But the yields have two components – expectations and term premia – and open questions remain: does QE reduce yields via expected rates or term premia? And which channel is more efficient in stimulating the economy? In our research paper, we find evidence that QE often worked through signalling and term-premia effects simultaneously. But the two main QE channels are transmitted to financial markets and the real-economy in different ways, and only signalling is found to have ultimately affected inflation significantly.
What we do
Building on and extending earlier work, we use high-frequency yield curve data combined with dynamic term structure models. These models allow us to decompose movements in the curve into two effects: policy rate expectations; and term premia. To extend the analysis to the period of QE, we use a shadow rate version of the model, imposing that the short-term interest rate has to remain above the effective lower bound.
To isolate the effect of policy moves from other factors affecting the term structure, we look at yield curve movements during a tight window around Monetary Policy Committee (MPC) meetings. This type of high-frequency event study also help us to guard against reverse causation (ie policymakers responding to moves in the yield curve). We decompose the high-frequency reaction of yields into reactions in policy rate expectations and term premia. These reactions to MPC announcements are then summarised by principal components (PCs), which distil the information from movements observed across term premia and expectations at different maturities into a small set of factors. These PCs are used to identify various types of monetary policy shocks.
What we find
First, we find that both yield components reacted to monetary policy announcements, often at the same time. This suggests that transmission channels working through expectations (signalling) and term premia have often operated together. While this finding seems unsurprising and, in fact, something that previous MPC communications has alluded to, here we show evidence of the contemporaneous and non-mutually exclusive nature of QE channels explicitly.
Second, comparing periods of pre and post-QE (see Table 1), the decomposition suggests that the expectations component has become less important at shorter maturities (with yields moving 0.019 percentage points in the window after MPC meetings post-QE versus 0.037 pre-QE). The reduced role of policy rates due to the ZLB proximity and, as a consequence, the reduced role of the path of monetary policy rates, could be a possible factor affecting the volatility of the expectations component, especially at shorter horizons. Instead, there was a larger role of term premium channels affecting long-term yields post 2009. In fact, while term premia were less important than expectations in driving yield reaction to MPC announcements before 2009 (0.008 versus 0.010), it is almost twice as reactive as policy expectations during the QE sample (0.012 versus 0.007). The relative importance of the term premia and expectations channels varies even during the QE though: UK term premia channels were dominant during 2009–15, whereas the signalling channel became more important during 2016–19, when the policy rate was an active MPC tool again.
Table 1: Volatility of yield components (term premia and expectations) on MPC announcements (percentage points)
Importantly, PC analysis of the decomposition across maturities shows that, during the QE subsample, to capture the term premia reaction to MPC announcements, an additional (QE-specific) factor is required. The need for the QE-specific factor is consistent with the additional channels through which QE is supposed to work. We find that this QE factor has largest moves coinciding with 5 March 2009, 7 May 2009, 9 July 2009, 6 August 2009, 6 October 2011 MPC announcements, each of which was heavily loaded with the information about the QE gilt purchases. These announcements came as large surprises to market participants, perhaps the largest across the sample (according to Reuters survey of QE expectations). Hence the QE-specific term premium factor can also be linked to the gilt supply, as forward-looking investors react to news of future asset purchases by offering higher prices (ie lower yield term premia) on long-term bonds, which they expect to be in shorter supply and availability after the Bank of England’s purchases.
The two main QE channels (signalling and QE-specific term premia) are transmitted to financial markets and then to the real economy in different ways. While the signalling channel has a pronounced and persistent impact across the whole yield curve, the effects of a QE-specific term premia channel are more localised at long maturities. We find that signalling has stronger effects on the real exchange rate and on equity prices than the pure QE-specific term premia channel. Although both have roughly similar impacts on the real economy (with maximum positive impact in 1–2 years), only signalling generates inflationary pressures (consistently with the associated stronger role of the exchange rate channel).
Finally, asset purchases can also lower term premia by reducing uncertainty, in particular when policy rates are close to zero and QE is paired with forward guidance. In line with this, the decomposition suggests that term premia associated with uncertainty about the future path for Bank Rate fell after QE was introduced in 2009, and subsequently reacted less to MPC announcements compared to before 2009. Feeding this decomposition into a macroeconometric model, we estimate that lower uncertainty translates into lower corporate bond spreads and improved real economic outcomes.
These findings emphasise that the QE has lowered yields via both expectations and term premia channels. The monetary transmission mechanism of these channels, and their eventual impacts on inflation and real activity, are not the same though. Therefore, to fully evaluate the overall QE impacts on macroeconomy and inflation, it is important to know not only by how much the QE policy reduces gilt yields but also via which channels the QE policy works at a particular time.
Iryna Kaminska works in the Bank’s Monetary and Financial Conditions Division and Haroon Mumtaz works at Queen Mary University of London.
If you want to get in touch, please email us at email@example.com or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.