VoxEU Column Europe's nations and regions Global crisis Monetary Policy

The economic impact of QE: Lessons from the UK

With the Bank of England recently announcing an additional £75 billion of quantitative easing, a reasonable question to ask is whether the last £200 billion has made any difference. This argues that QE may have helped boost real GDP by as much as 2% and inflation by 1.5%, similar to the effect from a drop in the base rate of around 300 basis points.

The decision by the Bank of England’s Monetary Policy Committee (MPC) to extend its QE programme follows a gap of more than a year and a half since the completion of its first round of asset purchases. The first QE purchases (henceforth QE1) were announced in March 2009, at the same time as the MPC cut the Bank Rate to 0.5%, its effective floor. The MPC judged that without additional measures nominal spending would be too weak to meet the 2% CPI inflation target in the medium term. The committee therefore also announced that it would begin a programme of large-scale asset purchases financed by central bank money, with the aim of injecting money into the economy, boosting nominal demand, and thereby helping to achieve the 2% inflation target. Over the period from March 2009 to January 2010, the Bank bought £200 billion of assets, consisting mostly of medium- and long-term UK government securities (gilts). This represented nearly 30% of the amount of outstanding gilts held by the private sector at the time and around 14% of annual nominal GDP.

Some economists have expressed scepticism about the effectiveness of QE. In conventional New Keynesian models, for example, QE is irrelevant unless it signals something about future policy that gets incorporated into expectations of future interest rates or inflation (see eg Eggertsson and Woodford 2003). But this kind of irrelevance result depends on strong Ricardian-like assumptions, which may not hold in practice. Ultimately, judging the effectiveness of QE is an empirical issue. This column reviews evidence on the impact of QE1 on the UK economy, drawing on our latest research (Joyce et al 2011b). We find that while there is considerable uncertainty about precise magnitudes, the evidence suggests that QE1 had economically significant effects.

How did it work?

The main effects of QE1 seem to have come through higher asset prices and the consequent reduction in borrowing costs and increases in wealth of asset holders.

Although signalling effects may have played a role, our recent work places more emphasis on the portfolio-balance channel, as the main mechanism at work. The portfolio-balance channel depends on the idea of imperfect asset substitutability. When the central bank purchases assets, this increases the money holdings of the sellers. But if money is not a perfect substitute for the assets sold, the sellers will attempt to rebalance their portfolios by buying other assets that are better substitutes. This shifts the excess money balances to the sellers of those assets who may, in turn, attempt to rebalance their portfolios by buying further assets — and so on. This process will raise the prices of assets (reducing their expected returns or risk premia) until investors, in aggregate, are willing to hold the overall supplies of assets and money. The consequent increase in asset prices/lower yields leads to lower borrowing costs and higher wealth.

Of course, there are a number of other possible ways QE may work. As well as affecting asset prices, QE may have broader effects on confidence and on bank lending, though the latter channel might be less material during a time of financial crisis when banks are repairing their balance sheets.

What is the evidence?

Most of the evidence comes from financial markets. Event studies show that there were significant falls in medium- to long-term gilt yields after important QE announcements, summing to just under 100 basis points (see Joyce et al 2011a). Using survey data of economists’ expectations regarding the total size of QE purchases, it is also possible to calculate the amount of asset purchase ‘news’ in each announcement and relate the latter to observed yield changes (Figure 1). A simple regression of yield changes against QE news calculated in this way suggests that medium to long-term gilt yields fell on average by 0.6 basis points in response to each additional £1 billion of unanticipated QE purchases that were announced. This would be consistent with an overall effect on yields of more than 100 basis points for the whole £200 billion programme.

Figure 1. Size of surprise and average gilt yield changes

Source: Joyce et al 2011b (Thompson Reuters Datastream and Bank calculations).

Figure 2. Changes in major UK asset prices

Sources: Bloomberg, Merrill Lynch and Bank calculations.

The portfolio-balance channel suggests that investors might have tried to reduce their money holdings by buying other assets, to the extent that they were regarded as closer substitutes for gilts than money. Similar event-study analysis of other asset prices shows that corporate bond yields fell broadly in line with gilt yields and sterling fell modestly in line with relative interest-rate movements. Equity prices did not react in a uniform way to QE news, but it seems plausible that any portfolio-substitution effects into equities would not immediately have been reflected in market prices. During 2009 UK asset prices all rose strongly (see Figure 2). Of course, this cannot be wholly attributed to QE1, as it accompanied a more general pickup in international asset prices, but it seems likely that the Bank’s asset purchases were a contributing factor.

What about the wider macroeconomic effects? Net equity and corporate bond issuance increased in 2009 (consistent with lower costs of capital-market borrowing). Measures of consumer and producer confidence improved. Short- to medium-term inflation expectations picked up. But it is difficult to isolate the incremental effects of QE1, given other relevant factors, not least the other policy measures taken both domestically and internationally over the same period. We need some kind of counterfactual analysis, though this is obviously not helped by the relative uniqueness of the policy.

We draw on new Bank of England research, which uses a number of different methods to estimate the macroeconomic effect of QE1 (Table 1). One approach uses as its starting point the event-study evidence that QE1 reduced medium- to long-term gilt yields by about 100 basis points. It then asks what effect a shock of this magnitude would be expected to have using a macroeconometric model. The first two rows of the Table show the peak effects on real GDP and inflation that have been derived from this kind of approach using, respectively, a small structural vector autoregression and a range of more sophisticated time-series models incorporating structural change in various ways (taken from Kapetanios et al forthcoming). Another method of estimating the effects of QE is to focus on its impact on the money supply. Bridges and Thomas (forthcoming) first calculate the impact of QE on broad money, allowing for various other influences over the period. They then apply their estimates to two econometric models that allow them to calculate how asset prices and spending need to adjust to make money demand consistent with the increase in supply. Their preferred model estimates are shown in the third row of the Table. As a cross check on these model-based estimates, we can also take a more bottom up approach. First, taking the impact on gilt prices and other asset prices, we use a range of models to link the changes in asset prices through to consumption and investment. This gives a figure for the impact on GDP, which can be translated into an inflation effect using a Phillips-curve relationship. The results from this analysis are shown in the fourth row of the Table.

Table 1. Estimates of the macroeconomic impact of QE, peak impact on the level of output and inflation

Method

Level of GDP

CPI inflation

SVAR

1 ½ %

¾ pp

Multiple time-series models average impacta

1 ½ %

1 ¼ pp

Monetary approachb

2 %

1 pp

Bottom up approach

1 ½ - 2 ½ %

¾ - 2 ½ pp

Range across methodsc

1 ½ -2 %

¾ - 1 ½ pp

a Kapetanios et al (forthcoming) (these estimates are based on the lower variant reported by the authors); b Bridges and Thomas (forthcoming). c Calculated using the centre of the reported range for the bottom up approach.

All the estimates shown in Table 1 are highly uncertain, particularly as none of the methods used to produce them fully capture all the likely transmission channels, but what is interesting is that they all produce figures that are in a similar ballpark. Taking the estimates together, they imply that QE1 could have boosted real GDP by as much as 1.5% to 2% and increased inflation by between 0.75 and 1.5 percentage points. Using a ready-reckoner from the Bank of England’s forecasting model suggests that this would be equivalent to a 150 to 300 basis-point cut in the Bank Rate, a significant reduction. Of course, there are large uncertainties even with this range and it is possible that the effects could have been larger or smaller. Nevertheless, they do seem economically significant and broadly comparable with the estimated effects of the Fed’s LSAP programmes (see eg Chung et al 2011).

What this implies for QE2 is less clear. The economic and financial circumstances in which further asset purchases are being made are different from those that prevailed in early 2009, so it cannot be assumed that the magnitude of the effects will necessarily be the same. We leave estimating the economic impact of QE2 for a future column.

References

Bridges, J, and R Thomas (forthcoming), “The impact of QE on the UK economy – some supportive monetarist arithmetic”, Bank of England Working Paper, forthcoming.

Chung, H, J-P Laforte, D Reifschneider, and J Williams (2011), “Have we underestimated the likelihood and severity of zero lower bound events?”, Working Paper 2011-01. San Francisco: Federal Reserve Bank of San Francisco.

Eggertsson, G and M Woodford (2003), “The zero bound on interest rates and optimal monetary policy”, Brookings Papers on Economic Activity 1:139-211.

Joyce, M, A Lasaosa, I Stevens, and M Tong (2011a), “The financial market impact of quantitative easing in the United Kingdom”, International Journal of Central Banking, 7(3):113-161.

Joyce, M, M Tong, and R Woods (2011b), ‘The United Kingdom’s quantitative easing policy: design, operation and impact’, Bank of England Quarterly Bulletin, Q3:200-212.

Kapetanios, G, H Mumtaz, I Stevens, and K Theodoridis (forthcoming), “Assessing the economy-wide effects of quantitative easing”, Bank of England Working Paper.

5,249 Reads