What's going on?

## What's going on?

Speech given by

Dave Ramsden, Deputy Governor for Markets and Banking, Bank of England

Barclays Inflation Conference, London

7 June 2018

I am grateful to William Abel, Harvey Daniell, Bianca Ginelli Nardi, Raf Kinston, Simon Kirby, Elliot Luciani-Kane, Rebecca Maule, Sanjay Odedra, Tom Smith, Pawel Smietanka and Brad Speigner for their assistance in preparing this speech, and Clare Macallan, Michael Saunders and other Bank staff for their helpful comments and suggestions.

## An unusual position

UK GDP growth has slowed following the Brexit referendum, and it slowed even more in the first quarter of this year. In the MPC's May Inflation Report we are forecasting that it will remain subdued over the forecast period. If our forecast materialises, annual growth will have been between 1¼ and 2% for five consecutive years from 2016 to 2020. To give some context, annual average GDP growth has been in this range in only eight of the 50 years before that period, and three of those were since the financial crisis.

It's been said before but it bears repeating: the UK economy is in an unusual position.

Over the past few years, meanwhile, unemployment has fallen from a high of 8.5% to a forty year low of just 4.2%. Employment has similarly picked up to record highs. Usually you would expect faster growth to be accompanied by stronger employment growth, and vice versa (Chart 1) - an empirical fact so robust that economists know it as Okun's Law. 1 Since the financial crisis, however, this law has, if not broken down, at least started working rather differently than before. Unemployment has fallen far more, and employment growth has been far stronger, than might have been expected given the subdued pace of GDP growth (Chart 2).

Chart 1: GDP and employment growth

Chart 2: Okun's Law and the financial crisis

-4

Arithmetically this is accounted for by a fall in the rate of labour productivity growth. Having grown steadily for a number of decades, the level of productivity fell sharply following the financial crisis. That was in line with historical experience. 2 What there is much less historical precedence for is the failure of productivity growth to recover to its pre-crisis growth rate. A cottage industry of economists - including me - have tried to explain the so-called productivity puzzle. 3

2 Carmen Reinhart and Kenneth Rogoff (2009), 'This Time is Different: Eight Centuries of Financial Folly'. Princeton University Press 3 I set out my views in more detail in a speech in February. See: Ramsden, D. 'The UK's productivity growth challenge', speech at Babraham Hall, Cambridge, 23 February 2018. Many other members of the MPC have also given their attention to the question - see for example: Tenreyro, S., 'The fall in productivity growth: causes and implications', speech at Queen Mary University of London, 15 January 2018; Haldane, A. G., 'Productivity puzzles', speech at the London School of Economics, 20 March 2017.

Chart 3: The productivity (growth) puzzle

Finally, with unemployment falling to near-record lows, and spare capacity being rapidly eroded, you might expect wage growth to have picked up at a similar pace. The existence of a wage Phillips Curve, its stability, its slope and its identifiability in the data are collectively one of the more contentious subject areas in economics - and for some of you my next chart (Chart 4) may not do much to persuade you of any of these. Nevertheless I think most practising economic policymakers would expect to see at least some link between falling unemployment and rising wages. 4 Indeed one reason for the fall in the curve over time is arguably policymakers' success at using that link to control inflation and bring down nominal wage expectations. At any rate, the stubborn failure of wage growth to rise much above 2% - that is, above half its pre-crisis rate and workers' surprising willingness to absorb that in real wages represents one more surprising feature of the British economy.

4 The original observation was made in: Phillips, A. W. (1958). "The Relationship between Unemployment and the Rate of Change of Money Wages in the United Kingdom 1861-1957". Economica, 25 (100): 283-299. For more recent MPC discussion, see: Tenreyro, S (2018). 'Models in macroeconomics', speech at the University of Surrey, 4 June 2018; Vlieghe, G (2018), 'From asymmetry to symmetry: changing risks to the economic outlook', speech at the Confederation of British Industry, Birmingham, 23 March 2018; Broadbent, B (2017), 'Brexit and interest rates', speech at the London School of Economics, 15 November 2017; Cunliffe, J (2017). 'The Phillips curve: lower, flatter or in hiding?', speech at the Oxford Economics Society, 14 November 2017.

Chart 4: The wage Phillips Curve

I want to use my speech today to focus on the implications of the UK's unusual economic position for inflation and for monetary policy, both in terms of what has happened over the recent past and of what I expect to happen in future. I will focus on three aspects of the economy in particular. First, the ongoing consequences of the vote to leave the EU, and what those have meant and could mean in future for monetary policy. Second, the continued evolution of the labour market and how I expect that to affect inflation and monetary policy in future. And third, developments in the economy at the start of this year: the unexpected weakness of GDP and evolution of inflation.

These aspects of the economy are unusual in terms of what they imply for the UK outlook. But as a policymaker I feel reasonably confident, even with the uncertainties, that, together with my colleagues on the MPC, I have a reasonable understanding of what's going on, and in particular what the implications are for inflation. Let me explain why in more detail, and then conclude with where that leaves my policy thinking.

## The Brexit vote

Turning first to Brexit.

The vote to leave the EU leaves the UK in a position which is not just unusual but unique. As my colleague the Governor said recently, the Brexit negotiations are the biggest challenge as well as the biggest opportunity facing the country at this moment. And the outcome of those negotiations remains the most significant influence on the economic outlook, as well as the greatest source of uncertainty about the future. As a member of the MPC, my task in the face of this uncertainty is to balance any trade-off between the speed at which inflation returns sustainably to target and supporting employment and activity, in line with our

remit from the Government, as the UK economy transitions to whatever end state results from those negotiations.

What does that mean in practice?

In the MPC's assessment, consistently held since before the referendum, Brexit is affecting the UK economy via three main channels - that is, through its effects on supply, demand and the exchange rate.

On the supply side we expect Brexit to have both a long and a short run effect on productivity.

The long run effect comes about through the effects on the UK's openness. 5 We know that on average being more open to trade makes an economy more productive. That comes about through a range of channels. For example, greater competition encourages UK firms to operate more efficiently. Access to larger export markets enables them to take more advantage of economies of scale. And the ability to trade allows the UK to specialise in sectors where it has a competitive advantage. To the extent that leaving the single market or the customs union makes it harder to do business with the EU, that will reduce the economy's productive capacity. (Equally if leaving the EU allows the other economies to trade with and invest more openly in the UK then these same channels are likely to support UK productivity). We don't know whether these long-run channels are already in operation. But it is possible that they have been contributing to reinforce the continued weakness of productivity growth. 6

In the short run what we do know is that uncertainty about the eventual outcome of the Brexit negotiations, and perhaps the expectation that Brexit will lower economic growth over time, has been depressing business investment. Businesses are waiting for more clarity before embarking on new capital projects, and that reduces their productive capacity relative to what it could have been, as well as hampering resource re-allocation across the economy.

We see this most clearly in business surveys. Brexit has consistently ranked at or near the top of the list of the risks identified in the Deloitte survey of Chief Financial Officers. 7 In the Bank's own quarterly Decision Makers' Panel, around 40% of firms consistently identify Brexit as a major source of uncertainty, with less than 20% viewing it as unimportant (Chart 5). Bank staff analysis of the results points to a drag of between 3 and 4% on nominal investment growth since the referendum as a result of this uncertainty 8 .

5 For more detail on these mechanisms, see the Box on 'Factors affecting the prospects for long-term supply following the EU referendum' on pages 29-30 of the August Inflation Report.

6 It is also possible that actual or anticipated changes in institutional arrangements due to Brexit might affect the outlook for net migration and therefore population growth.

7 Specifically, it has been at the top of the list for two years, falling back for the first time to second place in Q1, with its place at the top taken by weak UK demand.

8 For more information on the Decision Maker Panel, see Bloom, N. et al (2017), 'Tracking the views of British businesses: evidence from the Decision Maker Panel', Bank of England Quarterly Bulletin, 2017Q2. The Bank's 2018Q1 Agents' Summary of Business Conditions includes the latest published results from the survey.

Chart 5: Brexit as a source of uncertainty 9

On the demand side, while Brexit did not lead to an immediate fall in GDP, it is nonetheless exerting a drag. In addition to its effects on investment, it has had a significant impact on real household incomes and therefore on consumption. The main vehicle for this impact has been the fall in the exchange rate, which remains some 15% below its pre-referendum peak. The rise in import prices caused by sterling's depreciation has had a material and persistent effect on incomes: real household income has only risen by 0.2% in the seven quarters since the referendum, compared with averaged growth of over 1½ % per year in the five years prior to it ( Chart 6 ).

Given that slowing in real income growth, you could argue that it's surprising that households haven't adjusted consumption even more. Consumption growth has certainly slowed: it has averaged around 0.3% per quarter since the referendum, compared to 0.6% in the previous five years and nearly 0.9% in the pre-crisis decade. But households have also in aggregate reduced their savings - the household saving rate has fallen by some two per cent of income ( Chart 7 ). Whether households will maintain that lower rate of saving, as in the MPC's central forecast, remains to be seen - there must be a risk that consumption could adjust further.

9 Question: 'How much has the result of the EU referendum affected the level of uncertainty affecting your business?'

Chart 6: Consumption and income growth

## Chart 7: Household saving rate

Why didn't demand fall further following the referendum? One reason is that the fall in the exchange rate, while dampening consumption, will also have supported net trade. Another is the policy actions that were taken in response. 10 In addition, demand has been helped by an unexpectedly strong global recovery, which will have boosted trade and investment, as well as by more supportive financial conditions, particularly

10 The MPC's policy package in August 2016 - which consisted of a cut in Bank Rate, the introduction of the Term Funding Scheme to ensure that that cut was passed through, and an additional £60bn of asset purchases including corporate bonds - will have provided stimulus, as will the mild loosening in the fiscal stance in the subsequent Budget. In addition, the FPC amended the leverage ratio framework for UK banks by excluding central bank reserves, cut the counter-cyclical capital buffer to zero, and emphasised that banks' liquidity reserves are usable, while the PRA Board decided to use regulatory flexibilities to smooth insurers' transition to new regulatory standards in a very low interest rate environment.

intense competition in mortgage markets, which will have helped offset some of the effects on consumption of the fall in income. 11

Finally, the third channel through which Brexit is affecting the UK economy is through its effect on the exchange rate. I have already discussed its effects of the recent depreciation on demand, where it has contributed to a rotation away from consumption and towards net trade. Equally important for the MPC has been its effect on inflation.

I hardly need to remind an inflation conference that CPI inflation has overshot the MPC's 2% target for over a year now, peaking at 3.1% last autumn. That overshoot has almost entirely reflected the effects of the depreciation. UK non-fuel import prices have risen 7% since the referendum, a direct result of the lower exchange rate. If anything that is a little less than we had been expecting based on past experience, and we revised down our inflation forecast slightly in the last Inflation Report to reflect that. But it has nevertheless been enough to generate a significant overshoot.

Why did the MPC not act to prevent that overshoot? After all, the Committee knew it was coming.

One reason is that the lags from raising Bank Rate and lowering inflation are long enough that we would have been unlikely to be able to offset the full effect of import prices. But a more substantive reason is that we were faced with a trade-off between growth and inflation.

The MPC's remit gives us two very clear objectives. The first objective, is 'to maintain price stability', defined as an inflation target of 2%. Importantly that target is symmetric: there is no bias towards keeping inflation above or below the target. The second objective is 'to support' - subject to the first - 'the economic policy of Her Majesty's Government, including its objectives for growth and employment'.

Economic situations like the one the UK has been in create a conflict between these two objectives. Monetary policy can reduce inflation, but only at the cost of reducing output and increasing unemployment. If inflation is above target and output is below potential (or vice versa), there is a fundamental trade-off between achieving the two objectives. Since 2013 the MPC's remit has required us to manage significant trade-offs when exceptional circumstances cause them to arise. And that has been the basis for the Committee's assessment of the UK situation since the referendum. 12

To explain how our position on this has evolved, let me turn to the second unusual aspect of the economy that I want to focus on this afternoon: the UK labour market.

11 For more details see Box 5, 'How has the economy evolved relative to the February 2017 Report', in the Prospects for Inflation section of the May 2018 Inflation Report.

12 For a detailed explanation see: Carney, M (2017), 'Lambda', speech at the London School of Economics, 16 January 2017.

## The labour market

I said at the start of this speech that wage growth has been surprisingly weak. That wasn't a figure of speech. This chart ( Chart 8 ) shows the MPC's wage growth forecasts from the past few years against the actual data. As you can see, the MPC have been consistently surprised by the continued weakness of wage growth.

Chart 8: MPC wage forecasts

Why have wages been so persistently weak?

One reason is weak productivity growth. How much companies are prepared to pay their workers depends ultimately on how much those workers are able to produce. Weaker productivity has meant weaker wage growth.

But wages were weak even relative to productivity. Statistical models that take account of the effects of slack and productivity struggled to explain the full extent of the weakness in wages. 'Missing wage inflation' and the question of whether the Phillips curve was dead became a hot topic among economists. As with the productivity puzzle itself, there are a wide range of candidate explanations for this 'wage puzzle'. I won't go through the full list now. But I do just want to mention three that I find relevant.

One candidate explanation is that there may be lags between unemployment falling and wage growth starting to pick up. That might be the case for instance if wage expectations take time to adjust to a tighter labour market - any lasting effect of the post-crisis downward nominal wage rigidity should have worn off, but could re-emerge if 2% wage inflation has become accepted as more typical following the financial crisis. Or it

might be that pay initially only rises for workers moving jobs, and that pay for workers staying in the same job doesn't pick up until the stock of unemployment workers is used up. That would be consistent the fact that pay growth for 'switchers' has recovered by much more than pay growth for 'stayers' ( Chart 9 ). In the more recent period it might be that this is another effect of Brexit: as I argued last autumn, workers might have responded to uncertainties about the outlook by showing even more flexibility in their wage demands, or employers might be acting in a more precautionary way.

Chart 9: Pay growth for 'switchers' and 'stayers'

This chart contains statistical data from ONS which is Crown Copyright. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research datasets which may not exactly reproduce National Statistics aggregates.

A second explanation would be if technological or other developments have reduced individual workers' bargaining power. This would be a continuation of a long run trend rather than a development specific to the financial crisis. It would be consistent with those studies which find a longer-term flattening of the Phillips Curve that predates the crisis. But it might have intensified with the 'Uberisation' of, and other influences on, the UK labour market, which may have left workers more atomised. Greater contestability across international labour markets and the greater integration of the global labour pool would also fall into this category. 13

Finally, a third explanation would be that headline unemployment might be understating the amount of slack left in the labour market. At the same time as actual unemployment has been falling, there have been a number of structural changes in the labour market that are likely to have shifted the equilibrium

13 For more detail, see: Carney, M (2017), '[De]Globalisation and inflation', 2017 IMF Michel Camdessus Central Banking Lecture, 18 September 2017.

unemployment rate. For example, higher educational attainment, increased use of online job advertising, and increased flexibility in the labour market are likely to have increased matching efficiency and lowered job destruction rates. That would have been consistent with the fact that broader measures of domestically generated inflation also remained below levels that would be consistent with inflation at target in the medium term.

The idea that there was more slack than headline measures of unemployment were suggesting was one that I put particular weight on when I started on the MPC last autumn - in other words that there was more room for the economy to grow without generating inflation above target in the medium term. That was one reason why I voted (in the minority) for no change at the November policy meeting. 14 During the MPC's annual supply stocktake in February we made a judgement to revise our estimate of the economy's equilibrium unemployment rate from 4½% to 4¼%: in other words, a judgement that there was indeed some more slack in the labour market. That left me more comfortable with the balance of risks around the MPC's central assumptions than I had been previously.

At the same time, by the February Inflation Report, annual wage growth in the data had picked up materially relative to what we had been expecting at the November meeting. In November the most recent observation for whole economy regular pay had been of just 2.1% growth (in August). And we were forecasting it to only pick up very slightly over the next six months. But by the February Report it had picked up to 2.4% growth (in November) and we were forecasting a further pick-up to 2.8% over the next few months. That was corroborated by our Agents' pay survey which was pointing to a ½ percentage point increase in settlements this year, from around 2½% to around 3%, with the ability to recruit and retain staff the number one factor driving growth.

Since February the picture has remained the same. Wages did indeed pick up to 2.8% by the May Inflation report. And the very latest observation, of 2.9% regular pay growth in March, was slightly higher than our expectation even in May. Wage growth has now been rising steadily over the past six months, and unit labour costs have been rising towards growth rates consistent with overall inflation at target. The period of unusually subdued growth in wages appears to be coming to an end.

## The year so far

That brings me to the third unusual situation that that we have had to contend with - developments in the economy so far this year, and how the MPC have responded to them - most recently at our May policy meeting. At the time of the February report we judged that the trade-off between managing slack and controlling inflation was much diminished. We went so far as to say that 'were the economy to evolve broadly in line with the February Inflation Report projections, monetary policy would need to be tightened somewhat

14 See: Ramsden, D (2017), 'Monetary Policy from End to End: Define, Decide, Deliver', speech at King's College London, 20 November 2017.

earlier and by a somewhat greater extent over the forecast period than anticipated at the time of the November Report'. Given that, why didn't we vote in May to raise Bank Rate?

The key thing to note here is the dependence on the data. Did the economy evolve broadly in line with our February projections?

The simple answer to that question is no. Instead from early April we were faced with a string of much less encouraging data for the early part of the year. The March PMIs fell sharply and only partially reversed in April. Inflation surprised on the downside. GDP growth in Q1 was 0.1%. Consumer confidence fell. And consumer credit growth slowed sharply in March.

Of course as always with data, there were more layers to the story. Most importantly, our staff analysis suggested the low GDP growth figure was likely to be heavily influenced by the bad weather we had in March. That was partly because the weather is likely to have had a dampening effect on activity - and the pattern of activity was consistent with this. And partly because GDP is typically revised up more than usual after periods of heavy snow: this chart ( Chart 10 ) illustrates the size of revisions in previous periods of snow. From an economic perspective it was also hard to find other factors that could have contributed to such a sharp slowdown: the world economy had continued to grow, and financial conditions had remained accommodative.

Chart 10: Snow effects on GDP revisions

At the same time, the labour market continued to perform robustly. Employment continued to rise, which given its correlation with GDP, which if anything has become more contemporaneous in recent years - would

suggest all else equal that GDP was also growing. Underlying regular pay growth continued to pick up in line with our forecasts.

Given all this, it seemed to me and to the MPC that the most likely explanation for the apparent weak data in the run up to our May meeting was that it was a temporary, weather-related softening. And we continued to see the trade-off between slack and inflation as much diminished. In fact last month as part of our continuing efforts to make our thinking more transparent, we published our estimate of excess supply (or slack) in the economy for the first time. If you look it up in the Report you will see that according to our estimate only a very small amount of excess supply - ¼% of GDP - remained in the economy.

But we couldn't be certain of this. Even in the most benign interpretation the economy had gone through a temporary soft patch. The majority of indicators were pointing in one direction. And it felt like there was at least a risk that, perhaps reflecting the unusual position of the UK, something more material was going on perhaps one of the risks I highlighted earlier was crystallising. That was why I, along with the majority of the Committee, decided to vote again to leave Bank Rate unchanged.

How do I see the economy evolving from here?

It is still early days. We are still only two thirds of the way through Q2, less far through the Q2 data cycle, and only a month has passed since our last MPC meeting. Even so, the data we have had so far suggests our interpretation of the slowdown in Q1 as temporary looks to be being borne out. Consumer confidence and consumer credit both picked up in the latest data, as did retail sales and several business surveys. That included the latest services PMI output balance, representing 80% of the economy. So far at least our May judgement looks on track.

Looking ahead, my central expectation for the economy is in line with the MPC's best collective judgement as expressed in our inflation report forecasts. Global growth still looks solid, albeit a bit less rosy than it did before. The labour market is still robust. I expect GDP growth to resume at a steady but unspectacular pace ( Chart 11 ), and demand to continue to rotate away from consumption and towards trade and investment. Of course all this is conditional on a smooth transition to the eventual post-Brexit arrangements.

Chart 11: May IR central GDP forecast

Given our forecast for productivity growth, however, even this subdued pace of growth is enough to exceed the new, lower economic speed limit, which we estimate to be around 1½%. Unemployment falls to 4%, below our estimate of equilibrium, and a small margin of excess demand opens up. And so we expect wage growth and domestic inflationary pressures to continue the pickup we saw at the end of last year ( Chart 12 ): wage growth eventually picks up to 3½% - which given our productivity growth forecast implies unit wage cost growth of 2¼%.

Chart 12: Measures of domestically generated inflation

government

This is an important thing to bear in mind when looking at the current wage data. Before the financial crisis we were invariably in a 'two plus two is four' world, with two per cent productivity growth and two per cent

inflation yielding four per cent wage growth at least. Today we seem to be in more of a 'one plus two is three' world: given one per cent productivity growth, wage growth only needs to reach three per cent, not four, to be consistent with the two per cent target. In our central forecast wage growth achieves that. And so although imported inflation falls back as the upward pressure from sterling's depreciation comes to an end, domestic cost pressures keep inflation close to target in the medium term.

Relative to this central forecast I can see risks in both directions. I have already mentioned the possibility that consumption growth might take a further leg down as households continue to adjust to lower real incomes. In that case GDP growth would be lower, spare capacity could open up and inflation could drop below target. But there are also risks from the supply side. On the one hand, productivity growth has serially disappointed and could continue to do so in future. In that case excess demand could build more quickly and inflation could stay above target. That said, I am optimistic about the potential for Fintech and other new technologies to lead to productivity gains, either directly or by driving competition, and in turn to potentially higher, and non-inflationary, wage growth. 15

One thing to emphasise about our central forecast is that it is conditional on the market yield curve. At the time of the May Inflation Report that yield curve was pricing in roughly three rate rises over the three years covered by our forecast. And those three rate rises were one of the forces preventing excess demand rising faster in the forecast and inflation from remaining persistently above target.

To demonstrate what would happen without those rate rises, we also publish a version of the forecast in the Inflation Report showing what would happen if Bank Rate remained unchanged ( Chart 13 ). In that version of the forecast inflation does not fall back to target. Instead excess demand and domestic cost pressures build much more rapidly. And so inflation stays persistently at around 2.4%, little changed from where it is now and materially above our inflation target.

15 I set out my views on the opportunities presented by Fintech in a speech in March - see: Ramsden, D (2018), 'The Bank of England Open to Fintech', speech at HMT's International Finance Conference, London, 22 March 2018.

Chart 13: May IR central inflation forecast

-0.5

This does not seem a desirable outcome. Inflation persistently above target and a sustained period of excess demand would represent a failure to meet our remit. This is one of the reasons why I signed up to the Committee's collective judgement that 'were the economy to develop broadly in line with the May Inflation Report projections, an ongoing tightening of monetary policy over the forecast period would be appropriate to return inflation sustainably to its target at a conventional horizon'.

Despite the backdrop of uncertainty, and despite the unusual things going on in the economy, I am actually reasonably confident about our understanding of what's going on. Forecasts are never going to be perfectly accurate. But over the last year I think we have got the broad narrative, and our understanding of the key moving parts, about right, with the slowing in GDP growth, the rotation of demand and the overshoot in inflation. 16 Students of the UK labour market - and I have been one for over thirty years - should always expect surprises. But I feel my knowledge of what's going on has developed over the last year. And, lastly, while it is still early days, our judgements on the behaviour of the economy so far in 2018, and the effects of the snow, appear to have been borne out.

Nevertheless, our view of the outlook and the prescription for monetary policy remains conditional on the data, how they evolve and what pointers they give to the future. It is likely that supply, demand and the exchange rate will continue to move around as Brexit negotiations progress and external conditions change. 17 And the data and the outlook will change to reflect all of these.

16 Again, for more details see Box 5, 'How has the economy evolved relative to the February 2017 Report', in the Prospects for Inflation section of the May 2018 Inflation Report. In fact inflation overshot by a little more than we had expected, partly reflecting the rise in oil prices over the previous year, demand rotated by a greater degree than we had expected, and productivity growth slowed by more than we had expected. And, as discussed above, demand fell back by less than we had initially predicted in the August 2016 Inflation Report. But the broad shape of the economic response to Brexit was in line with what we had forecast in February 2017.

Whether in speeches such as this or in our collective communications we will continue to be as open and transparent as possible about our interpretation of the data and its implications for policy. And we will continue to respond as events evolve to bring inflation sustainably back to target, in line with our remit.