Introduction

UK economic growth has weakened since the EU referendum in June 2016, in contrast to a pick-up in global growth. UK economic prospects hinge on the future relationship with the EU and, as before, our forecast assumes a soft Brexit where the UK maintains a high level of market access to the EU. Under that scenario, GDP is set to rise by 1.4 per cent this year and 1.7 per cent next year (figure 1). Our GDP growth forecast for 2018 has been revised down from 1.9 per cent partly because of the disruption caused by adverse weather conditions in the UK in late February and early March, which resulted in a sharp slowdown in first-quarter real GDP growth to just 0.1 per cent according to preliminary official data. The outturn was broadly in line with our monthly GDP estimate for Q1, published on 11 April, of 0.2 per cent.

Figure 1. Real GDP growth (per cent per quarter)

Since our previous forecast in February, the UK government and the EU have made further progress by agreeing to a draft legal text of the Phase 1 agreement and alongside that a 21-month transition period after the UK exits the EU in March 2019 (see Box A). During this period the UK will have to comply with regulations of the single market and customs union, including free movement of labour, and also to maintain the budgetary contributions, but the UK is free to negotiate new trade deals. The agreement also includes a provision that helps prevent a hard border between the Republic of Ireland and Northern Ireland.

Although these developments represent a significant step forward, it is worth reminding ourselves that negotiations, that will help define the Withdrawal Agreement and are centred around a new trading relationship, are yet to begin and that nothing, including progress with Phase 1 and the transition period, is agreed until everything is agreed.

The UK government is looking to negotiate a new trading relationship outside the customs union and the single market that is likely to be more limited in scope. It is hard to know at this stage if parliament will accept a more restricted trading arrangement should that emerge. After all, the House of Lords voted in favour of a customs union^{Footnote 1} earlier this month and, what is more, the opposition Labour party also confirmed its preference for remaining in the customs union.^{Footnote 2}

Put differently, the soft Brexit that underpins our central forecast is not an outcome that we believe will necessarily materialise, it is simply the one to which we assign a higher probability. As such the risk to our GDP growth forecast is skewed to the downside and the risk to inflation to the upside (figures 2 and 3) to represent the risk of a collapse in negotiations and trade that is based on WTO rules.

Figure 2. GDP growth fan chart (per cent per annum)

Figure 3. CPI inflation fan chart (per cent per annum)

Important risks to our forecast are also coming from our supply-side judgements on productivity growth and the overall ‘speed limit’ of the economy. UK productivity has been particularly weak and disappointing since the financial crisis. So much so that the productivity gap relative to a pre-crisis trend is twice as big in the UK (18 per cent) as for the G7 (ex.UK) average (9 per cent).

More recently though, there have been some tentative signs of a recovery in hourly labour productivity in the second half of last year (see Supply Conditions section in this chapter). We have treated this recovery with caution for this forecast round and held our lower medium-term productivity forecast unchanged, but there are some good reasons to be optimistic. To start with, higher minimum wages and the very rapid increase in the living wage, together with lower levels of net migration that create labour scarcity, may well trigger business investment that can help raise labour productivity or raise total factor productivity (Reference Riley and BondibeneRiley and Bondibene, 2017).³ The outlook for wages in the UK and in other G7 economies depends most crucially on labour productivity (Box B).

The Chancellor did not announce any new policies at the 2018 Spring Statement. The Office for Budget Responsibility (OBR) published an update of its fiscal and economic forecasts and the main message for fiscal policy remained unchanged – the government is on track to meet its fiscal targets mainly by spending less each year as a fraction of GDP.

Box B. Decomposing wages: the productivity gap persists

Ten years from the crisis, average real wages are growing considerably below the pre-crisis average across the OECD and in the UK in particular. We estimate that UK real wages are around 20 per cent below a continuation of their pre-crisis trend, similar to the shortfall in productivity. This underperformance in real wages is even more striking against a backdrop of record high employment levels and a record low unemployment rate.

The gap has stirred a debate in the UK and elsewhere on the relevance of the Phillips curve, which illustrates the relationship between wages and unemployment. In this box we estimate a wage equation for the G7 economies. We show that a wage curve that is augmented with a measure of productivity helps explain the weakness in wage growth. Productivity grew by an annual average of 1.5 per cent in the five years before the crisis across the G7; the average of the past five years is half of that.^{Footnote 1} This entails a shift in the Phillips Curve – for an unchanged level of the unemployment rate we expect a lower wage growth. Our regression estimates suggest that other factors that contribute to wage developments in the G7 are labour market slack and inflation expectations. Taking all these competing explanations together, we show that productivity is by far the most important contributor to low wage growth in the G7 and also for the UK.

Data and model description

We use annual data for G7 countries between 1990 and 2016, collected from the OECD (www.data.oecd.org). As the dependent variable, we use the annual growth rate in nominal labour compensation per hour worked in local currency, which includes gross wages and salaries as well as employers' social security contributions. For regressors, we use the deviation of the unemployment rate from a five-year moving average. The inflation measure is the year-on-year growth rate of the harmonised index of consumer prices. The augmented model includes a productivity measure which is expressed as output per hour worked. Another model specification includes a labour market slack indicator, such as the share of involuntary part-time workers out of total employees. We perform the analysis at the aggregate level instead of differentiating by sectors because of the concern that people frequently move across sectors, preventing a clear separation between sectors.

The baseline model is as below:

(1)${\pi }_{i,t}^w={\alpha }_i+{\beta }_1{\pi }_{i,t-1}^w+{\beta }_2{\pi }_{i,t-1}+{\beta }_3\left(U_{i,t}-{\overline{U}}_{i,t}\right)+{\epsilon }_{it}$

where i is country and t is time, π^w is year-on-year growth in compensation per hour (wage inflation)^{Footnote 2}, π is price inflation, and U is the unemployment rate and $\overline{U}$ is a five-year moving average rate.

Our preferred specification augments the equation above with productivity growth (prod):

(2)${\pi }_{i,t}^w={\alpha }_i+{\beta }_1{\pi }_{i,t-1}^w+{\beta }_2{\pi }_{i,t-1}+{\beta }_3\left(U_{i,t}-{\overline{U}}_{i,t}\right)+{\beta }_{4}pro{d}_{i,t}+{\epsilon }_{it}$

Our results are shown in table B1. Consistent with IMF (2017) and Reference Anderton, Hantzsche, Savsek and TothAnderton et al. (2017), we find that an increase in unemployment by 1 percentage point leads to a decrease in wage growth by around 0.4 percentage points ceteris paribus. An equivalent increase in productivity growth leads to a rise in wage inflation of around 1 according to our key model specification. Using the share of involuntary workers for total workers in place of the unemployment rate^{Footnote 3} yields a smaller decrease in wage inflation compared to the unemployment rate. Finally, wages are relatively flexible and do not show too much persistence as captured by the coefficient on the lag of wages.

UK wage growth story

Using the results from the G7 panel regression (model specification II), Figure B2 shows the percentage point contributions of each of the competing factors to UK wage dynamics. Productivity made a large positive contribution to wage growth up until 2007 while it has been almost nil ever since. The recent uptick in wages in the second half of last year came hand in hand with productivity growth reaching its highest reading in six years. Unemployment dragged wage growth lower from 2008 until 2010 and added to wage inflation since 2014. The negative residuals in the years from 2010 are likely due to the cap on public sector wages introduced that year. Recent NIESR research has estimated that public sector wages had fallen more than 3 per cent below their equilibrium level in 2016–17 (Reference Dolton, Hantzsche and KaraDolton et al, 2018). The period from 2017 onwards is based on our newly published forecasts until 2022. Accordingly, we expect a gentle rise in nominal wages, in line with a soft recovery in productivity, while the unemployment rate stabilises at a slightly higher level than where it currently stands.

Figure B1. UK real wage growth, productivity growth and unemployment rate

Figure B2. Contributions to UK wage growth (percentage points)

Notes

1 For a detailed analysis of the underlying causes of the productivity puzzle see e.g. Reference Kazalova and NaisbittKazalova and Naisbitt (2018), Reference Chadha, Kara and LabonneChadha et al. (2017).

2 We use nominal wage growth instead of real in line with other studies such as IMF (2018), Reference Mojon and RagotMojon and Ragot (2018) and Reference Bell and BlanchflowerBell and Blanchflower (2018). We use lagged inflation as a regressor as a proxy for inflation expectations, in line with adaptive expectations.

3 We also include the unemployment rate together with our proxy for underemployment and the results are not significantly different.

This box was prepared by Amit Kara, Marta Lopresto and Rebecca Piggott.

In an important departure from the standard NIESR forecasting process, we have deviated from official spending plans and assumed as our central case that the government will relax the fiscal straightjacket somewhat later this year and in the 2019 Spending Review. The government's quest to lower the total managed expenditure-to-GDP ratio to a level below its long-run average is unsustainable in our view. We have, therefore, built into this forecast a higher path for government consumption that includes higher public sector wage growth (see Public Finance section and Box C in this chapter and the Commentary earlier in this Review).

Inflation is set to continue falling from an average of 2.7 per cent in 2017 to 2.4 per cent this year, before settling close to the target rate of 2 per cent in 2019 and beyond. As discussed above, there are countervailing risks to that central forecast from the two major risks identified in this forecast, namely Brexit and productivity. Taking both together, we judge the overall risk to inflation to be skewed to the upside.

Figures 2 and 3 are generated from our structural model, NiGEM, that allows the forecaster to apply judgement and convey a complete narrative. The fan is achieved by bootstrapping historical forecast errors around the forward path of the variables. These forecasts can be benchmarked against those published by the Warwick Business School Forecasting System, which combines state-of-the-art statistical models weighted solely by the forecasting performance of each model (Box D). On their forecasts, real GDP growth for the final quarter of 2018 is most likely to be somewhere between 1–2 per cent (NIESR = 1.4 per cent) but their forecast is skewed to the upside, while ours points to a downside risk mainly to allow for the possibility of a hard Brexit. The growth forecast for the final quarter of 2019 is expected to be between 2–3 per cent compared with the NIESR forecast of 2 per cent.

Similarly, the WBSFS model points to CPI inflation between 2–3 per cent as the most likely outcome for both the final quarters of 2018 and 2019 compared with our forecast of 2.1 per cent and 2.0 per cent respectively.

Monetary policy

As before, we maintain our recommendation that the Bank of England keeps interest rates on a path to normalisation (Reference ChadhaChadha, 2017). We see the policy rate rising by 25bp every six months until Bank Rate reaches 2 per cent by the end of 2020. We have, however, pushed back the timing of the next rate increase by three months from May to August this year, in response to comments by the governor of the Bank of England and the soft preliminary estimate of 2018 first quarter GDP of 0.1 per cent. Inflation has also surprised to the downside in the UK.

The decision to delay the next increase is finely balanced in our view but some caution might be warranted to allow for the fog of uncertainty around first quarter activity data to lift and for confirmation that economic growth has returned to its potential. Reference Barwell and ChadhaBarwell and Chadha (2013) argue that the Bank of England must provide explicit guidance on the take-off path in much the same way as the Federal Reserve. That will help reduce money market volatility in the face of timing or tactics and at the same time allow the markets to trade its own information against the central bank path.

Market expectations of Bank Rate have risen since our forecast in February (figure 4). Although market expectations have converged towards our view, our conditioning path remains more aggressive than market expectations. This is for three reasons.

Figure 4. Market-implied paths for short-term interest rates and NIESR forecast

First, our central forecast assumes a soft Brexit scenario where the UK maintains a high level of market access for both goods and services to the EU. Since our February forecast, talks with the EU have progressed and the government has reached a transition agreement but, of course, nothing is certain until the final agreement. The market view, by contrast, incorporates a spectrum of possibilities ranging from a soft to a hard Brexit.

Second, we assume that the recent downturn in hard data and leading indicators is temporary and largely attributable to adverse weather conditions in the UK and other parts of Europe in the first quarter of this year and the influenza epidemic in Northern Europe. Our forecast assumes that the UK economy recovers some of that lost output in the second quarter of this year.

Finally, in what is a major departure from standard NIESR forecasting process, we have assumed that the government will struggle to maintain the fiscal straightjacket that is embedded in the OBR's latest fiscal forecast (See Box C). Specifically, we have built into this forecast a higher path for government consumption. Crucially, from the perspective of monetary policy, some of that additional spending is to finance higher public sector wages which will raise the government consumption deflator with some spillover into the private sector (see Public Finance section below).

Box D. Forecasting with a benchmark: the Warwick Business School forecasting system

We provide benchmark forecasts to help understand and contextualise the forecasts presented in this Review. The box presents density forecasts for UK GDP annual growth and inflation, and reports the probabilities of a range of output and inflation events occurring, as calculated using the Warwick Business School Forecasting System (WBSFS).^{Footnote 1}

To reflect the uncertainties inherent in economic forecasting, and following the practice of the NIESR and other forecasters such as the Bank of England and OBR, the WBSFS provides probabilistic forecasts. The WBSFS forecasts are produced by explicitly combining density forecasts from a set of twenty four, statistically motivated, univariate and multivariate econometric models commonly used in the academic literature. The use of combination forecasts or model averaging reflects the view, supported by research (e.g., see Reference Bates and GrangerBates and Granger, 1969, Reference WallisWallis, 2011, Reference Geweke and AmisanoGeweke and Amisano, 2012, Reference Rossi, Elliott and TimmermannRossi, 2013), that because any single model may be mis-specified there may be gains from the use of combination forecasts for measuring probabilities.

Comparison of the Institute's forecasts with the probabilistic forecasts from the WBSFS may be interpreted as providing an approximate indicator of the importance of expert judgement, which may include views on the underlying structure of the macroeconomy. This is because the WBSFS forecasts are computed by exploiting regularities in past data with the aid of automated time-series models; they do not take an explicit, structural or theoretical view about how the macroeconomy works; and they do not rely on (subjective) expert judgement to the same degree as those presented by the Institute. The forecasts from the WBSFS are not altered once produced; they are deemed ‘simply’ to represent the data's view of what will happen to the macroeconomy in the future.

Figure D1 presents WBSFS's latest (as of 19 April 2018) probabilistic forecasts for real GDP growth and inflation – defined as year-on-year growth rates for 2018Q4 and 2019Q4 – as histograms. The information set used to produce these forecasts includes information on GDP growth up to 2017Q4 and the latest CPI inflation estimate for March 2018.

Figure D1. WBSFS forecast probabilities for real GDP growth and inflation, year-on-year

Table D1 extracts from these histogram forecasts the probabilities of specific output growth and inflation events. The events considered are the probability of output growth being less than 0 per cent, 1 per cent and 2 per cent, and of inflation lying outside the 1–3 per cent target range (i.e., the probability of the Bank of England's Governor having to write a letter explaining how and why inflation has breached its target range). Also reported are the individual probabilities of inflation being less than 1 per cent and greater than 3 per cent, to indicate which side of the target range is most likely to be breached.

Inspection of the forecasts for output growth for 2018Q4 in table D1 suggests that, compared with our forecasts made one quarter ago, relatively little has changed. The most likely range for the forecast remains for economic growth between 1 per cent and 2 per cent in 2018Q4. But looking out further to 2019Q4, a higher growth between 2 per cent and 3 per cent is now marginally more likely. As table D1 shows, the difference between the forecasts for 2018Q4 and 2019Q4 is explained by modest downward revisions to the risk of ‘low’ growth (growth less than 1 per cent); the probability of growth less than 1 per cent is 26 per cent for 2018Q4 and falls to 20 per cent for 2019Q4.

Similarly, for inflation, our forecasts are little changed relative to those published in this Review one quarter ago. An inflation rate between 2 per cent and 3 per cent is the most likely outcome in the year ending 2018Q4, with a 38 per cent probability (up by just 1 per cent from the previous estimate of 37 per cent). But the WBSFS predicts that inflationary pressures marginally dissipate in 2019Q4, with a probability of around 27 per cent of inflation falling in the 1–2 per cent range and 31 per cent in the 2–3 per cent range. In comparison with our previous forecasts, the probability of inflation above 3 per cent has increased from 33 per cent to 35 per cent in 2018Q4 and from 27 per cent to 30 per cent for 2019Q4.

The forecast update shifts the inflation distribution marginally to the right, but where the probability of being in the 1–3 per cent range stays the same at 59 per cent, as the downside risks to inflation have decreased by the same amount as the upside risks have increased. The implication drawn is that the lower CPI inflation estimate for March 2018 has made very little difference to our inflation forecasts.

Note

1 WBSFS forecasts for UK output growth and inflation have been released every quarter since November 2014. Details of the releases are available at https://www2.warwick.ac.uk/fac/soc/wbs/subjects/emf/forecasting/ and a description of the models in the system and of the indicators employed is available at https://www2.warwick.ac.uk/fac/soc/wbs/subjects/emf/forecasting/summary_of_wbs_forecastng_system.pdf.

This Box was prepared by Ana Galvão, Anthony Garratt and James Mitchell.

The MPC has long stated that it will continue to reinvest the proceeds from maturing bonds bought under its Asset Purchase Facility until the policy rate reaches the threshold of 2 per cent. That guidance has not changed and so we would expect the Bank's balance sheet to shrink only from mid-2021 as bonds mature, given that on our forecast the threshold is reached at that point. We assume that the Bank will not actively sell bonds back to the market. It is worth noting that this 2 per cent guidance was announced in November 2015 when the lower bound on the policy rate was thought to be 0.5 per cent. Since then the MPC cut Bank Rate to 0.25 per cent and also expanded the QE programme. The MPC should clarify its stance on the reinvestment threshold in light of the newer lower bound.

At its March meeting, the Financial Policy Committee (FPC) held the countercyclical capital buffer (CCyb) rate unchanged at 1 per cent. The CCyb rate was raised at the November 2017 meeting from 0.5 per cent to 1 per cent effective November 2018. Brexit remains a key risk for the FPC and in its judgement the UK banking system has the resilience to continue supporting the economy even through a disorderly Brexit.

Prices

Consumer price inflation is falling faster than expected, reaching 2.5 per cent in March. This faster-than-expected drop is the latest signal that the inflationary effects of the depreciation of sterling in 2016 are waning.

Figure 3 shows recent movements in inflation as well as our forecast for the years ahead, which is conditional on our forecasted monetary policy path. In view of recent data outturns, we have revised down our forecast for CPI inflation in 2018 to 2.4 per cent, 0.3 percentage point lower than our previous forecast. Thereafter, our view is that inflation will remain close to the target from 2019 onwards.

There are a number of risks to this forecast. One upside risk is a continuation of the recent pick-up in oil prices in sterling terms from roughly £20 a barrel to £50, which is likely to add to inflationary pressure in the UK (Reference Lennard and TheodoridisLennard and Theodoridis, 2018). Another is some cost-push pressure, emerging from a number of potential sources, such as from a further reduction in underemployment, a steepening of the Phillips curve, spillovers from a forecast easing of public sector pay restraint to the private sector, or from a trickle-up effect associated with the rising National Living Wage. A final upside risk is a hard Brexit, which might involve a repeat of the depreciation of sterling and the pass-through of 2016.

According to the Bank of England's Inflation Attitudes Survey, there has been little change recently in short-run inflation expectations. At the 12- and 24-month horizons, expected inflation was 2.9 per cent, unchanged in February from the November 2017 survey, while at the 5-year horizon, expectations fell from 3.5 to 3.4 per cent. Financial markets also expect inflation of roughly 3 per cent at the 1- to 5-year horizons (Bank of England, 2018). The Bank's survey of professional forecasters, on the other hand, expects inflation of 2 per cent at the 3-year horizon, which is more consistent with the Bank of England's target and our forecast.

Following the increase in Bank Rate in November, and the expectation that the Bank of England will continue on the path to interest rate normalisation (Reference ChadhaChadha, 2017), the idiosyncrasies of British price indices will become an important issue. Historically, there has been a positive wedge of roughly 1 percentage point between RPI inflation and CPI inflation. This wedge arises for a number of reasons, such as the treatment of mortgage interest payments, other differences in the coverage of goods and services, formula effects and differences in weights. The ONS calculates that mortgage interest payments, which are included in RPI but not CPI, have accounted for a trivial fraction of the difference between the two in recent years. However, between 1998 and the financial crisis, when interest rates were more variable than since the crash, mortgage interest payments were an important driver of the wedge between the two indices. In this period, we estimate that a 1 percentage point increase in Bank Rate raised the RPI relative to the CPI by 0.5 percentage points. This is important as interest rate normalisation could once again thicken the wedge, which in turn has complex implications for the finances of households and the government. The CPI, for example, is used to index tax allowances and thresholds, benefits and public service pensions, while the RPI is used to calculate interest on index-linked gilts and student loans (OBR, 2018). On balance, an increase in RPI relative to CPI would have a net negative impact on the public finances in the short run but a more negligible effect thereafter (OBR, 2017).

Components of demand

We have revised down our GDP growth forecast for 2018 from 1.9 per cent to 1.4 per cent. One reason for this revision is the disruption caused by extreme weather conditions in the first quarter of this year. According to the ONS, the economy expanded by just 0.1 per cent over this period, which is broadly in line with our monthly nowcast and some 0.3 percentage points lower than our February forecast.

There is reason to be cautious about the preliminary GDP estimate published by the ONS, as it is based on relatively little firm information. Back in 2010, when the UK suffered from extensive flooding, the first estimate for final quarter GDP growth was −0.5 per cent. That estimate has since been revised up by +0.6 percentage point to +0.1 per cent in subsequent data vintages.

Available indicators of demand have been mixed, though the pattern partly reflects a continuation of the rebalancing of demand away from consumers since the EU referendum. Consumer confidence reported by the GfK survey has dipped, dented by rising prices, sluggish household income, and the prospect of interest rate hikes. Also, retailers have been badly affected by the March snows as reported by the CBI Distributive Trades survey, which revealed a harsh drop in the sales balance, although the BRC survey gathered later in the month was less pessimistic. Looking at the health of the housing market, the RICS survey on sales per surveyor tells a downbeat story. Sales peaked in 2014 and have been softening ever since, in line with the weakness of the construction PMI. On the other hand, business surveys, such as those from the CBI and Markit PMI, continue to paint a healthy picture of output growth, while official data on growth have been volatile, and, if anything, a little weaker, especially in the construction and services sectors.

Turning to the expenditure components, the most important change relates to the treatment of government spending in our forecast. Instead of our usual practice of adopting the OBR's forecast for government spending, this time we have built in a higher path for both nominal and real government consumption (see Box C). As a result, we see government spending adding 0.2 and 0.3 percentage points to real GDP growth over the next two years (figure 5).

Figure 5. Contributions to GDP growth

Turning next to household spending, we see a modest slowdown this year in part because of the disruption to activity in the first quarter of this year and also because of the slowdown evident in surveys such as those from the Bank of England Agents and the British Retail Consortium. Real personal disposable income (RPDI), which tends to be the main driver of consumer spending, rose by 1.5 per cent year-on-year in the final quarter of 2017, the highest rate since early 2016. It is set to rise further over the course of this year as well as next year in part because of higher wages but also because inflation falls as the impact of the post-EU referendum currency depreciation fades. Against that, we expect the saving ratio to rise, from 5.3 per cent in the final quarter of last year to an average of 5.5 per cent this year and 5.7 per cent next year, as the unemployment rate nudges higher. Some of this recovery in the saving ratio is explained by the rolling out of auto-enrolment into workplace pensions.

House prices and consumption tend to move hand-in-hand. The Halifax and Nationwide price inflation measures are trending lower, while our preferred measure of prices, published by HM Land Registry, is still pointing to firm growth in house prices at around 5 per cent per year. As before, we expect house prices to start cooling off mainly on the back of the additional stamp duty on second homes, less generous tax allowance for buy-to-let properties and the squeeze in real incomes. Also, the gradual increase in Bank Rate, that we have in our forecast, will push up mortgage interest rates.

Business investment is expected to remain subdued over the forecast horizon mainly because of Brexit-related uncertainty. Historically, the investment-to-GDP ratio in the UK has been low relative to other G7 economies. Consistent with the view that uncertainty is holding down investment, that ratio was significantly below the three major Euro Area economies in 2017 (see figure 6). The Bank of England's Agents' score of Investment Intentions picked up from the second quarter of 2017 after a steep fall post-referendum but remains modest as a result of economic uncertainty. We expect business investment to grow by around 2 per cent this year and to stabilise at a growth rate of between 2 and 3 per cent in the medium term.

Figure 6. Business investment in the UK, France, Germany and Italy

Net trade contributed 0.6 percentage points to real GDP growth last year and is set to make another positive contribution to growth this year. On average, we expect net trade to make a positive contribution of 0.2 percentage point to growth in the three-year period from 2017 through to next year. This is a significant improvement from the past as net trade has on average subtracted some 0.3 percentage points from UK GDP growth since 1997. The main driver for the positive contribution is weak domestic demand and robust export growth that has been triggered by a weaker currency and a strong recovery in the Euro Area.

Supply conditions

Employment and labour costs

The unemployment rate fell to 4.2 per cent in February, its lowest level since 1975, down from 4.3 per cent in January. The Bank of England's recruitment difficulties index, which measures the scale of general recruitment difficulties across the economy, has reached its highest level since records began in 2005 (figure 7). This tightening of the labour market does not seem to have put much upward pressure on wages until recently. This has prompted some economists (e.g. Reference Farmer and NicolòFarmer and Nicolò, 2018) to argue that the relationship between wage growth and unemployment, as summarised by the Phillips curve, has broken down. Box B in this Review presents evidence to suggest that it has not done so but that weak real wage growth has been caused by dismal productivity growth.

Figure 7. BoE agents recruitment difficulties

The Bank of England's Agents' scores for labour costs point to a moderate uptick in the three months to February compared with the same period one year earlier (figure 8), while employment intentions are indicative of modest increases in hiring. Pay settlement data from XpertHR show a 2.5 per cent mean and median annualised pay increase in the three months to March 2018. The assumption underlying our forecast is that real labour costs will continue to increase at a moderate rate throughout 2018 as firms find it increasingly difficult to recruit workers, especially with the appropriate skills.

Figure 8. BoE agents labour costs

Productivity

The dismal productivity performance of the UK economy since the crisis means that the level of real GDP in the final quarter of 2017 was approximately 18 per cent below the level it would have been if pre-crisis trends had continued. This is twice the size of the equivalent gap for the average of the rest of the G7 countries (figure 9). However, there have been some tentative signs that productivity has started to accelerate. Output per hour picked up sharply in the final quarter of 2017, expanding by 1 per cent year-on-year which followed on from strong growth in the previous quarter. However, this was largely due to a fall in average hours worked in the second half of 2017, rather than stronger output growth. If hours worked data, which are sourced from the Labour Force Survey and can be erratic, were subject to an upward revision, this pick-up in productivity could be eradicated, as was the case in 2011. The ONS (2018) defends these data by highlighting that the industries that are experiencing higher employment growth tend to be those where average hours worked are already below average for the economy as a whole. Our underlying assumptions regarding productivity are unchanged since our February forecast and we expect modest growth in output per hour this year and an acceleration into the medium term. Thus, in the event that the most recent average hours data are in fact accurate, this presents an upside risk to our output forecast.

Figure 9. International productivity comparisons

The National Living Wage (NLW), which applies to those aged 25 and over, and the National Minimum Wage (NMW) for those aged 21–24 both rose by 33p in April, reaching £7.83 and £7.38 per hour respectively. The NLW must rise to 60 per cent of median earnings in 2020 and for that to be achieved the Low Pay Commission projects the NLW to rise by around 10 per cent over the next two years to £8.62 per hour. Reference Riley and BondibeneRiley and Bondibene (2017) exploit the introduction of the NMW to Britain and subsequent increases to identify the effects of minimum wages on productivity. They find evidence to suggest that labour productivity rose in response to these changes, and this was associated with increases in total factor productivity rather than a reduction in employment or capital–labour substitution. If these results hold for the NLW, which is broader in scope, we could see a large positive effect on productivity. Reference Forth and Rincon-AznarForth and Rincon-Aznar (2018) show that raising productivity in the UK's low-wage, high-employment sectors, such as retail and hospitality, to levels seen in other advanced economies could eliminate around 20 per cent of the total productivity difference relative to countries like the Netherlands, Germany and France.

In our November 2017 Review, we highlighted that one factor contributing to the observed productivity puzzle could be that, in a weak wage growth environment, firms have substituted labour for capital and failed to undertake total factor productivity improving measures such as training and organisational change. This has held down productivity and thus wage growth in a self-reinforcing cycle. Reference Chadha, Kara and LabonneChadha et al. (2017) also conclude that the financial crisis has led to a change in the mix of capital and labour employed and a sharp decline in total factor productivity. We expect that as labour costs rise due to increasing scarcity of workers, firms will be forced to increase labour productivity.

For further discussion of the productivity puzzle, see Reference LazarowiczLazarowicz (2018) who summarises the work presented at NIESR's 80th anniversary special session at the recent Royal Economic Society Conference. In addition, a recent virtual special issue of the National Institute Economic Review collects some of the best Review articles on the matter.

Public finances

Our fiscal forecast

Pressure to end fiscal austerity is high. We expect that the government will find it hard to resist lifting public sector wages to tackle elevated recruitment difficulties (see also Reference Dolton, Hantzsche and KaraDolton et al., 2018). Extra funding would be needed to stop the fall in public service quality and to respond to population ageing, which will progressively lead to higher demand for public services (see also this Review's Commentary). We therefore deviate from our standard practice of taking official spending plans as given and assume a pick-up in government spending from 2019. In particular, we forecast an annual rate of growth of nominal government consumption some 2 percentage points higher than in the OBR's forecast over the next five years. Figure 10 illustrates that this is due to a larger increase in real spending but also because we expect the cost of public services to rise, in particular public sector pay. As a result, we forecast total managed expenditure to remain stable as a share of GDP over the forecast horizon, just below its postwar average of 39.3 per cent. Given that we assume no adjustments to the government's taxation plans and no substantial productivity increases in the near term, our fiscal forecast implies that the public sector deficit remains just above 2 per cent of GDP. As a consequence, public sector net debt continues to stay above 85 per cent of GDP (figure 11). This would lead to a breach of the government's current fiscal rules, something that previous governments avoided by changing the rules.

Figure 10. Government consumption, annual growth

Figure 11. Public finances forecast

Fiscal risks

The risks around our fiscal forecast are balanced. If the Chancellor were to resist spending pressures, we would have to revise downwards our projections for public spending, deficit and debt. A hard Brexit continues to be a substantial negative risk to public finances. This is because recovering some of the current contributions to the EU budget is unlikely to be enough to offset increasing needs for public sector staff and infrastructure and to respond to the possibility of an economic downturn. An additional unknown is productivity growth, which has very recently surprised to the upside. In this forecast, we assume that this pick-up in output per hour, which is driven by a fall in average hours worked, will be revised away, as was the case in 2011, and have not adjusted our forecast for productivity growth since February. In the past, downward revisions to the OBR's productivity growth forecast were often accompanied by upward revisions to the employment growth forecast (figure 12). The opposite may hold if productivity indeed increases at the expense of employment (measured as total hours worked), leaving the effects on the public purse ambiguous: higher productivity could raise tax revenue but this may be offset by more welfare spending as a result of lower employment.

Figure 12. OBR productivity and employment growth surprises

Saving and investment

Sectoral balance: triple deficit

Table A9 shows the financial position of the private and public sectors of the economy and the resulting balance with the rest of the world. The private sector is further split into a household and a corporate sector. If investment is greater than saving for a sector, then this sector is a net borrower. The aggregation of these three sectors is the current account balance, which, if in deficit, implies that borrowing from the rest of the world is required in order to fund domestic investment plans. It is not possible to infer the optimality of the levels of capital from the current account but rather just the immediate financing needs of the economy. In 2017, all three domestic sectors of the economy – households, companies and government – were in deficit for the first time since at least 1987 (figure 13), and we forecast this pattern to carry on into the medium term.

Figure 13. UK sectoral financial balances

In the fourth quarter of 2017, household saving remained low, as we had forecast in the last Review: saving represented only 3.7 per cent of GDP, about the same level as the average over the whole year. Thus 2017 marked the year of lowest household saving to GDP since 1971. We expect households to continue to favour consuming over saving in 2018 and then gradually to increase saving as a proportion of GDP towards 5 per cent in the medium term as real personal disposable income recovers. An important driver of the increase in saving will be the ongoing auto-enrolment into workplace pensions. Household investment rose steadily from a trough of 2.9 per cent of GDP in 2009 to 4½ per cent of GDP in 2017, reaching the same level as the pre-crisis high of 2006–7. With demand for housing still growing strongly, we project household investment to increase in each subsequent year and to reach 5 per cent of GDP in 2022. The saving and investment positions of the household sector imply that in 2017 households required 0.8 per cent of GDP in funding from the rest of the economy. This represents the first time since at least 1987 that the household sector was a net borrower. With both household saving and investment growing slowly, we expect households to return to a near-balanced net position in the medium term.

On the corporate side, saving rebounded strongly in 2017, rising to 8½ per cent of GDP from 6½ per cent in 2016. This was driven by an increase in corporate profits both in manufacturing and in the oil and natural gas exploration and extraction sector. We forecast corporate saving to GDP to remain close to 9 per cent in the medium term as the headwinds from Brexit are balanced by stronger international growth. Similarly, corporate investment surprised on the upside reaching 10.3 per cent of GDP in the fourth quarter of 2017, up from a revised 10.1 per cent in the third quarter. Conditional on a ‘soft Brexit’ assumption, we forecast corporate investment to remain at about 10½ per cent of GDP in the medium term, and therefore the corporate sector still to require about 1½ to 2 per cent of GDP of net financing from the rest of the economy over the same time horizon.

Government sector dis-saving, which reached a peak in 2009 of around 5½ per cent of GDP, has vanished in 2017 as a result of the ongoing fiscal consolidation. We have revised our forecast for government saving down; we now expect saving to be about 1 per cent of GDP from 2018 to 2022, as opposed to increasing to 3 per cent over the same period in our February forecast. This represents a deviation from the OBR forecast and reflects our belief that the government will slow the pace of fiscal consolidation from 2019 onwards. Government investment stayed stable in 2017 at 2½ per cent of GDP and we have maintained our estimate of 2% to 3 per cent of GDP over the forecast horizon. As a result of reduced saving and constant investment, we now expect the government to remain in a net borrowing position of 1% to 2 per cent of GDP until 2022, after having borrowed 1.7 per cent in 2017. This is in stark contrast to our previous forecast that assumed a return to balance in 2022.

To finance the triple deficit of the household, corporate and government sectors, the domestic economy had to borrow 4.1 per cent of GDP in 2017 from the rest of the world. This was half a per cent less than previously expected thanks to the vitality of the corporate sector, and the lowest ratio since 2011. With stronger sterling, lower inflation and higher government spending, we now expect the rebalancing of the economy from domestic demand to net trade to grind to a halt. Hence the current account balance is forecast to decline only marginally to 3½ per cent of GDP in 2022, compared to 2½ per cent in our previous forecast.