Understanding Brexit-Related Uncertainties Exploration of the - - PowerPoint PPT Presentation

Understanding Brexit-Related Uncertainties Exploration of the Decision Maker Panel Survey

Nick Bloom (Stanford University), Phil Bunn (Bank of England), Scarlet Chen (Stanford University), Paul Mizen (University of Nottingham), Pawel Smietanka (Bank of England), Greg Thwaites (LSE Centre for Macroeconomics), Garry Young (National Institute of Economic and Social Research) ifo, Munich, December 2018 Disclaimer: Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent the views

• f the Bank of England or its Committees.

22

EU-UK relations

• 1969 – 3rd and successful application for membership in the EC
• 1973 – Entry to the EC
• 1974 – Harold Wilson’s (Labour) commitment to renegotiate Britain's terms
• f membership of the EC
• 1975 – National referendum on whether the UK should remain in the

European Communities (67.5% voted to stay, 37.5% voted to leave)

• 2013 – David Cameron’s (Conservative) promise to hold an EU referendum
• 2016 – National referendum on whether UK should remain a member of the

EU (48.1% voted to remain, 51.9% voted to leave)

• 2016 – Cameron’s resignation as PM, succeeded by Theresa May
• 2017 – Invocation of Article 50 to leave by March 2019

What next?

33

Probability of UK leaving the EU was low ahead of the referendum

44

Uncertainty indicators provided conflicting messages since the referendum

• 3
• 2
• 1

1 2 3 4 5 6 7 1997 2001 2005 2009 2013 2017

Standard deviations from average since 1997 Stock market volatility EU Referendum Policy uncertainty index Macro uncertainty Bank principal component index

55

Decision Maker Panel – a new survey of UK-based companies – allows the assessment of the impact of Brexit

• Decision Maker Panel was launched in August 2016 by the Bank of

England, Stanford University and the University of Nottingham.

• Used an approach pioneered by the Atlanta Fed (Altig, Barrero, Bloom,

Davis, Meyer and Parker, 2018)

• In UK, randomly contacted population of 31K UK firms with 10+

employees inviting them to join the monthly Decision Maker Panel

• As of October, around 6K have been part of the panel, providing a large

sample of timely firm data.

66

Key messages

• Brexit has been seen by most firms as large second moment (uncertainty)

shock.

• Firms with greater exposure to the EU, e.g. through exports, imports, and

more EU workers are more heavily affected.

• Uncertainties around Brexit are primarily about the impact on businesses
• ver the longer term rather than shorter term.
• Brexit-related uncertainty associated with around 1.5% lower employment

and 6% less investment

• Misallocation could reduce productivity by around 0.5% (likely to be negative

effect within firm effects too)

77

By October 2018 obtaining 2.6K responses per month spanning all industries and regions

500 1000 1500 2000 2500 3000 Aug-16 Sep-16 Oct-16 Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 May-17 Jun-17 Jul-17 Aug-17 Sep-17 Oct-17 Nov-17 Dec-17 Jan-18 Feb-18 Mar-18 Apr-18 May-18 Jun-18 Jul-18 Aug-18 Sep-18 Oct-18 Survey month

Number of responses

88

Respondents are spread across the UK

Notes: Data as of October 2018. The map shows location of businesses that have ever responded to the DMP survey since August 2016. The location of a business corresponds to the location of the registered office, hence it does not always match up with the actual location of the business.

99

There is not a strong Brexit-related bias in the survey

51.9 25.0 33.5 34.3 48.1 75.0 66.5 65.7 20 40 60 80 100 UK population DMP CFOs (a) BES managers as by work type (b) BES managers as by social grade (b) Percentage of voters/respondents Leave Remain

(a) Personal views of DMP members at the time of the June 2016 referendum taken from February to April 2018 surveys. The question asked respondents about whether they view Brexit in a positive or negative way rather than how they voted in the referendum. (b) To identify managers in the British Election Study by their stated work type, only participants doing professional or higher technical work/higher managerial work that required at least degree-level qualifications or who worked as manager or senior administrator/intermediate managerial/professional (company director, finance manager, etc.) were included. To identify managers in the BES by their stated social grade, only participants who identified themselves as in a higher managerial, administrative and professional or intermediate managerial, administrative and professional occupation were

• included. Only respondents working and aged 66 or lower for males or aged 60 or lower for females were included.

10 10

Sampling frame of 31K UK firms with 10+ employees: 20% responded, uncorrelated with Brexit vote share

Notes: Data as of October 2018. Two-digit UK SIC industry controls are included in all columns. Dependent variable equals 1 if a firm responded to any wave of the survey between September 2016 and October 2018 and 0 if it is part of the sampling frame but has never completed a survey. Firm characteristics are taken from Bureau van Dijk FAME data and are the latest available observations. ‘Leave vote share’ is the share of vote for leaving the EU in the local authority that a firm is headquartered in. There are 380 local authorities. Robust standard errors are given in parentheses. *** p<0.01, ** p<0.05, * p<0.1.

(1) (2) (3) (4) Leave vote share

• 0.022
• 0.026
• 0.020
• 0.018

(0.019) (0.019) (0.019) (0.019) Log of employment 0.017*** 0.011*** 0.011*** (0.002) (0.003) (0.003) Log of sales 0.007*** 0.004 (0.002) (0.003) Log of assets 0.003 (0.002) Observations 29,802 29,802 29,802 29,802 R-squared 0.010 0.013 0.014 0.014 Ever respond to a survey if in the sampling frame Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

11 11

The majority of DMP respondents are finance directors or senior managers

15 66 4 10 5 10 20 30 40 50 60 70 CEO CFO Finance Director Financial Controller/ Manager/ Executive Other Position of DMP respondents Percentage of respondents

Notes: Data as of November 2018. The question asked ‘Could you tell us the position of the person in your business that typically completes the Decision Maker Panel Survey?’

12 12

Data quality looks good – for example, comparing DMP to Company Accounts

2 4 6 8 10 DMP: log(Emp.) in 15Q4 2 4 6 8 10 BvD: log(Emp.) in 2015 6 8 10 12 14 DMP: log(Sales) in 2015 6 8 10 12 14 BvD: log(Sales) in 2015 .1 .2 .3 .4 5 10 15 ln(Num. of Emp.) DMP: ln(Emp) in 2015Q4 BvD: ln(Emp) in 2015 .1 .2 .3 .4 5 10 15 20 ln(Sales) DMP: ln(Sales) in 2015 BvD: ln(Turnover) in 2015

13 13

Data quality looks good – for example, comparing uncertainty to forecast errors

Note: Uncertainty defined as subjective uncertainty from the DMP 5-bin responses. Forecast errors defined as ABS(forecast - actual) growth over the following 12 month period. 1 2 3 Log of error in realized employment growth at t

• .5

.5 1.5 2.5 3.5 Employment 1 2 3 4 log of error in realized sales growth at t

• .5

.5 1.5 2.5 3.5 Log of uncertainty in expected sales growth at t-1 Nominal sales

• 2
• 1

1 2 Log of error in realized price growth at t

• 1.5 -1
• .5

.5 1 1.5 2 Log of uncertainty in expected price growth at t-1 Prices Log of uncertainty in expected employment growth at t-1

14 14

Data quality looks good – macro aggregates and outturns

1 1.5 2 2.5 3 3.5 2016 Q4 2017 Q2 2017 Q4 2018 Q2 2018 Q4 2019 Q2 Reference period Prices Past growth (DMP) Expected growth (DMP) CPI rate change over 12m (ONS) Percentage change

• n a year earlier

1 2 3 4 5 6 7 8 2016 Q3 2017 Q1 2017 Q3 2018 Q1 2018 Q3 2019 Q1 Reference period Sales Past growth (DMP) Expected growth (DMP) Total final expenditure (ONS) Percentage change

• n a year earlier

15 15

Hazard functions reveal no substantial difference between different cohorts

.2 .4 .6 .8 1 Probability 1 2 3 4 5 6 7 8 Quarter after joining the panel 2016Q3 2016Q4 2017Q1 2017Q2 2017Q3 2017Q4 2018Q1 2018Q2 2018Q3 Note: Based on panel members who joined between 2016Q3 and 2018Q3.

16 16

Brexit important source of uncertainty for around 50%

Note: The question asked ‘How much has the result of the EU referendum affected the level of uncertainty affecting your business?’. Respondents could select one of the options shown as response categories.

10 20 30 40 50 Not important One of many drivers One of the top 2 or 3 The largest current s Aug-Sep 16 Feb-Apr 17 Aug-Oct 17 Feb-Apr 18 Aug-Oct 18 Nov 18

17 17

In recent surveys, uncertainty was highest in wholesale & retail and manufacturing and lowest in human health & social work

Note: The question asked ‘How much has the result of the EU referendum affected the level of uncertainty affecting your business?’. Industries’ employment shares are shown in square brackets. DMP data from August to October 2018 surveys.

10 20 30 40 50 60 70 Wholesale & Retail Manufacturing

• Accom. & Food

Construction Transport & Storage

• Prof. & Scientific

Other Production Average

• Admin. & Support

Real Estate Finance & Insurance

• Info. & Com.

Other Services Percentage of firms with Brexit in top 3 uncertainty sources

18 18

Positive correlation between the share of firms in each industry viewing Brexit as an important source of uncertainty and exposure to the EU

5 10 15 10 20 30 40 50 60 70 Percentage of firms with Brexit in top 3 uncertainty sources Manufacturing Construction Percentage of workforce who are EU nationals 5 10 15 20 25 10 20 30 40 50 60 70 Percentage of firms with Brexit in top 3 uncertainty sources Percentage of output exported to EU Manufacturing Construction

19 19 Dependent variable: Brexit uncertainty (4 point scale) (1) (2) (3) (4) (5) Share of sales to EU 0.010*** 0.006** (0.002) (0.002) Share of sales to non-EU

• 0.003*
• 0.004**

(0.002) (0.002) Share of costs from EU imports 0.008*** 0.007*** (0.002) (0.002) Share of costs from non-EU-imports 0.005*** 0.004*** (0.002) (0.002) EU migrants 1-5% workforce (dummy) 0.207*** 0.178*** (0.064) (0.062) EU migrant 6-10% workforce (dummy) 0.339*** 0.291*** (0.083) (0.083) EU migrants 11-20% workforce (dummy) 0.286*** 0.243*** (0.090) (0.089) EU migrants > 20% workforce (dummy) 0.547*** 0.456*** (0.108) (0.110) Foreign owned (dummy) 0.173* 0.041 (0.092) (0.094) Industry dummies Yes Yes Yes Yes Yes Observations 1,213 1,213 1,213 1,213 1,213 R-squared 0.218 0.233 0.225 0.198 0.265 Note: Robust standard errors in parentheses. Dependent variable is defined as average uncertainty per firm in the two years after the referendum. *** p<0.01, ** p<0.05, * p<0.1.

EU exports/imports and use of migrant labour all help explain which firms are uncertain about Brexit

20 20

Differences in the importance of Brexit uncertainty by firm size and by region

20 40 60 10-49 50-99 100-249 250+ Percentage of firms with Brexit in top 3 uncertainty sources Employment 10 20 30 40 50 60 70 80 90 100 East Midlands North East South East East of England Northern Ireland West Midlands South West Wales North West Yorks and Humber London Scotland All firms Excluding large firms Percentage of firms with Brexit in top 3 uncertainty sources

Note: The question asked ‘How much has the result of the EU referendum affected the level

• f

uncertainty affecting your business?’. DMP data from August to October 2018 surveys. Note: The question asked ‘How much has the result of the EU referendum affected the level of uncertainty affecting your business?’. Large firms are defined as those with 250 or more employees. Region is based on the location of head office. DMP data from August to October 2018 surveys.

21 21

Firms’ own uncertainty lower than perceived overall economic uncertainty

10 20 30 40 50 60 Not important One of many drivers of uncertainty One of the top two

• r three drivers of

uncertainty The largest current source of uncertainty Uncertainty affecting own business Perceived uncertainty affecting other businesses

Note: The questions asked ‘How much has the result of the EU referendum affected the level of uncertainty affecting your business?’ and ‘How much do you think the result of the EU referendum is likely to have influenced the level of uncertainty affecting businesses in the UK economy other than yours?’ DMP data from November 2018 survey.

22 22

Businesses viewing Brexit as an important source of uncertainty find it more difficult to pin down the eventual impact of Brexit

2 3 4 5 6 7 Not important One of many sources One of top 2 or 3 sources Largest current source Eventual Brexit export impact (exporters

• nly)

Eventual Brexit labour costs impact Eventual Brexit sales impact Year-ahead sales growth (not just Brexit effects) Average within-firm standard deviations from expected change due to Brexit (three left) or overall (per cent)

Notes: The questions are as defined in notes to Figures 13 and 14. Graphs show mean standard deviation per firm. Brexit effects are calculated using mid-points of 0 per cent for no impact, 5 per cent for an effect of less than 10 per cent, and 15 per cent for an effect of 10 per cent or more. Eventual Brexit sales impact and year-ahead sales data are from August to October 2018 surveys, export data are from May to July 2018 surveys and labour costs data are from February to April 2017 surveys.

23 23

Businesses viewing Brexit as an important source of uncertainty also judge probability of a disorderly Brexit to be higher

10 20 30 40 50 60 Not important One of many sources One of top 2 or 3 sources Largest current source Probability of a disorderly Brexit, per cent Brexit as a source of uncertainty

Note: The questions asked ‘How much has the result of the EU referendum affected the level of uncertainty affecting your business?’ and ‘What percentage likelihood (probability) do you attach to a disorderly Brexit, whereby no deal is reached by the end of March 2019?’ DMP data from February to April 2018 surveys.

24 24

Brexit uncertainty is associated with lower firm employment…

Notes: Post Brexit data from Decision Maker Panel combined with pre-Brexit data from company accounts. All regressions include a data source dummy and are estimated from 2011 onwards (years are defined from Q3 to Q2 in next calendar year). Post Brexit defined as 2016 Q3 onwards. Standard errors are clustered by firm. *** p<0.01, ** p<0.05, * p<0.1.

Dependent variable: Annual employment growth (1) (2) (3) Uncertainty*Year 1 after referendum

• 0.732*

(0.445) Uncertainty*Year 2 after referendum

• 1.099***

(0.367) Uncertainty*Post referendum

• 0.960***

(0.340) Predicted uncertainty*Year 1 after referendum

• 0.546

(0.998) Predicted uncertainty*Year 2 after referendum

• 1.327*

(0.780) Year dummies Yes Yes Yes Firm fixed effects Yes Yes Yes Observations 12,602 12,602 12,602 R-squared 0.281 0.281 0.281

25 25

…and with less investment

Notes: Post Brexit data from Decision Maker Panel combined with pre-Brexit data from company accounts. All regressions include a data source dummy and are estimated from 2011 onwards (years are defined from Q3 to Q2 in next calendar year). Post Brexit defined as 2016 Q3 onwards. Standard errors are clustered by firm. Only firms with an investment growth rate between -100% and +100% are included. DHS growth rates are

• used. *** p<0.01, ** p<0.05, * p<0.1.

Dependent variable: Annual investment growth (1) (2) (3) Uncertainty*Year 1 after referendum

• 4.629**

(2.154) Uncertainty*Year 2 after referendum

• 0.739

(2.105) Uncertainty*Post referendum

• 2.675

(1.723) Predicted uncertainty*Year 1 after referendum

• 7.802*

(4.698) Predicted uncertainty*Year 2 after referendum 1.704 (4.719) Year dummies Yes Yes Yes Firm fixed effects Yes Yes Yes Observations 6,676 6,676 6,676 R-squared 0.237 0.236 0.236

26 26

Misallocation: More productive firms perceive a greater Brexit effect on sales

• 3.5
• 3
• 2.5
• 2
• 1.5

Expected eventual impact on sales (%) 2.5 3 3.5 4 4.5 5 Log of productivity (2013-15 avg)

27 27

As a result Brexit shrinks productive firms more

Notes: Robust standard errors in parentheses. Dependent variable is defined as self reported average eventual impact of Brexit on sales per firm in the two years after the referendum. *** p<0.01, ** p<0.05, * p<0.1. (1) (2) (3) (4) (5) (6) Log of pre-referendum productivity

• 0.553**
• 0.447**
• 0.463**
• 0.480**
• 0.523**
• 0.373*

(0.217) (0.220) (0.218) (0.211) (0.220) (0.217) Share of sales to EU

• 0.038***
• 0.027***

(0.009) (0.010) Share of sales to non-EU 0.008 0.012* (0.007) (0.007) Share of costs from EU imports

• 0.011
• 0.005

(0.007) (0.007) Share of costs from non-EU imports

• 0.016**
• 0.012*

(0.006) (0.006) EU migrants 1-5% workforce (dummy)

• 0.562*
• 0.468

(0.287) (0.287) EU migrant 6-10% workforce (dummy)

• 1.643***
• 1.476***

(0.367) (0.368) EU migrants 11-20% workforce (dummy)

• 1.582***
• 1.322***

(0.411) (0.421) EU migrants > 20% workforce (dummy)

• 1.730***
• 1.583***

(0.552) (0.550) Foreign owned (dummy)

• 0.370
• 0.104

(0.369) (0.379) Industry dummies Yes Yes Yes Yes Yes Yes Observations 1000 1000 1000 1000 1000 1000 R-squared 0.074 0.093 0.084 0.105 0.075 0.121 Dependent variable: Firms' expected eventual impact of Brexit on sales (%)

28 28

Estimate misallocation impact from Brexit at around -0.5% of TFP

Method: Calculate difference in Brexit sales effect for each firm if high productivity firms are more affected versus counterfactual where they are not. Sales weight productivity with and without this adjustment. Difference is an estimate of the misallocation effect

Winsorize at: Point estimate 1 & 99 pct

• 0.46%
• 0.11%
• 0.82%

2.5 & 97.5 pct

• 0.40%
• 0.09%
• 0.70%

Aggregate productivity effect, weighted by sales 95% Confidence Interval

29 29

Number of hours a week spent on preparing for Brexit (share) CEO CFO None 41% 38% Up to 1 hour 37% 39% 1 to 5 hours 14% 18% 6 to 10 hours 3% 3% More than 10 hours 1% 1% Don't know 4% 2%

Also likely negative within firm TFP impact - e.g. from wasted hours of senior management

Note: Growth in productivity has slowed to 0.45% a year since the referendum, compared to 0.7% between 2013 and 2015

30 30

Might also be a TFP effect if intangible investment (R&D and training) is reduced

5 10 15 20 Training of employees Software, data, IT, and website R&D Machinery, equipment, and buildings Investment type Net balance of respondents who report having reduced investment due to Brexit (per cent)

31 31

Key messages

• Brexit has been seen by most firms as large second moment (uncertainty)

shock.

• Firms with greater exposure to the EU, e.g. through exports, imports, and

more EU workers, are more heavily affected.

• Uncertainties around Brexit are primarily about the impact on businesses
• ver the longer term rather than shorter term.
• Brexit associated with around 1.5% lower employment and 6% less

investment

• Misallocation could reduce productivity by around 0.5% (likely to be negative

effect within firm effects too)

Back-Up

Firms report Brexit will cut sales, but also exports, while pushing up costs.

Notes: Self reported responses. In each case respondents were asked to assign probabilities to five different outcomes for each variables. Midpoints were then attached to each outcome to calculate mean expectations. Time horizon reported in parentheses. Data are expected percentage impacts of Brexit except for financing costs which are percentage point changes. Data are average values collected across all waves of the survey.

• 4
• 2

2 4 6 Sales (eventual) Exports (eventual) Unit costs (2020) Labour costs (2020) Financing costs (2020) Expected impact of Brexit (per cent, pp for financing costs)

Brexit Uncertainty Measure is Correlated with Stock Market Volatility

Notes: The graph plots firms’ reported uncertainty against increase in stock market volatility at industry level. It is plotted using binscatter with 50 bins. Each dot in the underlying graph is a 3-digit UKSIC industry. For the uncertainty measure in DMP, we take the average of a firm’s reported uncertainty among all waves, and then collapse to 3-digit UKSIC level by taking the mean of each industry. For the stock volatility measure, we use Compustat stock price data on all public listed firms in UK. We calculate the daily return, and then calculate the log of standard deviation in the 60 days right after Brexit and the 5 years before Brexit, and take the difference between the two (post minus pre). Then we winsorize this firm-level increase in stock volatility at 1 and 99 percentile, and collapse to 3-digit UKSIC level by taking the mean.

Increase in Brexit Uncertainty (DMP survey) Increase in Stock Market Volatility After June 22 2016

Expected impact of Brexit on sales

10 20 30 40 50 60 Large positive effect Small positive effect No impact Small negative effect Large negative effect Expected impact of Brexit on sales Aug-Oct 18 Feb-Apr 18 Aug-Oct 17 Feb-Apr 17

Results with investment level

Notes: Post Brexit data from Decision Maker Panel combined with pre-Brexit data from company accounts. All regressions include a data source dummy and are estimated from 2011 onwards (years are defined from Q3 to Q2 in next calendar year). Post Brexit defined as 2016 Q3 onwards. Standard errors are clustered by firm. Only firms with an investment growth rate between -100% and +100% are included. DHS growth rates are used. *** p<0.01, ** p<0.05, * p<0.1.

Dependent variable: Capex growth Capex growth Capex/assets(t-1) (1) (2) (3) Uncertainty*Year 1 after referendum

• 4.629**

(2.154) Uncertainty*Year 2 after referendum

• 0.739

(2.105) Uncertainty*Post referendum

• 2.675
• 2.103**

(1.723) (1.024) Year dummies Yes Yes Yes Firm fixed effects Yes Yes Yes Observations 6,676 6,676 5,819 R-squared 0.237 0.236 0.611