Uncertainty and Reallocation Nick Bloom CBO June 2020
Uncertainty and Reallocation Nick Bloom CBO, June 2020
Presentation based on two papers
Part I: Measuring Uncertainty Collect data on four uncertainty measures: financial, newspaper, subjective uncertainty & disagreement Picked because they are available: (A) almost real-time and (B) are forward looking Key findings: 1) Uncertainty rose to incredible levels – our series mostly at all time peaks during COVID pandemic 2) Financial (Wall Street) uncertainty measures falling but real (Main Street) still high → Uncertainty could slow the recovery by impeding investment, hiring and consumption rebound* * See prior evidence, for example: Bernanke 1983, Romer 1990, Caballero & Engel 1992, Dixit & Pindyck 1994, Ramey and Ramey 1995, Abel & Eberly 1996, Bertola, Guiso & Pistaferri 2005, Bloom 2009
Measure 1: VIX, Implied S&P 500 Stock Returns Volatility The “classic” uncertainty measure (e. g. I started using this in my Ph. D in the 1990 s) Main issue is stockmarket based, so long hi-tech, finance etc, and only 1 month ahead (although the 2 year VIX similar profile) Notes: Weekly implied volatility (over the next month) on the S&P 500 index from the Chicago Board of Options Exchange, expressed in annualized units. We plot data from 2 January 1990 to 27 May 2020. Values downloaded from: https: //fred. stlouisfed. org/series/VIXCLS
Measure 2: U. S. Economic Policy Uncertainty (newspapers) Relies on media so you might be worried about media bias, but highly correlated with other text measures – like the Economic Intelligence Unit quarterly reports or Twitter mentions of economic uncertainty. I think for the current crisis the EPU is a good indicator Notes: Weekly values from data from www. policyuncertainty. com/media/All_Daily_Policy_Data. csv. Based on data from around 2, 000 daily US newspapers accessed from AWN Newsbank. See Baker, Bloom and Davis (2016) for details of index construction. We plot data from 1 January 1985 to 26 May 2020.
Third measure uses survey data on firm subjective uncertainty 6 Note: For details see “Surveying Business Uncertainty” Altig, Barrero, Bloom, Davis, Meyer and Parker (2020, forthcoming Journal of Econometrics)
The large majority of respondents are CEOs and CFOs Percentage of respondents 66 70 60 50 40 30 20 15 10 5 4 10 0 CEO CFO Finance Director Financial Controller/ Manager/ Executive Position of respondents Other 7
Data looks very good – e. g. forecasts match realizations one year later 1 st Moment: Sales growth forecasts vs realizations 2 nd Moment: Sales growth forecast errors vs uncertainty 8 Note: For details see “Surveying Business Uncertainty” Altig, Barrero, Bloom, Davis, Meyer and Parker (2020, forthcoming Journal of Econometrics)
Measure 3 a: Firm Level Subjective Sales Uncertainty One year ahead sales uncertainty increased 120% in the US and 90% in the UK (noting the UK had a higher baseline already due to Brexit) Notes: Subjective uncertainty measured for the growth rate of 4 quarters ahead firm level sales expectations (details in Altig, Barrero, Bloom, Davis, Meyer and Parker 2020). US data form the Survey of Business Uncertainty conducted by the Federal Reserve Bank of Atlanta, Stanford University, and the University of Chicago Booth School of Business (https: //www. frbatlanta. org/research/surveys/business-uncertainty). UK data from the Decision Maker Panel Survey conducted by the Bank of England, Nottingham University and Stanford University (see Bloom, Bunn Chen, Mizen, Smietanka and Thwaites (2019) and www. decisionmakerpanel. com).
Measure 3 b: Another way to measure firm-level uncertainty % firms reporting Covid-19 as their top source of uncertainty Notes: Decision Maker Panel Survey conducted by the Bank of England, Nottingham University and Stanford University and Bloom, Bunn Chen, Mizen, Smietanka and Thwaites (2019) and www. decisionmakerpanel. com The survey is open for two weeks from the first Friday each month, with values reported in roughly four equal buckets (first day, and then three subsequent reminder intervals).
Weekly data from start of 2020 - see uncertainty slowly falling back Notes: Decision Maker Panel Survey conducted by the Bank of England, Nottingham University and Stanford University and Bloom, Bunn Chen, Mizen, Smietanka and Thwaites (2019) and www. decisionmakerpanel. com. Values linearly interpolated when the DMP survey was not in the field. Values of the Likert Uncertainty measure were extrapolated using information about firms’ sales expectations and uncertainty for the first five weeks. VIX 24 M, Likert Uncertainty, and Sales Subjective Uncertainty’s axes are hidden.
Part II: Reallocation - lots of discussion in the media
Using SBU find evidence for heavy job churn - for every 10 layoffs firms making 3 hires (also calculate 42% of layoffs are permanent) Source: Atlanta Fed, Chicago and Stanford Survey of business uncertainty, April 2020
As a result “excess reallocation” (a classic measure of reallocation) has tripled under COVID Source: Atlanta Fed, Chicago and Stanford Survey of business uncertainty
Finally, before COVID working from home (WFH) was pretty rare (accounting for about 5% of all working days) Source: BLS ATUS Job Flexibility Survey conducted continuously over 2017 and 2018 across all wage and salaried workers (excluded self employed) Collected around 10, 000 responses stratified across states, industries and geographies https: //www. bls. gov/news. release/flex 2. htm
Working from home (WFH) was balanced by gender and age Source: BLS data https: //www. bls. gov/news. release/flex 2. htm
But WFH was much higher for more educated higher-earners Source: BLS data https: //www. bls. gov/news. release/flex 2. htm
Under COVID there has been a massive reallocation to WFH: 56% (=39. 5/(39. 5+31. 3)) of earnings weighted activity now WFH 34 Working from home (full time) 39. 5 31. 4 Working on my business premises 31. 3 34. 6 Not working 29. 2 0 10 20 30 40 Percent of wages by May 2020 work situation Unweighted 2019 Earnings Weighted Source: Response to the question “Currently (this week) what is your work status? ” Response options were “Working on my business premises “, “Working from home”, “Still employed and paid, but not working “, “Unemployed, but expect to be recalled to my previous job “, “Unemployed, and do not expect to be recalled to my previous job “, and “Not working, and not looking for work “ Data from a survey of 2, 500 US residents aged 20 to 64, earning more than $20, 000 per year in 2019 carried out between May 21 -29, by Question. Pro on behalf of Stanford University. Sample reweighted to match current CPS and then working from home rations in the 2017/2018 American Time Use Survey. Shares shown weighted by earnings and unweighted (share of workers)
COVID WFH employees heavily drawn from offices in cities Source: Data from a survey of 2, 500 US residents aged 20 to 64, earning more than $20, 000 per year in 2019 carried out between May 21 -25 2020, by Question. Pro on behalf of Stanford University. Sample reweighted to match CPS and then current working from home rations in the 2017/2018 American Time Use Survey.
Reallocation Conclusions Key findings: 1) Very high levels of employment reallocation – we estimate about 3 x monthly values pre-COVID 2) Reallocation across other dimensions - e. g. work from office to home, spending from cities to suburbs
- Slides: 20