The British miracle: Explosive jobs growth during a recession

The endlessly perplexing British economy continues to post bizarre jobs numbers:

The employment numbers continue to be surprisingly strong, with a rise of 154,000 to 29.73 million in the October-December 2012 period, and a huge 584,000 increase over 12 months. Interestingly, the rise in employment was more than accounted for by an increase in full-time employment of 197,000 in the latest three months, with part-time employment down 43,000.

To put that in perspective for American readers, that would be the equivalent of 2.9 million jobs over 12 months in the US, far more than we’ve actually be able to generate.  (America has roughly 5 times the population of the UK.)

How did this jobs miracle happen?  Well let’s start with the fact that it may not have happened, the data might be wrong.  After all, RGDP has been flat, and Britain has been in recession during much of this time.

If it did happen, this might be one reason why:

Pay continues to be weak, up just 1.4% over the past year. More here.

If this refers to hourly nominal wages, it might help explain the jobs gains.  If it’s not hourly data, it’s meaningless.

However NGDP growth also seems to have been slow.  I say “seems” because Europe appears incapable of creating a St Louis Fred-type data set that doesn’t require a PhD in computer science to navigate.  And then there’s the question of whether the Brits even know how to calculate NGDP; last time I looked they calculated RGDP one quarter before NGDP, which is, well, mathematically impossible.

So if it’s not NGDP, what is the explanation?  Probably a combination of things.  For instance, falling North Sea oil output diverts production from areas using very few workers per dollar of NGDP, to areas using lots of workers per dollar of NGDP.

NGDP, hourly nominal wages, and employment are the key macro variables.  The goal is to stabilize employment growth (or more precisely to prevent suboptimal employment fluctuations due to sticky wages.)  Stable NGDP growth helps, but is not perfect.  In any case the UK could use a bit more monetary stimulus—it’s a pity that only three of their nine monetary policymakers understand that fact:

Also yesterday, three members of the Bank of England’s monetary policy committee – Sir Mervyn King, Paul Fisher and David Miles – voted to increase quantitative easing by £25 billion. This was a surprise. More here.

That doesn’t bode well for Mark Carney.

PS.  Keynesian models are completely silent on the question of how much real output we can expect from a given rise in employment.  That’s the supply-side of the economy, which the Keynesian multiplier model does not even attempt to explain.

Keep this in mind when you read prominent Keynesians telling us about the impact of the so-called “austerity” program adopted by the Cameron government.  They aren’t even addressing the real puzzle–why so many jobs?



33 Responses to “The British miracle: Explosive jobs growth during a recession”

  1. Gravatar of Nishant Tharani Nishant Tharani
    26. February 2013 at 07:29

    Regarding your point on monetary policy, hopefully there is some more hope – from the FT today:

    “Paul Tucker, deputy governor for financial stability, was one of the six members voting for no change. However, Mr Tucker signalled on Tuesday that he and others on the committee were open to more quantitative easing.”

  2. Gravatar of Bill Woolsey Bill Woolsey
    26. February 2013 at 08:18

    Along the lines of your oil story, what about a shift from finance to manufacturing or even other sorts of services like tourism?

    Instead of paying one guy $500 million per year to provide the service of bring together borrowers and lenders in some new and exciting way, instead, $500 million is paid for 20,000 cooks and waiters to serve German tourists in restaurants.

    Oh, better. Instead of paying $500 million to someone to help rich folks play some kind of fancy poker tied to borrowing and lending, $500 million is paid for 20,000 workers to serve drinks and food and sweep up in casinos were German tourists gamble in more obvious ways.

  3. Gravatar of ssumner ssumner
    26. February 2013 at 08:26

    Nishant, Thanks for that info.

    Bill, Yes, I mentioned finance in some other posts–that’s a good point.

  4. Gravatar of TravisV TravisV
    26. February 2013 at 08:28

    Some great stuff in Bernanke’s testimony today. On Japan: “I support their efforts to get rid of deflation.”

    Bob Corker said that the Fed is doing huge favors for the big banks, such as interest on reserves. Bernanke: “None of what you said is true……there is no subsidy involved.”

  5. Gravatar of Tyler Joyner Tyler Joyner
    26. February 2013 at 08:30

    That was quite the revealing comment, Bill.

  6. Gravatar of Britmouse Britmouse
    26. February 2013 at 08:31

    The ONS did update the hourly wage data in that bulletin (which covers 2012 Q4):

    Mean hourly wages rose 2.6% over the four quarters to 2012Q4.

  7. Gravatar of The British miracle: Explosive jobs growth during a recession | Fifth Estate The British miracle: Explosive jobs growth during a recession | Fifth Estate
    26. February 2013 at 08:32

    [...] See full story on [...]

  8. Gravatar of Geoff Geoff
    26. February 2013 at 08:45

    Today’s RGDP is just yesterday’s NGDP, isn’t it? If NGDP has been flat, so would RGDP be flat. RGDP doesn’t actually measure the output of real goods and services, it just measures the spending on them. RGDP is NGDP less inflation. All nominal statistics.

    Is this chart not a chart of NGDP in the UK? I searched “GDP UK” and both this data and “Real GDP UK” data were options.

    Assuming there was the kind of jobs growth reported, it could have been caused by a rise in overall demand for labor, at the expense of an overall decline in other spending, like capital goods, or consumer goods, or, shock!, government spending! That doesn’t need NGDP to rise.

  9. Gravatar of W. Peden W. Peden
    26. February 2013 at 08:58

    Geoff: that’s definitely real GDP data.

  10. Gravatar of Arthur Arthur
    26. February 2013 at 09:34

    I was trying to check this data, and also impressively the ECB database was the seasonally adjusted real GDP figures before they have the non seasonally adjusted.

  11. Gravatar of Geoff Geoff
    26. February 2013 at 09:42

    “Geoff: that’s definitely real GDP data.”

    Can I ask how you know?

    Why does one dataset include the word “Real” while the other does not?

    For the US, I’ve been posting “GDP” charts that do not contain the word “Real”, calling it NGDP. That’s at least right I hope?

  12. Gravatar of Ritwik Ritwik
    26. February 2013 at 09:50


    Not sure why you keep insisting on hourly nominal wages. Does the median worker get paid hourly? Do we get # of hours employed data? Why not by minute? Or by second?

    We simply have employment data. We should want to compare wage/employee data over a longer period of time, say average churn, which could be annual or even more. Or at least weekly wages.

    What’s so special about hours? Esp. since we don’t know how many hours the average worker works anyway.

  13. Gravatar of W. Peden W. Peden
    26. February 2013 at 09:50


    A good familiarity with periods of inflation, recession and boom in UK economic history.

    I don’t know why there are two datasets for RGDP on FRED.

    IIRC, the US “GDP” data on FRED is indeed NGDP.

  14. Gravatar of Doug M Doug M
    26. February 2013 at 10:10

    UK NGDP –

    12 months through 9/30/2012 — 2.2%
    2011 3.0%
    2010 4.2%
    2009 0.0%
    2008 -1.1%

  15. Gravatar of Tomasz Wegrzanowski Tomasz Wegrzanowski
    26. February 2013 at 10:10

    Here’s the ONS data –

    First, the distinction between unemployed and otherwise inactive is total bullshit, and we should completely abolish the concept of “unemployment” and speak about working age employment to population ratios only.

    Second, it’s mostly the end of early retirement behind this. Employment to population ratios rebound almost halfway to pre-recession levels, but workforce composition shifted towards older workers.

    Total hours worked weekly is at all time high, but not per capita.

  16. Gravatar of Mike Sax Mike Sax
    26. February 2013 at 10:14

    Despite the impressive job numbers-subject to revision as you say-unemployment dropped by only 14,000 over the last 3 months.

  17. Gravatar of Doug M Doug M
    26. February 2013 at 10:23


    Hours can be reflected in fractions of hours.

    In the US, we report change in payrols, average weekly hours and average hourly earnings.

    It would be possible for payrolls to increase by 150,000 workers (0.1% of the labor force) but hours worked to fall from 35.4 to 35.0 (-0.11%).

    Did employment increase or decrease?

  18. Gravatar of The UK employment “puzzle” | Historinhas The UK employment “puzzle” | Historinhas
    26. February 2013 at 11:08

    [...] Scott Sumner has done a post on “The British Miracle: exploding job growth during a recession. [...]

  19. Gravatar of ssumner ssumner
    26. February 2013 at 11:27

    Thanks for the data everyone, but what’s Q4 NGDP in the UK?

    Ritwik, Because hourly wage data always refers to pay for 60 minutes, but (unless I’m mistaken) weekly pay data does not always refer to 40 hours, or indeed any fixed period. If I’m wrong then weekly data would be fine, but not annual data.

    Geoff, You said;

    “RGDP is NGDP less inflation. All nominal statistics.”

    In which country did you study economics?

  20. Gravatar of Doug M Doug M
    26. February 2013 at 11:48

    Q4 NGDP will be released next month in late March.

  21. Gravatar of Mario Mario
    26. February 2013 at 11:57

    I haven’t thought this through so I might be way off, but couldn’t you use VAT data as a proxy for NGDP?

  22. Gravatar of Mario Mario
    26. February 2013 at 12:21

    fwiw, UK VAT collections in £m:

    2011Q4: 25403
    2012Q1: 24821
    2012Q2: 24469
    2012Q3: 24524
    2012Q4: 24805

    They report VAT by month, but clearly some payments are made quarterly (maybe even some annually), so the data is really lumpy.

  23. Gravatar of Mario Mario
    26. February 2013 at 12:22

    Oh, I forgot yearly totals:

    2011: 95208
    2012: 98619

    Kind of important considering my last parenthetical.

  24. Gravatar of Ritwik Ritwik
    26. February 2013 at 13:04

    Doug M

    You tell me. What are you interested in, decision-theoretically? Getting people to work, or getting people to work at exactly 35.4 hours? Is the true unemployment rate today is 45% because average weekly hours are now 35 vs 60 in the industrial revolution?


    So if I understand you, you choose hours because it is the smallest unit of time at which employers stop subdividing? As in, they vary # of hours/week but do not vary # of minutes worked per hour?

    Fair point. However, is that true for the median/marginal worker/job? As in, suppose I get paid a $1000/week and I work 40 hours this week an 50 hours next week. My nominal wages per hour just went down and my ‘employment’ went up. But neither I nor my employer made the choice to vary that lever based upon either expectations of future sales or utility/disutility of leisure. It is endogenous to my job.

    In other words, what if nominal wages/hour is an epiphenomenon? Wouldn’t you want to pick that time unit where nominal wages are actually the stickiest?

  25. Gravatar of Felipe Felipe
    26. February 2013 at 13:11

    Geoff: The data you posted is for chained 2009 GBP. That means it is also Real, but a different real than Real GDP. Just different ways of accounting inflation.

  26. Gravatar of Britmouse Britmouse
    26. February 2013 at 13:21

    Doug, no the month 2 estimate for Q4 is out tomorrow, that gives us the first look at NGDP.

  27. Gravatar of ssumner ssumner
    26. February 2013 at 13:31

    Ritwik, I’m not sure I follow—but I’m looking for a number that measures the cost of employing labor for a fixed span of time. That’s the criterion.

    Britmouse, I recall we discussed this earlier, but has anyone asked the British government why the NGDP numbers are delayed so long?

  28. Gravatar of Geoff Geoff
    26. February 2013 at 13:44

    Dr. Sumner:

    I said: “RGDP is NGDP less inflation. All nominal statistics.”

    You said: “In which country did you study economics?”

    I once visited Marketmonetaristan, a quite scary land with the death penalty for anyone who wants to engage in free market money activity external to the “official” taxed currency:

    “As long as NGDP growth is around 4%, long term nominal rates will remain relatively low. That’s the case regardless of whether the 4% NGDP growth is associated with o% RGDP growth and 4% inflation, 2% RGDP growth and 2% inflation, or 4% RGDP growth and 0% inflation.”

    If NGDP = RGDP + inflation, according to you, then

    RGDP = NGDP – inflation, according to you.

    Maybe you like to think RGDP is a real statistic so that you don’t end up concluding that RGDP in the latter years of an inflationary experiment is super high?

    RGDP is not a “real” statistic. It is a nominal statistic of past aggregate spending, because it is the difference between current NGDP and price inflation since that past.

    If anyone disagrees with this, then explain what RGDP would be for year 5, using year 4 as a baseline:

    Year 0: NGDP $10 trillion
    Year 1: NGDP $100 trillion, price inflation 1000%
    Year 2: NGDP $1000 trillion, price inflation 1000%
    Year 4: NGDP $1010 trillion, price inflation 1%
    Year 5: NGDP $1020.1 trillion, price inflation 1%

  29. Gravatar of Petar Petar
    26. February 2013 at 13:52

    Take a look at this, Vuk is onto something

  30. Gravatar of Ritwik Ritwik
    26. February 2013 at 14:36


    “..I’m looking for a number that measures the cost of employing labor for a fixed span of time”

    I agree, and so I’m saying that the choice of this fixed span of time should be a function of how labour is actually hired – per week, per month, whatever.

    So if someone hires me at $1000/week, and I report working 40 hours this week and 50 hours the next, my nominal wages/hour go up and down, but my nominal wages/week stay fixed. The nominal wages/(unit of time) series that shows the most stickiness is probably a good proxy for deciding what unit of time is labour typically contracted at, and hence what matters for nominal wage theories of unemployment.

    Not sure if I was clear there. Think about it as ‘unpacking’ M*V. Would you unpack M*V simply because M and V varies? Both are endogenous epiphenomena. The economic phenomenon is NGDP = M*V. Similarly, are wages/hr and hrs/week both epiphenomena, with the real phenomenon being wages/week? What about wages/week and weeks/year? Etc.

  31. Gravatar of W. Peden W. Peden
    26. February 2013 at 14:52


    “I once visited Marketmonetaristan, a quite scary land with the death penalty for anyone who wants to engage in free market money activity external to the “official” taxed currency”

    Not even accurate as hyperbole. I’m ok with currencies coexisting with the MoA for taxation, as I imagine are a few few others on here.

    The death penalty is restricted to those who draw general conclusions from the local fact of inflation and interest rate targeting.

  32. Gravatar of Britmouse Britmouse
    26. February 2013 at 14:59

    Scott, I did ask them. They produce monthly volume measures for the service and manufacturing sectors and the Month 1 RGDP estimate is based on those volume indicators only.

    They said they don’t have current price data in all the sectors surveyed – agriculture and government output were mentioned.

  33. Gravatar of ssumner ssumner
    27. February 2013 at 07:11

    Geoff, I have no idea what you are talking about, maybe someone else can figure it out. RGDP is a real variable.

    Ritwik, Here’s the problem. Suppose workers are paid hourly, but the government reports their weekly wage. Then hours worked per week changes. The government would show a change in wage rates, whereas actual hourly wages would be unchanged.

    Britmouse, Thanks. It is literally impossible to determine RGDP without prices. “Volume” would count a ton of iron and a ton of diamonds equally. Obviously the diamonds count for more in any sensible real GDP measure, hence you need prices. If they think they can measure RGDP without prices they are doing something wrong.

    What is the “volume” of computers? That makes no sense, as quality changes over time. Is an iPad a computer?

    Somehow I have the feeling that if I actually saw how NGDP were calculated in Britain, I’d be horrified.

Leave a Reply