his morning, I tweeted out Spencer Jakab’s WSJ column on NFP — It’s a Hard Job Predicting Payrolls Number, with the annotation “Its pointless, too.”

While I understand the obligation many economists have to their employers to make a jobs forecast, you have no such obligation. You don’t have to make a prediction, weigh in, make a guess, create a forecast model or even read other people’s forecasts.

Why not?

Here are three reasons:

1) People are really, really bad at making accurate forecasts:  Most forecasts are at best, an educated hypothesis and at worst, a blind guess. A glance at the history of these sorts of predictions reveals that everyone gets these things wrong. I have yet to see someone consistently forecast these things. Indeed, I have yet to see a good 3 month in a row streak forecast by any economist. We simply lack the ability to predict the future.

2) Modelling isn’t much better: The combination of a huge number of known variables, poor data assembly, and a number of unknown variables — plus a healthy dollop of unforeseen randomness — makes employment data forecasting at best slightly better than raw guessing.

3) Even if you could make an accurate forecast, it wont help you in the markets: That’s the funny part of all this — it is a meaningless exercise for investors, and a dubious one for traders. This is especially true in the present investment environment where the FOMC looms as large as they do. The next level analysis is whether the good news is bad (meaning less accommodation) or good (economic improvement) or conversely where bad news is good (meaning more accommodation) or bad (economic deterioration).

Our time would be better utilized trying to discern the current state of the labor market — what actually is (and recently was) rather than what might be. This is useful data for companies, policymakers and labor participants. It has actual utility. Predictions don’t.








Apprenticed Investor: The Folly of Forecasting (June 8th, 2005)

NFP Day: The Most Over-Analyzed, Over-Emphasized, Least-Understood Data Point (February 4th, 2011)



Consensus forecast is for 140,000 in April, with jobless rate at 7.6%.

Employment situation report released at 8:30am

Category: Data Analysis, Employment, Philosophy

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

18 Responses to “Predicting Jobs Data Is Hard — and Useless, Too.”

  1. all good
    165,000 jobs added; unemployment rate down to 7.5%. Real story is positive revisions to prior months http://on.wsj.com/18hpkpll

    • VennData says:

      And you believe these numbers?! Everybody in my neck of the forest is suffering since Romney wasn’t elected. Where do you guys believe that these “numbers” are real? You’re all smoking mirrors!

      – Jack Welch

    • manifest says:

      Ralph Wanger has a presentation where he takes a look at the history (and material nature) of revisions and chuckles about the attention given to data releases.

      But #3 is spot on — even if the figures were reliable and chiseled — in the absence of any correlation between reported figures and impact on stock prices, what would you do with them (outside of some short-term trading) anyhow?

  2. Not all good – participation rate unchanged and average hours/worker falling: “The average workweek for all employees on private nonfarm payrolls decreased by 0.2 hour in April to 34.4 hours. Within manufacturing, the workweek decreased by 0.1 hour to 40.7 hours, and overtime declined
    by 0.1 hour to 3.3 hours. The average workweek for production and nonsupervisory employees on private nonfarm payrolls decreased by 0.1hour to 33.7 hours.”

    Wages aren’t up much either, with average weekly earnings up only 1.9% year over year (and I think less after inflation).

    Until the Wages as a share of GDP returns to historically healthy levels, we’re stuck in an economy where the Fed prints money … to finance Uncle Sam’s deficits … which get spent wastefully … and then are immediately siphoned back out of the consumer economy as corporate profits and/or trade deficit… not enough sustainable gains to get us back on track…

  3. capitalistic says:

    “we’re stuck in an economy where the Fed prints money … to finance Uncle Sam’s deficits … which get spent wastefully … and then are immediately siphoned back out of the consumer economy as corporate profits and/or trade deficit… not enough sustainable gains to get us back on track…”

    The Fed is printing money to encourage market participants to take some long term risk. Long term risk/investments requires capex. Capex usually takes cycles/years to pay off.

    FOMC policies aren’t magically siphoned off to corporate profits. Corporate profits come from the employees willing to work for less or willing to do more work for the same pay.

    So, you can analyze the jobs data, but it doesn’t mean anything, unless you base your investments solely on sentiment.

    • “The Fed is printing money to encourage market participants to take some long term risk.”

      Sorry, but your understanding of the economy and the Fed is out of date. (1) The Fed is printing money because the banks blew themselves up in 2008 and still can’t generate healthy balance sheets without fudging their numbers. Otherwise we’d have more transparent accounting and fewer bank-subsidy policies. (2) The Fed is printing money to cover the U.S. federal deficit. The U.S. deficit is driven by military spending and an out-of-control healthcare system. (3) The Fed is printing money to suppress mortgage interest rates in order to prevent further deleveraging of an economy carrying about 350% “stated” debt/GDP (and more in unstated promises)… a historically almost-unprecedented and unsustainable level.

      If the Fed were genuinely concerned about wanting market participants to take more risk, they would be using their regulatory powers and jawboning other regulators to ensure that the risk-asset markets are worthy of taking risks in. When the futures market participants see folks like Corzine walk off scot-free with their money, you think they’re going to want to take more risk? When we have flash-crashes and hash-crashes, declining liquidity, mark-to-fantasy financial accounting and no accountability on Wall Street, do you think serious investors will be more or less likely to put money at risk on long-term investments? And don’t even get me started about Congress.

      Anyway, even if the Fed was trying to get more CapEx… how’s that policy working out, eh?

    • VennData says:

      Excuse me. Employees can work day and night for a dime an hour, but if you don’t have revenue you don’t have nothing.

      Once you understand that, you understand Keynsian stimulus. Until then. Sit on your cash hoard at Schwab with your arms folded with that red-faced embittered look of yours.

      Sorry you missed the rally.

  4. S Brennan says:

    Well here’s a counterpoint to all the happy talk on jobs;

    No surprise:
    Suicide Rates Rise Sharply in U.S.

    While NYT reporter TARA PARKER-POPE offers a multitude of explanations, the comment section following the article makes clear that 30+ years of thieving by the .01% has left blood in the streets.

    We have to start faking these numbers, they put a lie to official statistics…although jobless people killing themselves help the unemployment numbers…so perhaps our cruel masters can encourage the trend?

  5. rd says:

    With all of the revisions that keep dribbling out for months after the headline number comes out, it means that a forecaster needs to be able to forecast the final number AND the errors that it will have in the initial print. A random number table will probably be almost as effective.

  6. [...] Ritholtz, “Predicting Jobs Data is Hard — and Useless, Too”; The Big Picture (May 3, [...]

  7. Petey Wheatstraw says:

    Flaws in the wetware abound.

    Let’s simply ignore the absolutely pertinent facts that:

    a) The metrics used to calculate the reality of our situation have, and for purely political reasons, been changed to confirm the political bias of those needing to paint an acceptably false representation of the REAL “reality,” and;

    b) We really don’t want to see the reality — the bread line has been replaced by a shiny plastic card that can be used in any program-compliant transactional venue*, without stigmatizing the user or making anyone else uncomfortable on being confronted with the precariousness of their own societal value.

    “Useless” is an apt word for those conveniently dropped through the trap door used to dismiss the no longer counted/accounted for.

    The terminally unemployed need a new word coined for their cohort, or for any other similarly-discounted statistically significant metric (and don’t forget, you heard it here first): The anti-bubble.

    *Just yesterday, at the grocery store, there was a hang up with the customer in front off me (by all appearances a very normal looking middle class woman). She kept trying to slide her card through the scanner, but it wasn’t working. The cashier finally noticed that it was a welfare card, which had to be manually entered into the system. Oh, the fucking horror of it all.

  8. VennData says:

    BR claims “people are really, really bad at making accurate forecasts” …but that is only true half the time!

    Funny how the WSJ rolls that out after their partisan screeds just lost their reads billions of dollars if you bet with what they’ve been spewing for the last week. Funny, haha… Like you followers of the GIP Media Machine.

    That’s right GIP.

  9. mungobola says:

    What Nate Silver did with political polls could be done for employment data to get a better picture of the current state of the labor market. I’ve looked for this kind of thing, but haven’t been able to find it. Anyone seen such a thing anywhere?

  10. spooz says:

    Tyler Durden computed that total worker hours declined by 21,385,800 from March to April, which he then further calculates amounted to a drop of $323.2 in total compensation. There’s always a dark picture to paint if you look for it, I suppose.


  11. [...] Good Points Predicting jobs data is not only hard — it’s also useless – Barry Ritholtz [...]

  12. GeorgeBurnsWasRight says:

    The sideshow to the NFP report is how some people trade according to what ADP report says. Correlation between the two is weak.

  13. Richard W. Kline says:

    That graphic is awesome! And information rich, being completely accurate and metaphorically precise. Where’d you get it Barry, there’s no credit? I WANT this for a sticker or t-shirt; seriously this should be a buy opportunity or bonus for [x] for TBP.