Why bond yields rose on the disappointing payroll report

Normally you’d be better off looking at how someone like James Hamilton evaluates economic data, but I’ll take a stab at this one.  In early 2008 when we were in danger of slipping into recession, I got interested in the unemployment rate as a leading indicator (it is usually viewed as a lagging indicator.)  I asked myself how much of an increase in unemployment (during economic expansions) would be needed to indicate something more than just a blip in the recovery and that a new recession was underway.

I noticed that during expansions the unemployment rate never seemed to rise more than 0.6%, before falling again.  That is, unless we were actually going into a recession.  In that case unemployment rose much more, indeed at least 1.9% (in the mild 1980 recession.)  In between was a sort of donut hole, with almost no increases in unemployment of more than 0.6% that did not mean a recession was underway.  (Thus the US doesn’t have mini-recessions, even though all our economic theories predict we should see them more often than regular recessions.)  I say almost, because there was one exception; unemployment rose 0.8% during the 1959 nationwide steel strike, without any recession.  But the US economy is now far more diversified, and such an event is now very unlikely.  The share of employment in unionized industries like autos and steel has fallen sharply.  Here’s what I infer:

1.  Increases in unemployment of more than 0.6% are almost always indicators of recession.

2.  Increases in unemployment of more than 0.6% are ALWAYS economically significant.

The second statement is slightly more definitive, because I consider the 1959 steel strike to be an economically meaningful event.

If I’m right, then any change in the unemployment rate of more than 0.6% over a relatively short period of time, probably indicates that the actual unemployment rate changed—that it wasn’t all statistical noise.

Between November 2010 and January2011 the unemployment rate fell from 9.8% to 9.0%.  Let me emphasize that I strongly believe some of this was noise in the data, perhaps as much as 0.6% of the fall.  Thus the actual rate may have fallen from say 9.5% to 9.3%.  But I don’t think it was all noise, otherwise we would have seen previous episodes where the unemployment rate ticked up by more than 0.6% without triggering a recession.  And over the past 63 years (more than 750 months) we just don’t see meaningless blips that large.

Even an actual drop of 0.1% per month would be significant, that is faster than we had in 2010, and even faster than the pace of recovery that most forecasters expect over the next 5 years.  And remember that my argument suggests that 0.2% is the minimum plausible estimate for how much unemployment has actually improved; it may be a bit more.

I think this is why bond prices fell and yields rose.  The bond market knows that the payroll numbers are usually more reliable.  But they also know that both numbers are subject to error, especially with snowstorms disrupting data processing at some firms.  I actually know someone who expects to soon get a job, but the actual hiring has been held up by the recent storms in Boston, which have created a work backlog.  I think the bond market understands that while the payroll number is usually better, a change in unemployment that large is almost always economically significant.

BTW, my recession indicator worked pretty well in the 2008 recession.  By December 2007 unemployment was up 0.6% from the low, and in March 2008 it was up 0.7%.  Yes, the recession had already begun by March, but as late as mid-2008 some prominent forecasters still didn’t expect an outright recession.

PS.  I’ve been very busy and am way behind on comments.  I hope to get to the backlog this evening.

PPS.  Paul Krugman has a new post on the ECB in 2008 that is similar to arguments made by me and other quasi-monetarists about the Fed in 2008—they wrongly focused on high commodity prices.  Of course we tend to look at NGDP.

PPPS.  I just realized there is another (less appealing) possible interpretation.  It may be that the bond market thinks the lower unemployment numbers makes a premature Fed tightening more likely.  Let’s hope not.

Update (2/5/11):  Just as I predicted, James Hamilton’s analysis is better than mine.  He relies heavily on this excellent Rebecca Wilder post.


Tags:

 
 
 

20 Responses to “Why bond yields rose on the disappointing payroll report”

  1. Gravatar of Zarathustra Zarathustra
    4. February 2011 at 10:14

    Here’s my random thought: a 0.6% rise of unemployment would indicate problems, but how can I be sure that 0.6%, not 0.7% drop of unemployment, would indicate a recovery?

    If the latest number does confirm the recovery, then I think the bond market reacted by thinking that there will be no QE3.

    Also sprach Analyst

  2. Gravatar of StatsGuy StatsGuy
    4. February 2011 at 10:26

    Not sure what to make of the bond reaction given that NFP came in >100k below expectations. I also don’t know how to reconcile this with the large drop in unemployment rate, while the BLS claims labor force participation was steady. That was confusing…

    http://www.bls.gov/news.release/empsit.nr0.htm

    I struggled to reconcile what is seasonal adjustment, annual population adjustment, and measurement error.

    BTW, yields rising could be because of a couple things – belief that this allows the Fed to tighten sooner, or belief that the number was worse than expected (NFP was worse, inflation was better) and that the lack of recovery is going to require that the Fed resort to more inflation (thus driving up yields). So, I wasn’t 100% sure how to interpret this, though I’m inclined to agree with you.

    On the 2007 recession – I suspect that many of the folks who expected a “soft landing” in 2008 also expected the Fed to ease substantially (heh, I did – dumb me). My view of the commodity bubble in 2008 was this was in anticipation of QE which was expected as a response to the downturn. The Fed saw this as a dollar run and got cold feet, and did exactly the opposite of what the market was expecting. If the Fed had held to this expected path, we would have seen a quick burst in inflation which would have subsided as the dollar dropped, exports picked up, economic activity recovered, employment held steadier, and eventually the dollar would have recovered somewhat. The conventional view at the time was “Bernanke is a student of the great depression, he’ll know what to do.” oops.

  3. Gravatar of StatsGuy StatsGuy
    4. February 2011 at 10:27

    “NFP was worse, inflation was better” –> “NFP was worse, unemployment was better”

  4. Gravatar of ssumner ssumner
    4. February 2011 at 10:52

    Statsguy, I made exactly the same mistake in 2008.

    Regarding the bond market reaction, the view that a lack of recovery will force more QE, thus creating more inflation and driving up long rates is certainly possible, but the simplist explanation is that they focused on the household jobs number (was it plus 500,000?)

    I say that because even if your suggestion is right, it just begs a deeper question–why doesn’t the bond market usually react that way? In general, bond yields rise on weak NFP numbers, and they rise on strong numbers. It’s not clear why this time would be different. I think the simplest explanation is that the 0.8% drop in unemployment over 2 months was viewed as being large enough to be signficant. Note how I always assume markets look at things the way I do. That’s the rational expectations model. 🙂

  5. Gravatar of David Pearson David Pearson
    4. February 2011 at 14:10

    Scott,

    There are many potential explanations for yields rising, all of them vaild, but not necessarily consistent.

    First, the unemployment drop was primarily due to workers exiting the work force. The number of “Employed” persons in the Household survey rose only about 120k. Better than the Establishment Survey, but not great. In general, statistical relationships (such as “.8% reduction in UE in two months must be valid.”) imply actual employment growth, not more discouraged workers, as was the case this month.

    One explanation for yields backing up is that Bernanke signaled yesterday that he has been disappointed with the pace of job growth. Thus, the weak jobs number implies more QE ahead, and more QE implies higher inflation. Note that the 5yr TIPS yield is still negative (hardly indicative of real growth expectations), and the 10yr TIPS yield is about where it was pre-Jackson Hole. So almost the entire back up in yields since the Fed signaled QE2 has been due to an increase in inflation expectations.

    The chatter in the markets the past few days has been about how the Fed has signaled it will not pay attention to headline inflation or commodity prices. For traders this is a signal to bid up commodities without concern over Fed tightening. For bond investors, this is a reason to sell bonds.

  6. Gravatar of David Pearson David Pearson
    4. February 2011 at 14:29

    Scott,

    Correction: the Jan2016 5yr TIPS yield is at 13bp (positive) — I was looking at the Jul2015 instead, which is still negative. A real yield of 13bps, again, does not show much concern over real growth/tightening ahead.

  7. Gravatar of Dustin Dustin
    4. February 2011 at 15:14

    Number of employed people in the Household survey increased 117,000. But if you remove the “2011 population control effect”, then it increased 589,000.

    I don’t know what a “population control effect” is..

  8. Gravatar of scott sumner scott sumner
    4. February 2011 at 18:01

    Zarathustra, My argument is that random fluctuations never exceed 0.6%. I assume that’s true on both the upside and downside.

    I agree that a stronger recovery makes QE3 less likely.

    Nice China blog!

    David, There are flaws in the TIPS that make it difficult to measure real yields. I am referring to the lag in indexing, which can be a factor when commodity prices have risen or fallen sharply (as the TIPS work off headline inflation, not core inflation.

    My other comment is that the bond market reaction I referred to occurred right after the employment report, which suggests it was linked. Bond market reaction to Bernanke’s speech would have come earlier.

    But I do agree that there is substantial uncertainty about how and why bonds reacted to the report. For me, the most straightforward explanation (the number was stronger than expected) is the most appealing.

    Dustin, I don’t know what the effect is either, but I’ve learned over the years not to trust individual government numbers, but rather to look for a pattern. I thought the two month pattern in the unemployment rate was striking. I recall in early December when the 9.8% number came out, most economists said unemployment would rise to over 10% in early 2011, before falling. I wonder how they would have reacted if you suggested the rate would fall to 9.0% in two months. I’d guess they would have been astounded.

  9. Gravatar of steve steve
    5. February 2011 at 05:54

    Scot-H ow many of those prior 0.6% changes occurred during times with unemployment durations like we see now? Recoveries have generally been much faster.

    Steve

  10. Gravatar of Russ Anderson Russ Anderson
    5. February 2011 at 08:54

    Keep in mind with the Establishment Survey that the jobs numbers get revised. The January report reported 36,000 jobs in January but December’s total was revised upward by 18,000 and November by 22,000. November was first reported as 39,000 jobs but has been revised to 93,000. October was first reported as 151,000 but has been revised to 210,000.

    The upward revisions per month from July to Nov are 65,000, 53,000, 71,000, 59,000, and 53,000. My point is that the 36,000 job increase reported in January could end up closer to 90,000 if the current pattern of revisions holds.

  11. Gravatar of Mark A. Sadowski Mark A. Sadowski
    5. February 2011 at 08:59

    statsguy wrote:
    “I struggled to reconcile what is seasonal adjustment, annual population adjustment, and measurement error.”

    Yes indeed, me too. After reading Hamilton, Wilder, etc. I’m only more confused (as is DeLong). I think the safest thing to do in such situations is to look at the employment to population ratio and observe that at 58.4% it’s still only two tenths of a point from the bottom and it’s been at or below 58.7% since September of 2009. Wake me up when the labor market recovery really begins. (Yawn.)

    As for the bond market reaction I think the simplest explanation is they looked at the headline unemployment rate and assumed there just had to be good news buried in the BLS report somewhere. As for me, I’m still looking.

  12. Gravatar of scott sumner scott sumner
    5. February 2011 at 10:04

    Steve, I agree that recoveries have been much faster, and that this one is awful. But I do think the evidence suggests that it is picking up in the last few months.

    Russ, Yes, I think within three months it will be clear a corner was turned in late 2010.

    Mark, I’m surprised you don’t find their comments persuasive, I do. Where does DeLong post on this ? I’m going to add an update and link to Wilder. BTW, wouldn’t the employment to population ratio be distorted by the exact same adjustment that Wilder discussed? Or did the census claim the US population suddenly fell in January?

  13. Gravatar of Mark A. Sadowski Mark A. Sadowski
    5. February 2011 at 10:40

    Scott,
    You wrote:
    “Mark, I’m surprised you don’t find their comments persuasive, I do. Where does DeLong post on this ? I’m going to add an update and link to Wilder. BTW, wouldn’t the employment to population ratio be distorted by the exact same adjustment that Wilder discussed? Or did the census claim the US population suddenly fell in January?”

    I didn’t mean to imply DeLong was referring to Hamilton or Wilder, only that he was also perplexed. Here is the link for what it is worth. He simply throws his hands up in exasperation.

    http://delong.typepad.com/sdj/2011/02/employment-report-household-and-establishment-surveys-pointing-in-very-different-directions-this-morning.html

    As published the civilian noninstitutional population fell 185 thousand. The population control effect was -347 thousand. Thus removing the population control effect leads to a population increase of 162 thousand and increases the employment to population ratio only by 0.1% to 58.5%.

    Wilder herself concluded with:
    “Of course, there’s a slew of workers that are not in the labor force that may re-enter, which would undoubtedly drive the unemployment rate up (it probably will). And let’s remember this when talking about a “strong” employment report: 9% doesn’t represent the severity of the unemployment problem – the employment to population ratio does.”

  14. Gravatar of Mark A. Sadowski Mark A. Sadowski
    5. February 2011 at 11:08

    OK, maybe I’m being a little pessimistic. 900,000 jobs in two months is not trivial and it is movement in the right direction. But to put things in perspective, even taking into account the population correction factor roughly 40% of the 0.8 points drop in the unemployment rate was due to people leaving the labor force. So I’m not quite ready to break out the champagne just yet. Let’s see what next month’s numbers look like.

  15. Gravatar of It´s all very confusing « Historinhas It´s all very confusing « Historinhas
    5. February 2011 at 11:34

    […] Scott Sumner has a take: If I’m right, then any change in the unemployment rate of more than 0.6% over a relatively short period of time, probably indicates that the actual unemployment rate changed””that it wasn’t all statistical noise.  […]

  16. Gravatar of Lorenzo from Oz Lorenzo from Oz
    5. February 2011 at 13:59

    Mark gets the award for using statistical language correctly. Scott, I am betting when you say 0.6% you actually mean 0.6 percentage points. It is not the same thing.

  17. Gravatar of scott sumner scott sumner
    5. February 2011 at 16:08

    Mark, You said;

    “OK, maybe I’m being a little pessimistic. 900,000 jobs in two months is not trivial and it is movement in the right direction. But to put things in perspective, even taking into account the population correction factor roughly 40% of the 0.8 points drop in the unemployment rate was due to people leaving the labor force. So I’m not quite ready to break out the champagne just yet. Let’s see what next month’s numbers look like.”

    That would mean almost 0.5% of the drop was real. If you are right, I’m ready to break out the champagne.

    Lorenzo, You win your bet. I gambled my readers are smart enough to see through my time-saving sloppiness. On the other hand writing this note negates all the gains from being sloppy. 🙁

  18. Gravatar of Lorenzo from Oz Lorenzo from Oz
    7. February 2011 at 20:55

    Scott: I used to do be an on-tap statistical expert for politicians and their staff. You quickly get very careful about that sort of thing because they don’t get it unless you are. I remember vividly one Member of Parliament asking what ‘LHS’ meant on a graph (Left Hand Side). Or taking a moderately senior staffer for new Treasurer through, very carefully and basically, what Budget Papers were.

  19. Gravatar of D. Watson D. Watson
    9. February 2011 at 10:12

    Scott,

    I wrote up a supporting post trying to explain why unemployment could be a leading indicator when we typically think of it as a laggard. It boils down to breaking unemployment into its constituent flows with firms being forward thinking, discouraged workers backward looking, and differences in the costs of hiring and firing. Thoughts?

  20. Gravatar of Scott Sumner Scott Sumner
    10. February 2011 at 06:47

    Lorenzo, Good point.

    D. Watson. Nice post. I guess the best test would be to do a horse race between unemployment and other indicators like industrial production and retail sales.

Leave a Reply