Friday, November 7, 2014

October Employment

I thought this was a good, strong report.  Fixed income markets disagreed with me, I guess, with rates dropping 6 to 8 bp on the news.  I presume that is because of what is considered weak wage growth.  I argue that wage growth is right where we should expect it to be, considering current inflation and unemployment levels.  A supply shock in housing is causing wage gains to go to rent.  The consensus take on this seems to mostly take low growth in nominal wages as a sign against there being inflation pressures.  I think there are some subtle but important problems with that interpretation, but I suppose in the end, it is pulling the fixed income market toward my position where it counts - in forward rates.  I expected to get a small bump in rates today, but I suppose that I was expecting that bump to come from what I consider to be errors in the consensus paradigm.  I should probably be careful about that sort of position, since, as happened here, a differing paradigm may have its own ways of arriving at a pricing conclusion.

....Anyway, on to the data.  The decline in the unemployment rate was reported as a drop from 5.9% to 5.8%, but it was actually from 5.94% to 5.76%.  In fact, the last two months combined have seen a drop of 0.39% in the unemployment rate.  So, we are really only a 0.02% drop away from the 5.5% to 5.7% range I have been forecasting for the EOY unemployment rate, with some indication that we might arrive at the lower end of that range.  (There are some discrepancies between the unemployment duration data and the other data.  I'm not sure what the source is.  The unemployment durations add up to a rate of 5.83%.  But the flows data points to the same 5.76% as the headline number.)

First, the durations data.  My estimate of very long term unemployed (VLTUE) did find its downward trend again (though much of the noise here comes from my simple model), but long term unemployment held fairly steady.  The drop in unemployment didn't come from an obvious drop either in short or long duration unemployment in particular, although nearly 0.1% of the reported drop in unemployment isn't accounted for here.

The general rate of exits from 15+ week unemployment continues to grow, although this month was the latest in a string of months that has pulled back toward the moving average.

After a couple of months where continued unemployment claims pointed to lower unemployment, but total unemployment didn't follow, for three months now total unemployment followed the expected trend downward.  The 0.4% drop in unemployment in the last two months has basically been a reflection of the lower rate of job loss.  It appears to be a real, sustainable drop.  If anything, this still points to possible further drops in total unemployment.

The trends in flows continue to look stronger for workers moving into the labor force and workers moving into employment.

The moving averages for flows, with a longer time frame also show strong trends.  Net flows from unemployment into employment are at cyclical peaks.  And, net flows from not-in-labor-force into employment continue to look strong, mimicking the strong trends in the 2005-2007 time period.  Net flows from unemployment to not-in-labor-force continue to be slightly inflated, however.  I believe this UtoN movement reflects the remaining VLTUE workers who are marginally attached to the labor force.  There is some evidence that in previous cycles, these workers would have seen similar outcomes, but would have tended to report as not-in-labor-force.  The level of unemployment among "Re-entrants" tracks very closely with the level of flows between Unemployment and Not-in-Labor-Force.  Re-entrant unemployment is about 0.4% above long term lows, and it should continue to slowly move down over the next year or two.  This probably largely relates to my VLTUE measure, representing about half of them.

Looking at the last chart, most cyclical movement in unemployment comes from job losers.  The recovery among job losers has been surprisingly linear for the entire recovery.  This certainly establishes a reasonable trend, although there is no reason to expect the trend to continue indefinitely.

Before this report, I was beginning to question the sustainability of the strong trends downward in long term unemployment.  But, the continued strong trends in unemployment insurance together with the strong trends in flows this month suggest that the recent stagnation in long duration unemployment is probably the aberration.  Unemployment durations over 26 weeks were reported this month at 2.9 million.  This is where most of the remaining drops in unemployment will come from.  I think there is still reason to be confident that this will move below 2.5 million in the next 2 to 3 months as VLTUE continues to fall apace, regular flows from unemployment to employment remain strong, and some additional mean reversion from statistical variations all push the number down.

At that point, I expect the linear decline to finally kink, and to see further reductions in unemployment to come at a pace of closer to 0.5% per year.  If regulatory relaxations in mortgage credit lead to acceleration of construction employment, a faster paced decline might continue.  I am hoping for that outcome, although a lot of confusion over asset prices and inflation will lead to push back against that sort of progress.


  1. Kevin: for professional reasons I have a lot of exposure to institutional property markets and buyers of institutional quality property. In that world everyone is talking about a wall of capital that has to be invested --- too much money chasing too few deals. it seems to me minor fluctuations in interest rates could as much be affected by surges of capital into capital markets as by other factors.

    1. I appreciate your input, and I hear that. But, I also hear people talking about funding real estate funds because of the high returns compared to other investment options. Also, construction is stagnant, occupancy is high, and rents are increasing. If there was a wall of capital, then wouldn't we see increased building and falling occupancy and rent? It seems like there is a disconnect. I don't know what the answer is, but the wall of capital idea needs some counter-evidence to these other signals. Do you have any insight about that?

  2. TravisV here.

    Kevin, do you understand how the author below arrives at the following conclusion?

    "I don’t know exactly what’s going to happen, but simple math based on the current level of interest rates leads me to believe that these risk premiums will be much wider in the future over longer time frames than they’ve been in the recent past."

  3. Sure. Doesn't that make sense to you? If we model stocks and bonds on a CAPM framework, then there are two main factors: changing interest rates and fluctuations in operating results. Both have worked against equities in the past 15 years, and interest rates have worked for bonds for 35 years. Equities still have the operating risk, but the interest rate risk can't really help bonds at this point. A portfolio established in long term bonds today is locking in very low returns.

    Is there something about this that you have a different view on?

    1. TravisV here.

      I think the author has a very different sense of future probabilities than I do. What I'm really curious about is your own sense of future probabilities.

      Right now, "Risk Premium" is code for "Probability that a Recession is Coming"

      In the medium-term, I think there's a 25% chance that we'll consistently have inflation around 0% to 0.5%. In that case, equity risk premiums will increase, interest rates will fall, bonds will do well and U.S. stocks will do poorly.

      But that probably won't happen. Instead, I expect inflation to remain higher than 1.0%, equity risk premiums will fall and the S&P 500 will be 10%+ higher than where it currently is.

      While I think equity risk premiums are highly likely to fall, it's harder to say where the yield on the 10-year treasury will go. I think it's conceivable that a 10-year UST yield of 1.8% will be consistent with falling equity risk premiums. Of course, I'd like to see the 10-year yield go much, much higher than that but that might not happen.

      Another thing I'm pretty confident about: over the next five years, either the 10-year U.S. Treasury might do poorly or the S&P 500 might do poorly. But BOTH will not do poorly over a five-year period. That is extremely unlikely.

    2. Sorry, I'm going to be annoying and debate your priors.

      I think changes in stock prices will come more from changing expectations in real output, not from changing inflation expectations.

      In your consistent low inflation scenario, I think stocks will do quite well. Bonds won't suffer capital losses in that scenario, but their income relative to stocks will be very low.

      Your rising inflation scenario will also lead to good stock returns. It will lead to capital losses in nominal long term bonds if inflation expectations rise.

      I would need to see evidence of changing total returns to capital to change my expectations. If the 10 year has a yield of 1.8%, equity premiums will remain high. If the 10 year is below 2% and equity premiums fall significantly, either NGDP will be much lower, or stock prices will be through the roof.

      To be honest, I'm not well versed in the academic work here, but it seems like the causation is usually treated as going from low growth expectations to low interest rates. It seems more likely to me that low interest rates lead to both high equity premiums and low growth rates, because the low rates are a signal of low risk tolerance, and thus lower investment into transformative productive assets.

      I agree with your last paragraph. That would probably only come about with 1970's level inflation, which is unlikely.

    3. TravisV here.

      I feel like I have a simple model with a few variables that are really critical while you have a sophisticated model with lots of variables that are really critical.

      My prior is that monetary policy and sticky wages are really really really really important. Due to tight money (decreases in expected NGDP) and sticky wages, the U.S. economy has been like a factory operating at 60% capacity. Therefore, Bernanke's increases in expected NGDP have driven ENORMOUS gains in expected real growth. Sumner's key point is that large increases in expected NGDP result in large increases in RGDP plus teeny bumps higher in inflation.

      Even today, I still believe there's lots of slack out there. Even after all these gains, I still sincerely believe easier money could still create huge increases in expected real growth and therefore real stock prices.

      U.S. stocks seem to agree with my simple model. Just look how strongly they react to even small positive surprises like Japan's the other day.

      [As an aside, it's not quite clear to me how sticky wages fits into your thinking on all of this stuff]

      In my simple model, a high risk premium just means investors fear there's a high probability that a Lost Decade of low NGDP like Japan's is in our future. And a low risk premium implies the opposite: investors expect consistent 1%+ inflation and stability.

      If inflation does indeed fall and scrape along at 0.0% to 0.5% consistently, that would be result in a Lost Decade like 1990's Japan for the U.S. Pretty clear to me that that would only happen in we get NGDP that is far far below what is currently expected. With sticky wages, that would be horrible for stocks. The S&P 500 would fall significantly lower than where it is today. If investors believe we'll keep hitting the ZLB over and over and over, then risk premiums will be high.

      So I'm not quite understanding why you wrote "In your consistent low inflation scenario, I think stocks will do quite well.".......

    4. TravisV here.

      Just to add to my thoughts above, why do you think equity risk premiums were super-high all throughout the 1950's?

      My guess is it's because in 1950, investors thought the probability of another Great Depression was very high, with that probability gradually falling to a far lower number by 1959........

    5. Interesting ideas here. I don't think my model is really more sophisticated than yours. I think we might just have some assumptions or presumptions that are slightly different. And, mine might be wrong.

      Thinking through this might be worth a full post. I think there are interesting things going on here regarding equity premiums, economic growth, etc.

    6. TravisV here.

      Kevin, thank you so so much.

      You're clearly a brilliant guy and you have a phenomenal ability to evaluate theories by pulling together data. So I'm still very surprised by our difference in thinking on equity valuation here. Because you're so smart and uniquely skilled I really hope you'll follow up on this one.

      I feel like the core difference between us is "How much does the demand side matter?" My view is that it's so important that it overwhelms other factors.

      I think downward nominal wage rigidity is really strong, so therefore decreases in NGDP are really really harmful because huge losses in RGDP will always result. See here on wage rigidity:

      Negative demand shocks have been far far more harmful than negative supply shocks. See this post by Sumner:

      For a while, I personally flirted with the idea that the recent U.S. oil / fracking revolution was important but Mark Sadowski disabused me of that notion. See what Sadowski says in this comments section on "Petroleum Derangement Syndrome."

      Mark Sadowski and I might say the demand side (NGDP shortfalls) has been hugely hugely important and all the other supply-side harms have been teeny-tiny in comparison.

      Sumner might say that the demand side (NGDP shortfalls) has been hugely, enormously important while supply-side harms like the increases in unemployment insurance and the minimum wage have been far less (but still somewhat) important.

      Meanwhile, you might view supply-side harms, especially government housing policy, as even more important than Sumner does. Therefore, it appears that you're far less concerned than I am about how damaging it might be to equity risk premiums if NGDP expectations, inflation expectations and long-term interest rates fall significantly to consistently lower levels in the future.......

      P.S.: It is true that the demand side matters less today than it did in 2010. Eventually, wage expectations moderate, wages grow slower than other prices and we get closer to full employment / potential RGDP. However, I'm convinced that the demand side is still important because U.S. equities still respond strongly to small surprise news from the Fed, BOJ, ECB, etc......

    7. Thanks Travis. Thanks for all the links, too. I'll look at them soon. After my first glance at the issue, I think the data are going to end up surprising me. This should be interesting to work out. I don't think my answer to some of our forecast differences is what I thought it would be.

      Regarding some of your comments here, I think maybe one difference is that I see a lot of parallels between real interest rates, real wage growth, and production growth. These things will all help to reduce wage stickiness and the zero lower bound problem, and will tend to lead the Fed to err toward being more loose as the natural rate trends higher.

      I don't disagree that the secular trend of tight money could be raising risk premiums. But I think other factors may be important.

      My main concern with monetary policy is a negative demand shock. Steadily tight money might not look that different than the late 90s.

      But you have made me realize I need to think this through some more.

    8. For instance, I don't think low unemployment leads to inflation through rising wages. I think it's more likely that low unemployment coincides with rising natural interest rates, which cause Fed policy to err on the loose side, which then causes inflation.

      I don't think inflation is naturally pro-cyclical. It just looks that way because Fed policy errors are pro-cyclical.

    9. TravisV here.

      First, imagine if we have 1.1% inflation for the next four years. That will not look much like the 1990's. NGDP growth was far higher then. 1.1% inflation implies very low NGDP growth. RGDP growth would probably be pretty darn low in that case.

      On the other hand, if we had 3.1% inflation for three straight years, I think that would drive a lot faster RGDP growth and result in a lot higher stock prices.

      I agree with you that low unemployment doesn't necessarily result in high inflation and vice versa. FDR achieved 14% inflation or so in 1933 when the unemployment rate was around 20%, an example Sumner cites frequently against that line of thinking. High inflation or high interest rates can result from either demand or supply-side factors. Helpful to distinguish between the two.

      Marcus Nunes has an excellent analysis of how the Fed hugely mis-interpreted signals that were getting from inflation / interest rates from during the technology boom from 1996 to 2000:

      One concession: If you have high unemployment, then an acceleration in NGDP should result in a huge acceleration in RGDP + teeny-tiny increase in inflation. But if you're truly at full employment (very little slack), then an acceleration in NGDP should result in virtually no acceleration in RGDP + huge increase in inflation.

      I still think we're a long way from truly full employment. Lot of slack. Lots of people would be willing to work lots lots more hours if only aggregate demand were higher.

    10. TravisV here.

      Prof. Sumner glanced at our conversation........

  4. Travis-

    My main points have been that we have 12 million in the USA on dubious VA and SSDI "disability" and that cultures are mutable. More people will enter the labor force when demand is strong. Delayed retirement, finding babysitters etc.

    If labor is scarce enough, employers become flexible--"okay, you need Wednesday's off, we will work with it."

    BTW, check out Tim Duy's chart on unemployment and inflation. You give up a ton of employment to maybe cut inflation by one percent. It is sick.

    1. Yes. I think the whole employment/inflation trade off is very weak. I suspect that the main reason low unemployment correlates with high inflation is because the Fed is always chasing the natural rate of interest. Fed policy is, thus, pro-cyclical. Mostly, low unemployment, in and of itself, leads to higher real production.

      That's speculative. I'm not sure I'm able to tease that from the data or not,

  5. TravisV here.

    Benjamin Cole, right, I agree with you. I think we see the situation very similarly. In addition to your great points, lots of people working part-time would work full-time if only there were faster NGDP growth. That means there's tons of slack still left in the economy. Which means that if we had faster NGDP growth, we'd get a lot more RGDP and only a little more inflation.

    Benjamin Cole is also making an additional key point: faster growth in demand can often lead to better policies on the supply side and vice-versa. For example, if NGDP growth had been faster in 2009 and 2010, unemployment insurance would probably be much less generous.

    Benjamin Cole has also frequently warned about the possibility of consistent 0.0% to 0.5% inflation in the U.S. It could easily happen here. And if it does, I think almost everyone will be worse off, including U.S. stockholders. Benjamin Cole, hopefully you agree with me that the S&P 500 would be significantly lower in that scenario (consistent 0.0% to 0.5% inflation).

    Kevin, of course I agree with you about low unemployment / high inflation. It happens because inflation targeting is fatally flawed. Inflation targeting doesn't distinguish between demand and supply shocks. So if you commit to a strict 2% inflation target, that means millions of people must lose their jobs if a temporary oil shortage happens. That is perverse logic.

    And on the flip-side, if you have a technology boom, then 2% inflation targeting might lead to 7% or 8% NGDP growth, which is overshooting. When that technology and RGDP boom wears off, very high inflation will be required to keep NGDP growing at 7% or 8%.

    Marcus Nunes wrote a phenomenal post illustrating that something similar happened to that during the 1996 to 2000 period:

    P.S.: To reiterate, the key question here is how much the demand side will matter for stocks relative to the supply side over the next three to four years. I think a lot of Benjamin Cole's analysis indicates that the demand side will continue to really really really matter a lot........

    1. Hey Tavis. I just wanted to let you know I haven't forgotten about you. The question has led me down the rabbit hole. I revisited my data from the corporate leverage & interest rate posts, and I'm running into all sorts of interesting stuff. I'm not even sure how to organize it all at this point, but I might end up with several posts before I'm done.

    2. TravisV here.

      Glad to hear it! I tend to be reasonable on a lot of issues except........macro. On that issue, I'm a crazed monetary extremist. Money really really really really matters a lot compared to other stuff.

      This issue is not going away. Tight money is clearly the main factor explaining unemployment in the Eurozone. And it explains why the U.S. should experience far more real growth than Germany over the next 10 years. How do I know? Compare long-term interest rates for the two countries.

      Even in Japan, where wages have had decades to adjust to low NGDP, easier money can clearly still help a lot, as shown by the strong reaction to Kuroda's latest surprise.

      And here in the U.S., we're in a precarious situation with 5-year inflation expectations at 1.6%. That is a lot more dangerous than having those expectations at 2.1%. If 5-year inflation expectations fall to consistent levels of 0.0% to 0.5%, it is going to be enormously painful for almost everyone, particularly regular working people and stockholders.

    3. TravisV here.

      Also, a couple days ago, I wrote the following comment on a Sumner post:

      "Prof. Sumner,

      I'm a huge fan of this blog post. Very frequently, when explaining a concept, I'll exaggerate certain relationships in order to better get the point across.

      The hope is that the listener will say "Well, that's a bit exaggerated but his point still holds....."

      Perfect example: my analyses in the comments section of this blog post:

      I confess: I exaggerated the strength of a lot of relationships in an effort to quickly illustrate the workings of the model in my mind."

    4. I promise, I really haven't forgotten....still working on it.