Thursday, January 30, 2014

RDGP and NGDP

Market monetarists, who would have the Fed target an NGDP growth trajectory, often talk about the relationship between Nominal GDP (NGDP) and Real GDP (RGDP).  In short, it's not a fixed-pie relationship, where real growth goes up when inflation goes down.  It's not even a non-relationship, where they go up and down independently.  In fact, under conditions where inflation is not excessive, higher inflation tends to come with higher real growth.

Here, I have graphed RGDP on a scatterplot with an inflation measure.  What we can see here is that there appears to be some level of inflation - around 4% or so - above which the balance of positive and negative effects from inflation switches and becomes increasingly negative.  This was the case from 1973 to 1982.  When we combine all the periods since 1948, we can see the humped relationship.
If we take out the period before 1973, which had more volatile GDP behavior, the relationship is even more clear:
 
Demographics are certainly putting a damper on aggregate growth, and the correlations run in all directions here.  But, since 2009, the sideways movement in the Employment-Population Ratio has been about what one would predict, given the level of RGDP growth we have had.  Where would we be now, if the Fed had kept a 4% inflation target instead of a 2% target?  The nice thing about NGDP targeting is that they wouldn't have to make that distinction explicitly.  If they had a 5% NGDP target, inflation would find its own level, and wage and bond markets would be able to clear more easily.

In the meantime, it's worth noting that having an inflation rate consistently below 2% may be just as bad a policy as having inflation around 6% or 7%.  I don't see any reason to be afraid of 4% inflation, at least on occasion.  What caused us to be so afraid of 4% inflation?  What awful historical incident was triggered by 4% inflation?
 
 


Wednesday, January 29, 2014

A Couple More Minimum Wage Regressions

I've done a couple more sets of regressions that take economic growth into account, and I have found very strong relationships.  The strongest relationship uses all three trends from each episode (pre-MW, MW, and post-MW).  It attributes a decrease of 3.6% in teen Employment-Population Ratio and 1.4% in total EPR to the implementation of the typical series of minimum wage increases.  Results below the fold:

QE's and the end of the zero lower bound

Here is a graph that is a rough approximation of the date the market expects short term rates to escape the lower bound.

The date stabilized briefly, roughly coincident with QE2.  But generally, over the past 5 years, until the beginning of QE3, that future date, like a carrot on a stick, just kept moving out in the future, with some noise in either direction.

I think we can possibly call Operation Twist a failure.  It apparently didn't do anything to change the trajectory of expecations, but it saddled the Fed with a lot of duration risk that is now a cause of concern for the FOMC and some observers.

Here is a graph of the expected date of the first rate increase and the expected slope of the yield curve at that time:


Source: Authors calculations, based on daily prices of Eurodollars futures
Slope (left scale, in bp), date of first rate hike (right scale, in quarters from 1Q 2013)

The blue line (expected date of the rate hike) corresponds with the QE3 period in the graph above (it is roughly the inverse of that graph).  We can see that all of the permanent increases in interest rates over the past year have been the result of an increase in the slope of the yield curve, not a change in the expected date of the rate increase.

I would expect the slope of the yield curve to slowly trend up to around 40 or 50 bp per quarter, and this would cause a moderate amount of continued increases in interest rates over the next year or so.  (This is the typical slope coming out of a trough.)

I expect the date of the rate increase to arrive in the approximate time frame that the market expects.  The question now is what happens as the Fed tapers.  If some of the tail winds I am seeing in the labor market continue to play out, then we might see that date move up 6 months or so to late 2014.  If the disinflationary pressures of the taper cause the same sort of retardation that the previous QE tapers caused, then rates could crash, and we could have some difficult long-term problems.

The Fed has consistently underestimated the strength of the labor market, and with a current forecast of 6.3% - 6.6% for the 4th quarter of 2014, short of a massive correction in the direction of the economy, they have underestimated it again.  I hope this leads them to a dovish posture.  If that expected date starts creeping out again, we've got some problems.

Tuesday, January 28, 2014

North Carolina FTW!

The results from North Carolina continue to be extremely positive.  If this foretells the national experience, hold on to your hats, people, it's going to be a heck of a year.

This first graph is monthly changes, in thousands of people.  This is not a cumulative graph, folks.  The divergence in the employment and unemployment indicators is momentum.  That means that each month is improving at a faster pace than the months before.

Below, I have updates of the graphs I had done, comparing the North Carolina stats to the national stats.  The North Carolina unemployment rate is down a whopping 2% since Emergency Unemployment Insurance was terminated.

I still have some question about whether any of the labor movements before June, which weren't positive, might have been related to the policy.  But, here's what I said two months ago:
It looks plausible that the North Carolina experience will support both of my estimates that (1) unemployment is about 1% higher than it would be without EUI and that (2) LFP is slightly higher (less than 0.2%).
Holy cow, look at this graph, of unusual movements in the North Carolina data, compared to the national data.  LFP down about 0.2% and unemployment down more than 1% since the end of the policy.  I need to start charging you people some serious cash-ola for this information.

Here are all the graphs of North Carolina compared to the national numbers.





Policies are for identifying outsiders

A paper by Jeremy Greenwood, Nezih Guner, Georgi Kocharkov, and Cezar Santos outlines the overwhelming importance of assortive mating and female labor participation as causes of measured income inequality. (HT: Tyler Cowen).  Comparing 2005 to 1960, they find that a very large cause of inequality is the trend for women to become more educated and to enter the labor force, and to marry men who have similar earning power.  Essentially, they find that if you eliminate these cultural changes, the measured increases in inequality would disappear.

I have previously posted about how a large amount of household income variance is related to the number of earners in the household, and how much of the shift in inequality and median income levels is a result of having more one-earner and zero-earner households.

But, I think that the paper cited above is a great example to use to look at the implicit function of political policies.  Here are two lists of possible solutions to the inequality issue:

List A:
More progressive taxation
Wage controls
Restrictions and mandates on employers

List B:
Restrictions on mate selection
Forced birth control or adoption for unmarried parents
Return to 1960 level of female education
Return to 1960 level of female labor force participation
Restrictions on number of earners per household
Forced marriages


I suspect that you have a sour reaction to list B.  I know I do.  I think we could all agree, though, that the List B solutions would, for the most part, have a very direct and potent effect on measured inequality.  We could argue that some of those policies might not work so well in practice.  But, some of them, if they were really implemented, would clearly improve measured household inequality.

Your reaction to List A is probably very positive compared to List B.  These are commonly proposed solutions, and, in fact, they are solutions that have been implemented, at some scale.  We could also argue that these policies don't usually work as well in practice as they do in theory, but a reasonable argument can be made that they can be somewhat effective at creating marginal improvements in measured inequality.

If we only had one goal, though - to reduce inequality - and no set of principles limiting our solutions, List B would be overwhelmingly more direct and effective.


A Presumption of Rights

Bryan Caplan recently commented on the tendency for public policy to be implemented indirectly.  His conclusion is that citizens would see the moral trade-offs of policies more clearly if they were imposed directly, so policies are implemented through businesses and other focused agencies in order to obfuscate the moral and financial costs to the average citizen.

But, I think we need to take this a step farther than Bryan did.  My sour reaction to List B could be self-interested, but I feel just as strongly about opposing the specific limits on female personal development as I do about the other List B proposals.  I don't think self-interest is as operative here as is simply a basic notion of inalienable rights.  You just can't prevent people from becoming educated or marrying whom they choose.  Full stop.

So, there are two overriding influences on our policy constraints.  Regarding List B, there is no plausible social outcome that would be bad enough for us to implement these policies.  Regarding List A, there are a range of social outcomes that might plausibly lead to support for these proposals.  In fact, going back to the paper's starting point in 1960, considering the complexity of social development, we could expect with some certainty that between then and now, some list of problems would arise that would lead reasonable people to call for these policies as a solution.

We can see that the inevitability of List A comes not from its effectiveness, but from its philosophical availability.  Support for these policies is not a product of a search for solutions so much as it is an identification of social agents who can be reliably coerced without triggering a universal response of outrage.  It's an identification of non-affiliates - factional outsiders - a stand against bourgeois dignity.  This posture is a deep and foundational human tendency.


In Practice

I will close with a specific example of this process at work.  Last year, the New Mexico Supreme Court ruled that a photography business could be forced to photograph a gay wedding.  I would love to live in a country where discrimination of all types wasn't common.  But, note what solutions to this problem are not on the table.  You can still refuse to hire a gay photographer.  You can refuse to work for a gay photographer.  It seems obvious to me that more damage is done because of those two perfectly legal sources of discrimination than because gay couples might have a few less photographers to choose from when they marry.

Note also, that the photographer argued her case based on free speech.  The right of freedom of association, if you happen to be engaged in lowly, filthy commerce, is so out of favor in the Land of the Free, that it would damage your case to assert it.

That ruling was explicitly about the expansion of civil rights.  But, more accurately, it was about precisely the opposite.

(edit:  Please read the above paragraph carefully.  I do not support bigotry.  My point is that there are large areas in our personal lives where we demand the right to be bigots.  We don't say this explicitly.  But, in marriage, for instance, while a consensus of the population is against blatantly bigoted controls on marriage, we all take for granted that our own personal decisions can be as idiosyncratic and prejudiced as we like.  We would stand for no less.  I am attempting to make this distinction between our personal expectations and the controls we accept on commercial decisions explicit, so that we can think through the prejudices embedded in our own principles.)

Pride and Prejudice

Jane Austen didn't become famous by writing novels about aspirational young women winning the right to entrust their business ventures to the workers their hearts chose.  But talk about assortive mating!  Imagine the emotional reaction you would have to a Jane Austen story if the characters were as prejudiced with the hiring practices of their estates as they were in choosing their mates.  That sort of discrimination would have been a black mark, and it would lessen our fondness for the characters.  But, how we delight when they marry well.

Outrage and coercion are not about solutions.  They are about identifying outsiders and activities that we exempt from the protection of our principles.

Sunday, January 26, 2014

Minimum Wage - The Bad Luck Policy

I graphed minimum wage rises against the Fed's Yield Curve Recession Model.

Recession probability (left), MW/AW change (right)
1976-1982 shaded because high inflation/multiple MW hikes make binary distinctions about MW behavior difficult

The minimum wage's bad luck is even worse than we thought.  Not only do recessions tend to come after minimum wage hikes, coincidentally.  Not only do larger minimum wage hikes tend to precede lower real GDP growth and lower employment growth....coincidentally.  But, according to the Fed indicator, there were 3 times in the past 50 years when we avoided recessions, even though the probability of a recession topped 20%.  And, believe it or not, the world's unluckiest policy just happens to have coincided with two of them.

Now, I'm kind of joking about the bad luck.  Maybe it's a conspiracy.  This indicator looks ahead one year.  So, the interest rate spread, which is ostensibly managed by the Federal Reserve, that served as the indicator, would have happened a year before.  So, Congress would have known when they implemented these hikes that an economic downturn was possible.  And, unemployment in downturns is widely attributed to nominal rigidity in wages.  Our representatives might possibly be sadists.  (This would explain a lot, now that I think about it.)  And here's the smoking gun - in 2008, the second in a series of hikes was implemented in July.....the month after Congress passed the first Emergency Unemployment Insurance extension because of concerns that laid off workers were having trouble finding jobs quickly.  In July 2008, I wonder how many workers were thinking that walking into their boss's office and demanding a raise would be a good career move.


The hikes in 1961 and 1963 were implemented as the US was coming out of a recession, and managed to avoid this odd series of misfortunes.  But, strangely, even though real GDP was growing at more than 5% per year, the teenage Employment-Population Ratio and Labor Force Participation Rate managed to decline by 3%.  (You'll note that this means the teen unemployment rate remained fairly level during this period.)  Another amazing coincidence.  And, there again in 1967 is this strange coincidence of the teen labor force dropping sharply, right as the minimum wage is raised, during strong RGDP growth.  This policy can't catch a break!

So, there you have it.  Here's your headline:  "Before it hit a 50 year streak of bad timing, the minimum wage was raised twice in the 1960's, where analysts find that it was not associated with a rise in the teen unemployment rate."

PS.  Arindrajit Dube , (referring to this post) , says, reasonably, that we should be skeptical of glibly attributing these patterns in employment to the minimum wage.  Instead we should attribute the teen labor force patterns from 1961 to 1969 to the two short recessions in 1960 and 1970:
Well, here’s a reason why this evidence should be written off as a coincidence.......This (KE: recessions) tends to make the red post-MW trends smaller, providing a very simple explanation for why in 5 out of the 6 cases, the teen employment trend slowed down. And the 6th increase (in September 1961)  happened right after the recession officially ended in February, so one could make an argument that the business cycle can help explain that one too.... In other words, business cycles can explain the pattern of employment trend changes in at least 6 out of the 7 episodes, and maybe even partly in the 7th.
It can be stated with certainty that Arindrajit Dube has forgotten more about minimum wage research than I will ever begin to know, so it's hard to dismiss his point.

In other news, Robin Hanson had an interesting post today.

Friday, January 24, 2014

(Total and) Teen Employment, Minimum Wage Hikes, and Recessions

Added:  I ran the regression with total employment, and found higher significance than I did with teens.  It's added below.  But, it's hard to separate the cause and effect in the aggregate data.


Following up on this, and this, I figured at this point I could do an easy scatterplot or two to see if I could see how much the business cycle affected the changes in the employment trends here.  I figured one would expect to see a range of outcomes centered around a negative mean, with the deviation in the outcomes correlating to the effect of the business cycle - with worse outcomes during recessions.  Usually the correlation isn't so clean in practice.  But, here it looks surprisingly uniform.  And the contraction in teen EPR also has a surprisingly regular relationship with the scale of the MW hike:


Since I have simplified this into just 7 separate observations, I can see how the scale of the trend change relates to both the scale of the minimum wage hike and to the change in real GDP.

The change in the Minimum Wage, as measured, is the change, from the level the month before the first hike to the level after the last hike in the series, in the ratio of the minimum wage to the average wage.

The change in real GDP is measured over the 24 months from 1 year before the initial MW hike to 1 year after.  Declining unemployment tends to lag declining GDP, and this amount of lag seems to have the strongest correlation to the scale of teen EPR in these 7 episodes.

These two independent variables don't have a relationship with each other.

The dependent variable, Teen Employment to Population Ratio, is measured here as the difference between the pre-MW trend slope and the trend slope during the MW episodes, accumulated over 30 months.

Below is the regression of Teen EPR against both of these variables.  I was surprised to see the intercept so close to zero and such high correlations for both variables, together and separately.

Summary for Teen EPR:
MW/AW up by 10% ---> Teen EPR down by 10%
1 year RGDP up by 10% ----> Teen EPR up by 6%
Typical Episode:  MW/AW up 9%, RGDP up 4% ----> Teen EPR down by 6.3%


ADDED:  I decided to go ahead and make the same changes to the 7 observations for the total Employment-Population Ratio.  The coefficients are smaller, as one would expect, but the correlations are higher.  I suspect that this is partly because total EPR is just a cleaner, more stable data series.

Here are the individual correlations.

As with the teen data, the independent variables are not correlated with one another. (edit: Hm...after playing with this correlation, I see that there does tend to be a correlation between lagged RGDP and the size of the MW hike if I delete the 1956 observation.  So, on the one hand, there could be some colinearity here.  On the other hand, declines in RGDP are correlated with the size of the MW hike.  So, either the size of the MW hike is causing an RGDP decline, or, some set of factors is leading Congress to not only implement MW hikes at the end of a recovery, but to make MW hikes larger, coincidentally, when there are larger approaching economic busts.  For the teen data, when I used non-lagged RGDP growth, it lowered the strength of RGDP and the regression, so using the lagged RGDP seemed reasonable.  But, for total employment, using non-lagged RGDP strengthens the regression and the RGDP correlation with employment.  This fits with what I had found earlier, that the unemployment effect in the younger age groups tended to happen earlier and recover earlier than in the older age groups.  In general, it may be harder to capture a clean correlation in the total employment data compared to the teen data, because MW effects on the total labor force would show up in RGDP, while it would be possible for employment changes that were focused on teens to have relatively little effect on the aggregate numbers.)

Below is the comparison of actual EPR changes during the 30 month periods of the 7 minimum wage episodes to the levels predicted by the regression coefficients.

And, below that are the regression statistics.  The F-stat is .047, and the p-values for the two coefficients are .107 for the minimum wage hike and .069 for RGDP.  This is excel output, which is based on 2-tailed probabilities.
Here is the data:
Emp trend change*30 months
MW/AW%ch
2 yr RGDP
1956 -4.7% 13.0% 6.2%
1961 1.0% 7.5% 8.8%
1967 -1.1% 9.4% 6.6%
1974 -3.9% 10.0% -2.0%
1990 -3.7% 7.5% 1.9%
1996 0.7% 5.8% 8.7%
2007 -5.4% 9.4% 2.0%




Summary for Aggregate EPR:
MW/AW up by 10% ---> EPR down by 5.6%
1 year RGDP up by 10% ----> EPR up by 3.9%
Typical Episode:  MW/AW up 9%, RGDP up 4% ----> EPR down by 2.7%


PS.  Here is the correlation between the two independent variables.

The RGDP with a -1yr lag is what I used in the regressions.  But I noticed that when I removed 1956 (the blue dot at the top right), the correlation looked similar to the correlation between the change in MW and the concurrent change in RGDP (in red).  There is a slight correlation - there is actually just under a 10% p-value on a one-tailed test.

But, if the issue with minimum wage job losses was just a matter of bad timing, then, on this correlation, the trendline would be flat, but would be lower than the average 2 year RGDP growth (which is about 6.3%).  What we see instead are regression lines that start out at about the average 2 year RGDP growth rate and then decline steeply as the scale of the minimum wage hike increases.  This suggests that the declining RGDP is a product of the MW hikes.

The regression for EPR becomes very strong when I use the concurrent RGDP.  The MW variable doesn't improve the significance, but the concurrent RGDP is very likely to be signifying some employment loss from the minimum wage.  While the variance in the regression between the two independent variables is large, the effect is also large, so that the expected difference in 2 year concurrent RGDP between the smallest MW hike and the largest is more than 8%.

Wednesday, January 22, 2014

Some Great Graphs on Labor Force Participation

I wish I'd thought of these.

From the Federal Reserve of Atlanta. (via Pinetree Economics, with great added commentary)

Here, just to wet your whistle:
140117a


Excellent work.

PS.  OK.  I'm done.  With this incredible interactive chart, there is nothing left to be said about LFP.

Review of 60 Years of the Minimum Wage

(Added: I have an additional post with more on the effect of RGDP growth on this relationship.)

Thanks, everyone for the comments on this post.  I have addressed some issues in the comments there, and added some things in the body of that post.

Some commenters mentioned that using an employment rate instead of total employment might help to get rid of noise due to population shifts, etc.  That's a great idea.  It's funny how when you're working with data, you don't think of some obvious things.

Many comments have mentioned the effects of the business cycle.  Like many of them, I had kind of eyeballed it, and thought there was an unusually high coincidence of recessions following minimum wage hikes.

The comments spurred me to get more specific, and I think, in general, the coincidence isn't as strong as it seems like at first glance.  Here are all of the individual graphs from the first post, with the following changes:

1) I changed from teen employment level to teen Employment to Population Ratio (EPR).
2) I added  recession bars and also an inverted unemployment rate (for all ages) as signifiers of recession effects.



1956 - a recession coincides with a turndown in EPR, but only for the last few months of the MW period.

1961 - this is coming out of a recession, so there is a downtrend even though the economy should be recovering.  There is a recovery in teen EPR after the MW hikes have passed

1967 - no recession during the MW related downturn.  Then a brief recovery in teen EPR before a recession that follows

1974 - this MW hike appears to coincide precisely with a recession, so it is difficult to separate the signature of each effect.  However, teen EPR begins declining before unemployment.  EPR is not usually a leading indicator in recessions.

1990 - the recession begins 4 months after the MW hike.  However teen EPR begins declining in the months preceding the MW hike.  EPR is not usually a leading indicator in recessions.

1996 - no recession, no MW related decline

2007 - the recession begins 6 months after the MW hike.  However, teen EPR begins declining in the months preceding the MW hike.  EPR is not usually a leading indicator in recessions.

The Tally
7 Episodes:

1 - no MW decline

1 -  MW decline not discernible from concurrent recessionary effects, but does have signature of pre-MW downward shift in teen EPR

2 - MW decline precedes recession, with signature of pre-MW downward shift in teen EPR

2 - MW decline with signature pre-MW downward shift in teen EPR, no recessions near MW inception

1 - MW decline, in spite of economy in recovery phase after a recession


PS.  The recessions in 1957, 1960, and 1969 are not concurrent with initial MW hikes, but they appear in these graphs.  Note that the fall in the teen EPR does not lead the rise in the unemployment rate in these cases.


Here is the graph for the entire period.  Full tally of 10 recession and 7 MW episodes:
3 Concurrent recessions & MW, declining trend
7 recessions without MW, declining trend
3 MW without recessions, declining trend
1 MW without recession, climbing trend
All periods with no MW or recession, climbing or accelerating trend

Sunday, January 19, 2014

Could the Taper of QE3 Be Stimulative?

If we start with long term total bank assets:
FRED Graph

We see a fairly stable growth, with some slowdown during recessions.  There was a peak in 2008 relating to Fed liquidity maneuvers, including interest on reserves (IOR).

First, I want to propose that QE2 and QE3 may have had little effect on this growth rate.  My reasoning here is that with IOR and very low interest rates, reserves and treasuries are basically perfect substitutes.  If the main result of QE is to add reserves to the banking system, this should not affect total bank assets, because the banks should simply swap reserves with other low risk assets to suit their needs within the level of assets they already were intending to hold.

Here are some major portions of bank balance sheets, as a percentage of total bank assets:


FRED Graph
Real Estate Loans, Commercial and Industrial Loans, Treasury and Agency Securities, Cash
Treasury Securities and Commercial and Industrial Loans show typical cyclical behavior.  Reserves appears to have substituted only a very small amount for treasuries and agency securities.  The asset that most of the excess reserves appears to have replaced over the period as a whole is real estate loans.

There has been a huge increase in all-cash and institutional buyers in the housing market.  The net effect of QE appears to be that the Fed is buying treasuries and agency debt from non-banks.  Those treasury and MBS sellers, on net, are replacing their would-be treasury and MBS holdings with residential real estate, which is now being funded outside the banks instead of through mortgages.  And the cash ends up in the banks.  So, to a large extent, once the QE funds end up as bank reserves, they appear to have mostly served to move real estate assets from bank balance sheets to non-bank balance sheets.
 
But, I think there is a difference in character between QE2 and QE3.  Here is a chart of the change, in dollars, of various bank assets, compared to Fed Securities holdings:
 
The banks have absorbed the reserves with seemingly little adjustment, when viewed in absolute dollars.  Total low-risk assets (red line) move in concert with Fed purchases (grey background).

When we subtract cash, treasuries and agency securities from bank assets, the net bank asset level bottomed out during QE2, and then increased during the period between QE2 and QE3.  At the same time, Commercial and Industrial Loans bottomed out, and also started its cyclical recovery.
 
Before QE2, bank assets were stagnant and banks were reducing private credit in exchange for more treasuries.  These items leveled out during QE2 and total assets grew at roughly the rate of new reserves coming in from Fed actions.  After QE2, total bank assets continued to grow, and it appeared that bank assets, net of cash, treasuries, and agency debt, were starting to grow, but, this growth stopped when QE3 began.  For about 15 months between QE2 and QE3, bank assets increased by about $500 billion, and this was with a flat level of reserves + treasuries + agency debt.  In the 15 months since QE3 began, bank assets increased by about $1.2 trillion, but it has all gone to reserves.
 
QE3 has some weak inflationary effects, but I wonder if we have reached a point where reserves are crowding out bank balance sheets in a way that is stifling the development of private credit.  With that in mind, here is a graph comparing Commercial and Industrial Loans as a percent of Total Assets to Commercial and Industrial Loans as a Percent of (Total Assets less Cash).
FRED Graph
 
Here is a graph of Commercial and Industrial Loans since Dec. 2010:
 
Could the taper actually help free up bank lending?  Is it possible that one of the stimulative effects of QE1 and QE2 was that private markets were willing to trade some of the treasuries for higher risk assets while the banks were still trying to add low risk assets but were unable to add to real estate loans?  If, by QE3, banks are more willing to increase credit risk exposure, but are capital constrained by the inflow of cash, is it possible that the net effect of QE3 (by moving investments out of the banks and into non-banks) is inconclusive?  On the other hand, will banks be able to recapture some of the mortgage market if Fed money stops flowing into housing through non-bank sectors?  I think it might be worth watching these measures move as QE3 is stepped down.
 
To the extent that QEs are shifting investment from commercial banks to non-banks, does anyone have a handle on the net effect this shift has on economic development, risk premiums, etc.?  If you have read this far, and you have banking expertise, your fee for reading this blog is to comment here and tell me if I'm making any stupid errors in my analysis.

Side Note:
This does offer a caveat to my interpretation of the 2000's housing boom.  I believe the housing boom can generally be explained with reasonable market behavior, with homes behaving as very long term inflation-protected bonds.  One piece of evidence in my narrative has been the bullish real estate market, post-2009, which has happened in the midst of a sclerotic real estate banking sector.  But, if we treat the cash inflow from the QE's as a substitute for the bank mortgage market, then funding markets for real estate aren't as dead as they appear.

FRED GraphHere is a Case-Shiller Index for the period.  Home prices seem to have accelerated during QE's that included MBS purchases.  They didn't accelerate during QE2, which didn't include MBS purchases.  I don't believe that minor liquidity effects on mortgage rates could have this much effect.  I think it's more likely that when the Fed entered the MBS market, the MBS investors that they replaced used direct cash real estate investments as a substitute for some of the MBS assets that the Fed was buying away from them.  I still think there is room for home prices to rise over the next decade or so in an environment of low real long term rates.  I see this price reaction as a result of a dead banking system that is undermining traditional demand in housing, so that prices are below the equilibrium price that would come from healthy supply & demand with a healthy credit sector.  QE has allowed non-bank investors to capture the gains from that disequilibrium.

If tapering Fed purchases of MBS means that, on net, there is a small disinflation effect, less money flows into real estate, and banks increase their Commercial and Industrial Loans, that doesn't seem like a bad trade to me at this point.  But if the banks can't take over funding of home financing again, then maybe the end of QE will again lead to disinflation.

PS:  I must not be obviously wrong.  Soberlook has the same idea.
 

Friday, January 17, 2014

Long Duration Unemployment As We Begin 2014

Here is a chart of unemployment, by duration, each December, for the past two recessions.  Most of the damage from labor contractions comes, not from layoffs, but from a seizing up of labor churn.  Thus, there is only a slight increase in short-term unemployment at the height of the contraction.  Most of the excessive unemployment comes from increasing long-duration unemployment, due to a lack of new hiring.
 
In both of the last two recessions, long-duration unemployment didn't peak until 2 years after the peak in short-duration unemployment.  This is typical.  The unusual issue with this contraction has been the enormous size of long-duration unemployment compared to short-duration.
 
I have fingered Emergency Unemployment Insurance (EUI) for much of this anomaly.  Now, if the termination of EUI holds, I would expect to see some downward acceleration in the long-term unemployment rate.
 
However, it will probably be a long unwinding process, in any case.  In this graph, I compare the Unemployment Over 14 Weeks to the Unemployment Over 26 weeks with a 3 month lag.  This gives an estimate of the proportion of long term unemployed workers who are leaving unemployment every 3 months.  In a healthy economy, this is around 45-50%.
Here is a longer version of this measure, just for kicks.  The recent low turnover out of long-duration unemployment was clearly without precedent.  There is a caveat.  My concern might be overblown.  There is a clear long-term trend toward lower turnover in long-duration unemployment.  The trend at the peaks has declined by about 15% since the 1970's.  If we draw a similar trendline at the troughs, the current low level of long-duration unemployment would not be significantly below that trend.  So, it is possible that some of the seemingly unusual long-term unemployment comes from difficult comparisons to the previous two contractions, where long-duration unemployment churn was higher than normal, but was not immediately noticeable because it was fighting this long-term downward trend.
 
In any case, I would expect to see an acceleration of turnover in the long-duration unemployment measure with the termination of EUI.  But, even if I forecast a rebound of turnover to 40% by next quarter and 45% for the remainder of the year, the unemployment rate for duration over 26 weeks would look like this over the course of the year:
 
As can be seen in the first graph, there isn't much unemployment left to lose among the short-duration unemployed.  So, even if we are optimistic about a rebound in long-duration unemployment, an unemployment rate at 5.5% by the end of 2014 is probably the lower bound on what we could expect.  I don't know.  I guess we'd all be pretty happy with that.
 
A Side Note:
 
Here is the unemployment duration for the past 8 months:
 
We can see the consistent decline across durations that has led to the healthy decline in the overall unemployment rate.  The high short-duration unemployment in October was probably due to the government shutdown.  The low short-duration unemployment in December seems anomalous.  If that rebounds in January, we could see a pop back to 6.9%.  This is a very low level for short-duration unemployment.  On the other hand, long-duration unemployment dipped strongly in December.  This might have been in anticipation of the end of EUI.  I am looking for this to continue to fall by about 200,000 per month (reflecting the improved turnover discussed above.)
 
 

 

Thursday, January 16, 2014

More Evidence of Labor Market Strength, updated

Here is a graph comparing the unemployment rate to real wage growth:

There are several items I find interesting in this graph.

The Main Point:

Nominal wage growth reflects inflation, which is really a reflection of Fed policy, and real wage growth, which is a reflection of the myriad of influences on labor markets.  Wage growth is currently low, by historical standards:

FRED GraphThis is mostly the product of tight Fed policy and low inflation.  Since there is a widely held misperception that the Fed is being loose, I think this is a point that is being missed by most observers.

If we adjust for inflation, so that we are looking at real wage growth, we see that wage growth is quite strong.

That top graph reinforces my point that it is the unemployment rate that is giving us a false signal right now.  Real wage growth this strong reflects a labor market that would normally be at 5%.  If Congress doesn't reinstate EUI, I expect to see continued downward acceleration in the unemployment rate.

The question that remains is, will we see continued labor strength that pushes unemployment down to 5% in short order, or have there been a large enough number of frictions added to the labor market that unemployment will bottom out at 5.5% or 6%, like we saw in the 1970's.

Other Notes:

You can see the anomalous spike in real earnings growth in 2008.  This is a sign of how badly the Fed messed up during that time.  Late 2008 saw a deflationary collapse that doesn't show up anywhere else in the last 50 years' experience.  (The other spikes in real wages were during expansions.)  During this contraction, we were dealing with an unprecedented sticky wage issue, which has taken an excrutiatingly long time to work out of the system because the Fed has continued to keep inflation low.

It looks like, frequently, real wage growth spikes and then collapses in the period before a contraction.  I don't quite understand the underlying factors there, but it shows up enough here to warrant a viewing when it becomes time to forecast the next contraction. It appears to be part of a pattern where nominal and real wages grow during the mature portion of the expansion, and as the next contraction approaches, nominal wage growth is mostly inflation-related.

PS:  The PCE price index, less Energy and Food, appears to have the tightest pattern with the unemployment rate.  Here is a scatterplot showing the correlation.  The relationship has been tracking with the trend, but at an elevated unemployment rate.  I think that gap will close after we normalize the unemployment insurance policy.

Considering that one of the contributing problems during shocks to the labor market is the tendency for wage rigidity to lead to unemployment, it is probably not a coincidence that a labor crisis that has been worse than usual has seen real wage growth at levels far above the trendline for the given unemployment level.  Policies that have held wages too high have probably not been helpful in normalizing the economy.  Deflation followed by an extended period of low inflation has worsened this problem.  EUI worsened this problem.

The fact that real wages are trending higher suggests that much of the damage from tight monetary policy is a sunk cost at this point.  If we can keep EUI from being reinstated, I will be watching to see if that helps get us most of the way back to normal.  If the Fed stumbles into a moderate amount of inflation along the way, that probably wouldn't hurt, either.
x=YOY growth in real wages (AHETPI/PCEPILFE)


Teen Employment and the Minimum Wage, 60 years of experience

(Updated with trendlines at bottom chart, and better trendlines on top charts).
(Updated again to show the bottom chart in Employment to Population Ratio terms instead of total employment)

Added: For those visiting from Matthew Yglesias' Slate.com post, (1) I use teen employment because the higher % of teens at MW makes the trends show up more clearly, but I find the same trends in the total employment numbers, and (2) it's important to consider the difference between an event-based analysis and long-term trends.

And, for everyone, here is an updated post that adds recessions to the graphs. And a post with a regression of teen EPR against RGDP and the scale of MW hikes.
And here is my last post on the topic for now.


Here are the teen employment trends during all 7 of the series of minimum wage hikes (I exclude the high inflation period in the late 1970's, which included several nominal hikes that roughly matched inflation).



In every episode, except 1996 (which is the smallest hike relative to average wages), there was a distinct decline in the trend of teen employment, over the period of time covering from a few months before the initial hike until a few months after the follow-up hike.

Here is the entire period:
Pre-trend lines are for period from 27 to 3 months before MW hike.
MW Trend lines are for period from 3 month before to 27 months after initial MW hike.
[Y axis = Total teen employment (000's)]
 
ADDED:  Here is the graph in terms of the Employment to Population Ratio, which is probably a better way to do it.  The basic story is the same, though.  After 30 months, EPR, on average, is about 6% lower than it would have been based on the previous trend.  Using EPR does help to see that the scale of the effect is probably inflated by double counting, because the positive trends coming into some MW episodes probably reflect some amount of rebounding from the previous shock to teen employment.
Pre-trend lines are for period from 27 to 3 months before MW hike.
MW Trend lines are for period from 3 month before to 27 months after initial MW hike.
[Y axis = Teen Employment To Population Ratio]

Every kink down in the trend comes in the few months leading up to a minimum wage hike, except for a brief decline in 1969 and the contraction beginning in 2000. (The period from 1976 to 1982 is the period where frequent minimum wage hikes and high inflation rates make it hard to distinguish between categories.)

Is there any other issue where the data conforms so strongly to basic economic intuition, and yet is widely written off as a coincidence?

Wednesday, January 15, 2014

Minimum Wage and Labor Force Participation

There appears to be some relationship between nations with higher minimum wage levels and nations with lower labor force participation.

Here is a graph comparing the level of the minimum wage and labor force participation in the US over time:

I don't want to make too much of this.  There are so many factors that would influence LFP over time.  For instance, the growth in LFP in the 1970's and 1980's was mostly attributable to cultural changes in female labor participation.

But, even in that case, the reduction in the real level of the minimum wage must have facilitated the entry of unskilled women into the labor force.  Early minimum wage legislation had been applied only to women and children, sometimes with the aim of removing women from the labor force.

Considering the hand-wringing over voluntary decreases in labor force participation related to aging, proposals to raise the minimum wage back to around 50% of the average wage or more could reduce LFP by an additional 1% - 2%.  Low labor force participation is a common problem for marginalized populations, so the disinterest in this concern among minimum wage proponents is odd.

Tuesday, January 14, 2014

December 2013 Employment Report

I was surprised to see interest rates fall back on the release of the December Employment Report.  The markets seem to be discounting the 6.7% unemployment rate because of the low payroll jobs number and the drop in labor force participation.  All of these numbers have a lot of noise.  I would say that in the current environment, the unemployment rate is as relevant as the other two numbers.  In any case, I would have expected it to be kind of a wash.  But other analysts seem to share a consensus view that it was a poor report.

Here is a look at the trend in Labor Force Participation (LFP) and the Employment to Population Ratio (EPR).  I have also added the long term Labor Force Trend and the Employment Ratio that would produce a 6.0% unemployment rate (UER):


Even though the drop in the UER was due to a drop in the LFP rate, the EPR held strong.  The current trends in LFP and EPR would put the UER at 6.0% by early 2015, even if we consider the December report an aberration.

The labor market still appears weak even at that point in 2015, with an LFP below trend, but this is not unusual.  Here are graphs of the same measures during the 1990 and 2000 downturns.

The UER hit 6% in mid-1994, and LFP didn't recover to trend until early 1997.  The 2000 downturn was much more mild, with the UER topping out at about 6%.  But, even there, the UER started dropping below 6% by the end of 2003, while LFP stayed at the trend until mid-2005.

If we hit 6.0% UER this year, an LFP finally pulling back to trend in 2017, at a rate similar to today's rate of about 63% would be consistent with the current demographic trend and with the experience of previous downturns.

I think there is a case to be made for a more optimistic outlook.  I have posted previously on the three factors that I believe have had the largest impact on these labor market measures in the past 5 years - demographics (seen in the LFP trendlines here), minimum wage (which I believe reduced LFP by more than 1% at its peak, and still is responsible for most of the movement of LFP below trend), and Emergency Unemployment Insurance (EUI).

I have estimated that EUI added more than 1% to the UER and subtracted added about 1/4% from to LFP.  But, we now have two instances (North Carolina in June, and the U.S. currently) where LFP has dropped precipitously in the months leading up to the termination of the policy.  So, possibly EUI had more of an effect on LFP than I had expected.  I don't have a comprehensive narrative understanding of how EUI and LFP are interacting, but at a basic level, the tendency of EUI to inflate LFP should be obvious.

I would suggest that, to the extent that some of this movement is related to EUI, we should consider the new LFP level to be the natural level, and the LFP level from the past few years to be elevated.*  I realize that with all the LFP doomsaying, this seems like an odd thing to say, but I've discussed the LFP issue many times before.  The doomsaying is wrong.

The other prediction that would follow from the termination of EUI is an uptick in the EPR.  (North Carolina also saw a decrease in regular unemployment claims after June, but that could be related to other cuts they made in the program.)

In addition to expecting these changes, I also would expect to see LFP flatten out and start to slowly close in on the descending LFP trend line.  Part of this is simply strength that should eventually come from an improving labor market.  But, I think the distance we are getting from the minimum wage increases will help this rebound.  In 2007, 1.1% of the labor force worked at or below minimum wage.  After the 3 hikes from 2007 to 2009, that topped out at nearly 3%.  It is back around 2% now.  As this slowly falls back toward 1%, I would expect a rebound of more than 1/2% in LFP, relative to trend.  I doubt that we'll see LFP ever get much above 63%, though.

Taking all of these factors into account, and adjusting our forward expectations, an UER under 6.0% by the end of summer 2014 and a LFP back above trend in 2017 are feasible.

Of course, a new round of minimum wage hikes or a hawkish monetary freak out by the Fed would spoil these expectations.  Until either of these things happen, I continue to believe that forward interest rates in the 2 to 3 year range are about 0.5% below the expected eventual expiration levels, and they will generally bounce around a stable range as time passes.

* Addendum: Except for the long duration issue, the labor market is at healthy levels.  Continuing and Initial claims, and job openings, are all at levels associated with an UER of 5%.  While hiring is still at a level that would have previously been associated with UER over 6%, quits have returned to the level where they were when the UER previously topped out at 6%.  If, after the termination of EUI, we have an UER under 6% within a few months, this would be more in line with the other indicators.  I think we could expect to see a related increase in hiring and quits associated with the resulting reduction in labor market frictions as well.  I suspect that if that comes to pass, there will still be a bearish mood about the labor market, because while LFP might still be 1% below trend, there will still be a large contingent claiming that it is 3% too low.  I don't know what it will take to quiet that contingent, because no employment boom is ever going to get LFP back to 66%.

Sunday, January 12, 2014

Part Time Employment

Scott Sumner notes that the U-6 unemployment indicator (which includes part time workers who would prefer to work full time) seems to be stuck at an elevated level.  The other indicators seem to be moving in a more correlated direction.

I updated some data from the BLS concerning part time work, and found this:


This is full time employment (blue, left) and part time employment (red, right) as a percentage of the labor force. (The bump in 1994 is due to a change in method.)  There is a pretty stable long-term level of part-time employment that was not very cyclical, until 2008.

The next graph separates total part time employment (green line) between part time employment for non-economic reasons (red) and part time employment for economic reasons (blue).

Here we see that in 2008, there was a 2% jump in part time employment.  This appears to result from a shift from full time to part time employment for about 3% of the labor force (this has rebounded by about 1%).  This movement into part time employment displaced about 1% of the labor force that had been working part time for non-economic reasons.  These former workers could have either become unemployed or they could have left the labor force.

This suggests that some of the decline in the labor force could be the result of unemployment among workers who were previously marginally employed.

I have been forecasting a strong continuing fall in the unemployment rate, but this does suggest that there is some employment slack that is being "hidden" by an unusual amount of buffering within the part time employment space.  Some rebound in employment in 2014 could result in a reshifting of these workers from part time back to full time employment.  Inflationary pressures that I might be expecting from a tightening labor market may not be quite as close as I have suspected.

The U-3 Unemployment rate is at 6.7%, about 1 3/4% above a healthy rate of around 5%.  But, because of this unusual shift into part time employment, the full time unemployment rate remains about 3% above the healthy range.

Looking at causes, I have zoomed in on the above graphs for these two graphs.  What we can see is that the shift happened entirely in late 2008 and early 2009.  I think this is a time frame that is too early to blame on Obamacare, and there isn't anything about the patterns in this data series that points to the minimum wage as the culprit.


Why is this pattern unique to this cycle?  Maybe it's just a product of the depth of the employment crisis this time.  If that is the case, and full time employment is now back in the range of the bottoms of several previous cycles, we will finally see a snap-back in part time employment.  I don't have a ready intuition for how this will play out as an influence on interest rates, inflation, total employment, or GDP.