The frequency of economic statistics matters at turning points
by Rebecca Wilder
How are the data presented? At an annual, quarterly, monthly, or weekly frequency? At the onset of the New Year, you will undoubtedly see many charts illustrating records broken in 2009 using annual measures. This is always fun (from a data junkie’s point of view), but it only tells the reader where we were, on average in many cases, rather than where we are now! Quarterly data are the same story – often presented well after the culmination of the period.
Alternatively, monthly data are a little more telling but still lagged by at least one month. For example, the employment report is the first major economic release of the month, which sets the stage for many subsequent releases. However, by the release date, generally the first Friday of the month, the survey information is already one month old.
This leaves weekly, or even daily, data. High-frequency data can tell us “where we are now”, but are subject to substantial volatility. Nevertheless, high-frequency data are quite informative at economic turning points. So where are we?
Here are three high-frequency indicators that show an improving labor market, as illustrated by initial claims and daily tax receipts. However, the money multipliers remain at historically low levels, signaling that consumer spending and credit growth continues to elude monetary policymakers.
The weekly initial claimant count is dropping off quickly. So far, the 4-week moving average is 33% off its peak, a definite positive. And comparing to previous recoveries, the claimant count does suggest that this recovery will look more like a job-plus, rather than a job-less recovery.
However, don’t get too excited – an awful lot of jobs need to be created each month just to drop the unemployment rate.
The stabilization of the labor market is likewise seen in the Treasury’s daily tax receipts. Daily receipts have stabilized, and are now growing, off of their lows.
High-frequency monetary aggregate indicators show that traditional Fed policy – increasing bank reserves through open market operations – is not flowing into the economy as new money for spending on goods and services. Money multipliers of all types are half of what they were just two years ago (dropping even lower in recent months).
This is the bane of the Fed’s policy existence during and in the aftermath of the banking crisis. Inflation is not going to be a problem until this money clog frees up.
There is widespread stabilization, and even improvements, as shown by the high-frequency economic data. Perhaps I will follow up this post on the remaining weekly data, like on credit extension and housing.
Rebecca Wilder
No great argument. Three minor points:
1. For projection, we probably want to be looking at “leading indicators” not so much at frequently updated indicators
.
2. January is an especially awful time to be trying to make sense from statistical data. Seasonal adjustments in November, December and January are going to incorporate assumptions about holiday activity that are not especially true in deeply troubled economic times.
3. As Krugman, Roubini, et al point out at every opportunity, this is not a “normal” recession caused by excesses in the equity markets. It is a very deep recession caused by excesses in the financial and real property markets and accompanied by something strongly resembling a liquidity trap. Unfortunately, data on the Great Depression 80 years ago is spotty so the only available data from a somewhat similar situation is Japan in 1990. American economists (Krugman excepted) prefer to pretend that Japan didn’t exist, doesn’t exist, and couldn’t exist.
The employment situation looks like the period of job destruction being the problem is over.
Now the questions becomes, when do we get job creation.
Claims addresses the question of job destruction but does little to measure job creation.
Maybe the better measure of that is the ISM jobs survey.
Spencer, claims went up again, and we are hearing more talk about a double dip recession. Dunno, things better turn around soon.
Hi VtCodger,
You are right about projection – but I am trying to assess the current state of the economy. And that is consistent with improvement.
I agree – not seasonally adjusted data is likely more telling at this juncture of the business cycle.
Isn’t every recession “different”? I would argue that this is different, not because the US saw a massive banking crisis, but because the US saw a massive banking crisis and the dollar surged. That alone negates much of the history on banking crises (even during the Depression, since the central bank defended the value of the pegged dollar).
Thanks! Rebecca
Hi Spencer,
Completely agreed. However, history does suggest that when the job destruction plummets, the payroll grows fast enough to drop the unemployment rate. On a monthly basis, the ISM employment had been improving and above 50.
The even bigger question above that is, is the job creation sufficient enough to dent the unemployment rate? There are plenty of reasons that suggest job creation is anything but sluggish, one of which might definitely be the uncertainty in Washington (health care, financial reform, etc.). But, of course, we will see.
You made a very astute comment some time ago (my interpretation): economic forecasts tend to undershoot the magnitude of recoveries.
Rebecca
The real question is what will bring about job creation.
C R E D I T A V A I L A B I L I T Y is the key.
You’re correct, but as CR points out, the weekly data is noisy. You can’t read too much into any given number unless the change is horrendously large. The four week moving average is also up a bit, but not scarily so. I’d guess that the really scary stuff is over — there aren’t that many jobs left to lose in construction … or, sadly, in manufacturing.
http://www.calculatedriskblog.com/2010/01/weekly-initial-unemployment-claims_21.html
The question would be where the new jobs are supposed to come from. Retail? probably not. Energy? probably not. Agriculture? Maybe, but I wouldn’t bet on it. Health care? We’re supposed to be controlling costs, not hiring more overhead. Construction? You’re kidding, right? Manufacturing? Maybe ,,, somehow … but it’s gonna take a near miracle. State and local government? They won’t be hiring until 2013 or later.
Here’s an interesting and decidedly on-toppic comment from the comments at CR on the unemployment numbers
“Yalt (profile) wrote (in reply to…) on Thu, 1/21/2010 – 9:58 am
reply
No, I take that back. We had a similar pattern last year, with the heavy volume of claims causing a backlog, pushing the EOY bump back and skewing the seasonal adjustment, which is still primarily based on lower-volume years without the lag.
Here’s last year (SA):
12/6 552k
12/13 552k
12/20 564k
12/27 508k
1/3 488k
1/10 535k
1/17 575k
1/24 590k
1/31 624k
…and we then stayed above 600k through June.
Note the drop for the two holiday weeks. The seasonal adjustment expected a rush of claims, and indeed the three weeks from 1/3 to 1/17 were the three highest-volume weeks we’ve ever seen, unadjusted. But UE offices were at full capacity and they weren’t able to get the claims processed as quickly as usual. The adjustment expected heavy volume in 12/27 and 1/3, not 1/10 and 1/17.
I think you can make a case that it’s the good numbers we’ve seen the last three weeks that are the aberration, not today’s number.”
(and yes that’s way more than Fair Use allows, but I doubt the author will mind. Personally I think our Copyright laws are stupid, unenforcable, and ultimately doomed.)
Now we’ll see if I can conjure up a direct link to the comment. Nope … apparently I can’t get there from here At least not with basic HTML skills. I won’t bother anyone with my comments on modern web page design. I doubt anyone would care outside of a small circle of friends.
CoRev,
Claims also went up because of the extension to unemployment benefits.
Movie Guy,
Credit availability for some classes of borrowers might be a problem, but the larger problem is weak aggegate demand for credit. The Fed cannot lower the nominal interest rate below zero, and the NAIRU clearing interest rate is around negative 5 percent. We need to punch up aggregate demand. In the short run that means more aggregate demand from the govt. Over the longer run it means more private investment demand.
Rebecca,
“…traditional Fed policy – increasing bank reserves through open market operations – is not flowing into the economy as new money for spending on goods and services. Money multipliers of all types are half of what they were just two years ago (dropping even lower in recent months).”
But don’t you know…AB poster cantab assures us that we are not in a liquidity trap.
High frequency data also gives you the option of using spectral analysis techniques in lieu of more conventional time series analysis approaches.
Slugs,
Now cactus is complaining that I’m being unfair when I ask him to provide a source to expert opinion to back up his conclusion from his analysis. I would like to ask you for the same thing.
I’ve already laid out a markef. Find a statement by Ben Bernacke saying we’re in a liquidity trap and you win the point. Find a statement saying the same by anyone on the FED’s board of governors and I would say you have very strong evidence. After that come up with a frist rate economist that specializes in monetary economics that says were in a liquidity trap and i’ll agree that you have some very good evidence. Quote Krugman and I’ll laugh at you since we all know that Krugman just wants government spending any way he can get it and the liquidity trap lie is as good as any for this snake oil salesman.
Cantab,
You must have missed it because I did give you a Bernanke quote a week or so ago. It was his 1999 paper in which he defined a liquidity trap and said that conventional monetary open market operations were ineffective. That was his rationale for quantitative easing. I gave you the link to his paper. I also gave you a link to a JSTOR paper that defined liquidity trap. The definition is when the NAIRU clearing interest rate is negative.
state and local governments have just begun their layoff processes . . . .
Slugs,
So I take it that you can’t find a statement by anyone on the FED including Bernake who says we are in a liquidity trap.
Cantab,
Bernanke says that we are in a liquidity trap. That’s why he advocates quantitative easing….it’s his response to a liquidity trap. Oh…the Dallas Fed also says we’re in a liquidity trap.
Quantitative easing is the Fed’s vehicle for making the central bank relevant in a liquidity trap. It’s an unproven theory, but Bernanke’s argument is that creative schemes for flooding the banks with mountains of cash could have the same practical effect of a negative nominal interest rate. Also, Bernanke hints that a whiff of the threat of inflation could break the back of a liquidity constraint.
Cantab,
Bernanke says that we are in a liquidity trap. That’s why he advocates quantitative easing….it’s his response to a liquidity trap. Oh…the Dallas Fed also says we’re in a liquidity trap.
If the FED did not do quantitative easing would interest rates go up. You know they would. So much for the liquidity trap argument.
Would love to see the last graph going back farther in time (like, back to 1980 or before). I have no idea if the 9s and 10s of 9/08 are historically high, low, or… I could look it up and graph it myself, but…
Hi Steve, Contact me directly (newsneconomics@gmail.com), and I will send it to you!
Rebecca
***High frequency data also gives you the option of using spectral analysis techniques in lieu of more conventional time series analysis approaches. ***
As far as I know, there is nothing that prevents you from doing a Fourier transform or something more or less equivalent on any time series no matter what the frequency. All more frequent sampling does is lower the frequency of the shortest term components you can extract (Nyquist limit). But surely, we are looking for long term trends here, so losing the short term stuff should not be a problem.
What is a problem is that although Fourier guarantees that you will get an answer and that the answer will fit the past data, it doesn’t guarantee that it will predict the future worth damn.
Haven’t people been trying to spectrum analyze the equity markets for many decades with little success?
My guess is that a spectral analysis of either the weekly or monthly data will give you an annual component and a lot of meaningless “noise”.
But I don’t actually know anything about this. Am I wrong?
On further thought, this is seasonally corrected data(?) so maybe no annual component.
VtCodger,
But surely, we are looking for long term trends here, so losing the short term stuff should not be a problem.
I thought the point of Rebecca’s post was to find turning points rather than long term trends.
In the real world people don’t really bother with frequency domain techniques unless they have high frequency data. As a practical matter it just isn’t done.
cantab,
No, nominal interest rates would not go up. The reason is that the NAIRU clearing equilibrium point would continue to fall. With nominal rates stuck at zero the gap would widen. The effect is to increase the real interest rate. And I think that’s where your thinking has gone off track. You’re thinking in terms of the real rate increasing when the Fed reduces the money supply, and you’re thinking that not doing quantitative easy is sort of like contracting the money supply, so interest rates ought to go up. Am I right? Is that what you had in mind? The key to getting back on the right track is to keep in mind that the nominal rate is stuck at zero, and as investment demand falls off you need an ever lowering nominal rate. But you can’t get there. The effect is to widen the gap between the NAIRU clearing rate and the nominal rate, which is effectively an increase in the real interest rate. Understand?