Why doesn’t the government publish monthly GDP data?
A recent David Beckworth post linked to Macroeconomic Adviser’s monthly nominal GDP series. Please click on the Macro Adviser’s GDP link and take a look at the graph showing estimated monthly NGDP data. You will see that almost the entire drop in NGDP during this recession occurred during a brief 6 month period between June and December 2008. This fact is obscured by quarterly data, which show a very large drop in the average level of NGDP between 2008:4 and 2009:1. But again, by December 2008 it was almost all over, even though the official recession trough wouldn’t occur until June 2009, at a tad below December NGDP levels.
[Update, 10/25/10: I erred in telling you to look at the graph, which shows RGDP. The tab at the bottom opens another page which shows the nominal monthly data. This is even better—all of the decline in NGDP now occurs between June and December 2008, and almost all between July and December. This almost precisely aligns with the period of ultra-tight money.]
I wish the Federal government would publish monthly NGDP data. It can’t be that hard to get estimates; after all, GDP is mostly constructed out of other monthly time series. And even the quarterly GDP reports are estimates, often sharply revised years after the fact.
The period of June to December 2008 also saw one of the tightest monetary policies in American history. Between July and late November the real interest rate on 5 year TIPS soared from about 0.5% to 4.2%. The dollar soared against the euro. Stock, commodity, and commercial real estate prices plunged.
And all of this occurred before the Fed ran out of conventional monetary policy ammo. That’s right, it wasn’t until mid-December 2008 when the Fed reduced rates close to the zero bound (0.25%.) During the crash rates were always at least 1%, and mostly 2%. And that’s ignoring the contractionary impact of interest on reserves.
Frequent commenter “123” has studied this period more intensively than I have, and he recently sent me a post that sheds new light on the debate over whether the key problem was solvency or liquidity:
Brad DeLong ponders the issue of the root cause of the crisis. Is it that the full-employment planned demand for safe assets is greater than the supply, or is it that the full-employment planned demand for medium of exchange is greater than the supply? In other words, is it the flight to safety, or is it the flight to liquidity? Brad DeLong argues that we have the flight to safety:
“Thus we would expect a downturn caused by a shortage of liquid cash money to be accompanied by very high interest rates on, say, government bonds–which share the safety characteristics of money and serve also as savings vehicles to carry purchasing power forward into the future, but which are not liquid cash media of exchange.”
The problem is that government bonds serve both as savings vehicles and as medium of exchange. As Gary Gorton said, “it seems that U.S. Treasuries are extensively rehypothecated and should be viewed as money”. You purchase groceries with cash, and you purchase other financial assets with U.S. Treasuries, but in both cases we are dealing with media of exchange.
It is very hard to disentangle these two facets of U.S Treasuries, but we have a great natural experiment. After the collapse of Lehman, TIPS were a perfectly safe savings vehicle, but they were much less liquid than cash or other Treasuries. Lehman has mainly used TIPS as repo collateral, and after liquidation the pattern has shifted, the marginal holder of TIPS was more likely to use it as savings vehicle rather than for transactional purposes. The result was what FT Alphaville has called “The largest arbitrage ever documented”, as the price of TIPS fell by as much as 20 cents on the dollar as compared to much more liquid Treasuries. This gives us a clear indication that post-Lehman panic was a flight to liquidity, and not a flight to safety. This means the best policy response was the expansion of the quantity of liquid, rather than safe assets. After March 2009 QE announcement, the Fed has greatly enhanced liquidity properties of TIPS.
This does somewhat undercut my argument that the TIPS spreads showed inflation expectations falling sharply. They did fall, but not as sharply as the spreads suggest. But I’ve always acknowledged that the TIPS spreads might have been distorted by a rush for liquidity, and I’ve also argued that the rush for liquidity shows the Fed was behind the curve just as surely as would lower inflation expectations. Either way, money was much too tight.
It also seems to undercut my cherished belief in the EMH. Perhaps someone in the mutual fund industry can tell me why bond funds didn’t just construct a portfolio of TIPS plus inflation futures that was guaranteed to outperform Treasury note funds of equal maturity. 123 linked to an arbitrage paper that seemed to suggest there were lots of $100 bills lying on the ground.
PS. I’ve been so busy that I haven’t linked to this David Beckworth and William Ruger article on Milton Friedman in Investor’s Business Daily. Good stuff.
PPS. I also wanted to link to this good James Hamilton post on QE:
A related complication is the fact that the market has already anticipated substantial additional LSAP [i.e. QE]. My guess is that an additional trillion dollars in purchases is already priced into current bond yields and exchange rates.
For all these reasons, the key message of the November FOMC statement may not be the size of purchases that the Fed announces, but instead the framework it offers as guidance for exactly what such purchases are intended to accomplish.
Exactly. It’s all about the framework. In retrospect, the biggest problem in the second half of 2008 was the lack of a policy framework. The was no indication that the Fed would do anything to stop, or later reverse, the sickening plunge in NGDP.
Tags: TIPS spread
24. October 2010 at 17:45
Stephen Williamson (at “New Monetarist Economics”) has a post worth thinking about in response to Christy Romer’s piece in the NY Times today.
He’s basically pointing out that if QE is successful and inflation takes off higher than the Fed wants, the Fed might be hesitant to raise interest rates given the huge amount of government debt projected to exist by then.
“The up side of this is that some people will no longer be lying awake at night worrying about how we might end up looking like Japan. However, some others might be concerned that we end up looking like Argentina.”
24. October 2010 at 17:53
This post is a twofer.
1) Monthly data would have given the FOMC a much greater heads up in late 2008.
2) In retrospect there was no policy framework in place for a contingency that had already been forecast. This is a monumental failure from a strategic point of view.
I’m increasingly of the point of view that the problems we are experiencing are failures from the top. The Board of Governors only has one person at the moment who truly understands these issues: Ben Bernanke. And the Fed Bank Presidents are for the most part bankers and not macroeconomists.
It makes me sick to think that the most important decisions in this country are being made by people who lack the experience and knowledge to lead the charge against the current menace.
24. October 2010 at 19:58
One can see evidence for the flight to liquidity in the charts on p 13,14 from the Gary Gorton report to the U.S. Financial Crisis Inquiry Commission. The increase in haircuts in the repo market meant firms were forced to sell assets leading to the spread between Aaa-rated and Aa-rated corporate bonds flipping. Institutions were selling their Aaa-rated bonds to raise money throughout 2008, and it accelerated during September 2008. See page 14. If it was a flight to quality and not liquidity they would sell their Aa-rated bonds not the Aaa bonds.
http://online.wsj.com/public/resources/documents/crisisqa0210.pdf
24. October 2010 at 23:55
Mark: is what you are saying something of an argument against having such centralised control of money? If there is a real problem getting sufficiently knowledgeable people in the key positions?
Scott: Statistics Canada produces monthly GDP estimates. So, clearly it is not beyond the ken of national statistics bodies.
25. October 2010 at 00:38
“It also seems to undercut my cherished belief in the EMH. Perhaps someone in the mutual fund industry can tell me why bond funds didn’t just construct a portfolio of TIPS plus inflation futures that was guaranteed to outperform Treasury note funds of equal maturity.”
Because flight to liquidity is the same thing as a flight to safety, its just a flight to avoid liquidity risk.
1) At this point no one was in a position to do what you describe unilaterally.
2) Doing so as an ETF or something equally liquid for a large group would require months of waiting for SEC approval.
You have to face it, Scott, so long as we have a centrally planned monetary/financial system we do not have dynamic efficiency for large events, only for small ones.
25. October 2010 at 05:24
Doesn’t the EMH assume away liquidity problems? So if liquidity problems increased, it is unsurprising that the EMH would fail. EMH is only a rule of thumb.
For example, the fact that currency is rate of return dominated by nominal bonds is a glaring violation of EMH, but easily explained by the fact that you do spend currency in places where you can’t spend bonds, i.e. liquidity problems.
Another example is that “off the run” bonds trade at a discount on “on the run” bonds, even though risk and duration are identical. This violates EMH, but is easily explained by on the run bonds being more frequently traded and thus more liquid.
25. October 2010 at 06:07
JTapp, What I find so objectionable is when he argues there is no evidence that sticky wages and prices are a problem. Would he say the same about 1932? And what evidence did we have then, that we don’t have now?
Mark, I completely agree.
Richard, That’s a good point. It’s not an issue I’ve studied, but your analysis sounds correct to me. More evidence that liquidity was a problem.
Lorenzo, Yes, I knew about Canada, but forgot to mention it.
Doc Merlin, Yes, I could could see how SEC regulations could make the financial system less efficient. Indeed I think the SEC has done more harm that good—we would be better off if it had never been created.
Nick, Oddly, I’d take you argument as supporting the EMH. It would be no surprise if impressionist paintings tended to under-perform stocks of equal risk, they also provide services to the viewer. If cash provides liquidity services that bonds do not, then cash should have a lower rate of return in an efficient market. So perhaps I was too quick to dismiss the EMH. Thanks.
25. October 2010 at 07:29
Scott, you said “Oddly, I’d take you argument as supporting the EMH. It would be no surprise if impressionist paintings tended to under-perform stocks of equal risk, they also provide services to the viewer. If cash provides liquidity services that bonds do not, then cash should have a lower rate of return in an efficient market. So perhaps I was too quick to dismiss the EMH.”
You had me worried for a moment 🙂
From a practical perspective, it is very difficult to disentangle risk from liquidity. In many cases, illiquidity arises due to risk (and vice versa).
25. October 2010 at 08:22
Scott, thanks for the plug. Let me add another data issue: why doesn’t the Fed start publishing M3 again? Gary Gorton makes the case that M3 is a better measure of the money supply than the alternatives during this crisis. (Yes, I am sure a M3 divisia measure would be even better.) So why not start publishing it again? The “costs don’t justify the benefits” excuse that the Fed has up on its website may have been true in the past but is now dated. We would all benefit from seeing what really happened to money during the crisis.
Now there is no substitute for a real time measure of aggregate spending. Thus, your call for monthly GDP. Still, it would be nice to see what is truly happening to M3 over this crisis.
25. October 2010 at 08:36
Here is a very readable discussion of NGDP and QE that takes a contrary position to that of most contributors to this blog. I find Dr. Hussman’s approach to be very persuasive. See:
http://www.hussmanfunds.com/wmc/wmc101025.htm
25. October 2010 at 11:56
@thruth:
‘From a practical perspective, it is very difficult to disentangle risk from liquidity. In many cases, illiquidity arises due to risk (and vice versa).’
Totally true, illiquidity is just another form of risk.
25. October 2010 at 14:14
Might the refusal to track M3 (in the present) be a way to “hide” what gets done with the numbers, if deflation continues unabated in residential and commercial property?
25. October 2010 at 14:29
[…] “” Government should publish monthly GDP data. […]
25. October 2010 at 14:29
[…] “” Government should publish monthly GDP data. […]
25. October 2010 at 16:07
Mark A. Sadowski:
Does “Helicopter” Ben really understand it? He did 10 & 15 years ago, but not now. I wonder why. And he didn’t do squat about it while the problems were festering.
25. October 2010 at 16:18
It also seems to undercut my cherished belief in the EMH.
Krugman, Barry Ritholtz and others have have destroyed the EMH in the past 6 months, if not earlier.
25. October 2010 at 17:15
thruth, I agree.
David, They should do that. And for a truly “real time” NGDP measure, they really need to set up and subsidize a NGDP futures market.
Bob, Why is that persuasive? It seems like the old discredited “pushing on a string” view. The negative correlation between M and V is hardly surprising when between 1983 and 2008 the Fed was targeting something close to NGDP (the Taylor Rule.)
Rebecca, Yes, they may well be embarrassed by what it would show.
Phil, That’s funny, I’ve read Krugman for years and I don’t recall any persuasive critiques of the EMH. What is his evidence? More “anomalies?”
25. October 2010 at 18:34
@ Phil can you provide a time where he “destroyed the EMH.” I can’t recall any time.
The closest they got was the regression to the mean argument, but the possibility of bubbles removes that argument (yes bubbles are an efficient response to the possibility of regression to the mean arbitrage).
25. October 2010 at 19:19
We also publish a monthly US GDP figure, with our first estimate released the first Friday of the month on the previous month. Econbrowser (via Professor Menzie Chinn) has used it in several of his high frequency data posts. If you have an interest in it, please let us know and we’d be more than happy to share it with you.
26. October 2010 at 05:57
Doc Merlin, And those are still “anomalies” which don’t disprove the EMH.
Thanks Maria Simos. When the quarterly number come out, how close are they to your estimates of the previous three months? Do you have any data on average errors?
26. October 2010 at 08:50
Scott, I think I know why they don’t publish monthly data. Publishing it would show how horrible GDP data is. They have to go back for months after publishing GDP data and make corrections on, it… With higher frequency data it would become more obvious that they had to do this and would call into question the data.
Actually, i’m pretty sure the data is crap anyway,
Plot the official numbers for the national accounts summed up (with appropriate signs, so they should theoretically cancel out to zero. You see that after about iirc 82, they no longer do. You get massive multiple hundreds of billion dollars swings.
26. October 2010 at 09:56
In response to my link to Hussman Scott said:
“Bob, Why is that persuasive? It seems like the old discredited “pushing on a string” view. The negative correlation between M and V is hardly surprising when between 1983 and 2008 the Fed was targeting something close to NGDP (the Taylor Rule.)”
I find Hussman persuasive because I can follow all the logic and the equations and they all make sense. With NGDP targeting I cannot follow the logic. I have not figured out the mechanism by which NGDP targeting will help the economy. The only explanation I can remember is the one that talks about how NGDP acts on peoples expectations. I am an engineer not an economist and this seems too vague and unlikely.
I have this nagging suspicion that the level of NGDP is not a driver but the result of other factors that do the driving Thus, if the FED uses NGDP to try to drive up the economy it may not work. It may be like trying to drive using the rear view mirror.
26. October 2010 at 23:07
[…] – Why doesn’t the US publish monthly GDP data? […]
27. October 2010 at 02:26
Scott, you said:
“Nick, Oddly, I’d take you argument as supporting the EMH. It would be no surprise if impressionist paintings tended to under-perform stocks of equal risk, they also provide services to the viewer. If cash provides liquidity services that bonds do not, then cash should have a lower rate of return in an efficient market. So perhaps I was too quick to dismiss the EMH. Thanks.”
There are two issues here:
Is the liquidity risk premium priced efficiently?
Only a part of those 20 cents on the dollar can be explained by the changes of the liquidity services provided.
So the problems with EMH still remain.
27. October 2010 at 17:10
Doc Merlin, Yes, but they are no worse than the quarterly data–and what else do we have to go on?
Bob, Much of macro is very counter-intuitive, it goes against common sense. NGDP forecasts are roughly what businessmen mean by “business confidence.”
123, I’m not sure what you mean by 20 cents on the dollar. Was there some asset yielding 20%.
28. October 2010 at 02:27
If you believe inflation swaps market, Treasury – TIPS mispricing was approx. 20%.
28. October 2010 at 08:50
Scott, Thanks for the reply.
What I have the most trouble relating to is the mechanism by which the Fed can increase NGDP. Fed NGDP forecasting, FOMC activity, IOR and QE do not seem up to the task. When FDR set the price of gold in 1933 it may have worked but we do not have this tool at present.
29. October 2010 at 02:46
@Bob:
“When FDR set the price of gold in 1933 it may have worked but we do not have this tool at present”
The market sets the value of the dollar nowadays. The central planners just pick interest rates. Hopefully someday soon they won’t pick that either.
29. October 2010 at 14:34
Doc:
Obviously central planning the interest rate did not work. Do you think the Fed can increase NGDP by NGDP forecasting, FOMC activity, IOR and/or QE?.
30. October 2010 at 21:16
Without being mathematical, I suspect that differential capital / funding costs between Ts and TIPS drive apparent arbitrage. Longstaff et al didn’t seem to take these into account. If this is true, then they have to take a reality check and EMH lives.
4. November 2010 at 17:09
123, And how much of that was liquidity services? And how can you tell?
Bob, FDR was doing price level targeting, we can do the same. Permanent increases in the money base are just as effective as higher gold prices.
Manny, I just don’t know enough about that to comment.
5. November 2010 at 04:24
Using my priors, I’d say that 7% was related to liquidity risk premium, 7% was related to TIPS market inefficiency, and 7% was related to wrong inflation estimate in the swaps market.
If you believe EMH, you should give more weight to liquidity risk premium.