[This post has been updated as of 1/21/12]
Is the United States headed into a recession or not? There has been quite a bit of debate on this topic over the last few months. Naturally, the lack of consensus on this topic is due to conflicting data and indicators. I'm going to make an attempt to answer this question using some historical information as a guide.
Summary
Predicting business cycle turning points and recessions in timely fashion is extraordinarily difficult. The difficulty is compounded by the fact that there is no straightforward quantitative measure defining a recession. Currently, there's a debate in the United States as to whether the U.S. economy is tipping into recession heading into 2012. Single indicators like GDP, retail sales, industrial production, nonfarm employment, etc., at face value appear to be in non-recessionary territory, thereby leading some economists to conclude that the U.S. has avoided a recession. The Conference Board's Index of Leading Economic Indicators (LEI) has also been on a upswing in recent months, leading the Conference Board to predict that the U.S. economy will improve, rather than decline, in the future. In contrast, the Economic Cycle Research Institute (ECRI) has been vocal in saying that it's slew of short-, medium- and long-leading indicators are signaling a definite recession in the U.S. Who is right?
In this post, I examine some historical data and conclude that the ECRI is most likely correct and that the U.S. is very likely headed into a recession. Single indicators are generally not good at forecasting recessions and could be revised substantially in the future (see Section 2 below). The LEI's track record has been poor overall (Section 3A). ECRI's track record, while not perfect in its timing, has overall been stellar (Section 3B) and makes me more confident that their recession call should be taken seriously.
1. Predicting Recessions
Predicting recessions is no easy business. As The Conference Board observed in p. 16 of this report [I've emphasized sections in bolded text throughout this post]:
However, the fact is that peaks (or even troughs, for that matter) cannot always be recognized until months after they occur, especially during periods when the data are subject to significant revision. Therefore, a considerable amount of research has focused on finding a real-time turning point rule, which provides adequate warnings.
Unfortunately, it is imprudent to forecast a recession using a simple and inflexible rule. The U.S. economy is continually evolving, and is far too complex to be summarized by one economic series. Even official recession dates for the U.S. economy are determined by a committee of prominent economists that uses a multitude of indicators rather than a simple rule: “Why not replace all this agonizing over a multiplicity of measures with a simple formula—say, define a recession as two consecutive quarters of decline in GNP? Any single measure is sure to encounter special problems just when they matter the most.... We plan to stick with examining all of the data we can and making an informed judgment.” (Robert Hall, Chair, NBER Business Cycle Dating Committee.)
Predicting these turning points is a difficult task even for the best forecasters.
2. Single Indicators and Recessions
A number of recent data points on the state of the U.S. economy - I'm going to refer to these loosely as "single indicators" - have been interpreted as being positive or better than expected. For example:
- Industrial production rose 0.7% in October
- Capacity utilization improved
- Retail sales continued to hold up as did auto sales
- The government's estimate of Q3 GDP (~2.5%) was considered better than feared and was accompanied by a reassuring consumer spending figure
- Employment numbers were mixed
- Stock markets have recovered from a steep drop earlier this summer
Taken individually or considered together, one might therefore be tempted to conclude that the U.S. has avoided a recession. This would however not be a good conclusion. To see why, let's first examine why using single indicators to predict recessions is a bad idea and we'll then address the concept of considering multiple indicators together as part of a combined index.
(a) The first thing to recognize with many of the above single indicators is that they are coincident or lagging indicators of the economy and are not good predictors of future recessions.
For example, Doug Short at Advisor Perspectives looked at recessions since WWII and observed that GDP growth in the quarter corresponding to the starting month of each recession was positive in 6 out of 11 recessions and zero once. As he notes:
The U.S. has had eleven recessions since the earliest quarterly GDP calculations, which began in 1948. In the month declared by the National Bureau of Economic Research (NBER) as the beginning of the recession, quarterly GDP for that month has only been negative four times.
Of course, GDP is not the only single indicator that is not useful for predicting recessions. Mike Shedlock (Mish) at Global Economic Trend Analysis has also shown that stocks are not good leading indicators by themselves - nor is consumer sentiment or M2 money supply. This inference can be extended to several other single indicators.
(b) Secondly, many of the indicators are also subject to revisions - sometimes substantially - in the future.
Consider real GDP as an example. In a 2010 paper "The Income- and Expenditure-Side Estimates of U.S. Output Growth", Jeremy K. Nalewaik of the Board of Governors of the Federal Reserve System examined the two official measures of GDP - GDP (I) or GDI and GDP (E) or GDP - to figure out which of these two measures had a better track record of more accurately representing the business cycle in real-time. The paper is worth reading in full, but here are some of my key takeaways from the paper. In comparison to the usually reported GDP (E), GDP (I):
- Tends to better reflect the extent of growth & declines
- Is subject to less severe revisions in the future
- Correlates better with major real-time indicators
- Appears to be a better predictor of next-quarter GDP (E)
- Tends to better predict the direction of future GDP (E) revisions
What is most pertinent to this post is the fact that GDP is subject to quite a bit of revision - sometimes over a period of years. For example:
- The first estimate of Q4 2008 GDP came in at -3.8%
- The latest view of Q4 2008 GDP is -8.9%
This chart from the Nalewaik paper is helpful to illustrate the situation as of Feb 2010. The bottom line is that GDP could get revised quite a bit and using real-time estimates of GDP to predict recessionary conditions in the future is often not practical or reliable.
UPDATE 11/24: The BEA has published revised estimates for 2011 GDP. Here is a chart that summarizes the estimates from Oct 2011 and Nov 2011. Note the downward revision to Q3 2011 GDP and Q2 2011 GDI. Also notice that GDI is estimated as declining from 1H 2011 to 2H 2011, whereas GDP is estimated currently as increasing from 1H 2011 to 2H 2011. Just looking at the state of the economy, it is hard to argue that 2H 2011 has been better than 1H 2011 overall - I would not be surprised if Q2 and Q3 2011 GDP continue to be revised down in future releases. Further, given that GDI estimates are treading water so to speak, a recession in late 2011 or early 2012 is in the realm of possibility. As @justinwolfers rightly observed: "The untold story of this recession: The many false signals given by US GDP data which have given false hope, leading policy mistakes."
UPDATE 1/21/12: Here are a couple of charts that include the BEA's Dec 2011 revisions to GDP and GDI. First, GDP:
And here's the GDI:
The issues of revisions is true for many other indicators.
(c) Thirdly, many indicators could be positive or rise sharply in the middle of recessions.
GDP growth was once positive during the last two recessions (2008 Q2 at 1.3% and 2001 Q2 at 2.7% - per the BEA). Stocks could also rise sharply, especially early on in the recession, when it's not clear that the economy has already irrevocably fallen into a recession. This chart below, that I put together from the St. Louis Federal Reserve Economic Database (FRED), makes this point visually.
3. Leading Indicators and Recessions
To really get a sense for what is in store in the future, we need to focus on indices of leading indicators, as opposed to lagging or coincident indicators. Two of the most commonly cited indices are The Conference Board's Index of Leading Economic Indicators (LEI) and the Economic Cycle Research Institute's (ECRI's) collection of leading indicators, one of which is the more widely reported Weekly Leading Index (WLI).
(A) The Conference Board's Index of Leading Economic Indicators (LEI)
The LEI recently registered a strong positive number, leading some to assume that a recession is not in the cards. In fact, LEI growth has been positive through most of this year and the Conference Board is now sounding positive regarding the prospects for the U.S. economy in the months ahead:
The Conference Board reported Friday that its index of leading economic indicators rose 0.9% last month, significantly faster than the revised 0.1% rise in September and the 0.3% increase in August.
[...]
Economists said the strong October gain in the leading index and other positive reports recently had at least eased fears that the economy would be in danger of slipping into a recession.
Conference Board economist Ken Goldstein said that the latest leading indicators report was pointing "to continued growth this winter, possibly even gaining a little momentum by spring."
A natural question to ask in evaluating this forecast from The Conference Board: what is the prior track record of the LEI?
First, the LEI's individual components are not necessarily all leading indicators. Second, LEI's track record in predicting recessions has been underwhelming to say the least. For example, here's a snippet from an archived WSJ article by Patrick Barta from April 19, 2001 - inside the 2001 recession:
...the Conference Board in New York reported the latest results for its Index of Leading Economic Indicators, which is designed to forecast where the economy is headed in the next three to six months. While the index fell 0.3% in March to 108.5 -- the second consecutive monthly decline -- the board said the pace of the decline wasn't deep enough to prompt an economic contraction. "No recession is on the horizon," the group said. Defining a recession involves many factors, but a common rule of thumb is that it requires two consecutive quarters of decline in gross domestic product, the value of the nation's output.
[...]
The Conference Board's index missed the previous two recessions, in 1990-91 and 1981-82, prompting some economists to refer to it as the "index of misleading indicators." That may be too harsh, considering the index accurately called recessions in the 10 years prior to 1981. And the Conference Board revised the way it compiles the index in 1995 to eliminate a component that was sending false signals.
The Conference Board has also acknowledged in this report the issue of false positives in their indicators and the challenge of calling a recession ahead of time. In past decades, recession signals in their indicators sometimes emerged only well into the start of the recession. This problem was present in their data prior and during the Great Recession. As this Conference Board paper points out:
The three Ds—depth, diffusion, and duration—of a decline in the LEI can be a useful tool to analyze whether the economy is headed for a recession.5 According to this approach, a recession usually follows when the (annualized) six-month decline in the LEI reaches 4.0–4.5 percent and the six-month diffusion index falls below 50.0 percent. As of December 2007, the six-month decline in the LEI had reached 2.3 percent, which was not deep enough to signal a recession based on the three Ds criteria (Chart 3). At that time, the fall in the LEI was being dampened primarily by continued increases in money supply (M2).6 The six-month decline in the LEI quickened at the beginning of 2008, but it was not until October 2008 that it reached and exceeded the threshold decline called for in the three Ds rule (Chart 4).7 Revisions to the LEI in subsequent months were relatively small (Chart 5), and they did not alter the cyclical outlook as they unfolded in real time.
If you look at the chart in their paper, you'll notice that this problem was compounded by the fact that their metric dropped to ~-4.5% in Jan 2008 but quickly recovered to ~-1.5% by mid-2008, before dropping again, making it even more difficult for them to make a recession call at that time. All in all, the historical data on the LEI leads me to infer that it is *not* a good predictor of recessions.
However, what is quite striking this time is that the LEI is showing significant growth - and therefore predicting a positive economic trend for the U.S. in the coming months - at the same time as the other leading indicator, ECRI's WLI is predicting the exact opposite! In contrast, both the LEI and the WLI growth rates declined in the early period of the last two recessions.
(B) ECRI's Leading Indices
Unlike the LEI, ECRI appears to have a better track record of predicting recessions and growth rate cycles. Although the April 2001 WSJ article (above) claims ECRI only predicted two of the prior three recessions, ECRI claims they called all three of the prior recessions in advance.
ECRI's track record is not perfect. For example, as Mike Shedlock has pointed out (also see here), while ECRI called the Great Recession before many others did, they did so definitively only after the recession had started (a fact that was itself only established many months subsequently, thereby making ECRI's call a de-facto prediction in real-time). That said, I've been following ECRI for a long time and I've found that a less-than-perfectly-timed recession call that is still ahead of others is still a pretty good and valuable call. My view is that their track record on calling slowdowns, recessions and recoveries in the last 10+ years has been generally stellar. Currently, ECRI has been pretty vocal since late September 2011 in saying that the US is on a recession track:
The U.S. recovery wasn’t much to begin with, and now it’s dead.
That is the news from the widely respected Economic Cycle Research Institute that said the U.S. has already or is about to tip into recession. The announcement, made public Friday, was met with skepticism by other economists who are more hopeful the U.S. and the world will skirt recession.
Lakshman Achuthan, co-founder of ECRI, however says the hope is misplaced. He says the call is based not just on the weekly leading index that is disseminated to the public, but also on dozens of other ECRI leading indexes that are available mainly to ECRI’s clients.
Growth in the weekly leading index, released on Fridays, began to decelerate in April and turned negative in mid-August.
ECRI has tried to explain their recession forecast further:
The WLI, however, is not the only hammer in ECRI’s toolbox, says Achuthan.
To capture the macro-view, ECRI also puts together a long leading and short-leading index. The long-leading index, which has no exposure to equities, started falling back in December 2010 and is still falling.
In addition, ECRI puts together sector-specific indexes that cover areas including manufacturing, nonfinancial services, housing, credit and exports. These indexes are “overwhelmingly showing recession patterns,” says Achuthan.
In particular, the ECRI indexes are signaling the 3 “Ps” of a contraction: the decline has to be “pronounced, pervasive, and persistent.”
Achuthan makes clear that a recession is a “process” in which a negative loop feed backs on itself: Slowing demand lowers productions which results in dropping employment and incomes which in turn weaken sales further.
Ominously, Achuthan says policymakers will not be able to stop this recession.
There's a lot more in interviews posted on their website - including this flashback interview from April 2008 where ECRI maintained a recession call despite a rising stock market and other data.
UPDATE 11/24/11: I missed pointing out that Doug Short at Advisor Perspectives has two good charts showing the WLI growth rate vs. U.S. GDP growth - here and here.If you look at the first chart, you'll notice that GDP growth has generally dropped subsequent to negative periods of WLI growth. Short also links to an Oct 2011 report (PDF) by Hoisington Investment Management. Hoisington says:
Negative economic growth will probably be registered in the U.S. during the fourth quarter of 2011, and in subsequent quarters in 2012. Though partially caused by monetary and fiscal actions and excessive indebtedness, this contraction has been further aggravated by three current cyclical developments: a) declining productivity, b) elevated inventory investment, and c) contracting real wage income.
Given ECRI's past track record, I am much more inclined to believe they are right on their call and that The Conference Board is on the wrong track here. Only time will tell who was correct.
Comments