Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
Perma-bears are bombarding us with alarm bells, sounding the doom of the US economy. We find ourselves in yet another "summer slowdown scare," for the third year running. In 2010 and 2011, the purported slowdowns turned out to be soft landings. Investors who ran to the sidelines stared in disbelief as the stock market roared ahead, leaving them behind.
We are likely in the same position now.
Scary headlines attract more eyeballs than good news or a balanced perspective. Interviews tend to go to those proclaiming the end of the expansion – you rarely see the media interviewing someone with a more positive view to offer. As an outsider to the U.S. (I live in Cape Town, South Africa), I am amazed by how quick some Americans are to write off their own economy, which I still believe to be one of the most resilient, innovative, and adaptable on the planet.
In fact, I was so exasperated with the bleak picture painted in the US last fall that I penned an article, published on January 3, 2012, titled "U.S Recession – An Opposing View." It seemed like reputational suicide at the time – just about every reputable observer, inside the advisory industry and out had written off the expansion. Fortunately, as an outsider, I had nothing to lose. And neither did those investors who stuck it out – the SP-500 gained 7.6% in the first six months since my article appeared, peaking on April 2, 2012 with an 11.1% gain.
Among the bearish voices I most respect is John Hussman, whose work I read regularly. He is thorough and quantitatively rigorous. Whenever I am convinced there will be no recession, I temper my enthusiasm by re-reading his articles to make sure I maintain a balanced view. One day he will be right and I will be wrong, but at least I won't be blindsided.
Hussman, in this instance, acknowledged my research and approach even though he did not agree with it. This was hardly something he needed to do – he is a big fish, and I'm a comparative minnow. Hussman spent several paragraphs poking holes in my methods, but I did not mind that; I took it as a solid peer review. I'm grateful to him for keeping me honest.
Right now, the models I presented in my January 2012 article remain far from suggesting recession, but, I should add, they are also far from full-blown expansion. US economic growth remains tepid and vulnerable to the slightest external shock. Such risks are not negligible, but neither are they inevitable. The track of the co-incident SuperIndex versus the last seven business cycles aptly illustrates the tepid growth situation in which we find ourselves – we are on the verge of printing a new historical low-watermark for growth after 36 months of expansion. One can understand where the bearish sentiment comes from!
But the data don't show catastrophe. Looking at the Leading SuperIndex, we are a bit worse off than last summer and the summer before that. We just put in a leading SuperIndex peak on April 13 (10 days after the SP-500 peak) that is lower than the prior two peaks. This slowdown, if not checked in time, may well be the one that pushes us into recession. But even that worst-case scenario is still three to four months away, according to the SuperIndex recession-path projections in our regular weekly report.
So things aren't exactly looking rosy, but there's reason to have some patience. Let's reassess the rush to declare a recession, and consider some ways in which we might take stock of the current situation more objectively.
The problem with singular approaches
I have a deep suspicion of using any single indicator, or even any single composite indicator, to make a recession call. Recession calls are "high stakes" events; the costs of being wrong are high. I prefer to use an ensemble approach (recession forecasting ensemble or RFE for short) of six diversified leading, coincident and lagging composite systems, waiting for these to arrive at consensus before taking any defensive actions.. The longer the lead a system offers, the higher the likelihood of error. I do not use systems with nine- to 12-month leads for that very reason, and because my firm's research shows that any defensive stock market actions earlier than 5 months before recession are likely to be unproductive anyway. (See, for more on this, Recession – Just How Much Warning is Useful Anyway?)
We also do not advocate "all or nothing" reaction to recession signals. Because no system is perfect and nobody has yet managed to exactly model an economy and its likely behaviour, we prefer a staged-action approach with our RFE to adjust to increasing market risk. Typically, the longer-leading (least accurate) models in the RFE will flag recession first and the co-incident and lagging (most accurate) ones will trigger last. As more and more models in the RFE flag recession, we move more and more into cash (or hedge more and more). Hopefully, by the time the last coincident model flags recession, we are mostly in cash or fully hedged.
None of the RFE models are currently flagging recession, and, as regards the 9 SuperIndex sub-components, only the ECRI Weekly Leading Index (WLI), Philly Fed Business Outlook Survey and the ISM Purchasing Managers Index (PMI) are nearing recession territory. Of course, a couple more weeks' bad data and things could change quickly.
Additionally, if we look at the Philly Fed 50-state leading and co-incident indicators (which have a much shorter history than all the other models we use, hence we look at them separately), they do not show us to be in recession either. Not out the woods, but not in the clutches of the fox either.
What must the NBER be thinking?
Of course, the final word on recessions will ultimately fall to the NBER, the official arbiters of recession dating. The frustrating reality, however, is that the NBER takes forever to proclaim specific starts and ends to expansions, so all the revisions can "work their way through" and they can be dead accurate. Given that these proclamation lags can take up to 18 months, NBER announcements are good for historical, academic and back-testing use only. Now, given that many reputable people are claiming we (1) are already in recession or (2) are about to enter one, let us discard all our fancy models aside and look hard at what the NBER will have to chew over.
They will be examining four co-incident indicators:
- Industrial production
- Real personal income less transfers deflated by personal consumption expenditure
- Non-farm payrolls
- Real retail sales deflated by consumer price index
You can track these together on monthly charts at the Federal Reserve Bank of St Louis, which gives some idea whether we are in recession or not, but it offers limited information about how far from recession we might be. The latter question is a more important to investors, advisors, and fund managers than the first.
If we take real-time vintages of the above four co-incident components (no revisions included, just what the model would have observed in real time when data was published), we can use a Dynamic Factor Markov Switching (DFMS) process to build the data into an economic output index that will appropriately describe historical NBER-defined recessions and expansions. Seminal research by Chauvette, Hamilton and Piger have shown how DFMS models using real-time co-incident data are extremely accurate in estimating start and end dates to NBER recessions in near real-time. The result is the below chart:
Now, the problem with the co-incident model is that it has an average one-month lag to NBER recessions (some are leads, but most are two-month lags). The real-time data set has components that are up to two months lagged, and we also have a publish lag to consider, hence the two-month lags make sense. But considering that, using this measure, you get to know if we are in a recession at worst two months into it; that is a lot better than the average eight- to 12-month lag from the NBER committee's official pronouncements. Unlike the SuperIndex and the other RFE models we deploy, the purpose of this model is not to warn of a recession but to serve as a "confirmation of last-resort" if a recession has begun. The implied probability of recession within three months of the above model is about 6%.
Returning to the present situation, more damage has yet to come before we can declare a recession. This damage can be in the form of massive downward revisions of epic proportions (and our real-time DFMS model will not be affected by this) or continued weakened data of the next few months. When in the past we found ourselves at current levels and descended straight into recession are highlighted with red circles in the graph above. Apart from December 1973 (when we were indeed in recession) there was an average of six months before we entered recession, with a minimum of three months and a maximum 12 months. This means that, at worst, we could fall into recession between September and December 2012. That assumes, of course, that we are not in a December 1973 scenario, which history suggests might happen one out of every seven times – possible, but unlikely.
Let us turn to ECRI's own monthly co-incident index for the U.S. Ironically, this index does not support the ECRI's 9 month old call for recession (which is now "imminent" given their own deadline of no later than 2Q 2012).
If we track the current progress of the ECRI co-incident index growth versus the last 10 recessions, we can see that, barring revisions of a unprecedented magnitude, imminent recession is unlikely, with the 12-month correlation of the current track to the average recession approach path at -0.59 (negative correlation meaning it is in opposite direction to the average approach path).
No model is infallible. Recession forecasting is an art more than a science. We can try our best using the best models and adopt risk-adjusted stances to the market with our equity allocations. What the perma-bears are saying may come to pass, and we could fall into recession sooner. It's possible we may even already be in one. But that is unlikely according to my firm's models and methods, some of which were discussed in this article. You must make up your own mind, but don't blindly succumb to the perma-bear mindset. There's plenty of reason to believe we haven't reached doomsday just yet. The next few months however, will be telling indeed.
Dwaine van Vuuren is CEO of RecessionALERT.com, a provider of investment research. If you would like to receive the next 4 weeks Recession Reports for free, just email us at email@example.com with FREE FORECAST AP3 in the subject line.