Last week’s post ended with a bit of a cliff-hanger: I wrote about how the major stock market disasters are highly correlated with U.S. recessions. Since it doesn’t look like we’re anywhere close to a recession let’s not get too worried about the stock market volatility in early February! But I didn’t really elaborate on why I’m not that concerned about the U.S. economy right now. So, today’s post is about what indicators would I look at to reach that conclusion.
The broader context of this post and, hopefully, a few more followup posts in the coming weeks is the question that I’ve been grappling with for a while:
What would it take for me to reduce my equity weight?
You see, a lot of my safe withdrawal rate simulations assume either constant equity weights (e.g. 80/20) or a rising equity glidepath in early retirement (see the SWR series Part 19 and Part 20). But what would entice me to do the opposite? Throw in the towel and reduce my equity share as a Risk Control! Should I ever even consider that?
The broad consensus in the FIRE community seems to be to stoically keep your asset allocation through thick and thin. Physician on FIRE had a brilliant post, adequately titled “Don’t just do something. Stand there!” on why not to react to market swings. That was in 2016 and I very much agreed with that assessment back then. But that doesn’t have to be a universal truth. In my wedding vows, I swore to stay with my wife through “good times and bad.” But the last time I checked I’m not “married” to my equity portfolio, so I should have the right to at least consider scenarios that would convince me to pull the plug on stocks.
If nothing else, thinking about when would be a good time to dump stocks gives me the confidence not to lose my nerves when those conditions are clearly not present, such as during the volatility spike earlier this month. So, what would be the indicators I’m following? Today, Part 1 deals with the macroeconomic picture (but in a future post, I will also share my thoughts on momentum/trend-following etc. as requested by some readers). Among all the different macroeconomic indicators, here are my three favorites…
1: The slope of the Yield Curve
Side note: True, this is more of a financial indicator but the rationale behind this is a macroeconomic one!
There are different definitions for this indicator and I normally use the spread between the 10Y and 2Y U.S. Treasury yield. Whenever the spread drops below zero (i.e., the 10Y Treasury Bond yields less than the 2Y Bond) I would get nervous because over the last few decades and business cycles that’s been a sign of trouble! See the chart below. The spread dropped below zero before every single recession in recent history (shaded in orange). Why is that? Well, in normal times, a 10-year bond should yield more than a 2-year bond. Tying up money at a fixed nominal interest rate presents more risk over a ten-year horizon than a two-year horizon. So, if the yield curve inverts it signals that the Federal Reserve has likely gone too far with rate hikes. And once the next recession hits, likely 12-18 months after the yield curve inversion, the central bank will have to aggressively lower rates to stimulate the economy again. By the way, this raises the question of why the bond market foresees the trouble way before the equity market. Good question! My colleagues in the Fixed Income team tell me that they are simply smarter than the equity people! 🙂
Where to get the data:
- At the Federal Reserve, they have a data tool to pull all sorts of interest rate data going back to 1919s (!), see here.
- Even more convenient, the St. Louis Fed’s FRED database has a series that already calculates the 10-2 spread, see here.
Forecast efficacy: The Yield Curve signal seems to work pretty brilliantly. It’s a forecaster’s dream come true because it provides the trifecta of a good economic indicator:
- The yield curve normally inverts well before the start of the recession: 12-18 months!
- There are no Type 2 errors, i.e., the signal never missed a single recession over the last 50 years!
- There are (almost) no Type 1 errors, i.e., there were no false alarms. Well, I say almost because in 1998 the slope dropped very slightly below zero and then recovered again. The curve inverted again in 2000 and – like clockwork – the recession followed a year behind.
Any caveats? Sure! No single indicator can reliably predict everything. As I said, there was one false alarm where the curve inverted way too early (1998, 3 years prior to a recession). I can also envision circumstances under which the yield signal may not necessarily work going forward:
- A Type 2 Error (false negative), where the yield curve never inverts because the Federal Reserve is much more gradual in raising interest rates than in previous tightening cycles. Then the 10-year yield may never drop below the 2-year. But then some external catastrophic event, e.g., Rocket Man going crazy or a worldwide flu pandemic that paralyzes the world economy. The yield curve would likely still invert at that time (unless the flu wipes all the bond traders) but it would likely occur at the same time as the recession rather than before the recession.
- A Type 1 Error (false alarm), where the Federal Reserve is surprisingly aggressive in raising rates. The yield curve inverts because the 10-year yield is held down by demand for U.S. Treasury Bonds from abroad where interest rates remain artificially depressed. But the U.S. macroeconomy withstands the rate hikes and chugs along for another 5 years before going into another recession. I would probably not get too worried about just the yield curve inverting without any of the other indicators (see below) confirming the signal!
Where are we now? It doesn’t look like the yield curve is that close to inverting (yet). As of 2/20/2018, the spread was about 67bps (0.67%), which is slightly below the historical mean (~1%) but it actually increased slightly since January! So, subject to the caveats mentioned above it appears that there is no imminent danger of a recession. If the next business cycle looks like the last seven the curve would have to first invert and then it might take another year for the recession to start. So, in light of this yield curve correlation, when I heard that whole chatter over the last few weeks about how rising interest rates will sink the economy I realized that I would have been more worried if the 10Y had dropped and further narrowed the 10Y-2Y spread. Now, that would have been scary!
2: Unemployment Claims
In addition to the unemployment rate, there is also data on unemployment insurance claims. What I like about the Claims data is that it’s released every week and, unlike the Payroll Employment data, there are (almost) no revisions to past data. The weekly raw numbers are a bit noisy sometimes, but if I plot the 4-week moving average we get the chart below, again with recessions marked as orange bars. Notice the pattern? Unemployment claims spike around recessions! Who would have thought?
Where to get the data: I got the time series from the St. Louis Federal Reserve FRED database, see this link.
Forecast efficacy: The level of unemployment claims matter less than the direction. What I noticed is that around the beginning of each recession, the unemployment claims start going up, sometimes even well before the official start of the recession. Quite impressive! What’s more, the subsequent drop in the unemployment claims is also a pretty awesome indicator for timing the end of the recession! And again: all of this is with publicly (and free-of-charge!) available weekly data. You may remember the discussion last week about how the NBER is notoriously late in announcing the business cycle turning points. But there are a few (mostly) reliable macroeconomic indicators that provide much more timely information. If I had access to only one single macroeconomic indicator I would pick this one!!!
Caveats: Just for the record, there have been a few false alarms: In 1995 there was a bit a rise. But we never got a recession and the claims eventually dropped again in 1996. Also, 2005 brought a huge spike due to Hurricane Katrina, which then reverted back down very quickly. So one should probably disregard temporary spikes from local and short-lived events!
Where are we now? Currently, the unemployment claims are close to a 45-year low (!), in the low 200k range. We’d need to see an increase to at least 300k or even 350k before I’d get worried about an impending recession. And that rise has to be sustained and not due to a localized and short-term event like a hurricane. So when I saw the stock market drop by 10+% earlier this month, I looked at the Unemployment Claims data and concluded that this patch of volatility will likely pass!
3: The Manufacturing PMI
PMI stands for Purchasing Managers Index. It’s a poll that asks decisionmakers at manufacturing companies about their business outlook. A level of 50 indicates that the positive vs. negative responses are exactly even. Below 50 is considered contractionary and above 50 expansionary. What I like about the PMI data is that it is released in a very timely fashion on the first business day of the month (covering the previous month).
Where to get the data: I got a long historical time series from Quandl, but for some reason, it stops in 2016. I spliced the remaining data with the recent numbers from Ycharts.
Forecast Efficacy: The PMI time series is a bit noisy, I’ll admit that. The index also regularly drops below 50 into the “contraction” territory with no recession in sight. But a reading of 45 or below is a pretty sure-fire signal of a recession around the corner or already in progress. In fact, every single recession saw a drop below 45 and every drop below 45 occurred during or was followed by a recession. It’s also pretty good at timing the end of a recession! That’s a pretty sharp indicator!
Caveats: There have been two occasions where the PMI dropped below 45 only half-way through the recession (1974 and 2008). For example, during the Global Financial Crisis, the PMI dropped below 45 for the first in September 2008. That number would have been released on October 1, 2008 (the first business day of the next month). Nine months after the start of the recession! But keep in mind that the stock market dropped another 41% between October 1, 2008 and the March 9, 2009 equity market trough. So, the PMI would have been a bit late in timing the start of the recession but still excellent in timing the exit from the stock market!
Where are we now? The PMI is not just in expansion territory; it stood at 59.1 in January! That’s well above average and well above anything that would normally indicate a recession around the corner. With the recent tax cuts, more investment demand and a pretty rosy business climate it seems that there is no recession in sight. Which would signal to me that, according to last week’s results, volatile markets are always a possibility but they are likely not the type that will ruin everybody’s retirement!
Talking about macroeconomics…
… February is “Macro Month” on the ERN blog. So I wanted to give a shout-out to my blogging buddy Actuary on FIRE who launched the first part of a series on the impact of inflation on early retirees. This is a joint-venture with yours truly so watch out for Part 2 of the series here on the ERN blog next week!!!
Conclusion
I’ve been a passive investor and stuck with essentially 100% in equities through the last two major bear markets (2001-2003 and 2007-2009). It’s been much easier to so while accumulating assets because I could tell myself that I am picking up stocks at bargain basement prices. But risk control is much more important in retirement when the cash flows are reversed and we have to withdraw money from the portfolio: Say “Hello” to Sequence of Return Risk! And one way to deal with Sequence Risk is to actively (!) control the equity risk in my portfolio. It’s not a crazy idea! Todd Tresidder (Financial Mentor) mentioned the importance of managing risk in his ChooseFI Episode. And in my response to that episode (featured in Episode 52R), I agreed, though it took me until today to finally provide some details.
Out of my three favorite macro indicators, not a single one is screaming “danger” yet. Again, I like to stress that macroeconomics is not the only thing one should monitor but we have to start somewhere and this is all I could fit into a 2,000-word post. More to come in the following weeks!