I have been trying to write an all-encompassing post for the new year for two weeks now. But time has never allowed it. So I’ve decided to forego that and get something out to you on what I am thinking as 2020 begins.
First and foremost, let me say that I hope this post is emblematic of how 2020 will proceed. It’s not going to be all-encompassing. But, I will get it out there. And that’s what I intend to do for the rest of the year – write often, making Credit Writedowns much more of the weekday newsletter it used to be.
I appreciate your readership and feel I owe it to you to post much more often.
The 2020 Economy Framework
So, what am I thinking as 2020 starts? Well, it’s about the same as I wrote a month ago – that some soft data doesn’t spell recession. But, while that is actually a forward-looking statement, it is very much about the coming one to two quarters. After all, the last you heard from me, I was still saying recession in 2020 was a base case.
So, what about the rest of 2020? I don’t think anyone has a window into three or four quarters out. So, let me spell out how I think about it:
- First, you have to start with the present day, both in terms of what the current level of growth is and what the data say about growth prospects going forward. That gives you a sense of the vulnerability of the economy to growth slowing or a recession on the downside and to policy-tightening inducing growth or inflation on the upside.
- Then, you have to have a sense of the economy’s vulnerability to economic volatility. For example, when the last decade started at the end of an epic crisis, you would have been forgiven if you thought that the US economy was more vulnerable to recession because of the ‘output gap’, an expected lower level of growth below the economy’s potential. But, despite growth being relatively slow, it has also been very stable, meaning the amplitude of growth variation has been low. Both the push to the downside and upside in growth has been low. Why is that? It may have something to do with the fact that services, an increasing part of the economy, are less volatile than manufacturing. Whatever the source, there is no indication that we are moving into a phase of greater economic volatility yet, despite warnings about inflation I hear more and more.
- Finally, you have to have a sense of the potential for economic shocks and how much GDP ramp up they can create – or economic implosion they can cause. An example would be the phase one trade deal between China and the US. Does that really matter? My sense is no, it’s not a game changer for GDP; I don’t think it will have a meaningful impact on output or investment.
So that’s how I’m looking at this.
The Cycle Data: Jobless Claims
Now, having said all that, because we are so well-advanced in the business cycle in time, I do think about recession a lot. And so I want to present you with some data about past cycle ends and tell you what I think they mean in today’s context.
Let’s start with jobless claims since that release crossed the wires just as I was writing this. The number for initial claims was a seasonally-adjusted 204,000, bringing the 4-week average down 7,750 to 216,250.
Here’s the data set, using the year-on-year change, which is what I think matters regarding cycle troughs and peaks.
If you look for the data signaling cycle troughs, what you’ll notice is two-fold:
- Jobless claims have trended up before a recession or right when it happens. So this metric is coincident or leading.
- In terms of ‘false positives’ during mid-cycle slowdowns, the data have also trended up. But claims have never remained 50,000 more than year-ago periods for long. They passed that level in 1977 and 1996 but only briefly before heading back down.
My takeaway here then is that we can look at the year-on-year jobless claims change data as leading to coincident and that it doesn’t spell recession until we hit 50,000 on a consistent basis for several weeks.
If you look at the recent chart, we are nowhere near those levels. The data are more consistent with a mid-cycle slowdown in that they have trended up consistently since September 2018. But that’s it. Anyone telling you they see recession in those numbers is mistaken.
The Cycle Data: Non-Farm Payrolls
While we’re at it, let’s see how these numbers align with non-farm payrolls. Here’s the chart.
The data align well in my view. You can see the year-on-year top in January 2019, a 4-month lag in the trough in initial claims year-on-year data. Since then, payrolls have increased but at a marginally slower pace where we are now adding 2.1 million jobs every 12 months instead of 2.8 million.
So, as with the claims data, there’s nothing here that says recession. What you do notice with this time series though is that there are significant mid-cycle rollovers – ‘false positives’ if you will. You see that in the 1940s and the 1950s, at the 1966 equity market top, in 1984 and again in 1995. Moreover, the rollover decline into recession operates with a significant lag. The March 2006 top is a perfect example, coming a full 21 months before recession. And so, it makes interpreting a decline in GDP difficult from this data set.
The bottom line is that I don’t see a fall in non-farm payrolls – especially after such a long business cycle – as something to worry about.
The Cycle Data: Average Hourly Earnings
Now, my thesis here is that it is the combination of wages gained or lost from new jobs, jobs lost and wage levels that determine the pace of consumer spending, with a lagged effect. If these things goose household income long enough, consumption will rise. Conversely, if they crimp income, consumption will eventually fall.
So, let’s look at the average hourly earnings chart.
What I see is earnings growth topping before or right at cycle troughs every single time – with very limited false positive data. February 2003 is the best example of a false positive – and seems to support the Fed’s decision to leave rates at 1% for an extended period during that cycle, by the way.
This cycle, earnings growth is just beginning to roll over.
The top in year-on-year growth was February 2019, a lag of one month to the top in non-farm payroll growth and a 5 month lag to the bottom in year-on-year initial claims data.
This corroborates the intrinsic impulse behind the recent growth slowdown in the US because we have all three metrics showing consistent declines over an extended period. But again, I see nothing here that says recession is in the offing. I am certainly more alarmed by this data series than the non-farm payrolls rollover. But I remain sanguine about the prospect for recession.
Final Thoughts
So, as we head into 2020, I am cautiously optimistic. The data I am seeing corroborate a slowdown in growth. But they don’t definitively point to that growth continuing to slow going forward into 2020. At the same time, the policy mix in the US is not restrictive. And that’s where I see the most room for recessionary impulse.
Moreover, the Atlanta Fed GDPNow figure for Q1 2020 is 2.3% and the NY Fed nowcast for Q1 IS 1.24%. I would put that outside of the stall speed zone. It says that a so-called exogenous shock is unlikely to tip the US into recession from these levels. GDP growth would have to slow even more.
The one caution I have as the year begins is in the Fed’s balance sheet. They have really goosed the market with their repo liquidity injections. And the threat is that they unwind this injection early in 2020, now that year-end repo positioning is over. This could be the catalyst for a severe market correction if they do. And that would be an ‘exogenous shock’ with multi-quarter ramifications if it hits the credit markets hard enough.
On the whole, though, things look decent as we head into the new year. I am cautiously optimistic and am, therefore, moving away from thinking of recession as a base case. It is more of a reasonable worst case at this point. My expectation is for continued growth. Whether the pace of that growth turns up – or continues down – is still to be determined.