Jason Stipp: I'm Jason Stipp for Morningstar.
Our director of economic analysis Bob Johnson has been looking at the economic indicators for several years now, and he has a few guiding principles that can help you read those economic reports, too. He's here to share those with us today.
Thanks for joining me, Bob.
Bob Johnson: Great to be here.
Stipp: The first guiding principle is something that I hear you say a lot, and when you say it, you say, "There's a certain way that I like to look at the data and that's on a…"
Johnson: Three-month moving average, year-over-year basis.
Stipp: So, why do you look at the data that way. What does that do to help you get a better handle on the trends?
Johnson: Well, it strips out the monthly noise. So many of these reports are things that are mailed in, called in, and maybe one month somebody doesn't send something in, maybe some month somebody has a big sale they didn't have last year. … The economy moves and a lot of the data can get messed up in a single month.
So a three-month moving average is a great concept, [a great] way of looking at things, because it smoothes out some of those things. Sometimes there are strikes. There are a million things. So, that's one thing that I do for sure. So that's the "three-month" part.
The other part is the "year-over-year," and that deals with the seasonality issue. So many of the data are adjusted seasonally, and those seasonal adjustment factors have in many cases, in my opinion, been wrong. We've had a really volatile economy, we've had recessions, we've had auto industries completely change the way they account for their summer shutdowns, and all of those things make it a moving target. …
All of the government data assumes that everything's moving on a trend, and there's an underlying trend to everything, and then you strip out anything that's different from the trend, and a-ha--you've got the right answer. But when you've got things that really aren't trending, but instead have a bunch of really weird stuff happening …, the statistical data doesn't work out quite so well. So, you don't want to look at the individual pieces of data, because of the seasonal adjustment factors, and I can explain that a little bit more.
Stipp: So with the seasonal [adjustments], you think that some of them are off because we've seen some secular changes in the economy, and also, just calculating them--you are talking about some pretty big numbers here as well, right?
Johnson: Yes. The seasonal adjustment factors are done over a certain period of time, and so, if things happened in the past, but aren't happening today, it really hurts. [For example,] it used to be that people went back to school in September. Well now, people go back to school in August, and so that really changes how the seasonality between those two months work.
So, that seasonal factor is very difficult to start with, but you've got to remember what the seasonal [adjustments] are supposed to accomplish. … What happens is, you really can't compare retail sales in July to what they are at Christmastime in December, because they're just naturally going to be different numbers. And yet, maybe the economy has moved somewhere in there, so you've really got this desire to see what's changed. So you try to go in there and make the July sales and the December sales comparable by doing these seasonal [adjustments].
So it's good work. It's a good thing to try to do. In an economy that's trending, it makes a lot of sense. But it's very, very difficult to get them right, which is why I like to look at the data year-over-year without the seasonals to give me the correct flavor for the numbers.
Stipp: The second guiding principle that you have is somewhat related to the fact that you like to look year-over-year and use that three-month moving average. But that said, we get these data points, a lot of times we get the monthly, so we get unemployment every month, and you start to see these trends or you want to see these trends to figure out, are we picking up steam, are we losing steam? But there's an interesting thing where, if it looks too good to be true, or it looks too bad to be true, it probably is.
Johnson: Yes, that's absolutely right. Sometimes you just look at the number and say, that's too good. I know there's something wrong with it, and it's just going to revert next month. Somebody forgot to mail something in. And sometimes I know what it is, sometimes I don't.
For example, a couple of weeks ago, we got unemployment claims at 355,000, and I said, I've heard there have been more layoffs not less; that doesn't make any sense. And [it was because] we had the storm, and so lo and behold the 355,000, which should have been more like 380,000, then the following [week] came out at over 400,000. So, when you see those anomalies, you've got to really pay attention to them and think, what might be behind them?
Stipp: So if you're thinking about what that next month's unemployment might look like, you tend to look at the month before: If it really exceeded expectations, you might consider that to be somewhat of a headwind, perhaps, for the expectations for the coming month.
Johnson: Absolutely. When you see one number that's really good, sometimes you'll get a second month that's really good, but then once you get much beyond two [months], it's going to what we call "mean revert," it's going to dip back to something that looks more normal, and it usually overcompensates the other way. And people tend to interpret that as, oh things have been booming, and then we hit this slow spot, and now we're busting again! But the real truth of the [matter] is that the numbers weren't tight enough to make that conclusion. [They] probably were going flat that whole period.
Stipp: So we talked about how we can compare some of these reports to themselves and the trends, but let's talk a about what you've learned about the reports and when they are most meaningful. So, at different points of the economic cycle, some reports will be better indicators than other reports. What have you found there?
Johnson: At the bottom, there is absolutely two [reports] that are almost always invaluable. One of them is the purchasing manager data and the other one is unemployment claims. And you don't get a lot of [lead] time. You get maybe two or three months at the most on those two, but those two are clearly the most reliable.
The PMI--and manufacturing in general--is a very, very good indicator at the bottom. Let me do a quick example of why--and I'm going to play a little loose with the numbers: Typically, we did 15 million in auto sales, and we were down to maybe 9 million when things got really bad. Well, auto production got all the way down to 4.5 million. In other words, the [sales] number had been practically its worst point in an entire history, and we were producing half of that, which told you that things just had to stabilize a little bit, and you'd have a near doubling in your manufacturing sector, and that might be enough to lift the rest of the economy off the bottom, and lo and behold, that's what happened.
Manufacturing is much less useful in other parts of the cycle when it's not that dramatic.
Stipp: When the recovery really starts to get overheated, or when an up economy starts to get too hot, one thing that you especially look for--and you've called it the recovery killer--is inflation. How do you look at that and how can you gauge that at certain times to see that we could be due for a downfall.
Johnson: There it's probably an absolute number that makes sense, and again it's the year-over-year, three-month moving average basis, but it's about 4% [inflation]. When you get over that level, we have a recession. It's almost like clockwork.
And there may be other underlying causes of problems, but the final straw that breaks the camel's back is usually the inflation, and [the economy] usually falls relatively quickly after that--maybe three, four, five months after that.
There are a ton of great indicators for recession. The unfortunate thing is that they all come so early that you get faked out, and you think, oh gee, maybe it's not so bad after all. And you pile back in at just the wrong time. There are a lot of manufacturing data that tend to peak maybe as much as 12 to 14 months before you get a recession, and you'd think, oh I like early [indicators], but if you don't know if it's 12 or 14 or if it's 10 or if it's eight [months early], it can really get confusing. So, that's why I like inflation, because it tends to be the one [indicator] that's a little bit closer to the actual event.
Johnson: And not only are there certain times when certain indicators are more reliable or better indicators, the sequencing of indicators and what tends to happen first and what happens after that is really important to keep in mind as well.
Bob: Yes, and that's where graphing packages help, and one of the things where I might have an advantage, because you can just lay it out in a graph and see, here's one indicator and here is the other--which moves first? So many people at the recession were going, well we're not coming out of the recession--employment still down.
If anybody had drawn a graph, there has not been a recovery in history where employment started to move up until after the recovery was six, nine months under way before employment got better. And you can put it on a graph. You needn't have to think too hard about at it if you could see it graphically. So you always have to consider what moves what first.
And other one is a consumer confidence. That's one that tends to move in tandem with other things in the economy. It's not a leader--it doesn't really move either way [as a leading or a lagging indicator]. And so that's another one that's out there that I tend not to look at so much, because it's what they call a concurrent indicator. It really moves with other things, and it really isn't adding much new information.
Stipp: Another thing that you look at is the sizing of some of these indicators. It could be that a small move in something that's a big part of the economy can actually have a big impact, and then a bigger move in something that's a small part of the economy will have a smaller impact, even though the absolute number may be more dramatic.
Johnson: That's why I always keep an eye on the consumer, because that's 70% of the market. And that's one where some of the seasonals and the small factors in the retail sales report can get a little scary, because the numbers don't move a lot, but they are 70%. So you take 0.7 times a move, and it really amounts to something.
Whereas housing, which I am extremely excited about and will help the economy, right now is running under 3% of GDP--but that may swing. Right now housing starts are up something like 40% year-over-year. So even a large move on a small percentage still can mean a lot. That could add 0.8% to the economy.
So you've got to be careful. A 20% move in housing might not mean a lot, but maybe a 40% or 50% move does, and [housing] can swing 40% or 50%. [On the other hand,] the consumer is a huge part of the number--70%--but if it moves 2% or 3%, that would be somewhat unusual.
Stipp: So it's important to consider both: what's the magnitude and what's the absolute move in that indicator.
The last thing you said that's important to look at is not just the dollar changes that you see, but also the unit changes. What's the reason to look through that prism?
Johnson: That's an interesting question--the whole the whole question of units versus dollars, and it's where I think we've been able to make a little bit of a cutting-edge difference here lately. [For example,] housing starts are in units, but a lot of times those houses, the value of them has been … going up, so the impact on economy has been far greater than just looking at the unit level.
For example, unit levels in the existing home sales are only up 10%, but if you look at it in terms of dollar value, it's up something more like 18%, because the average value of the homes has gone up. There are some people like our cable analyst, who might only care about how many housing units we've sold, but in terms of the general economy, the higher-priced home you by, the nicer furniture you're are going to buy, and all the brokerage commissions [are higher] … a lot of good things start to happen with higher-value transactions. So you've got to be a little bit careful when you're counting units.
And the same thing can happen--and has happened--in employment, and sometimes it's both ways. But at the beginning, people with better jobs were doing a little bit better, so wages did a lot better than the employment count would suggest.
Now we're in a part of the cycle where some of the retail jobs are little bit hotter, and those are little bit less paying, so now some of the improvement isn't showing up as much in the wage data. So it's very important when people talk about units and dollars that you think about which is which, and what might the impact be if we put them both together.
Stipp: Bob, some great benchmarks for helping to understand all that economic data that we get. Thanks for joining me today.
Johnson: Thank you.
Stipp: For Morningstar, I'm Jason step. Thanks for watching.