Sober Look has two good, worth reading posts - "The story of two risk indices" and "The jump in U Michigan sentiment index is driven by consumer expectations". I'll try to return to the first post later but I thought I'd comment on the second one - which is about indicators tied to what we generally refer to as 'consumer sentiment' or 'consumer confidence'. SL makes some good points about these indices but he does not mention a larger point that I want to call out here - which is that these indices very clearly are not at all good gauges of absolute economic activity.
1) Let's start with SL's first point that the Thomson Reuters/University of Michigan's survey of consumer sentiment registered a reading in August 2011 that was as low as the reading in November 2008 - which, as SL says is "...a bit surprising and not at all intuitive, with anecdotal evidence suggesting that 08 felt far worse than 2011".
He's completely right about the non-intuitive nature of index values - but it is also completely non-intuitive when you look at non-anecdotal evidence. For example:
- The change in real GDP in Q4 2008 was -8.9% whereas in Q3 2011, GDP changed by +1.8%
- Retail sales changed at an annual rate of more than -10% in Q4 2008 compared to +6 to 8% in Q3 2011
- Industrial production and capacity utilization were much lower in Q4 2008 compared to Q3 2011
- The economy was losing jobs at the average rate of ~600,000 jobs/month in Q4 2008 as opposed to gaining on average ~150,000 jobs a month in Q3 2011
So, what exactly is the Thomson Reuters/University of Michigan Consumer Sentiment measuring?
2) Second, SL adds:
What drove the sentiment index to levels comparable to 2008? The index actually has two components: the Current Conditions Index and the Consumer Expectations Index. And it was the "expectations" component that pushed the whole index to this low point. In fact according to U Michigan, when it came to consumer expectations, the "end of the world" was coming in August of 2011 - a point on the index that was in fact lower than the expectations measure in 2008 (as shown in the chart below). This is puzzling. Were consumers reacting to all the market driven media frenzy in August that wasn't as prevalent in 2008? Were consumers not as focused on the state of the financial markets in the past as they are now?
On the other hand the "current conditions" component did not dive anywhere close to its 08 lows, which is more in line with other economic indicators.
So, basically, according to this measure of consumer sentiment, in August 2011 the Consumer Expectations index fell lower than the lowest value in 2008! Yet, all the above indicators measuring the real health of the economy (see above) were much worse in Q4 2008 than in Q3 2011 and Q4 2011. How is it that, not just the overall consumer sentiment index, but also its consumer expectations component, seems so out of whack with macroeconomic conditions (current and immediate future)?
3) Third, SL discusses another indicator on consumer confidence:
As a comparison let's take a look at another gauge of US consumer sentiment - the Conference Board Consumer Confidence Index. That index was behaving closer to the U Michigan Current Conditions component (above), with a dip that did not rival the 08 levels.
SL is only partly right about the Consumer Confidence index of the Conference Board. If you look at the chart posted on his website (see below), you'll see that it seems somewhat more intuitive overall in its behavior since the low in Q3 2011 was higher than the low in Q1 2009. But the Q3 2011 low was pretty much close to the low since in Q4 2008! So, this indicator is also, in many ways out of whack with the actual macro picture.
So, what gives?
Here's one clue.
First, here's the chart showing the Conference Board Consumer Confidence index.
Let's compare that qualitatively to this chart of the Dow Jones Industrial Average:
I think we can say a few things from this comparison:
- Prior to April 2009, the Conference Board Consumer Confidence Index seemed to roughly track the stock market. After April 2009, the index continued to approximately track the stock market as well, but the correlation (in terms of absolute values) was clearly weaker, even though qualitatively the trends have similarities.
- It could be that given the generally fragile economic environment in the aftermath of the Great Recession, the debt ceiling and Eurozone fears, and the fears of a double-dip recession, consumers were more spooked in Q3 2011 in some ways than in Q4 2011.
The most important takeaway though is that, although at a superficial level, the Conference Board survey of Consumer Sentiment has some relationship with the recession and recovery, in reality it is a very poor gauge of actual economic activity and is probably a much better gauge of stock market gyrations. After all, changes in GDP, employment, retail sales and industrial production in Q3/Q4 2011 have so far turned out to be positive and much higher than the changes to the same in Q3/Q4 2008 which were generally negative and often larger in magnitude. So, the fact that this gauge of consumer sentiment (the absolute value) was unable to quantitatively differentiate between a near-economic-depression and a slow-to-moderate growth economy indicates clearly that it is quite unreliable in gauging true economic activity, outside of any effects that stock market changes might have. I would also argue that the same is largely true of the Thomson Reuters/University of Michigan survey of Consumer Sentiment. As the charts posted by Sober Look show, the Consumer Expectations part of the index is more or less useless in understanding the underlying economic picture. It spiked up in the Q3/Q4 2008 timeframe (before dropping) even though output, employment, sales and income were plummeting or collapsing badly in that timeframe. It dropped to the 2008 lows in Q3/Q4 2011 despite a very different macro environment in 2011 - largely due to the perceptions tied to the Eurozone mess, US debt ceiling fiasco and stock market volatility. The Current Conditions index was only slightly better - the absolute numbers were still roughly comparable in Q3 2008 and Q3 2011 despite very different economic conditions.
If we want to use consumer sentiment numbers from these surveys, it might be worth looking at changes to the numbers and comparing them to the changes in economic activity in contemporaneous or future periods. Since I don't have all the raw data I can't do this myself but if someone else can, it might be worth a look. That still doesn't change the fact that the numbers as reported are fairly useless in understanding the true state of economic activity and their marginal value is also questionable if we already follow the stock market closely.
Comments