October 26, 2012

GDP Weak Once Again. Should We Rely on It?

October 26, 2012

GDP Weak Once Again. Should We Rely on It?

With GDP coming out today I am going to wait until next week to analyze it. Instead I want to present here some commentary written earlier this week by Rick Davis of Consumer Metrics Institute. Rick’s point is that GDP data 1) has, in the past, been unreliable sometimes reporting increasing GDP when in fact it was contracting and 2) after the report of 2nd Q2012 GDP sales tax data reported by the Census Bureau shows consumer sales contracting at a 2% rate year/year. BEA shows consumer sales increasing at a 2% rate year/year. They both can’t be right. Consumer spending being the largest component of GDP makes this an important point.

My point is simply that this is arguably the most important macroeconomic measurement and the folks in charge at BEA do an extremely poor job of measuring it. When I was a kid in grade school one of the nuns would have stamped these reports with a stamp saying “Must do better.” Rick’s comment at the very end spurred me to comment earlier this week that we could see growth as high as 3.8%.

The remainder of this section is Rick’s work.

At the Consumer Metrics Institute our mission has been to analyze changes in on-line consumer behavior for early signals about the macroeconomic health of the United States. So it makes sense that we are constantly asked to predict the direction that the GDP will be headed in the Bureau of Economic Analysis’ (BEA) next major GDP release (scheduled for Friday, October 26, 2012 and covering the third quarter of 2012).

In principal, we should be able to do that — given that we have a good quantitative handle on what consumers were actually doing as recently as yesterday. But unfortunately, it is not that easy to guess what kind of potentially election-turning numbers the BEA will publish on October 26th for a number of reasons — including myriad sampling and timeliness issues with their data, not to mention opaque “seasonal” adjustments and deflaters that profoundly change the numbers and get fixed only years after the fact.

So, what is actually happening with consumers now? And, what will be published on October 26th? Let’s visit those questions separately:

(1) What are consumers doing now?

Even that simple question can be difficult to answer unambiguously from official governmental sources. For example, for the 2nd quarter (2Q-2012) the government has reported two vastly different versions of the year-over-year growth in consumer spending:

– The BEA reported in their Table 1.5.6 Real Gross Domestic Product (under the “Section 1″ tab at their Interactive Data site) that “real” seasonally adjusted personal consumption expenditures (PCE — line 2 in their table) increased 1.9% year-over-year (2Q-2012/2Q-2011).

– The Census Bureau separately tallied actual total sales tax collections for the same two quarters (2Q-2012 and 2Q-2011) and reported a contraction in the “nominal” tax collections of about -0.46%. When converted into “real” dollars (using non-seasonally-adjusted year-over-year CPI-U from the Bureau of Labor Statistics (BLS)), the “real” quarterly sales tax collections actually contracted by something closer to -2.1% year-over-year.

Simply stated, for the same 2Q-2012/2Q-2011 reference period the BEA data and the Census Bureau data differed by an astonishing 3.9% in retail sales growth rates, with one source (the BEA) showing a modest growth while the other (sales tax collections) showing just the reverse — a modest contraction.

In general, at the Consumer Metrics Institute we are more inclined to trust data that includes a vast number of reporting entities and is both transparent and timely. In this case sales tax data has several things to recommend it: in essence it includes reporting from every merchant in taxable states, it is reported without seasonal “adjustments” and it is reported in nominal dollars (so we fully understand any deflaters used to convert the numbers into “real” dollars). On the other hand it is not as timely as we might like (the quarterly tallies are available about 90 days after the end of the quarter), the data is far noisier than the underlying commerce could possibly be (credit state bureaucracies for a certain level of timing nonchalance), and it completely misses 6 non taxing states (and Internet sales).

Nevertheless, the sales tax collections for the 2nd quarter of 2012 were still 5.5% (in “nominal” value) below the same quarter of 2008 — while in contrast after 4 years of “recovery” the BEA’s “nominal” PCE has reportedly risen 9.4% since 2Q-2008! However you want to slice it, the BEA is reporting cumulative consumer spending growth some 15% greater than states have seen in their tax collections. Blame the Internet if you will, but we suspect that the truth lies at least somewhere well south of the BEA’s numbers.

So, what did consumers do this past quarter? Unfortunately the Census Bureau sales tax data for the third quarter will not be available until late December. But for your amusement and bewilderment we do have another set of wildly conflicting data points to consider:

– The St. Louis Federal Reserve Economic Data (FRED) Real Retail and Food Services Sales data series RRSFS (which closely mirrors the data series used by the BEA for their PCE calculations) indicates that the annualized (and seasonally adjusted) “real” growth rate for the third quarter PCE was an eye-popping 7.5%.

– Our own data (based on measurements of on-line consumer demand for discretionary durable goods) suggests that the “real” PCE was actually contracting at about a -0.5% annualized rate (with the headline annualized GDP “growth” rate recording contraction at about a -0.4% rate).

Which number is right? Take your pick. One indicates continued and truly vigorous economic recovery, while the other indicates at least stagnation and possibly modest contraction. It all depends on which you want to believe.

2) What will the BEA report on October 26th?

First of all, the BEA has an abysmal record for accurately reporting economic growth data at times when the economic tide is turning. For the first full quarter of the past “Great Recession” (1Q-2008) the BEA initially reported growth to be +0.6%, and twice thereafter revised that number up — to a +1.0% annualized growth rate in their final revision. But some 40 months later they quietly revised the “growth” rate drastically downward — to a -1.8% contraction rate representing the onset of the “Great Recession.” So much for the accuracy of the “real time” prints from the BEA during a presidential election year.

One of the key points to remember about the headline number is that the math used is an annualization of the quarter-to-quarter changes — they essentially amplify those quarter/quarter changes by compounding them over the equivalent of a year (i.e., raising the quarter/quarter ratio to the fourth power).

As a consequence, the final “real” GDP aggregate number for the past quarter (in this case 2Q-2012) is just as critical to the new headline number as the new quarter’s “real” numbers. And the lower the prior quarter’s numbers are, the better the new headline looks.

And the final revision to last quarter’s (2Q-2012) “real” numbers was sharply downward. That was not a surprise in itself, since a weakening recovery was certainly plausible given all of the other domestic and global data releases. But those of you inclined to cynicism will note that the last revision considerably lowered the bar for the quarter/quarter ratio that creates the new headline number.

Furthermore, the BEA has their own set of deflaters, which they use to convert “nominal” numbers into inflation-adjusted “real” numbers. For most of the recent “recovery” those deflaters have reflected a lower inflation rate than similar data provided by their sister agencies (particularly the BLS). As a consequence the headline GDP growth rate has been consistently higher than what it might have been had they used (for example) BLS deflaters.

Curiously, this past quarter (2Q-2012) has been a notable exception to the BEA’s recent string of consistently low deflaters. In fact, the surprisingly high deflaters this past quarter arguably pushed the quarter/quarter comparison bar even lower — and could easily amplify the 3Q-2012 headline number if the deflaters simply reverted to historical form.

Given the recent retail sales data from FRED, the recent downward revision to the 2nd quarter numbers and the peculiar recent treatment of deflaters, we suspect that the BEA could report PCE and GDP significantly (maybe even shockingly) above consensus expectations.

What will they report on October 26th? Heaven knows. But don’t be surprised if the economy miraculously improves …