In Depth Analysis: CalculatedRisk Newsletter on Real Estate (Ad Free) Read it here.

Wednesday, February 02, 2011

Daily Color: D-List Data

by Calculated Risk on 2/02/2011 02:10:00 PM

Not all data are created equal.

For me, the ‘A List’ for understanding the current situation includes the monthly employment report from the Bureau of Labor Statistics (BLS), and the quarterly GDP report from the Bureau of Economic Analysis (BEA). My ‘B List’ usually includes several housing reports, the ISM manufacturing survey, retail sales and the monthly Personal Income and Outlays report from the BEA.

This brings up a key point: these lists are not static.

As an example, right now initial weekly unemployment claims is ‘B List’ data. This is a high frequency indicator for the labor market. I watch this closely when I think a recession is possible, during a recession, and then during the early stages of a recovery (like right now). During an expansion, initial weekly claims are ‘D List’ at best; “Don’t call us, we’ll call you!”

Watching initial weekly claims helped me call the 2007 recession in real time. However this data is not forward looking. At the end of 2006, when I predicted a recession would start in 2007, I wasn’t using weekly claims at all – I was mostly using housing data and my sense of how the housing bust would play out.

Deciding what data is important and when comes from experience.

The data released this morning – the Mortgage Bankers Association (MBA) Purchase Activity Index and the ADP Employment report – are pretty much ‘D-List’ data. They give us hints about other economic data (the MBA index about home sales, and ADP about the BLS employment report). Some people would argue either or both are a little more important – OK, call them ‘C-List’ data (I’m not here to quibble, but ‘D-List’ made for a better post title).

The MBA index has been very weak since the end of the housing tax credit last year. This weakness suggests that home sales will be weak for at least the next couple of months (homebuyers usually apply for a mortgage 30 to 60 days before closing on a home purchase). It is also important to remember that a fairly large percentage of recent homebuyers have been paying cash (many of these purchases are low end homes being bought by investors) and cash buyers aren’t captured by the MBA index.

It is also important not to use data in a vacuum, and the MBA index provides an excellent example. Here are a couple of articles quoting former Fed Chairman Alan Greenspan in 2006:

From Bloomberg in August 2006: Greenspan Says `Worst' May Be Past in U.S. Housing

Former Federal Reserve Chairman Alan Greenspan said the ``worst may well be over'' for the U.S. housing industry that's suffering its worst downturn in more than a decade.

Greenspan, speaking at a conference in Calgary today, pointed to a ``flattening out'' of weekly mortgage applications after they went down ``very dramatically.''
And from Reuters in October 2006: Greenspan: Housing market worst may be over
The U.S. housing market appears to be emerging from its recent travails and the “worst may well be over,” former Federal Reserve Chairman Alan Greenspan was quoted as saying on Friday.

“I suspect that we are coming to the end of this downtrend, as applications for new mortgages, the most important series, have flattened out,” Greenspan said at an event in Calgary, Canada ...
The housing downturn had just started, and I made fun of Greenspan's comments in 2006!

Here is a repeat of the MBA index graph from this morning:

MBA Purchase Index Click on graph for larger image in new window.

You can see the "flattening out" in the middle of 2006, and the increase at the end of 2006 and again in 2007.

This brings up a couple of points:

• The MBA data was NEVER the “most important series”.

• In mid-2006, the MBA index did flatten out, and in late 2006 the index increased (and increased further in 2007). At that time I spoke with some mortgage brokers, and there was clear evidence of homebuyers applying for mortgages with multiple brokers - this lead to some double counting by the MBA. And in late 2006 the increase was because mortgage brokers started going out of business (this skewed the data, because the MBA samples only certain large brokers – and the large brokers were getting more applications as the weaker companies went under). I identified these flaws and stopped using the MBA index, but Greenspan blindly used the index and drew the wrong conclusion.

The lesson: Never listen to Greenspan Always ask if the data is being impacted by changes in behavior or sample.

Now to the ADP employment report: this report is intended to help predict the BLS employment numbers, but the record on a monthly basis is very spotty. Just look at December: The ADP report showed 297,000 private sector jobs added, but the BLS report only showed 113,000 private sector payroll jobs added (103,000 Total). Not close.

Here is the ADP purpose and methodology:
Employment is an intrinsically important statistic. Furthermore, financial markets react, sometimes strongly, to “surprises” in the BLS estimates of establishment employment that might signal future changes in monetary policy. Hence, information that helps analysts anticipate monthly changes in employment is valuable. ... The ADP National Employment Report ... can be used, in real time, to improve upon consensus forecasts of the monthly change in establishment employment.
The report sure doesn't seem useful in "real time" to improve on forecasts. Sometimes the ADP report is close. Sometimes it is not (like last month). The ADP data is from a statistical black box based partially on the BLS data (as opposed to a completely independent report of payroll jobs added), and it is not as useful as some analysts had hoped.

Right now I think the ADP report is suggesting stronger job growth (a good thing). The ADP report has averaged 217,000 jobs added per month over the last two months, and maybe that indicates the BLS report will be higher than expected (current expectations are for an increase of 150,000 payroll jobs in January). But the ADP report really isn’t useful in predicting the BLS numbers on a month to month basis. I just use it as a “hint”.