Canadian Data Roundup

Recent Canadian data has been somewhat mixed. The Labour Force Survey for April was disappointing, but housing starts recovered. I discuss the economic data as well as the state of the Canadian inflation-linked bond market.

Commercial & Industrial Loans Holding Steady

The amount outstanding of Commercial and Industrial loans ("C&I loans") is an interesting indicator of the business cycle. It shows relatively smooth growth rates during an expansion, and contracts during recessions and sluggish growth periods (as seen in the last two cycles). And at present, the amount outstanding is increasing, which is consistent with the labour market data.

International Labour Market Comments

This is just a quick note related to the labour markets in Canada, the United States and Australia. The Canadian Labour Force Survey was strong in March, similar to the U.S. Household data. However, the housing market hangs over the Canadian economy. Australian labour data were weak in March, remaining on a sluggish trend.

How To Use Aggregate Indicators

The overall trend in U.S. economic data has wiggled lower recently. In this article, I discuss a few indicators of the overall direction of the economy, and I give an introduction to how economy-watchers should use these types of indicators.

U.S. Data: More Muddling

I have not seen a lot of reason to believe that the U.S. economy is accelerating or decelerating away from its recent trend of sluggish growth...

The Triumph Of The Straight Lines

The greatly-anticipated August 2013 U.S. Employment Situation report hit the markets today, and managed to convey no new information. The numbers which I view as the most important were exactly in-line (when rounded to 0.1%) with my highly sophisticated straight line projections.

The Treasury bond market recovered slightly in response, pulling the yield back into a trading range which is centered somewhere just below 3%. (I’ve seen technical analysts arguing that Armageddon will hit the Treasury bond market if the 10-year Note yield breaches 3%; I tend to think that the market could stay in a range even if the top end drifts a bit above 3%.) I see it as likely that the Fed will commence a slow tapering this month. But there is not enough information to say that the first rate hike will be before early 2015, which I view as being consistent with market pricing (with a reasonable risk premium).

I would characterise the Household Survey data for August as lousy, with 115 thousand jobs lost on the month; the aggregate employment ratio slipped back to trend at 58.6%. The rounded Unemployment Rate dropped by 0.1% to 7.3% as a result of the Participation Rate dropping by 0.2% to 63.2%. The Household Survey is a properly-designed random survey and is noisy month-to-month, but the underlying trends appear stable. The crux of the matter is whether the Participation Rate can keep dropping, as it explains why the Unemployment Rate is declining. However, the pool of workers who can drop out of the labour force is declining in line with the Unemployment Rate.

I would note that the Gallup-ing Unemployment Rate indicator did not prove too helpful this month.

Within the Establishment Survey, the hourly wage data looked good, but I will take a look at those numbers some other time. As for the headline Nonfarm Payrolls number, I think it’s a fairly silly statistic. Market participants like keying on that number, since it is much smoother than the Household employment number. That smoothness is the result of algorithms that can lead the number to miss turning points in the economy (however, later revisions will fix that problem, so it is not apparent to those who used the revised data in historical analysis). The market also seems to care about ridiculously small differences in the data; the difference between the actual outturn (169,000 jobs in August) versus an above consensus 219,000 is 50 thousand jobs, or 0.037% of the total number of employees. In other words, the same outcome if you followed normal rounding rules of 0.1% followed by other economic data. When you consider the difficulties involved in seasonally adjusting the labour force data in an environment of an evolving mix of industries, it is somewhat difficult to take the market behaviour too seriously.

On other fronts, the highly non-smoothed Canadian labour market data smashed expectations. I expect to discuss the situation in my home country next week.

(c) Brian Romanchuk 2013

That Gallup-ing Unemployment Rate

The spike in the Gallup U.S. Unemployment Rate poses some interesting questions heading into the August Employment Situation report (released September 6). (The unemployment data are available on the Gallup web site, and are copyright Gallup Inc.) Another muddle-through result seems the most likely outcome (which would imply that "tapering" by the Fed remains likely), but I feel that the risk of a surprise one way or another is a bit higher than usual. Given the laser-like focus of the FOMC on the labour market, the Employment Situation data will remain the main market-mover for bonds for some time to come.

The Gallup series is a 30-day moving average of a daily polling result. The average peaked on August 20-21 at 8.9%, and is now 8.8%. A quick eyeballing of the data seems to indicate that part of the rapid rise in the unemployment rate was a short-lived spike; and so some retracement will occur as the result of base effects in the coming weeks.

 The other Gallup series on underemployment and the percentage of the population working are not as straightforwardly scary looking, but they also paint a fairly negative picture. Importantly, the employment-to-population ratio is down year-on-year, and is roughly flat versus 2011. This is in-line with the official employment-to-population ratio (based on the Household Survey), which has roughly flatlined throughout the expansion.

However, the weakness in the Gallup series has not been corroborated by other high-frequency data. The weekly jobless number (released today) show no particular problems. It must be recognised that the unemployment claims data has a much longer history of use, and so there should be considerable analytical weight placed on the official claims data. The Gallup data only goes back to 2010, which is too short a period to do a proper seasonal adjustment, or create a model linking it to the official (NSA) Unemployment Rate. Moreover, the U.S. labour market has faced some important structural changes over the period, as the private sector extricated itself from the 2008-2009 meltdown (and the public sector launched some random fiscal policy changes). A changing employment mix should disturb the seasonal patterns, making it even harder to launch a seasonal adjustment.

As a final data note, the second estimate of 2013 second quarter GDP was released today. The real GDP number beat expectations, but I view the results are remaining mediocre. I prefer to follow nominal GDP growth, as the GDP deflator can be fairly loopy within a low growth environment on  a quarter-on-quarter basis. Nominal GDP was only 3.2% (annualised), which is below the sub-par 4% which has been the "normal" growth rate since the end of the recession.

(c) Brian Romanchuk 2013

Gallup Unemployment Rate Taking Off

Warren Mosler highlighted the ugly trend in the U.S. unemployment rate as measured by Gallup Inc.

The Gallup data provided is the 30-day moving average of daily polling data. It is not seasonally adjusted, so I am comparing this to the non-seasonally adjusted official (Bureau of Labor Statistics) U-3 unemployment rate data. (Most people are more familiar with the seasonally adjusted version of this series, which is trending down in a straight line.)

I am still getting to grips with the R programming language, so I'm not really able to do too much statistical analysis on this right now. In any event, if you wanted to seasonally adjust the data, it would be fairly unreliable (the usual algorithms want 5-7 years history). The structural changes that have resulted from the implosion of the U.S. labour market make those adjustments even harder.

But this bears watching, as the magnitude of this divergence appears unusual for mid-year. (The seasonal spike in unemployment that occurs after Christmas appears with a lag in the Gallup data, possibly because it is a moving average.) Since the dropping unemployment rate is arguably the most plausible piece of evidence in favour of tightening monetary policy, a reversal of that trend would provide very considerable relief to the embattled bond market.

(c) Brian Romanchuk 2013