When someone shares with you something of value, you have an obligation to share it with others.
Wednesday, 7 October 2009
Darkest Before The Dawn - BNP
Last week’s disappointing non-farm payrolls report rounded off a week of relatively tepid US macro-data that has challenged the bullish foundations on which US equities continue to precariously reside.
Last week’s disappointing non-farm payrolls report rounded off a week of relatively tepid US macro-data that has challenged the bullish foundations on which US equities continue to precariously reside. Certainly much of Friday’s labour market report made uncomfortable reading. Job losses in September were only in truth fractionally worse than expected (-263k vs. market expectations of around -200k) but the first estimate of the benchmark revisions to the data for the period April 2008 to March 2009 found that an extra -824k jobs were probably lost in this period than the BLS had previously estimated. For those of us that have long questioned the validity of the BLS’s so called ‘birth-death’ model persistently adding net jobs even in the depths of recession, these revisions represent vindication, albeit belatedly. Notably the benchmark revisions were even larger than the +717k jobs that the birth-death model magically created in the year to March 2009.
And this matters. Inflated non-farm payroll numbers in 2008 helped mask how rapidly the real economy was deteriorating even in advance of the intensification of the crisis last fall and so increased the chance of a serious policy mistake by the Federal Reserve last year. Better quality payroll data would have made it clearer when the recession began and would have greatly reduced the chance of the Fed hiking rates last summer. What is perhaps most alarming is that, despite these revisions the birth-death model nonetheless is still continuing to assume that more small businesses are adding jobs than shedding them adding to the likelihood of further downward revisions in the future!
The dubious nature of these adjustments is why many economists have come to favour the alternative, so called household survey of jobs which is actually used to calculate the unemployment rate and other key labour market ratios. This measure also has limitations being based on a considerably smaller sample than the payroll data and is much more volatile on a month-to-month basis. Still, this measure reported jobs fell by -785k in September; the survey’s biggest monthly drop since January. Despite this, the unemployment rate ‘only’ rose to 9.8% in September thanks to a further dip in the participation rate. The employment ratio (employment divided by population) however slumped to 58.8% from 59.2% in August; its lowest level since early 1984. And this ratio has now collapsed by a record 3.1% percentage points over the last year.
Discouraged workers dropping out of the labour force are captured in the broadest measure of labour underutilisation; the so-called ‘U6 measure’, which also includes frustrated part-time workers. This too continued to march higher in September, rising to a Depression-style 17.0% (vs. 16.8% in August). In terms of numbers, the rise in ‘U6’ over the last year equates to over 9 million workers! And if this were not bad enough, a final unsavoury aspect of Friday’s report was the continued rise in unemployment duration. In particular, the share of unemployed out of work for less than 5 weeks has now dropped to an all time low of just 19.4%. The flip side is that the share of the unemployed out of work for more than 26 weeks has now climbed to an all-time high of 35.6%, adding to the sense that many of the jobs lost in this downturn will be permanent. All told, these three metrics cement the sense that the US economy is wrestling with the biggest labour market shock since the 1930s.
While all these metrics are breathtakingly bad, there were nonetheless some glimmers of hope in Friday’s report. As we have highlighted before, given all the suspicion and confusion over the BLS’s adjustments to the payroll numbers, one alternative is simply to strip out both the seasonal adjustments and birth-death model changes that the statisticians apply and look at the raw, unadjusted payroll data. Given the lack of seasonal adjustment, the trend in these raw data can obviously only be looked at on a y/y basis but this can be thought as an useful and relatively clean alternative measure of payroll momentum. Importantly, this measure, which we first highlighted several months ago, turned positive in September for the first time since October 2007, suggesting that the labour market momentum is now turning. If this seems unnecessarily abstruse, a more orthodox lead indicator of total payrolls – temporary help employment – offered up similar grounds for optimism in September. Just 1.7k temporary help jobs were lost in September and over the last three months, temporary help employment has almost stabilised, suggesting that greater stability in aggregate non-farm payrolls may be closer to hand than typically thought. Finally, the historical record offers some additional encouragement. The last recession is officially dated as ending in October 2001; non-farm payrolls stabilised in early 2002 with the first positive print arriving in February 2002. The current recession will probably be dated as ending in July of this year. Dark as much of the September labour market report undoubtedly was, brighter news may therefore be just around the corner. But even if the biggest labour demand shock since the 1930s may soon flatten out, remember that the US needs to generate 100-150k jobs per months just to keep pace with rising labour supply and so prevent the unemployment rate from pushing higher.
2 comments:
Darkest Before The Dawn - BNP
Last week’s disappointing non-farm payrolls report rounded off a week of relatively tepid US macro-data that has challenged the bullish foundations on which US equities continue to precariously reside. Certainly much of Friday’s labour market report made uncomfortable reading. Job losses in September were only in truth fractionally worse than expected (-263k vs. market expectations of around -200k) but the first estimate of the benchmark revisions to the data for the period April 2008 to March 2009 found that an extra -824k jobs were probably lost in this period than the BLS had previously estimated. For those of us that have long questioned the validity of the BLS’s so called ‘birth-death’ model persistently adding net jobs even in the depths of recession, these revisions represent vindication, albeit belatedly. Notably the benchmark revisions were even larger than the +717k jobs that the birth-death model magically created in the year to March 2009.
And this matters. Inflated non-farm payroll numbers in 2008 helped mask how rapidly the real economy was deteriorating even in advance of the intensification of the crisis last fall and so increased the chance of a serious policy mistake by the Federal Reserve last year. Better quality payroll data would have made it clearer when the recession began and would have greatly reduced the chance of the Fed hiking rates last summer. What is perhaps most alarming is that, despite these revisions the birth-death model nonetheless is still continuing to assume that more small businesses are adding jobs than shedding them adding to the likelihood of further downward revisions in the future!
The dubious nature of these adjustments is why many economists have come to favour the alternative, so called household survey of jobs which is actually used to calculate the unemployment rate and other key labour market ratios. This measure also has limitations being based on a considerably smaller sample than the payroll data and is much more volatile on a month-to-month basis. Still, this measure reported jobs fell by -785k in September; the survey’s biggest monthly drop since January. Despite this, the unemployment rate ‘only’ rose to 9.8% in September thanks to a further dip in the participation rate. The employment ratio (employment divided by population) however slumped to 58.8% from 59.2% in August; its lowest level since early 1984. And this ratio has now collapsed by a record 3.1% percentage points over the last year.
Discouraged workers dropping out of the labour force are captured in the broadest measure of labour underutilisation; the so-called ‘U6 measure’, which also includes frustrated part-time workers. This too continued to march higher in September, rising to a Depression-style 17.0% (vs. 16.8% in August). In terms of numbers, the rise in ‘U6’ over the last year equates to over 9 million workers! And if this were not bad enough, a final unsavoury aspect of Friday’s report was the continued rise in unemployment duration. In particular, the share of unemployed out of work for less than 5 weeks has now dropped to an all time low of just 19.4%. The flip side is that the share of the unemployed out of work for more than 26 weeks has now climbed to an all-time high of 35.6%, adding to the sense that many of the jobs lost in this downturn will be permanent. All told, these three metrics cement the sense that the US economy is wrestling with the biggest labour market shock since the 1930s.
While all these metrics are breathtakingly bad, there were nonetheless some glimmers of hope in Friday’s report. As we have highlighted before, given all the suspicion and confusion over the BLS’s adjustments to the payroll numbers, one alternative is simply to strip out both the seasonal adjustments and birth-death model changes that the statisticians apply and look at the raw, unadjusted payroll data. Given the lack of seasonal adjustment, the trend in these raw data can obviously only be looked at on a y/y basis but this can be thought as an useful and relatively clean alternative measure of payroll momentum. Importantly, this measure, which we first highlighted several months ago, turned positive in September for the first time since October 2007, suggesting that the labour market momentum is now turning. If this seems unnecessarily abstruse, a more orthodox lead indicator of total payrolls – temporary help employment – offered up similar grounds for optimism in September. Just 1.7k temporary help jobs were lost in September and over the last three months, temporary help employment has almost stabilised, suggesting that greater stability in aggregate non-farm payrolls may be closer to hand than typically thought. Finally, the historical record offers some additional encouragement. The last recession is officially dated as ending in October 2001; non-farm payrolls stabilised in early 2002 with the first positive print arriving in February 2002. The current recession will probably be dated as ending in July of this year. Dark as much of the September labour market report undoubtedly was, brighter news may therefore be just around the corner. But even if the biggest labour demand shock since the 1930s may soon flatten out, remember that the US needs to generate 100-150k jobs per months just to keep pace with rising labour supply and so prevent the unemployment rate from pushing higher.
Post a Comment