Friday, September 2, 2016

BLS Lies to US Again, 204K Jobs Added in August

Seasonally Adjusted Non-Farm Jobs Change 2013-Present:  Comparison between Griz Method/Simple YoY and BLS Monthly Reports
Source Data:  BLS
Total Nonfarm 2016-9. xlsx




Seasonally Adjusted Non-Farm Jobs Change 2013-Present with 2016 Projections:  Comparison between Griz Method/Simple YoY and BLS Monthly Reports and my projections from June 2016 for monthly gains using the Griz Method and my expectations for the BLS reports.
Source Data:  BLS
Total Nonfarm 2016-6. xlsx

I thought there was a good chance the BLS would sandbag the August job gain number to ensure no September rate hike.  September/October is a historically terrible time for the stock market and a September rate hike certainly wouldn't help that situation.  However, I certainly wasn't expecting the number to be an anemic 151k.  But it just goes to show how hard it is to predict the magnitude and timing of a lie.

So here is some simple math to show what the real seasonal adjusted job gain was for August.  You can look at several of my earlier posts to see YoY non adjusted employment totals that show the annual employment pattern has matched for the last several years justifying simple YoY comparison.
([Aug 2106]-[Aug 2015])/12 = Seasonal Adjusted Gain for Aug 2016
Insert data from BLS not seasonally adjusted data set in thousands
(144424-141973) = 2451 thousand or 2.45 million more jobs in Aug 2016 than Aug 2015.   Divide that by 12
2451/12= 204.25 thousand monthly rate of gain

Again we see massive volatility and variability in the BLS seasonally adjusted number which is intended to remove variability.  But if we look at the data created with the Griz/YoY Method the resulting data is a smooth transition with little volatility.  If we compare my predictions from early June vs. the actual not seasonally adjusted jobs total/Griz Method job gains you get the following comparison.

Griz Projected Actual  Error %
June 200 207 -3.30%
July 200 202 -1.07%
August 195 204 -4.53%
Total 595 613.25 -2.98%

Notice my bearish/very conservative estimate for jobs gains made in early June have varied only 4.5% at worst from the actual reported total for the month.  And the worst error is in August, if the recent pattern repeats, August was overestimated and will be adjusted downwards in the coming 2 months.  Still the total error over 3 months is less than 3% in magnitude and will likely improve as adjustments come in.  3% error on a 3 month projection which constituted the absolute low bound is pretty good.  If a high, low and average of the range was predicted for each month you can see that the average would have almost no error from the actual reported data.  

So here is my question?   Why does the BLS monthly report swing so wildly?  Very conveniently it seems to have no guaranteed no stock market killing rate hikes until post election in December or possibly just 2 business days ahead of the election in the first week of November not leaving the markets enough time to crash and potentially derail Democrats chances in the election, given they are running on the strength of the economy.  A crashing stock and bond market doesn't exactly instill confidence in voters.

Also note:  An actual 151k monthly job gain rate is pretty ugly.  Indeed early in the year similar reports did a great job of capping stock market gains and pushed bonds up.  However, today this report is pushing up risky stocks and has treasury rates heading higher, signs of a strengthening economy/market, not one of weakness indicated greatly slowing employment gains.

Prediction:
My data shows about 244k jobs are now "banked" since January 2016 that will be used to over report monthly gains in the next few months.  The reality is job gains are about 200k/month so that means for September, October, November and December the average reported monthly gain is likely to be about  261k-265k.  If they decide to actually spread it out into January or even February, the average "reported" monthly gain for the next 9 months will be about 240k.