A Tale Of Two Distributions; Or Are These The Economic Numbers The BLS Now Openly Makes Up?
A funny thing happened on the way to American central planning: normal distributions became abnormal. To wit - the classical, Gaussian distribution chart, which lies at the basis of all modern statistics and offshoots thereof, such as quant theory, apparently is not good enough for the wonderful data aggregators and distributors at the all important Bureau of Labor Statistics, which is in charge of such key economic metrics as the unemployment rate, CPI, PPI, home sales, and virtually everything that tends to have a huge headline impact on stock futures (because let's face it, nobody trades during regular hours anymore). Curiously of the 25 or so data items tracked by the BLS, the vast majority have been revised over 50% of the time over the past decade. All, that is, expect for the most important, and politically critical ones: the unemployment rate (easily the only number the vast majority of the population understands), and consumer prices (which is the only number to direct impact the federal budget). In other words, as the chart below demonstrates, while the BLS has the track record of a blind and retarded monkey throwing a dart at a wall full of numbers when presenting initial economic data, something which Gaussian distributions would say is perfectly normal when running a $13 trillion economy, it has perfected confidence intervals and data estimates when dealing with the most economically sensitive and critical data. Whether the BLS has hired the same oracular prop traders that allowed Goldman to lose money on zero days in Q1 when calculating some numbers (but not others), or whether the BLS just spews forth what some excel model dictates (possibly the same used by Moody's that would crash upon imputing negative growth assumptions), while pretending to use traditional "ecostat" sampling, estimating and adjustments may be worthy of far more debate than currently afforded. Because the last thing an increasingly more cynical American public would want to believe is that the government is, gasp, lying to it.
Presenting Exhibit A:
So let us paraphrase: we have no complaint with the BLS for the massive amount of revisions on the bulk of the ecodata it collects and compiles. This is what Gaussian probability would dictate (although in all honesty we would prefer that prior revisions had a 50/50 distribution instead of the 90%+ downward revision as seen in some recent data series). What we do want to know, is just how does the BLS compile and analyze data on the following items which have a several standard deviation variation from the average revision rate for all other economic series:
- Unemployment Rate
- Core Consumer Prices
- Core Producer Prices
- Consumer Price Index
- Producer Price Index
Because anytime something is too perfect, especially emanating from the government, it isn't. Especially in this case, when these are precisly the numbers that have the biggest impact on the stock market and on economic policy. And the last thing we would want is to speculate that the five most important numbers tracked by the BLS have absolutely no credibility because they not only do not reflect a practical reality, but merely are the artefact of possibly the same computer used by the FRBNY to reroute stock orders into the Citadel dark pool system.
BLS- the podium is yours.
Courtesy of John Lohman