I received an interesting chart today from Ralph Dillon at Global Financial Data in regards to GDP consensus estimates.
GDP Consensus Estimates vs. Actual GDP
click on chart for sharper image
Dillon writes ...
I am by no means advocating that it is an easy task trying to predict what the economy will do in the next quarter or over a fiscal year, but when the forecasts compared to reality are so wrong time and time again, why do we place so much emphasis on forward looking data that is just flat out incorrect?
Over the last several years, I have followed and made notes of everything related to GDP because I am constantly perplexed as to how a broad range of economist (consensus) are constantly cutting and revising GDP estimates. For instance, just last Friday Goldman Sachs hiked Q2 GDP to 3.9%. This is on the heels of an awful Q1 that was blamed on everything from weather, deteriorating economic environment or even just a flat out surprise and getting caught flat footed. This morning, 2 business days later, they revised their Q2 GDP estimate to 3.5%. What changed in the last 2 days that would warrant a higher revision and then ultimately a lower one to 3.5%. If it’s going to be wrong, then why not just stick with your original estimate and not continue to move the bar up and down?
Let’s take a look at the last 21 quarters to help illustrate my point. After looking at all of the consensus estimates and comparing them to reality, I found that every single quarter has been wrong. Not saying that one or two economists didn’t get it right, but overall consensus has been flat out wrong! 21 quarters and 0-21. These are terrible statistics by any measure. In addition, the average miss in either direction has been $1.25 Billion dollars. All things considered, that is a huge miss from a percentage standpoint. I can’t help but think that perhaps a better way to forecast GDP going forward is to throw darts at GDP estimates on a dart board. If you recall from the internet bubble days, that was more accurate and successful way to pick stocks than most portfolio managers had at actively picking individual stocks. If that proves to be more successful than the current track record, then perhaps we should use that for forecasting. My point being, is that it can’t be any worse than 0-21 over the last 5 years.
How do you interpret the data? When they use terms like “probably, solid bounce back, grow strongly, gradually improve and distorted by one off numbers”, you would think that they are speaking with conviction to what they believe. But in reality, they are just words meant to reassure the investing public that all is good in the investment world when in fact that may not be entirely accurate. In the late 1990’s, Painewebber’s Chief Strategist Ed Kirschner wrote quite a bit about thematic investing. One of the things he spoke about often was all the revisions that take place on Wall Street with everything from GDP, earnings, employment to economic statistics. For every estimate, there was a revision. For every earnings report, there was a restatement. I recall him saying over and over again in his presentation, “ does anybody really know what time it is?”. Some 15 years later, I am afraid that no one still does.
Given that government spending adds to GDP by definition, How much of GDP is deficit financing of demand brought forward?
How much of GDP is in any way productive?
For the answer to the latter question, think about things like bombing Iraq and Afghanistan to smithereens. Also think about government contracts that pay people for doing nothing.
For a recent example of the latter, please see $1.2 Billion Obamacare Contract Pays Workers to Do Nothing.
If the government paid people to spit at the moon it would add to GDP by definition. Yet, paying people to spit at the moon makes as much sense as paying people to do nothing.
Worse yet, spitting at the moon actually makes more economic sense than many of the needless destructive wars we fought. Deficit spending and stupid wars realistically ought to subtract from GDP.
And what about hedonic pricing and imputations?
For those not familiar with hedonic pricing, the government accounts for the presumed value of goods sold, not actual prices. For example the prices of computers drop every year even though speed and memory increase continually. Rather than calculating the actual price of goods sold, the government adjusts prices to say computers sold for more than they really did.
Here's an example of imputations: The government adds the "imputed" value of free checking accounts to GDP as a valuable service rendered but not paid for.
I propose that makes as much sense as imputing the value of free sex rendered to married husbands on the grounds that if they weren't married, they may be paying prostitutes.
Yet nearly all economists focus on GDP as the be all and end all in economic measurements.
Mike "Mish" Shedlock