Every year in July, the BEA releases a set of annual revisions.
As noted previously, a portion of those revisions are related to a construction spending data error that goes back ten years.
Let’s take a look at the revisions.
GDP as Revised
- 2013 Q1: +0.9 percentage points
- 2013 Q2: -0.3 percentage points
- 2013 Q3: +0.1 percentage points
- 2013 Q4: +0.2 percentage points
- 2014 Q1: -0.3 percentage points
- 2014 Q2: -0.6 percentage points
- 2014 Q3: +0.7 percentage points
- 2014 Q4: +0.2 percentage points
- 2015 Q1: +1.4 percentage points
- 2015 Q2: -1.3 percentage points
- 2015 Q3: unchanged
- 2015 Q4: -0.5 percentage points
Because of construction errors I was pretty sure 2015 would be revised lower and I expected 2014 to go up. The latter appears to be flat.
For 2013, upward revisions to inventory investment, exports, and residential and nonresidential fixed investment were partly offset by a downward revision to personal consumption expenditures (PCE).
For 2014, a downward revision to inventory investment, an upward revision to imports, and a downward revision to state and local government spending were offset by upward revisions to exports, PCE, and residential fixed investment.
For 2015, upward revisions to state and local government spending and to residential fixed investment, a downward revision to imports, and an upward revision to PCE were partly offset by downward revisions to exports and nonresidential fixed investment.
Multiple revisions for multiple reasons are in play. Constant revisions make predicting GDP a real crap shoot.
Mike “Mish” Shedlock
When you can be off over 1% per quarter with errors (mistakes, omissions, etc.), as this latest revised series indicates, what sense does it make to have a Big Central Planning Bureaucracy like the FED playing around “fine-tuning” the economy via constant monetary and interest rate interventions?
No doubt there are many more interpretations or misinterpretations, such as which data to collect or omit, how to weight, correct and extrapolate the data, etc. Perhaps just by varying interpretative decisions, GDP, unemployment and other numbers could differ several percent, just like election polls. In short, the “real economy” and abstractions like GDP, inflation and unemployment numbers are the difference between fiction and nonfiction.