Note from dshort: With Thursday's release of the 2012 Q4 GDP Third Estimate, I've spent some time updating my occasional analysis of the government's calculation of "Real" GDP.
How do you get from Nominal GDP to Real GDP? You extract inflation from the numbers. The Bureau of Economic Analysis (BEA) uses its own GDP deflator for this purpose, which is somewhat different from the BEA's deflator for Personal Consumption Expenditures and quite a bit different from the better-known Bureau of Labor Statistics' inflation gauge, the Consumer Price Index.
The charts below show quarterly Real GDP since 1960 with the official and three variant adjustment techniques. The first chart is the official series as calculated by the BEA with the GDP deflator. The second starts with nominal GDP and adjusts using the PCE Deflator, which is also a product of the BEA. The third adjusts nominal GDP with the BLS (Bureau of Labor Statistics) Consumer Price Index for Urban Consumers (CPI-U, or as I prefer, just CPI). The forth chart, prompted by several requests, adjusts nominal GDP using the Alternate CPI published by economist John Williams at shadowstats.com.
I've calculated the latest GDP in all versions to two decimal places to help highlight the differences.
Suggestion: Click on any of the charts below and use the links at the top of the chart page to toggle between the versions for a closer comparison.
The PCE and GDP deflators are both the work of the Bureau of Economic Analysis, but as a comparison of official GDP and the PCE-deflated variant illustrate, the rate of inflation can vary significantly, depending on which series the BEA is deflating.
The CPI comes from a different government agency, the Bureau of Labor Statistics, and is calculated quite differently. As an inflation measure, it is much better known than the GDP and PCE deflators, and its growth rate has been higher than the two BEA metrics (see this illustration). If we use CPI as the deflator to compute Real GDP, we see a lower mean, a higher volatility. Note: For an apples-to-apples comparison, I'm using the seasonally-adjusted quarterly CPI (the CPIAUCSL in the FRED repository) since the PCE and GDP deflators are both seasonally adjusted.)
Ever since my original post of this series, I have received several requests for an additional version using the Shadowstats alternate CPI as the deflator.
I find this "alternate Real" GDP to be interesting (in a bizarre sort of way), but I personally see no credibility in the hyper-negative GDP it produces. On the contrary, I see this chart as further evidence that the alternate CPI, despite its popularity among many critics of government data, is a misguided concept.
- Quarterly GDP dates from 1947, CPI has been tracked since 1913, but PCE only goes back to 1959. Thus I used 1960 as the start date for the chart series. The 1959 inception of PCE explains why the 10-year moving average in the second chart starts later than the other two.
- The three deflators are available from the Federal Reserve Bank of St. Louis data repository: GDP Deflator, PCE Deflator, and CPI-U (seasonally adjusted)
- For the PCE and CPI-U data series I use quarterly averages to match the quarterly GDP deflator.
- I created the charts in Excel using a two-step process. I calculated the real quarterly GDP for each deflator against the nominal GDP series. I then used this function to calculate the individual quarters: ((Q/previous Q)^4-1)*100