The "base year"

I have a good feel for what a "base year" is. It is used a lot when "nominal" (actual-price) data is  converted to "real" (the-prices-never-change) data. For example, I use it when I divide the rising price level out of nominal GDP to get the "real" values -- values that exist because the economy actually grew, not because prices happened to go up. The "base year" is the year where the real GDP and the nominal GDP are the same. There is always one year like that for "real" data. If you look at a FRED graph of inflation-adjusted data (like Real GDP)  the vertical axis will be labeled something like "Billions of Chained 2017 Dollars". 2017 is the base year for that data. On a graph that shows both "real" and "nominal" GDP, the lines cross in the base year, because the values are equal in that year.

So I have a pretty good feel for the base year, but it comes from doing arithmetic. When I wanted a definition in words, I looked it up. The featured snippet comes from Investopedia, and it says "A base year is the first of a series of years in an economic or financial index."

The first of a series of years. I wish! If the base year was the first year for real GDP, the graph would show real GDP hanging low while inflation pushes nominal GDP up and up and up, like this:

Graph #1: The blue line is real GDP. The red line shows the result of inflation.

To make this graph, I took the shows both "real" and "nominal" GDP graph noted above, divided the real values by 1191.124 (the 1929 value of Real GDP in the FRED data), and multiplied the real values by 104.556 (the 1929 value of nominal GDP in the FRED data. Basically, I took the real data and scaled it down to make the first value -- the 1929 value -- equal to the nominal 1929 value.

Yes, I think the base year should, as a rule, always be the first year of the data on a graph. But that seldom happens. In fact, for Real GDP they move the base year frequently. At ALFRED they list the "units" used for the Real GDP in their archive:

• Billions of 1987 Dollars
• Billions of Chained 1992 Dollars
• Billions of Chained 1996 Dollars
• Billions of Chained 2000 Dollars
• Billions of Chained 2005 Dollars
• Billions of Chained 2009 Dollars
• Billions of Chained 2012 Dollars
• Billions of Chained 2017 Dollars

There is a move to a more recent base year every three to five years. Actual low prices become a more and more distant memory. This is how economists work. Standard practice, apparently.

The search that turned up the Investopedia definition also turned up one that I like better: this one, from  europa.eu:

In the calculation of an index the base year is the year with which the values from other years are compared. The index value of the base year is conventionally set to equal 100.

the year with which the values from other years are compared: yes.
the first of a series of years in an economic or financial index: no.

the base year is conventionally set to equal 100: yes
it is typically set to an arbitrary level of 100: no.

Oh, and both Investopedia and the europa (EuroStat) site think in terms of the price index having a base year. I think in terms of inflation-adjusted data having a base year. They are right. The data does have a base year, but it inherits the base year from the price index used to strip inflation out of the numbers. I can live with that.