Bacon Cheeseburger Inflation and Google Autofills

By in
Bacon Cheeseburger Inflation and Google Autofills

Continuing with our “Off The Grid” economic indicators in Data, we have 2 more on offer. Again, these are meant to fill in some of the blind spots created by standard economic data points with real-world observations.

#1: The bacon cheeseburger (BCB) inflation index. Our idea here is that consumers’ inflation expectations are grounded in everyday purchases. We all buy food – 13% of the CPI – more frequently than almost anything else in the remaining 87% of the index. So, even though this component is more volatile than many other items in the government’s basket of goods and services it weighs heavily on consumer inflation psychology.

The boring approach would be to just look at the “food at home” category of the CPI:

  • June showed just 0.9% annual inflation here versus the 1.6% headline number.
  • If that does not fit with your own experience, perhaps it is because “food away from home” inflation was +3.1% in June versus the same month last year, headlined by “food from vending machines/mobile vendors” of 4.5%.

Our playful heuristic for assessing US food inflation is to combine the price changes for ground beef, cheese and bacon. Bad for the arteries, yes, but a catchall measure of real-world consumer inflation expectations:

  • Using data from 1981 to the present, we even-weight the annual price changes for these 3 ingredients.
  • Over time BCB inflation index catches a variety of cyclical and secular factors. For example, when it hits negative 4% on a rolling 12-month average basis (1984, 1992, 2010) you know there has recently been a recession that causes deflation.
  • What’s interesting now is that BCB inflation turned quite negative in late 2016 (December reading of -3.7%) and after a lackluster bounce (+3.0% in 1H 2019) is once again showing declines (-1.3% as of June 2019).

The bottom line is that, for all its quirks, the bacon cheeseburger inflation index has correctly predicted both recent real-world declines in consumer inflation expectations (NY Fed Survey of Consumer Expectations) and declines in 10-year TIPS inflation break-evens. Moreover, the current BCB deflation signal says we’re not out of the deflationary woods just yet.

#2: Google search autofills. When you start typing a query into Google an algorithm will try to predict the rest of the search request. It offers up 5-10 guesses based on what other users actually searched for, so how Google autofills something like “I want to buy” or “I want to sell” is based on common finishes to those searches.

At present, the top autofills for “I want to buy…” are:

  • A house
  • A timeshare
  • A car
  • Stock

Here is how these compare to autofills since 2011, when we started logging them quarterly:

  • In Q2 2011, the top suggested completions were: house, car, something and timeshare.
  • Things took a turn towards financial assets in Q2 2012, with the top autofills being: car, stock, something, and Facebook stock.
  • For Q2 2013 the top items were: house, gun, car, and stock.
  • At the peak of crypto-mania (Q1 2018): bitcoin, house, timeshare, ripple.

As for “I want to sell my”, here are the top autofill suggestions at present:

  • Car
  • House
  • Furniture

And selected comparisons over the years:

  • Q2 2011: car, house, and eggs. The last one is not about poultry farmers, but rather human eggs used in fertility treatments.
  • Q2 2012: car, house, and hair. The last one relates to a market for long lengths of uncolored human hair, used in wigs and weaves.
  • Q2 2013: car, house, and comic books.
  • All through 2015: “kidney” made the top 3 items, even though it is illegal.

What we make of the current autofills as compared to the historical data: Google’s algorithmically driven automatic completions indicate a “normal” US economy. Searches to explore the purchase of a car or house top the list, not crypto currencies. The “sell my” searches are similarly mundane, rather than “eggs” or “hair” or “kidney”. All in all, a positive set of signals. Or, at least, not bad…