CPI? More Like C-R-A-P
The flaws of CPI vs actual inflation
My dear readers.
I failed you.
I don’t know how this flew under the radar for so long. It is one of my favorite rants. I assumed I shared it with you over the 100s of articles. But alas, while writing a different post I searched for it and was shocked to see the topic was not yet covered.
Today….Today we fix that. So without further ado, I present to you why the CPI is an atrocious measure of inflation.
What Is CPI?
You have probably heard of the consumer price index, or CPI, as a measure of inflation. It is tracked and watched and the FED uses it as a key economic indicator of inflation when making interest rate decisions.
But what exactly is CPI?
CPI is a “is a measure of the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services.” It is calculated by the Bureau of Labor Statistics (BLS) and published monthly.
The CPI has a wide range of uses like:
Adjusting gov’t benefits (ie- Social Security cost-of-living adjustments (COLA))
Tax bracket adjustments
Contribution limits on retirement accounts (Roth IRA) and Health Savings Accounts (HSA)
Rents
Wage adjustments
Additionally, CPI is indirectly applied to many prices that are modeled and is taken into account for government economic decisions like raising the FED interest rate.
So with how much is riding on accurately calculating inflation. And with CPI being the primary metric used to report inflation. You’d think the gubmit would want to be fairly accurate with its calculation right….
How CPI Gets Calculated
Before going into all the deficiencies, lets first quickly go into the CPI calc. If you want to see the full calc in obtuse language, check here.
There are 243 ‘items’ that used in the bottom level of the calculation (think items like rent, used cars, commodities, various food groups, etc).
There are also 32 separate geographic regions that each item is measured in.
This results in (243 x 32) 7,776 item-region data points at the bottom level.
These 7,776 data points get weighted and rolled up into different subsets of the total CPI (think rural vs urban measures) based on the relative spending.
Most of the item prices the CPI gets from surveys. Over 200 of the 243 items from from the C&S (commodities and services) survey. IE- they ask businesses and people what the prices are. A few of the items are calculated from secondary sources (other indices or industry measures). And 24 of the smallest weighted items are estimated based on price movements of related items.
The CPI is supposed to represent the items purchased by over 90% of Americans day-to-day.
After the data is collected, there are adjustments made for the following:
Discontinued items get replaced with next closest option
Hedonic (Quality) adjustments
Seasonal adjustments
Imputation (ie- estimating the results of any missing data)
6-month chained estimator for housing (‘average over trailing 6 months)
Vacancy changes and aging adjustments in housing
Price reductions, fees, or product incentives that would alter the price paid by consumers
For sampling error and survey error adjustments
Honestly, the BLS site and documentation of the formulas is well put together for what it is.
So where is the issues?