US stock markets were almost absolutely flat yesterday, as seen investors digested the Federal Open Market Committee (FOMC) December minutes.
Recall that this was the meeting where the Fed decided to slow down their monthly bond-purchasing program to $75 billion per month, so the minutes were particularly interesting.
There were no major surprises (hence the stock markets collective yawn), but it was good to see that there was widespread agreement on withdrawing stimulus and they wanted to be cautious as they proceed.
I was talking to a client yesterday who basically said that the Consumer Price Index (CPI) created by the Bureau of Labor Statistics (BLS) to measure the average price changes on a market basket of goods is manipulated by the government.
This is an important question, since ten percent of our bond allocation is invested in Treasury Inflation Protected Securities (TIPs) that are intended to protect us from unexpected inflation but are calculated based on CPI. If CPI is bunk, then this form of inflation protection is useless.
Personally, I don’t think that the government manipulates CPI for it’s own dastardly purposes, but I do think that it’s important and useful to understand what CPI does and doesn’t do well.
First, CPI is a proxy for inflation and by definition a proxy is representative and not the real thing. The reason you create a substitute is that the measurement can get really complicated.
Some things are easy to measure: a cup of flour, a gallon of gas or a quart of milk. Other things are impossible to measure: freedom of speech, good health or beautiful sunset.
The same is true for CPI – some things are easy to measure like the price of a Coca-Cola from one year to the next, but others are more difficult. Consider the largest single component of CPI: owner’s equivalent rent.
If you don’t own your home and your landlord raises the rent, which is surely a price increase that should be considered in inflation calculations. Most people own their homes, however, so unless we want to just forget about homeowners for the purposes of calculating inflation, some estimate has to be made.
The BLS ‘solves’ this problem by simply asking ‘if someone were to rent your home today, how much would it rent for monthly, unfurnished and without utilities?’ The home-related website, zillow.com, includes a ‘Rent Zestimate’ that estimates the rent on a home based on statistical measures.
This owner’s equivalent rent is a squishy measure and it’s a large component of inflation, but squishy isn’t equal to manipulation in my view.
Another factor that makes CPI controversial is that its purpose has changed over time. In the beginning, it was intended to simply measure the price of a basket of goods and services in two different periods.
Over time, however, CPI was repurposed from an index of goods and services to a cost of living index, designed to show the cost of maintaining a constant standard of living.
With this objective, the BLS has to change it’s methodology to account for changes in the quality of goods and substitutions. A substitution is a change in purchase behavior based on prices, so if the price on steak goes up, consumers are more likely to buy ground chuck.
All of the subjectivity and complexity makes CPI a bit like an MC Escher drawing, so what should we do?
First, remember that CPI is proxy and then use it in your calculations. Then, recognize that the calculations aren’t as precise as you think they are. Rounding is OK.
I also find it useful to look at multiple indicators to make your best estimate. In addition to CPI, there is the Gross Domestic Product deflator – the measure favored by the Fed (although if you don’t trust inflation, you don’t trust the Fed either). Other measures include income, commodity prices and currency values.
For fun, I’ve included an infographic from the New York Times from 2008. I can’t believe it’s this old, I feel like I was playing around with it just yesterday. The data is outdated, but the broad scope is right and the layout is entertaining. Click here to see the infographic.