Have we transitioned from the disinflationary trend of the past 40 years to a new inflationary regime? It's probably the most important question facing investors today. The types of investments one would want to own are quite different when inflation is rising than when it is falling (i.e., deflation; or even disinflation, which is still positive inflation but at decreasing rates).

Because of the importance of this topic, we've devoted the cover article of the upcoming May issue of SMI to the topic, which should be released a week from today. But as I've been writing that article over the past couple of weeks, I've realized there isn't space to cover everything. So given that a fair amount of material is going to be left on the cutting room floor, I thought I'd break some of that out here before (and after) the main article arrives.

CPI basics

As a starting point, I thought it would be helpful to dig into the primary measure of inflation, the Consumer Price Index or CPI. The U.S. Bureau of Labor Statistics offers this definition: "The Consumer Price Index (CPI) 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."

Fair enough. But as Joseph discussed a week ago, the way the government constructs its basket of goods and services may not be a particularly close match to the goods and services you consume. His post focused specifically on the huge weighting shelter gets in the CPI calculation and the fact that right now, home prices are vaulting higher while rents have been declining. This type of disparity happens a lot and is one reason people are often distrustful of the government's official CPI numbers.

Ultimately though, the government has to use some type of standard and no matter how they weighted such a basket, there would be plenty of people who feel it's not representative. Unfortunately, moving beyond that issue we run squarely into two more significant issues with the CPI calculation.

1. Substitution. The way the CPI is calculated assumes that consumers will substitute one good for another if prices rise. For example, if the price of beef rises, CPI assumes consumers will shift to eating chicken. To some degree, this is reasonable. The big problem that arises is at the lowest income levels, where the cheapest goods are already being consumed. At the other end of the spectrum, for those who aren't inclined to change their buying behavior in the face of higher prices, the degree of those higher prices may be understated because of this.

2. Hedonic Adjustment. This is a fancy term that means they adjust for the quality of the items in the basket. So when measuring the price increases of transportation, for example, they don't just look at how much the price of a car has gone up. They adjust for the quality increases in that car. This one is easier to think about over longer time intervals, so the addition of air conditioning, power windows, cruise control, and so forth.

Again, it's not unreasonable to make some adjustment for the fact that most products are, in fact, getting better and better. But as always, the devil's in the details. And when the standard being measured (the CPI basket of goods and services) is regularly being adjusted, it certainly opens the door to criticism about the methodology. Especially when the government has a vested interest in keeping the official inflation numbers low.

Flawed, but probably still reasonable

This skepticism has given rise to a number of alternative inflation sources. Interestingly, some of them, like MIT's Billion Price Project which tracks millions of online prices, have shown U.S. CPI to be a reasonably accurate depiction of overall price inflation.

Other approaches that try to maintain older CPI methodology come up with significantly higher rates of annualized inflation. But those are harder (at least for me) to believe, as they often seem to fail the common-sense test. If U.S. inflation had really averaged something closer to 8% over the past few decades, as some of these measures seem to imply, we would see massively inflated prices in many areas that simply aren't the case. For example, 25 years ago, a basic new Honda Accord cost roughly $15,000. Today, a basic new Honda Accord costs roughly $25,000. That's consistent with the 2-3% annual inflation the CPI has reported. If inflation had really averaged 8% over that time, that $15,000 price would have inflated to over $100,000!

Compounding takes these small increases in prices and really magnifies them over time. Ironically, as we've just seen, this provides some support for the idea that CPI — while clearly not perfect — probably isn't incorrect by orders of magnitude as some imply.

A big deal for investors — and governments

While such "common sense" tests seem to preclude the idea that inflation is running way over stated CPI, what about slightly higher? That's certainly possible. With bond yields as low as they've been in recent years, even a small gap higher could have significant implications. If CPI was even just half a percent higher than stated, the real (inflation-adjusted) rate on many bonds would already be negative. If that had happened, it's reasonable to assume interest rates would be higher, with all of the usual flow-through effects that typically exerts — lower stock and home prices, for starters. And, of course, there would be huge implications for the government's cost of financing its deficits/debt.

Because there's so much at stake, people are going to continue casting a skeptical eye on the official inflation figures. This isn't a particular focus of SMI's work on inflation, as our study of the subject hasn't given us reason to believe these numbers are wildly off.

Perhaps more scarily, they don't need to be wildly off to still have a significant impact over time. Even small inflation rates have a disturbingly large impact on the value of our dollars over time. But that's a subject for another day.