Note: This article is now also posted at iTulip.

A chorus of criticism has been rising, directed at how we measure inflation. For much of the past decade, no one really cared, because by basically all measures, inflation was low. But now, the balance sheet of consumers is deteriorating, and it looks very much like real median incomes are falling. There are also wider senses in which the extent of inflation seems to "change" the prognosis--e.g., what the GDP is, and therefore whether or not we are in recession. Thus, how inflation is measured has suddenly become of interest.

Taking a cue from Adam Hamilton, I decided to delve a little deeper into a specific analysis of what inflation really is, by looking at the costs of business. In specific, a non-manufacturing business, which essentially runs exactly at cost: The US Postal Service.

The US Postal Service only raises rates when it has to--it doesn't structurally turn a profit. I think there is legislation that essentially forces this; excess income is escrowed and the Postal Service has to explain how it needs to be retained for improvements and expansion if it is not to be confiscated by the rest of the Federal government.

Using first-class postage rates as a proxy for post office costs, I compared this measure to the M1 money level (hard currency plus money in demand accounts) and the BLS's all-items, nationwide CPI series. The result is given in the following chart:

(click for larger version)

This chart is pretty striking, and confirms the trend Hamilton was seeing: postal rates unmistakably shadow M1 money, not the BLS CPI. In fact, the only interval for which postage significantly diverges from the M1 curve is the 1974-1986 period, which precisely spans the energy crisis. Postage had to go up even more rapidly during this period because the post office's costs are transportation-heavy.

There are a couple of other provocative artifacts on this chart. The first is that M1 and CPI are essentially the same until about 1983, then they diverge markedly (with M1 to the upside). This confirms a suspicion I've long held: around this time, the US switched from "current" inflationary policies (printing money) to debt-based policies (issuing bonds) to fund the fiscally-unbalanced activities of government. That is, the policy became to take on more debt to fuel spending, spreading inflation out longer-term into the future (as taxes sure as heck weren't raised to pay the debt back). This was quite a stroke of genius, as people were pretty sick of rapid inflation by the early 80s, but of course, the government wasn't about to shrink itself. But take note that much of this debt still hasn't been worked off (in fact Bush II has doubled it), so it would seem to me that a lot of that inflation is just "stored up" and waiting to burst out.

There's another interesting event around 1995: the M1 level actually falls. Remember people thought we were going to have a recession here. This is when the Fed began worrying about "irrational exuberance" and began ostensibly "tightening" monetary policy--raising interest rates. It appears the Fed intentionally tried to reduce M1, possibly to stave off a bubble in stocks. But instead, what happened is that loosening in the fundamental controls in the banking system seems to have led to an explosion of broader money, which ended up fueling a stock market bubble anyway (high interest rates would attract creditors to corporate debt). At this point, I propose that M3 became a more central "kind of money", and has since evolved to play an even greater role in influencing the economy.

But back to the core point:

We can immediately conclude from this analysis that one government bureau is lying: either the Postal Service is squirrelling away vast quantities of accrued profits somewhere (from charging too-high postage), or the BLS is lying about the CPI serving as a realistic inflation/cost of living metric.

If you agree with me that the latter is more likely and that M1 makes a better baseline estimator for inflation than the BLS's CPI, then the real average rate of inflation since 1983 has probably been closer to 5.3%, rather than 2.4% as the official CPI figures would imply. So the BLS is understating inflation by almost two percent per year. The compounding effect of this methodical lowballing has resulted in a situation where the CPI today has become about 42% understated, and therefore the cost of living is being underestimated by at least that much.

What are some secondary implications for government and the economy? Glad you asked:

1. People on social security and inflation-indexed pensions and any other inflation-indexed payment are being swindled. Suddenly it becomes crystal-clear why non-wealthy retirees depend on senior citizen discounts, and rumors that the government knows it would be even more insolvent if it honestly accounted for inflation become a lot more plausible.


2. The CPI computation apparatus of the BLS is completely superfluous--price levels are and always have been a direct factor of the money quantity. This is a damning critique of the economic establishment, which expressly dismisses the money quantity theory of Austrian economics (because it is politically inconvenient). So feel free to add the budget of the CPI computation subunit of the BLS to your "government waste" roster.


3. The bond market is a complete joke. Shaving 2% off the inflation estimator (even worse if you're looking at core-CPI) puts most "risk-free" fixed-income investment in the null or negative real returns category. This explains why the activities of foreign central banks have become so important: they aren't really buying US Treasury securities for investment, they're buying them to manipulate trade or simply to park their reserves cash (which is in dollars thanks to the legacy of Bretton Woods). Real people and businesses can't possibly earn money or even preserve wealth using offical securities.


4. Even more ominously, undermining the "zero-risk" bond market undermines all other financial markets, which naturally calibrate themselves relative to it. Given that the real Dow is around 1.6% (rate of return), a 2% inflation haircut may mean that the average performance in the stock market is actually negative in real terms. Also, since inflation is lowballed like this, banks have more problems making money off of traditional lending activities, which I suspect has made them more aggressive about fees and more reliant on "exotic" lending (i.e. in housing), hedge funds, and proprietary trading.

Sadly, this is just the beginning of the problems. As pointed out above, M1 has diminished in economic importance compared to even "higher" forms of money (somewhat indicated by M2 and M3, but not totally). When you consider that people buy not only "real goods" using M1-level money, but also "quasi-real goods" using financing (which reflects in M2 and M3), it becomes apparent that there is more to the "cost of living" than just M1.

Kinds of "quasi-real goods" I'm thinking of are homes, automobiles, education, and big-ticket retail items which are typically financed. The very activity of financing these things both "creates more money" in the broad money sense, and makes the items more expensive (because more people can afford them, and afford to spend more on them). The financing itself has a cost, and "leaks" some expansion of money into the M1 level.

John Williams has done work in looking at inflation in an even broader sense, which I think includes these housing and financing aspects. While I think his work is excellent and provocative, I now think the truth is somewhere between the 5.3% indicated here and his present ~8%, which he obtains by adding adjustment factors for historical changes in the CPI computation methodology. As pointed out here, immediate inflation has indeed been lowered by shifting from the literal printing of money to debt-based financing. Note that I'm not saying this is actually a good thing--but it has succeeded in at least delaying a lot of immediate inflation and other forms of pain.