Inflation And Honest Data

’80s Flashback

In the course of inventing new tools you sometimes you find that the old tools weren’t so bad and might even have been better than your “improved” versions.

My friend John Williams at ShadowStats fervently believes that the changes that the Labor Dept. made in the ’80s and ’90s seriously distorted the Consumer Price Index. Reverting to the pre-1980 methodology, he estimates that CPI inflation is running close to 10% today.

As I’ve noted in recent letters, I think it’s likely that today’s official CPI understates inflation in some categories. But does it underestimate inflation as much as John Williams thinks it does? I think that is clearly not the case. The government changed its CPI methodology because the economy evolved. Experts believed the old methods were themselves misleading.

Looking at John’s chart above, you can see that most of the divergence happened in the 1990s. Official CPI-U growth ran in the 3–5% range, while John’s 1980 version rose from 5% to almost 10% by the year 2000. The growth rates have been nearly equal since then.

What was happening in the 1990s that might explain this? I can think of two possibly big factors:

• The internet and other rapid technological changes

• Globalization and trade agreements like NAFTA

Both of those influenced prices downward, and the 1980 CPI method wasn’t designed to capture the shift. I’m just speculating, but maybe applying the old rules in a new environment resulted in the overweighting of sectors with higher price inflation.

Notice in the chart below that it is the high-inflation items that are most influenced by government – things like health care and government-subsidized education. (If you think education is not influenced by the government, you are not paying attention.) The items that are not growing in price? Those are more purely market-driven.

Data from Afar

Speaking of inflation, you might recall that Argentina had a few rough years not long ago under socialist President Cristina Fernandez de Kirchner. The official inflation stats were more than suspiciously low, given the near-universal experiences of actual people. The International Monetary Fund threatened to expel Argentina at one point, before the government revised its data methods in 2013. I often traveled to Argentina during that time and wrote about the massive difference between the official dollar-peso exchange rate and what you could get on the street. Stores really did not want to take my credit cards. If you went to Argentina, you needed cash dollars to exchange for pesos. Inflation was running several hundred percent and climbing fast.

During that time, Alberto Cavallo, an Argentine economist at MIT, decided to estimate Argentina’s inflation rate from his remote perch in Massachusetts. He gathered price data from online retailers and constructed a new index which, not surprisingly, showed far higher inflation than Argentina’s government admitted to at the time.

From that experience Cavallo and another MIT economist, Roberto Rigobon (from Venezuela and no stranger to inflation himself), launched the Billion Prices Project (BPP). The number is not an exaggeration. Each week they get prices for thousands of goods from hundreds of retailers in dozens of countries. The resulting massive database reveals all kinds of interesting trends. There are numerous charts and free research on their website. (The data is available for commercial use through an associated company, PriceStats.)

Their methodology sounds ideal, but it has limitations. Some important prices aren’t readily available online in every location – rent, for instance, or college tuition. Those prices require different estimates and adjustments. Also, limiting the data to online sales channels omits the much larger proportion of offline economic activity. Lower-income people tend to fall into the offline category, so BPP may misjudge some aspects of their experience.