Measuring Inflation Revisited

A little over six months ago, there was quite a bit of debate on the adequacy of Malaysia’s inflation measures (see here and here). For many people, the official estimates of inflation didn’t seem to jive with their on-the-street experiences. I contended (and still believe) that its partly a perception issue, partly a factor of slow income growth (relative to inflation and productivity) and partly a problem of income inequality.

But before going further, what prompted this post was an article in Slate (excerpts):

Do We Need Google To Measure Inflation?

…Each month, the Bureau of Labor Statistics goes through all that hassle because knowing the rate of inflation is such an important measure of economic health—and it's important to the government's own budget…

But just because the government expends so much energy determining the rate of inflation does not mean it is tallying it in the smartest or most accurate way. The reigning methodology is, well, clunky…[and] archaic, given that we live in the Internet age. Prices are easily available online and a lot of shopping happens on the Web rather than in stores.

But there might be a better way. In the last few months, economists have come up with new methods for calculating inflation at Internet speed—nimbler, cheaper, faster, and perhaps even more accurate than Washington's. The first comes from the Massachusetts Institute of Technology. In 2007, economists Roberto Rigobon and Alberto Cavallo started tracking prices online and inputting them into a massive database. Then, last month, they unveiled the Billion Prices Project, an inflation measure based on 5 million items sold by 300 online retailers in 70 countries. (For the United States, the BPP collects about 500,000 prices.)

The BPP's inflation measure is markedly different from the government's. The economists just average all the prices culled online, meaning the basket of goods is whatever you can buy on the Web… Plus, the researchers do not weight certain items' prices, even if they tend to make up a bigger proportion of household spending.

Still, thus far, the BPP has tracked the CPI closely. And the online-based measure has additional advantages. It comes out daily, giving a better sense of inflation's direction. It also lets researchers examine minute, day-to-day price changes…And it has already produced some academic insights. For instance, Cavallo found that retailers change prices less often, but more, percentage-wise, than economists previously thought.

A second inflation measure comes from Web behemoth Google and is a pet project of the company's chief economist, Hal Varian. As reported by the Financial Times, earlier this year, Varian decided to use Google's vast database of Web prices to construct the "Google Price Index," a constantly updated measure of price changes and inflation…Google has not yet decided whether it will publish the price index, and has not released its methodology. But Varian said that his preliminary index tracked CPI closely, though it did show periods of deflation—the worrisome incidence of prices actually falling—where the CPI did not.

The new indices lead to the big question of whether the government needs to update its methods to account for changes in the economy—taking new pricing trends into consideration, rejiggering its formula, updating more frequently. The answer might be yes. (Economists have reformed CPI before.) But the CPI and its Stone Age method of calculation boasts one huge benefit: It's a stable, tested measure, consistent over time, since its methodology doesn't change much. Moreover, and somewhat remarkably, the Google and Billion Prices Project indices actually seem to confirm the accuracy of the old-fashioned CPI, tracking it closely rather than showing it to be off-base.

This article emphasises something I suspected (though didn’t articulate) – changing the methodology to make the Malaysian CPI more “accurate” might not yield a markedly different inflation measure. When you aggregate prices across many goods and services, there will be a muting effect on volatility – as any portfolio manager can tell you.

Which leads back to my conclusions about why people feel inflation is so much higher than the official data shows. Going back to my reasons:

1. The perception issue – it’s well established that people have an asymmetric response to good and bad events. You tend to remember when prices increase rapidly, but don’t really register it when prices come down or stagnate. Despite the doctrine of “sticky” prices, some goods categories do have price declines. Some seasonal food items have price volatilities exceeding 100% i.e. they’ll double then halve in any given year.
2. Slow income growth – average household income growth over the past decade has been a bare 2%-3%, compared to average CPI inflation over the past decade of around 2.8%. Take in productivity growth of around 2%, and Malaysians on average just aren’t being paid what they’re worth – hence the feeling of losing purchasing power over time.
3. Income inequality – there will be different perceptions of inflation, depending on where you are income-wise. If you’re earning a six figure annual salary, chances are your food bill will be somewhat less than 10% of your disposable income. On the other hand, if your salary is on the Malaysian median, food could take up 50% of your disposable income. And since food (and transport) have generally risen faster than the other components of the Malaysian CPI, then a Malaysian earning the median income will experience a faster rate of inflation then one higher up the income scale.

All the above leads me back to the recommendations I made in my earlier post on this – lets get alternate CPI measures based on geographical l0cation and income levels. That will help provide a more nuanced view of the price level and inflation. Just reforming the CPI for “accuracy” won’t necessarily provide the clarity people seem to be looking for.

(H/T Greg Mankiw for the article)