Here’s The Economist:

A forthcoming paper by Diego Aparicio and Roberto Rigobon of the Massachusetts Institute of Technology helps make the point. Firms that sell thousands of different items do not offer them at thousands of different prices, but rather slot them into a dozen or two price points. Visit the website for h&m, a fashion retailer, and you will find a staggering array of items for £9.99: hats, scarves, jewellery, belts, bags, herringbone braces, satin neckties, patterned shirts for dogs and much more. Another vast collection of items cost £6.99, and another, £12.99. When sellers change an item’s price, they tend not to nudge it a little, but rather to re-slot it into one of the pre-existing price categories. The authors dub this phenomenon “quantum pricing” (quantum mechanics grew from the observation that the properties of subatomic particles do not vary along a continuum, but rather fall into discrete states).

I think that’s right, but I’m not sure about the implication that many observers draw from this practice:

Central banks are starting to see the consequences. Inflation does not respond to economic conditions as much as it used to. (To take one example, deflation during the Great Recession was surprisingly mild and short-lived, and after nearly three years of unemployment below 5%, American inflation still trundles along below the Federal Reserve’s target rate of 2%.) In its recently published annual report the Bank for International Settlements, a club of central banks, mused that quantum pricing and related phenomena help account for such trends.

But firms’ aversion to increasing prices may be as much a consequence of limp inflation as a contributor to it. . . .

I agree with that final sentence, but not so much with the preceding paragraph.  Imagine a simple model where firms only changed prices when a 10% change was called for.  Now assume that aggregate spending rises enough so that the overall equilibrium price level increases by 2%.  In that case, roughly 20% of firms will now find themselves nudged into in a situation where a 10% price rise is called for.  Each of those firms will raise prices by 10%, while the other 80% of firms don’t change prices at all.  The price level rises by 2%.  In that simple model, prices are very sticky at the firm level, but the overall price level is quite flexible.

That’s not to say there isn’t also some price level stickiness—I suspect there is.  But I doubt this phenomenon explains the low inflation of recent years.  A better explanation is the roughly 4% annual NGDP growth since 2009.  That alone accounts for most of the roughly 1.6% inflation.  If you want to argue that inflation would have been higher with more flexible prices, you’d also have to argue that real GDP growth would have been lower, if we hold NGDP growth at 4%.  And real growth was already quite disappointing by historical standards.

In other words the surprising fact about the recovery is not so much the low inflation, it’s the low inflation combined with the low RGDP growth—in other words, low NGDP growth.  I am confident that had NGDP growth averaged 5% or 6% during the recovery, inflation would have been higher and unemployment would have fallen more quickly.  So while the sticky price theories are not wrong, they don’t tell us what’s truly important about the past decade.