Brad DeLong samples from an article in the Economist (access for subscribers only):
Today, however, consumer-price indices are arguably too narrow. Charles Goodhart, a former member of the Bank of England's Monetary Policy Committee, has long argued that central banks should instead track a broader price index which includes the prices of assets, such as houses and equities...
If the prices of goods and services and those of assets move in step, then excluding the latter does not matter. But if the two types of inflation diverge, as now, a narrow price index could send central bankers astray....
There are really two issues in play here. One has to do with the notion that monetary policy ought to battle "bubbles," or deviations of asset prices from their "fundamental" value. This motivation strikes me as problematic. Even supposing that we possessed the slightest facility at distinguishing fundamental from non-fundamental movements in asset prices, is it likely to be the case that all asset appreciations are the appropriate target of increases in policy rates?
More palatable is the presumption that asset prices give us a truer measure of the purchasing power of money. This concept was explored several years ago by my colleague Mike Bryan, along with co-authors Stephen Cecchetti and Roisin O'Sullivan, in a paper titled Asset Prices in the Measurement of Inflation.
The idea that asset prices should receive some consideration in the construction of aggregate price movements remained a largely dormant issue until Armen Alchian and Benjamin Klein, published their paper “On A Correct Measurement of Inflation" in 1973...
... they propose that we focus on measuring the purchasing power of money generally, rather than on prices of current consumption specifically. Instead of looking at the cost of a particular (carefully designed) basket of goods and services meant to measure current consumption, as is typically done by most consumer price indices, they suggest focusing on the current cost of expected life-time consumption.
Asset prices provide the requisite information on the price of expected future consumption.
Bryan, Cecchetti, and O'Sullivan construct several index measures designed to capture the common trend, or "core", in observed prices. They find:
... measures that include asset prices indicate that inflation has been somewhat higher than other measures would suggest in recent times.
A key question, then, is to ask how policy would have been different had it been based on these measures. Any attempt to estimate this would need to take account of the fact that history changes each time new data are added to this model and so only real-time information should be used. Overall, however, other simpler measures of “core” inflation such as the ex-food and energy approach seem to mirror the movements in [our price indexes] somewhat closely and as such, the inclusion of assets may not have produced dramatically different real-time policy responses. Indeed, much of the focus on asset prices appears to be on the unusual and somewhat dramatic run-up in certain asset prices in recent years. In our approach, which minimizes any idiosyncratic movement in component price data, we are led to the conclusion that such asset price movements contained relatively little information of a common inflation that is useful for month-to-month, or perhaps even year-to-year monetary policy choices.
On the other hand:
Nevertheless, failure to include asset prices appears to induce a bias in the estimate of the inflation trend that may have an impact on our understanding of the broader movements in real economic variables, such as labor compensation.
I'd be interested in hearing from others on the case (or non-case) for targeting inflation measures with asset prices included.
UPDATE: And read this, at William Polley's blog.