再Measuring inflation, May 11th 2013 P40 (各種物価指数)
In recent years inflation has been one of the few economic indices that has not caused much trouble to Americans. Contrary to repeated warnings, it has neither rocketed higher nor turned into deflation. But policy makers now face a different sort of inflation problem：determining which of many competing indices is giving the best picture of prices in America. The most popular measure, the consumer price index (CPI), is a representative basket of goods and services drawn from a survey of the spending habits of 12,200 households. The index assumes that consumers buy the same quantity of each commodity from one period to the next until the basket is updated, every two years. The change in the cost of that basket is the inflation rate. But this almost certainly overstates the cost of living, because consumers continually adjust their spending patterns to buy more of what is cheap and less of what is dear. America's statisticians have known about this substitution bias since at least 1961, when a commission urged the development of a better index. In 1996 the Bureau of Economic Analysis (BEA) adopted ゛chain-weighted゛indices - ones that are continuously updated to reflect changing spending patterns - to calculate real GDP. And in 2002 the Bureau of Labour Statistics (BLS) began calculating a chain-weighted version of the CPI. Inflation has run about a quarter of a percentage point lower using the chained CPI, and about a third of a point lower using another measure, the BEA's price index of personal consumption expenditure (PCE). This matters to almost no one except the Federal Reserve, which bases its forecasts and decisions on PCE inflation. Even for the Fed the distinction seldom matters, so long as the two inflation measures behave similarly. Lately, though, they have not：stripping out the volatile food and energy categories, underlying PCE inflation dived to 1.1% in March, wiile core CPI inflation has stabilised around 1.9%. If inflatition really were falling so fast, alarms would be ringing at the Fed, since deflation may not be far behind. But that seems unlikely. Apart from being chain-weighted, the PCE includes some things the CPI omits, such as the share of medical expenses borne by employers and governments and supposedly ゛free゛services such as no-fee bank accounts. Slower-rising medical and financial-service costs explain most of recent drop in PCE inflation. Neither effect seems to reflect broader economic trends and, once they fade, PCE inflation should rebound. Confusion over the right inflation index has spread from monetary to fiscal policy. In April Barack Obama's budget proposed using the chained CPI to index Social Security and federal military and civil-service retirement benefits, tax brackets and tax breaks. But Mr Obama is less interested in economic fidelity than in making a political point：by curbing Social Security benefits, a Democratic sacred cow, he hoped to entice Republicans into agreeing to raise taxes as part of a deficit deal. And almost half the $230 billion raised by the proposal over the next decade would come through increased revenue by pushing taxpayers into higher tax brackets and reducing the value of their tax breaks more quickly. Yet Mr Obama would stick with the original CPI for a host of other programmes, such as Medicaid and college grants - and for estimating the poverty level. So, under his proposal, the federal government would endorse two different inflation rates, just to lock in a higher rate of funding for favoured programmes. Mr Obama has long said Americans should pay more tax to finance a more generous safety net. Whatever the merits of that goal, cherry-picking inflation rates is a dubious way to achieve it.
