As the fiscal cliff talks evolve and devolve, the latest spat has been whether the arc of federal spending should be curtailed by changing the way that we assess costs. The proposal from the White House is to switch the way cost-of-living adjustments are made for Social Security benefits. Rather than pegging those to the Consumer Price Index as currently calculated, these would be pegged to a “chain-weighted” Consumer Price Index, which would save as much as $125 billion in additional benefits over the next decade.
Sounds wonky, and it is. But so is much of how the federal government accounts for spending, and these metrics intimately shape what we spend, how we spend, and how we think about the present and the future. The primary measure of inflation, the Consumer Price Index (CPI) uses a fixed basket of goods that resets periodically. Chained CPI uses a basket of goods that adjust more fluidly to account for what statisticians and economists call “the substitution effect.” A fixed basket of goods is easier to calculate: just define the basket and then measure the price changes. But in the real world, people don’t passively accept changing prices. They change their behavior. The price of gas goes up? People drive less; they carpool more; they buy more fuel-efficient cars and consume less gas. The price of a domestic flat screen television goes up? They buy a less expensive import. In short, people don’t necessarily bear rising costs passively; they react and shift to maintain their standard of living. The traditional CPI index doesn’t capture that.
For all its wonkiness, the proposal to change the benchmark used to determine Social Security and various other benefits has engendered attacks from all points on the political spectrum: the left assails it as a backdoor technicality that will increase burdens on the elderly and the less well-off; the right scoffs that Obama’s proposals don’t constitute true deficit or spending reduction but are simply accounting tricks, and the media treats it as politics as usual with the cynical corollary that because almost no one understands what these rules are, it makes it easier to enact them.
Despite ample media scrutiny over the past week, it’s safe to say that this debate is about as arcane as it gets. Almost no one knows the difference between CPI and chained CPI, even if for years the Bureau of Labor Statistics has been compiling multiple varieties of the index, which serves as a proxy for the rate of inflation. One variant is the chained index. There are others, including a special index for cost-of-living for the elderly and for CPI minus the volatile effects of energy and food (known as Core CPI). All of these use quarterly surveys of 7,000 families about prices paid for 211 different consumer goods in 38 different regions, for a total of 8,018 data points. That is how we assess inflation: whether those data points are going up or down, month-to-month. Every two years, that basket of goods and their weight in the index is adjusted.
Long before the current imbroglio, there have been questions about how inflation is calculated. Statisticians have understood the issues about “substitution effect” for years. In fact, they understood that when these indices were created. In the 1920s, Irving Fisher of Yale argued that the “ideal index” of prices would be a blend of fixed baskets of goods and something that captured how behavior changes in real time. The problem is that the fixed basket is easier to calculate and less costly. That is what the BLS adopted in the 1940s and what has been the standard ever since. In turn, Social Security benefits were pegged to the index, with cost-of-living adjustments based on what the official CPI reported.