Nathan Tankus — Are General Price Level Indices Theoretically Coherent?

In traditional neoclassical theory, the theoretical edifice behind a price index is that there is one generalized “value of money” and an index of current consumer goods and services prices successfully measures it. The fundamental forces of supply and demand are supposed to determine relative prices and the quantity of money straightforwardly determines the overall price level through determining “overall” demand. These theorists presumed that you can aggregate individual product, individual firm demand and supply curves into “market” demand and supply curves. You can then put all the product demand and supply curves into a general equilibrium system where the same forces determine relative prices and thus money has no role besides as a numeraire (the famous “Hahn problem”). Thus they had a product specific and price specific theory which they assumed was aggregatable to the macroeconomic level. However, none of these steps have proven to be true ( Andrews 1964, Jo 2016, Lee et al 2004 and Moudud 2012). Thus there is no neoclassical theoretical foundations for a monolithic “general” price level.

Since there is no unitary force moving "general prices" in contradistinction to relative prices, there is no unitary value of money. There is only a range of things you can buy at different prices. Why then do we persist in treating Consumer Price indices as fundamental? Why is the same index used to adjust everything from wage contracts to social security? In other words, if effective demand doesn't determine the price level but only impacts specific prices, and only weakly at that, what theoretical meaning does a price index have? So the consumer price index grew 3 per cent faster, so what? Does that say something about the value of money in general or did 6 industries increase their target profit margins? Did product quality change significantly across a number of products important to consumers? Neoclassicals try to get around the product quality issue by “quality adjusting” price level indices. However, once you don’t have some homogeneous substance which is the “essence” of all output (whether that substance is labor time, or utility) there is no theoretical basis for such adjustments.

Using developments in Post-Keynesian theory originating with Means I would argue that an index of consumer prices fixed relative to organized exchange prices and an index of administered goods and service prices could be constructed separately in a coherent way. I would further make the case for constructing an index of tradable administered price goods and services and non-tradable ones since they behave differently (whether because of baumol disease productivity factors or because non-tradable goods face localized and more geographically concrete competition). However, the reason to construct these indices isn’t to claim all the prices in the index will change for the same reason but simply to say that they are more similar to each other than the prices in other indexes. Nor would we any longer be driven to claim that faster growth in one more more of these indexes necessarily says anything about the state of effective demand and thus necessarily implies that x or y macroeconomic policy is worthwhile....

This is an issue for central banks in setting monetary policy based on inflation expectations and employment level. The inflation and employment rates are indexes that simply, and may oversimplify, a wide range of data.

Tankus Notes
Are General Price Level Indices Theoretically Coherent?

Nathan Tankus