E-piphany

I still think that “anchored inflation expectations” is a term that is devilishly difficult to define and measure, and therefore shouldn’t be a part of monetary policy planning. By the time you know that inflation expectations have become un-anchored, it’s too late to anchor them again because (theoretically) behaviors change when the inflation regime is perceived to have moved from “low and stable” to “higher and more volatile.” Generally left undiscussed is how behaviors change in this transition; it always sounds like the notion is that sellers can’t change prices, unless buyers expect them to do so…or unless sellers expect that buyer expect them to do so.

But there are some deeper mechanics of pricing that, we can illustrate, can produce a state-shift in pricing policy when expectations are for prices to generally rise. To the extent this happens, it could support the idea that once the state-shift happens, it is not trivial to shift it back.

Let’s first examine a case where a manufacturer expects cost increases to be followed by cost decreases, and vice-versa. Consider two pricing strategies: in one, the manufacturer passes through 100% of the cost increases to its price, so that its profit remains stable. The chart below shows costs per unit varying in blue, the producer’s price to its customers in red, and the net profit in green at the bottom.

An alternative pricing strategy is to pass through only a portion of the cost changes. In the chart below, the manufacturer adjusts prices only 25% of the movement in costs, absorbing the rest into its margin.

The second strategy results in a more-volatile earnings stream, but averages over time to the same level of profits. More usefully, it means a much more stable price to the customer, which customers clearly prefer. In the real world, where costs change more chaotically than this idealized oscillation, such price-dampening behavior likely leads to steadier end-customer demand and, over time, this potentially means more unit sales and greater total profit.

But now let’s turn to a case where instead of oscillating with no net change, we have costs oscillating on an upward trend. Now, when the manufacturer passes through 25% of the change, it sees a steady erosion in profitability since those costs never fully return to the prior level.

On the other hand, the manufacturer who is fully passing all cost changes along to the end customer sees steady profits, as before.

Moreover, the “full pass through” manufacturer no longer has the disadvantage of a more-volatile price. Because eventually, the partial-pass-through manufacturer will have to institute a large price change to become profitable again. It is not clear that a steady pricing policy punctuated with large step jumps is better than one that transparently passes through costs, from the end customer’s perspective.

The moral of the story is that a manufacturer who dampens cost swing pass-through in its pricing policy needs to be very good at knowing when the inflation environment changes, or be confident that inflation is low and stable generally. The manufacturer who passes costs through fully is already adapted for an environment of volatile inflation.

And, as more manufacturers move to the latter model, then inflation does become more volatile as the dampening behavior in the value-add processes goes away. Moreover, notice that the urgency to shift pricing regimes only works in one direction. A partial-pass-through producer can start losing money and feel a great urgency to shift to being a full-pass-through producer because it is losing money; but once a manufacturer has moved to full pass-through, there is not much urgency to move back. Ergo, once these shifts start to happen, inflation volatility gets somewhat institutionalized. In theory, we could measure the degree to which inflation expectations have become un-anchored by measuring the proportion of manufacturers’ changing costs that get passed through. As that percentage rises, it implies that manufacturers are growing more cautious about assuming mean-reversion in costs, and that they are moving closer to a model that works in an un-anchored environment and which tends to perpetuate that environment. I don’t know of anyone who has tried to do this, yet, but if policymakers are going to rely on “anchored inflation expectations” as a key component of their inflation models, they ought to examine ways to measure it better than just asking people whether their inflation expectations are anchored!