Pass-through of crude oil prices to domestic prices in Japan

How do domestic prices of goods and services respond to changes in world crude oil prices? A number of authors have documented declines in the “pass-through rate” of oil prices — defined as percentage responses of domestic prices to a percentage increase in oil prices — both for the US and elsewhere in the world. This finding is considered to be important, not only for the relevance of the subject per se, but also because it is believed to have great influence on how we understand workings of our nation economies. In the literature, there are three popular hypotheses on what has caused the decline in the pass-through rate. First is increased credibility of monetary policy: when domestic sellers of oil-related products expect that the central bank will try to suppress inflation by raising the interest rate as soon as it sees rising oil prices, they will hesitate to pass the increased cost onto prices of their own goods. Second is increased wage flexibility. Oil price surges, in the long run, should lower profitability of firms and induce them to cut their production and wage payments. If wages start decreasing fast, firms will find less reason to raise prices. The third explanation is related to the cost structure. It is believed that, after the painful experiences of the 1970s and the early 1980s, firms have learned the danger of depending too much on oil. They have thus transformed their production structure into a less oil intensive one, and thus there is less need to change their prices in response to fluctuations in oil prices.

My joint work with Taisuke Uchino (GCOE Fellow, Hitotsubashi University), titled “Pass-through of oil prices to Japanese domestic prices” (NBER Working Paper No. 15888, 2010 ¹) studies this pass-through issue for Japan. We argue that none of the above three hypotheses seem to be important, at least for Japan for the period up to the early 2000s. To be more precise, we find that the cost reason does explain much of the declining pass-through, but the story is a little different from the one told above.

In the paper, we start with estimation of “time varying parameter” VARs (VAR stands for vector autoregressions), which allow for the possibility that the impact of world oil prices on domestic prices in Japan might vary over time. This method is ideal if one wants to know when pass-through declined, and by how much. Our finding is that, although pass-through fell continuously between the mid 1980s and the early 2000s, major declines occurred in the second half of the 1980s and toward the end of the 1990s. We find that this was not so much due to changing composition of goods and services. Even when we limit our attention to domestic prices of a narrow range of oil-related products, such as petrochemicals or even “plastic hoses”, we find similar tendencies.

We then ask if the importance of oil in the cost of production fell during this period. If we find that the cost share of oil declined sufficiently to explain all of the reduction in the pass-through rate, leaving not much else to be explained, there would be no reason to resort to more complicated stories.

We utilize information from the input-output tables for this purpose. The tables allow us to take into account both direct use of oil for production and indirect ones. We use two types of input-output tables, “nominal” and “real”. The “nominal” tables evaluate both inputs and outputs at current market prices. We examine how the cost shares computed from those tables evolved over time. Note that those nominal shares can change for two reasons. First, they may become higher because firms decided to use larger quantities of oil to produce a given amount of output. Second, holding constant those quantities, the cost share may become higher simply because the price of this type of input became higher. On the other hand, the “real” or “constant price” input-output tables evaluate both inputs and outputs using prices in the base year, say in 1995. Those tables can be used to produce cost shares that are affected by changes in relative quantities but not by prices.

Our calculation shows that the nominal shares of oil in production in Japan (both sectoral and aggregate) declined considerably between 1980 and 2000. The magnitude of the decline is large enough to explain most of the reduction in the estimated pass-through rate for the same period. In that sense, our findings support the cost side story. However, we also find that, when we use the real input-output tables, there is virtually no change in oil’s cost share between 1980 and 2000. As differences between the nominal and real tables lie in prices, we conclude that the decline in the nominal cost share of oil was mostly due to price changes: to put it more simply, oil became less important in overall production cost simply because it became cheaper. This is different from the third hypothesis mentioned at the beginning of the article which emphasizes the quantity side.

At this moment, we do not have sufficient information for the 2000s to make reliable analysis. Our preliminary evidence, based on the data up to 2007, suggests that there was a revival of pass-through of oil prices. This is consistent with the fact that oil prices as well as the cost share of oil increased during this period. However, it appears that the pace at which pass-through came back was slower than what we would have expected from the speed at which the cost share of oil increased. If this tentative finding turns out to be robust, it may call for a new explanation. For example, it might simply be the case that firms regarded oil price increases to be purely temporary, and decided not to pass much of the cost increase onto prices of their products. Further studies on this period would be a fruitful area for future research.