Oil Price Pass-Through into Consumer Prices: Evidence from U.S. Weekly Data
One sentence summary: Oil price pass-through into consumer prices is through gasoline prices in the short-run and through ex-gasoline consumer prices in the long-run.
The corresponding academic paper by Hakan Yilmazkuday has been accepted for publication at Journal of International Money and Finance.
Abstract
Using U.S. data from Monday of each week, this paper estimates oil price
pass-through into consumer prices (PC) and oil price pass-through into
gasoline retail prices (PG) in a continuous way. The results show that
PC (PG) is about 0.5% (13%) after a week, 1.5% (37%) after three months,
and 4.2% (50%) in the long run. The estimated PC is further decomposed
into direct PC (representing oil price effects on consumer prices
through gasoline retail prices) versus indirect PC (representing oil
price effects on consumer prices through ex-gasoline prices), suggesting
that long-run oil price effects on consumer prices are mostly through
ex-gasoline consumer prices. Despite having distinct pass-through
estimates, about three-fourths of weekly volatility in both gasoline
retail and consumer prices are explained by oil price shocks in the long
run.
Non-technical Summary
Optimal monetary policy depends on the accurate prediction of domestic inflation that requires consideration of international shocks in an open economy. Since oil is a basic raw material at many production levels and its price is determined in global markets, changes in oil prices constitute a big portion of such international shocks. Accordingly, policy makers are interested in measuring the effects of an oil price shock on inflation, which can be achieved by estimating the oil price pass through into consumer prices (henceforth PC).
Oil price shocks can affect consumer prices through direct and indirect channels. The direct channel (that we consider in this paper) works through gasoline retail prices, because gasoline is the form of oil that is consumed the most as a final product by consumers (about 4% of overall expenditure), and thus developments in gasoline prices are salient to consumers. The indirect channel works through prices of products other than gasoline (i.e., ex-gasoline prices) in the consumption basket, since oil is used in the production and/or transportation of almost all products. Within this picture, oil price pass-through into gasoline prices (henceforth PG) represents the direct channel, while oil price pass through into ex-gasoline prices (henceforth PE) represents the indirect channel, both subject to the corresponding expenditure weights in the consumption basket.
By taking into account these direct and indirect channels, this paper estimates PC and PG by using U.S. data from Monday of each week on oil, gasoline retail, and consumer prices, where the estimation is achieved by a structural vector autoregression (SVAR) model. Using data based on Monday of each week is important to capture the dynamics of the oil market, since taking averages within a month, quarter or year as in the literature may suppress valuable information on weekly dynamics. The estimation by SVAR is also essential to identify weekly oil price shocks that are independent of weekly gasoline or weekly consumer price shocks.
PC (PG) is measured as the cumulative impulse response of consumer (gasoline retail) prices divided by the cumulative impulse response of oil prices, both following an oil price shock. Such a strategy not only results in having oil price pass-through estimates in a continuous way but also makes them robust to any endogeneity problem, since the response of oil prices following an oil price shock is also taken into account. When the estimated measures of PC and PG are combined with the implications of an economic model introduced in the Appendix, measures of PE are also obtained. Finally, when pre-shock expenditure share of gasoline for consumers is considered, PC is decomposed into direct oil price pass-through into consumer prices (DPC) versus indirect oil price pass-through into consumer prices (IPC).
The results show that PC (PG) is about 0.5% (13%) after a week, 1.5% (37%) after three months, 3.3% (50%) after one year, and 4.2% (50%) in the long run. When continuous PC estimates are further decomposed into those through direct channels (DPC) versus indirect channels (IPC), the results show that DPC (IPC) is about 0.5% (0%) after a week, 1.5% (0.1%) after three months, 1.8% (1.5%) after one year, and 1.9% (2.3%) in the long run. Therefore, short-run effects of oil prices on consumer prices are through gasoline prices, while their long-run effects are more through ex-gasoline consumer prices.
It is implied that gasoline prices should have higher weights in the short run, whereas ex-gasoline prices should have higher weights in the long run while conducting optimal policy through forming forward-looking monetary policy reaction functions. When consumer income is fixed, PC can also be used as a measure of welfare loss following an oil price shock (e.g., see implications of an economic model in the Appendix). Accordingly, following an oil price shock, consumers lose welfare in the short run more due to the direct effects of oil price shocks on gasoline prices, while their welfare loss in the long run is more due to the indirect effects of oil price shocks on ex-gasoline consumer prices.
Despite having distinct pass-through estimates, about three-fourths of weekly volatility in both gasoline retail and consumer prices are explained by oil price shocks in the long run. Compared to earlier studies, the contribution of this paper can be listed as follows: (i) weekly consumer prices are used for the estimation of pass-through measures, (ii) oil price pass-through estimates are obtained in a continuous way, (iii) effects of oil prices on ex-gasoline prices (PE) are obtained by using the implications of an economic model, (iv) PC is decomposed into DPC and IPC, which results in important policy and welfare implications, and (v) estimation strategy is robust to the consideration of any endogeneity problem due to considering the response of oil prices to their own shocks.
Optimal monetary policy depends on the accurate prediction of domestic inflation that requires consideration of international shocks in an open economy. Since oil is a basic raw material at many production levels and its price is determined in global markets, changes in oil prices constitute a big portion of such international shocks. Accordingly, policy makers are interested in measuring the effects of an oil price shock on inflation, which can be achieved by estimating the oil price pass through into consumer prices (henceforth PC).
Oil price shocks can affect consumer prices through direct and indirect channels. The direct channel (that we consider in this paper) works through gasoline retail prices, because gasoline is the form of oil that is consumed the most as a final product by consumers (about 4% of overall expenditure), and thus developments in gasoline prices are salient to consumers. The indirect channel works through prices of products other than gasoline (i.e., ex-gasoline prices) in the consumption basket, since oil is used in the production and/or transportation of almost all products. Within this picture, oil price pass-through into gasoline prices (henceforth PG) represents the direct channel, while oil price pass through into ex-gasoline prices (henceforth PE) represents the indirect channel, both subject to the corresponding expenditure weights in the consumption basket.
By taking into account these direct and indirect channels, this paper estimates PC and PG by using U.S. data from Monday of each week on oil, gasoline retail, and consumer prices, where the estimation is achieved by a structural vector autoregression (SVAR) model. Using data based on Monday of each week is important to capture the dynamics of the oil market, since taking averages within a month, quarter or year as in the literature may suppress valuable information on weekly dynamics. The estimation by SVAR is also essential to identify weekly oil price shocks that are independent of weekly gasoline or weekly consumer price shocks.
PC (PG) is measured as the cumulative impulse response of consumer (gasoline retail) prices divided by the cumulative impulse response of oil prices, both following an oil price shock. Such a strategy not only results in having oil price pass-through estimates in a continuous way but also makes them robust to any endogeneity problem, since the response of oil prices following an oil price shock is also taken into account. When the estimated measures of PC and PG are combined with the implications of an economic model introduced in the Appendix, measures of PE are also obtained. Finally, when pre-shock expenditure share of gasoline for consumers is considered, PC is decomposed into direct oil price pass-through into consumer prices (DPC) versus indirect oil price pass-through into consumer prices (IPC).
The results show that PC (PG) is about 0.5% (13%) after a week, 1.5% (37%) after three months, 3.3% (50%) after one year, and 4.2% (50%) in the long run. When continuous PC estimates are further decomposed into those through direct channels (DPC) versus indirect channels (IPC), the results show that DPC (IPC) is about 0.5% (0%) after a week, 1.5% (0.1%) after three months, 1.8% (1.5%) after one year, and 1.9% (2.3%) in the long run. Therefore, short-run effects of oil prices on consumer prices are through gasoline prices, while their long-run effects are more through ex-gasoline consumer prices.
It is implied that gasoline prices should have higher weights in the short run, whereas ex-gasoline prices should have higher weights in the long run while conducting optimal policy through forming forward-looking monetary policy reaction functions. When consumer income is fixed, PC can also be used as a measure of welfare loss following an oil price shock (e.g., see implications of an economic model in the Appendix). Accordingly, following an oil price shock, consumers lose welfare in the short run more due to the direct effects of oil price shocks on gasoline prices, while their welfare loss in the long run is more due to the indirect effects of oil price shocks on ex-gasoline consumer prices.
Despite having distinct pass-through estimates, about three-fourths of weekly volatility in both gasoline retail and consumer prices are explained by oil price shocks in the long run. Compared to earlier studies, the contribution of this paper can be listed as follows: (i) weekly consumer prices are used for the estimation of pass-through measures, (ii) oil price pass-through estimates are obtained in a continuous way, (iii) effects of oil prices on ex-gasoline prices (PE) are obtained by using the implications of an economic model, (iv) PC is decomposed into DPC and IPC, which results in important policy and welfare implications, and (v) estimation strategy is robust to the consideration of any endogeneity problem due to considering the response of oil prices to their own shocks.
The corresponding academic paper by Hakan Yilmazkuday has been accepted for publication at Journal of International Money and Finance.