各部门冲击的传播与部门价格的动态

The transmission of shocks acrosssectors and the dynamics of sectoralprices Francesca Monti, Leïla Van Keirsbilck Abstract This paper studies the dynamics of U.S. sectoral producer prices in a largeBayesian Vector Auto Regression (BVAR) model where the Input-Output (IO) ma-trix is used to structure their long-run relationships. The model provides evidenceof a sectoral spillover channel in driving headline inflation without imposing sucha mechanism in the model’s structure.Forecasts of headline inflation have accu-racy comparable to the Survey of Professional Forecasters’ and greater than thosegenerated by a standard BVAR with the Minnesota prior, confirming that the IOmatrix long-run prior conveys relevant information about the data.The study ofan oil price shock shows that adding the production network prior alters the trans-mission of the shock, amplifying headline inflation. Across sectors, the peak priceresponse to the oil shock increases with oil intensity. A narrowly sector-specific dis-turbance, such as a cereal price shock, has non-negligible aggregate effects once theproduction network is accounted for. Sectoral asymmetries are crucial for evaluatingthe macroeconomic consequences of macroeconomic shocks such as an energy priceshock and a monetary policy shock, as industries with slower price adjustmentsamplify inflation persistence, even after the shock dissipates. JEL Classification: C11; C55; E30Keywords: BVAR, production networks, sectoral shocks Non-technical summary The surge in inflation observed in 2021–2022, with its concomitant spikes in food andenergy prices, reignited fundamental questions about the origins and persistence of in-flation pressures.In particular, how do shocks that originate in specific sectors of theeconomy, such as food or energy, spread to other sectors and ultimately shape aggregateinflation? Modern economies are deeply interconnected through supply chains: firms relyon intermediate inputs produced by other industries, so price increases in one sector cancascade through the production network.This paper proposes a data-driven model tostudy how sectoral shocks propagate across industries, how they contribute to headlineinflation, and how the structure of production influences the persistence of inflationarypressures. To address these questions, we build a Bayesian Vector Autoregression model ofthe U.S. economy that includes 35 sectoral producer price indices and key macroeco-nomic variables such as industrial production, consumer prices, and the policy rate. Weincorporate information from the Input–Output tables on how sectors depend on eachother through intermediate input linkages, not imposing rigid theoretical restrictions, butrather using this production network information to guide the long-run relationships ina flexible statistical framework.In this way, the model allows sectoral prices to movetogether in a manner consistent with supply-chain linkages, while still letting the datadetermine the strength and importance of these connections. We then identify three dis-turbances—an oil price shock, a cereal price shock, and a monetary policy shock—andtrace their effects across sectors and onto aggregate inflation. Our results show that accounting for the production-network structure plays a centralrole in shaping inflation dynamics. For example, when the network structure is taken intoaccount, oil price shocks generate larger and more persistent increases in both producerand consumer prices than in models that ignore sectoral linkages.This heterogeneityacross industries proves crucial. The analysis also shows that shocks that appear narrowlysector-specific can have broader macroeconomic consequences once supply-chain linkagesare considered. A cereal price shock, for example, might seem confined to agriculture andfood production. However, when production linkages are incorporated, its effects spreadto other sectors and become non-negligible for aggregate inflation. Monetary policy transmission is likewise shaped by the production network. A con-tractionary policy shock reduces inflation and output, but its effects differ across in-dustries depending on their position in the network. Moreover, a counterfactual exerciseshows that fully offsetting the inflationary impact of an oil shock would require a strongermonetary tightening and would entail a more pronounced decline in industrial production.Production networks therefore amplify the persistence of inflationary pressures. 1Introduction The inflation surge of early 2021 in the U.S. and mid-2021 in the European Union igniteda heated debate among economists about its drivers and persistence. Concomitant spikesin commodity and energy prices brought the role of sectoral disturbances in shaping theaggregate price dynamics to the forefront. Understanding how sectoral shocks transmitacross sectors and ultimately affect aggregate inflation is crucial but challenging, given theinput-output links of the production net