Forecasting Chinese Steel-Chain Inventories with Leading Indicators and LASSO

Summary

This research summary describes a method for forecasting year-over-year inventory growth across six products in China’s ferrous commodity chain. It treats inventory as an indicator of relative supply and demand, while noting that true inventory is difficult to measure. The researchers select representative inventory series and organize potential drivers across macroeconomic conditions and product-level supply-chain factors, including upstream materials, substitutes, and downstream industry activity.

To identify predictors, they test macro and micro indicators at different leads and retain those with stronger relationships to inventory growth, with lead times spanning two to twelve periods. They then use LASSO, with a penalty selected by ten-fold cross-validation, to choose variables and estimate forecasts while addressing limited sample size and multicollinearity. The summary reports strong in-sample directional accuracy for several products and favorable rolling out-of-sample results for iron ore and hot-rolled coil. These results come from a specific historical study; the available text does not provide full model specifications or establish performance in other periods or markets.

Key ideas

• The study uses representative inventory series because complete inventory measurement is difficult.
• Candidate predictors combine macroeconomic conditions with commodity and industry factors.
• Leading indicators are selected by testing their relationships with future inventory growth.
• LASSO with cross-validation supports variable selection when samples are small and predictors overlap.
• Reported directional accuracy is historical evidence and does not guarantee future forecasting performance.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.