Forecasting Office Building Price Indices in Germany Using Regularized VARX Models

Construction price indices play a critical role in shaping construction activity and determining the economic success of building projects in Germany. This paper presents a detailed forecasting framework based on 35 sub-construction-price indices, providing granular insights for enhanced cost management and greater planning certainty. We employ regularized vector autoregressive models with exogenous variables (VARX) implemented via the BigVAR package in RStudio, estimating models across differing historical sample windows. We benchmark these high-dimensional BigVAR models against compact VARX specifications that forecast each target index jointly with macroeconomic variables (GDP, CPI, three-month interbank rate). Candidate models are evaluated using mean absolute percentage error (MAPE) and oot mean square error (RMSE), and final specifications are selected by MAPE to generate forecasts for the index components. Our findings show that regularized VARX effectively captures dynamic interdependencies among sub-indices and, relative to the macro-augmented VARX benchmarks, delivers more accurate forecasts. These results offer practical guidance for contractors, planners, and policymakers aiming to optimize budgeting and mitigate risk in the German construction sector.