Forecasting construction material prices in Egypt using vector autoregression: a decision-making approach

Abstract

Purpose Unpredictable construction material costs pose a major challenge for contractors, developers, and policymakers in Egypt. This study develops a forecasting framework for the Construction Material Cost Index (CMCI) that integrates material prices with key macroeconomic indicators. The aim is not only to predict cost fluctuations but also to provide an interpretable, context-specific tool for supporting budgeting, procurement, and risk management in volatile economies. Design/methodology/approach Using a Vector Autoregression (VAR) approach, the study analyzed monthly data from 2012 to 2022 covering five construction materials (steel, cement, bricks, gypsum board and ceramic tiles) and five macroeconomic indicators (GDP, CPI, ER, M2, and IR). A suite of statistical tests—correlation analysis, Augmented Dickey-Fuller (ADF) test, and Granger causality—was applied to identify suitable predictors. The VAR model, optimized with Akaike's Information Criterion (AIC), was validated using the Mean Absolute Percentage Error (MAPE), which averaged 10.96% across forecasts for early 2023. Findings The analysis revealed that exchange rates (ER), domestic liquidity (M2), and GDP exert the strongest influence on construction material costs, whereas inflation rate (IR) and CPI showed weaker predictive value. The resulting CMCI captures material price dynamics more accurately than traditional univariate approaches, offering a replicable framework for emerging economies. Originality/value Unlike generic indices or black-box machine learning models, this study provides a transparent, statistically rigorous framework tailored to Egypt's construction sector. Its originality lies in (1) constructing a localized CMCI using the Laspeyres index method with a carefully selected base year, (2) integrating macroeconomic indicators into a multivariate VAR framework, and (3) validating predictive performance with real market data. This research therefore contributes a replicable, context-sensitive forecasting tool that can be adapted to other developing economies experiencing high material price volatility.