Demand Signals for Capacity Planning: Using Employment Indicators to Forecast Office Vacancy in Singapore

Abstract

This study examines whether office-linked employment indicators function as useful demand signals for capacity planning by explaining and forecasting Singapore's office vacancy rate. The study addresses two questions: whether lagged office-linked employment is associated with office vacancy after controlling for office supply-side change and GDP, and whether adding employment improves out-of-sample forecast accuracy relative to a baseline model without employment. A quantitative quarterly time-series design is applied to Singapore over 1993Q3 to 2025Q3 using publicly available data. The dependent variable is the office vacancy rate. The main demand-side variable is a one-quarter-lagged office-linked employment indicator built from quarterly employment changes in office-using industries. Regression with ARIMA errors is used to model vacancy persistence alongside exogenous drivers. Forecast evaluation is conducted using rolling-origin, expanding-window testing at one-quarter-ahead and four-quarter-ahead horizons, with accuracy assessed using MAE, RMSE, and the Diebold-Mariano test. The results provide limited support for the explanatory role of employment. The baseline model without employment outperforms the model including employment at both horizons. This study contributes an operations management perspective on office vacancy as a capacity-slack metric and provides a transparent quarterly forecasting workflow.