Forecasting models of activity in industrial and commercial building

Skinner, D

Abstract

Despite its importance in national income, the level of activity in the construction
sector has received little attention in the economics literature. The lack of studies
attempting to forecast construction activity is surprising given that its volatility is
often regarded as destabilising to the economy. Here, we model an important and
growing component of construction, namely private industrial and commercial
building. Construction activity is typically measured by output. To the extent that new
construction output represents capital formation, output can be modelled as an
investment problem. The theoretical investment literature is disparate and confusing
but here, the leading models are presented in a unified framework in which the
similarities and differences between them can be easily identified. We then go on to
estimate a number of the models empirically. Some are econometric models
consistent with traditional theories of investment. Others are based on vector
autoregression (VAR) analysis which provides a largely statistical representation of a
set of variables with minimum use of a priori restrictions but in which long-run
relationships are preserved. The data required for model estimation is considerable
and complicated by the effects of investment incentives embodied in the tax system.
The forecasting performance of all the models is evaluated against forecasts generated
by a benchmark model suggested by the data rather than by economic theory. In terms
of forecasting performance, some of the investment models considered here are shown
to be superior to the benchmark model.

• Thesis Type: Thesis
• Deposit Date: Oct 3, 2012
• Award Date: Jan 1, 1999