The BCIS all-in TPI (Tender Price Index) has long been a trusted measure of price movements in the construction industry. This is based on a well-developed method of calculation and a reliance exclusively on tendered rates on construction projects as the source of data.
In recent years, changes in procurement practices in the industry have significantly reduced the number of tender documents available for analysis. This has led to increased volatility in the quarterly index which presents challenges to the users of the index. BCIS has been investigating over a number of years how to address this issue by exploring different methodologies to calculate or validate the TPI.
Firstly, the investigations have highlighted that the analysis of project tender documents is essential in calculating a reliable measure of long-term trends in construction pricing. No other data or method currently available will be reflective. However, additional sources of data are needed to represent and, where necessary, dampen any volatility recorded in the short-term fluctuations in the market.
Having explored several potential solutions, BCIS will be leveraging the support of an econometric model to reinforce the existing analysis of tender documents. This is in addition to the setting up of a panel of respected industry experts, which is already used within the calculation and the validation of the TPI.
The unique data set of thousands of analysed projects assembled by BCIS over nearly half a century has provided an excellent basis for modelling construction prices against available economic time series. After careful investigation of the published economic indicators, unemployment and new orders in the construction industry were found to be the most useful. The use of these indicators, along with other existing BCIS data inputs, give a relatively simple model that is effective at tracking short-term price changes.
The model and the panel data are used to rebase project data from nine quarters so that a weighted moving average can be calculated. By restricting the use of the new data sources to short range adjustments, the derived index cannot drift away from the observed tender prices in the market.
It should be noted that the value for the current quarter is based on the modelled results for the previous quarter adjusted with the panel data alone, as this is the most up to date measure.
BCIS will continue to report and measure the accuracy and volatility of the new methodology and will continue to investigate other sources of price measurements, including the adoption of AI and machine-learning techniques.