The data set we collected for our study pertains to tenders in different countries for various types of projects in a timeframe from 2013 to 2019. Specifically:
- 858 international public tenders
- 8 types of projects
Unlike previous studies, our data did not come from interviews with managers; instead, our work was based on empirical data on tenders and individual bidders.
By running a series of statistical analyses (including bivariate and multivariate regressions, and machine learning applications (LASSO)), we develop an econometric model that incorporates new variables (which are not addressed in extant literature), enabling us to predict the winners of the tenders in question.
In this case, in contrast with traditional research that focused on identifying a relationship between a single independent variable and the level of competitiveness, we built a predictive model based on correlations between the level of competitiveness and several independent variables. The aim of our work was to pinpoint which factors influence the chances of a bidder winning a contract, and the degree of this influence. We specifically looked at tenders for public construction projects which were awarded to the lowest bidders.
The statistical analysis for our research was particularly complicated due to certain limitations, chief among them information access. In fact, the data pertaining to any given contract/bidder are publicly available only when both the arranger of the tender and the bidder are public companies. Another limitation we had to contend with was inherent to the construction industry, where every contract is different, there are only a few bidders for each project, and for the most part the bidders are different for any given tender. Lastly, the effects of independent variables can only be verified on bid prices, not on cost estimates or profits because it is not possible to break down bids into these constituent elements.
Based on our data analysis, we developed three econometric models that incorporate five independent variables of the bidders:
- current work load in other projects or backlog
- experience in the tender country
- experience with the specific type of project similar to tender project
- level of internationalization
- age of the bidder
Generally speaking, our findings show that the biggest impact on the chances of winning tenders come precisely from the five factors above. Specifically, current work load in other projects and the bidder’s age reduce the chances of a successful bid, while the bidder’s experience in the tender country, project-specific experience and the level of internationalization have a positive effect. The relative importance of these variables changes from model to model. What’s more, choosing which model to use depends on two conditions: whether most of the bidders are international or local, and whether bidder data are available to calculate the independent variables.
Beyond the academic contribution of our study, our findings have useful practical implications for contractors, project owners and consultants.
- Contractors can take our results into account to make better-informed decisions as far as critical choices regarding bid/no-bid or markups, to avoid missed opportunities, to save resources, and to increase the chances of winning contracts, securing future profits.
- Project owners and consultants, on the other hand, can utilize this information in deciding when and how to arrange a tender, and how many and which bidders to invite to minimize procurement costs