To identify factors that influence the incidence of single bidding, an explanatory analysis is conducted that incorporates the variables detailed in Annex B. A binary logit model is run where single bidding is the dependent variable and the variables from the four indicator groups are the independent variables, whilst controlling for a host of predictors, such as year, sector or region. Through the control variables, the aim is to isolate their structural effect from the desired effect. The study stops short from claiming causation between the predictors and the dependent variable, however, significant (and substantial) correlations indicate that the practice takes place together with the higher likelihood of single bidding. While we the effect of each individual predictor variable on the likelihood of single bidding is explored (together with the control variables), the results are reported with all predictors in one model. To estimate the share of single bidding the following equation is used:
Zi is the log probability that the ith contract to be awarded by single bidding. X are the predictor variables and C stands for control variables (year, buyer type, buyer region, sector, identified based on the assigned product codes, tender year or nuts codes).
In order to estimate the share of single bidding for each category of the predictor variables, predictions on the final model for each variable are run. The results will show the predicted share of single bidding for each category with their 95% confidence intervals. Categories where the predicted share overlaps with the confidence intervals are not significant.
