Here, we examine whether the theoretical model modified on the basis of the Republic of Korea's PPP system is well in line with the real statistics on PPP projects. For this, we constructed a small sample of PPP projects using the InfraInfo database, which is constructed and managed by the Ministry of Economy and Finance and the Public and Private Infrastructure Investment Management Center. To control the different properties of each project as far as possible, we chose 33 road PPP projects that are in operation.
We focused on the possible impact that the initial composition and changes in the equity structure may have on the output indicators of each PPP project. To examine whether and to what extent the equity structure has changed, we gathered information from the InfraInfo database on the share of construction and financial investors at different project stages (for example, initial signing, construction, and operation). The focus of this analysis was to examine how the change of equity structure can induce cost-saving behavior by construction investors. For this, we compared the total private investment volume on the initial contracts with the corresponding amounts in the final audit reports for the PPP projects. We considered two dependent variables: the relative change of project cost compared with the cost estimation at the initial signing, and the absolute difference of audited private investment volume and initial private investment volume.
The main hypotheses that we are trying to verify are twofold. First, the higher the share of construction investors at the initial contract signing, the more prevalent will be the SPV's cost-saving efforts, since construction investors tend to focus on maximizing short-term profits by reducing construction costs. And, second, because construction investors leave projects after construction, which will be measured by the difference in the share of their equity structure before and after construction, they will show a great tendency for cutting construction costs.
We have already measured the composition of investors within each PPP project and followed its changes at different project stages. The dependent variable for the models is the difference in private investment volume between the initial agreement and the final audit report. The key explanatory variable is either the share of construction investors in the initial equity structure (models 1 and 3 in Table 9.2) or the change in share of construction investors before and after the construction (models 2 and 4), according to the design of regression models.
Table 9.2 shows the results of the regression models. Contrary to our expectations, we did not observe any statistically significant impacts from both the initial share of construction investors and the change in their share on reducing project costs when we used the share of the reduction in construction costs to the initial investment volume as a dependent variable (models 1 and 2). But, where the absolute change of investment volume was used as a dependent variable, the change in the share of construction investors has a negative and statistically significant effect on the dependent variable (model 4), while the initial share of construction investors has no significant effect on reducing construction costs (model 3). This means the greater the decrease of a construction investor's share in a project after the initial contract signing, the bigger is the cost reduction that can be expected by the time the project goes into operation. This finding seems to partially support our prediction on the efficiency-enhancing effects of contract bundling, but further examination using a greater number of cases should be done before this can be verified.
Another interesting result is on the minimum revenue guarantee.4 In all four models, the project with a minimum revenue guarantee showed a statistically significant cost reduction, and this is far more pronounced in models 3 and 4. Although further analysis may be needed to explain this result, it nevertheless seems to confirm that PPP projects with minimum revenue guarantees were often initiated by developers who are mainly financial investors. These developers, along with construction companies, tend to be involved in projects for far longer than those initiated by construction investors. In this respect, we interpret this finding as supporting our predictions from the theoretical models that strong bundling behavior, which was observed in the PPP projects with minimum revenue guarantees, can contribute to an increase in project efficiency.
Table 9.2: Results of Regression Models
| Explanatory Variable | Dependent Variable | |||
| Model 1 | Model 2 | Model 3 | Model 4 | |
| (% of investment volume) | (difference in investment volume) | |||
| Share of CIs | 2.73 | 22,948.76 | ||
| Change in share of CIs | -1.82 | -32,624.44* | ||
| Solicited (base: unsolicited) | 1.59 | 1.59 | 5,393.38 | -1,442.37 |
| MRG (base: none) | -5.56* | -5.27* | -32,304.36** | -35,065.06** |
| Competent authority dummy (base: region) | 3.22 | 3.56 | 14,050.44 | 26,807.86 |
| Initial investment volume | 0.0484* | 0.0541** | ||
| Constant | 5.88** | 6.48*** | -8200.81 | -12,533.78 |
| R2 | 0.1591 | 0.1466 | 0.4324 | 0.4629 |
| Number of observations | 33 | 33 | 33 | 33 |
CI = construction investor, MRG = minimum revenue guarantee.
*** p < 0.01 ** p < 0.05 * p < 0.10
Source: Authors.
The empirical results show that our hypothesis on the relationship between investor composition and cost reductions of PPP projects can only be partially verified. When using investment volume in relative terms as a dependent variable, we could not find a statistically significant effect, while the model using investment volume in absolute terms showed a significant yet weak effect. This rather disappointing result may be because of the availability of only a very small number of PPP projects for building the data. But this will be improved by more PPP project cases becoming available in the future.