An Approach To Optimize The Project Portfolio Of Venture Capital And Its Application Research

The venture capital (VC) has developed as an important intermediary in financial markets, providing capital to firms that might otherwise have difficulty attracting financing. These firms are typically small and young, plagued by high levels of uncertainty and large differences between what entrepreneurs and investors know. Moreover, these firms typically possess few tangible assets and operate in markets that change very rapidly. Venture capital organizations finance these high-risk, potentially high-reward projects, purchasing equity stakes while the firms are still privately held.To ensure the success of investment, VC project (denoted as project for short hereafter unless especially declared) portfolio(PP), which is composed of projects invested by VC, is proposed by making good balance between minimizing PP risk and maximizing PP return of investment (ROI). The literature describes two different ways of selecting PPs: Relying on the evaluation of an individual project (REIP); Relying on the evaluation of a group of projects (REGP).The REIP approach focuses on project evaluation criteria (PEC) and evaluation method (EM) for individual project. In this approach, it is supposed that projects could be well estimated as long as PEC and EM employed to evaluate projects are reasonable and scientific, and thus optimal PPs derived by evaluated ranking scores of projects. Though this approach can assess projects from different aspects and be easily applied in practice, the intercorrelativity among projects is ignored in it.The REGP approach pays more attention to intercorrelativity and uses the mean-variance (MV) model to select the optimal PP. In this approach, the optimal PP is obtained by the model, in which the expected ROI of a portfolio is considered as the investment return and the ROI variance of portfolios as the investment risk. Though the intercorrelativity is well reflected in this approach, yet it is a misusage to apply financial portfolio theory to the VC field, for VC capital is indivisible distinguishly different from the finance capital. In addition, unlike financial security investment, which ROI can be predicted reasonably by historical performance data, the ROI of VC PP can hardly be forecasted. As a result, it is not reliable and scientific to employ the MV model to derive the optimal VC PP.To overcome the drawbacks as mentioned above, in this paper two key concepts of PP preference rate (TPR) and variance of TPR (VTPR) are defined based on PEC, and a model named VCP-DEA/AR is proposed to derive pareto-optimal PPs using the technique of data envelopment analysis (DEA) with assurance region.Originally proposed by Markowitz, the MV methodology for the portfolio selection problem has been central to research activities of this area and has served as a basis for the development of modern financial theory. Its core is to take the expected return of portfolio as the investment return and the variance of the expected returns of portfolio as the investment risk. According to MV methodology, for a given specific return rate, one can derive a portfolio whose risk is minimum by minimizing the variance of portfolio; or for a given risk level which investors can tolerate, one can derive the portfolio whose return is maximum by maximizing the expected return of portfolio. On the other hand, the DEA introduced by Charnes et al. (1978) has gained a wide range of applications measuring comparative efficiency. It is a non-parametric method based on linear programming. The productivities of units, specified as the ratio of weighted sum of outputs and inputs, are compared with each other and efficient units are identified. According to above characteristics of the MV model and the DEA methodology, it is considered that there exists a close technical relationship between them, i.e. the MV model is a specific type of DEA, the reason of which is what problem does the MV model solve is relative efficiency evaluation with a single-type of input and a single-type of output, while the DEA solves is that with multi-type of inputs and multi-type of outputs. Therefore, simulating how does the MV model deal with the return and the risk of portfolio, two key concepts of PP preference rate (TPR) and variance of TPR (VTPR) are defined in this paper, we consider the PP as decision-making unit (DMU), the VTPR as input, the ATRP as output, and employ the DEA to select the pareto-optimal relative efficient DMUs, i.e. the pareto-optimal project portfolios.The VCP-DEA/AR has four advantages as follows: (1) With PEC, projects and PPs are able to be evaluated roundly; (2) With PPr, the drawbacks that VC project ROI is difficult to be forecasted but required to the MV model can be overcomed well. (3) With VTPR, the intercorrelativity existing in projects is able to be reflected. (4) With evaluating each PP, the issue of capital indivisibility can be solved. (5) With AR deriving from AHP with interval number, the solusion of VCP-DEA/AR can be certified to be scientific. The illustrative example shows, the VCP-DEA/AR is scientific, reasonable, and well applicable to real-world PP-selection problem.Note that, the optimal PP-selection problem that some projects in the choice set have several invested ways such as different investment proportion could also be solved by the present VCP-DEA/AR. The solution can be described as following three aspects: (1) Regard each invested way in one real project as a single'project', and let these'projects'originated from one real project be mutually exclusive, i.e. one portfolio can include no more than one'project'originated from one real project; (2) Define the set of PP by enumerating all feasible PPs from the choice set consisting of these so-called'projects'and other real projects; (3) Employ VCP-DEA/AR to select pareto-optimal PPs from the set of PP.