6. Caracterización del levitador
6.2. Toma de medidas
As explained in section 3.2.3.2, several aspects regarding the collection and the measurement of network data have to be considered before any network analysis is being performed.
Data collection: As to the data collection, there are several ways, in which network data can be gathered including questionnaires, interviews, experiments, observations, or
archival records.586 For the purpose of this study, i.e., in order to collect data on
investments of VCs,587 the most appropriate way is to use a database that contains this information. Of course, certain aspects regarding the validity and reliability of the information in the database have to be evaluated, which will be discussed further below. In the case of this study, the database VentureXpert from Thomson Financial has been used. VentureXpert contains information on investments of VCs and PEs from all over the world. Nevertheless, within this database, the user is able to retrieve data, based on certain selection criteria. For example, selections can be made regarding the nation of the portfolio company, the nation of the VC or PE, the investment date, or the investment stages that should be considered. As will be discussed below, these selection criteria will be used to determine the network boundaries, a necessary condition for any formal network analysis.588
Measurement of network data: As to the measurement of network data, the unit of observation, the modeling unit, and the quantification of the relations require attention. For the purpose of this study, the unit of observation is the individual VC and its investment relations to other VCs. On these entities, measurements are collected.589 The modeling unit refers to the level, at which network data is presented, i.e., it can be presented on the level of individual actors, of dyads or triads, of subgroups or actor subsets, or of the entire network.590 The choice of the modeling unit depends on the research question asked. In the case of this study, the impact of network measures on deal flow quantity and quality on the level of individual VCs will be focused on. Nevertheless, as discussed earlier, also the entire network structure will be looked at as well as the benefits that individual VCs have due to their network position with respect to certain subgroups. Regarding the quantification of relations, the two dimensions directionality
586 See Wasserman/Faust (1994), p. 45.
587 The information on investments will be collected given certain limiting criteria, i.e., information will be collected for a certain time period, region, and investment stages. This will be explained in detail in the following section.
588 See Jansen (2003), p. 71.
589 See Wasserman/Faust (1994), p. 43. 590 See Wasserman/Faust (1994), p. 44.
Data Collection 164
and numeration are relevant.591 The data on the investments of VCs that can be retrieved
from the VentureXpert database does not reveal the directionality of relations in the case of syndicated investments. That means, if two or more VCs jointly invested in a portfolio company, the data does not reflect, which VC had the role of the lead investor and which VC acted as co-investor.592 Neither does the database show, of course, which VC was the one, that initially originated the deal, i.e., that identified the investment opportunity. However, what the data shows are the individual VC(s) that jointly invested in a portfolio company. In addition, the names of the portfolio company as well as of the VCs that invested, are shown. That is, the dataset contains the information, in which portfolio company has been invested and who the one or more investors were, which are the necessary pieces of information for this study.
The second dimension regarding the quantification of relations refers to the numeration, which can either be binary, i.e., dichotomous, or valued. In the case of binary data, information only exists on whether a relationship between two actors is present or absent. In the case of valued data, information also exists about the strength, intensity, or frequency of the relation.593 The original data on investments of VCs retrieved from VentureXpert is not valued, i.e., one can only see if a VC invested in a portfolio company or if it did not. However, as will be explained in detail in the following sections, the data originally retrieved will be prepared and formatted in a way that, with the help of the software package UCINET 6, matrices can be generated that transform the basic information into a valued matrix. In the case of this study, this valued matrix then contains the information, how many times two specific VCs jointly have invested in portfolio companies.
However, as is common and necessary in any formal network analysis, the network boundaries have to be defined, being explained in the next section.
591 See Scott (2000), pp. 47 f.
592 In their study, Hochberg/Ljungqvist/Lu (2007) for example assume that the VC providing the largest amount is the lead investor. See Hochberg/Ljungqvist/Lu (2007), p. 259. However, in this study it is abstained from making this assumption because as discussions with managers from the private equity and venture capital industry have shown, it cannot be taken for granted that the lead investor always contributes the largest amount.