Directions and Paths of Knowledge Flows through Personnel Mobility: A Social Capital
4.4 Data and methods
4.4.2 Measures Dependent variable
Our dependent variable, the knowledge flows between companies through personnel mobility, is operationalized as the number of people that over time left company A to move to company B. In order to account for the value of the intellectual capital that these employees are embedding, we weighed every flow with the higher education degree of the person moving. In order to do that, we codified all the spectrum of possible Indian education levels into 7 layers, and we weighed each personnel movement accordingly.
Independent variables
Our three main explanatory variables have been operationalized through three four-layered categorical variables. Each layer of the variable behaves as an independent variable within the model. In order to measure the effect of gain from competitor or loss to a competitor, which in our model is operationalized as two different layers of the same variable, Somaya et al. (2008) have opted for two different dummy variables. We believe that our choice of three multi-layered categorical variables carries the benefit to explain the likelihood of different scenarios along the three different dimensions, adding in clarity and quality of the analysis.
Foreign affiliation. In order to test the effect of foreign affiliation on the probability of a personnel movement to occur from company A to company B in a directed dyad, we made use of a four-layer categorical variable. A simple dummy variable indicating either homogeneity or heterogeneity between the source and the recipient firm would have missed out the direction of the flows, which is of utmost importance to clearly understand the distribution of personnel mobility.
Considering our data on companies’ ownership, we thus coded every company holding a foreign affiliation as a MNC and every company owned by an Indian private or an Indian group as a domestic firm. Building on this, the four types of possible dyads have been coded, shaping the four layers of our categorical variable: 1. flow from domestic firms (DOM) to domestic firms 2. from domestic firms to MNCs 3. flow from MNCs to domestic 4. flow from MNCs to MNCs.
Cluster affiliation. To test the effect of the co-location on personnel mobility we coded the companies with headquarters or branches in Bangalore as a cluster company, while every other company has been generically considered an Indian company. The case of people moving cross-border, i.e. from India to some other country, was sought for, in order to add further variance to our scenarios. Yet, no case emerged from empirical evidence.
Considering the directed dyad, we used a multilayer categorical variable to portray the 4 types of employees’ flows as emerging from our coding: 1. flows within the cluster (from Bangalore to Bangalore) 2. flows from the cluster to outside (from Bangalore to India) 3.
flows from outside to the cluster (from India to Bangalore) 4. flows from outside to outside (from India to India).
Industry affiliation. Industry affiliation has been coded according to the companies’
economic activity. Our focal interest, like the Bangalore cluster for the geographical dimension, is the IT industry. Considering this as the reference point, companies have been coded as IT whether their economic activity is listed as computer software or ITES (Information Technology Enabled Services), else otherwise. Similarly to the procedure followed from the two previous variables, in order to portray the differences of industry affiliation in our directed dyads of companies, we used a multilayer categorical variable, distributed as follows: 1. flows within IT industry 2. flows from IT to other industry 3. flows from other industry to IT 3. flows from other industry to other industry.
Controls
We consider three main control variables that might be correlated to the probability of employees’ movements between two companies: age, size and financial performance, which translates into the differential of age, size and financial performance at dyad-level.
Age has been operationalized drawing on the year of incorporation for each firm: the differential has been computed for each dyad of companies. We believe that the age difference can be correlated to the likelihood of employees’ flows between firms for several reasons. A young firm is likely to be willing to attract new resources drawing on the knowledge of more experienced companies. Likely, older firms might be willing to encourage inflows of people previously employed in younger and dynamic companies, in order to renew and refresh the internal knowledge base. In both events, there are reasons to think that age differential might play a role in influencing the distribution of personnel mobility between companies.
Size has been measured through the total income of the company; the differential has been computed for each dyad. Size is likely to influence the probability of employees’
movement between two companies. A strong differential in size can intuitively lead to a strong attraction power of the larger company towards the employers of the smaller, because of reputation effects, sensible wage differentials and a stronger exposure to structured internal processes. On the other side, larger companies might be interested in attracting employees from smaller firms, because more keen on easily absorbing larger companies’ complex internal routines.
Likewise, profitability gap is intuitively conducive to a flow from less profitable to more profitable companies, because of the latter’s attraction power. Though, a steady policy will be enacted by the less profitable companies to attract employees from the more profitable ones, in order to enhance the internal management, skills and knowledge base, thus draining
some competitive advantage or best practices. Differential in profitability has been operationalized by computing the EBIDTA difference of the companies in the companies for each dyad on year 2007.