The Impact of Interfirm Labor Mobility on Innovation: Evidence from Job Search Portal Data

Abstract

Mobility of workers is one of the most important channels of knowledge spillovers between firms. In this study, we look at the relationships between interfirm labor mobility and technological innovation at firm level. For the analysis of labor mobility, we use a novel Estonian database from an online job search portal that includes detailed data on occupations and education. The employee level data is matched with the Community Innovation Survey data on business enterprises. We estimate various specifications of the knowledge production functions augmented using mobility indicators. In particular, the results indicate that product innovations and total factor productivity are associated with subsequent higher worker flows from other, especially innovative, firms. Among flows involving people in different occupations, the flows of managers and professionals and technicians are more important.

Appendices

Appendix 1 Overview of Selected Studies on the Links Between Labor Mobility and Firm Innovativeness

Author(s)	Data	Method	Main results
Maliranta et al. (2008)	Finland, 1995 and 2000, longitudinal matched employer–employee data	Regression for the relative changes in labor productivity, wages and profitability regressed in the hiring and separating rates and the share of staying workers	Hiring workers previously employed in R&D to one’s R&D department did not increase the hiring firm’s productivity; however, both productivity and profitability increased when workers working previously in R&D jobs were hired into non-R&D activities.
Graversen et al. (2002)	Sweden, Norway, Finland, Denmark: registers containing information on all employees, 1990s	Calculation of the number of people and respective rates of mobility between different sectors broken down by education of employees	The turnover is high, especially among employees who have entered the workplace recently and younger employees, there are large net flows from public sector to higher education, while there are also net flows from higher education and research institutes to the business sector
Tomlinson and Miles (1999)	CIS2 data for 2,400 firms of the UK, 1998; Employment in Britain survey among 3,855 respondents	Regression models of various learning indicators of employees on indicators of career history (number of jobs, tenure); regressions of organizational commitment variables in job shifts between and with the employer	Dynamic knowledge flows (individuals learning new competencies and firms enhancing their capabilities) are fostered more by intra-firm than interfirm mobility; external mobility may have some negative consequences in terms of organizational commitment
Møen (2005)	Matched employer–employee dataset from Norwegian machinery and equipment industry, 1986–1995, full-time males	Mincerian wage regressions augmented with the firms R&D intensity at current job and over previous career	Technical staff in R&D-intensive firms get wage payments for the knowledge acquired at work with lower wages during the early career, but gets later wage premium for the knowledge accumulated
Almeida and Kogut (1999)	Patent data on semiconductor industry of the US, career paths of 438 semiconductor engineers constructed from patenting record	Calculations of rates of intra- and interregional mobility (regions as semiconductor clusters), logistic regressions for localization of knowledge (major patent and citing patents being from the same region) regressed in mobility measurements and other variables	The interregion mobility is especially high in Silicon Valley; mobility of patent holders across firms has an influence on the local transfer of knowledge, i.e., higher mobility within the region increases localization of knowledge.
Lenzi (2006)	PATVAL survey from Italy on 106 holders of patents in pharmaceuticals	Duration model (Cox semiparametric approach) for the length of job spell	In addition to other determinants of job mobility, the interfirm mobility is enhanced by the inventive productivity of the inventor
Magnani (2006)	USA Panel Study of Income Dynamics data on 20,000 male technical employees, 1981–1992	Earnings regressions with additional independent variables of industry-level R&D and its interaction with experience and tenure	Some evidence that exposure to R&D activities allows workers to accumulate general human capital at the early stage of their career for the reason that they pay the price in terms of lower earnings
Fallick et al. (2006)	Current Population Survey of job movers in Silicon Valley computer industry, 1994–2001	Probit models for month-to-month job changes, independent variables region, education, family status, age, etc.	The job-hopping rates for males with college education are higher in Silicon Valley than in computer clusters outside of California; the higher mobility in California could be due to state laws restricting non-compete agreements
Müller and Peters (2010)	Mannheim Innovation Panel (innovation survey of German firms), 2005, 2006, 2008	Knowledge production functions for process and product innovations estimated as bivariate probit models	The probability of innovation first increases, but then decreases with labor churning (simultaneous hiring and separation of employees); in the case of non-duplicative knowledge churning, which has a more negative impact. The optimal churning is also larger for product than process innovations
Görg and Strobl (2005)	Ghana, enterprise survey, 1991–1997, 278 domestic manufacturing firms	Measurements of firm level total factor productivity (estimated from Cobb–Douglas production function) regressed in the indicator variables on the mobility and work experience of the firm’s owner	Domestic firms with owners having past employment in a foreign firm within the same industry are more productive than the other domestic firms
McCann and Simonen (2005)	CIS data for Finland augmented with R&D surveys and business register	Probit models for various types of innovations (knowledge production functions) with cooperation variables and variables for the share of new employees	Innovation is positively associated with a larger amount of new employees from other subregions, but negatively associated with new employees from the same location
Tambe and Hitt (2007)	Mobility data from the US leading career portal matched with several databases on publicly listed companies (Compustat, Compact Disclosure Database)	Productivity regressions with internal stock of process knowledge and knowledge spillover variables (constructed as the stock of knowledge of other firms weighted with interfirm flows)	Interfirm worker mobility creates significant IT-related process innovations. Spillovers are larger in the case of the mobility of skilled workers and the workers directly involved in the daily operations of the firm

Appendix B Definitions and Summary Statistics of Variables Used in Descriptive Tables and Regression Analysis

Variable name	Description	Mean	Standard deviation
Log number of employees	Natural log of the number of employees	3.269	1.188
Product innovation	Dummy, 1 if firm reports having introduced new or significantly improved product	0.213	0.409
Process innovation	Dummy, 1 if firm reports having introduced new or significantly improved production process	0.218	0.413
Innovation expenditure dummy a	1 if firm reports positive expenditure on innovation	0.198	0.399
International competition	Dummy, 1 if the firm’s most important market is the international market	0.531	0.499
Formal protection	Dummy, 1 if firm uses registration of design patterns, trademarks, copyright to protect inventions or innovations	0.085	0.279
Public funding	Dummy, 1 if firm received public funding for innovation projects	0.025	0.155
Sources within the firm or other firms within the group	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.553	0.350
Competitors	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.385	0.343
Customers	‘4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.519	0.365
Supplier	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.545	0.370
Lack of appropriate sources of finance	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.404	0.396
Innovation cost too high	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.382	0.403
Lack of qualified personnel	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.362	0.378
Lack of information on technology	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.232	0.298
Lack of information on markets	′4 values, 0, 1/3, 2/3, 1; higher value indicates greater importance	0.237	0.307
Share of new employees	Ratio of new employees (2001–2005) to total employment	0.047	0.096
Share of new employees with higher education	Ratio of new employees (2001–2005) with higher education to total employment	0.014	0.041
Share of new employees with past employment in innovative firm	Ratio of new employees (2001–2005) with past job in innovative firm to total employment	0.010	0.037
Share of leaving employees	Ratio of leaving employees (2001–2005) to total employment	0.026	0.059
Churning rate	Ratio of churning between 2001 and 2005 to total employment	0.012	0.030
Managers	Ratio of new managers (2001–2005) to total employment	0.001	0.007
Professionals	Ratio of new professionals (2001–2005) to total employment	0.000	0.006
Technicians	Ratio of new technicians (2001–2005) to total employment	0.001	0.009
Elementary occupations	Ratio of new workers with elementary occupations (2001–2005) to total employment	0.000	0.002
Service and sales workers	Ratio of new service and sales workers (2001–2005) to total employment	0.000	0.002
Operators, craft and trade workers	Ratio of new operators, craft and trade workers (2001–2005) to total employment	0.001	0.005
Clerks	Ratio of new clerks (2001–2005) to total employment	0.000	0.005