Master Thesis
Manager's social network and it's impacts on
job promotion
Supervisor: Dr. Tomislav Ladika
Hanying Song
MSc Business Economics, Finance track
Amsterdam Business School
University of Amsterdam
1
Abstract
This thesis contributes to exploration and understanding the role of senior managers’ social networks with CEO in senior managers’ job promotion. Using hand-collected data from S&P 500 firms, I run a linear regression of social network on promotion, controlling the effect of manager age, gender, managerial skill and working year, finding that social connections to CEO play a positive role in senior managers’ promotion. This positive correlation of connections and promotion stands even when I expand CEO to a wider scope, including CFO,COO, and other important chief position.
2
Contents
I. Introduction ... 3
II. literature review ... 6
III. Data and Variables ... 9
A. Firms and Managers ... 9
B. Social networks ... 10
C. Promotion and control variables ... 11
IV. Methodology: ... 12
V. Empirical Results: ... 14
VI. Conclusion: ... 20
3
I. Introduction
In recent years, the issues of firms' internal labor markets are becoming more and more important when we consider the field of corporate governance. More specifically speaking, senior managers play an increasing significant role in theories of internal labor market. Although it is widely known that human resource is one of the most important parts of a firm, and position system is an aspect which cannot be neglected when analyzing the future of a firm, issues related to firms' internal labor markets still have room for improvement. A number of studies find that division managers with better connections to CEO receive more capital (Dunchin and Sosyura 2013; Glasser, Silanes, and Sautner 2013 ) and some research examines role of social networks in efficiency of organizations (Gertner, Scharfstein, and Stein 1994). Yet we know relatively little about whether the social connections between CEO and senior managers effect managers’ promotion.
This paper cares about the relationship between managers’ promotions and their social connections to CEO. Or to say, do managers with social ties to CEO receive more promotions within firm? In this paper, evidence is provided on this question by constructing a hand-collected data set of senior managers and CEOs at S&P 500 firms and studying the effect of managers’ characteristics and connections to the CEO on promotion decisions. Due to the limitation of data, the measure of social networks is focus on networks formed via employment.
The main hypothesis is that: others being equal, more promotion happened to senior managers who have social network to the CEO. This scenario is consistent with the view that an
4
executive’s external social capital is positively associated with the likelihood of promotion (Kim, 2002).To test this hypothesis, I control the effect of senior managers’ age, gender, and education background, which are mentioned in literature review that could potentially affect the possibility of promotion. Then run a linear regression of social network (independent variable) on promotion (dependent variable), which is measured by promotion times (the number of promotion each manager got).
The second hypothesis, which looks deeper into the first one, expand senior mangers’ social network with CEO to a wider scale. Precisely speaking, besides CEO, i included other important position such as CFO, COO, and Chief-title managers. This hypothesis is aim to figure out whether social networks to important person in firm increase the likelihood of getting promoted.
The empirical results indicate that managers’ social network with CEO do have a significant positive effect on managers’ promotion. Managers who have social networks to CEO get more promotion than those who haven’t. And this positive correlation stands when we expand social networks to not only CEO, but also COO, CFO, and other chief title important person in firm.
An concern is that senior manager’s connections to CEO may proxy for managerial skill. For instance, if we assume CEOs are more likely to graduate from Top universities, a senior manager who also graduate from top university may be considered to gain better skill and receive more promotions on the basis of higher managerial ability. To account for this concern, I collect data on senior managers’ education background, as proxy variable for ability, then add it into control
5
variable. This method follows Chevalier and Ellison’s studies (1999) that the correlation between mangers’ average SAT scores and their managerial skill are positive and significant.
This paper contributes to the literature on internal labor markets, corporate governance and social networks. Although lots of existing paper already discussed much of social connections’ role in capital allocation, organization efficiency, information conduits and firms’ position system is often viewed as one of the most important parts of corporate governance, few research talk about the correlation of these two things— social connections’ role in manager’s promotion. In other words, my research is quite novel and explore a new direction in internal labor markets theory. This paper to some extent also improves our understanding of firms position system by studying the effect of social connection on promotion, which has been neglected for a long time.
My research also added to the literature on executive decision making and social connections. Despite managers’ key role in corporate governance, we know little about how lower-level managers will be promoted, this paper fill the gap by demonstrating that social connections do influence promotion decision. As for social connections, I collected data on social networks between top management and senior managers, and qualify the effects of social ties. With my research results, we’ll have a better understanding of firms’ internal labor markets.
The rest of the paper is organized as follows. Part Ⅱ is literature review; part Ⅲ describes the data and variables; part Ⅳ describes the methodology; part Ⅴ examines the role of social connections on promotion and arrived empirical results ; part Ⅵ summarizes and concludes.
6
II. literature review
The notion of social network, or social connections, has attracted considerable interests from social science community in recent decades, social networks can be defined as a set of socially relevent nodes (or network members) that connected by one or more types of relations(Wasserman and Faust 1994), and these units are most commonly persons or organizations. Social networks can also be defined as ties between individuals and/or groups that are not mediated by markets (Willians 2004).
The reason why I pay close attention to social networks is not purely social networks itself, but the influence of it in firm’s internal markets. A number of studies find that CEO-managers’ social connections do influence CEO’s decision-making. Duchin and Sosuyra (2013) find that divisional managers who have social connections with the CEO usually receive more capital than those who haven’t. In addition, managers’ connections with CEO even outweigh seniority and board membership, which are measures of managers’ formal influence, and have positive effect on both managerial appointments and capital allocations.
Willians (2014) explores the impact of social network on labor market, finding that social networks can play an important role as information conduits, and effect job seekers and employers’ decision-making, therefore they have substantial economic impact. Since social networks vary from person to person, they also play an important role in distribute different opportunities among individuals. This is consist with Albert Rees influential work in 1996, he studied the employ data of Chicago labor market and find that over 60 percent of jobs were
7
obtained through informal sources, which we now called social networks. Cohen, Frazzini, and Malloy(2010) have similar results that social connection role as channel of information transfer and softly, difficult-to-verify influence the corporate investment decisions(Petersen 2004).
Except for social networks’ influence on firms internal capital market and on employment as well as information transmission mechanism, earlier researchers also explored their effect on other areas. Cross and Parker (2004) studied the hidden power of social networks, and find that appropriate connectivity in well-managed social networks within organizations can have a substantial impact on performance. What’s more, Individuals who share social connections through mutual qualities or experience have been shown to have more frequent contact, a greater level of trust, and better mutual understanding.
Those papers greatly inspired me: if social networks attribute facilitate information sharing among CEO and managers, and cause distortion in CEO’s rational decision-making process, it is reasonable to hypothesis that they may also cause distortions in promotion decision.
Yangmin Kim (2002) shed light on executive social capital and its impact on job promotion, and arrived interesting conclusions: tie strength to the CEO increase the likelihood of promotion. Moreover, at the individual level, having a good relationship with the CEO is important for one’s career success, this also agrees with conventional argument.
Since what this thesis care about is social networks’ role in managers promotion, above literature already explained former researches that focus on social networks’ role in capital and
8
labor markets, the following literature is about the influential factors of job promotion, especially high level executive promotion.
Ruderman, Ohlott and Kram (1996) studied the gender difference in promotion decision in American Fortune 500 companies, find that women has slower progress to upper-managerial levels than men because bosses are more hesitant to promote women and tend to require women to show their personal strengths and to prove themselves extensively before they get a promotion. Cannings (1988) has similar results from Canadian Corporation, finding that women were only 80 percent as likely as their male colleagues to be promoted during their career within the firm. Diprete and Soule (1988) explored gender and promotion in segmented job ladder systems, no gender difference found in the promotion in upper grades, but significant gender difference exists in promotion from lower grade to upper grade, even control the effect of attributes and location of firms.
Landau(1995) examines the relationship of race and gender to managers’ promotion potential, controlling for age, education, tenure, salary grade, functional area, and satisfaction with career support, they found both race and gender were significantly related to promotion potential. Specifically, females were rated lower than males, and Blacks and Asians were rated lower than Whites. They didn’t found any interaction effects between race and gender.
What’s more, most organization use education as an indicator of personal ability or productivity.( Benson, Finegold, & Mohrman, 2004). There are substantial literatures agree that educational attainments are positively related to career outcomes: such as salary level and
9
likelihood of promotions. Chevalier and Ellison(1999) find that average SAT scores and managerial skill shows a strong positive correlation. However, Berg(1970) holds opposite opinion that many workers are overeducated for their jobs, better educated employees are often rated as less productive.
Robert D, Scott E and Carl (1984) look into the negative relationship between age and promotion, and addressed the behind logic of it. Adams (2002) hold similar opinion, he compared older workers to younger ones with similar demographic and job characteristics, finding that olders are more likely to experience a lower wage growth and early retirement, while younger workers are more favored in promotion decision. Bowen CE and Staudinger UM (2012) also reveal the negative relationship between age and promotion due to the perceived older worker stereotypes (POWS).
III. Data and Variables
A. Firms and Managers
I begin constructing the sample with all firms included in the S&P 500 index, then exclude firms with single division listed in the Compustat Segements file and firms with missing data on senior managers, 300 firms remained after the step. Next step is picking senior manager’s title, although senior manager has various of titles, the most representative ones are needed. To do this, I randomly pickd 30 firms and count how many times each senior manager title is used, finally chose the most high frequency senior manager title, including: vice president, division VP, division senior VP, division president, regional president, regional VP and regional senior VP.
10
Then randomly choose 10-15 managers under those titles for each of the 300 firm, collecting their information including individual profile and network profile from Boardex. Individual profile demonstrates manager’s past working experience and personal information such as age, gender and nationality etc. Network profile including senior manager’s social networks with others, most of the networks are build through common working experience, a small part from social clubs and education background. Finally my sample consists of 3748 people, including 297 CEOs and 3451 other senior managers. Among all the mangers, 2837 managers are male and 615 managers are female, and their average age is 53.78 years old. The database I use is Boardex.
B. Social networks
The best way I can imagine to define social networks agrees with Duchin and Sosyura’s paper(2012), that is connections via education, connections via employment and connections via nonprofit organization(for instance social clubs). However, the data I have collected is not balance with this three type, the majority of connection is via employment, therefore this paper focus on studying whether the social network formed by employment has effect on manager’s promotion. Since this paper is not trying to measure the strength of social network, but the effect of social network on promotion decision, in other works, it is interested in whether social network influence promotion and the effect is positive or negative, not to which extent it affect promotion, therefore the variable “social network” is defined as a dummy variable, equals to 1 if manger has connection to CEO in his firm, or 0 if hasn’t. Because CEO is top management and have overall responsibly for a firm, in theory he should be connected to all the managers in firm, however, some divisional managers have direct work overlap with CEO while others not(the information can be found in manager network profile and manager’s role description), this paper
11
define the first occasion as connected while the second as no connection, since more frequent contact contributes to better information sharing and mutual understanding.
C. Promotion and control variables
This paper use the most direct measure of promotion in all the tests: check all the positions each manager has worked, then count how many times promotion is happened. Data on managers’ positions come from Boardex database.
It is highly possible that the longer a manager works in the current firm, the more promotion he will get. What’s more, longer working years may also facilitate social connections with CEO, to eliminate the possible influence of working year, I include it as control variable “year”, which is the difference of each manager’s start year of first position and start year of last position.
Managerial skill is also important when consider the determinants of promotion. It’s easy to understand that a manager with better managerial skill has greater likelihood to be promoted. However, skill or ability is inherent difficult to measure, we need to find a proxy for it. In this paper, I use education information as a proxy for managerial skill, this approach follows earlier research that done by Chevalier and Ellison(1999),that average SAT scores and managerial skill shows a strong positive correlation. I picked top 25 ranked universities in US and top 5 in UK1,
1
The schools are: Ivy League( Brown University, Columbia University, Cornell University, Dartmouth College, Harvard University, Princeton University, the University of Pennsylvania, and Yale University. ) Massachusetts Institute of Technology, California Institute of Technology, Stanford University, University of California Berkeley, Duke University, University of Michigan, Johns Hopkins University,Carnegie Mellon University, Washington University in St. Louis, Vanderbilt University, Rice University, Emory University, University of Notre Dame, University of
Southern California, Georgetown University in US; and University of Oxford, University of Cambridge, London School of Economics and Political
12
and create a dummy indicator of attendance of those universities as control variable, proxy for managerial skill.
Age and gender are also control variables that cannot be neglected. As shown in literature review, lots of earlier studies find age has a negative effect on promotion, older workers are less likely than younger ones to be promoted, it is necessary to eliminate the influence of age when we look into the effect of social network. The same to gender, accidence or not, as Table I demonstrates, male account for a much larger percentage of managers than female, as well as the former studies point out gender difference do exists on promotion decision, the effect of gender is controlled in my test.
IV. Methodology:
This thesis uses a simple OLS regression to examine whether managers have social networks with CEO in firm get more promotions. The independent variable is social connections between CEO and managers, defined as dummy variable of connections between the divisional managers and CEO based on current employment. The dependent variable is the managers' position times. To get more accurate results, i also include some control variables: working years in current firm, education, manager age, and manager gender. The regression function is:
The dependent variable is each manager’s promotion times, measured by the number of times each senior manager be promoted in primary firms. The independent variable is social networks, a dummy indicator that equals to 1 if manager has connection to CEO in his firm (individuals’ all
13
connections could be found in his network profile ), and equals to 0 if manager has no connection to the CEO. There are four control variables : year, education, age and gender. Control variable year is defined as the difference of start year of manager’s first position in firm and start year of his last position, it measures how long does a manager takes all his promotion in firm,and aim to eliminate the influence of working years on promotion decision. Control variable education is a proxy for manager’s managerial skill, and is defined as a dummy variable equals to 1 if manager had attend the American top 25 / UK top 5 universities or equals to 0 if not. Control variable gender is also labeled as a dummy variable, equals to 1 if the manager gender is male, or equals to 0 is her is a female manager, controls the effect of gender difference. Last control variable age is simply equals to each manager’s age, aim to exclude age influence. Actually, it’s better to add another control variable race since earlier studies show race difference has effect on promotion decision, unfortunately I have no access to manager’s race information, this part can be improved by adding race.
An important concern in identifying the correlation of social connections on promotion is potential reverse causality, a scenario that senior managers who receive more promotion develop stronger connections with the CEO. If we want to figure out the causal relationship of social network on promotion, this would be a hidden trouble, the best way to address this concern is to exclude all the connections that were established after the manager arrived at the firm of interest, however, for this paper’s collected data, this filter is too aggressive that more than half of the data will be filtered in that way. Although the data isn’t so rigorous, regression results can also demonstrate whether social networks have positive effect on manager promotion, this part can
14
be improved if I have more data about connections that are formed via prior employment, education background and social clubs(or nonprofit organizations).
V. Empirical Results:
A. Identifying the role of social networks with CEO on manager promotion
To identifying social networks’ role in manager promotion, using collected data and following the equation listed in methodology section, I run several simple OLS regression and the aggregate results are shown in Table I.
Table I
T-values are given in parentheses under the coefficients, and the individual coefficients is statistically significant at the *10%, **5%, or ***1% significant level.
Dependent variable: Promotion times
Regressor 1 2 3 4 5 Social networks 0.2007** 0.1700** 0.1914* 0.1914* 0.1576 (2.44) (2.41) (1.94) (1.94) (0.93) Years 0.1444*** 0.1572*** 0.1572*** 0.1739*** (32.36) (26.84) (26.83) (19.34) Manager's age -0.0214*** -0.0214*** -0.0170* (-3.51) (-3.52) (-1.80) Manager's Gender 0.0087 0.1028 (0.07) (0.65) Top school -0.0711 (-0.51) Constants 1.5156*** 0.5629*** 1.8296*** 1.8237*** 1.5754*** (21.03) (8.98) (5.38) (5.15) (2.88) Observations 3344 2744 1658 1658 729 R-squared 0.0015 0.4561 0.4722 0.4722 0.5154
15
The dependent variable is promotion times, which measured by how many times each senior manager get promoted; and the independent variable is social network, a dummy variable equals to 1 if manager has connection (direct work overlap) to CEO in his firm or 0 if the manager doesn’t have connection to CEO. There are four control variable in total, namely: year, manager age, manager gender, and manager education.
At first, I just include one independent variable: social network, and run a regression on promotion times, results turn out to be consistent with main hypothesis, social network has significant positive effect on manager promotion. Obviously the regression doesn’t fit well due to omitted variables, as we can see from the table, the R-square is extremely low, which indicates social connection has quite low explanatory power of promotion times under this scenario.
Next, an important control variable year is added to the regression equation, which is measured by the difference between the start year of manager’s first position and start year of his current position, to control the effect of time, since the longer a manager worked in the firm, the more promotion he is likely to get. The results demonstrates year is an valid control variable because R-squared rise significantly to 0.456, what’s more, the coefficient of year is positive, means longer working year play a positive role in manager promotion, and the positive effect is statistical significant. With control variable “year”, independent variable “social network” is still positively significant under significance level of 0.05.
Then variable “manager age” is added to regression equation, which is actual age of each manger. In my sample, manager age range from 34 to 88, 53% of them are between 50 years old and 60
16
years old. As the table above shows, age has significant effect on promotion decision, in a negative way. The results follow earlier studies that reveal the negative correlation of age and promotion (Robert D, Scott E and Carl 1984; Bowen CEand Staudinger UM 2012; Adams 2002). The significance level of “social connection” drops slightly while coefficient of “social connection” and “year” rose.
After that I add “manager gender” into regression equation, a dummy indicator equals to 1 if manager gender is male or 0 if manager is female, aim to see the effect of gender. Although the coefficient of gender is positive, suggesting that male take advantage of female in promotion decision, the variable is statistical insignificant. The results indicate that gender doesn’t have significant influence on manager promotion. A possible reason for this result is 82% of managers in sample are male, only 18% are male, the unbalanced gender ratio itself may proof that male is more likely to be promoted to a manager position. And since majority managers are already male, it’s easy to understand why gender advantage is weak in our results.
Finally, education variable “top university” is added. It is also a dummy variable which equals to 1 if manger had attend US top 25 or UK top 5 universities, or equals to 0 if not. This control variable is a proxy or managerial skill, aim to control the effect of manager’s personal ability. However, the results turn out to be different from former hypothesis that better education background benefit to promotion. There is a serious problem with the education data — the number of observations drops sharply, among 3451 senior managers, only 1099 managers have education information, other 2352 managers’ education data is not available. This could partially explain the abnormal regression results, less than one third managers could not represent whole
17
sample, what’s more, it is possible that the other two third people follows earlier research but unluckily been excluded due to data deficiency. We can notice that the independent variable “social network” became insignificant after “top university” be added in regression equation, there are several possible reasons, one is education sample is too small to be a good control variable, it distort regression results; the other possible reason is variable “top university” is neither a accurate proxy for education background, nor a good proxy for managerial skill.
In general regression results agree with the main hypothesis that social networks’ play a significant positive role on manager promotion. And this results stay consistent when gradually add control variable to exclude the effect of working year, manager age and gender, but fail to keep significant when education variable is added, since education data is less than one third of manager number, it’s too small to draw any conclusion. If all the managers’ education data is available, we can get more accurate results.
B. Identifying the role of social networks with important person in firm on manager promotion
Then comes to check the second hypothesis, which is similar to main hypothesis, but expand connections to CEO to a wider scope, both CEO and other important person in firm, including CFO, COO, and other chief titles such as chief accounting officer and. The “important person” sample has 408 people in total, and manager size decrease to 3342. All the control variable has the same processing method as testing main hypothesis, the only difference is sample size. Regression results are shown in table II.
18 Table II
T-values are given in parentheses under the coefficients, and the individual coefficients is statistically significant at the *10%, **5%, or ***1% significant level.
Dependent variable: Promotion Times
Regressor 1 2 3 4 5 Connections 0.4695*** 0.1272* 0.1272* 0.0695 -0.0684 (5.96) (1.71) (1.71) (0.66) (-0.35) Years 0.1430*** 0.1429*** 0.1557*** 0.1709*** (31.10) (31.05) (25.76) (18.79) Gender 0.0476 -0.0184 0.0833 (0.64) (-0.15) (0.53) Manager's age -0.0178 -0.0136 (-2.92) (-1.44) Education -0.1913 (-1.54) Constants 1.1984*** 0.5730*** 0.5343*** 1.7159*** 1.6019*** (17.46) (8.57) (6.03) (4.74) (2.80) Observations 3341 2641 2641 1580 691 R-squared 0.0088 0.4567 0.4567 0.4751 0.5260
The dependent variable is each manager’s promotion times, and the independent variable is social network, a dummy variable equals to 1 if manager has social network with important person in firm, or 0 if hasn’t. Other independent variables include the following control variables: working year, manager age, manager gender and education, the same as former testing.
As Table II shows, social network with important person demonstrates a significant positive effect on manager promotion when regression alone, but the low R-square suggests that other important variable may omitted. Then we control the effect of promotion year— the difference of start year of manager’s first position and final position, both social network and working year are statistically significant and have positive impact on manager promotion. To eliminate the gender difference on promotion decision, variable manager gender is added to regression equation, gender is also a dummy variable equals to 1 if manager is male or 0 if is female.
19
Manager gender doesn’t have significant impact on promotion decision according to the results, although the positive coefficient suggests that male managers just potentially have slightly advantage over female managers on promotion decision.
However, big change happens when control variable “manager age” being added in. The coefficient of social network drop by half and become statistical insignificant, nor does manager age coefficient, while coefficient of manager gender abnormally turn from positive to negative, which indicates female manager is more likely to be promoted than male manager when other things being equal. This scenario is opposite to our conventional thoughts. The coefficient of manager age is negative but insignificant, which suggests older managers have fewer chance of being promoted, but the effect of age is quite weak.
At last education variable “top university” is included. But the results turn out to be unsatisfying. Except for variable year, none of variables are significant, what’s worse, social network with important person and better education seems to prevent manger promotion under this situation. Possible reasons are inadequate education information cannot represent whole sample, and “top university” isn’t a good proxy variable. It is easy to notice that in both model, “top university” doesn’t do a good job, if more information is accessible, better proxy for managerial skill could be found.
In summary the regression results indicate that social networks’ positive role in manager promotion stay consistent when we expand social networks to a wider scope, including CEO, COO, CFO, and other chief position. However, in the model of main hypothesis, the role keeps
20
significant even when we control the effect of working year, manager age and manger gender, in model of second hypothesis the role is generally less significant.
Both hypotheses are supported by empirical results above, social networks with CEO and other top management have positive effect on managers’ promotion. This result provide an extra perspective when consider firms’ position system, and point out the important role of social network in career success of senior manager.
VI. Conclusion:
The aim of this article is to examine the role of social networks to CEO in manager promotion, controlling possible effect of managerial skill, working year, manager gender and age. The empirical findings suggest a significant positive correlation between social network with CEO and manager promotion times. This correlation also holds when I expand social network with CEO to social network with CEO and other important person in firms.
The results of this study provided a fresh perspective for corporate governance by pointing out the ties to CEO influence promotion decision and contribute to managers’ career success. At the individual level, have a good relationship with CEO is beneficial for his career, but at the organizational level, it may not be beneficial and rational to make promotion decision based on managers’ relationship with CEO. This results make sense for validating the conventional argument above. If we fully noticed this point, we could check and revise firms’ position system and make an effort to neutralize influence of social network when making promotion decision.
21
However, this paper could be improved in several aspects if necessary data is available. The most important one is the proxy for managerial skill. Since earlier studies proved average SAT scores could proxy for managerial skill, i use adummy variable “top university” to replace SAT scores because in theory higher scores end up with better university. However, the unsatisfying results of both tables about variable “top university” suggest it’s not a good one, partly due to small sample size and partly due to indirect approach to control managerial skill. Managerial ability is difficult to measure directly, manager’s relative performance could be a better proxy than “top university”. The way to measure manager’s relative performance is analogical to (Duchin and Sosurya 2013), use the difference between the industry-adjusted ROA of manager’s division and the average industry-adjusted ROA of firm’s other divisions. Since ROA gives an idea as to how efficient management is at using its assets to generate earnings, it could be
appropriate to represent managerial skill.
Another concern is omitted variable bias. Due to limited data, this paper only focus on managers’ internal social network with CEO (i.e. social network formed via employment), therefore omitted
an important variable— external social network, including social network formed in School, in
previous employment,or in nonprofit organization such as social clubs and religious
organizations. Those connections can foster a sense of belonging and facilitate a stronger
network between manger and CEO in firm, the influence manager promotion. This is supported
by Useem and Karabel’s earlier research(1986), that external networking ability play an important role in executive promotion. If we have data on external social networks, we can
22
Firm size and Firm growth rate could be omitted variables too. Big firms have more opportunity
of promotion than small firms, and firms with high growth rate tend to have better chance of
promotion. In addition, the importance of social network with CEO varies in different firm size,
and varies when firm’s growth rate is different. If we control the effect of firm size and firm growth rate, we can better measure the impact of social network.
A question left unexplored in this thesis is the size of social capital. A dummy variable of 0 0r 1
can only show whether managers’ social network with CEO has significant influence on managers’ promotion, it do not measure the strength of social network. Tie strength to the CEO will be an important issue since most senior manager of a firm must have a nodding
acquaintance with the CEO, the strength of tie to the CEO may make a difference on promotion
decision. There are many possible ways to measure the strength, one is count the average number
of connections between senior manager and CEO based on internal and external social networks,
the other is count how many year manger and CEO know each other. Another natural extension
to the thesis is to figure out the role of external social network on manager promotion. Is
managers’ external social network positively associated with the likelihood of promotion? Do the strength of correlation between social network and the likelihood of promotion significantly
different between high growth firms and low growth firms? These are all areas for future
23
References:
Adams, S. J. 2002. Passed over for promotion because of age: an empirical analysis of the consequences. Journal of Labor Research, 23(3), 447-461.
Bowen, C. E., and Staudinger, U. M. 2013. Relationship between age and promotion orientation depends on perceived older worker stereotypes. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 68(1), 59-63.
Benson, G. S., Finegold, D., & Mohrman, S. A. 2004. You paid for the skills, now keep them: Tuition reimbursement and voluntary turnover.Academy of Management Journal, 47(3), 315-331. Berg, I. 1970. Education for Jobs; The Great Training Robbery.
Cannings, K. 1988. Managerial promotion: The effects of socialization, specialization, and gender. Indus. & Lab. Rel. Rev., 42, 77.
Chevalier, J. and Ellison, G. 1999. Are Some Mutual Fund Managers Better Than Others? Cross-Sectional Patterns in Behavior and Performance. The Journal of Finance 54: 875–899.
Cohen, L., Frazzini, A., & Malloy, C. 2010. Sell‐Side School Ties. The Journal of Finance, 65(4): 1409-1437.
Cross, R. L., and Parker, A. 2004. The hidden power of social networks: Understanding how work really gets done in organizations. Harvard Business Press.
DiPrete, T. A., & Soule, W. T. 1988. Gender and promotion in segmented job ladder systems. American Sociological Review, 26-40.
Duchin, R. and Sosyura, D. 2013. Divisional Managers and Internal Capital Markets. The Journal of Finance, 68: 387–429.
Gertner, R. H., Scharfstein, D. S., & Stein, J. C. 1994. Internal versus external capital markets (No. w4776). National Bureau of Economic Research.
Kim, Y. 2002. Executive social capital and its impacts on job promotion. Academy of Management Proceedings: J1–J6.
Landau, J. 1995. The relationship of race and gender to managers' ratings of promotion potential. Journal of Organizational Behavior, 16(4), 391-400.
Ruderman, M. N., Ohlott, P. J., and Kram, K. E. 1996. Managerial promotion: the dynamics for men and women. Center for Creative Leadership.
24
Wasserman S, Faust K. 1994.Social Network Analysis . Cambridge University, Cambridge. Williams, R. E. 2004. Social networks and labor market outcomes: Theoretical expansions and econometric analysis.
