Table 2. Interpreting worker flows Firm condition
Worker condition Possible interpretation
Predecessor category 1
F1 True : Firm exit
W1 True : 80 of workers go to successor
ID change or mergeracquisition 2
F1 True : Firm exit
W1 False : 80 of workers go to successor
Spinoffbreakout or reason unclear 3
F1 False : Firm continues
W1 True : 80 of workers go to successor
ID change or mergeracquisition 4
F1 False : Firm continues
W1 False : 80 of workers go to successor
Spinoffbreakout or reason unclear Successor category
1 F2 True
: Firm entry W2 True
: 80 of workers come from predecessor ID change or spinoffbreakout
2 F2 True
: Firm entry W2 False
: 80 of workers come from predecessor Mergeracquisition or reason unclear
3 F2 False
: Firm continues W2 True
: 80 of workers come from predecessor ID change or spinoffbreakout
4 F2 False
: Firm continues W2 False
: 80 of workers come from predecessor Mergeracquisition or reason unclear
1. The original administrative entity shut down. 2. More than 80 of the workers in the predecessor firm
moved to the successor firm. 3. The successor firm was a new entrant.
4. More than 80 of the workers in the successor firm came from the predecessor firm.
This sequence of events strongly suggests that the factors of production in the two firms are virtually the same, and that the
flows are the result of either an unlinked administrative edit or a change in ownership. The evidence is less strong for the 1, 3
category, where the successor firm was already in existence, but this is included because of the timing issue noted by Pivetz et al.
2001, and also because it is possible that the few employees in the successor firm before the event form part of a shell cor-
poration United States Department of Labor, Employment and Training Administration 2002. It is, of course, possible to in-
vent alternative scenarios, such as in the 1, 3 cell, whereby a startup firm is able to woo and attract large numbers of work-
ers from another firm. Similar arguments can be made for the combination of events in the 3, 1 and 3, 3 categories.
Some of the combinations identified in Table 3 are consistent with the sequence of events that occur during a merger or acqui-
sition. The 1, 2 cell, for example, identifies a firm shutdown combined with the move of 80 of workers into a newly born
firm, where those workers represent 80 of the new firm’s workforce. The 1, 4 cell represents the same transition, al-
beit for a successor that is an existing firm. Similarly, the 3, 2 and 3, 4 cells identify a continuing firm that has 80 of its
workers transitioning to a newly born firm column 2 or a con- tinuing firm column 4 also suggesting that either a merger or
an acquisition took place.
The final task involves identifying insourcing and outsourc- ing relationships. Typically, these types of arrangements in-
volve peripheral firm functions, such as payroll or human re- sources, and thus are not likely to make up 80 of either
the predecessor or successor firm’s employment. Therefore, the best candidates for these types of transitions are likely to be
found in the “reason unclear” cells. In each of these cells, a sub- stantial cluster of workers at least 5, but 80, move from the
predecessor to the successor firm and account for 80 of the
Table 3. Successorpredecessor flow and firm birthdeath combinations
Link description Successor category
1. 80
of successor
comes from predecessor and successor is entrant
2. Less than 80 of suc- cessor comes from pre-
decessor and successor is entrant
3. 80
of successor
comes from predecessor and successor was in
existence
4. Less than 80 of suc- cessor comes from pre-
decessor and successor was in existence
Predecessor cate
gory
1. 80 of predeces- sor moves to succes-
sor and predecessor exits
ID change Acquisitionmerger
ID change Acquisitionmerger
2. Less than 80 of predecessor moves to
successor and prede- cessor exits
Spinoffbreakout Reason unclear
Spinoffbreakout Reason unclear
3. 80 of predeces- sor moves to succes-
sor and predecessor lives on
ID change Acquisitionmerger
ID change Acquisitionmerger
4. Less than 80 of predecessor moves to
successor and prede- cessor lives on
Spinoffbreakout Reason unclear
Spinoffbreakout Reason unclear
Downloaded by [Universitas Maritim Raja Ali Haji] at 23:07 12 January 2016
workers at the successor firm. From the evidence presented in Table 1, clustered worker flows of this size are very rare events
.2 of worker movements, suggesting that the underlying transitions may be the result of an insourcing or outsourcing re-
lationship between two employers. We explore this hypothesis in various ways in the remainder of the article.
5. EMPIRICAL ANALYSIS OF CLUSTERED
WORKER FLOWS 5.1
Relative Frequency of Transitions Table 4 begins by documenting the relative frequency of the
16 worker flow classifications identified in Table 3. A brief analysis of Table 4 yields a number of interesting results. By
an overwhelming margin, the most numerically frequent set of links occurs in cell 4, 4, where the transitioning cluster ac-
counts for 80 of both the predecessor and successor firms’ total employment and where neither firm enters or exits. It is
noteworthy that although the “reason unclear” category domi- nates Tables 5 and 6 in terms of the number of links, the size
of the clusters tends to be relatively small. These results are quite robust; when we regenerated Table 4 for minimum flows
of 8 and 10 workers, we found that the relative distributions re- mained essentially unchanged. Not surprisingly, however, the
weakest link cell {4, 4} decreased in relative importance from the 73.49 reported in Table 4 to 72.79 for cluster size 8 and to
70.65 for cluster size 10.
One possible explanation for this phenomenon is outsourc- ing; another explanation is that the worker clusters simply rep-
resent transfers between establishments under some larger, sin- gle corporate identity. Yet a third possibility is that within-firm
networking leads groups of workers to move to new opportuni- ties together. We explore each of these possibilities later in the
article. For events defined by either the exit of the predecessor or the
entry of the successor firm, most transitions are accounted for by clusters of workers that make up “80 of employment.” In
contrast, the links that do not involve entryexit are dominated by the “80” condition. This suggests that clustered flows in-
volving entryexit are not associated with a fundamental change in firm activity and may well reflect a missed administrative edit
or ownership change. This is particularly true for the 1, 1 cell.
5.2 Using Industry to Shed Light on Firm Dynamics
The 16 firm relationship classifications identified in Table 3 differ vastly in terms of their effect on a firm’s workforce. Some
changes imply a large amount of structural change, whereas others have minimal impact. This suggests that certain types
of links, such as an ID change, where the factors of production are similar in both the successor and predecessor firm, likely
involve two firms in the same industry. Other links, in which relatively few workers have transitioned to a new firm, are more
likely to be in a new industry.
In particular, outsourcing and insourcing are two activities likely to involve a change of industry. For example, if a firm
outsources its information technology staff, then the new em- ployer is likely to be in the computer services business, which
is almost by definition not the primary industry of the predeces- sor firm. The temporary-helppersonnel supply industry Stan-
dard Industrial Classification 7363 is another important source of both insourcing and outsourcing transitions. For example, as
Table 4. Relative frequency of successorpredecessor combinations Successor category
1. 80 of successor come from predeces-