Tournament Theory

Rosenbaum’s (1979) tournament model (described in more detail in the Methodology section) argues that career mobility can be represented as rounds of a tournament, where each round has winners and losers and the results of each round affect subsequent career opportunities. We use 10-year career sequences of veterans beginning after their last military job to test hypotheses about career mobility.

We are particularly interested in two potential tournament “results” for veterans: early promotion (promotion within two years of leaving the military) and transitional unemployment (unemployment occurring directly after leaving the military). If these two results can be shown to be associated with positive or negative career outcomes, this could provide evidence in favor of certain types of career development interventions for veterans. As in the rest of our analysis, we use O*NET job zones (described further in the Data section) to represent different career levels; an increase in job zone is considered a promotion, and a decrease is considered a demotion. We also include “left” unemployment in our sequences to represent transitional unemployment after leaving the military.

Hypothesis: Promotion in the first period is associated with promotion in the second period

To test our first hypothesis, we group our sequences by individuals promoted in the first period (promoted after two years) and individuals promoted in the second period (promoted after four years). If this association exists, it would be evidence that promotion in early “tournament rounds” improves the chances of promotion later.

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  tblA
## X-squared = 0.23734, df = 1, p-value = 0.6261

The results of a chi-squared test (above) do not show a strong statistical assocation between promotion in the first period and promotion in the second period. This may be because our population (the DC MSA veterans) is already highly educated, and so a veteran is more likely to start in a higher job zone. If an individual starts in job zone four or five, there is not much upward mobility possible.

Hypothesis: Promotion in the first period is associated with employment in a later period

To test our second hypothesis, we group sequences by veterans who were promoted in the first employment period (promoted after two years), and individuals who were employed for the full remainder of the period (from year 3 to year 10, represented in the table below as Y3-Y10). If this association exists, it is evidence that promotion in early rounds of the tournament results in a positive career outcome (staying employed for the duration of the 10-year career).

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  tblB
## X-squared = 7.3905, df = 1, p-value = 0.006557

The results of the chi-squared test (above) show strong statistical evidence (p < 0.01) that early promotion is associated with employment for the remaining duration of the sequence.

Hypothesis: Transitional unemployment is associated with unemployment later in the sequence

To test our third hypothesis, we group sequences by veterans who were unemployed for two years following their last military job (the first two years of the sequence - transitional unemployment) and veterans who were unemployed sometime in the remainder of the sequence (year 3 to year 10, represented in the table as Y3-Y10). If this association exists, it is evidence that early unemployment (a “loss” in a tournament round) can cause “losses” in later rounds of the tournament.

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  tblC
## X-squared = 386.18, df = 1, p-value < 2.2e-16

The results of the chi-squared test (above) show a strong statistical association (p < 0.01) between transitional unemployment and unemployment in the later years of a career sequence; in fact, this is the strongest association we present.

Hypothesis: Transitional unemployment is associated with finishing unemployed or in a lower job zone (job zone 1 or 2)

To test our third hypothesis, we group sequences by veterans who were unemployed for two years following their last military job (the first two years of the sequence - transitional unemployment) and veterans who finished the ten-year period in either job zone one or two (or were unemployed). If this association exists, it is evidence that early unemployment lowers the career “ceiling” (highest possible career) of a given veteran.

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  tblD
## X-squared = 23.234, df = 1, p-value = 1.435e-06

The results of the chi-squared test (above) show a strong statistical association (p < 0.01) between transitional unemployment and finishing the sequence either unemployed or in a lower job zone.

Hypothesis: Promotion in the first period is associated with holding a job in the highest job zones at the end of a career.

To test our final hypothesis, we group veterans by those who were promoted in the first period (promoted after two years) and those who held a job in job zone four or five at the end of the 10-year period. If this association exists, it could be evidence that an early promotion results in a higher career “ceiling”; that is, higher potential outcomes for “winners” of the first round of the career tournament.

## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  tblE
## X-squared = 3.0829, df = 1, p-value = 0.07912

The results of the chi-squared test (above) show a weak statistical association (p < 0.1, but not < 0.05) between early promotion and a high final job zone. This may again be because our sample of veterans in the DC MSA tends to start out highly trained and educated, so there is not much additional mobility possible once a promotion occurs.

Policy Implications

In order to best improve veteran career mobility and outcomes, the first two years after the military exit are crucial. While early promotion does not have as strong of an association with positive outcomes, early unemployment has very strong associations with negative outcomes, such as additional unemployment or a lower career ceiling. Policy efforts that target “new” veterans should consider the strong effects of transitional unemployment on later career options and mitigate this unemployment where possible.