Antagonistic managers, careless workers and
extraverted salespeople: An examination of personality
in occupational choice
By Roger Hama, P.N Junankarab and Robert Wellsa*
Revised February 2011
a
School of Economics and Finance, University of Western Sydney. Locked Bag 1797
Penrith South DC NSW 1797 , Australia
b
School of Economics, University of New South Wales and IZA. Bonn, Germany
*
Corresponding Author. Email: R.Wells@uws.edu.au, Phone: +61 405 577 717 Fax: + 61 2 9685 9105
1
Antagonistic managers, careless labourers and extraverted salespeople:
An examination of personality in occupational choice
Abstract
This paper is an econometric investigation of the choice of individuals between a number
of occupation groupings utilising an extensive array of conditioning variables measuring
a variety of aspects of individual heterogeneity. Whilst the model contains the main
theory of occupational choice, human capital theory, it also tests dynasty hysteresis
through parental status variables. The focus is an examination of the relationship between
choice and personality with the inclusion of psychometrically derived personality
variables. The empirical model of occupational choice is a multinomial logit estimated
using the Household Income and Labour Dynamics in Australia (HILDA) survey data.
Human capital variables are found to exhibit strong credentialism effects. Parental status
has a small and limited effect on occupation outcomes indicative of only some small
dynasty hysteresis. On the other hand, personality effects are found to be significant,
relatively large and persistent across all occupations. Further, the strength of these
personality effects are such that they can in many instances rival that of various education
credentials. These personality effects include but are not limited to: managers being less
agreeable and more antagonistic; labourers being less conscientiousness; and sales people
being more extraverted.
JEL Classification: J24; J62; C25
Keywords: Occupational choice; personality traits; credentialism; dynasty hysteresis
2
Introduction1
Heckman and Rubinstein (2001) explicitly refer to the difficulty of analysing the
implications for labour market outcomes of non-cognitive skills, because of measurement
problems, “[M]any different personality and motivational traits are lumped into the
category of non-cognitive skills” (p. 145) and liken non-cognitive skills to the “dark
matter” of astrophysics. Nevertheless they cogently argue that “success in life” is
determined by non-cognitive traits as well as cognitive skills, but point out that:
“too little is understood about ...the separate effects of all these diverse
traits currently subsumed under the rubric of noncognitive skills. What
we currently know ... suggests that further research on the topic is likely
to be very fruitful.” Heckman and Rubinstein (2001, p. 149)
This paper uses well recognised measures of psychological traits, from the “Five Factor
Model” (FFM) of empirical psychology. These traits are contained within the category
non-cognitive skills to examine the effects of some psychological characteristics on
occupational outcomes.
1
This paper uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA)
Survey. The HILDA Project was initiated and is funded by the Australian Government Department of
Families, Housing, Community Services and Indigenous Affairs (FaHCSIA) and is managed by the
Melbourne Institute of Applied Economic and Social Research (MIAESR). The findings and views
reported in this paper, however, are those of the authors and should not be attributed to either FaHCSIA or
the MIAESR We are grateful for very helpful comments provided by anonymous referees of this journal.
The data used in this paper was extracted using the Add-On package PanelWhiz for Stata®. PanelWhiz
(http://www.PanelWhiz.eu) written by Dr. John P. Haisken-DeNew (john@PanelWhiz.eu). See HaiskenDeNew and Hahn (2006) for details. The PanelWhiz generated DO file to retrieve the data used here is
available from me the authors on request. Any data or computational errors in the paper are due to the
authors.
3
In economics it is often thought that differences between individual characteristics, such
as preferences, are far too numerous, and are not a sensible area, for the scientific
examination of economic behaviour (Caplan 2003). This is not the case in psychology
which has long sought to find and classify the differences between individuals’ consistent
and enduring behavioural characteristics; these are known as personality traits. The focus
of this paper is an examination of the influence of personality traits on occupational
choice by way of a standard multinomial logit model. The categorical variable,
occupation as measured by the Australian and New Zealand Standard Classification of
Occupations (ANZSCO) is used as the indicator of occupational outcomes. The eight
single digit occupations are used in this study. Whilst this may seem a high level of
aggregation it provides a larger number of categories relative to other studies. Further, the
use of two digit categories leads to computational problems, through narrower cell widths
with subsequent reduced variation leading to identification issues.
Whilst the focus of the paper is the conditioning of occupational outcomes by personality
measures, other sets of important conditioning variables are included in the model. In
terms of human capital, following Heckman et al. (2003) and Leigh (2008) the model
uses a series of binary variables associated with educational qualifications and
credentials. Parents’ characteristics by way of their occupational achievements and the
socio-economic index, AUSEI06, developed at the Australian National University
(McMillan et al., 2009) are included to capture dynastic hysteresis, The mechanism
underlying dynasty hysteresis is subject to debate. Laband and Lentz (1983) argue that
dynasty hysteresis is caused by human capital transfer which is more predominant in
some occupations than others. Fan (2008) argues that dynasty hysteresis may be
transmitted by religion and its associated characteristics. Akerlof (1997) puts forward a
theory of social distance in which individuals have both a desire to excel and a desire to
conform to their social group. Other potential mechanisms include the intergenerational
transfer of preferences (Doepke & Zilibotti 2005) and of non-cognitive factors (Bowles &
Gintis 2002). Irrespective of the transfer mechanism, dynasty is an important
phenomenon to be examined in occupational choice as it has a crucial effect on the ability
of an individual to have freedom to pursue any occupation they desire and the extent to
4
which the other determinants influence that choice (Bradley 1991; Mazumder 2005;
Bjerk 2007). Previous analyses of occupational choice have included parental variables to
attempt to control for dynasty hysteresis (Robertson & Symons 1990; Bradley 1991;
Tsukahara 2007). Further controls are also used in terms of demographic; personal;
industry; and location variables.
Whilst the psychological characteristics used in this empirical investigation of
occupational outcomes is only a subset of all those traits encompassed in non cognitive
traits, this investigation will help shed some light on Heckman and Rubinstein’s “dark
matter” of labour market outcomes. The next section deals at length with the personality
measures used to examine occupation outcomes and psychological traits.
Psychometrically Derived Personality Traits and Occupational Choice
McCrae and Costa (2003) provide a summary of the literature on personality psychology
to date and argue, along with others (Digman 1990; Goldberg 1993; Caplan 2003;
Terracciano et al. 2006; Cole 2007; Borghans et al. 2008a; Costa Jr & McCrae 2008;
Terracciano et al. 2010), that the general consensus model for personality traits is that of
the five factor model (FFM). The FFM states that personality traits are “dimensions of
individual differences in tendencies to show consistent patterns of thoughts, feeling, and
actions” (McCrae & Costa 2003, p. 25). These personality trait dimensions are typically
viewed as broad level dispositions or propensities, that is they are not the sole
determinant of behaviour but should be viewed as a contributing factor in a ceteris
paribus context. Caplan (2003) emphasizes that, whilst personality traits are an important
aspect of the examination of economic activity, the incentives of economics should not be
neglected. Personality traits consist of broad dispositional traits that can influence
behaviour, this does not mean that individuals with the same level of personality traits are
identical as the traits can manifest themselves in different specific mannerisms, or
characteristics adaptations, as stated by McCrae and Costa (2003). It should be noted that
these personality traits consist of subscales representing more specific behavioural
5
tendencies and further these subscales are in turn formed by facets which represent
increasingly more specific behavioural tendencies. Both facets and subscales covary
amongst each other and consequently form subscales and factors respectively, with
factors representing personality traits at the broadest and most consistent level. Subscales
might well have stronger influences on occupational outcomes. However, they are not
available in the Household Income and Labour Dynamics in Australia (HILDA) panel
used here, Further even if they were available, there are potential problems with their use.
Because of the high covariance between subscales within a factor there is a strong
potential for collinearity. Further, even though the occupational categorisation used here
is coarse, there are still a significant number of different occupations. Increasing the
number of personality measures may well lead to a low variation in some categorical
cells and identification problems.
These five traits consist of Openness to experience, Conscientiousness, Extraversion,
Agreeableness and Neuroticism2. Each of these traits have a corresponding negativeclosed to experience, carelessness, introversion, antagonism and emotional stabilitywhich correspond to low scores of the corresponding dimensions. It should be noted that
here emotional stability is used in the place of neuroticism to preserve the original
arrangement of the data. Openness to experience can be defined as a trait associated with
being accepting of new ideas and alternative points of view, appreciation of new
concepts, imaginative and creative and generally inquisitive and curious.
Conscientiousness is the trait that is associated with diligence, self discipline, punctuality,
organised and general competence3. Extraversion is the trait associated with being
talkative, friendly, energetic and outgoing. This trait has been previously examined in
economics by Krueger and Schkade (2008) and found to influence occupational outcomes
and wages. Agreeableness can be described as the tendency to be generous, warm,
altruistic, tender, and complaisant and tend to get along with others. Individuals with a
lack of agreeableness can conversely be aggressive, tough and quite adversarial.
2
3
This can be remembered with the mnemonic device OCEAN which corresponds to the order stated above.
This trait in a precursor to the FFM is referred to as ‘work’ (Goldberg,1993)
6
Neuroticism is the last trait in the FFM, it is often commonly referred to by its negative of
emotional stability. In this paper both terms are used interchangeably depending on
contextual appropriateness. However, the estimated empirical model uses the positive
counterpart, emotional stability, as the conditioning variable. Neuroticism can be
described as the tendency of experiencing negative emotions more frequently and
intensely. Neuroticism can be described as a trait associated with anxiety, worry, paranoia
and stress. Traits associated with these dimensions of personality can been seen in Table
1 from McCrae and Costa (2003).
7
Table 1: Characteristics of higher and lower scores in the personality traits of the
Five Factor Model
Personality trait
Low Scorer
Higher scorer
Openness
Favours conservative values
Values intellectual matters
Judges in conventional terms
Rebellious, nonconforming
Uncomfortable with complexities
Unusual thought processes
Moralistic
Introspective
Eroticizes situations
Behaves ethically
Unable to delay gratification
Dependable, responsible
Self-indulgent
Productive
Engages in fantasy, daydreams
Has high aspiration levels
Emotionally bland
Talkative
Avoids close relationship
Gregarious
Overcontrol of impulses
Socially poised
Submissive
Behaves assertively
Critical, sceptical
Sympathetic, considerate
Shows condescending behaviour
Warm, Compassionate
Tries to push limits
Arouses liking
Expresses hostility directly
Behaves in a giving away
Calm, relaxed
Thin-skinned
Satisfied with self
Basically anxious
Clear-cut personality
Irritable
Prides self on objectivity
Guilt-Prone
Conscientiousness
Extraversion
Agreeableness
Neuroticism
Source: McCrae and Costa, 2003
In psychology, research suggests that personality traits influence occupational outcomes
through choices made by individuals and the requirements of employers in occupations
(Barrick & Mount 1991; Larson et al. 2002; Barrick et al. 2003; Ozer & Benet-Martinez
2006; Furnham & Fudge 2008). Barrick and Mount (1991) put forward a number of
hypotheses in their meta-analysis of the effect of the FFM in regard to how personality
variables influence an individual’s productivity and consequently their occupational
achievement. Conscientiousness is argued to carry a ubiquitous positive effect on labour
market outcomes as individuals who possess this trait are often hardworking, productive,
punctual, organised and accepting of responsibility. It can be argued that openness has an
effect on the ability of individuals to be trained, being embracing of new ideas and
8
consequently can be positively valued. Emotional stability is argued to influence an
individual’s occupational achievement via the individual being less likely to experience
negative emotions and they can take on more stressful, non-routine and risky tasks.
Barrick and Mount (1991) posit that agreeableness may be valued in some occupations as
friendliness is desirable for interpersonal interaction; however, it has been argued that
antagonistic personality may be required for certain tasks by both the supplier of and
demander for labour (McCrae & Costa 2003; Borghans et al. 2008b). In addition, it has
been argued that a tendency to Machiavellian behaviour, that is a desire to manipulate
people for benefit, is positively valued in some labour markets (Bowles et al. 2001a;
Bowles et al. 2001b; Wakefield 2008). Extraversion may be valued in occupations that
involve a large amount of interaction with others, as extraverted individuals would gain
greater utility from these interactions (Barrick & Mount 1991; Krueger & Schkade 2008).
The results of Barrick and Mount’s (1991) meta-analysis show that conscientiousness and
openness behave as predicted, extraversion is valued, in terms of job productivity, in both
social jobs and training while agreeableness and neuroticism are observed to have no
effect on labour market outcomes. They argue that the lack of an observed effect with
regard to neuroticism may be due to a sample selection bias as you require a minimum
amount of emotional stability -the negative of neuroticism- to achieve a position in the
labour market.
There has been a neglect of the analysis of non cognitive skills on life outcomes in
economics. The limited literature sees certain non cognitive traits as being valued by
employers, Heckman and Rubinstein (2001) and Heckman et al (2006). Bowles et al.
(2001a) argue that non cognitive traits as well as cognitive skills are rewarded on the
labour market. Nyhus and Pons (2005) estimate earnings functions incorporating the
FFM personality measures. Borghans et al. (2008a) argue that personality traits, as
measured by the FFM condition many economic outcomes for individuals, including
occupation. Moreover, they discuss at length how best to incorporate personality into
conventional economic models. It could be argued that personality traits influence
individuals’ preferences towards various different occupations, that is, they are supply
side forces. Alternatively, personality traits will be valued by employers and therefore
9
they may be modelled as goods demanded by employers and will appear on the supply
side as a constraint to occupational choice. The empirical model that is proposed here is
not structural, recognising that outcomes are determined by supply and demand forces
and as such it is a reduced form which simply assesses how personality acts to sort
individuals into occupations.
The stability of personality traits is an important issue in the literature. McCrae and Costa
(2003) present a comprehensive review of a large body of work which suggests that from
the age of thirty personality, as defined above, is stable within the broad definition of the
FFM and some stability can even be seen in younger individuals. This study includes a
review of longitudinal research (pp. 108-115) which provide strong evidence for stability
citing a study reporting 3 year retest correlations for both men and woman in two groups
aged 25-56 and 57-84 ranging from 0.55 to 0.87. Alternatively, Roberts and DelVecchio
(2000) in an influential meta analysis of a wide mixture of studies find stability peaks at
50 to 70 years of age. These studies are based on group retest correlations and not
individual retest data and in their conclusion the authors state that it would be
inappropriate to make inferences about individual consistency from their results: “it
would be premature to render a final judgement concerning whether and when
personality traits are fixed” (Roberts and DelVecchio, 2000, p. 19). Further research
supports the claim for earlier stability (Terracciano et al. 2006; Terracciano et al. 2010),
where the latter uses individual rather than grouped data. Even in young adults some
retest correlations are as high as 0.3. Finally, Viinikainen et al. (2010), using a Finnish
data set, find correlations between childhood personality scores and adult labour market
outcomes 35 years later. Whilst the personality scores pre-date the FFM the personality
measures considerably overlap the FFM and these correlations are indicative of long run
stability in broad personality factors. Most studies find that stability decreases with time
between retest scores, that is the longer the period between retest, the lower are retest
correlations.
Personality traits, like cognitive skills are determined by “nature and nurture”, Cuhna and
Heckman (2008). They are not immutably fixed and can change and can be changed by
10
an individual’s occupation (Mueller & Plug 2006), (Groves 2005), (Heckman et al.
2006), (Cole 2007), (Semykina & Linz 2007), (Cunha & Heckman 2008). Yet, this
potential for the endogeneity of personality traits, measured by the FFM, in modelling
occupational outcomes is not necessarily large.
“[c]hange may be more difficult later in the life cycle, change may be more
enduring for some (such as more emotionally stable individuals) than for
others, change may require persistent and consistent environmental pressure
(as opposed to transient pressure from short-term interventions), and there
are powerful forces for stability (such as genes and habit) which make
change difficult.” (Borghans et al., 2008, p. 1021)
Whilst occupation might influence traits, that influence is one small part of adulthood
environmental influences which may themselves be marginal relative to the core
influences of environment in childhood and early adulthood and genetic endowment.
Further, occupation may not necessarily be seen as some persistent environmental force
in the HILDA sample. Over the seven annual waves used in this analysis, only 54% of
the sample recorded as “Managers” in Wave one were recorded as Managers in Wave
seven, Whilst the percentages for “Professional, “Technician”, “service”, “Clerical” and
“Operator” were somewhat higher at 75%, 62%, 63%, 63%, and 61% respectively, these
are still indicative of substantial movement between occupations. “Sales” and “Labourer”
experienced higher turnover with only 37% and 40% retention over the seven Waves.
Given the marginal nature of occupation determining change in the FFM and the fact that
there is considerable movement amongst occupations strong reverse causality is highly
unlikely.
The Data
The data source is the Household Income and Labour Dynamics in Australia (HILDA)
longitudinal data set. The HILDA survey is an approximately one in one thousand sample
11
of the Australian population consisting of 19,914 individuals in 7,682 households in
sample Wave one (Watson 2009). The HILDA dataset contains an extremely rich set of
variables which capture details on a large number of individual characteristics including
the focus of this study occupational and labour market outcomes, education, parental
status, and standard demographic variables. Particularly, it contains a comprehensive set
of personality measures which conform with the FFM. These were derived, following the
well established procedures of the FFM, using a factor analysis of underlying variables.
The sample used to estimate the model consists of pooled data from Waves 1 to Wave 7
(years 2001 to 2007) of the HILDA panel. Individuals were included in the sample if they
had a complete record of all the variables included in the model for any wave.
Occupational status is determined by recorded primary occupation; as such full time
students are excluded from the sample.
Occupations in the HILDA data are coded into the ANZSCO system. The HILDA data in
its general release provide the ANZSCO coding at both the one digit and two digit levels.
This research uses the one digit ANZSCO coding which consists of eight mutually
exclusive and exhaustive occupation outcomes. The eight one digit categories are
Managers, Professionals, Technicians and tradespersons, Community and personal
service workers, Clerical and administrative workers, Sales workers, machine operators
and drivers, and Labourers. Each of these categories describes a set of skill
specialisations that are relatively homogeneous when compared with other groups. These
categories give a good representation of occupations based on the view that an occupation
is a set of relatively homogenous tasks. A brief description of each of these eight
occupational categories and examples of jobs that fall into these categories can be seen in
Table 2.
12
Table 2: Definitions and examples of the ANZSCO coding of occupations.
Occupation (abbreviations underlined)
Description of tasks
Examples
Managers
Plan, organize, coordinate and review various
General manager, legislators, farm manager, finance
operations
managers, retail manager, & customer service manager
analytical, conceptual and creative tasks
Actors, airline pilots, Engineers, Physical and social
require the application of a body of
Scientists, Medical professionals, lawyers, IT professionals,
knowledge
and educators
Skilled tasks requiring broad or specific
Scientific Technicians, motor mechanics, construction
knowledge.
workers, chefs, florists and hairdressers
Provision of service to either individuals
Paramedics, child carers, baristas, waiters, security officer,
personally or the community as a whole that
military personnel, driving instructors and sportspersons.
Professionals
Technicians and tradespersons
Community and personal service workers
often requires interaction with others.
Clerical and administrative workers
Sales workers
Machinery operators and drivers
Labourers
Organize, store, manipulate and retrieve
Office managers, data entry clerks, receptionists, payroll
information
clerk, mail clerks and proofreader
Sell goods and services and provide sales
Sales representative, insurance brokers, retail supervisors,
support
checkout operator, models and telemarketers,
Operate machinery, plant vehicles and other
Industrial spraypainter, sewing machinist, motion picture
equipment
projectionist, crane operator, forklift driver, and train driver
Repetitive and routine tasks that may include
Cleaners, steel fixer, product assembler, packer, slaughter,
the use of hand or power tools.
farm worker, kitchen hand, freight handler and handypersons
While these groups are relatively homogeneous in terms of tasks, some occupations can
be expected to have differing desired characteristics; for example, security workers are in
the same category as baristas. Analysis of the two digit level of 43 categories would
capture some of the heterogeneity in the broader class. However, analysis at the two digit
level leads to difficulties given: the necessary number of parameters; a low frequency of
observations in particular states; and the possible violation of necessary conditions for the
model’s adequacy. For these reasons the current work is limited to occupational
categories at the one digit level.
As well as personality trait measures, the HILDA data give very detailed listings of
variables that theory elects as important in conditioning occupational outcomes. Along
with the standard demographic variables such as age, gender, location, marital status, etc.,
the HILDA survey data set contains data on the educational attainment of an individual in
terms of qualifications. In order to capture occupation specific human capital and to
13
introduce non-linearity into the influence of education on labour market outcomes, the
analysis uses a series of binary variables based on educational achievement.
Two sets of variables are used to capture parental status. The first is the labour market
success of an individual’s parents. This is measured by a set of binary variables which
reflect the occupation of an individual’s mother and father. Of particular interest is the
probability of an individual being in the same occupation as their parent, this would
provide support for the phenomenon of dynasty hysteresis. The second is the AUSEI06
measure of social status. This is an index ranging between zero and one hundred, to one
decimal place, which incorporates a variety of education, occupation, income and other
demographic effects that reflect an individual’s social success (Jones & McMillan 2001;
McMillan et al. 2008). The AUSEI06 allows for a continuous version of parental status to
be used in order to determine if it influences an individual’s occupational outcome.
Prior to presenting tables of measures and summary statistics which describe the data, it
would be useful to return to the personality data. The HILDA data set provides a rich
array of variables that are rare within nationally representative data sets to examine the
“dark matter” of the economics of personality factors (Heckman & Rubinstein 2001). The
survey administered a questionnaire based on the FFM in Wave five. The test takes 30
questions consisting of adjectives that describe the respondent’s typical behaviour from
the mini-marker test developed by Saucier (1994) and 6 questions from other valid
personality tests. These personality tests are usually subjected to standard tests in order to
ensure that they are valid psychometric instruments and can be used for meaningful
analysis of unobserved psychological phenomena and are not purely mathematical
constructs (Borghans et al. 2008a). The personality tests in HILDA are no exception and
Losoncz (2007) provides an assessment of validity tests of the personality tests and finds
that the derived psychometric instruments are valid.
14
Table 3: The Frequency of Occupations
Occupation
Frequency Percentage
Managers
Professionals
Technicians and Trades Workers
Community and Personal Service Work
Clerical and Administrative Workers
Sales Workers
Machinery Operators and Drivers
Labourers
4,211
8,063
3,749
2,641
4,883
1,911
1,560
2,216
14.40
27.58
12.82
9.03
16.70
6.54
5.34
7.58
Table 4: Summary Statistics
Variable
Mean
S.D.
Age
Age
40.45 11.69
Age Squared
1773.32 977.79
Personality
Agreeableness
5.3722 0.8792
Openness
4.3131 1.0197
Conscientiousness
5.1541 1.0052
Extraversion
4.4732 1.0840
Emotional Stability 5.1739 1.0885
Parental Status
Father’s AUSEI06 45.7487 22.718
Mother AUSEI06
42.2437 22.908
Min. Max.
15
225
893
6889
1
1
1
1
1
7
7
7
7
7
0
3.4
100
100
Table 3 gives the frequency and relative frequency counts for occupational status
categories used in the multinomial logit. The preponderance of professionals and clerical
and administrative workers reflects the distribution of occupations in the population and
mirrors outcomes of other Australian surveys. Table 4 contains summary statistics for
those variables where these measures are meaningful. Table 5 summarises the binary
variables used in the analysis. For each binary the score is the proportion of the sample
scoring one.
15
Table 5: Binary Variables
Variable
Proportion Variable
Gender (Base: male)
Female
Education (Base: year
12)
PhD or master
Graduate diploma
Bachelor
Advance diploma
Certificate 3 or 4
Certificate 1 or 2
Year 11 or Less
State (Base: New S
Wales)
Proportion Variable
Father’s occupation (Base:
Professional)
Year (Base: 2001)
Proportion
0.4946 2002
0.1315 Father Manager
0.2536
2003
2004
2005
2006
2007
Country of Origin (Base
0.2139 Australia)
English Speaking country of
0.0117 origin
0.2137 Non English country of origin
0.1371
0.1431
0.1602
0.1521
0.1478
0.2444
0.0346
0.0623
0.0480
0.1017
0.0494
0.0782
0.1787
0.1096
Marital Status (Base: single)
Father Technician
Father Service
Father Clerical
Father Sales
Father Operator
Father labourer
Mother’s occupation (Base:
0.1152 Professional)
0.0737 Mother Manager
0.0915
Mother Technician
0.0794
0.0993
Victoria
0.2507 Couple
0.5580 Mother Service
0.0809
Queensland
0.2084 Post Couple
0.1416 Mother Clerical
0.2311
South Australia
0.0846
Mother Sales
0.1206
Western Australia
0.0968
Mother Operator
0.0347
Tasmania
0.0317
mother labourer
0.1713
Northern Territory
Australian Capital
Territory
0.0074
0.0239
Estimation and Results
A multinomial logit model was estimated using STATA 10. The estimate was normalized
on the modal occupation, Professional. Using the test proposed by Long and Freese
(2006) in their SPOST suite of STATA commands (not reported), which is a test based
on the Cramer-Ridder test (Cramer & Ridder 1991), it can be concluded that none of the
occupational states can be pooled together, suggesting that no further aggregation is
possible without biasing the results. Based on this, one should not expect violations of the
independence of irrelevant alternatives (IIA) assumption .
Rather than report the rather large numbers of estimated coefficients the marginal effects
alone are reported. The analysis used the marginal effects estimated by the MARGEFF
module (Bartus 2005). Further, these are the mean marginal effects for the sample data
rather than the marginal effects at means. For brevity, the average marginal effects for the
personality traits, the focus of this study, are presented in Table 6 in the text. All other
16
average marginal effects are reported and discussed in separate tables in the Appendix. In
order to assess the relative impacts of discrete and continuous variables the paper
compares the average effect of a change in a discrete variable from its minimum to
maximum value to the same measure for a continuous variable4.
4
A comment on an earlier version of this paper suggested that comparisons of magnitudes in shift should
be on the basis of a standard deviation shift rather than minimum to maximum. This is problematical given
that a comparison may be personality against education credentials, where the latter are binary and
therefore discrete variables. The standard deviation is not a valid statistic for nomial variables (Stevens
1946)-and would produce poor estimates of the relative magnitudes of these effects especially in non-linear
models such as the multinomial logit.
17
Table 6: Average Marginal Effects Personality Traits for base model
Variable (Base)
Openness
Conscientiousness
Extraversion
Agreeableness
Emotional Stability
Female Specific
Openness
Conscientiousness
Extraversion
Agreeableness
Emotional Stability
Openness
Conscientiousness
Extraversion
Agreeableness
Emotional Stability
Female Specific
Openness
Conscientiousness
Extraversion
Agreeableness
Emotional Stability
Observations
Legend
Managers
0.0003
0.0177***
0.01128**
-0.0142**
0.0025
Professionals
0.0279***
-0.0043
-0.0008
-0.0115
0.0084
Technician
0.0035
0.0034
0.0057
-0.0027
-0.0005
Service
-0.0170**
-0.0004
0.0020
0.0121
0.0089
0.01485
-0.0100
0.0118
-0.0100
-0.0039
Clerical
-0.0003
0.0055
-0.0312***
0.0074
-0.0117
-0.0322***
0.0066
-0.0031
0.0034***
0.0024
Sales
-0.0044
-0.0089**
0.0115***
0.0038
-0.0025
0.0058
-0.0035
-0.0058
-0.0066
-0.0093
Operator
-0.0047
-0.0024
0.0013
0.004508
-0.0013
0.0143*
-0.0026
-0.0061
--0.0089
-0.0081
Labourer
-0.0051
-0.0107***
0.0001
0.0015
-0.0037
-0.0129
0.0089
0.0231***
0.0004
0.0031
29,234
* p<0.1
0.0054
0.0021
-0.0045
-0.0042
0.0012
0.0058
-0.0066
-0.0099
-0.0047
0.0102
Pseudo R2
*** p<0.01
-0.0015
0.0050
-0.0055
0.0005
0.0045
0.2444
** p<0.05
Personality traits, discussion: Table 6 indicates that personality traits generally seem to
have some influence on occupational outcomes with the average marginal effects
frequently being statistically significant for each trait and occupation.
Openness to experience, a trait related to receptiveness to training and accepting of new
and different ideas, significantly increases the probability of individuals being found in a
professional role but for males with the female specific effect perfectly negating the base
effect with the sum of effect failing to be significantly different from zero (p=0.48). This
is to be expected as both require individuals to deal with a variety of ideas, be accepting
18
of new ideas and concepts and require training, as can be seen from the definition of
professionals Table 1. The trait openness also tends to reduce the probability of a male
being a service worker with the female specific effects cancelling both effects (p=0.52).
The tasks of a service worker, as seen in Table 1, tends to be relatively more routine, thus
openness would not be highly valued by employers. It is possible that individuals who
tend not to be receptive to new ideas may choose jobs within these occupations as it
allows for a narrow specialisation. The effect of openness is relatively large and
important. For example, for a professional moving from the lowest possible score, one, to
the highest, seven, gives an increase of 0.17, which is slightly larger than the average
effect of completion of high school on the probability of being in a professional
occupation; see Table A1 in the Appendix.
Conscientiousness, the trait associated with hard work and effort, is found to significantly
increase the probability of an individual, both male and female, being in a management
position and decrease the probability of an individual being a salesperson or labourer.
One aspect of the personality trait conscientiousness is the ability to plan and be
organised, an aspect that is central to the tasks performed in management roles. Thus
despite the trait being considered valuable in all labour market outcomes, as in Roy’s
(1951) model, people tend towards an occupation where it is most valued and are drawn
away from those of salesperson, and labourers. The magnitude of the effects of
conscientiousness are quite sizeable: moving from the lowest level to the highest level of
this trait increases the probability of being in a management occupation by 0.097 which
has a greater effect than any education credential, while close to the influence of a PhD or
Masters, influence on the probability of being a manager. See Table A1 in the Appendix
for details of these scores. These results highlight the potential for personality factors to
matter more than traditional human capital variables.
Extraversion, the trait associated with being outgoing and desiring to engage with other
individuals is found to significantly affect some occupational outcomes. Individuals who
are observed with higher levels of extraversion have a higher probability of being in
management, with the effect larger for females, and sales while a lower probability of
19
being in a clerical occupation. These results make sense in that sales roles necessitate
social interactions. In addition, management entails, as can be seen in Table 1, the
organising of resources to complete tasks such as human labour and thus requires social
interaction. Conversely, clerical tasks are generally not focused on interacting with others
and may even inhibit social interaction thus individuals who prefer less social interaction
would tend to select these occupations. The effects of extraversion are also quite sizeable
with for example, the maximum effects for clerical workers, that is the effect of moving
from the lowest to the highest value of the personality trait, being larger in magnitude (0.19) than the effects of any education credentials.
Agreeableness has a negative effect on the probability of an individual being a manager
and a positive effect on the probability of females being professionals. As suggested
earlier, certain occupations would tend to favour individuals who are less concerned with
pleasing others. This includes occupations in which there is a greater focus on task
completion and competition rather than interacting with others such as management
positions. Conversely, females with higher levels of agreeableness have a higher
probability of being in the professions. The fact that this effect applies only to females
suggests there might be gender differences in the distribution of females in professional
occupations in that females select professions that require greater interpersonal actions.
Emotional stability, the negative of neuroticism, is a trait associated with being less likely
to experience negative emotions, and unlike all the other personality traits of the five
factor model, has no significant effect on occupational choice. The lack of a significant
effect simply suggests that this effect is relative unimportant in influence the sorting of
individual across occupations. Neuroticism may be associated with risk aversion and
therefore the sorting of occupations here might reflect risk preferences5. DeLeire and
Levy (2004) show that attitude towards risk affects occupational choice.
Personality significantly influences the probability of an individual choosing or being
chosen for a particular occupation, with each trait except emotional stability influencing
5
We are grateful to an anonymous referee for pointing out this issue.
20
at least one occupational outcome. Generally these effects are found to be modest in
comparison to that of human capital but they can rival educational credentials in certain
occupations. Due to the strength and magnitude of personality effects, it can be argued
that they are relatively more important than parental status; with the parental status
effects tending to be similar if not smaller in magnitude and less persistent. It is possible
that the findings with regards to personality are suppressed as personality and parental
status may also influence education and thus the indirect effect of personality is not
captured by the current model. To test for this, a model was estimated that placed
restrictions on the parameters associated with parental status to zero in order to see if the
effects changed; the parameters were virtually identical across both models.
In order to determine the robustness6 of the influence of personality on occupation
outcomes, a number of restrictions were made to the parameters of the model to see if
they altered the parameter values of the personality variables. Restricting parental status
parameters to zero caused little or no change in the parameter values including those
associated with personality. The restriction of education parameters to zero changed the
personality estimated parameter values but did not change their significance. In addition,
in order to examine if the age of an individual altered the effects of personality due to the
possible instability in personality traits, two models were estimated for those under thirty
and those thirty and over. Most parameters estimated for the sample restricted to less than
thirty years were not significant. The under thirties accounted for less than 25% of the
whole sample and it appears that the over thirties dominate the complete sample.
Because personality traits are the source of novelty in this research, discussion of the
results for other influences on occupational are left to the Appendix. However, in
summary, it can be found that labour market heterogeneity is important as individuals are
sorted between occupations based on the various other different characteristics nominated
by theory and incorporated in the multinomial logit model. The results broadly indicate
that education exhibits a non-linear effect in years, with arguably occupation specific
6
We would like to thank an anonymous reviewer for the suggestion of these robustness
checks. Detailed results are available on request.
21
education effects and that parental status has a small and limited effect on occupational
outcomes.
Conclusion
This paper examines the effect of human capital, parental status and personality on
occupational outcomes for a representative panel of Australian households. Economic
theory argues that individuals should select the occupation that grants them the highest
utility and that because both individuals and tasks in the labour markets are quite
heterogeneous individuals are sorted into occupations. Using a multinomial logit model,
it is found that human capital exhibits a non-linear effect on occupational attainment,
parental status has a limited influence and the broad personality traits of the highly
validated five factor model have a significant, relatively strong, persistent and expected
effect over occupational outcomes.
To conclude, we found in our exploration of the “dark matter” in economics a significant
role for personality in influencing occupational outcomes. We found that there were
significant gender differences in certain occupations for particular personality traits.
References
Akerlof, G. A. Social distance and social decisions. Econometrica 1997. 65; 5; 10051027.
Barrick, M. R. & Mount, M. K. The big five personality dimensions and job
performance: a meta-analysis. Personnel Psychology 1991. 44; 1; 1-26.
Barrick, M. R., Mount, M. K. & Gupta, R. Meta-analysis of the relationship
between the five-factor model of personality and Holland's occupational
types. Personnel Psychology 2003. 56; 1; 45-74.
22
Bartus, T. Estimation of marginal effects using margeff. The Stata Journal 2005. 5;
3; 309-329.
Bjerk, D. The differing nature of black-white wage inequality across occupational
sectors. Journal of Human Resources 2007. 42; 2; 398-434.
Borghans, L., Duckworth, A. L., Heckman, J. J. & ter Weel, B. The Economics and
Psychology of Personality Traits. Journal of Human Resources 2008a. 43; 4;
972-1059.
Borghans, L., ter Weel, B. & Weinberg, B. A. Interpersonal Styles and Labor
Market Outcomes. Journal of Human Resources 2008b. 43; 4; 815-858.
Bowles, S. & Gintis, H. The inheritance of inequality. Journal of Economic
Perspectives 2002. 16; 3; 3-30.
Bowles, S., Gintis, H. & Osborne, M. The determinants of earnings: a behavioral
approach. Journal of Economic Literature 2001a. 39; 4; 1137.
Bowles, S., Gintis, H. & Osborne, M. Incentive-enhancing preferences: personality,
behavior, and earnings. American Economic Review 2001b. 91; 2; 155-158.
Bradley, S. An empirical analysis of occupational expectations. Applied Economics
1991. 23; 7; 1159.
Caplan, B. Stigler-Becker versus Myers-Briggs: why preference-based explanations
are scientifically meaningful and empirically important. Journal of Economic
Behavior & Organization 2003. 50; 4; 391-405.
Cole, K. Good for the soul: the relationship between work, wellbeing and
psychological capital, PhD in Economics. University of Canberra: 2007,
Costa Jr, P. T. & McCrae, R. R. 2008. The NEO inventories In: Personality
Assessment. R. P. Archer and S. R. Smith (Eds.), Personality Assessment.
Routledge: New York; 2008. p.
Cramer, J. S. & Ridder, G. Pooling states in the multinomial logit model. Journal of
Econometrics 1991. 47; 2-3; 267-272.
Cunha, F. & Heckman, J. J. Formulating, Identifying and Estimating the
Technology of Cognitive and Noncognitive Skill Formation. Journal of
Human Resources 2008. 43; 4; 738-782.
23
DeLeire, T. & Levy, H. Worker sorting and the risk of death on the job. Journal of
Labor Economics 2004. 22; 4; 925-953.
Digman, J. M. Personality structure: Emergence of the five-factor model. Annual
Review of Psychology 1990. 41; 1; 417.
Doepke, M. & Zilibotti, F. Social class and the spirit of capitalism. Journal of the
European Economic Association 2005. 3; 2/3; 516-524.
Fan, C. S. Religious participation and children's education: a social capital
approach. Journal of Economic Behavior & Organisation 2008. 65; 303-317.
Furnham, A. & Fudge, C. The five factor model of personality and sales
performance. Journal of Individual Differences 2008. 29; 1; 11-16.
Goldberg, L. R. The structure of phenotypic personality traits. American
Psychologist 1993. 48; 1; 26-34.
Groves, M. O. How important is your personality? Labor market returns to
personality for women in the US and UK. Journal of Economic Psychology
2005. 26; 6; 827-841.
Heckman, J. J., Lochner, L. & Todd, P. E. Fifty years of Mincer earnings
regressions. Cambridge, Massachusetts; 2003
Heckman, J. J. & Rubinstein, Y. The importance of noncognitive skills: lessons from
the GED testing program. American Economic Review 2001. 91; 2; 145-149.
Heckman, J. J., Stixrud, J. & Urzua, S. The effects of cognitive and noncognitive
abilities on labor market outcomes and social behavior. Journal of Labor
Economics 2006. 24; 3; 411.
Jones, F. L. & McMillan, J. Scoring occupational categories for social research: A
review of current practice, with Australian examples. Work Employment
Society 2001. 15; 3; 539-563.
Knight, J.B. Job Competition, Occupational Production Functions, and Filtering
Down, Oxford Economic Papers 1979. 31; 2; 187-204.
Krueger, A. B. & Schkade, D. Sorting in the Labor Market. Journal of Human
Resources 2008. 43; 4; 859-883.
Laband, D. N. & Lentz, B. F. Like father, like son: toward an economic theory of
occupational following. Southern Economic Journal 1983. 50; 2; 474.
24
Larson, L. M., Rottinghaus, P. J. & Borgen, F. H. Meta-analyses of big six interests
and big five personality factors. Journal of Vocational Behavior 2002. 61; 2;
217-239.
Leigh, A. Returns to education in Australia. Economic papers 2008. 27; 3; 233.
Long, J. S. & Freese, J. Regression models for categorical dependent variables using
Stata. Stata Press. College Station, Texas; 2006
Losoncz, I. (2007). Personality Traits in HILDA. HILDA survey research
conference. University of Melbourne.
Mazumder, B. Fortunate sons: new estimates of intergenerational mobility in the
united states using social security earnings data. Review of Economics &
Statistics 2005. 87; 2; 235-255.
McCrae, R. R. & Costa, P. T. Personality in adulthood : a five-factor theory
perspective. Guilford Press. New York; 2003
McMillan, J., Jones, F. L. & Beavis, A. (2008). The AUSEI06: A new socioeconomic
index for Australia. The annual conference of The Australian Sociological
Association. T. Majoribanks et al. University of Melbourne, Victoria, TASA.
Mueller, G. & Plug, E. Estimating the Effect of Personality on Male and Female
Earnings. Industrial and Labor Relations Review 2006. 60; 1; 3-22.
Nyhus, E. K. & Pons, E. The effects of personality on earnings. Journal of Economic
Psychology 2005. 26; 3; 363-384.
Ozer, D. J. & Benet-Martinez, V. Personality and the prediction of consequential
outcomes. Annual Review of Psychology 2006. 57; 1; 401-421.
Roberts, B. & DelVecchio, W. The rank-order consistency of personality traits from
childhood to old age: A quantitative review of longitudinal studies.
Psychological Bulletin 2000. 126; 1; 3-25.
Robertson, D. & Symons, J. The occupational choice of British children. Economic
Journal 1990. 100; 402; 828-841.
Roy, A. D. Some thoughts on the distribution of earnings. Oxford Economic Papers
1951. 3; 2; 135-146.
Saucier, G. Mini-markers: a brief version of Goldberg's unipolar big-five markers.
Journal of Personality Assessment 1994. 63; 3; 506.
25
Semykina, A. & Linz, S. J. Gender differences in personality and earnings: evidence
from Russia. Journal of Economic Psychology 2007. 28; 3; 387-410.
Shaw, K. L. A formulation of the earnings function using the concept of
occupational investment. The Journal of Human Resources 1984. 19; 3; 319340.
Shaw, K. L. Occupational change, employer change, and the transferability of skills.
Southern Economic Journal 1987. 53; 3; 702.
Terracciano, A., Costa, P. T. & McCrae, R. R. Personality Plasticity After Age 30.
Personality and Social Psychology Bulletin, pp. 999-1009 vol. 32: 2006. 32;
999-1009
Terracciano, A., McCrae, R. R. & Costa Jr, P. T. Intra-individual change in
personality stability and age. Journal of Research in Personality 2010. 44; 1;
31-37.
Tsukahara, I. The effect of family background on occupational choice. Labour 2007.
21; 4-5; 871-890.
Viinikainen, J., Kokko, K., Pulkkinen, L. & Pehkonen, J. Personality and Labour
Market Income: Evidence from Longitudinal Data. Labour 2010. 24; 2; 201220.
Wakefield, R. L. Accounting and Machiavellianism. Behavioral Research in
Accounting 2008. 20; 1; 115.
Watson, N. (Eds.) HILDA user manual - release 7. Melbourne institute of applied
economic and social research, University of Melbourne. 2009
26
Appendix
Table A1: Average marginal effects of base Human Capital Variables
Variable
Age
Age squared
PhD or Masters
Graduate diploma
Bachelor
Advanced diploma
Certificate 3 or 4
Certificate 1 or 2
Year 11 or less
Managers
0.0634**
-0.0000
0.0096***
0.0380
0.0186
0.0114
-0.0589***
-0.0171
-0.0243
Clerical
Age
0.0043
Age squared
-0.0000
PhD or Masters
-0.1557***
Graduate diploma -0.1415***
Bachelor
-0.1047***
Advanced diploma -0.0235
Certificate 3 or 4
-0.0510**
Certificate 1 or 2
-0.2058***
Year 11 or less
-0.0292
Legend
* p<0.1
Professional
0.0018
0.0000
0.4174***
0.3781***
0.3335***
0.0701**
-0.0701***
-0.04922
-0.1257***
Sales
-0.0053***
0.0001***
-0.0774***
-0.0597****
-0.0439***
-0.0330**
-0.0066
-0.0487***
0.0177
** p<0.05
27
Technician
-0.0056***
0.0000*
-0.0746***
-0.0521***
-0.0434***
0.0309*
0.1724***
-0.0065
0.0038
Operator
0.0044***
-0.0001***
-0.0579***
-0.0437***
-0.0536***
-0.0420***
-0.0050
0.1029**
0.06685***
*** p<0.01
Service
-0.0011
0.0000
-0.0743***
-0.0539***
-0.0671***
0.0374
0.0301
0.0390
-0.0192
Labourer
-0.0072***
0.0001***
-0.0709***
-0.0652***
-0.0395***
-0.0286**
-0.0109
0.1855***
0.0717***
Table A2: Average marginal effects of female specific Human Capital Variables
Variable
Age
Age squared
PhD or Masters
Graduate diploma
Bachelor
Advanced diploma
Certificate 3 or 4
Certificate 1 or 2
Year 11 or less
Managers
-0.0001
0.0000
0.0141
0.0090
-0.0232
-0.0292
-0.0230
-0.0510
-0.0044
Clerical
Age
-0.0039
Age squared
0.0000
PhD or Masters
-0.0745
Graduate diploma -0.0559
Bachelor
-0.0904***
Advanced diploma -0.00857***
Certificate 3 or 4
-0.0567**
Certificate 1 or 2
-0.3973***
Year 11 or less
-0.0270
Legend
* p<0.1
Professional
0.0019
-0.0000
0.0986*
0.1109***
0.0942***
0.1132***
-0.0822**
-0.1937***
0.0589
Sales
-0.0018
-0.0000
0.0539
-0.0376**
-0.0327***
-0.0102
-0.0329***
0.0385
-0.0094
** p<0.05
Technician
0.0011
-0.0000
0.0515
0.0339
0.0403
-0.0088
0.0004
-0.0072
0.0156
Operator
0.0021
0.0000
-0.0576***
-0.0576***
0.0176
-0.0281
-0.0227
-0.0552***
-0.0370***
*** p<0.01
Service
-0.0016
0.0000
-0.0061
-0.0341
0.0162
0.0156
0.0245
-0.0642**
-0.0351**
Labourer
0.0028
-0.0000
-0.0799***
0.0314
-0.0220
-0.0060
-0.0178
-0.0645***
-0.0049***
Human capital variables seem to exhibit significant effects on occupational choice. Age
seems to increase the probability of an individual being a manager or operator while
decreasing the probability of an individual being a technician, sales worker or labourer
occupations significantly. These effects are observed to be quadratic with decreasing
returns for all occupational states except managers. These are expected as these
occupations relative to salespersons, labourers and service workers may require less
physical exertion relative to other occupation such as clerical workers, professionals and
managers. It can also be seen as promotion with individuals moving across these
occupations with experience. It should be noted that there is no statistically significant
different between the effects of age on females and males. These findings confirm
previous research that potential experience influences labour market outcomes such as
occupational choice.
28
Education is specified as a series of education credentials in order to capture any possible
non-linearity and occupational specific capital that would not be captured using the
standard specification of education measured in years. It should be noted that all the
effects of education are relative to that of an individual with a completed high school
education. University level education, consisting of PhD or masters, graduate diploma
and the bachelor degree, have a strong positive effect on the probability of an individual
being in a profession and, to a less extent, manager. In addition, the effect of a PhD or
Masters for females being professionals is significantly larger than that of their male
counterparts as also generally are their probability of being in other occupational states.
The estimated model suggests that the possession of university level education credentials
draws individuals away from all other occupations fairly equally. This result is expected
as professionals, Table 1 in the text, require the completion of conceptual, creative and
analytical tasks based on a body of knowledge. This body of knowledge is typically
gained from university education and is very important for professionals. The effects on
management are also expected but are due to individuals ‘rising through the ranks’ to
management positions. Education credentials also exhibit some non-linearity, in terms of
years, and some occupation specific human capital. Certificates 3 or 4 from technical
colleges associated with the acquisition of a trade, technical or other applied skills are
after a high school qualification. This credential has a significant positive effect on
technician or service occupations but having a certificate 3 or 4 actually decreases the
probability of being in a profession or management occupation relative to that of a high
school graduate. This suggests that education does not just exhibit a constant linear effect
and may possess occupational specific components.
Other possible occupation specific effects can be the increased probability of being either
an operator with a certificate 1 or 2. This highlights the importance of occupation when
examining labour market outcomes and possible flaws with standard specification of
human capital theory as have been highlighted in previous literature (Shaw 1984; Shaw
1987; Heckman et al. 2003; Leigh 2008). Education has its largest effect on the outcomes
of professional occupations, with a university degree increasing the probability of a
29
professional occupation by approximately 0.3. Lack of education has a significant
influence, with people who fail to complete the final year of high school having reduced
probabilities, compared to high school graduates, of being a professionals and they are
more likely to be operators or labourers.
Table A3: Average marginal effects of the base Dynasty Hysteresis Variables
Variable
Father is manager
Father is technician
Father is service worker
Father is clerical worker
Father is sales worker
Father is operator
Father is labourer
Mother is manager
Mother is technician
Mother is service worker
Mother is clerical worker
Mother is sales worker
Mother is operator
Mother is labourer
Father’s AUSEI06
Mother’s AUSEI06
Father is manager
Father is technician
Father is service worker
Father is clerical worker
Father is sales worker
Father is operator
Father is labourer
Mother is manager
Mother is technician
Mother is service worker
Mother is clerical worker
Mother is sales worker
Mother is operator
Mother is labourer
Father’s AUSEI06
Mother’s AUSEI06
Legend
Manager
-0.0391
-0.0655***
-0.0544
-0.0530**
-0.0728***
-0.0672***
-0.0683**
0.0024
-0.0095
-0.0161
0.00079
0.0110
-0.04300
-0.0171
-0.0012***
-0.0001
Clerical
-0.0277
-0.0095
-0.0207
-0.0090
0.0138
0.0077
-0.0132
0.0169
-0.0459
-0.0037
-0.0306
-0.0087
-0.0055
0.01734
0.0005
-0.0008
* p<0.1
Professional
0.0026
0.0127
-0.0129
-0.0241
0.0256
-0.0328
-0.0328
-0.0406
0.0111
0.0236
-0.0034
-0.0335
-0.0407
-0.0227
0.0006
0.0001
Sales
-0.0094
-0.0167
-0.0037
-0.0203
-0.0122
-0.0083
-0.0121
0.0214
0.0342
0.0274
0.0196
0.0416
0.0419
0.0423
-0.0004.
0.0008
** p<0.05
30
Technician
-0.0147
0.03575
0.0573*
0.0639**
-0.0146
0.0128
0.0019
0.0125
0.0059
0.0086
0.0409*
0.0158
-0.0075
0.0415
0.0000
0.0005
Operator
0.0047
0.0063
0.0133
0.0008
0.0213
0.0312
0.0250
-0.0278
-0.0402***
-0.0287
-0.0373**
-0.0355*
-0.0346
-0.0451**
0.0001
-0.0007
*** p<0.01
Service
-0.0033
0.0030
-0.0007
0.0279
0.0071
0.0239
0.0347
0.0229
-0.0086
0.0223
0.0114
0.0192
0.0073
-0.0162
0.0003
0.0003
Labourer
0.0087
0.3200
0.0218
0.00137
0.0320
0.0332
0.0477
-0.0077
0.0001
0.0139
-0.0085
-0.0099
0.0007
0.00040
0.0001
-0.0001
Table A4: Average marginal effects of the female specific Dynasty Hysteresis
Variables
Variable
Manager Professional Technician Service
0.0083
0.0005
0.0547
0.0022
Father is manager
0.0729
-0.0347
0.0580
-0.0092
Father is technician
0.0737
-0.0185
0.0452
0.0064
Father is service worker
0.0687
0.0427
-0.0087
-0.0352
Father is clerical worker
0.0352
-0.0331
0.1058*
0.0026
Father is sales worker
0.0787
-0.0145
0.0912
-0.0270
Father is operator
0.1009
-0.0246
0.1006
-0.0534**
Father is labourer
0.0607
0.0087
-0.0179
-0.0569**
Mother is manager
0.0188
-0.0217
-0.0881**
-0.0429
Mother is technician
0.0181
0.0168
-0.0681
-0.0399
Mother is service worker
0.00203
-0.0137
-0.0711**
-0.0344
Mother is clerical worker
-0.0106
-0.0200
-0.0846**
-0.0366
Mother is sales worker
0.0843
-0.0552
-0.0965**
-0.0307
Mother is operator
0.0118
-0.01500
0.0955**
-0.0195
Mother is labourer
0.0019***
-0.0006
0.0016**
-0.0007
Father’s AUSEI06
0.0005
0.0003
-0.0012
-0.0010
Mother’s AUSEI06
Clerical
Sales
Operator
Labourer
0.0290
-0.0050
-0.0612*** -0.0284
Father is manager
0.0372
0.0010
-0.0618*** -0.0634***
Father is technician
0.0436
-0.0164
-0.0691*** -0.0648
Father is service worker
0.0254
0.0188
-0.0541*** -0.0575***
Father is clerical worker
0.0037
-0.0006
-0.0552*** -0.0584***
Father is sales worker
-0.0119
0.0023
-0.0649*** -0.0540**
Father is operator
0.0024
0.0080
-0.0681*** -0.0658***
Father is labourer
-0.0053
-0.0327
0.0288
0.0147
Mother is manager
0.0865
-0.0478*
0.1162
-0.0212
Mother is technician
0.0021
-0.0417
0.1307
-0.0179
Mother is service worker
0.0724
-0.0526**
0.0959
-0.0168
Mother is clerical worker
0.0598
-0.0465
0.1526
-0.0142
Mother is sales worker
0.0347
-0.0747***
0.1177
0.0203
Mother is operator
-0.0008
-0.0628**
0.1687
0.0132
Mother is labourer
-0.0006
0.0006
-0.0014
-0.0008
Father’s AUSEI06
0.0013
-0.0015**
0.0016*
0.0001
Mother’s AUSEI06
Legend
* p<0.1
** p<0.05
*** p<0.01
Parental status also seems to have significant effects on occupational outcomes, albeit
these effects are much smaller than those of education. Generally, social status measured
by the AUSEI06 seems to be insignificant; however, father’s AUSEI06 has a significant
and positive effect for males only on the probability of being a manager. Father's social
31
status significantly increases the probability of females being technicians and mother
social status increases the probability of being an operator and reducing the occurrence of
sales. These effects do suggest that parents with high social status do generally improve
the labour market outcomes of their offspring and supports dynasty hysteresis.
The binary variables for parental occupation provide some support for the hypothesis of
dynasty hysteresis. If the father of an individual is in any occupation other than a
manager then the probability of being a manager relative to a professional decreases. This
suggests fathers being in service occupations seem to have a significant positive effect on
the probability of their offspring being in a clerical occupation or for females in service
occupations. In addition a mother being a technician, clerical worker, sales worker or
labourer reduces the probability of that individual being an operator. Mother’s social
status has a number of interesting effects which can be generally summarised by if the
mother is in any occupation other than a profession it reduces the probability of an
individual being a technician, an operator or for a female being a sales worker. Generally
these results suggest some evidence of dynasty hysteresis; however, the effects are not
large in magnitude or systemic.
32
Table A5: Average effect of Base Demographics variables
Variable
Couple
Post couple
English
speaking
Non-English
speaking
Couple
Post couple
English
speaking
Non-English
speaking
Legend
Managers
0.0510***
0.004236
Professional
-0.106
-0.0515**
Technician
-0.0321***
-0.0080
Service
0.0132
0.0429
-0.0029
-0.0027
0.0078
-0.0045
-0.0489***
Clerical
0.0161
-0.0074
-0.0536***
Sales
-0.0150
0.0090
0.0479***
Operator
-0.0056
-0.0053
0.0152
Labourer
-0.0170**
0.0160
0.0010
0.0002
-0.0101
0.0115
0.0168
* p<0.1
0.0074
** p<0.05
-0.0238***
*** p<0.01
0.0391**
Table A6: Average effect of Female Specific Demographics variables
Variable
Female
Couple
Post couple
English
speaking
Non-English
speaking
Female
Couple
Post couple
English
speaking
Non-English
speaking
Legend
Managers
-0.2093***
-0.0347*
-0.0171
Professional
-0.1258
0.0295
0.0527
Technician
-0.0622
0.0298
0.0105
Service
0.2361
-0.0222
-0.036
-0.0234
0.0001
0.0171
0.0096
-0.0198
Clerical
0.0955
-0.0115
0.0117
-0.0000
Sales
0.0145
0.0109
0.0273
-0.05324**
Operator
-0.0584**
-0.0173
-0.0299**
0.0074
Labourer
-0.0207
0.0155
0.0050
0.0156
0.0004
-0.0189
-0.0005
-0.0078
* p<0.1
*0.0009
** p<0.05
0.0617
*** p<0.01
0.0126
33
Table A7: Average Effect of Location, Date and Industry for base variables
Variable
Victoria
Queensland
South Australia
Western Australia
Tasmania
Northern territory
Australia CT
2002
2003
2004
2005
2006
2007
Production
industry
Victoria
Queensland
South Australia
Western Australia
Tasmania
Northern territory
Australia CT
2002
2003
2004
2005
2006
2007
Production
industry
Legend
Managers
-0.0054
-0.0349***
-0.0096
-0.0127
-0.0127
0.0391
0.0077
0.0019
-0.0050
-0.0042
-0.0090
-0.0023
-0.0169
Professional
0.0070
-0.0175
0.0161
0.0360
-0.0201
-0.0699
0.0139
-0.0145*
-0.0287***
-0.0205**
-0.00253
-0.0321***
-0.0313***
Technician
-0.0135
0.0240**
-0.0213
0.00150
0.0004
-0.0263
-0.0068
0.0069
0.0041
0.0123**
0.0095*
0.0105*
0.0200***
Service
0.0085
0.0040
-0.0232
-0.0360**
-0.0016
0.0213
-0.0225
0.0174**
0.0234***
0.0191**
0.0298***
0.0263***
0.0299***
0.0876***
Clerical
-0.0022
0.0115
0.0102
0.0080
0.0146
0.0673
0.0556
0.0065
0.0025
0.0012
0.0028
0.0028
0.0063
-0.01111***
Sales
0.0124
0.0023
0.0054
0.0115
0.02311
-0.0177
-0.0373***
-0.0089
0.0038
-0.0044
-0.0059
-0.0090
-0.0067
0.1478***
Operator
-0.0022
-0.0012
0.0001
0.0043
0.0024
0.0008
-0.0163
0.0013*
-0.0009
-0.0019
0.0019
0.0017
0.0020
-0.1099***
Labourer
-0.0046
0.0117
0.0221
-0.0126
-0.0028
-0.0157
-0079
0.0.105*
0.0010
-0.0048
-0.0022
0.0020
-0.0033
-0.0608***
* p<0.1
-0.0667***
** p<0.05
0.0185***
*** p<0.01
0.0947***
34
Table A8: Average Effect of Location, Date and Industry for female specific
variables
Variable
Victoria
Queensland
South Australia
Western Australia
Tasmania
Northern territory
Australia CT
2002
2003
2004
2005
2006
2007
Production industry
Victoria
Queensland
South Australia
Western Australia
Tasmania
Northern territory
Australia CT
2002
2003
2004
2005
Managers
0.0008
0.0039
-0.0092
0.0359
-0.0626**
-0.0114
-0.0063
-0.0085
0.0110
0.0177
-0.0019
-0.0006
-0.0031
0.0412*
Clerical
0.0094
-0.0277
-0.0059
-0.0253
-0.0452
-0.0561
0.0100
0.0094
0.0099
0.0009
0.0178
2006
0.0148
2007
0.0048
Production industry 0.0922
Legend
* p<0.1
Professional
-0.0137
0.0074
-0.0039
-0.0245
0.0618
0.0140
0.0025
0.0125
0.0170
0.0083
0.0142
0.0159
0.0138
-0.1147***
Sales
-0.0002
-0.0023
0.0035
-0.0087
-0.0313*
0.0129
0.0164
0.0070
-0.0081
0.0134
0.0076
Technician
0.0223
-0.0285
0.0425
-0.0257
-0.0103
0.1676
-0.0142
-0.0159
-0.0327***
-0.0341
-0.0255**
-0.0281**
-0.0316**
-0.0632***
Operator
0.0137
0.0327
-0.0229
0.0058
0.0025**
-0.0524
0.0309***
0.0038*
0.0190
-0.0075
-0.0039
-0.00203
0.0136
0.0161
0.0214
** p<0.05
0.0057134
0.0249
-0.0042
0.0393
-0.0274
*** p<0.01
35
Service
-0.0176
0.0054
0.0197
0.0622
0.0255
-0.0490*
-0.0011
-0.0124
-0.0149
-0.0077
-0.0183*
-0.0192*
-0.0206**
0.0113
Labourer
-0.0147
0.0090
-0.0237
-0.0197
0.0374*
0.0254
0.03827
0.0116
-0.001131
0.00898
0.0099
Table A9: Average effect of a one standard deviation change in personality traits
Variable
Openness
Conscientiousness
Extraversion
Agreeableness
Emotional Stability
Managers
0.0061
0.0125
0.0162
-0.0126
0.0019
Clerical
Openness
-0.0098
Conscientiousness 0.0066
Extraversion
-0.0144
Agreeableness
0.0065
Emotional Stability -0.0084
Professionals
0.0117
-0.0006
-0.0020
-0.0016
0.0071
Sales
-0.0005
-0.0082
0.0060
0.0035
-0.0020
Technician
0.0015
0.0014
0.0012
-0.0032
-0.0026
Operator
-0.0044
-0.0032
-0.0000
0.0054
-0.0004
Service
-0.0006
-0.0008
-0.0004
0.0031
0.0027
Labourer
-0.0041
-0.0077
-0.0066
-0.0012
0.0016
Table A10: Average effect of one standard deviation change in education credentials
Variable
Openness
Conscientiousness
Extraversion
Agreeableness
Emotional Stability
Managers
0.0061
0.0125
0.0162
-0.0126
0.0019
Clerical
Openness
-0.0098
Conscientiousness 0.0066
Extraversion
-0.0144
Agreeableness
0.0065
Emotional Stability -0.0084
Professionals
0.0117
-0.0006
-0.0020
-0.0016
0.0071
Sales
-0.0005
-0.0082
0.0060
0.0035
-0.0020
36
Technician
0.0015
0.0014
0.0012
-0.0032
-0.0026
Operator
-0.0044
-0.0032
-0.0000
0.0054
-0.0004
Service
-0.0006
-0.0008
-0.0004
0.0031
0.0027
Labourer
-0.0041
-0.0077
-0.0066
-0.0012
0.0016