RDP 1999-07: Job-Search Methods, Neighbourhood Effects and the Youth Labour Market 3. Data and the Basic Framework

The data used for the following analysis are from the AYS which covers the period from 1989 to 1994. The first wave, sampled in 1989, consists of 5,350 16 to 19 year olds. In each subsequent year roughly 1,500 16 year olds are interviewed for the first time, and all other panel members are re-interviewed where possible. Table 2 summarises the main activities of respondents of different ages.

Table 2: Main Activity by Age at the Time of Interview, 1989–1994
Per cent of total age group in parentheses
Age Main activity at the time of interview Total
School Employed Unemployed Other study Other
16 6,705 (84.3) 812 (10.2) 303 (3.8) 60 (0.8) 78 (1.0) 7,958
17 4,644 (61.9) 1,894 (25.3) 598 (8.0) 206 (2.7) 155 (2.1) 7,497
18 1,253 (17.4) 3,915 (54.3) 1,058 (14.7) 747 (10.4) 233 (3.2) 7,206
19 144 (2.2) 4,348 (66.1) 1,057 (16.1) 777 (11.8) 253 (3.8) 6,579
20–24 23 (0.2) 9,788 (73.5) 1,745 (13.1) 1,047 (7.9) 705 (5.3) 13,308

The sample used in the following analysis includes all 16 to 19 year olds who reported that unemployment was their main activity at the time of their interview, and who provided information on all the variables used in the estimation. The restriction of the sample to teenagers ensures that the age composition of the sample does not vary significantly over time. The results presented in what follows do not rely on restricting the sample in this way.

The dependent variable used in the following analysis indicates which search methods were chosen by different individuals as their main search method. Given the information in Table 1, we restrict the choice set to include ‘direct employer contact’ or ‘friends and relatives’ (taken as one category), visiting the CES and searching in newspapers. Sample sizes do not permit us to consider ‘friends and relatives’ as a separate job-search method.

The respondent's age, marital status and their number of siblings are included in the analysis to control for personal characteristics. These variables may capture the extent to which the respondent faces financial constraints or is likely to be independent, and consequently the costs of unemployment faced by the individual.

To control for the effects of family background on behaviour, several characteristics of the respondent's parents are included. First, there is a variable indicating whether each parent was present in the household when the teenager was 14 years old. For each parent who was present, questions are asked about that parent's work experience and educational attainment. Work experience is captured by two variables. The first indicates if the parent was not employed (i.e. unemployed or not in the labour force) when the teenager was 14 years old. The second is an index of occupational status, ranging from 0 to 100, constructed for each employed parent. Educational attainment of each parent is captured by variables indicating the highest educational qualification achieved.

Previous education and labour market experience are also likely to be important variables for explaining job-search behaviour. A dummy variable indicating whether the teenager left school in Year 10 or earlier, and a dummy indicating whether the job seeker attended a government school are included to capture the education experience of the individual.

The current duration of unemployment has been included to capture the possibility that the effectiveness and availability of different search methods may change over the course of a spell of unemployment. To ensure that this is not picking up some measure of the time spent in the labour force, the number of years since leaving school has also been included in the specification. A dummy variable which indicates whether the individual receives unemployment benefits or the Job Search Allowance (JSA) is included to control for the fact that receipt of this benefit will affect the costs of unemployment.

As already discussed, another potentially important influence on the job-search behaviour of individuals is their local environment. The AYS provides information on the state of residence and the section of state the respondent spent most time in until the age of 14 years.[8] The information provided in the AYS also allows us to identify the postcode where the interview took place. This is more disaggregated than most neighbourhood level data which are available: the average postcode has 5,558 residents over the age of 15 years; the largest postcode has a population of 62,885; and the smallest has less than a hundred residents. The distribution is highly skewed with 90 per cent of postcodes having fewer than 15,131 residents.

Using postcode information, it is possible to match individuals with information about the average characteristics of all the other residents living in that postcode area from the 1991 Australian Census. This includes information about education attainment, household and personal income, and labour force status by gender. Of special interest, given the model developed in Appendix A, is the local unemployment rate which can be thought of as a proxy for the effectiveness of local job-information networks.

Table 3 summarises the mean values of the variables used in the estimation in Section 4. A more detailed description of data definitions is provided in Appendix B. For comparison, the average characteristics of teenagers who obtained employment in the survey year and reported their successful job-search method have also been included. Based on our observation that direct search methods are more successful than the CES or newspapers, we would expect that characteristics which are more prevalent in the employed sample would also be associated with an increased probability that an unemployed teenager would choose to use a direct search method.

There is a slightly higher proportion of males in the unemployed sample. The employed sample come from families where the parents have higher skill levels on average, as indicated by the higher proportion of parents with graduate qualifications and trade qualifications. Parents of the employed sample also have higher average occupational status, and a higher probability of being employed. Members of the employed sample are also more likely to have been living with both parents when they were 14 years old.

Table 3: Sample Averages
Standard errors in parentheses where appropriate
  Sample of unemployed* Sample of employed**
Personal background
Male 0.52 0.51
Age 17.93 (0.99) 17.90 (0.99)
Married 0.06 0.04
Number of siblings 2.42 (1.76) 2.29 (1.57)
Parents' characteristics
Father's occupational status @14 22.89 (22.84) 28.26 (23.08)
Mother's occupational status @14 14.95 (20.20) 17.95 (20.89)
Father not employed @14 0.08 0.04
Mother not employed @14 0.45 0.38
Father not present @14 0.22 0.15
Mother not present @14 0.05 0.04
Father has:
Degree 0.10 0.12
Trade qualifications 0.13 0.18
Other post-school qualifications 0.08 0.09
Secondary education 0.34 0.37
No qualifications 0.35 0.24
Mother has:
Degree 0.08 0.09
Trade qualifications 0.04 0.05
Other post-school qualifications 0.13 0.14
Secondary education 0.56 0.58
No qualifications 0.19 0.14
School/work experience
Attended government school 0.79 0.74
Left school in year 10 or earlier 0.30 0.26
Years since leaving school 1.57 (1.13) 1.56 (1.00)
Current unemployment duration*** 29.18 (31.13) 9.94 (14.45)
Receives unemployment benefits 0.53 0.04
Neighbourhood
Average personal income 17.27 (3.01) 17.88 (3.31)
Unemployment rate 12.85 (4.81) 11.75 (4.31)
Per cent with vocational qualifications 14.51 (2.99) 14.95 (3.11)
Per cent with post-graduate qualifications 10.67 (6.40) 11.45 (6.65)

Notes: @14 indicates that the variable takes the characteristic of the parent when the respondent was 14 years old. Occupational status is an index ranging from 0, for low-status jobs, to 100 for high-status jobs.
* These are the average characteristics of those unemployed individuals used in the following estimation, i.e. 2,284 observations; ** these are the average characteristics for all the employed who became employed in the previous year and answered the question regarding how they obtained their current job. The actual sample size used will vary as some variables have missing information; *** this is the length of the previous unemployment spell for the employed.

Employed teenagers in the sample are less likely to have left school in Year 10 or earlier, and are less likely to have attended a government high school. Perhaps the most marked difference between the two samples is the average unemployment experience. The unemployed sample are experiencing an average incomplete duration of unemployment of 29 weeks, whereas the employed sample experienced an average completed duration of unemployment of 10 weeks. Roughly half the unemployed report that they receive unemployment benefits or the Job Search Allowance. A very small percentage of the employed report that they are also receiving benefits which is possible if they are earning a sufficiently small amount.

The neighbourhood composition variables tell a similar story to the family background characteristics. On average, respondents in the employed sample come from neighbourhoods where the unemployment rate is lower, and the average skill level measured as the proportion of the adult population with vocational or graduate qualifications is higher.

Footnote

Section of state is categorised as either capital city, other city, country town or rural area. [8]