RDP 1999-02: Reservation Wages and the Duration of Unemployment 3. The Data Source and Reservation Wages

Before discussing the empirical model, it is useful to get a feel for the data which will be used. Section 3.1 describes the Survey of Employment and Unemployment Patterns (SEUP) and the definition of our sample. Section 3.2 considers some descriptive evidence about the minimum acceptable wage rates reported in the SEUP.

3.1 The Sample

The SEUP is a longitudinal survey; information is collected from the same individuals over a number of years. The panel of individuals was established between April and July 1995 and includes people who were aged 15 to 59 at that time, and who were living in private dwellings in both urban and rural areas. A time line of the important events in the compilation of the SEUP data is presented in Table 1.

At the first interview, three sub-groups were defined: Jobseekers; a Population Reference Group; and persons known to have been labour-market participants. Jobseekers are defined as those who were unemployed, marginally attached to the labour force (i.e. discouraged job seekers) or under-employed in May 1995, and are the focus of this paper. The Jobseeker panel provides a far greater number of observations on episodes of unemployment than the Population Reference Group, and has a greater proportion of long-term unemployed.

At each interview, individuals are asked to divide their time over the relevant survey period (column 3 of Table 1) into periods of work, looking for work, or absence from the labour market. Clearly, these categories are not mutually exclusive as it is possible to be both working and looking for work. In this paper we focus on people who were looking for work and who were not employed at the end of the period covered in the third interview i.e. in September 1996. This definition corresponds more closely to the more familiar concept of unemployment, than the Jobseeker definition used in the SEUP, although, given the different survey questions in the SEUP and the Labour Force survey, it satisfies less stringent activity and availability criteria.

From this sample, we retain for analysis all individuals with valid answers to all the questions used in our econometric estimation in Section 6. This resulted in a sample of 1,063 unemployed job seekers.^[3] Of these 1,063 individuals, around 32 per cent reported that they had been retrenched or their employer had gone out of business, 23 per cent reported that they had previously been employed in temporary or seasonal jobs, 10.5 per cent reported that they had left unsatisfactory work arrangements, pay or hours and 17 per cent reported that they had left for child care, health and other reasons. The remaining 17.5 per cent either had never worked or had not had a job since 1985.^[4]

Of those unemployed job seekers who have had a job, 75 per cent have been in relatively low-skilled professions: intermediate and elementary clerical, sales, service, transport and production workers, and labourers and related workers. In contrast, only 47 per cent of employed people at that time worked in these occupations.

By educational attainment, 50 per cent of the unemployed sample had not achieved the highest level of secondary school. Another 19 per cent had completed the highest level of secondary school but had no further qualifications, with 21 percent completing vocational qualifications and just 10 per cent a degree or post-graduate diploma. By comparison, the proportion of employed respondents who had not completed high school was 34 per cent and the proportion with graduate qualifications was 22 per cent, suggesting that the unemployment pool is less well educated on average.

The duration of unemployment is measured as the number of days of unemployment up to 1 September 1996.^[5] Figure 1 plots the distribution of unemployment durations.^[6] The distribution is positively skewed with 44 per cent of the sample having an unemployment duration longer than 12 months, and 4 per cent having a duration in excess of five years. The average unemployment duration is 520 days.

3.2 Reservation Wages

Reservation wages are collected by asking job seekers how many hours per week they would like to work and, based on those hours, what is the minimum weekly take-home pay (or reservation wage) that they would be prepared to accept if they were offered a job. Respondents indicated ranges within which their desired hours and reservation wage fell; there were 7 bands for desired hours and 30 bands for wages. For each respondent, the hourly reservation wage is calculated by dividing the midpoint of the indicated band for the reservation wage by the midpoint of the band for desired hours.

To gain some understanding of the pattern of hourly reservation wages, Table 2 gives the medians and means of the ratio of hourly reservation wages to previous hourly pay for different groups of people who were unemployed at September 1996. The first thing which is notable about this table is that the means are much greater than one. It is hard to understand why people would only accept a job if it paid more than their previous job, especially if they did not leave their last job voluntarily. The median values reported in Table 2 are, however, much closer to one.^[7]

One possible explanation for the high mean ratios is that previous pay has not been indexed. We also report the median and average unemployment duration for each group to gauge the extent to which this is likely to be an issue. As the mean and median durations are not typically more than a couple of years and cover the relatively low inflation 1990s, this is unlikely to be a large effect. Measurement error may also go some way towards explaining these unrealistically high mean ratios. The dollar ranges for pay in jobs held prior to September 1994 are different to the ranges provided for reservation wages. As such, the same response to both questions may result in different hourly rates and hence, a ratio that differs from one.

Although the levels of the ratios themselves are puzzling, their relative values accord with prior intuition. The young have higher ratios than older age groups and those who were previously retrenched have lower ratios than those who left their job due to dissatisfaction with working conditions. This suggests that reservation wage data are meaningful.

It is also of interest to consider the relationship between reservation wages and minimum wages. Figure 2 plots the distribution of hourly reservation wages for unemployed job seekers as at 1 September 1996. The minimum wage reference point on this graph is an hourly wage calculated from the C14 classification in the Metal Industry Award, which was used as a benchmark minimum in the 1996 Safety Net Review. The gross level of this award is $349 per 38 hour week. Assuming that there are no transfer payments, nor deductions, and that the individual works full-time, this translates to an after-tax hourly minimum wage of $7.90, and is therefore comparable to the reservation wages which are also net figures.^[8]

The striking feature of the graph is that around one quarter of the sample have hourly net reservation wages below the legal minimum of $7.90 per hour. One qualification to this is that young workers are likely to face lower minimum wages as Australia has a system of age based discounts to the adult award wage that decline with age, up to the age of 21. The discount is as much as 50 per cent for 16-year-olds in some occupations and around 52 per cent of workers under 21 are employed on youth wages. Training wages and apprenticeships also provide a discount to employers with the understanding that structured training will be provided to the employee. Around 13 per cent of young workers are paid in this manner.

Figure 3 plots the distribution of reservation wages for 15 to 19-year-olds and for those older than 19 years separately. The benchmark minimum wage used for the youth is $5.30. This is the minimum wage paid to 18-year-old metal workers and is close to the Retail Trade Award for 18-year-olds, which covers a large number of teenage workers. Twelve per cent of teenagers report reservation wages below this benchmark, and the proportion of adult workers reporting reservation wages below $7.90 falls to 18 per cent.

This evidence suggests that the minimum wage, rather than the reservation wage, may be acting as a constraint on employment, at least for some individuals.^[9] It should also be stressed, however, that the relationship between minimum wages and reservation wages is only one element in a relatively complex relationship between legal minimum wage rates and employment levels. In particular, Figures 2 and 3 only present the supply side and have no information about the demand response of employers to lower legal minima. There are also other factors such as equity and income distribution, as well as employment consequences of a minimum wage which need to be considered when assessing its appropriate level. This is beyond the scope of the current paper.

Although we have used a benchmark minimum, it is also important to bear in mind that there is a system of award wages, which in general sits on top of the minimum wages presented in Figures 2 and 3. In many cases the relevant award wage will be the constraint facing job seekers rather than the benchmark minimum shown in the graph.

Another factor which is likely to affect the reservation wages reported by the unemployed who are expecting to earn low incomes is the interaction of the tax and transfer systems. For example, a couple earning $349 per week from one full-time income with two children under 5 and one child aged between 5 and 12 paying $120 per week in rent could receive as much as $312 in social security payments.^[10] As family income increases, means tested social security payments will fall and tax payments will rise. If the non-working partner starts working 15 hours per week at adult benchmark minimum wages, family social security payments will fall to $207, and taxes will rise from $46 to $51. Consequently, job seekers in families with significant transfer payments may report higher reservation wages because they need to be compensated for the loss of welfare payments.

Footnotes

Another 337 unemployed job seekers answered all questions but didn't know their hourly reservation wage. The definition of reservation wages used in our analysis is discussed in Section 3.2. [3]

The SEUP asks unemployed people whether they have had a job since 1985. There is no labour-market information before this date. [4]

International studies have indicated that, at successive interviews, respondents tend to change their answers to questions concerning their labour-market status at a particular point in time. To try to ensure the reliability of retrospective data, the ABS used dependent interviewing whereby the interviewer reminds the respondent of their labour-market activity at the end of the previous period. To further facilitate accuracy in recalling the dates of events, each respondent was supplied with a calendar to record their labour-market activity; use of the calendar, however, was not compulsory. [5]

All distributions have been estimated using kernel density estimation. [6]

This is consistent with a highly skewed distribution of these ratios. [7]

The SEUP asks for reservation wages in net terms and usual weekly earning in gross terms. ABS internal research suggests that respondents answer these questions with the correct measures, although using these two different concepts is likely to increase respondent error. The idea that respondents mistakenly answer these questions using the same concept is at least tentatively supported by the ratios of hourly reservation wages to hourly pay reported in Table 2 which tend to be around 1. [8]

Miller and Volker (1987) argue that it is the minimum wage and not reservation wages which is the binding constraint on youth employment. This is based on the observation that actual youth wages are on average 25 per cent higher than reservation wages and that award wages constitute, on average, around 95 per cent of the full-time wages bill. [9]

Information provided by the Department of Social Security. [10]