What are the characteristics of the employers of the low paid in Australia?

• Author: McGuinness, Seamus
• Position: Invited Article

Abstract

Service-sector employers from Communications; Accommodation, cafes and restaurants; Personal and other services; Retail trade; and Manufacturing have the highest densities of low-wage employees in Australia. In addition, the majority of all employers of the low paid were service-industry employers. A multivariate analysis found that employers of the low paid were more likely to be small firms employing a disproportionate share of casual labour. Employers of the low paid were more likely to be located in industries where labour costs form a relatively smaller proportion of turnover (possibly because of high material costs) and where there were higher rates of businesses recording a loss of profit. In addition, substantial regional differences remain in the presence of industry controls, indicating that state-level differences in the incidence of low-paying employers cannot be fully accounted for by the other factors.

1. Introduction

   While much is assumed with respect to the principal characteristics of employers of the low paid no serious attempt has been made (to date) to provide objective data on the matter. This paucity of evidence mostly reflects a lack of accessible plant-level data a situation common to both Australia and other countries.

   This paper seeks to address this lack. While the flurry of research activity that accompanied the introduction of the national minimum wage in the United Kingdom in 1998 did examine the employment implications for low-wage workers generally (Stewart 2001; Dickens and Manning 2002) and within particular industries (Machin and Wilson 2001), very limited attention was paid to employers. The one exception to this was a study by Forth and Millward (2001) who--using data from the 1998 Workplace Employee Relations Survey--found that, on average, employers of the low paid had higher densities of part-time staff, were smaller, were domestically owned, operated in uncompetitive markets, and were non-unionised.

   In this paper, we provide a descriptive analysis of low-pay employers. Section 2 describes the data and methods used, together with our definitions of low wages and of the employers of the low paid. Section 3 presents data on the average densities of low-wage employees within workplaces by industry, firm size and region. Section 4 reports the number of employers of the low paid, as defined by us, and presents their distribution by industry, firm size and State. A probit model of the probability of being an employer of the low paid is estimated using both firm and industry-level data. The results from this regression are given in Section 5. Finally, Section 6 provides a summary and conclusions.

2. Data and Methods

   Data for this study are derived from the 2004 Survey of Employee Earnings and Hours (SEEH)which is carried out on a biennial basis by the Australian Bureau of Statistics (ABS). The SE E H collects information on approximately nine thousand non-agricultural businesses and, after weighting, the data reflect the structure of employment within the Australian economy. Both part-time and full-time employees are included, but information on owner/managers and junior employees is excluded from this study The nine thousand management units in the sample have been weighted to reflect the full Australian population of 837,078 non-agricultural employing management units (employers). (1)

   Three low-wage thresholds are considered: $12, $15 and $17 per hour. The lowest represents the minimum wage in May 2004 (Fair Work Australia 2011) and is the main threshold we use; we also included two alternative wages, in selected cases, as a sensitivity analysis. For this study, an enterprise is defined as being an 'employer of the low paid' if more than 50 per cent of its employees are paid at or below the designated low-wage threshold.

   Wages for permanent employees relate exclusively to ordinary-time earnings and for casual employees they relate to hourly wages multiplied by 0.8 to account for the absence of sick leave and holiday pay. (2) For comparative purposes, however, we also present the unadjusted casual wage rates in some instances.

   Since the SEEH dataset is drawn from a sample of management units, and within these units from a sample of employees, our estimates are subject to sampling error. (3) These errors are not easy to detect, but future work could present confidence intervals based on bootstrapped standard errors.

3. Percentage of Low-wage Employees per Employer (low-wage density)

   In this section, we report the proportions of low-wage employees per employer averaged across all employers for Australia as a whole and by industry, employer size and State. (4) We call this a low-wage density. It is important to stress that this is not the average percentage of low-wage employees. That is, it is a simple average of low-wage percentages in each management unit. It is not weighted by employer size. Since about 99 per cent of all employers are small or medium-sized enterprises (SME) employing between one and 199 persons, these densities predominantly reflect the SME situation. The overall incidence of low-wage employment within Australia is given in McGuinness, Freebairn and Mavromaras (2007).

   Table 1 gives the average low-wage densities across employers for each low-wage threshold before and after adjustments have been made for the casual loading. Dealing first with the unadjusted figures, we see that the average low-wage density at $12 per hour was 6.7 per cent. This density rises to 30.2 per cent for the $15 threshold and to 50.9 per cent for the $17 threshold. The relatively steep increase in average density between the $15 and $17 thresholds suggests that the distribution of pay within firms is concentrated within a range of between 25 and 45 per cent above the federal minimum wage.

   After adjusting wages to remove the estimated loading for casual employees (row 2 of Table 1) the average density figures increase dramatically for the $12 threshold: from 6.7 per cent to 13.6 per cent. If the appropriate casual loading is 15 per cent--and not the conventionally assumed 25 per cent-13.6 per cent will be an overstatement of the average low-wage density; however, if the appropriate casual loading is 50 per cent, we will understate the true low-wage density.

   Figure 1 presents estimates of the $12 low-wage density levels across all employers after adjustment is made on casual wages to remove the estimated casual loading. Kernel density charts are similar to histograms, in that they show the distribution of employing firms with different proportions of low-wage employees. They a re especially useful for illustrating continuous variables such as low-wage densities. Since they involve a n artificial smoothing process, however, they should not be read literally but, rather, be interpreted as showing the relative preponderance of different x-axis variables.

   That said, according to Figure I the vast majority of firms pay all or most workers more than $12 per hour, with a further small proportion of firms employing most or all of their employees at or below $12 per hour. Thus, it would seem that employers are more likely to engage either almost all or almost none of their workers below a particular wage level than to engage them more evenly across a range of scales; however, this pattern varies considerably when the data are disaggregated by industry. The same pattern is also observed when Figure I is produced for the unadjusted hourly wage distribution. (5)

   [FIGURE 1 OMITTED]

   The distribution of low-wage densities across employers is also illustrated by Lorenz curves. Lorenz curves show the cumulative frequency of firms ranked according to their low-wage density. Taken as a whole, these curves illustrate how evenly (or unevenly) low-wage employment is distributed across employers. If all employers had the same low-wage density, the Lorenz curve would be a 45-degree '/'-shaped line. This 45-degree line is called the line of equality. If one employer accounted for all of the low-wage employees, then the Lorenz curve would be '[??]'-shaped.

   ...