This article appears in the following Oxford Review of Economic Policy issue: CAPITALISM AND INEQUALITY [View the issue table of contents]
Capitalist economies and wage inequality
* Amsterdam Institute for Advanced Labour Studies of the University of Amsterdam, e-mail: w.salverda{at}uva.nl
** Pembroke College, Oxford and SKOPE, Oxford University, e-mail: ken.mayhew{at}pmb.ox.ac.uk
 |
Abstract |
|---|
This article presents new stylized facts on the incidence of low pay and mobility out of low pay for 13 European countries and the USA. Women, the young, the less skilled, and part-timers are generally most at risk, as are those who work in retail, hotels, catering, and personal services. However, the relative importance of these characteristics can vary from country to country. The incidence of low pay varies considerably across countries, as does its trend. No direct link is found to aggregate employment or to the employment rate of the less skilled. Nor does the industrial structure of employment have much effect. However, differences between the low-wage production of goods and of services are important. ‘Inclusive’ labour relations are central in containing the incidence of low-pay. By inclusiveness is meant the existence of mechanisms, formal or informal, to extend terms and conditions negotiated by workers with strong bargaining power to workers with less bargaining power. In some countries a national minimum wage is an essential accompaniment. The article considers the extent to which countries can maintain the more benign institutions that limit low pay.
Key Words: low-wage employment • earnings mobility • wage inequality • labour-market institutions
We are grateful to Maite Blázquez at Universidad Autonoma Madrid for developing the probability and mobility analysis on the basis of Jenkins and Cappellari (2004), and to Daniella Brals at AIAS for her work on ECHP and PSID.
1 This included employees working part-time. In many countries much, if not most, of recent job growth has been in part-time work which, as we will see, is often low paid and thus contributes to growing inequality.
2 We are grateful to our partners in the project and to the Russell Sage Foundation, New York, who initiated and financially supported the project. Schmitt and Gautié (2009) present the comparative results. The original national studies are Bosch and Weinkopf, Caroli and Gautié, Lloyd et al., Salverda et al., and Westergård-Nielsen (all 2008).
3 ‘Demand Patterns and Employment Growth: Consumption and Services in France, Germany, the Netherlands, Spain, the United Kingdom and the United States’, an EU 5th Framework Programme project. Results are reported in Gregory and Schettkat (2007).
4 See www.eu-klems.net, available from the Groningen Growth and Development Centre. The data project was aimed at studying productivity and was supported by the European Union's 6th Framework Programme. It offers detail of 32 industries, it accounts for working-time developments, and covers the long period, 1979–2005 (data start in 1970 but seem more reliable from 1979 onward).
5 This section draws on Lucifora and Salverda (2009).
6 The alternative is to choose either a summary measure across the distribution as a whole, such as the Gini coefficient, or a fixed quantile, for example the bottom 20 or 30 per cent, of the distribution. The disadvantages of the former are that different measures stress different aspects of the distribution, that they do not always allow decomposition, and that, as a single measure, they assume a single mechanism affecting the distribution, while the latter implies that the incidence of low pay is always the same, leaving only the composition for further study.
7 It eliminates the effects of outliers and fat tails, to which we will return.
8 In the OECD dataset the correspondence across 15 countries to the two ratios is still stronger for the bottom half (R2 = 0.87) than for the top half (R2 = 0.67) but differences are small. However, definitions and years of observation are widely different. Note that levels may also differ from the Russell Sage project outcomes (see Table 3), especially for the Netherlands, in spite of a similar approach. It may be that the ECHP's household-based data are better at capturing small jobs than the establishment survey of Dutch Statistics.
9 From this perspective better pay and especially high pay, defined for example symmetrically to low pay as over and above 1.5 times the median wage (cf. Salverda et al., 2001), may deserve separate scrutiny.
10 The incidence below a threshold situated well above the minimum wage may be less sensitive to spikes and allow a rough estimation through linear interpolation, as in the case of the USA.
11 With somewhat higher levels for Denmark and Germany, since apprentices are included.
12 Website version of September 2008, last updated October 2007.
13 Where enough detail is available, a few parts of manufacturing are also included, especially textiles.
14 Blázquez and Salverda (2009) discuss these issues in more detail.
15 Unfortunately, PSID has no waves for the years 1998 and 2001 and therefore 2-years transitions had to be used. A quick check that dropped the same years for the UK analysis on ECHP found no fundamental differences. In addition, the US occupational classification is not identical to the International Standard Classification of Occupations (ISCO) used by ECHP and in the table the US results correspond to categories 1–19, 20–21, 30–39, 40–49, 50–59, 60–64 and 90–99, 70–89 (of the CNEF dataset).
16 Unfortunately, retail trade and wholesale trade are not treated separately in ECHP.
17 Note that the competitiveness indicator shows substantial volatility. Simple correlation between the two variables is never strong and in the USA, Germany, and the Netherlands it is even negative. In Denmark and France it is positive but small. In the UK the correlation is positive and larger (0.19), but its significance is minute (R2 < 0.10).
18 But compare, for example, Nickell and Bell (1995), who showed that the employment performance of the American low-skilled was no better than the European. See also Glyn and Salverda (2000).
19 For goods production: agriculture and fishing, textiles and leather, other manufacturing and recycling, and for services: car sales and retail trade, hotels and catering, and other community, social, and personal services.
20 In a purchasing power parity (PPP)-based comparison. Note that goods consumption per hour worked in retail provides an internationally comparative measure of the sector's productivity. It appears to be uncorrelated with the incidence of low pay.
21 In combination with the Russell Sage Foundation country studies. In the case of France we corrected the values of both the OECD indicators, for centralization and coordination respectively. Industrywide collective bargaining is predominant and there are also tripartite national agreements which point to a value of 3.5 instead of 2 for all three periods, while the high level of industry bargaining and its extension by the state point to 3.5 instead of 2 for coordination.
22 Note that for the Netherlands the adult minimum wage is used. This implies that many youths are included who earn above their age-related youth minimum wage. The employment incidence of the formal minimum wages declined from 9 per cent in 1979 to 4 per cent in 2005. Note also that in the USA, with the decline of the federal minimum wage, various state minimum wages have increased and their coverage of the workforce has shot up very quickly in recent years, up to 60–70 per cent (Mishel et al., 2009, Figure 3AC).
23 Effects are shown for FTE employment to control for the exceptional growth of part-time jobs.
24 Salverda (2008a, p. 103) criticizes the indicator for its fixed weighting scheme which allows no variation over time or across countries.
25 Germany could not be included as data problems owing to unification thwart long-run comparisons.