RBA Annual Conference – 1998 Discussion

1. Claudia Goldin

The central point of this informative paper is that the unemployed in Australia are disadvantaged in several ways. They are, first of all, out of a job – a circumstance that often makes for a dispirited and even depressed condition. Second – and this is what much of the paper concerns – the unemployed are poor. Not only are they poor, but they are considerably poorer than are the working poor – those working at the minimum wage. (Frankly, I would be rather nervous about any benefits system that made the unemployed better off, on average, than those working at the minimum wage.)

Although there is no doubt that the unemployed are disadvantaged, the question I must first raise is whether the authors measure the income of the unemployed appropriately. The measure they use is the reported weekly earnings of the family during the week in which the household member in question experiences unemployment. Such a measure vastly overstates the difference in economic well-being between those experiencing unemployment and those who do not. The reason is because the income measure applies only to the week during which unemployment is experienced. Most families smooth consumption over intervals as brief as a week. Many of the currently unemployed will be employed at some point during the year and some of those currently unemployed will experience a spell of unemployment at another time. By measuring income for only the week during which unemployment is experienced, the difference in economic well-being between families with an unemployed member and those without, will be greatly magnified. Note that the odd result in the paper that the long-term unemployed fare better economically than do the short-term unemployed is most probably a consequence of this method of measuring income. A better measure of economic well-being would be to use annual or quarterly income (were it available) for the family.

The paper spurred me to consider differences between the Australian and US systems of unemployment compensation, and the importance of the incentives and disincentives that governments often inadvertently create. Is the relationship between unemployment and income influenced by the rules of the Unemployment Benefits system (UB)? That is, if the Australian UB rules were different, would the relationship between unemployment and income be less strong?

There are many reasons why the rules of the UB system affect who, in the Australian household survey, reports being unemployed. Those that are of the most quantitative importance and are likely to affect the income distribution concern:

  • the long-term unemployed;
  • married women; and
  • youths living at home with their parents.

In contrast to the Australian UB system, consider the rules of the US Unemployment Insurance system. In most American states – UI is a state system, although set up by the 1935 (national) Social Security Act – unemployment insurance benefits run out at 26 weeks. The long-term unemployed often take a lower-paying job after their benefits run out. Some, however, stop searching for a job and drop out of the labour force. Because the definition of unemployment in the United States requires that real action be taken by the agent – one must be actively searching for work – some of the long-term unemployed are simply not considered unemployed. If they stop searching, they are not in the labour force. In contrast, in Australia where the long-term unemployed are on the UB rolls, they remain in the labour force and unemployed. I am not claiming that one system is better than the other. I am simply pointing to the effect the rules of the system have on who is considered unemployed. The difference in the two systems influences who becomes re-employed and also who remains in the labour force and is considered unemployed.

Consider, as well, the fate of married women who are unemployed. Here the Australian system is less kind than is the American system. In the United States, where the system is one of insurance, rather than benefits, and is not means-tested by income or assets, an unemployed woman whose husband has a job – even if he is a stockbroker – may still be eligible for UI benefits, if she worked in the covered sector for a long enough period of time. But in Australia, where UB is means-tested, the unemployed married woman with a working husband may not be eligible for UB. And, if she does not receive UB, she may be less likely to consider herself unemployed in the household survey. Therefore, in Australia only the more indigent married women would receive UB and, thus, they would be more likely to consider themselves unemployed. Once again, the rules of the Australian system guarantee that the unemployed will be drawn disproportionately from the lower tail of the family income distribution.

The third case concerns unemployed youth living with relatives, generally at home with their parents. In the US, the youth living at home is no more likely to be eligible for UI than is the youth not living at home. Qualification would depend on the length of time the youth had been employed, whether the sector was covered, and the general accessibility of the UI office in the youth's city or town. More important for the issue at hand, is that a youth in the United States is no more or less likely to be considered unemployed than is his twin who is unemployed but married to an employed stockbroker. In Australia, the unemployed youth living at home with parents is more likely to be covered than is an unemployed youth who is married to an employed youth. By considering the youth living at home to be a separate household, we have added to the number of households living at, near, or below the poverty line and containing an unemployed household member.

The simple point is that in all three cases, Australian unemployment is increased at the lower end of the income distribution in comparison to the United States because of the different set of rules governing unemployment benefits. That is not to say that unemployment in the United States is evenly parcelled out across the income distribution. On the contrary. In the United States, as well, the unemployed are disproportionately the less educated and the less well off. That is to be expected, of course, because the unemployed have characteristics that led them to be laid off or terminated from a job and because their incomes suffer mechanically from not having a job.

What is the remedy for the poor economic position of the unemployed? UI and UB were initially designed to ease the transition from being out of work to locating new employment, providing sustenance during the period of search. Training programs alter the composition of the unemployed by changing their characteristics. If we believe that all low-wage workers and the unemployed are identical in their underlying characteristics and that the unemployed are, each period, simply a random draw from the lower skilled, then a high minimum wage would redistribute income from one (random) portion of the low skilled to another. The remedy might then be wage subsidies, if the polity remains committed to a high minimum wage. Alternatively, one could greatly reduce the minimum wage, shorten unemployment benefits, let the labour market adjust, and institute an earned-income tax credit.

In summary, although I have questioned the use of the weekly income variable as a measure of the economic well-being of the unemployed, this paper has provided a welcome introduction to the subject of unemployment and inequality in Australia, and it raised, in my mind, differences in how the United States and Australia treat, and thus define, their unemployed populations.

2. General Discussion

While the basic features of the wage distribution are known, prior to Harding and Richardson's paper, little was known about the position of those earning low or minimum wages, or the unemployed, in the distribution of household income. Given the new information presented in this session about these issues, the discussion centred on:

  • how reducing minimum wages would affect income distribution; and
  • what type of credits to low-wage earners should be used to achieve the desired distribution of income.

Participants were not surprised with the finding that the unemployed are concentrated in the bottom of the income distribution. More surprising, however, was the finding that minimum-wage earners are placed in the middle of the household income distribution, with low-wage earners placed somewhat higher.

A number of participants noted that the results of the paper imply that reducing the minimum wage would not necessarily have adverse effects on income dispersion. For those participants arguing that substantial changes to wage relativities are important to solve unemployment, this was viewed as a reassuring result; for others it was viewed with some scepticism and further analysis of the data was recommended. In particular, some suggested that the results may be sensitive to the use of weekly earnings data – in which current weekly income (transfers) relates to the week in which they are employed (unemployed). Annual earnings data were considered to better capture accrued income. Others noted that the analysis of the household income distribution does not give insights into how many new jobs might be created in the event of a wage cut, nor where they might be concentrated.

Given concern about the growing dispersion of wages and income, an important question is whether some type of superior distribution mechanism can be found. Some participants argued that tax credits for the low paid are superior, since they can be well targeted, and create an incentive to exit unemployment. But the incentive to exit unemployment makes it a different distribution mechanism to the existing system of family payments. Consequently, there was debate over whether income supplements should only be earned. Some argued that an exclusive focus on tax credits creates a ‘deserving poor’ who work, and an ‘undeserving poor’ who do not. All agreed however, that for those who do not work, it matters that targeted assistance has often been associated with very high effective marginal tax rates during the subsequent transition to employment, with consequent adverse implications for income dispersion and the incentives to work.

One participant also noted that there are political economy issues in the choice of distribution tool. In the absence of clear evidence about the effectiveness of tax credits paid to workers versus employers, the final choice may be influenced by the impact of a given scheme on the budget. Will it appear as a spending program or as a tax cut, given that tax cuts are typically more saleable politically?

Finally, there was discussion about the need for the Harding and Richardson paper to be complemented by a study on income mobility. For how long are people poor and why? Do low-income people move up the income distribution through time? The answers to these questions are important in considering policies to address growing wage and income dispersion.