Chapter 2: Welfare Based on Categories

A chapter of Globalisation and Welfare State

Keywords: Social Policy;

Gosta Espinger-Anderson characterized the New Zealand welfare state as a `”liberal” welfare state, in which means-tested assistance, or modest universal transfers or modest social-insurance plans predominate.’ (1) He put New Zealand with Australia, the United States, Canada, Ireland and the United Kingdom, in the most primitive group of the 18 nations he looked at. However his classification procedure was based solely upon entitlements to old-age pensions, sickness benefits, and unemployment insurance, and evaluated under the assumption that contributory systems were superior. Frank Castles argued that this categorization is inappropriate because it does not distinguish between those welfare states which are minimalist (most typically the United States), commonly called `residual’ welfare states, and those like New Zealand and Australia where delivery is more through the labour market and other non-social security mechanisms. (2)

While Castles is right to separate off the early post-war New Zealand (and Australian) his account is vague on the actual delivery mechanisms, which makes it difficult to trace what has been happening in recent years. Chapter 1 has already argued that a key feature of the early post-war period was non-stressful unemployment, which was the labour market component of the delivery system. That resulted in a different principle underpinning the social transfer system, which Espinger-Anderson failed to recognize.

There will, of course, be private means of support: the person’s private provision from savings and insurance; the family; perhaps the voluntary actions of a village, community, or employer if there is one; or a private charity. These have not always proved successful, and over the years some state provision has evolved in all rich capitalist economies. There are three general principles by which a person may be eligible to an entitlement to public support:
– dire need, as in the case of the British poor law, usually after all sources of private support are exhausted;
– previous contributions or work performance, an approach which developed out of the insurance tradition which first began in Germany;
– membership of a category, with the underlying principle of entitlement as a right of citizenship, which is known in the international literature as the Beveridge principle, following his report of 1942.(3)

This citizenship principle has been the foundation of the New Zealand welfare state since at least the 1938 Social Security Act (that is before the Beveridge report), and indeed it was argued in an early form in the 1880s and 1890s when the issue of state provision for the elderly was first discussed. It is a noble ideal but we need to explore how it works.

Obviously not every citizen of a country is practically entitled to every form of state assistance at any point in time. In practice their entitlement applies only when they are in need. However unlike the British poor law, the identification does not depend upon demonstrable and abject need, but membership of a broad category. Thus, in principle,
– state pensions are available to everyone over a specified age of retirement;
– a woman whose husband had died was entitled to a widow’s benefit;
– unemployment benefits are available to everyone who is unemployed and unable to find a job;
– sickness and invalids benefits are available to everyone who is unable to work because of sickness or disability;
– state provided education is available to everyone in particular age groups;
– state funded health care is available to everyone who is demonstrably in need of the particular treatment;
– the mothers of all children were entitled to a family benefit;
– and so on.

In practice there are a series of complications, which moderate these principles, but not drastically. For instance there was usually a residency test (to deal with recent immigrants), while those with economically active married partners were not generally entitled to a social security benefit (other than the family benefit), an issue elaborated in chapter 5. Another complication was that no categorical system could be completely comprehensive, so there was a backup needs based provision as well, in which entitlement was more discretionary. (Sometimes those in need would be a large enough identifiable group to evolve to a designated category in their own right. Chapter 5 describes this occurring for single mothers.)

Note that at this stage we have not stated what the level of entitlement should be. This was not merely a matter of the minimum level of the social security benefit, but also what would happen if the individual had other income – the abatement problem. These issues are set aside here to be returned to in Chapter 9.

In the 1960s, it was common to state – by those of a right wing political persuasion – that the New Zealand economy could not afford to fund the category based welfare state which existed then. It would be easy to say that they not only were the critics right, but the subsequent pressures on the welfare state demonstrated the correctness of their analysis. However this is to ignore a central but subtle feature of a category based welfare state. To explain this we begin with the analogy of the stockyard.

The Stockyard

Stockyards usually have races down which the sheep travel, coming to a swinging gate at which the race bifurcates. The gatekeeper quickly switches the gate back and forth to sort out which sheep should go down one race (say to the paddock) and which should go down another (say to the freezing works). Basically the system allocates the sheep into one of two categories (or more if there are also further gates). How does the system works so efficiently?

Obviously it requires some skill on the part of the gatekeeper to be able to identify the appropriate category for each sheep. It also requires that there is not a great deal of ambiguity as to which category – paddock or works destined – each sheep belongs. (In practice too, the costs of making an error are not great.) But there are a couple of other requirements.

First the gatekeeper’s sole interest is in correctly identifying the category of each sheep. Suppose that he (or she, but the ones I have seen are always male) was paid a dollar for each sheep that went to the works, and nothing for those that went to the paddock. Were there any ambiguity, one might expect the gatekeeper to send such sheep down the work’s race, since that was in his financial interest. The technical terms for this in economics is `adverse selection’, but we shall call it `selector discretion’, which is easier to remember. (5)

I originally used the phrase `selector bias’, but many people would think the term `bias’ has an element of fraud (it does not in its statistical use), so I have chosen a more neutral term. Selectors often have an incentive to choose a group according to a criteria other than that which the system intended. For instance a private medical insurance company prefers to deal with patients who are well, since sick ones cost them money. (Another source of selector bias would be if there was some limit on the number who could go down a particular race – suppose the paddock could hold only a certain number of sheep. This situation is analogous to there being funding limitations, so we shall come back to it.)

A second requirement is that the sheep do not take actions which influence the selection process. That does not happen in practice, but suppose there were some very clever sheep who worked out sometime before they were destined for the works rather than the paddock. By reducing their grass consumption they might look a bit straggly coming down the race and be sent to the paddock for further fattening. The technical term for this in economics is `moral hazard’. It happens when one signs up for private medical insurance, knowing that one needs some treatment. (The company is likely to try to counter it by requires a medical checkup before it agrees to the applicant on.)

The term `moral hazard’ is not a very good one, especially in New Zealand where morality tends to be most focused on sexual issues. Some cases of moral hazard involve sexual issues, but mostly they do not. For instance the most common form of moral hazard is probably tax avoidance, where people arrange their affairs to reduce the amount of income tax they pay. So we will call the phenomenon `behavioral response’. I did think of using the term `deceit’ or even `corruption’ instead of `response’, but either might suggest that the phenomenon is always fraudulent. As the case of tax avoidance shows it may not be. The underlying notion of behavioral response is that a person with some knowledge of the system may be able to manipulate circumstances in their favour, in ways which were not originally intended.

In summary the stockyard selection process works because

(1) there is little ambiguity (or the costs of a wrong decision are small);

(2) there is little selector discretion (adverse selection);

(3) there is little behavioral response (moral hazard);

(4) there are no limits on the numbers which go into either category.

How does the story apply to an entitlement in a category based welfare state?

Selection in a Category Based Welfare State

Suppose someone in the 1950s applied for an unemployment benefit. Think of the corridors of the (then) Department of Social Security as the stockyard race, and the benefit manager as the gatekeeper. He (in the 1950s it would most likely have been a he) has to decide whether the applicant is entitled to go to receive a benefit (the paddock), or go out and find a job (the works).

In the 1950s the four criteria for an effective gatekeeping applied pretty well. In particular there was not a lot of opportunity for behavioral response. The gatekeeper could easily check up if there were jobs available so the works option was viable. Indeed because it was relatively easy to obtain work, most potential applicants would not even visit the Department because they know they would be turned down. Moreover the cost of the scheme was small, so the gatekeeper was not under much pressure to save money by forcing the handful of unemployed to seek work when that was not practical. If it was deemed that the person was unemployed because they were sick or invalided, the benefit payment was exactly the same so the cost of allocating them to one benefit (category) rather than another was negligible. (That is not true today.)

Thus the category approach worked beautifully for the unemployed, as it did for most other social security applicants. True one could forge an age certificate or the husband’s death certificate, but as I have emphasized behavioral response is more about working the system legally, than fraud. In any case it did not happen often. Note that for a sickness beneficiary or an invalid there was more opportunity to obtain a misleading doctor’s certificate, but apparently that did not happen a lot. and yes, a woman could get pregnant to obtain a family benefit, but its value never covered the entire cost of a rearing and caring for a child. (We will get to the Domestic Purposes Benefit for solo mothers in due course.)

The education entitlement could work as long as it applied to everyone of a particular age (which removed ambiguity) or there was a realistic examination hurdle for post-compulsory education, and as long as the public were willing to fund the assistance.

The health service entitlement was a little more complicated, but it seems to have depended on the simplicity of health care in the early part of the post-war era. In those days there was not a lot of discretion. For instance elective surgery was rare, (5) and those terribly ambiguous states such as requiring psychiatric were rare or ignored. Costs of treatment were low too.

In summary there was not a lot of category ambiguity – including opportunities for behavioral response – and in the early post-war period the cost of the welfare state did not seem onerous. For instance in the 1959/60 fiscal year (which I have chosen because some benefit levels had just been increased in real terms and there was some unemployment pressure), the total public outlays on social security, war pensions, health and education came to around 15 percent of GDP. The comparable proportions in 1995/6 was nearer 27 percent. While the increase is exaggerated by the income tax beneficiaries now paid, and the GST on education and health services, there can be no doubt that 35 years later the welfare state was substantially more expensive relative to the productive capacity of the economy.

The point of drawing attention to the difference is not to judge whether one was superior to the other – later chapters will discuss why this happened. At this stage it is to remind us that the early post-war welfare state, with its low unemployment, its unambiguous categories and its low funding requirements was simpler to manage. But it was a system from the past to which it is not obvious we may readily return.

And yet the category based welfare state deserves to be recognized as an elegant, efficient, and equitable way to handle welfare issues, not the primitive and uncaring method which the Espinger-Anderson grouping seems to make it.

Next Chapter Ch 3:The Progress of Poverty

1. Espinger-Anderson (1990), p.26
2. Castles (1985, 1994).
3. Beveridge (1942).
4. Another reason is that what appears “adverse selection” from one perspective, may be “moral hazard” (see below) from another.
5. For instance the 1944 Shorter Oxford English Dictionary (3rd edition) does not associate `elective’ with `surgery’, while the smaller 1982 Concise Oxford Dictionary (7th edition) does.