The welfare state has been heavily debated in recent years. Should we roll back the spendings of the welfare state or not? Most of the focus has been on whether the welfare state improves or erodes overall efficiency in the economy.
''Growth and Employment: The Scope for a European Initiative'' by Drèze and Malinvaud (1994) conclude that ''the agenda should be to make the Welfare State leaner and more efficient'' (p.82). ''Turning Sweden around'' (Lindbeck et al 1994) argues that the welfare state has ''resulted in institutions and structures that today constitute an obstacle to economic efficiency and economic growth because of their lack of flexibility and their one-sided concerns for income safety and distribution, with limited concern for economic incentives'' (p. 17). Martin Feldstein was concerned with the adverse effects of social security spending on economic performance. He concluded that ''the social security program [in the United States] approximately halves the personal savings rate, [which] implies that is substantially reduces the stock of capital and the level of national income'' (Feldstein 1974, p. 922)
These views are not the only ones however. Maddison (1984) stated that judgements of the influence of the welfare state on economic development were ''influenced mainly by ideological positions, or predictions about what might happen in the future''. Sandmo (1995) reviewed the aggregate empirical evidence between growth and social security and concluded that there was no clear connection either way.
Atkinson (1999) presents a powerful defence of the welfare state. He concludes that '' there can be little doubt about [the welfare state's] importance in providing income support''. He finds no conclusive evidence between the aggregate relationships of economic performance and the size of the welfare state.
Efficiency is just one aspect of the discussion of the welfare state. Another is legitimacy. Rolling back the spendings of the welfare state may very well affect the legitimacy of the welfare state as well as the efficiency. In a democracy, the scope of the welfare state is not decided by economists, but by the population through political channels. A welfare state that is opposed by a large fraction of the population has a low degree of legitimacy and is not sustainable over time. A change in the welfare state, which affects the majority's view of the welfare state, may thus have wider consequences than originally planned. When political support of the welfare state is affected by the scope of the welfare state, there may exist a conflict between efficiency and legitimacy. The welfare state may be a redistributive tool, with the less fortunate being subsidised by the more fortunate members of society. How efficient the welfare state is as such a tool is clearly dependent on its legitimacy.
Moene and Wallerstein (1999) investigate the effect of changes in the inequality in pre-tax and transfer income on the political support for welfare policy. In contrast to the findings of Romer (1975), Roberts (1977) and Meltzer and Richard (1981), they find that ''The impact of increasing inequality on the political support for redistributed \ policies depends critically on the way in which benefits are targeted when targeting is exogenous.'' (p.26). They conclude that ''When benefits are mostly targeted to those without earnings,... ,greater inequality of income reduces support for redistributive policies.'' and that ''Political support for benefits targeted to those without earnings goes down as inequality increases.''.
Casamatta, Cremer and Pestieau (1998) present a model with majority voting to examine the impact of the redistributive degree of the social insurance on the political sustainability of the welfare state. They find that there is a potential trade-off between efficiency costs and the political sustainability. They find that ''it may be appropriate to adopt a system which is less redistributive than otherwise optimal, in order to ensure political support for an adequate level of coverage'' (p. 27). They also examine the effect of allowing supplemental private insurance. They find that private insurance does undermine the political support for social insurance, but that this nevertheless may increase the welfare of the poor.
Casamatta, Cremer and Pestieau (1999) extend the model by allowing for two overlapping generations. They examine the effects of changes in the population structure, Pay as You Go-systems vs. Fully Funded-systems and take tax distortions into consideration.
Casamatta, Cremer and Pestieau (1998), (1999), Moene and Wallerstein (1999) and (2001) all examine the need for a degree of universalistic welfare policies in contrast to means-tested welfare policies. They all find that a degree of universalistic welfare policies may be needed for political support. M & W (2001) conclude that '' a limited welfare state that pays benefits only to the poor may be politically unsustainable in the absence of altruistic voting.'' Casamatta, Cremer and Pestieau (1999) find that universalistic policies are not desirable for the social planner. They find however that universalistic policies may be desirable when majority voting is taken into consideration due to the need for political support.
I present a model of how political support for welfare policies depend on whether there are private alternatives to the goods offered by the welfare state. The welfare policies here are insurance against income loss, with all welfare spendings being received by those without other income. The question asked is: What is the optimal amount of consumption of a private good, if this good is supplied by the government in an equal amount for all, and is tax financed? The trade-off is that the higher the tax, the higher the consumption of this good, but this means lower consumption of other goods as well. The answer is clearly dependent on preferences and income. In addition, it depends on whether there are private alternatives to the publicly supplied good that may be less expensive for some.
My focus is on the government as a collective supplier of a private good. I disregard other aspect of the government. The policy space then becomes one-dimensional. The only question is how large it should be, i.e. how high should the taxes be. Political competition may drive the level of the public supply towards the ideal tax rate for the median voter, i.e. the voter with the median ideal tax rate. If more than 50% of the population prefers a lower, or higher, tax rate, political competition may drive the tax rate up, or down. The higher the median ideal tax rate, i.e. the more people who prefer a high tax rate, the higher the realized tax rate and supply of the good in question.
The good in question could be any private good. I have chosen insurance as an example, but only minor changes are needed to capture the essence for other commodities, e.g. health care, pensions or education. The Von Neuman-Morgenstern utility function is a quasi-concave utility function, strictly quasi-concave in the presence of risk aversion, while the role of the coefficient of relative risk aversion is filled by the elasticity of substitution.
That the welfare state is a collective supplier implies that the scope of the welfare state, i.e. the amount of goods supplied, is collectively decided through political channels. Given that people are different, some may be dissatisfied with this amount. Some may find it too extensive, some may find it insufficient. I apply median voter theories for the analysis of the support of the welfare state.
I assume that people vote strictly according to self-interests. It is possible to allow for some degree of altruism, as done in Moene and Wallerstein (1999).
The welfare state described here, is only a social insurance system, with the government as a non-profit insurance agency. The scope of the welfare state (i.e. the tax rate) is assumed to be decided through political competition. The decisive voter is the median voter and the realised tax rate is the one that the median voter favours. The median voter is not necessarily the median income earner. This is shown in section four. An introduction of a private alternative may create a means-against-ends situation.
What are the effects of allowing for an optional additional private insurance on the scope of social insurance? I use social insurance as the term for the insurance supplied by the government and private insurance as the term for insurance supplied by the private insurance company. It is easy to conclude that private insurance can never hurt someone as long as it is optional. This is not straightforward however. An introduction of a private insurance may very well affect the preferences towards the social insurance for some groups of the population, and thereby affect the scope of the social insurance as well. This may imply a welfare loss for other groups.
There are several major differences between having the good supplied by the government in the above mentioned manner or by a private firm:
- While the amount of goods is socially decided through a political process in the public scheme, the amount is individually decided in the private scheme. The former promotes equality, while the latter promotes individual choice.
- If the consumption of the publicly supplied good is equal for all, and thereby unrelated to the amount of tax paid, the public scheme works in a redistributive manner, with consumption being more costly to the ones paying high taxes. In the private scheme there are no such aspects, with everyone paying the same price.
- With a tax financed public supply, there might be some dead-weight costs, which may increase with the supply. There are no dead-weight costs associated with private supply, but there may be rents, or profits, and other costs.
I do not consider informational aspects. Blomquist and Christiansen (1995) investigate how public supply of a private good may weaken the information constraints of non-linear taxation. I comment only very briefly on information aspects, even though there are many insights to be found here.
There may be reasons for wanting a mix of a public and a private supply. We may want to limit the dead-weight costs of taxation or allow for more individual freedom in the consumption of the good in question. However, an introduction of a private supplement to the social scheme may have effects on the preferences of the population and through this have effects on the amount of the public supply. This may have undesired equality and redistribution effects.
The private supply may be more desirable for some parts of the population than for other parts. If the amount of consumption is unrelated to tax payments in the collective scheme, the higher-than-average taxpayers are subsidising the lower-than-average taxpayers in the public system. Thus it may be preferable for some parts of the population to reduce the public part of the mix. If individuals are voting in accordance with their self-interests, this may effect the realised mix of the supply scheme.
If there are redistribution goals to the public supply, the social scheme is more favourable to some. With no price discrimination, the price in the private marked is identical for all. The private scheme is thus relatively favourable to some.
Additional private consumption for some only makes these individuals better off if the amount of public consumption remains unchanged. However the private supplement may affect the preferences about the amount of public consumption and lead to a change in the public sector through political processes. This may have consequences for the other consumers in either a negative or positive manner. If the consequences are positive or zero, the private supplement leads to a Pareto-improvement. Someone is made better off without anyone being made worse off. If the consequences are negative, there is a trade-off between individual and social welfare. The optimal mix of publicly and privately consumption then depends on the social weights put on the losses and gains of the different individuals. The object here is to analyse under which conditions the private supplement may lead to a loss for someone, and to highlight some factors that may influence a potential loss.
I present the basic model, with a homogenous population and no private alternative. Government spending is a tax-financed transfer to the unemployed. I then allow for a private supplemental insurance and analyse the effects of this. Next I make the population heterogeneous in the way that the risk of being unemployed is heterogeneous. Translated to another good, this would be that preferences over the various goods are heterogeneous. I again allow for a private insurance. Next I keep risk homogeneous, but introduce heterogeneous income and analyse the situation with and without a private insurance. I then add social mobility to the situation of heterogeneous income, with and without the private alternative.