This paper sets out three methods for calculating the gains and losses that households would experience at the point of implementation and/or the total net cost of a Citizen’s Income scheme.
A. The typical household method
There is no such thing as a typical household, of course: but what we can do is list a variety of typical households by specifying the values of a number of variables: single or couple; number and ages of children; housing tenure; earnings levels; etc.. However large the set of typical households that we construct there will of course be numerous households that don’t fit any of them (three generational households are often difficult to fit into categories; shared houses containing couples and single adults can be difficult to define, especially when couple relationships are flexible; and a woman and two men, or a man and two women, living together in a household, can mean lots of things). But still, to construct a wide diversity of household types, and to work out for each of them whether they are likely to suffer losses on the implementation of a Citizen’s Income scheme, might at least give some indication as to whether the scheme will be what we might call ‘household financially feasible.’
Take the simplest example: a working age adult living alone. For each earned income level, from zero to a (somewhat arbitrary) high income, applying the current tax and benefit regulations will generate a disposable income for each earned income level. A second calculation can then be done by applying the new tax and benefit regulations that would accompany a Citizen’s Income scheme, and then adding the working adult Citizen’s Income. The calculation generates a second set of disposable incomes that can then be compared with the first set to determine the gain or loss at each earnings level.
Complications occur if any means-tested benefits are related to housing costs. For each earnings level a range of calculations will need to be done for each of a range of rent levels. Further complic-ations arise if households contain more than one employed adult, as for each of one individual’s earnings levels a separate calculation will have to be done for each of the other individual’s earnings levels. If all of a country’s tax and benefits calculations are based on a household’s aggregated earnings then this step is not required: but in most countries the calculations of at least some benefits and/or taxes relate to individual earned income, so the complication applies.
If a researcher is interested in the way in which a particular Citizen’s Income scheme would affect the disposable income of a particular typical household in which the number and ages of children, earnings levels, and housing costs, are already closely specified, then this method can be useful. As a guide to whether a particular Citizen’s Income scheme will impose a high number of unacceptable losses across particular earnings deciles it is less useful. It will often not be clear to what proportion of households a particular household specification might apply, and so even if calculations generate a list of expected gains and losses for a wide variety of household types, no overall picture of gains and losses will be delivered.
- The method cannot calculate the total net cost of implementing a Citizen’s Income scheme
- The method can be useful for illustrative purposes ( – the graphs in the CIT introductory booklet were created using this method)
B. National statistics
A country’s national accounts, population statistics, and other national statistics, can be used to calculate the cost of giving to every member of the population a Citizen’s Income, the money saved by abolishing tax allowances and means-tested and other benefits, and the additional revenue that would be collected if tax rates were raised. The net cost of the Citizen’s Income scheme is then the total cost of the Citizen’s Incomes less a) the money saved by abolishing allowances and benefits and b) the additional tax revenue collected. If a revenue neutral scheme is required then a process of trial and error can reduce the net cost to zero.
If means-tested benefits are retained rather than abolished, and the amounts of means-tested benefits received by households are reduced by taking into account households’ Citizen’s Incomes when the means-tested benefits are calculated, then the amount saved by reducing means-tested benefits will be different for each household. A method that uses the national accounts to calculate the net cost of a Citizen’s Income scheme would not be able to calculate the total saving in means-tested benefits that would result from every household on means-tested benefits having its means-tested benefits reduced, so using national statistics to calculate the net costs of Citizen’s Income schemes is only appropriate when each existing means-tested benefit is either abolished completely or retained in its present form.
An inescapable complexity is that if a Citizen’s Income replaces a variety of means-tested benefits, and other means-tested benefits are retained, then a method that employs the national accounts will only generate an accurate net cost for the Citizen’s Income scheme if retained means-tested benefits are calculated in exactly the same way as they were before the scheme’s implementation. In practice, any retained means-tested benefits would take into account households’ Citizen’s Incomes when they were calculated, whereas before implementation of the scheme they would have taken into account the means-tested benefits that have now been abolished. The difference between the cost of a retained means-tested benefit before and after the implementation of the Citizen’s Income scheme will be different for every household, and the total difference across the population could be either a saving or an additional cost, depending on how the retained means-tested benefit is calculated both before and after the scheme’s implementation. Aggregated national figures are all that are available if national statistics are employed to calculate the net cost of a Citizen’s Income scheme, so an assumption has to be made that any change in the total cost of a retained means-tested benefit will be zero. This might or might not be accurate.
- The method cannot calculate the gains and losses that would be experienced by individual households at the point of implementation.
- The method cannot calculate accurately the net cost of a Citizen’s Income scheme if means-tested benefits are retained and the method of their calculation changes at the point of implementation of the scheme.
- The method can be used to calculate the net cost of a Citizen’s Income scheme if the scheme abolishes all means-tested benefits, and it can also give an accurate figure for the net cost if any means-tested benefits that are retained are calculated in the same way both and after the scheme’s implementation.
As well as enabling new benefits to be created and their regulations and levels to be specified, the computer programme enables the regulations of existing benefits to be amended, and, in particular, enables households’ Citizen’s Incomes to be taken into account when existing household means-tested benefits are recalculated on the implementation of a Citizen’s Income scheme. This means that this method – unlike the method that employs national statistics – can cope with means-tested benefits being retained and recalculated on the implementation of a Citizen’s Income scheme. Microsimulation programmes can also provide a certain amount of detail if that is required. Suppose that a handful of households with low disposable incomes experience massive losses. The programme’s output will generally enable the particular circumstances of the household’s individuals to be studied in order to provide an explanation of the losses. A household of two parents and three children with twice the disposable income of a household containing just one adult will not be as well off as that individual adult. Therefore if we study the gains and losses for the one tenth of households with the lowest disposable incomes, we shall be studying some small households that are relatively well off, and we shall not be studying some larger households that are not well off. It would be possible to employ household weights, but I am not aware of this being done.
In spite of the problem, we are fortunate that modern microsimulation methods enable us to produce the results that they do: and as far as I know they offer the only way of providing those results.
Clearly we shall be particularly interested in the losses that would be experienced by low income households: but ‘household financial feasibility’ also requires that no losses will be unacceptable. A particular Citizen’s Income scheme might impose no losses at all on low income households, but if it imposes large losses on households elsewhere in the earnings range then it might still not be financially feasible in the way in which either the public or a government would understand that term. For household financial feasibility we require minimal losses in the lowest disposable income decile, and only acceptable losses higher up the income range.
For each household in the sample, the programme generates a gain or loss. This can then be turned into a percentage of the household’s original disposable income. We can then order the list of households to generate the results in which we are interested. Suppose that we want to know how many households in the lowest disposable income decile (the lowest tenth of disposable incomes) suffer losses in disposable income greater than 5% on the implementation of a Citizen’s Income scheme. The households can first be ordered according to original disposable income, and the bottom tenth of the list selected; and then that selected list can be reordered according to the magnitude of gains and losses. The number of losses over 5% can then be counted, and the number turned into a percentage of the size of the sample. This gives for the population as a whole the percentage of losses over 5% for households in the lowest disposable income decile.
Microsimulation undertakes calculations at the level of the household or the individual rather than at the national level. A survey of individual and household data relating to wages, benefits, other income, income tax, and social security contributions paid, is carried out, preferably for a random sample of at least 0.1% of the country’s population. A computer programme then uses the income and expenditure information collected to calculate each individual’s and each household’s disposable income. Tax rates, tax allowances, benefits levels, and other variables, can then be changed in the programme – and entire new benefits can be created – and the programme can then be run again to generate a new set of individual and household disposable incomes. An increase in a household’s disposable income represents an increase in public expenditure, and a decrease represents a saving: so the total of all of the new disposable incomes, minus the total of all of the original disposable incomes, represents the net cost of the Citizen’s Income scheme for the population sample in relation to the time period assumed by the microsimulation programme. If the programme uses monthly tax and benefits levels, then for an annual figure for the sample the net cost is multiplied by 12; and to obtain a total net cost for the population as a whole the sample net cost is multiplied by the ratio of the population to the sample.
- Microsimulation enables the net cost of any Citizen’s Income scheme to be calculated, including schemes that retain means-tested benefits and calculate them in different ways before and after implementation of Citizen’s Income.
- Microsimulation enables us to evaluate Citizen’s Income schemes of any kind for household gains and losses.
D. Unresolved problems
A further problem now relates to the roll-out of Universal Credit. Only when UC has been in place for long enough for the Family Resources Survey to have caught up with the number of people on it rather than on JSA, Working Tax Credits, etc., will we be able to use the UC provisions in the Euromod programme. It will also be some time before the national accounts catch up with expenditure on UC.
A further problem is the localisation of Council Tax Support. Euromod will struggle to incorporate every individual borough’s Council Tax Support regulations, so presumably a ‘typical’ set of regulations will be employed. Results will therefore be less accurate than before.
Unfortunately none of the methods can model contributions to the funding of Citizen’s Incomes achieved by altering the detail of tax allowances relating to expenditures. Take, for example, the UK tax allowance that enables the money spent on private pension contributions to be regarded as non-taxable income. For anyone paying only the basic rate of Income Tax, the saving in Income Tax will be the money spent on pension contributions multiplied by the basic rate of tax. For anyone paying a higher tax rate, the saving will be the money spent on pension contributions multiplied by the higher rate. A proposal might be made to reduce the tax relief to the basic rate for everyone. The Income Tax calculation for any individual paying the higher rate of tax would provide the information that would enable us to calculate the additional tax that they would have to pay and that would therefore be available to fund Citizen’s Incomes, but this amount will depend on individual circumstances, so no aggregate figure will be available. Neither is the information available in the Family Resources Survey data employed by microsimulation programmes. An estimate of the total amount of additional revenue can be made – as in our 2013 introductory booklet – but it might not be accurate.
Microsimulation appears to be the most useful of the three methods for generating the kinds of information on which academic debate and government-level consultations will continue to rely: but the national statistics will still have their uses for certain types of scheme, and household-level calculations will continue to be useful for illustrative purposes.