Yesterday I made the point that the way in which the IFS presents its estimates as to how progressive or not government measures are risks giving an over-precise impression bearing in mind the data and methodological issues the IFS has acknowledged.
That is an issue which comes up again in the answers from the IFS in this post, where the IFS’s position can be summarised as ‘yes, our calculations are not perfect but they are the best that can be reasonably done’.
Which means there is plenty of scope for people to place undue weight on their results given some of the reasons for imperfection:
What allowances do you make in your calculations of the impact of the Budget on the increase in benefits that will follow from the increase in inflation caused by the VAT increase?
When we analyse the impact of changes to taxes and benefits on household incomes and expenditures, we hold constant all other aspects of the economy (for example, the price level, patterns of employment, earnings, pension income, and what people buy). This approach is consistent with that taken by the Treasury when estimating the cost of tax and benefit changes.
The increase in VAT due in January 2011 is highly likely to increase measures of inflation during 2011, and this will almost certainly lead to increases in benefits, tax credits and tax allowances in April 2012 which will be higher than would have been the case had VAT not risen. Some benefit recipients will find that this offsets their losses from the rise in VAT. However, we do not think that it would be helpful to allow for this in our assessment of the impact of changes to taxes and benefits on household incomes, for two main reasons.
First, if we were to allow for this knock-on effect of higher inflation, consistency would require us to allow for other knock-on effects of inflation. For example, some employees will have wages or salaries which are contractually linked to measures of inflation, and some private pension payments and other forms of unearned income are automatically linked to measures of inflation. It is simply not possible for us to model precisely the impact of higher inflation on household incomes, and so the most consistent and transparent approach is to hold constant the price level when analysing the impact of changes to taxes and benefits on household incomes.
Second, it is relatively straightforward to assess the impact of a VAT rise on inflation (though there is some question as to how far firms will pass on the VAT rise to customers in higher prices). But many other tax and benefit reforms could affect inflation in ways that are much harder to assess. Similarly, the VAT increase and other policies could have affect a much wider range of economic outcomes, ranging from employment rates to exchange rates. It would be impossible for us to incorporate all the knock-on economic effects of all reforms into a distributional analysis. Rather than selectively incorporate some effects of some policies, but not others, we – like the Treasury – believe that the most consistent and transparent approach is to estimate the direct – almost “arithmetical” – effect of reforms on different households holding all other aspects of the economy constant. Where possible we often try to complement our distributional analysis with discussion of the possible wider economic effects such policies might have.
What allowances do you make in your calculations of the impact of the Budget or the Spending Review on the impact in changing the number of people out of work on the overall progressive / regressive effect?
The analysis published by us after the June 2010 Budget and the October 2010 Spending Review did not allow for any form of behavioural response to the tax and benefit changes. Assuming no behavioural change is a natural starting point, and is a good approximation to the welfare gain or loss caused by the tax and benefit reforms. This is again consistent with the approach taken by the Treasury in its distributional analysis.
When we analyse the distributional impact of tax and benefit reforms, we often analyse (separately) their effect on financial work incentives as well, to allow people to judge the likely impact on employment (and earnings). We will be doing this in detail when we publish an analysis of the Universal Credit.
But estimating the impact on employment (or other behavioural responses) and incorporating it into the distributional analysis would be much more difficult to do, would require more assumptions and judgement/guesswork, and would be much less transparent. It is also far from clear that incorporating behavioural responses would make the distributional analysis a better guide to the impact on people’s well-being, since (for example) the extra effort of working harder is a cost to the individual as well as bringing the benefit of extra earnings – otherwise they would presumably have chosen to work even before the reform in question.
Check back tomorrow for the final part of this series.