Tuesday, December 11, 2018

The Distribution Impact of Budget Changes

After each budget the ESRI produce excellent analysis on the distributional impact of the announced measures on disposable income.  This tells us whether the budget is regressive or progressive.  The analysis published for Budget 2019 contains the following chart:

ESRI Impact of Budget 2019 changes

It is commonly used to infer the impact the budget had on disposable income.  This is not necessarily true.   The budget did not reduce the disposable income of all households as might be inferred from the above chart.

To understand this it should be noted that these losses shown above are relative to a benchmark scenario, in which welfare payments and tax bands and credits are indexed in line with wage growth.

The chart above shows whether households are better or worse off compared to this benchmark; not relative to their starting position.  In the ESRI benchmark scenario for Budget 2019, wages tax bands and credits and welfare payments are increased by 3.0 per cent.

Let’s try to set up some sample households to see the impact of this.

Example 1
Consider a household with an earned income of 30,000 (and no other income). Assume the single tax rate is 20 per cent and the tax credit is 3,000.

Pre-budget disposable income in 27,000. The benchmark scenario assumes a 3.0 per cent rise in wages and the tax credit. Thus income is assumed to rise to 30,900 and the tax credit becomes 3,090. Disposable income in the benchmark is 27,810 – a rise of 3.0 per cent.

If the budget does nothing to this household (by keeping the tax rate at 20% and the tax credit at 3,000) the distribution analysis will look at the impact of these (non)changes after the assumed 3.0 per cent increase in wages.

Thus, their assumed income of 30,900 will continue to be taxed at 20 per cent with the credit staying at 3,000.  Actual disposable income is expected to rise to 27,720 – a rise of 2.67 per cent

According to the distribution analysis the household is worse off by 0.33 per cent – and that’s even though the budget changed neither the tax rate nor the tax credit they get on their income.

Intuitively we would say the budget has no impact on them, but they are relatively worse off because tax credits did not increase in line with the increase in wages. A greater proportion of their income is now taxed.

The households "effective tax rate" has gone from 10.00 per cent to 10.29 per cent. This is called fiscal drag. They are said to be worse off even though the household disposable income has increased from 27,000 to 27,720. It would have increased to 27,810 if tax credits were increased in line with wage growth. The household are worse off relative to this benchmark by 0.32 per cent but in reality their disposable income has increased by 2.67 per cent (because of the expected wage growth).

Example 2
Consider a household who receive transfer income of 15,600 (and no other income). This one is more straightforward. The benchmark scenario assumes a 3.0 per cent increase in the transfer payment. So the benchmark is a disposable income of 16,068 – a rise of 3.0 per cent.

Let’s assume that the budget sees this family receive am increase in their transfer payments of 6 per week, i.e., an annual gain of 312. This means the household’s disposable income increases to 15,912 – a rise of 2.0 per cent.

According to the distribution analysis this household is worse off by 1.0 per cent – and that’s even though the budget increased their welfare payments by 312. Intuitively we would say that the budget has made them better off but they are relatively worse off because their welfare payment did not increase in line with the increase in wages.

The gap between this household on welfare and a household on wages is now greater (as wages are assumed to have increased by 3.0 per cent). They are said to be worse off even though their disposable income has increased from 15,600 to 15,912.

It should have increased to 16,068 if transfer payments were increased in line with wage growth. The household are worse off relative to this benchmark by 1.0 per cent but their disposable income has actually increased by 2.0 per cent (because of the increased transfer payment).

Distribution Analysis
If our economy is made up of just these two households and we have a budget that does nothing except increase the transfer payment to the household on welfare the distributional analysis where wages are expected to grow by 3.0 per cent is:

  • Welfare Household: –1.00%
  • Wage Household: -0.33%

The budget is regressive and everyone is worse off. Yet the only thing in the budget was an increase in the transfer payment. The budget gave income to the low-income household and did nothing to the high-income household yet is labelled regressive.

This is because the budget failed to offset the impact of the assumed 3.0 per cent increase in wages. It is because of this that the high-income household is relatively better off compared to the low-income household – not because of any changes in the budget.

Before the budget the high/low income ratio was 27,000/15,600 = 1.73. After the budget the ratio is 27,720/15,912 = 1.74 (based on the assumed increase in wages). 

Inequality has increased and we shout at the minister for introducing a regressive budget. The minister looks down at her notes and wonders what sort of eijits is she dealing with as she is after announcing a budget where the only change is an increase in transfer payments to the low-income household. How could that be regressive? It is regressive because income inequality in the economy has increased.

It is also possible that the same budget could be judged as progressive.  If we put in a wage increase of 1.5 per cent the conclusion changes – assumptions matter.  Here is the distribution impact for our simplified scenario with an assumed 1.5 per cent increase in wages.

  • Welfare Household: +0.50%
  • Wage Household: -0.17%

The same budget is now progressive.  This is because the income gap narrows.  The wage household is a little worse off because of the budget, again because tax credits were not increased in line with the expected wage growth (which this time is 1.50 per cent).  The “effective tax rate” goes from 10.00 per cent to 10.15 per cent.  The welfare household have gained because of the increase in the transfer payment and have gained even more relative to wage earners.

By working through the numbers it can be seen that the post-budget income ratio here is 27,360/15,912 = 1.72.  Income inequality has been reduced relative to the starting ratio (1.73).  So the same budget can be both regressive and progressive depending on what initial assumptions are put in.

This might seem like a massive complication of what should be a relatively simple concept – what impact did the budget have on household disposable income. However, a budget is not set in isolation. The impact of the dynamics of the economy (wage growth etc.) have to, or least should, be taken into account when the budget is being formulated so it is right that the impact of the budget is compared to some benchmark of what is expected to happen in the economy rather than the way it is now. We do not live a world where things are static.

This makes the analysis of the ESRI a bit more nuanced to interpret. The key thing we can take from it is whether a budget is regressive or progressive in terms of the income gap between households in a changing economy. Hence the simple contrived example above shows a regressive budget even though all the budget did was give a higher transfer payment to the low income household – it should have given them more to keep pace with the assumed 3.0 per cent growth in wages.

Budget 2019 did not make anybody worse off. Just looking at the measures introduced would show that. What the ESRI are saying is that inequality will increase if the assumed wage growth inputted to the analysis materialises. Hence the budget was regressive – or at least it was from decile three up to decile ten.

This does not mean inequality will rise in 2019. This is because the analysis by the ESRI is micro-based. So, yes, the micro analysis points to greater inequality as a result of government policy relative to a scenario where wages rise by 3.0 per cent.  However, the distributional analysis does not reflect macro factors such as employment growth which obviously has a large impact on inequality – and probably a greater one than the budget measures.

A continued increase in employment in 2019 will likely reduce inequality. A priori, the net effect on inequality between the offsetting micro and macro factors is impossible to determine.  For that we need something like the Survey on Incomes and Living Conditions. SILC 2017 is due to be published by the CSO next week.

1 comment:

  1. Nice post Seamus.

    The distributional analysis depends on a whole range of assumptions about wage and inflation growth, all of which involve judgements about which measures of wage and inflation growth to use, around which there is not an obvious consensus. They are also a huge hostage to fortune: budget-day forecasts are often wrong, and a regressive budget can turn progressive if inflation turns negative (and vice versa).

    The other issue is that the differences in impacts by household decile are really, really small - in the order of 20c in ever 100 euro of income. These are of course averages, and the volatility in year-to-year income for most households is large. It is not unusual for working households to move decile entirely from one year to the next - and these changes generally swamp any relative impact of discretionary fiscal policy.

    The SWITCH model is interesting and useful, but has its drawbacks in this context. I think you need to look at budgets cumulatively, ex ante to have any idea how it is really impacting income groups on a relative basis.