Thinking a little about indexation

The Minister for Social Protection wants to index many social protection payments to a cost of living index as an anti-poverty measure. This makes sense on the face of it, as long as that cost of living index is going up, and as long as the level of benefits fall when the cost of living falls. It’s also worth thinking about the virtues of indexation, as this was one of the main criticisms IFAC had of the fiscal space calculations during the last election.

Let’s say you index benefits to the consumer price measure of inflation.

Here’s what happened to that reading over the longer run.

Screen Shot 2016-07-22 at 11.29.28Just messing about with the idea a little more, imagine we ‘begin’ the Irish economy in year 1 with a CPI reading of 100, and grant benefits of €100. Then we can add in (say) the last 20 years of real CPI data from 1995 to 2015 to get a sense of what would have happened to benefits in a year-on-year basis as a result.

The line is the increase in benefits as a result of the indexation, and the bars are the changes in euros to the benefits as a result of the cost of living increase or decrease, measured on the right hand axis. The excel sheet I used to knock this up is here.


Hopefully you can see two things. First, the measure is highly pro cyclical. Precisely when we want benefits to decrease a bit, because the economy is growing strongly, they go up, and when we want benefits to increase a bit to cover the cost of living during a crash, they go down. Second, in recent years inflation has either stagnated, or fallen, so you wouldn’t see a huge increase or decrease in benefits either way. Now you could smooth out some of these effects out with a moving average of, say, 3 years, but this little exercise shows, I think, that it’s worth looking carefully at indexation proposals.

(Updated with thanks to commenter Tony_Eire.)

By Stephen Kinsella

Senior Lecturer in Economics at the University of Limerick.

18 replies on “Thinking a little about indexation”

The CPI is not a cost-of-living index. It is a pure price index. The two terms should not be used interchangeably.

A cost-of-living index allows for substution in response to price changes.

The CPI basket remains essentially unchanged for five years.

It rises probably faster than the true cost of living for this as well as other reasons.

Unfortunately there isn’t one.

Inflation in Ireland is unusually volatile due in large part to the high share of imported goods in final consumption.

It is also very difficult to forecast.

Take 2009. Forecast HICP in October 2008 was +2.5. Outturn was -1.5. The Department of Finance got the price level wrong by 4 percentage points.

You can index to outturn or to forecast but either approach will create big deviations from what you are trying to achieve. Unless you carry out multiple intra-year adjustments. This is not practical.

To maintain the real value of benefits you would need to have a correction mechanism. This would also need to allow for negative adjustments. Politicians will be wary of this!

As per Colm’s paper, the approach by the CSO to measuring the price of housing is flawed conceptually and practically.

Forecast error is more of a practical problem for any indexation approach but should not be underestimated.

In the 1980s social welfare payments were formally indexed to price increases and it was just as pro-cyclical as you might expect. Then the link was broken and in the 1990s and 2000s social welfare increases out-paced inflation by a distance – see page 58 of the 2001 report on benchmarking and indexation here: In the context of the proposal it is interesting that the report notes “At a time when incomes are increasing at a rate significantly ahead of inflation, linking social welfare rates to the CPI would leave social welfare recipients progressively further behind the living standards of the majority of the population.”

Social welfare is (for mostly good reasons) not amenable to being managed like some other state expenditures: that is it cannot easily be reduced in bad times. Therefore if the bad times co-incide with high inflation (say because of Ireland’s imported energy costs) the Government of the day would have to just ratchet up the welfare payments (pension, jobseekers, disability for certain and likely child benefit and too).

Best not to link to any hard figure but to have a ‘soft’ target perhaps related to net household income (e.g. a % of median income after tax/transfers etc.)

Above all don’t let them index it to GDP growth!

Hi Stephen,
1) Column E is % change in Benefits not euro change in benefits.
2) Your benefit column (C) seems to be Previous AMOUNT + percentage change in inflation. i.e. mixes euros and percentages. This does not correspond to any indexing formula I am familiar with. Is the point of indexing not to ensure % change in benefits = % change in CPI?? so formula in C3 would be =C2+C2*((B3-B2)/B2)
and the same for the other formulae in this column

Fixing this, this all becomes trivial as column C equals column B and the curves will lie on top of each other? Or are you pointing out that the formula being used somewhere is incorrect?

I agree that the change in benefits will be pro-cyclical to the extent that CPI is. Perhaps I am missing the point of this post. A point I would highlight with indexing is that by indexing benefits to CPI, we essentially ignore that there is likely to be a reinforcement effect where the increased benefits lead to increased CPI in future periods leading to a cycle of increasing benefits and prices, which is presumably undesirable. I think we’d need to make some assumptions about the speed of transmission fo the benefit rises into prices to have some idea of the short term effects. To be clear, this is not an argument against indexing by me, failing to index leads to a diminishing value of benefits, reducing the relative well-being of presumable the more vulnerable members of society which is itself undesirable. I guess the question then is can we create a mechanism to increase benefits without having a knock on effect on inflation during a boom.

I guess one starting point would be to explore how tax revenue responds to inflation vis-a-vis the increased cost to the govt. of benefits and costs on other govt. expenditures – I imagine this shows whether in real terms the increase in benefits is affordable in some sense, although I have not fully thought this through (a key point being that all govt tax and expenditure does not increase uniformly with inflation).


Thanks very much for taking the time to go through this. You’re right, it is unclear as I presented it, so I re-did the calculations above. The major point about pro cyclicality and lack of impact in low-inflation environments are now much clearer. I take your point about transmission mechanisms. Another commenter Brannofloin makes the really important point that social protection shouldn’t be considered the same as other government spending, simply because of its automatic stabiliser mechanisms. And of course you *never* want to index to GDP!

Thanks again for taking the time to comment.

Thanks Stephen.
To quibble a little, the purpose of indexation as I understand it is to preserve the real value of benefits. The real value of benefits is constant in all years in your graph. We did actually want nominal benefits to increase during the boom and reduce during the recession in contrast to your first point. Had benefits decreased during the boom as your first point would propose, the real value of benefits would have been eroded, increasing poverty. (albeit absolute numbers in poverty would probably decline due to the boom, but that is a separate point).

As political kite-flying exercises go, this has to be one of the best!

Gratitude is due on the part of all taxpayers for the speed with which it has been shot down, especially through this Irish Economy blog contribution.

Meanwhile, elsewhere in the forest!

cf. Budgetary Oversight Arrangements.

Before consideration is given to elements of the overall budgetary envelope, the size of the envelope has to be decided. On the evidence, this penny has yet to drop even at an exalted political level.


“Gratitude is due on the part of all taxpayers for the speed with which it has been shot down, especially through this Irish Economy blog contribution.”

If only Ireland’s EU contribution was as disconnected form economic performance, however measured, as you aspire that social welfare payments should be, Ireland would not have to fork out an extra €380 million (it is €380, I understand) next year.
I fear that you are inclined to discriminate as to who are the worthiest recipients within the flocks of geese.

In ‘normal’ modern economic times, services inflation has tended to exceed goods inflation, which benefits low-income people. There are of courses losses as well from globalisation. This is from the Economist:

Clothes now cost the same as they did in 1986; furnishing a house is as cheap as it was 35 years ago. More trade brought more choice, too. Robert Lawrence and Lawrence Edwards, two economists, estimate that trade with China alone put $250 a year into the pocket of every American by 2008. The gains from cheap stuff flowed disproportionately to the less well-off, because the poor spend more of their incomes on goods than the rich.

Durable goods are also becoming more durable: the average age of a car in the US is about 11.5 years and almost 10 years in Europe. An Electrolux domestic washing machine is likely to last longer than in the past and have lower requirements for maintenance. The price of PC type equipment has plunged and the quality/output has jumped.

Cars are also much safer than in the past.

Property-related costs such as hotel prices in main cities have jumped but communication costs have plunged.

The map below shows the ratio of annual income spent on food across the world:

It should be possible in Ireland in indexing the old age pension, to target the most vulnerable in say Dublin in particular where an old person’s biggest outlays maybe on food, rent and clothing.

Call me a cynic, but I assumed that this was an attempt to head off demands for real increases in benefits now that the good times are a-comin’.

Even if you could get a good measure of inflation, there is a problem with indexing to it, which is that if real earnings are rising through productivity growth, you end up with the bottom of the income distribution falling further and further behind – as mentioned by Tony. The ESRI, for example, made this argument very strongly during the boom. Preventing this re-starting would be good for the fiscal position but bad for income inequality.

Formal indexation involves a big trade-off. It introduces a level of counter-cyclicality which is often absent in Irish public finance policy. Welfare increase are in essence permanent spend as it is so difficult to reduce it. But it can be financed from transitory expenditures as we saw in the years to 2007.
The downside is that it reduces flexibility. There may be good reasons to cut spending in response to a structural deterioration in the public finances and indexation makes that very hard. Or there may be good reasons to share the benefits of an economy-wide leap in labour productivity with those who are out of work.

There is also a large choice of things you can index too, none of which are straightforward.

CPI: the most commonly understood measure of inflation. It is influenced a lot by mortgage interest though which most welfare recipients don’t pay. It also includes cigarettes and alcohol which welfare recipients consume proportionately more of. But compensating welfare recipients for (say) a policy-induced change in tobacco prices would be very difficult in political economy terms.

CPI excluding certain items: For example welfare dependent households spend less on mortgage interest and motoring but more on energy, food, alcohol and cigarettes.

Special CPI: you could construct a bespoke CPI which is most representative of the consumption basket of welfare-dependent households (see previous point).

Minimum wage: This is the reservation wage for the unemployed. Indexing unemployment benefits to it would keep this ratio fixed at least. However the ratio between out-of-work benefits and the minimum wage should not be fixed in aspic, it may need to vary according to the business cycle, societal preferences, etc.

Average wages: This is probably the most sensible thing to link pensions to, given that most future recipients will get the contributory version which is a function of how long you’ve worked over a lifetime. But again, setting the ratio in stone for a very long period may not be wise. There may be sustainability, inter-generational equity concerns.

The institutional setting would also need to be considered. As per the minimum wage you could outsource the decision to a body or an agency. But given it involves voted expenditure it would still need Oireachtas approval.

Putting in place a mechanism for downward nominal adjustment would need to be considered. Again the political economy of this is very difficult. Take the 2008-2010 period. Consumer prices fell 3%, tax revenues fell by about 25% but the state pension was left untouched in nominal terms.

There is no simple way of doing this. One would hope a lot of research on how it is done abroad would be done before any policy measures are announced.

I think income Inequality is coming to the end of its Kondratieff style cycle. The next economic sequence will be less capital friendly. Income Inequality is deflationary and a huge component of the current crisis.

Comments are closed.