C. New Zealand faces a growing fiscal challenge - continued

What assumptions underpin our projections in the "Resume Historic Cost Growth" Scenario?

We used the Treasury's Long-Term Fiscal Model to put the projections in Table 1 together. We started by using the 2013 Budget Economic and Fiscal Update spending forecasts until the end of the 2014/15 fiscal year.[30] This is shorter than the forecast period in the Budget Economic and Fiscal Update, and it represents a change in practice from our 2006 and 2009 Statements, in which we used the most recent Budget Economic and Fiscal Update forecasts for the whole of the forecast period.

The reason for this change is that the 2014/15 fiscal year is the last year that the current Government will set the Budget. 2015/16 is the start of a new parliamentary term. Regardless of the composition of the Government in 2015/16, legally it will be a new Government. It may be that the new Government directly continues the current Government's fiscal strategy, or adopts a strategy that is very similar. But we cannot assume that. Accordingly, from 2015/16, we project the spending pressures that would build over time under the "Resume Historic Cost Growth" scenario. This scenario assumes that different spending categories will revert to their historic rates of growth per recipient (which are different for different spending categories), taking into account likely future economic and demographic factors, and assuming that current legislative settings do not change. The projections are intended to be a "what if" scenario.

As is implicit in the idea of a "what if" scenario, the projections are not a prediction for how expense areas will actually grow.

In terms of tax revenue, which makes up most of core Crown revenue, we assume that tax revenue will rise gradually between now and 2020, when we project it will return to its 2001 to 2012 average of around 29% of GDP (tax revenue is somewhat depressed right now for cyclical reasons, hence our projecting it to rise again to its recent historical average). From 2020, we assume that tax revenue will remain constant at 29% of GDP. A totally stable tax take each year is in fact very unlikely - tax revenue fluctuates owing to the economic cycle and also owing to changes in government policy. However, we assume tax revenue stays at its long run average in order to show the growing gap between expenses and revenue that arises if we assume that our tax take will be broadly similar in size (in relative-to-GDP terms) to what it has been in recent history.

Since our projections hold tax revenue constant as a percentage of GDP, where revenues are insufficient to cover expenses the implication is that governments must borrow. Accordingly, we project "Debt-financing costs" as an item of core Crown expenses, which increase over time in our projections. We also project the path of core Crown net debt, which also increases over time in our projections.

Both expenses and revenues are affected by the assumptions we make about demography (like longevity and fertility) and the economy (such as how many people will be working, and how productive we will be). The projections in this table are very sensitive to those assumptions.

In terms of demography, we use Statistics New Zealand's median projections to make assumptions about the birth rate and life expectancy. We also use Statistics New Zealand's assumptions about net migration - the number of people moving to New Zealand minus the number leaving in any one year.

For example, Statistics New Zealand's median projections show life expectancies continuing to rise over time and reaching 88.1 years for men and 90.5 years for women in 2061. This life expectancy assumption we use here is higher than that used in our 2009 Statement, as since then Statistics New Zealand has revised its life expectancy assumptions upwards. It may be that even the new assumptions turn out to be too low. That would make our government spending projections too low, as people living longer would boost the costs of NZ Super. It may be that our adjustment task is harder than we think.

In terms of economic assumptions, our projections incorporate assumptions about average hours worked, productivity growth, workforce participation, inflation, and interest rates. All of these assumptions affect the path of our projections.

For example, we assume annual productivity growth of 1.5%. Treasury research suggests this is a reasonable assumption for the future.[31] However, we might be wrong. Productivity growth could be lower than what we project. Although less tax would be collected, lower productivity growth would also slow the projected growth of many expense categories. For instance, a major driver of future expenses is the wages of public servants - doctors, teachers, and others. And lower productivity growth implies lower wage growth.

Our projections are sensitive to the assumptions we made in producing them. But changing our assumptions within reasonable parameters does not make much difference to the overall projections. In fact, others have come to essentially the same conclusion as us: over time we will see a growing gap between government expenses and revenue if we make no policy changes.[32]

Annex 2 to this Statement contains a table of our key modelling assumptions and also notes whether those assumptions have changed from those we used for our 2009 Statement.

The problem of actual numbers versus projected numbers

The 2010 column in Table 1 is different from the 2020-2060 columns. The numbers in the 2010 column reflect what our actual measures of expenses, revenue, and debt were in that year. The 2020 onwards columns reflect projected

numbers. One important difference between historical years and projected years is that historical years take into account economic cycles, whereas projected years cannot (as we do not know when cycles will occur in the future or how big they will be).

Some of the numbers in the 2010 column reflect the impact of cycles. For example, welfare expenses tend to be higher in the downturn part of a cycle, as more people need to be supported through unemployment. Also, tax revenues tend to be lower in a downturn.

Because many of the numbers in the 2010 column reflect cyclical factors, for some lines in Table 1 the change between 2010 and 2020 looks more significant than it really is. For example, Table 1 shows welfare expenses declining from 6.7% of GDP in 2010 to 4.8% of GDP in 2020. However, we spent an unusually high amount on welfare in 2010, owing to being in an economic downturn. The decline in spending in 2020 is best thought of as a return to a longer-term average, rather than a particularly significant drop.