Part 3: The value of the long-term financial projections

Commentary on He Tirohanga Mokopuna 2021.

During the next 40 years, New Zealanders will face a wide range of complex challenges and opportunities. Understanding what challenges and opportunities could significantly affect the government’s long-term financial sustainability is critical to supporting and informing the government’s strategic decisions in the Budget. It is also important for improving the quality and depth of public information and engagement.

Long-term projections are never precise, because the uncertainties and risks involved are greater the further into the future we look. To be helpful in understanding the implications of future challenges and opportunities, long-term projections need to provide realistic information that can be understood and potentially acted on.

As the International Monetary Fund stated in 2016:

Comprehensive analysis and management of fiscal risks can help ensure sound fiscal public finances and macroeconomic stability … and [c]ountries need a more complete understanding of these potential threats to their fiscal position.35

In this Part, we:

The Treasury’s two long-term projection models

For the first time, the 2021 Statement used two financial models to project the Government’s long-term financial position and explore issues that may affect the Government’s financial sustainability. These were:

  • the established Long-term Fiscal Model (LTFM), updated to 2021; and
  • a new Neoclassical Growth Model (NCGM).

The LTFM is the model that the Treasury used in its previous statements. The LTFM provides a reasonable approach to identifying a “benchmark” future financial path that possible changes in circumstances can be compared against.

However, the LTFM’s modelling has the following two main limitations:

  • It has a lack of feedback effects – for example, large projected increases in net debt levels should have substantive effects on interest rates that the government can borrow at.
  • The model cannot project regular short- to medium-term shocks that might knock the economy and government budget off their long-term tracks.

The Treasury developed the NCGM to address these weaknesses. The NCGM is a completely different and more economic-based projection model. This approach allows for numerous feedback effects and includes the ability to model short- to medium-term shocks, such as major recessions and natural disasters.

Using different models for different reasons is a well-recognised principle of economic modelling. This is because no single approach can answer all financial sustainability questions. Relying solely on the new and relatively untested NCGM for the 2021 Statement’s modelling would not have been prudent.

The Appendix provides a graphical summary of the two models.

The Long-term Fiscal Model’s assumptions and projections

The LTFM is an accounting-based model that projects the Government’s financial statements into the future. The first five years of the LTFM’s projections are the same as the Treasury’s shorter-term economic and fiscal update projections.

The projections assume that most spending and all tax revenue grows at the same rate as GDP over the long term. As in previous statements, the two exceptions that lead to a projected increase in the operating deficit and net debt relative to GDP are:

  • increasing healthcare costs per person arising from population ageing, which raises health spending from 6.9% of GDP in 2021 to 10.6% of GDP in 2061; and
  • growing superannuation entitlements arising from population ageing, which raises New Zealand Superannuation expenses from 5% of GDP in 2021 to 7.7% of GDP in 2061.

This additional spending is assumed to be entirely funded through debt, with government borrowing rising each year so that net debt increases from 34.0% of GDP in 2021 to 196.9% by 2061. Of all expense categories, debt financing costs show the biggest increase, rising by almost 8 percentage points over the 40-year period.

Figure 2 summarises the projections arising from the LTFM.

Figure 2
Projections from the Long-term Fiscal Management historical trends scenario (% of GDP)

2021 2030 2045 2061
Healthcare 6.9 6.8 8.6 10.6
NZ Superannuation 5.0 5.6 6.6 7.7
Education* 4.7 5.0 5.4 6.4
Debt-financing costs 0.6 1.6 3.7 8.4
Other expenses 15.9 12.1 12.1 11.9
Total expenses 33.1 31.1 36.4 45.0
Total revenue 29.3 29.5 29.5 29.6
Operating balance** -2.6 -0.3 -5.1 -13.3
Net debt 34.0 42.9 84.2 196.9
Net worth 11.7 7.7 -30.3 -137.1

* The Treasury’s 2021 LTFM background paper explains that spending on education per child has tended to increase over time and that this is assumed to continue over the projection period. See the Treasury (2021), Demographic, economic and fiscal assumptions and logic in the 2021 Long-term Fiscal Model – Background paper for the 2021 Statement on the Long-term Fiscal Position, pages 40-42.

** The operating balance is the difference between total revenues and total expenses, plus gains or losses in the market values of government assets and liabilities.

Source: The Treasury (2021), He Tirohanga Mokopuna 2021, page 19.

The 196.9% net debt projection in 2061 is slightly lower than the equivalent value in the 2016 long-term fiscal statement (206% in 2060). This difference is mainly from the following changes in the assumptions since 2016:

  • Updated Statistics New Zealand population projections now expect net immigration of 25,000 per year (previously 12,000). This raises GDP, reducing the debt to GDP ratio.
  • Productivity growth is lower at 1% per year instead of 1.5%. This reduces wage growth in the health sector (as elsewhere), reducing future healthcare costs.
  • The long-run interest rate on government bonds is down from 5.3% per year to 4.3% (and even lower during the next one to two decades). This reduces government debt service (interest) payments.

These changed assumptions may be reasonable to make currently. However, when looking ahead 40 years, financial projections that are sensitive to these adjustments raise questions about the reliability of the policy choices in each long-term fiscal statement.

For example, a change in the projected productivity growth rate will have a significant effect on projected education and health spending. So, should New Zealand’s slower average productivity growth in the 10 years since the global financial crisis be treated as permanent (as it is in the LTFM projections) or temporary over the projection period?

The Neoclassical Growth Model’s assumptions and projections

The NCGM uses a framework to capture the economic decisions of individuals – as consumers, workers, investors, and taxpayers. Like the LTFM, the NCGM allows for trend growth in productivity and population ageing, which affects the growth of New Zealand Superannuation and healthcare expenditures.

This type of economic model remains controversial among economic forecasters.36 However, compared to the LTFM, it offers greater capacity to investigate the effects of future economic and demographic trends and economic shocks on New Zealand’s financial sustainability.

Several assumptions distinguish the NCGM from the LTFM.

The NCGM incorporates many economic decisions by individual New Zealanders, such as decisions about how much to consume or to invest. These are affected by wage levels and interest rates, which are themselves affected by debt or tax levels set by the government. On the other hand, the LTFM simply projects each of its main economic variables forward, based on observed historical trends.

The NCGM assumes that rising superannuation and healthcare costs from population ageing are paid for by tax rate increases. This means that no new borrowing is needed and net debt remains the same, at around 48% of GDP over the projection period. Three different tax rates are increased (labour, capital income, and consumption). These increases slow GDP growth, as taxpayers react by reducing spending, investing, and working.

The NCGM’s starting value for net debt to GDP in 2021 is set at 48% of GDP,37 and this is relatively constant over the projection period. The LTFM adopts the current net debt value for 2021 of 34% of GDP, and this increases over the projection period.38

The Treasury acknowledges that the starting point of 48% in the NCGM is an arbitrary choice and that holding it constant over the projection period is an important but debatable assumption underlying the NCGM. The 2021 Statement shows that, if this 48% is reduced to 20%, the additional needed rise in the tax rate could be as much as 8% of GDP. This is because the lower borrowing means that additional funding through higher taxes is needed.

The NCGM scenarios include economic responses to unanticipated events, including global recessions, earthquakes, and severe droughts or floods. We discuss some of these in the next section.

Figure 3 summarises the projections arising from the NCGM.

Figure 3
Projections from the Neoclassical Growth Model increased tax rates scenario (% of GDP)

2021 2030 2045 2061
Healthcare 6.0 6.8 8.6 10.6
NZ Superannuation 5.0 6.2 7.2 8.3
Debt-financing costs 1.6 1.5 1.5 1.4
Other expenses (incl. education) 17.6 18.0 18.4 19.3
Total expenses 30.1 32.5 35.7 39.7
Total revenue 29.1 31.6 35.2 38.9
Operating balance -1.0 -0.9 -0.5 -0.7
Net debt 48.0 48.3 47.4 46.7

Source: Adapted from the Treasury (2021), He Tirohanga Mokopuna 2021, page 21.

The projected expenses for the NCGM differ from the LTFM because the NCGM models a long-run equilibrium projection from 2021 onwards. Despite the Government’s current response to Covid-19, the NCGM treats 2021 as if we were on a long-term trend (for example, projected welfare payments will be lower in the NCGM than they are currently).

The NCGM’s relatively constant 48% net debt to GDP assumption particularly affects 2021 debt interest payments and some other expenses.

The NCGM cannot project net worth because the model’s design does not include annual values for government assets and liabilities. This also means that the financial assets “netted” off in the relatively constant 48% net debt assumption are not separately projected.

Our observations about the models and their projections

We did not carry out an audit or line-by-line review of the LTFM or the NCGM. Previous versions of the LTFM have been reviewed in detail, and the NCGM consists of a complex set of inter-related equations that would take more time and specialist expertise to review in detail.

When we compare the projections from the two models, we usually start at 2030. This is because, as we mentioned earlier, the NCGM assumes a long-run equilibrium perspective. This means that its early projections from 2021 do not always reflect current conditions.

From 2030 on, both models reflect approximately the same conditions and treat the economy as fully recovered from the effects of Covid-19.

We set out the main findings from our review below.

The 2021 models and projections represent a significant change

In 2016, we questioned the rationale and various assumptions that the previous model and projections relied on. The Treasury has done considerable work to improve the modelling and projections for the 2021 Statement. There is evidence that the models’ developers worked together to align the models.

The NCGM is a major departure from the previous approach and provides a comparison to the established LTFM.

Figure 4 summarises our 2016 suggestions and the changes that the Treasury made for the 2021 Statement.

Figure 4
Modelling changes made in response to the Auditor-General’s 2016 commentary

Suggestions in our 2016 commentary 2021 Statement response
Prepare a set of plausible scenarios in support of its financial projections. The new Neoclassical Growth Model (NCGM) assesses demographic changes as well as different scenarios – for example, earthquakes, some climate change events, and economic shocks.
Establish a clear rationale and uniform approach to projecting these scenarios. For both models, there is a clearer rationale and a more uniform approach to the projections.
Reassess the consistency and reasonableness of the key projection assumptions. Difficult to assess because there is still a lack of clarity and little discussion about some of the main assumptions – for example, how taxes are assumed to respond in the NCGM and how these responses impact on well-being and financial sustainability.
Reconsider the rationale for excluding New Zealand Superannuation Fund assets from the primary financial sustainability indicator of net debt. Not addressed – in our view, excluding New Zealand Superannuation Fund assets from the calculation of net debt should be reconsidered. According to a background paper on the 2021 Statement, excluding New Zealand Superannuation Fund assets reduces net debt in 2060/61 by about 38%.*
Explore different approaches to measuring financial sustainability. Not addressed – the NCGM shows how much taxes could rise. However, this is not linked back to financial sustainability, and the NCGM cannot project net worth.
Obtain expert financial modelling advice. The Treasury obtained additional expert modelling advice to build and operate the NCGM.

* The Treasury (2021), Demographic, economic and fiscal assumptions and logic in the 2021 Long-term Fiscal Model – Background paper for the 2021 Statement on the Long-term Fiscal Position, page 56.

Compared to the 2016 long-term fiscal statement, the Treasury has made progress in modelling shocks and scenarios, sensitivity analysis, and incorporating feedback effects.

However, it needs to do more work on developing a wider view of financial sustainability and on some of the model’s assumptions, such as the rationale for excluding New Zealand Superannuation Fund assets from the primary financial sustainability indicator of net debt.

To more clearly show the benefit of having the New Zealand Superannuation Fund, we suggest describing the net debt indicator with and without the value of the New Zealand Superannuation Fund assets in the main document. As we discussed in previous commentaries, we remain unconvinced that New Zealand Superannuation Fund assets should be excluded from the calculation of net debt.

Projections show consistent revenue and spending paths

The two models are constructed in different ways, and their projections assume different things. A key difference is that the LTFM projections hold tax revenue constant, with any deficit funded through increasing debt. The NCGM’s projections hold debt (roughly) constant, with any deficit funded through increasing tax revenue.

Despite these differences, reviewing the two models’ projections from 2030 to 2061 in Figures 2 and 3 shows the following:

  • In both models, total government expenditure (excluding debt interest) increases by about the same amount from 2030 to 2061.39
  • If tax revenue were allowed to rise in the LTFM instead of net debt, the projected rise in tax revenue would be only slightly lower than the rise in tax projected through the NCGM.40

The models are constructed differently, use different assumptions, and provide different insights about the future financial profile and capacity of the government. However, from 2030 they have similar underlying revenue and expenditure projections. In our view, this indicates that there is a good underlying consistency to both models.

The two models tend to amplify the rise in debt or taxes

Both models rely on assumptions that, in our view, represent a pessimistic view of the government’s future financial management. Over the long term, this produces a particularly strong upward tendency for net debt (in the LTFM) or tax rates (in the NCGM).

For example, the LTFM assumes that all future operating deficits are only funded by borrowing money and that any interest on that borrowed money is also funded in the same way. The compounding effect of this assumption over 40 years means that, by 2061, debt financing costs will be the second-highest cost for the government – behind healthcare but above spending on superannuation and education.

As referred to in Figure 4, the assumption that New Zealand Superannuation Fund assets are excluded from the calculation of net debt reflects a view that these financial assets could not be used to avoid increasing debt to unsustainable levels.

The Treasury states in a background paper that the 2061 net debt to GDP value of 197% could be reduced to 159% if New Zealand Superannuation Fund assets were recognised and included in the calculation of net debt.41 In our view, it would be helpful if this were also included in the main document.

Whether and by how much projected interest rates on government debt exceed the growth rate of GDP is also critical for the size of operating deficits or surpluses that the government can afford. Although the models use different interest rate assumptions, projected interest rates for both models are above GDP growth projections in most (LTFM) or all (NCGM) years.

The 2021 Statement discusses the reasons for the level of projected interest rates in both models. The Treasury recognises that the gap between interest rates and GDP growth is important for long-term debt sustainability. It notes that “the interest rate has an impact on the level of debt but does not fundamentally change its trajectory during the next 40 years”.42

When projected interest rates remain above projected GDP growth, debt servicing (interest) costs grow more quickly, allowing less public spending to be funded through revenue. This means that more public spending must be funded by increasing debt or taxes. It also means that governments cannot “grow themselves out of debt” because the debt and interest costs increase faster than GDP and taxes.

In our view, the assumption that interest rates will always exceed the growth rate of GDP is an important, but debatable, assumption for a 40-year projection.43 For many governments, the reverse took place after World War 2 and the global financial crisis in 2008. Interest rates have been especially low since the global financial crisis.

He Tirohanga Mokopuna 2021’s focus is primarily on net debt

Long-term projections of government net debt to GDP are a useful indicator of governments’ future financial health. However, by itself, the measure does not cover the full extent of other projected changes that could influence the government’s financial sustainability over time.

The Public Finance Act 1989 provides a list of other factors that should be considered as part of responsible financial management. These include having predictable and stable tax rates, managing the Crown’s resources effectively and efficiently, thinking about the likely impact on present and future generations, and achieving and maintaining levels of net worth that can act as a buffer against adverse events.

Taken together, both models’ projections show what tax rates, net worth, and spending on other well-being domains could do during the next 40 years. Greater focus on these other indicators, alongside net debt to GDP, would provide a more comprehensive assessment of, and way to address, financial sustainability.

For example, tax revenue is fundamental to the government’s financial sustainability. The principles of responsible financial management (under the Public Finance Act 1989) suggest that large, unexpected fluctuations could be detrimental to taxpayers’ willingness and ability to pay.

The NCGM’s projections of a 7.3 percentage point increase in tax revenue from 2030 implies average corporate taxes rising from 30% to 40%,44 average personal tax rates rising to 26% (they are currently about 19%),45 and GST rising to 20%. The 2021 Statement may benefit from some discussion about what tax rates are likely to be financially responsible and sustainable.

Another example is the LTFM’s large projected deterioration in net worth from +11.7% of GDP at the start of the projection to -137.1% of GDP in 2061. Net worth is an important long-term measure of the strength of the government’s balance sheet and the stewardship of its assets and liabilities. A net worth that is -137.1% of GDP means a weak balance sheet where the government owes substantially more than it owns.

This projected deterioration in net worth is because the modelling for the LTFM assumes that core Crown assets increase at a far smaller rate (8% in total to 2061) than core Crown liabilities (278% in total to 2061). The larger increase in liabilities is a result of the modelling assumption that all operating deficits are funded through additional debt.

A large negative net worth indicator suggests that, in the future, the government’s balance sheet will provide little support for New Zealanders in times of need. It also raises questions about the effective and efficient management of the Crown’s resources (particularly its asset base). Both the above considerations are fundamental to responsible financial management and the Government’s long-term financial sustainability.

The Treasury told us that it is thinking about the role of the long-term fiscal statement and long-term insights briefing in the wider suite of stewardship reporting it is responsible for. There may be opportunities to think more broadly about other indicators and influences on long-term financial sustainability in other reporting, such as the upcoming well-being report or the investment statement.

Good transparency but the Neoclassical Growth Model is difficult to understand

The models and their projections are important for understanding the logic and evidence behind the many issues and policy options that the 2021 Statement raises.

The Treasury has always published the LTFM in full alongside the corresponding long-term fiscal statement. In 2021, the Treasury also published a background paper that set out the underlying equations for the NCGM.

Although this transparency is positive, because of the overall complexity of the NCGM, many of the assumptions and internal logic that sit behind the model are difficult to understand. For example, the internal equations that the NCGM uses to describe underlying economic relationships rely on more than 30 input assumptions, with limited testing of their reliability. Additional scenario analysis could help clarify these uncertainties.

Including new and different scenarios is positive but raises questions about wider financial sustainability

For the first time, the 2021 Statement includes an analysis of the financial effects of various possible events, such as economic shocks, earthquakes, droughts, and floods. We consider that this is a positive step.

According to the NCGM, the impact of these separate events on long-term financial sustainability is low. The 2021 Statement states that the Government’s financial position is “relatively resilient” to natural disasters and that the “uncertainty about the future does not change our analysis of the long-term financial trends that governments will need to manage”.46

However, this conclusion does not consider the possibility of these events occurring together and relies on the government increasing taxes after each crisis to pay for the cost of the shocks.47

Figure 5 shows how each scenario is funded, according to the 2021 Statement.

Figure 5
Scenario funding assumptions in the 2021 Statement

Scenario Funding assumptions
Economic shock/recession The government increases tax revenue after each recession to bring debt back down to its pre-recession levels.
Earthquakes The government increases tax revenue after the disaster to bring debt back down to its pre-earthquake levels. For example, the government gradually increases tax after the earthquake, and, after 10 years, the tax-to-GDP ratio is about 2% higher than the level expected if the earthquake had not happened.
Climate change: droughts and floods Nothing is noted in the 2021 Statement, but a background paper* states that taxes will increase after any floods or droughts to bring debt back down.

* The Treasury (2021), Shocks and scenarios analysis using a Stochastic Neoclassical Growth Model – Background paper for the 2021 Statement on the Long-term Fiscal Position, page 53.

Figure 5 shows that, for each event, significant adjustments in expenditure and revenue settings may be needed.

Despite this, the 2021 Statement has little recognition or discussion of the wider financial sustainability implications of these possible tax changes for the government or the public, separately or combined.

Our comments on the value of the financial projections

Using two projection models is a change from the single model that the Treasury used in previous statements. Both models show consistent revenue and spending paths over the projection period.

However, each model is built differently, and each measures the financial implications of a range of scenarios in a different way. For example, the original LTFM holds taxes constant and shows how net debt could increase over time, and the new NCGM holds net debt constant and shows how taxes could increase over time.

Although these projections are an improvement, we consider that the different measures, combined with some untested assumptions, result in:

  • a lack of clarity and confidence about whether a problem could exist with the Government’s long-term financial sustainability; and
  • a limited view of the costs and benefits needed to decide whether policy action is advisable or when choices between tax rises, increased debt, and/or spending restraints might be needed.

Although useful, the measures of uncertainty presented in the NCGM’s projections also do not allow for the model’s inherent (un)reliability. In our view, the insights from the projections are probably best interpreted as a start to estimating long-term financial impacts and with large margins of error.

In terms of possible policy actions, the Treasury acknowledges that:

While we do not know exactly how large a policy adjustment will be necessary … the scale of the long-term fiscal challenges will make a significant adjustment necessary.48

In our view, of the possible conclusions that the financial projections could support, this potentially overstates the magnitude of the adjustment. Particularly so, given that the various modelling assumptions result in an upward tendency for net debt or tax rates over the long term (including the New Zealand Superannuation Fund assets not being part of the calculation of net debt).

The findings of Furman and Summers also suggest that there should be caution about making important policy changes too early, particularly in times of significant uncertainty, because this makes assessing the policy costs and benefits of policy adjustment difficult.49

One way of mitigating this uncertainty might be to present different scenarios over different time horizons within the 40-year projection. For example, shorter-term challenges and opportunities could be modelled as one scenario and matched with a range of shorter-term policy choices.

The 2021 Statement’s projections offer new insights into the financial implications of some challenges and risks. However, in our view, the Treasury needs to do more work to provide a comprehensive view and to highlight the relevant financial implications and available choices in a realistic and meaningful way.

In our view, without this information, it will remain difficult for future long-term fiscal statements to inform and influence the government’s strategic policy decisions.

35: International Monetary Fund (2016), Analyzing and managing fiscal risks – best practices.

36: This is largely because the NCGM seeks to measure how individuals respond to different economic incentives, which needs different assumptions. Many of these assumptions have little supporting evidence. For more details about the use of these types of models, see Blanchard, O (2017), “Do DSGE models have a future?” in Gürkaynak, R S and Tille, C (eds), DSGE models in the conduct of policy: Use as intended, CEPR Press, London. See also Romer, D (2016), “The Trouble with Macroeconomics”, Commons Memorial Lecture of the Omicron Delta Epsilon Society, Stern School of Business, New York University.

37: This is the peak projected debt level taken from the Treasury’s latest Budget economic and fiscal update.

38: This is the current level taken from the Treasury’s latest Budget economic and fiscal update.

39: The LTFM projects an increase of 7.1 percentage points, and the NCGM projects an increase of 7.2 percentage points.

40: If tax revenue were allowed to rise in the LTFM instead of debt, then the projected increase in debt interest of 6.8% from 2030 to 2061 in Figure 2 would not occur – because taxes would rise by 6.8% instead. Adding the 0.1% tax revenue increase projected by the LTFM in Figure 2 gives a total tax revenue increase of 6.9%. The projected rise in tax from the NCGM from 2030 to 2061 is only slightly higher at 7.3 percentage points.

41: The Treasury (2021), Demographic, economic and fiscal assumptions and logic in the 2021 Long-term Fiscal Model – Background paper for the 2021 Statement on the Long-term Fiscal Position, page 56.

42: The Treasury (2021), He Tirohanga Mokopuna 2021, page 23.

43: For example, see Kirdan Lees’ presentation to the RBNZ-Treasury Workshop, “Fiscal and monetary policy in the wake of Covid”, Wellington, 22 June 2021.

44: The 2021 Statement uses the term “taxes on capital”. Taxes on capital are similar to corporate taxes but do not include the taxation of interest and dividends in individual hands.

45: The 2021 Statement uses the term “labour taxes”. Average labour taxes are similar to personal taxes. They include the taxation of wages and salaries but do not include any taxation of interest, dividends, or property rental income received by individuals.

46: The Treasury (2021), He Tirohanga Mokopuna 2021, pages 30 and 34.

47: The modelling assumes increasing operating deficits and/or debt in response to the event. This is paid for afterwards by increasing taxes.

48: The Treasury (2021), He Tirohanga Mokopuna 2021, page 44.

49: Furman J and Summers L (2021), A reconsideration of fiscal policy in the era of low interest rates – Discussion draft, Harvard University, page 31.