Part 5: The Treasury's model used to prepare the 2013 Statement

5.1
In this Part, we consider the underlying model, assumptions, and the financial projection that support the 2013 Statement.

5.2
Because the Treasury's main financial indicator uses net debt, our observations in this Part focus on the core rather than the consolidated activities of government.²⁷

Description of the model used to prepare the 2013 Statement

5.3
The Treasury uses a set of spreadsheets that were originally designed before 2006 to model medium-term financial strategies. The 2013 model:

• uses the latest Economic and Fiscal Update (EFU) forecasts as its base to project revenue inflows (mainly tax) and expense outflows (such as health and education) for the Government, using a set of demographic, economic, and other drivers; and
• shows how those revenues and expenses affect the operations, investments, and financing of the Government, using a statement of financial performance and position every year for 48 years.

5.4
Figure 5 summarises the structure of the 2013 model.

Figure 5
The elements and flows of the Treasury's 2013 model

5.5
The Treasury has used core net debt as the main measure of the Government's financial position. In line with current policy settings, the Treasury has excluded the New Zealand Superannuation Fund's assets from the calculation of net debt. The 2013 Statement notes that this approach is consistent with how net debt is defined in the Financial Statements of Government and reflects that these assets are "held for a specific future policy objective".²⁸ We understand that the policy objective is to help ease the financial pressure of the long-term costs of superannuation.

5.6
Although we understand the reasons given in the 2013 Statement, in our view, the Treasury should give further consideration to the exclusion of these assets in the calculation of net debt. If the New Zealand Superannuation Fund's assets were included in the 2013 model when calculating net debt, the percentage of net debt to GDP would reduce from 198% to about 170%. Alternatively, if these assets were all used to offset superannuation costs from 2032 to 2060 (through increased annual payments back to the Government), then the percentage of net debt to GDP in 2060 would also decline from 198% to about 170%.

5.7
The main purpose of the 2013 model is to estimate the funding needed to meet the operating and investing needs of the Government. This is primarily done through a single projection called the "Resume Historic Cost Growth" scenario, which assumes that tax revenues are held constant as a percentage of GDP and that government expenses grow from 2016 in line with their historical trends. The 2013 Statement also includes the "Sustainable Debt" scenario – a projection that assumes a spending path that maintains net debt to GDP at 20%.

5.8
Annex 2 of the 2013 Statement contains the main assumptions used in the 2009 and 2013 models. In our view, the main changes in assumptions between the two models are:

• a reduction in projected tax income the assumed rate has changed from 30% of GDP in 2009 to 29% in 2013; and
• a reduction in the level of government spending and forecast spending for the next few years in the 2013 model, government spending forecasts are based on current spending until 2016 and are lower than levels in the 2009 model.

5.9
As with earlier models, the 2013 model has maintained a focus on demographics, spending, and debt. Although these factors are important, many of the other wider drivers of financial sustainability mentioned in the 2013 Statement are not incorporated into the 2013 model.

Balancing simplicity and functionality

5.10
Although a financial projection will never fully reflect the real world, it is fundamental to supporting the Treasury's advice about the potential sustainability of government finances.

5.11
We reviewed how the 2013 model produces the long-term financial projection. We noted that the Treasury has made various judgements about what to include and how. These judgements tend towards more certainty, control, linearity, and simplicity more towards the left side of the cone in Figure 1. There is little recognition of the spectrum of possible outcomes for the wider, less certain, less controllable, and, sometimes, more catastrophic drivers of financial sustainability.

5.12
In our view, the structure and functionality of the 2013 model does not fully complement the Treasury's newer initiatives, including more public engagement and the comprehensive Living Standards Framework.

5.13
We consider that the Treasury should consider whether a better balance could be achieved in future between the simplicity and the functionality of the model. We think that this will help the Treasury to:

• explain the factors that affect the long-term financial outcomes of government;
• clarify whether the Government's long-term financial position represents a problem and what the size, profile, and nature of that problem is; and
• evaluate the options available to manage and/or control the long-term financial position.

Evaluating inherent uncertainty in the financial projections

5.14
Because the future is uncertain, a key function of any long-term financial model is to help understand the uncertainty that surrounds the financial projections.

5.15
Sensitivity analysis is a way of understanding how the uncertainties in the model and in the assumptions that drive the model affect possible outputs or measures.

5.16
Sensitivity analysis allows readers to better understand which assumptions matter most, how the profile of the projection might change over time, and how well the model reflects reality. Without this understanding, it will be difficult to prepare relevant and durable policy responses.

5.17
Early on, the Treasury recognised the risks of readers of the 2013 Statement not understanding the uncertainty inherent in the projections. In May 2012, the Treasury's Executive Leadership Team noted that updated economic and financial projections "suggested that a one percentage point increase in the tax GDP ratio could solve New Zealand's long-term fiscal challenges" and that this "understates the extent of the fiscal pressures and risks".²⁹

5.18
The Executive Leadership Team paper suggested addressing this by using more realistic parameters for some spending areas, such as health and "communicating more clearly the risks around the projections", such as the influence of another large economic shock on the financial outlook.³⁰

5.19
The linearity and simplicity of the 2013 model means that "what if" sensitivity analysis is limited. As a result, there is no provision to look at major contingencies, such as future shocks of the scale of the Canterbury earthquakes or the global financial crisis. Examples of the potential scale of these are the estimated public cost of the Christchurch rebuild (10% of GDP, spread over five years), and the estimated cost of mitigating climate change (2% of GDP a year).³¹

5.20
A simple way of appreciating the long-term potential effect of shocks is to look at how the projected net debt position changed from the 2006 Statement (when no shocks were expected) to the 2013 Statement (when two shocks had taken place). Without any radical changes in financial stance, the net debt to GDP projection (at the end of the projection period) increased from about 98% in 2006 to 198% in 2013.

5.21
The Treasury has included some sensitivity analysis in the accompanying research papers to the 2013 Statement. For example, one paper looks at how changing an individual assumption could affect the main financial debt indicator.³² Another paper refers to work by the Treasury in 2011 that attempted to assess the appropriate fiscal buffer for New Zealand by analysing the impact of fiscal and economic shocks on the fiscal position.³³

5.22
The Treasury's 2011 work concluded that "having a starting level of net debt below 20% of GDP is an important condition for ensuring these shocks would be manageable".³⁴

5.23
We recognise that estimating the effect of more uncertain events is difficult and involves making assumptions that will, most likely, later be shown to be wrong. Explaining this analysis could, as the Treasury has said, distract readers from the main messages in the 2013 Statement.

5.24
However, we consider that there are important advantages in including sensitivity analysis within a document like the 2013 Statement. On balance, we consider that the advantages, if presented properly, should outweigh the potential disadvantages.

Reviews of the 2013 model

5.25
There has been no independent detailed review of the 2013 model (or earlier long-term fiscal models).³⁵ However, the 2006, 2009, and 2013 versions of the model are available on the Treasury's website. Compared to other countries, this transparency is very good.

5.26
Although there has been no detailed review of the model, we were provided with some documents that discussed the model, including:

• A referee report on a technical paper that a Reserve Bank economist wrote in November 2009 to summarise the 2009 model.³⁶ The report found the paper to be clear in describing the technical and structural aspects of the 2009 model. The report suggested better connecting and modelling the relationships between the assumptions and better analysing the uncertainties in the 40-year projections.
• High-level comments in June 2012 by the Emeritus Professor of Economics at Victoria University's School of Economics and Finance, on two papers about the Treasury's draft 2013 model.

5.27
We carried out a high level review of the 2013 model. We found the model difficult to understand because of, for instance, the use of complicated formulae. Although we found no evidence of any errors in arithmetic, we consider that readers might struggle to understand how the Treasury prepared the financial projection and how it supports what is said in the 2013 Statement. From a financial modelling perspective, many of these points also mean that potential errors are more easily introduced and might be more difficult to identify.

5.28
We consider that future long-term financial models could benefit from the Treasury getting independent financial modelling advice.

5.29
The Treasury has made some assumptions to simplify the 2013 model relating to how the operations, investments, and financing of government are accounted for in the period from 2018 to 2060. However, as explained below, these assumptions create some uncertainty about how the future financial strength or weakness of the Government is measured and portrayed.

5.30
The values of government assets and liabilities are expected to change because of wear and tear, the money spent on them, or the money received from them. The values can also change with changes in the surrounding market environment. Although the values of some assets and liabilities change along pre-set (fixed) tracks in the 2013 model, the values of most assets and non-debt liabilities only change with movements in external general price indices, such as GDP or the consumer price index.

5.31
Projected debt will rise and fall depending on the operating and investing cash flow needs of government. However, because the 2013 model does not differentiate between cashflows and non-cashflows, it is unclear how much of the projected increase in net debt results from non-cash movements, such as market value gains or losses.

5.32
The New Zealand Superannuation Fund is expected to be used to offset future superannuation costs and, therefore, plays an important role in reducing future funding needs of the Government. However, removing these assets from the 2013 model makes almost no difference to the percentage of net debt to GDP in 2060. This is because most of the cash flow benefits of the New Zealand Superannuation Fund (that is, the withdrawals the Government will receive to offset future superannuation costs) are expected after the end of the projection period in 2060.

5.33
The 2013 model included a "sustainable debt" scenario, whereby government expenditure is reduced by a total of about 21% between 2012 and 2060. A reduction of this magnitude would affect all aspects of government, including the balance sheet. However, because the financial statements in the model are not fully integrated, government assets and non-debt liabilities do not change under this scenario, despite the significant reduction in spending.

5.34
We have discussed these matters with the Treasury. We understand that the Treasury plans to improve the financial model, particularly in relation to assets and non-debt liabilities.

27: The 2013 model projects both the core and consolidated activities of government. Core activities exclude, for example, Crown entities and State-owned enterprises.

28: The Treasury (2013), Affording Our Future: Statement on New Zealand's Long-Term Fiscal Position, page 12.

29: The Treasury (May 2012), Executive Leadership Team paper.

30: The Treasury (May 2012), Executive Leadership Team paper.

31: Doherty E. (2011), Economic effects of the Canterbury earthquake, New Zealand Parliament and former World Bank economist Lord Stern quoted in the Guardian newspaper (26 June 2008), "Cost of tackling global climate change has doubled, warns Stern", available at www.guardian.co.uk.

32: Rodway, P. (2012), "Long-term Fiscal Projections: Reassessing Assumptions, Testing New Perspectives", pages 35-40.

33: Buckle, R. and Cruickshank, A. (2012), "The Requirements for Long-Run Fiscal Sustainability", page 22.

34: Buckle, R. and Cruickshank, A. (2012), "The Requirements for Long-Run Fiscal Sustainability", page 22.

35: The Treasury Model was reviewed in 2003, but the review noted that this model was different to the long-term financial model.

36: Bell, M. et al (2009), Modelling New Zealand's Long-term Fiscal Position: The 2009 Statement.