Capex delays likely to greatly reduce state borrowing in FY2023 by $8bn to $16bn alone

Our research indicates that issuance of state government debt is likely to be between $8bn and $16bn lower than the market expects in FY2023 (or 2022-23 in budget terminology)  because history demonstrates that infrastructure projects regularly take longer to complete than planned. This possibility was reinforced on Monday by remarks from the NSW Infrastructure and Cities Minister, Rob Stokes, who declared that several multi-billion dollar infrastructure projects will be delayed because "volatility in cost of materials, equipment, and the shortages in skilled labour means it’s really not the right season to commence more mega projects immediately". Sharply higher debt servicing costs could also restrain public spending with the interest rate on NSW's benchmark 10-year bonds more than doubling from as little as 1.5% in August 2021 to 3.3% today.  

State and territory governments will deliver their 2022-23 (i.e., FY2023) budgets in a few months – most likely in May and June – and borrowing authorities will update their term funding tasks for FY2023 at that time. Based on the updated mid-year official issuance forecasts in December 2021, the estimated term funding task for all jurisdictions is expected to be about $84bn in FY2023.  

Coolabah Capital Investments' (CCI) research suggests that the FY2023 term funding task is, however, likely to be substantially less simply as a function of delays to state investment plans, such that infrastructure projects take longer to complete. This analysis is separate to the impact of any upside shock to state revenue from stronger-than-expected consumer spending if households draw on record savings buffers boosted by pandemic-related fiscal stimulus or the windfall from higher commodity prices. Our analysis suggests: 

• In a central scenario, a typical 10% deferral of state investment would reduce the term funding task in FY2023 by an additional $8bn to around $76bn, with spending delayed to the following two to three years; and 
  
  
• History suggests that the risks around this central scenario are skewed to the upside, where a 20% deferral of investment would reduce gross issuance by about $16bn to around $68bn. This risk is amplified by current supply-chain blockages coupled with Australia's tight labour market.   

Signs of this risk being realised are clear from recent media reports, with the Sydney Morning Herald noting on Monday that, "the NSW government is preparing to delay several multi-billion dollar mega projects ... Beaches Link motorway and an extension of the Parramatta light rail line. ... as a senior government minister warns that rushing ahead with major builds would be “reckless”. 

CCI's assessment that a significant share of state investment is likely to be deferred reflects two reinforcing issues:

1. Actual investment being smoother than government forecasts of investment because governments regularly overestimate investment in the short term and underestimate investment in the medium term; and
   
   
2. The practical difficulty in achieving the largest forecast increase in state public investment in about a decade.
   

This pattern of over- and subsequent underestimation is clear from government forecasts of cash non-financial public sector capex – which comprises investment by both general government and public corporations – in New South Wales, Victoria, Queensland and Western Australia, which together account for nearly all state and territory investment.

We have analysed actual versus forecast budget capex outcomes for the past couple of decades for New South Wales, Queensland and Western Australia, and the past decade or so for Victoria given a shorter forecast horizon in that state. State budgets typically report forecasts for:

• the current financial year, where the final budget outcome is not known for some time;
• the financial year corresponding to the date of each budget, called the "budget year" in this work; and 
• three subsequent financial years, termed “outyears” by state treasuries.

We find that state public investment in the current financial year is typically overstated by 2-3% across most states and by about 5% in Western Australia. Forecasts in the budget year typically overestimate investment by a much larger margin of around 10%. The forecast miss for the first outyear is more variable, ranging from a median overestimation of 7% in Queensland to an 8% underestimation in Western Australia. 

However, the pattern of underestimation is obvious in the second and third outyears, with investment underestimated by about 10-25% in the second outyear in most states, except Queensland, where there is a median overestimation of 2%. In the final outyear, investment is overstated by very large amounts, with the median degree of overestimation ranging between about 10% and 50%.

In many ways, the underestimation in later years is more understandable given the overestimation in the budget year points to delays in spending, particularly when new projects will also boost future investment given budget estimates are generally prepared on the assumption of no change in current policies.

The persistent overestimation of investment in the budget year is much harder to interpret given all treasuries periodically review their forecast performance, but it probably reflects the triumph of hope over experience, as well as promising ministers that new projects will be quickly under way and finished on time. 

Having said that, sharply higher debt servicing costs may temper the enthusiasm of the state treasuries to issue substantial new debt to underwrite discretionary projects; for example, the yield on New South Wales's 2032 bond has more than doubled from 1.5% in August 2021 to 3.3% today.

In the past few years, the degree of overestimation in the budget year has been larger at about 10% in New South Wales and Queensland and about 20% in Victoria and Western Australia. This larger overestimation – which is likely accentuated by successive outbreaks of COVID and some shortages of skilled labour given border closures – suggests that in current circumstances total state and territory investment in FY2023 could be delayed by as much as 20%.

A risk to the central case and upside scenarios is the potential for higher costs. These have been most apparent in residential construction, where private construction costs are currently increasing at an annual rate of 9%.

However, infrastructure accounts for nearly all state investment and the total price deflator for state and local investment has increased by only 3½% over the past year. This measure of inflation will no doubt accelerate given the sector will be at capacity, but it seems unlikely at this stage that it will reach the 10% rates posted in the 2000s.  

The risk of a larger-than-normal delay to capex in FY2023 also reflects the sheer scale of the investment planned by state governments, which is forecast to peak at about 3½% of national GDP that year. This almost matches the 4% high reached in 2009-10, which itself was the largest increase in investment since the late 1980s. 

Every state has a substantial programme of infrastructure investment for the next four years, although New South Wales had begun to significantly increase its spending prior to the pandemic, with Victoria dramatically boosting its investment plans more recently, along with Western Australia.   

In considering the broader risks around the central and upside scenarios, the main issue is the ongoing uncertainty created by COVID, where future  health outcomes are assumed to relatively benign even though another variant could emerge given very low vaccination rates in many emerging economies. There is also a large tail risk for the global economy from Russia's invasion of Ukraine, and ongoing geopolitical ructions, including the possibility of conflict in the Indo-Pacific region.

Locally, the cost of flood-related spending by New South Wales and Queensland mainly has a timing impact on state issuance insofar as the Commonwealth government will eventually reimburse 75% of state disaster payments. There is also potential for the Commonwealth to make an advance on this reimbursement depending on the outcome of the federal election, as was the case with the large natural disasters in Queensland in 2010-11.

Significantly and as mentioned above, this research does not account for potential upside revenue surprises to state budgets from stronger-than-expected consumer spending and higher-than-expected commodity prices. This will be the subject of another future note.