Published in Quadrant (Sydney), 29 October 2025
We all know the Greek myth about Sisyphus, condemned to roll his boulder endlessly uphill, only to watch it tumble down each time he nears the summit. It comes to mind when I look at Louise Upston, the minister responsible for welfare (or “social development” as we call it in New Zealand).
Few people have a soft spot for any minister of welfare. Almost by definition, they have only enemies. Reformers are attacked by the Right for being too slow and lenient, while the Left call them heartless ideologues. Bleeding-hearts get the same criticism in reverse.
I have a soft spot for Upston, though. I have known her for a long time. She wants to do the right things. She prepared meticulously for this job over long and tough years in opposition, which makes it hard to watch as she struggles to bring New Zealand’s welfare payments under control.
I first met Upston when she was a minister in Bill English’s government. Over the years, particularly during her time in opposition from 2017 to 2023, we talked about welfare reform. She was never one of those politicians who simply bided their time until the electoral wheel turned. She read voraciously, studying international models, and asked sharp questions about what did and didn’t work.
When the National-led coalition returned to office in late 2023 and Upston became Minister for Social Development and Employment, I was optimistic. She had done the homework, understood the evidence, and genuinely cared about moving people from welfare dependency into fulfilling work.
Two years later, the numbers tell a sobering story. Working-age New Zealanders on main benefits reached 406,000 by mid-2025, up more than 25,000 from the year before. New Zealand’s main benefits include Jobseeker Support (for those seeking work or with temporary health issues), Supported Living Payment (for those with or caring for someone with long-term conditions that severely limit work, or their carers) and Sole Parent Support. (There are similarly named programs in Australia, but there are differences in their precise design, so do not rush to direct comparisons.)
Yet this is not a tale of ministerial incompetence; it is something more troubling, a demonstration of how quickly a welfare system can be structurally engineered to foster dependency and how difficult it is to reverse once entrenched.
For Australia, it offers a stark cautionary tale. When Labour took office in late 2017, approximately 280,000 working-age New Zealanders received main benefits. Six years later, that figure stood above 351,000, a 25 per cent increase.
The Covid pandemic accelerated the trend, but what made it structural rather than merely cyclical was a deliberate policy guided by the 2019 Welfare Expert Advisory Group, which set out to transform a system focused on work activation into one prioritising “dignity” and “adequacy”.
The mechanics of this transformation deserve close attention. The intention, at least to its supporters, was laudable: to lift incomes and reduce hardship through the largest increases to base benefit rates in a generation.
But the truly consequential change was legislative: in 2020, benefits were indexed to annual wage growth rather than inflation. This meant the real value of benefits rose over time, systematically narrowing the gap between welfare and low-wage work. Benefits became relatively more attractive than work. (The Luxon government has reversed this, indexing benefits to inflation again.)
The Ardern government also removed many of the conditions for receiving welfare. Financial sanctions for failing to meet work obligations fell 58 per cent between 2017 and 2023 as the result of an explicit ministerial direction to use sanctions “sparingly”. The message to beneficiaries and frontline staff was unmistakable: mutual obligation was no longer a priority.
The administrative culture shifted accordingly. Case managers, already burdened by soaring demand for hardship assistance due to New Zealand’s acute housing crisis, diverted their time from employment-focused work to crisis management. By 2019, well before the pandemic, employment-focused engagement with clients had fallen to roughly one-third of the rate five years earlier.
The results were predictable and dispiriting. Most telling was the decoupling of benefit numbers from the official unemployment rate. During 2021 and 2022, as the economy recovered strongly, and unemployment fell to a record low of 3.2 per cent, benefit numbers remained stubbornly elevated. In a labour market crying out for workers, with businesses reporting acute shortages across sectors, hundreds of thousands remained on welfare. This was not a story of insufficient jobs. It was evidence of a system that had become “sticky”—easier to enter and less urgent to leave.
The long-term fiscal implications became clear through actuarial modelling. By 2023, it was estimated that the average time a person would spend on a benefit in their lives had risen to 13.4 years, and for people aged sixteen to twenty-four, it was 20.4 years. These were not temporary safety net users but individuals who faced the prospect of spending a significant portion of their working lives dependent on state support.
This shambles was Upston’s inheritance. Her response was “Welfare that Works”, a textbook activation policy focused on incentives and requirements to move recipients into work, anchored by a target to reduce Jobseeker Support recipients by 50,000 by 2030. A “traffic light” system tracks compliance with work obligations. Beneficiaries must reapply every six months rather than annually. Financial sanctions have been reinstated with vigour, more than doubling in late 2024. New non-financial penalties, including mandatory community work and restricted payment cards, provide alternatives to simply cutting payments.
These are rational, evidence-informed policies. International experience demonstrates that well-designed activation measures can reduce welfare dependency. Yet the numbers continue to climb.
Here lies the cruel irony. Upston’s policies are having measurable behavioural effects. Sanctions are up 126 per cent (July to December 2024 compared with the same period a year earlier). Exits to work rose 22 per cent over the same period. By any intermediate metric, the reforms are working.
But exits are overwhelmed by new claimants. In the March 2025 quarter, 23,268 people left benefits to take up work, but 39,618 new claimants received Jobseeker Support. The minister is overseeing a churn of beneficiaries, pushing people off benefits more quickly without reducing the stock because the influx is too large.
She thus faces a double bind. The first constraint is cyclical and deeply unfortunate in its timing. New Zealand’s economy entered a protracted downturn in 2023. Per capita GDP has fallen 4.6 per cent since mid-2022. Unemployment climbed from 4.0 per cent to 5.2 per cent, with Treasury forecasting it may peak around 5.4 per cent.
Upston is applying pro-cyclical policy (increasing pressure on beneficiaries to find work) in a counter-cyclical moment (when jobs are scarce and competition is intensifying). This is not a recipe for immediate success, however sound the underlying framework.
The second constraint is structural and even more intractable. The composition of the caseload has undergone a fundamental change. The number of recipients on Supported Living Payment—New Zealand’s equivalent of Australia’s Disability Support Pension—has grown relentlessly from about 93,000 in 2018 to over 105,000 by mid-2025. Disabled recipients (as opposed to carers) grew from about 84,000 to over 96,000.
Worse, the number of recipients on Jobseeker Support specifically because of health conditions has surged. By June 2025, there were 95,178 people in this subcategory, a 15.4 per cent year-over-year increase. These individuals exist in a policy netherworld; medical certificates state they cannot work, but they do not meet the stringent criteria for the Supported Living Payment, leaving them subject to work obligations they may genuinely be unable to meet.
This represents a slow-moving public health crisis manifesting as a welfare problem. The causes are multiple: an ageing population, rising mental health diagnoses, and the accumulation of complex chronic illnesses. But activation tools designed for work-ready job seekers are increasingly being applied to a caseload with severe barriers to employment.
Upston’s work-first approach is struggling to address this structural shift. She can sanction non-compliance and mandate job searching, but if a growing proportion of recipients are genuinely incapable of work, no amount of administrative pressure will move them into sustainable employment.
Could alternative approaches fare better? International evidence suggests promising avenues. Individual Placement and Support programs, implemented in Britain and Australia, integrate employment services with mental health treatment, achieving employment rates double or triple those of traditional approaches. The partial capacity framework used in the Netherlands moves beyond binary “fit” or “unfit” determinations to calculate remaining earning capacity and tailor obligations accordingly. But these sophisticated interventions require significant upfront investment, institutional capacity and time to build skilled workforces, and would face the same hostile economic environment and structural caseload challenges.
Even the most brilliant imported innovation cannot conjure up jobs in a recession or restore work capacity to those with genuine long-term health conditions. This is the essence of Upston’s predicament. She is unquestionably capable, armed with evidence-based policies and working diligently with clear intent. Yet success or failure is not primarily within her control. She is hostage to the economic cycle, to the legacy of policies that made welfare more attractive and less conditional and to public health trends, gradually transforming the caseload.
The tragedy, in the classical sense, is that doing the right things at the wrong time, in the wrong circumstances, yields disappointing results.
For Australian policy-makers, the lesson should be sobering. Once a welfare system’s core logic shifts from work towards unconditional accommodation, the damage compounds quickly. The financial incentives change. The administrative culture adapts. The caseload grows, not just in size but in duration and complexity. When a subsequent government attempts to restore mutual obligation and work focus, the reversal is exponentially harder than the original liberalisation.
Prevention, it turns out, is much easier than cure. A government that weakens the link between welfare and work, reduces enforcement of mutual obligation, and indexes benefits to grow faster than inflation, may increase incomes and reduce hardship in the short term, but it leaves successors facing the near-impossible task of re-establishing work incentives in a system structurally recalibrated against those goals.
Upston may yet succeed if economic conditions improve. A strong recovery, falling unemployment and renewed employer demand would create the favourable environment her policies need to demonstrate their full effect. But the fact that her success depends so heavily on forces beyond policy control is the cautionary tale.
Reflecting on our conversations in opposition, she was determined to move people from dependency to self-sufficiency, to break cycles of intergenerational welfare and prove that activation policies grounded in mutual obligation could transform lives. I am sure she still believes this, and she is right to believe it. What New Zealand demonstrates is not that welfare reform is impossible, but that the window for reform can close quickly. Wait too long and the task becomes Sisyphean. This tragedy is not a result of personal failure. It demonstrates that even capability, preparation and sound policy struggle against the gravitational pull of structural reality that, once established, proves remarkably resistant to change, however much we might wish otherwise.
