When the Commonwealth Government reorganised its innovation responsibilities for the fourth time in a decade, public servants made jokes about updating their email signatures again.

The humour masked a deeper problem: innovation remains everybody’s concern and nobody’s clear responsibility. This institutional restlessness reflects something fundamental: policy machinery lacks the analytical tools to handle what innovation actually is – a deeply political process that determines who prospers, who loses and how societies reorganise themselves around new technologies.

As artificial intelligence reshapes labour markets, Australian policymakers find themselves being reactive rather than strategic. Recent IMF analysis warns that up to 40% of global employment faces exposure to AI. Geographic isolation intensifies the risk: by the time policy responses crystallise, the terms of technological change may already be set elsewhere.

Political economy thinking was designed to address precisely these situations, yet it has been marginalised in Australian policy circles for decades, effectively eliminated during the neoclassical consolidation of the 1980s and 1990s. Understanding innovation and its importance to growth and productivity cannot be derived from neoclassical static equilibrium models where technological change is a residual. This requires Schumpeterian models capable of analysing the dynamic process of capital accumulation, particularly within national innovation systems.

Why standard economic tools fail

Contemporary innovation policy typically relies on frameworks borrowed from neoclassical economics. These treat technological change as an unexplained residual in growth equations, somehow accounting for 90% of economic expansion, yet remaining analytically opaque. When pressed on deployment questions, the standard response defaults to market mechanisms and corrective interventions after problems have emerged.

This market-solution approach produces short-term reactive decision-making rather than strategic shaping of technological transitions. The loss of Australian manufacturing capability after the removal of protectionism illustrates this problem. The same mistake is not being made with the transition to renewable energy.

The political dimensions of technological change are hardly new. David Ricardo recognised in 1821 that machinery deployment could be detrimental to workers’ interests. French legislation in 1623, restricting printed calicoes, triggered political unrest. The 19th century Luddites destroyed textile machinery threatening their livelihoods. These were political contests over how technological change would be governed and whose interests would prevail.

Technological transitions create winners and losers. The real issue is whether policy actively shapes these transitions or manages their aftermath. Standard economic frameworks lack the conceptual apparatus to address questions of directionality: innovation towards what purposes, benefitting whom and governed by what mechanisms?

What political economy offers

Political economy thinking recognises that markets are political constructs shaped by legal frameworks, power relationships, and institutional designs that favour certain interests over others. Technology does not arrive as an external force; it emerges from specific institutional contexts and embeds particular assumptions about how work, exchange and social relationships should be organised.

This lens changes which questions are asked. Rather than treating automation as inevitable, political economy analysis asks who benefits from framing AI deployment as automation, rather than augmentation. What institutional arrangements would favour augmentation approaches? How do procurement rules, funding criteria and regulatory frameworks channel innovation in particular directions?

The Science Policy Research Unit at Sussex University has worked on this analysis for decades. Freeman and Soete’s work on information and communication technologies showed how the organisation of work might be transformed, projecting technology as offering greater autonomy and skill development within workforces, but only if appropriate institutional frameworks are designed. Later, Soete argued that the net employment impacts of technologies like AI depend on macroeconomic compensation mechanisms, diffusion timing, and institutional factors.

South Korea provides evidence that this thinking can translate into effective policy. Its economic transition drew heavily on innovation systems analysis that understood technology as something to be actively governed rather than passively received.

The directionality challenge

This is where innovation policy becomes clearly political. Choosing to orient innovation around societal challenges, rather than merely economic expansion, requires sustained political commitment. It means redirecting resources, establishing new priorities and building coalitions around goals that do not receive the automatic political support that economic growth commands.

Eminent economist Carlota Perez argues we are approaching a situation where directional choices become unavoidable. Each techno-economic cycle passes through stages of increasing inequality before navigating a turning point towards broader prosperity. How harsh the turning point proves depends on how well public policies manage the transition through proactive structural changes.

What this means for practice

Australian policymakers face current challenges from an exposed position. Geographic isolation limits direct participation in global innovation networks. Institutional diffusion of innovation responsibility prevents concentrated capability building. The tendency to treat substantial research evidence as mere academic output delays responses to emerging challenges.

Recent CSIRO research indicates that meaningful change in policy direction requires sustained boundary-spanning collaboration between researchers, political actors and policy implementers. Without this, transformative change becomes impossible, regardless of how compelling the evidence or how urgent the challenge.

Historical precedent demonstrates that such coalitions can succeed. The OECD’s Directorate of STI facilitated learning between academics and policymakers during the 1980s through continuous interaction and joint publications. These efforts culminated in National Innovation Systems becoming the favoured OECD policy framework by 1991.

The current situation differs markedly. Academics and policymakers operate from different epistemic silos, where academic work becomes disconnected from policy needs while policy lacks analytical depth.

What made the OECD experience work was structured institutional spaces where researchers and policy insiders engaged in sustained learning. Australia now faces an opportunity to create similar conditions through political will and bureaucratic activism:

• Creating institutional anchors for innovation policy to ensure stability and clear ministerial responsibility. The pattern of reorganisation destroys institutional memory.
• Developing public service expertise in innovation systems thinking that goes beyond standard economic frameworks.
• Establishing sustained boundary-spanning engagement between researchers, policymakers and political actors.
• Making explicit political choices about innovation priorities, rather than maintaining the fiction that market mechanisms alone should determine technological directions.
• Building coalitions around mission-oriented approaches that maintain commitment across electoral cycles.
• Examining how AI and other technologies are deployed in Australian contexts, rather than accepting automation narratives imported from elsewhere.

Beyond the valley of policy death

The gap between academic research and policy implementation exists partly because researchers and policymakers operate according to different logic and face different constraints. But it also reflects a deeper problem: the analytical frameworks dominating policy discourse cannot adequately grasp what innovation is, or how it can be governed strategically.

Political economy thinking offers better tools, but these remain outside mainstream policy practice. Bringing them back requires more than publishing papers or holding conferences. It requires political action: building institutions, allocating resources, establishing new priorities and challenging existing distributions of influence.

Treating innovation as political means recognising these dynamics and working within them. It means understanding that choices regarding research funding, regulatory design, procurement rules and institutional structures are never purely technical; they inevitably prioritise some interests over others and either open or foreclose particular futures.

Australian policymakers could continue to treat innovation as a technical optimisation problem, deferring to market mechanisms and reacting to disruptions as they emerge. Or they could recognise innovation as the political process it actually is and develop the institutional capacity to govern it strategically. The first path is familiar, but increasingly inadequate. The second is challenging, but necessary.

Now is the right time for Australian politicians, policymakers and academic theorists to come together, learn and set new directions for policy. The emergence of National Innovation Systems as a major OECD policy framework, at the height of neoliberalism, demonstrates what is possible. Circumstances are now such that a new wave of policies to address societal challenges is urgently required.

The choice itself is political. Making it wisely requires better analytical tools than those provided by mainstream economics. Political economy thinking offers those tools, but only if policymakers will use them.

The authors appreciate comments by Professor Roy Green on an earlier draft of this Innovation Insight.

The views expressed in this article may or may not reflect those of Pearls and Irritations.