Editorial illustration for The Machinery of the Middle

Good evening, reader.

In today's Innovation Commons: the country has, for the third time in three years, discovered the firms it is missing — and once again proposed to fix the gap by naming it.

Let's get into it.

On 24 February, Industry Innovation and Science Australia (the government's standing advisory body on innovation and science) published Mobilising Australia's missing middle. It is a careful, unarguable paper. Medium-sized enterprises, it says, diffuse technology downstream and connect large and small firms, yet they represent a smaller share of Australia's firms and workers than OECD peers such as Canada and Germany. They are, in the paper's own framing, often overlooked. The central proposal is to identify these firms as a distinct cohort in policy and industry strategy — a first step, IISA writes, towards realising the benefits of an uplifted economic middle.

Everything in that sentence is true. The problem is the word first. This is the third time IISA has arrived at roughly the same doorstep. Its 2023 report, Barriers to collaboration and commercialisation, already flagged the missing band of mid-sized businesses. The new paper is billed as the opening instalment in a series of thought-leadership. And the mid-sized segment it describes has, on InnovationAus's reading of the paper, sat at about 2.6 per cent of Australian firms for more than two decades. A cohort that has not moved in twenty years is not waiting to be recognised. It is waiting to be built.

Take the case for the middle at full strength, because it is strong. IISA lists what mid-sized firms are: strong innovators; more productive, more export-oriented and higher value-adding than small firms; supply-chain connectors that specialise to compete at scale; more resilient to shocks; and disproportionately regional, holding scale in places where large companies are absent and distance makes it essential. The highest rate of high-growth manufacturing firms, the paper notes, sits inside the mid-sized cohort. This is the part of the economy where industrial capability actually compounds — where a good idea becomes a production line, a production line becomes an export, and an export becomes a supply chain that smaller firms can plug into. A country thin in the middle is a country that converts its inventions into someone else's industry.

So the diagnosis is settled. What is not settled is the mechanism. Recognising a cohort is a measurement. It changes a spreadsheet, not a firm. IISA is honest about this: it calls recognition essential but not sufficient. The sufficiency question is the whole question, and it is the one that keeps being deferred to future analysis. Future analysis is where Australian industrial policy goes to be admired. Recognition recurs because it is the cheapest move on the board — consensual, uncontroversial, and free of any obligation to spend or to choose. It lets everyone agree that the middle matters without anyone having to build it.

Look at the same gap from the money side and it stops being abstract. The Strategic Examination of Research and Development built its case on a single, ugly number: gross R&D spending has fallen to 1.66 per cent of GDP, against an OECD average of 2.73 per cent, down from a peak above 2.2 per cent in 2009. Business expenditure on R&D, the part that is supposed to live inside firms, has been stuck near 0.9 per cent of GDP and flat since 2017-18, on the Australian Bureau of Statistics' figures. Almost three-quarters of it comes from just three industries: professional and technical services, manufacturing, and finance. Large firms fund only 45 per cent of business R&D here, against an OECD average of 61 per cent, and their R&D has fallen by roughly a quarter over the past decade, on the Business Council of Australia's numbers.

Read those two findings together. The missing middle is the cohort where diversified, industry-led R&D would naturally sit; not the miners, not the banks, but the mid-sized manufacturer refining a process or the specialist exporter improving a product. Its absence and the collapse in business R&D are not two problems. They are one problem, described by two agencies who have not yet been asked to describe it together. The R&D Tax Incentive has cost around $3.5 billion a year and has not shifted the business R&D line. Rebating experimental development that firms were going to do anyway is not the same as building firms that do more of it.

It helps to look at a country that has the middle Australia says it wants. Germany's Mittelstand is not a classification; it is an apparatus. It is held up by dual vocational training that supplies skilled labour to firms too small to train alone, by the Fraunhofer network of applied-research institutes that lets a mid-sized manufacturer buy into development it could never staff in-house, by patient regional banks that lend on decades rather than quarters, and by a standards culture that turns a good component into an exportable one. Germany runs R&D above 3 per cent of GDP. The middle is thick there because the machinery that manufactures and holds mid-sized firms exists, and has for generations. Australia has recognised the cohort. It has not built the machine.

What would the machine look like here? Three moves, none of them a new definition.

First, demand rather than description. IISA's own strongest claim is that mid-sized firms are supply-chain connectors — so give them chains worth connecting. Government procurement, and the supply chains of large anchor customers, are the most powerful scale-up instrument the country already owns, and they are largely pointed at the incumbents. Procurement written to reward a firm for crossing from twenty employees to two hundred, rather than to minimise unit price this quarter, would do more than any cohort label. Demand is an industrial policy the country can run without a single new grant.

Second, an applied-research bridge. The missing institution in the Australian stack is the intermediary between a public research base and a mid-sized firm. A tax rebate is not intermediation. Standing up applied-research institutes that mid-sized firms can commission, at a price they can bear, is the difference between subsidising R&D and producing it. It also fixes the concentration problem: intermediation is how R&D capability spreads beyond the three sectors that currently hold three-quarters of it.

Third, patient capital and vocational depth aimed squarely at the crossing from small to medium. The twenty-to-two-hundred transition is where firms in this country stall, and it stalls for want of financing that tolerates the wait and skills that tolerate the scale. Both are institutional choices, not market accidents. If we are going to measure the cohort at all, measure the crossing rate, not the static count. A policy that cannot see the crossing cannot tell whether it is working.

There is a strategic edge here that the productivity framing understates. The mid-sized firm is the layer that holds capability in place when a shock arrives — the manufacturer that can retool, the exporter that can re-route, the regional employer that does not fold the moment a large customer leaves town. IISA calls this resilience, and it is the quiet argument for the middle in a decade of broken supply chains and contested industrial policy. A country that offshores its middle keeps its inventors and its incumbents and loses the firms in between that would have turned one into the other. Sovereignty, in practice, is a distribution of firm sizes, and ours is hollow where it most needs to be full.

None of this is exotic. All of it has been recommended somewhere, by someone, in one of the reviews Australia commissions at a rate no other outcome can match. The point of the missing middle is not that we cannot see it. We see it clearly, repeatedly, in high-quality prose. The point is that seeing has become the substitute for building. Recognition is where the last three reports stopped, and Mobilising Australia's missing middle will be judged on one thing only: whether mobilising becomes machinery, or becomes the fourth paper to recognise the problem.

A middle is not a measurement error. It is an institution the country has chosen, so far, to describe instead of assemble. The commons is built in the middle, or it is not built at all.

Editor's note: We have a soft spot for IISA, and this is a good paper making an argument we have made ourselves. The frustration is not with the finding. It is with a policy culture that treats a diagnosis as a deliverable. Watch what follows this paper, not the paper: a cohort definition that arrives with a procurement rule, an applied-research institute, or a financing vehicle attached is a start. One that arrives with a commitment to further analysis is a bookmark.

— The Editor

Sources