Measuring industrial policy: key KPIs for Kazakhstan’s growth

Industrial policy in Kazakhstan has been living an active life for years. Strategies get refreshed. Programs get relaunched. Presentations look sharper. Reported numbers go up. Yet one question stays oddly quiet: how do we measure results? Not budget absorption. Not opening a facility. Not the number of signed MoUs. But real economic impact – productivity, exports, technological upgrading, private investment, and tax returns.

Without a clear answer, the industrial policy debate quickly turns into a contest of nice slides.

The trend of recent years is simple. Governments worldwide are returning to active industrial policy. The reasons vary: geopolitics, supply-chain shocks, competition for technology, fear of losing production and jobs. China became the most visible example of this approach: massive, aggressive, and uneven in outcomes. And that unevenness is exactly the point for Kazakhstan. It shows something basic. Industrial policy can deliver growth in some sectors and fail in others at the same time. What matters is not the slogan or the volume of money. What matters is policy design and measurement.

China increased its share in global manufacturing for decades. In some industries this led to dominance: shipbuilding, EVs, solar panels. In others, chronic problems emerged: overcapacity, weak investment returns, projects that never reached the promised technology level despite billions in support. Wherever analysts looked deeper, one detail kept showing up. Higher output and higher sales do not automatically mean higher efficiency. Not higher productivity. Not more innovation. Not stronger business resilience.

This is where Kazakhstan should be careful. Industrial policy here is often judged with binary logic: launched or not, spent the budget or not, met the plan or not. That logic is convenient for reporting. But it misses real outcomes. A factory can run. People can be employed. Output can rise. Yet productivity can stagnate. Exports may never appear. Private capital stays away. Technology levels do not change. In this mode, industrial policy starts living its own life: formally “successful”, economically empty.

A key problem is indicator choice. When the state backs an отрасль, it usually declares several goals at once: output growth, import substitution, exports, technology upgrading, employment. These goals often conflict. You can raise output quickly with subsidies and tax breaks, but kill incentives to improve efficiency. You can preserve jobs, but lock in low productivity. You can grow sales without profitability.

The practical rule for Kazakhstan is simple: one indicator proves nothing.

• ⁠ ⁠Output growth without productivity growth is a warning sign.
• ⁠ ⁠Export growth without higher value added is also a red flag.
• ⁠ ⁠R&D spending growth without products and revenue is another.

A KPI framework must be multi-dimensional and strict. It must make it impossible for a project to be called successful on one metric while failing completely on others.

Another lesson is the mismatch between central and regional incentives. In China, the center often pursued strategic goals: technology leadership, global markets. Regions focused on immediate targets: jobs, local GDP, today’s tax base. As a result, the same programs were executed differently. Sometimes competition and efficiency were strengthened. Sometimes capacity was simply expanded. That produced distortions: excess investment, falling profitability, market pressure.

In Kazakhstan, this tension can be even sharper. Regions want a visible launch: a facility, a ribbon-cutting, a report. The center wants strategy delivery. In the end, KPIs shift toward form, not substance. If evaluation does not include direct indicators of productivity, export revenue, tax returns without privileges, and private investment alongside state support, the outcome is predictable. Projects will optimize for reporting.

China’s experience highlights another point. State support works better when it reinforces market dynamics, not when it substitutes for them. Where support came through demand, infrastructure, and standards, results were more durable. Where the bet was on direct production subsidies, effects were often short-term. Firms grew while money flowed, then faced reality: competition, oversupply, collapsing margins.

For Kazakhstan, that means KPIs must track not only the volume of state support, but private-sector behavior. If an industry survives only because the state feeds it, it will show up clearly:

• ⁠ ⁠no private investment next to public money
• ⁠ ⁠exports sustained by administrative measures
• ⁠ ⁠low tax returns

All of this is measurable. The question is whether we choose to measure it.

Innovation is its own trap. In China, R&D spending figures rose. But deeper analysis showed that higher spending did not always produce technological outcomes: patents did not grow, products did not reach markets, productivity did not move. For a KPI system, this is decisive. Innovation without market outcome is just process. If the evaluation framework cannot separate process from results, money disappears into sand. For Kazakhstan, any “technology bet” should come with hard market-facing KPIs: export, revenue from new products, share of private orders. Without that, innovation policy becomes a showcase.

Another lesson concerns state-owned firms. In China, they often received more support, more privileges, and easier financing. Yet in many sectors they lagged private firms in efficiency. That created a double effect: rapid scaling, but falling overall sector productivity. A KPI framework must be able to detect this – not by ownership type, but by performance. For Kazakhstan, this matters because state and quasi-state players are central to industrial policy. Without strict KPIs, they can become anchors. With them, they can become drivers. The difference is measurement.

China’s experience also shows that the early years of almost any support program look successful. Growth. Launches. Investment. Problems appear later: when the market saturates, subsidies shrink, and firms must compete without crutches. A KPI system must look beyond the first year. It should evaluate years three, five, and seven – and be ready to record not only success, but the need for the state to exit. In Kazakhstan, this rarely happens. Projects are either labeled successful and live forever, or quietly frozen without public analysis. A KPI framework changes that. It turns industrial policy from a set of initiatives into a managed portfolio.

The most unpleasant part about KPIs is that they remove comfort. They expose mistakes. They make it harder to hide behind vague language. That is why they face resistance. But without them, industrial policy becomes a debate about presentations: whose slides are brighter, whose forecast is more optimistic, whose numbers are bigger. China’s experience is useful not because it can be copied, but because it shows the cost of mistakes. The cost is not only money. It is lost time, lost opportunities, and lost business trust.

Kazakhstan is at a moment when industrial policy is again in focus. This is the right time to ask a simple question: which numbers will prove the policy worked, and are we ready to admit when it did not? Industrial policy without KPIs is belief. With KPIs, it is management. The choice between them shapes not only program effectiveness, but the quality of economic governance as a whole.