Capital intensity and the UK's productivity problem

The Office for National Statistics' May 2025 market-sector productivity bulletin described capital services per hour after 2008 as weak by historical standards. This is a dated statistical assessment, not a claim that every subsequent release or industry followed the same pattern. Its significance is the distinction it makes between capital deepening, workforce composition, and multifactor productivity.

The Bank of England's February 2026 Monetary Policy Report likewise treats capital deepening and total factor productivity separately in its discussion of potential productivity. Its outlook is a conditional projection. It should not be read as an observed improvement already secured by investment announcements. Both sources direct attention toward the services that assets can provide rather than investment expenditure alone.

The ONS quality and methodology report explains that capital services are a flow into production, while the capital stock is a wealth measure at a point in time. Its market-sector coverage excludes general government and nonprofit institutions serving households, as well as imputed rental. The coverage matters when comparing these estimates with whole-economy output per hour or public-service productivity.

The same release's Table 1 provides a concrete illustration for 2024: market-sector capital services increased by 0.9%, while hours worked increased by 1.8%; capital deepening fell by 0.9% on the release's growth-accounting basis. Output per hour fell by 0.5%. The coexistence of rising capital input and falling capital intensity is therefore an observed feature of this specified vintage, not merely a theoretical possibility. These estimates carry the source's statistical-development designation and subsequent correction notice.

Gross investment adds assets, but some of that expenditure replaces equipment and software that have worn out or become obsolete. An economy can undertake substantial gross investment without a correspondingly large increase in productive capital. A longer-lived structure and a rapidly obsolescing digital asset also require different patterns of renewal.

A simple illustration makes the distinction visible. Suppose an enterprise begins with productive assets equivalent to 100 service units and loses 8 units through retirement or deterioration. New investment adds 10 usable units. The net increase is 2 units, not 10. If hours worked increase by 4%, capital services per hour can still decline. These units are hypothetical and do not represent an estimate of the UK capital stock.

The example also shows why asset prices should be separated from quantities. A higher price for the same machine increases nominal investment but does not by itself add capacity. A more capable machine can provide additional services even if its purchase price is unchanged. Measuring the difference requires appropriate price and quality adjustment, with assumptions about the asset's productive life.

Capital intensity depends on hours worked as well as capital. When a business responds to demand by hiring more workers while postponing equipment purchases, output can expand without capital deepening. When hours fall suddenly, measured capital per hour can rise without any new investment. Neither movement, taken alone, establishes a durable change in productive efficiency.

This denominator issue is particularly important when labor-market measurement is uncertain or when the industry mix changes. An increase in the share of employment in relatively labor-intensive activities can reduce aggregate capital services per hour even if individual industries maintain their own capital intensity. Within-industry changes and movements between industries should therefore be examined separately.

A fall in measured productivity can also arise when firms retain staff during a temporary demand slowdown. The resulting underutilization is different from a permanent loss of technical capability. Yet persistent weak demand can eventually affect investment and skills, so the cyclical and structural explanations are connected. Distinguishing them requires a time series and information about capacity utilization, rather than one quarter's output-per-hour figure.

Growth accounting adjusts labor input for differences in the composition of the workforce. More hours supplied by groups with different observed labor compensation alter the estimated labor input, even if total hours are unchanged. This adjustment does not mean that compensation perfectly measures every worker's productive contribution. It is a modeling convention that makes the accounting more informative than a headcount alone.

After measured capital and labor inputs are accounted for, multifactor productivity is the remaining part of output growth. It can reflect technological progress, management, organization, utilization, measurement error, and factors not captured in the inputs. Describing the entire residual as innovation would therefore overstate what has been identified.

For the UK debate, this matters because an investment policy and a skills policy can affect several components simultaneously. New software may be ineffective without training and changes to working practices. Better management can raise the return on existing equipment. A measured rise in the residual following investment does not prove that capital spending and organizational improvement are independent alternatives.

The composition behind these measures deserves attention. A warehouse, a laboratory, software, and a vehicle supply different services. Their contribution depends on the production process. There is no general economic rule that a larger quantity of one asset should compensate for a shortage of another. Additional servers may do little for a firm whose binding constraint is reliable electricity or access to skilled engineers.

Intangible assets also complicate interpretation. Some expenditure generates reusable knowledge or software; other expenditure is a current operating input. The national accounts classification and the firm's financial reporting may differ. A comparison across businesses or countries should establish which assets are included before attributing a measured investment gap to behavior rather than coverage.

For an enterprise, additional capital is attractive when its expected contribution to revenue or cost reduction exceeds the full acquisition and operating cost. Financing conditions affect that calculation, but so do demand, installation delays, regulation, and the availability of complementary inputs. A lower tax cost of investment can improve the calculation without resolving a constraint on electricity connections or specialist labor.

The expected utilization of the asset is especially important. An automated production line requires sufficient throughput to justify its fixed cost. A demand outlook with high uncertainty can favor incremental improvements to existing equipment over a large new facility. This response may be commercially rational even while aggregate investment remains below a policy ambition.

Public policy can improve the environment for these decisions through reliable infrastructure, consistent rules, competition, and access to relevant skills. The appropriate instrument depends on the constraint. A subsidy intended to correct a financing problem may have little effect where projects are delayed by planning or where the expected market is too small. Evidence about the constraint should precede the choice of support.

A convincing improvement would show up across several linked observations: a stronger flow of real asset acquisition, an increase in capital services relative to hours, improved utilization, and ultimately more output from the combined inputs. The sequence need not be immediate. Installation, learning, and organizational change can postpone measurable gains, while disruption during installation can temporarily reduce output.

Evaluation should therefore follow cohorts of investment as well as aggregate spending. It should ask whether the equipment is operating, whether employees can use it, and whether improvements persist after initial demand conditions change. An asset purchase is an input to this process, not the final productivity outcome.

The UK's productivity problem cannot be resolved by selecting a single aggregate as the definitive explanation. Capital intensity is a necessary part of the diagnosis because it distinguishes investment spending from the resources available to each hour of work. Its interpretation becomes stronger when labor composition, industry change, utilization, and measurement are considered alongside it.

Installation can also draw skilled employees away from ordinary production before an asset supplies additional services. Comparing the purchase date with that period's output can consequently understate the eventual contribution, while ignoring training and conversion work can overstate the immediate benefit. A useful investment cohort records the date of acquisition, the transition period, and normal operation separately. This connects the UK capital-intensity diagnosis with the business process through which newly acquired assets become usable inputs to production.

The ONS bulletin cited above is retained as a specified historical release. This study does not present its estimates as a newly extracted 2026 series. The Bank's projection is identified as a projection, and the capital arithmetic is hypothetical. For the public investment decisions that can complement private capital, see Public investment and the British fiscal framework.