Why infrastructure regulators should rethink efficiency modelling

In summary

• Efficiency models used by regulators to assess infrastructure providers’ performance are being undermined by modern-day realities  
• Their assumptions fail to reflect the challenges of decarbonisation, digitalisation, climate resilience, cyber risks and ageing infrastructure 
• Practice varies across UK regulators, and most show signs of a more multidimensional approach to efficiency modelling  
• There is much further to go, however, to develop modelling approaches that reflect the requirements of the modern world 

Efficiency modelling has always been at the heart of UK economic regulation. It determines allowed revenues, shapes incentives and influences how companies invest, maintain assets and deliver service quality.  

But the assumptions underlying the traditional toolkit risk being outdated. Econometric benchmarking to compare company performance, frontier shift factors used to estimate how the most efficient companies are likely to reduce their costs over time, and comparative cost analysis used to judge value for money: All were built for a world of stable technologies, predictable demand and incremental improvements. That world has changed.  

Across water, energy, telecoms, transport and post, regulated sectors face structural pressures: decarbonisation, digitalisation, climate resilience, cyber risk and a need to replace ageing infrastructure. These forces reshape cost structures in ways that legacy models cannot capture.  

The result is a widening gap between regulatory assumptions and operational reality. 

The standard econometrics methods, such as OLS (ordinary least squares) and COLS (constrained OLS) regressions, stochastic frontier analysis (SFA), data envelopment analysis (DEA) and simple productivity adjustments, rest on assumptions that are increasingly hard to defend: 

• The homogeneity of firms – Benchmarking assumes companies face similar operating environments, but in the water industry, for example, legacy asset conditions vary dramatically. Similarly, in telecoms, build density and topology drive structural cost differences. These variations are not noise; they are fundamental drivers of efficient cost, the cost justified by real operating conditions 
• Stable production technology – Frontier shift factors assume incremental productivity improvements, but sectors like water and energy are undergoing step changes in environmental obligations, digitalisation and resilience requirements 
• Short-term optimisation – Traditional models reward cost minimisation in the current control period, even when long-term asset health or resilience would justify higher expenditure 
• Data sufficiency – Many sectors have small samples. There are only ten UK water companies, six distribution network operators groups for electricity, and a handful of airports. This makes econometric inferences on the available data fragile and sensitive to specification choice 

In fact, the evidence suggests regulators recognise this and are already shifting their approach, experimenting with new methods. 

Ofwat continues to rely heavily on econometrics but has introduced cost adjustment claims and resilience metrics to capture structural differences, for example. Ofgem increasingly uses engineering-based assessments for net zero investments, recognising that historical data cannot predict future efficient costs. 

Ofcom, meanwhile, has moved away from cost benchmarking for fibre networks, focusing instead on competitive dynamics and investment incentives, while ORR and CAA use hybrid approaches. These combine benchmarking with bottom-up engineering analysis and long-term asset stewardship 

Progress varies by sector, but the direction of travel is clear: Efficiency modelling is becoming more multidimensional, with greater emphasis on engineering evidence, long-term value and risk-adjusted cost assessment. 

Moving forward, a modern framework should aim to integrate economics, engineering and risk analysis. Three components are essential. 

Regulators need models that reflect the underlying drivers of cost: asset age, climate exposure, network topology, service quality and resilience requirements. Structural models can incorporate nonlinearities and interactions that traditional econometrics cannot capture. 

Efficiency should not be measured only by current period cost. A company that underinvests in maintenance may appear efficient today but impose higher costs tomorrow. Regulators need tools to measure asset health trajectories, resilience benefits, long-term cost avoidance and whole-of-life cost optimisation.  

Instead of a single deterministic model, regulators should use scenario-based benchmarking that tests efficiency under different assumptions about demand, climate, technology and risk. This approach is common in financial regulation but underused in utility regulation. 

A comparative view across UK regulators reveals divergent philosophies. Ofwat emphasises econometrics and frontier shift; Ofgem emphasises engineering evidence and risk; Ofcom emphasises competition and investment incentives; and the CAA and ORR emphasise hybrid models and long-term asset stewardship. 

These differences reflect sectoral characteristics, but they also reflect methodological choices. A future-proofed approach would blend the strengths of each. 

The next decade of UK regulation will be defined by how well regulators adapt their efficiency frameworks to a world where costs are increasingly driven by resilience, decarbonisation, digitalisation and long-term asset stewardship. Traditional benchmarking will remain part of the toolkit, but it can no longer be the backbone of regulatory decision-making.  

Several strategic questions require consideration: 

• How should efficiency be defined when long-term value matters more than short-term costs? Regulators must distinguish between legitimate investment in resilience or asset health and inefficient overspend. This requires integrating asset management science with regulatory economics 
• How should regulators treat heterogeneity when geography, climate exposure, network topology and legacy asset condition are structural cost drivers? Efficiency models must reflect this rather than penalise companies for factors outside their control 
• What is the right balance between econometrics and engineering evidence? Econometric models provide comparability; engineering evidence provides realism. A mature regulatory framework needs both, with transparent rules for how they interact 
• How should regulators incorporate uncertainty and risk? Climate change, cyber threats and technology transitions create asymmetric risks. Scenario-based benchmarking and risk-adjusted cost assessment should become standard tools 
• How should regulators ensure incentives support long-term outcomes? Price controls that reward short-term cost savings at the expense of asset health or resilience are economically inefficient. Incentive design must evolve to reward behaviours that create durable value 
• How should regulators maintain legitimacy and predictability? As methodologies become more complex, transparency becomes more important. Regulators need to explain not just the results of their models, but the economic logic behind them 

The UK’s regulatory model has always been adaptive, but it faces an unprecedented challenge. The pressures on modern networks are environmental, technological, financial and political. To address these, efficiency modelling must evolve from a narrow statistical exercise into a broader economic assessment of long-term value, risk and resilience.  

Regulators that do not evolve their modelling approaches will see increased appeals, weaker investment signals, and price controls that become progressively harder to administer and defend. Regulators that embrace the shift, however, will be better equipped to deliver outcomes that are efficient, sustainable and aligned with the public interest.