Reform the Market Before We Plan the Infrastructure

From Cost Recovery and Central Planning to Targeted Infrastructure Investment

Britain is creating an increasingly sophisticated system for deciding what energy infrastructure should be built and where. But what if we are doing this in the wrong order?

Britain needs an enormous amount of infrastructure investment.

Electricity demand is expected to grow as transport, heating and industry electrify. Distributed generation, batteries, EVs, heat pumps and data centres are changing where electricity enters and leaves the network.

At the same time, Britain is developing an increasingly extensive institutional architecture for planning the energy system.

The National Energy System Operator (NESO) is now responsible for a suite of strategic planning exercises including:

the Strategic Spatial Energy Plan (SSEP);
the Centralised Strategic Network Plan (CSNP);
the Regional Energy Strategic Plans (RESPs);
the transitional RESP and CSNP arrangements;
the Future Energy Scenarios (FES) that inform assumptions about future demand and supply.

The objectives are understandable.

Someone needs to think about where electricity demand might grow, where generation might be located, where networks could become constrained and what infrastructure Britain will require over the coming decades.

But I think we are in danger of doing this in the wrong order.

We are building an increasingly sophisticated machinery for planning around the electricity market before fixing the electricity market itself.

That matters because a well-designed market should already be producing some of the most important information these planning exercises are trying to discover:

Where is electricity scarce?

Where is network capacity scarce?

What is that scarcity costing us?

What would another MW of network capability at that location actually be worth?

Would building infrastructure create more value than it costs?

If the electricity market cannot answer those questions, we are asking planners, models, forecasts, consultations and institutions to reconstruct information that the operation of the physical system itself could continuously reveal.

The problem with paying for inputs

Electricity networks remain regulated natural monopolies.

That is unlikely to change.

The question is not whether networks should be regulated.

It is what regulation should reward.

The danger of any regulatory system centred around expenditure allowances is that the conversation naturally becomes:

Tell the regulator what you intend to build → justify the expenditure → negotiate an allowance → build it → recover the cost.

What we actually want is:

Identify an infrastructure deficiency → quantify its economic impact → find the most efficient way of removing it → verify the improvement → reward the value created.

Those are subtly but importantly different objectives.

A Distribution System Operator should not necessarily make more money because it builds another substation.

It should make more money because it increases the useful capability of the electricity system.

Sometimes another substation will be exactly the right answer.

Sometimes it might be:

reconductoring;
batteries;
demand flexibility;
dynamic line ratings;
voltage optimisation;
network reconfiguration;
lower-loss transformers;
improved control systems;
strategically located generation.

The regulatory framework should not unnecessarily predetermine which of these wins.

It should create incentives to find the intervention that produces the greatest system value for its cost.

The new architecture of strategic energy planning

This question is becoming increasingly important because Britain is dramatically expanding the role of strategic energy planning.

NESO's Strategic Spatial Energy Plan is intended to provide a pathway for the location, capacity and timing of electricity and hydrogen generation and storage.

The Centralised Strategic Network Plan will recommend future electricity, gas and hydrogen transmission infrastructure over a long-term horizon.

At the regional level, NESO is developing 11 Regional Energy Strategic Plans: one for Scotland, one for Wales and nine English regions.

The RESP methodology contains several components, including:

regional contexts;
pathways for future supply and demand;
consistent planning assumptions;
spatial context;
strategic investment need.

The spatial context is particularly important.

It overlays projected energy pathways with existing network capacity to identify areas where network investment may be required.

The Strategic Investment Need process then seeks to identify locations where network investment could accelerate decarbonisation, support economic development and create wider system value.

The transitional RESP is already feeding into DNO business planning for the RIIO-ED3 period from 2028 to 2033.

There is also a national transmission equivalent.

The Centralised Strategic Network Plan is intended to recommend which transmission reinforcements should proceed and when, considering infrastructure alongside alternative solutions.

These are serious attempts to solve a genuine problem.

My concern is not that Britain is doing no infrastructure planning.

It is almost the opposite.

We are building an enormous infrastructure-planning machine without first creating the market signals that would tell that machine, continuously and empirically, where infrastructure is economically valuable.

Forecasting scarcity versus observing scarcity

Consider how regional infrastructure planning must work without sufficiently granular dynamic market signals.

We forecast:

EV adoption;
heat-pump deployment;
population growth;
industrial development;
data-centre demand;
renewable generation;
storage deployment;
hydrogen demand;
economic growth.

Those forecasts become pathways.

The pathways become spatial assumptions.

Those spatial assumptions are compared against network models.

Potential constraints emerge.

Those constraints become strategic investment needs.

Eventually they become network investments.

There is nothing inherently wrong with this.

Infrastructure necessarily requires forecasting because infrastructure takes years to deliver.

But forecasts should not be the only source of information.

There is another extraordinarily valuable source:

the actual operation of the electricity system.

Every second of every day, electricity is flowing through the network.

Assets are being utilised.

Constraints are binding.

Electricity is being curtailed.

Flexibility is being activated.

Energy is being lost.

Demand is being shifted.

Some locations have abundant infrastructure.

Others have scarce infrastructure.

The physical system is continuously generating information about what it needs.

The market should reveal it.

Let the network tell us where investment is required

This becomes particularly interesting when combined with the electricity market architecture I have been developing through Enleashed.

Under an Automatic Market Maker (AMM), electricity prices continuously reflect the physical state of the network.

Prices can exist at different levels of the electricity system — from local networks through feeders and substations to wider regional and national levels.

When network capability becomes scarce somewhere in the system, the local price diverges from the price immediately upstream.

That price performs an immediate economic function.

It tells flexible demand:

Consume somewhere else, consume less, or wait.

It tells batteries:

Discharge here.

It tells generators:

Electricity is particularly valuable here.

But it tells the infrastructure planner something as well:

Network capability is particularly valuable here.

This is the crucial connection.

The market does not simply coordinate the use of existing infrastructure.

It can help reveal where new infrastructure is economically justified.

Temporary scarcity should be priced. Persistent scarcity should be built out.

Imagine a distribution feeder becomes constrained for several hours on a particularly cold evening.

That does not necessarily mean we should build another feeder.

The whole point of dynamic electricity markets is that flexible resources can respond to temporary scarcity.

But imagine the same feeder experiences substantial congestion every day.

Its local electricity price is persistently different from the price immediately upstream.

Now the market is telling us something different.

This is no longer simply temporary scarcity.

It may represent a structural infrastructure shortage.

For network location (i), define:

\Delta P_i(t) = P_i(t) - P_u(t)

where:

(P_i(t)) is the local AMM price;
(P_u(t)) is the relevant upstream price.

When the relevant network interface is unconstrained, the congestion component of this differential should be zero or small.

When the constraint binds, the prices separate.

A simple annual congestion-value measure is:

C_i^{rent} = \sum_{t=1}^{T} \left(P_i(t)-P_u(t)\right) Q_i(t)\Delta t

Suppose a feeder experiences:

an average congestion premium of £20/MWh;
10 MW of relevant flow;
congestion for 1,000 hours per year.

Then:

C_i^{rent} = 20 \times 10 \times 1000 = £200,000/\text{year}

This is useful.

But it is not necessarily the complete economic cost of the constraint.

It is closer to the observed congestion rent or scarcity value.

For investment, we can do something considerably more powerful.

Ask the counterfactual

The economically important question is:

How much better would the system perform if additional infrastructure capability existed here?

Suppose the current capacity of a feeder is:

K_i = 10\text{ MW}

Now rerun the market with:

K_i' = 11\text{ MW}

Everything else remains the same.

Define:

W(K_i)

as total annual economic welfare with the existing network.

Then:

\Delta W_i = W(K_i+\Delta K)-W(K_i)

measures the economic value created by additional network capacity.

This captures consequences including:

renewable curtailment;
unserved or shifted demand;
inefficient battery dispatch;
expensive replacement generation;
export restrictions;
consumer utility lost through rationing;
network losses;
balancing requirements.

Now we can ask a much more meaningful infrastructure question:

What is another MW of capability here actually worth?

The marginal value of network capacity

The marginal economic value of network capacity is:

MV_i = \frac{\partial W}{\partial K_i}

In practice:

MV_i \approx \frac{W(K_i+\Delta K)-W(K_i)}{\Delta K}

Suppose increasing a feeder from 10 MW to 11 MW creates:

£150,000/\text{year}

of additional welfare.

Then:

MV_i \approx £150,000/\text{MW-year}

We have transformed:

This feeder might need reinforcement.

into:

Another MW of capability at this precise location is worth approximately £150,000 per year.

That is a genuine economic investment signal.

Build the investment curve

We can repeatedly run the counterfactual:

Feeder capacity	Additional capacity	Annual welfare gain
10 MW	Existing	—
11 MW	+1 MW	£150k
12 MW	+2 MW	£280k
15 MW	+5 MW	£600k
20 MW	+10 MW	£680k

The value of additional infrastructure eventually exhibits diminishing returns.

The first few MW remove severe congestion.

Eventually:

\frac{\partial W}{\partial K_i} \rightarrow 0

Further reinforcement creates little additional value.

This gives us a natural answer to:

How much network should we build?

Not:

Build whatever a central forecast says might eventually be required.

And not:

Build until utilisation falls below an arbitrary threshold.

Instead:

Build until the marginal economic value of additional capability no longer exceeds its marginal cost.

Compare value with cost

Suppose adding 5 MW requires an £8 million reinforcement.

Convert the investment into an Equivalent Annual Cost:

EAC_i = CAPEX_i \times CRF + OPEX_i

where (CRF) is the capital recovery factor incorporating asset life and financing costs.

Suppose:

EAC_i = £600,000/\text{year}

and the counterfactual AMM simulation shows:

\Delta W_i = £1.4m/\text{year}

Then:

NetBenefit_i = £1.4m-£0.6m = £800,000/\text{year}

The investment rule is:

\Delta W_i > EAC_i

Or marginally:

\frac{\Delta W_i}{\Delta K_i} > \frac{\Delta Cost_i}{\Delta K_i}

In plain English:

Invest where the value of additional capability exceeds its cost.

A market-generated Network Investment Signal

We can now construct a continuously updating Network Investment Signal.

For a candidate intervention (x):

NIS_i(x) = W_i(x)-W_i(0)

This could produce an infrastructure dashboard such as:

Network area	Utilisation	Congestion hours	Congestion rent	Value of +1 MW	Investment signal
Feeder A	43%	20	£10k	£4k/MW-year	Low
Feeder B	71%	700	£180k	£90k/MW-year	Moderate
Feeder C	91%	2,700	£1.2m	£420k/MW-year	Very high

This is where I think the relationship with RESP becomes particularly interesting.

A RESP should absolutely contain forecasts, local knowledge and strategic objectives.

But imagine overlaying those things with a continuously generated map of:

the observed economic value of additional network capability.

The RESP would no longer begin primarily with:

What do we think will happen?

It could combine that with:

What is the system actually telling us?

RESP should sit on top of the market signal

This suggests a different architecture.

The AMM provides the bottom-up revealed signal.

RESP provides the regional strategic layer.

SSEP provides the national spatial layer.

CSNP provides the long-term transmission investment layer.

They are complementary.

But the order matters.

The market should continuously reveal the economic state of the system.

Strategic planning should then incorporate information that the market cannot know.

For example:

a government decision to build a new town;
a proposed industrial cluster;
a major data centre;
future railway electrification;
housing targets;
a new nuclear station;
offshore wind development;
defence requirements;
resilience against low-probability events.

This is where strategic planning is genuinely necessary.

The planner knows things that current market operation cannot yet reveal.

But the planner should augment the market signal, not substitute for its absence.

The danger of planning before market reform

Without this underlying economic signal, there is a danger that Britain's energy transition becomes increasingly dependent on administrative optimisation.

We forecast where generation will be.

We forecast where demand will be.

We forecast what technologies people will adopt.

We forecast how much flexibility will appear.

We forecast network requirements.

We optimise those forecasts.

We consult stakeholders.

We create strategic needs.

We approve investment.

Then consumers pay for the resulting infrastructure for decades.

Every individual step can be technically sophisticated.

But sophisticated optimisation does not remove the fundamental uncertainty of predicting the future.

There is a profound difference between:

A model predicts that this location will require infrastructure.

and:

The electricity system is persistently revealing that another MW of infrastructure at this location is worth £420,000 per year.

The first is a forecast.

The second is an economic signal grounded in the operation of the physical system.

We need both.

But we should not confuse them.

Before building more, use what we already have properly

There is another reason market reform should precede enormous infrastructure programmes.

The cheapest infrastructure is often the infrastructure we do not need to build.

Before concluding that a network needs reinforcement, we should ask whether existing assets are being used efficiently.

A properly designed dynamic market can coordinate:

batteries;
flexible demand;
EV charging;
heat pumps;
distributed generation;
local storage;
controllable industrial loads.

If flexibility can economically manage a constraint for 20 hours per year, building a new substation may be absurd.

If the constraint persists for 4,000 hours per year, reinforcement may be obvious.

The market helps reveal the difference.

This is why:

Temporary scarcity should be priced. Persistent scarcity should be built out.

Without an effective scarcity signal, we risk solving operational coordination problems with capital expenditure.

Efficiency is infrastructure too

Nor should we assume infrastructure improvement necessarily means building more infrastructure.

Electricity networks lose energy.

Define:

\eta_i = \frac{E_i^{delivered}}{E_i^{input}}

and:

L_i = E_i^{input}-E_i^{delivered}

Suppose a network receives:

1,000\text{ GWh/year}

and loses 8%.

Only:

920\text{ GWh/year}

is delivered.

If an intervention reduces losses from 8% to 5%, delivered electricity becomes:

950\text{ GWh/year}

The intervention has effectively created:

30\text{ GWh/year}

of useful electricity without constructing generation to produce another unit.

That is infrastructure value.

The AMM can value losses

Not every lost MWh has the same value.

The annual economic value of network losses can be represented as:

C_i^{loss} = \sum_{t=1}^{T} P_i(t)L_i(t)\Delta t

Losing electricity when it is worth £10/MWh is different from losing electricity behind a severe constraint when its local value is £300/MWh.

Spatial price differences can therefore contain information about both congestion and losses.

Conceptually:

\Delta P_i = \Delta P_i^{loss} + \Delta P_i^{congestion}

A persistently high downstream price may therefore be telling us:

We need more transfer capability here.

or:

We are wasting too much energy getting electricity here.

The solution might be a new circuit.

But it might instead be:

reconductoring;
lower-loss transformers;
voltage optimisation;
reactive power management;
topology optimisation;
better control.

Again, the objective should be system value, not CAPEX.

Don't reward capacity. Reward useful capability.

This suggests that even "network capacity" is slightly too narrow a concept.

Define a candidate intervention (x).

It might represent:

+5 MW physical capacity;
storage;
flexibility;
reconductoring;
loss reduction;
voltage optimisation;
dynamic line rating;
topology optimisation;
improved network control.

Then:

NIS_i(x) = W_i(x)-W_i(0)

and invest where:

W_i(x)-W_i(0) > EAC_i(x)

Now conventional reinforcement has to compete economically with every other credible solution.

That is exactly what we should want.

Turn network constraints into investable opportunities

Imagine the AMM identifies:

Feeder 17
Congestion: 2,400 hours/year
Congestion rent: £3.1m/year
Electrical losses: 6.8%
Value of relieving constraint: £6.8m/year

Instead of automatically concluding:

Build £40 million of regulated infrastructure.

publish:

Network Constraint Opportunity: Feeder 17 — solutions invited.

A DSO might propose reinforcement.

A battery developer might propose storage.

An aggregator might propose flexibility.

An engineering company might propose dynamic network management.

Another provider might propose lower-loss equipment.

The question becomes:

What is the cheapest reliable way of eliminating this economically significant infrastructure deficiency?

This creates something approaching a market for solving network bottlenecks.

Reform the DSO business model

The final part of the architecture is to change how network companies make money.

I would retain a regulated base return.

Infrastructure operators need predictable revenues to maintain, finance and replace critical assets.

But additional returns should increasingly depend on measurable infrastructure performance.

Create an Infrastructure Value Dividend:

IVD = w_AA + w_RR + w_CC + w_EE + w_LL + w_FF

where:

(A) = availability;
(R) = reliability;
(C) = economically valuable congestion reduction;
(E) = efficient utilisation;
(L) = loss reduction;
(F) = future readiness.

Suppose an intervention reduces the annual economic cost of a deficiency from:

£10m \rightarrow £2m

Then:

\Delta W=£8m/\text{year}

A proportion could become a performance return:

Reward_i=\alpha\Delta W_i

If:

\alpha=0.2

then:

Reward_i = £1.6m

The DSO now has a strong commercial reason to find the cheapest way of creating the improvement.

It does not necessarily make more money by building more.

It makes more money by making the system better.

Strategic planning still matters

None of this means abolishing RESP, SSEP or CSNP.

Quite the opposite.

Long-lived infrastructure inevitably requires anticipatory planning.

If a major industrial cluster will connect in five years, waiting until the network becomes congested would be ridiculous.

The same counterfactual framework can therefore operate prospectively:

\mathbb{E} \left[ W_{future}(x)-W_{future}(0) \right]

Strategic planners can introduce information about expected future developments and calculate the expected value of infrastructure under different scenarios.

But now there are two sources of evidence:

Revealed need

The operating market continuously reveals existing infrastructure scarcity.

Anticipated need

RESP, SSEP and CSNP incorporate credible information about future developments.

The infrastructure plan should reconcile the two.

That is much stronger than relying principally on either one.

A closed-loop infrastructure system

The overall architecture becomes:

\mathrm{Physical\ System} \rightarrow \mathrm{AMM} \rightarrow \mathrm{Economic\ Signal} \rightarrow \mathrm{Investment\ Signal} \rightarrow \mathrm{RESP/SSEP/CSNP} \rightarrow \mathrm{Investment} \rightarrow \mathrm{Physical\ System}

The loop closes.

The physical system generates information.

The market converts physical scarcity into economic signals.

Persistent signals reveal potential infrastructure deficiencies.

Counterfactual analysis quantifies the value of possible interventions.

Strategic planning adds information about the future that current operation cannot reveal.

Investment follows where expected value exceeds cost.

The resulting infrastructure changes the physical system.

Prices change.

The investment signal changes.

Planning updates.

This is much closer to a feedback-controlled infrastructure system than a sequence of periodic administrative planning exercises.

Water demonstrates the same principle

The same philosophy applies beyond electricity.

Suppose a water network introduces:

1,000\text{ ML/day}

but only:

750\text{ ML/day}

reaches legitimate consumption because of leakage.

There are two ways of creating another 100 ML/day.

Build another 100 ML/day of supply.

Or recover 100 ML/day currently being lost.

Economically, those options should compete.

Define:

\eta_i^{water} = \frac{Water_i^{delivered}} {Water_i^{introduced}}

and:

L_i^{water} = Water_i^{introduced} - Water_i^{delivered}

The value of reducing leakage should include avoided:

abstraction;
treatment;
pumping;
storage;
transfer;
environmental costs;
scarcity costs.

If recovering 100 ML/day costs £200 million while creating 100 ML/day of new supply costs £1 billion, regulation should make leakage reduction extraordinarily attractive.

But neither should we pursue zero leakage regardless of cost.

The rule remains:

\text{Marginal value recovered} > \text{Marginal cost of recovery}

From infrastructure planning to infrastructure intelligence

Britain does need strategic energy planning.

But we should be careful not to mistake more planning for better information.

RESP, SSEP and CSNP are attempts to answer extraordinarily difficult questions about what Britain's energy system will need decades into the future.

We should give those institutions the best information possible.

That means reforming the market beneath them.

A properly designed network-aware electricity market should continuously tell us:

where electricity is scarce;
where network capability is scarce;
how persistent that scarcity is;
how much congestion is costing;
where losses are economically significant;
how valuable another MW of capability would be;
whether flexibility can solve the problem;
whether reinforcement creates more value than it costs.

Strategic planners can then concentrate on what markets cannot know:

the future.

New towns.

Industrial strategy.

Future generation.

Defence requirements.

Major infrastructure projects.

Long-term resilience.

Political choices about the kind of country we want to build.

That seems a much more sensible division of labour.

Reform the market before spending hundreds of billions around it

The danger otherwise is that Britain spends enormous sums building infrastructure around the deficiencies of the existing electricity market.

We could end up simultaneously:

overbuilding some parts of the network;
underbuilding others;
curtailing renewable generation;
paying for flexibility;
failing to exploit flexibility properly;
accepting unnecessary network losses;
reinforcing constraints that better coordination could solve;
and asking consumers to finance all of it.

The answer is not to stop infrastructure investment.

The answer is to make the economic value of infrastructure visible.

The principle for regulated utilities should be:

Infrastructure owners should not primarily earn more because they spend more. They should earn more when they create more useful, efficient, resilient and economically valuable infrastructure capability.

The principle for infrastructure planners should be:

Forecast what the market cannot yet know. Measure what the physical system can already tell us.

And the investment rule becomes:

\text{Invest where the value created by an intervention exceeds its cost.}

Sometimes that means building a cable.

Sometimes a substation.

Sometimes storage.

Sometimes flexibility.

Sometimes reducing losses.

Sometimes better control.

Sometimes doing nothing.

The objective is not to maximise infrastructure investment.

It is to maximise the useful capability and economic value of the infrastructure system.

That gives us two very simple rules:

Temporary scarcity should be priced. Persistent scarcity should be built out.

And:

Wasted resources should be recovered wherever the value of recovering them exceeds the cost.

RESP, SSEP and CSNP could then become substantially more powerful.

Rather than attempting to determine the future infrastructure system principally from forecasts, scenarios and administrative optimisation, they could sit on top of a continuously operating economic feedback mechanism.

Fix the market. Let the market reveal the system. Then plan the infrastructure around what it tells us.

Reforming the Utility Business Model: From Cost Recovery to Targeted Infrastructure Investment

Reform the Market Before We Plan the Infrastructure

From Cost Recovery and Central Planning to Targeted Infrastructure Investment

The problem with paying for inputs

The new architecture of strategic energy planning

Forecasting scarcity versus observing scarcity

Let the network tell us where investment is required

Temporary scarcity should be priced. Persistent scarcity should be built out.

Ask the counterfactual

The marginal value of network capacity

Build the investment curve

Compare value with cost

A market-generated Network Investment Signal

RESP should sit on top of the market signal

The danger of planning before market reform

Before building more, use what we already have properly

Efficiency is infrastructure too

The AMM can value losses

Don't reward capacity. Reward useful capability.

Turn network constraints into investable opportunities

Reform the DSO business model

Strategic planning still matters

Revealed need

Anticipated need

A closed-loop infrastructure system

Water demonstrates the same principle

From infrastructure planning to infrastructure intelligence

Reform the market before spending hundreds of billions around it

Comments (0)