Author: Houman Asefi
Date: 2026
Externalities are the simplest explanation for why “private choice” fails to deliver “social optimality.” They are also the most underestimated governance problem of modern technological economies, where innovation is no longer a sectoral event but an ecosystem shock. This essay reconstructs, in full lecture-scale detail, a canonical introduction to externalities—negative and positive; production and consumption—using the standard marginal framework, welfare geometry, and the logic of government intervention. It then extends that foundation into a technology-agnostic mathematical model suitable for AI-era, quantum-era, and neurotech-era disruption. The core thesis is institutional: superior societies and firms are those that operationalize ownership of spillovers—across time horizons, across organizational layers, and across stakeholder boundaries.
The starting point is blunt: markets are not automatically optimal. The “celebration” that often ends basic supply-and-demand teaching—where the intersection of demand and supply is treated as efficient—depends on assumptions that can be violated by a small number of failures. Externalities are among the most important of these failures because they strike at the core condition required for market efficiency: that decision-makers face the full marginal consequences of their actions.
An externality exists whenever:
The action of one party affects another party (making them better off or worse off),
and the first party does not bear the full costs or receive the full benefits of that effect.
This is not a moral statement. It is an accounting statement. Externalities occur because the price system fails to transmit some costs/benefits to the decision-maker. The result is a mismatch between what private agents maximize and what society would maximize if it could internalize spillovers.
Global warming is the cleanest motivating example because it is a negative externality with planet-scale spillovers and non-trivial intergenerational consequences. The mechanism (at the level of intro economics) is straightforward: fossil fuel use increases greenhouse gases such as carbon dioxide and methane, altering atmospheric heat retention and raising global average temperatures. The critical economic detail is not atmospheric chemistry; it is incentive structure.
When a person drives a car, they typically do not compute the marginal contribution of that trip to Bangladesh’s flood risk, agricultural disruption, or climate-related migration pressures. Not because they are immoral, but because the private decision is not priced to reflect the social harm. That is the essence of a negative externality: private benefit, social cost, and no internalization.
This is why public-sector involvement emerges in environmental policy: the market alone does not spontaneously price the damage unless forced via institutions, taxes, permits, regulation, or property-right reforms. Environmental agencies exist precisely because “leave it to the market” fails when the market does not include the external damage function in the choice calculus.
Economics becomes sharp when it becomes marginal. For any activity producing quantity q, define:
Externalities appear as wedges:
SMC(q) = PMC(q) + MEC(q)
SMB(q) = PMB(q) + MEB(q)
where:
A negative externality typically implies MEC(q) > 0, so SMC(q) > PMC(q). A positive externality typically implies MEB(q) > 0, so SMB(q) > PMB(q).
The classic teaching case is a negative production externality:
This scenario can be represented with the standard supply-and-demand apparatus for the market for steel. Let quantity of steel be on the x-axis and price on the y-axis. The demand curve represents marginal willingness to pay and therefore PMB(q). The private supply curve represents the private marginal cost PMC(q).
Private market equilibrium occurs where:
PMB(q) = PMC(q)
Call this private equilibrium quantity q1 (and price p1). Intro economics often stops here, praising efficiency. But the presence of sludge breaks the equivalence between private and social cost. Since each unit of steel creates additional harm (dead fish), social cost is higher. Let marginal damage be MD(q). Then:
SMC(q) = PMC(q) + MD(q)
In the lecture case, marginal damage was treated as linear/constant per unit, producing a vertical shift in the marginal cost curve. The social optimum quantity q2 satisfies:
SMB(q) = SMC(q)
Under typical assumptions here, SMB(q) = PMB(q) (consumption benefit is correctly reflected), so the correction is purely on the cost side. Thus:
PMB(q) = PMC(q) + MD(q)
Since SMC(q) > PMC(q), the socially optimal quantity is lower: q2 < q1. The market overproduces steel by (q1 - q2).
The welfare loss arises from units produced where social marginal cost exceeds social marginal benefit. Formally, the deadweight loss (DWL) is:
DWL = ∫q2q1 (SMC(q) - SMB(q)) dq
Graphically, DWL is a triangle bounded by the social marginal cost curve, the demand curve, and the vertical line between q2 and q1. A practical rule emphasized in the lecture:
This is not just a drawing trick. It encodes the conceptual message: misalignment between private decisions and social consequences produces pure welfare losses.
Externalities are not limited to production. Consumption can generate negative spillovers. Smoking is canonical:
Here the distortion is typically on the benefit side:
SMB(q) = PMB(q) - MD(q)
So, in the market equilibrium, individuals consume where:
PMB(q) = PMC(q)
But the social optimum requires:
SMB(q) = SMC(q)
If supply reflects true marginal production cost, then correction is largely on the benefit side: social marginal benefit is lower than private marginal benefit. This yields overconsumption relative to the social optimum and a deadweight loss wedge similar in shape to the production-externality case, but driven by a downward shift of the marginal benefit curve rather than an upward shift of the marginal cost curve.
SUVs provide an unusually vivid example because the externalities are multi-channel and interact socially. The lecture highlighted that changes in vehicle composition and weight matter economically because externalities are not only “pollution” but also infrastructure and safety.
Three negative consumption externalities from SUVs were emphasized:
Heavier vehicles generally consume more fuel to travel the same distance, increasing greenhouse gas emissions. Thus SUV adoption amplifies climate-related marginal damages relative to smaller vehicles.
Road damage rises more than proportionally with vehicle weight per axle. Heavier vehicles impose higher road maintenance costs on the public budget. Importantly, the lecture stressed a key conceptual point: externalities can be physical and financial. Road wear is both: it manifests as potholes (physical) and as higher taxation or public spending burdens (financial).
A safety externality arises because larger vehicles increase pedestrian fatalities and collision severity. A specific mechanism emphasized: vehicle geometry—especially high front grills—raises lethality because impact points shift toward the head and torso.
A deeper dynamic was also identified: SUVs induce a strategic arms race. Individuals buy larger vehicles for personal safety, which makes the road environment more dangerous, pushing others to buy SUVs in response. This is not merely a static externality but an endogenous amplification mechanism.
The lecture also treated an objection seriously: could higher capacity imply fewer vehicles per passenger and thus fewer externalities? The answer given was empirical: in practice, switching to SUVs does not meaningfully increase occupancy on average, so the capacity argument does not offset the external costs. The deeper point: externality assessment requires measurement, not vibes.
The framework is symmetric. Externalities can be positive: actions that benefit others without compensation. Markets then underprovide the activity because private decision-makers capture only a fraction of total benefit.
Research and development generates spillovers: knowledge diffuses through labor mobility, reverse engineering, publications, supplier learning, and competitive imitation. This means a firm’s R&D can raise productivity outside the firm—even benefiting rivals. From society’s perspective, that is still benefit.
Let q denote R&D investment. Then:
SMB(q) = PMB(q) + MEB(q)
Market equilibrium underprovides R&D because firms choose q such that:
PMB(q) = PMC(q)
But the social optimum requires:
PMB(q) + MEB(q) = PMC(q)
Hence qsocial > qmarket. Deadweight loss points toward the right: society wants more R&D than the market delivers.
The lecture referenced empirical work showing that private returns to R&D can be substantial (e.g., around 15% per year), while social returns can be multiples larger (e.g., roughly three times), implying large uninternalized spillovers. The exact numbers are not the conceptual core; the conceptual core is that innovation ecosystems are systematically underbuilt by private incentives alone.
To make positive consumption externalities concrete, the lecture used an intentionally mundane example: a neighbor can remove an unsightly pile of dirt for $1,000. To him, the private benefit is only $600, so privately he does not do it. But the action also yields $500 in benefit to a neighbor (a positive externality). Thus the total social benefit is $1,100, exceeding cost, and removal is socially desirable.
Formally, let action a be “remove dirt.” Let private benefit be B_p(a), external benefit be B_e(a), and cost be C(a). The private decision rule is:
Choose a if B_p(a) ≥ C(a)
The social decision rule is:
Choose a if B_p(a) + B_e(a) ≥ C(a)
This illustrates the general underprovision logic for positive spillovers: the market fails not because actors are irrational, but because the incentive calculus omits benefits enjoyed by others.
A naive framing treats the government as a neutral fixer of market failures. The lecture conversation rejected that simplification: governments can generate externalities too.
Thus the institutional question is not “market or government” but “which governance design minimizes negative spillovers and amplifies positive spillovers.”
Externalities are unavoidable in a connected world. The differentiator is not whether spillovers exist, but whether institutions internalize them. This conversation converged on one word: ownership.
Ownership, in this context, means that the decision-maker (firm or government) builds mechanisms so that:
This is why leadership questions matter. Leaders who ask “when is the next deal?” signal a short horizon. The organization adapts accordingly: it optimizes for immediate revenue, shortcuts, and cosmetic compliance. The result can become toxic: internal gaming, customer harm, and long-run fragility.
Leaders who ask “how do we improve this system over three years?” signal institutional time. They create conditions for process improvement, capability building, and durable customer value.
The practical leadership method proposed is an “altitude cycle”:
Once the wedge between private and social incentives is identified, the classic corrective instruments appear:
If marginal damage is MD(q), the corrective tax is:
t(q) = MD(q)
This shifts private marginal cost toward social marginal cost, reducing overproduction/overconsumption. A legitimate critique raised in the conversation: broad-based taxation can raise costs and feed inflationary pressures if applied bluntly, and it can be politically constrained. Taxation cannot be the only lever.
If marginal external benefit is MEB(q), the corrective subsidy is:
s(q) = MEB(q)
This raises private marginal benefit toward social marginal benefit, increasing underprovided activities such as R&D.
Negotiation matters when centralized pricing is hard, measurement is uncertain, or stakeholders are diverse. However, bargaining requires enforceable rights and manageable transaction costs; otherwise it collapses into power contests. The pragmatic stance is hybrid: use negotiated compacts where measurement is hard, and use taxes/subsidies where externalities are measurable and scalable.
Modern technological forces (AI, quantum computing, neurotechnology, VR) behave like ecosystems: they link R&D, startups, enterprise adoption, compute infrastructure, data centers, land, energy grids, talent markets, and geopolitics. A usable national or firm strategy therefore requires an abstract model that can be re-parameterized for any technology wave.
Let there be a set of technological activities indexed by i ∈ {1, …, N}. Each activity chooses a scale/output q_i ≥ 0.
Let externality types be indexed by j ∈ {1, …, M}. Examples include energy draw, emissions, labor displacement, safety risk, knowledge spillovers, security risks, and regional inequality impacts.
Define the externality generated by activity i of type j as:
E_{ij} = f_{ij}(q_i, x_i)
where x_i is a vector of design choices (e.g., model architecture, compute intensity, safety constraints, data center location). Importantly, E_{ij} may be positive or negative:
Let the policymaker (government or governance body within a firm ecosystem) control:
These instruments can be modeled as affecting private payoffs, feasibility constraints, or the externality functions themselves (via design requirements).
Let each activity’s private payoff (profit, utility, political payoff) be:
Π_i(q_i, x_i; τ, σ, ν) = R_i(q_i, x_i) - C_i(q_i, x_i) - \sum_{j} τ_{ij} \cdot g^-_{ij}(E_{ij}) + \sum_{j} σ_{ij} \cdot g^+_{ij}(E_{ij}) + \sum_j ν_{ij}
where:
Let social welfare be:
W(q, x) = \sum_{i=1}^N Π_i(q_i, x_i; τ, σ, ν) + \sum_{i=1}^N \sum_{j=1}^M \alpha_j \, E_{ij}
where α_j represents society’s valuation weight for externality type j (which can encode distributional preferences, national security priorities, climate priorities, etc.).
To encode “ownership,” the model must be dynamic. Let time be indexed by t = 0, 1, 2, …. Let the state of technology and institutions evolve via a state vector s_t, which includes infrastructure capacity, talent pool, energy grid, regulatory credibility, and geopolitical conditions. Externalities and outputs are time dependent: q_{i,t}, x_{i,t}, E_{ij,t}.
Define intertemporal social welfare:
\max_{\{q_{i,t}, x_{i,t}, τ_{ij,t}, σ_{ij,t}, ν_{ij,t}\}_{t \ge 0}} \sum_{t=0}^{\infty} \beta^t \, W_t(q_t, x_t, s_t)
subject to institutional, fiscal, and political constraints and the state transition:
s_{t+1} = F(s_t, q_t, x_t, u_t)
where u_t captures shocks (breakthroughs, crises, wars, supply-chain failures, sudden compute cost drops).
In this representation, “ownership” is not a slogan; it is the choice of β and the robustness of the transition system:
The leadership question dynamic becomes mathematically interpretable: leaders who reward only near-term deals implicitly set institutional β low and distort F toward fragility. Leaders who institutionalize altitude cycling and long-run incentives increase effective β and stabilize F.
Consider a nation planning a 40-year technology leadership strategy. The model classifies externalities into “encourage” vs “discourage” components without being married to one technology. AI is simply the most immediate case:
The policy implication is not “tax everything” or “subsidize everything.” It is a balanced schedule: price what is measurable and harmful; subsidize what is spillover-rich and underprovided; negotiate where measurement is ambiguous but stakeholder alignment is possible.
Even with dynamic structure, several blind spots remain—because they are not merely mathematical but political and epistemic.
Therefore, the model is best understood as a disciplined scaffold: it forces clarity about variables, spillovers, time horizons, and levers, but it cannot substitute for institutional design, legitimacy, and enforcement capacity.
The significance of this framework is strategic. As technological forces increasingly act directly on the economy—rather than indirectly through sectoral diffusion— societies will be separated by their ability to internalize spillovers at scale.
In the long run, the winners are not the most optimistic about technology, nor the most fearful. They are the ones that encode ownership—measurement, feedback, and long horizon—into institutions.
Externalities do not disappear. They scale. In a world of tightly coupled technological ecosystems, externalities become the primary unit of governance. The practical and theoretical frontier is therefore not invention alone, but the institutional capacity to internalize consequences. The difference between “progressive” and “behind” is not information. It is ownership: whether a society or firm builds mechanisms that force itself to face what it causes.
APA:
Asefi, H. (2026). Ownership of Externalities: From Steel Sludge to AI Ecosystems—A Dynamic Framework with Formal Notation. Unpublished manuscript (web essay).
MLA:
Asefi, Houman. Ownership of Externalities: From Steel Sludge to AI Ecosystems—A Dynamic Framework with Formal Notation. 2026. Web essay.
Chicago:
Asefi, Houman. 2026. “Ownership of Externalities: From Steel Sludge to AI Ecosystems—A Dynamic Framework with Formal Notation.” Web essay (unpublished manuscript).
BibTeX:
@misc{asefi2026ownership,
author = {Asefi, Houman},
title = {Ownership of Externalities: From Steel Sludge to AI Ecosystems---A Dynamic Framework with Formal Notation},
year = {2026},
note = {Web essay / unpublished manuscript},
}