Earth-Economy Modeling

Putting it all together

Resources

Slides 22

Entering the Final Week: Context, Relevance, and Framing

As the class enters its final full week, the lecture begins by situating the conversation in both personal and institutional context. The instructor reflects on an unusually busy semester and uses that transition point to emphasize that the themes now being covered are not peripheral topics, but central analytical tools. This framing is reinforced by recent travel to Santiago, Chile, where similar material was presented in a keynote role at a conference hosted by the Chilean Central Bank. That connection is not presented as anecdotal decoration; it is used to demonstrate that the questions raised in class are active concerns for major public institutions.

The lecture underscores that interest in nature-economy relationships now extends across central banks, financial institutions, policy organizations, and broader governance systems. The implication is that environmental-economic analysis has moved beyond niche environmental discourse and into the core of macroeconomic, regulatory, and development decision-making.

A brief project-management interlude addresses draft feedback timing and reinforces that written analytical work remains central to the course process. This practical note connects to the larger pedagogical structure: students are expected not only to understand concepts, but to communicate them clearly in policy-relevant form.

Position in the Course Arc

Recent Guest Lectures as Conceptual Inputs

The lecture then integrates the previous week’s guest presentations as substantive foundations for the current material. The biodiversity lecture by Colleen Miller is recalled as an effort to widen the class perspective beyond strictly economic metrics and into ecological measurement. The key pedagogical move here is not simply to list biodiversity indicators, but to foreground the contested question of what biodiversity means and how its value should be represented in analytical frameworks.

The land-use lecture by Siford Rungi is recalled for clarifying a structural tension: agricultural production and environmental protection often pursue goals that are both legitimate and directly conflicting. Rather than treating this as a temporary policy disagreement, the lecture frames it as a recurring governance problem rooted in competing objective functions, historical land-use trajectories, and institutional incentives.

A further conceptual refinement asks whether biodiversity was explicitly framed as the foundational substrate from which ecosystem services emerge. This matters because valuation discussions can become detached from ecological structure if biodiversity is treated as one service among many rather than as an enabling condition for the service portfolio itself.

Why Sequencing Matters

The lecture acknowledges that topic order is itself a teaching decision with analytical consequences. Biodiversity could plausibly be taught earlier as first principles, yet practical constraints and guest timing altered the sequence. This reflection is relevant for manuscript purposes because it reveals an important meta-point: the order in which frameworks are introduced can shape how students interpret causality between ecological systems and economic outcomes.

The Earth-Economy Modeling Framework

Bidirectional Linkages: Impacts and Dependencies

The lecture re-centers the course around a two-system model: Earth systems and economic systems. The essential claim is that these systems are coupled in both directions. Economic activity affects environmental conditions through extraction, land conversion, emissions, and waste. At the same time, economic performance depends on ecological functions that support productivity, risk buffering, and human well-being.

Traditional environmental economics is characterized as often emphasizing one direction of causality, namely economy-to-environment damage reduction. The lecture argues this is necessary but insufficient. A complete framework must also account for environment-to-economy dependency pathways, because policy errors arise when dependence is under-modeled.

InVEST, Dependency Models, and Value Representation

Prior course tools such as InVEST are located within this broader architecture. Assigning monetary values to ecosystem services is presented as one mechanism for rendering dependencies visible in decision spaces that are otherwise dominated by market prices. However, monetization is described as only one representational method, not a full substitute for biophysical and distributional analysis.

GTAP and CGE as Core Analytical Infrastructure

What GTAP Represents

GTAP is introduced as more than a single model. It is described as a shared analytical language and data architecture used by major institutions in finance, trade, and policy analysis. The significance of GTAP in this lecture is twofold: it provides operational structure for economy-wide modeling, and it creates comparability across analyses performed in different organizations.

The lecture notes that institutions such as McKinsey and the World Bank rely on this modeling family. The point is not primarily vocational, although employability is mentioned. The deeper point is that GTAP-style modeling has high policy leverage because it informs consequential decisions in trade, fiscal design, and strategic planning.

Defining CGE: Computable, General, Equilibrium

The lecture clarifies the three terms in computable general equilibrium. Computable means the system can be solved numerically in finite time using real data constraints. General means the model links multiple sectors, agents, and markets simultaneously rather than isolating one market at a time. Equilibrium means behavior and market outcomes are jointly determined through utility maximization, profit maximization, and price-mediated market clearing conditions.

This definition is important because it distinguishes CGE logic from partial-equilibrium intuition. The manuscript should preserve this distinction: partial analysis can be directionally useful, but it cannot fully capture cross-market feedbacks, reallocation effects, and policy spillovers.

From Simple Circular Flow to Real Economic Structure

Households, Producers, and Intermediate Goods

The lecture revisits the circular flow diagram and then expands it into a more realistic production network. Households still supply factors and demand goods, while firms still transform inputs into outputs. The major extension is that firms are not only producers of final goods for households; they are also buyers and sellers of intermediate goods to one another.

This intermediate layer is not a technical footnote. It is the structural core of modern production. Agricultural output depends on machinery. Machinery depends on metals, energy, logistics, and specialized services. Each sector’s output is another sector’s input. Any realistic model must track these internal linkages to avoid misrepresenting both policy incidence and shock propagation.

Government, Taxation, and Non-Idealized Markets

The lecture emphasizes that government cannot be omitted without losing realism. Taxes, transfers, procurement, and regulation shape prices, quantities, and investment trajectories in every contemporary economy. Simplified perfect-competition examples are useful pedagogically, but they are inadequate for policy analysis when public institutions are major demanders and allocators.

A vivid classroom example involving large public cheese stocks is used to illustrate that policy-driven storage and support programs can sustain outcomes that differ sharply from frictionless market equilibria. The example is less important for its commodity details than for its methodological message: observed economies include institutional interventions that must be represented explicitly.

Savings, Investment, and Growth Dynamics

The lecture also reintroduces savings-investment mechanisms as the channel linking current income allocation to future productive capacity. Savings fund investment, and investment expands capital stock. This dynamic matters for environmental-economic analysis because growth pathways alter both resource demand and the feasible policy frontier over time.

Why Trade Is Essential to Earth-Economy Analysis

Autarky as Baseline and Limiting Case

The concept of autarky is introduced as a useful analytical endpoint representing no international exchange. The lecture stresses that most real economies are far from autarkic conditions, even where political rhetoric emphasizes self-reliance. Therefore, trade linkages are not optional model features; they are foundational to accurate attribution of production, consumption, and environmental burden.

Historical and Policy Uses of GTAP Trade Modeling

Trade-model applications are framed through salient policy episodes, including U.S.-China trade conflicts and Brexit. In such cases, model outputs are used to estimate distributional outcomes, aggregate welfare changes, and sectoral adjustment pressures. The lecture also warns that model configuration choices can be used to support contradictory narratives, which creates an ethical and interpretive burden for analysts.

The manuscript should retain this warning in full force. Model outputs are only as credible as the assumptions and closures that produce them. Responsible use requires transparency, skepticism, and capacity to interrogate scenario construction.

Comparative Advantage and the Logic of Trade

Distinguishing Comparative from Absolute Advantage

The lecture returns to a foundational principle: comparative advantage, not absolute advantage, is the engine of gains from trade. Even when one country is more productive across all goods, gains remain possible if relative productivity differs by sector. Countries can benefit by specializing where their relative efficiency is highest, or where their relative disadvantage is smallest.

This principle justifies specialization across countries and sets up the formal trade example that follows.

Spectrum of Trade Regimes

The lecture places policy choices on a spectrum from autarky to highly open trade. This framing avoids false binaries and captures the practical reality that countries choose combinations of tariffs, quotas, regulatory barriers, and strategic exemptions. The post-2016 shift toward stronger protectionist sentiment is presented as a reminder that trade openness is politically reversible, not historically guaranteed.

Two-Country Soybean Model: Mechanics and Welfare

Autarky Equilibria and World-Market Formation

The lecture develops a two-country soybean case in which Country 1, labeled Soylandia, has comparative advantage in soy production, while Country 2 is relatively less efficient. Under autarky, each country clears its own market and thus exhibits different equilibrium prices.

When trade opens, a world market emerges. Export supply from the low-price country and import demand from the high-price country interact to determine a world price between the two autarky prices. Transport costs matter but are generally small relative to many commodity values, allowing substantial arbitrage and volume movement.

Distributional Effects Within Countries

In the exporting country, domestic price rises relative to autarky. Producers gain from expanded market access and higher returns. Domestic consumers lose due to higher prices and reduced local consumption. In the importing country, domestic price falls. Consumers gain from lower prices and higher consumption, while domestic producers lose market share.

The lecture emphasizes that these distributional effects are politically powerful because gains and losses are unevenly organized. Producer groups are often concentrated and politically mobilized; consumer losses or gains are diffuse and less likely to generate organized advocacy.

Aggregate Surplus Result

Despite internal winners and losers, total surplus across both countries rises under standard assumptions. The welfare gain is generated by reallocating production toward sectors and locations with higher relative efficiency. This is the classic efficiency argument for trade and remains central to mainstream economic analysis.

Political Economy of Trade Policy

The lecture explicitly connects analytical welfare results to political process. Trade policy outcomes do not follow welfare arithmetic automatically. They emerge through institutional bargaining, lobbying asymmetries, and narrative framing. This is why policies can persist that are politically stable yet economically inefficient in aggregate terms.

The manuscript should preserve this political-economy layer because it explains the gap between what models predict as efficient and what governments actually implement.

Embedded Environmental Impacts in Trade

Embedded Flows and Attribution

A major conceptual transition occurs when the lecture connects trade theory to environmental accounting. If production generates externalities, then traded goods carry embedded environmental impacts. The location of consumption and the location of damage can diverge sharply.

Palm oil is used as a paradigmatic case. Consumption may occur in high-income importing countries that show little domestic deforestation, while production expansion in exporting countries drives forest conversion and biodiversity pressure. Domestic environmental indicators in importing countries can therefore understate global impact linked to domestic demand.

Embedded Biodiversity Loss

The lecture names a specific consequence: embedded biodiversity loss. Consumption in one jurisdiction can induce species loss elsewhere by expanding production in ecologically sensitive regions with strong comparative advantage. This reframing is essential for policy design because it challenges territorial accounting systems that ignore supply-chain causality.

Leakage: When Domestic Protection Shifts Damage Abroad

Definition and Mechanism

Leakage is defined as the increase in environmental degradation elsewhere that follows from protection measures in one country. The mechanism is price-mediated. If domestic environmental policy constrains supply, prices can rise in world markets, creating incentives for expanded production in other jurisdictions.

Policy Example and Model Logic

The lecture uses forest and riparian-buffer policy examples to illustrate this mechanism. Local environmental outcomes can improve through reduced sediment export and stronger ecological protection. Yet reduced local timber supply can raise world prices and stimulate logging in other regions, including high-value ecosystems such as parts of the Amazon.

The key analytical point is that local policy success may overstate global success if leakage is substantial. Reported benefit rates can be partially offset by displacement effects outside the regulated jurisdiction.

Magnitude and Policy Implications

Leakage magnitude is context-dependent, but the lecture highlights that offsets can be large enough to materially alter cost-effectiveness rankings across policies. Therefore, policy design should prioritize interventions with lower leakage potential and stronger international coordination.

Why General Equilibrium Is Necessary

The lecture concludes by asserting that these questions cannot be answered reliably with single-market intuition. Embedded impacts, trade-mediated reallocation, and leakage are system phenomena. They require models that capture intersectoral and international feedback loops.

General equilibrium analysis is presented not as academic complexity for its own sake, but as a practical necessity for credible environmental policy. The forward link to the next lecture is clear: once the class understands why these feedbacks matter, CGE tools become justified as the appropriate framework for quantifying them.

The final message is constructive rather than fatalistic. Environmental protection remains both possible and necessary, but effective design depends on evaluating policies in a globally connected economic system rather than in isolated domestic snapshots.

Lecture Context and Immediate Stakes

This lecture opens on day three of the earth economy modeling sequence and immediately frames the topic as both academically and professionally urgent. The instructor explains that the day sits at the intersection of end-of-semester teaching demands, recent international presentation obligations, and a manuscript submission deadline for the Proceedings of the National Academy of Sciences. That context is not presented as personal drama for its own sake. It is used to highlight that the concepts in the lecture are not abstract classroom exercises, because the same framework and figures discussed with students are being finalized for high-stakes scientific communication.

The instructor also notes that the full integrated model had been running overnight for approximately twenty hours and had to complete successfully for submission to proceed. This detail underscores an important methodological point that recurs later in the lecture. Real-world computable general equilibrium and earth economy model workflows are computationally intensive, and practical constraints around run time and model stability shape research pacing, policy relevance, and publication timelines.

The lecture agenda is stated clearly. The class will begin with trade and use it as an entry point into the distinction between partial and general equilibrium. It will then move to the role of general equilibrium in sustainability analysis and conclude with practical modeling structure, toy-model exploration, and a transition toward land-integrated earth economy methods.

Partial Equilibrium and General Equilibrium: Core Conceptual Distinction

Why the Distinction Matters

The lecture emphasizes that the partial-versus-general equilibrium distinction may sound technical but is foundational for policy analysis. Partial equilibrium is introduced as the standard framework students encounter in introductory economics. A single supply-and-demand market is analyzed while all other conditions are held constant under the ceteris paribus assumption. This approach remains useful because it is tractable, transparent, and often adequate when spillovers are limited.

General equilibrium is then defined as the framework in which all major markets are linked and virtually no domain is held exogenously fixed. Instead of asking what happens inside one market while the rest of the economy remains unchanged, general equilibrium asks how changes in one market alter behavior, prices, and resource allocation throughout the entire system.

The Mechanism of Linkage

To make this concrete, the lecture uses shared inputs, especially labor, as an intuitive linkage mechanism. If workers can move between sectors such as paper production and automobile manufacturing, a shock in one sector affects wage pressure, hiring decisions, and output composition in others. What appears to be an isolated market intervention under partial equilibrium can therefore produce indirect effects that are first-order, not secondary.

The lecture argues that partial equilibrium remains appropriate in many contexts, especially where firms are small relative to market size and behave as price takers. However, once shocks propagate through labor markets, trade channels, cross-sector input dependencies, or demand complementarities, partial methods can miss central dynamics.

Ethanol as a General Equilibrium Case Study

Policy Origins and Market-Level Effects

The lecture introduces U.S. ethanol policy to show why market-level analysis can be misleading when used alone. Ethanol is described as fuel produced from feedstocks such as corn, sugarcane, and potentially cellulosic materials. The class is reminded that blending mandates and E15 and E85 infrastructure are now familiar policy realities in places like Minnesota.

At the partial equilibrium level, the logic is straightforward. A mandate shifts ethanol demand outward, which increases equilibrium price and quantity in the ethanol market. Producers benefit from higher prices and stronger demand conditions, and the policy appears to succeed from the perspective of those represented by politically powerful producer groups.

Economy-Wide Spillovers and Distributional Effects

The lecture then pivots to what partial analysis misses. If corn demand expands for fuel, land is reallocated away from alternative uses such as soy production and pasture. Food prices rise not only for corn but for downstream goods that rely on corn-intensive inputs. This introduces domestic distributional consequences in which some producers gain while many consumers face higher costs.

The central argument is that these effects are not accidental side notes. They are structural consequences of interlinked markets and therefore require general equilibrium treatment. The policy can no longer be evaluated only by ethanol-sector output gains because the full incidence is spread across land use, food systems, household budgets, and trade patterns.

Environmental Feedback Through Global Land Use

The lecture then adds the environmental general equilibrium channel. As agricultural prices rise, land conversion becomes more profitable globally, not just domestically. This can induce deforestation and land-use change in other countries, including Brazil, even when the originating policy is framed as domestic climate action.

This mechanism is described using terms such as conservation leakage and induced land-use change. The policy lesson is explicit. A measure can look environmentally beneficial in one accounting frame while generating offsetting harm elsewhere through price-mediated global responses. Without a general equilibrium framework, that displacement remains hidden.

Research Controversy and Institutional Stakes

The lecture places this debate in institutional history. Researchers in the University of Minnesota’s applied economics community were among those producing early evidence that corn ethanol could have substantial negative side effects through food markets and land systems. At the time, these findings were highly contentious and drew intense industry pressure.

This historical note is used to reinforce why academic freedom and tenure matter in politically charged policy domains. The instructor states that what was once controversial has moved toward broad consensus, namely that corn-based ethanol is likely a poor policy on net once full-system effects are considered. At the same time, the lecture preserves nuance by noting active research on alternatives such as sugarcane and cellulosic pathways.

From Circular Flow to Computable Structure

Circular Flow as a Starting Mental Model

The lecture returns to the circular flow diagram and treats it as a pedagogical starting point for computable general equilibrium modeling. The key transition is from a stylized representation of firms and households to a quantitatively calibrated system that can support policy counterfactuals.

Regional Income Allocation and Macroeconomic Shares

The first structural addition is what GTAP labels the Regional Household, reframed in the lecture as regional income allocation. The model must represent how national income is distributed across private consumption, savings, and government spending. This is presented as realism rather than theoretical ornament. Different countries allocate these shares differently, and those differences shape macroeconomic behavior, distributional outcomes, and policy response.

Examples are used to anchor this point. The United States is described as relatively consumption-heavy, Sweden as having larger government shares, and China as having comparatively high savings rates. These patterns enter the model through calibrated coefficients chosen to reproduce observed expenditure data. The lecture explicitly links this to familiar macro accounting identities such as C + I + G and, with external trade, C + I + G + X - M.

Production Networks, Intermediate Goods, and Sectoral Depth

The next structural layer is a full production system in which firms use household-supplied factors and trade intermediate inputs across sectors. This means producers do not only supply final demand. They also buy from and sell to one another in input-output networks. The lecture notes that GTAP-style systems often include approximately sixty-five sectors, which immediately increases dimensionality and interdependence.

Savings are tied to investment and capital accumulation, while government demand is represented as an active component of total demand. The lecture emphasizes that this complexity is intentional. The purpose is not minimal elegance. The purpose is policy-relevant realism in a world where sectoral dependencies, public spending patterns, and capital formation materially affect outcomes.

Multi-Region Replication and Data Demands

The circular-flow-plus-macro structure is then scaled across roughly one hundred sixty countries, subject to data quality constraints. Input-output matrices are required to represent how much each sector uses from every other sector and how these relationships differ across regions. The lecture uses examples such as corn in ethanol and wood in paper to make matrix logic tangible.

This explains why full model runs are computationally expensive and can require many hours. It also reinforces that model quality depends on the depth and reliability of underlying data, not only on equation form.

The 3x3 CGEMES Toy Model as Pedagogical Bridge

Why a Toy Model Is Used

Because full systems are difficult to run interactively in class, the lecture introduces a browser-based 3x3 CGEMES model as a pedagogical bridge. Students are directed to access it through the course site under games and circular flow content. The instructor explains the teaching strategy directly. Complex methods are easier to learn when students can manipulate a simplified system and see immediate equilibrium responses.

The toy model is presented as a genuine CGE implementation, not a decorative mock-up. It is inspired by early GTAP-era demonstrations associated with Tom Hertel and is designed to capture the core logic of trade and policy shocks in a tractable setting.

Structure of the 3x3 System

The model uses three regions, the United States, the European Union, and the rest of the world, and three sectors, agriculture, manufacturing, and services. Despite this simplification, it preserves the main theoretical insight that trade barriers alter prices, production allocation, welfare, and cross-country incidence.

The lecture highlights a key advancement over older blanket claims about free trade. CGE frameworks can show aggregate welfare implications while also identifying winners and losers across sectors, regions, and factor groups.

Trade Liberalization, Distribution, and the NAFTA Illustration

Aggregate Gains and Concentrated Losses

The lecture uses NAFTA as a modern example of why distribution must be integrated into welfare interpretation. The instructor argues that evidence strongly supports aggregate gains from liberalization in standard surplus terms while also showing meaningful losses concentrated among specific worker groups and places, especially in manufacturing-centered communities.

A recent study is referenced to emphasize that these distributional shocks can have long-run health consequences, including increased mortality among more exposed populations. The policy implication is that aggregate efficiency gains do not erase social harm when adjustment burdens are concentrated.

Why This Matters for Model Interpretation

The lecture treats this as a corrective to simplistic messaging. Saying that freer trade raises total welfare can be true and still be incomplete. CGE analysis is valuable precisely because it can retain both truths simultaneously, aggregate gains and uneven losses, and force policy discussions to confront compensation, transition support, and equity design.

Running Policy Counterfactuals in the Toy CGE

Interactive Controls and Equilibrium Re-Solution

The interface allows students to modify tariff rates, productivity parameters, and utility-related preferences through sliders. Every change triggers rapid equilibrium re-solution, making visible the comparative statics that are otherwise hidden in algebraic systems. The lecture stresses that this rapid feedback is computationally enabled but conceptually grounded in standard profit and utility maximization with market-clearing constraints.

Tariff Example and Intersectoral Reallocation

A featured scenario imposes a twenty-five percent U.S. tariff on manufacturing imports. The model shows increased U.S. manufacturing output, illustrating the domestic-protection argument for tariffs. At the same time, output and welfare shifts appear across other sectors and regions, demonstrating spillovers and trade-offs that cannot be inferred from one market alone.

In the scenario discussed, U.S. welfare rises by approximately 0.67 percent while other regions decline. The lecture emphasizes that unilateral gains can coexist with external losses, which is a recurring theme in strategic trade policy.

Retaliation and Reciprocal Tariffs

The lecture then introduces retaliatory responses through reciprocal tariffs. Once trading partners match barriers, prior unilateral gains can disappear and all regions can incur welfare losses. This supports a standard free-trade conclusion with strategic nuance. Tariffs may create local, short-run gains under asymmetric conditions, but retaliation risks can reverse those gains and reduce welfare broadly.

A humorous aside about tariffs being listed for unpopulated territories such as the Heard and McDonald Islands is used to illustrate how broad policy templates can be applied mechanically. The anecdote serves a serious methodological point. Policy design and implementation details matter, and poorly specified categories can produce absurd but revealing outcomes.

Policy Relevance and Institutional Use of CGE Models

The lecture closes the toy-model segment by noting that these are not merely educational exercises. Comparable model classes are routinely used by administrations, advisory teams, and opposing policy analysts during trade conflicts. Different actors may frame results differently, but CGE methods remain a common technical language for evaluating trade interventions and their distributional consequences.

Students are encouraged to use the model exploratorily rather than as a graded assignment, with the goal of developing intuition for how equilibrium systems respond when policy, productivity, or preference parameters move.

Earth Economy Modeling Beyond Standalone CGE

Coupling Economic and Natural Systems

The lecture then returns to the full research model and explains why it is referred to as an earth economy model rather than only a CGE. The distinction is coupling. The economic core is linked to ecosystem service models and land-use change systems covered elsewhere in the course.

Land is identified as the primary coupling channel because land allocation sits at the interface of production, environmental quality, carbon dynamics, and biodiversity outcomes. This is also why the prior lecture on economic treatments of land is treated as foundational.

Expanding Production Functions to Include Resources

The lecture presents a conceptual shift from production functions of capital and labor alone to functions that include resources and land. In notation, this means moving beyond F(K, L) toward forms like F(K, L, R), where R captures resource constraints and land-dependent productivity.

This change is characterized as analytically necessary rather than optional refinement. Without explicit resource channels, growth and policy models can overstate substitutability and understate ecological constraints.

Historical Evolution of Land Representation in CGE Systems

From Value-Only to Area-Based Representation

The first generation of land integration treated land primarily as a dollar-valued factor, consistent with broader monetary accounting conventions. The lecture notes that this was a useful starting point but physically incomplete because land is spatially bounded.

Subsequent model generations added physical extent, typically measured in hectares, to constrain land availability in biophysical terms.

From Area to Heterogeneous Land Quality

Later refinements recognized that one hectare is not equivalent to another. Productivity, accessibility, and ecological context vary across locations, so effective land supply and opportunity cost are heterogeneous. The lecture uses intuitive contrasts, such as highly productive agricultural regions versus locations where production potential is low.

This move toward heterogeneity marks convergence between economics and Earth system data practices. Remote sensing and geospatial datasets become essential inputs for constructing more realistic land supply relationships.

Transition to Spatially Derived Land Supply Curves

The lecture previews the next session by noting that land supply curves can now be derived from detailed spatial data rather than inferred only from aggregate market observations. This transition substantially improves realism and enables tighter coupling between economic counterfactuals and environmental process models.

Closing Course Logistics and Feedback Requests

The lecture ends with two practical course items that are directly relevant to instructional continuity. Students are asked to complete official course evaluations, with the instructor emphasizing that this feedback is taken seriously and can influence future course offerings in the department. The instructor frames the course as an uncommon and still-evolving effort to teach integrated earth economy modeling at the undergraduate level.

The second item concerns rough-draft timing and feedback availability in Canvas. Students are reminded that the rough draft is intended as a low-stakes opportunity to receive substantive guidance before the final deadline. The class is encouraged to verify that feedback is visible and to follow up by email if publication delays occur.

The session closes by reiterating that students are broadly on track and by encouraging continued progress toward final submission.

Course Closure, Final Presentations, and Framing of the Day

This lecture opened with practical course logistics that also framed the broader intellectual arc of the semester. The instructor clarified that the final class meeting on Monday would be dedicated to student presentations and requested that slides be submitted by email, or through Canvas, at least ten minutes before class. This early submission requirement was presented as a coordination device to allow all decks to be merged into a single sequence, ensuring smooth transitions and equal treatment across presenters.

The format for these presentations was specified as lightning talks, a format widely used in academic and professional settings. The key principle of the lightning talk was not rapid speech for its own sake, but disciplined synthesis. Students were asked to communicate only the most important findings, assumptions, and implications in a highly constrained time window. Each student would have five minutes, with a likely slide range of roughly three to six slides depending on individual style and content density. The strict five-minute cutoff was emphasized as non-negotiable, not as punitive classroom management, but as training in a core professional skill: the ability to distill complexity without losing conceptual integrity.

A brief clarification followed regarding the rough draft due date. The expectation was substantial progress toward the final project rather than complete coverage of every section. In this framing, the draft functioned as a progress checkpoint rather than a high-stakes completeness audit, and the evaluation criterion centered on meaningful advancement.

Research Context and Why This Lecture Matters

The instructor then situated the lecture within live research activity. A multi-author manuscript involving twenty-six contributors had just been submitted shortly before the deadline, and this context was used to illustrate the practical realities of scientific collaboration. Beyond analysis and writing, the process involved intensive coordination, consensus building, and procedural completion among many coauthors. This transition into current research was not incidental. It provided a bridge between classroom theory and frontier application, signaling that the lecture material was not retrospective pedagogy alone, but active methodological practice connected to ongoing scholarly production.

The day’s agenda was organized around three substantive goals. First, the course would close the unit on land as a foundational production input and a central sustainability variable. Second, students would engage with an interactive land-use simulation and game to build intuition about system dynamics. Third, the lecture would present new quantitative results from an integrated modeling pipeline linking economics, land-use change, and environmental feedbacks.

Land in the History of Economic Thought

Classical Centrality and Modern Neglect

A core conceptual thread in the lecture was the historical trajectory of land within economics. Early economic thought treated land as a fundamental source of value and production capacity. Over time, however, mainstream macroeconomic modeling often shifted attention toward labor and capital, and in some growth frameworks toward capital in particular. This intellectual shift was interpreted partly through geopolitical history. During and after World War II, and throughout the Cold War, economic planning and theory were heavily oriented toward industrial capacity, strategic production, and capital accumulation. In that context, land often appeared less analytically urgent than industrial capital stock.

Re-entry Through Ecological Limits

As ecological constraints became more visible and policy attention moved toward planetary boundaries, land returned to center stage. The lecture emphasized that computable general equilibrium models were among the earlier economic frameworks to reincorporate land constraints in explicit, policy-relevant ways. While broader economics has increasingly moved in this direction, CGE work remains one of the most operationally mature traditions for integrating economic and biophysical constraints at scale.

GTAP-AEZ and the Spatial Differentiation of Land

Countries, Agro-Ecological Zones, and Productive Heterogeneity

The modeling architecture discussed in the lecture used GTAP and particularly its GTAP-AEZ variant. This framework disaggregates the world first by country and then by agro-ecological zones. AEZs are defined by climatic and agronomic conditions, including humidity regimes, tropical-temperate-boreal distinctions, and growing-period characteristics. The implication is that land is not treated as homogeneous acreage. Instead, each zone is associated with distinct productive potentials and constraints.

This differentiation matters because economic outcomes depend not only on how much land exists, but also on what that land can produce under specific climatic and ecological conditions. In this approach, hectares are tracked physically and economically, and land quality enters as a determinant of sectoral feasibility and profitability.

The Initial Limitation: Fixed Land in Economic Use

Despite the sophistication of GTAP-AEZ, the lecture highlighted a crucial limitation in standard implementations: the total amount of land in economic use was effectively fixed. Sectors could compete for land shares, but aggregate land conversion at the frontier was not endogenously represented. This meant that the model could redistribute land among uses but could not directly represent the expansion of economic land use through conversion of natural landscapes.

Endogenizing Land Supply and Modeling Conversion

Why an Endogenous Land Supply Curve Was Necessary

To address the fixed-land limitation, the research introduced a land supply curve within each AEZ. In this structure, land supplied to economic use becomes responsive to returns, analogous to standard supply logic but interpreted through conversion rather than creation. The model does not assume that new terrestrial area can generally be manufactured. Rather, it represents the transition of land from natural states into agricultural, urban, or other economic categories under changing price and policy conditions.

The empirical basis for this extension came from geospatial data and yield information across grid cells. By associating profitability thresholds with incremental conversion potential, the model estimated how much additional land would enter economic use at different return levels.

Policy Translation Through Supply Shifts

This change enabled direct policy analysis for protected areas. When policy designates land as protected, available land for economic conversion is reduced. In the language of supply curves, this shifts effective supply leftward. With demand held in view, the new equilibrium implies lower quantity of land in economic use and higher shadow prices for land. The lecture framed this as a straightforward supply-and-demand mechanism whose importance had nevertheless been underrepresented in earlier fixed-land model structures.

Distributional Consequences and Equity Exposure

The lecture emphasized that higher land and food prices create unequal welfare effects. Low-income households, which spend larger shares of income on food, face stronger burden under price increases associated with land scarcity policies unless compensating mechanisms are introduced. The analytical point was that conservation policy cannot be evaluated on ecological outcomes alone; incidence and distribution matter.

Substitution, Consumer Preference, and Green Product Differentiation

A student question prompted an important extension: how the framework can represent environmentally differentiated products and changing consumer behavior. The lecture responded by describing a two-product structure in which conventional production and certified or biodiversity-friendly production are treated as substitutes. If consumers shift demand toward environmentally preferred goods, demand for harmful production falls. This mechanism remains fully consistent with foundational microeconomics while opening practical pathways for certification policy, labeling systems, and market-mediated ecological improvement.

Linking Economic and Environmental Models Through Land-Use Change

The Missing Link Problem

The lecture identified a central bottleneck in Earth-economy modeling: translating coarse economic projections into high-resolution land-use maps suitable for ecological impact models such as InVEST. Economic models often produce regional or coarse outputs, whereas ecosystem service models require fine spatial detail. Without a credible downscaling bridge, the integrated policy workflow remains incomplete.

Computational Scale and Practical Constraints

This challenge is not conceptual only. It is computationally severe. Large-country analyses can involve millions to billions of cells, and full global exercises exceed ordinary desktop capacity. The research effort therefore required intensive programming and high-performance computational design, with prototyping on fast local systems and production runs on supercomputing infrastructure.

SEALS: Spatial Economic Allocation Landscape Simulator

Purpose and Naming

The land-use downscaling model introduced in the lecture was SEALS, the Spatial Economic Allocation Landscape Simulator. The model was designed to convert coarse land-use change signals into high-resolution spatial allocations that preserve aggregate consistency while generating plausible local transition patterns.

Two-Step Downscaling Logic

SEALS operates with two linked stages. The first stage allocates regional change totals into coarse grid patterns guided by economic logic and relative profitability. The second stage performs fine-scale allocation using high-resolution predictors and machine-learning pattern recognition from historical land-use transitions. This second stage relies on covariates such as market access, slope, soils, topography, and bioregional structure.

The result is a map that simultaneously respects aggregate change constraints and produces locally realistic transition geography. If one class contracts in a location, other classes can expand in the same location in ways that remain globally and regionally coherent.

Policy Counterfactuals and Leakage Dynamics

A major strength of this structure is policy counterfactual capability. The lecture used protected-area policy as the core example. Under a baseline scenario, urban or cropland expansion may occur inside a candidate conservation zone. Under policy, those transitions are blocked within the protected area, but total demand pressure can reallocate expansion to surrounding spaces. This displacement effect, often discussed as leakage, becomes visible and quantifiable in the spatial outputs.

Adjacency, Spatial Dependence, and Transition Probability

Why Adjacency Is Predictively Powerful

The lecture identified adjacency as a key predictor in land-use transition systems. Cropland often expands at existing cropland edges, producing perimeter-led growth rather than random emergence. The model generalized this intuition into class-pair adjacency functions that estimate how nearby land classes affect transition likelihood.

Nonlinear Distance Effects

The influence of nearby urban land was presented as non-monotonic. Areas very close to cities may lose cropland probability due to suburban expansion pressure, while areas too far from urban markets may also be less attractive because of transport and market access costs. Intermediate distances can therefore be most favorable. This kind of nonlinear spatial dependence increases realism and improves predictive performance.

From One-Dimensional Relationships to Suitability Surfaces

By extending these relationships into two-dimensional suitability surfaces across the landscape, the model generated transition maps with strong explanatory and forecasting value. The broader claim was that this approach creates a practical forward-looking lens for biodiversity risk, deforestation trajectories, and policy-sensitive spatial planning.

Interactive Simulation as Conceptual Infrastructure

The lecture moved from formal model architecture to pedagogical simulation. Students were introduced to a land-use simulation app containing forest, agriculture, urban, and protected categories, each associated with carbon and revenue implications. Forest included age-related carbon dynamics, agriculture provided intermediate revenue and lower carbon retention, and urban land produced high private revenue with low carbon storage.

Suitability layers explained why different classes appear in different locations, while period-by-period simulation updated land allocation under specified prices and returns. Under default settings, cropland tended to expand first under food pressure, followed by stronger urban expansion once food requirements were met. Over time, revenue increased while biodiversity commonly declined, and carbon outcomes depended on the sequence and composition of transitions.

The simulation was used to illuminate tragedy-of-the-commons dynamics. Private conversion incentives can dominate when ecological externalities are not priced. The lecture argued that this is precisely where economics is useful, not to justify degradation, but to diagnose incentive misalignment and evaluate corrective instruments.

A carbon tax example illustrated policy trade-offs. A sufficiently strong tax can suppress harmful conversion, but extreme values can also collapse modeled economic activity in a simplified setting. The policy lesson was that environmental objectives and livelihood objectives must be jointly optimized rather than treated as separable.

Land Use Game, Social Planning, and Budget-Constrained Governance

The game version introduced scenarios and explicit welfare scoring to simulate social-planner decisions under finite intervention budgets. Welfare combined economic revenue with conservation value from carbon and biodiversity, thereby distinguishing social value from individual private returns. Players could manually allocate limited land-use interventions over a sixty-step horizon.

Class interaction showed that strategy choices, including tax calibration and strategic placement of protected areas near expansion fronts, could substantially change outcomes. The instructor’s own run highlighted a common governance reality: conservation budgets are often insufficient relative to the scale of conversion pressure. The game therefore functioned as an accessible representation of constrained policy optimization.

Full Earth-Economy Model Linkage and Multi-Model Integration

The lecture then returned to the complete modeling chain. Economic modules generated demand and price signals. SEALS translated these into high-resolution land-use trajectories. Environmental modules, including InVEST-style components, evaluated ecosystem-service implications. Additional integrated assessment components handled climate and macroeconomic linkages. A growth dynamic was included through endogenous investment behavior, with next-period capital determined by depreciation and new investment, consistent with Ramsey-style accumulation logic.

The integration challenge was substantial because five model systems had to be linked in a coherent computational pipeline. This was framed as both a technical achievement and a practical necessity for policy-relevant quantification.

NGFS Scenarios and Nature-Related Financial Risk

Scenario Architecture

The quantitative results were organized around Network for Greening the Financial System scenarios. These scenarios reflect an emerging institutional recognition among central banks and financial authorities that ecological destabilization translates into macro-financial risk. NGFS pathways add policy specificity, including carbon pricing assumptions, to broader socioeconomic narratives.

Dependency Assumptions and Comparative Results

The lecture reported comparisons across four dependency structures: an unrealistic no-nature-dependency baseline still common in some modeling practice, a climate-only dependence case, an ecosystem-services-only case, and a combined climate-plus-ecosystem-services case. The combined case was presented as conceptually superior because climate and ecosystem channels interact rather than operate in isolation.

Under current-policy trajectories, projected GDP losses were substantial. Alternative policy pathways moderated losses to varying degrees but did not uniformly eliminate negative outcomes. The lecture emphasized that these aggregate trajectories conceal strong distributional heterogeneity.

Equity, Regional Asymmetry, and Migration Pressure

A central finding was that lower-income countries bear disproportionately large damages across scenarios. Some higher-income regions appeared less harmed or occasionally benefited in narrow channels due to climatic productivity shifts, but such asymmetries were not interpreted as good news. Instead, they imply intensified global imbalance, with heightened migration pressure, social stress, and geopolitical risk. The lecture treated equity as a first-order analytic and policy variable rather than a secondary ethical add-on.

Land Outcomes Across Policy Pathways

Spatial and aggregate land outputs reinforced the macro findings. Under current policies, natural land tended to decline while cropland expanded. Under stronger mitigation pathways, especially those aligned with net-zero framing and conservation support, natural land decline could be reduced, stabilized, or partially reversed. The lecture underscored that conservation outcomes depend not only on nominal targets but on modeled incentives, financing structures, and implementation constraints.

Stress Testing Abrupt Climate Acceleration

In addition to baseline pathways, the lecture introduced a stress-test design inspired by financial risk practice. The stress test imposed an abrupt climate-damage acceleration around 2035 by switching from a milder to a harsher trajectory. The key insight was nonlinearity. Damages grew more than proportionally relative to forcing changes, indicating compounding risk and feedback amplification. This finding was used to argue that central banks and financial institutions should treat ecological and climate risks as dynamic systemic risks rather than smooth trend variables.

Closing Synthesis for the Semester

The lecture closed with two integrative claims. The first was that modern economics must operate within a bounded framework resembling the donut concept, in which societies must remain above a social foundation while staying below ecological ceilings. Endless growth narratives are inadequate, yet pure degrowth framing without social provisioning is also inadequate. The practical challenge is governed transition under dual constraints.

The second claim was cautiously optimistic. Institutional momentum is increasing among central banks, investors, and finance ministries regarding nature-related risk. Although timing uncertainty remains and delayed response is a serious concern, the field is expanding rapidly in both research and professional domains. This implies rising demand for expertise in nature dependency analysis, sustainability risk assessment, integrated modeling, and policy design.

The final message returned to the course purpose. Students were encouraged to see environmental and natural resource economics not as a narrow technical specialty, but as part of a broader sustainability framework that links incentives, institutions, ecological systems, and distributional outcomes. The lecture ended with appreciation for student engagement and a practical reminder to submit presentation slides before class for final-session coordination.

Transcript (Day 1)

Welcome to the final full week of class. We are almost at the end of the semester. It has been an incredibly busy semester for me. I’m looking forward to today very much.

Last week, I was in Santiago, which was beautiful. It was fall there, though the city is somewhat polluted. It was wonderful because Santiago is one of the biggest cities with mountains really close by—the Andes Mountains are right there. We got to hike up to high places and look out over the city, though the pollution did make it less pretty than I would have hoped.

What’s interesting is that the content we’re discussing today is literally what I was presenting there. I was invited as the keynote speaker at a conference hosted by the Chilean Central Bank, and the content was very similar to what we’ll cover today. It’s a good sign that this material is relevant across global institutions.

There is broad interest across central banks, financial institutions, and throughout the economy and government about how nature affects our well-being. It’s quite encouraging.

One of you asked if I received feedback on the first draft report that was due Friday. I’m hoping to get through those today or maybe Wednesday and will send out some emails with my feedback. Once I have a chance to review what you’ve written, I’ll be able to give more detailed comments.

For today’s class, we’ll cover several topics. We’ll start with a discussion of the guest lectures from last week, then move into the detail of the computable general equilibrium model. I’ll emphasize trade theory and show you two web app games I’ve created. We’ll also discuss land use, building on what Professor Siford Rungi covered.

Let’s start by reviewing the guest lectures. Last Wednesday and Friday, you heard from two speakers. I want to check in on what they covered and what you learned.

First, we had Colleen Miller, our senior biodiversity scientist at NatCap. She discussed how we measure biodiversity. What stood out to you?

It was interesting to hear different perspectives on measuring biodiversity and to dive into the ecological aspects, since we’ve focused more on economics so far. She presented several metrics and we had a class discussion about what biodiversity means and different approaches to measuring it.

For the second lecture with Siford Rungi, many of you noted that a key takeaway was the tension between environmental and agricultural perspectives. Each perspective has its own agenda—either maximizing production or protecting the environment—and these often conflict. He also discussed the history of land use and how resource usage and policies have changed over time. The point about these two conflicting perspectives was particularly striking.

Did Colleen emphasize that biodiversity is the fundamental foundation from which ecosystem services derive?

I was wondering if she would make that point. It’s something I debated whether to emphasize more at the beginning of the semester. The existence of diverse ecosystems is the reason we can derive value flows from ecosystem services. I was planning to go to Santiago at the end of the semester, so I wanted to have Colleen speak about biodiversity rather than having me cover it early. I may reorganize that for next year.

One amusing side note: when I was preparing my keynote speech in Santiago, I received a call saying that Siford Rungi wasn’t in class. Apparently, he went to the wrong room. I emailed him several times trying to help, and he ended up contacting our academic support staff, but the database still showed the old room. So I was literally sending emails as I was walking up to the podium to give my talk. I’m glad it worked out and he found his way to the correct location.

Just a reminder of where we are: we have the Earth and the economy as our two core models. We have discussed how they are linked through impacts and dependencies. We have already worked through key tools, including InVEST and dependency models, and one way we have represented dependence is by assigning dollar values to ecosystem services. That is only one part of the story, and we will continue expanding it.

Today we will spend more time on the economy side. That is expected in an economics class, but the details matter because actions inside the economy drive major impacts on the Earth. Traditional environmental economics often focused on only one direction: the economy harms the environment, and policy should reduce that harm. That is important, but incomplete, because it omits the other direction: the economy is also dependent on Earth systems.

This broader framing raises the question of the total value of nature. We have one planet, and we cannot simply relocate off Earth and expect economic life to function normally. I used this point in my keynote by comparing The Martian and Project Hail Mary. They are by the same author, but they offer two perspectives on how hard life becomes without Earth’s support systems.

Last class, we began with GTAP. The key point is that GTAP provides a shared modeling language used widely across global finance, trade, and policy institutions. If you scan the list of users, you will see major organizations such as McKinsey and the World Bank. Career incentives aside, the important takeaway is that this is a heavily used, high-impact tool.

We also briefly defined CGE: computable, general, and equilibrium. Computable means it can be solved in finite time with real data. General means all major parts of the economy are linked. Equilibrium means behavior follows familiar economic structure: utility-maximizing consumers, profit-maximizing producers, and market clearing through prices where supply equals demand.

I have been showing an intentionally dense GTAP flow diagram. We are not memorizing it, but it is useful because it tracks value flows in detail. Conceptually, it is a circular flow diagram with many additions. In Econ 101, we draw factor and product markets and show money flowing opposite to goods and services. GTAP keeps that foundation, but adds institutional and sectoral detail.

At the core are still two agents: households and producers. Households supply labor and capital, producers use those inputs to make goods, and households buy the outputs. The major addition is that producers do not only sell to households. They also sell to other producers. Those are intermediate goods, and they are central to real economies.

This is not a minor detail. Nearly every production process depends on other sectors. A farmer buys machinery. Machinery producers buy steel. Steel producers buy mined inputs and energy. The economy is an interdependent production network, not just final consumers buying goods at retail stores.

That is why GTAP and other CGE models track intermediate demand in detail. They also include government, taxes, and public expenditures, which basic competitive models abstract away from. Real economies do not operate without government. Governments tax, spend, transfer income, and purchase large quantities of goods and labor.

A colorful class example made this concrete: the U.S. strategic accumulation of dairy surplus, including very large stocks of stored cheese. Whether one finds the example amusing or not, it illustrates the same point: real economies include policy-driven purchasing and storage decisions that do not emerge from simple frictionless competition.

We also include savings and investment, which are fundamental to growth. Savings finance capital formation, which increases productive capacity over time. Finally, and most relevant for today, countries do not operate in isolation.

This brings us to autarky, meaning no economic exchange with other countries. In practice, that is rare. To model actual economies, we must model cross-border trade. Each country has its own circular flow, but those national systems are linked through imports and exports.

Historically, this is why GTAP began as the Global Trade Analysis Project. It has expanded since then, but trade remains foundational. When major policy events happen, such as U.S.-China trade conflicts or Brexit, analysts turn to models like GTAP to estimate winners, losers, and aggregate effects.

One caution is crucial: complex models can be configured in ways that support opposing political narratives. Different assumptions can produce very different conclusions. That is exactly why model literacy matters. You need to evaluate assumptions, not just headline results. Even so, the broad professional consensus is clear: trade wars usually reduce total surplus, even if particular sectors gain.

Before discussing trade mechanics, we need to connect trade to Earth-economy analysis. The key distinction is between consumption and production impacts. Consumption itself often generates little direct environmental harm relative to production. Production is where many externalities occur: pollution, land conversion, and biodiversity loss.

That asymmetry means wealthy importing countries can appear environmentally clean while shifting environmental damage abroad. Palm oil is a classic example. It is used in many processed goods, and expanded production has often required forest clearing in countries such as Indonesia and Brazil.

So if a country consumes palm-oil-intensive goods but does not clear forests domestically, it may look environmentally better in domestic accounting. But that framing is incomplete. Consumption is still causally linked to production impacts abroad. Trade analysis is therefore essential for assigning responsibility and predicting where damage occurs.

The logic of trade starts with comparative advantage. You learned the distinction between comparative and absolute advantage early in economics: even if one country is more productive in everything, gains from trade can still arise because relative productivity differs across goods. Countries gain by specializing where their relative efficiency is greatest, or where their disadvantage is smallest.

That implies specialization across countries, not just within them. The standard free-trade argument compares outcomes along a spectrum from autarky to very open trade. On one end, no trade. On the other, minimal barriers: low tariffs, limited quotas, and fewer restrictions.

This spectrum has always been politically contested. Before 2016, many countries were moving toward greater openness. Since then, protectionist policies have become more prominent in many places. North Korea remains one of the clearest modern examples of near-autarky, though even it has external trade ties. Economies such as the U.S. and Hong Kong have historically represented more open regimes, though positions shift over time.

Now consider a two-country soybean example. Country 1, call it Soylandia, has comparative advantage in soy production. Country 2 is relatively less efficient. Under autarky, each country has its own supply-demand equilibrium. Soylandia has lower equilibrium prices. Country 2 has higher prices.

When trade opens, a world market emerges. Producers in Soylandia can sell abroad when foreign buyers pay more than domestic buyers, net of shipping costs. Shipping is often cheap relative to production value, so those arbitrage opportunities matter.

In this setup, we derive world-market equilibrium from excess supply in the exporter and excess demand in the importer. Graphically, this can be built through horizontal summation, but the intuition is enough: exports from the low-price country meet import demand from the high-price country.

The resulting world price lies between the two autarky prices. For Soylandia, domestic price rises relative to autarky, domestic consumption falls, and production rises; the difference is exports. Consumers in Soylandia lose from higher prices, while producers gain from expanded sales and higher returns.

In Country 2, the domestic price falls relative to autarky. Consumers gain from lower prices and higher consumption. Domestic producers lose because import competition displaces some local output.

When we aggregate consumer and producer surplus across both countries, total surplus increases. That is the central efficiency argument for trade. Resources shift toward higher-productivity production, and global output composition better matches comparative advantage.

At the same time, distributional conflict is real. In Soylandia, consumers may oppose freer trade because they pay more for soy. Producers support it. In the importing country, consumers support openness and producers resist it. Politically, producers often have stronger lobbying capacity because they are concentrated and organized, while consumers are diffuse. That asymmetry helps explain why trade policy is persistently contentious.

Now return to environmental implications. A crucial concept is embedded flows. If producing a traded good causes externalities, then importing that good embeds those impacts in the consumption bundle of the importing country. The damage occurs where production happens, not where final consumption is observed.

In the soy example, if trade-induced production expansion is achieved through deforestation, the exporting country experiences forest loss and associated ecosystem service declines. Those losses are often hard to value in markets, which is exactly why ecosystem-service modeling is difficult and necessary.

The same applies to biodiversity. Consumption in one place can drive species loss elsewhere through production expansion in regions with high comparative advantage. This is commonly called embedded biodiversity loss.

A second major concept is leakage. Leakage occurs when environmental protection in one country leads to increased environmental degradation in another country.

Suppose a country protects forests or requires stronger riparian buffers that reduce sediment runoff. Those policies can produce genuine local benefits. In Minnesota, for instance, buffer-style policies have been used to reduce sediment loading, and models like InVEST have helped evaluate those benefits.

However, if domestic protection reduces local timber supply, domestic and world prices can rise. Higher world prices can make logging profitable elsewhere, including in sensitive ecosystems such as parts of the Amazon. In that case, some of the intended environmental gain is offset by increased damage abroad.

Leakage rates vary by context, but the policy implication is consistent: domestic success can overstate global success. If substantial leakage occurs, then apparent gains may be much smaller than expected. Policies with lower leakage can therefore be far more cost-effective in global environmental terms.

This is the practical reason we need general equilibrium analysis. We cannot evaluate these questions reliably with partial, single-market intuition alone. We need full-system models that trace price and quantity adjustments across sectors and countries.

That is our bridge to Wednesday. We now have a clear justification for using general equilibrium methods: they allow us to quantify embedded impacts and leakage rather than guessing. So the message is not pessimistic. Environmental protection still matters deeply, but we must design and evaluate policy with the right economic framework.

Transcript (Day 2)

All right, let’s get started. Welcome to day three of our lecture set on earth economy modeling. Today we’re focusing on why general equilibrium is central to sustainability analysis.

I have a new figure on screen, and I’ve been working on it a lot. Side note: back in 2025, I projected that today and tomorrow would be the busiest period of my career. Classes are ending, I just got back from Santiago, and, most importantly, the most important paper I’ve written is due to the editor tomorrow.

So you are seeing this content first. It is the same material I presented in Santiago and the same material going into a submission to the Proceedings of the National Academy of Sciences tomorrow.

I’m feeling okay now, but it was close. The model was running all night. It takes about 20 hours to run all components, and if it had failed, I would not have had results to submit. It succeeded this morning, so I am relieved, and we can talk through what came out of it.

We’ll start with trade, because that leads directly to general versus partial equilibrium. Then we’ll discuss why general equilibrium has become so important in sustainability conversations, including at the multilateral level.

First, what is the difference between partial and general equilibrium? It sounds technical, but it is fundamental.

Partial equilibrium is what most people first learn in economics. In Econ 101, when you draw a supply and demand graph, that is, by definition, a partial equilibrium model. You likely saw this with the ceteris paribus assumption: all else held constant.

If you are looking at one market and just its supply and demand, you are holding much of the rest of the economy fixed. Partial equilibrium is often used because it is much easier to do.

So, partial equilibrium: one market, all else fixed.

You can go beyond a single market and still be in partial equilibrium. A broader definition is that some subset of the economy is fixed. You might model linked markets, like forestry and paper, but you are still not modeling the whole economy.

General equilibrium is the opposite: all markets are linked. In principle, nothing is fixed.

In a partial equilibrium model of timber and paper, we assume everything else stays unchanged, such as trade relationships or unrelated sectors. In general equilibrium, the choice to produce in timber and paper is affected by everything else.

The easiest linkage to think about is shared inputs. Labor can move across sectors. Someone working in a paper mill could work in automobile manufacturing, so what happens in cars affects paper.

That is general equilibrium.

Partial equilibrium is fine when spillovers are small. That is often true at the level of an individual firm. In microeconomics, we call firms price takers, meaning they do not individually set market prices. So most everyday analysis works under partial equilibrium.

But when a shock in one sector ripples through labor prices, trade, or complementary products, you need general equilibrium.

I’m emphasizing this because many environmental examples are exactly this case: many markets, many ripple effects. Let me give a specific story from Minnesota, our department, and ethanol.

Who knows ethanol? You see it at the pump. You can make fuel from corn, sugarcane, or potentially cellulosic feedstocks like tree fiber. In Minnesota, we have policies requiring ethanol blending, including E15 and E85 pumps.

My anti-environmentalist alter ego loves E85 because if an engine is tuned for it, it can produce much more power. But politically, ethanol is also a story of successful lobbying. The corn industry in the U.S. pushed very effectively for mandates such as blending requirements.

This is like other producer-supported policies: politically powerful producers can push rules that increase demand and raise prices.

Now suppose the government mandates ethanol in fuel. What does standard partial equilibrium say?

In an ethanol supply-and-demand graph, the mandate shifts demand outward. Price goes up, quantity goes up. This is textbook Econ 101.

That is why producer groups favored it: higher prices and higher output for ethanol producers.

Partial equilibrium can also show distributional effects in that market: producers gain, consumers of corn-based products pay more. You can trace this through producer and consumer surplus.

But the bigger point is that this has system-wide effects.

If the policy is large enough, the all-else-fixed assumption fails. More corn production pulls land from other uses, such as soybeans or pasture. Land use shifts.

One early concern was food prices. Corn gets more expensive, and goods using corn get more expensive.

Many environmental advocates argued that this cost might still be worth it if environmental benefits were large enough.

The problem is that there is also a general equilibrium environmental mechanism. Rising agricultural prices increase the profitability of land conversion globally. That induces land-use change in other countries.

In this case, ethanol mandates can increase deforestation in places like Brazil. That is a general equilibrium story: conservation leakage, spillovers, induced land-use change. You cannot analyze it correctly without a general equilibrium framework.

That is why earth economy modeling is built on general equilibrium logic. Biofuels are one example, but energy, land, water, and food markets are tightly coupled. Ignore linkages, and you can be very wrong.

This is personal for me because many environmentalists went all in on corn-based ethanol as a climate solution, and that conclusion was, in important ways, wrong. Researchers in Applied Economics at the University of Minnesota were among the early groups publishing findings on these broader market effects: food prices, land-use change, and more.

At the time, this was highly contentious. Industry groups pushed hard against those findings. That episode is a reminder of why tenure and academic freedom matter.

There was a lot of money behind the claim that ethanol was a win-win. General equilibrium analysis weakened that claim: possibly good for the industry, but harmful for the environment and for many consumers through higher food prices.

That argument has largely moved toward consensus. Most informed observers now agree that corn-based ethanol is likely a poor policy on net, for these reasons.

There is still active research on alternatives. Sugarcane may be more promising, and cellulosic ethanol, from residues or non-food biomass, remains a potential breakthrough. But corn into fuel instead of food is a canonical case where general equilibrium effects can overwhelm claimed benefits.

I still like E85 for horsepower.

So that is the motivation. Now let’s talk about implementation.

We have repeatedly discussed the circular flow diagram, and that is the mental starting point for a CGE model. The question is how to move from a stylized economy diagram to a model detailed enough for specific policy recommendations.

First, we add a component that GTAP calls the Regional Household. I’ll call it regional income allocation.

Given all income generated in a country, how much goes to private consumption, how much to savings, and how much to government?

CGE modeling prioritizes realism over simplicity, so this added detail is intentional.

We assign coefficients for each country that represent expenditure shares across private consumption, savings, and government. For example, the U.S. has relatively high consumption shares, Sweden has larger government shares, and China has high savings shares.

How do we set those shares? Calibration. We use expenditure data and choose coefficients so the solved model reproduces observed reality.

If you have taken macro, this corresponds to the C + I + G structure, with GDP often written as C + I + G + X - M.

At this point, it still does not look like circular flow, so we add production.

Now we connect private households to a full production system. The arrow from household to production is the value of domestic private purchases, a financial flow. In a complete circular-flow diagram, you would also show physical flows in the reverse direction.

Then we add sectoral detail: Producer 1 through Producer N. In GTAP, N is around 65 sectors.

Firms receive factors from households, such as labor, and they also buy and sell intermediate goods to each other.

Savings feed investment and capital accumulation. Government spending must also be represented accurately, including what sectors public demand supports.

Again, the point is not elegant theoretical minimalism. It is representing real-world complexity with enough fidelity to be policy relevant.

Then scale it globally: that circular-flow-plus-macro structure is replicated across about 160 countries, depending on data quality.

To do that, we need detailed input-output matrices, such as how much corn goes to ethanol or how much wood goes to paper.

That is why these models get computationally heavy. The one I ran overnight is this type of model.

We are not running the full model today. Instead, I built a smaller one that is easy to run in a browser and should work on computers, tablets, and probably phones.

On the course website, go to Games, then Circular Flow, then open the 3x3 CGEMES model.

This semester is a bit experimental pedagogically. The material is complex, closer to master’s-level methods, but I think upper-level undergrads can do it if we pair complexity with interactive toy models.

The slider-based tools seem to help, based on class feedback.

So here is what this toy CGE includes.

It is a miniature computable general equilibrium model inspired by early work from my mentor, Tom Hertel, and the original GTAP-style demonstrations of why trade data matters.

Instead of 160 countries, we have three regions: the United States, the European Union, and Rest of World.

Instead of 65 sectors, each region has three: agriculture, manufacturing, and services.

The original lesson of this 3x3 setup is that trade matters, and barriers to trade create inefficiencies. Trade liberalization can raise total welfare.

But there is a major asterisk: equity and distribution.

For a long time, economists from Adam Smith onward emphasized welfare gains from freer trade without adequately emphasizing who gains and who loses.

CGE models let us do both: estimate aggregate welfare effects and identify winners and losers.

Take NAFTA. You have probably heard it described as either a major success or the worst trade deal ever.

The evidence strongly supports that liberalization increased total economic surplus.

At the same time, there were serious distributional harms, including concentrated losses among manufacturing workers in specific places, such as parts of Detroit.

Recent work using quasi-experimental variation across cities suggests that exposure to NAFTA had measurable long-run health consequences, including increased mortality in harder-hit populations.

So the discontent tied to trade shocks is not imaginary; it appears in real outcomes.

That is exactly the nuance CGE analysis should keep front and center: aggregate gains can coexist with severe localized harm.

Now, in the model interface, scenarios are created by moving sliders.

One set controls tariffs, such as U.S. tariffs on imported agriculture, manufacturing, or services.

Another set controls productivity, capturing comparative advantage changes.

Others affect utility and production preferences, such as consumption weights and factor use.

Notice the equilibrium timer at the top. Every time you move a slider, the model re-solves.

Behind the scenes, it solves multiple linked market systems: domestic supply-demand blocks and international trade blocks.

Outputs include prices, welfare (for example real income), and bilateral trade flows calibrated to observed data.

Try a policy: set a 25% U.S. tariff on manufacturing imports.

Compared with baseline, U.S. manufacturing output rises. That is the core political argument for tariffs: revenue and domestic competitiveness.

But this comes with trade-offs. Other regions’ manufacturing falls. Within the U.S., resources shift across sectors, so some domestic sectors lose while others gain.

In one scenario, U.S. welfare rises by about 0.67%, while other regions lose.

This is the classic CGE result: a country can gain under some unilateral policies while imposing losses elsewhere.

But if others retaliate with reciprocal tariffs, those initial gains can disappear. In the tit-for-tat scenario, everyone can end up worse off.

That supports the standard free-trade result: retaliation risk can turn apparent unilateral gains into global and even domestic losses.

As a humorous real-world aside, policy lists have even included places like the Heard and McDonald Islands, which have no permanent human population. That is what happens when broad policy rules get implemented mechanically.

Still, the core economic point stands: reciprocal trade barriers can erase gains.

So this toy model is policy relevant. These are the kinds of tools analysts run when trade wars are announced.

Different teams may interpret scenarios differently, but the shared modeling language is often CGE.

No assignment on this specific tool today. It is there so you can experiment directly with general equilibrium logic.

When we scale to many countries and sectors, run times become long, but we gain much more detailed insight into who wins and who loses.

Now to the model I ran overnight. It is a CGE, but we call it an earth economy model because it is coupled with natural-system models from this course: ecosystem services and land-use change.

The main coupling channel is land.

That is why Professor Seiford-Rungie discussed land in economic thought. We are writing a related paper together.

The punchline is that growth models are increasingly moving beyond just capital and labor.

Instead of production as F(K, L), we need something like F(K, L, R), where R represents resources, including land.

Adding land is the key bridge between economic and environmental systems.

Historically, land entered CGEs first as a dollar value, which was natural because economic accounting is value-based.

But land is physical and spatial. You cannot have infinite dollars of land without area.

So next-generation models tracked physical extent, such as hectares.

Then models improved further by recognizing heterogeneity: one hectare is not equivalent everywhere.

Land quality and location matter. Agricultural productivity differs across space, and so does economic value.

That led to convergence between economics and Earth science, including satellite remote sensing and GIS data sources like those you have used in InVEST work.

Tomorrow we will pick up the technical piece: building land supply curves from detailed spatial data, not just market aggregates. That substantially improves realism and accuracy.

I’ll end a few minutes early for course evaluations.

Please check your email for the student response form and fill it out.

This course matters to me. We are trying something unusual by teaching earth economy modeling as an integrated framework, not only standard environmental economics. Your feedback will help shape how we improve and whether the department supports future versions.

There is no pressure, but I would sincerely appreciate your responses.

Last reminders: your rough draft is due soon. Check Canvas feedback on prior work. Most grading weight is on the final version, so the rough draft is your chance to get actionable comments early and reduce stress before final submission.

If your feedback is not visible yet, send me an email and I will make sure it is published.

Thanks, everyone. Have a great Wednesday.

Transcript (Day 3)

All right, let’s get started. As people wander in, they can catch up.

This is our last lecture day. You probably already know this, but presentations will happen on the last actual class day, Monday. One quick note: please email me your presentation slides, or submit them in Canvas if you prefer and I will post a submission box, at least 10 minutes before class on Monday. That gives me time to combine everything.

We are doing lightning talks. These are very common in academia: you condense a lot of information into a short time, not by talking fast, but by focusing on highlights. You will each have 5 minutes. Based on your own style, that is probably about 3 to 6 slides. I will cut you off at exactly 5 minutes so everyone gets equal time. This is an important professional skill; I have done hundreds of lightning talks in my career.

Any questions about that, the presentation, or the rough draft due tonight? I am fairly flexible on the draft deadline.

Student question: Do we need something for every section, or can it be partial?

Answer: Partial is fine. For the draft, I just need to see substantial progress toward the final version. You do not need every piece complete. This is mainly a progress check, and you will get full credit if you have made strong progress.

Any other questions?

All right.

Agenda for today: first, a quick update from my life because it has been all-consuming. This is the manuscript I submitted three minutes before yesterday’s deadline. There are 26 authors. Beyond the writing, the real act of science often feels like herding cats, getting 26 people to sign off at the end, confirm affiliations, and finalize details. But it worked, so that is good.

The timing is great because I was writing this while we were discussing related concepts in class. We will end today with some of the newest results from that paper. This is close to the cutting edge right now, so it is fun to share.

Before we get there, we will close out our discussion of land as a key input to production and to sustainability questions. Land is central to many environmental debates, so it deserves substantial attention. We will discuss how land is modeled in general equilibrium frameworks.

I also spent too much time making a land-use-change game. This one will not be graded, because I do not want to add another assignment while you are finishing presentations, but we will look at it in class. Then we will finish with results from the latest paper.

So where did we leave off?

Last class, we discussed how land had a long history in economics. Early economists treated it as a fundamental source of value, but later economics often shifted to labor and capital, and in some models mostly capital, such as Ramsey-style growth frameworks. Partly this reflects historical context: modern economics developed through World War II and the Cold War, with intense focus on industrial production and capital accumulation. Tanks, planes, and military output made industrial capital the priority, and land seemed less central.

As we moved beyond that era and began confronting ecological limits, land came back into focus. Computable general equilibrium models (CGEs) were among the first economic models to take this seriously again. Other areas are catching up now, but CGEs were early leaders.

In our work, we use GTAP, specifically GTAP-AEZ. This framework breaks the world into countries and then subdivides countries into agro-ecological zones (AEZs). AEZs are based on climatic conditions and growing characteristics: arid versus humid conditions, tropical versus temperate versus boreal zones, and growing-season length. Within an AEZ, growing conditions are relatively similar.

Each AEZ tracks land availability in physical area terms (hectares), not just dollars, and it tracks land type and quality: what can actually be produced there. So land is differentiated. Different places are better at producing different bundles of goods. That is obvious in reality, and AEZs provide a practical way to encode it in general equilibrium modeling.

One surprising thing when I started this work was that I assumed there would already be a clear economic treatment of deforestation. It is so important in environmental debates that it seems obvious. But even GTAP-AEZ effectively treated total land used by the economy as fixed. Sectors could compete for that fixed land, soy versus wheat, for example, but there was no mechanism making total land use endogenous.

By endogenous, we mean determined by supply and demand conditions. So one key modification we made was introducing a land supply curve in each AEZ.

Think of this as supply and demand, but now for land supplied to economic use. We no longer fix hectares in use. Instead, land enters economic use as a function of returns.

You generally cannot create new land in the usual sense, though there are exceptions like land reclamation in places such as the UAE and the Netherlands. Here, the key is not creating land, but converting natural land into economic uses.

This was feasible because we had rich geospatial information on AEZ land attributes. We estimated crop yields across grid cells and used that to infer the shape of land supply: how profitable conditions need to be for the next increment of land to enter economic use.

This mattered because it let us analyze protected-area policies. There has been major policy discussion about protecting 30% of Earth by 2030. We are now close to 2030 and not close to that target. Some proposals go further, suggesting Half-Earth protection to preserve ecosystem integrity.

What was often missing from these large protection goals was cost analysis. In our framework, one way to represent policy is with a land supply curve asymptotic to total usable land. Some land is effectively non-arable. A protection policy can be modeled by removing specific pixels from availability, which shifts the asymptote left and shifts supply from S to S-prime.

What happens in equilibrium? Econ 101 gives the answer. With a demand curve, equilibrium moves to lower land use in the economy and a higher land price. Reduced land use is expected under protection constraints, but higher prices create distributional effects, especially for low-income consumers who spend a larger share of income on food.

This mechanism sounds straightforward, but earlier versions of these models lacked it because land use was fixed.

Student question: Does this framework account for environmentally friendly products and more aware consumer spending?

Answer: Yes. Suppose we distinguish conventional soybean production, including expansion into sensitive forests, from certified forest-friendly soy. Then we can model green soy as a substitutable product. If demand rises for the environmentally preferred substitute, demand falls for the harmful alternative. This is still basic Econ 101: higher demand for one substitute shifts demand away from the other. So the framework can evaluate exactly that kind of policy and preference shift.

That is the economic side. But Earth-economy modeling links economic and environmental systems, including models like InVEST. The missing link was land-use change at high enough resolution.

Historically, we lacked realistic methods to take coarse global estimates, for example regional changes from GTAP-AEZ, and translate them into high-resolution maps suitable for InVEST and related tools. Some of you saw this computational issue in your projects: once countries are large, millions or billions of grid cells become heavy even on gaming laptops. The hard part was programming.

We built a land-use-change model called SEALS: Spatial Economic Allocation Landscape Simulator. Yes, we chose the name because I like seals and reverse-engineered the acronym.

What does SEALS do? It downscales coarse inputs to high-resolution spatial outcomes. Inputs might be land-use change totals by AEZ, or coarse gridded outputs from Earth system models, often around one arc degree (about 110 km).

Conceptually, imagine coarse cells showing where grassland expands or contracts. SEALS produces high-resolution maps consistent with those aggregate constraints. If grassland contracts somewhere, another class like cropland may be expanding there. Land transitions are jointly allocated.

Under the hood, there are many details, but the structure is two-step downscaling using covariates.

Step one is coarse allocation from regional totals into coarse grids, guided by economic logic and relative profitability.

Step two is fine allocation using high-resolution predictors. This stage uses machine learning based on historical land-use and land-cover time series: learn transition patterns from the past and project likely locations conditional on features such as market access, slope, topography, soils, and bioregions.

The goal was not only accuracy and global scalability, but policy relevance. For example, if you designate a new protected area, you can run scenarios with and without that policy. Without protection, urban or cropland might expand inside that area. With protection, expansion there is blocked, but total change still has to occur somewhere else, so pressure shifts outside boundaries.

A key variable in most land-use models is adjacency. Cropland often expands at existing cropland edges. If you have two agricultural patches, expansion probabilities are highest around their margins, with overlapping influence where both are nearby.

We generalized this using adjacency functions for each class pair, estimating how nearby classes affect transition probabilities. For example, proximity to agriculture often increases odds of more agriculture. Proximity to cities can have a non-monotonic effect: very close to urban areas, cropland may decline due to suburban expansion; very far from cities, cropland may also be less likely because market access is weak. So expansion can peak at intermediate distance.

We extended these effects from one dimension to two-dimensional suitability surfaces and applied them across the landscape. That generated strong predictive performance and gave us a practical lens for forward-looking analysis: where might deforestation-driven biodiversity loss occur if patterns continue?

At global scale, this required calculations over 8.4 billion grid cells. The next version targets 10-meter resolution, meaning well over a trillion grid cells. That is why we prototype on fast machines and run full analyses on high-performance computing systems.

Still, intuition matters more than technical details in this class, so let’s use a simpler simulation.

I built you a land-use-change simulation app. You can access it from our site under Games and Apps, in the Land Use section. There are two new entries: a simulation and a game. Open the simulation first.

This app is not as feature-rich as SEALS, but it captures the core dynamics and is increasingly relevant for conservation planning.

You are looking at a land-use and land-cover map. Forest (green), agriculture, urban, and protected areas are represented. Forest stores carbon and now includes age dynamics, older trees store more carbon. Agriculture stores less carbon but produces revenue. Urban stores much less carbon but produces much higher revenue. That is the core Earth-versus-economy trade-off.

Under the hood, each land-use class has a suitability map explaining why a city appears in one location and not another. Suitability might represent soil, terrain, or other conditions. Turn off suitability and you see implied conversion pressure based on relative returns.

Given food prices, carbon prices, and urban returns, pressing Next Period simulates one step forward. With defaults, food prices are high enough that cropland expands. Once food demand is sufficiently met, urban expansion tends to accelerate, following urban suitability.

The model tracks multiple outcomes: revenue, carbon stocks, food security, and biodiversity. In one run, revenue rose with urban growth, carbon eventually fell, food security stayed stable, and biodiversity declined.

If you run Auto into the future, you often see a tragedy-of-the-commons pattern: private incentives favor conversion of natural land into higher-revenue uses, while biodiversity and carbon decline.

This is why externalities matter. Economists are often caricatured as anti-environment, but economic tools can diagnose and manage these failures. If market incentives ignore climate and biodiversity costs, private returns dominate social welfare. Policy should correct that gap.

A carbon tax is one example. If high enough, it can price part of the externality and slow harmful conversion. But if set too high in a simplistic model, you can collapse economic activity. That illustrates trade-offs: maximizing carbon preservation alone is not the same as maximizing social welfare with people included.

Student experiment: If food prices, urban returns, and GDP growth are all high, a given tax rate may no longer be enough.

Answer: Exactly. If private returns rise, the tax required to offset them rises as well.

You also observed path dependence: without nearby farms, new farms may not emerge because isolated conversion is less attractive. Add one farm, and local dynamics can trigger more expansion nearby. This reflects the adjacency mechanisms discussed earlier.

In the full workflow, GTAP provides demand and prices for land uses; the land-use model allocates where changes occur; and resulting land-use maps feed into environmental models like InVEST. This closes the Earth-economy loop and supports policy optimization rather than unmanaged externalities.

The simulation is useful, but the game version is more engaging.

In the Land Use Game, you get scenarios, balanced baseline, heavy deforestation pressure, Green New Deal-style settings, and no-law settings, and you try to maximize score.

Your score is social welfare over time: economic revenue plus conservation value from carbon and biodiversity. That differs from individual private profitability, and the whole point is to think like a social planner.

You have a budget of limited placements, manual land modifications, and a 60-step horizon. Strategy matters.

In one run, I spent placements adding forest and still fought underlying growth pressure. I ran out of budget quickly, which mirrors real conservation underfunding. I scored 3972.

Class discussion showed different strategies: adjusting tax rates, placing protected areas near expansion frontiers, and tuning constraints produced different outcomes, with some scores above mine.

This is a simplified game, but the underlying dynamics are close to what serious models compute. It is an intuition builder for linking economic and environmental systems.

Now, stepping back to the full model chain:

We covered the land-use-change component and earlier covered InVEST. The broader pipeline links multiple models, including integrated assessment components, to produce quantitative estimates of environmental and macroeconomic impacts.

We added growth dynamics as well: agents choose investment, and next-period capital equals depreciated current capital plus new investment, the core Ramsey logic.

The implementation challenge was substantial: we coupled five models, and making them run together was largely a programming and systems integration task. The payoff was informative scenario results.

We ran Network for Greening the Financial System (NGFS) scenarios. NGFS includes many central banks that increasingly recognize nature-related systemic risk: if natural systems destabilize, economies and finance destabilize.

These scenarios sit on top of SSP-style pathways but add more policy detail, including climate policy assumptions such as carbon pricing.

For each scenario, we compared dependency assumptions: no nature dependency (the simplifying assumption still common in many models), climate-only dependency, ecosystem-services-only dependency, and combined climate-plus-ecosystem dependency, which is the most realistic.

Key result: under current policies, projected GDP losses are large. Other policy pathways reduce losses but often remain negative. Distributional effects are severe: lower-income countries are hit hardest.

Some higher-income countries can appear less harmed, or even slightly benefited in specific channels, due to shifting productivity patterns under warming, but globally this creates instability, migration pressure, and equity problems.

Land-use outputs align with this story. Under current policies, natural land declines while cropland expands substantially. Under stronger mitigation and conservation pathways, including net-zero style strategies, natural land can stabilize or increase, partly through conservation incentives.

We also ran a stress-test scenario, common in finance practice. Stress testing asks whether systems break under abrupt shocks.

In our case, we imposed an unexpected acceleration of climate damages around 2035, shifting from a milder to a harsher climate trajectory. Damages increased more than linearly, showing compounding risk and nonlinear feedbacks that banks and policymakers should care about.

So, some concluding thoughts for the semester.

First, thank you for your feedback and engagement. This was rewarding and also difficult, first-time teaching is much more work because content is built as you go. Over time, this course will be refined into an interactive online textbook with integrated web apps.

If I leave you with two core points:

One, modern economics must operate within the donut: above the social foundation and below the ecological ceiling. Endless growth framing is no longer sufficient, but neither is ignoring growth. The task is managing complex trade-offs while preserving social stability and ecological integrity.

Two, there is real momentum. Central banks, investors, and finance ministries are beginning to treat nature-related risk seriously. It is early, and timing is uncertain, but this is a rapidly growing field with expanding job opportunities in sustainability risk, nature dependency analysis, and policy design.

I know that sounds utilitarian, and I assume most of you are not here just for labor market outcomes. My hope is that this course gave you a clearer picture of how environmental and natural resource economics fits into broader sustainability practice, and where the field is heading.

Thank you all. Please email your presentation slides before class so I can merge them into one deck and keep transitions smooth.

Any last questions?

Great semester. It has been fun.