RESEARCH DOSSIER

Market Clearing and Partial Equilibrium Comparative Statics

Abstract. Market clearing is a conditional consistency relation between desired purchases and desired sales at a specified price, under specified institutions and over a specified period. Partial equilibrium comparative statics derives the response of that relation to changes in its determinants while treating the surrounding economy through an explicit closure. This monograph develops the scalar excess-demand approach, the existence and uniqueness of market-clearing prices, local derivatives and global monotonicity, stability under alternative adjustment mechanisms, incidence, constrained allocation, welfare, and empirical identification. It extends the analysis to capacities, administered prices, rationing, trade wedges, imperfect competition, selection, information friction, and linked markets. Historical foundations and comparative studies from the United States, India, Spain, and Germany establish the institutional variation of the analytical objects. Reproducible illustrative calculations distinguish a change in potential demand from a change in actual allocation and from a change in social surplus. The original synthesis integrates mathematical conditions, measurement, allocation rules, and the distinction between equilibrium response and causal identification. Its principal finding is that the defensible scope of partial equilibrium depends on the closure of the market, the regularity of the equilibrium, the allocation mechanism, and the identification of the initiating shock. These conditions determine what a comparative-static estimate can establish and what additional evidence is required. The work synthesizes established theory, independent derivations, and reported empirical findings; it supplies no newly estimated empirical parameters or named theory.

Keywords: Market clearing; partial equilibrium; excess demand; comparative statics; elasticities; tax incidence; pass-through; rationing; consumer surplus; producer surplus; equilibrium stability; market definition; causal identification; welfare economics; microeconomics

The analytical object and market boundaries

A market is an institutional domain within which participants can exchange specified claims, goods, services, or factor inputs under defined rules. An economically meaningful market requires a unit of quantity, a price convention, a geographic or network boundary, a time interval, and a characterization of the product. A megawatt-hour delivered at a congested node this evening differs from a megawatt-hour delivered tomorrow at a different node. An occupied rent-regulated apartment differs from a vacant apartment offered on the unregulated rental market. A kilogram of fresh fish at a landing site is perishable stock, whereas the annual harvest is a flow supported by capital and labor. Combining these quantities into one undifferentiated demand schedule may erase the very mechanism that determines the response to a shock.

Let D(p,x) denote aggregate desired purchases and S(p,y) aggregate desired sales over the same time interval, where p is a common transaction price, x contains demand determinants, and y contains supply determinants. Both schedules count compatible physical units per period. Excess demand is defined by

Z(p,x,y)=D(p,x)−S(p,y).Z(p,x,y)=D(p,x)-S(p,y).

An interior competitive clearing price p* solves Z(p*,x,y)=0. The associated volume q* equals D(p*,x)=S(p*,y). These are behavioral plans evaluated at the same price. Observed completed purchases and sales are equal as an accounting identity even in a shortage. A queue, an unfilled order, and a rejected application may coexist with that equality. The identification of a clearing price therefore requires evidence about intended quantities, access constraints, or the institutional process that matches supply and demand.

Partial equilibrium specifies which elements of x and y remain fixed and which are allowed to move. A standard closure fixes incomes, competing goods' prices, factor prices, expectations, contractual rules, and the number of firms over the relevant horizon. A richer closure can make some of these objects endogenous. The economist must identify the boundary between the selected market and its environment, rather than presume that the adjective partial denotes a universally adequate simplification.

Competitive behavior means that individual participants take the relevant price as given when deciding how much to buy or sell. Aggregate supply can be generated from firms maximizing profit given their technology and constraints; aggregate demand can be generated from consumers maximizing preferences over feasible bundles. Their summation assumes that the same price applies after accounting for legitimate delivery or quality differences. Heterogeneous transactions can instead require a vector of prices and a correspondence of feasible matches.

The clearing equation is an equilibrium condition, not a theory of exchange institutions by itself. Double auctions, brokers, posted prices, procurement platforms, negotiated contracts, and rationing authorities can implement different mechanisms. Some institutions discover a common price; others clear a network of quantities through differentiated prices or explicit allocation rules. The scientific inquiry begins with the conditions under which reducing these institutional objects to one scalar price is justified.

Historical foundations and economic time

Alfred Marshall's 1920 edition of Principles of Economics distinguishes the equilibrium of a market with an already existing stock from normal supply under a longer period of production adjustment. His discussion of normal demand and supply (book V, chapter III) connects the direction of production adjustment to the difference between demand price and supply price at a quantity and recognizes the possibility of declining normal supply prices. The historical point is that the shape of supply belongs to a specified economic time horizon. An increase in demand during a market day may largely increase price; a persistent increase can induce capacity expansion and entry. These are different conditional experiments.

The mathematical relation between quantity and price adjustment is central. A price-adjustment mechanism takes quantities desired at a price as given and moves price toward a point of equality. A quantity-adjustment mechanism compares the price buyers would pay for an output level with the price that sustains that output. Both can represent market coordination when their schedules and rules are compatible. Their stability requirements can diverge when supply or demand has unusual slopes, when long-run supply reflects changing industry costs, or when the inverses are multivalued.

The development of general equilibrium gave the closure problem an explicit mathematical form. Arrow and Debreu (1954) investigate the existence of equilibrium in an integrated economy where consumers, producers, and markets are determined simultaneously. A scalar crossing of supply and demand is a special analytical operation with the rest of that system held fixed or summarized. It does not establish that the full vector of equilibrium prices and allocations exists or is unique after a change in the focal market. The difference matters most where input demands, incomes, and expenditure shares react substantially.

Modern comparative statics broadens the available tools. The opening discussion in Milgrom and Shannon (1994) examines the limitations of differential methods and motivates order-based comparative statics. For this dossier, an elementary scalar monotonicity argument is sufficient to handle many nondifferentiable market schedules. The content of generalized lattice methods belongs to a distinct mathematical-economics investigation; the present use is limited to the market-clearing implications and their boundaries.

These historical developments support a unified reading of economic time: a market schedule is a description of feasible behavioral responses at a specified stage of adjustment. Movement to a new intersection, entry of new producers, and revision of expectations should be distinguished. The initial equilibrium calculation frequently remains informative as a conditional benchmark even when the subsequent trajectory requires a richer model.

Existence, uniqueness, and equilibrium derivatives

Existence and boundaries

Assume the price lies in a closed interval [l,u], that Z is continuous there, and that Z(l) is nonnegative while Z(u) is nonpositive. The intermediate value theorem yields at least one clearing price. If Z is strictly decreasing throughout the interval, the root is unique: two distinct roots would contradict strict monotonicity. Existence and uniqueness are separate claims; continuity does not imply either endpoint signs or monotonicity. At a zero lower price, free disposal can allow excess supply. If prices are bounded above by law, an excess-demand condition at the upper endpoint can identify scarcity rather than an interior clearing price.

For zero-price free disposal, one can write p >= 0, S(p)-D(p) >= 0, and p times [S(p)-D(p)] = 0. This complementary relation requires the capacity to discard or retain unwanted goods without violating the resource description. It applies to a zero-price surplus, not to a shortage at a binding ceiling. A price ceiling normally requires an explicit rationing variable or shadow price.

Consider discrete units: D falls in steps as reservation prices are crossed and S rises as additional sellers become willing to trade. At a reservation-price tie, participants can be indifferent and quantities may be set-valued. An equilibrium may be represented by a price and selected quantities from demand and supply correspondences with equal total volume. A unique price is not necessary for a unique efficient transaction count. Conversely, indivisible quantities and restrictions on allocation can obstruct clearing even if a continuous approximation crosses zero.

Local comparative statics

Let theta be a scalar exogenous parameter affecting the schedules. If Z is continuously differentiable near an interior root and Z_p is nonzero there, the implicit function theorem supplies a locally differentiable price branch:

dp∗dθ=−ZθZp=Dθ−SθSp−Dp.{dp^*\over d\theta}=-{Z_\theta\over Z_p} ={D_\theta-S_\theta\over S_p-D_p}.

The volume derivative follows from either side of the identity:

dq∗dθ=Dθ+Dpdp∗dθ=SpDθ−DpSθSp−Dp.{dq^*\over d\theta}=D_\theta+D_p{dp^*\over d\theta} ={S_p D_\theta-D_p S_\theta\over S_p-D_p}.

All partial derivatives are evaluated at the reference root. Under the ordinary signs D_p < 0 and S_p > 0, a positive demand intercept shock raises both price and quantity if the supply curve does not move. A positive supply intercept shock lowers price and raises quantity when demand is fixed. If both schedules shift, their net horizontal change determines the price response; their weighted combination determines the quantity response.

With A=-D_p>0 and B=S_p>0, the first-order approximation for small shifts dD0 and dS0 is dp=(dD0-dS0)/(A+B) and dq=(B dD0+A dS0)/(A+B). This decomposition identifies the slope weighting of identical physical shifts. A price rise alone cannot distinguish increased demand from reduced supply. A quantity rise alone cannot distinguish increased demand from increased supply. Under the single-shock assumption, the joint signs of price and quantity distinguish the four elementary shifts. With known slopes and purely additive shifts, the two observations identify the two intercept changes locally: dD0=dq+A dp and dS0=dq−B dp. Without those restrictions, changes in slopes, unobserved prices, composition or simultaneous structural disturbances can produce observationally equivalent outcomes.

A slope has dimensions of quantity per currency unit. At a positive reference price p and volume q, define the signed demand elasticity epsilon_D = D_p p/q and supply elasticity epsilon_S = S_p p/q. A positive proportional demand shifter d ln a entering multiplicatively gives d ln p / d ln a = 1/(epsilon_S-epsilon_D), and d ln q / d ln a = epsilon_S/(epsilon_S-epsilon_D). The elasticities are local and horizon-specific. Replacing one by an uncompensated estimate from a different population can invalidate even the sign prediction in richer linked systems.

Global responses and singular points

If Z is continuous, strictly decreasing, and has a unique root for every parameter in an interval, and an increase in theta raises Z at every price, its equilibrium price is nondecreasing in theta. Proof: at the old root the new excess demand is nonnegative. Because the new function is decreasing and crosses zero, its root is weakly to the right. This argument does not require derivatives and also covers kinks. Strict upward shifts imply strict price increases when both roots are interior. A uniform shift is sufficient, not necessary.

At Z_p=0, the implicit derivative may fail. For example, let u=p-p0 and Z=a-u^3, with p0 sufficiently positive locally. The unique root is p0 plus the real cube root of a. At a=0 the price map has no finite derivative although direction and continuity survive. For Z=a+u-u^3, three roots exist for a near zero; at a=0 they are u=-1,0,1. Under positive-speed price adjustment the outer roots are locally stable and the center unstable. A small parameter change can preserve several equilibria; a sufficiently large one may eliminate a branch. The point of these illustrative excess-demand functions is that a derivative obtained on one branch cannot determine selection among branches.

Stability and alternative adjustment mechanisms

A comparative-static result compares equilibria. Stability adds a law of motion. For continuous price adjustment dp/dt=k Z(p), k>0, linearization about a regular root gives d(delta p)/dt=k Z_p delta p. The root is locally asymptotically stable when Z_p<0 and unstable when Z_p>0. In the scalar continuous case, the local rate of convergence is governed by k times the slope. With a globally decreasing continuous Z and a unique root in an invariant interval, the price vector field points toward the root on both sides.

For discrete adjustment p_(t+1)=p_t+k Z(p_t), the multiplier is 1+k Z_p. Stability requires its absolute value to be strictly below one. A market can possess a unique root that is stable under gradual continuous adjustment and unstable under an aggressive discrete adjustment. Market institutions determine whether there are auction rounds, price change limits, delays, inventories, or learning processes that make either model credible.

A distinct agricultural cobweb example uses planned supply based on the preceding period's expected price, qS_t=c+d p_(t-1), where d>0, and current demand qD_t=a-b p_t, b>0. Clearing in each period produces p_t=(a-c)/b-(d/b)p_(t-1). The steady-state multiplier is -d/b. Prices converge with alternating deviations only when d/b<1. This system clears at every transaction date but can fail to converge. The time interpretation of a supply schedule is thus indispensable: the predetermined supply for current trading and the steady-state supply curve encode different decisions.

The continuous quantity rule dq/dt=g[P_D(q)-P_S(q)], g>0, is locally stable when P_D'(q)-P_S'(q)<0. With strictly downward demand and strictly upward supply the price rule and quantity rule agree in their stability sign. If industry supply slopes downward, their conditions need not agree. A negative normal supply slope associated with external economies is not automatically the supply behavior of a price-taking firm operating on a fixed cost curve. Distinguish the industry equilibrium relation from an individual firm's decision technology.

None of these rules follows logically from the equilibrium condition. The speed of adjustment is an empirical question; transaction costs, inventories, expectations, and contracts govern it. An interpretation of comparative statics as a time path requires specifying a law of motion and establishing that the relevant equilibrium is reachable under it.

Policy wedges, price controls, and constrained allocation

Unit taxes and pass-through

Let p_b be the buyer's gross price and p_s the seller's net price. For a unit tax t, impose p_b=p_s+t and D(p_b)=S(p_s). At a regular interior competitive equilibrium with A=-D'(p_b)>0 and B=S'(p_s)>0,

dpbdt=BA+B,dpsdt=−AA+B,dqdt=−ABA+B.{dp_b\over dt}={B\over A+B},\qquad {dp_s\over dt}=-{A\over A+B},\qquad {dq\over dt}=-{AB\over A+B}.

The signs follow by differentiating the clearing identity and wedge. The fractions sum to one when expressed as the buyer's price increase and the seller's price decrease. A flatter demand curve in quantity-price coordinates means larger A; buyers reduce purchases more readily and bear less of a marginal tax under the stated closure. When quantity is fixed by perfectly inelastic demand, A=0, the gross price bears the entire marginal wedge if B>0. Fixed supply B=0 gives the converse when A>0.

For the infinitesimal tax at a zero wedge, elasticities evaluated at the common price yield the familiar buyer share epsilon_S/(epsilon_S+abs(epsilon_D)). At a positive wedge, elasticities evaluated at distinct gross and net prices require conversion back to slopes. A finite change requires integration of the derivatives along the relevant path and may pass through capacity constraints or entry thresholds. For an ad valorem tax, p_b=(1+tau)p_s; the derivative includes the changing tax base. Applying the unit-tax formula directly to a percentage rate is dimensionally wrong.

Statutory placement and economic incidence coincide only in special cases. Under the competitive model with full salience, equivalent gross-to-net wedges and unchanged enforcement generate the same clearing solution irrespective of which side remits the tax. The equivalence can fail when payment changes attention or compliance. Chetty, Looney, and Kroft (2009) report an 8% decrease in demand when grocery shelf labels included the sales tax. This experiment supports taking the perceived price mechanism seriously; it does not establish a universal eight-percent response to tax inclusion.

Price ceilings and floors

At a ceiling p_c below the unique unconstrained clearing price, D(p_c)>S(p_c) under ordinary slopes. If sellers supply the scheduled amount and no public supplement or unlawful sale occurs, actual transactions cannot exceed S(p_c). The amount of excess desired purchases is D(p_c)-S(p_c). Allocation becomes an additional variable. First-come-first-served queues, lotteries, eligibility priorities, political allocation, and resale can place the same physical goods with different users and generate distinct welfare outcomes.

A price floor above the clearing price analogously creates excess intended supply. With no government procurement, realized volume is at most D(p_f). Government acquisition can raise the volume produced but creates a fiscal expenditure and a stock that must be used, stored, or destroyed. The resulting resource and budget consequences must be included in welfare. A mandated minimum price with free disposal is institutionally different from a binding commitment to purchase any amount offered.

A capacity constraint q<=K can be represented by a competitive supply schedule capped at K or by a complementarity condition. For strictly increasing marginal cost MC(q), at positive output q the efficient scarcity price satisfies p=MC(q)+lambda, lambda>=0, lambda(K-q)=0. The scarcity rent lambda is the marginal value of capacity under the simplified planner's objective. It differs from a regulatory price cap, which can prevent the observed monetary price from reflecting that value. If the cap and capacity are both binding, rationing and unserved demand matter.

Trade and quotas

For a small importing economy facing fixed world price p_w and no other trade costs, domestic imports equal D(p_w)-S(p_w) when positive. A tariff t increases domestic price to p_w+t while imports remain positive and world price is unchanged. Domestic production rises, consumption falls, and government revenue equals t times imports. In a large economy, foreign export supply has finite elasticity and the world price may move. For a fixed import quota, domestic excess demand equals the quota; the wedge between domestic and world prices is a scarcity rent whose recipient depends on licensing, auctions, and contracts. Identical domestic prices and quantities can yield different national welfare when the rents accrue to different parties.

In a spatial network, transaction price differences can persist because of delivery costs and capacity constraints. If a quantity can move between two nodes at marginal cost c and there is an unconstrained competitive flow, the price difference equals c for a positive flow. If the route is saturated, it can exceed c by a congestion value. If no trade occurs, the relevant directional price difference cannot exceed c in a frictionless entry benchmark. These relations extend clearing from one nominal market to physically coupled locations.

Welfare and distribution

Surplus representation

For a homogeneous divisible good, let inverse demand P_D(q) describe marginal willingness to pay, and inverse supply P_S(q) describe marginal social opportunity cost under the maintained assumptions. With efficient sorting, no externalities, and a transferable monetary numeraire, total surplus is

W(q)=∫0q[PD(v)−PS(v)] dv.W(q)=\int_0^q[P_D(v)-P_S(v)]\,dv.

If P_D is decreasing and P_S increasing, an interior crossing maximizes W since W'(q)=P_D(q)-P_S(q) and W''(q)<0. The same condition gives the competitive clearing quantity if private motives coincide with social opportunity costs. The result is conditional on the interpretation of both curves. Marshallian willingness-to-pay integrals exactly represent money-metric welfare for quasilinear preferences under the usual fixed-numeraire closure. With substantial income effects, compensating and equivalent variations generally differ from the Marshallian area. Distribution also matters because a dollar to one person need not have the same social weight as a dollar to another.

At a common price p, consumer surplus is the integral of inverse demand above p up to q, while producer surplus is revenue minus the integral of marginal cost up to q. Producer surplus need not equal profit when fixed costs are incurred. Entry viability and long-run economic profit require fixed and sunk costs to be treated explicitly. With taxes, adding government revenue reconstructs resource surplus if tax receipts are treated as transfers and collection consumes no resources. Under an alternative social value of public funds, the revenue weight must be explicit.

At a preexisting unit tax t0, a further marginal increase changes resource surplus by W'(q)dq/dt=t0 dq/dt if there are no additional distortions. Starting from zero tax the first-order loss is zero; the second-order approximation is (1/2) t squared AB/(A+B). This is the ordinary local deadweight-loss triangle. It fails as a sufficient measure when other distortions have first-order effects, where agency costs matter, or where taxes correct external harms. A corrective tax can improve the resource allocation when private costs omit social damage and the tax aligns them at the relevant margin.

Rationing and transfers

A binding ceiling offers lower money prices to successful purchasers. Its welfare consequences depend on which potential purchasers receive the goods. Assume two potential buyers value an indivisible unit at 12 and 8, respectively; one unit costs 4 to produce. At a regulated price of 6 both demand it. Allocation to the high-value buyer yields resource surplus 8; allocation to the other yields 4; an equal lottery yields expected surplus 6. All three outcomes share the posted price and realized quantity. Monetary surplus goes to different individuals. Nonmonetary search costs reduce total benefits further. This example is an original illustrative calculation, not an empirical estimate.

Pareto improvement is a stronger statement than a positive sum of money-metric changes. A price change can increase total surplus while harming incumbent buyers or sellers. Actual compensation must be distinguished from hypothetical compensation. Distributional analysis requires baseline endowments, income changes, membership in the market, and a transparent normative choice about aggregation. An empirical study of prices does not supply those weights.

Extensions and counterexamples

Market power and curvature

A monopoly facing inverse demand P(q) and marginal cost C'(q) chooses interior output by P(q)+qP'(q)=C'(q)+t for a unit cost or tax t. Assume the objective is locally strictly concave so 2P'(q)+qP''(q)-C''(q)<0. Differentiation yields

dqdt=12P′+qP′′−C′′,dpdt=P′2P′+qP′′−C′′.{dq\over dt}={1\over 2P'+qP''-C''},\qquad {dp\over dt}={P'\over 2P'+qP''-C''}.

For linear demand and constant marginal cost, pass-through is one half. For isoelastic demand q=p^(-e) with e>1 and constant positive unit cost c+t, the optimal price is e(c+t)/(e-1); pass-through is e/(e-1)>1. The difference comes from demand curvature and the markup condition. Producer burden cannot be inferred by applying a competitive elasticity share to either example. Weyl and Fabinger (2013) extend incidence analysis to imperfect competition; the two explicit examples above reconstruct the basic curvature issue directly.

A monopsonist hiring labor faces an inverse labor supply w(L) and marginal revenue product R'(L). With a uniform wage, its first-order condition is R'(L)=w(L)+Lw'(L), while a competitive wage-taking employer uses R'(L)=w. A minimum wage between the monopsony wage and competitive wage can increase employment in a suitable model because the marginal hiring cost changes. For illustration, take w=2+L and R'=14-L. The competitive outcome is L=6, w=8; the unconstrained monopsony outcome solves 14-L=2+2L, yielding L=4 and w=6. A floor of 7 lets the firm hire up to five at that wage; the marginal product at five is 9, and the marginal expense immediately beyond five is 12, so employment rises to five. This result depends on uniform wage setting and an upward labor supply, not on a universal labor-market law.

Selection and product quality

Insurance provides a case where the cost of serving an additional participant differs from the average cost of the enrolled pool. If customers sort themselves according to willingness to pay and risk, premium variation affects both enrollment and the distribution of claims. A break-even average-cost schedule is not the same curve as the marginal resource cost of extending insurance to another person. Einav, Finkelstein, and Cullen (2010) use price variation together with cost observations to study selection and welfare in employer-provided insurance. They detect adverse selection but report a relatively small welfare cost in their particular application. An ordinary supply-demand diagram can organize this question only when it labels the different cost objects correctly.

A similar distinction applies to quality and matching. When a higher price changes product quality, delivery reliability, contract duration, screening intensity, or platform ranking, observed quantities refer to different goods. A demand shift estimated from such observations may combine willingness to pay and a change in the object purchased. A credible counterfactual specifies whether product quality responds endogenously and how that response is valued.

Linked markets and closure

Suppose a focal excess demand Z(p,r,theta)=0 is coupled to another market H(p,r,theta)=0; r is the surrounding price. Local elimination of r is possible where H_r is nonzero. The resulting price derivative is

dpdθ=−Zθ−ZrHθ/HrZp−ZrHp/Hr.{dp\over d\theta}=-{Z_\theta-Z_r H_\theta/H_r\over Z_p-Z_r H_p/H_r}.

This follows by solving dH=0 for dr and substituting in dZ=0. Holding r fixed sets both feedback terms to zero, which is a restrictive closure. Even if both own-price derivatives are negative, cross-price effects may weaken or reverse the denominator. For example, take Z=10-2p+r+theta and H=8+p-2r. At theta=0, r=(8+p)/2, so 14-1.5p=0; p=28/3 and r=26/3. The full price derivative is 2/3, compared with 1/2 under fixed r. The difference is entirely induced by feedback. If the cross terms have opposite signs, feedback can dampen rather than amplify the response.

The example establishes a practical test: identify the other market variables likely to respond, estimate or bound the feedback terms, and compare the induced change with the target effect. An apparently narrow sector can propagate through essential inputs; a large sector may sometimes admit a useful partial analysis over a limited horizon if critical outside determinants are fixed by contract or institutions.

Measurement and causal identification

Data dictionary and comparability

The following data requirements define the minimum information for measuring a market response. Each observation needs a time stamp and market identifier. Prices require currency, unit, tax treatment and deflation convention; quantities require physical or contract units, period and coverage. Demand-side records include population, income, substitute prices, eligibility and availability. Supply-side records include input costs, capacity, inventory, entry, weather and contractual obligations. Where a constraint binds, unmet orders, waiting times, cancellations, dispatch instructions, or stock changes measure margins omitted from completed transactions.

Object Operational measure Principal threat
Transaction price Gross or net price per comparable unit Fees, quality, discounts, salience
Quantity traded Matched or delivered units per interval Censoring, inventory, stock-flow mismatch
Desired demand Bids, reservation prices, orders, applications Strategic bids, missing unsuccessful buyers
Capacity Available and deliverable physical units Outages, congestion, calendar effects
Shock Policy, input cost, weather, eligibility Anticipation, simultaneous confounding
Welfare Willingness to pay, marginal cost, transfers Income effects, selection, externalities

A current-dollar per-unit price cannot be compared across decades without a price-index convention. Exchange-rate conversions compare monetary amounts under one convention; purchasing-power adjustments answer a different question. Elasticities are dimensionless but may still differ because of composition, aggregation, adjustment horizons, and institutional definitions. A national measure of housing stock cannot identify the volume of units currently offered to rent. The researcher should publish inclusion rules, missingness, filters, sampling weights where relevant, and the price and quantity definitions before comparing results.

Simultaneity

Write a stylized linear system qD=a-bp+u and qS=c+dp+v, b,d>0. Market clearing gives p=(a-c+u-v)/(b+d). A regression of q on p mixes supply and demand shocks. If u varies while v is fixed, observations trace the supply curve; if v varies while u is fixed, they trace the demand curve. Both shocks varying produces a covariance determined by their relative variances and any dependence between them. In the uncorrelated zero-mean case,

Cov⁡(p,q)=dVar⁡(u)−bVar⁡(v)(b+d)2.\operatorname{Cov}(p,q)={d\operatorname{Var}(u)-b\operatorname{Var}(v)\over(b+d)^2}.

The variance of p is [Var(u)+Var(v)]/(b+d)^2. The population slope can range from -b to d. A positive empirical price-quantity slope is therefore compatible with downward-sloping demand. Price endogeneity is a structural consequence of the clearing system.

An excluded supply shifter can identify a linear demand slope if it changes price, is independent of the demand disturbance, and has no direct effect on demand conditional on controls. The first-stage instrument must be relevant; the exclusion restriction must be justified institutionally. Angrist, Graddy, and Imbens (2000) use conditions at sea as instruments in a Fulton fish-market application and explain why heterogeneous nonlinear responses alter the interpretation of an instrumental-variable coefficient. A weighted average derivative is not automatically the slope of every buyer's demand curve. Weather may also affect demand or fish quality in other markets; the instrument must be reevaluated in each setting.

Policy evaluation and uncertainty

A policy change can serve as an exogenous shifter only under a defensible assignment mechanism. Difference-in-differences comparisons require an appropriate counterfactual trend, attention to event timing, no uncontrolled anticipation, and a specification of spillovers. A new rule may change participation; comparing only surviving participants conditions on a post-treatment outcome. Event-study estimates can help diagnose pretrends, but statistical insignificance does not prove the identifying assumption. A structural model instead estimates or calibrates primitives and solves counterfactual equilibria. Its credibility depends on the information identifying those primitives, assumptions about conduct, and validation against outcomes beyond the fitted moment.

Uncertainty in derivatives can be magnified near a singular equilibrium. In the scalar formula dp/dtheta= -Z_theta/Z_p, the uncertainty of the ratio can be large when the denominator is near zero. A scientific report should distinguish sampling uncertainty in estimated slopes, specification uncertainty from market closure, and scenario uncertainty from future shocks. Reporting a confidence interval around one conditional estimate is not a substitute for comparing plausible allocation mechanisms.

International evidence and institutional comparison

Fuel taxation in the United States

Davis and Kilian (2011) distinguish the response to gasoline tax changes from ordinary observed price movements and address price endogeneity through several empirical approaches, favoring historical policy variation. Their preferred estimate associates a ten-cent-per-gallon tax increase with an approximately 1.5% reduction in carbon emissions from US vehicles. The estimate concerns a policy-specific vehicle-emissions response, not an invariant gasoline-price elasticity. Neither total national emissions nor gasoline consumption itself is the denominator of the quoted percentage.

The lesson for the clearing equation is that a change in tax can be more persistent or more visible than a transitory price movement. Expectations about future travel costs, fleet replacement, and location choices can change the response. Estimating a single price coefficient from monthly price and consumption data and multiplying it by a permanent tax increase can miss those mechanisms. Distinguish a tax-induced gross price change from a generalized price shock and preserve the time unit of the outcome variable.

Housing in San Francisco

Diamond, McQuade, and Qian (2019) exploit assignment under the 1994 rent-control expansion. The March 2019 author manuscript reports that protected renters' mobility fell by 20% and that treated landlords reduced rental supply by 15%, including responses through conversion or redevelopment. The two estimates refer to different populations and outcomes. They do not describe a nationwide fall in total residential construction. The institutional treatment affects the status of an existing rental contract and the landlord's ability to reposition the property.

A static shortage calculation can describe excess intended demand at a controlled price. The housing case adds enduring eligibility, tenure protection, contract-specific benefits, supply conversion and the distinction between incumbent and prospective tenants. The welfare calculation must value incumbent security alongside the effect on access, resource use, and future rental-market supply. A finding about one regulatory design should not be transported mechanically to jurisdictions with different exemptions, enforcement, new-construction treatment or initial rental supply.

Labor-market boundaries

Dube, Lester, and Reich (2010) compare contiguous counties on opposite sides of US state borders between 1990 and 2006 and report no adverse employment effect in the sectors studied. Their identification uses geographically close comparisons to address spatial differences in trends. The result pertains to the design, period and affected sectors. It does not establish that all wage floors at all levels have no negative employment effects. A competitive upward floor, monopsony, search frictions, firm adjustment through hours, and geographic spillovers generate different predictions about the measured outcomes.

The key methodological issue is the definition of a labor market. Employees can commute across legal boundaries, businesses can relocate, and hours may adjust separately from headcount. The price variable can represent a legal minimum, an actual average wage, or the marginal compensation of an additional unit of labor. Measuring one object cannot automatically establish a comparative-static derivative for another.

Fisheries in Kerala

Jensen (2007) studies the introduction of mobile phone service between 1997 and 2001 in Kerala's fisheries sector. The reported evidence links adoption to a sharp reduction in price dispersion and unsold fish waste. The study illustrates the importance of information and coordination across landing sites. A change in information access can increase trade efficiency without changing the usual biological production capacity at the moment of trade. A single aggregate equilibrium price misses the spatial dispersion and the value of choosing a better destination.

The appropriate closure includes time-sensitive transport, alternative buyers, and the perishability of inventory. Spot prices in separate locations need not coincide if delivery costs or information constraints separate those locations. After communication improves, existing supply can be allocated more effectively even with limited changes in effort. The observed outcome therefore requires a mechanism of coordination in addition to a market-clearing condition.

Spanish electricity auctions and reliability

The final publication record for Fabra and Reguant (2014) reports near-complete transmission of emissions costs to Spanish electricity prices, supported by reduced-form and structural bidding analyses. High-frequency auctions and relatively unresponsive demand are important in their reported interpretation. This evidence concerns a marginal cost shock in a particular auction setting. An emissions permit price per ton and an emissions-cost change per megawatt-hour are different units; their relation varies by the technology setting the price.

A power market is physically constrained. Borenstein, Bushnell, and Mansur (2023) discuss reliability institutions and externalities of electricity shortfalls. Generation and consumption must be balanced continuously at the system level, with losses and network conditions accounted for. A posted wholesale clearing price does not by itself prove that reserve requirements are adequate or that private bidding internalizes outage externalities. Capacity payments, operating reserves, demand response, network constraints and retail risk contracts expand the market specification. A price discontinuity during a scarcity interval may represent an operating regime change rather than smooth movement along a conventional curve.

German energy substitution

Bachmann et al. (2024) compare macroeconomic approaches to the consequences of stopping Russian energy imports into Germany. Their revised analysis gives a short-run scenario range of 0.5% to 3% of GDP relative to a counterfactual without the import interruption. This is a modeled loss range, not an observed causal difference between two realized economies. Substitution possibilities and reallocation across sectors drive the sensitivity. The case identifies an economically consequential limit to single-market closure: even a correctly modeled energy price and import volume do not determine the economy-wide output effect when energy is an intermediate input in interconnected production.

Across these studies, the appropriate comparison is between mechanisms, not between raw numerical coefficients. Fuel taxation measures a policy-specific behavioral response; housing regulation changes protected status and supply; wage rules affect labor allocation and employer incentives; information changes matching across markets; emissions costs change auction bids; energy interruption induces substitutions through production chains. They are distinct conditional experiments. Their numerical results cannot be combined into one pooled elasticity.

Numerical experiments and robustness

All quantities and monetary amounts in this section are synthetic, measured in units per period and currency units per physical unit. No empirical calibration is asserted.

Linear equilibrium and taxes

Take D=100-2p_b and S=10+p_s over the region where both quantities are positive. With no wedge the solution is p=30, q=40. A tax of six implies p_b=p_s+6. Solving gives p_s=26, p_b=32, q=36, obtained from 100-2(p_s+6)=10+p_s, or 78=3p_s. Consumers bear two currency units per unit and sellers bear four relative to baseline. Tax revenue is 216.

The inverse curves are P_D(q)=50-q/2 and P_S(q)=q-10 for q>=10. To avoid assuming negative marginal cost for lower quantities, set marginal cost zero for q from zero to ten; this is consistent with S(p)=10+p at positive p. At q=40 the consumer surplus is (50-30)40/2=400. Producer surplus is 3040-(40-10)^2/2=750. At q=36 the buyer surplus is (50-32)36/2=324 and seller surplus is 2636-(36-10)^2/2=598. Tax revenue 216 produces total surplus 1138, compared to 1150 before the tax. The loss of 12 equals one half of the six-unit wedge times four units of lost trade. These identities validate the local linear formula for this finite change.

Demand and supply shocks

Scenario Gross price Net price Quantity
Baseline 30 30 40
Demand intercept +12 34 34 44
Supply intercept +12 26 26 48
Both intercepts +12 30 30 52
Unit tax 6 32 26 36

The table uses exact linear equilibria. The simultaneous equal shifts leave price unchanged while raising volume. A researcher observing price alone would miss the combined change. For a fixed initial price and volume, the same tax can produce different incidence when slopes differ. With A=1 and B=3 the buyer share is 0.75; with A=3 and B=1 it is 0.25. With both slopes summing to four, the same demand intercept shock produces the same price change while traded quantities have different responses.

At a ceiling of 24 under baseline schedules, desired demand is 52 and supply is 34, leaving excess demand of 18. Assuming all 34 supplied units trade, the physical shortfall relative to desired purchases at that price is 18, while traded volume falls by only six relative to the unconstrained benchmark. These two measures answer different questions. At a capacity of 35 with no price cap, price rises to 32.5 and the unconstrained supply at that price would be 42.5. The marginal cost at 35 is 25, yielding a scarcity rent of 7.5 per marginal unit. The shadow price is consistent with the capacity constraint and the inverse demand.

Stability and feedback sensitivities

At the linear root, Z_p=-3. A discrete iteration with k=0.2 has multiplier 0.4, so small deviations decay without sign changes. With k=0.5 the multiplier is -0.5, so deviations alternate and converge. With k=0.8 the multiplier is -1.4 and the fixed point is locally unstable. These values alter adjustment but leave the same equilibrium prices and volume. They show why a static equilibrium table is insufficient to establish stability.

In the linked-market example, an exogenous unit increase in theta changes focal price by 2/3 with the feedback market solved and by 1/2 when its price is held fixed. The full system changes the surrounding price by 1/3. The fixed-closure estimate understates the full-system marginal response by 1/6 price unit, or 25% of the full response. This difference demonstrates the scale of closure sensitivity in a simple solvable system; actual cross-market derivatives require evidence.

Analytical synthesis and research boundaries

A clearing condition is scientifically useful when its objects are aligned across theory, institutions and measurement. The independent contribution of this dossier is the integration of these three levels into a single analysis of conditional market outcomes and the construction of counterexamples that isolate their possible divergence. This is an analytical synthesis of established economic tools, not a claim of a newly discovered theorem or a new theory of markets.

First, the existence of a clearing price is a mathematical question about excess demand and the feasible price domain. It precedes the calculation of comparative statics. At a regular root, the implicit derivative gives a conditional local response. At a kink or singular root, order arguments, one-sided derivatives, or branches are required. If a policy binds a price or quantity directly, the relevant equilibrium conditions change rather than merely acquiring new parameter values.

Second, the adjustment mechanism determines whether the computed equilibrium is attainable and which branch is selected. Discrete adjustment can destabilize the same crossing that is stable under continuous adjustment. Inventory and contractual delay can produce a sequence of market-clearing trades while the system fails to converge. A static price prediction therefore has no intrinsic timing claim.

Third, the transaction quantity differs from the potential quantities on either side of the market when an allocation constraint binds. The excess-demand measure records the incompatibility of desired quantities at a price. Welfare depends on who receives the goods, who incurs costs, whether externalities are present, and which transfers enter the objective. A reduction in observed price dispersion may be associated with better allocation, but equal prices alone do not guarantee efficient allocation when quality, rationing or external harms differ.

Fourth, structural derivatives and estimated causal effects are distinct objects. A supply shifter generates variation suitable for demand identification under an exclusion restriction. A policy intervention can simultaneously change attention, composition, entry and relative prices. The resulting estimate is legitimately a policy-specific equilibrium response if the causal design supports that interpretation. Recovering an invariant demand or supply derivative requires additional identifying restrictions. An observed coefficient can be scientifically valid at its original scale and yet inadmissible for a different counterfactual.

Fifth, the value of partial equilibrium should be assessed by the economic magnitude of feedback, not by a generic preference for small or large models. When linked-market terms are small or contractually fixed, the scalar closure may have substantial explanatory value. When feedback approaches the magnitude of the own-price slope, predictions can change dramatically. When a market is subject to a system-wide reliability constraint or vital intermediate-input bottleneck, broader equilibrium and network analysis may be indispensable.

The research hypotheses receive conditional support. The formal examples establish H1: comparative-static directions require explicit restrictions on slopes, the shock, and the feasible equilibrium. The reported international studies provide mechanism-specific support for H2: institutions and economic horizons affect observed responses. The measurement and identification analysis establishes H3 as a logical comparability requirement: distinct price and quantity definitions or different policy estimands need not yield comparable coefficients. These are heterogeneous forms of support; the case studies are not a statistical test of a common population hypothesis.

Further empirical work should use registered market definitions and shock labels, compare alternative estimates of own and cross-market slopes, and match institutional records of rationing or capacity to completed transactions. The empirical review in this manuscript is selective: full-text access and primary-data replication are incomplete, and source-specific access is recorded in the evidence ledger. No current administrative dataset was downloaded, no new regression was estimated, and no universal historical time series was constructed. These limits constrain the claim to a theoretical and comparative research synthesis with independently derived illustrative calculations.

Conclusion

Partial equilibrium is a disciplined method of conditional reasoning. Its central operation is to identify the market-clearing relation and examine how the solution varies as the determinants of supply, demand, or institutional constraints change. Its mathematical strength arises from explicit assumptions on the equilibrium's existence, regularity and selection. Its empirical strength arises from a market definition that aligns transactions, prices, quantities and causal variation with those assumptions.

The ordinary competitive model yields clear signs under downward-sloping demand, upward-sloping supply and a specified single shock. Taxes, capacity limits, market power, salience, information, and linked markets alter the relevant conditions. Welfare requires a further accounting of allocation, costs, transfers, external effects, and distribution. The analytical value of the method lies in stating these dependencies explicitly and matching the scope of the conclusion to the evidence that supports it.

Market clearing therefore serves as the starting point for a scientific investigation of economic coordination. The completed investigation establishes the conditions that make the comparative-static result meaningful, the institutions that implement the allocation, and the evidence needed to identify the response.

References

  • Marshall, Alfred. 1920. Principles of Economics. 8th ed. London: Macmillan, book V, chap. III, ‘Equilibrium of Normal Demand and Supply.’ Source.

  • Arrow, Kenneth J., and Gerard Debreu. 1954. ‘Existence of an Equilibrium for a Competitive Economy.’ Econometrica 22 (3): 265–290. Source.

  • Milgrom, Paul, and Chris Shannon. 1994. ‘Monotone Comparative Statics.’ Econometrica 62 (1): 157–180. Source.

  • Angrist, Joshua D., Kathryn Graddy, and Guido W. Imbens. 2000. ‘The Interpretation of Instrumental Variables Estimators in Simultaneous Equations Models with an Application to the Demand for Fish.’ Review of Economic Studies 67 (3): 499–527. DOI: 10.1111/1467-937X.00141. Source.

  • Weyl, E. Glen, and Michal Fabinger. 2013. ‘Pass-Through as an Economic Tool: Principles of Incidence under Imperfect Competition.’ Journal of Political Economy 121 (3): 528–583. DOI: 10.1086/670401. Source.

  • Chetty, Raj, Adam Looney, and Kory Kroft. 2009. ‘Salience and Taxation: Theory and Evidence.’ American Economic Review 99 (4): 1145–1177. DOI: 10.1257/aer.99.4.1145. Source.

  • Davis, Lucas W., and Lutz Kilian. 2011. ‘Estimating the Effect of a Gasoline Tax on Carbon Emissions.’ Journal of Applied Econometrics 26 (7): 1187–1214. DOI: 10.1002/jae.1156. Source.

  • Diamond, Rebecca, Tim McQuade, and Franklin Qian. 2019. ‘The Effects of Rent Control Expansion on Tenants, Landlords, and Inequality: Evidence from San Francisco.’ American Economic Review 109 (9): 3365–3394. DOI: 10.1257/aer.20181289. Source.

  • Dube, Arindrajit, T. William Lester, and Michael Reich. 2010. ‘Minimum Wage Effects Across State Borders: Estimates Using Contiguous Counties.’ Review of Economics and Statistics 92 (4): 945–964. DOI: 10.1162/REST_a_00039. Source.

  • Jensen, Robert. 2007. ‘The Digital Provide: Information (Technology), Market Performance, and Welfare in the South Indian Fisheries Sector.’ Quarterly Journal of Economics 122 (3): 879–924. DOI: 10.1162/qjec.122.3.879. Source.

  • Fabra, Natalia, and Mar Reguant. 2014. ‘Pass-Through of Emissions Costs in Electricity Markets.’ American Economic Review 104 (9): 2872–2899. DOI: 10.1257/aer.104.9.2872. Source.

  • Einav, Liran, Amy Finkelstein, and Mark R. Cullen. 2010. ‘Estimating Welfare in Insurance Markets Using Variation in Prices.’ Quarterly Journal of Economics 125 (3): 877–921. DOI: 10.1162/qjec.2010.125.3.877. Source.

  • Borenstein, Severin, James Bushnell, and Erin Mansur. 2023. ‘The Economics of Electricity Reliability.’ Journal of Economic Perspectives 37 (4): 181–206. DOI: 10.1257/jep.37.4.181. Source.

  • Bachmann, Rüdiger, David Baqaee, Christian Bayer, Moritz Kuhn, Andreas Löschel, Benjamin Moll, Andreas Peichl, Karen Pittel, and Moritz Schularick. 2024. ‘What if? The Macroeconomic and Distributional Effects for Germany of a Stop of Energy Imports from Russia.’ Economica 91: 1157–1200. DOI: 10.1111/ecca.12546. Source.