Title: Prices and Concentration: A U-shape? Theory and Evidence from Renewables Authors: Michele Fioretti, Junnan He, Jorge Tamayo Date: April 7, 2025 Focus: The report centers on understanding how firm concentration and technology diversification affect pricing and market power in electricity markets, specifically using Colombian renewable energy sector data. The paper advances both a theoretical supply function equilibrium model and empirical analyses with structural estimation to dissect these dynamics. Core Thesis: The study proposes a U-shaped relationship between electricity prices and market concentration in contexts where firms operate diversified technologies. Transferring high-cost capacity to the dominant firm can initially reduce prices by intensifying competition but beyond a threshold, increased concentration becomes anticompetitive and raises prices. This tension is resolved in the context of renewable intermittency and capacity heterogeneity.
1. Metadata and Overview
The paper tackles how technology diversification and industry concentration interplay to affect electricity market prices, with an innovative theoretical contribution based on supply function equilibrium (SFE) and a detailed empirical investigation exploiting natural variation in hydropower availability in Colombia.
Motivation: Rising costs and producer heterogeneity due to different production technologies (hydropower, thermal, etc.) create complex competitive dynamics, affecting market power and prices.
Main Innovation: Demonstration of a U-shaped relationship between market concentration (measured via capacity-based Herfindahl-Hirschman Index, HHI) and prices in technology-diversified renewable energy markets.
Key Findings:
Small reallocations of high-cost capacity to a dominant, more efficient firm can lower prices by expanding overall supply.
Large reallocations lead to capacity imbalances that raise prices, confirming a U-shaped price-concentration curve.
Estimated counterfactuals show reallocating about 30% of rival thermal capacity to the leader reduces prices by 10% but larger transfers increase them.
The authors highlight policy implications on merger control and regulation sizing caps given these nuanced effects on market power. [1](#paper-page-1) [2](#paper-page-2) [3](#paper-page-3)
2. Section-by-Section Deep Analysis
2.1 Introduction
Key argument: The allocation of production technologies across firms shapes competitive market dynamics. Diversification may attenuate traditional concentration-driven price increases by aligning marginal costs among firms.
Context: Colombian electricity market, where firms own mixed portfolios including hydropower (renewable) and thermal plants.
Research question: Does diversification of the market leader offset concentration’s anticompetitive effects?
Approach: Theoretical modeling with supply function equilibrium and empirical identification leveraging hydrological shocks and resulting capacity shifts.
Preliminary evidence: Hydropower availability shocks imply that concentration changes can increase or decrease prices, depending on whether the dominant or smaller firms face shocks. This contradicts standard monotone concentration-price relationships predicted by classical Cournot or Bertrand models without technology considerations. [2](#paper-page-2)
2.2 Theoretical Intuition and Model Setup
Firms operate multi-technology portfolios with different marginal costs. The dominant firm faces a trade-off: undercut rivals when demand is low, but may withhold supply to exercise market power when demand is high.
A "small" transfer of high-cost capacity to the leader improves diversification: it stimulates output expansion by the leader in low-demand states (using cheap technology) and capacity withholding for high-demand periods with high-cost technology.
This increased aggressive supply induces rivals to respond strategically by expanding supply to defend shares —a phenomenon of strategic complementarity—which overall reduces prices.
However, if transfers are "large," rivals lack capacity to respond, enabling the dominant firm to restrain supply and increase prices.
Thus, price responses to concentration changes are non-monotonic, U-shaped, conditioned by relative capacities and cost gaps across technologies. [3](#paper-page-3)
Notable: The U-shape vanishes if firms specialize in one technology, highlighting the critical role of diversification. [3](#paper-page-3) [4](#paper-page-4)
2.3 Empirical Setting: Colombia’s Wholesale Energy Market
Market Overview: ~170 GWh daily production, highly concentrated: 6 firms hold 50%+ units, 75% total capacity.
Technology mix: 60% hydropower capacity, 30% thermal, with small shares for run-of-river, wind, solar.
Generation Variability: Hydropower dominates total production (~75%) but varies with climate (drought and wet spells). Thermal compensates when hydro is scarce.
Institution: Spot (day-ahead uniform-price auctions) and forward markets, with firms bidding quantities and prices per unit/hour.
Data richness: Ownership, geolocation, unit capacity, bids, weather data (rainfall, temperature), and commodity prices for thermal fuel inputs are available.
Price Patterns: Wholesale prices more than double during scarcity episodes, especially during El Niño and dry seasons (2015–2017). [7](#paper-page-7) [8](#paper-page-8) [9](#paper-page-9) [10](#paper-page-10)
Figure 1 and 2 clearly illustrate installed capacities and respective production shares over time, with shaded vertical bars representing hydrological shocks affecting supply and prices.
2.4 Firms’ Behavioral Responses to Water Forecasts
Empirical Setup: Firms’ bids (both price and quantity) adjust in anticipation of inflow forecasts at 1 to 5 months ahead. Regression models use dummies for adverse (low inflow) and favorable (high inflow) forecasts. Controls for market demand, water stocks, forward contracts, and fixed effects remove confounders.
Results:
Hydropower plants adjust mainly quantity bids, decreasing output by about 7.1% one month before adversity, and increasing 3.7% before favorable forecasts.
Sibling thermal units adjust price bids upwards before favorable forecasts and downwards before adverse forecasts, responding earlier (likely due to being less flexible).
No significant response to forecast errors confirms forecasts convey relevant predictive information.
Figure 3 visualizes these bid adjustments demonstrating distinct strategic channel responses by asset type.
These patterns indicate diversified firms manage hydro and thermal units complementarily across expected droughts, preserving hydropower while ramping up thermal generation as backups. [11](#paper-page-11) [12](#paper-page-12) [13](#paper-page-13)
2.5 Price Implications of Inflow-Driven Concentration Changes
The authors define a net adverse inflow indicator combining forecasts across firms weighted by squared capacity shares to estimate capacity-based HHI changes, denoted as (\Delta_{t+3}).
Regressing log market-clearing prices on (\Delta_{t+3}) reveals a U-shaped pattern:
Negative (\Delta_{t+3}) (lower concentration, large firm faces drought) leads to higher prices.
Positive (\Delta_{t+3}) (higher concentration, large firm benefits from water abundance) also increases prices.
Control variables include lagged HHI, water stocks, demand, forward contracts, and fixed effects to isolate the effect.
Figure 4 illustrates this result with piecewise linear fits showing the negative slope at lower values and positive slope at higher values of (\Delta_{t+3}).
This supports the theory’s novelty vis-à-vis traditional monotonic concentration-price relationships. The effect is robust to seasonality controls and alternative specifications. [14](#paper-page-14) [15](#paper-page-15)
3. Theoretical Model with Diversified Firms (Section 4)
3.1 Setup
Firms choose supply functions, non-decreasing supply schedules mapping prices to quantity offered, committing before demand (\epsilon) is realized.
Market clearing price is lowest (p) such that aggregate supply matches realized demand (D(\epsilon)).
A Supply Function Equilibrium (SFE) exists when firms’ supply schedules are mutual best responses maximizing expected profits conditional on others’ schedules.
The model extends classical frameworks (Klemperer & Meyer 1989) allowing firms to own multiple technologies with different marginal costs and capacities, yielding stepwise cost functions.
Firms face capacity constraints and uncertain demand inverse to the SFE concept of commitment prior to demand resolution.
where (s_i(p)) is firm (i)’s market share, (\eta) price elasticity, and (S_i'(p), D_i^{R}'(p)) derivatives of supply and residual demand. The markup depends strategically on competitors’ responses due to strategic complementarities in supply slopes.
Small vs. large capacity transfers:
Small transfer: leader’s supply constraint relaxes near a critical price (p^0), prompting it to expand supply; rivals respond by increasing supply to defend market share. Market supply grows and price falls.
Large transfer: rivals face capacity constraints, reduce supply; leader exercises market power by shrinking output, prices rise.
Numerical Illustration (Figure 5): Two-firm example with low-cost and high-cost technologies; transferring some high-cost capacity from smaller to larger firm can either increase or decrease prices depending on resource abundance or scarcity.
Generalization/Proofs (Prop 2 & 3):
SFE exists uniquely for all capacity reallocations.
Price-concentration curve is U-shaped conditional on capacity-weighted efficiency.
This elegant mathematical framework provides a rigorous foundation supporting the empirical U-shaped price-concentration relationship. [16](#paper-page-16)–[22](#paper-page-22)
4. Empirical Application and Estimation (Section 5)
4.1 Data Structure and Market Specificities
Firms own multiple units of hydropower and thermal plants; hydropower capacities vary with water stocks (w_{it}) per firm and time.
Profits include spot market revenue, forward contracts, and reliability charges.
Key dynamic: water stocks evolve according to hydrological inflows minus hydro output.
Focus on firm-level water stocks (dams tend to spatially cluster per firm).
4.2 Empirical Strategy & Estimation
The first-order conditions (FOCs) for optimal quantity bids link marginal revenue to technology-specific marginal cost and dynamic water value.
The dynamic value function (V(w)) is approximated with basis functions (e.g., B-splines).
Hydrological inflows modeled by autoregressive distributed lag (ARDL) models with asymmetric error distribution (Pearson Type IV) to fit dry vs. wet season tails.
Instruments for thermal and hydro cost parameters rely on local temperature and global gas prices, exploiting exogenous shifts in input costs affecting marginal costs.
Regressions are two-stage least squares on bids data from 2010-2015 with fixed effects controlling for units, time, and technology-season interactions.
4.3 Key Estimation Results (Table 1)
Thermal marginal costs ~140,000 COP/MWh align with observed scarcity prices.
Value function parameters capture dynamic opportunity costs of water storage.
Robustness verified with different smoothing and distributional assumptions.
This rich structural approach allows simulation of firm behaviors under alternative capacity allocations directly linked to industry fundamentals. [22](#paper-page-22)–[30](#paper-page-30) [77](#paper-page-77)–[79](#paper-page-79)
5. Simulation and Counterfactual Analysis (Section 6)
5.1 Simulation Setup
Focus simulation on the largest firm, EPMG, which holds ~80% hydro capacity and largest thermal capacity.
Uses mixed-integer linear programming to solve firm's intertemporal profit maximization under capacity and stochastic inflows.
Aggregate hourly markets weekly to reduce computational load.
The model well replicates observed price patterns and volatility except during extreme events (El Niño drought spike underpredicted).
5.2 Counterfactual Reallocation Exercises
Reallocate varying fractions (\kappa%) of competitor firms' thermal capacity to EPMG to observe price effects.
Adjust competitors’ supplies proportionally to capacity constraints.
Key findings (Figures 8 and G1-4):
Reallocations up to about 30% reduce prices up to 10%, especially during drought periods (low water stocks for EPMG).
Transfers exclusively from fringe/non-hydro firms produce more pronounced price reductions than transfers from all competitors, supporting model predictions about symmetry and competition erosion.
Gains are more significant in scarcity regimes.
These simulations confirm the quantitative importance of the theory and suggest policy relevance for capacity regulations and allocation strategies to balance concentration and diversification. [31](#paper-page-31)–[35](#paper-page-35)
6. Policy and Broader Discussion (Section 6.4 and Conclusion)
Firm-level diversification in renewable and thermal technologies enables more efficient market responses to supply intermittencies reducing price spikes.
Existing capacity caps (e.g., 25%) may prevent firms from achieving such efficient diversification, potentially backfiring by increasing market power.
Regulators should incorporate technology portfolio considerations, not only aggregate size or concentration indices, in merger and competition analysis.
The framework extends beyond electricity to industries with technology heterogeneity and capacity constraints: aluminum production, automotive platforms, labor skill mixes.
Symmetric diversification across firms yields the most competitive outcomes, guiding industrial and antitrust policy design.
The paper pioneers the empirical study of supply function equilibria with asymmetric stepwise cost functions informed by detailed market bid data.
Final takeaway: technology diversification attenuates but does not eliminate market power; optimal policy balances diversification gains against concentration risks—leading to nuanced, non-monotonic outcomes. [36](#paper-page-36)–[37](#paper-page-37)
7. Detailed Analysis of Figures and Tables
Figure 1: Installed Capacity and Production Volumes
Panel (a) shows total capacity by technology (hydro largest, then thermal, run-of-river small), stable and rising over 2008-2016.
Panel (b) shows production volume varying over time with drought/wet periods shaded; hydropower output fluctuates more, consistent with natural variability, with thermal ramping in droughts.
Shows average weekly electricity prices with significant spikes during droughts (El Niño). Prices more than double during scarcity, evidence of inelastic demand and supply constraints.
Provides motivation for need to understand capacity and technology effects on pricing. [10](#paper-page-10)
Figure 3: Firms’ Bid Responses to Inflow Forecasts
Top panel: Dams respond mostly with quantity bid reductions in adverse forecasts about 7% at one-month horizon; minimal price bid responses.
Bottom panel: Sibling thermal units respond primarily with price bid increases before favorable forecasts (about +10%) and price decreases before droughts, responding earlier at firm level.
These patterns confirm the dynamic supply adjustment behavior predicted by the model. [13](#paper-page-13)
Empirical evidence of nonlinear relationship between forecast-driven concentration changes ((\Delta_{t+3})) and log prices.
Slopes negative when (\Delta<0), positive when (\Delta>0), confirming theory. [15](#paper-page-15)
Figure 5: Model Equilibrium Supply and Price Effects
Panels compare supply functions and residual demand before and after reallocating high-cost capacity to market leader under two regimes: abundance (top) and scarcity (bottom) of low-cost technology.
Under abundance, market price increases with capacity transfer; under scarcity, price decreases, illustrating U-shaped effect.
Visualizes how capacity transfers affect firm behavior and market price depending on underlying capacity/cost structure. [20](#paper-page-20)
Figure 6: Dam Locations by Firm
Maps dam locations, showing spatial clustering by firm, justifying aggregation to firm-level water stocks for analysis. [24](#paper-page-24)
Figure 7: Model Fit of Simulated vs. Actual Prices
Simulated weekly prices closely track real price volatility over 6 years except for extreme drought spike in 2016.
Validates structural model's performance and suitability for counterfactuals. [33](#paper-page-33)
Figure 8: Counterfactual Price Effects of Capacity Transfers
Heatmaps of price changes as function of (y) fraction capacity transferred and (x) water stock percentile (scarcity).
Blue zones: small transfers reduce prices, especially in drought (left bottom corner).
Red zones: large transfers increase prices, especially in scarcity and high concentration.
Consistent with mechanism of competition enhancement via small transfer and market power dominance for large transfers. [34](#paper-page-34)
Table 1: Estimated Marginal Costs and Value Function Parameters
Thermal cost ~140,000 COP/MWh, consistent with engineering estimates and scarcity pricing.
Hydro cost considerably lower.
Value function parameters characterize dynamic intertemporal opportunity cost of hydropower.
Hydrological Variability: Extreme droughts or wet spells could disrupt price patterns and firm behavior beyond model scope, as seen in less accurate 2015–16 simulation.
Modeling Assumptions: Simplifications like ignoring cross-firm hydrological correlations, limiting future water stock impacts to own firm (for tractability), and holding competitor bids fixed during counterfactuals may restrict generalizability.
Regulatory Changes: Unanticipated policy shifts or changes in capacity caps could alter firm incentives and market structure significantly.
Data Limitations: Forward contract data limited; unobserved cost components or behavioral strategic bidding nuances could bias parameter estimates.
Model Complexity: Solving large-scale SFEs with stepwise cost functions is computationally intensive, constraining full equilibrium re-computations in counterfactuals.
The authors tackle some risks via robustness checks, instrumental strategies, and structural estimation frameworks. However, large shocks or regulatory shifts remain exogenous risks outside model control. [31](#paper-page-31)[32](#paper-page-32)[34](#paper-page-34)
9. Critical Perspective and Nuances
The SFE framework elegantly captures strategic interactions with uncertainty and capacity constraints but can be sensitive to assumptions (e.g., strictly increasing supply functions, symmetry conditions) which may not always hold in real-world markets with regulatory intervention or collusion risks.
While the U-shaped price-concentration curve is intuitive in this framework, disentangling the pure effect of diversification vs. concentration in practice is demanding; endogeneity in capacity transfers (e.g., mergers) is discussed but still challenging.
The focus on dominant firms (EPMG) and partial counterfactuals fixing competitor strategies may underestimate feedback effects and second-order responses.
The aggregation of capacity and water stocks at firm level ignores within-firm heterogeneity or spatial transmission constraints potentially affecting residual demand curves.
Policy implications suggest loosening capacity caps to allow diversification but do not fully address potential risks of large firm dominance or other market power abuses.
The model’s dynamic and stochastic features are advanced, but operational realities like network constraints, ancillary services, and demand flexibility are abstracted from.
Overall, the study is a major empirical and theoretical advance but requires cautious interpretation when transferring findings to policy design. [27](#paper-page-27)[34](#paper-page-34)[36](#paper-page-36)
10. Conclusion and Synthesis
This research presents a comprehensive analysis combining novel theory and rich data to illuminate the nuanced interaction between firm market concentration and technology diversification on electricity market pricing.
Key Theoretical Contribution:
Under supply function competition with capacity constraints and heterogeneous technology portfolios, prices exhibit a U-shaped relationship with market concentration.
Small capacity transfers to a dominant, diversified firm increase competition and lower prices as they enable profitable aggressive supply expansion and trigger rival responses.
Large transfers reduce competition, constraining rivals and raising prices as dominant firms wield market power.
Symmetric capacity and technology portfolios yield the most competitive outcomes.
Empirical Validation:
Using Colombian wholesale electricity market data, the paper identifies bid responses consistent with theory, showing hydropower and thermal units behave strategically in anticipation of inflow shocks.
Reduced-form price analysis confirms the nonlinear (U-shaped) price-concentration relationship.
Structural estimation quantifies costs and dynamic opportunity costs, enabling simulation of counterfactual capacity reallocations.
Simulations confirm theory: moderate transfers (~30%) of competitors’ thermal capacity to dominant firm reduce prices by 10%, especially in scarcity, while larger reallocations reverse gains.
Policy Insights:
Current regulatory caps on firm size may inadvertently restrict beneficial diversification, raising prices inadvertently.
Antitrust evaluations should incorporate technology diversification and capacity portfolios rather than solely market share or HHI metrics.
Findings have broader implications across sectors where firms manage heterogeneous input portfolios under uncertainty.
The detailed examination of all key tables and figures demonstrates the robustness of theoretical findings to empirical realities and highlights the operational mechanisms driving these phenomena.
This work sets a new benchmark for understanding oligopoly dynamics with technology heterogeneity and provides policymakers with critical new tools to assess and regulate concentrated markets amid ongoing energy transitions. [1](#paper-page-1)–[37](#paper-page-37)