A Mexican industrial site that buys power through the Mercado Eléctrico Mayorista pays for two products, not one. The first is energy, priced hourly at the Precio Marginal Local of its node. The second is potencia, settled once a year through the Mercado para el Balance de Potencia and priced in pesos per MW-year. For production year 2018, CENACE settled the closing capacity price in the Sistema Interconectado Nacional at MXN 4,114,245.28 per MW-year (CENACE, Informe Ejecutivo MBP 2019, año de producción 2018, 28 February 2019).
Finance teams model the first product carefully and often ignore the second. That is a mistake, because the rule that sets a facility's capacity charge rewards a different behavior from the rule that sets its energy charge. Energy cost follows total consumption. Capacity cost follows average demand across just 100 hours of the year.
This piece explains how each price is formed, what drives the differences between regions and seasons, and where a plant can act on the result.
How the wholesale energy price is actually set
Generators submit offers to CENACE stating how much energy they will supply and at what price. CENACE dispatches in economic merit order and the clearing price is the offer price of the last unit dispatched to meet demand, the marginal unit. It is not the lowest price that fills demand, and the distinction matters commercially. A plant's cost is driven by the offer of the most expensive unit that had to run, which in most of Mexico is a gas-fired unit. Movements in delivered natural gas cost therefore pass through to the energy price with very little lag.
The resulting price is nodal. The Precio Marginal Local at a NodoP decomposes into exactly three components: the componente de energía, the componente de pérdidas and the componente de congestión (CENACE, Diccionario de Datos Abiertos, referencing the DOF note of 4 July 2016). The energy component is common to the system reference. The losses and congestion components are what make the same hour cost different amounts at different nodes. Treating the PML as a single national price hides the two components a site can do nothing about and the one it can hedge.
Why regional and seasonal spreads persist
Prices are persistently lower in the north of the country, where generation sits close to direct connections with United States natural gas pipelines. Prices in the Yucatán Peninsula run higher. On our read, that gap is mostly the congestion component doing its job: the Peninsular region has limited transmission capacity to the rest of the system and constrained gas supply, so local units set the price more often and the losses component is larger.
Seasonally, prices rise through the summer as air conditioning load builds, and temporary spikes to several times the average level are more common in those months. Cyclical fuel cost is the other driver. Because the marginal unit is usually gas-fired, a period of higher delivered gas cost raises the energy component across the whole system, independent of any change in local demand.
The capacity market a CFO should be modeling
The Mercado para el Balance de Potencia operates under Base 11 of the Bases del Mercado Eléctrico (DOF, 8 September 2015) and the Manual del Mercado para el Balance de Potencia (DOF, 22 September 2016). It is an annual, ex post market. CENACE runs it after the production year closes, to settle the difference between the capacity each load-serving entity was obliged to hold and the capacity it actually had under contract.
The obligation is calculated from a narrow slice of the year. Under the Manual, CENACE identifies the capacity each Entidad Responsable de Carga drew from the Sistema Eléctrico Nacional during the Horas Críticas of the production year, and the average of those figures becomes that entity's Capacidad Demandada in MW. Applying the minimum planning reserve to that average produces the Requisito Anual de Potencia. CRE sets the requirement parameters. For MBP 2019, CENACE identified the 100 critical hours as the 100 hours with the lowest generation reserve level in the period, a change from the earlier practice of using the highest demand hours.
The settled prices give the charge its scale. For production year 2018, CENACE recorded closing capacity prices of MXN 4,114,245.28 per MW-year in the SIN, MXN 3,191,968.00 per MW-year in BCA and MXN 5,594,248.40 per MW-year in BCS, against net capacity obligations of 6,447.06 MW-year, 473.55 MW-year and 127.28 MW-year respectively (CENACE, Informe Ejecutivo MBP 2019, 28 February 2019). Baja California Sur is the outlier in both price and scarcity, which is consistent with an isolated system carrying its own reserve.
An illustrative calculation, using the SIN closing price for production year 2018 and no reserve uplift: a plant whose average demand across the 100 critical hours is 5 MW carries a capacity requirement of roughly 5 MW-year, which at MXN 4,114,245.28 per MW-year is close to MXN 20.6 million for the year before the planning reserve is applied. This is illustrative only. The capacity price is settled fresh each production year, the reserve uplift raises the requirement above the measured average, and a supplier's contract determines how much of the charge is passed through as a separate line and how much is folded into the energy rate.
What this changes about load management
Two facilities with identical annual consumption can carry materially different capacity charges if one of them happens to be running flat out during the system's tightest hours and the other is not. That is the operational point. Capacity cost is a demand problem, not a consumption problem, and the levers are different: shifting a shift pattern, staging start-ups, running behind-the-meter generation or storage during tight system conditions.
The complication is that the 100 critical hours are identified after the fact. A plant cannot schedule against a list it will not see until CENACE publishes the executive report the following February. What it can do is manage against the observable proxies, which are periods of low system reserve and high summer demand, and instrument the site well enough to know its own demand in those hours. A facility without interval demand data cannot participate in this conversation at all, because it has no measurement of the quantity that sets the charge.
What solar does to each price, and what it does not
Photovoltaic output tracks irradiance. Module output falls as cell temperature rises above the 25 degrees Celsius standard test condition reference, so the hottest hours are not the highest output hours, and output ends at sunset regardless of what demand does afterwards. The common claim that solar output rises with temperature and therefore offsets air-conditioning demand has the mechanism backwards.
The market consequence follows from the corrected mechanism. Solar capacity suppresses the energy component of the PML across daylight hours, which is real and valuable to a daytime industrial load. Its effect on the hours that set capacity cost is weaker, because system reserve conditions frequently tighten in the late afternoon and evening as output declines. Wind and solar also reduce exposure to imported fuel cost, though they do not eliminate imports from the system: the marginal unit setting the price in most hours still burns gas, and the equipment itself is largely imported.
For a plant evaluating onsite solar or a renewable supply contract, that split is the useful frame. Model the energy saving on a daylight-hours basis. Do not assume the same asset reduces the potencia line.
Where the near-term risk sits
Regulatory action is now a direct input into both prices. On 29 April 2020 CENACE issued an acuerdo suspending pre-operational tests for photovoltaic and wind plants, barring new test requests, imposing operation of transmission corridors under remedial action limits, and authorizing the startup of must-run units for voltage regulation. Delaying new intermittent capacity and dispatching must-run thermal units both push in the same direction for a buyer, toward a higher energy component and a tighter reserve position in the hours that set the capacity charge.
The coronavirus pandemic cut demand sharply, and lower demand reduces congestion and the price instability that comes with it. Our expectation is that this is the temporary half of the picture. Recovering industrial demand, combined with any relocation of manufacturing into Mexico, restores the pressure on both the energy and the potencia side, and the capacity side responds to peak demand rather than to volume.
Model both halves of your delivered power cost
Mexico Energy Partners can separate the energy and potencia components of a facility's delivered cost, test how much of the capacity charge a supplier's contract passes through, and identify where load in high-risk hours is driving the requirement. A review needs 12 months of CFE billing, interval demand data at 15-minute resolution where the meter records it, and the site's node or tariff class. Related reading sits in who you can contract with in Mexico's electricity market, energy procurement and energy monitoring.