| Description: | Residual supply index analysis |
See also RSI Property Reference for a detailed list of properties for this class of object.
Residual Supply Index (RSI) is a technique developed by the California ISO. The following explanatory material is adapted from "Market Based Simulation in the CAISO Transmission Evaluation Assessment Methodology (TEAM)", a presentation by Mingxia Zhang, June 14, 2004. This technique has general applicability, thus the references to California in the below are for explanation only.
Modelling strategic bidding can be approached from multiple angles. Game theoretical approaches such as Nash-Cournot game (physical withholding) or Supply Function Equilibrium (economic withholding) are difficult to implement in a complex network model. Empirical approaches can be based on backcast simulation and calibration (or regression analysis) and offer a simpler alternative.
Residual Supply Index "RSI" is an empirical approach to modelling strategic bidding. It develops a historical relationship (regression) between Price-cost Mark-up and certain system conditions. It uses the regression results to predict Bid-cost Mark-up under future system conditions and applies the bid-cost mark-ups to the supply bids and runs the model to determine dispatch and market clearing prices.
Price-Cost mark-up regression results estimate the relationship between price-cost mark-ups and system conditions using hourly data ( e.g. in California the regression study covered November 1999 to October 2000 and 2003). The price-cost mark-up is expressed as the Lerner Index. The Lerner Index at region i and hour t is:
PMU it = (P it - C it) / P it where P it = Actual price in region i and hour t C it = Estimated competitive price i region i and hour tThe system conditions are represented by several key variables ( e.g., RSI, % of Unhedged load, etc.)
A Residual Supply Index provides a good measure on the extent to which the largest supplier in the market is "pivotal" to meeting demand.
| RSI = | Total Supply - Largest Supplier's Supply |
| Total Demand |
An RSI value less than 1 indicated the largest supplier is pivotal in meeting demand. In the CAISO markets, RSI values less than 1.2 have generally been associated with market prices in excess of estimated competitive levels.
To create the regression equations do the following:
You can now enter the regression terms using the RSI properties.
Example of Regression Results
| Dependent: Lerner Index | Model 1 | Model 2 |
|---|---|---|
| Intercept | 0.14[11.08] | 0.57[14.77] |
| RSI (gross RSI specification) | -0.53[72.76] | -1.81[35.55] |
| RSI_Square (RSI*RSI) | 0.54[27.75] | |
| Pct_load_unhedged | 0.65[70.98] | 0.06[66.77] |
| Normalized Load (hourly load/annual average load) | 0.4[32.89] | |
| Dummy for Peakhour | 0.086[23.77] | 0.018[4.00] |
| Dummy for Summer Months | 0.15[48.19] | 0.1[30.83] |
| R Squared | 0.46 | 0.49 |
| Number of Observations | 31333 | 31333 |
RSI currently isn't compatible with Step Link Mode of Parallel.
We can apply regression results prospectively to predict hourly price-cost mark-ups in future years, and then use predicted price-cost mark-ups as bid-cost mark-ups.
The Bid-Cost Mark-up functionality is incorporated directly into PLEXOS. The RSI and other determinants of predicted bid-cost mark-ups can be computed internally in PLEXOS. The projected bid-cost mark-ups can be automatically incorporated into the market-price run.
Each RSI object in PLEXOS is associated with a single Region object. The Lerner Index defines the amount of mark-up applied in the Region via this formula:
Bid Cost Mark-up = Bounded Lerner Index / (1 - Bounded Lerner Index)
The Bid Cost Mark-up is applied automatically to the Generators in the Region associated with the RSI. The Competition setting controls how the Region-wide BMU is applied to individual Generator objects. By default the mark-ups vary according to the location of the Generator in the merit-order, with marginal generators receiving the full mark-up and inframarginal generators receiving less mark-up.