Investigating the Price Formation of the Californian Wholesale Electricity Market

Abstract

In this paper, the California Independent System Operator's (CAISO's) day-ahead market (DAM) wholesale price and fundamental variables are considered in an attempt to model the DAM price distribution for CAISO zones NP15 and SP15, given the substantial increase in intermittent renewables. We develop fundamental quantile regression models for the DAM wholesale electricity price in each trading period for data from 08.01.2013 to 24.09.2016. The fundamentals include gas prices, GHG allowance prices, load forecast and the DAM price lagged by one and seven days. In addition, we extend previous research by including forecasts of renewables as fundamentals and investigate their impact on the Californian DAM price formation. We find that the price sensitivity to different fundamental factors are dependent on the trading period, as associated coefficients vary significantly throughout the day. The DAM price follows the same trends as the residual load curve, with low prices mainly occurring during sunshine hours. Extremely high price occurrences are clustered in summer months with high cooling requirements and thus high load. Lagged prices tend to be significant and seem to reflect marginal costs in hours that are off-peak. Gas prices and load forecasts have a significant effect on DAM prices and are mainly positive. We find that GHG allowance prices and DAM price volatility coefficients for the most part have an insignificant effect on the DAM price. For the majority of DAM price quantiles, solar and wind forecasts are significant and have a dampening effect on the DAM price. When considering quantile regression across 2015 to 2016, we find that natural gas prices are no longer significant across trading periods. Further, in SP15, forecasted solar production has a larger negative impact on the DAM price when compared to values for 2013 to 2016. When modeling summer and winter months separately in SP15 for trading period 12, the DAM price shows larger sensitivity to wind forecasts in winter months and solar forecasts in summer months. In NP15, the renewable production tends to have an insignificant effect on price. Through the use of scenario analysis using our quantile regression models, we show how altering the values of fundamental variables influences the DAM price distribution.