The effect of virtual bidding on forward premiums in the New York wholesale energy market

In many parts of the United States, the power industry has been deregulated and replaced with regional wholesale energy markets, where utilities purchase electricity from generators at competitive market rates for subsequent distribution to customers. Numerous studies have shown that in each of these markets, the price of energy purchased in the Day Ahead (futures) market exceeds the price in the Real Time (spot) market on average. The existence of this "forward premium" is evidence of market inefficiency and may indicate participants' aversion to risk in the Real Time market or the exercise of market power by generators. To address this inefficiency, the New York Independent System Operator introduced a virtual bidding system within its wholesale market, which permitted participants to engage in purely financial transactions and hedge their exposure to risk. The new policy was expected to promote price convergence by allowing bidders to arbitrage expected differences between Day Ahead and Real Time prices.;This study examines whether the presence of virtual bidding was associated with a change in the mean value and magnitude of forward premiums in the NYISO energy market. The study applies a GARCH model to hourly pricing data from 2001 to 2009, controlling for temperature and economic activity. The results indicate that prior to 2005, virtual bidding was associated with significantly lower and less volatile forward premiums in New York's five most congested zones but with increased premiums in the remaining less congested zones. However, when the entire period from 2001 to 2009 is examined, the results suggest that prices have become significantly more divergent in the presence of virtual bidding. Closer examination of the data reveals a dramatic increase in forward premium volatility across all zones beginning in 2005 that is not accounted for by temperature or economic activity and may have biased the results. This study attempts to account for this unexplained shift in price behavior by limiting the analysis to off-peak hours and by substituting an adjusted measure of forward premium, but neither approach yields results consistent with the pre-2005 effects. Further study is needed to identify the potential causes of the sudden increase in forward premium volatility and isolate the effect of virtual bidding on NYISO price convergence.