Working Papers

If large natural gas firms also generate electricity, is this detrimental for competition and trade? I consider eastern Australia, where both electricity and gas markets are operated as uniform-price auctions, and large firms participate across both. I estimate a supply function equilibrium model for the gas market, allowing for firms' asymmetric information about electricity. Two potential issues arise. First, when gas firms generate electricity from non-gas fuels, such as coal and renewables, they may have incentive to raise the gas costs for rival gas-fired generators, because this is passed through to higher electricity prices. I estimate that this has raised the price of gas on average, but by a modest 0.8 percent. Second, the realized gas price reveals rivals' signals about gas-fired electricity generation levels, so at higher gas prices firms shade their gas supply more. This leads to steeper gas-market supply schedules, exacerbating market power. I estimate that this adverse selection has reduced gains from trade in the gas market by 10 percent on average. Increasing the frequency with which the gas market is cleared could improve market outcomes.

We study how a major hydroelectric scheme operator values its stored water. Using wholesale electricity market bidding data, we estimate a descriptive function of the opportunity cost of generation, which follows from the operator's implicit water valuation. The function exhibits near-complete passthrough of an exchange-traded electricity futures price, and decreases with greater reservoir volumes and snow depth. It remained stable even when the value of releases to an alternate, dependent river was high, as it ceased-to-flow. These results may motivate development of futures-linked administered water prices for hydropower schemes to provide water services for non-energy uses.

We estimate pass-through of international fossil fuel prices to wholesale electricity prices in Australia, where approximately 85% of coal and 75% of natural gas is exported. Using station-level offer prices from the National Electricity Market over the period 2018-2023, we find strong pass-through. For black coal power stations, long-run pass-through of coal export prices is approximately 70%, though short-run variations remain insulated. For natural gas power stations, the relevant international benchmark shifts, depending on spare LNG export capacity. Before 2021, the Asian spot price (JKM) is passed through in full. Thereafter, it is passed through only when export terminals have spare capacity: during Australian winters. Otherwise, Brent is passed through, consistent with reported Brent-linked natural gas contract prices. Storage operators - both reservoir hydro and short-cycle storage - also pass through international fossil fuel prices, at rates comparable to thermal generators. This reflects the dependence of their opportunity costs on fossil fuels. These findings have direct policy implications. Our results suggest that a proposed gas reservation policy may not reduce electricity prices during peak winter periods, and that international commodity futures offer electricity market participants a practical tool for hedging wholesale price risk.

Solar generation reduces wholesale electricity prices during the day but can raise them in the evening, as fossil-fuel plants shut down in response to low daytime prices and face costs to restart. This mechanism has been documented in grids where natural gas dominates dispatchable generation, but our findings indicate that this is not an inevitable consequence of high solar generation. We study a coal-heavy grid with high solar penetration. While solar substantially reduces daytime prices, it has at most a small effect on evening prices. The key is how coal responds: rather than shutting down, coal plants ramp down to minimum stable operating levels and ramp back up to meet the evening peak, avoiding startup costs and limiting the impact on evening prices. Hydro, acting as long-duration storage, further buffers the system by shifting generation from daytime to nighttime hours. Our results suggest that solar-induced coal exit may occur later than conventionally assumed.