When my colleagues and I developed the theory of real time electricity prices back in the dark ages (1982), we were amused to see that our equations allowed for the optimal price to be negative. Power companies would pay consumers to use more electricity! At the time, we thought it was a paradoxical case that was unlikely in practice, except possibly in the middle of the night in systems with lots of nuclear units.

Fast forward 30 years, and negative prices are a regular occurrence in real systems, including in Texas and California. And now they are even happing in the middle of the day. But there is still a puzzle: why don’t generators stop generating the moment the price goes negative?

Several blog posts from Berkeley’s great Energy Institute, and my response to one of them, show that real power systems can have a lot of unanticipated phenomena. Take together, these probably explain these apparently strange behaviors.

Source: Is Solar Really the Reason for Negative Electricity Prices? – Energy Institute Blog.   and from Catherine Wolfram, Is the Duck Sinking?

Here is my solution to the puzzle proposed by Lucas David, which is “Why do hydro generators seem to keep generating when prices are negative? Why don’t they just spill water over the dam, without going through the generators?”

Part of the explanation for negative prices seems straightforward: subsidies for wind generation.
The answer to the spillway question may also be straightforward: dam engineering.

Disposing of water is not free! More precisely, building a spillway is expensive, so dams are built with finite spillways. Look at discussions of dam spillway design, and it’s clear there are tradeoffs.

Many dams were built with seemingly adequate spillway capacity when they were originally constructed. In some cases, more recent data and improved hydrologic models dictate the need for increased spillway capacity. [In other words, spillway capacity is a constrained resource.]

When a dam overflows, it can catastrophically destroy the entire dam. Spillways provide a deliberate and safe route for excess water, but if the flow in a river is too large, the spillway itself can fail. This suggests that when a reservoir is close to full and inflow is high, safety requires running the generators in order to not overload the spillways. See this, for example . 
The huge Oroville dam almost failed in this way a few months ago, so it is not just a hypothetical problem.

What about wind – why do wind generators keep operating when prices are negative? Wind subsidies were discussed extensively for Texas when it had negative prices in 2015. The TL;DR from what I read was that federal and state subsidies for wind were the main culprit:

” But wind operators have another advantage over generators that use coal or natural gas: a federal production tax credit of 2.3 cents per kilowatt-hour that applies to every kilowatt of power produced. And that means that even if wind operators give the power away or offer the system money to take it, they still receive a tax credit equal to $23 per megawatt-hour.

It appears that wind generation still receives a subsidy of about $.02 /kwh or $20/Mwh. See https://energy.gov/savings/renewable-electricity-production-tax-credit-ptc. So prices will have to go more negative $20/MWh than this to lead to curtailed wind generation.

Are these two factors enough to explain the negative prices? Not clear. But whenever a dam’s spillways are are operating at full capacity, its operators must run the generators, even when prices are negative. Perhaps that plus the wind subsidy together explain the level and extent of negative prices this year. A term paper project for an MS student at UCSD or Berkeley?