Deregulation in the US actually got its start way back in 1992 with a directive to FERC, the Federal Energy Regulatory Commission.  The Federal government began pressuring the states to do away with generation monopolies.  You can get a good idea of the background that drove this whole process by reading a three part article by Dick Mills.  I had links to the three parts and the whole thing, but they have broken; they were at, and seems to have gone the way of the Dot Bombs.

Instead of that, here is my own `capsule summary' (not nearly as good as Dr Mills' articles):

You might think, `Great!  Newer plants will be way more efficient!  Instead of creeping up a quarter to half percent per year, the new plants leap 5, 10, even 15 percent over the old ones!'  But if you think that, you are not thinking like an accountant.  To an accountant, this is the same as saying: `Those plants you built ten years ago, that you expect to run for another 30 years?  They're junk!  They should be torn down now.'  This is a disaster.  This is like buying a new delivery truck and having it half-dissolve in the rain after you have made the first payment, with the rest of the payments still to be made.  You must still make the deliveries, and you now have a shambling rust heap instead of a shiny new reliable truck.

Who was going to pay for the massive upheaval required to replace all those old, inefficient plants?  Enter deregulation.  Get rid of the monopolies, kill off new investment in `gold plated' reliability, scrap the old plants or let private companies buy them for pennies on the dollar, try to recover the `stranded' costs elsewhere, etc.

Before Deregulation

The situation before deregulation was pretty much as Dick Mills described.  In California, there were three big utilities: Pacific Gas and Electric (PG&E), Southern California Edison (SCE), and San Diego Gas and Electric (SDG&E).  Each utility worked with the California Public Utilities Commission (CPUC) and the California Energy Commission (CEC) to figure out what plants would be needed, where they should be built, and how rates should be set to pay for them.

Power plants are expensive.  Large plants cost many millions, or -- in the case of the large nuclear plants -- billions of dollars.  The reasons are simple enough.  The plants are physically very large, so they need a lot of real estate.  They contain a lot of expensive parts (turbines and plumbing and wiring and controls and transformers and generators ad nauseam) and they take a long time to build.  Nuclear plants in particular take very long to build, because they are held to much higher safety standards than, e.g., gas-fired plants.  If there is a gas leak and a big explosion and a bunch of people die, it is considered very tragic, but nobody really takes that much notice; but if there is a nuclear accident with even slightly measurable radiation levels, it makes world headline news.  I make no arguments as to whether this is or is not a good thing; I merely note that it happens.

In any case, the point remains that plants are expensive to build.  In general -- and this was also true for the CPUC -- most states only allowed a utility to begin to recover the cost of building a plant once the plant came on line.  In other words, if a plant was supposed to cost $500 million, the utility could not bank the $500 million first, then build the plant; instead, they had to borrow the $500 million first, build the plant, and only then start paying the loan.  Plants were expected to last 30 years or more, and were financed accordingly (like buying a house on a 30-year mortgage).

Remember also that utilities were granted a monopoly on power generation, transmission, and distribution, but in return, they had to make a `deal with the devil': the PUC would get to control how much profit they made.  In return for investing (and collecting) all this money, they were guaranteed a return of about ten percent.  This is the situation that made utility stocks so reliable, giving them their nickname of `widow and orphan' stocks.  Thus, the utility would build a big plant, and run it for 30 or 50 years in order to pay off their loan and make their ten percent or so.  All of this was controlled by a complex web of state law.

In the 1980s, the utilities built a lot of plants.  In the early 1990s, California fell into an economic recession.  Southern California was particularly hard-hit, with the collapse of the aerospace and defense industries.  That left the state with a huge oversupply of electric generation plants, not enough people using electricity, and a high cost per kWh for the remaining users (who, after all, still had to pay off those big utility mortgages -- they had a guaranteed return).

A Few Words About Money

The utility business is big.  Really, really big.  Mind-bogglingly big.  How big?  Glad you asked.

The California electricity market is so big today that every penny that someone pays per kWh for electrical energy winds up adding up to nearly three billion dollars a year.  This money is, of course, split across many buyers and sellers, including direct contracts between generators and heavy industrial users, which bypass the utilities almost entirely.   But this still leaves a stark fact: five cents per kWh for energy represents something like ten billion dollars a year in cash flow at PG&E and SCE (SDG&E is smaller than the other two, and still solvent).  Paying a nickel (wholesale) for running a 1000-watt electric heater for one hour may not seem like much, but those nickels add up (and of course if you are a retail customer, it was more like eleven or twelve cents, of which a nickel or so went to the wholesaler).  Keep this in mind when you see numbers like `$14 billion in debt' -- 14 billion is also just 7 cents.

The Deregulation Scheme

In California, deregulation came about through Assembly Bill 1890 (AB1890).  AB1890 was written by various legislators headed by Sen. Steve Peace, with lots of input from many hands, including the three large California utilities.  AB1890 was signed into law in 1996 by Pete Wilson.  Its goal was to use the overabundance of generation in a more effective manner, so that costs could drop.

AB1890 created a number of new entities.  Chief among these were:

The CalPX and CalISO day-ahead, hour-ahead, and real-time spot markets used the so-called `market clearing pricing' (MCP) mechanism to decide which generators to run.  Under this system, the idea is that a company that owns a generator can `bid' the ability to run that generator for an hour, cranking out (say) 50 megawatt-hours (MWh) during that hour, for a given price.  The price would be whatever it actually costs to run that generator plus some small profit margin.  Suppose a very large number of generator companies make a very large number of bids, and the utilities schedule almost all of their anticipated load.  Then, well in advance, CalPX could go through all the bids and say: Hm, here's the cheapest one.  It will meet the first part of the load.  Here's the next cheapest, and it will meet the next part.  They repeat this until they have found all the cheapest generators that suffice to meet the anticipated load.  Then -- here is the crucial part -- they pay everyone who won a bid the highest price they came up with, i.e., the price of the last accepted bid.

This design was taken more or less straight out of the UK's 1988 deregulation (where it did not work very well either, but that is another matter entirely).  The MCP sounds stupid at first blush.  If Joe was willing to sell 50 MWh for $50 ($1/MWh or $0.001/kWh), why pay him more?  The answer lies in `strategic bidding' and what is called `seller's remorse'.  Suppose Joe sold his 50 MWh for $50 and actually got $50, but Bob down the street sold another 50 MWh for $500.  Joe looks over at Bob and says to himself: Gee, I should charge $10/MWh too.  By having an MCP, Joe can rest assured that even if he `bids cheap', he will get as much as Bob, if Bob gets anything at all.  In theory, this removes Joe's incentive to charge extra.  If everyone plays fair -- if everyone bids their own cost plus some profit -- then the lowest-cost generators will always run, and as more power is needed, higher-cost generators will run only as needed.  No one would ever have any reason to `bid high' because someone else would come in and undercut them.  After all, there was (in 1995) way more generation than load, so there was always plenty of spare capacity.

Where this falls apart is when someone -- anyone! -- has `market power'.  Suppose Bob is greedy and realizes that there is so little power to go around, his power will get bought no matter how high he prices it.  All he has to do is set his price to a million dollars per MWh.  CalPX will go everywhere else first, but eventually only Bob will remain, and they will buy his million-dollar megawatt-hours -- and pay everyone that same price.  Even if Bob is only slightly greedy, he can raise prices quite a bit.  (There are also strategic `gaming' tricks Bob can try; I will describe these later.)

The obvious solution is to switch to `as-bid pricing'.  But that just puts us back at the `strategic bidding' above: Joe gets $1/MWh but sees that Bob is getting more, so Joe raises his price.  In fact, Joe raises it to whatever he figures he could have gotten, if CalPX were using the MCP.  To a great extent, MCP is a red herring -- it has nothing to do with the real problem.  The real problem lies in the ability of generators to use `market power', as if they were a monopoly.  In other markets, someone has to have a huge share to be a monopoly, but because of the real-time load/generation balance problems of electricity, someone only has to control `just enough' generation to be able to unbalance the system.  One single `bad egg' using this ability will send the price up even if everyone else is a `good egg'.  Again, if there are enough generators and transmission lines, this is not a problem.  The one bad egg trying to drive up the prices will get undercut by someone, somewhere.  And in 1995, there was that huge oversupply of generation.

In economics, these problems all really boil down to good old `supply and demand'.  If the supply is large, the prices fall until demand rises or supply falls, or both.  If the demand is large and the supply is small, the price rises until the supply increases -- because people see how much money there is to be made -- or demand falls off, because no one can afford to buy the stuff.  The keys to making this work are pricing (and to some extent, demand `elasticity' -- how quickly demand responds to price changes), and the ability for anyone to jump in and sell the desired item, in this case electric power.  In 1995, things looked okay here.  There were occasional long-distance transmission limitations, but rarely anything serious.

Anyway, the above describes the `goal state', as it were: generators would sell into or through CalPX, who would find the most efficient, least-cost way to make generation meet load (and vice versa).  The utilities would pay out the wholesale cost, and would collect the same wholesale price from their retail customers, plus various fees for transmission and distribution.  To make sure that no one had market power, the utilities would have to sell off their generating plants.  They would become `utility distribution companies' (UDCs) that simply acted as a carrier.  Retail customers could choose to buy power from the UDC (i.e., through CalPX), or through some other retailer who would use the UDC to distribute the power (hence the name).  UDCs would be monopoly distributors, heavily regulated as before, but generators would compete in a free market.

Transition Period

In the past, of course, the utilities were guaranteed a return on investment.  The utilities had made various investments (i.e., built various plants) that had some particular costs.  They could sell off those plants, but if they did not get enough money for them, they would not get the return they had been guaranteed.  Most of the plants they had to sell were old -- the last major plants had been built in the 1980s, and it was already 1995 -- and newer plants were much more efficient, so the old ones were seen as inefficient and not worth much.  The costs would be `stranded' after the sale, i.e., there would be no way for the utility to make up the difference.  All of this meant there had to be a transition period during which the utilities would be `partly regulated', collecting enough money to pay off all those `stranded costs'.

So, suppose you are a utility.  Along come the Federal and state deregulators, and they tell you: `We want you to sell off your plants.  Old, new, whatever, we don't care, we just want things set up so that no generator has a monopoly anymore.'  But you still have plants with big mortgages left on them.  Not only that, you think that you cannot sell them for anywhere near enough money to pay off the loans you are still carrying.  (It turns out that you are wrong, but never mind that yet.)  So what do you do?  How do you get those things paid off?  By law, you -- and those widows and orphans who own your stock -- were guaranteed a payoff.  To suddenly rip that away would hardly be fair.  So you, and the deregulators, had to come up with a way to pay off these stranded costs.

In this case, the utilities came up with an idea.  They took a guess as to what wholesale electricity prices might be.  They figured -- correctly -- that those prices would start out around 3 or 4 cents per kWh ($30 to $40 per MWh) on average.  They had a pretty good idea what it was costing them to generate electricity, after all; they had been doing it for decades.  They agreed with the legislators to set up a fixed wholesale price: an average of 5.5 cents.  If electricity actually cost 3 cents, then, this would give them `headroom', as they called it, of 2.5 cents.  Since every penny becomes billions, this `headroom' would give them a way to pay off the stranded costs.  They could sell their tired old plants cheaply, yet still make their previously-guaranteed profit.

(I should add here that the California PUC apparently had some say over how this `headroom' money was to be allocated.  Its main purpose was to pay stranded costs, but it could get tapped for other things.)

Now, here things get a little tricky.  As the utilities sold off their plants and paid off their stranded costs, they would go through this `transition period' and collect a fixed 5.5 cents, no matter how much they had to pay out for wholesale electricity.  Once they finished the process, they would go to `fully deregulated' status, and collect the actual wholesale price.  To make sure that everything happened on schedule, they would lose their `headroom' in 2002, even if they had not paid off all their stranded costs by then -- everyone, regardless of status, would become fully deregulated in 2002, or when they were done, whichever occurred first.  (Note: the exact amount of headroom, and fixed price, was different for PG&E and SCE, and no doubt for SDG&E too.)

To get that fixed price, every bill would include something called a `competition transition charge'.  If a utility bought 500 kWh for you, and paid $17.50 (3.5 cents per kWh), they would bill you for $17.50 in `energy' and $10 or so in `CTC'.  This way you would pay about 5.5 cents per kWh, giving them their `headroom'.  (On top of that $27.50 they would add other fees to cover their distribution costs and so on -- the $27.50 covered just the wholesale electricity itself, including the headroom.  Your actual bill would be a bit over twice that.)

The Financial Mess

As it turns out, SDG&E was the first, and so far only, utility to complete the process.  In 2000, they earned their new status, and began passing wholesale prices through unchanged.  PG&E and SCE remained in the `transition' state at that time, still collecting money to pay off stranded costs, and still selling power plants.  But in 2000, something went wrong.  Electricity prices, which had been averaging well under the 5.5 cent `headroom' region, suddenly ran up -- a lot.  Suddenly the CTC numbers were negative -- PG&E and SCE had to pay more than they were collecting.  Their headroom was gone and the ceiling was pressing in.  SDG&E, on the other hand, got to pass the new, higher costs through to ratepayers, whose bills shot up amazingly.  In August, the legislature put in a wholesale price cap of 6.5 cents for residential SDG&E customers, and since then the difference has been piling up -- with a promise that SDG&E will be able to collect it later.
Supply and Demand (or, Why So Expensive?)
The total amount of generation available inside California is around 48000 MW [1] .  This puts an upper limit on the supply, at least from within the state itself.  (California can import and export fairly large amounts of power over the major interties that connect it to Arizona, Nevada, and Oregon; and those connections connect on to Washington, Utah, Colorado, Idaho, and so forth.  There are limits to this, and grid transmission limitations play a big part in today's problems too.)  What about demand?  This, it turns out, is seasonal.  Demand in winter is as low as about 18000 MW at night, and as high as about 33000 MW during the day.  Demand always bottoms out around 3 AM, and there is always a daily power curve.  As the weather warms up, demand drops at first, until the weather gets hot.  At that point the air conditioners turn on and demand goes way, way up.  Last year (2000) summer demand often reached 45000 MW, and peaked as high as 55000 MW now and then.  When demand exceeds supply on hot summer afternoons, California must buy from out of state suppliers (who are of course subject only to Federal, not state, law).  In-state generators can also sell to out-of-state buyers.  In the past, there were not that many, but the population of states outside California has been growing rapidly too, and those states now not only want to keep more of their own power, but also buy more of California's.

For reliability reasons, the ISO likes to have about a 10% `buffer zone' of extra generation ready to be used at a moment's notice (or a moment's light switch, as it were).  If these reserves drop below 7%, the ISO declares a `stage 1 emergency'.  At 5%, they go to `stage 2', and at 1.5% they go to `stage 3' and start ordering utilities to turn out the lights.  Adding the 10% reserve to the typical 30000 MW winter load (at 7 or 8 PM when all the lights are on and many businesses have not yet closed), California needs about 33000 MW available all winter.  33000 out of 48000 leaves 15000, which should be plenty.

Unfortunately, a lot of that 48000 MW is hydroelectric generation, much of which is fed by snowmelt.  In the dead of winter, the snow is not melting, so some of that 48000 disappears.  Some tiny bit is wind-power that mostly runs in summer.  Another tiny bit is solar power, e.g., from the Luz plant in the Mojave desert, which also mostly runs in summer.  No problem -- even subtracting all the summer-only plants, that still leaves plenty, even if they are mostly natural-gas fired.  In any case, the need is highest in summer when that snow is melting nicely, so that part works well too.  So -- why did the lights go out in January?

Out of State, Out of Mind
California demand in 2000 was actually just about the same as it was in 1999 -- slightly higher, especially in mid-summer and at the end of winter, but really not much higher at all.  I suspect  the difference is due to out-of-state supply and demand.  In the past, states like Arizona and Nevada had excess power to sell most of the year.  Washington, Oregon, and Idaho had excess hydroelectric power most of the summer (they have a lot of electric heat, and colder weather than California, so they actually tend to buy power from California all winter long).  But Las Vegas, Phoenix, and Tucson have all been growing rapidly.  So have Washington and Oregon.  While 2000 was not a drought year, it was not a particularly wet one either.  All of these mean that supplies outside California have been steady or shrinking, while demand outside California has been growing.  That means less left over, and hence the very high prices in summer 2000.  It also creates that market power situation.

If you own a bunch of generators, you may not know whether you have market power.  As it turns out, though, you can find out pretty easily in a CalPX or CalISO style setup.  You test for market power by using two different kinds of withholding, called physical withholding and economic withholding.  The process is fairly simple.  You set the price on one of your generators -- such as a small, inefficient peaker -- higher than you really need to.  If it is used, you have not really learned anything, although you know that this price is not yet too high, and of course all your other generators get at least this much money per MWh.  In that case, you crank the price up for the next round, and/or take one of the larger generators off line for maintenance -- physical withholding.  As soon as the small, expensive generator is not used, you have achieved economic withholding.  By using a spectrum of pricing, and taking a small number of generators off line, and repeating until you find the `sweet spot', you can discover whether you have market power, and price your generators so that you exercise it effectively.

It is very hard to prove whether someone is doing this, and in a market clearing pricing system, only one generator company has to do it to raise prices.  The amount gained this way may be just a few percent, but because of the size of the electricity market, that can mean millions of dollars.  (What this really means is that, to prevent easy exercise of market power, most electricity should be bought and sold via contracts.  Operating a residual market with an MCP system can leave pricing problems built into that system, but since it is just a residual market, it does not really matter much.)

Natural Gas Prices
Even if there has been no gouging, or only a tiny bit of gouging in a few rare cases, natural gas prices in California have been remarkably, even suspiciously, high.  Because of the market clearing pricing and the need to run gas-burning plants all winter, winter electricity prices will be correspondingly high.  When gas cost $3/MMBtu, the fuel at an old generator might cost $35/MWh (3.5 cents per kWh).  But prices were often ten times that, so electricity really did cost $350/MWh in some cases.  Because of the market clearing pricing mechanism, these apparently insane prices may well have been real.

In other words, rather than generator price gouging, perhaps the state should be looking into natural gas price gouging.  (And in fact, they are -- but only FERC have actual legal authority here.)

The Financial Mess, Part 2

AB1890 was signed in 1996, and went into effect in 1998.  When it was to have finished taking effect in 2002, the UDCs would all buy power from the CalPX pool, at whatever it cost, using the market clearing pricing mechanism to get the cheapest possible price.  They would then tack on regulated fees, and customers would pay the wholesale price plus the fees.  In the meantime, the CTC would provide headroom for paying off stranded costs.  In 2000, SDG&E finished and went to `fully deregulated', and wholesale prices shot up so much that the headroom went seriously negative.  Instead of collecting an extra cent or two, PG&E and SCE had to pay out an extra 15 cents.

The two utilities went to the California Pubic Utilities Commission and asked for their blessings in making contracts.  Traditionally, if a utility makes a contract on their own to buy electricity, a state's PUC comes along later and takes a look at whether the contract was a good idea.  If it was, they praise the utility; if the utility overpaid, they say instead: `bad dog, no biscuit!'  They can even slap the utility with a fine.  So this discourages utilities from getting contracts without getting `pre-approval' from the PUC.  PG&E in particular have said that they felt such pressure and needed explicit approval.  Today, there is finger-pointing by both sides -- the utilities say `we asked for contracts', the PUC says `we said okay', and the utilities say `you did not say okay in strong enough terms to make us feel comfortable'.

Now what?

As soon as the CTC turned negative, the utilities began to lose money.  But for two years before that, the CTC was positive -- and the utilities earned billions off that headroom.  Moreover, even after the CTC went negative, the utilities still owned many generating plants, and were still selling those plants' output through CalPX and earning the market clearing price.  (In December 2000, for fairly complicated reasons, FERC shut down CalPX, and this probably changed.)  That helps muddy the financial picture.

Remember those stranded costs?  When PG&E and SCE sold generation plants, they got far more money than they had expected.  That means lower stranded costs to begin with.  Then wholesale prices went crazy.  Remember those market clearing prices?  That means higher income from the remaining plants.  Even though the utility distribution business had gone sour, with the CTC eating up any distribution profits and then some, the utility generation business was still going great guns.  That means quicker payoffs.  In early 2001, PG&E and SCE brought two separate lawsuits to the courts, claiming that they had paid off all their stranded costs and that they should now get in on SDG&E's deal.  The PG&E case was referred to the Los Angeles court that was handling the SCE case.  This was one prong of the utilities' plan.

The other was bankruptcy.  Bankruptcy might be good for the distribution company, because it holds off the creditors, and pits a bankruptcy judge against the PUC.  Even if the judge cannot set rates, this at least buys time -- either enough to reach the `deregulation finish line' and stick the ratepayers with the tab, or enough to let a court decide that they had already reached that finish line.

One of the key elements the state extracted in the bailout deal was a promise that SCE would drop the lawsuit mentioned above.  The state was trying to do the same with PG&E; I believe this was a key sticking point in the discussions.  (Another sneaky trick, this time on the part of the CPUC, was to change the accounting methods and/or rules for the `headroom' -- the CPUC is apparently trying to make sure that those stranded costs continue, to keep the negative CTC around.)

update (3 May 2001):  A federal judge has dismissed the PG&E lawsuit for now.  The suit can be resurrected later, but for the immediate future, the utilities will have to depend on the PUC to implement rate hikes.
Wholesale costs in the last few months have remained high, and projections are that they will go much higher in the summer.  The state's Department of Water Resources (DWR) is buying energy through a combination of contracts and spot market purchases.  Only the state know exactly what is in the contracts, but even if half the energy is bought amazingly cheaply -- say, at $20/MWh -- if the other half costs $400/MWh, that produces an average price of $210/MWh, or $0.21/kWh.  More realistically, half the energy might cost $70/MWh.  If the other half costs as little as $350/MWh, the average is still $210; if the other half goes to $500/MWh, the average rises to $285.  That would put the retail price around $0.35/kWh.  Projections are that the state may spend $70 billion this year, which works out to about $0.27/kWh.

The state intends to finance the electricity cost over many years.  The `buy now, pay later' plan can hold the price increase down, at the expense of making it last longer.  This is not such an unreasonable idea -- after all, it is how a home mortgage works.  But nobody knows, yet, what the actual cost will be.

The PUC proposed a new four-tier rate structure.  Now they are talking about a five-tier rate structure.  Assembly Bill 1X, passed back in January 2001 or so, has some special provisions for residential customer rate increases: anyone using under `130% over baseline' must not have their rates increased.  That means the entire increase has to go to the so-called `heavy users'.

To pull a few numbers out of a hat, suppose that the ten-year average cost will actually be around $0.09/kWh.  Add a few billion for financing to get $.11/kWh.  The proposed 46% increase would run collections up to about $0.095/kWh.  That means the state would need another 1.5 cents per kWh -- all of which has to come out of Tiers 3 through 5, which only cover about half the electricity, so we should double that to $0.03 more.  The PUC's proposal set Tier 4 for SCE at around $0.25/kWh, so that suggests a ten-year retail price for Tier 4 of perhaps $0.28/kWh.


[1] You can find lots of different numbers, most of which are not identified as to exactly where they apply.  The CalISO web site has a figure closer to 44 GW, but they cover only their own control area, which excludes some of California.

All contents are copyright © 2001 Chris Torek.