00:00:00
Arthur van Benthem: So that would suddenly tell me
that it's five p.m. in
00:00:04
Philadelphia, and on August 21,
it's extremely hot outside. But
00:00:08
now my power is very expensive.
So maybe instead of setting my
00:00:11
thermostat to, you know, 70, I
might set it to 74, and that's a
00:00:15
big difference. The
issue is that people don't like
00:00:19
unpredictable bills. If you
don't pay attention, all of a
00:00:23
sudden, you hadn't really looked
at your thermostat settings, and
00:00:26
you get a bill the end of the
month, and it turned out you
00:00:29
consumed power during really,
really high price hours.
00:00:33
Dan Loney: Welcome to the Ripple Effect,
the podcast that takes you on a
00:00:36
journey through the minds of
Wharton faculty. I'm your host,
00:00:39
Dan Loney, and in each episode,
we'll be diving deep into the
00:00:42
inspiration behind the ground
breaking research that Wharton
00:00:46
professors have conducted, and
exploring how their findings
00:00:49
resonate with the world today.
Dan Loney: Well, another area of innovation
00:00:55
to be looked at involves that in
and around the energy sector,
00:00:58
more specifically, electricity.
Research takes a deeper dive
00:01:02
into the new ways to potentially
charge consumers for the
00:01:06
electricity they consume.
Pleasure to be joined by Arthur
00:01:09
van Benthem, who is an Associate
Professor of Business, Economics
00:01:13
and Public Policy here at the
Wharton School. He's been part
00:01:16
of research looking into this.
Arthur, great to talk to you.
00:01:19
How are you today? - I'm doing
fine. Thanks for asking. - Thank
00:01:23
you. I guess, let's start with
what was the driver to take a
00:01:26
look at this component of how
consumers deal with the costs
00:01:30
around the electricity they use.
00:01:33
- Righ, yeah. I heard so many
stories and energy podcasts
00:01:37
about how great it would be if
people charged their cars and maybe
00:01:40
put their clothes in a dryer
during hours when there's
00:01:43
lots of clean, cheap, renewable
electricity. But then I thought
00:01:47
to myself, right? I mean, I
always pay the same price
00:01:50
per kilowatt hour of
electricity. What's in it for
00:01:53
me? Like, what incentive do I
have to actually use electricity
00:01:56
at the right time?
00:01:58
- So you talk about a mispricing
component that's out there in
00:02:02
this industry, meaning what
specifically? - Yeah,
00:02:05
let me start by saying that
electricity is kind of a special
00:02:09
product. It's priced in a so
called wholesale market, where
00:02:12
generators sell their power to
large customers. You can think
00:02:16
of a utility. And the price in
the market changes every few
00:02:20
minutes, right? When demand is
very low, then electricity is
00:02:23
cheap. But all of a sudden, a
couple hours later, perhaps when
00:02:28
air conditioning demand is very
high, because the temperature
00:02:32
has risen, prices can suddenly
skyrocket, as we need expensive
00:02:36
power plants to supply that
power, but we as end consumers
00:02:40
don't see any of that action. We
just pay the same constant price
00:02:45
of electricity. And that's kind
of the source of the mispricing.
00:02:48
You always pay the same price
for your power, but in reality,
00:02:52
the power is sometimes very
cheap to generate, and sometimes
00:02:55
it's actually very, very
expensive.
00:02:57
- How much of a component are we
talking about here in terms of
00:03:01
the annual costs associated with
this pricing?
00:03:04
- Yeah, that was one of the things
that we tried to get a handle on
00:03:06
in this study. And sort of our
estimates suggest that it is
00:03:10
several billion dollars a year
across the US, probably much
00:03:15
higher. I can get to that. But
the main reason for the $2
00:03:19
billion is that, you know, the
flat pricing for end consumers,
00:03:23
it leads to inefficient energy
use. We use it at the wrong time
00:03:27
of the day, right? So there's
sort of two things that are
00:03:32
costly. One is, you know,
during peak hours and hot
00:03:34
weather, the wholesale prices
are very high for electricity,
00:03:39
but consumers don't have much of
an incentive to increase their
00:03:43
thermostat settings and help the
power system, so to speak. So
00:03:47
they consume way too much power
exactly when it's very expensive
00:03:51
to generate. And then the flip
side is, is that there's also
00:03:54
hours of the day where
electricity is very cheap. It
00:03:57
would be perfect to charge your
electric vehicle, say, but you
00:04:02
still pay that same flat price,
or you're not seeing that very
00:04:05
low price in the wholesale
market. So we're also
00:04:07
inefficiently not charging our
cars and doing our clothes
00:04:12
washing when it would be great
to do so. And that altogether is
00:04:17
sort of a total loss of several
billions a year.
00:04:20
- For the utilities, is this price
fluctuation probably their
00:04:25
biggest element that they have
to deal with because of these
00:04:29
conditions? - Yes,
00:04:30
yes, indeed. I mean, we as
end consumers are shielded from
00:04:34
all those fluctuations in the
wholesale market, right? The
00:04:37
prices for electricity can be 10
times as high late afternoon
00:04:41
compared to, you know, two a.m., but
the utility, of course, you
00:04:45
know, faces that variation. So
they can handle that in
00:04:48
various ways. They can sign long
term power contracts with
00:04:54
generators to hedge that price
risk. But still, they need to
00:04:57
buy some of their power on that
wholesale market. And
00:05:00
what's happening, has happened
actually many times before, is
00:05:03
that during extreme peak hours,
when prices that are
00:05:06
normally in, say, the $40 per
megawatt hour, are suddenly 1,000,
00:05:12
utilities might find themselves
buying at very high prices in
00:05:16
that real time market, but still
only selling it to consumers at
00:05:19
that fixed, much lower
rate. So it's a real issue
00:05:23
for utility.
00:05:24
- So what are some of the type of
ideas that need to be
00:05:27
considered in terms of trying to
find a solution that both sides
00:05:31
see the benefit from this
relationship?
00:05:35
- Yeah, well, there's an
idea that a lot of
00:05:40
economists have proposed, which
hasn't gotten much traction yet,
00:05:45
but it's sort of useful to start
out with. It's called real time
00:05:47
pricing, which is essentially
letting consumers face that
00:05:53
price that changes every five
minutes, right? So that would
00:05:56
suddenly tell me that it's it's
five p.m. in Philadelphia, and on
00:06:01
August 21, it's extremely hot
outside, but now my power is
00:06:05
very expensive. So maybe
instead of setting my thermostat to, you
00:06:08
know, 70, I might set it to 74,
and that's a big
00:06:11
difference. The issue is that
people don't like unpredictable
00:06:15
bills. If you don't pay
attention, all the sudden you
00:06:19
hadn't really looked at your
thermostat settings, and you get
00:06:22
a bill the end of the month, and
it turned out you consume power
00:06:25
during really, really high price
hours. And the extreme example
00:06:29
was Texas a couple years ago
when there was this big winter
00:06:32
storm. Texas doesn't regulate
prices very much. So some
00:06:36
consumers who were unlucky
enough to have signed up for one
00:06:40
of those programs ended up
paying a monthly power bill in
00:06:45
excess of $10,000, because the
wholesale price had spiked to
00:06:49
$9,000 per megawatt hour. The
other sort of big issue with
00:06:55
why this is so difficult to
do in practice is that there's
00:06:58
vested interest, right?
Generators love those high
00:07:01
prices, and you know, real time
pricing would probably make
00:07:04
people a little bit more
sensitive to them, and might
00:07:07
actually soften those peaks. So
there's
00:07:10
pushback from all kinds of
sides.
00:07:12
- Are we close, though, to being
able to do that in the first
00:07:15
place, that idea of real time
pricing? - Yeah,
00:07:18
so it's not about technological
barriers. I mean, you need
00:07:24
an electricity meter that
can sort of, you know, handle
00:07:27
that. That's not the biggest
obstruction. What's really hard
00:07:32
is that people would need to pay
active attention all the time
00:07:37
to, you know, what is happening
to the power press. How am I
00:07:39
going to move my activities
around during the day? And
00:07:45
that's where, I think, you know,
innovation in home automation
00:07:50
software and smart EV charging
is going to help, right? If an
00:07:53
app could basically receive the
real time electricity price and
00:07:59
program for you when your car
will be charged if you don't
00:08:03
need it to happen straight away,
for example, that will take away
00:08:07
a lot of that risk that people
would be paying very high prices,
00:08:11
not being aware they're using
that power. So it should
00:08:14
actually become quite a bit
easier over time. But
00:08:17
nevertheless, utilities have
used very different pricing
00:08:20
systems so far to try to
approach it, which our study
00:08:23
finds have not been very
effective.
00:08:25
- But for the consumer, it sounds
like it would be a little bit of
00:08:29
a challenge for them to kind of
deal with real time pricing and
00:08:33
understanding how, you know, the
dynamics play here. And it
00:08:38
almost feels like it's like
you'd have to be watching your
00:08:41
thermostat over the
course of 24 hours to make sure
00:08:45
you're maximizing the
benefits. - Yeah, your
00:08:48
software,
00:08:49
you would essentially have
to, in the beginning, be very
00:08:54
careful and add -- you know,
tell your software what you're
00:08:57
willing to pay and how it
should sort of -- you know, what
00:08:59
kind of rules it should it
use to handle it. But
00:09:03
it's especially for like --
if you own an electric
00:09:06
vehicle, that's a huge
percentage of your electricity
00:09:10
bill. That's typically something
that can, in fact, be charged at
00:09:14
very low prices if you were in a
real time pricing plan, as long
00:09:18
as you're a little bit patient
and are happy to charge it
00:09:21
overnight, which most people
would do anyway. So people
00:09:25
always worry about the high
prices, but those plans also
00:09:28
allow you to take advantage of
very low prices exactly because
00:09:32
you have some electricity demand
that you can schedule during
00:09:35
those off peak hours. So I think
it's a real opportunity for a
00:09:38
lot of people to also save
money. - In
00:09:40
terms of doing this research,
we're able to determine that
00:09:44
with some of these off peak
hours, there's enough of a
00:09:47
pattern that plays out so that
people can understand that this
00:09:51
would be a time, two a.m., three a.m.,
whatever it might be, that this
00:09:57
would be the time that I can
take advantage of this on a
00:09:59
frequent basis. Not just
on maybe one day a month, but
00:10:03
maybe all week or five days out
of a month, that they can
00:10:08
develop a pattern with this.
- Yeah,
00:10:10
so that's the way utilities have
tried to tackle it so far,
00:10:15
because the move to real time
pricing has been so hard. The
00:10:20
most common approximations of
real time pricing, although they
00:10:24
don't get very close, the first
one is called time of use
00:10:29
pricing. It's something along
the lines of what you suggest,
00:10:32
right? There might be sort of a
lower off peak tariff that you
00:10:36
have during the night. You have
a peak tariff, could be late
00:10:39
afternoon, for example. And
then, because it's only two
00:10:43
tariffs, it's not that hard
to remember that you should
00:10:46
charge your car outside of the
peak hours. Now, you can make
00:10:50
those more sophisticated. Some
utilities have actually
00:10:52
introduced something like 20
different tiers. It varies by
00:10:55
season. It varies by day of
week, time of day. Thinking that,
00:11:01
as long as you keep adding more
and more of those tiers, you
00:11:06
will approximate more and more
closely what would happen if you
00:11:10
put people on real time pricing.
And what we find in this research,
00:11:13
which is kind of
surprising to us, and you know,
00:11:18
we got some generous help
here and research and support from
00:11:22
Wharton's Mack Institute for
Innovation, is that those
00:11:27
pricing systems really only
capture maybe 10 percent of the
00:11:32
benefits that real time pricing
could achieve. It turns out you
00:11:36
can estimate that, last year, it
happened to be that Tuesday
00:11:40
afternoons had high prices, but
then out of sample, projecting
00:11:44
it into next year, you're very
often going to miss what's
00:11:48
happening to next year. So in
fact, the more categories you
00:11:50
use, the more you over fit the
data, and the worse these,
00:11:54
systems perform. So
the current trend that the
00:11:59
utilities are trying doesn't
seem to solve the issue, really,
00:12:04
as much as real time pricing
could. - That's
00:12:06
what I was going to say next,
because it feels like, with all
00:12:09
of these dynamics at play, that
trying to fight -- find, excuse
00:12:13
me, what the right formula is
for both sides to benefit here,
00:12:18
it's still got a lot in the mix
to truly understand what that
00:12:21
path is going to be.
00:12:23
- Yeah, and that's -- it's very
difficult, and especially if you
00:12:27
need to forecast it based on
past data, it makes it very
00:12:31
difficult, right? It's actually --
it's going to -- you're going to
00:12:35
get more alignment between the
price that the consumer
00:12:40
pays, and the cost that it takes
for generators to generate the
00:12:44
power, if you can set the price
close to the actual event. And
00:12:51
that's the second system that
utilities have been playing
00:12:54
with. It's called critical peak
pricing, where they say
00:12:56
something like, you know, "You
pay this flat price most
00:12:59
of the year, but we are allowed
to call, say, 20 events every
00:13:05
year." So a day in advance, you
get a text message saying, "We
00:13:09
anticipate prices to be very
high tomorrow because it's
00:13:12
soaring temperatures." You give
people some advanced warning so
00:13:17
they know that tomorrow they
have to be careful during a
00:13:20
certain time window, and they
need to adjust their thermostat
00:13:23
settings, otherwise they would pay
a high price. Or you can
00:13:26
subsidize them for reducing
consumption. That's sort of the
00:13:28
more carrots version. The first
one was the sticks version. So
00:13:33
that system has been used
throughout the US. It's, you
00:13:38
know, the vast majority of
people don't have it, but some
00:13:40
people do. And again, it seems
like that can cover about -- it can
00:13:43
close about 10 percent of the gap
between flat pricing and real
00:13:46
time pricing. But again, it
doesn't come close to something
00:13:51
that directly connects the
wholesale market with what
00:13:54
consumers pay.
00:13:55
- So what do you think are the
ways that maybe need to be
00:13:58
considered in order to try and
increase that percentage of
00:14:02
seeing benefit on real time
pricing as we move forward? And
00:14:05
I'm guessing maybe technology is
going to play a role in this.
00:14:10
You know, innovation, obviously,
but technology will play a
00:14:13
role in this.
00:14:14
- Yeah, to me what was really
interesting about the
00:14:19
results that we found is that
you could do -- you could take the
00:14:23
sharpest edges off of real time
pricing and say, "Let's just have
00:14:28
the price vary, you know, minute
by minute or as frequently as
00:14:31
you want to. But let's avoid
these Texas stories, where
00:14:36
people get totally caught by
surprise, and that leads to
00:14:39
really serious problems
for the customers." So let's just
00:14:45
say that the price can only vary
but up to some price cap. And
00:14:50
the price cap doesn't even have
to be that high. You could say,
00:14:53
if a typical wholesale market
price is 40, you could set the
00:14:55
price cap at 80 or 100, which is
far removed from the
00:15:00
craziness that we've seen in
some of these markets. Turns out
00:15:04
that most of the time the price
still tracks the wholesale
00:15:08
markets correctly, and that
actually recovers about two
00:15:12
thirds of that gap. So
you would -- of the two
00:15:16
billion a year, you would get
two thirds back. When you
00:15:21
run simulations with that,
monthly bills are, you know,
00:15:25
certainly not higher. If
anything, they're going
00:15:27
to be lower. And
the variation in
00:15:29
those bills isn't all that
spectacular, either. So what,
00:15:33
we find in this research
is actually that
00:15:36
all this emphasis on extending
time of use and critical peak
00:15:40
pricing might, in fact, be
pushing really hard on levers
00:15:44
that aren't doing all that much,
whereas we could do a much more
00:15:47
modest version of real time
pricing, which would lead to
00:15:50
much bigger benefits. I think
that's what I would propose.
00:15:54
- Arthur, great to talk to you
again. Thanks very much for your
00:15:56
time. All the best. - Thanks for
having me. - Thank you. Arthur van
00:16:00
Benthem from here at the Wharton
School. - Thank you for listening
00:16:03
to the Ripple Effect. We hope
you found this episode
00:16:06
informative and engaging. Don't
forget to subscribe and leave us
00:16:09
a review so that we can continue
to bring you the best insight
00:16:13
from the Wharton School.
