Okay, good
afternoon, everyone. There's a little bit of housekeeping
to start with. The next homework
is due on Sunday, and I'm actually
going to release the next homework before
this one is due. So on Friday, we'll
finish this topic, and because that one
is due the following Sunday, which is the
day before the exam, I want to give you the
most amount of time I can to finish it.
So I'll release that on Friday afternoon.
So don't stress if you haven't done the
current homework. There's plenty of time
to do both. I just know some people are
ready to click this one. A raring to go
for the next homework. In that regard
the exam not next Monday the Monday
after is going to be up until the
end of this topic. So it's going to be rationality and utility, consumer optimization,
and this topic which is producer optimization
as well as costs. so what we'll cover
next on Monday next week that's not going to
be on the next exam so I've got plenty to test
you on on these three topics any questions
regarding any of that great so I feel like
I can do a better and do a little bit better
in explaining what's going on here in terms
of cost minimization so when we flip
the W and the NPK around like this we know
this is our tangential point when the slope
of the the isoquant is equal to the
slope of the isocost. This is similar to the
intuition we had in consumer
optimization but this is a little
bit different. So just think
about it this way. Let's normalize the
price of wages to $1 and the price of
capital to $1. So the denominators
simply disappear. So this will help
with the intuition. So we know the marginal
product of something decreases over time.
So at the price of $1, if the marginal product
of labour is let's say 20, and the
marginal product of capital is let's say
5, what that means is you're putting too much
resource into capital and not enough into
labour. The last resource you put into
capital produced 5 units of output, and the
last resource you put into labor, produce
20 units of output. So by shifting units
that you're putting into capital into
labor instead, you can essentially produce
more at no cost. So you want them to
equal each other, which means once again
this idea is kind of a Goldilocks zone,
you're leaving no money on the table,
in this case you're leaving no output on
the table. So that's when the prices are
normalized to one. When the prices
have a range of anything, in, we want
to take into account two things. The
marginal product, so how much output
can we get if we invest one into the
next unit of labor? The higher this is,
the better for us. And the wage rate is
how much this costs. The higher this
is, the lower this number will be.
So the higher this is, the worse it is
for us as a producer, the lower this
number will be. The same thing with
the marginal product of capital and the rate
of capital, the rental rate of capital up.
so if one of these is larger than the other
it's the same idea that we're over investing
in the other we should shift some of our
inputs from capital to labor or from labor
to capital so take our mower example here so
once again this was the the idea of the
tangential property the slope of the isocost
curve is the tangent of the isoquant so when
you think about it this is the cost we want to
be producing 100 units you could produce here
at a with this cost line but you can make
it cheaper you can keep bringing down
the cost line with the same slope until it's
just touching q0 which is the tangential
point then we had this example here the marginal
product of a small push mower is three
lawns per day the large mower is six lawns per
day the rental price of a small push lower
is $10 per day and the rental price for a
large one is $25 per day so as you can see here
the marginal product of the small mower
is three the price is $10 and this is greater
than the price of sorry the the marginal
product of a large mower divided by the price
of it so the question doesn't give us much
in terms of how they would adjust their
their quantities but essentially this isn't
minimizing costs between these two products if
they're not equal each other if this you
know person could they should shift some of
the the inputs from the large mower to the
small mower if this was possible that's what
that is saying all right so this is also
relevant to consumer optimization i didn't
really talk about it but this is something known
as corner solutions. So a corner solution is
when you put everything into one but not the
other. So in this case you're putting
everything into labor, not capital. In consumer
theory this person only purchases one good
but not the other. And there is I believe a
question in the homework that kind of has that
flavor a little bit as well. So as you can
see here this occurs when the isoquant is
steeper than the isocost line, we can have a
corner solution here. So imagine this also
went into the negatives and continued like
this, we could find a tangential point
as we usually do, but because this is going
to occur somewhere down here, like this curve
will go like that, our tangential point
will be somewhere down here, you can't have
negative quantities. So it kind of stuck
here at the corner. So in this case this person
only hires labor and uses zero capital and
we can have the same situation as well with
capital if the iso quant is flatter than
the iso cost lines we can have this example
as well so just be aware of corner solutions
and corner solutions i think are kind of
important when thinking about the world because
for example as capital gets cheaper or as labor
gets more expensive it shifts whether
these slopes become steep or flat and as
a result it can change where the solutions are
and someone might only have capital and not
have labor so okay so as we just talked about
to minimize the cost of producing a given
level of output the firm should use less
of an input and more of other inputs when that
inputs price rises. So this kind of has
a similar feeling to substitution
income effects and slightly different
in how we go about it. So our original
budget line is FG, so this here, and on, sorry
not budget line, isocost line. On
this original isocost line, FG, we can see
that's tangential to the isoquant of
Q0 at A. So we can produce Q0, let's say
Q0 equals 100 units of output. We can
produce 100 units of output at the lowest
cost being here on this isocost line,
which is point A. Let's say labour
increases as some policy passed wages
increase. What this is going to do is
it's going to make labour more expensive,
so it changes the isocost curve like
this, similar to our budget lines. We've
seen this before. What this means is we
can no longer produce 100 units of output
at the same cost, at the same cost. If we
want to produce 100 units of output, we're going
to have to increase our cost. We could
produce less than 100 units if you wanted to
produce at this point is tangential to an
indifference curve, which could be somewhere
around here, for example. But what we can
do to produce the same amount of output is
take the slope of the new isocost line and
move it all the way up until it's tangential
with the isoquant Q0, which is 100 units
of output. And as you can see, this occurs
at this point here, B. So to produce at the
same output, it's going to cost us more, and
we're substituting away labor for capital.
As you can see, labor decreases from l1 to l2
and capital increases from k1 to k2 and
this is more expensive than before relatively
more expensive uh so i want to take a
little bit of a detour here just talking about
this idea of of how you know as labor
becomes more expensive or as capital becomes
cheaper we can see var shifts away from one to
the other which is as you can see here if
labor increases there's much more investment
in capital than labor vice versa if capital
becomes a lot cheaper we'll also see something
similar as well and I think we're
currently in the midst of you know a mini
revolution right now which could turn into more
than a mini one in the near future and that's
the idea of AI as a form of capital which
is already having vast effects on the labor
market which I'll talk more about in a second
but I think it would be remiss of me not to
get us to discuss this because so many of
you are about to go on to the job market in
a couple of years to figure out like hey like
what's going on here and to get each other's
insights so what i want you to do for the
next few minutes is you can do this on your
own if you really want i'd encourage you to
discuss with people next to you can be
pairs threes fours i i don't really care so
much you can discuss it with people and i want
you to write down or just think about the
three jobs the three professions you think
will be the most likely to be replaced by ai
in the next three years also you could you
know give me a you know a further on time
horizon if you're like yeah three years this
might be okay but 10 years it's close you
can do that as well also what you think
the least like the jobs that are the least
likely to be replaced by ai and kind of
similar what skills do you think are going to
be the most important for employment and
the skills that are the least important
for employment given a lot of you uh in the
midst of these these decisions right now
i think it would just be good to take a
step back and think about it also we didn't
actually talk about this in the other class
in the end, but I'm curious to get
people's unemployment figures in the next
three years caused by AL. So, take a few
minutes just to discuss amongst
yourselves, and then we'll bring it all
back as a class. Logan, if you want to
join those guys. Thank you. Okay, let's go to the
back. So in the previous class, we had a
pretty hot discussion about these people who
were quite passionate. I'm curious. Let's
start off with what jobs do people feel
like are the most in danger in the
next few years? Like, what would be a
bad profession of getting to right now?
Yeah. Data entry. Why data entry? AI can do it faster and
cheaper. Okay. That's a great answer.
So yeah, data entry, if it hadn't already
been, like, started to get replaced
over the previous 20 years with, like,
other software now is probably, you know,
it's going to happen soon. Yeah. I'm talking
about animation. Animation, okay,
can you tell me more about that? I have
no knowledge of animation, so. So, it
was already so big, like, the way I can
bring it into this. Yeah. And it was kind
of, like, sort of, basically, as an
enterprise. Yeah. So, I feel like, didn't
that change a bit, like, 15 years ago as
well? Like, I know, for example, South
Park used to, like, draw everything and
actually move the stuff around, and then it
changed to more digital, like, 15 years ago,
and we're probably you're going to see
something similar in that regard with
animation as well. You don't need to hire
animators. You can have one person do everything.
Is that the idea? It's like AI
can do anything. Ervin, I might come
to you because you said something
really interesting. Animation is like a
form of entertainment. I think you had a really
interesting perspective on entertainment as
an industry here. Yeah. Yeah. Yeah, so like,
you know, art, for example, like, yeah,
like, AI might be able to do art better than
humans, but do we really care about AI and art?
Like, why do we care about art? Why do we
care about music, for example? Is it the
end product, or is it like the emotion and
the story behind it, et cetera? So I found that
really interesting. And I know there's a
lot of people pushing back on AI animation
and things like that. It's all a bit of
a hot mess right now, but I could
definitely see it going in either of
these directions. Yeah? Paralegals. Okay,
what paralegals? You've got to
tell me more about the paralegal
industry, the lawyers. The actual lawyer that goes in for instance. Okay. The guy's
not paralegal. Fair enough. The majority of people in a big one. Yeah. Daniel? That's a good one. So,
like, a lot of people talked about data
analytics and things like that, even things like
coding as well. So, I don't know if you
take this with a grain of salt or not,
but there's, like, people saying that
the people at these AI companies, so
Anthropic and OpenAI, aren't doing any
of that when coding anymore, they're
just getting AI to code for them. And
I think there's a lot of pushback on
a client like that as well, but I
think that's really interesting where
we're at right now. Any others about
professions that are in trouble? OK, what about
professions that are most robust, do you
think, to be able to apply? So do you think
everyone's going to go eventually and
we'll have zero labor? Yeah? I think that's a really
good point and it seems like a lot of
people are predicting their roles that
have like some form of either human oversight,
human connection or something that's
trust based as well. It might take
a long time if not, you know, at no point
for AI to overtake that. So something
like therapists, someone actually
mentioned HR positions. Yeah, like you can
get rid of a lot of stuff but you might
still need some people to validate things as
well. So to have some human oversight as
well. So those sort of things are I think
are really interesting, but there are people
using high eyes There's a therapist
out there to my understanding. I don't
know how effective it is I think it's worth
testing, but yeah, there's a lot of like
human AI interaction You guys spoke about
some jobs that you think are robust right
now The trades so plumbers, electricians.
Does anyone know what we call the
electrician in Australia? Call them a sparky.
That's what we call them. Yeah, so it
trades people for now that they seem okay
but I don't see a reason why we can't
have AI's in the future that can inhabit some
humanoid body that are able to do those
things as well. Probably talking far
future here, but I don't know why that
would be more safe than the white-collar
work in the future. And also, all of you
here, because presumably you want to end up
in white-collar jobs, this is a trade
class, for example, so does anyone have any
other white-collar positions, for example,
in business that you think are going to be
thriving or safe in the future. Well, not
necessarily in business, but I feel like doctors
are pretty well. I mean, AI can
do so much to an extent, but it's
going to need to be doctors in
the long term. I think this is
really interesting. I'm going to tangent
for a bit to come back to this point.
Has anyone here been in a Waymo
before by any chance? Tell us about
your experience in a Waymo. How
did you feel? So the safety
features have made it like too slow,
kind of? It wasn't like that slow. I
didn't like it. I'd rather have a
driver. Interesting. So in general, this
is the idea. There's a lot of people that
argue that if every car was self-driving, we
could have a much safer system, for example.
A lot of accidents happen due to human
error. but if one death happens due to like
artificial intelligence or an algorithm rather
than human error people probably weight
that more than the death due to human error so
people are much less forgiving about death
due to like a machine than they are of
humans so I think the same thing will apply
to doctors it might actually be the case that
like an AI can perform a surgery in a much
more efficient manner but if the rate of
an accident is 1% for the AI and 2% for the
human I feel like a lot of people will
still choose the human because psychology is
kind of weird in that regard. But that's a
really interesting one. Okay, what else do we
have here? Okay, yeah, what do people think
about the university? What's going to
happen to college and university in five,
ten years? Does anyone know about
your predictions? Should I start
looking for a new job? No predictions here?
I think there's going to be a huge upheaval
in the education industry. Yeah, Ivan?
I was going to say, AI can, I mean, university
might be a little different, but, like,
just in terms of, like, teaching in general,
a lot more beyond, like, just the knowledge
you're learning is, like, also kind of,
like, the connection you make with,
like, your teachers, especially, like, people,
like, in elementary school or in middle
school as well. Yeah. I feel like having, like, learning straight
from AI, there'd be no point of, like,
going to school. Yeah, yeah. So yeah, we kind of
mentioned this at the start of the
semester, this thought experiment,
like do you come to school because you
want the piece of paper which is some
sort of signal to employers, or is
it what you learn? And at the end of the
day it's kind of a mix of both, and
there's kind of a third thing in there as
well, network effects. You meet a ton of
people here within your courses, and
you build connections which will take you
through your career. One of my hot
takes, I'm being a recorder right now,
so hopefully it doesn't come back
to bite me, is I think MBAs are
a crock of shit. In terms of what you
learn in an MBA, the content is not
different from the stuff we teach in first
and second years in economics or in basic
probability theory or some of your management
classes as well, which are more like
case-based studies. However, that doesn't mean you
shouldn't do an MBA because the evidence
suggests if you do an MBA, your salary
doubles within a year. That seems kind of nice. The reason for that
is A, the prestige of the MBA itself. It's
a signal to employees. Two, the network
effects are so strong. If you do an MBA, like
a good MBA school, you have networks of
all the previous MBA students, your
current crop of MBA students as well, who
all end up in pretty high-achieving places.
So, you know, it's much easier to get
your foot in the door. So there's a lot of
reasons why, you know, institutions like
universities are successful. if we really wanted
to just maximize our knowledge it could
be a better way of doing it rather than
our current system here I'm kind of curious
about research I think AI is doing a lot
like in research right now and also if we
see massive shifts in the job market that
might mean less people attend university I
don't know the future is going to be pretty I
think up in the air so I wanted to give
you a little bit of information of of what
I'm seeing in regards to where AI is currently
at and where it's going to help give
you a little bit more information about how
you should spend the remainder of your years
here at college to maximize your opportunities
in the future. I'm actually on the
AI task force for the business school, and
it's a bit weird. So currently, the
business school is trying to find ways
to get AI put into the courses so you learn
how to use it, etc. I don't know if AI's
good in terms of a complement. I think
it can be useful, but I think in the future,
it's just more of a substitute delay
within a complement. However, from my
understanding, right now, there's a lot of
firms coming to Purdue and saying, hey, we
want students who know how to use AI.
We have no idea how to implement it
ourselves. We want our graduates to be able
to do that for us. So I know for a fact right
now, there's a lot of firms out there
looking for people who have an understanding
of this. And I think there's only like
20 to 25% of firms that are actually
utilizing these skill sets. So that's something
to keep in mind. So this is a company
called METR evals and what they've done for
each AI model that's been released is to see
how capable they are essentially and the
trend of capabilities so what this is on the y
-axis is a human-based task how much time
it takes a competent human to complete this
task so 30 minutes would be a simple task for
a human to complete whereas 12 hours is
something that takes a human 12 man-powered
what hours to complete. And what this is
measuring is each model's ability to
complete a human task at at least
50% probability. So as you can see, the early models
in 2019, 2020, GPT-23, were like, okay, not
that correct. They could, you know,
50% likely of completing like a 20
-30 minute human task. But as you can see
recently, especially in the past year,
this has kind of exploded a bit.
And this was Opus 4 .6, which was
released, what, or two weeks ago now or
something like that. And this is saying
essentially that this can complete
a task that takes a human 15 hours with
50% probability. But it's the trend
here that's really important as you
can see. So a lot of people arguing that
there's this explosive growth in the ability
for AI to actually do things. So
whether you think the current capabilities
are any good or not, that doesn't matter
if there's gonna be this explosive
growth in the future. One caveat here is
there's huge error bars on this. So as you can
see, the little gray error bars here go
down here and actually go so far up you can't
see it on the graph. What this means is
they estimate it can take anywhere between
do a task of six hours all the way
up to 98 hours. So it's a very imprecise
estimate. So I take this with a
grain of salt. I just think it's a really
interesting graph showing how the
capabilities are growing. This is a paper
written last year, not published yet, by Eric
Brynjofelsen and co-authors. eric is the head
of the like econ ai institute at stanford
they do a lot of really cool work and they've
started looking into employment and the key
headline here is that they um find that early
career workers ages 22 to 25 in ai exposed
occupations experienced 16 relative employment
declines controlling for firm level shocks
while employment for experienced workers
remained unstable so the the the argument
here as you've seen in like more traditional
media as well is that it's the entry-level
jobs that are starting to disappear and i
know this kind of sucks here because you're
all about to try and get entry-level jobs but
i think it's important to see the reality
on the ground as well and how you can approach
that so um this is some of the data as
you can see the gray line is firms that
aren't adopting ai right now and red lines of
firms adopt in AI and they find at this point
here in 22 there's a stark difference between
the two in terms of entry-level jobs or
entry-level highs but this decreases by like
10 percentage points compared to the other
once again the data is messy at the moment
there could be a lot a lot going on like there's
a lot of uncertainty if you're talking
like AI jobs and the economists I took a
screenshot of this last night and you have like
three very different articles already within
like a week of each other how big a threat
is AI to entry level why AI won't wipe
out white collar jobs stop panicking about
AI start preparing so everyone has different
predictions for example some people are arguing
that actually entry level jobs are going
to be more likely in the future they're cheaper
than more experienced people if you need
one or two people to help with oversight
or with trust or with human properties entry
level people are going to be better for
that because they're cheaper workers and more
experienced people so that could be a trend
that you see as well. Alright, I really like
this. So Alex Imas has a really cool
blog post and a blog. He's an economist
at the University of Chicago. His name
might sound familiar because I've already
seen one of his papers. You remember all
the way back when we discussed incentives,
one of the papers we looked at was how
people put in effort when the outcome of
the effort is given to them versus given
to like a charity. That's Alex's paper
called working for the one globe and over
the past year and a half he's only like
written about ai he's completely pivoted and
he has a whole bunch of stuff on on on the
on the future of the economy and on on
employment etc there's some cool stuff out
there that i'll share and he said here here's
the question i get asked most how ai will
affect art and artists so we talked about
animation we talked about other like kind
of artistry before and this is his like
summary answer ai will be incorporated into
art but it will not replace artists in
fact as more and more human labor becomes
automated there will likely come a day when most
if not all of us will end up doing work
that right now would be classified as art
or at least art artisan and he goes into
his argument why so there's a lot of economists
thinking we really are on the precipice
of something here um cool so before we
move on i have a guest speaker coming next
wednesday for my experimental economics
class which is 11 30 to 12 30. however i've
booked a bigger room if anyone else is
interested is an assistant professor at Carnegie
Mellon and he is an expert in experiments
in AI so he has a paper he's just written
for example which is a field experiment
using a hundred thousand you know people
to see what's more effective AI recruiters
or human recruiters he finds AI recruiters
much more efficient and people are
actually happy with AI recruiting compared to
human recruiting so he's going to talk about
that and then he's going to talk about
other ideas that we are pursuing right now
and figuring out what AI is better and worse
at than humans. So if you have any interest
in that I'll post the details on it for
you. He's great. I highly recommend but
I know you have a ton of other things on so
no pressure to come. Okay so I felt that
was an important detour but a detour nonetheless
none of this AI stuff will be on the
exam. It's just something to think about and
talk about. My door's always open if you
want to talk about this more. so we're talking
about cost minimization but let's talk more
about the cost function this is the mathematical
relationship that relates cost to the
cost minimizing output associated with an iso
quantity so we have short-run costs divided
into fixed and variable costs so a fixed cost
is defined as a cost that doesn't change
as output changes so you can have zero
units produced one unit produced or a billion
units produced, fixed costs don't change in
any of these scenarios. Variable costs,
on the other hand, are a function of
quantity. So as quantity increases
or decreases, variable costs
increase or decrease. Then total cost
is just our fixed cost plus our variable
cost together. In the long run, all costs are variable,
you can make changes, etc., so there are
no fixed costs. How we define the long
run is like an interesting question,
but like usually it's around six months as
a convention okay i'm not going to play this
video um just because i've already detailed
enough but in this breaking bad scene
walt jesse and mike are talking about
their profits and then mike keeps taking
away money from walt's pile talking about
all these costs and we can talk about which
costs are fixed and which are variable yeah
so graphically this is what they look
like so quantity is on the x-axis price or
cost is on the y-axis You can see fixed
costs are horizontal. What this means is,
as you can see, if we produce zero, if we
produce here, or if we produce an infinite
amount of quantity, the fixed cost is
always the same price. It doesn't change
as quantity changes. Whereas variable
cost, you can see, starts at zero
when we produce nothing. There's
zero cost. And as we produce more, the
cost increases. However, as we can see,
the slope is changing over time. So as we
can see it originally has this concave shape.
So what we say is that it's increasing
at a decreasing rate So the next unit of output
cost less than the previous unit But once
we get to like this point here, you can see
we start having this Exponential
convex shape so the next unit
of output here Increases the variable
cost by more than the previous unit. So we
can have both these functions here and as you
can see the relationship between total cost
and variable and fixed cost is that it's
just the exact same shape as the variable
cost curve it's just shifted up by the fixed
cost so the variable cost starts at zero
we know the fixed cost starts here so the
total cost starts here follows the same shape
as the variable cost and at any point in
output the difference between the variable cost
and the total cost is just the amount of the
fixed cost that's how to interpret that
yeah so we have a few different definitions
here so the average fixed cost is just our
fixed cost divided by the quantity of output
so when you think about this this only
decreases only decreases fixed cost always stays
the same so as the quantity of output
increases the denominator becomes larger the whole
number becomes low. The next is the
average variable cost. So this is your
total variable cost divided by the quantity.
And whether this is increasing or
decreasing depends. As I showed you before, if we're at this point
here, as we produce more, the average
variable cost becomes less and less
and less. But as we start producing here,
the average variable cost becomes more
and more and more. Average total cost
is the same sort of thing, just
the total cost divided by the
quantity of output. Finally, our good old
margin, our marginal cost, is just how much
does it cost to produce one extra unit? Or
how does the change in units produced affect
the change in cost? Hopefully, I've kind
of drilled the idea of marginal thinking
all into your head right now. So this is
a table format of it. so you can see the
output here is increasing the fixed cost is
the same at every point the variable
cost increases by 400 at each point the
total cost increases by 400 at each point
after it's got the 2000 you know fixed cost in
the first um scenario of producing zero
as we talked about the average fixed cost
is always decreasing always decreasing
however However, as you can see with the
average variable cost, it's decreasing,
decreasing, decreasing until here, until we have 2
,124 units, and then we produce
more, the average variable cost
is increasing. And it's the same
thing for the average total cost as well,
decreasing, decreasing, decreasing up until
this point here. And this is because of
the relationship between the marginal cost and
the average variable cost and by the average
total cost so the first thing you can see
here is this average fixed cost curve starts
off extremely high and as soon as you
start adding more output quantity which is on
the x-axis it drops a lot and then like each
extra unit once you get down here only
changes it a little bit so as you can see it goes
from five to three sorry it goes from 26 to
eight to four but then all the changes here
are very marginal as well it has this
asymptotic nature the next thing to notice is that
the average variable cost and average total
cost are happy. They have this, you know,
smile type curve here. And kind of the same
with the marginal cost as well. It looks
like this. And there's a relationship between
the marginal cost and the average variable
cost and the average total cost curve. So
before talking about it in an economic
sense, let me talk about it in a sports sense,
a basketball sense. So LeBron James' average,
this was in 2021, so he's probably
five more years since then, but in 2021 he's
his average was 27 .05 the average points
per game is the same thing as like an
average cost when you think about it and each
game he plays is either going to score more
than 27.05 or less than 27.05 this is the
marginal points per game the next unit
of games so as we can see here he started
off his career with a very low average of
like 20 but you can see each of these
dots are the points he scored in each game,
the line is his average, and when he scored
more points in a game, it dragged the average
up, then as you can see here, when he
scored less points than the average, it drags
the average down. So this is the relationship
between the two. So if his average is
currently 27.05, and in the next game he
drops 50 points, this is going to drag his
average up. If he scores three points in the
next game it's going to drag his average
down this is the exact same relationship that
the average cost has with the marginal cost
so what this is saying is every time lebron
scores less than his average it's going
to drag it down every time he scores more
it's going to drag it up so everywhere between
this point a and up here the marginal cost
is below the average cost it's matter if
the marginal cost is decreasing or increasing
you can see it's dragging down the average
cost as long as the marginal cost is below
it each extra unit you produce is cheaper
than the average so it drags it down all
the way up until point a where they equal
each other and then in this quadrant here the
marginal cost is greater than the average cost
lebron is shooting more than 27 points
per game he's dropping 30s and 50s etc so
every time that happens it starts dragging
the average cost up so the relationship here
is as long as the marginal cost is lower
than the average cost this is decreasing the
average cost and as soon as it goes above then
the average cost is increasing so I feel
you know this obligation whenever an appropriate
clip arises to show you a scene from
the office and this talks about the
relationship between fixed cost average cost and
how this affect how this changes with more
output how much can we afford to pay a delivery
well these numbers you gave me are correct
they are correct sir then you can't afford
to pay them anything okay a lame attempt
at humor swing and a miss your prices are
too low lowest in town why do you think
staples and dunderbiflin can't match your prices
corporate greed look our pricing model is
fine i reviewed the numbers myself over
time with enough volume we've become
profitable With a fixed cost pricing model,
that's correct, but you need to use a variable
cost pricing model. Okay, sure. Right.
So, why don't you explain what that is
so they can understand? Just explain what that
is. Explain what you think that is. Okay.
Just explain it. As you sell more paper and
your company grows, so will your costs. For
example, delivery man, health care, business expansion. Whatever. Yeah. At these prices, the
more paper you sell, the less money will
make. Our prices are the only thing keeping
us in business. They're actually putting you out of business. Okay, okay. Hold on. Hold on. Ty, I would like you to crunch those
numbers again. It's a program. There's
no such thing as... Just crunch them. Just
crunch them, please. Crunch. Did it help? Okay, so let's go
back to you to talk about what's going
on. So those of you who aren't
familiar with that episode or storyline
from The Office, Michael, Ryan and Pam,
they break away from the main company,
Dunder Mifflin, they start their own paper
company, and the fixed costs for this paper
company are really low, like they work out
of one of their vans, etc, they do their own
deliveries, etc, they have really low prices
because of this, the fixed cost is fine,
but because of these low prices, like they
need to start expanding to be able to become
more profitable, but as we can see here,
They probably have very low average costs
like if they're at this point But as I
start selling more the average cost will start
to increase So their current prices which are
really low They can't afford to keep that
model as these average costs start to increase
because the marginal costs are increasing
He talks about things like getting more
delivery people paying for health care the
more labor you hire as That's the idea
behind using a fixed cost pricing
model compared to the variable cost
pricing model So, let's finish up
today with a return to opportunity costs
and sunk costs. So, if you recall,
the opportunity cost is the
value of the next best alternative
that you give up. So, we gave a ton of
examples of this, so this is both included
explicit and implicit costs. So, at the
individual level, coming to class today,
you're essentially giving up whatever you
could do within an hour or 50 minutes of your
time instead. You could be having a nap
right now, studying for another class,
doing anything really. On the weekend,
you might want to watch a movie on a
Saturday night or go to a party.
You can't do both. Whichever, you
know, alternative that you don't
do, which is worth the most, is your
opportunity cost. These are more
implicit than explicit. We gave the example
of attending college that has
that explicit cost as well of tuition,
book fees, etc, etc as well. As
long as the implicit cost of what you
could do instead. At the firm level,
for example, you could open
up a business, it could be a cafe or
any business, and the opportunity
cost of that is all the resources you
spent, you know, getting the business
up and running, plus the implicit
cost, which could be, for example,
you could have got a job at, I don't know,
KPMG or something like that instead.
So you've got to take that wage into
account as well. Also, how should
a firm use its resources, sell or
use as an input? So should a firm use
its machinery to produce cars or
should it sell the machinery? The
machinery has value. We talked about these
examples of a firm that produces both
trucks and cars. How much of its resources
does it put towards making trucks? How
many to make in cars? Whatever you don't
use it for, that could potentially be the
opportunity cost as well. Depending on your
outside source, the opportunity cost differs.
So I want to give a an example of a
famous person and how their opportunity cost
can change over time Ah, um, so as you can
see here, we have a very young mr. Beast
a very young mr. Beast So this is a video he
released. I'm gonna make this nine years
ago I don't know if anyone's seen this
has 33 million views 20 is 23 hours, so we're not gonna watch all of it Okay, so that gives
you kind of an idea of the mixture of
emotions coming through. Also, as you said,
I think that the video is 24 hours,
but actually it took him 48 hours, and you
have to like narrow it down. You couldn't
publish more than a 24-hour video or
something back then. So the question
I have for you here is, what
is Mr. Bass's opportunity cost
in this situation? What are the explicit and implicit costs here? Ivan, do you
have any ideas? I mean, he can
literally do anything else for this
time. Yeah, so 48 hours of time is
the implicit cost. What are the
explicit costs here? I mean, I guess he's not really purchasing
anything. Yeah, so like a camera
and a microphone. I mean, yeah, I
got to think about that. But yeah, I
guess that would be as implicit. Yeah,
explicit. So these are the low explicit
cost. And his implicit cost, what do you
think back then, nine years ago, what do
you think the best thing he could have
done with his time? It's a tough question. I would have
probably, I don't know, had to
go see Drake. Okay, okay. So yeah, back
then, he was quite unknown. As you can
see, he's shouting out, please other YouTubers
share my videos, etc. So, if we go back
to here, that's not what we want. This
is what we want. so how do you think
his opportunity cost for this video would
differ back then in 2017 to today
i'm going to pick on you again do you
think it would be worthwhile him making
this same video today 48 hours i'm doing
something that's not entertaining like this
is probably not the right depending on what
your outside option is the opportunity cost
can differ especially on the explicit
cost the explicit cost would be very
similar if not the same and we can actually
look at one of his more recent videos so um
if people of here have seen squid game i
assume i don't know if some of you have
seen this video in particular um but this
has 908 million views i think is this his most
viewed video of all time we can have a
look oh it's not oh yeah it is this is most
viewed video of all time so remember the
explicit cost of the previous video was a
webcam and it may be a microphone here he's
recreated the entire set of squid game
you can see here and then yeah so the
explicit cost for this is so much more than
the first video so he gives away to people
plus all the crew that he has to pay
probably insurance and stuff as well I'm sure
he's had mishaps on set in recent times
as well he's probably had to pay out legal
etc So the cost from that first, one of
the first videos you ever created for this
is just exponential. And there's a
question about, was this the
best actual way to spend time
making a video? Probably yes, given it
got 900 million views, but, I mean, you
could think about the explicit cost, and the
implicit cost has probably taken a while to
plan and shoot. You could have probably
made 10 videos in that time. And for a lot of
content creators, this is a trade-off they
didn't make. Like, how often they release videos,
like you can probably make five short
videos for one larger video. so if you notice
content creators will usually either release
something once a week on YouTube once
a month or some people like do a whole bunch of
shorts every day these are all different
strategies about what they think is the best
way to optimize their viewership and profit
because there's an opportunity cost at
each point what you're giving up by producing
five videos what you're giving up by producing
you know one video a month etc okay we're
going to end up here and we'll speak back on
on on sunk costs next week yeah on Friday
and we'll probably have occurred on Friday
I should mention that Yes, yeah, I wonder
if this game actually makes a problem, like
do you know how many, apparently it's very
highly viewed, but I, yeah, oh really, oh
yeah, did he do okay, I didn't watch it,
but like the final, yeah, yeah, he's in
the final, I didn't watch the episode, oh
is it still going, or, I think the last
episode just came out, and he probably had
an NDA or something like that, oh yeah,
yeah, yeah, yeah, I wonder what happens if
someone breaks an NDA, like does he take the
money back, I mean he's a ruthless
business family oh yeah but no i i think he's
fascinating so well from asking people
to like share on twitter and youtube is
where he is now 90 it's pretty funny thank
you have a good one yeah i'll release
practice well let me write the actual
exam yeah yeah so so i know i
know this is the last minute this
is just a fun show I'll make sure
you guys have a practice exam,
but, yeah, we'll probably be next
week sometime. Thank you. No worries. Have a good one,
everyone. I just have a question. Yeah,
what's up, David? Did you just work at, like, the
business school? The tour? Yeah, the
business school, yeah. Like, what do you mean? Like, I do,
like, the tours. Yeah. And, like, you're,
like, helping the business
that I helped. And you want me to?
I can give you a tour of the Econ department.
I don't know if I'm going to do
the whole business school, but I know
people that do like tours in the business
school for like their job. So I can put me
in touch with them as well. Okay, yeah.
So what do you do to do this for? You
have to use a player. Oh, are you counting
to 100,000 or? Yeah, like 10%
there. Okay, so like you just do tours
in the store? No, I do that. Oh, nice, nice. Do you want to have a game with me sometime? Oh, yeah. Yeah, yeah. I'm just running
Wildcat the entire time, I'm just
letting you know. I'm running 26. Since I moved over from
Australia, I'm having my Xbox with me. Are
you a PS or an Xbox? I'll have to buy a new
one, get acquainted just so I don't
embarrass myself. Yeah, that'll be
cool. Happy to do a tour for
you if you want. I can organize that
for you. Okay, cool. So what do you do
with the engineers pulling your videos?
I do most of the ECEs. Yeah, so I can do a
video with you as well showing the experimental
economics lab.