Decision Analysis: Part II

Transcript

Slide 1: Welcome to "Decision Analysis: Part II."

Slide 2: In this presentation, we're going to calculate the Expected Monetary Value
or the EMV, and also perform Sensitive Analysis.

Slide 3: Let's consider a case where we need to make a decision under risk. There
are several states of nature and the associate probabilities for each state is known.
So, we don't know exactly what's going to happen, but we know what's possible and
what the probabilities are for each one of those outcomes. For example, the
probability of the weather tomorrow, we're going to assume that this, it's either
going to rain or not going to rain. And the probabilities are 0.2 and 0.8.

Another example, let's say you're trying look at the sales of iPhone, you know,
there's three possible possibilities here. The sales are going to be high, high sales,
assume 30%, probability the sales are medium, or medium sales is 50%, and the
probability of low sales is 20%. The assumption here is there's only three outcomes,
high, medium, low, and we have a probability for each one of those outcomes.

Slide 4: We're going to calculate the EMV or the Expected Monetary Value. Typically,
in the decision-making scenario, we calculate the EMV and we choose the
alternative with the highest EMV. Money is really the most common measure we
use in a lot of decision-making criteria for a variety of reasons. And EMV is probably
the most common way to assess each alternative.

This is by far the most easiest and popular method to calculate the EMV for each
alternative, it's simply the summation of each possibility or alternative, and the
probability of each alternative. So, typically it's the outcome or the payoff of each
alternative and the probability of each one of those alternatives. And basically it's
the expected value of that decision.

Slide 5: Let's consider this scenario, a new product production. A fictitious

company, ACME company, must decide on the level of production for its new
smartphone. The alternatives are, you could opt to do a high volume, medium
1
volume, or low volume production. The actual sales, though, we don't know exactly
what's going to happen, but there's two alternatives or possibilities. The sales could
be strong, strong sales, or sales could be weak, weak sales. And let's say we know
the probabilities, we have the probability for each possibility. The probability of
strong sales, let's say it's 40%, the probability of low sales is 60%. So, we have two
possibilities, and we have the associate probabilities for each possibility.

Slide 6: We have the probabilities, now let's get the pay off the table. So, we have
three alternatives, high volume, medium volume, and low volume, and then based
on the State of Nature, whether you have strong sales or weak sales, we have
different payoffs. I do want to note here, the do nothing option is technically here
as well. I'm going to hold it here for now, but technically you can opt to not do any
kind of production.

So, saying we're not going to produce that at all. And the assumption there will be
strong sales and weak sales, the payoff will be zero and zero. And okay, so based on
this payoff table, we have to decide which alternative to actually pursue under
Expected Monetary value or EMV criteria, either high volume, low volume, or
medium volume. So, we're going to calculate the expected monetary value for each
criteria. So, we're going to take the payoff at strong sales times the probability of
strong sales, plus the payoff at weak sales times the probability of weak sales.

So, the EMV for the high-volume scenario is going to be 14 billion times 40%, plus
the weak sales scenario, negative 3 billion times 60% so EMV ends up being 3.8
Billion, for this alternative. Now, we could repeat for the other two alternative. So,
quickly do that. So, for medium volume, I'll just do right it over here, For medium
volume, we're going to take the P of the strong sales, so medium here, the payoff is
6 billion. The probability of strong sales is 0.4 plus the payoff for a medium volume
weak sales is 1 probability is 0.6, and that results to 3 billion. Okay?

And then for low volume production alternative, okay, the payoff is 4 billion if the
sales are strong, which has the probability of 40%, the payoff is 3 billion if the sales
are weak and the probability of weak sales is 60%. So, monetary value here ends up
being 3.4. So, we have the EMV for all three alternatives.

Slide 7: We have completed the Expected Monetary Value calculations for each
alternative, and the results are shown on the right most column on the EMV
column. So, we are actually now ready to make a recommendation. Since we are
looking at profits here, we want to maximize the EMV. So, the alternative that ACME
should pursue under EMV is going to be high volume at 3.8 billion. If this is were
not profit but cost, then we'll be seeking the lowest value here Since we're just
profits, assumption is we're going to try to maximize the EMV.

2
Slide 8: Continuing on with our analysis due to EMV. I think the more interesting
question is, how do our decision change if the state of nature probabilities change?
Okay, so the 40% probability for strong sales and 60% probability weak sales, what
if those probability change or they're not you know, it turns out to be very
inaccurate. How sensitive is our decision to that parameter? So, what if the state of
nature probabilities change, what happens?

Yeah, we can perform Sensitivity Analysis, see whether our decisions robust to
solve these changes. We can do this by simply looking at the way we calculate the
Expected Monetary Value or the EMV. For the EMV for the high volume, we did 14
billion times the probability of strong sales plus minus 3, which is the payoff for
weak sales times the probability of weak sales, which is one minus the probability
of strong sales. And that's how we got the values here.

So, if you look at the equation, okay, what we ended up doing, what we actually did
was 14 times P, where P represents probability a strong sales, right? Okay, let me
be very specific here. P is the probability of strong sales, right? Okay, so the EMV
high volume was 14 times P minus 3 times 1 minus P. Okay? So, you simplify this it's
14P minus 3, plus 3P. 17P minus 3, okay? So, if you look at this equation, this, you
can actually plot this because you know, again, you go back to, you know, maybe
your algebra days Y equals to MX plus B. That's the questionable line. So, my Y here
is my EMV equals 17P minus 3. So, my M here, this is my slope. - 3 is my Y-intercept,
right?

Okay, so you can actually plot this and see what this looks like, right? And then you
can do the same thing for the other two alternatives. So, I have the Expected
Monetary Volume for medium volume, and I could do the same thing for a low
volume. So, essentially I have three different lines with three different slopes and
three different intercepts, okay? And so, if plot these and see what they look like.

Slide 9: So, if you plot this out, the blue line represents the EMV for the high-volume
production. The red line represents the medium volume production, the green line
represents the low-volume production. And so those are the three lines for three
alternatives. So, in our case, the value we, the initial probability for the strong deals
is 40%. So, basically, you kind of imagine that's basically over here, right? And so
that's why, you know, if you want to maximize EMV, we ended up picking this value
here, which is high-volume production, right?

So, that's the, you know, that was the solution deemed right for us. The question,
the more interesting question is though, is what if, you know, this probabilities
change, right? So, at what point do we change our decision from to something else?
You know, based on what was provided earlier, our best decision was to go with
high-volume production. But if the probability has changed, at what point do we

3
change our decision, okay? Then really, you know, if you look at this graph, okay,
you're always trying to maximize the expected monetary value. So, for any given P
value here, okay, you're trying to pick the best EMV, right? So, really you're looking
at this line right here, right? Okay. Initially this green and then blue here, right? So,
it's, if you look at this graph here, you never choose the red line, the medium line
production, right?

For any given probability you have another alternative or a higher EMV value, okay?
For a lower probability, our best decision is to go with the green line here, the low
wide production. Once we hit this point, A, we switch over from low-volume
production to high-volume production, okay? So, you know, we have to decide
where this point A lies, right? So, and that's what we're going to do next.

Slide 10: So, at point A, we know this is a point where the Expected Monetary Value
or the EMV of the low volume equals the EMV of the high volume, okay? And we
know the equation for each one of these lines, right? So, EMV for the low volume,
we know it's going to be it was P plus 3, and the expected value of the high volume
was 17P minus 3, okay? Going to move them over and solve for P here.

Okay, point A, these two lines are equal to each other or these values are equal to
each other. So, this ends up being, solve over here, so I'm going to move that here,
so it's going to be 16P equals to 6, then P ends up being 6 over 16 which is 3 over 8.
So, basically P is 0.375. Okay? So, that means at point A P is 0.375, okay? Which
means that we can conclude if P value is less than 0.375, the low-volume alternative
yields highest EMV. So, we'll pick the low-volume production. If the P value is
greater than 0.375, the high-volume alternative yield was the highest EMV. So, we
would choose high-volume production. At no point and no value of P, we will pick
the medium volume production.

Slide 11: Summarizing our results here. So, again, this is our graph here showing all
three alternatives. When we did this Sensitivity Analysis, our best alternative is high
volume for any P value greater than 0.375. Any P value less than 0.375, the best
alternative low volume. The medium volume alternative is never the best
alternative low value of P we'll pick the low volume production and at point A itself.
So, when P value is 0.375, we're indifferent between high-volume and low-volume
production. Final note here is, you can do a Sensitivity Analysis on basically any
parameter on any of these analysis. We just happened to do it on the property of a
strong sales, but we could have done it on other parameters. But if this was a much
bigger problem with more parameters, we could basically run a Sensitivity Analysis
on different parameters that we see appropriate.

Slide 12: That concludes this presentation. Thank you for watching.