DM: Monte Carlo Simulation and Revenue Forecasting

The next in a series of blog posts for my Decision Modeling class examining the use of the tools learned in class as applied to the outside world.  This one is one how some of the techniques learned in class might apply to my current job.

One of the potential uses of the tools learned in this class that I’ve been mulling over in my current job is the potential usefulness of Monte Carlo simulation to improve Revenue Forecasting.

But first, a little background: I’m a financial analyst for a teleconferencing and webcoferencing company.  I work on the expense-forecasting side of the equation, but the revenue forecasts sit right next to me, so I overhear a lot about their challenges in developing an accurate forecast.  There are a couple different ways to categorize our largest revenue sources.  First there is the difference between “attended” and “unattended” calls. “Attended” calls require an operator to be on the line during the conference call to provide assistance and help handle Q & As.  A typical example of this kind of call would be an Investor Relations call, in which the a client company’s CFO, CEO, or other executives present information to stock analysts.  “Unattended” calls don’t require an operator, and tend to be shorter calls.  Another method of dividing the revenue up is based on the size of the customer: Small and Medium Businesses (SMBs) are more numerous, but each individually generates a relatively low volume of traffic.  But much larger corporate accounts can each generate a very large volume of traffic.  There is a lot of empirical, historical data on all of these factors that we could use to build a more accurate revenue forecasting model.

There are several uncertainties involved in forecasting revenues from these sources.  The first is the total number of SMB clients currently using our services.  On a monthly basis we may cycle through SMB clients as the sales team signs new clients up and old clients fall off or switch to competing services.  The second is the total number of minutes each SMB client might be expected to use in a month.  Without having looked at the data, I can only guess that both of these will likely form some sort of normal distribution.  But other analyses would need to be done on the empirical data to look for trends over time and other clues that might help us refine our model.  Finally, there seems to be some variability in the rates that SMBs may be charged for these services.

The second set of uncertainties concern the large corporate client.  In any given month, there is a risk that we may lose any of our corporate clients to a competitor.  The exact risk might be uncertain, but some idea of that risk might be gleaned from examining the historical data to get an idea of the turnover.  These are usually rare because long-term relationships with large clients are bound by contracts that may have penalties for breaking the contract.  The next uncertainty concerns the probability of whether we will “win” new large corporate clients.  Usually our revenue forecasting team is apprised of current deals in the hopper, but often are given unrealistic assessments of the probability of a deal being signed, with the sales group often reporting “We are definitely getting this sale” only to find, all too often, that we will not be getting the sale after the forecast including that sale has already been finalized.  The final uncertainty with large clients is the volume they will generate.  Usually, these contracts include a minimum obligation of minutes to be used by the clients, but large clients don’t always meet these minimum monthly usage obligations.  Sometimes, they also may greatly exceed these minimum obligations.

With all of these uncertainties playing a role in our revenue forecasts, it starts to seem absurd that we are using point estimates.  Clearly, a Monte Carlo simulation would be the superior method of providing useful data for Revenue Forecasts.  (Somewhat less so for most of the cost forecasting that I do.  Some variable costs are driven by our traffic volume, but the Revenue Forecasting team handles these costs.  Most of the rest of the Sales, General, and Administrative costs are theoretically controllable, though experience has shown this not to really be the case either.)

But then I started to consider what value this would really add, based on how this superior forecast might influence decision making.  To some degree, a better long-term revenue forecast might improve long-term capacity planning, but generally our forecasts are focused on short-term fluctuations and short-term accuracy more than for the long term.  Which lead me to wonder what decisions, if any, our forecasts are being used to inform.  Other investment and R&D decisions are primarily made based on the CEO’s “vision”, and these forecasts seem to have very little to do with that.  In fact, the most prominent use of our regular monthly forecasts that I have seen are two: monthly departmental meetings with the Finance group in which the CFO or Finance Directors rail and bemoan the lack of control various departments have on their finances, and point to dismal revenue forecasts as a sign that these departments need to bring costs under control, and our own investor relations calls, in which the executives attempt the opposite spin on the numbers to try to talk up our long-term profitability to stock analysts.  In that case, the decision that is being driven being driven in part by our forecasts, ultimately, is whether stock analysts will recommend to investors either to buy, sell, or hold our stock.

Given the uncertainties inherent in what value can be gained from a more textured and nuanced revenue forecast (or expense forecast, for that matter) that is more accurate than a simple point estimate, it would be hard to recommend to the company that they invest in purchasing licenses for a risk analysis module such as @Risk.  It then becomes clear that what the company (particularly the finance department) really needs to do is invest some time and effort in analyzing what are the objectives of our regular monthly forecast, what we will be trying to achieve with them, and why – and to communicate those objectives clearly throughout the company and especially throughout the Finance department.  Once that quality discussion has been had, the question of the accuracy of those forecasts can be re-addressed.


DM: One of These Days… To the MOON!

Before the end of the semester, I have to put up several entries on my Decision Modeling professor’s class blog that relate to the topic of the class: instances of the use of (or the need for the use of) decision modeling that we see in the world around us, in our own lives, or in the news.  Here is the next entry in that series:  
To the MOON!  Or to MARS!  Or an Asteroid!  Or maybe just the International Space Station, at least until it falls out of the sky! 
This morning [or yesterday morning, by the time my readers get to see this] I was reading about President Obama’s trip to Florida today, and his plan to announce a change of focus for NASA’s mission.  It’s a plan that’s been widely criticized by astronauts old and new, as well as by Space Enthusiasts (such as myself, I being a nerd; I don’t mean to be political in this blog, and have only done so once before, but Space Exploration is a sufficiently nerdy topic to warrant my posting something about it; besides which, this is meant to be a semi-humorous exploration of the topic).  Comedian and fellow space enthusiast Stephen Colbert summarized the criticism in a humorously witty remark as he commented on the shift in focus away from a new manned mission to the Moon or Mars and toward more robotic exploration missions:

Sending robots into space does not win glory for Americans.  It wins glory for Roombas.

In reading these two linked articles, it made me think about what are the objectives of the US Space Mission, and how those objectives, in retrospect, ought to have been considered before making a decision this significant.

First, we can say that deciding what NASA’s mission will be is hardly a trivial decision.  The implications for how NASA spends its time are immense, for the future of the country.  For one, the jobs of thousands of space administration employees hangs on the line of this decision: not a trivial consideration at all.  Secondly, it is not at all obvious what the right decision should be.  There are numerous possible options that can be selected in this decision, and the implications of each of those are different.

So, in making this decision, the administration would first have needed to consider what our fundamental objective for NASA will be.  Space enthusiasts might have said that our fundamental objective would be to launch a successful manned mission to Mars, enabling an American to be the first human to plant foot on the soil of the Red Planet.  But would that have really been our fundamental objective?  Whether intentionally or unintentionally, Colbert’s joke makes a point about the nature of the objective of manned space flight: is the success of a manned space flight mission the real objective, or is there something else to be gained by the success of that mission, such that the manned space flight mission is really just a means objective?  Colbert seems to suggest that the real fundamental objective is to “win glory for Americans”.  If that is the real fundamental objective, then, is a manned mission to Mars, or the Moon, or to somewhere else, really the best means objective?  Some astronauts contend that returning to the Moon would be a step down in glory, because we’ve already achieved that (though we haven’t been back since before I was born).  The only viable means to win glory, they contend, is to put a human on Mars. 

But is winning glory really our fundamental objective?

As a matter of fact, there are a very large number of objectives for which NASA’s human spaceflight program can play a role as an important means objective.  Previous human spaceflight programs in this country have lead to an incredible leap in technological development, much of which filters down to the average American in the form of an improvement in their standard of living.  Manned space exploration has also been a source of inspiration for America’s school students, providing the kind of motivation to improve their math and science education that few other outside motivators can – and it is a stated goal of the Obama administration to improve Math & Science education in the U.S.  And the push to develop the technologies needed for manned space flight has helped to keep America at the cutting edge of technological development, all of which improves the American economy.

So, ultimately, the manned spaceflight program has the true fundamental objective of improving economic development in the U.S.  It can be viewed as an investments that pays economic dividends in a variety of ways.  Having determined that, the next step would be to seriously consider how the different potential missions for NASA influence those goals.  Which was summed up well in one quote from the first article, in which the commander of the Apollo 13 mission, Jim Lovell, said:

The whole idea of any program is you have to set a goal.  You don’t just build technology and figure out what to do with it.

Obviously I don’t have access to all the data.  But I do worry that a loss in focus on manned spaceflight, and lacking a clear program objective (like “landing a human on mars”), NASA’s value to the country, in terms of the potential economic development resultant from spaceflight as well as in terms of national pride, will begin to wane.  I just don’t believe that’s a frontier we really want to cede.

DM: Choosing an Infant Car Seat

Before the end of the semester, I have to put up several entries on my Decision Modeling professor’s class blog that relate to the topic of the class: instances of the use of (or the need for the use of) decision modeling that we see in the world around us, in our own lives, or in the news.  Here is the first of my entries:


My wife and I had decided some time ago on a car seat for an impending “bundle of joy” due to arrive in about month’s time: the Graco Snugride 35 “Racer”. 

Graco Snugride 35 "Racer"

Graco Snugride 35 "Racer"

In retrospect, reaching this decision was the kind that can benefit from some values-based decision analysis.  There are a variety of options to choose from, and the right decision is not obvious.  In addition, the consequences of making a poor decision are potentially very material: there is a risk to the life and health of our baby, which is not a risk we can take.

Our criteria on this decision were pretty simple.  We were concerned with Safety first, the comfort of the baby, and ease of use.  Price as an issue, but it was not the dominating decision criteria.  On the measures of safety, comfort, and ease of use, we understood the Graco Snugride line of products to be the superior choice.  While a little more expensive than other brands, we felt that the value on the earlier criteria made the snugride the better option.  We decided next to get either the “32” or the “35” models, instead of the regular models, which can hold older and heavier babies.  Our reasoning on this decision was that we could make the car seat last longer before we had to buy a larger “toddler” seat.  The last criterion was appearance.  On this criterion, our primary interest was matching the red stroller we had already received as a gift (which is able to hold the car seat using a special adapter).  In looking at the various designs of size 32 and 35 models, we determined that the “Racer” design matched our stroller closest. 

This weekend, though, we were forced to revisit this decision. 

Graco Snugride "Lotus"

Graco Snugride "Lotus"


Our goal this weekend was to install the infant car seat.  My wife will be considered “full term” this week, and we wanted to be ready.  However, when we went out to the car to start the installation process, we found that we couldn’t install the car seat in the middle seat, as we had assumed we could – it would have to go on one of the sides, either behind the driver or behind the passenger, because of the way these car seats have to be installed.  The problem with that, however, was that the car seat wouldn’t fit in either spot unless either the driver’s seat or the passenger’s seat were pushed all the way forward. 

We realized quickly that it would be exceedingly difficult for either of us to drive with the driver’s seat pushed all the way forward.  The alternative, installing the car seat behind the passenger, gave us pause.  Our understanding was that passenger-side impacts are among the most common types of accidents.  This put us back in square-one, again: considering our first priority of safety.  We knew that the regular Snugride was several inches smaller than the “35” model Snugride, and it was our belief that we could fit this smaller model in either position 

So, in reconsidering our earlier decision, we now had to decide how much this new risk weighed against our other decision criteria.  The right choice in this case wasn’t immediately obvious.  However, what helped us make our decision was considering the possibility of a second child in the future, and what would happen if we had one child in a toddler seat and another child in a baby seat: would they both fit?  Because cost was still an issue, we realized we would not want to purchase a new, smaller baby seat at that time.  Because of that, we decided that it would be best to return our larger baby seat and purchase a new, smaller car seat. 

Reconsidering our earlier criteria, once we had satisfied our safety need, we stopped to think about our other criteria.  The ease-of-use would be the same from one Snugride model to another, so that would not be an issue.  We weren’t sure whether the baby’s comfort would be an issue on this move; because we were looking at our options online, we couldn’t compare the quality of the materials, or gauge whether one material would be more or less uncomfortable for the baby.  That left color, which lead us to consider the “Lotus” design.  The only problem with this model was that it was not available at a local store.  However, that minor inconvenience was not high on our decision criteria.  In the end, we did decide to order the “Lotus” model and return the “Racer”.

Decision Modeling & Corporate Responsibility

One theme from my last Decision Modeling class was the question of corporate responsibility.  Our case for the week was on one company’s attempt to organize a volunteer effort within the company to do service in the local communities where the company operated.  It was an unusual case, and the professor knew it.  He clearly wanted us to learn to apply decision analysis tools in situations without an obvious quantitative focus.  But it soon became obvious that he wanted us to be thinking about the role of corporate responsibility in general.

There are, he asserted, generally two schools of thought on this subject.  The first holds that the primary, and only, responsibility of a corporation is to its shareholders.  This school regards anything that detracts from that sole responsibility as detrimental to the shareholders.  Thus, corporate giving and philanthropy are stealing from the shareholders.  The second school believes that corporations are part of communities and affect the communities around them.  Insofar as communities are external stakeholders to the corporation, therefore, the corporation has a responsibility for the wellbeing of the community.  Both of these points of view, the professor believes, are ill-conceived.  The reasoning behind this assertion is that he does not believe that members of either “school” have thoroughly thought out their positions.  He wants us to think and analyze the pros and cons before ascribing ourselves to either school.

So, here’s where I stand on the subject.  It’s all well and good to say that the corporation’s responsibility is to its shareholders.  In a strictly legal sense, that’s true.  But in a market where shareholders have few if any actual rights – in which shareholders do not hire and fire their managers directly, but operate through the proxies of “Boards of Directors” the Chairpersons of which regularly anoint themselves as “CEO” and approve their own irrationally high compensation packages and perks based on the unstudied and unproven justification that these lavish salaries are necessary to “retain top talent” – this is an argument that is at best specious and at worst a crass appeal to corporate greed.  It’s also so narrowly focused on short-term profitability that it ignores the larger, long-term implications of corporate social responsibility.

The fact is, over the long-term, corporate philanthropy and social responsibility can have a positive impact on shareholder wealth.  The degree to which this has been studied (and to which we might be able to site empirical evidence) I do not know, but the logic is sound.  As much as economists and their ilk prefer to believe that human beings (including potential future customers) are purely “rational” beings that make decisions solely based on expected monetary value, this economic view is simply contrary to reality.  One of the recurring themes in our Decision Modeling class is that people value things to which we cannot easily apply a monetary value.  It is for this reason that we learned the concept of “utility”.  In fact, economists, as a general rule, have largely fallen into the trap of using expected monetary value as a shorthand for utility, where utility itself is a convenient but arbitrary measure of an individual’s values. 

In this vein, it should come as no surprise that many (even if not all, I would still assert that a majority) people place a great deal of value in certain non-monetary factors.  In fact, this is the very basis of business strategy: were it not so, the only viable strategy in any situation would be a low-cost, high-volume strategy.  But no; people value the intangibles, and one such intangible is the level of corporate and social responsibility.  Witness, for example, the rise of the green movement.  A large enough portion of the population values the protection of the environment that they are willing to pay more for a product that can be deemed environmentally safe and constitute a viable and exploitable market segment.  Companies are making a profit selling “green” products to these people.

Having thus done away with the myth that corporate or social responsibility somehow drains value away from shareholders, the question then becomes not whether corporations should act in a socially responsible way, but how they should act, and how much should they focus their attentions on this matter.  The answer will vary from corporation to corporation.  There are ways to go about investing in communities and social responsibility that add to shareholder value significantly over the long term; there are ways to do it that are neutral to shareholder wealth; and, yes, there are ways that would decrease shareholder value.   Corporations should be considerate in how they pursue these activities to ensure they are acting in ways that benefit both the community and their shareholders.  For example, a technology company might find that investing in education is the best way for them, because an educated population living in a developed economy are the sorts of people who will make the best employees and provide for a customer base with more needs for the company’s technology solutions.  This is the kind of values-based thinking that the professor of our Decision Modeling class is advocating, I believe.

Projects Out the Ears

My apologies for the shortness of today’s post.  I’ve got projects coming out of my ears over here, and I anticipate that will slowly eat up all of my available free time over the coming weeks.  In Project Management we’ve got a project studying the application of Project Management techniques throughout the classes in our MBA program.  My role will be to try to gather some data to benchmark our findings against other top-rated MBA programs.  In Decision Modeling the project will be to try to develop some decision-support in the form of some kind of model to help MBA students interested in a career change decide what career path to pursue.  That latter project was partly inspired by my own difficulty figuring out what to do with my MBA, career-wise.  My reasoning was that it’s a non-trivial decision, in that the decision will have far-reaching consequences both for me and for my family.  Additionally, it is decision where the right choice isn’t obvious: there are several options, and it’s not clear how they each stack up on my personal preferences.

In addition to these projects, I may be taking on another project, unrelated to class, as a kind of mini-internship.  Basically, if I get picked, I’ll be doing a small project in a field outside my own for another company besides where I work.  My personal goal will be to gain resume-worthy experience.

So, with that stuff going on, I may be making my posts in the coming weeks a little more light-weight.  With luck, the updates here will still be interesting, well-written nuggets.

What probably will suffer is progress on the revision of the short story I’m working on (a progress bar for which you’ll see in the sidebar).  I’m about a third of the way through the short story, revising as I go.  I’ve had to add a little to the story – moreso than what I’ve taken out.  But with projects and things going on, I really need to focus the next few weeks on my classes.

I can set aside work on the short story primarily because I’d already set a deadline out past the end of the semester: June 15.  Since the May period Accelerated Course options this year are all Study Abroad, which I can’t do right now, that means I’ll basically be taking May off.  So once this semester ends, I may have a little more free time to work on the story, finish my revisions, and get it sent off to the first market.  I do say may.  Because as I’ve been sharing throughout my blog, I’m soon to be a father, and sometime between now and then there’s a very good chance that baby B.T. will have joined us.  Once that happens, all bets are off, because I’ll have a new project coming out my ears: helping Dear Wife care for the wee bairn.

A Short Note On Modeling Class

Decision Modeling, that is, not that other kind of modeling (I don’t have the appropriate assets for that other kind of modeling).  Class last week had a very interesting lesson embedded in it, and we’ll be learning more this week about that, I think.  It’s was about weighted averages, and the hidden logical fallacy buried within them.  It’s this: when we apply a “weight” to a weighted average in which we are assigning relative importance to various attributes of a decision we are trying to measure, we are implicitly assigning a constant exchange rate between the various attributes.  The weight basically means this: I would trade this much of attribute X in order to gain this much of attribute Y.

Most people, the professor contended, don’t think that deeply about what the weights they are using mean.  Further, he suggested, most of us don’t really have a constant trade-off that we’re willing to make.  The more we have of something, often, the more of it we’re willing to trade to get something we have less of.  As the relative difference between the two diminishes, so to does the amount we’re willing to trade.  Weighted averages don’t take this fact into consideration.  Whether there’s an answer to this problem-solving conundrum, I don’t yet know.

For this week, there’s a chance I may be late or non-existant with a few of the normal daily updates.  I haven’t missed a day in over two months, but there are a number of deadlines and other higher-priority tasks hitting this week that will require my more immediate attention than this blog.  So, if I miss a day or two this week, my sincerest apologies, in advance.  It is only because it is unavoidable, if so.

DM: How Much for that Risky Bet in the Window?

Now that Spring Break is well and truly done, it’s time for another “What Have I Been Learning In School” update.

The big news from Thursday’s Decision Modeling class is actually what happened before class.  Before we went on our merry way to Spring Break (mine consisted of being at work every day except on weekends, and spending most nights studying) we got a take-home “Quiz” in our Decision Modeling class. 

I’ve loved this class so far, but here I have to take issue with the professor’s use of terminology.  “Quiz” implies a short, easily completed review.  This particular “Quiz”, which was a take-home, happened to roughly coincide with the mid-term period, and happened also to be not short nor easily completed.  Most of it was not overly complicated or difficult, and was a good review of the course material so far, but there was one problem that was particularly vexing.  I’ll get back to that in a moment.

At any rate, I completed this quiz one week prior to its due date, on the day of class Thursday last week.  Then, on Wednesday night, I e-mailed myself all of the various files, Excel spreadsheets, and documents that had answers and supporting material for the questions, so that I could print them and turn them in.  As I opened the files up, with only a few hours remaining (and an hour of that guaranteed to be eaten by my commute to class) I discovered that one of the files did not have the completed answer I had just finished the week before.  And the file in question was the answer for, you guessed it, that one particularly vexing problem.  I fairly quickly realized what happened (and confirmed it after investigating).  I’d worked on the problem at work, during lunch break and a moment of down-time.  As I often do when I am forced to work on things on multiple computers, I e-mailed a copy of the file to myself.  At home, I opened the file from e-mail, finished the job, and saved it down.

Except: Microsoft’s Windows has the terrible penchant for saving files opened from e-mail into a Temporary directory.  Which, of course, makes no sense on this g0d-given-green-earth!  Because if I hit SAVE, I think that is a pretty clear indication that I intend to retrieve what I have saved at a later date.  If it ends up in a temporary directory, and is therefore irretrievable at a future date, well – that pretty much defeats the purpose of my hitting the Save button, doesn’t it?

So, that’s what happened, and in a mad rush I tried to recreate the work I  had done.  Luckily, I had a slightly more recent version of the relevant file on my work computer than what was last really-and-truly-not-just-temporarily-saved on my home computer.  So I’d only lost about an hour or so of work.  I was able to finish the job, get everything printed, and submitted on time.

Once the mid-term was in, class was mostly a review of certain concepts that serve as an introduction to where we’re going next.  We were talking about risk analysis in our modeling and how EMV (expected monetary value) is a fine tool to use if the size of a risky proposition is small relative to our resources.  He calls this the “Billionaire’s Perspective”.  A Billionaire doesn’t really care much if an investment has a 50/50 chance of either making  him $200,000 or costing him $100,000.  Over the long term, making investments like these will average out and earn him $50,000 each.  That’s EMV.  But for the rest of us, it’s not possible to look at things that way.  We have a very low “risk tolerance” because, sure, we’d like to make back $200,000 but we can’t afford the risk of losing $100,000 because we don’t have $100,000 to lose.  So, our models should take our risk tolerance into account.  Doing this involves transforming the values we measure into an “Expected Utility” instead of “Expected Monetary Value”, and then calculating back to our “Certainty Equivalent”, or the certain amount we would value our risk at.

This, he explained, is basically how insurance works: we have a risk that, if whatever it is happens, we can’t afford it.  So, we actually pay some certain amount to an entity that has a Billionaire’s Perspective in order to rid ourselves of the risk.  The insuring entity is able to pool a lot of risk together to get the long-term EMV advantages.  (I could very easily now wax political about this topic but this isn’t the place for that.  If you happened to be really interested in that topic, well, I am willing to share, just not here.)

So, that’s all well and good.  Next class we’re going to learn more about how to figure out the right Risk Tolerance to use in our modeling.

Other than that, nothing interesting to share about writing, the process or writing, or life in general at the moment.  But, as boring as that probably was for some people, for today that’s enough.  See you here same time tomorrow.