Thursday, November 30, 2006

How can I apply my new MBA skills?

December 2006 - Regular readers of my column (which by the looks of it are few and far between) may notice that I’m currently getting a masters degree in business administration from a fairly respected local university. Recently I just finished up with one of the tougher courses called “Data and Statistical Analysis for Managers.” As a side note, one would think “for Managers” in the title would make the class a bit less intensive and compared to a PHD class it might be. That being said, for the average student, which is a group I include myself in, the class is pretty challenging so beware.

While I faired alright on the homework and exams, the true question in my mind is how can I apply these new concepts to my job? This question only makes sense as my employer paid the outrageous price of tuition. Now I’m not proud of it nor tell even my dearest friends, but I’m pretty close to an expert on the format of a seven step hypothesis test. How in the world does this help me as a software engineer though? Let me try to explain how we’ve taken a simple concept from my statistics course work and tried to apply it to work.

I’m leading up a project for migrating our source code repository from a fragile, debilitating, and antiquated system to a new (hopefully robust) tool set. One of the outcomes of this effort is the migration of years of change history that is a critical part of source control management. While the migration tool we’re using is “suppose” to flag us with migration errors, there always is a chance where we could run into a migration error which never got caught (if you’re wondering, I’m a skeptic of most things). Here is a question that I often hear upper management asking us worker bees before pushing a button that could potentially bring the company and all it’s stakeholders to our knees, which this migration could easily do. “How confident are you everything is correct?” Never being an overconfident engineer, I’ve always said “we’re as confident as we can be with the information at hand.” As one can guess this never goes over very well as from what they tell me, it’s not quantifiable. The follow-up questions are always based around confidence numbers (e.g. 90% or 99.99% confident) of which I have no clue how to talk my way out of.

As luck would have it, in my difficult “for Managers” statistics class, levels of significance and confidence intervals were discussed on week four. After scouring my notes, we’re thinking we can use some of the techniques learned to “quantify” how confident we are (or are not) to impress upper management. How though you might ask? While we’ve not started the migration yet, after a “successful” migration we plan on taking a random sampling of source files in the new system and comparing the history with the original source file history in the old system. The result will give us a proportion of the file history that was fully migrated verses the total number of files in the sample. Based on the arbitrary assumption that 95% of all the history is migrated (we simply picked 95% as we’re comfortable losing 5% of history during the migration), we can in return run a proportion z-test using a specified value of alpha (i.e. .05) and figure out the probability of migrating 95% of all our history. The specified value of alpha will give us our level of confidence.

Now I have no grand elusions of impressing upper management with our results. If anything, I’m sure the question will come up as why we spent so much time on this migration and totally forget the fact that we did the migration, saved close to 95% of the history dating back from the early stone ages, and quantified our confidence. Putting upper management aside, the true outcome is that we’ve taken a very important (and difficult I might add) concept from our learning (which upper management paid for) and used it to quantify our confidence level. When upper management then asks, how confident are we that the new system will work and contains all the history we’ve created for the past 10,000 years, we can simply say “at the .05 level of significance (alpha is .05) we’re confident (or not confident) we’ve migrated 95% the history to the new system.” In addition to comforting upper management, this experiment will actually comfort us as well and since we’re ultimately responsible this is probably a good idea. So we’ll see how it turns out.

And the interesting thing is after we’re done with this experiment I’ll be on to my next class (business ethics) learning some new exciting things to annoy upper management with! Shouldn’t they be so lucky to pay me for this!

Saturday, November 04, 2006

Do stock prices matter?

November 2006 - I make no secrets about my love for the morning business section in the local fish wrap (i.e. newspaper) or my unsuccessful run at stock price speculation. My stock picking is so bad of late my colleagues and friends have found investing against me to be quite profitable. This all being said, the curious mind does know one thing for sure; no mater what the list price is, stock prices do not matter. Let me explain.

A few years back while waiting in line for an overpriced latte, I overhead a young man brag to a half listening acquaintance that a stock (of which I’ve never heard of) just split and he was taking his wife out for dinner to celebrate. Being anti-cerebral at the time, I cordially congratulated the gentlemen under my breath and wished to someday be as lucky as he apparently just was.

Five or so years after this experience, I was fortunate to struggle through my first financial accounting course and during an especially arduous homework assignment found that a stock split has no, let me repeat no, economic affect on the stock. Every company issuing stock has a market value, which is simply the total number of issued shares of stock multiplied by the price of the stock. For example, at the time of this writing Google (GOOG) has a market cap of around $143 billion and is trading at $471 a share. This means Google has roughly a little over 300,000 outstanding shares.

A student loan ridden, large mortgage owner, future family man like me has little expendable cash lying around, but if I did I could probably afford ten shares of Google. Let’s say after breaking open my piggy bank and purchasing ten shares of Google, Google decides to split its stock two for one (i.e. two shares for every one share). The result is 600,000 outstanding shares trading at $235.50 and instead of only having ten shares, this curious mind has twenty! Hurray, right?

Now I grew up directly across the street from a corn field in the middle of the upper Midwest and have never once been praised for abnormal intellect. But, I was taught (in the first grade I believe) that ten multiplied by $471 is $4,710 and oddly enough equal to twenty multiplied by $235.50. Consequently, while I may feel wealthy having doubled my stock holdings in Google, in reality I’ve gained absolutely nothing in economic terms.

Of course if you don’t take my word for it, just ask Mr. Warren Buffet whose company “Berkshire Hathaway” was trading (at the time of this writing) for around $105,000 a share. Warren (as I like to call him in person conversing over a Coke) has never split his company’s stock. While I don’t know the exact reasons why, I’m pretty sure he sees a split as needless as I and avoids dealing with superfluous details as such.

In conclusion, do stock prices matter? No, market capitalization matters and your percentage ownership of it. Stock prices are simply arbitrary numbers which can be manipulated to any value wanted.