Well, our methods assignment was due today, and the good thing is that I have never felt so good about most of my methods!!! Hopefully the feeling will stay with me throughout the rest of my research project.
After looking at various other research papers, I am particularly confident, as I was not before, with the statistical analysis that I will be performing on the comparable firm firms as well as the sample limitations that I must justify within my methods. The statistical analysis of the different comparable firm multiples, as I found out, is implemented through two closely related methods in the literature. The older method estimates price base off of a slope-intercept type equation, disallowing any y-intercept, with the slope as the comparable multiple. This method is "old" because in order to estimate the comparable multiple in the first place, it uses a mean or median estimate of the multiple based off of a company's comparable firms. This essentially means that comparable multiples for "similar" or comparable firms are calculated and used to determine the multiple for the company being analyzed. The newer method, instead of using mean or media, uses a harmonic mean estimate for the multiple, claiming that studies have shown the harmonic mean method to work more effectively. A harmonic mean is essentially an inverse mean of inverses, so essentially the entire equation for the mean is "inverted twice" to achieve a harmonic mean.
The second thing I am fairly confident about is limiting my sample within my methods for validity precautions. Because the validity of a sample is essential in data analysis (because it can potentially skew results), it is essential to provide a systematic way of limiting the sample. In the context of my research project, this means providing a systematic way for selecting comparable firms for a firm I want to analyze. Luckily, I found a good study done on biotech IPOs that showed me an efficient method of finding comparable firms. Essentially firms in the industry are divided into 9 classes based on their total assets and EBITDA, a financial variable for a firm. Then, the firm being analyzed is classified in the same manner, and once in the specific class, certain firms having the closest total assets as the firm are chosen as comparable firms. Doing so ensures that the best, most representative firms are picked each time.
One part of my methods that I am somewhat unsure about is which multiples to use within my research. There are several different multiples and narrowing these down is difficult. Although one source I was reading did provide me with a starting point, by looking towards those most commonly used in practice and academia, I still have a long ways to go. I am currently reading a source describing certain multiples based off of financial variables as well as their uses. This will help me in this part of my methods. However, because non-financial variables may also play a large role in valuing social media IPOs, I don't want to discount these and need to find a way to include these as well. I anticipate most of the coming weeks to be concerned with this issue in particular.
I'm constantly thinking about these problems, but at least thanksgiving is here so I can gorge myself with Turkey!! See you next week! (560)
Akash
After looking at various other research papers, I am particularly confident, as I was not before, with the statistical analysis that I will be performing on the comparable firm firms as well as the sample limitations that I must justify within my methods. The statistical analysis of the different comparable firm multiples, as I found out, is implemented through two closely related methods in the literature. The older method estimates price base off of a slope-intercept type equation, disallowing any y-intercept, with the slope as the comparable multiple. This method is "old" because in order to estimate the comparable multiple in the first place, it uses a mean or median estimate of the multiple based off of a company's comparable firms. This essentially means that comparable multiples for "similar" or comparable firms are calculated and used to determine the multiple for the company being analyzed. The newer method, instead of using mean or media, uses a harmonic mean estimate for the multiple, claiming that studies have shown the harmonic mean method to work more effectively. A harmonic mean is essentially an inverse mean of inverses, so essentially the entire equation for the mean is "inverted twice" to achieve a harmonic mean.
The second thing I am fairly confident about is limiting my sample within my methods for validity precautions. Because the validity of a sample is essential in data analysis (because it can potentially skew results), it is essential to provide a systematic way of limiting the sample. In the context of my research project, this means providing a systematic way for selecting comparable firms for a firm I want to analyze. Luckily, I found a good study done on biotech IPOs that showed me an efficient method of finding comparable firms. Essentially firms in the industry are divided into 9 classes based on their total assets and EBITDA, a financial variable for a firm. Then, the firm being analyzed is classified in the same manner, and once in the specific class, certain firms having the closest total assets as the firm are chosen as comparable firms. Doing so ensures that the best, most representative firms are picked each time.
One part of my methods that I am somewhat unsure about is which multiples to use within my research. There are several different multiples and narrowing these down is difficult. Although one source I was reading did provide me with a starting point, by looking towards those most commonly used in practice and academia, I still have a long ways to go. I am currently reading a source describing certain multiples based off of financial variables as well as their uses. This will help me in this part of my methods. However, because non-financial variables may also play a large role in valuing social media IPOs, I don't want to discount these and need to find a way to include these as well. I anticipate most of the coming weeks to be concerned with this issue in particular.
I'm constantly thinking about these problems, but at least thanksgiving is here so I can gorge myself with Turkey!! See you next week! (560)
Akash
Akash -- while I'm glad that you have clarity, your explanation was a little bit above my head (specifically in your second paragraph). I think that you're right, though, that the choosing multiples will happen partially through what's commonly used in the field and partially through your reading through and discerning which are most relevant to your particular research. I still, however, lack a clear picture as to what exactly you'll be doing, so it's hard for me to conceptualize how all of the articulated parts of your methods fit together. I think it'll be imperative to explain this clearly, as if you don't, it'll be impossible for a reader to judge the strength of your methods.
ReplyDeleteHey Akash! So, regarding your third paragraph, when you talk about dividing by classes, are you essentially saying that the companies are divided first by general size/industry and then divided once more so that the comparable financial details create a new class of similar companies? Just looking for a bit of clarification there. As to the part that you aren't so sure about, I think there are a few variables that are always going to be out of anyone's control, and should be addressed as limitations (as I think youll get too bogged down if you try to address them all). Some of the ones that I think you need to at least mention include: public perception of the company, the situation of the market at the given time of IPO (for example IPOs after Brexit probably did slightly worse overall than they might have at some other point). I'm just throwing ideas out there, but I'm sure there are others that you know of. Just something to think about. Don't fill up on too much turkey, and have a great break!
ReplyDelete(183)
Hey Akash! I want to start by saying that it's wonderful to see how you are gaining more and more confidence about how you will approach the methods of you research. The way you talked about the linear and harmonic regressions in detail and the comparable firm selection speaks to the extent to which you are really delving into your project! I love it! I do agree with Mrs. Haag that a lot of what you are saying, while detailed, is at times hard to follow. The biggest area I struggled with was understanding what multiples are and what they mean in regards to IPOs. Without really knowing what they are, it's difficult for me to fully understand what you aim to analyze. Also, is the EBITDA and total assets information readily available for Snapchat and other similar firms? I guess I'm trying to say, is this the most universal way of classifying and categorizing these companies? You hint at other sources that claim this to be true, but just make sure that you completely justify this section! Keep up the awesome work!! And have a wonderful break :)
ReplyDelete(189)