|
Sometimes it is best to post examples of the usage of tools to exhibit knowledge of both the tools and the understanding of the topics. At the right are some examples of analysis and usage of multiple regression tools to conduct the analysis and market research.
|
|
|
Risk
written by : William F Bryant
|
Analysis
written by : William F Bryant When most people think of analysis they probably associate it with charts and data plots or perhaps raw data dumped into applications; these applications spitting out an answer in regards to inventory level or ordering amounts or some other assist to a decision evaluation process. If you have ever worked with data or had someone explain the possibilities for predictive power in data analysis then you understand why companies that hold data can command such a high price for access to their databases.
Data can streamline capital management, operation's raw materials and labor usage, credit terms, project valuations, examine levels of advertising and promotional expenditures and even identify target markets to maximize sales potential. Some of this may seem fantastical, but there is significant company benefit to understanding data. Some data you will keep on hand to assist in tracking operations and profitability, your financial statements. Some data you may get from your customers through rewards programs and company specific applications. And, of course, you can always seek out data from external sources to examine the economy, your industry, your competitors and additional markets. In these times, everyone collects data on nearly everything and hold databases of information. The true power of analysis comes from knowing what questions need to be asked, what data needs to be gathered, how to gather it and what tools or methods will be most effective to analyze this data to get an answer. This analysis can, in fact, be active or passive depending on the complexity of the former thanks to patterns in data. Active analysis, is just that. The active compiling or mining of data to actively process it with the hopes of determining some conclusive answers. This is most often done when planning new strategies . Passive analysis can be done at a glance. A product line manager can glance at an accounting report and see if his or her numbers are trending at par or if there are issues. Passive analysis is often done without recognizing that it is a detective risk mitigation procedure. Passive analysis is simply, "part of the job", but passive analysis doesn't stop at the numbers. If someone hears that a machine isn't functioning to tune or a delivery truck is leaning, these are all passive analysis techniques that recognize there is an issue. Education and training of employees on properly functioning operations is, itself another method of analysis that carries with it benefits of both risk mitigation and, hopefully, more capable employees. Of course, this passive training, extends to managing staff, as mentioned, in addition to the recognition of patterns in their data. A passive recognition of an issue in the data, can be an effective risk alert before things get out of hand and become costly. Many companies, actually, employ 'sensors' to track acceptable parameters for data in operations that have tight spec controls, but this same premise can be extended to any operation through the training of employees. The employees may not have the training, nor the time and responsibilities, to conduct an active analysis, but at least you can now be assured that there will be a positive benefit to the accompanying cost. But active analysis is not the end of the analysis, nor is it a final definitive answer. When drilling down into the numbers and conducting active analysis, understanding assumptions to your chosen methods is necessary to properly assess and analyze data using the chosen methods is essential to get the most reliable answer. The biggest issue people often have in analysis is not understanding that many techniques come from statistical principles that may cause the analysis to fall apart or reveal an unreliable answer. Like most things, it is important to consider what is "good enough", or put another way, when does the marginal cost exceed the marginal benefit. I have always found that understanding the basis, (theoretical background) for any analytical tool chosen, lends to a much more thorough understanding of the analysis and defines when the marginal benefit has been exceeded. This depth of knowledge helps choose the correct way to formulate the question, the data needed and how to gather the data. What this all comes down to is better information for you, the business owner, at the end of an analysis. Following is some sample analysis using different techniques. *Econometric Analysis (log-log) on the relationship of the change of a selected beverage stock's daily returns regressed on the change in national consumer revolving and nonrevolving credit from credit union and depository institutions. (Time-Series, Lagged, Log-Difference, Multiple Regression Model) *Dow Study - Optimized Portfolio Pre & Post Housing Crisis Time-Series data must be evaluated for stationarity. The following analysis displays why this is important. Through multiple methods, the significant difference the economic environment made to the risk and returns before and after the housing crisis and displays the optimized portfolio difference through the frontier model. I would suggest that two entirely different markets existed pre and post crisis. |