The Right Tools:
Some time ago my brother-in-law did not have his buzz saw so he improvised by attaching the saw blade to a sander. The results were predictable, with a real cool scar as a bonus. Choosing the right tool is important wherever tools are used.
In the financial world the right tools are a synthesis of database, analysis, and presentation. My expertise is in Excel and it is the best tool in many circumstances. There are other spreadsheets available but Excel has the advantage of being the flagship of Microsoft and, as such, the R&D resources dedicated to keeping it at the forefront of business computing are probably unique. Spreadsheets, however, should never be the only tool.
The tools and the job, i.e. the right tool for the job:
The topic of this post is “The Right Tools” (plural) because there are times when Excel is not the right tool for the job.
The job is singular because the whole process should be counted as a single process:
Data Gathering -> Compiling -> Transcribing -> Analysis ->
Modeling -> Presentation -> Decision Making -> Implementation
The bulk of my experience is in Finance but stretches back to the pre-digital, analog, era with mud-logging. I have a fair amount of experience as an observant consumer. I have worked along the whole continuum of the “job” and my expertise is strongest in the Analysis, Modeling, and Presentation segment.
LESSON # 1: DATA INTEGRITY
Without Data Integrity there really is no point to the whole project and, for that reason, this post stops with Data Integrity.
- Accuracy (sine qua non)
Is the data accurate at every point of the process?
Has it been passed correctly from process to process?
- Consistency (identify process variations)
Are all pieces of each process equivalent (apples to apples)?
What adjustments need to be made to keep the data consistent?
- Completeness (“The Truth, The Whole Truth, Nothing but The Truth”)
Are pieces of the puzzle missing because they were not identified?
Do source changes require process adjustments?
There are, unfortunately, a multitude of examples of Data Integrity fails, some well known, some buried by skillful statisticians, some with major consequences that are never uncovered.
Within a specific enterprise, all of the above are likely, with varying degrees of failure. Two specific pain-points: (1) conversion process, especially when Data Specialists are not familiar with the company; (2) project specialists (e.g. PhD candidates) do not know their tools or the weaknesses of a given tool.
Fails (or possible fails):
(Accuracy) Reinhart Rogoff overstating differences at the margin because a spreadsheet did not pick up two countries.
(Consistency) a landmark Acemoglu and Card study on the minimum wage that has been challenged because they chose the wrong people to interview.
(Completeness) Thomas Piketty’s best selling book on income inequality lacks crucial historical data because it simply was never gathered.
My next post on the topic of tools: Data Integrity is a natural byproduct of good tools, skillful technicians and routine trouble-shooting.