This blog series explores the stages of the Model Development Lifecycle (MDLC) so that you understand the components of model development and maintenance as well as where potential sources of model risk can reside. Read the first post here for an introduction to the series and why we are likening this process to the creation of a cookie recipe.
In this blog about the MDLC the focus is on: Defining Business and Model Objectives.
In this post, and with others in the series, we will first look at an example of how one baker goes through the steps of this process to achieve business success, and then tie her story into the larger Model Development Lifecycle.
Cookie’s Story: A New Recipe
Cookie N. Lac, the founder and CEO of Cookie’s Bakery, stared out her window one Saturday morning. She was thinking of chemistry and supply chains.
Baking, unlike cooking, needed to be exact. At its heart was chemistry. Deviations in the amounts could be ruinous. An extra half-cup of flour in a cookie recipe would result in a product that could be used as a doorstopper. An extra half-cup of broccoli in a casserole recipe would just make an already unappetizing dish just … well, more unappetizing.
And supply chains were how the company got its inventory. COVID showed how fragile those could be. With more frequent and devastating natural disasters occurring in the US, the bakeries the company had scattered across the country could be cut off from vital supplies. Perishable ingredients—eggs, butter, milk— were particularly at risk. The bakeries used those ingredients quickly. The fresher the ingredients, the better the cookie.
Cookie turned from the window and sat down at her desk. Across from her was Schopenhauer’s Law of Entropy, framed in mahogany. It was about how a teaspoon of wine in sewage and a teaspoon of sewage in wine both resulted in sewage. Entropy, sure, but she used it as a reminder that all her ingredients needed to be of the highest quality to ensure a high-quality product.
If devastating floods, wildfires, or another pandemic occurred, there could be delays in getting the perishable ingredients her bakeries needed. That would result in a failure to produce a quality product. If her bakeries couldn’t deliver on that, the company would lose market share and, if it got really bad, then…well, she might be baking out of a tree with a bunch of elves.
Cookie smiled. Hold up. She wasn’t just tough, but she was also smart. You didn’t build a multi-million-dollar business if you couldn’t innovate.
She drummed her fingers on the desk. Her bakeries needed to be able to pivot during a supply chain disruption. Was it possible to create a recipe that used some lower-quality ingredients but somehow maintained her high-quality standards? Schopenhauer’s law had driven her for so long that she considered her current line of thinking almost disloyal. But having a backup plan was crucial for her company’s success.
Cookie let out a long sigh. Technically, if she could pull this recipe off, it would make business sense to use it all the time. Cost savings on ingredients. However, that bothered her. Doing that did feel disloyal—to herself and to her customers. No, she would only have the bakeries use this recipe during supply chain disruptions.
She tapped her knuckles on her desk and stood. Now, time for some R&D. Cookie headed to the entrance of her test kitchen.
Relating it to the MDLC
The first two stages of the MDLC are about objectives. There is the business objective: what is the problem that needs to be addressed? And there is the model objective: how to reach the desired business outcome?
Why are the first two steps in the MDLC about objectives?
Clear objectives ensure the business need will be addressed and provide an expectation on what the model should do.
Cookie’s objectives arose because she identified a problem. She needed to ensure the high quality of her cookies remained in the event of a supply chain disruption (i.e., business objective). To do this, she needed to create a recipe (i.e., model) that allowed for different grades of ingredients while still producing a high-quality cookie (i.e., model objective).
Jonathan Leonardelli, FRM, Director of Business Analytics for the FRG, leads the group responsible for business analytics, statistical modeling and machine learning development, documentation, and training. He has more than 20 years’ experience in the area of financial risk.