Memorandum -Recommendation for Adopting an Advanced Method for Causal Inference

Memorandum -Recommendation for Adopting an Advanced Method for Causal Inference

To: The Supervisor

From:

CC: Other recipients

Subject: Causal Inference

Greetings

Bounded rationality creates a need to use mental models and the associated heuristics to make appropriate strategic choices relevant to an organization’s business industry. We, therefore, need to adopt a strategic group framework to study the competitive structures in our industry. We will focus on the sociocognitive processes that are interactions with other organizations, which will allow us to establish a shared understanding of the best competition strategies and the process of social imitation and comparison, enabling us to categorize rival firms into groups and copy the behavior of organizations similar to ours. We, however, need to first establish causal inference based on the fact that a better understanding of causation is important for advancing strategy research and facilitating the formulation of an effective strategy.

The casual inference will help us analyze the impact of intervention or investment, which is referred to as a treatment-effect problem whereby the treatment or intervention has a causal effect on an outcome (Geraldo Bastías & Brand, 2020). We will use causal inference to bridge the gap between decision-making and prediction because prediction models cannot be relied on to effectively reason what may happen if we alter a system or take action, even for prediction models with exceptionally high accuracy. We may also use a supervised machine learning algorithm to optimize the difference between predicted and actual values, but the prediction may not maximize the intended outcome when the action is taken. Making decisions based on a prediction model may alter the environment in ways that push the organization to untested territory hence limiting the model’s predictive power. Therefore, I recommend using instrumental variables to improve the causal inference process.

Instrumental variables analysis incorporates various steps. The first step includes observing a variable called an instrument connected with the outcome variable. The main assumption is that the instrument does not have a direct causal effect on the outcome variable. The correlation we observe between the outcome and the instrument is not because the instrument has a causal impact on the outcome but because the correlation selects the effect of a confounding variable (Baiocchi et al., 2017). The second step is making two assumptions that the instrument has a causal impact on the treatment variable and that the instrument is assigned to units randomly (Baiocchi et al., 2017). Based on these two assumptions, the causal effect of the instrument on the treatment is their relationship in the data.

The instrument variable analysis is the most appropriate strategy for our organization’s strategy formulation because it will allow us to make causal inferences using observational data. We can also make necessary adjustments for both unobserved and observed confounding effects. Other methods used to adjust confounding effects, including matching, stratification, and multiple regression, can only adjust for the observed cofounders. Instrument variables are also ideal because controlling biases does not require knowing the source of the bias (Baiocchi et al., 2017). We, however, need to ensure that the selected instruments are not weak. A weak instrument does not strongly predict exposure. Such an instrument can create various problems. One problem is increasing bias because standard instrument variable estimators have a finite sample bias that is inversely proportional to the statistics used to determine whether the included instruments make a significant contribution to the first-stage model. A weak instrument also increases any residual bias arising from a confounded instrument.

References

Baiocchi, M., Cheng, J., & Small, D. S. (2017). 2. Instrumental variables methods. Methods in Comparative Effectiveness Research, 39-106. https://doi.org/10.1201/9781315159409-3

Geraldo Bastías, P., & Brand, J. E. (2020). Causal inference. Sociology. https://doi.org/10.1093/obo/9780199756384-0240

ORDER A PLAGIARISM-FREE PAPER HERE

We’ll write everything from scratch

---

In the ninth phase of the business strategy formulation, you have determined that the organization needs to adopt a formal, advanced method for establishing causal inference.

After meeting with your supervisor, you were asked to provide a recommendation for the organization’s adoption of one of the two advanced methods—instrumental variables or panel data methods.

Write your recommendation to your supervisor in the form of a business memorandum. In your memorandum, provide at least one example of how your recommendation could be applied within the organization. Be sure to use information from our readings to support your thoughts. Chapter 8 of your textbook, Predictive Analytics for Business Strategy, by Jeff Prince

Submission Requirements:
This assignment should be at least 500 words in length.
Submit as a Microsoft Word document attachment.

**I can add in information specific from the textbook if unable to acquire**