Replication and Adaptation of Model For Creation of Multi-Factor Economic Indexes

I need a Python and/or R developer to implement in Python/R a full working replication of the Aruba-Diebold-Scotti (ADS) Business Conditions Index, with some small modifications/customizations. A complete description of the index, its component data and the methodology used to create the index can be found here: [login to view URL] Under the “Resources” heading at the bottom of this page, there are two papers which fully describe the methodology. Essentially, the paper describes an index created with a Kalman filtering technique (a dynamic multi-factor model) that uses inputs with variable frequencies: daily, weekly, monthly and quarterly. You will be provided with the complete (extensively annotated) RATS code that the program is currently implemented in. Furthermore, you will be provided with all necessary input data and the complete output data generated by the model so that you can cross-check that the output has been correctly replicated. The model that you deliver in Python or R should be able to easily create new custom indexes (based on the ADS dynamic multi-factor Kalman filter model) which can be created with an unlimited number of different variable inputs at daily, weekly, monthly and quarterly frequencies. Furthermore, the input interface should be set so that user can choose whether the input enters the model as a raw numeric value or as a growth rate. The model delivered should be a fully working tool that creates indexes based on a dynamic multifactor model Kalman filter which is able to: A) Quickly process historical input data to generate a historical time series; B) Produce a continually updated nowcast with mixed frequency inputs from real-time sources – FRED and my database. Model outputs and deliverables: 1. All outputs currently published on the ADS links provided. 2. A standard index with base 100 in a given year. (Currently, the index oscillates around it’s mean growth rate. The additional output variant I am requesting is a standard continuous index base 100). 3. Bonus: An additional output which the freelancer can provide at his discretion, if it is not too difficult: A) A running tally of the forecast for each variable in the model. B) The impact of each piece of news on the overall index and the updated forecast of each variable in the index. This output is similar to the NY Fed Nowcast model that I have discussed with the freelancer. The freelancer should provide also provide the following: 1. Code should be well annotated so we can understand exactly how the model was constructed and how this relates to the RATS model that was replicated. 2. A separate written document describing important details and key features necessary and helpful for implementation and flexible use of the model. 3. Written instructions on how to implement the model, (how to query the inputs, how to retrieve the outputs, etc.) 4. Written answers to any questions we may have as we go through implementation and testing.