# Computational Public Policy: Proposal Overview Prepared for James Savage by the [QuantEcon](https://quantecon.org) team. December 2021 ## Introduction **Phase one** of this project will develop course material for a world-leading graduate program in computational public policy and computational development. Phase one of the project will also provide training in computational methods, while assessing capacity at a broad range of policy institutions, building connections and identifying highly talented individuals, with the primary aim of facilitating phase two of the project. **Phase two** of this project will found a center, institute or university focused on analyzing, developing and teaching public policy from a computational and cross-disciplinary perspective, combining economic theory, computer science, applied mathematics, statistics, machine learning and data science. The institute will attract and facilitate collaboration between the most outstanding academics, business and policy experts in these fields. ## Motivation Policy makers routinely face highly complex challenges, such as planning national economic infrastructure, managing the impact of climate change and stabilizing economic and financial environments. We believe that these challenges are best faced by applying data, computational power, and high-level technical expertise from a broad range of disciplines. Graduate level courses are not always well aligned with this growing need for cross-disciplinary training. One reason is that academic incentives are skewed towards publication, which pushes research and training towards narrow niches. In contrast, the courses and programs developed through this project will be strongly collaborative and cross-disciplinary, based on open scientific reasoning and founded in a common language of computation, mathematics and statistics. ## Case Study This case study underlines the motivation for the project provided in the previous section. The Chilean Central Bank (CBC) has contacted QuantEcon for assistance with macroeconomic modeling. One motivation was that standard tools for macroeconomic modeling are typically created in stable, high-income countries and are solved using linearization around the steady state. CBC views this modeling framework as incompatible with the Chilean experience, which has included periods of intense financial instability. The figure below helps to illustrate this difference in experiences by comparing inflation rates for Chile and the US since 1971. While inflation in the US was regarded as high and volatile in the 1970s, these numbers are dwarfed by Chilean inflation over the same period. ![](https://i.imgur.com/ZPwB3bz.png) CBC regards financial crises as inherently nonlinear in nature, and wishes to develop their capacity for modeling such phenomenda. Realistic nonlinear modeling of macroeconomic dynamics is more computationally demanding than standard, linearized solutions, requiring powerful hardware and high-level skills in scientific computing. The CBC recognizes the need to build modeling capacity that combines cutting edge technical expertise in software development, high-performance computing, mathematical modeling, statistics and economics. As part of this process, the CBC has independently made the decision to shift towards high quality open source scientific computing tools based around Python and Julia. It now seeks international expertise to build its capabilities. Our team is in the process of preparing specialized training. ## Phase 1 Phase one will implement 18 months of intensive remote and local training. The purpose of phase one is twofold. One objective is to transmit modern scientific computing skills to policy professionals. The second is to prepare for phase two by listening, learning and seeking feedback, while writing course material, building contacts and identifying talent. Key questions include * What are the interesting / hard / common problems? * What are important and representative case studies? * What skills are needed? * What capacities currently exist on the ground? ### Workshop Locations * Central banks * Treasuries, departments working on industrial policy, trade, natural resources, data collection * International agencies: World Bank, ADB, IMF ### Conference Phase one will also include a gathering of policy professionals, with the objective of sharing expertise in computational public policy and identifying common problems and constraints. ## Phase 2 Phase 2 is not yet planned in detail. Some notes are collected here. ### Prior Examples * UPF in Barcelona -- Tom Sargent has already discussed this with co-founder [Ramon Marimon](https://www.ramonmarimon.eu/) * [di Tella University](https://en.wikipedia.org/wiki/Torcuato_di_Tella_University) ### Contacts * Danny Quah, dean of Lee Kuan Yew School of Public Policy, NUS, is a friend and student of Tom's ## Scope See the separate file https://hackmd.io/yY66K7bkS5G5-Ts4tPA2tw ## Budget (Phase One) The proposed budget is available [here](https://docs.google.com/spreadsheets/d/1j6uF5TtF2ebAo4feIMx_G0fLlKSOjagc/edit#gid=1961790218). ## Why QuantEcon? The QuantEcon team is ideal for implementing Phase 1 and preparing for Phase 2, due to - world-leading expertise in economics, mathematical modeling, statistics and high performance computing. - existing public goods including extensive [lecture series](https://quantecon.org/lectures/) and code libraries - deep connections with policy professionals - very broad connections with experts in academia - good will from extensive provision of public goods - track record (e.g., FRBNY) ## Possible Endorsements This section gives examples of people and institutions we could request endorsements from, should they be necessary. * Marco at the FRBNY * Central Bank of Chile * Treasury / RBA / Dept of Industry in Australia?