Posts

The Indian Agarbatti exports case study - effect of import control

Image
What happens if we restrict cheap imports coming into the country in an employment intensive sector? While a free trade economist would squirm in agony, policymakers are faced with such choices on daily basis. Navigating multilaterally agreed rules in order to create space for domestic industry to fight and survive is a tight balancing act. I present a case study where the policymakers made a choice under similar circumstances.  Agarbatti (Incense sticks) industry in India is small in terms of turnover at around Rs 7000 crores (under USD 1 Billion). However,  it is large in terms of employment per Rupee it generates. It employs close to 400,000 people, around 80 percent of whom are women (https://bit.ly/3oyNysO). Agarbatti industry also exports a significant amount of production and maintains leadership position in the world market. The leading competitor is China.  Agarbatti manufacturing needs raw material - a premix powder popularly known as 'masala' - and bamboo sticks. Ove

Framework for Artificial Intelligence Regulation – Questionnaire (FAIR-Q)

  Introduction: Artificial Intelligence (AI) tools used in governance require a great degree of scrutiny before they are launched. This is due to the potential of AI tools to create a disproportionate impact when used to achieve a public policy goal.  To mitigate the risks, a Framework for AI Regulation – Questionnaire (FAIR-Q) is proposed, which can be used by regulators or government departments implementing these AI based solutions. Recommendation: FAIR-Q should be adopted by any department that proposes use of any AI tool for governance. FAIR-Q covers questions related to the need for AI, algorithms, processes, fairness, accountability, and ethical issues. The government department and the AI team should be able to satisfactorily answer these before the tool is allowed into public service.  The questions are designed to remain simple, yet effective. Relevance: AI tools can embed and exacerbate biases and inequalities found in the data that is used to train the tool. This raises imp