Multi-objective optimisation is that step in any design process which tries to make a system suitable for several objectives at the same time. This concept is applied in several branches of science like engineering, economics and logistics.
In engineering, this process of multi-objective optimisation translates into design constraints. Some common design constraints are: performance, cost, reliability and usability. The whole design problem then is about coming up with a solution that is optimal on all these counts. For example, a hardware engineer designing a chip tries to optimise it for higher speed, smaller size, a wide temperature range of operation and low costs.
Often, increasing design constraints both in terms of their number and their strictness makes the system design so complex that it becomes impossible to construct it. This is because objectives are often conflicting and trying to optimise for one leads to a degradation with respect to the other. In such a case, system design can only proceed if one objective is traded-off to some extent. In other words, for a nontrivial multi-objective optimisation problem, there does not exist a single solution that simultaneously optimises each objective.
Multi-objective optimisation is particularly applicable to government policies. Apart from the usual design constraints of equity, efficiency and costs, there are several other constraints like political feasibility and ease of implementation. Thus, designing good policies is essentially a case of multi-objective optimisation.
Now, the problem with many policies is this: governments try to optimise a policy or an agency for several objectives at the same time. Just like an engineering system design returns a null solution when strict conflicting objectives are applied at the same time, public policies trying to optimise for several objectives end up failing.
Now, the argument that this blog post makes is that the reason some government policies in India fail is because they try to do hyper multi-objective optimisation, ultimately creating a system that meets none of the objectives. Let’s consider a few cases:
The first illustrative example is that of India’s tax policy. India’s tax policy is extremely complicated, with several layers of rebates and raises across sectors, income levels and geographic areas. The reason behind this complexity is that India’s tax policy has been burdened with several objectives. And hence, it is no surprise that such a system does not function as desired. Dr. M. Govinda Rao summarises this condition best when he says:
Although many countries’ tax policy is used as an instrument to accelerate investment, encourage savings, increase exports and pursue some other objectives, Indian’s obsession is perhaps unique. In addition to the above, India’s tax policy is loaded with objectives such as industrialisation of backward regions, encouraging infrastructure ventures, promotion of small scale industries, generation of employment, encouragement to charitable activities and scientific research, and promotion of enclave-type development through Special Economic Zones (SEZs). These objectives are pursued through various exemptions, differentiation in rates and preferences which enormously complicate the tax structure and open up avenues for evasion and avoidance of tax and create rent-seeking opportunities.
The second illustration is that of National Rural Employment Guarantee Act (NREGA). It was originally meant to be a scheme to augment the income of households by providing wage employment opportunities in rural areas. However, several new objectives were subsequently added. For instance,
creating sustainable rural livelihoods through regeneration of the natural resource base, and strengthening rural governance through decentralisation and processes of transparency and accountability. Thus, far from being optimised for increasing wages, this is also seen as a process of regeneration of natural resources and for strengthening rural grassroots democracy. This hyper multi-objective optimisation thus is the bane of MGNREGA.
Third, an urban example. The traffic police system was created with the objective of upholding the rule of law on roads i.e. ensuring that the traffic rules, whatever they may be, are adhered to on roads. But this same police force is also tasked with an objective of reducing traffic congestion i.e. ensuring a smooth flow of vehicles. Often, the two objectives of faster vehicular traffic movement and upholding of traffic rules conflict with each other. The result is that neither objectives are met.
Thus, hyper multi-optimisation is a challenge for policymaking. There are broadly two responses to this challenge.
The first one is augmentation. This involves creating separate agencies or policies, each of which is optimised only for one or two objectives. This is a common response observed in India. For example, in pursuance of the objectives of promoting a rapid rise in the standard of living of the people, increasing production and controlling the direction of the economy, the Planning Commission instituted in 1950.
The second response is that of withdrawal. This involves a realisation that a few objectives just cannot be optimised efficiently by government policies. They can be best handled by the market or by the society. This would mean that policies could leave some objectives unoptimised or only marginally optimised. For example, the traffic police can return to its original duty—ensuring that traffic rules are adhered to. The objective of managing vehicle flows can be left to automated traffic signals. Beyond that, it is for individuals to assess and build consensus for reducing travel times. Similarly, given that absolute poverty is its biggest concern, the government may choose to leave the moral question of relative poverty and the pursuit of zero inequality to a future date.
The second response is definitely the tougher one. Not only does it require a projection of what policies can do, it also needs the humility to accept and explain what government policies cannot do.