## Thursday, December 08, 2016

### Designing the payout phase of the National Pension System

by Renuka Sane.

A pension program must ultimately be judged by the consumption delivered in retirement. In defined contribution (DC) pension systems, such as the NPS, we accumulate wealth over our working life, and draw down this wealth after retirement. When savings are gradually drawn down in order to pay for consumption, there is the possibility of living too long and running out of savings. Buying an annuity eliminates this risk, but it may yield a low consumption per year of life.

In the early years, the focus in pensions policy thinking was on investment. Our task was to get participants going with regular contributions, and sound asset management, so as to build up pension wealth. This thought process has given us the National Pension System (NPS). The prospect of exiting the workforce, and using pension wealth to obtain a stream of consumption, was buried somewhere deep in the future. As a consequence, this part of the pension process has thus far been under-emphasised.

### Why do we care about the draw-down phase of NPS?

In India, the NPS forces individuals to annuitise 40% of their pension wealth, and take the remainder as a lump-sum. The subscriber can delay the annuity purchase or lump-sum withdrawal by 3 years. If the accumulated corpus is less than or equal to Rs.200,000, the subscriber can withdraw the entire amount and forgo the annuity purchase. Phased withdrawals are prohibited.

Now that more than 10 years have passed since the first entrants into the NPS, and we are getting closer to the first full cohort retiring, it is time to evaluate the draw-down policy. If there is more clarity and improved design for the draw-down phase, this may induce increased enrollment into the NPS.

### The questions

There can be many reasons for exit from the NPS - voluntary retirement, untimely death of the subscriber, and exit at the prescribed retirement age. In this article, we focus mostly on the latter i.e. draw down policy on retirement. Two questions are faced here:

What is the optimal level of mandatory annuitisation?
Different countries have approached the question of mandatory annuitisation differently, and are largely influenced by the existence of a state funded pension which offers protection from poverty in retirement. The Chilean approach, for example, has been to restrict lump-sum distributions, and mandate the use of fixed inflation-indexed annuities or lifetime phased withdrawals. The Australians are more flexible in allowing lump sum withdrawals. Most recently, the UK has done away with its rule of mandating the purchase of an annuity by the age of 75, and allows for programmed withdrawals. The US has very little mandatory annuitisation. Is our mandatory 40% annuitisation optimal, and if not, what should be done?
How do we make the market for annuities work?
Life insurance companies are often reluctant to enter into annuity markets because of the lack of availability of good mortality tables as well as instruments for hedging longevity and inflation risk. Customers may like a survivor annuity that includes the spouse, children and dependent parents, but from a pricing perspective, this may not be feasible. Customers may like inflation indexed annuities, but this requires that the insurance company is able to trade in a market for long dated inflation indexed bonds. We need to understand what are the requirements for enabling an annuity market that is able to provide competitive pricing on its products. The problems of the Indian Bond-Currency-Derivatives Nexus inhibit the emergence of an efficient market for annuities.

### Gaps in our knowledge

In order to arrive at an optimal level of annuitisation, we need to have an understanding of what our objective is from the annuitisation. For example, if our objective is to ensure a minimum consumption in retirement, then annuitisation can be mandatory only to the extent that is required to buy the minimum annuity. This requires us to take a view on what constitutes minimum consumption. A nominal annuity may not be able to buy a minimum consumption basket if inflation surprises occur over the lifetime of the retiree. In this case, annuitisation should mandate the purchase of an inflation indexed annuity instead of a nominal annuity. If, on the other hand, we believe that RBI will deliver on its 4% CPI inflation target, this changes the way we think about this. We have not had a larger policy discussion on this question.

Minimum consumption can be thought of either in terms of a minimum replacement rate relative to the average of the last few years of wage or contributions made, or a value that is linked to some consumption index. It may also vary depending on the age at which draw-downs are expected to begin. While the age of retirement is fixed for salaried employees, this may not be the case for informal sector workers. We, therefore, need to take a view on what will be the retirement age, or access age for informal sector workers.

If insurance companies do not take into account person-specific mortality differences, this is unfair on the poor as they die sooner. Under these conditions, mandatory annuitisation may be problematic for poor people. In this case, it might be more prudent to allow for a policy of programmed withdrawals. The trade-offs between an annuity and programmed withdrawal, and the design features of programmed withdrawals need to be better understood.

Finally we need to understand what impedes the development of annuities products. Why is it that insurance companies are reluctant to offer multiple products? What market and regulatory failures need to be addressed so that this market can take off?

### What is to be done?

Given the gaps in our knowledge the following elements of work are now required:

Design of annuitisation policy and phased withdrawal policy
The level of annuitisation needs to be thought through from the perspective of consumption as well as the ability of the market to provide such an annuity. This includes questions such as the access age, the level of mandatory annuitisation, the type of annuity, as well as the design of the programmed withdrawal product.
Developing annuity markets
Processes for solving market failures that may impede the functioning of annuities markets need to be set up. For example, an important policy measure that might be in the domain of the PFRDA is the development of mortality tables. PFRDA needs to establish a position on the requirements that the BCD Nexus has to satisfy in order to achieve PFRDA's objectives, and advocate this position with the Ministry of Finance. This includes dealing with questions about long dated bonds, inflation indexed bonds, interest rate derivatives, and instruments to hedge longevity risk.
Procurement procedure for annuity providers
The provision of annuities also depends on the competition in the annuity market, and the price at which the annuity is offered to the subscriber. A focus on low-cost annuity provision needs to be developed. For example, the procurement of annuity service providers should be done via auction, which leads to the lowest prices. This was the approach taken for the appointment of pension fund managers and has led to some of the lowest fund management costs in the world.

The author is a researcher at the Indian Statistical Institute, Delhi.

#### 1 comment:

1. The author argues that imposing annuity prices which ignores individual level mortality difference would create bias in favour of rich as they tend to live more.

In order to remove the pro rich bias, survival probabilities need to be a function of income. As per my understanding, insurance company include current health report in their pricing. Including income as an input for calculation of survival probability is relatively unheard of.

An easy and practical way is to incorporate the accumulated corpus as a proxy for rich/poor. This would be like a progressive annuity price system where large lumpsums are priced at higher rate.

Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticising me is perfectly okay; uncivilised language is not. I delete any comment which is spam, has personal attacks against anyone, or uses foul language. I delete any comment which does not contribute to the intellectual discussion about the blog article in question.

Please note: LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.