The decumulation problem

By Lionel Martellini and Shahyar Safaee

printer-friendly version

Lionel Martellini, Shahyar Safaee

Decumulation investing – An important question

In a nutshell, the decumulation problem is defined as the challenge involved in efficiently turning wealth into income, which stands in direct contrast with the accumulation problem, which is instead about efficiently turning income into wealth. In individual money management, the most typical decumulation problem starts when an individual retires, ceases to receive labor income, and starts to draw down on accumulated assets to generate the level of replacement income needed to sustain the target consumption level in retirement. The problem actually extends beyond individual money management to encompass all situations where an institutional investor is tasked with the complex challenge of managing assets while facing pre-committed outflows, as would be the case for example with a mature defined benefit pension fund or an endowment. 

Given the broad relevance of the decumulation problem – after all income generation is the end goal of most if not all saving processes – it is somewhat surprising that it has received very little attention from the asset management industry. It is striking that very few asset management products, even those presented as retirement products, embed explicit mechanisms designed to provide retirement income security. The inadequacy of existing retirement products and lack of focus on income generation in retirement has been precisely identified as one of the most fatal flaws of investment practice by Bob Merton (2014)[1], who offers the following one-sentence summary of the issue: “Our approach to saving is all wrong: we need to think about monthly income, not net worth.”

Fortunately, we have reasons to believe that the situation is slowly but surely starting to evolve in the right direction, and a much-needed change in focus in favor of an income-oriented mindset is at least emerging in the US with the Setting Every Community Up for Retirement Enhancement (SECURE) Act, a bipartisan bill enacted on December 20, 2019 that is designed to improve the US retirement prospects. Amongst a large number of provisions affecting individuals, pension plan sponsors, and small businesses, the SECURE Act does indeed help shift the focus of all stakeholders, especially individuals, towards replacement income and away from retirement capital. This is mainly achieved through two main provisions related to (i) lifetime income disclosure and (ii) the annuitization of assets held in a DC plan. Under the former, DC plans are now required to provide participants with an annual estimate of their affordable monthly lifetime income. Two illustrations are mandatory, representing the income streams of a single life annuity and of a qualified joint and survivor annuity. This will help individuals understand how much income might be derived from their account balances. It will consequently trigger a shift towards an income-oriented mindset when participants assess how far they are from their retirement income objectives. Regarding the annuitization of DC plan assets, the SECURE Act now provides a safe harbor for determining the capacity of the insurer to satisfy its obligations, thereby lessening the liability of plan sponsors. Additionally, the Act requires plan sponsors to deem the cost of the selected annuity products “reasonable” but does not require them to necessarily select the lowest-cost provider.

While placing much-needed emphasis on income, the SECURE Act provides renewed incentives for asset managers to rethink their approach to the decumulation problem, building on the fact that annuities, which are otherwise worthwhile decumulation products marketed by insurance companies, suffer from a number of severe shortcomings which undoubtedly explain their low penetration. In practice the demand for such products is actually extremely low despite their relevance for income generation in a retirement context. Using the RAND Health and Retirement Study dataset for the cohort aged 65–75 in 1998, Pashchenko (2013)[2] for example reports that only 5% of individuals receive income from annuities, with a peak at 12.2% among the highest income quintile and a low at 0.4% for the lowest quintile. Common explanations of the "annuity puzzle" are related to the fact that annuities involve counterparty risk and high levels of fees, and also that they do not contribute to bequest objectives. One additional key drawback of annuity products is their severe lack of flexibility. Annuitization is an almost irreversible decision, unless one is willing to bear the costs of extremely high surrender charges, which can amount to several percentage points of the invested capital. This lack of flexibility is a major shortcoming in the presence of life event uncertainties such as long-term care needs that arise due to health-related issues in the later stages of retirement. In this context, we argue that a Copernican revolution is needed in investment management, calling for a paradigm shift in favor of proper treatment of the decumulation challenge.


Decumulation investing – A difficult challenge

A partial explanation for the lack of focus on income in investment products is perhaps the fact that the decumulation problem is a very difficult one which involves a much higher degree of complexity than the accumulation problem on three key dimensions, namely sources of uncertainty (risk factors), decision-making process (decision variables) and preferences (objectives and tradeoffs).

Starting with an analysis of the sources of uncertainty, the list of relevant risk factors includes the standard investment risks found in the accumulation problem but also risk factors impacting the present value of income cash-flows, including most importantly interest rate and inflation risks, as well as personal risk factors such as long-term care and longevity risks. In the face of these sources of uncertainty, the control variables available to decision makers include not only investment decisions, as in accumulation, but also withdrawal decisions. In other words, individuals or institutions in decumulation have to assess optimal investment decisions for a given stream of decumulation cash-flows, but they also need to envision optimal income withdrawal decisions and their impact on investment decisions. To make things more complex, preferences involved in this dual optimal decision-making process in the face of extended uncertainty cannot be simply summarized in terms of the standard risk versus performance tradeoff typical of investment decisions. They actually involve more complex tradeoffs related to the presence of essential versus aspirational levels of replacement income, as well as the presence of a possible conflict between replacement income and bequest objectives.

The intrinsic difficulties involved in retirement investing have been identified by two Nobel prize winning economists, one of whom – Richard Thaler[3] – was quoted in 2019 as saying “decumulation is a more difficult challenge than accumulation”, almost as an echo to Bill Sharpe’s[4] emphatic view expressed a couple of years earlier that “decumulation is the nastiest, hardest problem in finance.”

Bill Sharpe can actually be credited for developing a conceptual framework suitable for addressing the decumulation problem in the form of a generic lockbox strategy defined as a way to frame and manage risks in a retirement income generation context. More precisely, a lockbox retirement income strategy consists in segregating assets by retirement year, and designing a specific investment plan for each targeted year. Let us quote Sharpe’s own words: “The idea is to assess the individual's preferences for various amounts of consumption in each future year, his or her risk tolerance vis a vis spending at various times in the future, current wealth and other sources of income, and then determine an overall plan. Part of this plan involves allocating current funds to a series of "lockboxes," each of which is designed to provide spending in a given future year. The box would also include instructions for the management of the money from the present to the terminal year. (…) Equipped with this framework, you have the key if you need to access the funds, but the idea is that, once a year, you would sell the assets in that year's lockbox.”

In this overarching framework, there are two main degrees of freedom, namely the decision on dollar budgets, or how to best split the investor’s initial wealth across the lockboxes, and the decision on risk budgets, or how to best split the investor’s portfolio across performance-oriented and hedging-oriented assets. By default, the investor can favor an equal allocation of retirement savings to each lockbox, and an identical level of risk-taking in all lockboxes, but more general choices can of course be made. In Bill Sharpe’s words: “Different boxes could well have different investment management strategies as well as different amounts of initial funding.”


Decumulation investing – A new frontier

The decumulation problem opens a whole new frontier for financial engineering. A key distinction is between secure decumulation lockbox solutions with no upside for well-funded institutions/individuals, which would allow investors to find in each lockbox the exact amount needed to match the replacement income needs based on an investment in inflation-linked pure discount bonds, and decumulation solutions with upside for less well-funded institutions/individuals. In the latter situation, investors would not find a guaranteed amount in the lockbox but instead a lottery ticket. Here, the remaining question of how to design optimal investment strategies for each lockbox raises a number of interesting challenges regarding the kind of lottery ticket, or portfolio payoff, to be found in each one. Given the intrinsic correspondence between the set of dynamic portfolio strategies and the set of affordable payoffs, one can in particular use the key principles of goal-based investing (GBI) to build risk-managed strategies allowing investors to secure essential levels of replacement income while maximizing the probability of reaching target levels of replacement income – see Martellini and Milhau (2021)[5] and Deguest et al. (2021)[6] for recent references.

While each strategy will have to be specified as a function of the specific problem at hand, a general taxonomy can already be sketched out with three main categories. The first category includes buy-and-hold GBI decumulation strategies where each lockbox contains a lockbox-dependent allocation to performance-seeking assets, e.g., equities, and an inflation-linked pure discount bond with the relevant maturity. For practical purposes, these strategies will aggregate up to some combination of an equity portfolio and a retirement bond with cash-flows of potentially unequal sizes. The second category includes glidepath GBI decumulation strategies, where each lockbox contains a revised-over-time allocation to performance-seeking assets versus safe goal-hedging assets. Finally, the third category embeds all dynamic GBI decumulation strategies, where each lockbox contains a state-dependent allocation to performance-seeking versus income goal-hedging assets.

In addition to opening a new frontier in financial engineering, the decumulation problem also opens a whole new frontier for EDHEC-Risk Institute. Building upon academic expertise developed during the last 15 years of research in the area of investment solutions for institutions and individuals, and following up on the Scientific Beta and Scientific Infra strategic initiatives, we are launching a new venture, Scientific Retirement, in partnership with EDHEC. We aim to position ourselves as architects and engineers of the next generation of retirement solutions to be implemented with index providers and asset managers for the benefits of institutional and individual investors. In addition to the aforementioned GBI strategies, which extend the liability-driven investing (LDI) concepts to individual money management, we also intend to explore cash-flow-driven investing (CDI) strategies, where real assets generating attractive inflation-linked revenue streams such as real estate and infrastructure are used to meet some of the cash-flow needs of investors in decumulation.

As Bill Sharpe eloquently put it, the decumulation problem is indeed a hard and nasty problem, but its importance is so overwhelming that this cannot be used as an excuse for inertia. We do believe that dedicating our time, energy, enthusiasm and resources to this question would be a most effective way to live up to the core ambition of EDHEC Business School, which is summarized in the motto “make an impact!”  


[1] Merton, R., 2014. The Crisis in Retirement Planning. Harvard Business Review, July/August 2014.

[2] Pashchenko, S., 2013. Accounting for Non-Annuitization. Journal of Public Economics, 98: 53–67.

[3] Thaler, R., 2019. Financial Advisors and Retirement: The Decumulation Dilemma. PIMCO Insight, October 28, 2019.

[4] Sharpe, W., 2017. Tackling the 'Nastiest, Hardest Problem in Finance‘. Bloomberg Opinion, June 5, 2017.

[5] Martellini, L. and V. Milhau, 2021. Advances in retirement investing. Cambridge University Press.

[6] Deguest, R., L. Martellini and V. Milhau, 2021 (forthcoming). Goal-based investing: Theory and practice, World Scientific Publishing.