Scenario analysis tool predicts U-shape

A U-shaped recovery is the most likely economic outcome in the US for the next two years, but stagflation has a higher than anticipated chance of occurring according to a new paper about scenario analysis co-authored by State Street and GIC researchers.

The study, which revolutionises scenario analysis by reorienting it towards a path rather than a single period outcome, examines six different economic paths in the US between 2020-2022.

It finds that a U-shaped recovery is most likely with a 30 per cent probability, followed by a shallow V (24 per cent), a baseline V (22 per cent), stagflation (16 per cent), a W-shaped recovery (6 per cent) and a depression (2 per cent).

One of the authors, Dave Turkington who is senior managing director and head of portfolio and risk research at State Street Associates, says the U-shaped forecast is based on about 90 years of historical experience from the US. “Stagflation was a higher probability than I expected going into it,” he added.

The authors propose a new approach to scenario analysis that enables investors to consider sequential outcomes, which it argues is not only more intuitive but provides better predictions.

They define prospective scenarios, not as average values of economic variables but as paths for those variables. So multiple values are given for each economic variable representing the early, middle and late stages of a pattern which might cover multiple months or years.

In order to estimate the relative likelihood of a prospective scenario, the authors work out the statistical similarity of the scenarios to the most recent economic experience using a method called the Mahalanobis distance.

“Effectively, we are asking: given the recent economic experience, how unusual would it be for one scenario to prevail going forward versus an alternative scenario?”

The analysis is progressive because it looks at the path, not just the end of the investment period, which means decision-making about the likelihood of alternative scenarios is more meaningful. By defining scenarios as paths, the relevance of historical observations can be better determined and this is important because the historical data is relied upon to map economic scenarios onto asset class returns.

“When investors focus on single-period outcomes, such as expected return or expected utility, they look beyond the metaphorical storms to the end of their investment horizon, which gives only a limited view of a portfolio’s exposure to loss,” the authors say in the paper, Portfolio Choice with Path-Dependent  Scenarios, which is forthcoming in the Financial Analysts Journal.

Co-author Mark Kritzman says the method shows that sometimes using a smaller, but more relevant sample can lead to more reliable results.

“You can get a more reliable prediction from a smaller sample if you have the right observations, the relevant ones. And the relevant ones are those observations that in past history are similar to what’s happening today and what’s relevant, everyone does that. But what they don’t do is ask how unusual these past observations are; the more unusual, the more informative they are.”

According to the GIC researchers, Ding Li and Grace Qiu who are also co-authors of the paper, defining scenarios as paths rather than single period averages has a number of benefits.

“Traditionally in scenario analysis the emphasis is on the destination – how the market looks 10 years later, or the average output through the horizon they are interested in. It only defines the scenario with one single number,” Li says. “In our discussion we decided the most important and intuitive way is to focus on the path that can lead to the destination. We think it’s very helpful at GIC as a long term investor that we have a better understanding of the past and it can give us some insights. It’s looking at the sequence of economic events to trigger that outcome. The tool is trying to emphasise getting more relevant information from history to extrapolate the future.”

GIC’s asset allocation toolkit

GIC has many tools that it uses for its asset allocation determinations, and is constantly adding to its toolkit. Another recent collaboration, with PGIM, resulted in the development of a sophisticated and timelyasset allocation framework that explicitly models the impact of private assets on total portfolio liquidity, incorporating both the top-down allocation view and the bottom-up cash flow view.

GIC’s Qiu says the main application of the new scenario analysis-based portfolio construction technique, with path dependent research, is part of the toolkit to drive asset allocation decisions and help address the long-term uncertainties.

“We used it in the strategic asset allocation process to help build a portfolio that is resilient across a wide range of potential scenarios, especially given we face more wider, and differentiated outcomes,” she says. “The path information is very informative and important to understand the downside risk of various scenarios and their paths. We look for new vulnerabilities and trends, with discipline to invest when attractive.”

She says one of biggest challenges for asset allocators today is the potential very wide range of outcomes.

“We have high valuations as a starting point, so there is more downside risk and a more volatile path along the way. It’s more challenging so we need better tools.”

Kritzman, who is an MIT academic, has had a long standing relationship with GIC, including previously sitting on its international advisory board, and his research and publications have been an ongoing source of knowledge for the GIC team.

Qiu says the collaboration with State Street Associates is similar to its other research projects in that it is motivated by the challenging and uncertain economic environment.

“With a wide range of potential outcomes and downside risk it is very important for long-term investors like GIC to construct a portfolio that is resilient across a broad range of market and economic conditions,” she says. “‘Prepare not predict’ is one of our investment beliefs.”

In conducting scenario analysis in 2019, GIC applied some of the techniques in one of the earlier scenario papers by Turkington and Kritzman, adjusting  and customizing their process to meet the fund’s specific needs.

“We made some changes and brought our adaption of their technique to discuss with Mark and Dave and this led to the new paper collaboration. One of the adaptions – the main novel contribution we made – was to look at paths rather than a single point.”

Qiu says the benefit of this new technique compared to more traditional approaches is it brings intuition, and while the tool itself is systematic investors should apply further quantitative or qualitative assessments that are suited to their own investment beliefs.

“It includes a nice economic narrative and as a result it can be more intuitive and understood by key stakeholders like board members,” she says.

Similarly, Li says the authors tried to emphasise in the paper that a framework is a starting point for a conversation around different scenarios.

“I think the main contribution of this tool, is it’s the objective opening point for management to debate and make decisions. Rather than say the tool gives us the correct answer, it provides a platform and starting point for the discussion.”

The underlying technique used in this paper was also used to predict the outcome of the US election and correctly predicted a Joe Biden victory. Read the story here.