mast image

Special Report

ESG: The metrics jigsaw


How AI could change manager selection at world’s biggest pension fund

Artificial intelligence (AI) could significantly improve how the world’s largest pension fund selects and monitors its fund managers, according to a report.

The report’s authors explained that a collaboration between Japan’s ¥158trn (€1.2trn) Government Pension Investment Fund (GPIF) and Sony Computer Science Laboratories (Sony CSL) developed a proof-of-concept prototype system to test the principle of “using deep learning to detect the investment style of managers from trading behaviour data collected daily by GPIF”.

It found that the system could properly detect different fund managers’ styles and drifts, “enabling evidence-based, prompt analysis of investment styles”.

Introducing the prototype AI system within GPIF would have multiple effects, according to the report, which GPIF commissioned last year.

The pension fund should be able to conduct more prudent and data-driven selection and monitoring of fund managers, which could in turn “foster more constructive and in-depth dialogue” between GPIF and its fund managers.

This would improve the robustness and performance of the pension fund’s investment practices in the long run, the authors said.

GPIF’s current approach relied on the track records and qualitative explanation of candidates and commissioned fund managers.

The authors also outlined a series of effects on asset management companies. 

“When asset management companies recognise that GPIF has the ability to independently analyse their investment styles and intends to continue development of even more advanced technology, they will recognise that they cannot justify their results with only qualitative explanations,” they wrote.

This would cause asset managers to introduce more sophisticated technologies to explain their behaviour and be accountable for their investment practices, which would accelerate the use of AI-assisted asset management.

Associated benefits could include high cost transparency, as a result of reduced reliance on individual persons from management strategies “whose fair value is ambiguous at best”.

“This sequence of developments will further promote the science and technology of asset management,” the report added.

Concerns about the cost of active investment management and manager selection recently prompted GPIF to link management fees with active return.

The report on the AI study can be found here.

GPIF joins Climate Action 100+

Separately, GPIF has joined Climate Action 100+, the institutional investor-led initiative to engage with major greenhouse gas emitting companies to improve corporate governance on climate change, curb emissions and strengthen climate-related financial disclosures.

Anne Simpson, investment director at CalPERS and inaugural chair of the initiative, said GPIF’s support would be “vital to our success”.

Related images

  • Blockchain

Readers' comments (1)

  • Interesting area of development with AI and selection. However, care does need to be taken as the aggregation of data can reproduce existing bias and prejudices in decision-making. An example of this was when Amazon used AI in selection and used all pre-existing data to inform the process because of previous bias to make workers, AI concluded that male workers were automatically better than female workers and was discarding all female workers so the whole scheme had to be scrapped. While this example is about historic patterns on inequality could bias in historical patterns of financial decision-making be reproduced? And would this be necessarily what is needed looking towards the future as investing parameters may change.

    Dr Susan Sayce
    University of East Anglia, UK

    Unsuitable or offensive? Report this comment

Have your say

You must sign in to make a comment


Your first step in manager selection...

IPE Quest is a manager search facility that connects institutional investors and asset managers.

  • QN-2575

    Asset class: Core Real Estate Muli-Manager Separate Managed Account.
    Asset region: Global.
    Size: CHF 150m.
    Closing date: 2019-12-20.

  • QN-2578

    Asset class: Sovereign Local Currency Emerging Market Debt.
    Asset region: Local emerging markets.
    Size: EUR 950m.
    Closing date: 2019-12-19.

Begin Your Search Here