El Niño, Hurricanes, and the Bias We Don’t Talk About

NOAA has confirmed the emergence of El Niño, but what does that mean to decision-makers on the ground? A closer look at seasonal hurricane predictions, how re/insurers respond to them, and a structural blind spot that may be costing everyone.

Introduction

On June 11th, NOAA officially confirmed that we have entered an El Niño (warm) phase of the ENSO cycle. For the insurance industry, the instinctive reaction may well be relief. As Daniel Swain explored in a recent Inigo blog, El Niño years generally correlate with a lower count of hurricanes across the North Atlantic basin — music to the ears of those managing catastrophe exposure.

But the picture is more nuanced than that. Any reduction in risk associated with lower basin-wide counts may, this year, be tempered by warm ocean temperatures near the US coastline. Warm oceans generally tend to increase the probability of intense hurricanes — the ones that do the most damage.

Both statements above come with the important caveat of “generally.” In reality, there is an ongoing and lively scientific debate around the interplay between complex ENSO cycles, ocean temperatures, many other transient variables, and how these all interact with North Atlantic hurricane activity. That debate grows even thornier when you attempt to limit predictions to exposed & vulnerable communities along North Atlantic coastlines.

Re/insurance connections

Over the past decade or two, re/insurers have been pressured to develop bespoke views of risk that not only reflect their strategic appetite for volatility, but that are also forward-looking with respect to a non-stationary climate. This pressure has been reinforced by regulatory exercises including the EU’s Solvency II (ORSA), the PRA’s CBES 2021 and SS5/25, and the NAIC’s Climate Scenarios 2025.

The positive outcome of these exercises is that many major re/insurers have now embedded near-term, forward-looking views of risk for weather perils that are affected by climate. To provide stability — both in client premiums and internal modelling/reporting — these views are typically smoothed over a minimum of five years to remove year-to-year volatility.

For North Atlantic hurricanes, the scale of that year-to-year volatility is striking. Research by Caron et al. (2020) (plot below) shows annual hurricane counts from 2000 to 2018 ranging from as few as 2 (in 2013) to as many as 15 (in 2005), against a long-term climatological average of around 6 per year, and a ten-year moving average (persistence) that trends slightly above this.

However, the annual volatility presents a potential opportunity for those that can react to it. Thus, beyond simply building forward-looking views of climate-adjusted risk, some re/insurers have begun flexing their smoothed, multi-year baselines in response to seasonal hurricane predictions. ILS and alternative capital firms have accelerated this trend, and the market has followed.

The question is: how to respond to these seasonal predictions coherently, and how to align them with the many different forward-looking climate adjusted baselines that exist across the reinsurance market?

A potential downside bias in the system

There is a common narrative worth naming — one that appears regularly in ENSO and seasonal hurricane prediction commentary, and one that this author has contributed to in the past.

In La Niña phases, when North Atlantic hurricane activity is generally expected to rise, headlines tend toward the dramatic: “risk is off the charts — low wind shear, background ocean warming, brace yourself.” The implied message is clear: reduce your exposure, buy more reinsurance.

The inverse is rarely communicated with equal force. In an El Niño year, the statement that “risk is lower” is seldom given the same weight. Where are the calls to deploy capital and expand exposure?

There may be legitimate reasons for this asymmetry — and all are worth debating:

• Is it socially perceived as opportunistic (the word “icky” was thrown at me at an event recently, and it’s a good one) to expand exposure and capitalize on events (or non-events) that are associated with destroying lives and livelihoods?
• Do people in risk advisory roles have a natural predisposition toward caution over risk-seeking behaviour?
• Who wants to advocate for lower rates when the upside is likely marginal compared to the catastrophic downside of structural under-pricing?

On the surface, a bias toward downside risk management may look prudent. But it may not be optimal — for insurers or the insured.

From an insurer’s perspective, a board may not understand the cautionary attitude embedded in a seasonal estimate, and may miss meaningful opportunities for capital deployment. It also creates a confusing message for anyone who has implemented a stable, multi-year forward-looking view of climate risk: if risk is never communicated as being below baseline, what is the baseline actually for?

Perhaps more importantly, from an insured’s perspective, this potential downside bias would feed into pricing and capital management decisions that ultimately raise the cost of cover. As has been documented extensively, rising insurance costs disproportionately impact the vulnerable. That is the core reason this issue warrants more open debate.

If advisors in the insurance chain are systematically overcautious, society may paradoxically end up less protected than it could be — not more.

In the market today

The current insurance market continues to soften, with pressure to cut rates bearing down on underwriters who are simultaneously trying to find adequately priced business. In this context, not allowing risk management teams the opportunity to flex their view of risk downward in years that are predicted to be lower-activity is inconsistent and indefensible. If pricing can flex upward in high-activity years, the reverse ought to be possible too.

The conservatism built into seasonal messaging rarely translates into optimal resilience outcomes. It is worth asking, honestly and collectively, whether the way we communicate seasonal risk — and the hypothesized asymmetric weight we give to downside versus upside signals — is serving the industry, and the people it exists to protect, as well as it should.

The open question: Where does bias exist in the risk advisory chain, and are we doing enough as a community to identify and address it?