EuroFinance

Managing foreign-exchange risk at scale requires consistency more than conviction. For a global manufacturer operating across dozens of currencies, predictability has long been a central objective. That remained true at Avery Dennison, a multinational manufacturer and distributor of pressure-sensitive adhesive materials based in America even as the external environment shifted.

“As the zero interest rates were gone, hedging costs started increasing,” said Sandeep Nene, director of treasury. “And we were under pressure to reduce costs.”

According to Nene, the question treasury faced was not whether to hedge, but how to balance cost, risk and governance in a way that could be sustained every month. The result was a gradual move towards a value-at-risk (VaR)–based FX portfolio, supported by automation and tightly defined controls.

A portfolio view of FX exposure

Avery Dennison operates in close to 60 countries and is exposed to around 25  currencies. That geographic breadth introduces volatility—but also natural offsets.

“We are a global company,” Nene said. “Which means there are some natural correlations and natural hedging that we can utilise to reduce our hedging costs.”

Treasury began analysing FX exposure at the portfolio level rather than each currency. VaR provided a way to quantify that exposure consistently, he added.

“Value at risk is just a statistical measure based on a confidence interval which tells you what will be your risk in that portfolio if you leave it unhedged,” Nene explained.

At a 95% confidence interval, the numbers were clear. “If Avery Dennison decided to not hedge any of its exposures, we would be having a plus 2 million or minus 2 million FX result every month,” he said. “That means in one in 20 months the result is actually going to exceed that.”

That was not a position treasury intended to take. “No CFO is going to be willing to take a plus 2 million minus 2 million every month,” Nene said.

Defining acceptable risk

Rather than aiming for either full hedging or no hedging, treasury focused on defining a tolerable level of residual risk. “We said we are not going to do that,” Nene said, referring to remaining unhedged. “We are going to take 10% of that and then start our modelling.”

That decision anchored discussions with senior management. “We agreed with the CFO saying 250k a month is a risk he can live with,” he said. “That’s where we started.”

The initial model was built on Excel and reviewed externally. “We got it reviewed by two of our relationship banks,” Nene said. “We did a back-testing period over two years, and we used a couple of black swan events to see how the model performs.”

The concept held—but the process did not scale. “When we ran that module, one of my analyst’s laptop was frozen for three hours,” Nene said. “We could run one iteration and then had to live with it.”

Automation as an enabler, not a goal

At the same time, treasury was rethinking its systems architecture. In 2019, Avery Dennison moved away from a heavily customised, hosted treasury management system.

“Every two to three years we needed to upgrade it,” Nene said. “And anyone who has done an upgrade on a hosted solution knows how costly it can get.”

The move to a SaaS model allowed treasury to standardise processes instead of customise systems. “We said we are not going to do any more custom reports,” he said. “We will modify our process if needed to match the system.”

When FX exposure management was automated, treasury insisted that the VaR model had to be embedded into the platform. “We gave them our Excel model,” Nene said. “And we told them that’s the minimum that we need out of the system.”

Once live, the process became repeatable. Forecasts from 120 business units are submitted centrally, netted, and hedged by treasury.

“The BU (business unit) is always 100% hedged,” Nene said. “Any risk that is taken is taken on the treasury entity and never the BU.”

That structure was intentional. “It also reduces the noise in the organisation,” he said.

 Rules over discretion

The VaR engine runs Monte Carlo simulations to map the efficient frontier between risk and cost. “They run a Monte Carlo simulation of a thousand points for any one iteration,” Nene said. Treasury runs multiple iterations each month to generate a broad set of outcomes.

Governance is embedded through constraints. “It cannot be that me or my boss decides the portfolio every month,” Nene said. Exposure caps, minimum savings thresholds and portfolio rules narrow the outcomes to a small number of viable options.

“That makes sense for us to put in the effort and get the savings,” he said. “And then we still put in more rules which then ends up with one portfolio—maximum two—between which we can choose.”

The approach has delivered consistent results. “We have been now doing this for three years and it’s been saving us a million annually,” Nene said. “A million is a million.”

Risk remains constrained. “Our maximum risk is 250k a month,” he said. Losses are monitored daily. “Every morning, my analyst sends out an email showing what is the mark-to-market on the current portfolio.”

Escalation is predefined. “If a loss reaches 500k, then we’ll have a call with the CFO and cut the portfolio,” Nene said.

Incremental progress

Backtesting remains central. “We took a couple of black swan events,” Nene said, including Covid, the Ukraine war and historical stress periods. “We wanted to see how the model behaves in that scenario.”

That discipline has built internal confidence. “Delivering such a project has also given us a lot of credibility within the organisation,” he said.

The next phase will be incremental. “We don’t want to go big bang,” Nene said. “It’s also optically better to keep improving every year.”

For Avery Dennison, VaR has become less about optimisation and more about clarity—making explicit how much FX risk matters, how it is measured, and how it is governed. As Nene put it, “You have to put in time and effort. You have to really dig into it if you want it to perform—and not leave it to hope.”