Glass's: ‘We are always looking for greater accuracy’

Pricing guides are an easy target for organisations unhappy about fluctuations in residual values. They are accused of being too slow to react, or reacting too quickly; their changes are too severe, or not severe enough. They just can’t win.

One fleet view that has tended to hold more than a grain of truth is the belief that valuations are largely based on personal experiences and not cold hard facts generated from analysing reams of in-depth data.

That may have once been the case, according to some in the know, but not any longer. The main price setting protagonists are far more transparent about their processes, opening their doors and their software programmes to greater scrutiny. They are also happy to be challenged about their historical accuracy.

However, there is still a place for experience when it comes to setting prices, according to Glass’s director of valuations Rupert Pontin.

The editors’ dashboard

Glass’s bases its car prices on eight million observations a year from retail and another 1.5m from the trade, covering 80-85% of market records. That data is crunched by its Forecast Data software but is then overlaid with evidence from its car editor experts, who spend half their time out in the field gathering information.

“This helps to steer the hard data; we can’t just rely on stats and analysis,” says Pontin. He estimates that as much as two-thirds of forecast prices have been tweaked in this way.

He adds: “We are always looking for greater accuracy so we have a dashboard to rate each editor’s performance, to help them to identify where we are outside of our valuations window.”

The target is 1% either side of the actual price; globally Glass’s is averaging 98.8% – 1.2% below the selling price. Pontin claims that, over the past nine months, Glass’s has been the most accurate forecaster.

“We overlay this with our standard deviation, which has reduced from 20% to 13% in the past year,” he says. “We have done this by drilling down by editor, manufacturer and model range into our observations to see where the shortfalls are and which models they are not as accurate on.”

The variations are often the result of manufacturer policy. For example, for mainstream brands, it is usually individual models that are out rather than an entire range and that can often be due to short-term activity to push surplus supply into the market.

“If the activity is a one-off we can’t change our values, so the difference could be short-term,” Pontin says. “This is where the benefits of our understanding of the market as well as the stats are so important.”

Variations can also appear in Glass’s valuations when cars appear to have been sold when they haven’t.

“We can see the hammer drop and then the car appears at auction next week: we don’t know if it has actually sold and that can affect the accuracy of the data we base forecasts on,” explains Pontin.

Glass’s has around 62,000 vehicles on Forecast Data, up from 46,000 in 2012 when the dashboard was introduced. It enables quicker, more accurate valuations but is currently only used for cars for which Glass’s covers 99.6% of the parc.

For vans it covers just 77% – equivalent to 12,500 vehicles, although that has risen by 20% over the past three years. In reality, that equates to the majority of the market, as chassis cabs are excluded from valuations because there is no way of knowing what body is going on the back.

Pontin wants to create a version of Forecast Data for commercial vehicles and motorcycles, but needs to resolve the issue of observations which are more limited in number.

He expects to have a system in the market within the next 12-18 months which will also enable Glass’s to extend its reach beyond its current 7.5-tonne limit to the heaviest trucks. “It will definitely facilitate better forecasting,” he adds.

Irrespective of Glass’s forecasting accuracy, Pontin believes leasing companies need to take better stock of activity in the auction hall. He accuses them of “almost dictating to the trade”, because they are setting their own residual values rather than reacting to what is happening.

“In other countries, leasing companies rely on the bid for the car and then they sell it – it’s driven by the dealer’s knowledge of what a car will sell for plus their own costs,” Pontin says.

“But in the UK, it’s still ‘that’s the money on the sheet and that’s what we will sell for’. So it is driven by the trade rather than the retail. I believe that this will change.”

In some cases it already is changing, as dealers improve their understanding of market data. Leasing companies, too, are taking much deeper levels of intelligence, looking at regional datasets, number of days to sell, options and colours, to understand what the retail customer is willing to pay.

Manufacturers are also getting in on the act, recognising that, in cases where they encourage dealers to register as many vehicles as possible, they need to understand how that type of tactical activity will affect the market.

“We are definitely seeing change with the larger leasing companies – they have to get the best out of the asset to make money,” Pontin says. “Shifting prices by 1-2% and ramping up volume can make significant savings. But you can only do that by understanding the marketplace. It also impacts fleets’ buying policies: tweaks can add up to big savings, especially as new car volumes increase. You have to choose the right cars.”

That selection can be contrary to a fleet’s priorities. For example, a job-need car will often be selected with small wheels to minimise emissions and maximise fuel economy, but small wheels aren’t as attractive to a retail buyer which will affect the appeal of the car when it reaches auction.

One way to overcome this is by being more aggressive when negotiating prices with the manufacturer. And there are “enormous” deals available, according to Pontin.

“There can be more than 40% discount for some cars – it’s not been like that for a while,” he says, although fleets need to act quickly: “Most manufacturers front-loaded the year in case anything went wrong . Because of this, I am sticking with my prediction that will be up 3% at the end of the year.”

‘Auction jockeys’ have negative impact on values

Remarketing is changing, according to Glass’s, fuelled by the need to sell vehicles as quickly as possible – ideally starting the process before the car has even been defleeted.

So-called upstream selling will see an inspection take place six weeks before defleet to establish the work required and to agree costs. The car can then be advertised in advance or go straight onto the line at auction.

Rupert Pontin says: “The focus is getting cars into the market as soon as possible. Having all the information on the website gives the ability to sell cars within 48 hours of defleet.”

Minimising days to sell improves cashflow for fleets or leasing companies and protects them from a potential drop in book values.

However, some remarketing processes can have a negative impact on auction times, according to Pontin.

“There are a number of companies that rely on auction jockeys who are paid on performance,” he says. “The risk is that they lead the market with vehicles sold not for their value but for what someone wants in their back pocket at the end of the month.

“It impacts on days to sell and vendor income and is the wrong way to do it; it doesn’t get the best return for the customer. But it has been acknowledged by some vendors that it isn’t the best way to do it.”