Executive Pay Inflation Rates

Executive pay inflation is something that worries us.  We are acutely aware, as are the board chairmen and remuneration committees we deal with, of the intense scrutiny and criticism levelled at the rate executive remuneration is increasing.  Twenty-nine percent in 2004, 16.5% in 2005 and 11.5% last year: all considerably above the CPI and AWE rates.  Our concerns are exacerbated when we acknowledge the role the remuneration consultants have in increasing executive pay.  We want to make sure that everyone lifts their game, including our competitors, to apply the same rigour we do in estimating what these increases may be.  This article examines some of the methodology that can be applied for more valid results.  We explain this in plain English for the most part, although the latter part of this article gets a little “mathy”.  Up to then we trust it is a good read for most general readers!

After decades in this business, we are also uncomfortably aware of the role that consultants can play in inflating rates of increase.  We worry that many advisers tend to apply the method known as “last year’s increase, but different” approach.  That is, they take the prior year’s increase, but tweak it up or down by a fraction on an entirely subjective basis, only reasoning that the rate of increase should be similar to last year’s.  It should not.  This is dangerous, and boards should demand to know more of how these inflation factors were arrived at.

Board remuneration committee decisions on CEO pay take into account current market rates.  However, as there is no ready source of current market data in Australia, the committee has to either:

• Rely on data from company disclosures, or
• Commission a survey of current market rates

Most companies rely on public company disclosures.  But these tend to be up to 12 to 18 months out of date.  Therefore, to compare their CEO’s current remuneration to the market, the market data from company disclosures has to be inflated by an expected rate of increase to a common and more current date.  This is called “aging” the data.

The application of aging factors to disclosed data is, or should be, a major concern for all board remuneration committees.  Last year disclosed ASX 300 CEO remuneration increased by 11.5% on the year prior (on a like for like “same incumbent” basis). See here.

Why?  Was this the result of supply and demand factors, generous incentive pay as a result of buoyant company earnings, or some other factors?  This is important to know, because any rational company would not be expected to apply similar rates of increase if the labour productivity is not there to support the increased costs, not to mention unwelcome media attention, governance downgrades and a reduction in shareholder trust. 

Nevertheless, this variable tends to be badly estimated – often without any sort of credible justification. To address this problem, Guerdon Associates develops and applies valid statistical models that can be applied to disclosed remuneration data.

A generic model is described in this article to illustrate how this is done.  In practice we prefer to be more specific by taking into account peers and industry type.

Basically, we’re looking for a way to estimate what the remuneration level should be in the near future. From there, we can use it as an actual remuneration forecast. Because we are assuming that there is some sort of connection between an executive’s compensation and their performance, the models contain at least one variable that is used as a measure for the company’s performance.

Some of the variables are also what are called lagged, which just means that they’re taken from the previous year’s data. (So if we’re forecasting the 2007 remuneration, a lagged variable will use its 2006 value even if its 2007 value is available.) Reliance on lagged variables has many advantages.  Chief amongst these is that estimates of future remuneration are not reliant on other estimates, such as economists’ inflation forecasts, analysts’ expectations of earnings growth, etc. To the extent that this can be achieved the more valid, we believe, is the remuneration increase forecast.  It is fortunate, then, that Guerdon Associates has developed robust methods for forecasting executive remuneration increases and levels on lagged (i.e. known) data. 

We examined many “lag” factors impacting increases.  Among the most valid is company market capitalisation.  Many may be uncomfortable with the fact that remuneration increases are “just” related to company size.  But in effect this is a measure of company performance.  An increase or decrease in market value can be attributed, to varying degrees in most cases, to executive performance.  This makes intuitive sense as well.  That is, the remuneration increase applied for next year relates to the achieved market capitalisation of this year.  Investors should take some comfort in this.  That is, overall market executive pay increases are not the outcome of self-interest and greed (although certainly there may be odd instances of this), but are the rational outcome of a robust relationship between shareholder wealth and executive pay.  As further comfort, our research is backed up by similar findings for the US and UK.

Of course, it does not end there.  Once we have verified a robust relationship, the next is to test various models of the relationship in order to provide the best possible estimate of future executive pay increases.  This can (and does) vary by industry and other factors.  But, by way of illustration, we show alternative “total market” models.  These use lagged variables (e.g. market capitalisation) and variables that in themselves are other experts’ estimates (e.g. EPS growth).

From here on the article gets a little into the math, for those so inclined.

0106 math1.png
A simple model to use since it only requires the previous year’s data to do any forecasting. The market capitalisation is present to provide some feel for how the firm will perform, while the previous year’s remuneration is present to provide some scale to work on. Since it relies on the previous year’s figures, this model may not be efficient if it is believed that the company will not have the same performance pattern the next year. Some form of professional judgement (such as external advice from Guerdon Associates) should be applied when using this model.
0106 math2.png

Instead of using the previous year’s market capitalisation, this model uses the current year’s EPS as the measure of the firm’s performance instead. This model is useful when a reliable estimate of the coming year’s EPS exists. It does not assume any sort of similarity between the previous year and the coming year. However, this model will only be as good as the EPS estimates that are used.

0106 math3.png

Instead of the EPS used in the previous model, this model uses the forecasted market capitalisation of the firm in the coming year. This model is useful when the market capitalisation forecast is thought to be more reliable than the EPS forecast, or if the EPS forecast does not exist.

0106 math4.png

The final model is similar to the first model, but includes the previous year’s EPS as an additional factor. This model is useful if neither the EPS nor market capitalisation forecasts are reliable enough to be used. Choosing between this model and the first model is a matter of judgement on whether the firm’s EPS is a good proxy for the firm’s performance.

Choosing between the models is dependent on two things: the available data and judgement in choosing which model seems to be the most appropriate. If multiple models are available, judgement should be applied as to which model(s) are to be used, or in what combination the models will be used. We recommend using a weighted average of the models if more than one model is used.

To convert the estimated remuneration value into a dollar format, use this formula:
0106 math5.png

To obtain the aging factor for a particular year, two consecutive remuneration figures are required. Say that we have used one of our models to forecast the 2007 remuneration figure and we then wish to calculate the 2006-2007 aging factor in percentages, it will be:

0106 math6.png

From there, the aging factor can then be used to convert partial information into a more useable form such as a situation where an executive resigned from their post in August 2006, with the next financial year being 11 months away on July 2007. When we are at July 2007, we may wish to compare this executive’s pay with other executives, but we have only information on their compensation package at August 2006. To convert or “age” the August 2006 information to a comparable July 2007 data, we can then apply our newly calculated aging factor in the following manner:

0106 math7.png

We have multiplied the Aging Factor inside the exponential by eleven twelfths because there are 11 months between the start of August 2006 and end of June 2007. You can use more exact fractions if you wish. For example, you can use the fraction of the days of the year remaining from the day of the executive’s resignation until the end of the current financial year. We now have an appropriately “aged” dollar figure for this executive’s compensation, which we can now use directly in whatever way we wish.
These are just examples of how aging factors may be calculated in a more credible and consistent manner, based on numerical methods. The models provided here are of a simpler nature and are intended for easier application on spreadsheets or statistical programs.  For more information of the models we use in consulting applications please do not hesitate to call us.

The important aspect to take away from all of this is to make sure that the source of your advice for executive remuneration increases relies on more than the “wet finger in the breeze” approach.  A company’s bottom line, board reputation, and shareholder interests demand more.

Enquiries or comments regarding this article can be directed to Peter McAuley, +612 9270 2912, or email Peter at peter.mcauley@www.guerdonassociates.com.

© Guerdon Associates 2024
read more Back to all articles