Change-in-control payments have been pruned way back. Historically, stock market conditions and regulatory changes were the big drivers of executive pay trends. But in the past few years, ISS (Institutional Shareholder Services) and other proxy advisors have taken the wheel. Pay complexity is at a bewildering new peak, so participant discounting — the “wastage” of executive pay — is high as well.
We see this not merely as an era of quirky, rapid change nor just one of heightened complexity. It is also an era of widespread, deep misunderstandings and strange decision-making. Today, large swaths of the landscape appear to be ruled by innumeracy or delusion – by myth. Myths are making pay policy less effective, efficient and compelling. To make progress, step one is to expose the myths.
Myth #1 Financial performance targets are too hard to set in an LTI context
ISS is often blamed for the market “mandate” to add performance conditions to long-term incentive grants. But this admonition has been around a long time. Corporate interest groups and watchdogs have been demanding this sort of accountability for years (a range of opinions of governance advocates are summarized in The New Standards by Richard N. Ericson).
Shortcuts are in demand, for LTI target-setting. Relative TSR is a popular one — peer median TSR often can act as target. So are one-year performance periods, threshold-only or “gate” style goals, and trailing averages of annually-set bonus goals. These methods often are used for good but temporary reasons. However, none is faithful to the key idea of performance-based LTI, to the basic governance contract in which executives make long-term business performance commitments in exchange for potential long-term wealth.
Companies have something to gain from true goal-based LTI. Concentrating pay-outs on tangible goals and long timeframes as it does, it can create stronger, more explicit linkages between decisions, long-term results, value creation and reward. Many companies should not be going to so much trouble to avoid setting long-term goals for LTI. Goals are not actually that hard to set and run. Here’s what we mean:
There are plenty of sources of help. Sector performance norms are worth a close look, applying balanced and complete metrics and focusing on persistent trends. Stock analyst forecasts are worth a look, though the user must be on guard for sell-side bias. Capitalization multiples, too, can indicate high or low future earnings growth expectations (or higher or lower risk or free cash flow).
The LTI plan structure itself can be helpful as well, to ensure payouts align with economic performance over time. New goals are set for every cycle so there are entry points for any needed course corrections. Variable LTI expense accruals tend to reduce payout variances too (example later). Adjustments for big events like M & A can tame their volatile effects while preserving accountability. Long-term gains are earned across a series of multi-year grants, with heavy averaging effects that continually extend the timeframe. Most performance-based LTI plans are stated in shares, so the share price acts as arbiter of earnings quality.
No one gets a “Mulligan,” but there are plenty of ways to manage targeting variances and keep outcomes proper. Uncertainty per se does not make it harder to set workable goals. Normally, it just requires a wider margin of error — that is, wider performance ranges. Goals themselves can be set up to work flexibly and adaptively over time, making plans responsive to business conditions as they play out. Most companies simply use the business plan as a quarry for performance goals. “Sandbagging” is a concern here, but one often offset by the chance of optimistic bias.
Here’s the key thing about LTI plans: you do not pay the awards out until the period is over and you know what happened. This creates tolerances much wider than in other types of forecasting. LTI pay is meant to vary across a broad central range of likely results.
Myth #2 We do conventional pay benchmarking, so our pay is competitive
Traditional methods do not assure proper benchmarking for competitive pay. Benchmarking means comparing the complex bundle of claims issued by your company with those issued by peer companies. We encounter issues all the time. Examples:
Valuation methods can distort granting. Relative TSR shares, for example, tend to be valued at levels roughly 25% higher than a comparable grant of traditional performance shares (that is, ones based on pre-set financial goals, not on relative TSR). A traditional performance share grant with an expected payout of $1,000 in shares would be valued, by benchmarking convention, at $1,000. A relative TSR grant on the same $1,000 in shares would be valued around $1,250. A company could switch, issuing 25% more in traditional performance shares — a very large pay increase — and it would be unrecognized in a typical pay assessment.
Traditional performance cash grants, on the other hand, are overvalued routinely. Customary benchmarking states their value at 100% of target, ignoring risk and deferral. They would have to be discounted by at least ¼ to be put on a consistent basis with the stock-denominated LTI grants that make up the bulk of the market data (stock prices are already discounted — they are risk-adjusted present values). In this case, replacement with traditional performance shares increases grant sizes by a gigantic 1/3 or more, again undetected in normal market pay assessments.
Performance standards can be skewed, quickly causing large competitive gaps in pay. Half of top officer pay hinges on pre-set goals. Payouts are leveraged, so if goals were low overall by 5% relative to market norms, payouts could be high by 20% (4x pay/performance leverage example). Companies often think of competitive pay as just one data point. But the solid line shown in the chart above is a better objective — it expresses competitive pay at a range of different performance levels. The dotted line to the right, in contrast, shows the impact of above-market performance standards — uncompetitive pay at every performance level.
The perceived value of pay at many companies has been impaired, reducing real competitiveness. LTI is subject not only to some actual valuation haircuts cited earlier, but to large perceived value discounts as well. These stem from complex grant mixes, convoluted metrics and standards, and unusual deferrals.
Pay is linked to organization size, so problems in determining “scope” of peer groups can distort market pay estimates. Some companies continue to use revenue too heavily as an indicator of scale. Factors like margins, risk, and growth potential are important in judging economic scale alongside revenue. Market capitalization reflects these drivers so it can be a good companion indicator of overall economic scale, where practicable. We find that market capitalization predicts pay more reliably than revenue in many samples.
Overwrought peer groups are unproductive. Highly studied groups of peers often generate results statistically indistinguishable from simple survey data or general industry samples. Stock prices of business “peers” should correlate unusually strongly with the company’s, as one test of peer validity. Often, they connect no more strongly than randomly-chosen stocks. This problem calls into question some peer groups used for relative TSR purposes as well.
Cost deserves more attention at benchmarking time. In the incentive plan below left, for example, an increase of $5 over target drives payouts from $5 up to the maximum level of $10. That is, incentives cost 100% of marginal income. Companies often come unintentionally close to this cost level.3 The second example shows the same outcome, but based on 5X leverage and a 20% bonus cost at target (pay increases by 100 percentage points relative to target when performance exceeds target by 20 points, so leverage is 5X). Actual effects would be under 100% since bonus cost variances tend to normally affect income.
Why do surprising pay cost effects go unexposed? A typical pay benchmarking exercise is not looking for them.
Myth #3 For Most Companies, Relative TSR Should Be a Prime Incentive Metric
We believe relative TSR should be a prime metric only in special situations. When the company and its peers have particularly strongly shared business influences, for example and, often relatedly, when alternative methods for performance measurement are especially problematical. Instead, use of the metric appears to cluster mainly in response to two drivers: a perceived mandate to put goals in LTI plans coupled with unwarranted fears about setting long-term financial performance targets.
The end result often looks like a structured equity interest netted by a large random variable — the peer TSR array. Discounting is an obvious risk. Speaking anecdotally, participants often tell advisors that they do not like relative TSR plans. Data is the plural of anecdote, as they say, and we have some formal data as well. Working with the London School of Economics, we undertook a global study of executive attitudes about pay. One of the more startling findings was the extent to which executives reject and discount pay plans that are unusually complex, risky or deferred. As noted in The Psychology of Incentives (2012). Relative TSR plans are perhaps the poster child. In many cases, relative TSR should be moved into a more limited “award modifier” role.
Myth #4 Companies Should Bend Over Backwards to Comply with ISS
Executive wealth almost always falls sharply when company prospects turn down. ISS somehow missed that when it devised the pay-for-performance tests used in making say-on-pay voting recommendations. Real pay-performance disconnects are irksome and they often do signal problems with “the tone at the top.” ISS procedures have been arbitrary, however.
Concerns about ISS can warp pay policy. For example, companies sometimes act as if adopting relative TSR plans is helpful from an ISS approval standpoint. But we saw that relative TSR grants carry higher costs at target than traditional performance shares, causing higher pay costs in the proxy Summary Compensation Table (SCT). This higher pay would increase the chances of failing one of ISS’s tests (the alternative, pay cuts, would reduce competitiveness).
The irony does not end there. Relative TSR grants have been unhelpful in regard to the performance aspect of ISS testing as well. Why? Relative TSR grants are presented in the proxy statement at a fixed value determined at the grant date. Those are the quantities ISS uses for its pay/performance scoring. But the grants are reported years before they have done their work to align pay and relative TSR. The grant might just as well have been based on relative shoe size. Actually, the main thing ISS and other bodies have militated for was not relative TSR but goal-based LTI more generally. Traditional metrics like operating income qualify for this purpose just as the relative TSR metric does.
There remain plenty of ways for prudent companies get tripped up by ISS. Having sensible, compelling and competitive pay is the best solution on this front and many others, of course. From that starting point, though, one still needs to be on guard for periodic and arbitrary problems from ISS. And a question companies should ask is whether it is worth the trouble in all instances to comply with ISS to ensure their “yes” recommendation. That last tranche of votes might come from understaffed compliance functions at asset managers, reflecting no close look at the company’s pay policy.
Obviously pivotal, executive pay is an area of corporate policy where companies should innovate and compete for advantage. Against this, ISS has been running a “race to the middle” for years. The results are in. Instead of innovation, we got a focus on compliance and, failing that, special pleading. Companies now go directly to their owners, disintermediating ISS. This outreach is called a “best practice” rather than what it is — a costly reaction to a predicament, to a market with a discredited intermediary. If a bridge is out, does it become a “best practice” for travellers to swim the river one by one?
There is a widespread “sameness” of pay trends today, a bit like it was before the onset of option expensing. Pay trends should be showing much more dispersion, more adaptation to business differences. Companies should refocus their pay policies for strategic impact. They should regard pay not as a politically reactive policy but as a proactive instrument of business governance. The current focus on pay governance should shift to pay as governance.
Myth #5 Companies Need to “fix” LTI Pay Costs
The desire to fix pay costs seems partly an artifact of an earlier era in which “fixed,” for the most part, meant “no option expense.” Today, companies can avoid expense variances by issuing relative TSR grants settled in shares. However, as noted earlier, doing so causes an increase in target cost of around 25%. That seems a high P & L “price” for indemnity from variances. Restricted share and option grants would fix expenses as well, but the former forfeits the desired performance element of LTI design and the latter is well out of favor (options are addressed next).
Performance shares based on operating metrics have costs that can vary based on performance but this should not be seen as unfavorable. Variances are driven by actual financial results, so they run high when earnings are high and low when earnings are low. This acts to shave the peaks in earnings and to make the troughs shallower. In the left panel above, LTI cost is fixed at 5% of income. On the right, the cost is variable and so the residual variation in earnings is much lower.
Companies habitually worry about line-item expense variances, in a budget context. Companies should put these worries aside when considering incentives.
Accounting aside, incentives actually shift a remarkable amount of business risk to employees. And, often, they shift risks onto people who don’t mind — people who favor having an upside stake and who are not scared off by accountability. Ceteris paribus, shareholders want less risk. This risk shift is an economic “win win.”
There has always been a bias toward “fixing” accounting expense. As it did in the past with “free” stock options, accounting rules can bias incentive policy. As always, the accounting portrayal of incentives should not predetermine design choices.
Myth #6 Stock Options Are Mostly Useless
LTI policy was driven wholly by accounting rules a decade or more ago. At-the-money option grants carried no expense so they were, predictably, wildly overused. Now, options seem to have “leper” status. This is an absurd overreaction. Options have some serious, standing advantages. They convey stock-based upside, linking with investor expectations and outcomes. They say to investors, “I don’t gain, from today, if you don’t gain.” They do not require explicit performance targets. Utility can be high, as they allow recipients to control exercise, timing of income and stock exposure. Their cost has come far down post-crisis along with volatility, and the cost impact can be lessened with shorter terms.
Options usually are not strong in terms of “perceived value.” They do provide a good ongoing fit in high-growth settings (e.g., venture-stage tech) where traditional metrics are hard to apply, where cash is short and where deep grants are known to be helpful at hiring time and strongly engaging afterward.
Myth# 7 Our change-in-control terms are shareholder friendly
A shareholder-friendly CIC policy would encourage a takeover at a premium price and when the board desires it. One of the most pro-shareholder dynamics is a liquid market for corporate control and asset redeployment. Company CIC plans have gone the other way.
“Gross-ups” have mostly been eliminated. They’re often replaced by “best net” arrangements in which the executive gets the better of two choices: the after-tax amount assuming all CIC payments are made, or the “safe harbor.” In the former case, the employer loses a tax deduction of amounts over 1X.
Single trigger equity vesting acceleration is no longer the norm. Acceleration is now more often subject to board discretion. Grants made within 6 months or a year prior to a CIC often do not accelerate. CIC severance continues to be two or three times base salary and bonus, but that is before the best-net effect. Much of the change seems prudential, aimed at avoiding absurdly high tax costs or executive windfalls. But now, actual payouts are often very unpleasant news. For many, CIC actually is one of the worst ways out.
In the example below, removing the gross-up reduces CIC payments to the level of ordinary CEO severance — about two years’ pay. Are top officers routinely trying to get terminated so they can receive a 2X payment? Then why would they favor a CIC?
Tactical methods can help mitigate the reduction in after-tax proceeds. But overall there is no 280(g) magic bullet, three decades years now after enactment.
CIC gains now go to fewer, in smaller amounts and quite unevenly — basically, to those having the largest vested stock holdings. With secularly lower acquisition premia, those effects are reduced. Stock holding requirements are 5X salary, typically, for a CEO; that is a bit over 1x against total pay, since salary runs as low as 1/5 of pay. A 25% control premium would convey, what, a few months’ pay? Expect no deal here.
Companies do have choices in the CIC area, and these policies need some rethinking. For example, cash CIC/severance terms often make sense for a new hire. Do they make sense years later? And at a low CIC price? And must pay accelerate upon a CIC, to avoid the rare post-close repudiation? If buyer promises were unenforceable as a rule there would not be much of a deal market anyway. Should boards more often have discretion, then?
The rationale for long-term incentives can be assessed together with those for long-term retirement benefits and CIC and regular severance protections. Taken together, these are mostly--performance-dependent long-term capital accumulation arrangements, ones that provide some indemnity for loss of employment under a range of circumstances. Their terms can be converged, as needed, to meet proper pay objectives like desired CIC effects. Long-vesting, contingent equity grants? Stock-based SERPs? To encourage CIC when proper and to avoid a 70% marginal tax rate, we should be willing to break some eggs.
Myth #8 Taxes Don’t Matter Much
The bulk of executive pay in the U.S. is taxed as ordinary income. Capital gains taxation normally means giving up a corporate or pass-through deduction. Tax deferral per se is not as valuable, with interest rates so low.
So taxes are not a big variable, right? Wrong. Tax rates in the U.S. are high, complex and uncertain. Deductibility of top officer pay is in the cross-hairs for congressional revenue-raisers. Losing a corporate or pass-through deduction can mean increasing an after-tax cost by 60% or more. Companies can be expected to use capital-gains-generating forms of pay for top officers, as practicable, if they are losing the deduction anyway.
The picture is dicier overall for the dozens or hundreds of other key decision-makers in the company. They are the bulk of the management payroll and they are the big targets for taxing authorities, too. They make good money pre-tax but face often-surprising marginal rate effects. Many are HENRYs — the high-earning, not-rich-yet. For them, a key preoccupation is whether their high incomes will allow them to build wealth or just to live nicely today. Tax-effective and wealth-enabling pay is a big win, increasing the perceived value of pay plans for this key cohort without raising costs much, if at all.
Most companies should consider QSERP. This set of tactics makes much greater use of qualified retirement plans. Such plans have remarkable tax benefits, being currently deductible, deferred to the individual and often effective in reducing or eliminating payroll and income taxes. Companies may increase qualified deferrals by tens or even hundreds of thousands per executive per year. To a HENRY, that is a big deal.
Non-qualified deferrals are now more constrained under 409A and they convey no current deduction. But participants can still ladder their deferrals and re-deferrals, controlling the timing of much of their taxable income. Deferrals still can allow late-career executives to move income into the lower federal tax rates that may apply in retirement and, in some cases, into low-tax locales. Stock options can be helpful as well, when timing of income has sharp effects. Tax education is a key to success in this area generally.
Conclusion
Executive pay is highly quantitative and strategic subject matter. It should be driven by fact, principle and method, and not by myth. Here are some proper ingredients based our experience
• Alignment with business strategy, industry conditions and shareholder value creation
• Proper and judicious use of market data
• Reasoned responses to proxy activism
• Efficacy and prudence, with attention to agency effects, plan costs and all relevant taxes
• Parsimony — streamlining pay to a small number of contingencies and moving parts
• General skepticism, diligence and numerical sense
• Thorough testing and measurement, with intensity befitting the high stakes
Richard Ericson is a managing director with PwC in chicago, Paul Perry is a partner with PwC and Scott Olsen is US Leader of PwC's Human Resource Services (HRS) practice The research assistance of Andrew Reuter, Alexander Widlak and Kamal Chakravarti is acknowledged.