Chapter 2: Get the Right Facts

Overview

The second principle of human capital management concerns facts. For decision makers, facts are like candles in a dark place: They illuminate portions of the landscape with a degree of certainty and cast some light onto the murky path ahead. There are, of course, no facts about the future, and so every decision contains some level of uncertainty. However, facts about the past and the present help reduce that uncertainty and increase the probability of desirable outcomes.

Unfortunately, many decisions that are intended to improve the productivity of human capital, some costing hundreds of millions of dollars, are made without sufficient facts or without the right facts. For some reason executives who refuse to spend a dollar on a new piece of equipment until they have studied all its costs and potential benefits will make multi-million-dollar decisions about people programs with little more than hunches or irrelevant facts dragged in from other companies to guide them.

Consider your own decision-making approach. If you were faced with a decision about replacing your current e-commerce infrastructure, you’d insist on substantial facts and financial projections. You probably would ask for a net present value analysis of the best-case, worst-case, and most-likely-case outcomes. The assumptions underlying that analysis would be scrutinized or challenged. In some cases you’d go a step further, conducting a “sensitivity analysis” to determine how those forecasted outcomes would change if certain critical assumptions failed to hold. You also would want to know the internal rate of return from the proposed investment. If that rate is positive, does it exceed the returns available from alternative investments? If too few facts were available to support those analyses—and assure a good decision—you probably would ask staff people to chase them down to reduce the level of uncertainty.

Such fact-based rigor almost never is applied to decisions about human capital, even when they involve huge sums of money. More often one sees decisions based on the following:

• No facts. “Common sense tells us that our employees will be more productive if they have a stake in the profits of the business.”

• Unreliable facts. “Employees say that they are more likely to stay if we offer profit sharing.”

• Irrelevant facts. “We benchmarked three world-class companies with variable pay plans: a bank, a hotel chain, and a defense contractor. All reported good results.”

These types of “facts” won’t help you. What you need to make high-quality decisions are relevant and reliable facts. And decision quality matters. Think how much better your company would perform if the decisions made by lower-level and mid-level managers improved by just 10 percent. You’d see a notable boost in bottom line results. Now think what would happen if the CEO and other senior executives improved their decisions by 10 percent; you’d see a huge effect on the bottom line. David and Jim Matheson made this observation in The Smart Organization noting the leverage that decisions made at the top have on outcomes.^[1]

Not all facts are created equal. Some facts are weak or too unreliable for important decisions. Facts drawn from anecdotes fall into this category. Facts drawn from benchmarking (internal and external) and statistical correlation are generally stronger; they can help you form useful insights. As Figure 2-1 indicates, facts about cause-and-effect relationships have the greatest power. These facts not only provide a better understanding of the role of human capital in organizations but allow you to predict the impact of decisions about human capital. Here is an example: “If we redirected bonus pay from group performance to individual performance, the likely result would be a 6 percent increase in workforce productivity and a 10 percent reduction in the turnover rate of high-performing employees.”

Figure 2-1: Some Facts Are More Powerful than Others 2003, Mercer Human Resource Consulting LLC

Establishing causality is always a challenge and seldom assured solely by means of statistical analyses. At a minimum, three conditions must be met to infer causal relationships among variables. The first is simple correlation. The variables must move together in a consistent, orderly way, as occurs when increases in labor productivity are associated with increases in target rates for incentive compensation. Second, changes in the presumed causal factor must precede changes in the outcome variable (i.e., when a change in bonus eligibility occurs before increases in productivity are observed). Otherwise it is plausible that the chain of causation is actually reversed or reciprocal. Productive organizations tend to have more resources at their disposal for generous pay and benefits. Thus, increased productivity may account for higher bonus payments, not the other way around. Understanding the chronology of events is essential in discriminating between cause and effect.

Third, it is necessary to take account of other factors that affect the outcome of interest, factors that may be changing simultaneously with the variable of interest. It is known, for instance, that general economic conditions affect economic productivity and that the correlation between the two is strongly positive. Thus, an observed increase in productivity could well be attributable to better macroeconomic conditions, not to changes in incentive compensation.

In attempting to establish causality, therefore, it is imperative to introduce appropriate “control” variables in any statistical analysis and to do that on the basis of a solid theoretical or empirical understanding of what drives the outcomes of the interests—in this example, workforce productivity. Such knowledge about human capital has grown enormously during the last several decades as a result of research in the fields of economics and organizational psychology. That research has provided a firm foundation for empirically evaluating hypotheses about causal relationships. It provides the basis for selecting appropriate control variables in statistical models to better evaluate whether and how particular human capital factors are influencing outcomes that are of interest to managers. Thus, the “facts” one seeks can be generated through a disciplined process of hypothesis testing. To return to our earlier example, a plausible theory of how tighter labor markets influence pay and employee turnover can enhance the predictive power of a model that relates productivity to bonus pay. Theory-based statistical controls are far superior to the unreliable “data-mining” approach sometimes used in making decisions about human capital.

Several dimensions of facts affect their power to inform. This chapter introduces the four dimensions of facts that are especially pertinent in making decisions about human capital: what we call say versus do, time, magnitude, and inside-outside balance. Each is illustrated by a brief case study.

^[1]David Matheson and Jim Matheson The Smart Organization. Boston: Harvard Business School Press, 1998, 6.