March, James G. & Shapira, Zur Managerial Perspectives on Risk and risk Taking, Management Science, 33:11 Nov 1987 p.1404 - 1418.

This paper compares decision theory with risk attitudes actually held by managers. They conclude the managers use different rules for risk decisions than standard decision theory. Managers are usually unaffected by estimates of probabilities of possible outcomes, but more on their attention to critical performance targets. They make a sharp distinction between taking risks and gambling.

1. Risk as a factor in decision making.

Definition of Risk

In classical decision theory, risk is the variation in distribution in possible outcomes, likelihoods, and subjective values. Risk is also a non-linearity in the perceived utility of money. or Variance in distribution of possible gains or losses associated with a particular alternative.

Risk is thus seen as a choice based on the expected value of return of an alternative. Standard theories assume decision makers prefer larger expected values.

But real people tend to ignore possible events that are unlikely or remote, regardless of their consequences. p. 1405 They look at only a few alternatives and make estimates from them. They are more comfortable with verbal characterizations of risk than numbers. Likelihoods and values may be considered separately rather than as their product. p. 1405 Different people will evaluate risk differently.


Attitudes Toward Risk

Most behavior theories see people as risk-averse. They avoid situations of high variance, even if the expected value is somewhat higher. But it is not certain the managers consider risk and return as highly corelated

Standard theory assumes risk-preferences to be relatively stable and factor of personality. Possibly a person's mood or the way in which problems are framed has an impact. p. 1406. Kahneman and Tversky found that if the outcomes are generally good people are risk averse but if they are generally poor they are risk-seeking (willing to take more risk to avoid a really bad outcome). Thus risk behavior may be context-dependent.

Risk taking also doesn't seem connected to adversity (i.e. lower-success people are willing to take more risks).

Dealing With Risks

In conventional theory risk averse people avoid variation and are willing to sacrifice some expected value to reduce variation. Risk seeking people prefer relatively high risks and are willing to sacrifice some expected value to increase variation. The theories assume people first calculate the expected returns.

But empirical studies have shown that people ignore normal decision calculations. Sometimes they deny risk. Sometimes they accept the risk but don't believe it applies to themself. It may be they reject the probabalistic reasoning for a single event or actually have a stronger belief in the causal basis of events instead. Many studies have shown people to assign causality to random events.

2. Managerial Perspectives

The definition of risk

Managers don't equate risk with variance in potential outcomes. Three obvious differences are:

1. Managers don't treat uncertainty about positive outcomes as an important aspect of risk. Risk is seen with negative outcomes only. Risk is more seen as a danger or hazard.

The managerial definition is that a risky choice is one that contains a threat of a very poor outcome. p. 140

Risk is also not a probability concept for managers. While half saw uncertainty as a factor in risk, the magnitude of bad outcomes was more important. They thought risk is better defined as the amount potentially expected to lose. "I take large risks regarding the probability but not the amounts". This is probably better called loss aversion (Kahneman & Tversky) or regret aversion. It's a tendency for less risk taking when greater stakes is involved. They looked for the "worst case scenario" and didn't use statistics in their evaluation.

Managers don't quantify risk in alternatives, but more get a "feel" for it. They feel risk can't be captured ina single number. p. 1408

Attitudes Toward Risk

Managers tend to think that risk is a preference based on one's personality. They see incentives and norms as influencing risk taking even more. Middle level managers said riskiness disappers as you move up the hierarchy, while upper level managers feel that risk taking is important and want to encourage lower level managers to take more risk. Higher level managers score higher on risk taking than lower level managers.

Managers also see risk taking is sustained more by personal incentives, and that organizations generally inhibit risk taking. Most of their risk advice was conservative. "Don't gamble" "Get others to particpate in your decision". This is common among managers who don't see risk as uncertainty.

However, most managers saw themselves as higher risk takers than others. They also feel they are bigger risk takers than they really are. They justify a particular choice if big potential losses are balanced by big potential gains. p. 1409

There are three motivations for risk taking:

They first believe that risk taking is essential to being successful.

They believe risk taking is part of being a manager. It's an affirmation of the role. They are more likely to take a risk when framed as a business decision than as a personal decision. p. 1409

They also see risk taking as thrilling, involving danger.

Managers also feel they take different risks depending onthe situation. They believe that fewer risks should be taken when things are going well. Risk taking is affected by the current position and critical reference points.

They compare their performance versus an aspiration level. They take more risk when below aspiration level than above it. But they also feel that organizational survival should not be risked. They tend to risk more when they have more assets to play with. Risk seems to depend on which of two targets the person is focusing upon.


Dealing With Risk

Early studies show managers tend to avoid risk. They avoid risk by not anticipating future risks (look at short-term), negotiating uncertainty absorbing contracts, and delay and delegate risk decisions. p. 1410.

They also sometimes see risk as something they can control. They believe their skills can help reduce risk. They believe risk is manageable. They see a big difference between risk taking and gambling. They try to reduce the potential for failure without sacrificing the potential gain. They try to change the odds. They look for risk controlling strategies.

Most managers believe they can do better than expected, by "eliminating the unknowns" or "controlling the risk". Managers accept risks because they don't think they will have to bear them. p. 1411


3. Implications for Understanding Risk Taking By Managers

There are three main differences between manager risk taking and theory:

1. Insensitivity of Risk Taking to Probability Estimates

Managers don't look at probability statistics. They ignore unlikely outcomes regardless of high negative significance (like potential for accidents). Thus they are persistently surprised and unprepared for low probability events that come true.

They judge risk by outcome value than by likelihood x value. Thus they accept greater risk when the probability distribution of outcomes if more rectangular than with long tails.

Their "confusion" echoes the literature, where risk is increasingly seen as a cost in engineering terms. Risk then becomes a hazard p. 1412. It's also because measuring risk is very difficult. It's hard to determine possible outcomes and estimate values and likelihoods. In fact, estimates of likelihood are seen as more uncertain, so they focus on outcome value estimates instead.


2. The importance of attention factors for risk taking

Risk preference varies with context and aspiration level. Risk taking stems from a change in focus among a set of inconsistent and ambiguous preferences. Focus is shifted away from the dangers involved and toward the opportunities. p. 1412

People can only focus on a few items at a time. Choice behavior is driven by attention instead of preferences. Order presentation and agenda effects become an issue as well. The aspiration and survival levels also define success/failure points that aid in attention focus. They define success, failure, and extinction. p. 1413

When above the target, people want to avoid going below it and are more risk averse. Thus successful managers just above their target are more risk averse. They take more risk only when they are far above their target.

Those just below their target take more risk to get above it. Their desire causes them to focus attention on theopportunities than the potential losses. However, if they are near the survival point they are much more risk-averse.

Risk Taking, Gambling, and Managerial Conceit

Interestingly, risk taking is both desired and yet differentiated from the gambling that it really is. Good outcomes are rewarded, not good decisions. They do it because society rewards outcomes not decisions and that experience leads them to believe they can control fate. p. 1413

Gambling becomes risk taking that turns out badly. "You have to be a risk taker, but you have to win more than you lose". Post-hoc explanation events allows chance to be removed from consideration. Risky choice that turn out badly are seen as mistakes.

The experience of successful managers tells them that the probabilities of life don't apply to them. Thus they are more prone to accept risks than they would otherwise be. p. 1414. Managers are expected to take "good" risks. Managerial ideology portrays good managers as risk takers. Managers come to believe they can tell good outcomes from bad outcomes and to "manage risk" to improve the odds.


4. Conclusion

Thus it might be more effective to affect risk taking behavior by changing attention behavior than probability estimation ability. p. 1415 Training in decision theory may also be useful. But the risk behavior is embedded in society norms and values. There is also a danger that such training might make decision makers more passive. We might be better off having managers think they can control destiny.