Performance feedback plays an important role in indicating when a firm needs to change its management strategy. It doesn't, however, indicate just what this new strategy should be, and firms do not always respond appropriately, says Henrich Greve, INSEAD Professor of Entrepreneurship and Organisational Behaviour.

Performance feedback indicators are based on the goals and aspirations that managers set for their firms. They get those usually by looking at how other firms are doing firms they feel similar to, for example in the same industry or by looking at their own record ... And so there are expectations, both for internal and external reasons because their audience is looking at how well they're doing, analysts and so on. It's a clear signal when the firm is falling below these aspiration levels ... And that means something has to happen, says Greve.

But in terms of answering the question 'what happens now?' that's where things get complicated.

When a firm fails to meet its goals and aspirations, senior management may become more open to taking greater risks, potentially leading to greater innovation, says Greve, citing the example of Nichia Chemical. It's the firm that nobody thought would be able to make the blue diode that much larger firms had tried to make and failed. Having previously entered several markets with semiconductor products, Nichia had been beaten back by larger firms in every instance. The turning point was when, at the brink, they took the huge risk of investing in furthering the research already abandoned by larger firms. There were two ways to make the blue diode that were known for scientific and engineering reasons. One of those ways just looked so much more likely to succeed than the other one. Well, Nichia picked the other one and it turned out to work, against expectations.

Risk-taking can, however, lead to recklessness. Greve warns that performance feedback needs a complementary system of checks and balances. Without some checks and balances, you can have managers pick things that prior analysis says the expected value looks negative but if you're really lucky you can overcome the problems. Should you be taking those kinds of bets? Under most circumstances, no. In fact, a year ago, the same management team wouldn't have taken it, and then suddenly it becomes okay; and then you wonder if that's really a good decision or not. One antidote to excessive risk-taking, according to Greve, is board oversight.

Willingness to take risks can vary by firm size. While Nichia was a small firm which took a great risk, many small firms are not similarly inclined. Large firms are more likely to take risks because they feel further from the failure point. When you're close to the failure point, you want to take risks because you're not doing as well as you thought you should be, but you're also close to the place where too much risk can lead to firm failure ... In fact, small firms can really be caught in a trap because they have nowhere to go. To go up, you should be taking risks, but you can't really handle the risk.

That said, smaller firms such as Nichia may actually be more innovative. Smaller firms face greater variation and find themselves performing below the level they aspire to, more frequently than larger firms. This situation, says Greve, creates a constant challenge there's always some little firm out there that needs to improve its position, and that is going to try and use innovation or other strategic moves.

Research has also shown that instead of leading to innovation, risk-taking can sometimes involve fraud. There are some gray zones there, but performance below the aspiration levels seems to drive accounting restatements, and of the bad kind. Instead of improving the actual performance, the firms tried to report a performance level higher than the real one.

Another form of 'overresponsiveness' is when companies receiving positive feedback cut back on investment and innovation, and risk falling behind those firms that have increased investment after receiving negative feedback, says Greve. Firms with low performance will ratchet up investment in facilities, they will speed up their product development, but just looking at their current performance you might suspect that they would be less successful in any of those activities than the firms that are actually doing well. And so it makes sense from the viewpoint of those firms trying to catch up, but then you do wonder about the others that are doing well: aren't they neglecting a push to exploit the advantage that they appear to have?

Conversely, Greve says that companies may not be responsive enough if they don't trust the numbers. Larger companies that have a long history of high performance face a 'paradox of success'. It's as if the record of being successful is so long and so convincing that they look at how poorly they're doing and they just can't believe it, Greve told INSEAD Knowledge. It's not certain how long this lasts, but it certainly seems to be long enough for firms that are leading in an industry that's undergone regulatory change, for example, to completely lose their leadership and become another medium size, not very successful firm.

Some key things should be kept in mind when designing a performance measurement system, Greve says. It's important for firms to understand that it's a warning tool and a search tool: it warns when something isn't working well; it's a way of searching for exactly where the problem is, and I'm not sure that traditional budgeting and accounting systems can do that necessarily ... There seems to be not enough of an engineering way of thinking about the firm as a number of moving parts, where you have to look at the health of each part, plus how they add up. Being able to assess firms in such a way is crucial, because, Greve says, even when performance management systems are not designed this way, that is how they are used when there is a problem.

When there's a crisis that is how they're going to use it, whether or not it's designed that way, so when the design of the performance measurement system does not fit this use, it complicates the search for the source of the problem.

SL 09/08