Linking financial data to other measures the next necessary step
The process in which companies monitor and report their performance is due for an injection of analytics. Over the past 15 years, the most important innovation in this process has been the balanced scorecard, popularized by Robert Kaplan (http://en.wikipedia.org/wiki/Robert_S._Kaplan) and David Norton (http://www.juergendaum.com/news/07_18_2001.htm). This innovation consists of reporting multiple nonfinancial performance indicators along with financial performance indicators on one sheet or screen. The idea of “balanced” suggests that all indicators are of equal importance, although this is seldom true in reality. Financial measures still receive much more attention, both inside and outside of companies.
Many corporate reporting programs today describe results on financial and nonfinancial performance metrics, and often present key measures in a unified scorecard or dashboard format. However, the particular nonfinancial measures that are presented are usually chosen arbitrarily, and the quantitative relationships among nonfinancial and financial performance factors are not understood. That is, companies have no idea whether such metrics as employee engagement, customer loyalty, and innovation are truly driving or even related to their financial performance metrics such as revenue growth, earnings per share, and stock price. In short, their scorecards are essentially arbitrary, as academics Chris Ittner and David Larcker argued in a Harvard Business Review article. (http://maaw.info/ArticleSummaries/ArtSumIttnerLarcker03.htm)
I see a time in the near future, however, when analytics will inform scorecards. In this situation, companies would relate nonfinancial to financial metrics using statistical “path models” that incorporate input, output, and intermediate variables, with quantitative path coefficients for each significant relationship. The path models would be based on multiple regression analysis or a more recent variation on it, structural equation modeling. Kaplan and Norton have described this step as “testing” the balanced scorecard, though they haven’t talked about it much.
If such a statistical model were available, companies would be able to state with confidence that certain factors in their business are reliable drivers of financial performance. They would be able to take such an explanatory model, and also use it for prediction of future performance. If key nonfinancial indicators showed declines, the companies could perhaps intervene to address the declines before financial performance suffered. They could begin to report to external bodies—Wall Street analysts, for example—their nonfinancial performance variables, which would be useful in predicting future stock prices. Over time, external regulators might require that these nonfinancial variables would be broadly reported so that investors would also be able to predict future performance.
Some of this has begun to happen, though not enough. In my next post I’ll describe a few early adopters.
