Forecasting in Times of Uncertainty: It Can Be Done

The Covid-19 emergency has caused a strong change in the business models of companies in many sectors. For some of them it has simply been an accelerator, for others something absolutely new. It is inevitable that, faced with this path of evolution, control systems, that have the goal of supporting decision-making processes and orienting management behavior, cannot be left out of this change.

In particular, in a context characterized by a high level of discontinuity with respect to the past and by high levels of uncertainty, above all traditional tools for forecasting results and setting goals are called into question. Specifically, we refer to the sum of tools that support planning cycles: budgeting and forecasting.

With reference to those systems, the current emergency has accelerated some trajectories of change already present before the crisis, that seem appropriate to pursue with conviction in order to guarantee greater efficiency for company forecasting processes. In particular, there are three dimensions - that are strictly linked - in regard to which it is necessary to rethink traditional systems for forecasting results:

1. the time horizons, to be understood both on a structural level (the forecast horizon) and in terms of the process (the forecasting cycles);
2. the measures with respect to which forecasting efforts are to be focused;
3. the "organizational" role assigned to the forecasts, to be understand first of all in terms of the goals attributed to them.

As regards the first aspect - time horizons - the uncertainty that characterizes many of the competitive situations in which companies operate determines the impossibility of forecasting results according to traditional multi-year time horizons (from three to five years). Planning based on excessive time horizons de facto risks being costly on the one hand, and not very useful, on the other. It would be preferable to construct forecasts around alternative scenarios, updating the forecasting logic as a function of the evolution of the situation.

This is the viewpoint in which we suggest faster and more responsive forecasting processes with respect to the traditional long planning and budgeting cycles. That need for responsiveness inevitably has consequences also on the level of depth of the forecasting: to paraphrase a famous motto dear to the supporters of lean management strategies, in this case it appears appropriate to say that less (detail) is more (precision).

This leads us to address the second dimension of planning: the choice of the measures on which to focus forecasting efforts. In this regard, traditional forecasting systems have often been open to criticism due to their excessive concentration on economic-financial results and little emphasis on the underlying determinants. In order to orient plans of action, though, in situations of discontinuity of the competitive and/or organizational context, the determinants of the results are precisely what must be the object of forecasting, while the results become a mere logical and algorithmic consequence of the same. In contexts characterized by high levels of uncertainty, the most critical determinants to monitor and foresee are external: environmental factors and their evolution facilitate or make it difficult for the company to act, and thus forecasting efforts must be concentrated there. To the contrary, the structural and internal functioning determinants, on short-term time horizons, act principally as a constraint, since they are difficult to modify with extreme flexibility. This leads to the necessity to orient forecasting systems more towards measures and variables external to the company, in antithesis to the way forecasting is still conducted in many businesses, i.e. starting from past results and simply assuming rates of growth/contraction with an incremental logic.

Finally, the changes commented on above imply a rethinking of the very purposes attributed to forecasting systems in particular, and to the company control system in general. First of all, the main goal of those systems cannot be the static forecasting of results, but rather the activation of learning cycles within the company. The greater attention to the determinants of results inevitably generates a greater capacity for the company to make explicit the relationships between actions (and the related efforts, in terms of use of resources) and their impact on results. The comparison between forecasts and what actually happens, both on determinants and results, increases management ability for understanding. Secondly, the changes described implicate a change in the style of control adopted by many companies. Traditionally, it is defined as planning influence or control influence, depending on the strength in orienting management decisions that is expressed in the phase of defining goals or that of measuring and assessing results. Based on what has been described above, a planning influence style that is characterized by the definition of specific goals and the use of long and detailed planning and budgeting processes does not fit well with the needs of speed and responsiveness that forecasting systems require. The true moment of learning is de facto during control of the results.

As already mentioned, the health emergency is an important accelerator of the processes of evolution of results forecasting systems, but the systems represent trajectories of evolution that, if pursued, can allow companies to have more effective systems not only in periods of extraordinary turbulence such as the current one. Traditionally, planning and control systems have been criticized for the strong delay in responding to the stimulus deriving from the introduction of management innovations or changes in the competitive context in which companies operate. The adoption of forecasting tools and processes that are faster, more dynamic and more attentive to the determinants of results can certainly contribute to overcoming this limit and allowing companies to have operational mechanisms able to effectively link the company's strategy, actions, and results.

Gianluca Meloni is Associate Professor of Accounting and Control Practice at the SDA Bocconi School of Management; Marco Morelli is Associate Professor of Accounting and Control Practice at the SDA Bocconi School of Management and Lecturer in Planning and Control at the Department of Accounting of the Bocconi University.