Earnings and business activity reports, as well as quarterly and annual financial reports, only rarely separate the impact of exogenous shocks to the company from its actual performance. Exogenous risk factors are many: the overall economic climate, the price of raw materials, calendar effects, etc. A growing number of companies have also become aware of the significant impact of weather on their business activity. Without claiming to predict the exogenous factors, explicitly accounting for them in budget scenarios easily allows a company to separate out questions of company performance, management decisions (marketing, pricing, etc.), and the unforeseen development of exogenous risk factors. The approach contributes to better progress reports as well as better management: identification of business unit performance, management decisions, and, potentially, hedging strategies.
Exogenous Factors or Risk Factors ____________________________
Whether we are talking about turnover and company operating income or about different decentralized profit centers, it is only natural for executive management to know their fundamentals.
Description of the company’s business model entails nothing more than this fundamental analysis. When available, factors (“drivers”) characteristic of company demand are carefully studied. Those affecting supply are also important in certain business models.
Factors may be quite disparate; most often, it is preferable to establish a composite indicator that summarizes the information rather than to describe and model demand too precisely. For it is especially important to be able to follow the indicator over time without waiting for all the information to become available1. In this way, when the customers are individuals, indicators like the unemployment rate or consumer confidence can “composite” more structural information based on household revenue or consumption and, more specifically, information from the segment concerned (socio-professional group, geographical area, etc.).
Other factors affect demand, often in a quite mechanical way: calendar effects for individual customers (number of days open each month, timing of school holidays in the year, etc.). Another kind of exogenous factor is increasingly followed and calculated: the weather (temperature, rainfall, sunshine, etc.).
The effects of each of these factors are uncertain, and they are themselves for the most part difficult to predict. They represent demand risk factors. Furthermore, demand may be affected by aspects of company management (marketing expenditure, communication, discount amounts, etc.) or, directly, by pricing policy.
Moreover, company business activity is sometimes determined by supply. Its production capacity may limit business activity. Checks on a factory’s capacity or on the supply chain or, in the service industry, constraints on personnel may come into play.
A company may thus see its expansion checked by the diminished demand or by its own capacity constraints. These different growth “regimes” sometimes arise as a function of the overall economic circumstances in which the company finds itself or as a function of company history. Income items (GOP, EBITDA, etc.) are affected in any case by factors stemming from supply and demand, along with additional factors arising from costs (input prices, currency, interest and tax rates, etc.).
It is at once easy and advantageous to account for changes in turnover, profit and the different components of business activity in terms of changes in the company's fundamentals represented by these different risk factors. Such an approach aims at calculating the contributors to performance and at making reports clearer. If the budget is in line with the hypotheses of evolving fundamentals, then the performance gaps between realized and budgeted figures can be calculated and analyzed through “performance attribution.”
(1) Of course, it is as important to carry out precise analyses, although the latter should be done less frequently, on an annual basis, for example, and not as part of ongoing infra-annual monitoring.
Business Reporting and Performance Attribution
________________________________________
Suppose that company turnover is a function of two exogenous factors:
> The overall economic climate (consumer confidence index); > The weather (temperature, for example);
and of a management variable (measured by expenditures, or by the amount of price discounts accorded, etc.).
Turnover is written: according to the values taken each month by the three explanatory variables of the turnover. Often, the chosen function is simply linear (see below).
Analysis of month over month turnover is written as a first-order approximation:
Of course, rarely do the three variables explain all the changes in turnover. Other factors – in particular, internal company performance (team performance) or, simply, chance – end up filling out the picture every month (we will return to this below). However, analysis of business activity is in this case simply interpreted – with orders of magnitude – in function of contextual developments and of company decisions.
Let us assume that, knowing this function of turnover, the budget has been created by defining trajectories for the three explanatory variables for the 12 coming months. Let be the monthly forecast retained for the budget, while the function CA = CA ( . ; . ; . ) stays the same. Performance attribution will in this case consist in considering the variation observed in turnover from one month to the next, in comparing it to that anticipated in the budget, and in explaining the difference with the following formula:
According to the type of function and the significance of “forecasting errors,” the approximation leaves a larger or smaller residual term, and one may then seek to determine to which “points” the partial derivatives apply.
In the majority of cases, most of the performance gap can be explained by deviations in budget forecasts of the exogenous variables (the context) and the decisions taken by the company (possibly in reaction to changes in the context). The deviations are weighted by partial derivatives and represent different “effects” that have an impact on business activity and “explain” over- or underperformance.
Analysis and Forecasting: Econometrics
_________________________________
Supply and demand functions are often complex. Management control mechanisms are increasingly sophisticated. The goal here is not to take aim at complexity. On the contrary, the goal is rather to identify the principal factors at stake for the company. Whether other factors affect the company doesn’t matter, as long as they don’t finally have an underlying impact and the company’s history allows one to believe that they will cancel each other out.
Certain econometric techniques (see Annex), simple to use but particularly theory-based, enable one2 to find the happy medium between relevance and efficiency. Econometric estimates are not intended to aid in forecasting. A temporal scenario for each of the explanatory variables simply ensures the coherence of the turnover and income budget scenario. In certain cases, it is possible to use these models in forecasting. In any case, they enable risk analyses through simulation3.
(2) The same techniques that enabled C. W. J. Granger to win the Nobel Prize in Economics in 2003!
(3) See the article “Cash at Risk,” p. 17, or our article, “Risk Budgeting and Managing through Risk.”
Performance and Management
_________________________
Often, certain variables appear clearly explanatory (expenditure, marketing, pricing, personnel, etc.). In this case, it is possible, over the course of the year, to implement responses to developments in the overall economic situation. Operational management is one of the sources of company performance. Certainly, as far as competitiveness is concerned, there remains everything that will never be measurable and that depends on the intelligence and effort provided by employees.4 This last point is of utmost importance with regards to managing different units of a company, beyond the mechanical outcomes anticipated in the budget.
The forecast error is a good indicator of internal performance. It includes, of course, other non-measurable exogenous factors, but it can be normalized and should serve as an alarm should thresholds be exceeded.
(4) See the literature regarding Principal/Agent relations and our articles on “Management Decentralization.”
Planning and Managing ___________________
Budgetary management is first of all a planning then a management exercise. It is often part of a long-term vision for the company, but it does require communication – with shareholders, markets, employees – which should be handled infra-annually: on what bases are the budget hypotheses constructed? Why is the company ahead/behind on its operating schedule?
Similarly, operational management must measure company risks5 and account for company performance. The approach proposed in this article is easy to implement. It can help strengthen – without requiring an overhaul of – the reporting tools that are already a part of management control. It opens the door to company management truly attuned to performance and risk.
(5) It might in certain cases choose to hedge them (climatic derivatives, for example).
Econometric techniques, to which we turn here, rely on the identification of a “long-term relationship” between variables of interest (turnover, net pretax earnings, etc.) and risk factors. This linear relationship is written6 in our example as a function of turnover:
To make things simple, this function is valid if variable Z is “stationary,” which means that turnover changes over time “around” its underlying “value” represented by
Z is not a “white noise,” although the sources of the deviation are sufficiently offset over time so that there is a “restoring force” toward underlying equilibrium. This is written by way of a second relationship reckoned econometrically, called a “short-term equation”:
Such models are called “Error Correcting Model”
The Δs symbolize the differences between dates. The “restoring force” or “Error Correction Mechanism” is expressed by the last term; a positive Z (in t-1, turnover is above its underlying value) will negatively impact variation in turnover in t.
It should be noted that the coefficients are not necessarily all significantly different from zero: in order to simplify things, we have not included time lags in the explanatory variables; occasional indicator variables (specific events in a given month) or more structural7 indicator variables may have to be included. Of course, it is best to have a certain amount of experience in order to determine the right specifications.8 Still, the required tools are those of least ordinary squares regressions, available in Excel!
Yet more sophisticated specifications can improve on this linear version of things, particularly through time-varying-coefficients or even as dependant on other exogenous factors. These enhancements are nonetheless only rarely necessary to start off with.
(6) Most often we are dealing with logarithms of variables so that the difference between two periods can be interpreted as a growth rate.
(7) Non-stationary econometrics with structural breaks identifies the existence of several “regimes” in the dynamics of variables of interest (See Bruneau, C., Duval-Kieffer, C., Nicolaï, J.-P. “Managing Funds in the US Market: How to Distinguish between Transitory Distorsions and Structural Changes in the Stock Prices,” European Journal of Finance, 6, 146-162, Spring 2000).
(8) Selection of variables, number of time delays, frequency, etc.
Print
back to articles list
according to the values taken each month by the three explanatory variables of the turnover. Often, the chosen function is simply linear (see below).
be the monthly forecast retained for the budget, while the function CA = CA ( . ; . ; . ) stays the same. Performance attribution will in this case consist in considering the variation observed in turnover from one month to the next, in comparing it to that anticipated in the budget, and in explaining the difference with the following formula:
