Managerial Accouting - Activity Analysis, Cost Behavior & Cost Estimation

Profit planning has two main question: (1) how much sales and (ii) what cost? First question can be answered by simply multiplying Quantity with Price. Second question is a bit cumbersome to answer as firstly (i) cost data would be analyzed to ascertain the cost-behavior ( fixed, variable or mixed), (iI) the cost behavior would be reduced to mathematical formula for each cost item, and finally (iii) cost prediction would be made at X level of activity.

This hub would cover cost estimation (cost analysis), cost behavior and cost prediction. The following behaviors have been identified and briefly explained:

Costs which vary directly in proportion to change in a certain activity. For example, as number of units produced would increase so would be the direct material & labor costs. As sales increase so the sales commission which is based on certain percentage of sale say 5%.

If a company want to decrease its variable cost, it should start from reducing wastage as far as possible. Further decrease can be made through introduction of modern technology through size reduction ( as in printed circuits ), switching over to cheaper raw material (paperboard from garbage) or recycling to reduce waste to zero.

Unlike Variable costs, fixed costs do not change with any movement in some other activity. For example, lease rental would be same whether a plant is operative or not, Similarly, salaries of permanent staff, depreciation on straight-line methods, plant insurance etc would remain the same irrespective of capacity level or capacity utilization or production efficiency.

Fixed costs are becoming more prevalent in many industries because of automation. Modern technology enable a plant to reduce to a negligible amount and to use cheap materials.  A plant in Pakistan was successful using river-grass for the production of fine paper. It was later closed down due to paper dumping by some industrialized countries like China. Also, the automation is replacing direct labor with  engineers which are mostly permanent. Direct labor is gradually become non-existent in capital intensive industries.

There are certain costs which are partly fixed and partly variable. These can be described as semi-variable costs or semi-fixed cost or even mixed costs. All utility bills have a fixed factor and usage factor. One would receive a telephone bill even if not a single call was made during a particular month. A club would charge a minimum subscription even if one does not enter the club for months together.

In industrial units, there are setup costs whether 10 items are produced or 1000. Semi-variable costs are split into two component i.e. variable and fixed.  So ultimately, only two types, variable and fixed, would emerge.

In some cases, costs increase in small steps. Teachers salaries is a step-variable function. If it is decided to keep the class size limited to 35, another teacher would have to be inducted when number of student increase beyond 35, Similarly, bus hire or compensation to be paid to invigilators would increase in steps.

Step-variable function deals with operational side. There are small steps which makes it different from Step-Fixed as explained in the next capsule.

Like step-variable costs, step fixed cost increase in steps but these steps are much bigger. In fact, step fixed cost relates capital investment. If demand for a product has increased, more capacity is added which would result in more fixed costs by way of depreciation and mark-up. Similarly, as number of students go up, the university would be hard pressed to construct additional rooms or even new campus.

The relevant range is the range of activity within which managers expect a company to operate and within which managers can predict cost behavior with some certainty. Within the relevant range, even curvilinear costs may behave in a linear fashion. In a car, the owner changes the engine oil every three months only because the owner plans to drive 1,900 to 2,100 km a month and, let us suppose, engine oil is changed after every 6,000 km. This would fixed cost for the owner but not if the owner speeds up drive to 6,000 km a month or locks up the car for three months in a row. So variable and fixed costs classification is applicable if one remains within the relevant range.

Having ascertained the cost behavior, cost prediction can be made. Segregating into fixed and variable costs is a must before many cost prediction or forecasting techniques can be used such as (i) accounting method, (ii) high low method, (iii) linear regression and finally (iv) curvilinear regression. All these techniques are explained as under:

Annual growth rate is widely used in cost prediction as in large number of cases where the growth or decline is gradual. Population Growth is one example, inflation is another and annual production of citrus fruit is still another. If one has data for the past years, one can find out annual growth rate and work out what would be the level next or say after five years.

In the preceding capsule, past sales data is given. If we work out annual growth rate, it comes to 10%. (This has been explained in the side-chart. If we want to know what sales could be expected in 2009, the calculation would be:

153,731 (Sales in 2008) x (1+10%)=169,104

An EXCEL FUNCTION RATE (nper, pmt, pv, (fv), type, guess) would give the result as ACGR.

A case where cause and effect relation is certain, one can use High low method. Some arbitrary data is given in above capsule. There is apparently a correlation or cause and effect relationship between units produced and electricity charges. If we divide electricity cost by units produced, we can find out cost of electricity consumed for producing one unit in various years. If such a cost is identical, we would say that it is variable. If the unit rates is not uniform, the relevant cost is partly variable and partly fixed. The fixed part can be segregated and a cost formula can be arrived as:

Total Cost = Fixed Cost +(Numbers of Units Produced x Electricity Rate Per Unit).

Linear Regression is a powerful tool for forecasting. It uses a formula i.e Y = a + bX which is the same as described in High Low Method. The difference is that How-Low is a crude way which cares for only the highest and the lowest. The linear regression takes into account each and every figure, calculate average and standard deviations and only then processes it further.

Curvilinear is a much advance tool which takes into more than one factors. In my previous example, it was stated that electricity cost depends upon unit produced. We may introduce more factors like quality of units ( course, medium and fine) and capacities 2hp, 4hp and likewise

Let us taken another example: Suppose a university has enrolled 30 fresh students for 2-year MBA program. The university is interested to know what would be GPA of each student at the end. It would decided on factors or causes effecting GPA, the dependent variables and may would arrive at the following equation:

Y = a + bX + cW + dZ + E, where Y= GPA, bX= Weighted Average of Marks in Matric, Intermediate and Graduation, cW is marks in Aptitude Test and dZ is Interview score while E is error factors based previous actual GPA and predicted GPA.

TASK ANALYSIS

When past data is not available and some one wants to how how much 100 chairs would cost, one can resort to task analysis whereby a item would be broken into its components, each component would be measured and its cost obtained from the market sources. This is called bottom up approach. (On the contrary, in bottom-down approach, an estimated is obtained from experts to total budget and later is it is distributed to various factors. One may want to know how much a university for 2,500 students would cost. The experts can tell us that it would cost around Rs. 350 million. This amount can be split into land, building, laboratories, libraries and play grounds etc based on past pattern or proportion ( land to be 20% of total cost) to be given by the same experts.

ACCOUNTING CLASSIFICATION METHOD

An accountant studies the past records and classifies each cost component into fixed, variable, semi variable, step fixed and step variable. One this done, fixed component is extracted from semi-variable, step-fixed and step variable. This leaves only two components i.e. fixed and variable. Later, same approach as in Linear Regression can be adopted.

VISUAL FIT METHOD

This is graphical presentation where a pattern can be noted and extend for prediction.

An understanding cost behavior helps manager in anticipation of changes in cost when there is a change in their activities like production, sales, inventory pile up etc. It provides good assistance in planning, cost management and decision making.

A number of behavior patterns exist ranging from fixed to variable and from linear to curvilinear. Many cost predictions techniques are used to forecast sales and costs or any other factor.