Printer Friendly

Managing meal costs: Variance generation, analysis, and interpretation.

Times are tough in the restaurant industry. Because of the weak economy, the number of people eating out has fallen at the same time that the costs of producing and serving meals have increased. To coax more customers into dining away from home, restaurants are exploring and implementing marketing and promotion efforts. This article, however, will focus on meal costs by examining an integrated cost control analysis.

First, we will explain the traditional cost variance framework, then apply it to the control of labor and food costs in a restaurant operation. Next, we will discuss the considerations in setting cost standards, illustrate calculations of cost variances, and propose possible interpretations of the calculated variances. Throughout the effort, Legends of Notre Dame Restaurant and Alehouse Pub (LEGENDS) serves as the primary example, and there are five supporting examples.

Cost variance analysis begins with the accounting processes of determining theoretical costs, setting cost standards, collecting actual costs, and ending with evaluating performance. Properly determining theoretical costs in the restaurant business involves setting standard recipes and preparation procedures for every menu item. Setting cost standards is the next step in this process. The standard or expected cost is the total cost that should occur for an actual level of activity within the restaurant. An example dealing with Irish Nachos at LEGENDS will reveal in greater detail that standard costs vary from theoretical costs as the result of a certain degree of unavoidable occurrences in the kitchen.

The analysis rounds out with collecting actual costs and evaluating performance evaluation. Expected costs result primarily through analyzing theoretical costs and looking at historical operations. Ideally, this deliberate process involves many people, including purchasing personnel, budget administrators, managers, and possibly kitchen or wait staff. Although actual costs are simply the actual cost of goods or labor during a certain period, collecting actual cost data can be quite difficult. We will discuss this topic thoroughly later. Finally, performance evaluation compares budgeted and actual costs during the period. We can separate the difference between actual costs and budgeted costs (i.e., variance) into subcomponents so we can pinpoint, explore, and address the causes of and responsibilities for the variances.


We developed a variance analysis model to evaluate costs that occur in the meal preparation process. To begin, we can express the actual cost of meal preparation, defined as the exact amount the restaurant expended, as the actual quantity of resources (AQ) times the actual per-unit cost rate (AR), or AQ X AR. On the other hand, the budgeted cost represents the desired cost of the preparation process given the actual number of prepared meals. In other words, this is the amount that the restaurant should have expended to produce the actual output. We can express this as the standard quantity of resources used (SQ) for the actual meal preparation level times the standard per-unit rate (SR), or SQ X SR. The total variance is the difference between the actual cost and the standard cost, expressed mathematically as (AR X AQ) - (SR X SQ). If the actual amount exceeds the budgeted amount, we label the variance as unfavorable, and we label the opposite result as favorable. See Table 1 and Example 1 for a summary and application of the traditional cost variance framework.

Example 1

Use the following values and the information from Table 1 for Brian's Better Burgers, which sold 1,000 half-pound burgers in the time period being examined:

Actual rate/cost          $5.95/pound for 580 pounds
Actual quantity           0.58 pound per burger
                          (i.e., 580 divided by 1,000)

Standard rate/cost        $6.00/pound for 500 pounds
Standard quantity         0.50 pound per burger

Total Variance

Actual (AR X AQ) or $5.95 X 580 = $3,451
Standard (SR X SQ) or $6.00 X 500 = $3,000
Unfavorable variance              ($451)

Rate Component

AQ (SR - AR) or 580 ($6.00-$5.95)   $29 Favorable

Quantity Component

SR (SQ - AQ) or $6.00 (500-580)   ($480) Unfavorable
Total variance                    ($451)

In a restaurant setting, the expected or budgeted amount in the variance analysis model is a flexible value that depends on the actual activity level rather than a static projection of costs based on expected activity. This flexible amount is part of the post-meal-preparation analysis and is a function of the actual activity level. At higher activity levels, the total flexible budget figure is larger, and it is smaller at lower levels. Example 2 illustrates the flexible budget procedure used in an analysis of labor costs.

Example 2

Kelly's Kitchen developed the following standard information for its upcoming year:

Standard labor cost: $25 per hour

Standard hours/meal: 12 minutes or 0.2 hours

The static budget amount to prepare 3,000 meals, the expected number for a particular time period, is $15,000:

3,000 X 0.2 hours/meal X $25

The restaurant served 2,750 meals during the time period. Based on the actual information, the flexible budget is $13,750:

2,750 X 0.2 hours/meal X $25

Clearly, the flexible amount differs from the static amount because the expected level of activity did not occur. Although we may use the difference between the static and flexible budget amounts to evaluate the budget forecasting process, we must compare the evaluation of actual operations through cost variance analysis to the flexible or recalculated budget amount.

The total variance (budgeted or standard costs minus actual costs) includes two elements: the rate or price component and the quantity component. The rate component finds the product of the standard rate and the actual quantity being calculated and compared to the actual cost of production. The formula for rate variance of (SR X AQ) - (AR X AQ) is shortened to AQ (SR - AR), as Table 1 shows. A rate or price variance results when the actual quantity of input is incurred at a rate or price that is different from the standard rate or price. Once again, if the actual cost exceeds the standard, the variance is unfavorable, revealing that the company spent more than it should have for this amount of input. If the opposite holds true, the variance is favorable.

The second element of the total variance, the quantity component, results from comparing a new, recalculated value of standard rate times actual quantity with the flexible budget amount. As Table 1 shows, we express this in a shortened formula of SR (SQ - AQ). This variance results from incurring an actual rate per unit for a quantity of input that differs from the expected standard quantity. As before, if the actual amount exceeds the standard, the variance is unfavorable, and the opposite is favorable.


To provide a more usable end result and to illustrate the ultimate goal of cost variance analysis within a restaurant setting, let's look at one menu item from start to finish and use that as a framework to discuss the overall restaurant operation. Once the LEGENDS chef establishes exactly how many ounces of chips, cheese, and other toppings go into an order of Irish Nachos, he/she will be able to properly calculate the theoretical cost of that particular dish. Note that the food cost does not, and should not, take into consideration indirect materials such as the oil to fry the chips. Table 2 illustrates that the food cost associated with making a plate of Irish Nachos is $2.87 under normal conditions in a perfectly run kitchen at LEGENDS.

The phrase "perfectly run kitchen" is obviously a major assumption. A restaurant kitchen will never operate with such efficiency that will permit actual food costs to align with theoretical costs. While a good kitchen manager will work to reduce the cause of these variances, there will inevitably be miscues in the kitchen (e.g., food will spoil, measurements will not be as precise). This is where setting expected costs comes into play. Developed primarily from historical data analysis, the expected cost will take into account inevitable operating situations that cause costs to fluctuate. For example, it is nearly impossible to measure out exactly four ounces of shredded cheese, and some tortilla chips in every bag are unusable because they are broken into crumbs. Careful observation of the Irish Nachos preparation and costing through the years has revealed that the expected cost of making Irish Nachos is $2.92, not $2.87.

Now that we have theoretical and expected costs for the Irish Nachos, it is time to determine the actual cost of the dish during a certain period of time-one month, for example. This step can be very difficult, especially if the recipes are complex or if a menu cross-utilizes many different items. We will discuss overcoming these complexities later. For simplicity's sake at this juncture, we will assume that LEGENDS uses the ingredients for Irish Nachos only and depletes inventory levels at the end of each month.

To calculate the actual cost of the Irish Nachos, LEGENDS takes the cost of all the ingredients (i.e., salsa, cheese, guacamole, etc.) the restaurant purchased during the given month and divides it by the number of Irish Nachos sold in that month. For example, if the cost of ingredients was $5,008 and there were 1,600 orders of the Nachos, the actual cost of Irish Nachos would be $3.13 ($5,008/1,600). This actual cost of $3.13 is higher than the theoretical cost of $2.87 and also the expected cost of $2.92, meaning there is an unfavorable variance.

An operator should be concerned with the variance between actual and expected. In this case, the Irish Nachos have produced an unfavorable variance of $0.21 ($3.13- $2.92) per order on average. Because customers ordered 1,600 servings of this appetizer, this equals a total unfavorable variance of $336 ($0.21 X 1,600) for the month. If the manager ignores the variance, this could create a negative variance of more than $4,000 for the year.

Naturally, the next step is to question why the variance occurred. Did the cost of cheese increase? Did a case of tortilla chips get smashed? Are the cooks not portioning the ingredients properly? The numbers simply become a tool that allows the kitchen manager to detect and address the problem. This is just one item on a menu containing more than 100 different food and beverage offerings, so it becomes evident why using cost variance analysis in a dining operation is important.


The first step in cost variance analysis is to set standards by which to judge performance. This is an essential element of the restaurant's budgeting process. There are three major aspects of operational budgets: planning, motivation, and evaluation. For planning purposes, the operational budget, or standard, should represent the expected, most likely outcome for an operation, but such a budget may not serve as an adequate motivator to those responsible for the budgeted activity.

When a standard is set, the operation gets a specific target. If this target is too easy to attain, the budget is inadequate for the purpose of motivation, so the planning and motivation roles conflict. A budget based on the most likely outcome will be exceeded half of the time merely by chance if the standards were determined in an unbiased manner. Such a budget is too easy to achieve. Potential conflict may also exist between the budget's motivational purposes and the budget's purpose of evaluation, which obviously happens after the operator has collected the actual results. It is not, however, as simple as looking at actual-vs.-expected results because there must be some differentiation between controllable and uncontrollable (i.e., the causes of a variance between actual and budgeted cost at the end of a period may be classified into two components-those causes that are and should have been controlled by the manager and those that are not). Unexpected inflation is an example of an event in the latter category that should have caused the standard to change during the period. The manager then should be evaluated on a sort of prorated standard to extract the uncontrollable components from the budget variance. At the same time, managers may not be totally committed to achieving budget objectives if there is some possibility that the standards by which they are judged may change. On the other hand, managers may become frustrated if they believe that uncontrollable factors may trump the effect of their actions on budget variances. Example 3 illustrates this type of situation.

Example 3

A LEGENDS manager knows that an upward spike in petroleum costs drives up dairy and fresh fruit prices, which generates higher costs than anticipated. The impact of the increased dairy and fruit prices more than offsets efficiencies the manager recently introduced in the kitchen operation.

For these reasons and to standardize the budgeting process to some degree, we recommend that restaurants use a statistical approach to setting the cost standards that will be used later in the cost variance analysis.

In developing labor and food cost standards, the restaurant manager should note that labor and food costs are by nature variable within a relevant range of activity. Labor costs vary with the number of labor hours consumed, and food costs generally vary with dollars of sales. Thus, calculating an average is a convenient and simple way to approximate the costs incurred per unit of activity. Averaging the 16 observations in the first two columns of Table 3 yields a standard labor rate of $8.534 per labor hour. Similarly, dividing total food cost by total food sales for the 16-week period yields a standard food cost of 28.44% of sales:

Standard labor rate

$8.534 = $131,522/15,412

Standard food cost

28.44% = $118,822/$417,738

Many costs, however, are not purely variable since they include both fixed and variable elements. In this case, a second statistical technique, linear regression, can approximate both the rate of variable costs and the component that remains fixed with the activity level. In developing quantity standards, the standard number of labor hours per customers served apparently exhibits such a relationship.

Figure 1 plots the relationship between the number of customers the restaurant served and the labor hours to serve them. The plotting reveals that labor hours do not relate to the number of customers served in a purely variable fashion. As a result, the line drawn to approximate the relationship between the number of customers and the number of labor hours does not begin at 0,0. In fact, calculation of the least-squares regression line, which is intended to make the best fit of the plotted points shown in Figure 1, yields the following estimate of the relationship: Y = 119.86 + 0.4631x. Spelled out, this means that the total number of labor hours for any given number of customers will be a minimum of 119.86 plus an additional 0.4631 labor hours for each customer served. In other words, it takes approximately 120 hours of direct labor simply to open the doors at LEGENDS, regardless of the number of guests.


The very basic Irish Nachos example poses many difficulties when trying to evaluate an entire menu portfolio made up of more than 100 menu items that cross-utilize ingredients at every turn. Furthermore, the example does not evaluate labor cost variances, which is where the pragmatics of cost variance analysis come into play.

Table 3 shows a weekly cost data sample for LEGENDS. With the information in the table, we can use statistical techniques such as averages or linear regression to create price and quantity standards. Once we develop historical standards, we can analyze labor and food costs by using basic statistical methods to develop standards for both. The rate standard for labor costs is the cost of using one labor hour, while the quantity standard is the number of labor hours to provide food and service for one restaurant guest. For food cost, the price standard is the food cost per dollar of sales, otherwise known as the food cost percentage, and the quantity standard is the dollars of sales that the restaurant should earn per guest. A restaurant operator should have or should develop the ability to capture this data from either a point-of-sale system or some other recordkeeping system (see Point-of-Sale Systems, p. 6).

Data that might be somewhat difficult to obtain would be the weekly food cost because restaurants usually maintain some food in inventory. To determine food costs, the most accurate method is to adjust the purchases for a period for the costs of beginning and ending inventory (i.e., Cost of Meals Served = Beginning Inventory + Purchases - Ending Inventory). The calculation requires taking a physical inventory or applying an approximation technique each time the analysis occurs, which is the most accurate and highly recommended way to calculate food costs. If inventory levels are stable, however, we can approximate food costs satisfactorily using the value of the period's purchases. Direct labor costs in Table 3 encompass the wages that direct labor employees earn and do not include indirect labor costs (e.g., salaries of the manager or chef). Presumably, the restaurant can control other labor costs (e.g., wages of cooks, dishwashers, and bartenders) by scheduling staff as a function of forecasted activity levels. In the case of a restaurant, the appropriate activity level to forecast is commonly the number of customers.


Now that we have established standards, we can effectively analyze cost variances for a set period. For example, we want to analyze week X, which Table 4 features.

Using the standards we already calculated and the data from week X, we can determine the labor and food cost variances in Tables 5 and 6.

Table 5 indicates an unfavorable direct labor variance of $369 that we can further break down:

Rate Variance = AQ (SR - AR), or 975 ($8.534 - $8.44) $88 Favorable

Quantity Variance = SP (SQ - AQ), or $8.534 (921.02 - 975) ($461) Unfavorable ($373)

Table 6 reports a favorable $72 food cost variance. The variance cannot be segmented into cost and quantity components because of a lack of information.


The traditional analysis of food and labor costs in a restaurant operation typically begins with determining the food cost percentage (food cost divided by food sales) and the labor cost percentage (labor cost divided by food sales). The more astute restaurateur realizes that, for the operation to be profitable, these ratios should remain and report at some optimal level, given the restaurant's specific characteristics and goals. Percentages above the optimum suggest a higher level of service or food quality, but also a degeneration of profits unless increased sales volume offsets the higher costs. Higher-than-expected percentages may also signal the existence of theft, excess spoilage, or generally poor internal controls.

On the contrary, lower percentages suggest less service, lower quality, and smaller portions, but higher profits unless sinking sales outstrip reduced costs. Restaurateurs are right to examine food and labor cost percentages, but examination at this level only fails to suggest or uncover possible factors that affect deviations from normal percentages.

Food Cost Variances

The food cost variance is the responsibility of the individual or individuals performing the purchasing function for the restaurant. Unfavorable cost variances result from paying more for food items than expected. An analysis must determine:

* Did the purchaser obtain the best prices?

* Did the purchaser buy better quality items than would be commensurate with the prices in the restaurant?

* Did the purchaser overlook discounts?

* Did avoidable rush orders cause higher costs?

* Did out-of-stock items necessitate substitution of better quality items to meet demand?

On the other hand, the purchaser may justify the variance as being a result of inflationary price increases or rush orders that could not have been foreseen.

Favorable cost variances result from paying less for food items than the restaurant anticipated. The manager should investigate significant favorable cost variances, however, to ensure that inferior goods are not being substituted because this practice may improve short-run profitability but have an adverse impact in the long run. If the cost variance is justifiable and continues, the restaurateur should alter the standard and take action to extend the favorable purchasing outcome to other items. As a final consideration, it might be worth passing some savings on to restaurant patrons in the hope of increasing volume.

Food Quantity Variances

The food quantity variance is the responsibility of the production personnel. Unfavorable quantity variances result from using more food than the standard allowance for the number of meals served. Possible causes include waste, spoilage, insufficient portion control, pilferage, and poor preparation techniques. The unfavorable result, however, might not be within the control of the production personnel. Perhaps an incorrect end-of-period physical inventory or an unfavorable food sales mix led to the quantity variance. As not all menu items are expected to result in the same food cost percentage, a greater ratio of high food cost at times results in a higher cost percentage.

Favorable quantity variances result from using less food than anticipated, given the number of meals served. This may be an unpopular outcome if the guest has received less than expected. Alternatively, if the variance resulted from production efficiencies, the manager should encourage such practices to not only continue but expand as well. If the cause is a favorable sales mix, the manager should take note of it. Example 4 illustrates the calculation and analysis of a food cost variance.

Example 4

Urlacher Fish Organization (UFO) operates a seafood restaurant where tilapia is a very popular item. UFO has established the following standards for each tilapia dinner served:

Standard cost         $5.80 per pound
Standard quantity      0.4 lb. per meal

During the past quarter, the restaurant served 7,500 tilapia dinners. Actual tilapia cost of $18,125 is based on:

2,900 pounds X $6.25 per pound

Thus, a total food cost variance for tilapia served was $725 unfavorable:

Standard 7,500 dinners X 0.4 X $5.80 = $17,400

Actual (above)               $18,125
                            ($725) Unfavorable

Analyzing the above variance into its component elements shows the following:

Cost/rate variance

AQ (SR - AR) or 2,900 ($5.80 - $6.25) = ($1,305) Unfavorable

Quantity variance

SR (SQ - AQ) or $5.80 (3,000 - 2,900) = $580 Favorable

Total variance               ($725)

Labor-Rate Variances

To the degree it can and should be controlled, the labor-rate variance is the responsibility of the person doing the scheduling. The standards in this area combine a number of different wage rates taken together to determine the average cost of labor to serve a customer (e.g., servers, cooks, bussers, bartenders, dishwashers). An unfavorable variance will result either from assigning higher-paid workers to perform tasks that lower-paid staff should do or from employing standard wage rates that do not reflect the current wage scale accurately. The manager can rectify the former cause by proper scheduling, whereas the latter should result in revising standards. The favorable rate variance may result when the manager schedules lower-paid employees to perform jobs that should command a higher wage rate or when rate standards are higher than the current employee mix suggests they should be.

Labor-Efficiency Variances

The labor-efficiency variance applies to the average quantity of labor to serve one customer. Theoretically, an unfavorable variance in this area is the responsibility of the scheduler or production supervisor. For example, for the number of customers served, fewer labor hours should have been required, or, given the number of hours worked, more customers should have been served. In a typical manufacturing setting, such an allegation could be made legitimately. The restaurant environment, however, poses a particular problem in this area because the purpose of production (food preparation) is to fill sales orders, not to produce inventory. Scheduling is based on expected demand; production is the result of actual demand. Unfavorable labor-efficiency variances are perhaps as likely to be the result of significant forecasting errors and, therefore, bad scheduling as they are of inefficient operations. Better forecasting will lead to better scheduling, which should result in a lower labor-efficiency variance. On the other hand, a favorable variance is probably the result of an underestimate of demand, causing the scheduled employees to do more than they should reasonably be asked to do. Example 5 illustrates the calculation and analysis of a labor cost variance.

Example 5

Muffet and Mike's Magnificent Meals (4M) has developed the following labor standards for its restaurant operations:

Standard labor cost = $30/hour

Standard hours per meal = 15 minutes or 0.25 hours

During a specific time period, 4M served 90,000 meals, and the actual labor costs were $666,400 for 23,800 hours (i.e., $28/hour).

Using a flexible budget figure of $675,000 (i.e., 22,500 hours X $30) results in a favorable labor variance of $8,600:

Flexible budget             $675,000
Actual                      $666,400
                            $8,600 Favorable

Analyzing this variance further, 4M finds the following components of the favorable variance:

Rate variance       AQ (SR -AR)
                    23,800 ($30 - $28) = $47,600 Favorable

Efficiency variance SR (SQ - AQ)
                    $30 (22,500 - 23,800) ($39,000) Unfavorable
                                           $ 8,600


The traditional variance analysis model has a number of limitations for restaurant operations. The first limitation is the one just discussed. To a large degree, production depends on actual demand, not capacity. Production capacity sets limits on how much can be produced and, in a manufacturing setting, provides a standard to evaluate actual productivity. Restaurants, however, do not produce inventory; they produce meals for immediate sale. If there is no demand, there is no production, so productivity or efficiency in the restaurant is limited not only by productive capacity, but also by actual demand. Therefore, we must view labor efficiency in the traditional sense of efficiency and in relationship to the accuracy of forecasting.

Second, using statistical techniques to develop cost standards has both advantages and problems. On the positive side, one can combine large amounts of data in an efficient, objective manner and revise standards periodically, based on experience and further accumulation of data, without much additional cost. The main disadvantage of a statistical approach is that a standard setter must assume no biases arise from using past data to approximate future relationships. It is not always valid to assume that the environment generating these observations has remained stable from period to period, and we can extrapolate this environment to determine what the future relationships between costs and levels of activities should be.

The third limitation involves the actual nature of cost behavior over changes in volume of activity (i.e., increases and decreases in the number of customers). The variance model assumes a constant linear relationship between the cost items and the activity level. For food cost, this is a reasonable assumption since each additional sale requires a proportional increase in the amount of food the restaurant uses. Perhaps some inefficiencies result at very low volumes, but the linearity assumption appears reasonable within steady activity levels.

For labor cost, the case is somewhat weaker. Labor cost is described more accurately as a step variable cost. Up to a point, each additional customer does not require an increase in the labor force, but when volume increases past a particular point, labor costs also must increase. Labor hours, too, are related positively to customers served, but this relationship is not strictly linear. As volume increases, labor hours, like labor costs, increase in a step-like fashion, and only when volume reaches the next step does one expect the labor force to increase. A particular size of staff may accommodate a range of output levels, but at some point the workforce must increase or decrease to reflect larger or smaller volumes.

This description suggests that, even with accurate forecasting, the relative efficiency of labor still depends on sales volume. With all other variables constant, labor will be most efficient at volumes just before the step and least efficient just beyond the step. Because of the step nature of labor costs, the distortion is less critical than the forecasting problems noted earlier. Nevertheless, the combination of forecasting difficulties and step variable labor costs causes the interpretation of labor variances to be a bit muddled. Therefore, the manager may need to conduct an in-depth investigation of the causes of a significant variance before taking corrective action.

The fourth limitation deals with the aggregation of food items and labor grades in generating the variance statistics. Menu items are neither priced identically nor do they have the same production costs, and, similarly, all laborers do not earn the same wages per hour nor do they work with the same degree of efficiency. As a result, out-of-control cost components may be hidden by offsetting errors. Having a service staff that is too small and a dish crew that is too large may never become apparent if the manager aggregates labor costs for the two staffs. By the same token, charging too much for one menu item and too little for another may result in statistics that appear appropriate but hide menu price flaws.

The final limitation concerns the lack of independence among the variances. Food price variances may not be independent of food efficiency variables because, for instance, buying a cheaper cut of meat may generate a favorable price variance only to yield an unfavorable quantity variance resulting from shrinkage or spoilage. This is also true of labor. Employees earning lower wages will produce favorable rate variances but may also cause unfavorable efficiency variances. This situation is particularly likely when significant turnover of staff occurs or when the manager asks employees skilled in one job to perform another. In general, dishwashers may not perform well when they work as low-paid waiters, and they also do not always make good low-paid cooks. This potential lack of independence among variances simply requires an investigation of variances that goes beyond number crunching and immediate finger-pointing. Variance analysis is only the first step in the evaluation process.


Cost variance analyses are useful to restaurateurs to signal which activities are functioning as planned. Although the implementation of a standard cost system in restaurants poses unique problems that do not exist in a typical manufacturing or service setting, the system has strong potential to provide high-quality information that restaurant managers have historically overlooked. LEGENDS, however, has successfully adopted these techniques and will continue to integrate this system of cost variance analysis into its regular profit-and-loss examinations. If another restaurant plans to do the same, it should ensure all standards are in place for a reasonable period of time prior to beginning analysis. It is also important to have the entire management team on board and educated about the new analysis system. Finally, conversations with other restaurant operators and possibly even manufacturing management personnel will go a long way to pose the proper questions as management generates and analyzes standards and variances. After one has become a champion of the cost variance analysis system and starts to see the usefulness of the information unfold, the natural next steps will involve plans to make data collection, standard computations, and variance calculations more intuitive and instinctive to the individual operation.


Point-of-sale (POS) systems can be extremely valuable tools in a restaurant. The most advanced software on the market can automatically calculate some of the information we discussed. These machines are not financially or logistically feasible for all organizations, and they vary greatly in functionality. Additionally, it is important for all restaurant operators to effectively understand the intuition involved in cost variance analysis because the numbers tell only a part of the story. Furthermore, the numbers from POS systems are frequently skewed because they rely entirely on employees accurately inputting thousands of pieces of data throughout any given period.


Author Ken Milani knows firsthand about the restaurant business. For nearly 20 years, his parents owned and operated a pizzeria and sandwich shop attached to a tavern, which his folks also operated. Ken's summers, Christmas breaks, and spring breaks were spent working as a cook in the pizzeria and sandwich shop. While working at the restaurant, he took an interest in ensuring food did not go to waste. Located in Chicago, Ill., for more than 14 years before moving the business to a suburb, the restaurant combined fun and work with the emphasis definitely on the latter.

This article is dedicated to the memory of Grace and Albert Milani, Ken's parents, who owned and operated a restaurant for almost 20 years.

Ken Milani, Ph.D., is an accountancy professor at the University of Notre Dame in Notre Dame, Ind., and is a member of the Michiana Chapter of IMA[R]. You can reach him at (574) 631-5296 or

Aaron Perri is the executive director of Downtown South Bend (DTSB). At the time the article was submitted, Perri was general manager of Legends of Notre Dame Restaurant and Alehouse Pub.

Table 1. Traditional Cost Variance

Total Variance = (AR X AQ) - (SR X SQ), or
Actual Cost - Standard Cost

This variance measures the total difference between
the actual cost and the standard cost for
a given output.

Rate Component = (SR X AQ) - (AR X AQ), or
AQ (SR - AR)

This variance subcomponent measures the effects
that result from utilizing units of input (labor
or goods) at a price or cost different from the

Quantity Variance = (SR X AQ) - (SR X SQ), or
SR (SQ - AQ)

This variance subcomponent measures the effect
of using a quantity of input (goods or labor) that
is different from the amount predicted for the
level of output.

Table 2. LEGENDS' Irish Nachos Recipe and Theoretical Cost

Amount/Item                      Cost         Total

6 oz. Tortilla Chips          $0.08/oz.       $0.48
4 oz. Shredded Cheese         $0.24/oz.       $0.96
6 oz. Salsa                   $0.15/oz.       $0.90
1 oz. Guacamole               $0.13/oz.       $0.13
1 oz. Black Beans and Corn    $0.08/oz.       $0.08
1 oz. Sour Cream              $0.22/oz.       $0.22
1/2 oz. Jalapenos              $0.20/oz.      $0.10


Table 3. 16-Week Data for LEGENDS of Notre Dame

Week    Labor     Direct       Food   # Customers       Food
        Hours      Labor       Cost        Served      Sales

1         958     $8,124     $7,340         1,795    $25,882
2         999     $8,298     $7,608         1,873    $26,886
3       1,026     $8,938     $7,947         1,944    $27,828
4         963     $8,257     $7,451         1,820    $26,139
5       1,006     $8,189     $7,342         1,864    $26,210
6       1,009     $8,485     $7,842         1,946    $26,957
7         960     $8,237     $7,395         1,815    $26,114
8         948     $8,255     $7,353         1,801    $25,928
9         968     $8,347     $7,791         1,896    $26,773
10        909     $7,972     $7,010         1,728    $25,183
11        914     $7,992     $7,062         1,716    $25,126
12      1,001     $8,147     $7,821         1,885    $26,345
13        926     $8,054     $7,143         1,746    $25,423
14        947     $8,079     $7,187         1,771    $25,731
15        932     $8,065     $7,180         1,757    $25,454
16        946     $8,083     $7,350         1,781    $25,759
       15,412   $131,522   $118,822        29,138   $417,738

Table 4. Data for Week X

 Week    Labor     Direct      Food    # Customers     Food
         Hours   Labor Cost    Cost      Served       Sales

X          975     $8,229     $7,350      1,730      $26,098

Table 5. Labor Cost Variance

Actual Rate x Actual Quantity

$8.44 * 975 = $8,229

Standard Rate x Standard Quantity

$8.534 ** 921.02 *** = $7 ,860

Standard - Actual =  Total Variance
$7,860 - $8,229 = ($369)

Because actual labor costs were higher than the
standard costs, we have determined an unfavorable
variance of $369 in labor costs.

* This is Direct Labor Cost/Labor Hours

** Average from  Table 3

*** Regression relationship formula:
0.4631(1,730) + 1 19.86 = 921.02

Table 6. Food Cost Variance

Actual Food Cost

$7,350 *

Standard Food Cost

$7,422 **

Standard - Actual = Total Variance

$7,422 - $7,350 = $72

Because actual food costs were lower than standard
(expected) food costs, we have determined a
favorable variance of $72 in food costs.

* From inventory

** 28.44% Standard Food Cost $26,098
Actual Food Sales


Please note: Illustration(s) are not available due to copyright restrictions.
COPYRIGHT 2013 Institute of Management Accountants
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Milani, Ken; Perri, Aaron
Publication:Management Accounting Quarterly
Date:Jun 22, 2013
Previous Article:Building accounting students' communication skills.
Next Article:Do lean implementation initiatives have adequate accounting support? The debate of duality.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters