A

dva
n ce
sinA
ccoun
tin
g38(2017
)9–
910
5

Contents lists available at ScienceDirect

Advances in Accounting
journal homepage: www.elsevier.com/locate/adiac

Lattice allocations: A better way to do cost allocations

Kirke Benta,1, Dennis Caplanb,⁎
a
Prudential Insurance Company, Chatham, NJ, United States
b
University at Albany – SUNY, 1400 Washington Ave., Albany, NY 12222, United States

ARTICLEINFO ABSTRACT

Keywords: This paper describes a new cost allocation technique that we call “lattice allocation.” The technique uses matrix
Lattice allocation algebra operations such as in standard spreadsheet software to perform complex cost allocations. To illustrate
Cost allocation the power and versatility of the technique, we provide two examples. First, we show how lattice allocations can
Matrix algebra allocate service department costs to fully recognize mutual services among service departments, similar to the
Service department costs
reciprocal method but without the complexity of solving a set of linear equations. Second, we show how lattice
Reciprocal allocations
allocations can implement two-stage activity-based costing in a single step. The lattice allocation method is easy
Activity-based costing
to apply, requires no special software for small organizations, easily scales up to medium and large organiza -
tions, and provides managers better information about the origins of allocated costs. Also, lattice allocations are
likely to produce more accurate results than traditional techniques and are well-suited as a pedagogical tool for
cost accounting courses. The paper suggests future research topics, including the implications of lattice allo -
cations for Fast Close, Continuous Reporting, and government contractor cost accounting requirements.

1. Introduction easier to detect, and easier to correct. An error in a particular service

We introduce a new method, “lattice allocation,” for computing cost
allocations. We illustrate the method first for service department cost
allocations, and then for two-stage activity-based costing. The tech-
nique is quite different from methods currently used, and has sig-
nificant advantages over them. The lattice allocation method (LA) uses
matrix algebra operations such as in standard spreadsheet software to
perform complex cost allocations, thereby making sophisticated costing
methods accessible to organizations of all sizes, as well as for pedago-
gical purposes.
There are four important advantages of LA over current methods.
First, LA is easy to learn and easy to implement. The user creates a
single matrix that shows the allocation rules, runs a simple matrix
manipulation, and the transformed matrix provides the user the per-
centages of the cost category (e.g., service department or cost pool) to
allocate to each cost object (e.g., production department or product).
The allocation rules only reflect allocations from each service depart-
ment, and individually contain no information about downstream re-
allocations. A single matrix can handle multi-stage allocations, such as
allocating from cost categories to cost pools and then to products.
The second advantage of LA is that it is less prone to error. LA
produces none of the errors inherent in the use of the direct and step-
down methods. Because LA is simple, errors are less likely to be made,
department allocation rule can be corrected without reference to the
allocation rules in the other service departments. Also, some types of
errors are easier to correct. For example, when LA is used for service
department reciprocal cost allocations, if a service department dis-
covers additional department costs that should have been allocated,
there is no need to revise the matrix or rerun the matrix transformation.
As long as the measures of usage by downstream departments are un-
changed, the percentages in the transformed matrix are still correct.
These percentages can be applied to the revised department cost total.
Importantly, there is no need to revisit the allocations from the other
service departments.
The third advantage of LA is that it highlights for downstream
managers (e.g., product managers or production department managers)
precise information about the original source of the costs allocated to
them. For example, in two-stage activity-based costing where costs
originate in cost categories, are allocated to cost pools, and then to
products, LA provides the product manager the costs allocated from
each cost category, not the total from the intermediate cost pool. If the
same cost category is allocated to multiple cost pools and then to pro-
ducts, the product manager sees the total costs allocated from the cost
category summed across all cost pools. This information has the po-
tential to focus the product manager's efforts on cost reduction at the
source, and away from efforts to shift costs in the cost pool to other

⁎
Corresponding author.
E-mail addresses: kirkebent@icloud.com (K. Bent), dcaplan@albany.edu (D. Caplan).
1
Prudential Insurance Co., Retired, USA.

http://dx.doi.org/10.1016/j.adiac.2017.07.008
Received 26 June 2017
Ava
ilableonlin
e1 4A
ugust20
17
0
882 -6
1 10
/© 2
017TheAuth
ors.P
ublish
edb
yElse
vie
rLtd
.Th
isisa
nope
nacce
ssa
rticleu
nde
rth
eCCBY-NC
-NDlice
n se(h
ttp
:/cre
ativ
ecomm
ons.org
/lice
n se
s/BY-NC
-ND
/4.0
/).
K. Bent, D. Caplan A
dva
ncesinA
ccou
ntin
g38(2
017
)9–1
9 0
5

product managers. A recent survey of management accountants iden- Table 1

tified cost reduction as their top priority, followed by generating cost Example 1 – intrinsic rules.

information (Clinton & White, 2012). Lattice allocations are well-suited Maintenance IS Machining Assembly
to helping organizations address both priorities.
The fourth advantage of LA is that it scales well, from the smallest Maintenance 0.00 0.20 0.30 0.50
classroom problem to the largest industrial problem. What students are IS 0.10 0.00 0.80 0.10
Machining 0.00 0.00 1.00 0.00
taught can be directly applied in practice.
Assembly 0.00 0.00 0.00 1.00
The lattice allocation method is an entirely new technique for al-
locating costs in accordance with the underlying provision of services.
It is not a new principle of where costs should fall. Lattice allocations Systems (IS), and two production departments, Machining and As-
can be beneficially applied to any type of allocation, but it is most sembly. The support departments provide services to the production
beneficial when current methods for implementing the allocation are departments and to each other. Each support department tracks which
complex, such as when there is a hierarchy of allocations (as in two- departments receive its services and in what proportions.
stage activity-based costing) or when there are reciprocal relationships For Maintenance, 20% of its services are provided to IS, 30% are
(as often occurs in service department cost allocations). Although ma- provided to Machining, and 50% are provided to Assembly. For IS, 10%
trix algebra is currently used in some cost allocation applications, LA of its services are provided to Maintenance, 80% are provided to
uses matrix manipulations in an entirely novel way, and by doing so, Machining, and 10% are provided to Assembly. In the terminology of
generates information that is not available from existing techniques. Lattice Allocations, these two sentences describe the “intrinsic” rules of
The next section of the paper illustrates lattice allocations in the the problem. Table 1 provides a comprehensive description of these
context of service department cost allocations. Section 2 applies LA to a intrinsic rules:
relatively simple example drawn from a leading cost accounting text- Some points to note:
book. Section 3 provides a step-by-step overview of lattice allocations.
Section 4 applies LA to a more complicated service department cost 1. The percentages have been restated to fractions.
scenario involving four service departments and three operating de- 2. Each row sums to one.
partments. Section 5 illustrates lattice allocations for two-stage activity- 3. Production departments are shown as allocating 100% to them-
based costing. Section 6 provides concluding remarks. selves.
4. The information in Table 1 can be obtained from each service de-
2. Service department costs: example 1 partment, based on that department's information about services it
provides to other departments, without requiring any information
There are currently three common methods for allocating service about subsequent downstream re-allocation.
department costs: the direct method, the step-down method, and the
reciprocal method. Of these three techniques, only the reciprocal We refer to the 4 × 4 matrix of fractions in Table 1 as the Lattice
method recognizes all services provided by service departments to other Allocation matrix. The central technique of LA is to multiply this matrix
service departments. Thus, it is the most accurate method of allocating by itself until the transformed matrix converges to a stable solution and
service department costs, but it is also the most complicated. According final rules emerge. In other words, one takes this matrix to a sufficiently
to Zimmerman, “while the reciprocal allocation method has certain high power to obtain the desired degree of accuracy for the organiza-
theoretical advantages, it is not widely used” (Zimmerman, 2014, 345). tion's costing purposes. Simply squaring the matrix often provides the
Christensen and Schneider (2017) note that “often the prospect of al- desired accuracy. When the LA matrix in Table 1 is squared, the result is
locating service department costs can lead to feelings of frustration or the matrix of percentages (rounded to two decimal places) shown in
dread in accounting practitioners or students” (51). Table 2.
There are several ways to apply the reciprocal method (Keller, The matrix in Table 2 can be partitioned into quadrants. The four
2015). One method solves a set of simultaneous equations using alge- cells in the upper-left quadrant represent services provided by each
braic substitutions. Another way uses a matrix approach for solving service department to the other service department. The four cells in
these simultaneous equations. The algebraic substitution method be- the upper-right quadrant represent services provided by each service
comes unwieldly as the number of service departments increases. department to the production departments. The eight cells in the lower
Hence, many cost accounting textbooks illustrate the reciprocal method quadrants reflect the fact that no costs are allocated out of the pro-
using just two service departments and two production departments, duction departments. Going from Tables 1 to 2, the total amounts al-
similar to the example in Horngren, Datar, and Rajan (2012) used located in the upper-left quadrant are reduced, the amounts allocated in
below. The matrix approach works for a more realistic number of ser- the upper-right quadrant have increased, and there are no changes in
vice and production departments, but requires a level of mathematical the lower quadrants. Each row continues to sum to one. These char-
ability beyond skills typically possessed by accounting students or even acteristics are always true in the Lattice Allocation method.
many practitioners. Concerning use of matrix methods for traditional Table 3 reports the results of the LA matrix taken to the third power
allocation methods, Christensen and Schneider (2017) recently ob- (i.e., Table 3 multiplies the matrix in Table 1 by the matrix in Table 2),
served “students and financial managers find the use of matrix algebra rounded to two decimal places.
to be quite a challenge” (52). Christensen and Schneider (2017) illus- In Table 3, the amounts in the upper-left quadrant have been re-
trate how Excel can be used to facilitate implementation of a third duced to zero, while the amounts in the upper-right quadrant sum to
method for deriving reciprocal cost allocations, the iterative method.
This method is easier to explain than the matrix approach, but also
becomes unwieldly when there are numerous service departments. We Table 2
Example 1 – the squared matrix.
present another way to allocate service department costs that captures
the reciprocal relationships among service departments. Our method Maintenance IS Machining Assembly
uses matrix algebra, is fundamentally different from matrix methods
Maintenance 0.02 0.00 0.46 0.52
described in the literature, but obtains results identical to those
IS 0.00 0.02 0.83 0.15
methods. Machining 0.00 0.00 1.00 0.00
We start with the example from Horngren et al. (2012, 550–557). A Assembly 0.00 0.00 0.00 1.00
firm has two service departments, Maintenance and Information

1
0
0
K. Bent, D. Caplan
Table 3
Example 1 – the LA matrix taken to the third power.
Maintenance IS Machining
Maintenance 0.00 0.00 0.47
IS 0.00 0.00 0.85
Machining 0.00 0.00 1.00
Assembly 0.00 0.00 0.00
one along each row. Table 3 provides what we call “direct” rules, which
is the “solution” to the allocation problem.2 The appropriate allocation
is to allocate 47% (53%) of the Maintenance service department costs to
Machining (Assembly), and to allocate 85% (15%) of the IS service
department costs to Machining (Assembly). These direct rules result in
the allocations in Table 4.
These allocated costs agree to the results for the reciprocal method
as calculated in Horngren et al. (2012, p. 555, Exhibit 15-6), to within
3/10's of 1%. Greater accuracy could be obtained by taking the LA
matrix to a higher power.
3. A step-by-step description of lattice allocations
In this section, we provide a general step-by-step summary of LA,
sufficiently detailed to be applied by practitioners and students.
Although we provide this example in the context of service department
cost allocations, it applies to any cost allocation problem. In general
terms, the “service department” is where the costs originate, and the
“production departments” are the cost objects. Furthermore, as shown
in Section 5, there can be intermediate allocations between the source
of the costs and the ultimate cost objects, such as with two-stage ac-
tivity-based costing.
o Step 1: Service departments prepare intrinsic rules. By intrinsic
rules, we mean just those aspects of allocation that are known di-
rectly by the department providing the service. The department
answers the questions: (1) what departments do I provide services
to? and (2) What fraction of my costs should be assigned to each of
these departments? In answering these questions, the department
may ignore services that it provides back to itself, so that the per-
centages assigned to the other departments sum to 1.0. Each pair of
answers constitutes one “element” of that service department's rule.
None of these rules individually specify what further reallocations
there might be. However, that information is inherent in the totality
of the rules.
o Step 2: Form a matrix. The matrix has a column for each service
department and each production department. Although the ordering
is arbitrary, an important optimization is facilitated by listing the
service departments first. The rows are listed in the same order as
the columns. Each row for the service departments contains the
intrinsic rules from Step 1. For example, if Service Department F
provides 30% of its services to Service Department C, 0.3 is recorded
in the cell at the intersection of the row for F and the column for C.
Any element of the matrix without a fraction representing an allo-
cation is set to 0.0. Each row for the production departments con-
tains zeros except for the intersection of that production department
with the column for itself, which contains a 1.0. This last convention
implies that each production department allocates 100% of costs to
itself. Each row sums to 1.0, for both the service departments and
trivially for the production departments.
2
Tables 1, 2 and 3 illustrate the following with respect to the cells along the diagonal
at the intersection of each service department with itself. Although these cells start out
with values of zero (Table 1), and converge to zero in the solution matrix (Table 3), they
can take on positive values in the intermediate matrices (Table 2).
A
dva
ncesinA
ccou
ntin
g38(2
017
)9–1
9 0
5
Table 4
Example 1 – allocation of service department costs.
Assembly Service department costs Machining Assembly
0.53 Maintenance $6,300,000 $2,961,000 $3,339,000
0.15 IS 1,452,150 1,234,328 217,822
0.00 Total $7,752,150 $4,195,328 $3,556,822
1.00
o Step 3: Multiply the matrix by itself. This step is easily performed in
Excel for even large matrices using the MMULT function. 3 The re-
sulting matrix has the same shape as the original matrix but dif-
ferent numbers. The total of the fractions in the upper left quadrant
(the quadrant comprising the intersection of the service department
columns with the service department rows) will be reduced, while
the total of the fractions in the upper right quadrant (the quadrant
comprising the intersection of the production department columns
with the service department rows) will be increased. The lower
quadrants will be unchanged.
o Step 4: Multiply the original matrix times the matrix resulting from
Step 3. (Or alternatively, multiply the matrix from Step 3 by itself,
effectively taking the original matrix to the 4th power.) Repeat Step
4 until the cells in the upper left quadrant are all zeros (or suffi-
ciently close to zero to obtain the desired accuracy). At this point,
the service department rows can be used as rules for directly allo-
cating each service department's costs to the production depart-
ments. The resulting allocation will be identical to the results ob-
tained from conventional methods of applying the reciprocal
method.
The technique described here is very efficient with respect to
computing power. For example, we conducted a highly optimized ex-
perimental implementation that allocated 3,000,000 costs from 2700
service departments to 1800 production departments in 9 seconds on a
$659 laptop. Two-thirds of these 2700 service departments provided
services to other service departments. The computational efficiency of
LA suggests that it can be implemented in small to medium-sized or-
ganizations using spreadsheet software.
Large organizations can take advantage of specialized software and
hardware. Matrix multiplication is commonly done with highly-opti-
mized software. The industry-standard Basic Linear Algebra
Subprograms (BLAS) defines linear algebra algorithms. The algorithms
are implemented by hardware vendors for their architectures and by
software vendors that market math packages. The BLAS is implemented
by and available from MATLAB, Mathematica, R, and NumPy, among
others.
4. Service department costs: example 2
We next apply the LA method to a more complicated and realistic
service department example provided in Togo (2010). We then compare
the ease of implementing LA to Togo's step-by-step explanation of how
to use spreadsheet software to apply the reciprocal method.
The Maile-Ann Company has four service departments and three
3
Instructions for using the Excel MMULT function are available from numerous sources
on-line. We provide our own explanation: Assume the LA matrix occupies a 10 × 10
matrix on the spreadsheet, and that it is located such that B2 is the upper-left corner and
K11 is the lower-right corner of the matrix. In an empty portion of the spreadsheet, paint
an area the same size as the original matrix and then hit the F2 key. For example, click on
M2 and holding the curser down, sweep down and to the right to V11, then hit F2. This
should provide a curser in one cell in the new matrix. In that cell, type:
= MMULT(B2:K11, B2:K11). Then hit the Control, Shift, and Enter keys simultaneously.
This should generate values in the new matrix. Note that the MMULT formula is multi-
plying the original matrix by itself. For greater accuracy, create a third matrix that
squares the second matrix.
1
0
1
K. Bent, D. Caplan A
dva
ncesinA
ccou
ntin
g38(2
017
)9–1
9 0
5

Table 5
Maile-Ann Company.
Intrinsic rules.

Dept A Dept B Dept C Dept D Dept X Dept Y Dept Z Total

Dept costs $900,000 $600,000 $500,000 $300,000 $8,500,000 $5,000,000 $4,200,000 $20,000,000

Dept A 0.00 0.08 0.07 0.05 0.40 0.25 0.15 1.00

Dept B 0.09 0.00 0.03 0.06 0.30 0.35 0.17 1.00
Dept C 0.05 0.06 0.00 0.04 0.40 0.25 0.20 1.00
Dept D 0.03 0.07 0.04 0.00 0.32 0.40 0.14 1.00

Table 6 Table 7
Maile-Ann Company Maile-Ann Company.
The LA matrix. LA matrix to the 4th power.

Dept A Dept B Dept C Dept D Dept X Dept Y Dept Z Dept A Dept B Dept C Dept D Dept X Dept Y Dept Z

Dept A 0.00 0.08 0.07 0.05 0.40 0.25 0.15 Dept A 0.00 0.00 0.00 0.00 0.48 0.33 0.19
Dept B 0.09 0.00 0.03 0.06 0.30 0.35 0.17 Dept B 0.00 0.00 0.00 0.00 0.38 0.42 0.20
Dept C 0.05 0.06 0.00 0.04 0.40 0.25 0.20 Dept C 0.00 0.00 0.00 0.00 0.46 0.31 0.23
Dept D 0.03 0.07 0.04 0.00 0.32 0.40 0.14 Dept D 0.00 0.00 0.00 0.00 0.38 0.45 0.17
Dept X 0.00 0.00 0.00 0.00 1.00 0.00 0.00 Dept X 0.00 0.00 0.00 0.00 1.00 0.00 0.00
Dept Y 0.00 0.00 0.00 0.00 0.00 1.00 0.00 Dept Y 0.00 0.00 0.00 0.00 0.00 1.00 0.00
Dept Z 0.00 0.00 0.00 0.00 0.00 0.00 1.00 Dept Z 0.00 0.00 0.00 0.00 0.00 0.00 1.00

operating departments. Table 5 presents the cost data and support
services shown in Togo (2010) in Table 1, Panel A. (We altered Togo's
original table slightly by replacing −1 with 0 for the allocation of each
service department to itself.) Table 5 provides what we call the intrinsic
rules for LA:
From Table 5, we construct the lattice allocation matrix shown in
Table 6.
Some intuition for why squaring the LA matrix iteratively converges
to the reciprocal cost allocations can be illustrated as follows. Consider
the cell in Table 6 at the intersection of the row for service department
B and the column for production department X. When Table 6 is
squared, the value for this cell is computed as follows:
precision could have been obtained by raising the LA matrix to a higher
power. Recall that from the point of view of numerical analysis, any of
these equation solutions or multiplication processes is subject to minute
errors.
Togo concludes that the method presented in his paper “demon-
strates the ease in using spreadsheet matrix functions to solve reciprocal
cost allocations of support departments” (Togo, 2010, 39). Togo asserts
that “this intuitive spreadsheet matrix approach for the preferred re-
ciprocal cost allocation method can be easily adapted for more service
and operating departments” (2010, 39). The following four steps
paraphrase Togo's explanation of how to apply the reciprocal method.
Starting from Table 5:

(0.09 × 0.40) + (0.00 × 0.30) + (0.03 × 0.40) + (0.06 × 0.32)
+ (0.30 × 1.00) + (0.35 × 0.00) + (0.17 × 0.00) = 0.3672
The fifth term in this series (0.30 × 1.00) captures the services
provided by service department B directly to production department X.
The first term (0.09 × 0.40) captures the services provided by service
department B indirectly to production department X through service
department A. Specifically, 9% of department B's services are provided
to department A, and then 40% of department A's services are provided
to department X. Similarly, the third and fourth terms in the series
capture the services provided by department B indirectly to department
X through service departments C and D, respectively. All other terms in
the series equal zero.
Squaring the matrix does not provide the precise solution because
some cells in the upper-left quadrant of the squared matrix (not shown)
still have positive values. Squaring the matrix a second time (thus
taking the matrix in Table 6 to the 4th power) produces the matrix in
Table 7. Now all cells in the upper-left quadrant of Table 7 contain
zeroes (when amounts are rounded to two decimal places), and the cell
at the intersection of the row for service department B and the column
for production department X shows the solution (accurate to two dec-
imal places) of 0.38, slightly larger than the value calculated in the first
iteration of 0.3672.
Table 8 applies the fractional equivalent of percentages in Table 7 to
the costs in each service department, and then derives total costs for the
operating departments for ease of comparison to Table 1, Panel B in
Togo (2010):
Table 8 presents total operating department costs that differ from
the solution in Togo (2010) by less than 1/10 of 1%. Still greater
1. Write down the set of four independent linear equations with four
unknowns. For example, the equation for service department A is
derived from the column for Department A in Table 5 as follows:
900,000–1.00A + 0.09B + 0.05C + 0.03D = 0.
2. Transform the set of four independent linear equations into a matrix
relationship S × X = K, where S is a 4 × 4 matrix of the coefficients
on the variables A, B, C and D for each linear equation, X is a 4 × 1
matrix representing the service departments' unknown reciprocated
costs, and K is a 4 × 1 matrix of the costs originating in each service
department.
3. Multiply both sides of the matrix relationship S × X = K by the
inverse of the matrix S, in order to solve for the matrix X.
4. Create a 4 × 4 matrix with the values from the matrix X along the
main diagonal and zeros everywhere else. Multiply this 4 × 4 ma-
trix by the 4 × 7 matrix of percentages in Table 5 (ignoring the top
row of Table 5). The result is a 4 × 7 matrix of the allocation of
reciprocated costs of service departments to the support depart-
ments, as shown in Table 9 (replicated from Table 1, Panel B in
Togo, 2010).
Stinson (2002) provides another step-by-step illustration of using
spreadsheet software to apply the traditional matrix algebra equation
solution for the reciprocal method. Table 6 in Stinson (2002) provides a
nine-step guide to reciprocal allocation with Microsoft Excel. We leave
it to the reader to judge which method is simpler: the LA method or the
solution techniques explained in Togo (2010) and Stinson (2002).
In addition to simplicity, the lattice allocation method provides
another advantage. Table 8 shows operating department managers the

1
0
2
K. Bent, D. Caplan A
dva
ncesinAccou
ntin
g38(2
017)9–
910
5

Table 8
Maile-Ann Company.
LA allocation of service department costs.

Department costs Percentages from the LA matrix Final costs Grand total

Dept X Dept Y Dept Z Dept X Dept Y Dept Z

Dept A $ 900,000 0.48 0.33 0.19 $ 433,192 $ 294,423 $ 171,547

Dept B 600,000 0.38 0.42 0.20 227,770 249,299 122,401
Dept C 500,000 0.46 0.31 0.23 230,874 154,546 114,204
Dept D 300,000 0.38 0.45 0.17 113,770 135,313 50,706
Dept X 8,500,000 1.00 0.00 0.00 8,500,000
Dept Y 5,000,000 0.00 1.00 0.00 5,000,000
Dept Z 4,200,000 0.00 0.00 1.00 4,200,000
Total $20,000,000 $9,505,606 $5,833,581 $4,658,858 $19,998,045

Table 9
Maile-Ann Company.
Solution as reported in Togo (2010).

Dept A Dept B Dept C Dept D Dept X Dept Y Dept Z Total

Dept costs 900,000 600,000 500,000 300,000 8,500,000 5,000,000 4,200,000 20,000,000
Dept A − 1,010,300 80,824 70,721 50,515 404,120 252,575 151,545 0.00
Dept B 67,212 − 746,799 22,404 44,808 224,040 261,380 126,955 0.00
Dept C 30,496 36,595 − 609,914 24,397 243,965 152,478 121,983 0.00
Dept D 12,592 29,380 16,789 − 419,720 134,310 167,888 58,761 0.00
Total 0.00 0.00 0.00 0.00 9,506,435 5,834,321 4,659,244 20,000,000

true origin of the costs allocated to their departments from the service
departments, as illustrated by the fact that the amounts allocated from
each service department to the three operating departments sum to
each service department's original costs. As has been noted elsewhere
(e.g., Livingstone, 1968), this is not true of traditional methods of cal-
culating reciprocal service department costs that use the gross services
model.
In the gross services model, the solutions to the system of equations
are not the allocated costs, but rather are augmented values that sum to
more than the actual costs to be allocated. While they can be “ex-
plained,” these values do not represent any quantities involved in the
provision of services by the service departments. Rather, they are only
needed to adjust the “solution” to provide the correct allocations.4
5. Two-stage activity-based costing
Lattice allocations can execute two-stage activity-based costing in a
single calculation. Relative to the traditional two-stage ABC calculation
method, the LA calculation method generates the same results but is
simpler and provides product managers better information about the
origins of allocated overhead.
We illustrate lattice allocations for activity-based costing by
adapting the Classic Brass example from Chapter 8 of Garrison and
Noreen (2000). In that example, detailed information is provided for
two products sold to a single customer: standard stanchions and a
custom compass housing. These two products account for a small
fraction of the company's total production. For expositional con-
venience, we assume that a single product accounts for all remaining
production. Thus, we do not change any of the numbers in the Garrison
and Noreen example, but we fill in for missing information.
Overhead costs are provided in Exhibit 8-2 of Garrison and Noreen
(p. 330). We reproduce those costs in Table 10, omitting shipping costs
which is included in Exhibit 8-2 but is later determined to be a direct
cost:
4
Keller (2015) develops a matrix-based technique for calculating the reciprocal
method using the net services model that facilitates tracing costs between service de-
partments. Although Keller's method may have advantages over the gross services model,
it is not simple.
These overhead costs are assigned to five cost pools:
1. Customer orders (CO)
2. Product design (PD)
3. Order size (OS)
4. Customer relations (CR)
5. Other (O).
Using information provided by Garrison and Noreen (p. 333) in
Exhibits 8-3, 8-5, 8-8A and 8-8B, we construct the Lattice Allocation
matrix as shown in Table 11.
The first nine columns and rows in Table 11 list the individual
overhead cost categories. Column and row headings for these cost ca-
tegories use the abbreviations provided in Table 10. The next five
columns and rows list the overhead cost pools. Column and row
headings for these cost pools use the abbreviations shown after
Table 10, and are shown in italics. The last three columns and rows list
the final cost objects: stanchions (s), compass housing (c.h.), and all
other products (a.o.). The abbreviations for these three products are
lower case.
Many of the cells in the matrix have values of zero by design. The
cells that potentially have nonzero values are the cells contained in the
9 × 5 box in the upper-center of the matrix under the columns for the
cost pools, the cells in the 5 × 3 box in the center-right of the matrix in
the rows for the cost pools, and the three 1's along the diagonal of the
3 × 3 box in the lower-right of the matrix.
The 9 × 5 box represents the first-stage allocation of the nine cost
categories to the five cost pools. Each row sums to 1 (100%), indicating
that all costs are allocated to those five cost pools. These percentages
come directly from Exhibit 8-3 in Garrison and Noreen.
The 5 × 3 box represents the second-stage allocation of the over-
head cost pools to the three products. These percentages were calcu-
lated from information provided in the exhibits and text in Garrison and
Noreen. For example, Exhibit 8-5 indicates that there were 1000 cus-
tomer orders, and Exhibit 8-8 indicates that stanchions account for two
customer orders and the compass housing accounts for one order.
Therefore, stanchions account for 0.2% of total orders, the compass
housing accounts for 0.1% of total orders, and the remaining products
account for 99.7% of customer orders. The two cost objects identified in

1
0
3
K. Bent, D. Caplan A
dva
ncesinAccou
ntin
g38(2
017)9–
910
5

in the last three columns under the column headings for the three
Table 10
Classic brass overhead costs. products. The first nine rows of those columns provide the percentages
of each cost category to allocate to each product. Hence, the last three
Production Department columns of each of those rows sum to 100%). That portion of the matrix
Indirect factory wages (FW) $500,000 is reproduced in Table 12, columns (3) to (5). Column (2) reproduces
Factory equipment depreciation (FD) 300,000
Factory utilities (FU) 120,000
the costs from Table 10. The allocation of overhead costs to stanchions
Factory building lease (FL) 80,000 is calculated by multiplying Column (2) by the percentage in Column
1,000,000 (3), with the result recorded in Column (6). The costs for the compass
General Administrative Department housing and the “other” category are calculated analogously.
Administrative wages and salaries (AW) 400,000
Although the LA matrix was taken only to the second power (i.e., it
Office equipment depreciation (OD) 50,000
Administrative building lease (AL) 60,000 was only squared), the resulting allocation agrees to the dollar to the
510,000 costs for stanchions and the compass housing in Exhibits 8-8A and 8-8B
Marketing Department in Garrison and Noreen (p. 339). LA correctly allocates overhead in a
Marketing wages and salaries (MW) 250,000 two-stage ABC system without the necessity of explicitly calculating
Selling expenses (SE) 50,000
overhead rates. This enables avoiding a category that is unnecessary for
300,000
Total overhead costs $1,810,000 calculating correct values, removing a possible source of confusion. The
only inputs that were required to create the LA matrix were the per-
centages that each cost pool uses of each cost category, and the per-
Garrison and Noreen account for only a small fraction of the overhead centages that each product uses of each cost pool. The only other inputs
allocated, and our third “catch-all” product category receives most of are the cost category amounts, in order to convert percentages in the LA
the allocated overhead. If Garrison and Noreen had provided informa- matrix to dollars in Table 12, columns (6)–(8).
tion for all of the other products, there would be an additional row and
column in the matrix for each product. 6. Discussion and conclusion
The diagonal of three 1's in the lower-right represents the fact that
the products are the final recipients of costs, and no costs are allocated We have presented a new, matrix-based technique for allocating
from those final cost objects. This diagonal of 1's in the lower-right costs, which we call lattice allocations. We illustrated the technique for
corner of the matrix is a recurring feature of the LA matrix. two types of cost allocations: service department reciprocal cost allo-
When the matrix in Table 11 is squared, the only nonzero entries are cations, and two-stage activity-based costing. We believe the technique

Table 11
Classic brass lattice allocation matrix.

FW FD FU FL AW OD AL MW SE CO PD OS CR O s c.h. a.o.

FW 0 0 0 0 0 0 0 0 0 0.25 0.40 0.20 0.10 0.05 0 0 0

FD 0 0 0 0 0 0 0 0 0 0.20 0 0.60 0 0.20 0 0 0
FU 0 0 0 0 0 0 0 0 0 0 0.10 0.50 0 0.40 0 0 0
FL 0 0 0 0 0 0 0 0 0 0 0 0 0 1.00 0 0 0
AW 0 0 0 0 0 0 0 0 0 0.15 0.05 0.10 0.30 0.40 0 0 0
OD 0 0 0 0 0 0 0 0 0 0.30 0 0 0.25 0.45 0 0 0
AL 0 0 0 0 0 0 0 0 0 0 0 0 0 1.00 0 0 0
MW 0 0 0 0 0 0 0 0 0 0.20 0.10 0 0.60 0.10 0 0 0
SE 0 0 0 0 0 0 0 0 0 0.10 0 0 0.70 0.20 0 0 0
CO 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.002 0.001 0.997
PD 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.005 0.995
OS 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.01 0.0002 0.9898
CR 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
O 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
s 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0
c.h. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0
a.o. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1

Table 12
Classic brass allocation of costs to products.

Cost category Costs Percentages from the LA matrix Costs allocated

Stanchions Compass housing Other Stanchions Compass housing Other

(1) (2) (3) (4) (5) (6) (7) (8)

FW $500,000 0.00250 0.00229 0.99521 $1250 $1145 $497,605

FD 300,000 0.00640 0.00032 0.99328 1920 96 297,984
FU 120,000 0.00500 0.00060 0.99440 600 72 119,328
FL 80,000 0 0 1.00000 0 0 80,000
AW 400,000 0.00130 0.00042 0.99828 520 168 399,312
OD 50,000 0.00060 0.00030 0.99910 30 15 49,955
AL 60,000 0 0 1.00000 0 0 60,000
MW 250,000 0.00040 0.00070 0.99890 100 175 249,725
SE 50,000 0.00020 0.00010 0.99970 10 5 49,985
Total $1,810,000 $4430 $1676 $1,803,894

1
0
4
K. Bent, D. Caplan
is potentially useful for other types of complex allocations, including
allocating corporate costs to reportable segments.
Lattice allocations are executed using matrix algebra and standard
spreadsheet software. The learning curve for using lattice allocations is
not steep. We believe that LA can be used by practitioners in organi-
zations of all sizes. Bradbard, Alvis, and Morris (2014) surveyed
spreadsheet usage by management accountants, and finds that, on
average, basic Excel spreadsheet functions are used several times a
week or daily. Although matrix manipulation was not explicitly in-
cluded in the survey, the mean respondent uses Excel functions that
seem equally difficult (e.g., logic functions and basic statistical func-
tions) almost weekly. We also believe that lattice allocations can be
used in cost accounting classes to provide students a more complete and
realistic understanding of cost allocation techniques.
In the context of two-stage activity-based costing, lattice allocations
provide the managers of the final cost objects (e.g., product managers)
information about the original source of costs. Traditional two-stage
activity-based costing focuses the product manager's attention on the
product's use of cost pool cost drivers: reduce your product's use of the
cost driver, and fewer overhead costs will be allocated to your product
from the cost pool. But as noted by West and West (1997): “Cost shifting
is not cost containment” (26). Lattice allocations applied to activity-
based costing shifts the manager's focus to the original source of the
costs, which we believe is more likely to result in cost reductions for the
organization.
Performing cost allocations contributes to the effort required to
close the books and compute profits at period-end. Two characteristics
of LA bear on the speed and ease with which profits can be computed.
First, LA is expected to increase the speed of executing complex cost
allocations. Second, the ability to develop transformed rules (the matrix
of intrinsic rules) before costs are known can remove some processing
time from the critical path of the close. The degree of improvement is a
subject for future research.
We conjecture that some companies would like to generate con-
tinuous measures of earnings and its components, or at least frequent
measures. We further conjecture that the difficulty and time required
for cost allocations are barriers to continuous reporting. The same two
qualities of LA mentioned above can help eliminate these barriers. The
speed of LA means that allocations can be calculated daily or even more
frequently.
The speed of LA applies to the development of a set of rules as well
as to the application of those rules to costs. The matrix of intrinsic rules
can be updated at frequent intervals, as it only requires each service
department to forward to Accounting its data on the proportion of its
services used by each service and production department.
In the 1970s, the Cost Accounting Standards Board (CASB) stated a
preference for reciprocal allocations over direct or step-down alloca-
tions (Christensen & Schneider, 2017, 52), presumably because the
1
0
5
A
dva
ncesinAccou
ntin
g38(2
017)9–
910
5
reciprocal method is more accurate and less susceptible to manipula-
tion. However, the perceived difficulty of implementing reciprocal al-
locations led the CASB to accept the step-down method
(Kaplan & Atkinson, 1998, 74). Today, Federal regulations still permit
government contractors to use the step-down method (CASB, 2017,
§9904.418.50(e)(4)). Future research can study whether it would be
appropriate for the CASB to revisit this issue and possibly mandate
government contractors to use LA in order to generate more accurate
cost data than is currently required.
Theoretical research can be conducted on the transformed matrix
generated by the LA method. Characteristics of the matrix might be
useful in further processing or in understanding particular sets of rules.
Case study research might examine the application of LA in a real-world
setting. Finally, pedagogical research can determine whether the use of
LA in the classroom leads to better learning outcomes for management
accounting students.
Acknowledgements
We would like to thank participants at the 2017 Advances in
Accounting Research Conference, and especially Chris Akroyd, Richard
Schneible, and Joe Ugrin for helpful comments.
References
Bradbard, D. A., Alvis, C., & Morris, R. (2014). Spreadsheet usage by management ac-
countants: An exploratory study. Journal of Accounting Education, 32(4), 24–30.
Christensen, D., & Schneider, P. (2017). Allocating service department costs with Excel.
Strategic Finance, 98(11), 50–55.
Clinton, B. D., & White, L. R. (2012). The role of the management accountant: 2003–2012.
Management Accounting Quarterly, 14(1), 40–74.
Cost Accounting Standards Board (2017). Electronic code of federal regulations: Title 48:
Federal acquisition regulations system: Part 9904—Cost accounting standards.
Washington, D.C.: U. S. Government Publishing Office. Available at: https://www.
gpo.gov/fdsys/pkg/CFR-2003-title48-vol1/content-detail.html (accessed June 22,
2017) .
Garrison, R. H., & Noreen, E. W. (2000). Managerial accounting (9th edition). Boston, MA:
Irwin McGraw-Hill.
Horngren, C. T., Datar, S. M., & Rajan, M. V. (2012). Cost accounting: A managerial em-
phasis (14th edition). Upper Saddle River, NJ: Pearson Education.
Kaplan, R. S., & Atkinson, A. A. (1998). Advanced management accounting (3rd edition).
Upper Saddle River, NJ: Prentice Hall.
Keller, A. C. (2015). Service department cost allocations using the net services model and
the MDTERM function in Excel. Journal of Accounting Education, 33, 241–255.
Livingstone, J. L. (1968). Matrix algebra and cost allocation. The Accounting Review,
43(3), 503–508.
Stinson, J. B. (2002). Cost allocation—From the simple to the sublime. Management
Accounting Quarterly, 4(1), 1–10.
Togo, D. F. (2010). Maile-Ann Company: A matrix approach to reciprocated support
department cost allocations. Journal of Business Case Studies, 6(2), 35–40.
West, T. D., & West, D. A. (1997). Applying ABC to healthcare. Management Accounting,
78(8), 22–33.
Zimmerman, J. L. (2014). Accounting for decision making and control (8th edition). New
York, NY: McGraw Hill.