Misleading allocations and how to fix it–Part 1

Two things about which I warn my clients who buy manufacturing software are these:

1. Manufacturing software is capable of capturing, storing and reporting on reams of data
2. If you are not careful, you will find yourself taking “as fact” the data produced by the system and being mislead in your decision-making

Why is this so?

Because ERP systems allow the users to create allocations of overhead based on manufacturing “drivers.” In Sage 500 ERP’s case (as shown in the screen image below), the chosen driver is “labor hours”—for run time and set-up time.

In the Sage 500 ERP Set Up Work Center screen there are places for “Fixed Setup” costs and “Fixed Run” costs. The values placed here are used to absorb “Fixed” overhead costs at the rate supplied based on each hour of “Setup” or “Run” time calculated for production utilization of the Work Center.

The problem is that these “absorption rates” must be calculated based on historical (or prognosticated based on expected future) utilization rates of each Work Center. These calculations must make assumptions about product mix, work center utilization rates and operating expense levels. As soon as any of the these factors change

• Product mix
• Work center utilization rates
• Overhead expenses

The data supplied by the calculations will be wrong.

And, since either the product mix or the total of operating expenses will certainly be different than the numbers used in the calculations, the data resulting from the calculations will (virtually) always be wrong.

A simplified example

We are going to look at two different allocation methods and the decisions that might be derived from such calculations.

• Standard overhead allocations by Job (equivalent to allocation per work order in a manufacturing operation)
• Activity-Based Costing (ABC) allocation based on production hours

In order to make the allocations easy to follow, you will see that the company is a service company and that the firm has three partners (administrative overhead) and some relatively fixed overhead in the form of vehicle leases, maintenance and so forth.

The direct labor (production labor) comes from five employees who—to make it simple—all work exactly 200 hours per month and all make exactly the same rate—$10 per hour. This also gives “production” a known capacity—1,000 hours per month.

The partners have kept good track of their history over the last six months and have also done enough market research to have a good handle on the size of the market they are serving. They know, therefore, how many of each kind of job they have done each month (on average), as well as the market potential for the kinds of jobs they do.

STANDARD COST ALLOCATIONS (by Job)

In an attempt to leverage what they have learned by capturing data about past performance and, of course, to improve profitability, the partners do an analysis that includes a standard allocation of overhead to each job.

From this analysis, they discover that their most profitable jobs are landscaping jobs ($35 per job), followed closely by window cleaning jobs ($30 per job). So, they decide to satisfy the market demand in that order, using the resources they have (1,000 hours of production time).

Before we move on, note that with their present product mix, the company is producing a profit of $4,100 per month ($49,200 per year).

The results of this action are shown here:

Upon first glance, it appears that this has been a great move. Based on the calculations in the table, profit has moved from $4,100 per month to $7,200 per month!

Again, the problem is that since NO plumbing or gutter guard jobs were done, some of the overhead (allocated at $90 per job) was not absorbed in the calculations. The total overhead is $18,000 plus $9,000, or $27,000. But the 220 jobs only absorbed 220 times $90, or $19,800 in overhead. That leaves $7,200 in overhead NOT absorbed. Take that $7,200 away from the calculated profit of $7,200 and the company is actually worse off (zero profit) after having reallocated its resources to what appeared to be the “most profitable jobs.”

---

[To be continued—be sure to watch for Part 2!]

Consider the possibilities (especially now, in these challenging times)

A recent survey of published results by manufacturing and service companies[1] that have applied constraint management methods effectively shows:

• A mean reduction in lead time of 70%
• A mean reduction in manufacturing cycle times of 65%
• A mean improvement in due-date performance of 44%
• Mean inventory reductions of 49%
• A mean combined financial improvement (revenue, throughput, profit) of 76%

[1] Mabin, Victoria J. and Steven J. Balderstone, The World of the Theory of Constraints: A Review of the International Literature, St. Lucie Press, Boca Raton, FL, 2000

[Excerpt from Schragenheim, Eli and H. William Dettmer, Manufacturing at Warp Speed – Optimizing Supply Chain Financial Performance, St. Lucie Press, Boca Raton, FL, 2001]

---

If you would like help getting started with apply constrain management to your business for rapid ROI and ongoing improvement, please contact me. Find me on LinkedIn.

The Problem with Manufacturing Software

Maybe I am just too cynical. It is likely that I am. However, whenever I work with clients who are implementing (or thinking about implementing) ERP software that includes a manufacturing suite and further, when this client is implementing the manufacturing suite because they "need to get a better handle on manufacturing costs," I warn them: "The problems with implementing software that helps you calculate your cost of manufacturing are two-fold: first, the system will produce a lot of numbers and reports for you and, second, you will believe the reports!"

"So, why is this a problem?" I hear you ask.

In order to answer that question, let me take you through the steps that a client is going to go through in order to implement their new manufacturing suite before they are going to start getting reports and numbers coming out of the system.

In even a relatively simple manufacturing suite, there are dozens of parameters involved. These parameters are used in various places. However, for our purposes, we will just concern ourselves with the few parameters that might be found on a typical "routing" or "router" – the part of the data that tells the system what steps must be executed – and in what sequence the steps must occur – in the manufacture of any given item. Consider the following table:

	Parameter	Purpose	Source and Comments
1	Move Hours	Used by APS (advanced planning and scheduling) to calculate the time it will take to move a unit of production (piece) between operations	Most organizations have never tracked this, nor even given much consideration to "move time" in their operations. Therefore, this is usually a very round "guess-timate" provided to the new manufacturing suite from "tribal knowledge."
2	Queue Hours	Used by APS to calculate the time a unit of production will sit in its queue waiting for the pending operation to actual work on this particular piece	Again, this is usually a very round "guess-timate" provided to the manufacturing suite from "tribal knowledge."
3	Set Up Hours	Used by APS for scheduling purposes, but also used by manufacturing costing to calculate the labor costs and fixed overhead that should be allocated to a production run for the time spent setting up to run a particular operation	The values provided to the new manufacturing suite for the time it takes to set up for a particular operation will probably be pretty close, even if they do come from "tribal knowledge" with no formal calculations underlying them. Likewise, the dollar-costs for the variable labor will also probably be pretty close to the dollar-amounts per set-up. The problem area is going to be fixed overhead absorption rates. These are problem because the actual amount of fixed overhead dollars that should be absorbed will be dependent upon the number of set-ups that occur. If there are more actual set-ups than the number of set-ups used to make the calculations, fixed overhead (and variable labor) will be over-absorbed, and if there are fewer actual set-ups, fixed overhead (and variable labor) will be under-absorbed. One thing of which you may be pretty certain – the number will never be the right number. That is, the actual number of set-ups and the actual duration of the set-ups will virtually never coincide precisely with the numbers used to set the parameters in the software.
4	Reset Hours	See Set Up Hours above	The problems with Reset Hours are precisely the same as with Set Up Hours above.
5	Pieces per Reset	Used by APS and manufacturing costing to determine how many "resets" were performed during each reported production run	This is just one more parameter that contributes to the calculation of manufacturing costs. The relative accuracy of the calculated cost versus the true cost will be entirely depend on the following factors: Was the number of "resets" actually performed exactly as estimated in the creation of the routing? Did the "resets" actually take the precise number of hours allotted for each "reset"?
6	Scrap Pieces	Used by APS and manufacturing costing to determine how many pieces had to be "processed" in order to produce the number of usable pieces required	This parameter also is based on averages. Therefore, if the actual scrap was different from the averages, incorrect costs will be assigned to WIP and, ultimately, to finished goods. If methods are provided by the manufacturing software to capture actual scrap by production run then, most likely, the differences will show up in variances – yet another confusing data point to unravel for decision-making.
7	Production Rates (pieces/hour or hours/piece)	Used by APS for scheduling purposes, but also used by manufacturing costing to calculate how much labor and fixed overhead should be calculated into the manufacturing cost of each operation	Even if the averages used for this parameter are pretty accurate, they are just that – averages. This means that, while they may be reasonably accurate on average over a large sampling of data, the factor will be actually wrong (inaccurate) for virtually every actual production run (which will almost never hit precisely on the average used to set the parameter in the software).
8	Production Effective Rates (percent of "standard" rates)	Companies frequently have "standard" rates per manufacturing operation based on some (frequently unknown) factors. However, they realize that in the process of actual manufacturing the operations generally do not hit this rate. As a result, they may apply a "effective rate" factor to the "standard rate." This factor is used by both APS and manufacturing costing.	Since this factor is taken into account in the calculation of manufacturing costs, it is subject to the same weaknesses previously listed for other parameters.

So, let me get this right…

All right. Let us start with this sampling of eight data points in a typical routing for manufacturing. Remember, these eight data points are repeated for each labor step in the routing. So, if a complex item has, say, 30 labor steps, that means that these data points are going to be used in 240 calculations associated with determining what will appear on reports and may be affecting the value of inventory, as well.

These eight data points – most of which are "guess-timates" or averages to begin with, will next be mathematically compounded against other data points related to:

• Variable labor absorption rates
• Variable labor overhead absorption rates
  
• Fixed overhead absorption rates
  

Since these absorption rates must be calculated based on other "averages" or "guess-timates" as to production quantities per period (e.g., month, quarter, year), we may safely assume that these absorption rates themselves will never be correct. We may say this in a mathematical sense that, in order to be mathematically correct the actual production during the period must match precisely the estimated production during upon which the absorption rates were calculated for the whole organization. Furthermore, the actual fixed overhead or variable labor expenses destined for absorption must match precisely the estimated fixed overhead or variable labor expenses used to calculate the absorption factors. The statistical likelihood of this occurrence is so close to zero as to be the statistical equivalent of zero. Therefore, we may properly say, these calculations will never be "correct" in actuality.

Unfortunately, having populated their new manufacturing suite with "averages" and "guesses," when the official-looking reports come out the other end, far too many executives and managers actually believe what the reports say. Worse! They actually begin taking action on the results presented by their costly manufacturing software as though the data reported is "God's truth" in print.

A simpler solution

Before you invest from $100,000 to $1 million or more in the purchase and implementation of a manufacturing suite of software, allow me to suggest some calculations that you and your management team can do simply in a spreadsheet (or on a napkin at lunch).

Try this simple formula for any item in your system:

T = R – TVC
where T = Throughput,
R = Revenue, and
TVC = Truly Variable Cost

Now, for almost any item in your manufacturing operations, you – or someone on you management team – can come pretty close to calculating the Throughput (T) value without getting up from the table. Many times this calculation can be done with reasonable accuracy without ever going to your present accounting system to get "costs."

Forget about all those "allocations" and "absorptions" of fixed overhead! They don't really happen and they can make you believe things that aren't true. Your payroll and fixed overhead do not vary directly with each unit of production – even though that's what most manufacturing software wants to make you believe in their costing and variance reports.

The previous formula is good for looking at a individual product or product family, but how about looking at overall operations?

Here's another simple formula to help you do that:

P = T – OE
where P = Profit,
T = Throughput, and
OE = Operating Expenses

Operating Expenses are, essentially, everything that you pay out that is not TVC.

We will look into this further in another post. Stay tuned.

©2010 Richard D. Cushing

The New ERP – Part 35

Death by data

Writing for the Aberdeen Group, Matthew Littlefield and Shah Mehul suggest, "The only way for manufacturers to achieve world-class performance [is] by providing greater visibility into what [has] long been the black box of production. And the only way to do that [is] to start collecting a lot more data on work in process (WIP)." (Littlefield and Mehul 2009) This is an all-too-common misconception that originated long before the inception of the computer, but has been dramatically augmented and expanded since computing power was made available to the business at low cost and on an unprecedented scale with the introduction of the personal computer.

A never-articulated, but oft-held, belief amongst business executives and managers is that more data leads to better management. This thought has been sometimes carried to the extreme in the minds of some executives – and fully supported by their all-too-willing IT departments – to the point that the concept may be formulated along the following lines:

1. More data will help me make better decisions
   
2. Better decisions means that, as a manager, I will be more effective and make fewer mistakes
   
3. If I can know "everything" – have all the data – about my operations, I can manage flawlessly
   

Even as I write this, I am certain that there are business owners, executives and managers busily scouring the Web for new "business intelligence" tools as the next real wave in ERP.

Nevertheless, all the data can tell an executive is what has happened. Data, by its very nature, is entirely historical. (Yes, there are "forecasts," but forecasts – if they are known for anything – are best known for being wrong. Not a reputation likely also sought by executives and managers in pursuit of "flawless" management.

What the historical data cannot tell the executive is, "What lever should I push or pull to produce some particular outcome in the future – an outcome that assures improvement and not just added cost, expense or consumed capital?" Only a sound theoretical framework about how the executive's "system" – read: whole organization – works (or fails to work) can aid him or her in finding "the right lever" and applying the correct amount of force in the proper direction.

Employing reams of data will not keep you and your management team from spending precious time, energy and money optimizing the efficiency of departmental silos while reducing the efficiency of the organization as a whole. Investments in business intelligence in the absence of a sound theoretical framework will not prevent you and your managers from building work-arounds to keep work moving instead of solving problems that repeatedly delay revenues or disrupt operations. In fact, data – wrongly understood and improperly applied – may actually move your management team to take actions that sacrifice quality and lead time in a mistaken attempt to increase production or meet standard cost goals.

What's wrong here?

As H. Thomas Johnson, professor of Business Administration at Portland State University, puts it, "Causing [such] destructive practices is the assumption that financial information not only defines the purpose of the business, it also provides the primary means to control the financial outcomes of a business…. A key reason [that] American companies fail to emulate Toyota's long-term financial results is their belief that managers can use financial targets as 'levers' to control those results." (Johnson 2006)

Professor Johnson's argument is precisely the reverse of that stated by Littlefield and Mehul. Johnson argues that U.S. executives and managers tend to believe that they can employ relatively linear and one-dimensional data – the data they use to report on the financial performance of operations – to "understand, explain, and control" the results of those operations, "even though the results emerge from nonlinear and multidimensional operations." Toyota's executives and managers do not make this same mistake.

In fact, while Littlefield and Mehul state that "world-class performance" can only be achieved by companies developing systems to give them "greater visibility into… the black box of production," Toyota has, in fact, achieved "world-class performance" by virtually assuring that accounting has no visibility into "the black box of production." In Toyota's arrangement, corporate finance knows only two things about "the black box of production": 1) what goes in, and 2) what comes out. Everything else is invisible to "accounting." In fact, it may be because "Toyota makes virtually no use of management accounting targets (or 'levers') to control or motivate operations" that they have achieved financial performance levels that are "unsurpassed in its industry." (Johnson 2006)

Understanding your operations

Inside "the black box of production," Toyota's managers are highly visual in their management style. They do not believe that they "know" or "understand" what is happening on the shop floor simply because they have worked in the plant ten years, or 20 years, or more. They believe that to understand how to improve again and again, they must thoroughly understand what is happening today – everyday. Toyota managers employ genchi genbutsu ("going to the place") to see first-hand where and why there is any delay or disruption in production of quality products. These managers understand that the sought-after financial "results ultimately emanate from, and are explained by, complex processes and concrete relationships, not by abstract quantitative relationships…." (Johnson 2006)

Whether you and your management team choose to employ the Toyota method of genchi genbutsu and asking "Why" five times to get to the root of what needs to change, or if you choose to employ the Thinking Processes (as we have discussed elsewhere on this site and in this series on The New ERP – Extended Readiness for Profit), do not fall for the line that "more data will help you manage better." Avid IT staffers aided by value-added resellers (who genuinely believe the mantra to be true) are more than happy to have you spend your money on systems to collect, organization and report on more and more data. However, if you do not yet understand your "system" thoroughly – if you have not yet developed a sound theoretical framework by which to manage your enterprise – most or all of what you spend to obtain "more data" will be wasted.

©2009 Richard D. Cushing

Works Cited

Johnson, H. Thomas. Manage a Living System, Not a Ledger. December 2006. http://www.sme.org/cgi-bin/find-articles.pl?&ME06ART83&ME&20061210&&SME& (accessed November 18, 2009).

Littlefield, Matthew, and Shah Mehul. Operational Excellence in the Process Industries: Staying Profitable Through the Downturn. White paper, Boston, MN: Aberdeen Group, Inc., 2009.