- The Demand Driven Institute proposes a master’s thesis at MIT that proves DDMRP’s efficacy.
- We analyze this thesis.
We analyzed DDMRP in the article Repackaged Lean as DDMRP and concluded that it is not an improvement on MRP and is just repackaged Lean with a few tweaks. In this article, I analyze the validity of an MIT master’s thesis, which is used to demonstrate the validity of DDMRP.
Our References for This Article
If you want to see our references for this article and related Brightwork articles, see this link.
I have to state that I have had aggressive debates with Chad Smith and Carol Patak – the Demand Driven Institute or DDI leaders, and various DDMRP devotees. I have also been on the receiving end of a large number of personal attacks from DDMRP proponents. The intent appears to be to censor those that oppose DDMRP or that are unconvinced by DDMRP. This is somewhat similar to debates with Six Sigma proponents but dialed up a few notches. However, in this article, I analyze each web page on DDMRP vendors and consulting firms on their own merits and based on what they publish on the topic of DDMRP.
The Validity of DDMRP MIT Research
Chad Smith’s research refers to the master’s thesis titled Investigation of Potential Added Value of DDMRP in Planning Under Uncertainty at Finite Capacity. He states that this partially proves the benefits of DDMRP.
We reviewed this publication in the article How Accurate is the Criticism of Lokad DDMRP Video?
Repetition of What DDI States About DDMRP
This is one of the first quotations in the research.
In 2011 a new planning methodology called Demand Driven MRP (DDMRP) was introduced in response to the new dynamics of supply chain complexity. DDMRP is a multi-echelon supply chain planning approach that combines the best of lean, MRP, six-sigma and the theory of constraints. It relies on the idea that ROI comes from emphasizing the flow of product to the market rather than mere unit cost reductions. DDMRP proposes an intuitive way to manage flows of products and relevant information by strategically positioning decoupling points and managing those with clear inventory policies. DDMRP has a particular focus on managing variability and planning and execution priorities.
This is taken directly from the DDI website. So this is DDI’s assertion.
The Demand Driven Institute (DDI) has published results of DDMRP implementations that show an increase in service levels by 13%, reduction in inventory by 31%, and a decrease in lead times by 22% (Camelot, 2019). However, these are median results and different industries can have different results.
First, no research produced by any IT consulting company can be trusted. Camelot has a consulting practice that sells DDMRP services.
Consulting firms cannot do research. They are continually introducing “research,” which is in reality promotion. In the article Why PwC’s Research Fellows are Fake and Pretend to be Academic, we cover how PwC creates fake academic titles for people posing as researchers. It is well known that any audit results that one desire can be purchased from PwC. On the consulting side, PwC does not produce research — they produce marketing collateral.
I have never come across an IT consulting firm that would not rig any result to increase sales. I was pressured to rig results for several consulting companies. Early in my career, when I had very little leverage in the job market as I had little work experience, I buckled and presented information that I knew was false for KPMG and Accenture. Beyond this, both firms would present assertions that they did not have any support for making, but which they said when they presented
“No one seemed to challenge.”
Therefore, both KPMG and Accenture would make strongly declarative statements that they knew they had no support for, and which sounded like they did have evidence to present — and if challenged, they would say
“This is what we have seen at clients.”
As far as I could tell, these companies present false results as part of their normal operations. I would never take a study by any company like this seriously.
Secondly, the study in question was for only a small number of product locations and allowed the effort to be placed into the DDMRP run, but for the MRP run to be performed without any assistance. This thesis does not point out these observations.
Long Term Forecasts are Impossible?
The founders of DDMRP, Ptak and Smith (2011), have emphasized that for an MRP system to run, actual customer requirements is required. However, due to lead time, it is impossible to only base the plan on actual demand. This requires the use of forecasted demand. Burbridge (1980) states that it is impossible to make accurate forecasts for long periods. Therefore, incorrect forecasts are fed into MRP systems in place of actual demand causing nervousness.
The forecasts are not incorrect. Each forecast has a specific inaccuracy. MRP and other supply planning systems account for this inaccuracy with safety stock, which is not mentioned in this quotation. There are just many problems with the assumptions accepted by the authors Leo Ducrot and Ehtesham Ahmed in this quotation.
Safety Stock Can Amplify Instability?
Among the proposed solutions to handle nervousness, the most commonly used are safety stock or safety lead-time or safety capacity. Ho et al. (1995), Whybark and Williams (1976), and New (1975) maintain that safety stock is the preferred technique to control quantity uncertainty and is the primary protection against overall uncertainty in the system. However, a study showed that safety stock could also, in certain circumstances, amplify the variability and the instability in the system (Sridharan and LaForge, 1990).
How would safety stock amplify variability? Safety stock accounts for variability. One might be able to find some study that shows this under “certain circumstances,” However, that is not the norm.
APS Improves Outcomes Over MRP?
In the literature, we find that APS provides better results than MRP. In a study conducted by Moscoso, Fransoo, and Fischer (2010), the APS implementation had a positive result. Backlogs were reduced by 84% (in three months) and 97% service levels were achieved. However, they also found that average production lead time increased by 15%. Hvolby and Steger-Jensen (2010) in their study found that delivery accuracy went up from 79% to 99% after implementing an APS system.
Do Leo Ducrot and Ehtesham Ahmed have any work experience on supply planning projects — because if they did, they would probably have a view on the topic. I have over twenty years working on supply planning projects. Why am I reading a master’s thesis where the authors have to work entirely on their reading?
APS systems have been quite problematic, with most cost optimization projects (so APS) underperforming MRP. Heuristics for supply planning essentially emulate MRP, and allocation is only applicable for companies, such as high tech, where order allocation is required. However, allocation is not particularly logical or intelligent. As for multi-echelon inventory optimization, the results have been generally poor as the applications are considered too challenging to implement. My experience indicates that most APS implementations did no pay back their investments.
Is Conventional Planning Obsolete?
Ptak & Smith (2011) stated that the hypotheses and rules used to design ‘conventional planning’ were no longer valid because they rely on low complexity, low variability, and high customer tolerances.
I don’t think this is true. If one has a higher variability, that is managed with more stock. There is no evidence that all conventional planning has been made obsolete.
Buffer Stock…Not Safety Stock?
DDMRP proposes to reduce the variability transferred between the levels by strategically positioning dynamic buffers and promoting a flow-centric approach.
That is precisely what safety stock does. DDMRP seems to be renaming things. “Flow centric” is code for Lean, or not using forecasts.
Lower Stock Levels are Better?
It can be observed that financial performances of companies with lower stock levels better than that for companies with higher stock level (Obermaier 2012).
The authors are making an immense amount of assumptions and then pointing to one study to support each assertion. As an aggregate, that may or may not be true — but it isn’t easy to see how that relates to a specific situation. There could also be other reasons for this relationship.
MRP Better for Stable Demand?
In cases where demand is constant, MRP performs better with real demand and few forecasts for a short period of time, and is able to accurately absorb spikes. With seasonal variations, DDMRP is more suitable (Miclo et al., 2016). In any case, MRP requires safety stock to account for forecast variability over production lead time (Shofa & Widyarto, 2017) but Miclo et al. (2016) observed that with DDMRP the stock levels are flat instead of following a normal distribution.
This makes little sense. The authors are confusing constant demand or the demand pattern with predictability. If a seasonal pattern is relatively predictable – which many are, MRP has no problem with the pattern. Secondly, MRP systems can also use reorder points, which are activated as the planning method if the forecast error is too high. The authors seem to be leaving out these, and they are available in all supply planning systems.
MRP Has Poor Cash Flow?
MRP has poor cash flow, and service levels keep on declining despite high levels of inventory; revenues also keep on declining for the company (McCullen & Eagle, 2015). Shofa & Widyarto (2017) found that DDMRP compressed the lead time by 94% for a company, McCullen & Eagle (2015) observed that service levels were increased from 90 to 99% for a company and there was a 35% reduction of inventory levels. Shofa, Moeis, & Restiana (2018) observed an average inventory reduction by 11% and stability in inventory levels with DDMRP.
What? Why are service levels continuing to decline with high levels of inventory? Is this for poorly maintained MRP systems?
The Rest of the Study
The study includes a survey as well as a simulation.
The authors put a lot of work into this study — that was clear. However, the study is problematic because the authors don’t appear to have experience in the field. Naturally, they are young, as this is a master’s thesis. After I graduated with my master’s degree in Logistics, I don’t recall many people caring much about what I thought about supply chain planning. When I went into consulting firms, I was just a quantitative analyst. Outside of running numbers, no one asked me what I thought I did not have work experience.
The authors make several assertions that are not true — and they point to individual studies to support their assertion — but the problem is these studies contradict my work experience with supply chain planning systems.
The study was well written, but it critiques MRP systems in a way that is not believable. I have managed MRP systems in precisely the environment that the authors say won’t work — so high variability, problematic or intermittent demand, with a high number of new product introductions. My biggest challenges were not MRP per se. They were related to educating my client, forecast error testing, to parameter optimization. This may seem incredible to many, but most companies have no way of performing a comparative forecast error measurement. They don’t realize their forecast error measurement does not serve them. They set safety stock at a product location combination without considering the impact on the other product locations. Supply planning systems are poorly maintained. Earlier in the quote by Mike Bradshaw, he made similar observations about companies’ poor state of affairs. However, he seemed to be referring to smaller companies that did not have automated planning systems of any kind. But my observation is that measuring MRP and forecasting systems that have extensive underinvestment will not tell you much about how effective the methods are. If, for example, you want to critique the effectiveness of hand gliding equipment, you can’t learn this by measuring the outcome of taking out a hang glider that was never maintained and with a pilot who has never taken a class in hang gliding. Observing that the pilot crashed as soon as they took off from the mountain top is not a proper measurement of modern hang gliding effectiveness.
MRP systems are victims of underinvested — with companies preferring to hire inexpensive and inexperienced resources. Any MRP system I have seen can be dramatically improved by simply investing in the system. And DDMRP requires similar, and I would argue even higher investment (particularly around the management of subcomponents.