dot 3rd eye: information + insight = innovation

Buying Flexibility to Manage Growing Supply Chain Risk

2010-08-03T16:05:00.000-07:00

A Fresh Approach to Outsourced and Offshore Manufacturing

Amar Singh, CEO, Amitive, Inc.

Narayan Venkatasubramanyan, President, dot 3rd eye, inc.

Idea in brief

· Cost-cutting drove companies to outsource and offshore manufacturing. The resulting longer lead times, and delays and distortion of information made the supply chains inflexible, thereby lowering their ability to respond to demand volatility. Recent changes in consumer behavior (driven by recession) and shorter product lifecycles are driving higher demand volatility. This inflexibility in the face of increasing volatility magnifies supply chain risk and actually increases cost.

· Traditional decision-making criteria for supply chain design (outsourced and/or offshore manufacturing decisions) underplay the impact of demand volatility on cost. They mostly focus on the trade-off between manufacturing and logistics costs.

· The proposed approach goes beyond cost comparison; it offers a framework to classify products based on demand volatility and product lifecycles. This results in a differentiated strategy for supply chain design that balances cost against supply chain risk. Readers familiar with the theory of real options may recognize this as a practical application of that idea.

Current Approach to Supply Chain Design

The analysis that is done to justify outsourcing of production and moving it offshore is largely based on the trade-off between the cost differential per unit of production and the incremental cost of logistics. This thinking is illustrated in the article titled “Time to rethink offshoring?” (The McKinsey Quarterly, September 2008). The authors argued that the increase in energy prices during the summer of 2008 has reduced the cost advantage of manufacturing in the Far East. Their argument can be summarized as follows:

v If the cost of manufacturing in the US a product is high, the likely savings from manufacturing it near- or offshore is likely to be high

v If the product weight is high, its logistics cost per unit is likely to be high

v The decision to move manufacturing offshore is the result of the relative values of these costs:

Ø If the manufacturing cost is high and the logistics cost is low, it makes sense to build this product in Asia

Ø As logistics cost increase relative to the cost of manufacturing, it may be more economical to build this product in Mexico

Ø Only for low-cost products that are bulky do they recommend manufacture in the US.

Based on their analysis, they have developed breakeven curves that carve up the 2-dimensional space of manufacturing cost and product weight into “Build in Asia” vs. “Build in Mexico” vs. “Build in US”. The increase in energy costs between 2005 and 2008 has shifted these curves in a direction that favors the US and Mexico options over Asia. This is captured in the following graphic:

The reasoning is simple and linear: if it is cheaper to build and ship one unit from China than to build it in the US, it must mean that we ought to shift it all production to China. If we can find a contract manufacturer who can do it for even less, then it must mean that it is time to outsource all production.

Mitigating Risk through Flexibility: An Example

The above decision model overlooks one important reality: demand volatility. More specifically, it overlooks the reality that demand forecasts are uncertain and, the further out we look, the more uncertain is the forecast. In this section, we provide an example of how accounting for that effect leads to a very different and non-intuitive conclusion.

Imagine a widget supply chain that consists of suppliers who are contract manufacturers of widgets based in China, a widget brand owner who buys them in bulk from these Chinese factories and then resells them into widget retailers in the US. Until recently, this brand owner used to manufacture widgets in the United States and, to some extent, in Mexico but the lower cost of manufacture in China has led to a restructuring of the supply chain and the outsourcing of all production to Chinese suppliers.

Given their low margin, widgets have to be transported by sea and then land. Assume that the lead times are as follows:

	Lead Time (weeks)
Supplier	3
Brand owner	8
Retailer	2

The supplier’s lead time includes the lead time in production and transportation to the port in China. The brand owner’s lead time includes the time across the ocean, clearance through customs, and over-land transportation to their distribution centers. Finally, the retailer’s lead time includes the movements of widgets from the brand owner’s DCs to their warehouses and finally to the store-shelf.

Given this, it is clear that a widget that commences production today does not find its way to its final customer for 3 months. In other words, the supplier is starting production today for a demand that is forecasted to be realized a quarter out.

It should come as no surprise that forecasts of demand get progressively poorer the further out we look. Therefore, any brand owner who builds to stock based on a forecast is exposed to the risk implicit in making a commitment to a number that is fraught with uncertainty. If the cost of a lost sale is considered to be high, the brand owner is more prone to err in the direction of having the supplier make too much than too little. On the other hand, if the loss due to obsolescence (or price protection) is too high, the brand owner will tend to shoot low. In any case, the brand owner works under the misguided belief that some mix of excess/obsolete inventory and expedited production is an inevitable consequence of forecast error.

Notice how the length and inflexibility of the lead times that are baked into the design of this supply chain guarantee that forecast error translates into increased cost in the form of lost sales or excess/obsolescent inventory.

Analysis of an Alternative Supply Chain Structures

Clearly, the logic for outsourced and offshore manufacture is inescapable but, we argue, not overwhelming. Unfortunately, simple linear, deterministic models have driven decision-makers to extremes. What we argue for is a strategy that balances the low-cost, long lead-time option with the high-cost, responsive option. A careful analysis of the lead times and delays involved in outsourced, offshore manufacture, and a quantification of the resulting increase in noise in the signal that is used to start manufacture will naturally lead to such a conclusion.

Using a simplified numerical example, we will reveal the opportunity to borrow a key principle from other disciplines that have had to contend with uncertainty: the use of options.

Imagine that the widget in question has a life-cycle of 1 quarter, after which it is replaced by a newer, better widget. Assume that the season for the widget begins on July 1, 2009 and ends on September 30, 2009.

Given the lead times involved, the brand owner has to place orders upon the supplier on April 1, 2009 so that the first batch of product arrives on the store-shelf at the beginning of the season on July 1. As a result, this order will be based on a forecast of demand 3 months out. Assume that on April 1, 2009, the most likely estimate of demand for widgets is 100 units per week over the 13 week period starting July 1, 2009. Clearly, this is a guess. It is understood that the actual demand could turn out to be higher or lower.

For the purpose of this example, we assume that, on April 1, all we know about the demand starting July 1 is this: the demand will be 100 units/week with a 50% chance but it could be 80 units/week or 120 units/week with equal probability. This is captured in the table below:

Demand (units/week)	Probability
80	25%
100	50%
120	25%

For simplicity, we assume that these probabilities remain unchanged until the season begins. As soon as the season begins, the weekly demand is known with certainty[1]. Unfortunately, by then it is too late to do anything in response because of the lead times involved.

Assume that it costs $150/unit to manufacture the widgets at the supplier and another $50/unit to ship widgets to the retailer. The brand owner gets $300/unit for each widget sold by the retailer (i.e., a margin of $100 per widget sold). Widgets that remain unsold at the end of the season have no residual value[2] (i.e., a margin of -$200 per widget left unsold).

The brand owner is faced with a multitude of possibilities, three of which are summarized in the following table:

Brand owner asks supplier to make (units/week)	Demand turns out to be (units/week)	Units sold per week	Margin on units sold – Cost of unsold units ($/week)	Probability	Expected result ($/week)
80	80	80	8,000	0.25	8,000
	100	80	8,000	0.50
	120	80	8,000	0.25
100	80	80	4,000	0.25	8,500
	100	100	10,000	0.50
	120	100	10,000	0.25
120	80	80	0	0.25	6,000
	100	100	6,000	0.50
	120	120	12,000	0.25

Here is an alternative way of thinking about this:

· Every one of the first 80 widgets made each week is guaranteed to be sold; hence each will earn $100.

· Every widget beyond the 80^th and up to the 100^th, has a 25% probability that it will remain unsold, i.e., cost $200, and a 75% that it will be sold, i.e., earn $100. This means that the expected returns from building any unit between the 80^th and the 100^th is a net profit of $25.

· Every widget beyond the 100^th has a 75% probability that it will remain unsold, i.e., cost $200, and a 25% that it will be sold, i.e., earn $100. This means that the expected returns from building any unit beyond 100 is a net loss of -$125.

Based on this analysis, we can conclude that the brand owner ought to place orders with the contract manufacturer for 100 widgets every week, starting April 1, 2009. The simplicity of this analysis is rooted in the inflexibility of the supply chain.

This reasoning tends to overlook the effect of long lead times on the quality of forecast signal used as the basis for launching production.

Now, consider how the situation is altered fundamentally if there were a more responsive (albeit expensive) alternative to the long lead-time but inexpensive option. To make the illustration stark, we assume that the brand owner has an onshore factory that can get widgets to the store-shelf instantaneously, starting the first week of the season[3]. Assume that the cost of production at the onshore factory is 50% higher, i.e., $225/widget. Because the factory is onshore, the logistics cost are only $25/widget. This means that every widget made at this location and sold earns $50.

How does this alter the reasoning?

Clearly, there is no improvement in the quality of the forecast information prior to the season, so there is no incentive to shift any production to the higher cost alternative before the season begins. But (in our world of extremes) the forecast picture clarifies dramatically at the beginning of the season: we know the demand with certainty. This means that any shortfall that results from the demand exceeding the flow of widgets from China can be made up by local production. Let us see how that alters the choices laid out above:

Brand owner asks supplier to make (units/week)	Demand turns out to be (units/week)	Units sold per week	Margin on units sold – Cost of unsold units ($/week)	Probability	Expected result ($/week)
80	80	80	8,000	0.25	9,000
	100	80+20[4]	9,000[5]	0.50
	120	80+40	10,000	0.25
100	80	80	4,000	0.25	8,750
	100	100	10,000	0.50
	120	100+20	11,000	0.25
120	80	80	0	0.25	6,000
	100	100	60,000	0.50
	120	120	12,000	0.25

Notice that the availability of a more responsive option has tipped the balance in favor of local production, even though it is more expensive!

The question that now arises is this: what is the cost of maintaining an option to produce locally? In this example, the availability of a local option increased the bottom line by $500/week, even though a superficial analysis may dismiss this alternative as being more expensive.

Recommendations

Although the numerical example was unrealistic, it is thought-provoking because it defies intuition. This is because most “intuitive” analysis is limited when it comes to understanding how forecast errors play out over time. Before decisions regarding outsourcing and offshore production are made, one needs to ask a few questions:

· What is the effect of outsourcing and offshore production on the ability of the supply chain to respond quickly? Keep in mind that outsourcing introduces delays in the flow of information that generally tend to go unaccounted for.

· How much more expensive is a more responsive alternative to outsourcing and offshore production? How much more responsive is a more expensive alternative to outsourcing and offshore production?

· How much is responsiveness worth? This is a particularly important question for products with short life-cycles. This is because such products live and die by the success of their launch and phase-out; these are the periods when the ability to respond to the market is crucial.

Conclusion

As supply chains experience a growing volatility in demand, they must make adjustments to improve their ability to respond rapidly. We contend that the conventional structure of supply chains tends to magnify the effects of volatility in demand because of long lead times and information delays. Lead times grow due to outsourcing and offshore manufacture; demand becomes more volatile due to changing customer behavior; shorter life-cycles alter the calculus of managing uncertainty in supply chains through inventory. We need to structure supply chains differently in the light of these factors. Tools available to practitioners of supply chain design need to embrace the power of real options to help dampen the effects of demand volatility. Much of what we propose here is well within the realm of the possible, even though some of it may appear non-intuitive at first sight.

[1] This assumption simplifies the analysis significantly. In reality, the revised forecast of demand once the season begins continues to remain uncertain, but it generally tends to be far better than the pre-season forecasts. Early sales of a product are usually a very good indicator of the response of the marketplace to that product. Once the early sales figures are in, there is a significant improvement in the quality of the forecast for the remainder of the season. For simplicity, we take this phenomenon to the extreme by assuming that all uncertainty in demand disappears once the season begins.

[2] This too is a simplifying assumption; incorporating mark-downs will only complicate the model without fundamentally altering the argument.

[3] Clearly, this is an unrealistic assumption. We use it purely for ease of exposition.

[4] 80 units made in China + 20 units made locally.

[5] Explanation: In this case, the brand owner asks the Chinese supplier to make 80 units/week starting April 1, 2009. By July 1, 2009, first 80 units reach the store shelves. Each subsequent week, another batch of 80 units arrives. It costs $200 to get each of these widgets to the store-shelf. Since the demand turns out to be 100, the brand owner triggers an additional 20 units/week from his local facility at a cost of $250/unit. In total, the cost is $21,000/week (i.e., 80x$200 + 20x$250). These 100 widgets earn $30,000 in revenue. This results in a margin of $9,000 per week.

ackoff's fables

2010-05-09T19:00:00.000-07:00

in a prior entry titled "Operations Research is dead", i had proposed to write a series of entries each touching on a particular part of a pair of papers written by ackoff in the late 70s.

the early entries will focus on his critique of OR as it appears in the paper titled "The Future of Operational Research is Past".

i will begin with what his students called one of "ackoff's fables". it focuses on a review of the work done by a team of highly-educated practitioners of OR. i have reproduced below a slightly edited version of what appears in a section of that paper and interspersed it with my comments:

"... they showed me what they referred to as their "evaluation of the results." Over a reasonably long period of time they had recorded the decisions actually made and implemented by the responsible managers and had fed these decisions into their model and calculated the total related operating costs. They had then compared these costs with those associated with the optimal solution derived from the same model. ... The optimal solutions were consistently better than those of the managers. Using these differences they had estimated an annual saving which they had used successfully in convincing management to adopt their model and optimizing procedure."

so far, this sounds like a text-book approach to justifying the cost incurred in implementing an OR model.

"... The optimal solution of a model is not an optimal solution of a problem unless the model is a perfect representation of the problem, which it never is. Therefore, in testing a model and evaluating solutions derived from it, the model itself should never be used to determine the relevant comparative performance measures. The only thing demonstrated by so doing is that the minimum or maximum of a function is lower or higher than a non-minimum or non-maximum."

ackoff starts by making a point that many practitioners tend to lose sight: in order to justify their work, the research team ought to make their case by checking if that was indeed the best solution in the real world, not just in the fictitious world of the model. quite often, because of the shortcomings of the model, the solution is found to be infeasible. because of approximations in the measure of goodness used, the solution, even if feasible, is the best with respect to a "wrong" objective function.

nevertheless -- and this is where i find ackoff's judgment a trifle harsh -- the optimal solution to an imperfect model is often pretty close to a good solution in the real world. bridging this gap is a tricky problem that is rarely addressed well by the system; users who see the potential of a good starting point can often work wonders with it. one must presume that the team that was under ackoff's microscope had missed this subtle but all too important point. this is apparent from what he says next:

"All models are simplifications of reality. ... Therefore, it is critical to determine how well models represent reality. In this project the team had not done so."

what makes this matter even more curious is what he says next:

"... The researchers went on to tell me how the managers had invariably modified the optimal solution to take into account factors that were not taken into account in the model. Furthermore, the team confessed that it had not carried out any analysis of the nature of the adjustments made by managers or their effects. When I pressed for an explanation of this oversight, I was told that the nature of the factors considered by the managers precluded their inclusion in a mathematical model. Voilá again!"

one can't help asking: weren't these researchers even mildly curious?

this is another recurring problem with the application of optimization in the real world. inebriated by the power of their models and algorithms, OR professionals dismiss the accumulated knowledge of their intended users. instead, they go into head-to-head competition with them. they fail to recognize that their models are imperfect. but even imperfect models can be useful but not if they disregard what the users of those models know over and above what the models do.

after this, the story plays out as you would expect it to:

"Then came the climactic revelation. After about six months the managers had discontinued use of the model because of a significant change in the environment of the system. This change was political in character. When I asked why they had not tried to incorporate the relevant politics into their model, I was told that such changes are neither quantifiable nor predictable.

It should be apparent by now that if the researchers had, in fact, solved a problem, it was not the problem that the managers had."

i believe that ackoff's fable is extremely valuable but not as a nail in the coffin of OR. this is more a cautionary tale for all practitioners, a tale that ought to teach us to be humble in the face of knowledge that our models do not, and quite often cannot, comprehend.

in subsequent entries, i will touch upon some of the more serious attacks on OR that ackoff launches.

Operations Research is dead

2010-04-27T20:22:00.001-07:00

those are the words of Russell Ackoff who, according to wikipedia, is "an American organizational theorist, consultant, and Anheuser-Busch Professor Emeritus of Management Science at the Wharton School, University of Pennsylvania. Ackoff is a pioneer in the field of operations research, systems thinking and management science."

about three decades ago, ackoff -- then an operations researcher himself -- wrote a pair of papers titled "The Future of Operational Research is Past" and "Resurrecting the Future of Operational Research". (for those who are curious about minutiae, the acronym OR translates to Operations Research in the US but into Operational Research in Europe, probably for historical reasons.) these articles were published in successive issues of the Journal of Operational Research Society.

ackoff makes several damning indictments of the field of OR. it is not my intent to defend OR. it could be argued the fact that OR survives thirty years after those papers were written is defense enough. nevertheless, there are several points that ackoff makes that ought to be of interest to this audience. practitioners of OR may recognize that some of the criticism he offers is valid to this day.

the two papers are worth reading in their entirety. given that you're reading this blog, i suspect that you may not be inclined to read something quite as heavy as a pair of journal papers. in my opinion, ackoff's paper is better nibbled on and chewed carefully rather than swallowed whole. to make it easier for you, i plan to produce excerpts from those papers and comment on each of them. each entry in this series is intended to highlight an aspect of ackoff's pronouncement of the death of OR or an element of his recommendation for its resurrection. stay tuned ...

back to the roots

2010-04-10T19:32:00.000-07:00

for those unfamiliar with pune (or poona as it was known before the anglicized version was dropped), it is a city in western india that has, over the last decade or so, quietly established itself alongside bangalore and hyderabad as a hotspot for tech activity. it is the home of PuneTech.com a site that "brings together in one place all information about interesting software technology in Pune - the companies, the tech user groups and organizations, the people, and the technologies, the news and the events."

last year, amit paranajape, a friend, former colleague, and co-founder of PuneTech.com encouraged me to write a piece about "optimization" for publication in PuneTech. for reasons that may have something to do with the fact that i've made a living for a couple of decades as a practitioner of that dark art known as optimization, he felt that i was best qualified to write about the subject for an audience that was technically savvy but not necessarily aware of the application of optimization. it took me a while to overcome my initial reluctance: is there really an audience for this? after all, even my daughter feigns disgust every time i bring up the topic of what i do. after some thought, i accepted the challenge as long as i could take a slightly unusual approach to a "technical" topic: i decided to personalize it by rooting it in a personal-professional experience. i could then branch off into a variety of different aspects of that experience, some technical, some not so much.

what followed was a series of postings that began with "Optimization: A case study". in this entry, i recounted the details of one of my first engagements as a professional. it was an experience that was rich with potential for learning, perhaps made more so by the combination of boundless enthusiasm and naivete that are hallmarks of a young man who is has been recently unleashed on the real world.

i followed this with a series of articles, each of which was intended to expand on a particular aspect of the learning experience.

in the first of the follow-up articles titled "Architecture of a decision-support system", i shared my ideas on how one ought to think about decision-support systems. the ideas i present there are not exactly earth-shattering. instead, they are born from the many scars that i bear from the times when i let my technical training get ahead of the needs for decision-making in the real world.

in the next piece titled "Optimization and Organizational Readiness for Change", i reflected on the pitfalls of diving head-first into implementing a technology-based solution without ensuring that the seeds are being planted into fertile, receptive soil.

in the last of the pieces titled "Optimization: A Technical Overview", i touched upon the areas of study that have contributed to the body of knowledge and software that are the tools of my trade.

i'm sure there is much more to write about. i hope that the discussions that this entry will spawn will generate fresh insights.

why a blog?

2010-02-23T22:59:00.000-08:00

when i first started toying with the idea of having a functioning website for my fledgling company, i had this vague notion that a blog would be an integral part of it. i wasn't entirely sure why -- i had no experience with blogs back then -- but i was convinced of the idea.

upon further reflection, i realized that my instinctive decision to blog was well-founded. a functioning blog embodies a couple of things that i strongly believe in:

the act of writing down one's ideas is immensely powerful. the idea of putting things down on paper (or a screen) forces one to think through one's ideas carefully and critically. it exposes any gaps in the premise, any inconsistencies in the logic, any loose ends in the train of thought; in short, it forces us to rid our thinking of the intellectual sloppiness that one can carry around in our heads.
publishing is an obvious (though not always necessary) follow-on to writing. because of my professional network, i can publish my thoughts to an audience that can be counted on to be knowledgeable, thoughtful, expressive, and constructive. i expect my ideas to be rejected, challenged, supported, and built upon. i hope that the output of the collective will amount to something more than what any one of the participants could have achieved.

this blog is expected to work in a variety of ways:

it will serve as a sand-box in which i will play around with ideas that may have crossed my mind during the course of my professional engagements.
i hope that the people in my professional network will engage actively in the discussions my ideas initiate.
i intend to invite more active participation by encouraging the more adventurous and forthcoming members of this network to chip in with ideas of their own.
besides being a breeding ground for new ideas, this blog will also serve as a repository for interesting ideas published elsewhere as long as they are accompanied by thoughtful commentary on those ideas. i'm sure there is a lot that we can learn from works published elsewhere. this may take the form of interesting books and articles that one comes across. this may even take the form of recent events (as long as there is a bigger point to be made -- i'm sure that there are better forums for raw news).

i'm sure that the possibilities for a blog like this are endless. i'm sure my thoughts on this blog will grow and change over time. for now, this is what i think it is.

i must admit that there is no dearth of blogs written by people with backgrounds like mine. many of my former colleagues have blogs of their own. each of them brings their own personal touch to their contributions. i hope my blog will add a new dimension into this mix. it is my intent to help make this a greenhouse for new ideas. i realize that this is a lofty goal. i know i have taken a step in the right direction if i can point to one good idea that is born here ...