Electronics Production | September 24, 2009
Supply Chain Constraints – Real and Otherwise
How “over learning” and poor communications are creating the next major supply chain crisis.
In 2002 I presented a paper at an industry conference with a title that I thought was quite clever, if not the least bit descriptive of the actual contents to be presented. The paper’s title was a quote, or rather a commonly used misquote, from the comic strip character Pogo who was lamenting the devastation man has wrought upon mother earth. My paper entitled “I have seen the enemy and he is us” pondered an issue not quite as weighty as global warming, but one where the victims and the perpetrators are one and the same nonetheless. The specific subject of my paper was the massive and very painful inventory correction in the electronics industry that followed the bursting of the tech bubble in 2000 and how most of us industry participants were complicit in creating this event. Today, in an industry environment much different from that which prevailed in 2000 – 2001, a similar set of behaviors on the part of various market participants is creating a whole new problem in the supply chain. The Tech Wreck Inventory Debacle Rewind After several years of explosive growth fueled by as of yet unproven internet business models, many segments of the industry such as networking and telecom had mostly convinced itself that 30% to 50% growth was the new norm. As the growth continued capacity became a constraint in numerous supply channels. Lead times pushed out, prices shot up, and the most feared and hated word in all of electronics became as much a common part of the daily vernacular as “hello”. The dreaded word I’m referring to………allocation! Struck with fear that their revenue growth would be constrained by component supply, thus possibly plunging their stock prices to some ridiculously low valuation like 100 times revenue, many OEMs and EMS providers alike began increasing their orders in an attempt to get a larger slice of the allocation pie. Double ordering became commonplace, as did a practice of loading MRP with “independent demand”, which was simply a euphemism for “Extra Flash Memory for whatever product I might be building.” And so the bullwhip cycle developed: OEMs inflated product demand signals to EMS providers, who in turn inflated demand signals to component suppliers who were generally willing to expand capacity to sell allocated components at inflated prices. Then the roof started to cave in. Figure 1: Quote from Sloan Management Review Article on the “Bullwhip” Effect. Dr. H. Lee et. al. As many start-ups and VCs alike began to realize that “JustDogFood.com” was really not a business, public investors began to realize that the issuing of new shares by a plethora of Cisco and Amazon want-to-be’s did not actually constitute a business model. Funding began to slow, restricting the ability of companies to invest in new tech hardware. OEMs, reluctantly at first, began reducing their orders to EMS providers who began reducing their orders to component manufacturers. As a “temporary lull” began to turn into the new norm, the electronics industry was soon confronted with the stark realities – and dire consequences – of their own behavior. It was painful enough that real product demand was contracting significantly in many segments. But compound this real decline with the fact that up and down the supply chain OEMs and EMS providers were now scrambling to cancel orders and push back shipments for the artificial demand that so many companies had put into the tightly constrained supply chain. What should have been simply a really bad dream turned into a full blown nightmare as the downturn of 2000 was greatly intensified and prolonged in many segments due to the pervasive supply chain overreaction to the upside that had preceded it. Having fully participated in the tech wreck, and with plenty of scars to show for it, I personally learned a number of very hard lessons about life, economics and the meaning of personal character from the bursting of one of modern history’s greatest financial bubbles. On the supply chain side, the greatest lesson I learned from the bubble is that data is silver, information is gold, and that the various systems and communication channels employed to manage both of these failed miserably in many, many cases in late 2000. The various forecasting, planning and procurement systems employed – whether fully integrated ERP systems or homegrown tools – did not have any meaningful vertical connectivity such as is available today through systems like One Networks and others. There was no real-time visibility from the original source of demand, or from the end source of supply, and thus nobody along the supply chain had a true understanding of what the actual demand was. OEMs regularly overstated demand, fearful that they might leave upside on the table should more orders materialize. Many drove independent demand signals to component suppliers without informing their manufacturing partners, thus driving up lead times and in some cases creating a bidding war for components with their own EMS provider. Meanwhile, some EMS providers, unaware that the OEMs had inflated their numbers, took the previously rare action of adding additional demand on top of the OEM’s inflated numbers since the OEM’s forecast was increasing every single month anyway. On highly constrained or allocated components such as flash memory, some EMS providers would also load their own independent demand and decide later which high growth customer account would be the beneficiary of this component allocation manna from heaven. Figure 2: A supply chain house of cards – circa 2000 As demand started to decline, OEMs and EMS providers alike were reluctant to acknowledge that the party was over – clinging instead to the notion that demand was temporarily soft. By the time reality had set in, there were so many layers of padded demand numbers, and demand coming from so many different inflated, arbitrary or independent sources, that it was very difficult to understand the real situation in terms of either supply or demand. Through unlinked (or under-linked) systems, and by specifically and intentionally withholding information from each other, OEMs and EMS players found themselves awash with inventory. And as with every great problem, the process soon began of pointing fingers, blaming the other guy, and apportioning the financial liability for a mountain of material that was not only excess, but much of which had been purchased at inflated prices that no longer represented the market value of the material. Lessons Not Learned, and Lessons “Over-learned” Although it has been nearly a decade since the tech bubble burst, the memory of this event was still very fresh in the minds of most industry executives when the air started to leak out of another type of bubble in late summer of 2008. Having learned their lesson – or perhaps over-learned their lesson from the prolonged decline and billions in inventory write-offs from the last downturn – many supply chain participants were very quick to act at the first sign of trouble. As early as July of 2008 we began to see significant cuts to demand signals in some areas. By September, several EMS companies were making broad, across-the-board cuts to MPS and component orders irrespective of the demand they were seeing from their customers. By early November OEMs were slashing orders and forecasts to their EMS providers, some of whom reacted again by sending another wave of massive demand cuts into the supply chain. And just like that, the negative “bullwhip" effect was on. The world economy was falling apart! Cash, which had long been king, was suddenly promoted to ruthless dictator. And as every good CFO knows, cash’s number one enemy is inventory – which simply had to go to protect the balance sheet. After all, lessons were learned in the tech crash of 2000, right? In fact, the real lessons were perhaps not learned at all in the outsourced manufacturing world, and the false lessons of 2000 – 2001 were learned well beyond their usefulness. In 2000, many OEMs and EMS providers did a poor job of working together to formulate a single, cohesive plan for tackling the massive growth and continuous upside – choosing instead to act unilaterally to secure supplies, conceal information, and protect their own interests. Eight years later as global storm clouds were forming, numerous EMS companies both large and small began acting unilaterally in the supply chain to protect their own interests. This is not the least bit surprising given the amount of inventory that they were stuck with – legitimately and otherwise – the last time around. MPS and demand signals were slashed across the board and massive cuts were put into the supply chain. The semiconductor industry’s monthly order book declined a staggering 25% from August to September. Many EMS companies were just not going to wait for their customers to update forecasts before acting. As OEMs began to drop their forecasts aggressively later in the fall, another wave of cuts flooded into the supply chain. December saw semiconductor monthly order book plunge by 26.1% followed sequentially by a truly staggering 52.1% month-on-month drop in January. By March 2009 the order book had dropped to the lowest level in my lifetime, with orders coming in at just 1/5th the pace they were 12 months before. Figure 3: Semiconductor bookings percent monthly change The chart in figure 3 clearly shows the incredibly rapid pace of order declines in the semiconductor industry as demand was slashed. The rate of decline in orders is by far both the steepest and largest in my career and makes an unequivocal statement about many participants’ views on excess inventory and inventory liability. As the chart also clearly shows, these cuts were a dramatic overreaction and the industry is now scrambling to catch back up. The preliminary July 2009 data on semiconductor bookings shows a relative monthly change in that indicator of 62% - the single largest monthly increase witnessed over my career. Figure 4: Semiconductor Book-to-Bill ratio. The New Emerging Problem – The Reverse Bullwhip Although the process of wholesale unilateral demand tweaking that began in late 2008 has mostly abated, we continue to see both EMS providers’ and ODMs’ pipeline materials at what appears to be below the level of their aggregated customer demand. Executives remain cautious of the recovery, skeptical of customer demand, and highly protective of the balance sheet. The process of getting caught up to real demand is starting to manifest itself in expanded lead times for some parts as well as missed product shipments from some EMS providers. Many economic pundits are touting industry metrics such as those charted above and expanding component lead times as proof of a strong economic recovery. In reality, the situation seems to be made up of three different factors, in the following order: 1) A realignment of demand and forecasts more consistent with reality. In short, some of the unrealistically pessimistic and overly bearish “CYA on the downside” forecasts are being adjusted to reflect actual customer demand. 2) A modest recovery in numerous sectors is creating real incremental demand. 3) Capacity reductions implemented in a number of key component areas is reducing the ability of some suppliers to respond to the tremendous “bullwhip” effect of the past six to nine months, the end result of which is reflected in the chart shown above in figure 3. These issues are simultaneously mildly encouraging and mildly concerning, depending on your particular view. The real concern today lies in the current actions of many OEMs in forecasting demand to their EMS providers. Bitten badly by inventory carrying costs, forced inventory buy-backs, and massive inventory write-offs in the last downturn, many OEMs are currently being overly conservative in both forecasting demand and in placing actual POs with their EMS providers. Questioning the staying power of what may (or may not) be a nascent economic recovery, up until recently many OEMs were discounting demand signals and being extraordinarily cautious with placing EMS demand commitments that may possibly impact their balance sheet. The nearly universal chorus I had been hearing from OEMs over the past six weeks sounds something like this: "Things are soft out there! So I know that if I'm a little short on my forecast, [insert EMS provider's name here] will be able to pull in some material and deliver some upside to my numbers." The problem with this view in the current economic environment is that numerous OEMs are under forecasting demand slightly, and thus they are all counting on the exact same upside, from the exact same group of component suppliers, to meet their actual demand. This creates the front end of a negative bullwhip effect where demand actually contracts the further up the supply chain a demand signal flows. This negative bullwhip is exacerbated by the previous systemic under-driving of the pipeline by EMS providers previously discussed. Now suddenly over the past six weeks or so, OEMs are faced with potential component shortages and significant lead time increases in a supply chain that in many areas actually has some considerable slack in it. But this slack capacity cannot respond instantaneously, and just 20 to 25 weeks ago suppliers were still cutting capacity as quickly as they could. In recent weeks lead times have pushed out on a number of key components types across broad categories, and this trend is likely to continue through at least the remainder of this seasonally busy period in manufacturing. Particularly hard hit have been high density synchronous SRAM, Linear and Analog IC’s, DSPs, Standard Logic, and some Interconnect products, but the list continues to grow. Once again it seems as though a real and significant shift in demand for electronics hardware has created a problem in the supply chain that has been amplified in magnitude several times over by poor communications and inadequate data flow. As EMS providers, ODMs and OEMs respond to tightening supply conditions in the coming weeks and months, it will be interesting to watch key industry metrics in an attempt to gage how quickly demand signals find an equilibrium point near the level of real customer demand. Based on the data shown in figure 5 below, it’s just too early to tell. Figure 5: Semiconductor Books vs. US Electronic Equipment By Ron Keith, Chief Executive Officer, Riverwood Solutions