Supply/Demand Ratio: At What Price?

Electronic News, June 7, 1999 by Frank Dickson

Cahners In-Stat Group

Scottsdale, Ariz.-Second only to the book-to-bill ratio, the ratio of supply-to-demand is the most common barometer for gauging the future health of our beloved semiconductor industry.

In concept, the ratio is straightforward. First we add up all the capacity (or supply) for a particular device or an aggregate of devices. DRAM seems to be the focus of most of the attention. Then we add up all the demand for the device or aggregate of devices. Finally we divide the supply by the demand to arrive at a ratio. If the ratio equals one, the industry is in balance. Ratios more than one or less than one indicate over- or under-supply respectively.

The simplicity of the supply/demand ratio is what causes us to gravitate toward using it. However, that simplicity disguises some fundamental problems with the ratio, problems that cause the uninformed user to jump to some dangerous conclusions.

The first problem comes in the enormity of the task of defining supply. First, this task involves counting every fab globally. Then, one needs to figure the full capacity of each fab, and then calculate the true capacity actually achievable in the real world. One then must estimate the product mix of each fab, the die size of each product, and corresponding yields of each. If that was not enough, rate of die shrinks and corresponding, yield impacts have to be calculated on almost a daily basis.

If navigating the incredible complexities of supply were not enough, we also need to actually measure supply. The day and age of dedicated fabs committed to manufacture one product line are over. The proliferation of the fabless/foundry model speaks well to this point. As a result, supply in today's world is freer flowing than in the past, moving to areas that yield the highest margins and, of course, profits.

The second problem is in the measurement of demand. It was hard enough to measure demand when almost the entire semiconductor world revolved around the electronic data processing sector. All of the experts, including Cahners In-Stat Group, forecast and monitor PC shipments. However, no can agree as to exactly how many PCs are actually shipping, far less agree on a forecast. Additionally, the measurement of semiconductors depends on accurately measuring and forecasting the exact configuration of each on almost a daily basis. If the enormity of measuring EDP semiconductor consumption were not enough, the use of semiconductors is proliferating in applications such as automobiles, consumer appliances, etc. When the next generation of home game machines has 32Mbytes of DRAM, the task of measuring demand becomes further daunting.

The final problem is that the fundamental assumptions of the ratio violate the fundamental economics. The laws are fairly straightforward. As the price of an item rises, there are more suppliers willing to produce a product. As the price of an item falls, there are more buyers willing to consume a product. In the end, the forces of supply and demand will balance out and equilibrium will result. This concept is illustrated in Chart 1.

The concept of a demand/supply ratio is predicated upon the assumption that there is an imperfection in the market dynamics that prevents prices from adjusting and ensures that demand and supply are relatively fixed and unwavering. As you can see in Chart 2, when a price resides above the equilibrium point, the quantity demanded exceeds the quantity supplied, resulting in a surplus. When a price resides below an equilibrium point, a shortage results.

The problem is that prices do not remain fixed; they are constantly in flux, along with the amount supplied and demanded. Don't believe me? Ask someone in the DRAM market. Chart 3 is a graph of Worldwide DRAM Monthly Mbit Shipment Change (percentage) and Worldwide DRAM Monthly ASP per Mbit Change (percentage) from February 1997 through December 1998. The first point that jumps-out is that the Mbits shipped and the ASP per Mbit are constantly in flux. Additionally, the fact that when one goes up the other goes down not only illustrates that the market dynamics are constantly forcing prices to an equilibrium price, but that memory is elastic and people add memory because it is cheap.

In the end, a supply/demand ratio with a corresponding price qualifier is useless!

A side note needs to be added here. The economic theory of equilibrium is a word that solicits pictures of calm. However, equilibrium in a market may only exist for a brief second. Markets are dynamic; thus, they are constantly moving. The drive for equilibrium is a constant, on-going and often very painful process.

Does this mean the exercise of supply and demand analysis is wrong? Of course not! The value of the supply/demand exercise comes in the evaluation of the trends. The evaluation of the demand and supply trend and the interactive nature between them supplies the insights into the future that we desire. Just as one datum point provides little information, one static ratio does the same.