This Is the End


Markets, Risk and Human Interaction

November 22, 2010

Secular Deflation

This reflects my personal opinion, not the views of the SEC or its staff.
My last post posited a future world where technology-driven production and consumption lead to a secular drop in demand, and with it a drop in employment, resulting in what I called a consumption trap:
We are in the year 2025. Because of advances in production technology, much of the path from extracting the required renewable resources through to the production and distribution of most of the items we demand can be accomplished with automated methods overseen by a small cadre of engineers.
The main items we demand, beyond food, clothing and shelter, are the nth generation game systems. Computer games have progressed to the point where they are approaching the level of Nozick’s experience machine. They allow us to be anyone we want in whatever world we want, accompanied by whomever we want, all with full sensory feedback. Given our evolved interests, most of us are spending a fraction of our income on consumption. There just isn’t a lot that we demand.
We are a society that basically eats, sleeps, works and then veges out. Many of us now have our own prefabricated SmallHouse® (McMansions are a thing of the past; no one needs all that space, and, like mink stoles post-Mad Men, social norms regard these as the extravagances of a bygone era). That plus a car, food (the former rarely used, and both produced very inexpensively), our two-hundred dollar experience machine games, and we are happy as a clam.
This world creates consumption trap because we are not consuming enough to generate full employment. And such a world comes with a corollary: secular deflation.
We all know that technology is perpetually in deflation mode. On a per unit of consumption basis – whether measured by the cost of processing one instruction or storing one bit of data – the cost of speed and storage is cut in half every year or so. That makes even roaring inflation look like a crawl.
And, it makes life difficult for the Bureau of Labor Statistics folks who have to figure out CPI. This deflation in technology can overwhelm the inflation in other areas of the economy, where a ten or twenty percent rise is something of note. There are two ways the BLS can keep the relentless price deflation in technology from dominating CPI.
One way is to act as if computers do not represent much of consumption. This route is hard to take, given, first of all, that it is obvious that we all spend a reasonable amount on computers and their peripherals, and second, that with prices dropping by orders of magnitude, consumption would have to be suppressed to a rounding error in order to staunch its effect. It would be one thing to understate our consumption of computers, another to act like computers didn’t even exist.
The second way is not to look at the per unit price of computations and data storage, but look at the price of computers without regard to how much they can do. This is the route that the BLS has taken. The BLS defines high, mainstream and low quality computers, and prices each of those. No matter that the low quality computer now is a hundred times better than the high quality one of ten years ago. And, of course, what constitutes these quality levels is up for grabs. Maybe in the bowels of the calculations, the mainstream computer is defined as the one that costs a bit more than the mainstream one of five years ago. Or, maybe I’m being too cynical.
And to be fair, this approach is not totally foreign in other products. Both houses and cars have improved over the past hundred years, and they also are priced based on the standard item at any given time. But a car continues to provide a way for a small group to travel from place A to B, albeit more comfortably and safely, and a house still keeps a roof over the head of one family, even if it does so with fewer drafts and more efficient floor plans. If a car morphed over time into a vehicle that could transport thousands of people simultaneously to their various destinations in fractions of a second, I doubt that anyone would be able to keep it in the “car” column of the CPI calculation with a straight face.
Whatever the mechanism used to corral the inherent deflation in computers, it starts to break at the seams the more the world becomes integrally related to the cost of units of CPU and memory, where CPU and memory permeate what we consume directly, through our experience machines and the like, and indirectly as a principal factor in producing whatever else we need to keep us alive while we stay plugged in. In that world, the trick of using high, medium and low cost computers to hide the inexorable decline in the prices of computations and memory won’t work anymore. If these prices drop by orders of magnitude, we are going to have a negative number for CPI even if health care, housing and food go up by ten or twenty percent.


  1. The proliferation of broadband, CPUs/GPUs capable of processing HD/HTML5/3D, etc., has created an explosion of the market and profits for companies like Google/Hulu/Oracle. Years ago, it would have been impossible for it to be cost effective to deliver TV over the Internet, and yet, here we are. The efficiency of Mapquest compared to atlases is allowing more transactions, not less.

    It is certainly possible for PCs to sell for $100, just like it's possible to sell a car for $5000, but the $15,000 MSRP has been a pricetag that people have come to accept for a while now.... Netbooks are slipping to less than $250, but they aren't exactly flying off the shelves anymore. As long as technology improves, and demands from new technologies requires more, people will tend to buy more for the same/more price (dumbphones to smartphones, processed to organic), rather than buy the same capability for less price. If you make a new, good market (iPad), they will come. We'll have a long time, if ever, until technology impedes growth. I think the bigger threat is the glut of housing.

    In any case, Moore's Law is hitting a limit. We're already at energy, but there's no promise that we'll have another revolution in energy like we did with computing.

  2. I followed the link across from FT AV.

    I believe that the BLS does take account of quality changes (eg the speed of a computer) in producing the CPI, by hedonic pricing:

    Are you saying that such quality changes are not taken into account?

  3. Hedonic pricing is not used now ( ):

    "From January 1998 to September 2003 the CPI program used hedonic regressions, developed in a cooperative effort with the Producer Price Indexes (PPI) and International Price Program (IPP) programs, as a basis to determine appropriate quality adjustments amounts for personal computers. While this endeavor was viewed as successful and worthwhile, the CPI program decided to adopt a different approach. It should be noted that the hedonic quality adjustments regarding chip speed were deemed unreliable and were never applied to CPI data".

    But I don't know if that would necessarily deal with the problem.

  4. Whats wrong with this picture?December 13, 2010 at 2:03 PM

    Its funny that all this talk of deflation seems to apply to categories that are a smaller and smaller percent of consumption.

    Government waste (they call it spending?) continues to grow much faster than the economy. The economy has off years (aka recessions), while government spending hasn't had an off year in decades. Government waste is now close to 50% of GDP. That means at least half the economy isn't even close to deflation

    Actual healthcare costs might be helped by computer technology -- however lawyers and hospital administration expands like government waste, because in large part that is what it is. Another 15% (and rising) of GDP has no deflation.

    Now add in education (local property taxes or college tuition) and you have another 10% of the economy that continues to grow 2-3x as fast as CPI. All the computers in academia have done nothing to slow the growth of education costs.

    And 40 years after the Department of Energy was created by Jimmy Carter to get the US off foreign oil -- we are MORE dependent, not less. And we still have no energy policy. Energy costs go all over the economy.

    Government (45% of GDP) + health (15%) + education (10%) + energy ... we are talking at least 75% of the economy being uneffected by this alleged technology productivity miracle.

    Sorry Mr Bookstaber... it is great to have an economic theory to explain away the failings of CPI, but like much of Wall Street -- the models don't tie out with reality.

    The cost of living in the real world is going up high single to low double digits, no matter what the government's models say "should" be happening

    The only reason anyone is talking about deflation is because The Bernank happened to write his thesis on the subject. And to a man with a hammer, everything looks like a nail.

    Or do you still think the subprime contagion is well contained?

  5. Here's a new story I found supporting my thesis that people aren't going to continue to pay less and less for computing.

  6. Here's a research piece on the slowing growth of tech advances.

    Microprocessors improved in speed by a factor of 10,000 during the 1980s and 1990s. But two obstacles could mean computing power hitting a wall in the next decade.

    As transistors have become ever smaller and more tightly packed, the speed at which microchips are clocked has levelled off, reaching around 3 gigahertz in 2005. That's because such fast chips generate too much heat to be used in smartphones and personal computers.

    The plateau in clock speed threatens to end the trend we call Moore's law – the doubling of the number of transistors on a chip every couple of years.

    On the bright side, I do forecast significant advances in health (figuring out what the genome means, cheaper drugs, smarter healthcare system), so that old people will spend more money on vacations and exercising than being bound to an IV.

  7. In my experience, the maxim: "The computer you want is always $3,000," generally holds true (for a hobbyist, not necessarily mainstream where entry level prices have come down significantly). People generally buy a computer once every few years and the cost is the cost regardless of it's increased power. That said, we're hitting a point where increased computing power doesn't bring all that much benefit to the average user. There hasn't been a hardware intensive killer app in a while.

  8. This comment has been removed by a blog administrator.


Note: Only a member of this blog may post a comment.