Investors, and others who try to predict the future, are often fond of thinking in terms of megatrends. Changes in demographics, climate, technology and culture are often discussed. One that I believe to be the most significant of all is rarely mentioned, and I have never seen the implications discussed. It is also hard to accept, because it appears to be disproved by experience.
The pace of innovation is slowing (hard though it may be to believe). I have looked at this from a social and political point of view already. I now want to look at the implications a bit more, and also why it seems unbelievable at first. The reality is that we are living on past innovation, and it looks as though we are losing the key driver of economic growth and potential solutions to our most pressing problems.
The idea that the innovation is slowing is hard to accept because it contradicts both the received wisdom of an ever accelerating pace of change, and the experience we have of lives transformed by technology. However if we look at this more carefully, the explanation becomes apparent.
First, to sum up my previous argument: there have been very few big new inventions in the last forty years. The previous forty years saw radar, electronic computers, transistors, lasers, nuclear reactors and many more huge breakthroughs, the peak of a rate of innovation that had been increasing since the start of the industrial revolution. That is probably an understatement as the pace of progress (with some setbacks when civilisations fell) has tended to increase throughout human history, and the the upward trend was continuous in Western Europe (where modern technology as born) since the fall of the Western Roman Empire (and, from what I understand of history, even that setback is now regarded as popularly overstated).
What I did not discuss properly before, was why we perceive the pace of change to be accelerating. If asked for examples of new technologies have had a significant impact in the last forty years, most people will suggest computers, the internet and mobile phones. All these were invented more than forty years ago (even computers with mice and graphical user interfaces). What has happened in the last forty years is that we have made them smaller and cheaper. However this rests on improvements (not breakthroughs) in a very narrow area.
We may now be able too combine a computer and a mobile phone into a pocket sized unit and sell it at a price that makes it affordable to consumers. However this, and most of the other advances of the last forty years are the result not just of progress in a single industry, but are the result of improvements to a single manufacturing process.
The rapid progress has been made in the manufacture of semiconductors, which has continued to follow Moore’s Law. Most of what has been changed is simply the result of being able to fit 30 times as many transistors on a silicon chip at the end of each decade as we could at the beginning. As smaller transistors are faster this meant a thousand fold increase in performance every twenty years, and a million fold increase in the last thirty years.
There are three aspects to the limits of this:
- The physical limits of what a computer can do,
- the limits of the underlying technology, including problems lithography and heat dissipation problems,
- diminishing economic returns from further improvements.
The first is not much of a problem. We are centuries from hitting the ultimate physical limits. To get there we need the invention of whole new technologies — which is exactly what has has not happened.
The limits of further improvements to the processes we have been polishing for the last forty years are likely to be much more immediate. We could see a significant slow down this decade. We are already seeing some changes even in consumer devices that reflect these limits: multi-core processors have become standard (all new computers, and even high end phones) because of the amount of heat produced by ever faster single core processors. We are very close to hitting this.
The diminishing economic returns are also something that is evident, at least in part. The utility of further increases in processing power, and the effect of further increases to increase productivity are becoming limited. We saw huge increases in productivity from early automation: banks once employed vast numbers of clerks just to process cheques. This process was first automated (making it much cheaper) and has now largely been replaced by purely electronic transactions making the margin cost of actually processing a transaction near zero). Those of us who heavily use online information were one of the the last large groups to make large productivity gains, and there are still large cost savings to come in areas such as replacing physical media distribution with electronic. What then?
It is also worth considering what has not happened. We have still not got fusion power (green, and virtually limitless), real breakthroughs in artificial intelligence elude us (much needed given the shortage of the natural kind), and almost every form of transport uses the same underlying technology (jet aircraft, petrol and diesel piston engined cars, electric and diesel powered trains). It would not be hard to extend this list with a many more optimistic expectations of the seventies.
So without innovation, what do we have? Much slower economic growth. If the improvements in semi-conductors, out magic of the last few decades, slows, everything else that has depended on it will stagnate, and that means the whole global economy. It is technological progress that drives productivity, and it is productivity that drives growth. Does anyone question that?
There is room to for global growth by rolling out even more slowly developing technology to developing countries, but this will lead to resource shortages: we need new forms of energy and new materials to make that scale of development possible. I am not even considering the problems of food production and water supply.
The implications just for investors are staggering. It will mean much lower returns on equity investments. This in turn means that the the era of equities hugely out performing bonds may be over. Alternatively, it may lead to lower bond yields as well, as the need for investment capital to roll out new technology diminishes. There may be opportunities in emerging markets, natural resources, and food, but the market growth cannot be expected to continue at past levels for more than another decade or two. This is long term by most standards, but not if you are saving for your retirement.
Opportunities in technology sectors will be very limited. With a slower rate of progress they will become mature industries. There may be some gains for consumers as the focus moves from innovation to quality and reliability or price — as has happened in other mature industries, but this will mean lower margins and limited expansion into new product categories, which means lower returns.
My previous post looked at some of the possible social and economic consequences, such as increasing income inequality. There are also wider consequences. Ultimately, this megatrend affects all the others. Fusion power (or cheap solar power, or any of many other possibilities) would solve out most pressing environmental problems — and the alternative solutions will restrict, or even reverse, growth. We face food shortages that would be relieved by the reduction in the birth rate that accompanies affluence, and affluence spreads to the poor more slowly without technology to drive growth. The developed countries need medical innovation to cope with ageing populations. The unacknowledged megatrend is the biggest of them all. It is perhaps, better called a gigatrend.
A provocative and well-argued thesis, but are you really comparing like with like? If much of the innovation is now happening in computers, perhaps it looks like ‘more computers’ but is really as radical as the personal car or the telegraph that happened to operate in the ‘real world’.
Your comment about declining population is interesting – while I am all for it, I seem to recall reading somewhere that rapidly rising populations in the past 300 or so years is a big reason why we had so much invention.
Anyway, thanks for making me think.
Yes, its not over yet. For example, I am sure that cheaper and better mobile electronics (phones, tablets, ebook readers, hopefully wearable computers etc.) will have a huge impact over the next decade, BUT in the longer term progress is stagnating. If you are young or middle ages and saving for your retirement the impact on your returns of slower growth in ten of fifteen years time will be quite significant.
We are running low on fuel and have no plans to fill the tank. If you prefer an investment analogy, the world is like a pharmaceutical company that has just launched a block buster drug that will drive a few more years of growth, but has nothing more in its drug pipeline.
I just cannot see what will drive growth once Moore’s law runs out of steam.
I am not sure which way causality runs between technology and a rising population: most likely both ways. A rising population provided the workforce for the industrial revolution, and the resources for subsequent development, but sustaining a growing urban population has only been possible because of more efficient agriculture and transport.
Thanks for the comment and he compliment.
Interesting thesis, and I’ve seen it elsewhere connected to Great Stagnation lines, but you need to do a lot more work to establish this.
Prima face, tech patent numbers are increasing all the time. Information density is increasing all the time. You completely ignore biotech and agrotech and materials science, which are making huge (if mostly invisible) strides in human welfare.
Did you notice your car became twice as reliable in the last decade and a half?
A little thought experiment for subjectivity control to consider is how much you would want to be paid to LIVE IN 1990 rather than today. How much to live in 1980? 1970? Compare your answers, what does the trend look like.
I think it is well established that the reason for the increase in numbers of patents is due to a lowering of patent quality. A more litigious environment makes patents more valuable, and lower standards for obviousness make them easier to obtain.
Your counter examples are mostly improvements to existing technology. My point is that without big breakthroughs the benefits of incremental improvements will reach limits as every technology has its limits.
I have not looked explicitly at bio-tech or materials science, but the question is how many big breakthroughs can you find in those fields?
As for your last point, I would say that comparing the change from 1930 to 1970 is far larger than the change from 1970 to now.
I am not claiming that technological advance has stopped, but that without the big breakthroughs it is slowing.