What would an advanced economy look like if technological innovation began to dry up? Economic growth would begin to slow, and we’d expect the shortage of opportunities for new, lucrative investments to lead to a period of persistently lower rates of return on capital. The prices of existing income-yielding assets would rise, and as wealth-holders hunted out increasingly rare higher yielding investment opportunities we’d expect to see a series of asset price bubbles. As truly transformative technologies became rarer, when new technologies did come along we might see them being associated with hype and inflated expectations. Perhaps we’d also begin to see growing inequality, as a less dynamic economy cemented the advantages of the already wealthy and gave fewer opportunities to talented outsiders. It’s a picture, perhaps, that begins to remind us of the characteristics of the developed economies now – difficulties summed up in the phrase “secular stagnation”. Could it be that, despite the widespread belief that technology continues to accelerate, that innovation stagnation, at least in part, underlies some of our current economic difficulties?
Growth in real GDP per person across the G7 nations. GDP data and predictions from the IMF World Economic Outlook 2014 database, population estimates from the UN World Population prospects 2012. The solid line is the best fit to the 1980 – 2008 data of a logistic function of the form A/(1+exp(-(T-T0)/B)); the dotted line represents constant annual growth of 2.6%.
It’s easy to be ambivalent about nuclear power, as my last post illustrated. Nuclear power does provide a low carbon source of energy at scale – if we are serious about decarbonising our energy systems we are going to need a new wave of nuclear power stations, not least to replace an earlier generation of ageing reactors. But nuclear enthusiasts seriously underestimate the scale of the problems that need to be overcome to achieve a large scale expansion of nuclear power. Civil nuclear power has a troubled history everywhere; in Japan consequences of the Fukushima disaster are very much part of current affairs, whose repercussions have spread to countries like Germany. To move beyond this troubled history, to a future in which nuclear power does provide safe and affordable low-carbon energy, we need to understand how this technology got to its current state.
The way in which the technology of civil nuclear power has unfolded was not inevitable; it was the result of the specific circumstances in which it was born and developed. In this sense nuclear power is a great example of the way in which technological trajectories are not pre-ordained; there are many possible paths that nuclear energy could have gone down. What has happened is an example of “technological lock-in” – the particular historical environment in which nuclear power was born put the technology on one particular trajectory, from which it is difficult to make a big jump (as argued in this article by Robin Cowan). This is important because it explains why the current state of the technology is probably not the best place to be given the problems we need to solve.
One slightly dispiriting feature of the current environmental movement is the sniping between “old” environmentalists, opposed to nuclear power, and “new” environmentalists who embrace it, about the relative merits of nuclear and solar as low carbon energy sources. Here’s a commentary on that dispute, in the form of a pair of graphs. In fact, it’s two versions of one graph, showing the world consumption of low carbon energy from solar, nuclear and wind over the last forty years or so, the data taken from the BP Statistical Review of World Energy 2013.
The first graph is the case for nuclear. Only nuclear energy makes any dent at all in the world’s total energy consumption (about 22500 TWh of electricity in total was generated in the world in 2012, with more energy consumed directly as oil and gas). Although nuclear generation has dropped off significantly in the last year or two following the Fukushima accident, the experience of the 1970’s and 80’s shows that it is possible to add significant capacity in a reasonable timescale. Nuclear provides the world with a significant amount of low-carbon energy that it’s foolish to imagine can be quickly replaced by renewables.
The second graph is the case for solar. It is the same graph as the first one, but with a logarithmic axis (on this plot constant fractional growth shows up as an increasing straight-line). This shows that world solar energy consumption is increasing at a greater than exponential rate. For the last five years, solar energy consumption has been growing at a rate of 66% a year compounded. (Wind-power is also growing exponentially, but currently at a slower rate than solar). Although in absolute terms, solar energy is only now at the stage that nuclear was in 1971, its growth rate now is much higher than the maximum growth rate for nuclear in the period of its big build out, which was 30% a year compounded in the five years to 1975. And even before Fukushima, the growth in nuclear energy was stagnating, as new nuclear build only just kept up with the decommissioning of the first generation of nuclear plants. Looking at this graph, solar overtaking nuclear by 2020 doesn’t seem an unreasonable extrapolation.
The case for pessimism is made by Roger Pielke, who points out, from the same data set, that the process of decarbonising the world’s energy supply is essentially stagnating, with the proportion of energy consumption from low carbon sources reaching a high point of 13.3% in 1999, from which it has very gently declined.
Of course, looking backwards at historical energy consumption figures can only take us so far in understanding what’s likely to happen next. For that, we need to look at likely future technical developments and at the economic environment. There is a lot of potential for improvement in both these technologies; not enough research and development has been done on any kind of energy technology in the last few years, as I discussed here before – We sold out our energy future.
On the economics, it has to be stressed that the progress we’ve seen with both nuclear and solar has been the result of large-scale state action. In the case of solar, subsidies in Europe have driven installations, while subsidised capital in China has allowed it rapidly to build up a large solar panel manufacturing industry. The nuclear industry has everywhere been closely tied up with the state, with fairly opaque finances.
But one thing sets apart nuclear and solar. The cost of solar power has been steadily falling, with the prospect of grid parity – the moment when solar generated electricity is cheaper than electricity from the grid – imminent in favoured parts of the world, as discussed in a recent FT Analysis article (£). This provides some justification for the subsidies – usually, with any technology, the more you make of something, the cheaper it becomes; solar shows just such a positive learning curve.
For nuclear, on the other hand, the more we install, the costlier it seems to get. Even in France, widely perceived to have been the most effective nuclear building program, with widespread standardisation and big economies of scale, analysis shows that the learning curve is negative, according to this study by Grubler in Energy Policy (£).
What is urgent now is to get the low-carbon fraction of our energy supply growing again. My own view is that this will require new nuclear build, even if only to replace the obsolete plants now being decommissioned. But for nuclear new build to happen at any scale we need to understand and reverse nuclear’s negative learning curve, and learn how to build nuclear plants cheaply and safely. And while the current growth rate of solar is impressive, we need to remember what a low base it is starting from, and continue to innovate, so that the growth rate can continue to the point at which solar is making a significant contribution.
Medical progress will have come to a complete halt by the year 2329. I reach this anti-Kurzweilian conclusion from a 2012 paper – Diagnosing the decline in pharmaceutical R&D efficiency – which demonstrates that, far from showing an accelerating rate of innovation, the pharmaceutical industry has for the last 60 years been seeing exponentially diminishing returns on its research and development effort. At the date of the anti-singularity, the cost of developing a single new drug will have exceeded the world’s total economic output. The extrapolation is ludicrous, of course, but the problem is not. By 2010 it took an average of $2.17 billion in R&D spending to introduce a single new drug, including the cost of all the failures. This cost per new drug has been following a kind of reverse Moore’s law, increasing exponentially in real terms at a rate of 7.6% a year since 1950, corresponding to a doubling time of a bit more than 9 years (see this plot from the paper cited above). This trend is puzzling – our knowledge of life sciences has been revolutionised during this period, while the opportunities provided by robotics and IT, allowing approaches like rapid throughput screening and large scale chemoinformatics, have been eagerly seized on by the industry. Despite all this new science and enabling technology, the anti-Moore’s law trend of diminishing R&D returns continues inexorably.
This should worry us. The failure to find effective therapies for widespread and devastating conditions – Alzheimer’s, to take just one example – leads to enormous human suffering. The escalating cost of developing new drugs is ultimately passed on to society through their pricing, leading to strains on national healthcare systems that will become more acute as populations age. As a second-order effect, scientists should be concerned in case the drying up of medical innovation casts doubt on some of the justifications for government spending on fundamental life sciences research. And, of course, a healthy and innovative pharmaceutical industry is itself important for economic growth, particularly here in the UK, where it remains the one truly internationally competitive high technology sector of the economy. So what can be done to speed up innovation in this vital sector? Continue reading “Decelerating change in the pharmaceutical industry”
I have a post up on the blog of the Sheffield Political Economy Research Institute – The failures of supply side innovation policy – discussing the connection between recent innovation policy in the UK and our current crisis of economic growth. Rather than cross-posting it here, I tell the same story in four graphs.
1. The UK’s current growth crisis follows a sustained period of national disinvestment in R&D
Red, left axis. The percentage deviation of real GDP per person from the 1948-1979 trend line, corresponding to 2.57% annual growth. Sources: solid line, 2012 National Accounts. Dotted line, March 2013 estimates from the Office for Budgetary Responsibility.
Blue, right axis. Total R&D intensity, all sectors, as percentage of GDP. Data: Eurostat.
In a speech at the Royal Society last November George Osborne said that, as Chancellor of the Exchequer, it is his job “to focus on the economic benefits of scientific excellence”. He then listed eight key technologies that he challenged the scientific community in Britain to lead the world in, and for which he promised continuing financial support. Among these technologies were synthetic biology, regenerative medicine and agri-science, key examples of what a recent report from the Nuffield Council for Bioethics calls emerging biotechnologies. Picking technology winners is clearly high on the UK science policy agenda, and this kind of list will increasingly inform the science funding choices the government and its agencies, like the research councils, make. So the focus of the Nuffield’s report, on how those choices are made and what kind of ethics should guide them, couldn’t be more timely.
Everyone should know that the industrial society we live in depends on access to plentiful, convenient, cheap energy – the last two hundred years of rapid economic growth has been underpinned by the large scale use of fossil fuels. And everyone should know that the effect of burning those fossil fuels has been to markedly increase the carbon dioxide content of the atmosphere, resulting in a changing climate, with potentially dangerous but still uncertain consequences. But a transition from fossil fuels to low carbon sources of energy isn’t going to take place quickly; existing low carbon energy sources are expensive and difficult to scale up. So rather than pushing on with the politically difficult, slow and expensive business of deploying current low carbon energy sources, why don’t we wait until technology brings us a new generation of cheaper and more scalable low carbon energy? Presumably, one might think, since we’ve known about these issues for some time, we’ve been spending the last twenty years energetically doing research into new energy technologies?
Alas, no. As my graph shows, the decade from 1980 saw a worldwide decline in the fraction of GDP major industrial countries devoted to government funded energy research, development, and demonstration, with only Japan sustaining anything like its earlier intensity of energy research into the 1990s. It was only in the second half of the decade after 2000 that we began to see a recovery, though in the UK and the USA a rapid upturn following the 2007 financial crisis has fallen away again. A rapid post-2000 growth of energy RD&D in Korea is an exception to the general picture. There’s a good discussion of the situation in the USA in a paper by Kamman and Nemet – Reversing the incredible shrinking energy R&D budget. But the largest fall by far was in the UK, where at its low point, the fraction of national resource devoted to energy RD&D fell, in 2003, to an astonishing 0.2% of its value at the 1981 high point.
I’ve long suspected that physical scientists have occasional attacks of biology envy, so I suppose I shouldn’t be surprised that the US government announced last year the “Materials Genome Initiative for Global Competiveness”. Its aim is to “discover, develop, manufacture, and deploy advanced materials at least twice as fast as possible today, at a fraction of the cost.” There’s a genuine problem here – for people used to the rapid pace of innovation in information technology, the very slow rate at which new materials are taken up in new manufactured products is an affront. The solution proposed here is to use those very advances in information technology to boost the rate of materials innovation, just as (the rhetoric invites us to infer) the rate of progress in biology has been boosted by big data driven projects like the human genome project.
Mark Henderson’s book “The Geek Manifesto” was part of my holiday reading, and there’s a lot to like in it – there’s all too much stupidity in public life, and anything that skewers a few of the more egregious recent examples of this in such a well-written and well-informed way must be welcomed. There is a fundamental lack of seriousness in our public discourse, a lack of respect for evidence, a lack of critical thinking. But to set against many excellent points of detail, the book is built around one big idea, and it’s that idea that I’m less keen on. This is the argument – implicit in the title – that we should try to construct some kind of identity politics based around those of us who self-identify as being interested in and informed about science – the “geeks”. I’m not sure that this is possible, but even if it was, I think it would be bad for science and bad for politics. This isn’t to say that public life wouldn’t be better if more people with a scientific outlook had a higher profile. One very unwelcome feature of public debate is the prevalence of wishful thinking. Comfortable beliefs that fit into people’s broader world-views do need critical examination, and this often needs the insights of science, particularly the discipline that comes from seeing whether the numbers add up. But science isn’t the only source of the insights needed for critical thinking, and scientists can have some surprising blind-spots, not just about the political, social and economic realities of life, but also about technical issues outside their own fields of interest.
In 1981 the UK was one of the world’s most research and development intensive economies, with large scale R&D efforts being carried out in government and corporate laboratories in many sectors. Over the thirty years between then and now, this situation has dramatically changed. A graph of the R&D intensity of the national economy, measured as the fraction of GDP spent on research and development, shows a long decline through the 1980’s and 1990’s, with some levelling off from 2000 or so. During this period the R&D intensity of other advanced economies, like Japan, Germany, the USA and France, has increased, while in fast developing countries like South Korea and China the growth in R&D intensity has been dramatic. The changes in the UK were in part driven by deliberate government policy, and in part have been the side-effects of the particular model of capitalism that the UK has adopted. Thirty years on, we should be asking what the effects of this have been on our wider economy, and what we should do about it.
The second graph breaks down where R&D takes place. The largest fractional fall has been in research in government establishments, which has dropped by more than 60%. The largest part of this fall took place in the early part of the period, under a series of Conservative governments. This reflects a general drive towards a smaller state, a run-down of defence research, and the privatisation of major, previously research intensive sectors such as energy. However, it is clear that privatisation didn’t lead to a transfer of the associated R&D to the business sector. It is in the business sector that the largest absolute drop in R&D intensity has taken place – from 1.48% of GDP to 1.08%. Cutting government R&D didn’t lead to increases in private sector R&D, contrary to the expectations of free marketeers who think the state “crowds out” private spending. Instead the business climate of the time, with a drive to unlock “shareholder value” in the short-term, squeezed out longer term investments in R&D. Some seek to explain this drop in R&D intensity in terms of a change in the sectoral balance of the UK economy, away from manufacturing and towards financial services, and this is clearly part of the picture. However, I wonder whether this should be thought of not so much as an explanation, but more as a symptom. I’ve discussed in an earlier post the suggestion that “bad capitalism” – for example, speculations in financial and property markets ,with the downside risk being shouldered by the tax-payer – squeezes out genuine innovation.
The Labour government that came to power in 1997 did worry about the declining R&D intensity of the UK economy, and, in its Science Investment Framework 2004-2014 (PDF), set about trying to reverse the trend. This long-term policy set a target of reaching an overall R&D intensity of 2.5% by 2014, and an increase in R&D intensity in the business sector from to 1.7%. The mechanisms put in place to achieve this included a period of real-terms increase in R&D spending by government, some tax incentives for business R&D, and a new agency for nearer term research in collaboration with business, the Technology Strategy Board. In the event, the increases in government spending on R&D did lead to some increase in the UK’s overall research intensity, but the hoped-for increase in business R&D simply did not happen.
This isn’t predominantly a story about academic science, but it provides a context that’s important to appreciate for some current issues in science policy. Over the last thirty years, the research intensity of the UK’s university sector has increased, from 0.32% of GDP to 0.48% of GDP. This reflects, to some extent, real-term increases in government science budgets, together with the growing success of universities in raising research funds from non UK-government sources. The resulting R&D intensity of the UK HE sector is at the high end of international comparisons (the corresponding figures for Germany, Japan, Korea and the USA are 0.45%, 0.4%, 0.37% and 0.36%). But where the UK is very much an outlier is in the proportion of the country’s research that takes place in universities. This proportion now stands at 26%, which is much higher than international competitors (again, we can compare with Germany, Japan, Korea and the USA, where the proportions are 17%, 12%, 11% and 13%), and much higher now than it has been historically (in 1981 it was 14%). So one way of interpreting the pressure on universities to demonstrate the “impact” of their research, which is such a prominent part of the discourse in UK science policy at the moment, is as a symptom of the disproportionate importance of university research in the overall national R&D picture. But the high proportion of UK R&D carried out in universities is as much a measure of the weakness of the government and corporate applied and strategic research sectors as the strength of its HE research enterprise. The worry, of course, has to be that, given the hollowed-out state of the business and government R&D sectors, where in the past the more applied research needed to convert ideas into new products and services was done, universities won’t be able to meet the expectations being placed on them.
To return to the big picture, I’ve seen surprisingly little discussion of the effects on the UK economy of this dramatic and sustained decrease in research intensity. Aside from the obvious fact that we’re four years into an economic slump with no apparent prospect of rapid recovery, we know that the UK’s productivity growth has been unimpressive, and the lack of new, high tech companies that grow fast to a large scale is frequently commented on – where, people ask, is the UK’s Google? We also know that there are urgent unmet needs that only new innovation can fulfil – in healthcare, in clean energy, for example. Surely now is the time to examine the outcomes of the UK’s thirty year experiment in innovation theory.
Finally, I think it’s worth looking at these statistics again, because they contradict the stories we tell about ourselves as a country. We think of our postwar history as characterised by brilliant invention let down by poor exploitation, whereas the truth is that the UK, in the thirty post-war years, had a substantial and successful applied research and development enterprise. We imagine now that we can make our way in the world as a “knowledge economy”, based on innovation and brain-power. I know that innovation isn’t always the same as research and development, but it seems odd that we should think that innovation can be the speciality of a nation which is substantially less intensive in research and development than its competitors. We should worry instead that we’re in danger of condemning ourselves to being a low innovation, low productivity, low growth economy.