Matt Ridley points me to a paper published a few days ago, which finds that, despite the impact of climate, variability in crop yields has not changed. As the authors Tom Osborne and Tim Wheeler conclude
This study has taken a broad-scale view of crop yield variability and its relationship with climate. Whilst general conclusions are difficult to draw when considering diverse crop production systems across the globe it appears that, for the majority of crop–country combinations examined, crop yield variability has not increased. Indeed, in half of the cases yield variability has decreased. Direct attribution of the drivers of change in variability, as in previous studies, was not attempted. However, for the first time, the potential role of climate variability was elucidated. The variation of several growing season climates was shown to have changed significantly during the past 50 years. The use of empirical relationships between crop yield and climate identified several countries, in particular maize in Indonesia and rice in India, where significant changes in climate variability have led to the observed reductions in yield variability. In the remainder of cases it was found that climate has not contributed.
They also say:
At the global scale the variability of wheat and rice yields have declined significantly between 1960 and 2009, whilst the variability of maize peaked during the middle of the time period (figure 5). Consideration of the observed relationship between yield and climate suggests that a significant reduction in the variability of rainfall may have contributed to the reduction in rice yield variability, but that wheat yield variability declined despite an increase in the variability of associated climate.
The authors seem to be somewhat bemused by these findings, suggesting that 'it is unclear how rising CO2 concentrations might have significantly impacted the variability of yield'. This is an odd thing to say; I thought it was fairly well established that high carbon dioxide levels tended to reduce water loss in plants, thus protecting them from drought.
Matt also points out this 1991 paper by Jesse Ausubel, which notes that even 20 years ago it was becoming clear that technology was profoundly reducing mankind's vulnerability to climate:
In many respects we seem to be climate proofing society, making ourselves less subject to natural phenomena. For centuries and millennia we relied mainly on behavioural and social adaptation. We took siestas when the sun was high and sought refuge in hill stations in the monsoon season. Large pastoral and nomadic populations followed the seasonal availability of resources and avoided climatic stresses. Much of the planet remained seasonally or entirely uninhabitable for climatic reasons. With current technology many people can live in virtually any climate that now exists. Modern water supply and heating and ventilation systems, along with medicines (for example, quinine and vaccines) and public health measures, have enabled large populations to inhabit formerly uninhabitable regions
It is worthy of note that Ausubel's comments came even before the GM and telecommunications revolutions, which have had such profound effects on agriculture. So much for the dumb farmer assumption. One wonders just how many of these innovations - particularly the system innovations such as telecoms - are factored into the impacts assessments and the economic models.
Perhaps one of the Richards can help here.