Australian Beer Prices set to Double Due to Global Warming?

Earlier this week Nature Plants published a new paper Decreases in global beer supply due to extreme drought and heat

The Scientific American has an article “Trouble Brewing? Climate Change Closes In on Beer Drinkers” with the sub-title “Increasing droughts and heat waves could have a devastating effect on barley stocks—and beer prices”. The Daily Mail headlines with “Worst news ever! Australian beer prices are set to DOUBLE because of global warming“. All those climate deniers in Australia have denied future generations the ability to down a few cold beers with their barbecued steaks tofu salads.

This research should be taken seriously, as it is by a crack team of experts across a number of disciplines and Universities. Said, Steven J Davis of University of California at Irvine,

The world is facing many life-threatening impacts of climate change, so people having to spend a bit more to drink beer may seem trivial by comparison. But … not having a cool pint at the end of an increasingly common hot day just adds insult to injury.

Liking the odd beer or three I am really concerned about this prospect, so I rented the paper for 48 hours to check it out. What a sensation it is. Here a few impressions.

Layers of Models

From the Introduction, there were a series of models used.

  1. Created an extreme events severity index for barley based on extremes in historical data for 1981-2010.
  2. Plugged this into five different Earth Systems models for the period 2010-2099. Use this against different RCP scenarios, the most extreme of which shows over 5 times the warming of the 1981-2010 period. What is more severe climate events are a non-linear function of temperature rise.
  3. Then model the impact of these severe weather events on crop yields in 34 World Regions using a “process-based crop model”.
  4. (W)e examine the effects of the resulting barley supply shocks on the supply and price of beer in each region using a global general equilibrium model (Global Trade Analysis Project model, GTAP).
  5. Finally, we compare the impacts of extreme events with the impact of changes in mean climate and test the sensitivity of our results to key sources of uncertainty, including extreme events of different severities, technology and parameter settings in the economic model.

What I found odd was they made no allowance for increasing demand for beer over a 90 year period, despite mentioning in the second sentence that

(G)lobal demand for resource-intensive animal products (meat and dairy) processed foods and alcoholic beverages will continue to grow with rising incomes.

Extreme events – severity and frequency

As stated in point 2, the paper uses different RCP scenarios. These featured prominently in the IPCC AR5 of 2013 and 2014. They go from RCP2.6, which is the most aggressive mitigation scenario, through to RCP 8.5 the non-policy scenario which projected around 4.5C of warming from 1850-1870 through to 2100, or about 3.8C of warming from 2010 to 2090.

Figure 1 has two charts. On the left it shows that extreme events will increase intensity with temperature. RCP2.6 will do very little, but RCP8.5 would result by the end of the century with events 6 times as intense today. Problem is that for up to 1.5C there appears to be no noticeable change what so ever.  That is about the same amount of warming the world has experienced from 1850-2010 per HADCRUT4 there will be no change. Beyond that things take off. How the models empirically project well beyond known experience for a completely different scenario defeats me. It could be largely based on their modelling assumptions, which is in turn strongly tainted by their beliefs in CAGW. There is no reality check that it is the models that their models are not falling apart, or reliant on arbitrary non-linear parameters.

The right hand chart shows that extreme events are porjected to increase in frequency as well. Under RCP 2.6 ~ 4% chance of an extreme event, rising to ~ 31% under RCP 8.5. Again, there is an issue of projecting well beyond any known range.

Fig 2 average barley yield shocks during extreme events

The paper assumes that the current geographical distribution and area of barley cultivation is maintained. They have modelled in 2099, from the 1981-2010 a gridded average yield change with 0.5O x 0.5O resolution to create four colorful world maps representing each of the four RCP emissions scenarios. At the equator, each grid is about 56 x 56 km for an area of 3100 km2, or 1200 square miles. Of course, nearer the poles the area diminishes significantly. This is quite a fine level of detail for projections based on 30 years of data to radically different circumstances 90 years in the future. The results show. Map a) is for RCP 8.5. On average yields are projected to be 17% down. As Paul Homewood showed in a post on the 17th, this projected yield fall should be put in the context of a doubling of yields per hectare since the 1960s.

This increase in productivity has often solely ascribed to the improvements in seed varieties (see Norman Borlaug), mechanization and use of fertilizers. These have undoubtably have had a large parts to play in this productivity improvement. But also important is that agriculture has become more intensive. Forty years ago it was clear that there was a distinction between the intensive farming of Western Europe and the extensive farming of the North American prairies and the Russian steppes. It was not due to better soils or climate in Western Europe. This difference can be staggering. In the Soviet Union about 30% of agricultural output came from around 1% of the available land. These were the plots that workers on the state and collective farms could produce their own food and sell surplus in the local markets.

Looking at chart a in Figure 2, there are wide variations about this average global decrease of 17%.

In North America Montana and North Dakota have areas where barley shocks during extreme years will lead to mean yield changes over 90% higher normal, and the areas around have >50% higher than normal. But go less than 1000 km North into Canada to the Calgary/Saskatoon area and there are small decreases in yields.

In Eastern Bolivia – the part due North of Paraguay – there is the biggest patch of > 50% reductions in the world. Yet 500-1000 km away there is a North-South strip (probably just 56km wide) with less than a 5% change.

There is a similar picture in Russia. On the Kazakhstani border, there are areas of > 50% increases, but in a thinly populated band further North and West, going from around Kirov to Southern Finland is where there are massive decreases in yields.

Why, over the course of decades, would those with increasing yields not increase output, and those with decreasing yields not switch to something else defeats me. After all, if overall yields are decreasing due to frequent extreme weather events, the farmers would be losing money, and those farmers do well when overall yields are down will be making extraordinary profits.

A Weird Economic Assumption

Building up to looking at costs, their is a strange assumption.

(A)nalysing the relative changes in shares of barley use, we find that in most case barley-to-beer shares shrink more than barley-to-livestock shares, showing that food commodities (in this case, animals fed on barley) will be prioritized over luxuries such as beer during extreme events years.

My knowledge of farming and beer is limited, but I believe that cattle can be fed on other things than barley. For instance grass, silage, and sugar beet. Yet, beers require precise quantities of barley and hops of certain grades.

Further, cattle feed is a large part of the cost of a kilo of beef or a litre of milk. But it takes around 250-400g of malted barley to produce a litre of beer. The current wholesale price of malted barley is about £215 a tonne or 5.4 to 8.6p a litre. About cheapest 4% alcohol lager I can find in a local supermarket is £3.29 for 10 x 250ml bottles, or £1.32 a litre. Take off 20% VAT and excise duty leaves 30p a litre for raw materials, manufacturing costs, packaging, manufacturer’s margin, transportation, supermarket’s overhead and supermarket’s margin. For comparison four pints (2.276 litres) of fresh milk costs £1.09 in the same supermarket, working out at 48p a litre. This carries no excise duty or VAT. It might have greater costs due to refrigeration, but I would suggest it costs more to produce, and that feed is far more than 5p a litre.

I know that for a reasonable 0.5 litre bottle of ale it is £1.29 to £1.80 a bottle in the supermarkets I shop in, but it is the cheapest that will likely suffer the biggest percentage rise from increase in raw material prices. Due to taxation and other costs, large changes in raw material prices will have very little impact on final retail costs. Even less so in pubs where a British pint (568ml) varies from the £4 to £7 a litre equivalent.

That is, the assumption is the opposite of what would happen in a free market. In the face of a shortage, farmers will substitute barley for other forms of cattle feed, whilst beer manufacturers will absorb the extra cost.

Disparity in Costs between Countries

The most bizarre claim in the article in contained in the central column of Figure 4, which looks at the projected increases in the cost of a 500 ml bottle of beer in US dollars. Chart h shows this for the most extreme RCP 8.5 model.

I was very surprised that a global general equilibrium model would come up with such huge disparities in costs after 90 years. After all, my understanding of these models used utility-maximizing consumers, profit-maximizing producers, perfect information and instantaneous adjustment. Clearly there is something very wrong with this model. So I decided to compare where I live in the UK with neighbouring Ireland.

In the UK and Ireland there are similar high taxes on beer, with Ireland being slightly more. Both countries have lots of branches of the massive discount chain. They also have some products on their website aldi.co.uk and aldi.ie.  In Ireland a 500 ml can of Sainte Etienne Lager is €1.09 or €2.18 a litre or £1.92 a litre. In the UK it is £2.59 for 4 x 440ml cans or £1.59 a litre. The lager is about 21% more in Ireland. But the tax difference should only be about 15% on a 5% beer (Saint Etienne is 4.8%). Aldi are not making bigger profits in Ireland, they just may have higher costs in Ireland, or lesser margins on other items. It is also comparing a single can against a multipack. So pro-rata the £1.80 ($2.35) bottle of beer in the UK would be about $2.70 in Ireland. Under the RCP 8.5 scenario, the models predict the bottle of beer to rise by $1.90 in the UK and $4.84 in Ireland. Strip out the excise duty and VAT and the price differential goes from zero to $2.20.

Now suppose you were a small beer manufacturer in England, Wales or Scotland. If beer was selling for $2.20 more in Ireland than in the UK, would you not want to stick 20,000 bottles in a container and ship it to Dublin?

If the researchers really understood the global brewing industry, they would realize that there are major brands sold across the world. Many are brewed across in a number of countries to the same recipe. It is the barley that is shipped to the brewery, where equipment and techniques are identical with those in other parts of the world. This researchers seem to have failed to get away from their computer models to conduct field work in a few local bars.

What can be learnt from this?

When making projections well outside of any known range, the results must be sense-checked. Clearly, although the researchers have used an economic model they have not understood the basics of economics. People are not dumb  automatons waiting for some official to tell them to change their patterns of behavior in response to changing circumstances. They notice changes in the world around them and respond to it. A few seize the opportunities presented and can become quite wealthy as a result. Farmers have been astute enough to note mounting losses and change how and what they produce. There is also competition from regions. For example, in the 1960s Brazil produced over half the world’s coffee. The major region for production in Brazil was centered around Londrina in the North-East of Parana state. Despite straddling the Tropic of Capricorn, every few years their would be a spring-time frost which would destroy most of the crop. By the 1990s most of the production had moved north to Minas Gerais, well out of the frost belt. The rich fertile soils around Londrina are now used for other crops, such as soya, cassava and mangoes. It was not out of human design that the movement occurred, but simply that the farmers in Minas Gerais could make bumper profits in the frost years.

The publication of this article shows a problem of peer review. Nature Plants is basically a biology journal. Reviewers are not likely to have specialist skills in climate models or economic theory, though those selected should have experience in agricultural models. If peer review is literally that, it will fail anyway in an inter-disciplinary subject, where the participants do not have a general grounding in all the disciplines. In this paper it is not just economics, but knowledge of product costing as well. It is academic superiors from the specialisms that are required for review, not inter-disciplinary peers.

Kevin Marshall

 

Climate Interactive’s Bogus INDC Forecast

Summary

Joe Romm wrote a post in early November claiming UNFCCC Executive Secretary Christiana Figueres had misled the public in claiming that the “INDCs have the capability of limiting the forecast temperature rise to around 2.7 degrees Celsius by 2100”. Using Climate Interactive’s figures Romm claims the correct figure is 3.5°C. That Romm had one of two sources of the 2.7°C staring at him is a side issue. The major question is how Climate Interactive can achieve a full 1.0°C reduction in expected temperature rise in 2100 and a reduction of 40% in 2100 GHG emissions from pledges covering the period 2015, when the UNFCCC estimates will have a much smaller impact in 2030? Looking at the CO2 emissions, which account for 75-80% of GHG emissions, I have found the majority answer. For OECD countries where emissions per capita have been stable or falling for decades, the “No Action” scenario forecasts that they will rise for decades. For Russia and China, where per capita emissions are likely to peak before 2030 without any policy action, the “No Action” scenario forecasts that they will rise for decades. This is largely offset by Climate Interactive assuming that both emissions and economic growth in India and Africa (where there are no real attempts to control emissions) will stagnate in the coming decades. Just by making more reasonable CO2 emissions forecasts for the OECD, Russia and China can account for half of the claimed 2100 reduction in GHG emissions from the INDC. Climate Interactive’s “No Action” scenario is bogus.

 

Joe Romm’s use of the Climate Interactive projection

A couple of months ago, prior to the COP21 Paris climate talks, Joe Romm at Climate Progress criticized the claim made in a press release by UNFCCC Executive Secretary Christiana Figueres:

The INDCs have the capability of limiting the forecast temperature rise to around 2.7 degrees Celsius by 2100, by no means enough but a lot lower than the estimated four, five, or more degrees of warming projected by many prior to the INDCs

Romm’s note to the media is

If countries go no further than their current global climate pledges, the earth will warm a total of 3.5°C by 2100.

At a basic level Romm should have done some proper research. As I found out, there are two sources of the claim that are tucked away at the end of a technical appendix to the UNFCCC Synthesis report on the aggregate effect of INDCs. One of these is Climate Action Tracker. On their home page they have a little thermometer which shows the 2.7°C figure. Romm would have seen this, as he refers in the text to CAT’s page on China. The significance may not have registered.

However, the purpose of this post is not to criticize Romm, but the Climate Interactive analysis that Romm uses as the basis of his analysis. Is the Climate Interactive Graph (reproduced in Figure 1) a reasonable estimate of the impact of the INDC submissions (policy pledges) on global emissions?1

Figure 1. Climate Interactive’s graph of impact of the INDC submissions to 2100

What struck me as odd when I first saw this graph was how the INDCs could make such a large impact beyond the 2015-2030 timeframe that they covered when the overall impact was fairly marginal within that timeframe. This initial impression is confirmed by the UNFCCC’s estimate of the INDCs

Figure 2. UNFCCC’s estimate of emissions impact of the INDCs, with the impact shown by the yellow bars. Original here.

There are two things that do not stack up. First is that the “No Action” scenario appears to be a fairly reasonable extrapolation of future emissions without policy. Second, and contrary to that is the first, is that the “Current INDCs” scenario does not make sense in terms of what I have read in the INDCs and is confirmed by the INDCs. To resolve this requires looking at the makeup of the “No Action” scenario. Climate Interactive usefully provide the model for others to do their own estimates,2 With the “User Reference Scenario” giving the “no action” data3, split by type of greenhouse gas and into twenty regions or countries. As about 75-80% of emissions with the model are CO2 Fossil Fuel emissions, I will just look at this area. For simplicity I have also reduced the twenty regions or countries into just seven. That is USA, Other OECD, Russia, China, India, Africa and Rest of the World. There are also lots of ways to look at the data, but some give better understanding of the data than others. Climate Interactive also have population estimates. Population changes over a long period can themselves result in changed emissions, so looking at emissions per capita gives a better sense of the reasonableness of the forecast. I have graphed the areas in figure 3 for the historical period 1970-2010 and the forecast period 2020-2100.

Figure 3 : Fossil Fuel Emissions per capita for six regions from the Climate Interactive “No Action” Scenario.

Understanding the CO2 emissions forecasts

In the USA, emissions per capita peaked at the time of 1973 oil embargo. Since then they have declined slightly. There are a number of reasons for this.

First, higher oil prices gave the economic incentives to be more efficient in usage of oil. In cars there have been massive improvements in fuel efficiency since that time. Industry has also used energy more efficiently. Second, there has been a growth in the use of nuclear power for strategic reasons more than economic. Third is that some of the most energy intensive industries have shifted to other countries, particularly steel and chemicals. Fourth, is that growth in developed countries is mostly in the service sector, whereas growth in developing countries is mostly in manufacturing. Manufacturing tends to have much higher energy usage per unit of output than services. Fifth, is that domestic energy usage is from cars and power for the home. In an emerging economy energy usage will rise rapidly as a larger proportion of the population acquire cars and full heating and lighting systems in the home. Growth is much slower once most households have these luxuries. Sixth is that in the near future emissions might continue to fall with the development of shale gas, with its lower emissions per unit of power than from coal.

I therefore cannot understand why anyone would forecast increasing emissions per capita in the near future, when they have been stable or falling in for decades. Will everyone start to switch to less efficient cars? When these forecasts were made oil was at $100 a barrel levels, and many thought peak oil was upon us. Would private sector companies abandon more efficient energy usage for less efficient and higher cost usage? The USA may abandon nuclear power and shift back to coal for political reasons. But in all forms of energy, production and distribution is likely to continue to become more efficient in all forms.

In the rest of the OECD, there are similar patterns. In Europe energy usage was never as high. In some countries without policy CO2 emissions may rise slightly. In Germany they are replacing nuclear power stations with coal for instance. But market incentives will increase energy efficiency and manufacturing will continue to shift to emerging nations. Again, there appears no reason for a steady increase in emissions per capita to increase in the future.

Russia has a slightly different recent past. Communist central planning was highly inefficient and lead to hugely inefficient energy usage. With the collapse of communism, energy usage fell dramatically. Since then emissions have been increasing, but more slowly than the economy as a whole. Emissions will peak again in a couple of decades. This will likely be at a lower level than in the USA in 1970, despite the harsher climate, as Russia will benefit from technological advances in the intervening period. There is no reason for emissions to go on increasing at such a rapid rate.4

China has recently had phenomenal growth rates. According to UN data, from 1990 to 2012, economic growth averaged 10.3% per annum and CO2 emissions 6.1%. In the not too distant future economic growth will slow as per capita income approaches rich country levels, and emissions growth will slow or peak. But the Climate Interactive forecast has total emissions only peaking in 2090. The reason for China’s and Russia’s forecast per capita emissions exceeding those of the USA is likely due to a failure to allow for population changes. In USA population is forecast to grow, whilst in China and Russia population is forecast to fall.

India has the opposite picture. In recently years economic and CO2 emissions growth has taken off. Current policies of Prime Minister Narendra Modi are to accelerate these growth rates. But in the Climate Interactive forecast growth, economic growth and CO2 emissions growth plummet in the near future. Economic growth is already wrong. I am writing on 30/12/15. To meet the growth forecast for 2010-2015, India’s GDP will need to drop by 20% in the next 24 hours.5

For the continent of Africa, there have been encouraging growth signs in the last few years, after decades when many countries saw stagnation or even shrinking economies. Climate Interative forecasts similar growth to India, but with a forecasts of rapid population growth, the emissions per capita will hardly move.

Revised CO2 emissions forecasts

It is extremely difficult and time consuming to make more accurate CO2 emissions forecasts. As a shortcut, I will look at the impact of revisions on 2100, then at the impact on the effect of the INDCs. This is laid out in Figure 4

Figure 4 : Revised Forecast CO2 Emissions from Fossil Fuels

The first three columns in pale lilac are for CO2 emissions per capita calculated, from the Climate Interactive data. In the 2100 Revised column are more realistic estimates for reasons discussed in the text. In the orange part of the table are the total forecast 2100 Climate Interactive figures for population and CO2 emissions from fossil fuels. In darker orange is the revised emissions forecast (emissions per capita multiplied by forecast population) and the impact of the revision. Overall the forecast is 10.2GtCO2e lower, as no calculation has been made for the rest of the world. To balance back requires emissions of 11.89 tonnes per capita for 2.9 billion people. As ROW includes such countries as Indonesia, Bangladesh, Iran, Vietnam, Brazil and Argentina this figure might be unreasonable 85 years from now.

The revised impact on the INDC submissions

The INDC submissions can be broken down.

The USA, EU, Japan and Australia all have varying levels of cuts to total emissions. So for the OECD as a whole I estimate Climate Interactive over estimates the impact of the INDCs by 8.4GtCO2e

The Russian INDC pledge it is unclear, but it seems to be saying that emissions will peak before 2030 at below 1990 levels6. As my revised forecast is above this level, I estimate Climate Interactive over estimates the impact of the INDCs by 3.2GtCO2e

The Chinese INDC claims pledges that its emissions will have peaked by 2030. This will have happened anyway and at around 10-12 tonnes per capita. I have therefore assumed that emissions will stay constant from 2030 to 2100 whilst the population is falling. Therefore I estimate that Climate Interactive over estimates the impact of the INDCs by 19.5GtCO2e

Overall for these areas the overestimation is around 31 GtCO2e. Instead of 63.5GtCO2e forecast for these countries for 2100 it should be nearer 32.5GtCO2e. This is about half the total 2100 reduction that Climate Interactive claims that the INDC submission will make from all types of greenhouse gases. A more rigorous forecast may have lower per capita emissions in the OECD and China. There may be other countries where similar forecast issues of CO2 emissions might apply. In addition, in note 7 I briefly look at the “No Action” CH4 emissions, the second largest greenhouse gas. There appear to be similar forecast issued there.

In summary, the make-up of the CO2 emissions “No Action” forecast is bogus. It deviates from an objective and professional forecast in a way that grossly exaggerates the impact of any actions to control GHG emissions, or even pledges that constitute nothing more than saying what would happen anyway.

Notes

  1. The conversion of a given quantity of emissions into average surface temperature changes is outside the scope of this article. Also we will assume that all policy pledges will be fully implemented.
  2. On the Home page use the menu for Tools/C-ROADS. Then on the right hand side select “Download C-ROADS”. Install the software. Run the software. Click on “Create New Run” in the centre of the screen.


    This will generate a spreadsheet “User Scenario v3 026.xls”. The figures I use are in the User Reference Scenario tab. The software version I am working from is v4.026v.071.

  3. The “User Reference Scenario” is claimed to be RCP 8.5. I may post at another time on my reconciliation between the original and the Climate Interactive versions.
  4. The forecast estimates for economic growth and emissions for Russia look quite bizarre when the 5 year percentage changes are graphed.


    I cannot see any reason for growth rates to fall to 1% p.a in the long term. But this is the situation with most others areas as well. Nor can I think of a reason for emissions growth rates to increase from 2030 to 2055, or after 2075 expect as a contrivance for consistency purposes.

  5. The forecast estimates for economic growth and emissions for India look even more bizarre than for Russia when the 5 year percentage changes are graphed.


    I am writing on 30/12/15. To meet the growth forecast for 2010-2015, India’s GDP will need to drop by 20% in the next 24 hours. From 2015 to 2030, the period of the INDC submissions, CO2 emissions are forecast to grow by 8.4%. India’s INDC submission implies GHG emissions growth from 2014 to 2030 of 90% to 100%. Beyond that India is forecast to stagnate to EU growth rates, despite being a lower to middle income country. Also, quite contrary to Russia, emissions growth rates are always lower than economic growth rates.

  6. The Russian Federation INDC states

    Limiting anthropogenic greenhouse gases in Russia to 70-75% of 1990 levels by the year 2030 might be a long-term indicator, subject to the maximum possible account of absorbing capacity of forests.

    This appears as ambiguous, but could be taken as meaning a long term maximum.

  7. CH4 (Methane) emissions per Capita

    I have quickly done a similar analysis of methane emissions per capita as in Figure 2 for CO2 emissions. The scale this time is in kilos, not tonnes.

    There are similarities

  • OECD emissions had been falling but are forecast to rise. The rise is not as great as for CO2.
  • In Russia and China emissions are forecast to rise. In Russia this is by a greater amount than for CO2, in China by a lesser amount.
  • In Africa, per capita emissions are forecast to fall slightly. Between 2010, CH4 emissions are forecast to rise 3.1 times and population by 4.3 times.
  • In both the USA and Other OECD (a composite of CI’s categories) total CH4 emissions are forecast in 2100 to be 2.778 times higher than in 2010. In both China and India total CH4 emissions are forecast in 2100 to be 2.420 times higher than in 2010.



UNFCCC Massively Overstates Impact of INDCs on 2100 Emissions

At the end of October UNFCCC Executive Secretary Christiana Figueres was reported by the BBC as saying

The INDCs have the capability of limiting the forecast temperature rise to around 2.7C by 2100, by no means enough but a lot lower than the estimated four, five, or more degrees of warming projected by many prior to the INDCs.

In the context of the objective of limiting prospective global warming to 2C this statement gives encouraging news. Already the policy proposals are most of the way towards that objective, so a final push at COP21 in Paris is all that is required.

Summary

The analysis by the UNFCCC shows that the policy proposals contained within the INDCs will make very little difference to trends in global emissions of greenhouse gases to 2030. In the accompanying literature, the UNFCCC makes no projections of the difference the INDCs will make beyond 2030. The claim that policy will limit forecast temperature rise to the 2.7C by 2100 is claimed by two other organisations, and is only referenced in a table at the very end of a separate technical annex without any discussion or endorsement. One of these, the IEA, achieves the projection by, post 2050, replacing forecasts contingent on the policy impact of the INDCs with an average of modelled RCP emissions pathways. The RCP website explicitly states that they are not forecasts of potential emissions or climate change, whether with or without policy action. It also states that any of the differences between the pathways be directly attributed to policy differences. The IEA thus replaces real emissions forecasts with data that is unrelated to the real world. The other claim, by Climate Action Tracker, has no explicit statement of how the increasing global emissions through to 2030 start tracking downwards post 2030. Contributing factors may include understating the emissions impact of India and China, along with excluding the likely increasing emissions in the coming decades from the poorest nations.

The claim that any agreement reached in Paris based on the INDCs will constrain to global average temperature rise to 2.7C by 2100 through constraining GHG emissions is therefore unsupported by any rigorous forecast of the policy impact in the referenced documents. Such forecasts are based on making a forecast without policy, then modelling the impact policy will make, stating the assumptions. With 40,000 people attending a conference, the UNFCCC could surely have set aside a couple of million dollars to obtain such a forecast from genuine experts.

In Detail

If Christiana Figueres is correct, the INDC submissions, covering the period 2015-2030 have dramatically changed the course of prospective warming getting two-thirds of the distance between the non-policy and the target of limiting warming to two degrees. Bjorn Lomborg’s recent paper “Impact of Current Climate Proposals” published in the Global Policy journal stated

All climate policies by the US, China, the EU and the rest of the world, implemented from the early 2000s to 2030 and sustained through the century will likely reduce global temperature rise about 0.17°C in 2100. These impact estimates are robust to different calibrations of climate sensitivity, carbon cycling and different climate scenarios. Current climate policy promises will do little to stabilize the climate and their impact will be undetectable for many decades.

Having read the policy proposals on a large number of INDCs I concur with Lomborg. There is very little in the INDCs that will alter the future course of warming. So why the difference between my reading and the UNFCCC? The Executive Secretary has the World’s leading experts behind her, so there must be substantial support for the claim. The BBC article provides a link to the UNFCCC Synthesis report on the aggregate effect of INDCs. The link is to a number of documents. The main document makes no attempt to project forward the policy impacts to 2100. In fact if it did, the prognosis would be similar to Lomborg’s. The main graphic in Figure 2, also as a separate file, is shown below.

The orange is the pre-INDC pledges, the yellow the INDCs and the blues various scenarios to stay below two degrees.

To the right is two graphics for 2025 and 2030. The yellow arrow is “Reduction due to INDCs” and the blue arrow “Remaining reduction for least-cost mitigation“. For 2030 the INDCs seem to get a quarter of the way to the desired reduction. There is nothing about trends beyond 2030. The graphic could not be clearer. If the INDCs are to obtain constrain emissions consistent to the 2C of warming, the increasing trend from 2010 to 2030 would have to be rapidly turned into a decreasing trend post 2030, with global emissions reduced by half in two decades. As the non-policy trend is for about 4.5C of warming, then to obtain a 2.7C forecast requires the INDCs to collectively cause emissions to peak and then start a downward trend.

It is clear that there is no mention at all of the 2.7C of warming by 2100. No bridge of the 70 years from the period covered by the INDCs to the end of the century. What is more there is nothing in the aggregate policy contained in INDCs that would cause global emissions to first peak, then be set on a downward trend. So where is the reference?

For that you need to look in the Technical Annex section M. Summary of results from other studies. Even then the text does not mention 2100, but table 6 does.

Instead of the UNFCCC making projections to 2100 on the basis of the INDCs for themselves, they use those of others. Yet the UNFCCC should have the expertise in projecting the impact of policy. I will look at three – that of another UN organisation and the two estimating 2.7C resulting from the INDCs.

UNEP Gap Report

The link within the footnote to table 6 is to the Executive Summary of the UNEP Emissions Gap Report 2015. The proper reference should have been to all the documents related to the Gap Report found here. The Executive Summary states

Full implementation of unconditional INDC results in emission level estimates in 2030 that are most consistent with scenarios that limit global average temperature increase to below 3.5 °C until 2100 with a greater than 66 per cent chance.

There is no actual projection from the INDCs. Rather, it looks at the emissions levels and emission trends in 2030 and compares them with modelled estimates that are similar. It is these modelled estimates that produce the 3.5C of warming in 2100. There is no reconciliation between the country-by-country INDCs and the overall global emissions scenarios. Rather it is just picking estimates that seem to fit at a global level. In terms of assessing the impacts of policy it is useless, as the modelled estimates may be markedly different from a forecast based on the latest information.

International Energy Agency (IEA) World Energy Outlook 2015

The link within the footnote to table 6 is to a press release for the IEA’s World Energy Outlook 2015. The footnote d. to table 6 gives an explanation of how the 2.7C projection was arrived at. In particular is the final point

To assess the impact on global average temperature increase, we used MAGICC with an emissions pathway post-2050 in between the representative concentration pathways (RCP) 4.5 and (RCP) 6 scenarios from the IPCC’s Fifth Assessment Report as this was interpreted as representing the best available trajectory compatible with IEA’s INDC Scenario.

The RCP (“Representative Concentration Pathways”) scenarios are explained on the RCP website

RCP 4.5: … is a stabilization scenario where total radiative forcing is stabilized before 2100 by employment of a range of technologies and strategies for reducing greenhouse gas emissions.

RCP 6.0: … is a stabilization scenario where total radiative forcing is stabilized after 2100 without overshoot by employment of a range of technologies and strategies for reducing greenhouse gas emissions.

Under “Characteristics and guidance” the website states (bold mine):-

The RCPs are named according to their 2100 radiative forcing level as reported by the individual modeling teams. …….

The RCPs are not forecasts or boundaries for potential emissions, land-use, or climate change. They are also not policy prescriptive in that they were chosen for scientific purposes to represent the span of the radiative forcing literature at the time of their selection and thus facilitate the mapping of a broad climate space. They therefore do not represent specific futures with respect to climate policy action (or no action) or technological, economic, or political viability of specific future pathways or climates. …..

The RCPs are four independent pathways developed by four individual modeling groups. The socioeconomics underlying each RCP are not unique; and, the RCPs are not a set or representative of the range of potential assumptions. …. The differences between the RCPs can therefore not directly be interpreted as a result of climate policy or particular socioeconomic developments. Any differences can be attributed in part to differences between models and scenario assumptions (scientific, economic, and technological).

The IEA has therefore used a hybrid of emissions scenarios as emissions forecasts to assess the impact policy when the group producing them has specifically said that these scenarios are not directly comparable and should not be used for policy purposes. In so doing, they implicitly make a set of assumptions about policy that may not relate to the real world and are definitely not related to the policies proposed within the INDCs.

Climate Action Tracker (CAT)

CAT tracks the INDCs from 32 nations that currently have about 80% of global emissions. Their estimate of the INDC impact until 2030 is broadly consistent with the UNFCCC. However, I am not sure where they obtain the historical emissions figures. For countries they appear to be from the World Resources Institute CAIT2.0. CAIT2.0 figures which are slightly different from those in the UNFCCC Country Briefs, but not markedly so in total. CAT’s methodology does have the advantage of providing a breakdown by country until 2030 between the BAU, but beyond 2030 details are distinctly hazy. The 2.7C claim is made on a briefing of 1st October 2015. The main graph behind the 2.7C estimate is reproduced below.

The impact of the INDCs is to see emissions peak about 2030, then gently fall through to the end of the century. From the detailed explanation it is not possible to determine how the emissions peak at the level as collectively the INDCs appear to show an upward trend. The reasons for this are probably from the following:-

India – The INDC that estimates a tripling of emissions between 2014 and 2030 was not available when the briefing was published, which only assumes a doubling. The country assessment for India then tries to reconcile the difference by some fancy assumptions (lower economic growth rates and a higher emissions intensity reduction than in the INDC), without adjusting the overall assessment. If CAT’s 5GtCO2e estimate of 2030 emissions turns out to be correct (9% of global emissions), it will only be emissions growth delayed not a lower emissions peak. It is unlikely that India’s emissions will peak at less than 10GtCO2e, equivalent to over 20% of 2010 global emissions.

China – will likely reach peak emissions prior to 2030, but that peak will be likely higher than the 13.6GtCO2e forecast The CAT country assessment admits this is the case, but makes no allowance in the emissions forecast.

Missing Countries – Most of Africa, along with Pakistan, Vietnam, Bangladesh, Thailand and Myanmar are missing from the sample. Collectively their current emissions are small, but in the coming decades the share will rise. Africa’s emissions will rise as most of the global population increase in the latter half of the century is forecast to be within the continent. In South Asia there is already economic growth above the world average that will likely continue as the poorer countries follow in the wake of India. By 2100 these countries could collectively have emissions greater than current emissions of the OECD and China combined.

Ambiguities in the INDCs – Many of the INDCs are highly ambiguous. The historical figures are inconsistent; the forecasts are opaque; some key figures are missing; and it is not clear if some pledges in the INDC are in addition to the others, or part of the whole. CAT maximises the impact, rather than trying to frame questions for the submitting countries to clarify. There should be an assessment of these pledge risk factors. These will likely reduce the estimated policy impact.

Without any other hidden assumptions, CAT’s methods are likely to massively overstate the impact of policy. Critically is how increasing global emissions though to 2030 become decreasing global emissions post 2030. As inferred above, I believe it is due to systematic understating emission projections in the sample countries and ignoring the growth in the other countries. It may also be due to making further policy assumptions for the period beyond 2030. We will only be able to assess the impact if CAT provide a full country-by country projections of emissions in 2100 for the sample countries for both BAU and with INDC scenarios, along with projections for the rest of the world. Putting the figures on a table, rather than spending time creating graphs from which figures have to be estimated, would ease the process. If proper forecasts have been generated (that is making a forecast without policy, then modelling the impact policy will make) then the outline figures will be available already.

Concluding Comments

The UNFCCC presents no evidence that policy contained within the INDC submissions will make more than a small difference to global emissions in 2100. Instead they rely on external organisations. One quite clearly substitutes real world forecasts with emissions scenarios that do not relate to real world situations, and assume implementation of policy quite different to that contained with the INDCs. The other is likely to have massively overstated the policy impacts, but a lack of any clear statements as to how the conclusions were arrived at means quantification is not possible. The claims that the policy pledges within the INDCs will massively alter global emissions levels in the latter part of this century (and, subject to the climate models being broadly correct, the rise in global average temperatures) are without any proper foundation. The UNFCCC Executive Secretary Christiana Figueres is has made a misleading statement to drive through policies that are both costly and ineffective.

Such forecasts are based on making a forecast without policy, then modelling the impact policy will make, stating the assumptions. With 40,000 people attending a conference, the UNFCCC could surely have set aside a couple of million dollars to obtain a rigorous forecast from leading experts in that field. The methodology is fairly straightforward. It requires making a forecast for each country without policy, then modelling the impact policy will make, stating the assumptions. The important parts are data gathering, adhering rigorously to a consistent method and leaving an audit trail.

Kevin Marshall