Knowledge for better food systems

FCRN Blogs

The implications of empirical and 1:1 substitution ratios for consequential LCA: using a 1 % tax on whole milk as an illustrative example

October 27, 2015
Neil Chalmers

This blog-post is written by FCRN member Neil Chalmers who is studying a PhD titled “Demand for low carbon food products” at the University of Edinburgh. Neil was educated at the University of Stirling where he received a BA (Hons) in Economics. He then moved to Denmark and received an MSc in Agricultural Economics from the University of Copenhagen. His main interests are the economics of consumer behaviour and carbon policy and modelling the likely effects of agricultural and environmental policy. Neil has also completed an internship with the Scottish Agricultural College focussing on modelling policy implications for Scotland.

This post discusses a recent paper Neil has co-authored together with Matthew Brander (University of Edinburgh Business School) and Cesar Revoredo-Giha (Scotland’s Rural College (SRUC) published in The International Journal of Life Cycle Assessment.


What is the difference between the carbon footprint of whole milk (3.5 % fat) and low fat milk (1 % fat)?  The initial expectation might be “Not much”, especially given that all milk goes through a skimming process, and then both whole and low fat milk have fat added back in. Most existing LCA studies of milk don’t consider the effects of different amounts of fat, and so there may be an expectation that the difference isn’t very significant. However, the choice of life cycle assessment (LCA) adopted will in fact give rise to different answers.  There are two kinds of LCA: An attributional LCA (ALCA) estimates the greenhouse gas (GHG) emissions associated with the processes used in the life cycle of a product. Consequential LCA (CLCA) estimates the consequence of decisions such as increasing the price of whole milk and how this affects GHG emissions due to consumer substitutions with low fat milk.

This study adopted a CLCA approach and we found that the greenhouse gas emissions associated with consuming whole milk could in fact be up to 120% higher than low fat milk.  The reason is that the fat co-product from low fat milk can be used as a substitute palm oil for other foods (e.g. processed foods), - and as palm oil production can generate very high emissions due to its role in deforestation and cultivation on peatlands. Our study therefore concluded that switching from whole to low fat milk could, by displacing palm oil production potentially create a large reduction in emissions. Note that we only consider these two milk products and a future study could incorporate other milk based foods.  This is something of a (milk) shake-up for existing life cycle studies on milk, which haven’t previously considered this effect.

A further shake-up for existing CLCA practice comes if we explore the effects of a hypothetical tax on whole milk, aimed at reducing greenhouse gas emissions. Let’s milk this example further, since taxing food products for health reasons is not a new concept- recent examples include the Mexican and French soft drinks tax. We don’t consider how realistic the imposition of such a milk tax might be from a policy making perspective – we only use it for exploring the consequences for emissions arising from a change in demand.

 

Traditionally LCA assumes a one-to-one substitution ratio between alternative product options.  For example, if the tax causes a switch to low fat milk, then it is assumed that for every 1 litre of whole milk reduced an extra 1 litre of low fat milk is produced. While this may seem simplistic, this kind of 1:1 substitution ratio relationship is often considered as standard practice for both ALCA and CLCA. The 1:1 substitution ratio allows the effect of the tax to be estimated by subtracting the emissions associated with whole milk from the emissions associated with the functionally equivalent quantity of low fat milk.

However, using an econometric model (a linear approximated almost ideal demand system, or LA-AIDS[1]) and data for Scottish household purchasing behaviour, we found that the substitution ratio between whole and low fat milk is approximately one-to-half rather than one-to-one (this is derived from the price elasticities[2]). Depending on the emission values assumed for palm oil, using a one-to-one substitution ratio could underestimate the greenhouse gas reductions caused by the tax by over 400%.  This illustrates the importance of estimating the actual substitution effects from interventions, such as a tax, and not using the traditional LCA convention of a one-to-one ratio. If consequential LCA is to fulfil its purpose of estimating the actual change in emissions caused by specific decisions or actions then it seems the existing guidance and standards (e.g. ISO 14044 and the ILCD Handbook) should be amended, with the 1:1 substitution ratio treated as a rule-of-thumb or default value, and not a methodological principle.

 

The citation for the paper is as follows:

Chalmers, N. G., Brander, M. & Revoredo-Giha, C. (2015). The implications of empirical and 1:1 substitution ratios for consequential LCA: using a 1% tax on whole milk as an illustrative example. The International Journal of Life Cycle Assessment, 20(9): pp.1268-1276.

 

To read the full paper, please go to:

http://link.springer.com/article/10.1007/s11367-015-0939-y

 

We welcome your comments on this post in the comments box below. You will need to be logged in as a member on the FCRN website to post a new comment.  

 

[1] More information on the AIDS which was developed by the recent Nobel prize winner (Angus Deaton) can be found on Wikipedia: https://en.wikipedia.org/wiki/Almost_ideal_demand_system

[2] Price elasticities measure the responsiveness of a change in quantity demanded to a change in price. The substitution ratio is obtained by first calculating the change in consumption (induced through a 1% price increase of whole milk) of the two milk products by applying the price elasticities to the volume of milk purchased in Scotland. Then the ratio is obtained by dividing the change in low fat milk consumption by the change in whole milk consumption which equals 0.52.

Comments

John Kazer's picture
Submitted by John Kazer (not verified) on

You raise some interesting points Neil, and parallel my view that the proper use of consequential LCA is to support strategic decision making rather than address comparability or supply chain engagement.

However, I think you fall between two stools somewhat.  At one end being the attributional approach, which in my direct experience generated differential footprints for skimmed, semi-skimmed and whole milk at Tesco of 1.2, 1.4 and 1.6 KgCO2e/Lt respectively (full lifecycle).  The main differences being the use of fats for other products (e.g. butter, cream) and subsequent allocation of GHGs.

I presume these differences would be larger if we had included the consequences of using milk fat rather than palm oil (for example).  However, are milk fats a suitable substitue for palm oil?  I don't know enough about product formulation to say.  I may also suggest that purchasing palm oil which did not cause deforestation is actually a potentially more efficient way to get fat into our foods (or other products) than milk.

Also, I think that using attributional footprints within a full diet analysis can provide the same insights as a consequential analysis - which in this context can perhaps be seen as a sub-set of that dietary approach?

So I will follow the conclusions of a consequential analysis *only* if I can see how the assumptions would fit within the context of a full diet.  For example, you have used Scottish household purchasing information which gives the total amount of each milk type purchased, but presumably also the amount of butter, cream etc. which will have been made as a co-product from skimmed and semi-skimmed milk.  The calculation, *I think* then becomes how much fat is left over to displace palm oil?  Perhaps not so much...

Miguel Astudillo's picture
Submitted by Miguel Astudillo (not verified) on

This blog post has been written a while ago, but since I am doing a PhD using consequential LCA I think it is still worth commenting :-).

I think that using attributional footprints within a full diet analysis can provide the same insights as a consequential analysis - which in this context can perhaps be seen as a sub-set of that dietary approach?

So I will follow the conclusions of a consequential analysis *only* if I can see how the assumptions would fit within the context of a full diet.

In this point I quite disagree. If you want to assess the effects of a dietary change you need to consider market driven substitution “indirect” effects that are neglected in attributional analyses. All the problematics with the first generation of biofuels are a good example of the consequences of neglecting substitution effects. The study presented by Neil illustrates how different marginal vs average and allocation vs substitution can be. To put it other way, attributional and consequential analyses respond to different questions, so we should not expect the same answer.

It is true that consequential analyses require more explicit assumptions and these should be transparently presented. Same goes for attributional LCA, and assuming attributional data is representative of the effect of a dietary changes is indeed a strong assumption.

Finally, I quote here “How low can we go?” from Audsley et al (2009). That can be found in the FCRN website.

Attributional LCA (ALCA) is useful for allocating "responsibility" for emissions, based as closely as possible on the causal relationship between the emissions and the entity to which they are allocated. It is also the appropriate approach for consumption-based carbon accounting as it avoids double-counting emissions. However, it does not capture all the complexities and consequences of specific mitigation actions or policies.

in order to quantify the full GHG consequences of an action, consequential LCA (CLCA) is required. CLCA looks at marginal changes arising from actions and quantifies all the consequences which flow from this. The attributional approach is therefore useful for estimating the size of LUC emissions attributable to UK food consumption, and it can indicate possible  agricultural system, beef has more embedded emissions than poultry meat. The attributional approach is essentially a system of accounting emissions and attributing them to commodities as currently produced and consumed. However, it does not say what the full consequences of a significant shift from beef to poultry would be. For example, a reduction in beef consumption may increase reliance on male calves from the dairy herd reducing the burdens from beef production. It should be noted that this limitation with attributional analysis arises for most emissions sources across the economy. For example, a grid average emissions factor is used when allocating emissions from electricity consumption (within an ALCA). However, when quantifying the actual emissions reductions from reducing electricity consumption the grid margin should be used, and other consequences from the action should also be taken into account. The relationship between the attributional figures for LUC and the emissions reductions achieved by specific mitigation options is likely to be less close than for other emissions sources, given the complexity of the causal interactions between demand for a food commodity and LUC (particularly indirect land use change). Attributional figures help to indicate possible mitigation options, such as switching from foods which have high land area requirements to those that have lower land area requirements. However, such options should be investigated in greater detail using consequential analysis, in order to accurately assess the emissions reductions achieved.

Animal logic's picture
Submitted by Animal logic (not verified) on

Neil,

This is an interesting contribution to the task of reducing the carbon foot print of livestock. Your thesis I think would be improved by a better treatment of the substitution side. The costs and benefits of Palm Oil cultivation are much debated and often one sided. It does not follow that the use of Palm oil necessarily results in rain forest loss and according to one study; WWF Germany, in cooperation with WWF Swizterland and Netherlands, even where it does the real cost is in biodiversity loss, not carbon or energy losses (Rain Forest for Bio Diesel, the ecological effects of using palm oil as a source of energy, WWF 2007) [This biodiversity loss is a very serious problem to be sure but not central to your thesis]

Palm Oil as a source of nutrition is similarly subject to over simplification, too much for this note to summarise, but a reading of this paper might help you with this. (Palm Oil Contraversies. Oil Palm and Development Challenges Rival A and Levang P. Center for International Forest research (CIRAD) 2014).

As an aside, the whole question of Livestock as a negative impact on the carbon cycle is also more complex than often presented. Livestock are produced under many food systems with greatly differing climate (and other) impacts. As one example I refer you to (Challenges and Opportunities for Carbon Sequestration in grassland systems. A Technical Report on Grassland Mangement and Climate Change Mitigation, FAO Intergrated Crop Management Vol 9- 2010) . Properly managed grass lands can, and have been, a long term net carbon sink, Milk is but one (very important) traditional product of such livestock systems that very large numbers of quite poor people depend on for their livelihood. 

Add new comment

Member input

Plain text

CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.