By Arnaud Costinot, Dave Donaldson and Cory Smith
Many fear that climate change will have severe effects on the global economy, particularly through the threat to food production and farmers’ earnings. This column suggests that much of the potential harm could be avoided if farmers can switch their crops in response to changing relative yields. But it is ‘intra-national trade’ – trade among farmers and between farmers and consumers within countries – rather than international trade that will be crucial in alleviating the consequences of climate change.
A substantial body of research by crop scientists has analysed the likely implications of climate change for agriculture in a variety of locations around the world (Intergovernmental Panel on Climate Change (IPCC), 2007). These micro-level predictions – one for each crop and location – are an essential ingredient for understanding what warmer climates could do to global agricultural markets. But they are of little use on their own since in a globalised world, the impact of micro-level shocks depends not only on the average level of these shocks, but also on their dispersion over space.
For example, if climate change were to affect all crops in all countries uniformly, then there would be little incentive for farmers to adjust which crops they grow or for countries to adjust which crops they import and export. By contrast, if climate change were to have a differential effect on crop yields both within and between countries, then adjustments through production and trade patterns could significantly dampen the adverse consequences of climate change: a country may stop producing a crop for which yields have fallen and import it in exchange for another crop for which yields have remained constant at home.
In short, the macro-level consequences of climate change in a global economy are inherently related to how it affects comparative advantage – the relative capabilities for crop production across crops and locations – around the world. But to date, there has been little research on whether climate change will affect comparative advantage – either within or between countries – and hence the aggregate harm that climate change will cause to agriculture remains an open question.
The impact of climate change on comparative advantage
To shed light on the relationship between climate change and comparative advantage, we draw on a rich set of micro-level data from the Global Agro-Ecological Zones (GAEZ) project at the Food and Agriculture Organization (FAO). The GAEZ dataset uses agronomic models and high-resolution data on geographical characteristics such as soil, topography, elevation and, crucially, climatic conditions to predict the yield that would be obtainable for 10 of the world’s most important crops – bananas, cotton, maize, oil palm, rice, soybeans, sugarcane, tomatoes, wheat and white potatoes – across 1.7 million ‘grid cells’ (small parcels of land) throughout the world.
The GAEZ dataset is available under both contemporary growing conditions and a climate change scenario used by IPCC of the United Nations. By comparing productivity for a given crop under the two scenarios at each of these 1.7 million grid cells, we can therefore directly observe the evolution of comparative advantage across space, as predicted by climatologists and agronomists.
Even a brief glance at these predictions demonstrates a striking degree of heterogeneity in the predicted effects of climate change across both crops and location. For example, many regions are expected to see a differential productivity change in the world’s two most important crops – wheat and rice – and this relative productivity change differs substantially across the 1.7 million grid cells.
Furthermore, the contours of the predicted effects of climate change on crop yields pay no attention to country borders. Within-country heterogeneity is therefore a central feature of the predicted impact of climate change on agriculture.
An economic model of agricultural markets
To go beyond the evolution of comparative advantage documented in the agronomic GAEZ data and quantify the economic macro-consequences of climate change, we need an economic model of agricultural markets that can predict three outcomes:
Where crops are produced and, in turn, which productivity changes are relevant.
How shocks to the supply of crops affect prices around the world.
How changes in productivity and prices map into changes in consumption and income.
We therefore develop a model of production, consumption and trade in which each country consists of a large number of ‘fields’ with heterogeneous productivity across multiple crops. These are the theoretical counterparts of the 1.7 million grid cells in the GAEZ data. In this model, comparative advantage – relative productivity differences across crops and fields – determines the pattern of specialisation within and between countries. Furthermore, potential barriers to trade, such as tariffs or transport costs, prevent the full integration of agricultural markets across countries.
Naturally, such a model contains a number of unknown parameters the values of which crucially determine the strength of each economic force at work. For example, an important parameter governs the extent to which consumers around the world are willing to substitute one crop for another – or one country’s version of a particular crop for another country’s version – in their consumption choices. If climate change were to make local wheat scarce in a given region, how willing would consumers there be to substitute imported wheat, or rice, instead?
A similarly important phenomenon concerns farmers’ abilities to substitute their production of one crop for that of another crop. Finally, an essential unknown is the extent to which international trade appears to be impeded by trade barriers, since this will govern, all else equal, the geographical reach of any given country’s trade.
We use publicly available data, again from the FAO, on international trade, production and prices in 2009 to estimate the unknown parameters of our model. As is always the case, estimation of demand-side substitution patterns is challenging due to the fact that prices paid and quantities consumed in the data move in a correlated manner as a result of unobserved supply- and demand-side variation.
We tackle this problem by using the GAEZ predictions about exogenous determinants of local productivity as a source of supply-side variation that allows us to arrive at unbiased estimates of the true demand-side parameters. After applying a similar procedure to estimate the other model parameters, we find that the within-sample fit of our model for output levels, land use and trade flows is strong.
Armed with estimates of the crucial economic parameters at work, as well as with detailed knowledge of how scientific experts expect the pattern of comparative advantage (across fields and crops) to change around the world, we then simulate the impact of climate change.
To do so, we solve for the economic equilibrium that we would expect to occur once yields for each crop and location around the world change from their current (or ‘pre-climate change’) levels to the agronomists’ predicted post-climate change levels. This is clearly a crude and discrete approximation to the continuous dynamics of actual climate change slowly evolving over many decades. But we believe that it nevertheless cuts to the heart of the matter and provides a useful approximation to reality.
Climate change scenarios
In our first climate change scenario, we imagine that countries are free to trade (subject to trade frictions consistent with our estimates of these parameters) and that farmers face no barriers to changing their output decisions in the face of new yields or prices. Under this scenario, we find very heterogeneous effects across countries, with some countries – such as Malawi, a major agricultural producer and consumer located on the frontline of climate change – experiencing dramatic income losses.
But our overall estimates predict that climate change will account for only a 0.26% decrease in world GDP (gross domestic product). This relatively low number merely reflects that fact that agriculture is a relatively small share of global output: the value of output in the 10 crops in our study is equal to just 1.8% of world GDP. This implies that the predicted impact of climate change is about one-sixth of total crop value, a relatively large impact within agriculture itself.
A potential source of adjustment to climate change will be the ability of farmers to change the crop that they grow in the face of new productivities and prices. To shed light on this mechanism, we consider a second scenario in which countries can trade, but farmers cannot adjust the crop mix that they use on each field.
Under this scenario, the adverse consequences of climate change are significantly larger than in the previous scenario. For the world as a whole, the loss would be three times as large: 0.78% of GDP or two-thirds of total crop value. This illustrates how farmers’ abilities to substitute crop production in response to changes in their comparative advantage – which our micro-level dataset gives us a unique opportunity to study – will play an essential role in our ability to mitigate the negative effects of climate change.
The restriction on farmers’ abilities to adapt to climate change embodied in this scenario is clearly extreme and unrealistic. But so is that in the first scenario, in which we assume that farmers could freely change their crop allocations. A comparison of these two extremes helps us to understand the range of possibilities, but a great deal of future research is needed to understand which end of the range is more realistic.
Another potential source of adjustment at the macro-level is the ability of countries to change what they trade with the rest of the world. To explore the quantitative importance of this economic channel, we consider a final scenario in which export patterns – the shares of crop output exported to the rest of the world – are held fixed before and after climate change (but unlike in the second scenario, farmers can reallocate production).
In contrast to the scenario in which farmers are assumed to be unable to adjust their crop growing patterns, the consequences of climate change in the case of no trade reallocations remain very similar to those obtained under full adjustment. We predict just a 0.27% loss in world GDP when global trade shares cannot adjust, compared to 0.26% if they can.
In other words, imposing a hypothetical constraint that prevents countries from changing what they trade has little bearing on the overall consequences of climate change. This is in strong contrast to the hypothetical constraint on farmer’s abilities to change what they grow, a constraint that would prove detrimental.
Contrasting roles of international and intra-national trade in alleviating consequences of climate change
Putting together these two findings tells us that international trade is likely to play only a minor role in alleviating the consequences of climate change. But ‘intra-national trade’ – trade among farmers and between farmers and consumers within countries – seems to be crucial. This is not surprising given the nature of the damage that scientists expect climate change to cause: strongly heterogeneous across crops and locations, even within countries that occupy a relatively small land area.
Of course, there is a great deal of uncertainty about future climate change and how it will affect crop yields at various locations around the world. We therefore have devoted significant effort to an exploration of how our results change as we consider different climatological and agronomic assumptions about future climatic conditions as well as the contemporary growing conditions used in the GAEZ data.
Not surprisingly, the large uncertainty about future crop yields leads to large uncertainty over the economic consequences of climate change. But the relative importance of production adjustments relative to trade adjustments remains of similar magnitude across all the scenarios that we explore. So while the overall impact is uncertain, the central lesson about mitigation strategies that emerges from our research appears to hold regardless of the extent of climate change uncertainty.
Arnaud Costinot is a Professor of Economics at Massachusetts Institute of Technology. Dave Donaldson teaches and carries out research on topics at the intersection of international/intra-national trade, development economics, economic history, and environmental economics. Born in Chicago, Illinois, Cory Smith graduated with a degree in economics from the Massachusetts Institute of Technology (MIT) in June, 2011 and is currently working as a research assistant.
This column previously appeared on Ideas For India. It summarises ‘Can International Trade Mitigate the Impacts of Climate Change in Agricultural Markets: Evidence from 1.7 Million Fields around the World’ by Arnaud Costinot (MIT), Dave Donaldson (Stanford) and Cory Smith (MIT).
Featured Image Courtesy: Instarix
Stay updated with all the insights.
Navigate news, 1 email day.
Subscribe to Qrius