The bounty of plenty, in the twenty-first century, is questioned not due to its productive capabilities but rather because of its lack of a palatable narrative of place and transparency to which it could inform, and reassure, its consumers and citizens. This conundrum is eloquently summed up by Connerton : “As natural ecosystems became more intimately linked to the urban marketplace of Chicago, they came to appear ever more remote from the busy place that was Chicago. Chicago both fostered an ever-closer connection between city and country, and concealed the very linkages it was creating.” Problems and issues associated with modern agricultural production, for instance is to a large degree perceptual. This is not to say that these developments are not indeed real, but that the ultimately overall meanings and thus societal importance and individual significance rest, to a large degree, in the perceptions of such developments and their perceived effects, as one cannot know everything for sure, and even what sure, or the truth, means might be debatable as we move from “Authority to Authenticity” . Perceptions about agriculture, food and people, therefore, does not just serve to navigate, interpret and/or internalize the symbols imposed by other agents and structures, but also, in their appropriation, circumvents and re-invents meanings and understandings of these. In other words, the importance of perceptions lies both in their use as “interpreters” of the surrounding world and as creators of the surroundings – the change perspective. These must, surely, not be underestimated, as is evident in the burgeoning food movements and its spatial outlets,precio de macetas de plastico as well as the increasing organic production worldwide. Also, important is the fact that participation in the local sphere is for many in the Western World a choice made out of want and less so out of social and economic necessity as in the past – perhaps reflected in an increase in informalization .
This “choice,” it should be noted still, often, excludes and eludes people with low socio-economic status, who still depend very much on locality, as their mobility and often also skills are more limited and less “mobile.” Context continues to matter, greatly, if for different reasons, and with different effects, than in yesteryears. Lastly, I would like to add that this paper is not a rebuff of the criticism levelled at the food markets and its actors in regards to its, at times, exploitative attitudes towards the nature of resources , workers governance and politics , among others. It is rather an attempt to show how these held perceptions came about, and how structural and historical developments can contribute to explaining their emergence, without retorting to complete structural determinism or complete rational-choice “free” actor perspectives, the two preferred methodological and philosophical lenses that the oppositional camps of these issues wear. Rather, reality is formed by perceptions that both influence and are influenced by individual choices and structural developments, whose autonomy and power structures vary depending on context. There is mounting evidence that the global climate has already changed and it is projected to continue changing for the coming centuries . The world has experienced many new record highs that suggest that the mean temperature is increasing. For example, Munasinghe et al. examine the frequency of new record temperatures across the global landmass and find that the frequency of extremely high temperatures increased tenfold between the beginning of the 20th century and 1999–2008, the most recent decade for which they obtained gridded weather data. At the same time, the frequency of new record lows has also increased, suggesting that the variance and not only the mean may have increased. A spatially disaggregate analysis reveals that the tropics experienced a larger increase in the frequency of record highs during the last 100 years than higher latitudes. This is consistent with forecasts of global circulation models . Looking across 23 circulation models, the authors find that countries in the tropics have a probability greater than 90% of experiencing average summer temperatures by the end of the 21st century that are larger than the hottest summers on record in 1900–2006.
In higher latitudes, the average seasonal temperature will be about equal to the hottest on record for the period 1900–2006. On the other hand, Hsiang and Parshall plot the distribution of absolute changes in predicted temperatures for a number of global circulation models and emphasize that the higher latitudes have larger predicted increases in temperature. While this might at first seem like a contradiction, the reason for this finding is that there is less historic variation in the tropics than in the higher latitudes, and more of the increased warming in the higher latitudes will occur during the winter time. The key features of observed trends as well as future warming are the observed and predicted non-uniformity of warming as well as sharp increase in record highs, especially in lower latitudes that generally have less institutional capacity to adapt to these new records. The predicted change in the mean and variance of weather has direct implications for agriculture, since weather is a direct input into the production function. Unlike many other sectors of the economy that are shielded from weather fluctuations through buildings, agriculture is still at the direct mercy of weather fluctuations . As we will discuss below, the relationship between yields and weather is highly nonlinear and concave. The best predictor of corn yields is a measure of extreme heat over the growing season that only incorporates temperature above29 °C , and slightly higher thresholds apply to soybeans and cotton. Future impacts crucially depend on how often and by how much this threshold will be passed, which can both occur due to an increase in the mean or the variance. As Munasinghe et al. have shown, the observed trend is fairly large. It is generally easier to adapt to shifts in the mean than to shifts in the variance, as optimal crop varieties have to be chosen and planted before the unknown weather is realized. An anticipated change in the mean can be incorporated at the time the planting decision is made, while a change in the variance increases the uncertainty of what will happen after the crop is planted.
Adequate adaptation to an increase in the variance hence has to allow for flexible adjustments during the growing season, e.g., the construction of irrigation systems that can counterbalance fluctuations in temperatures, which increase water demands, as well as fluctuations in precipitation. The majority of studies so far have examined the effects of changes in the mean climate, while estimates of the effects of an increase in the variance are just starting to emerge. It is therefore paramount that empirical studies as well as integrated assessment models move away from impact evaluations that only look at changes in average global temperature or rely on a single global circulation model . Further, reliance on average temperature in these modeling exercises does not properly capture the spatial and seasonal heterogeneity in predicted temperature changes. This reasoning carries over to predicted changes in precipitation, for which there is much less agreement across models. There is a myriad of studies examining the effect of weather/climate on agriculture, both structural integrated assessment models and reduced-form empirical studies. Chetty sees the advantages of reduced-form strategies in “transparent and credible identification”, while the important advantage of structural models is “the ability to make predictions about counterfactual outcomes and welfare.” This paper discusses the issues involved in identifying the impact of climate change on agriculture both on the intensive and extensive margins. We put a special focus on the role of extreme temperatures. Hertel and Lobell in this issue discuss the literature on structural modeling approaches for this important sector. The paper is not meant to be a universal overview of the literature,macetas para viveros but as a survey of issues facing empirical researchers interested in identifying impacts in this important coupled human/natural system. The remainder of our paper is organized as follows. Section 2 summarizes the issues involved in identifying the impact of climate/ weather on agriculture, emphasizing the importance of extreme weather outcomes. Section 3 discusses issues involved in identifying evidence of adaptation in the agricultural sector. Section 4 concludes.There is a long history of empirical estimates of the effect of weather on agricultural outcomes. For example, Fisher developed the concept of maximum likelihood estimation by linking wheat yields to precipitation outcomes.Weather impacts agricultural outcomes, yet humans traditionally have not been able to influence year-to-year weather fluctuations. Only recently have cloud seeding experiments been used to influence precipitation. While it is impossible to summarize the entire history of empirical studies, we focus our attention to the most recent studies. Advances in computer power and data availability have made it possible to fit models with a huge number of observations, which allow for the identification and estimation of a more flexible relationship between weather variables and agricultural outcomes.
One of the most important differentiating factors between econometric studies is the source of variation the study uses to link agricultural outcomes to weather/climate: one has to either rely on time series variation, cross-sectional variation, or a combination of the two in a panel setting. Each source of identifying variation will be discussed in turn. Agronomic field experiments have linked agricultural outcomes to various weather measures in both controlled laboratory settings as well as real-world settings that rely on farm-level data. The number of plots or parcels has traditionally been very limited. For example, McIntosh outlines how time-series variation over two or more field experiments can be combined in a statistical setting. Such field experiments have been used to examine not only the effects of weather variables, but more generally of all sort of inputs, including fertilizer, CO2 concentrations, etc. The estimated weather parameters have been used to predict the effects of changes in climate. This approach has been criticized as “dumb farmer” scenario, as it implicitly assumes that farmers continue to grow the same crop even if the climate is permanently altered. One extension is hence to derive predicted yields under various climate change scenarios and then model the effect of inputs, crop choice, and prices . In their seminal paper, Mendelsohn et al. use a cross sectional analysis that links county-level farmland values in the United State to climatic variables as well as other controls . The advantage of the cross-sectional approach is that farmers in different climatic zones had time to adjust their production system to different climates. For example, if it were to become permanently warmer in Iowa, farmers could adjust their production systems to cope with the hotter climate, just as farmers in Florida have done in the past. Florida farmers currently face higher average temperatures than farmers in Iowa, and hence might be a good case study of how farmers will adapt. There are, however, at least three significant drawbacks to cross sectional studies of this type. First, any cross-sectional analysis is subject to omitted variable bias, as statistical correlations do not imply causation. For example, Schlenker et al. show that access to highly subsidized irrigation water is positively correlated with hotter temperatures. The benefits of higher temperatures in a cross-sectional analysis are upward biased as they also include the beneficial effect of access to subsidized irrigation water. Second, Timmins shows that within-county heterogeneity and endogenous land use decisions can bias Ricardian analyses by allowing for use-specific error terms in his cross-sectional analysis of county-level Brazilian farmland values. Farmers endogenously select the crop they are best suited to grow. The effect of climate on land values hence depends both on how a particular land use responds to climatic conditions, as well as what land use is selected as a function of climate. Third, traditional cross-sectional analyses of farmland values are partial-equilibrium studies. If weather were to make farming either greatly more or less productive, prices for agricultural goods would adjust, and so would farmland values. This is evident in the recent sharp increase in commodity prices that led to a significant increase in the US farmland values. Consumer surplus decreased while producer surplus increased. A decrease in farm productivity might in some circumstance even be good for farmers as demand for agricultural products is highly inelastic and weather-induced yield reductions increase the price of agricultural commodities.