Adjustment costs from environmental change
Introduction
Climate change is clearly one of today's leading global environmental problems, brought on by an increase in atmospheric levels of carbon dioxide and other greenhouse gases. Climate change may manifest itself in many ways, including changes in average temperature and precipitation, variability of weather (such as more frequent hurricanes), and sea levels.1
One of the most important issues in the climate change debate is the quantification and monetization of the market and non-market impacts (damages) from a change in the climate. Most quantitative estimates of damages are modest, which tends to contradict the intuition of many that a significant change in the climate must have very serious consequences.2
Early damage studies measured the short-run response from climate change: if temperatures rise rapidly, what are the losses to a sector (e.g., agriculture), assuming the sector has no time (or inclination) to change practices and adapt?3 Assuming away adaptive actions is known as the “dumb farmer” assumption.4 This was the norm for estimates of damage in the late 1980s and early 1990s.
Recognizing the inappropriateness of assuming away agent adaptive response to climate change, other authors have assumed agents have time to react and adapt to climate change. This implicitly quantifies the long-run impact from climate change, for instance after farmers adapt to the changed climate and after coastal areas adapt to newly defined shorelines. In fact, in the long-run, climate change could have a positive effect, at least on certain sectors and regions. Adaptation is now recognized as an important factor in moderating potential impacts of climate change, a factor which must be incorporated into estimates of damage.5
A third issue, which has received very little attention in the literature (and which is the subject of this paper), concerns the path from short-run to long-run—the adjustment process with associated adjustment costs. When a shock occurs, agents cannot instantly adapt. More precisely, adjustment costs are the extra costs incurred relative to the (counterfactual) case of instantly adapting to changed circumstances.6
The nature of adjustment costs can most easily be understood in the context of a firm subject to a one time price shock for one of its inputs. The literature on the theory of the firm subject to price shocks (beginning with Lucas [21] and nicely summarized in [4]), distinguishes between short-run response, long-run response and the additional costs of adjusting from the short- to the long-run. After the price shock, if the firm makes no attempts to adjust inputs, it will suffer the maximal profit loss. However, the firm can adapt, changing production technique, moderating the loss in profit. But the firm may not be able to adjust its inputs quickly—capital for instance takes time to replace. Additional losses will occur, compared to the case of complete flexibility and ability to instantly respond to the shock. These extra costs are termed “adjustment costs.”
An analogous process occurs with climate change, though instead of a price shock we have a technology shock—the change in climate. When the climate changes, maximal profit loss occurs if the firm makes no changes in its production techniques; i.e., remains a “dumb farmer.” This loss can be moderated by adaptation, reducing the profit loss. Agents are slowed in their ability to instantly adapt to the changed climate for two reasons: input (e.g., capital) fixity and incomplete knowledge of the climate change. In this paper, we focus on the later source of adjustment costs.
There are two purposes of this paper. One is to provide a theoretical structure for viewing adjustment in the context of environmental change, specifically climate change. The second is to apply this theoretical structure and measure adjustment costs; in particular, we consider agriculture and examine the costs of adjustment in US agriculture in the Midwest, a relatively homogenous region agriculturally.
In this paper, the inability to instantly adjust to a changed climate arises because the economic agent (think of a farmer) does not perfectly observe the climate change: the farmer only slowly realizes that the climate has changed. The farmer forms her subjective assessment of what weather might be for the coming year based on currently available information (IPCC predictions and historical weather). Given this structure, we simulate how a farmer would react to an unobserved change in the climate (i.e., the distribution of weather). The farmer would initially think she is seeing unusual weather, only slowly updating her estimate of the true climate. Eventually she learns the true nature of the change in the climate; in the meantime, her production decisions are suboptimal (relative to perfect information) and thus her profits suffer, generating adjustment costs.
To implement this concept of adjustment costs, we first econometrically estimate how US farm profit is influenced by a variety of factors, including realized weather and expectations about weather. Expectations are assumed to be derived from observing past weather. We then simulate how farmers would respond to a change in climate (mean temperature and precipitation). We consider two cases, one where the farmer is fully informed of the change in the climate and another where the farmer only realizes the climate has changed by observing unusual weather and slowly updating her prior on the climate. The difference in profits for these two cases is the adjustment cost.
We argue that adjustment costs are likely to be quantitatively important for many cases of environmental change. Environmental change results in many winners and losers, which tend to cancel each other out. But adjustment costs are a cost to everyone, regardless of whether or not eventually environmental change is a benefit or a cost.
Indeed, we show that incomplete knowledge of the climate can indeed lead to a loss in profits. We examine agriculture in the Midwest using data from the 1976–1997 period, and then simulate what might happen over a century from gradual unobserved climate change.7 Although we find that this change in climate eventually increases annual expected profits by 3.7%, the net present value of adjustment costs is 1.4% of annual land rents. In absolute net present value terms, in our sample climate change results in a long-run gain of $0.54 billion but adjustment costs of $0.24 billion, for a net gain of $0.30 billion.
Section snippets
Background
If farmers know the climate has changed, how do they respond in the long-run and what is the effect on welfare? We call the equilibrium response the profits or welfare under environmental change under full information less profits or welfare with no environmental change. The equilibrium response includes any adaptation that economic agents may make to the environmental change. Understanding the transition to a different climate is somewhat more subtle. Under our assumption that all inputs are
A model of adjustment costs
In our method of modeling adjustment costs, agents learn about an unobserved environmental change by observing the weather and updating using Bayes rule. More importantly, we use the Ricardian framework and assume that decision making by farmers under uncertainty is implicit in agricultural profits. Specifically, we show that profits given uncertainty about environmental change in a county with a given climate variance is equivalent to an otherwise similar county with no uncertainty about
Estimation
In implementing the theory of the previous section, we focus on agriculture in a portion of the US. Our goal is to estimate an ex post profit function, an empirical version of Eq. (3). The ex post profit function is the primitive that permits us to calculate the equilibrium response (Eq. 8) and adjustment costs (Eq. (13)). This function specifies profits as a function of input prices, output prices, climate, weather, and other technological factors. In our application, profits will be
Adjustment costs for climate change
The estimated profit function presented in the previous section gives us the information we need to compute adjustment costs. As we have argued here, the adjustment costs are a direct function of the agent's information regarding the true distribution of weather and the speed with which the agent acquires information. In order to compute meaningful adjustment costs, we compute them for a change in the climate commonly used in the literature. We also make the rational expectations assumption
Conclusions
The goals of this paper have been to develop a theoretical and empirical structure to conceptualize, examine, and compute adjustment costs from incomplete information about a change in the distribution of technology shocks. This is an extension of the well-known work involving adjustment costs from price shocks due to physical rigidities in inputs. In contrast to the factor-fixity basis for price-induced adjustment costs, we have focused on the incomplete information about a shift in the
Acknowledgments
Work supported in part by USDA cooperative agreement 43-3AEL-6-80050 and USDOE Grants FG03-96ER62277 and FG03-00ER63033. Able research assistance from Catherine Dibble is acknowledged. Comments by two anonymous referees, Peter Berck, Hadi Dowlatabadi, Joseph A. Herriges, Bill Nordhaus, John Reilly, Jean-Charles Rochet, Kerry Smith, Douglas Steigerwald, Jim Sweeney, David Zilberman and workshop participants at Duke University, Harvard University, MIT, Stanford University, the University of
References (54)
- et al.
Adjusting to climate change: implications of increased variability and asymmetric adjustment costs for investment in water reserves
J. Environ. Econ. Manage.
(1997) - et al.
Bayesian learning, pollution, and growth
J. Econ. Dynam. Control
(1999) - et al.
Climate impacts on aggregate farm value: accounting for adaptation
Agri. Forest Meteorol.
(1996) - et al.
Value of improved long-range weather information
Contemporary Econ. Pol.
(1995) - et al.
Economic effects of climate change on US agriculture
- et al.
Global climate change and US agriculture
Nature
(1990) - et al.
Dynamic models of energy demand: an assessment and comparison
- et al.
Economic growth and business cycles
Optimal Statistical Decisions
(1970)- et al.
Flexible functional forms and global curvature conditions
Econometrica
(1987)
Rules of thumb for social learning
J. Polit. Economy
Information and the speed of innovation adoption
Amer. J. Agr. Econ.
Farmer response to changes in climate: the case of corn production
J. Agr. Econ. Res.
Agricultural land price differentials and their relationship to potentially modifiable aspects of the climate
Rev. Econ. Statist.
A farm-level analysis of economic and agronomic impacts of gradual climate warming
Amer. J. Agr. Econ.
Linking adaptation and mitigation in climate change policy
Climatic Change
A biophysical model of corn yield: integrating climatic and social determinants
Amer. J. Agr. Econ.
Stochastic trend, weather and US corn yield variability
Appl. Econ.
Adjustment costs and the theory of supply
J. Polit. Economy
Welfare analysis of incomplete adjustment to climatic change
The impact of global warming on agriculture: a Ricardian analysis
Amer. Econ. Rev.
The impact of climate variation on US agriculture
Cited by (131)
Adapting to climate change accounting for individual beliefs
2024, Journal of Development EconomicsSufficient statistics for climate change counterfactuals
2024, Journal of Environmental Economics and ManagementA unifying approach to measuring climate change impacts and adaptation
2023, Journal of Environmental Economics and ManagementThe impact of climate risk on earnings management: International evidence
2021, Journal of Accounting and Public PolicyDaily weather only has small effects on wellbeing in the US
2020, Journal of Economic Behavior and OrganizationClimate change and index insurance demand: Evidence from a framed field experiment in Tanzania
2020, Journal of Economic Behavior and OrganizationCitation Excerpt :Our results are consistent with these findings in that participants slow response to probability changes is due to the considerable weight put on prior beliefs. Furthermore, Kelly et al. (2005) employ a Bayesian learning model to show that agricultural firms incur adjustment costs when learning about a changing distribution. While we do not model adjustment costs directly, we show that the utility loss due to learning results in sub-optimal demand for insurance when compared to the full-information scenario.