Skip to main content

On the evaluation of heterogeneous climate change impacts on US agriculture: does group membership matter?


The Ricardian literature has only a handful of contributions addressing the presence of spatial heterogeneity in the marginal effects of climate change on agriculture. Although the majority of these studies offer models with group-specific slope parameters to account for spatial heterogeneity, large discrepancies on which grouping should be preferred still exist. This paper evaluates the extent to which expected future agricultural profits is sensitive to the four pre-determined groupings currently used in the literature. The results indicate that accounting for grouping uncertainty greatly increases the confidence interval around projected climate impacts. In addition, we do not find that one type of grouping is superior to any other. We suggest two potential solutions and emphasize the importance of explicitly controlling for grouping uncertainty in future studies.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Data Availability

We will make the replication data available to the public.


  1. The nonlinear effect of climate can also be modeled through Mth order polynomials (Schlenker and Roberts 2009) or an M-piecewise “binned” function (Barreca et al. 2016). Since our focal point is grouping uncertainty, it is beyond the scope of this paper to investigate whether adding nonlinear effects would lead to an efficiency gain.

  2. In Timmins’ (2006) approach, the categories of land use are forest, pasture, temporary crops, and permanent crops.

  3. For simplicity, we assume here that the effect of precipitation and all other exogenous determinants of net return or profits are appropriately controlled for.

  4. Figure 1 is adapted from Mendelsohn et al. (1994). We choose temperature on the x-axis to illustrate the rationale of varying slope in the Ricardian framework. A similar conceptual framework could have been done for precipitation.

  5. All figures are in 2012 constant US dollars.

  6. Some states are grouped together though. One group is composed of AZ and NV. Another one includes CT, DE, MA, MD, ME, NH, NJ, RI, and VT. The authors do not explain the reason(s) for such groupings.

  7. The rural counties are defined as the counties having a population density below 400 inhabitants per square mile (Schlenker et al. 2006; Dall’erba and Domínguez 2016). We drop 34 rural counties that have missing socio-economic or soil data for 1 or more years. In addition, there are 143 instances missing agricultural profit observation, which are replaced with the average of the remaining years’ values. The sample restrictions are imposed to provide a balanced panel of counties from 1997 to 2012 for the subsequent analysis.

  8. Recent research shows that the threshold to calculate growing season degree-days is broadly representative of major crops such as corn, soybeans, and cotton (Ritchie and NeSmith 1991). Some crops and varieties have different growing seasons. For example, winter wheat is planted in September through November and harvested in late Spring and early Summer.

  9. The soil data and the weather conditions are initially two raster datasets. We aggregate each of them to the county level using the ArcGIS toolbox.

  10. Under this assumption, slope coefficients are treated as completely different across different individual units, and each βi is estimated by a separate regression. Traditional panel models frequently assume complete slope homogeneity. Extension to other covariates is straightforward.

  11. The discount rate is the ratio of net farm income to aggregate farm value. It allows us to convert annual profits into land values.

  12. The results in panel are available from the authors upon request.

  13. Tables S5–S8 report the results of robustness checks based on a log-linear functional form and the presence of temperature-precipitation interactions. These results confirm our main findings.

  14. Because the numerical algorithm in implementation of C-Lasso involves separate regressions at the initial step and half-panel jackknife method to correct bias, a panel with long time series is recommended. Because no county-level measures of the dependent variable, profit per acre, exist beyond agricultural census years, we obtain raw data at the state level to construct annual profits per acre from USDA Agricultural Resource Management Survey (ARMS). We then aggregate all the climate and socioeconomic measures of all counties in a state. Using the time spanning 1993–2002, we are able to compile a balanced panel of 48 states. Note that the paucity of state-level panel data results in a classification on a coarser geographic scale than a county-level panel.

  15. A similar analysis can be done for any other climate variable.


  • Adams RM (1989) Global climate change and agriculture: an economic perspective. Am J Agric Econ 71(5):1272–1279

  • Aghion P, Howitt P, Mayer-Foulkes D (2005) The effect of financial development on convergence: theory and evidence. Q J Econ 120(1):173–222

    Google Scholar 

  • Baltagi BH, Fingleton B, Pirotte A (2014) Spatial lag models with nested random effects: an instrumental variable procedure with an application to English house prices. J Urban Econ 80:76–86

    Article  Google Scholar 

  • Barreca A, Clay K, Deschenes O, Greenstone M, Shapiro JS (2016) Adapting to climate change: the remarkable decline in the US temperature-mortality relationship over the twentieth century. J Polit Econ 124(1):105–159

    Article  Google Scholar 

  • Blanc E, Schlenker W (2017) The use of panel models in assessments of climate impacts on agriculture. Rev Environ Econ Policy 11(2):258–279

    Article  Google Scholar 

  • Brücker H, Siliverstovs B (2006) On the estimation and forecasting of international migration: How relevant is heterogeneity across countries? Empir Econ 31(3):735–754

    Article  Google Scholar 

  • Brunsdon C, Fotheringham AS, Charlton ME (1996) Geographically weighted regression: a method for exploring spatial nonstationarity. Geogr Anal 28(4):281–298

    Article  Google Scholar 

  • Bun MJ, Harrison TD (2019) OLS and IV estimation of regression models including endogenous interaction terms. Econ Rev 38(7):814–827

    Article  Google Scholar 

  • Burke M, Dykema J, Lobell DB, Miguel E, Satyanath S (2015) Incorporating climate uncertainty into estimates of climate change impacts. Rev Econ Stat 97(2):461–471

    Article  Google Scholar 

  • Cheng H (2014) Analysis of panel data. Cambridge University Press

  • Chow GC (1960) Tests of equality between sets of coefficients in two linear regressions. Econometrica 28(3):591–605

    Article  Google Scholar 

  • Dall’erba S, Dominguez F (2016) The Impact of climate change on agriculture in the Southwestern United States: the Ricardian approach revisited. Spat Econ Anal 11(1):46–66

    Article  Google Scholar 

  • Dall’erba S, Chen Z, Nava N (2021) Interstate trade will mitigate the negative impact of climate change on crop profit, American Journal of Agricultural Economics, forthcoming

  • Darwin R (1999) The impact of global warming on agriculture : a Ricardian analysis : comment. Am Econ Rev 89(4):1049–1052

    Article  Google Scholar 

  • Davidson R, MacKinnon JG (1981) Several tests for model specification in the presence of alternative hypotheses. Econometrica: Journal of the Econometric Society 49(3):781–793

    Article  Google Scholar 

  • Davis J, Heller SB (2017) Using causal forests to predict treatment heterogeneity: an application to summer jobs. Am Econ Rev 107(5):546–550

    Article  Google Scholar 

  • Deschênes O, Greenstone M (2007) The economic impacts of climate change: evidence from agricultural output and random fluctuations in weather. Am Econ Rev 97(1):354–385

    Article  Google Scholar 

  • Fisher A, Hanemann M, Roberts MJ, Schlenker W (2012) The economic impacts of climate change: evidence from agricultural output and random fluctuations in weather: comment. Am Econ Rev 102(7):3749–3760

    Article  Google Scholar 

  • Fotheringham AS, Brunsdon C, Charlton M (2003) Geographically weighted regression: the analysis of spatially varying relationships. John Wiley & Sons

  • Goodman AC, Dubin RA (1990) Sample stratification with non-nested alternatives: theory and a hedonic example. Rev Econ Stat 72(1):168

  • Herwartz H, Xu F (2009) Panel data model comparison for empirical saving-investment relations. Appl Econ Lett 16(8):803–807

    Article  Google Scholar 

  • Heutel, G., Miller, N.H. and Molitor, D. (2017) Adaptation and the mortality effects of temperature across US climate regions (No. w23271). National Bureau of Economic Research

  • Hoerling M, Eischeid J, Perlwitz J, Quan XW, Wolter K, Cheng L (2016) Characterizing recent trends in US heavy precipitation. J Clim 29(7):2313–2332

    Article  Google Scholar 

  • Hsiang S (2016) Climate econometrics. Ann Rev Resour Econ 8(1):43–75

    Article  Google Scholar 

  • Karl, T.R., and W.J. Koss (1984) Regional and national monthly, seasonal, and annual temperature weighted by area, 1895-1983

  • Knittel CR, Stolper S (2019) Using machine learning to target treatment: the case of household energy use (No. w26531). Research, National Bureau of Economic

    Book  Google Scholar 

  • Lancaster T (2000) The incidental parameter problem since 1948. J Econ 95(2):391–413

    Article  Google Scholar 

  • LeSage J, Pace K (2009) Introduction to spatial econometrics. Taylor and Francis/CRC, Boca Raton, FL, pp 33487–32742

    Book  Google Scholar 

  • Mendelsohn R, Dinar A (2003) Climate, water, and agriculture. Land Econ 79(3):328–341

    Article  Google Scholar 

  • Mendelsohn RO, Massetti E (2017) The use of cross-sectional analysis to measure climate impacts on agriculture: theory and evidence. Rev Environ Econ Policy 11(2):280–298

    Article  Google Scholar 

  • Mendelsohn BR, Nordhaus WD, Shaw D (1994) The impact of global warming on agriculture: a Ricardian analysis. Am Econ Rev 84(4):753–771

    Google Scholar 

  • Miller S (2020) Causal forest estimation of heterogeneous and time-varying environmental policy effects. J Environ Econ Manag 103:102337

    Article  Google Scholar 

  • Pesaran MH (2015) Time series and panel data econometrics. Oxford University Press

  • Plantinga AJ, Lubowski RN, Stavins RN (2002) The effects of potential land development on agricultural land prices. J Urban Econ 52(3):561–581

    Article  Google Scholar 

  • Polsky C (2004) Putting space and time in Ricardian climate change impact studies: agriculture in the US Great Plains, 1969–1992. Ann Assoc Am Geogr 94(3):549–564

    Article  Google Scholar 

  • Polsky C, Easterling WE (2001) Adaptation to climate variability and change in the US Great Plains: a multi-scale analysis of Ricardian climate sensitivities. Agric Ecosyst Environ 85:133–144

    Article  Google Scholar 

  • Ritchie JT, NeSmith DS (1991) Temperature and crop development. In: Modeling plant and soil systems, (modelingplantan). Madison, WI, pp 5–31

    Google Scholar 

  • Schlenker W, Roberts MJ (2009) Nonlinear temperature effects indicate severe damages to U.S. crop yields under climate change. Proc Natl Acad Sci 106(37):15594–15598

    Article  Google Scholar 

  • Schlenker W, Hanemann WM, Fisher AC (2005) Will U.S. agriculture really benefit from global warming ? Hedonic for irrigation approach. Am Econ Rev 95(1):395–406

    Article  Google Scholar 

  • Schlenker W, Hanemann WMM, Fisher AC (2006) The impact of global warming on the U.S. agriculture: an econometric analysis of optimal growing conditions. Rev Econ Stat 88:113–125

    Article  Google Scholar 

  • Su L, Shi Z, Phillips PCB (2016) Identifying latent structures in panel data. Econometrica 84(6):2215–2264

    Article  Google Scholar 

  • Timmins C (2006) Endogenous land use and the Ricardian valuation of climate change. Environ Resour Econ 33(1):119–142

    Article  Google Scholar 

  • Vanschoenwinkel J, Mendelsohn R, Van Passel S (2016) Do Western and Eastern Europe have the same agricultural climate response? Taking adaptive capacity into account. Glob Environ Chang 41:74–87

    Article  Google Scholar 

  • Wager S, Athey S (2018) Estimation and inference of heterogeneous treatment effects using random forests. J Am Stat Assoc 113(523):1228–1242

    Article  Google Scholar 

  • Wendling T, Jung K, Callahan A, Schuler A, Shah NH, Gallego B (2018) Comparing methods for estimation of heterogeneous treatment effects using observational data from health care databases. Stat Med 37(23):3309–3324

    Article  Google Scholar 

  • Wooldridge JM (2010) Econometric analysis of cross section and panel data. MIT Press

Download references

Code availability (software application or custom code)

We will make the replication code available to the public.


This study was funded by USDA Hatch grant.

Author information

Authors and Affiliations



All authors contributed to the study conception and design. The data collection and econometric analysis were performed by Chang Cai. All authors contributed to the writing of the manuscript. All authors read and approved the final manuscript. Sandy Dall’erba secured funding for this study.

Corresponding author

Correspondence to Chang Cai.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information


(DOCX 2.61 mb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Cai, C., Dall’Erba, S. On the evaluation of heterogeneous climate change impacts on US agriculture: does group membership matter?. Climatic Change 167, 14 (2021).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • Climate change
  • Agriculture
  • Spatial heterogeneity
  • Grouping uncertainty