Policy Research Working Paper               11015




              The Mismeasure of Weather
       Using Remotely Sensed Earth Observation Data
                  in Economic Contexts

                          Anna Josephson
                         Jerey D. Michler
                            Talip Kilic
                         Siobhan Murray




Development Economics
Development Data Group
January 2025
Policy Research Working Paper 11015


  Abstract
 The availability of weather data from remotely sensed Earth                        in Earth observation data in the context of smallholder
 observation data has reduced the cost of including weather                         agricultural productivity. The paper shows that different
 variables in econometric models. Weather variables are                             Earth observation sources use different measurement meth-
 common instrumental variables used to predict economic                             ods. The findings are not robust to the choice of Earth
 outcomes and serve as an input in modeling crop yields                             observation dataset, and the outcomes are not simply affine
 for rainfed agriculture. The use of Earth observation data                         transformations of one another. Thus, the paper suggests
 in econometric applications has only recently been met                             that researchers should exercise caution in using these data
 with critical assessment of the suitability and quality of                         and include robustness checks that test alternative sources
 these data in economics. This paper quantifies the signif-                         of Earth observation data.
 icance and magnitude of the effect of measurement error




 This paper is a product of the Development Data Group, Development Economics. It is part of a larger effort by the
 World Bank to provide open access to its research and make a contribution to development policy discussions around the
 world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may
 be contacted at jdmichler@email.arizona.edu or tkilic@worldbank.org.




         The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
         issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
         names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
         of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
         its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                       Produced by the Research Support Team
       The Mismeasure of Weather: Using Remotely Sensed Earth
               Observation Data in Economic Contexts∗
           Anna Josephson1 , Jeffrey D. Michler1 , Talip Kilic2 , and Siobhan Murray2
                1
                    Department of Agricultural and Resource Economics, University of Arizona
                                   2
                                    Development Data Group, World Bank




JEL Classification : C38, C81, D83, O13, Q12
Keywords : Remote Sensing Data, Socioeconomic Data, Measurement Error, Weather, Sub-Saharan Africa


   ∗
     A pre-analysis plan for this research has been filed with Open Science Framework (OSF): https://osf.io/8hnz5/.
We gratefully acknowledge funding from the World Bank Living Standards Measurement Study (LSMS) and the
Knowledge for Change Program (KCP). This paper has been shaped by conversations with Marc Bellemare, Leah
Bevis, as well as seminar participants at the Methods and Measurement Conference 2021, the AAEA annual meetings
in Chicago and Atlanta, the 31st triennial ICAE conference, and participants in presentations at Arizona State
University, the University of Minnesota, the World Bank, and Virginia Tech. We are especially grateful to Reece
Branham, Alison Conley, Kieran Douglas, Rodrigo Guerra-Su, Emil Kee-Tui, Brian McGreal, and Jacob Taylor for
their diligent work as research assistants and to Oscar Barriga Cabanillas and Aleksandr Michuda for early help in
developing the Stata wxsum package. Thanks also to Laura Josephson for logistical support. We are solely responsible
for any errors or misunderstandings.
                                                     Science must be understood as a social
                                                     phenomenon... not the work of robots programmed
                                                     to collect pure information. Scientific findings
                                                     should not be elevated to the status of immutable
                                                     truths.

                                                        — Stephen Jay Gould, The Mismeasure of Man



1    Introduction
There are 37 weather stations in the World Meteorological Organization’s (WMO) database on the
entire continent of Africa. These stations provide weather data for 1.1 billion persons living on the
continent (Tzachor et al., 2023). This figure stands in contrast to the 636 weather stations across
the European Union and the United States, which provide weather data for 1.2 billion persons
in those two regions (Tzachor et al., 2023). Further, because of their uneven dispersion across
the African continent, the weather stations that exist cover only about 40 percent of the African
population. Weather stations on the continent are often so far apart that the data collected are of
limited use, a condition exacerbated by under-investment in their maintenance, which results in a
deterioration of the frequency and quality of data reporting. Only one in five weather stations in
Africa met the WMO’s reporting standards as of 2019.
    The lack of station data across Africa means that, for much of the continent, the truth of
the weather is unmeasured. The emergence of Earth Observation (EO) and improvements in
weather modeling have enabled the development of remotely sensed weather datasets to fill this
gap. These products provide an estimate of the on-the-ground conditions of weather, albeit detected
from a distance. These EO products can provide myriad measures of weather phenomenon, like
precipitation, temperature, wind speed, or humidity. Each EO product uses a different combination
of weather sensors and methods for interpolating and interpreting the data from those sensors. And
so, while each EO product endeavors to measure the objective truth of weather conditions, each
produces its own “truth.” In theory, these “truths” should all be the same. In practice, they are
not. Figures 1 and 2 show these different “truths” across six remotely sensed EO precipitation
products and three temperature products. One precipitation product reports rainfall of less than
five mm while a different product reports rainfall of more than 47 mm for the same location on the
same day. Temperature also varies by EO product. One product reports a maximum temperature
of 23◦ C while another reports the maximum temperature that day as 27◦ C.
    Where does this disparity leave researchers who want to incorporate EO weather data into
economic contexts? If using ground-based weather station data, the gold standard for weather
measurement, is not possible, then we are left to use EO sources. But, their mismeasurement of
weather can introduce two issues. First, if economists are unaware of these measurement issues, we

                                                 2
may be using the “wrong” data, or a poor proxy, to measure the truth of what weather occurred
where and when. This can introduce either classical or non-classical measurement error into our
estimates and predictions. Second, if economists are aware that different EO products report
different numbers, then we can engage in p-hacking, HARKing, and other unethical behavior by
choosing the EO product that gives us the results we want.
   To determine the magnitude of this disparity and its impact on the use of EO data in eco-
nomic contexts, we integrate nine EO geospatial weather datasets with georeferenced longitudinal
household survey data. The survey data comes from six Sub-Saharan African countries and was
collected under the World Bank’s Living Standards Measurement Study - Integrated Surveys on
Agriculture (LSMS-ISA) initiative. We assess the degree to which mismeasurement matters by
modeling the relationship between weather and smallholder agricultural productivity, as captured
in the LSMS-ISA survey data. We use a simple heuristic to compare coefficients from regressions
of yield on weather as measured by one EO product to coefficients when weather is measured by
a different EO product. Our goal is to provide guidance to researchers looking to integrate EO
weather data with socioeconomic survey data regarding the degree to which their results may be
dependent on their choice of remote sensing EO product.
   We find that the different measurement methods by EO products affect not only the cardinality
of coefficients across the products but also coefficient ordinality. The first result is unsurprising.
Obviously, continuous variables measured differently by different sources will result in different
effect sizes. In applied economics, we are typically not too concerned about differences in relative
size (cardinality) across specifications, as we understand that different choices in data cleaning and
modelling will affect the exact size of a coefficient. What we care about is getting the order correct
(ordinality). That using different EO products affects the ordinality of coefficients means that EO
products are not simply affine transformations of one another. One can completely reverse the
ordering of coefficient size across EO products by simply changing the specification, country, or
dependent variable. A researcher can essentially get whatever result she or he wants through the
judicious choice of which EO weather product to use.
   That different EO sources produce a different orderings of outcomes suggests that many results
in economics that rely on EO weather data, or any model-derived EO product, may not be robust
to the choice of EO product. As Jain (2020) has pointed out, this is a concern for the large body
of economic research that has relied on EO weather data for identification of causal effects. This
literature includes important contributions to understanding human capital formation (Maccini
and Yang, 2009; Shah and Steinberg, 2017; Garg et al., 2020), labor markets (Jayachandran, 2006;
                                                                          uckner and Ciccone,
Chen et al., 2017; Kaur, 2019; Morten, 2019), conflict and institutions (Br¨
                      onig et al., 2017), agricultural production and economic growth (Miguel
2011; Sarsons, 2015; K¨
                   ene and Greenstone, 2007; Barrios et al., 2010; Dell et al., 2012; Yeh et al., 2020),
et al., 2004; Deschˆ
intra-household bargaining power (Corno et al., 2020), technology adoption (Taraz, 2018; Jagnani



                                                   3
                  on et al., 2021; Tesfaye et al., 2021), and extreme weather impacts (Wineman
et al., 2021; Arag´
et al., 2017; Michler et al., 2019; McCarthy et al., 2021). Our findings suggest that economists
should exercise caution when seeking to combine EO data with socioeconomic survey data, as
the selection of different EO sources may produce very different results, thus contributing to this
literature on the use of EO data in economic applications.
    Our paper also contributes to the nascent, but growing, literature on the economic implications
of measurement error. Much of the research in this stream has focused on mismeasurement in the
context of agricultural production and focuses on mismeasurement in self-reported data (Carletto
et al., 2017; Abay et al., 2019; Gourley et al., 2019; Kosmowski et al., 2019; Lobell et al., 2020; Abay,
2020; Abay et al., 2021; Gollin and Udry, 2021; Kilic et al., 2021; Michler et al., 2022). A related
literature focuses on mismeasurement in everything from prices (Fitzpatrick, 2023), to consumption
(Abate et al., 2023), to time use (Field et al., 2023), to willingness to pay (Jack et al., 2022). A
smaller literature has begun to document, for economists, the presences of measurement error in
EO data (Gibson, 2021; Gibson et al., 2021; Alix-Garcia and Millimet, 2023), an issue long known
and discussed in GIS, remote sensing, and geophysical science (Wilkinson, 1996). We add to this
literature as the first attempt, to our knowledge, at understanding the impact of mismeasurement in
EO products in estimating policy relevant relationships between climate and development outcomes
in low- and middle-income countries.
    Understanding the differences in results that arise when a researcher chooses one EO product
over another also has practical consequences, due to the importance of smallholder agriculture for
rural livelihoods in low- and middle-income countries. Agriculture is a source of both income and
employment, and as such understanding how weather affects agricultural productivity on small-
holder farms is of policy interest. Thus, these results are important for informing policies to de-
velop and advise on improved agricultural technologies that can mitigate the risks posed by climate
variability and extreme weather events, and to provide social protection measures to smallholder
farmers that are exposed to climate shocks.


2    Measuring the Truth: Gauge Station Data and Earth Observa-
     tion Data in Africa
The goal of weather data products is to measure and report on the objective fact that is the volume
of precipitation and the temperature in degrees in a given location at a given time. In theory, this
goal is easily accomplished using technology that has existed for centuries. One simply needs a
rain gauge to collect precipitation, a mercury thermometer to measure temperature, and someone
to record this data with pencil on paper. What could be simpler than measuring the weather?
Humans have been reporting on the weather since the invention of writing and have been using
scientific instruments to record precipitation and temperature since the 1600s (Lundstad et al.,


                                                   4
2023).
       The problem with measuring the truth of the weather is not the technology but the personnel.
It is insufficient to simply have the instruments to capture the data. One requires a person to record
and report the data. In our setting, where we use LSMS-ISA data that provides tens of thousands
of observations across 13 years and six countries, one would need to provide each household with
the appropriate instruments and sufficient compensation to record the daily record of the weather.
Additionally, one would need to employ individuals to collect the information from around the
country and do quality checks on the data. The cost of a rain gauge and a thermometer might
be trivial. The cost of converting that data into a geospatially explicit time series spanning across
decades and countries is where measuring the truth becomes a problem.
       Given the prohibitively high costs of recording and reporting the weather for every household
in a nationally representative survey, two alternatives exist. The first alternative is to rely on gauge
data from nationally run meteorological stations. This is considered the gold standard of weather
data, as the quality control process in capturing, recording, and reporting the weather should be
pristine. However, not everyone lives near a weather station. And in low-income countries, weather
stations can be sparsely distributed (Tzachor et al., 2023). Figure 3 maps the locations of LSMS-ISA
households with locations of Global Historical Climatology Network (GHCN) weather stations.1
These maps show the limited density of these stations in Africa. Consider the case of Uganda,
in which there are no in-country stations currently reporting to the GHCN. This means that the
closest weather station for LSMS-ISA households in Uganda are one station in Kenya and one in
Tanzania. And so, the second alternative (and the one most widely employed in economic research
at this time) is to rely on remotely sensed EO data. Using EO data provides global-scale data on
the weather at various spatial and temporal resolutions, overcoming the sparseness of station data.
       Spatial datasets of weather variables, like precipitation and temperature, that are produced
using remotely sensed EO data, are not direct measurements of the variable of interest. Satellite
sensors provide spatially continuous observation of reflectance from the Earth’s surface in different
parts of the magnetic spectrum. These values are used to estimate related phenomena, such as
cloud presence, cloud top temperature or earth surface temperature. The continuous datasets are
then used in combination with directly observed, but often sparsely distributed, gauge data to
produce weather variables. Some inputs are common across products, but there are differences in
other inputs as well as modeling techniques.
       The type of analysis matters in assessing weather datasets for use in economic contexts. Is the
goal to understand climate trends; to capture characteristics of a particular agricultural season;
or to identify extreme weather events occurring in near real-time? This can help determine the
   1
    GHCN is a database which addresses the critical need for historical daily weather records. It is a composite of
climate records from numerous sources, merged together and assessed for quality. At its best, the data includes more
than 40 meteorological elements, including temperature daily maximum and minimum, temperature at observation
time, precipitation, snowfall, snow depth, evaporation, wind movement and maximums, soil temperature, and more.



                                                         5
relative importance of different dataset characteristics, such as spatial detail, temporal frequency
and length of record, with respect to the intended analysis.
    In an ideal setup, we would compare the objective truth of rainfall and data measured on the
ground, with remote sensing EO sources, and simply determine which EO product best matches
with the measured truth. However, as we allude to, it is not possible to measure the truth, through
direct measurement or through gauge station data. This leaves us with our EO weather data.
We discuss the particular EO sources which we use and their various interpolation methods in Sec-
tion 3.1. It is this second-best and its outcomes for the measurement and potential mismeasurement
of agricultural production and productivity that we explore in this paper.
    With this said, we acknowledge that EO data are not intended for field-level analyses - that
is, they are not explicitly designed to fix the problems identified. And so, we are pushing these
data to the limits of precision. While this is justified due to lack of availability of higher resolution
alternatives, we appreciate that spatial aggregation likely itself creates some messiness within the
data.


3    Data
To understand the possible sources of measurement error associated with integrating spatial, re-
motely sensed EO weather data with socioeconomic data, we combine publicly available EO weather
data products with publicly available unit-record survey data that have been generated as part of
the World Bank LSMS-ISA initiative and that are made available through the World Bank Micro-
data Library.
    The publicly available data contains obfuscated GPS coordinates to protect the privacy of sur-
vey participants. Specifically, spatial anonymization in the LSMS-ISA is achieved by only releasing
GPS coordinates for enumeration area (EA) clusters, not households (Blankespoor et al., 2021).
Further, following a method pioneered by the Demographic and Health Surveys (DHS), the EA
cluster coordinates are displaced. Urban clusters are displaced in any direction within a two kilo-
meters radius while rural clusters are displaced in any direction within a five kilometers radius.
Additionally, one percent of rural clusters are displaced up to ten kilometers.
    In trying to assess measurement error in EO weather data, the displacement of coordinates
coming from spatial anonymization obviously can have a confounding effect, since one would be
unable to disentangle mismeasurement in the EO product from mismeasurement introduced by the
mismatch between the true household location and the displaced EA cluster coordinates. However,
for this analysis, members of the authorship group were granted access to the true household
coordinates - not just the un-displaced EA cluster coordinates - and matched the EO data to
the true household location data. Thus, the temperature and precipitation reported by each EO
product is measured at the exact location of the household. While this gives our analysis an added
degree of accuracy in that we can precisely match household point location to EO grid cell, Michler

                                                   6
et al. (2022) show that, in similar analyses, using the obfuscated coordinates has no meaningful
impact on results.
      In the remainder of this section, we describe the sources of weather data and household data.
We then discuss descriptive statistics on the combined weather-household datasets.

3.1     Earth Observation Weather Data
In this study, the most important data are our EO weather data. To align with the use of these data
by economists, while also representing a wide range of the types of available data in this domain, we
use several public weather data sources that represent different modeling types, input sources, and
spatial resolutions. To ensure consistency and to ensure the creation of common metrics across the
analysis, we imposed two inclusion criteria. The source had to have (1) high temporal resolution
(daily), and (2) a minimum 30-year length of record at least up to 2021.2 Table 1 describes each
data source, including the length of record, spatial and temporal resolution, and the type of data
recorded.
      We classify EO weather data products by the method through which they generate precipita-
tion and temperature estimates. The first type of product we use merges gauge data with data
from meteorological satellites, which provide indirect information at full coverage. EO products of
this type include the African Rainfall Climatology version 2 (ARC2), the Climate Hazards group
InfraRed Precipitation with Station Data (CHIRPS), and Tropical Applications of Meteorology
using SATellite data and ground-based observations (TAMSAT) (Novella and Thiaw, 2013; Funk
et al., 2015; Tarnavsky et al., 2014).
      ARC2 is a merged gauge data and EO product that provides daily rainfall outputs for the
African continent. The dataset, produced by the National Oceanic and Atmospheric Administration
(NOAA) Climate Prediction Center (CPC) provides improvements over ARC1 and a longer length
of record compared to the rainfall estimate (RFE), the operational dataset of USAID’s Famine
Early Warning Systems Network (FEWSNET) program. Inputs are Global Telecommunications
System (GTS) rain gauge data over Africa, geostationary Meteosat infrared (IR) imagery, and polar-
orbiting microwave Special Sensor Microwave Image (SSM/I) and Advanced Microwave Sounding
Unit (AMSU-B).
      Validation efforts by Novella and Thiaw (2013) found that low reporting rates for some GTS
stations degrades model performance in those regions. Other findings are a general tendency to
underestimate rainfall, which is enhanced in areas of high relief or complex topography.3
      Like ARC2, the CHIRPS rainfall dataset builds on established techniques for merging gauge and
  2
     These criteria resulted in the exclusion of some EO products that have been frequently used by economists,
including the monthly Terrestrial Air Temperature and Precipitation from the Center for Climatic Research at the
University of Delaware.
   3
     Data and technical documentation are available for download from https://www.cpc.ncep.noaa.gov/products/
international/data.shtml.



                                                       7
EO data. Produced by the Climate Hazards Group at the University of California, Santa Barbara
this dataset is designed for monitoring of drought and environmental change at a global level. To
minimize latency, there are two products, a preliminary version with two day lag, and final output
available at three weeks. Outputs are available at time-steps from six hours to three months. As
inputs, CHIRPS makes use of a monthly climatology CHPclim, Tropical Rainfall Measuring Mission
Multi-satellite Precipitation Analysis version 7 (TMPA 3B42 v7) and global Thermal Infrared Cold
Cloud Duration (TIR CCD) from two NOAA archives. The EO data are then merged with gauge
data from five public archives, including the Global Historical Climatology Network (GHCN) and
GTS, several private sources, and meteorological agencies. While targeted gauge data collection
efforts resulted in a greater number of input stations for years prior to 2010, the number of stations
going forward is more limited, particularly in Sub-Saharan Africa. Detailed metadata by country is
available and may be a useful reference to determine if coverage for a region of interest is sufficient
for the analysis.
      Validation for select countries found that the climatology input CHPclim outperformed other
climatology datasets in data sparse regions and complex terrain (Funk et al., 2015). Further, in an
evaluation of wet season statistics CHIRPS showed less bias than other rainfall sources and good
correspondence with Global Precipitation Climatology Centre (GPCC) estimates.4
      The TAMSAT rainfall dataset is the highest spatial resolution gridded dataset used in this
analysis. Inputs are similar to other merged gauge and EO products: Meteosat TIR imagery,
purposefully collected archival (1983-2010) rain gauge data from meteorological agencies and other
sources and GTS gauge data. Rainfall estimation is based on cold cloud duration (CCD) inferred
from TIR and calibrated using gauge data within discrete calibration zones.
      Validation of TAMSAT found a mean underestimation of rainfall of approximately four mm
per dekad, though the bias was not always negative (Tarnavsky et al., 2014). Due to differences
in methodology from CHIRPS and ARC2 precipitation products, TAMSAT is not affected by
inconsistency in gauge data inputs. This makes it suitable for placing rainfall variability in the
context of a long-term climatology and thus detecting unusually wet or dry conditions.5
      The second type of product uses assimilation models to combine a large number of observations
from different sources (e.g., satellites, weather stations, ships, aircraft) to produce a model of the
global climate system or a particular atmospheric phenomenon. Outputs are inferred or predicted
based on the system state and understanding of interactions between model variables. EO prod-
ucts of this type include the European Centre for Medium-Range Weather Forecasts ERA5 and
the NASA Modern-Era Retrospective analysis for Research and Applications (MERRA-2) (Hen-
nermann and Berrisford, 2020; Bosilovich et al., 2016).
      ERA5, based on the European forecasting model ECMWF, is one of two assimilation model
  4
     Data and technical documentation are available for download from https://data.chc.ucsb.edu/products/
CHIRPS-2.0/.
   5
     Data and technical documentation are available for download from http://www.tamsat.org.uk/data/.



                                                   8
datasets used in this paper. The inputs are far too numerous to mention but include a range of
satellite inputs as well as gauge datasets. There are a wide range of outputs as well, including
2-meter air temperature and rainfall, available at sub-daily intervals and differentiated vertically.
ERA5 is coarser spatial resolution than the global and regional merged rainfall datasets, but more
detailed than MERRA-2.
      The sheer number and complexity of outputs can be a deterrent to the use of weather variables
from assimilation models. Uncertainty or lack of understanding about inaccuracies associated with
individual output variables of assimilation models, compared to other types of models, is another
reason to carefully consider their suitability for particular research (Parker, 2016). Nevertheless,
reanalysis datasets are used in a broad range of applications and even outperform other gridded
climate datasets in some settings (Zandler et al., 2020).6
      The second reanalysis dataset used in this analysis is MERRA-2, a product of NASA’s Goddard
Earth Observing System, version 5 (GEOS-5) assimilation model. Specifically we make use of the
variables T2MMEAN from the statD daily statistics collection, and PRECTOTLAND from the
Land Surface Diagnostics collection.7
      Last, we consider an EO data product produced primarily from gauge data, using only spa-
tial interpolation techniques to produce a continuous surface from observed measurements. The
NOAA Climate Prediction Center (CPC) Unified Gauge-Based Analysis of Daily Precipitation
and Temperature datasets were created using all information sources available at CPC and under-
goes extensive pre-processing and cleaning, including comparison with contemporaneous data from
satellite and other sources (Chen et al., 2008).
      NOAA’s Climate Prediction Center (CPC) Unified Gauge-based (CPC-U) datasets for daily
temperature and precipitation do not incorporate EO data in the estimation of weather variables.
Instead, an optimal interpolation (OI) technique is used on gauge data for precipitation, and
Shepard’s algorithm for temperature. CPC-U provides systematic global datasets for validation
and climate monitoring. GTS is a primary input data source, with some national collections, but
density is most sparse over Africa.
      As to be expected, even though the OI interpolation performs better than other techniques,
a cross-validation exercise shows performance to degrade significantly with increasing distance to
nearest station (Chen et al., 2008). As a result, this dataset may not be suitable for analysis in
some parts of Africa, with high spatial variation and low density of stations.8
  6
     Data and technical documentation are available for download from https://cds.climate.copernicus.eu.
  7
     Data and technical documentation are available for download from https://disc.gsfc.nasa.gov/.
   8
     Data and technical documentation are available for download from https://psl.noaa.gov/data/gridded/data.
cpc.globalprecip.html for precipitation and https://psl.noaa.gov/data/gridded/data.cpc.globaltemp.html
for temperature.




                                                     9
3.2   Household Survey Data
The World Bank Living Standards Measurement Study - Integrated Surveys on Agriculture (LSMS-
ISA) is a household survey program that provides financial and technical assistance to national
statistical offices in Sub-Saharan Africa for the design and implementation of national, multi-topic
longitudinal household surveys with a focus on agriculture. Our analysis uses data from multiple
rounds of panel household surveys from Ethiopia, Malawi, Niger, Nigeria, Uganda and Tanzania.
Table 2 provides a summary of the countries, years, and observations used in the analysis.
   In Ethiopia, we use data from the 2011/12, 2013/14, 2015/16, 2018/19, and 2021/22 rounds
of the Ethiopia Socioeconomic Survey (ESS), which has been conducted by the Central Statistical
Agency of Ethiopia (CSA, 2014; CSA, 2015; CSA, 2017; CSA, 2020; CSA, 2024). The Wave 1 data
is representative at the regional level for the most populous regions in the country while Waves 2
and 3 expanded to include 1,500 households in urban areas. The 2018/19 round represented a total
refresh of the panel, making it nationally representative as well as representative of each of the 11
regions in Ethiopia. The 2020/21 round followed up with the households from 2018/19, creating
a new panel distinct from the three-round panel that ran from 2011 to 2016. Because of COVID
and civil conflict in Ethiopia during the 2021/22 data collection effort, this fifth wave contains no
households from Tigray and is thus not nationally representative, though it remains representative
of ten of the regions. After data cleaning to remove urban and non-agricultural rural households,
we are left with 10,674 observations from 5,333 distinct household across five survey waves.
   In Malawi, the LSMS-ISA data includes two separate surveys that form three different data sets:
the cross-sectional Integrated Household Survey (IHS), and the longitudinal Integrated Household
Panel Survey (IHPS) (NSO, 2012; NSO, 2015; NSO, 2017; NSO, 2020a; NSO, 2020b). These two
surveys form three data sets: three cross sections (2010/11, 2016/17, 2019/20), two waves of panel
data known as the short-term panel (2010 and 2013), and five waves of panel data known as the
long-panel (2010, 2013, 2016, and 2019). This analysis relies on the short-term and long-term data
sets from the IHPS, which are representative at the national-, urban/rural-, and regional-level.
The long-term panel tracks households that move or split off, meaning the sample size grows from
round-to-round. After data cleaning to remove tracked and non-agricultural households, we are left
with 8,897 observations from 3,833 distinct households across four survey years.
   In Niger, we use two waves, the first from 2011 and the second from 2014 (NIS, 2014; NIS, 2016).
Unlike the other countries from the LSMS-ISA used in this paper, there has been no follow-up on
data collection in Niger since 2014. The sample is representative at the national and urban/rural-
level. Data cleaning and removal of non-agricultural households gives us 3,913 observations from
2,320 distinct households across two survey waves.
   In Nigeria, we use the data from the 2010/11, 2012/13, 2015/16, and 2018/19 rounds of the
General Household Survey - Panel, which is representative at the national and urban/rural-level
(NBS, 2012; NBS, 2014; NBS, 2016; NBS, 2019). Similar to Ethiopia, there was a panel refresh in


                                                 10
2018/19 that started a new panel. Unlike in Ethiopia, in Nigeria the refresh was partial, with the
survey continuing to track 1,425 of the original households. Because there is not yet a second round
of data for the refreshed households, we exclude them from the analysis Data cleaning and removal
of non-agricultural households yields 9,145 observations from 3,412 distinct households across four
survey waves.
   In Tanzania, the data come from the 2008/09, 2010/11, 2012/13, 2014/15, 2019/20, and 2020/21
rounds of the Tanzania National Panel Survey (TZNPS) (TNBS, 2011; TNBS, 2012; TNBS, 2015;
TNBS, 2017; TNBS, 2021; TNBS, 2023). The sample is representative for the nation, and provides
estimates of key socioeconomic variables for mainland rural areas, Dar es Salaam, other mainland
urban areas, and Zanzibar. As in Nigeria, the fourth wave (2014/15) was a partial refresh, with
data being collected from a sub-sample of the original panel households (known as the extended
panel) plus the addition of completely new households (known as the refresh panel). The extended
panel households were re-interviewed in 2019/20 in what is known as the Sex-Disaggregated Data,
which added improved individual-level data. The refresh panel was followed up with in 2020/21.
The 2020/21 survey round also included a booster sample, that will continue to be followed in
subsequent rounds. However, since at this time there is only one round of observations for the
booster sample, we exclude them from the analysis. Focusing on rural, crop producing households
we have 9,916 observations from 4,804 distinct households across five survey waves.
   In Uganda, we use the data from the 2009/10, 2010/11, 2011/12, 2013/14, 2015/16, and 2019/20
rounds of the Uganda National Panel Survey (UNPS) (UBOS, 2012; UBOS, 2014a; UBOS, 2014b;
UBOS, 2016; UBOS, 2019; UBOS, 2021). The original households were followed for three waves and
then starting in 2013/14, a subset of households were cycled out and replaced with new households.
This continuous, partial refresh process means that households remain in the sample for different
lengths of time, though by 2019/2020 all of the original households have been replaced. The original
sample, as well as the refresh, was designed to be representative at the national-, urban/rural- and
regional-level. Focusing on rural, crop producing households we have 11,692 observations from
4,003 distinct households across six survey waves.
   For the analysis, we combine data from the six countries and all waves to generate a single
cross-country panel dataset which includes 54,237 observations from 23,705 distinct households.
For estimation, we include two measures of agricultural productivity: yield (kg/ha) of the primary
cereal crop and the value (2015 USD/ha) of all seasonal crop productivity on the farm. Summary
statistics for these outputs as well as measured input are in Table A3 in the Appendix.

3.3   Descriptive Statistics
Our pre-analysis plan specifies that we will examine 22 different ways to measure precipitation and
temperature. These variables and definitions are presented in Table A1 in the Appendix Section 6.
For parsimony, we focus all of our analysis in the paper on only four of these 22 variables: (1) total


                                                 11
seasonal rainfall, (2) number of days without rain, (3) mean seasonal temperature, and (4) growing
degree days (GDD). These four variables are indicative of a number of different ways to measure
precipitation and temperature (volume/degree v. count). We spend what may be more space than
is typical in an economics paper discussing the descriptive statistics around these measures, but
this is essential for understanding the way in which each EO product differs in its measurement
of rainfall and temperature. In summarizing the descriptive statistics, we highlight a trends or
stylized facts about differences in measurement (a) between EO products within a country and (b)
within an EO product between countries.
       Figure 4 presents the distribution of total rainfall (measured in mm) during the growing season,
by country and EO product.9 Not surprisingly, there are substantial differences in the distribution
of rainfall across countries. Ethiopia and Uganda average around 650mm of rainfall a season while
Malawi, Nigeria, and Tanzania average around 900mm of rainfall a season. By contrast, Niger
averages only around 350mm of rainfall in a season. These level differences in the quantity of
precipitation in a country are clearly due to geographic characteristics. When we consider all
countries together, some general trends emerge with respect to EO sources. A first trend (T1) is
that ERA5 tends to have long tails, indicating this reanalysis data product reports dramatically
more rainfall than other EO source.10 For example, in Ethiopia and Tanzania, ERA5 reports some
households receiving 6,000-7,000mm of rainfall in a season - six to seven meters - while no other
EO product reports even half that amount. A second trend (T2) is that CPC tends to measure less
rainfall relative to other EO products. For example, in Ethiopia and Malawi, CPC reports only
half as much rainfall as the other products.
       However, there are important exceptions to these trends. In Ethiopia, Malawi, and Uganda,
both T1 and T2 hold but in Niger, neither holds. In fact, nearly the opposite is seen: ERA5
measures lower rainfall, with a distribution below those of all other EO products. CPC’s distribution
is nearly identical to the distribution of the non-ERA5 products. Next, in Nigeria, T2 holds (CPC
reports less rainfall than the other products) but T1 does not hold. The largest volume of rainfall
reported by ERA5 and is almost the exact same amount as reported by CHIRPS, MERRA-2,
and TAMSAT. Finally, in Tanzania, T1 holds (ERA5 reports values more than double any other
product) but T2 does not. Here CHIRPS and TAMSAT report less rainfall than the other products
and both report zero rainfall for some households, something that does not happen for any other
products in any country but Niger. The differences in rainfall reported by each EO product are
likely due to a combination of the country’s geography (topography, hydrology) and the method
each product uses to convert EO signals to estimates of the volume of precipitation.
       Figure 5 further explores these differences by estimating the mean number of days without rain
   9
    We discuss our calculation of growing season in the Appendix Section 6.
  10
    These long tails are shown by Hennermann and Berrisford (2020). A few times per year, the rainfall can be
extremely large in small areas, in an event known as a “rain bomb”. This can create the long tails we observe on the
high rainfall end.



                                                        12
reported by each EO product for each country. Mean estimates are generated using a fractional-
polynomial and graphs include 95% confidence intervals on the mean estimates. We again make
note of several trends. One trend (T3) is that ARC2, CHIRPS, and CPC form a cluster, measuring
more days without rain, relative to the other EO sources. Another trend (T4) is that the two
reanalysis data sets, ERA5 and MERRA-2 form a different cluster, measuring fewer days without
rain, relative to other EO sources. A final trend (T5) is that TAMSAT reports substantially fewer
days without rain than the first group but substantially more than the second, ending up somewhere
in the middle.
   Again, there are exceptions to these trends. In Ethiopia, Nigeria, and Tanzania, T3, T4, and T5
hold. While MERRA-2 reports the fewest days without rain in all four countries, the EO product
that reports the most days without rain varies from country to country. Additionally, the size
of the gap between the products reporting drier conditions and TAMSAT varies, with TAMSAT
reporting nearly as little rain in Tanzania as the others while it reports nearly half the number
of days without rain relative to the other products in Ethiopia and Nigeria. Turning to Malawi,
T4 and T5 (TAMSAT in the middle) hold, though T3 does not. Instead, CPC measures about 20
percent more days without rain than ARC2 or CHIRPS. Next, looking at Niger, T3 and T5 hold.
Unlike in other countries, in Niger ERA5 does not agree with MERRA-2 but rather reports very
few days without rain, similar to ARC2, CHIRPS, and CPC. Finally, in Uganda, only T4 (ERA5
and MERRA-2 cluster) holds. TAMSAT and CPC report the fewest days without rain while ARC2
and CHIRPS fall in between this high reporting cluster and the low reporting re-analysis cluster.
   In Figure 6 we present the distribution of mean daily seasonal temperature (measured in ◦ C),
by country and EO product. Compared to the distribution of mean daily rainfall, the figures show
much tighter distributions around mean temperatures. All EO sources in all countries produce
essentially the same results for temperatures. This results is our first temperature trend (T6): there
is relatively low variation across EO sources within a country. This is largely due to the relatively
low temperature variation in these tropical and sub-tropical countries. While there is some variation
country-to-country in terms of which EO product reports the hottest average temperature, these
differences are not agronomically meaningful.
   Figure 7 estimates the mean GDDs in a year using a fractional-polynomial and includes 95%
confidence intervals on the mean estimates. As with number of days without rain, GDD represents
a relative coarsening of the data by converting measured temperature into a count variable for
the number of days in which temperature fell within a given range. Compared to mean daily
temperature, there is less overlap in the values each EO product produces, though relative to the
rainfall measures there is again substantially less variation across temperature products. This gives
us our final trend (T7): there is relatively low variation in GDD across EO sources.
   Examining each country individually, we see T7 most clearly in Ethiopia and Uganda where
there is overlap in the confidence intervals for all three products across almost all years. Nigeria also



                                                  13
adheres to T7 in that several products overlap and all have downward trajectories in the number
of GDD per year. By contrast, Malawi, Niger, and Tanzania all deviate from T7, though each
deviates in its own way. In Malawi, there is effectively no overlapping in the confidence intervals
of the three products and while CPC and ERA5 have flat slopes, MERRA-2 produces a strongly
divergent downward trajectory. Niger is similar to Malawi in that there is almost no overlap in
confidence intervals and MERRA-2 has a strong downward trajectory. However, in Niger, CPC
and ERA5 have strong upward trajectories. Further, in Malawi MERRA-2 produces the fewest
GDDs while in Niger it produces the most. Finally, in Tanzania, there is almost perfect agreement
between ERA5 and MERRA-2, both of which show no change in GDD over time. By contrast,
CPC is in almost complete disagreement, producing fewer and fewer GDDs over time.
    Summarizing the descriptive evidence: there can be substantial differences in the weather re-
ported by each EO product. Again, this is not surprising since each product uses different data
and different methods to produce gridded weather data. If all that these descriptive figures showed
was one EO product consistently reporting higher levels of rainfall and temperature than other
products, researchers could easily adjust their methods to account for this over reporting. How-
ever, Figures 4 through 7 document not just differences in cardinality (one product reports a few
more mm of rain) but in ordinality. In most countries ERA5 reports the most rain but in Niger it
reports the least. In most countries TAMSAT falls in the middle in terms of how many days are
without rain, but in Uganda it reports the driest conditions. And while compared to rainfall, there
is much more agreement among the EO temperature products, there are still significant deviations,
particularly in the number of GDD produced by the products. In Malawi, MERRA-2 produces the
fewest GDD but in Niger it produces the most. In most countries, CPC is in agreement with ERA5
but in Tanzania CPC disagrees with both ERA5 and MERRA-2. Based on these changes in the
ordering of EO products across countries, we conclude that researchers’ selection of EO product
cannot be country-agnostic.


4    Analysis Plan
A common feature of research on measurement error is that the researchers have a measure of
objective truth. They can trace the boundaries of a plot of land (Abay et al., 2019), they can weigh
the quantity of harvest (Lobell et al., 2020), they can determine the exact location of a household
(Michler et al., 2022). In our setting, we lack an objective measure of the truth regarding what the
weather was like for a given household in a given year. And, given the paucity of gauge station data
on the Continent, the typical gold standard for weather data does not exist. This lack of an objective
measure for quantifying mismeasure in EO weather data informs how we implemented our research
design. First, we developed a pre-analysis plan and registered it at Open Science Framework
(Michler et al., 2019). While pre-analysis plans have become common in experimental economics,
they are still relatively uncommon for binding researchers’ hands when using observational data

                                                 14
(Janzen and Michler, 2021). The use of a pre-analysis plan allowed us to pre-define the sources
of data for inclusion in the study, what metrics would be tested using what functional forms, and
how we would compare results across models in the absence of formal statistical tests. Second, we
adopted a blinding strategy to help ensure objectivity in the implementation of the pre-analysis
plan. As such, the authors were divided into two groups: the Data Generating Group and the
Data Analysis Group. Authors Kilic and Murray were in the Data Generating Group and had full
responsibility for extracting the EO data and matching it to the household records in the household
survey data to create a number of different paired EO-survey datasets. In these datasets, the
source of the EO data was anonymized prior to sharing with the Data Analysis Group. Authors
Josephson and Michler made up the Data Analysis Group and had full responsibility for cleaning
the agricultural productivity data, running the regressions, and conducting and writing the analysis.
The pre-specified analysis was carried out on the blinded datasets and these results were posted
to arXiv.org prior to unblinding (Michler et al., 2021a). The generation of datasets in this manner
preserves the objectivity of any findings regarding differences in outcomes between different EO
products.
   The following analysis, and the associated results, were all pre-specified. If methods, approaches,
or inference criteria differ from our plan, we highlight these differences. Results arising from these
deviations in our plan should be interpreted as exploratory.

4.1   Estimation
                                  ene and Greenstone (2007):
We follow the basic model of Deschˆ

                                                 J
                               Yht = αh + γt +        βj fj (Wjht ) + uht                        (1)
                                                 j

where Yht is our outcome variables from the LSMS-ISA-supported household surveys, described
above, for household h in year t, log transformed using the inverse hyperbolic sine. We control for
year fixed-effects (γt ) and include household fixed-effects (αh ) in some specifications. The function
fj (Wjht ) represents our weather variables of interest where j represents a particular measurement
of weather. Finally, uht is an idiosyncratic error term clustered at the household-level.




                                                 15
       From this general set-up, we estimate three versions of the model:11



                                     Yht = α + β1 Wht + uht                                                       (2a)
                                     Yht = αh + γt + β1 Wht + uht                                                (2b)
                                     Yht = αh + γt + Xht π + β1 Wht + uht                                         (2c)

Model (2a) examines the simple correlation between a weather metric (Wht ) and crop productivity
(Yht ), measured as either yield of the main crop or the total value of on-farm production per hectare.
This model, which lacks any control variables, implicitly treats the data as if it was a pooled cross
section. We refer to this model as the “weather only” model in our discussion. Model (2b) accounts
for the fact that we have multiple observations from a household over time by including household
fixed effects (αh ). This model also includes year fixed effects (γt ) that control for country-wide
changes, such as civil conflict, the pandemic, or changes in government, that vary over time. We
refer to this model as the “weather with fixed effects” model. Model (2c) adds measured inputs
as covariates (Xht ). These include quantity of fertilizer per hectare, the amount of labor days per
hectare, and indicators for if the household applied pesticide, herbicide, or irrigation. We refer to
this model as the “weather with fixed effects and inputs” model.
       All of the regression models are estimated for each permutation of the data. This is a substantial
number of regressions, given the number of models (three), countries (six), variables defined (14
rainfall, eight temperature variables), EO products (six rainfall, three temperature), and the number
of outcomes (two). This gives us a total of 3,888 regressions: for each of our three models and two
outcomes on the 648 different versions of the data. By varying specifications and data we define a
robust set of outcomes regarding mismeasurement in EO data absent objective measure of rainfall
and temperature.

4.2      Inference
In the majority of economics research, empirical results are presented in tables, with coefficient
estimates and statistics for inference. In our case, because of the large number of regressions
estimated, standard modes of inference and traditional presentations of results are not appropriate.
Instead, we rely on a series of heuristic methods and criteria to make inference, evaluate the results,
and present our findings.
       In discussing our method for inference and interpretation of results, it is useful to reflect on
the characteristics necessary for our purposes. First, a heuristic should be agnostic about the sign
  11
    In our pre-analysis plan we defined three additional models that include weather and its square to allow for
nonlinearities in the way weather is related to crop productivity. Our results do not change when we use the
quadratic specifications, so in this paper we focus solely on the linear version of the model. All results are available
in the populated pre-analysis plan through the World Bank (Michler et al., 2021b).



                                                          16
of a coefficient, as some weather metrics that we test are likely to be positively correlated with
outcomes (e.g., mean rainfall) while others are likely to be negatively correlated (e.g., days without
rain). While we are agnostic about sign generally, we are interested in the ordering of coefficients
and so, do not completely disregard this dimension in our heuristics. Second, a heuristic should
be able to determine whether or not a weather metric is significantly correlated with outcomes.
Our prior is that weather is significantly correlated with outcomes, regardless of direction. This
maintained assumption is based on the frequency with which weather is used in the economics
literature to predict all sorts of outcomes, from crop production to migration to economic growth.
Thus, identification of statistical significance is an important dimension to be included in our
heuristics. Finally, a heuristic should be able to measure the amount of unexplained variance in
the model after controlling for weather. This is also in line with our prior, as we expect weather to
reduce the amount of unexplained variance in a model, all else being equal.
    The full set of heuristics developed in our pre-analysis plan and corresponding results are pre-
sented in Michler et al. (2021b).12 Because of our interest in the order of coefficients, the most
relevant heuristics for this analysis is the first, related to directionality and ordering. To this end,
we use specification charts to examine the actual regression coefficients and estimated confidence
intervals for a subset of regressions.
    An important caveat for consideration with respect to our results: the significance of a point
estimate does not necessarily imply that the model is correctly specified, that the point estimate
is agronomically meaningful, or, in particular, that the point estimate has the correct sign. These
results and the associated figures simply allow us to visualize the variability in the number of
significant coefficients, their magnitude, and ordering across specifications. Variability within these
domains may suggest that measurement error exists.


5     Results
As with the descriptive statistics, we focus our results on four variables, which are representative of
the overall trends across variables. They are (1) total seasonal rainfall, (2) days without rain, (3)
mean seasonal temperature, and (4) growing degree days. These four variables present a series of
different ways to measure precipitation and temperature (e.g., volume/degree v. count) and allow
us to focus our discussion on the differences in measurement (a) between EO products within a
country and (b) within an EO product between countries.
  12
     The three metrics are (1) mean log likelihood values, (2) share of coefficient p-values significant at standard levels
(0.01, 0.05, and 0.10), and (3) coefficient size with 95% confidence intervals.




                                                          17
5.1   Coefficient Size
Following our described heuristics, we present our results in specification charts. We include visual
representations of both point estimates and 95% confidence intervals for single regressions, with 36
regressions per rainfall metric per country and 18 regressions per temperature metric per country.
The markers in the space below the coefficients indicate which specific version of a regression
the coefficient comes from. The gray circles at the bottom indicate the econometric specification
(weather only, weather with FE, and weather with fixed effects and inputs). The gray squares
indicate the dependent variable (either total farm value or the quantity of maize harvest). The
diamonds indicate which EO product the weather data comes from as well as if the sign and
significance of the coefficient. Gray diamonds indicate not significant, blue diamonds indicate
positive and significant, and red diamonds indicate negative and significant. Each chart is divided
into three regions which correspond to the econometric specification that the results come from.
This is because we expect the magnitude of the coefficients to differ depending on the inclusion or
exclusion of controls. Within specification, results are sorted by coefficient size.
   In understanding and interpreting the specification charts, one wants to look for patterns in
the data. If differences in the values of rainfall and temperature that different products reported
were simply affine transformations of each other, then we should see a consistent pattern in the
location of the diamonds (though not in their color). Said another way, if all that the differences
in measurement produced was a change in cardinality, then the size and significance of coefficients
would change (their color) but not the ordering of coefficients (the location of the diamonds). If
instead the differences in EO products are such that it leads to changes in ordinality, then we should
see no consistent pattern in the location of the diamonds. At one extreme, in which results are
wholly robust to the choice of EO product, then the pattern in the location of the diamonds should
repeat in each section of the chart and across charts. At the other extreme, in which results lack
any robustness to the choice of EO product, then the diamonds should look like a cloud of noise.
   Figures 8 and 9 present specification charts for mean daily rainfall and number of days without
rain, by country. It is immediately obvious that the location of the diamonds looks more like a cloud
of noise than a repeated pattern both across specifications within a country and across countries.
Consider, as an example, Ethiopia in Figure 8: in the specification in which only weather variables
are included, all coefficients are positive and significant with ARC2 producing small but positive
and significant results and MERRA-2 reporting large, positive and significant results. But, when
fixed effects are included (with and without inputs) the pattern changes. It is not just that all
coefficients get larger or smaller. Rather the ordering changes. ARC2 now produces the largest
coefficients, and the only that remain positive and significant, while MERRA-2 produces some of
the smallest, which tend to be negative and significant. Similar re-orderings occur in each country
and for days without rain, though not all are as stark as mean daily rainfall in Ethiopia.
   Next, we turn to temperature in Figures 10 and 11. As with rainfall, results differ by EO


                                                 18
product. And again, as with rainfall, coefficient ordering shifts across models. This is more evident
with the smaller number of EO sources used in the temperature regressions. Most frequently,
countries follow a trend of one set of ordering for the weather only model specification, which shifts
to a different ordering for the models which include fixed effects. Consider, as an example, Niger
in Figure 10: in the first panel ERA5 produces large negative and significant coefficients while
MERRA-2 produces small coefficients that are either negative and significant or not significant. In
the models with FE, MERRA-2 produces the largest negative coefficient while ERA5 produces the
largest coefficient, now positive and significant. While we see that similar reorderings appear in
other countries for other temperature metrics, the most striking element of the temperature results
are that we see strong patterns in the specification charts for a couple countries, specifically Nigeria
and Uganda. As we saw in the descriptive statistics, there is less variation in the temperatures that
the different EO products report relative to the temperature products and thus less changes to the
ordinality of coefficients.
   Before moving to a closer examination of changes in ordinality in regression results, two other
regularities in the specification charts bear commenting upon. First, for both rainfall and tempera-
ture, regardless of the EO product or the weather metric, weather becomes much less important in
predicting agricultural productivity once household FEs are included. Over and over again, regress-
ing output on just rainfall or temperature shows a significant relationship between the two. But
once household FEs are included to control for time-invariant unobservables, the relationship goes
away. Weather is a time varying phenomenon, often considered exogenous or as good as randomly
determined. It is not immediately obvious why the inclusion of a time-invariant control destroys
the relationship between weather and agricultural production. A likely reason, which we leave
unexplored, is that weather in a given location does not vary that much over time. Thus, including
household FEs, assuming a household does not move, is essentially a location FE. Once location is
controlled for, weather varies very little year-to-year and small deviations from the time average in
a location has little to no impact on agricultural production in the year. This raises questions about
the use of weather as an instrument (as it is correlated with time-invariant household heterogeneity,
specifically location) as well as the ability of economists to use weather to predict the effects of
climate change (by definition, a slow process) on economic indicators.
   A second regularity is that the sign on the weather coefficients can vary substantially. This is
related to, but distinct from the ordinality issue. Within a single econometric specifications, rainfall
from one EO product will be positively correlated with agricultural production while in rainfall from
another product will be negatively correlated. Shifts in significance across specification can be the
result of changes in relative size (cardinality). But what should raise concerns among applied
economists is that in every country a researcher could produce whatever relationship they wanted
between weather and the dependent variable by judiciously choosing the EO product, the weather
metric, and the econometric specification. Essentially, a researcher using EO weather data has



                                                  19
almost unlimited degrees of freedom to p-hack their way to his or her desired results.

5.2   Coefficient Ordering
While the specification charts summarize a large amount of information in a compact form, it can
be difficult to detect the extent to which the diamonds indicating EO product form a pattern or are
patternless. To dig further into coefficient ordering, we present a series of figures which represent
the order (ranking) of coefficients within a country, across models, by EO source using bumpline
functions (Naqvi, 2024). These are presented for total seasonal rainfall in Figure 12 and mean daily
temperature in Figure 13. Results for days without rain and GDD are functionally similar and
so we present figures for those variables in the Appendix. Within each country panel, the x-axis
(each column) presents a different outcome variable (value or quantity) and a different econometric
specification (weather only, weather with FE, and weather with fixed effects and inputs). The
y-axis (each row) represents the ranking of the coefficient size for each EO source within each
regression. They are ordered from one to six, with one representing the largest coefficient and six
representing the smallest coefficient. These results give greater insight to the changes in ordinality
across specifications and across countries.
   We first consider the findings from total seasonal rainfall (see Figure 12). Going country by
country: in Ethiopia, we see that in the weather only specification, CPC and TAMSAT have large
coefficients while ERA5 has the smallest. This ordering changes once FE are included, with ERA5
producing the largest coefficients and TAMSAT and CPC producing the smallest. In Malawi, the
same trend does not hold. Rather, there appear to be trends based on the dependent variable,
with ARC2 and TAMSAT producing large coefficients when the value of total farm production
is the dependent variable and small coefficients when the dependent variable is quantity of maize
yield. Thus, what drives heterogeneity in results in Ethiopia differs in Malawi. Results in Niger
appear almost chaotic, with order of EO products jumping around with any change in specification
of dependent variable. It is difficult to identify any trends other than a lack of trends. Next, we
consider Nigeria, which does not demonstrate a specification- or dependent variable-based pattern,
but generally has ARC2, CPC, and ERA5 ordered in the last three places with MERRA-2, CHIRPS,
and TAMSAT in the top three places. Compared to the previous countries, the ordering of EO
products in Nigeria appears robust, suggesting one would get similar results regardless of which EO
product one chose. This can be confirmed by referencing the specification charts in Figure 8. A
similar result holds in Tanzania, where we observe that ARC2 generally falls into the fourth position
and TAMSAT into the sixth. Finally, we consider Uganda which somewhat resembles Ethiopia. In
the weather only specifications MERRA-2 and ERA5 produce the largest coefficients while ARC2
and TAMSAT produce the smallest. This order reverses when FEs are included. Overall, based
on the findings for total seasonal rainfall, we can draw some general trends about ordering or re-
ordering of coefficients based on EO products within countries, but we cannot draw trends across


                                                 20
countries. Which EO products produce the largest or smallest coefficients varies from country to
country suggesting the how well an EO product captures the truth about weather depends on the
geography and climate of where it is looking.
   Moving to mean daily temperature (see Figure 13), the figures are presented with the same
column and row structure, but instead of having six ranking places, there are only three. This
somewhat simplifies the view. Beginning with Ethiopia, in the weather only specification, ERA5
produces the largest coefficients while CPC and MERRA-2 which places for the smallest. When
we include FEs, CPC and MERRA continue to switch places based on the dependent variable but
ERA5 now always produces the smallest coefficients. Next, in Malawi, we see a clear and consistent
ordering of outcomes: MERRA-2 always produces the largest coefficients while CPC and ERA5
switch regarding which one produces the smallest based on the dependant variable. Like with
Nigeria and rainfall, one get similar results in Malawi regardless of what EO product was selected.
Turning to Niger, MERRA-2 produces largest coefficients in the weather only specifications while
ERA5 produces the smallest. This order changes once FEs are added to the specification. We
next consider Nigeria, which does not present a consistent pattern in the weather only specification
but produces a consistent ordering of EO products once FEs are added. For Tanzania, a similar
but less consistent pattern exists. There is very little stability in ordering of coefficient size in
the weather only specification but a fair amount of consistency in the specifications that include
FEs, with MERRA-2 always producing the smallest coefficients. Finally, in Uganda, there exists
a pattern very similar to Malawi. CPC always produces the largest coefficients while MERRA-2
and ERA5 swap places for the smallest coefficient. Overall, while there is must fewer changes in
ordinality for EO temperature data compared to rainfall data within a country, there again is little
consistency across country. In Ethiopia and Uganda, CPC tends to produce the largest coefficients
while in Malawi and Nigeria it is MERRA-2 and in Niger and Tanzania it is ERA5. But ERA5
produces the smallest coefficients in Ethiopia, Malawi, and Uganda, while MERRA-2 produces the
smallest coefficients Niger and Tanzania. The EO product that produces the largest or smallest
coefficients varies from country-to-country and again suggests that the accuracy of these products
is a function of geography and climate. Not all products perform equally well across space. This
means what results a researcher gets will depend on their choice of EO product as well as their
country of study.
   Taken together, the preponderance of evidence from our regressions leads us to conclude that
our results are not robust to the choice of remote sensing EO weather product. The correlation
between rainfall or temperature and agricultural productivity varies not only across countries but
within a country based on the econometric specification or how productivity is quantified. Our
findings show that different measurement methods from different EO sources result in different
ordering of coefficients and not just a change in the relative size of the coefficients.




                                                21
6    Conclusions and Recommendations for Researchers
The use of remotely sensed EO weather data in economic contexts has become commonplace.
Researchers use a variety of EO products to produce a host of different weather metrics to predict
numerous different economic outcomes. If each EO product reported the same values for rainfall
and temperature for a given location for a given time on a given day, then the choice of EO product
would be immaterial. Unfortunately, this is not the case. Each product, while attempting to
measure the objective truth of the weather, produces its own “truth.” The findings produced are
not immutable truths.
    Using LSMS-ISA data from six African countries in combination with nine EO weather data
products, we demonstrate the lack of consistency in how EO weather products report on the
weather. Importantly, the differences across products do not simply affect the cardinality, or relative
size, of coefficients from different EO products. The shifts we see cannot be ascribed to an affine
transformation of one EO product by another. Rather, what we document are changes in ordinality,
the ranking of coefficients by size. In some countries one EO product may produce large, positive
and significant coefficients while another EO product produces negative and significant coefficients.
This order can swap if one moves to another country or if, within a give country, one changes the
econometric specification or how the agricultural productivity is quantified. The unwary researcher
may end up with results and draw conclusions that are a function of their choice of EO product
and could have ended up with completely different results and conclusions if they had chosen a
different EO product. Knowing this, the unscrupulous researcher can select the EO product that
measures the “truth” in the way that produces the results and conclusion they want.
    Given the lack of gold standard gauge station data in many countries, where does this disparity
in how EO products measure the weather leave researchers who want to incorporate this type of
data into economic contexts? As we cannot know the truth of what weather occurred where, we
are left in a second-best world where more work may be required to demonstrate the accuracy of
our scientific findings. Based on the results in this paper, we recommend that researchers carefully
choose which remotely sensed EO source to use in their analysis. More justification needs to be
given in papers for the choice of one EO data set over another. In EO product documentation there
is often a list of known issues, such as over-estimation of rainfall events or difficulty in capturing
precipitation in rugged terrain. In the same way aseconomists already document and justify choices
in terms of variable selection, we also need to document and justify our selection of EO product.
Additionally, researchers may need to demonstrate that key results are not a function of their
choice of EO product. Economists already verify the robustness of results to changes in data and
specification. Checks for robustness to alternative EO weather products may also be called for,
particularly if weather is a key component of the story or plays a role in causal identification.
    Potentially the most important recommendation resulting from our findings is the need to
recognize, as Stephen Jay Gould wrote, that science is a social phenomenon, not the collection and


                                                 22
production of pure information. And, the results of scientific inquiry should never be interpreted
as immutable facts. For years the economics profession has used remotely sensed EO data, weather
or otherwise, as if it perfectly reported on objective facts. We have done this ourselves, having
chosen one EO product over another simply because it was easy to download or work with. By
keeping Gould’s caution about science and scientific findings in the forefront of our minds, we reduce
the probability of blindly accepting the truth or accuracy of the data we use. As researchers, we
should cultivate skepticism, but we should also cultivate empathy for all of us engaged in this social
phenomenon called science.




                                                 23
References
Abate, G. T., A. de Brauw, K. Hirvonen, and A. Wolle (2023). Measuring consumption over
 the phone: Evidence from a survey experiment in urban ethiopia. Journal of Development
 Economics 161, 103026.
Abay, K. A. (2020). Measurement errors in agricultural data and their implications on marginal
 returns to modern agricultural inputs. Agricultural Economics 51 (3), 323–41.
Abay, K. A., G. T. Abate, C. B. Barrett, and T. Bernard (2019). Correlated non-classical mea-
 surement errors, ‘second best’ policy inference, and the inverse size-productivity relationship in
 agriculture. Journal of Development Economics 139, 171–84.
Abay, K. A., L. E. Bevis, and C. B. Barrett (2021). Measurement error mechanisms matter:
 Agricultural intensification with farmer misperceptions and misreporting. American Journal of
 Agricultural Economics 103 (2), 498–522.
Alix-Garcia, J. and D. L. Millimet (2023). Remotely incorrect? accounting for nonclassical mea-
  surement error in satellite data on deforestation. Journal of the Association of Environmental
  and Resrouce Economists 10 (5), 1335–67.
Arag´on, F. M., F. Oteiza, and J. Pablo Rud (2021). Climate change and agriculture: Subsistence
  farmers’ response to extreme heat. American Economic Journal: Economic Policy 13 (1), 1–35.
Barrios, S., L. Bertinelli, and E. Strobl (2010). Trends in rainfall and economic growth in Africa:
  A neglected cause of the African growth tragedy. Review of Economics and Statistics 92 (2),
  350–66.
Blankespoor, B., T. Croft, T. Dontamsetti, B. Mayala, and S. Murray (2021). Spatial anonymiza-
  tion: Guidance note prepared for the Inter-Secretariat working group on household sur-
  veys. UN Inter-secretariat Working Group on Household Surveys Task Force on Spatial
  Anonymization in Public-Use Household Survey Datasets. https://unstats.un.org/iswghs/task-
  forces/documents/Spatial Anonymization Report submit01272021 ISWGHS.pdf.
Bosilovich, M., R. Lucchesi, and M. Suarez (2016). MERRA-2: File specification. GMAO Office
  Note No. 9 (Version 1.1). http://gmao.gsfc.nasa.gov/pubs/office_notes.
  uckner, M. and A. Ciccone (2011). Rain and the democratic window of opportunity. Economet-
Br¨
  rica 79 (3), 923–47.
Carletto, C., S. Gourlay, S. Murray, and A. Zezza (2017). Cheaper, faster, and more than good
  enough: Is GPS the new gold standard in land area measurement. Survey Research Meth-
  ods 11 (3), 235–65.
Central Statistics Agency of Ethiopia (CSA) (2014). Ethiopia Rural Socioeconomic Survey 2011-
  2012. Public Use Dataset. Ref: ETH 2011 ERSS v01 M. Downloaded from https://microdata.
  worldbank.org/index.php/catalog/2053 on 6 September 2019.
Central Statistics Agency of Ethiopia (CSA) (2015). Ethiopia Socioeconomic Survey 2013-
  2014. Public Use Dataset. Ref: ETH 2013 ESS v02 M. Downloaded from https://microdata.
  worldbank.org/index.php/catalog/2053 on 6 September 2019.

                                                24
Central Statistics Agency of Ethiopia (CSA) (2017). Ethiopia Socioeconomic Survey 2015-
  2016, Wave 3. Public Use Dataset. Ref: ETH 2015 ESS v02 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/2783 on 6 September 2019.

Central Statistics Agency of Ethiopia (CSA) (2020). Ethiopia Socioeconomic Survey 2018-
  2019. Public Use Dataset. Ref: ETH 2018 ESS v03 M. Downloaded from https://microdata.
  worldbank.org/index.php/catalog/3823 on 9 February 2024.

Central Statistics Agency of Ethiopia (CSA) (2024). Ethiopia Socio-Economic Panel Survey
  2021-2022. Public Use Dataset. Ref: ETH 2021 ESPS-W5 v03 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/6161 on 15 January 2024.

Chen, J. J., V. Mueller, Y. Jia, and S. K.-H. Tseng (2017). Validating migration responses to
  flooding using satellite and vital registration data. American Economic Review 107 (5), 441–45.

Chen, M., W. Shi, P. Xie, V. B. Silva, V. E. Kousky, R. W. Higgins, and J. E. Janowiak (2008).
  Assessing objective techniques for gauge-based analyses of global daily precipitation. Journal of
 Geophysical Research 113, D04110.

Corno, L., N. Hildebrandt, and A. Voena (2020). Age of marriage, weather shocks, and the direction
  of marriage payments. Econometrica 88 (3), 879–915.

Dell, M., B. F. Jones, and B. A. Olken (2012). Temperature shocks and economic growth: Evidence
  from the last half century. American Economic Journal: Macroeconomics 4 (3), 66–95.

Deschˆene, O. and M. Greenstone (2007). The economic impacts of climate change: Evidence from
  agricultural output and random fluctuations in weather. American Economic Review 97 (1),
  354–85.

Field, E., R. Pande, N. Rigol, S. Schaner, E. Stacy, and C. T. Moore (2023). Measuring time
  use in rural india: Design and validation of a low-cost survey module. Journal of Development
  Economics 164, 103105.

Fitzpatrick, A. (2023). Which price is right? a comparison of three standard approaches to mea-
  suring prices. Journal of Development Economics 163, 103106.

Funk, C., P. Peterson, M. Landsfeld, D. Pedreros, J. Verdin, S. Shukla, G. Husak, J. Rowland,
  L. Harrison, A. Hoell, and J. Michaelsen (2015). The climate hazards infrared precipitation with
  stations—a new environmental record for monitoring extremes. Scientific Data 2, 150066.

Garg, T., M. Jagnani, and V. Taraz (2020). Temperature and human capital in India. Journal of
 the Association of Environmental and Resource Economists 7 (6), 1113–50.

Gibson, J. (2021). Better night lights data, for longer. Oxford Bulletin of Economics and Statis-
  tics 83 (3), 770–791.

Gibson, J., S. Olivia, G. Boe-Gibson, and C. Li (2021). Which night lights data should we use in
  economics, and where? Journal of Development Economics 149, 102602.

Gollin, D. and C. Udry (2021). Heterogeneity, measurement error, and misallocation: Evidence
 from African agriculture. Journal of Political Economy 129 (1), 1–80.

                                                25
Gourley, S., T. Kilic, and D. B. Lobell (2019). A new spin on an old debate: Errors in farmer-
 reported production and their implications for inverse scale - productivity relationship in Uganda.
 Journal of Development Economics 141, 102376.

Hennermann, K. and P. Berrisford (2020). ERA5: data documentation. Last modified Nov 18,
  2020. https://confluence.ecmwf.int/display/CKB/ERA5%3A+data+documentation.

Jack, B. K., K. McDermott, and A. Sautmann (2022). Multiple price lists for willingness to pay
  elicitation. Journal of Development Economics 159, 102977.

Jagnani, M., C. B. Barrett, Y. Liu, and L. You (2021). Within-season producer response to warmer
  temperatures: Defensive investments by kenyan farmers. Economics Journal 131 (633), 392–419.

Jain, M. (2020). The benefits and pitfalls of using satellite data for causal inference. Review of
  Environmental Economics and Policy 14 (1), 157–69.

Janzen, S. A. and J. D. Michler (2021). Ulysses’ pact or Ulysses’ raft: Using pre-analysis plans in
  experimental and nonexperimental research. Applied Economic Perspectives and Policy, forth-
  coming.

Jayachandran, S. (2006). Selling labor low: Wage responses to productivity shocks in developing
  countries. Journal of Political Economy 114 (3), 538–75.

Kaur, S. (2019). Nominal wage rigidity in village labor markets.          American Economic Re-
 view 109 (10), 3585–616.

Kilic, T., H. Moylan, J. Ilukor, C. Mtengula, and I. Pangapanga-Phiri (2021). Root for the tubers:
  Extended-harvest crop production and productivity measurement in surveys. Food Policy 102,
  102033.

 onig, M. D., D. Rohner, M. Thoenig, and F. Zilibotti (2017). Networks in conflict: Theory and
K¨
 evidence from the Great War of Africa. Econometrica 85 (4), 1093–1132.

Kosmowski, F., A. Aragaw, A. Kilian, A. Ambel, J. Ilukor, B. Yigezu, and J. Stevenson (2019).
 Varietal identification in household surveys: Results from three household-based methods against
 the benchmark of DNA fingerprinting in southern Ethiopia. Experimental Agriculture 55 (3),
 371–85.

Lobell, D. B., G. Azzari, M. Burke, S. Gourlay, Z. Jin, T. Kilic, and S. Murray (2020). Eyes in
  the sky, boots on the ground: Assessing satellite- and ground-based approaches to crop yield
  measurement and analysis. American Journal of Agricultural Economics 102 (1), 202–19.

Lundstad, E., Y. Brugnara, D. Pappert, J. Kopp, E. Samakinwa, A. H¨          urzeler, A. Andersson,
  B. Chimani, R. Cornes, G. Demar´  ee, J. Filipiak, L. Gates, G. L. Ives, J. M. Jones, S. Jourdain,
  A. Kiss, S. E. Nicholson, R. Przybylak, P. Jones, D. Rousseau, B. Tinz, F. S. Rodrigo, S. Grab,
  F. Dom´ ınguez-Castro, V. Slonosky, J. Cooper, M. Brunet, and S. Br¨ onnimann (2023). The global
  historical climate database HCLIM. Scientific Data 10 (1), 44.

Maccini, S. and D. Yang (2009). Under the weather: Health, schooling, and economic consequences
 of early-life rainfall. American Economic Review 99 (3), 1006–26.

                                                26
McCarthy, N., T. Kilic, J. Brubaker, and S. Murray (2021). Droughts and floods in Malawi: Impacts
 on crop production and the performance of sustainable land management practices under climate
 extremes. Environment and Development Economics 26, 432–49.
Michler, J. D., K. Baylis, M. Arends-Kuenning, and K. Mazvimavi (2019). Conservation agriculture
 and climate resilience. Journal of Environmental Economics and Management 93, 148–69.
Michler, J. D., A. Josephson, T. Kilic, and S. Murray (2019). Empirically estimating the impact
 of weather on agriculture. OSF Registries. July 1. https://doi.org/10.17605/osf.io/z3snh.
Michler, J. D., A. Josephson, T. Kilic, and S. Murray (2021a). Estimating the impact of weather
 on agriculture. arXiv.org. August 19. https://arxiv.org/abs/2012.11768v2.
Michler, J. D., A. Josephson, T. Kilic, and S. Murray (2021b). Estimating the impact of weather
 on agriculture. World Bank Policy Research Working Paper, No. 9867.
Michler, J. D., A. Josephson, T. Kilic, and S. Murray (2022). Privacy protection, measurement
 error, and the integration of remote sensing and socioeconomic survey data. Journal of Devel-
 opment Economics 158, 102927.
Miguel, E., S. Satyanath, and E. Sergenti (2004). Economic shocks and civil conflict: An instru-
 mental variables approach. Journal of Political Economy 112 (4), 725–53.
Morten, M. (2019). Temporary migration and endogenous risk sharing in village India. Journal of
 Political Economy 127 (1), 1–46.
Naqvi, A. (2024). Stata package “bumpline” (version 1.21).
National Bureau of Statistics (NBS) (2012). Nigeria General Household Survey (GHS), Panel
  2010-2011, Wave 1. Public Use Dataset. Ref: NGA 2010 GHSP-W1 v04 M. Downloaded from
 https://microdata.worldbank.org/index.php/catalog/1002 on 6 September 2019.
National Bureau of Statistics (NBS) (2014). Nigeria General Household Survey, Panel 2012-2013,
  Wave 2. Public Use Dataset. Ref: NGA 2012 GHSP-W2 v02 M. Downloaded from https://
  microdata.worldbank.org/index.php/catalog/1952 on 6 September 2019.
National Bureau of Statistics (NBS) (2016). Nigeria General Household Survey, Panel 2015-2016,
  Wave 3. Public Use Dataset. Ref: NGA 2015 GHSP-W3 v02 M. Downloaded from https://
  microdata.worldbank.org/index.php/catalog/2734 on 6 September 2019.
National Bureau of Statistics (NBS) (2019). Nigeria General Household Survey, Panel 2018-2019,
  Wave 4. Public Use Dataset. Ref: NGA 2018 GHSP-W4 v03 M. Downloaded from https://
  microdata.worldbank.org/index.php/catalog/3557 on 9 February 2024.
National Statistical Office (NSO) (2012). Malawi Third Integrated Household Survey 2010-2011.
  Public Use Dataset. Ref: MWI 2010 IHS-III v01 M. Downloaded from https://microdata.
 worldbank.org/index.php/catalog/1003 on 6 September 2019.
National Statistical Office (NSO) (2015). Malawi Integrated Household Panel Survey 2010-2013
  (short-term panel, 204 EAs). Public Use Dataset. Ref: MWI 2010-2013 IHPS v01 M. Down-
  loaded from https://microdata.worldbank.org/index.php/catalog/2248 on 6 September
  2019.

                                              27
National Statistical Office (NSO) (2017). Malawi Fourth Integrated Household Survey 2016-2017.
  Public Use Dataset. Ref: MWI 2016 IHS-IV v04 M. Downloaded from https://microdata.
  worldbank.org/index.php/catalog/2936 on 6 September 2019.

National Statistical Office (NSO) (2020a). Malawi Fifth Integrated Household Survey 2019-2020.
  Public Use Dataset. Ref: MWI 2019 IHS-V v06 M. Downloaded from https://microdata.
 worldbank.org/index.php/catalog/3818 on 9 February 2024.

National Statistical Office (NSO) (2020b). Malawi Integrated Household Panel Survey 2010-2013-
  2016-2019 (long-term panel, 102 EAs). Public Use Dataset. Ref: MWI 2010-2019 IHPS v06 M.
  Downloaded from https://microdata.worldbank.org/index.php/catalog/3819 on 21 May
  2024.

Novella, N. S. and W. M. Thiaw (2013). African rainfall climatology version 2 for famine early
  warning systems. Journal of Applied Meteorology and Climatology 52 (3), 588–606.

Parker, W. S. (2016). Reanalyses and observations: What’s the difference? Bulletin of the American
  Meteorological Society 97 (9), 1565–72.

Ritchie, J. and D. NeSmith (1991). Temperature and crop development. In R. J. Hanks and J. T.
  Ritchie (Eds.), Modeling Plant and Soil Systems, pp. 5–29. American Society of Agronomy, Crop
  Science Society of America, Soil Science Society of America.

Sarsons, H. (2015). Rainfall and conflict: A cautionary tale.      Journal of Development Eco-
  nomics 115, 62–72.

Shah, M. and B. M. Steinberg (2017). Drought of opportunities: Contemporaneous and long-term
  impacts of rainfall shocks on human capital. Journal of Political Economy 125 (2), 527–61.

Survey and Census Division, National Institute of Statistics, Niger (NIS) (2014). National
  Survey on Household Living Conditions and Agriculture 2011. Public Use Dataset. Ref:
  NER 2011 ECVMA v01 M. Downloaded from https://microdata.worldbank.org/index.
  php/catalog/2050 on 6 September 2019.

Survey and Census Division, National Institute of Statistics, Niger (NIS) (2016).   Na-
  tional Survey on Household Living Conditions and Agriculture (ecvma) 2014. Public Use
  Dataset. Ref: NER 2014 ECVMA v02 M. Downloaded from https://microdata.worldbank.
  org/index.php/catalog/2676 on 6 September 2019.

Tanzania National Bureau of Statistics (TNBS) (2011). Tanzania National Panel Survey 2008-
  2009, Wave 1. Public Use Dataset. Ref: TZA 2008 NPS-R1 v03 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/76 on 6 September 2019.

Tanzania National Bureau of Statistics (TNBS) (2012). Tanzania National Panel Survey 2010-
  2011, Wave 2. Public Use Dataset. Ref: TZA 2010 NPS-R2 v03 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/1050 on 6 September 2019.

Tanzania National Bureau of Statistics (TNBS) (2015). Tanzania National Panel Survey 2012-
  2013, Wave 3. Public Use Dataset. Ref: TZA 2012 NPS-R3 v01 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/2252 on 6 September 2019.

                                               28
Tanzania National Bureau of Statistics (TNBS) (2017). Tanzania National Panel Survey 2014-
  2015, Wave 4. Public Use Dataset. Ref: TZA 2014 NPS-R4 v03 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/2862 on 6 September 2019.
Tanzania National Bureau of Statistics (TNBS) (2021). Tanzania National Panel Survey
  2019-2020 - Extended Panel with Sex Disaggregated Data.   Public Use Dataset. Ref:
  TZA 2019 NPS-SDD v06 M. Downloaded from https://microdata.worldbank.org/index.
  php/catalog/3885 on 9 February 2024.
Tanzania National Bureau of Statistics (TNBS) (2023). Tanzania National Panel Survey 2020-
  2021, Wave 5. Public Use Dataset. Ref: TZA 2020 NPS-R5 v02 M. Downloaded from https:
  //microdata.worldbank.org/index.php/catalog/5639 on 9 February 2024.
Taraz, V. (2018). Can farmers adapt to higher temperatures? evidence from India. World Devel-
  opment 112, 205–19.
Tarnavsky, E., D. Grimes, R. Maidment, E. Black, R. P. Allan, M. Stringer, R. Chadwick, and
  F. Kayitakire (2014). Extension of the TAMSAT satellite-based rainfall monitoring over Africa
  and from 1983 to present. Journal of Applied Meteorology and Climatology 53 (12), 2805–22.
Tesfaye, W., G. Blalock, and N. Tirivayi (2021). Climate-smart innovations and rural poverty in
  Ethiopia: Exploring impacts and pathways. American Journal of Agricultural Economics 103 (3),
  878–99.
Tzachor, A., C. Richards, M. Gudoshava, P. Nying’uro, H. Misiani, J. Ongoma, Y. Yair, Y. Mu-
  lugetta, and A. Gaye (2023). How to reduce Africa’s undue exposure to climate risk. Nature 620,
  488–91.
Uganda Bureau of Statistics (UBOS) (2012). Uganda National Panel Survey 2005-2009. Pub-
  lic Use Dataset. Ref: UGA 2005-2009 UNPS v01 M. Downloaded from https://microdata.
 worldbank.org/index.php/catalog/1001 on 6 September 2019.
Uganda Bureau of Statistics (UBOS) (2014a). Uganda National Panel Survey 2010-2011. Public
  Use Dataset. Ref: UGA 2010 UNPS v01 M. Downloaded from https://microdata.worldbank.
 org/index.php/catalog/2166 on 6 September 2019.
Uganda Bureau of Statistics (UBOS) (2014b). Uganda National Panel Survey 2011-2012, Wave
  III. Public Use Dataset. Ref: UGA 2011 UNPS v01 M. Downloaded from https://microdata.
  worldbank.org/index.php/catalog/2059 on 6 September 2019.
Uganda Bureau of Statistics (UBOS) (2016). Uganda National Panel Survey 2013-2014. Public
  Use Dataset. Ref: UGA 2013 UNPS v02 M. Downloaded from https://microdata.worldbank.
 org/index.php/catalog/2663 on 9 February 2024.
Uganda Bureau of Statistics (UBOS) (2019). Uganda National Panel Survey 2015-2016. Public
  Use Dataset. Ref: UGA 2015 UNPS v02 M. Downloaded from https://microdata.worldbank.
  org/index.php/catalog/3460 on 9 February 2024.
Uganda Bureau of Statistics (UBOS) (2021). Uganda National Panel Survey 2019-2020. Public
  Use Dataset. Ref: UGA 2019 UNPS v03 M. Downloaded from https://microdata.worldbank.
  org/index.php/catalog/3902 on 9 February 2024.

                                               29
Wilkinson, G. G. (1996). A review of current issues in the integration of gis and remote sensing
 data. International Journal of Geographical Information Systems 10 (1), 85–101.

Wineman, A., N. M. Mason, J. Ochieng, and L. Kirimi (2017). Weather extremes and household
 welfare in rural Kenya. Food Security 9, 281–300.

Yeh, C., A. Perez, A. Driscoll, G. Azzari, Z. Tang, D. Lobell, S. Ermon, and M. Burke (2020).
  Using publicly available satellite imagery and deep learning to understand economic well-being
  in Africa. Nature Communications 11, 2583.

Zandler, H., T. Senftl, and K. A. Vanselow (2020). Reanalysis datasets outperform other gridded
  climate products in vegetation change analysis in peripheral conservation areas of Central Asia.
  Scientific Reports 10, 22446.




                                               30
                                                             Table 1: Sources of Weather Data


     dataset                                                                       Length of record      Resolution          Time step   Data            Units
     Precipitation
     -Africa Rainfall Climatology version 2 (ARC2)                                 1983-current          0.1 deg             daily       total precip    mm
     -Climate Hazards group InfraRed Precipitation with Station data               1981-current          0.05 deg            daily       total precip    mm
     (CHIRPS)
     -CPC Global Unified Gauge-Based Analysis of Daily Precipitation                1979-current          0.5 deg             daily       total precip    mm
     -European Centre for Medium-Range Weather Forecasts                           1979-current          0.28 deg            hourly      total precip    m
     (ECMWF) ERA5
     -Modern-Era Retrospective analysis for Research and Applica-                  1980-current          0.625x0.5 deg       hourly      rain rate       kg m2 s1
     tions, version 2 (MERRA-2) Surface Flux Diagnostics
     -Tropical Applications of Meteorology using SATellite data and                1983-current          0.0375 deg          daily       total precip    mm
     ground-based observations (TAMSAT)
     Temperature
     -CPC Global Unified Gauge-Based Analysis of Daily Temperature                  1979-current          0.5 deg             daily       min, max temp   C
     -European Centre for Medium-Range Weather Forecasts                           1979-current          0.28 deg            hourly      mean temp       K
31




     (ECMWF) ERA5
     -Modern-Era Retrospective analysis for Research and Applica-                  1980-current          0.625x0.5 deg       daily       mean temp       K
     tions, version 2 (MERRA-2) statD

     Note : The table summarizes the Earth observation sources and related details for precipitation and temperature data.
                              Table 2: Sources of Household Data


Country      Survey Name                                                    Years    Original n      Final n
Ethiopia     Ethiopia Socioeconomic Survey (ESS)                      2011/2012           3,969         1,689
                                                                      2013/2014           5,262         2,865
                                                                      2015/2016           4,954         2,718
                                                                      2018/2019           7,527         1,996
                                                                      2021/2022           4,999         1,406
Malawi       Integrated Household Panel Survey (IHPS)                 2010/2011           3,247         2,250
                                                                           2013           4,000         2,472
                                                                           2016           2,508         1,845
                                                                           2019           3,178         2,330
Niger            ete Nationale sur les Conditions de Vie des
             Enquˆ                                                         2011           3,968         2,223
               enages et l’Agriculture (ECVMA)
              M´                                                           2014           3,617         1,690
Nigeria      General Household Survey (GHS)                           2010/2011           4,916         2,674
                                                                      2012/2013           4,716         2,768
                                                                      2015/2016           4,581         2,783
                                                                      2018/2019           4,976           920
Tanzania     Tanzania National Panel Survey (TZNPS)                   2008/2009           3,280         2,001
                                                                      2010/2011           3,924         2,013
                                                                      2012/2013           5,015         1,889
                                                                      2014/2015           3,352         2,127
                                                                      2019/2020           1,184           312
                                                                      2020/2021           4,709         1,564
Uganda       Uganda National Panel Survey (UNPS)                      2009/2010           2,975         1,883
                                                                      2010/2011           2,716         1,886
                                                                      2011/2012           2,850         2,020
                                                                      2013/2014           3,119         2,190
                                                                      2015/2016           3,305         1,868
                                                                      2019/2020           3,098         1,845
Total        6 countries                                                27 waves        105,945       54,237

Note : The table summarizes the household data details for each country, per LSMS Basic Information Documents.




                                                   32
                         Figure 1: Varying Resolution of Rainfall Measurement




Note : The figure captures rainfall as measured by all six precipitation products for the same 100km x 100km area on
a single day (7 January 2010).




                      Figure 2: Varying Resolution of Temperature Measurement




Note : The figure captures temperature as measured by all three temperature products for the same 100km x 100km
area on a single day (7 January 2010).




                                                        33
                                   Figure 3: In-Country Weather Stations




Note : Maps display LSMS-ISA household locations along with locations of Global Historical Climatology Net-
work (GHCN) weather stations. Small black dots represent the approximate location of all surveyed communities
(anonymized public coordinates). Blue and organge triages represent the GHCN stations located within or near the
country border, where blue symbols are stations with at lesat some rainfall or temperature data reported during the
time period of interest. Note that for some households the closest station is outside the national boarders. Blue dots
with green halos are the stations that continue to report data to the GHCN network. In the case of Uganda, no in-
country stations currently report to the network, meaning the closest weather station data for LSMS-ISA households
in Uganda are one station in Kenya and one in Tanzania.


                                                         34
                                                                   Figure 4: Distribution of Total Seasonal Rainfall


                                                    Ethiopia (n = 10,674)                                                                  Malawi (n = 8,897)
                   .0025                                                                            .004

                    .002
                                                                                                    .003
         Density



                   .0015
                                                                                                    .002
                    .001

                                                                                                    .001
                   .0005

                          0                                                                           0
                                  0   2000                  4000                6000         8000                      500            1000               1500            2000   2500



                                                     Niger (n = 3,913)                                                                      Nigeria (n = 9,145)

                   .008                                                                              .002


                   .006                                                                             .0015
         Density




                   .004                                                                              .001
35




                   .002                                                                             .0005


                     0                                                                                     0
                              0       200                   400                 600          800                   0                1000                          2000                 3000



                                                    Tanzania (n = 9,916)                                                                   Uganda (n = 11,692)
                    .002                                                                            .004


                   .0015                                                                            .003
         Density




                    .001                                                                            .002

                   .0005                                                                            .001

                          0                                                                           0
                                  0          2000                        4000           6000                   0             1000                 2000                   3000          4000
                                              Total Season Rainfall (mm)                                                             Total Season Rainfall (mm)



                                                     ARC2                CHIRPS        CPC             ERA5              MERRA-2                   TAMSAT



     Note : The figure presents the distribution of total seasonal rainfall pooled across all years, disaggregated by Earth observation source. Each distribution
     (Earth observation product) in each panel is constructed using all household-year observations (n). Variation in distributions do not come from variation
     in the household data that is paired with the Earth observation data. Rather, variation in distributions within a panel is solely due to differences in the
     value of precipitation reported by the Earth observation source.
                                                                Figure 5: Average Number of Days Without Rain


                                                 Ethiopia (n = 10,674)                                                          Malawi (n = 8,897)
         Days without Rain   100                                                               140

                                                                                               120
                              80
                                                                                               100
                              60
                                                                                                80

                              40                                                                60

                                                                                                40
                              20
                                   2010              2015                       2020                 2008          2010        2012              2014          2016   2018



                                                  Niger (n = 3,913)                                                            Nigeria (n = 9,145)
                             80                                                                80
         Days without Rain




                                                                                               60
                             60

                                                                                               40

                             40
36




                                                                                               20


                             20                                                                 0
                                  2011    2012                           2013           2014        2010              2012            2014              2016          2018



                                                 Tanzania (n = 9,916)                                                          Uganda (n = 11,692)
                             120
                                                                                               80
         Days without Rain




                             100

                              80                                                               60

                              60
                                                                                               40
                              40

                              20                                                               20
                                   2005   2010                           2015           2020                2010                          2015                        2020
                                                         Year                                                                         Year



                                                  ARC2                CHIRPS      CPC                ERA5            MERRA-2           TAMSAT



     Note : The figure presents the mean number of days without rain (< 1mm) in a year, disaggregated by Earth observation source. Means are calculated by
     fitting a fractional-polynomial line to the data, with 95% confidence interval represented by shaded area. Each line (Earth observation product) in each
     panel is constructed using all household-year observations (n). Variation in lines do not come from variation in the household data that is paired with the
     Earth observation data. Rather, variation in lines within a panel is solely due to differences in the number of days without rain reported by the Earth
     observation source.
                                                                 Figure 6: Distribution of Mean Daily Temperature


                                                  Ethiopia (n = 10,674)                                                                         Malawi (n = 8,897)
                    .2                                                                                .4


                   .15                                                                                .3
         Density




                    .1                                                                                .2


                   .05                                                                                .1


                        0                                                                             0
                            10   15                20              25               30    35                    20                  22            24              26       28        30



                                                   Niger (n = 3,913)                                                                            Nigeria (n = 9,145)
                   .6                                                                                 .8


                                                                                                      .6
                   .4
         Density




                                                                                                      .4

                   .2
37




                                                                                                      .2


                   0                                                                                  0
                            28        30                   32                  34         36               20                                       25                          30



                                                  Tanzania (n = 9,916)                                                                         Uganda (n = 11,692)
                   .25                                                                                .4

                    .2
                                                                                                      .3
         Density




                   .15
                                                                                                      .2
                    .1
                                                                                                      .1
                   .05

                        0                                                                             0
                            15               20                           25              30                         18              20            22                 24   26        28
                                           Mean Seasonal Temperature (C)                                                                  Mean Seasonal Temperature (C)



                                                                                    CPC        ERA5                       MERRA-2



     Note : The figure presents the distribution of mean daily temperature pooled across all years, disaggregated by Earth observation source. Each distribution
     (Earth observation product) in each panel is constructed using all household-year observations (n). Variation in distributions do not come from variation
     in the household data that is paired with the Earth observation data. Rather, variation in distributions within a panel is solely due to differences in the
     temperature reported by the Earth observation source.
                                                                     Figure 7: Average Growing Degree Days (GDD)


                                                     Ethiopia (n = 10,674)                                                                  Malawi (n = 8,897)
                                                                                                          182
                                154
         Growing Degree Days

                               153.5                                                                      180

                                153
                                                                                                          178
                               152.5
                                                                                                          176
                                152

                               151.5                                                                      174

                                       2010               2015                           2020                   2008          2010          2012              2014          2016   2018



                                                      Niger (n = 3,913)                                                                     Nigeria (n = 9,145)
                               75                                                                         145
         Growing Degree Days




                               70
                                                                                                          140
                               65

                               60
                                                                                                          135
38




                               55

                               50
                                                                                                          130
                                    2011      2012                          2013                2014            2010                 2012          2014              2016          2018



                                                     Tanzania (n = 9,916)                                                                   Uganda (n = 11,692)
                               184                                                                        146
         Growing Degree Days




                               182                                                                        144

                               180                                                                        142

                               178                                                                        140

                               176                                                                        138
                                     2005     2010                          2015                2020                   2010                            2015                        2020
                                                            Year                                                                                   Year



                                                                                   CPC             ERA5                 MERRA-2



     Note : The figure presents the mean growing degree days in a year, disaggregated by Earth observation source. Means are calculated by fitting a fractional-
     polynomial line to the data, with 95% confidence interval represented by shaded area. Each line (Earth observation product) in each panel is constructed
     using all household-year observations (n). Variation in lines do not come from variation in the household data that is paired with the Earth observation
     data. Rather, variation in lines within a panel is solely due to differences in the daily temperatures reported by the Earth observation source.
                                      Figure 8: Specification Charts for Total Seasonal Rainfall

                                                   Ethiopia                                                                                                          Malawi
                                       p < 0.05         n.s.         p < 0.05                                                                        p < 0.05          n.s.        p < 0.05

                                                                                               .001
                                                                                                                                                                                                                 .002
                                                                                               .0005
                                                                                                                                                                                                                 .001
                                                                                               0
                                                                                                                                                                                                                 0
                                                                                               -.0005
                                                                                                                                                                                                                 -.001
                                                                                               -.001
    *Weather Product*                                                                                             *Weather Product*
              TAMSAT                                                                                                        TAMSAT
                 ERA5                                                                                                          ERA5
                 ARC2                                                                                                          ARC2
             MERRA-2                                                                                                       MERRA-2
                  CPC                                                                                                           CPC
               CHIRPS                                                                                                        CHIRPS


            *Dep. Var.*                                                                                                   *Dep. Var.*
                 Value                                                                                                         Value
               Quantity                                                                                                      Quantity


              * Model*                                                                                                      * Model*
  Weather + FE + Inputs                                                                                         Weather + FE + Inputs
          Weather + FE                                                                                                  Weather + FE
               Weather                                                                                                       Weather

                          0   4   8   12          16            20       24     28   32   36                                            0   4   8   12          16            20        24    28   32   36



                                                       Niger                                                                                                     Nigeria
                                       p < 0.05          n.s.        p < 0.05                                                                        p < 0.05          n.s.        p < 0.05

                                                                                                                                                                                                             .0015
                                                                                                   .005
                                                                                                                                                                                                             .001

                                                                                                   0                                                                                                         .0005

                                                                                                                                                                                                             0
                                                                                                   -.005
                                                                                                                                                                                                             -.0005
    *Weather Product*                                                                                             *Weather Product*                                                                          -.001
              TAMSAT                                                                               -.01                     TAMSAT
                 ERA5                                                                                                          ERA5
                 ARC2                                                                                                          ARC2
             MERRA-2                                                                                                       MERRA-2
                  CPC                                                                                                           CPC
               CHIRPS                                                                                                        CHIRPS


            *Dep. Var.*                                                                                                   *Dep. Var.*
                 Value                                                                                                         Value
               Quantity                                                                                                      Quantity


              * Model*                                                                                                      * Model*
  Weather + FE + Inputs                                                                                         Weather + FE + Inputs
          Weather + FE                                                                                                  Weather + FE
               Weather                                                                                                       Weather

                          0   4   8   12          16            20        24    28   32   36                                            0   4   8   12          16            20       24     28   32   36



                                                  Tanzania                                                                                                       Uganda
                                       p < 0.05         n.s.         p < 0.05                                                                        p < 0.05          n.s.        p < 0.05

                                                                                               .0005                                                                                                         .001

                                                                                                                                                                                                             .0005
                                                                                               0
                                                                                                                                                                                                             0

                                                                                                                                                                                                             -.0005
                                                                                               -.0005
                                                                                                                                                                                                             -.001
    *Weather Product*                                                                                             *Weather Product*
              TAMSAT                                                                                                        TAMSAT
                 ERA5                                                                                                          ERA5
                 ARC2                                                                                                          ARC2
             MERRA-2                                                                                                       MERRA-2
                  CPC                                                                                                           CPC
               CHIRPS                                                                                                        CHIRPS


            *Dep. Var.*                                                                                                   *Dep. Var.*
                 Value                                                                                                         Value
               Quantity                                                                                                      Quantity


              * Model*                                                                                                      * Model*
  Weather + FE + Inputs                                                                                         Weather + FE + Inputs
          Weather + FE                                                                                                  Weather + FE
               Weather                                                                                                       Weather

                          0   4   8   12          16            20       24     28   32   36                                            0   4   8   12          16            20       24     28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                           39
                                  Figure 9: Specification Charts for Number of Days With No Rain

                                                       Ethiopia                                                                                                         Malawi
                                           p < 0.05          n.s.        p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                       .02                                                                                                      .02

                                                                                                                                                                                                                .01
                                                                                                       .01
                                                                                                                                                                                                                0
                                                                                                       0
                                                                                                                                                                                                                -.01

                                                                                                       -.01                                                                                                     -.02
    *Weather Product*                                                                                                *Weather Product*
              TAMSAT                                                                                   -.02                    TAMSAT
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



                                                           Niger                                                                                                    Nigeria
                                           p < 0.05          n.s.        p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                       .05                                                                                                      .02

                                                                                                                                                                                                                .01

                                                                                                       0                                                                                                        0

                                                                                                                                                                                                                -.01

                                                                                                       -.05                                                                                                     -.02
    *Weather Product*                                                                                                *Weather Product*
              TAMSAT                                                                                                           TAMSAT
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



                                                      Tanzania                                                                                                      Uganda
                                       p < 0.05             n.s.         p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                                                                                                                                .01
                                                                                                   .005

                                                                                                   0                                                                                                            0

                                                                                                   -.005                                                                                                        -.01

                                                                                                   -.01                                                                                                         -.02
    *Weather Product*                                                                                                *Weather Product*
              TAMSAT                                                                                                           TAMSAT                                                                           -.03
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                              40
                                      Figure 10: Specification Charts for Mean Daily Temperature

                                                           Ethiopia                                                                                                   Malawi
                                            p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                    .4
                                                                                                                                                                                                               .1

                                                                                                    .2                                                                                                         0

                                                                                                    0                                                                                                          -.1

    *Weather Product*                                                                                            *Weather Product*                                                                             -.2
                                                                                                    -.2
             MERRA-2                                                                                                      MERRA-2                                                                              -.3
                 ERA5                                                                               -.4                       ERA5

                  CPC                                                                                                          CPC



            *Dep. Var.*                                                                                                  *Dep. Var.*

                 Value                                                                                                        Value

               Quantity                                                                                                     Quantity



              * Model*                                                                                                     * Model*

  Weather + FE + Inputs                                                                                        Weather + FE + Inputs

          Weather + FE                                                                                                 Weather + FE

               Weather                                                                                                      Weather

                          0   2   4     6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                            Niger                                                                                                     Nigeria
                                            p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                                                                                                                               .6
                                                                                                    .5

                                                                                                                                                                                                               .4
                                                                                                    0
                                                                                                                                                                                                               .2

    *Weather Product*                                                                               -.5          *Weather Product*                                                                             0

             MERRA-2                                                                                                      MERRA-2                                                                              -.2
                                                                                                    -1
                 ERA5                                                                                                         ERA5

                  CPC                                                                                                          CPC



            *Dep. Var.*                                                                                                  *Dep. Var.*

                 Value                                                                                                        Value

               Quantity                                                                                                     Quantity



              * Model*                                                                                                     * Model*

  Weather + FE + Inputs                                                                                        Weather + FE + Inputs

          Weather + FE                                                                                                 Weather + FE

               Weather                                                                                                      Weather

                          0   2   4     6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                       Tanzania                                                                                                       Uganda
                                            p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                    .2                                                                                                         .2

                                                                                                    .1
                                                                                                                                                                                                               0
                                                                                                    0

                                                                                                    -.1                                                                                                        -.2
    *Weather Product*                                                                                            *Weather Product*
                                                                                                    -.2
             MERRA-2                                                                                                      MERRA-2                                                                              -.4
                                                                                                    -.3
                 ERA5                                                                                                         ERA5

                  CPC                                                                                                          CPC



            *Dep. Var.*                                                                                                  *Dep. Var.*

                 Value                                                                                                        Value

               Quantity                                                                                                     Quantity



              * Model*                                                                                                     * Model*

  Weather + FE + Inputs                                                                                        Weather + FE + Inputs

          Weather + FE                                                                                                 Weather + FE

               Weather                                                                                                      Weather

                          0   2   4     6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                          41
                                  Figure 11: Specification Charts for Growing Degree Days (GDD)

                                                         Ethiopia                                                                                                    Malawi
                                          p < 0.05          n.s.        p < 0.05                                                                      p < 0.05          n.s.        p < 0.05

                                                                                                  .1
                                                                                                                                                                                                                  .15
                                                                                                                                                                                                                  .1
                                                                                                  .05                                                                                                             .05
                                                                                                                                                                                                                  0
                                                                                                  0                                                                                                               -.05
    *Weather Product*                                                                                           *Weather Product*

             MERRA-2                                                                                                     MERRA-2                                                                                  -.1
                                                                                                  -.05
                 ERA5                                                                                                        ERA5

                  CPC                                                                                                         CPC



            *Dep. Var.*                                                                                                 *Dep. Var.*

                 Value                                                                                                       Value

               Quantity                                                                                                    Quantity



              * Model*                                                                                                    * Model*

  Weather + FE + Inputs                                                                                       Weather + FE + Inputs

          Weather + FE                                                                                                Weather + FE

               Weather                                                                                                     Weather

                          0   2   4   6              8             10       12     14   16   18                                       0   2   4   6              8             10        12    14   16   18



                                                          Niger                                                                                                      Nigeria
                                          p < 0.05          n.s.        p < 0.05                                                                      p < 0.05          n.s.        p < 0.05

                                                                                                  .04                                                                                                             .04

                                                                                                  .02                                                                                                             .02

                                                                                                  0
                                                                                                                                                                                                                  0
    *Weather Product*                                                                             -.02          *Weather Product*
                                                                                                                                                                                                                  -.02
             MERRA-2                                                                                                     MERRA-2
                                                                                                  -.04
                 ERA5                                                                                                        ERA5                                                                                 -.04
                  CPC                                                                                                         CPC



            *Dep. Var.*                                                                                                 *Dep. Var.*

                 Value                                                                                                       Value

               Quantity                                                                                                    Quantity



              * Model*                                                                                                    * Model*

  Weather + FE + Inputs                                                                                       Weather + FE + Inputs

          Weather + FE                                                                                                Weather + FE

               Weather                                                                                                     Weather

                          0   2   4   6              8             10       12     14   16   18                                       0   2   4   6              8             10        12    14   16   18



                                                     Tanzania                                                                                                        Uganda
                                          p < 0.05          n.s.        p < 0.05                                                                      p < 0.05         n.s.         p < 0.05

                                                                                                  .15                                                                                                         .005

                                                                                                  .1
                                                                                                                                                                                                              0
                                                                                                  .05

                                                                                                  0                                                                                                           -.005

    *Weather Product*                                                                             -.05          *Weather Product*
                                                                                                                                                                                                              -.01
             MERRA-2                                                                              -.1                    MERRA-2
                                                                                                                                                                                                              -.015
                 ERA5                                                                                                        ERA5

                  CPC                                                                                                         CPC



            *Dep. Var.*                                                                                                 *Dep. Var.*

                 Value                                                                                                       Value

               Quantity                                                                                                    Quantity



              * Model*                                                                                                    * Model*

  Weather + FE + Inputs                                                                                       Weather + FE + Inputs

          Weather + FE                                                                                                Weather + FE

               Weather                                                                                                     Weather

                          0   2   4   6              8             10       12     14   16   18                                       0   2   4   6              8             10       12     14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                         42
                                                  Figure 12: Bumpline: Total Rainfall Growing Season
43




     Note : The figure presents a bumpline plot, where each panel represents a different country, where each column represents a different models and a different
     outcome variable for each model; each row represents the rank of each coefficient (one to six, with one representing the largest number and six representing
     the smallest number) within those regressions. Each panel includes 36 regressions. The bumpline package was developed by Naqvi (2024).
                                                         Figure 13: Bumpline: Mean Temperature
44




     Note : The figure presents a bumpline plot, where each panel represents a different country, where each column represents a different models and a different
     outcome variable for each model; each row represents the rank of each coefficient (one to six, with one representing the largest number and six representing
     the smallest number) within those regressions. Each panel includes 18 regressions. The bumpline package was developed by Naqvi (2024).
    Appendix to “The Mismeasure of Weather: Using Remotely
                   Sensed Weather in Economic Contexts”

Defining Growing Season
We define growing season following the FAO crop calendar for each country. Table A2 presents
details for each country on the growing season used, as well as whether that season spans years and
whether it is unimodal or bimodal. Earth observation data used in our analysis follows the defined
growing season in each respective country.
   It is important to note that both Malawi and Tanzania span calendar years. This means that
the growing season begins in one year and extends into the year that follows. Consider Malawi.
The growing season begins on 1 October and ends on 30 April. This means that it would begin 1
October 2024 and would end 30 April 2025.
   Similarly, both Nigeria and Uganda have bimodal seasons. Season modality designates whether
different regions within the countries have different growing seasons. In both Nigeria and Uganda,
the northern part of the country has a different growing season from the southern part of the
country. In these cases we designate the modality of the season, and also provide the growing
season dates for both regions.

Weather Variables




                                                45
                          Table A1: Weather Variables & Transformations


Panel A: Rainfall
Daily rainfall                            In mm
Mean                                      The first moment of the daily rainfall distribution for the growing
                                          season†
Median                                    The median daily rainfall for the growing season†
Variance                                  The second moment of the daily rainfall distribution for the grow-
                                          ing season†
Skew                                      The third moment of the daily rainfall distribution for the growing
                                          season†
Total                                     Cumulative daily rainfall for the growing season†
Deviations in total rainfall              The z-score for cumulative daily rainfall for the growing season†
Scaled deviations in total rainfall       The z-score for cumulative daily rainfall for the growing season†
Rainfall days                             The number of days with at least 1 mm of rain for the growing
                                          season†
Deviation in rainfall days                The number of days with rain for the growing season minus the
                                          long run average∗
No rain days                              The number of days with less than 1 mm of rain for the growing
                                          season†
Deviation in no rain days                 The number of days without rain for the growing season minus
                                          the long run average∗
Share of rainy days                       The percent of growing season days with rain†
Deviation in share of rainy days          The percent of growing season days with rain minus the long run
                                          average† ∗
Intra-season dry spells                   The maximum length of time (measured in days) without rain
                                          during the growing season†

Panel B: Temperature
Daily average temperature                 In ◦ Celsius
Daily maximum temperature                 In ◦ Celsius
Mean                                      The first moment of the daily temperature distribution for the
                                          growing season†
Median                                    The median daily temperature for the growing season†
Variance                                  The second moment of the daily temperature distribution for the
                                          growing season†
Skew                                      The third moment of the daily temperature distribution for the
                                          growing season†
Growing degree days (GDD)                 The number of days within bound temperature for the growing
                                          season, following Ritchie and NeSmith (1991)†
Deviation in GDD                          GDD for the growing season minus the long run average† ∗
Scaled deviation in GDD                   The z-score for GDD
Maximum temperature                       The average maximum daily temperature

Note : The table presents definitions for included weather variables and transformations from weather sources defined in
Table 1. † Growing season determined for each country following FAO crop calendar (see Table A2). ∗ For variables when
“long run” is referenced, long run is defined as the entire length of the weather dataset. While each weather source has
a different start date, to ensure blinding all datasets were shortened to 1983, which is the latest start date of the data
sources.




                                                         46
                                      Table A2: Growing Seasons


                              Growing Season                Span Calendar Years        Season Modality
         Ethiopia     1 March - 30 November                         no                    unimodal
         Malawi       1 October - 30 April                          yes                   unimodal
         Niger        1 June - 30 November                          no                    unimodal
         Nigeria      North : 1 May - 30 September                  no                     bimodal
                      South : 1 March - 31 August
         Tanzania     1 November - 30 April                           yes                  unimodal
         Uganda       North : 1 April - 30 September                  no                    bimodal
                      South : 1 February - 31 July

         Note : The table presents the growing season ranges, as defined by following FAO crop calendar for
         each country, respectively.




Household Data Summary Statistics

Specification and Bumpline Charts




                                                      47
                      Table A3: Household Data Summary Statistics


                                       Ethiopia        Malawi    Niger     Nigeria   Tanzania   Uganda

Panel A: Total Farm Production
Total farm production (2015 USD)         340.3          395.0     188.5     639.3     223.3      145.5
                                        (538.8)        (3913)    (201.7)   (689.3)   (369.2)    (195.8)
Total farmed area (ha)                   0.959          0.852     11.14     1.999     1.748      3.872
                                        (4.709)        (0.858)   (15.59)   (3.092)   (3.691)    (8.939)
Total farm yield (2015 USD/ha)           605.0          367.3     60.93     728.9     233.1      98.90
                                        (909.7)        (393.9)   (169.0)   (929.3)   (306.0)    (136.4)
Total farm labor rate (days/ha)          480.1          241.4     94.82     161.9     299.2      286.2
                                        (987.5)        (187.6)   (219.1)   (218.4)   (359.6)    (363.8)
Total farm fertilizer rate (kg/ha)       67.02          126.8     5.120     109.2     23.87      0.376
                                       (108.35)        (279.7)   (23.83)   (240.8)   (79.05)    (2.601)
Total farm pesticide use (%)             0.072          0.058     0.080     0.211     0.087      0.069
                                        (0.259)        (0.234)   (0.271)   (0.408)   (0.282)    (0.253)
Total farm herbicide use (%)             0.229          0.016     0.037     0.307     0.085      0.036
                                        (0.420)        (0.125)   (0.190)   (0.461)   (0.279)    (0.185)
Total farm irrigation use (%)            0.070          0.016     0.000     0.028     0.036      0.015
                                        (0.255)        (0.126)   (0.000)   (0.165)   (0.187)    (0.123)
Observations                            10674           8897      3913      9145      9916      11692
Households                              5333            3833      2320      3412      4804      4003

Panel B: Primary Crop Production
Primary crop production (kg)            330.1        1043         471.5     607.8     660.9      44.38
                                       (492.2)     (3034.8)      (488.5)   (962.2)   (881.0)    (75.11)
Primary crop farmed area (ha)           0.295        0.712        4.930     0.677     0.984      1.121
                                       (1.083)      (0.760)      (6.451)   (0.999)   (1.691)    (2.002)
Primary crop yield (kg/ha)              1711         1491         219.8     1840      1160       82.06
                                       (1893)       (1379)       (363.5)   (1992)    (1322)     (115.6)
Primary crop labor rate (days/ha)       493.3        237.8        70.71     167.2     259.2      328.2
                                       (712.5)      (193.8)      (121.1)   (232.2)   (302.6)    (435.1)
Primary crop fertilizer rate (kg/ha)    91.08        118.7        1.945     150.0     28.23      0.471
                                       (159.9)      (250.6)      (7.754)   (296.5)   (89.41)    (3.450)
Primary crop pesticide use (%)          0.044        0.027        0.055     0.185     0.077      0.037
                                       (0.204)      (0.163)      (0.228)   (0.388)   (0.266)    (0.189)
Primary crop herbicide use (%)          0.044        0.009        0.028     0.390     0.072      0.033
                                       (0.206)      (0.094)      (0.164)   (0.488)   (0.259)    (0.178)
Primary crop irrigation use (%)         0.050        0.013        0.000     0.022     0.020      0.012
                                       (0.219)      (0.115)      (0.000)   (0.146)   (0.141)    (0.110)
Observations                             5904           7866      3492      5349      7341       6156
Households                               3490           3594      2079      3055      3837       2872
Note : The table presents the mean and (standard deviation) of total farm production and primary crop
production. Statistics are calculated for each country, aggregated across all waves used.




                                                  48
                                      Figure A1: Specification Charts for Mean Daily Rainfall

                                                       Ethiopia                                                                                                        Malawi
                                           p < 0.05          n.s.        p < 0.05                                                                      p < 0.05          n.s.        p < 0.05

                                                                                                       .2                                                                                                      .4


                                                                                                       .1                                                                                                      .2


                                                                                                       0                                                                                                       0


                                                                                                       -.1                                                                                                     -.2
    *Weather Product*                                                                                               *Weather Product*
              TAMSAT                                                                                   -.2                    TAMSAT
                 ERA5                                                                                                            ERA5
                 ARC2                                                                                                            ARC2
             MERRA-2                                                                                                         MERRA-2
                  CPC                                                                                                             CPC
               CHIRPS                                                                                                          CHIRPS


            *Dep. Var.*                                                                                                     *Dep. Var.*
                 Value                                                                                                           Value
               Quantity                                                                                                        Quantity


              * Model*                                                                                                        * Model*
  Weather + FE + Inputs                                                                                           Weather + FE + Inputs
          Weather + FE                                                                                                    Weather + FE
               Weather                                                                                                         Weather

                          0   4   8   12              16            20        24    28   32   36                                          0   4   8   12          16            20        24    28   32   36



                                                           Niger                                                                                                   Nigeria
                                           p < 0.05          n.s.        p < 0.05                                                                      p < 0.05          n.s.        p < 0.05

                                                                                                       1                                                                                                       .2

                                                                                                       .5                                                                                                      .1

                                                                                                       0                                                                                                       0

                                                                                                       -.5
                                                                                                                                                                                                               -.1
    *Weather Product*                                                                                               *Weather Product*
                                                                                                       -1
              TAMSAT                                                                                                          TAMSAT
                 ERA5                                                                                                            ERA5
                 ARC2                                                                                                            ARC2
             MERRA-2                                                                                                         MERRA-2
                  CPC                                                                                                             CPC
               CHIRPS                                                                                                          CHIRPS


            *Dep. Var.*                                                                                                     *Dep. Var.*
                 Value                                                                                                           Value
               Quantity                                                                                                        Quantity


              * Model*                                                                                                        * Model*
  Weather + FE + Inputs                                                                                           Weather + FE + Inputs
          Weather + FE                                                                                                    Weather + FE
               Weather                                                                                                         Weather

                          0   4   8   12              16            20        24    28   32   36                                          0   4   8   12          16            20        24    28   32   36



                                                      Tanzania                                                                                                     Uganda
                                       p < 0.05             n.s.         p < 0.05                                                                      p < 0.05          n.s.        p < 0.05

                                                                                                   .1                                                                                                          .3

                                                                                                   .05                                                                                                         .2

                                                                                                   0                                                                                                           .1

                                                                                                   -.05                                                                                                        0

                                                                                                   -.1                                                                                                         -.1
    *Weather Product*                                                                                               *Weather Product*
              TAMSAT                                                                                                          TAMSAT
                 ERA5                                                                                                            ERA5
                 ARC2                                                                                                            ARC2
             MERRA-2                                                                                                         MERRA-2
                  CPC                                                                                                             CPC
               CHIRPS                                                                                                          CHIRPS


            *Dep. Var.*                                                                                                     *Dep. Var.*
                 Value                                                                                                           Value
               Quantity                                                                                                        Quantity


              * Model*                                                                                                        * Model*
  Weather + FE + Inputs                                                                                           Weather + FE + Inputs
          Weather + FE                                                                                                    Weather + FE
               Weather                                                                                                         Weather

                          0   4   8   12              16            20        24    28   32   36                                          0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                             49
                                      Figure A2: Specification Charts for Median Daily Rainfall

                                                       Ethiopia                                                                                                            Malawi
                                       p < 0.05             n.s.         p < 0.05                                                                          p < 0.05          n.s.        p < 0.05

                                                                                                   .15                                                                                                             1

                                                                                                   .1
                                                                                                                                                                                                                   0
                                                                                                   .05

                                                                                                   0                                                                                                               -1

                                                                                                   -.05                                                                                                            -2

    *Weather Product*                                                                              -.1              *Weather Product*
                                                                                                                                                                                                                   -3
              TAMSAT                                                                                                          TAMSAT
                 ERA5                                                                                                            ERA5
                 ARC2                                                                                                            ARC2
             MERRA-2                                                                                                         MERRA-2
                  CPC                                                                                                             CPC
               CHIRPS                                                                                                          CHIRPS


            *Dep. Var.*                                                                                                     *Dep. Var.*
                 Value                                                                                                           Value
               Quantity                                                                                                        Quantity


              * Model*                                                                                                        * Model*
  Weather + FE + Inputs                                                                                           Weather + FE + Inputs
          Weather + FE                                                                                                    Weather + FE
               Weather                                                                                                         Weather

                          0   4   8   12              16            20        24    28   32   36                                          0   4   8   12              16            20        24    28   32   36



                                                           Niger                                                                                                       Nigeria
                                           p < 0.05          n.s.        p < 0.05                                                                      p < 0.05              n.s.        p < 0.05

                                                                                                       1                                                                                                           .1

                                                                                                       .5                                                                                                          0

                                                                                                       0
                                                                                                                                                                                                                   -.1
                                                                                                       -.5
                                                                                                                                                                                                                   -.2
    *Weather Product*                                                                                  -1           *Weather Product*
              TAMSAT                                                                                                          TAMSAT
                 ERA5                                                                                                            ERA5
                 ARC2                                                                                                            ARC2
             MERRA-2                                                                                                         MERRA-2
                  CPC                                                                                                             CPC
               CHIRPS                                                                                                          CHIRPS


            *Dep. Var.*                                                                                                     *Dep. Var.*
                 Value                                                                                                           Value
               Quantity                                                                                                        Quantity


              * Model*                                                                                                        * Model*
  Weather + FE + Inputs                                                                                           Weather + FE + Inputs
          Weather + FE                                                                                                    Weather + FE
               Weather                                                                                                         Weather

                          0   4   8   12              16            20        24    28   32   36                                          0   4   8   12              16            20        24    28   32   36



                                                       Tanzania                                                                                                        Uganda
                                           p < 0.05          n.s.        p < 0.05                                                                      p < 0.05              n.s.        p < 0.05

                                                                                                       .2                                                                                                          .3

                                                                                                                                                                                                                   .2
                                                                                                       .1
                                                                                                                                                                                                                   .1
                                                                                                       0
                                                                                                                                                                                                                   0

                                                                                                       -.1                                                                                                         -.1

    *Weather Product*                                                                                               *Weather Product*                                                                              -.2
              TAMSAT                                                                                                          TAMSAT
                 ERA5                                                                                                            ERA5
                 ARC2                                                                                                            ARC2
             MERRA-2                                                                                                         MERRA-2
                  CPC                                                                                                             CPC
               CHIRPS                                                                                                          CHIRPS


            *Dep. Var.*                                                                                                     *Dep. Var.*
                 Value                                                                                                           Value
               Quantity                                                                                                        Quantity


              * Model*                                                                                                        * Model*
  Weather + FE + Inputs                                                                                           Weather + FE + Inputs
          Weather + FE                                                                                                    Weather + FE
               Weather                                                                                                         Weather

                          0   4   8   12              16            20        24    28   32   36                                          0   4   8   12              16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                             50
                                      Figure A3: Specification Charts for Variance of Daily Rainfall

                                                         Ethiopia                                                                                                            Malawi
                                             p < 0.05          n.s.        p < 0.05                                                                          p < 0.05           n.s.        p < 0.05

                                                                                                         .1                                                                                                               .1

                                                                                                                                                                                                                          0
                                                                                                         0
                                                                                                                                                                                                                          -.1
                                                                                                         -.1
                                                                                                                                                                                                                          -.2

                                                                                                         -.2                                                                                                              -.3
    *Weather Product*                                                                                                 *Weather Product*
              TAMSAT                                                                                                            TAMSAT
                 ERA5                                                                                                              ERA5
                 ARC2                                                                                                              ARC2
             MERRA-2                                                                                                           MERRA-2
                  CPC                                                                                                               CPC
               CHIRPS                                                                                                            CHIRPS


            *Dep. Var.*                                                                                                       *Dep. Var.*
                 Value                                                                                                             Value
               Quantity                                                                                                          Quantity


              * Model*                                                                                                          * Model*
  Weather + FE + Inputs                                                                                             Weather + FE + Inputs
          Weather + FE                                                                                                      Weather + FE
               Weather                                                                                                           Weather

                          0   4   8     12              16            20        24    28   32   36                                          0   4   8   12              16             20        24    28   32   36



                                                             Niger                                                                                                           Nigeria
                                             p < 0.05          n.s.        p < 0.05                                                                          p < 0.05           n.s.        p < 0.05

                                                                                                         1                                                                                                                .2

                                                                                                                                                                                                                          .1
                                                                                                         .5

                                                                                                                                                                                                                          0
                                                                                                         0
                                                                                                                                                                                                                          -.1

    *Weather Product*                                                                                                 *Weather Product*
                                                                                                         -.5
              TAMSAT                                                                                                            TAMSAT
                 ERA5                                                                                                              ERA5
                 ARC2                                                                                                              ARC2
             MERRA-2                                                                                                           MERRA-2
                  CPC                                                                                                               CPC
               CHIRPS                                                                                                            CHIRPS


            *Dep. Var.*                                                                                                       *Dep. Var.*
                 Value                                                                                                             Value
               Quantity                                                                                                          Quantity


              * Model*                                                                                                          * Model*
  Weather + FE + Inputs                                                                                             Weather + FE + Inputs
          Weather + FE                                                                                                      Weather + FE
               Weather                                                                                                           Weather

                          0   4   8     12              16            20        24    28   32   36                                          0   4   8   12              16             20        24    28   32   36



                                                        Tanzania                                                                                                         Uganda
                                         p < 0.05             n.s.         p < 0.05                                                                      p < 0.05              n.s.         p < 0.05

                                                                                                     .05                                                                                                              .2
                                                                                                                                                                                                                      .15

                                                                                                     0                                                                                                                .1
                                                                                                                                                                                                                      .05

                                                                                                     -.05                                                                                                             0
                                                                                                                                                                                                                      -.05
    *Weather Product*                                                                                                 *Weather Product*
              TAMSAT                                                                                 -.1                        TAMSAT
                 ERA5                                                                                                              ERA5
                 ARC2                                                                                                              ARC2
             MERRA-2                                                                                                           MERRA-2
                  CPC                                                                                                               CPC
               CHIRPS                                                                                                            CHIRPS


            *Dep. Var.*                                                                                                       *Dep. Var.*
                 Value                                                                                                             Value
               Quantity                                                                                                          Quantity


              * Model*                                                                                                          * Model*
  Weather + FE + Inputs                                                                                             Weather + FE + Inputs
          Weather + FE                                                                                                      Weather + FE
               Weather                                                                                                           Weather

                          0   4   8     12              16            20        24    28   32   36                                          0   4   8   12              16             20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                               51
                                      Figure A4: Specification Charts for Skew of Daily Rainfall

                                                    Ethiopia                                                                                                   Malawi
                                        p < 0.05          n.s.        p < 0.05                                                                 p < 0.05           n.s.        p < 0.05

                                                                                                2                                                                                                       4


                                                                                                1                                                                                                       2

                                                                                                0
                                                                                                                                                                                                        0
                                                                                                -1
    *Weather Product*                                                                                       *Weather Product*                                                                           -2
              TAMSAT                                                                                                  TAMSAT
                 ERA5                                                                                                    ERA5
                 ARC2                                                                                                    ARC2
             MERRA-2                                                                                                 MERRA-2
                  CPC                                                                                                     CPC
               CHIRPS                                                                                                  CHIRPS


            *Dep. Var.*                                                                                             *Dep. Var.*
                 Value                                                                                                   Value
               Quantity                                                                                                Quantity


              * Model*                                                                                                * Model*
  Weather + FE + Inputs                                                                                   Weather + FE + Inputs
          Weather + FE                                                                                            Weather + FE
               Weather                                                                                                 Weather

                          0   4   8    12          16            20        24    28   32   36                                     0   4   8   12          16             20        24    28   32   36



                                                        Niger                                                                                                  Nigeria
                                        p < 0.05          n.s.        p < 0.05                                                                 p < 0.05           n.s.        p < 0.05

                                                                                                4                                                                                                       2


                                                                                                2                                                                                                       0

                                                                                                0
                                                                                                                                                                                                        -2

                                                                                                -2
    *Weather Product*                                                                                       *Weather Product*                                                                           -4
              TAMSAT                                                                                                  TAMSAT
                                                                                                -4
                 ERA5                                                                                                    ERA5
                 ARC2                                                                                                    ARC2
             MERRA-2                                                                                                 MERRA-2
                  CPC                                                                                                     CPC
               CHIRPS                                                                                                  CHIRPS


            *Dep. Var.*                                                                                             *Dep. Var.*
                 Value                                                                                                   Value
               Quantity                                                                                                Quantity


              * Model*                                                                                                * Model*
  Weather + FE + Inputs                                                                                   Weather + FE + Inputs
          Weather + FE                                                                                            Weather + FE
               Weather                                                                                                 Weather

                          0   4   8    12          16            20        24    28   32   36                                     0   4   8   12          16             20        24    28   32   36



                                                   Tanzania                                                                                                Uganda
                                        p < 0.05          n.s.        p < 0.05                                                                 p < 0.05           n.s.        p < 0.05

                                                                                                2                                                                                                       6


                                                                                                1                                                                                                       4

                                                                                                                                                                                                        2
                                                                                                0

                                                                                                                                                                                                        0
                                                                                                -1
    *Weather Product*                                                                                       *Weather Product*
                                                                                                                                                                                                        -2
              TAMSAT                                                                            -2                    TAMSAT
                 ERA5                                                                                                    ERA5
                 ARC2                                                                                                    ARC2
             MERRA-2                                                                                                 MERRA-2
                  CPC                                                                                                     CPC
               CHIRPS                                                                                                  CHIRPS


            *Dep. Var.*                                                                                             *Dep. Var.*
                 Value                                                                                                   Value
               Quantity                                                                                                Quantity


              * Model*                                                                                                * Model*
  Weather + FE + Inputs                                                                                   Weather + FE + Inputs
          Weather + FE                                                                                            Weather + FE
               Weather                                                                                                 Weather

                          0   4   8    12          16            20        24    28   32   36                                     0   4   8   12          16             20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                     52
                                      Figure A5: Specification Charts for Deviation in Total Rainfall

                                                      Ethiopia                                                                                                          Malawi
                                          p < 0.05         n.s.         p < 0.05                                                                        p < 0.05          n.s.        p < 0.05

                                                                                                  .0015                                                                                                             .002
                                                                                                  .001
                                                                                                                                                                                                                    .001
                                                                                                  .0005
                                                                                                                                                                                                                    0
                                                                                                  0
                                                                                                                                                                                                                    -.001
                                                                                                  -.0005
                                                                                                                                                                                                                    -.002
                                                                                                  -.001
    *Weather Product*                                                                                                *Weather Product*
              TAMSAT                                                                                                           TAMSAT
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8     12           16            20       24     28   32   36                                            0   4   8   12          16            20        24    28   32   36



                                                          Niger                                                                                                     Nigeria
                                          p < 0.05          n.s.        p < 0.05                                                                        p < 0.05          n.s.        p < 0.05

                                                                                                      .005                                                                                                      .0015
                                                                                                                                                                                                                .001

                                                                                                      0                                                                                                         .0005
                                                                                                                                                                                                                0
                                                                                                      -.005                                                                                                     -.0005
                                                                                                                                                                                                                -.001
    *Weather Product*                                                                                                *Weather Product*
                                                                                                      -.01
              TAMSAT                                                                                                           TAMSAT
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8      12          16            20        24    28   32   36                                            0   4   8   12          16            20       24     28   32   36



                                                     Tanzania                                                                                                       Uganda
                                          p < 0.05         n.s.         p < 0.05                                                                        p < 0.05          n.s.        p < 0.05

                                                                                                  .001                                                                                                              .003

                                                                                                                                                                                                                    .002
                                                                                                  .0005
                                                                                                                                                                                                                    .001

                                                                                                                                                                                                                    0
                                                                                                  0
                                                                                                                                                                                                                    -.001
                                                                                                  -.0005                                                                                                            -.002
    *Weather Product*                                                                                                *Weather Product*
              TAMSAT                                                                                                           TAMSAT
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8     12           16            20       24     28   32   36                                            0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                              53
                                      Figure A6: Specification Charts for z-Score of Total Rainfall

                                                     Ethiopia                                                                                                    Malawi
                                         p < 0.05          n.s.        p < 0.05                                                                  p < 0.05          n.s.        p < 0.05

                                                                                                 .4                                                                                                      .4

                                                                                                                                                                                                         .2
                                                                                                 .2
                                                                                                                                                                                                         0
                                                                                                 0
                                                                                                                                                                                                         -.2

    *Weather Product*                                                                            -.2          *Weather Product*                                                                          -.4
              TAMSAT                                                                                                    TAMSAT
                 ERA5                                                                                                      ERA5
                 ARC2                                                                                                      ARC2
             MERRA-2                                                                                                   MERRA-2
                  CPC                                                                                                       CPC
               CHIRPS                                                                                                    CHIRPS


            *Dep. Var.*                                                                                               *Dep. Var.*
                 Value                                                                                                     Value
               Quantity                                                                                                  Quantity


              * Model*                                                                                                  * Model*
  Weather + FE + Inputs                                                                                     Weather + FE + Inputs
          Weather + FE                                                                                              Weather + FE
               Weather                                                                                                   Weather

                          0   4   8     12          16            20        24    28   32   36                                      0   4   8   12          16            20        24    28   32   36



                                                         Niger                                                                                               Nigeria
                                         p < 0.05          n.s.        p < 0.05                                                                  p < 0.05          n.s.        p < 0.05

                                                                                                 .4                                                                                                      .2

                                                                                                 .2                                                                                                      .1
                                                                                                 0
                                                                                                                                                                                                         0
                                                                                                 -.2
                                                                                                                                                                                                         -.1
                                                                                                 -.4

    *Weather Product*                                                                            -.6          *Weather Product*                                                                          -.2
              TAMSAT                                                                                                    TAMSAT
                 ERA5                                                                                                      ERA5
                 ARC2                                                                                                      ARC2
             MERRA-2                                                                                                   MERRA-2
                  CPC                                                                                                       CPC
               CHIRPS                                                                                                    CHIRPS


            *Dep. Var.*                                                                                               *Dep. Var.*
                 Value                                                                                                     Value
               Quantity                                                                                                  Quantity


              * Model*                                                                                                  * Model*
  Weather + FE + Inputs                                                                                     Weather + FE + Inputs
          Weather + FE                                                                                              Weather + FE
               Weather                                                                                                   Weather

                          0   4   8     12          16            20        24    28   32   36                                      0   4   8   12          16            20        24    28   32   36



                                                    Tanzania                                                                                                 Uganda
                                         p < 0.05          n.s.        p < 0.05                                                                  p < 0.05          n.s.        p < 0.05

                                                                                                 .2                                                                                                      .4

                                                                                                 .1                                                                                                      .2

                                                                                                 0                                                                                                       0

                                                                                                 -.1                                                                                                     -.2
    *Weather Product*                                                                                         *Weather Product*
              TAMSAT                                                                                                    TAMSAT
                 ERA5                                                                                                      ERA5
                 ARC2                                                                                                      ARC2
             MERRA-2                                                                                                   MERRA-2
                  CPC                                                                                                       CPC
               CHIRPS                                                                                                    CHIRPS


            *Dep. Var.*                                                                                               *Dep. Var.*
                 Value                                                                                                     Value
               Quantity                                                                                                  Quantity


              * Model*                                                                                                  * Model*
  Weather + FE + Inputs                                                                                     Weather + FE + Inputs
          Weather + FE                                                                                              Weather + FE
               Weather                                                                                                   Weather

                          0   4   8     12          16            20        24    28   32   36                                      0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                       54
                                      Figure A7: Specification Charts for Number Days with Rain

                                                        Ethiopia                                                                                                             Malawi
                                            p < 0.05          n.s.        p < 0.05                                                                           p < 0.05           n.s.        p < 0.05

                                                                                                        .02
                                                                                                                                                                                                                          .02

                                                                                                        .01                                                                                                               .01

                                                                                                                                                                                                                          0
                                                                                                        0
                                                                                                                                                                                                                          -.01

                                                                                                        -.01                                                                                                              -.02
    *Weather Product*                                                                                                 *Weather Product*
              TAMSAT                                                                                                            TAMSAT
                                                                                                        -.02
                 ERA5                                                                                                              ERA5
                 ARC2                                                                                                              ARC2
             MERRA-2                                                                                                           MERRA-2
                  CPC                                                                                                               CPC
               CHIRPS                                                                                                            CHIRPS


            *Dep. Var.*                                                                                                       *Dep. Var.*
                 Value                                                                                                             Value
               Quantity                                                                                                          Quantity


              * Model*                                                                                                          * Model*
  Weather + FE + Inputs                                                                                             Weather + FE + Inputs
          Weather + FE                                                                                                      Weather + FE
               Weather                                                                                                           Weather

                          0   4   8    12              16            20        24    28   32   36                                           0   4   8   12              16             20        24    28   32   36



                                                            Niger                                                                                                            Nigeria
                                            p < 0.05          n.s.        p < 0.05                                                                           p < 0.05           n.s.        p < 0.05

                                                                                                        .05
                                                                                                                                                                                                                          .02

                                                                                                                                                                                                                          .01
                                                                                                        0
                                                                                                                                                                                                                          0

                                                                                                                                                                                                                          -.01

                                                                                                        -.05                                                                                                              -.02
    *Weather Product*                                                                                                 *Weather Product*
              TAMSAT                                                                                                            TAMSAT
                 ERA5                                                                                                              ERA5
                 ARC2                                                                                                              ARC2
             MERRA-2                                                                                                           MERRA-2
                  CPC                                                                                                               CPC
               CHIRPS                                                                                                            CHIRPS


            *Dep. Var.*                                                                                                       *Dep. Var.*
                 Value                                                                                                             Value
               Quantity                                                                                                          Quantity


              * Model*                                                                                                          * Model*
  Weather + FE + Inputs                                                                                             Weather + FE + Inputs
          Weather + FE                                                                                                      Weather + FE
               Weather                                                                                                           Weather

                          0   4   8    12              16            20        24    28   32   36                                           0   4   8   12              16             20        24    28   32   36



                                                       Tanzania                                                                                                          Uganda
                                        p < 0.05             n.s.         p < 0.05                                                                       p < 0.05              n.s.         p < 0.05

                                                                                                    .015                                                                                                              .01

                                                                                                    .01                                                                                                               .005

                                                                                                                                                                                                                      0
                                                                                                    .005
                                                                                                                                                                                                                      -.005
                                                                                                    0
                                                                                                                                                                                                                      -.01

    *Weather Product*                                                                               -.005             *Weather Product*                                                                               -.015
              TAMSAT                                                                                                            TAMSAT
                 ERA5                                                                                                              ERA5
                 ARC2                                                                                                              ARC2
             MERRA-2                                                                                                           MERRA-2
                  CPC                                                                                                               CPC
               CHIRPS                                                                                                            CHIRPS


            *Dep. Var.*                                                                                                       *Dep. Var.*
                 Value                                                                                                             Value
               Quantity                                                                                                          Quantity


              * Model*                                                                                                          * Model*
  Weather + FE + Inputs                                                                                             Weather + FE + Inputs
          Weather + FE                                                                                                      Weather + FE
               Weather                                                                                                           Weather

                          0   4   8    12              16            20        24    28   32   36                                           0   4   8   12              16             20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                               55
                                  Figure A8: Specification Charts for Deviations in Days with Rain

                                                       Ethiopia                                                                                                         Malawi
                                           p < 0.05          n.s.        p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                       .02                                                                                                      .04


                                                                                                       .01                                                                                                      .02


                                                                                                       0                                                                                                        0


                                                                                                       -.01                                                                                                     -.02
    *Weather Product*                                                                                                *Weather Product*
              TAMSAT                                                                                   -.02                    TAMSAT                                                                           -.04
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



                                                           Niger                                                                                                    Nigeria
                                           p < 0.05          n.s.        p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                       .04                                                                                                      .03

                                                                                                       .02                                                                                                      .02
                                                                                                       0                                                                                                        .01
                                                                                                       -.02                                                                                                     0
                                                                                                       -.04
                                                                                                                                                                                                                -.01
                                                                                                       -.06
    *Weather Product*                                                                                                *Weather Product*                                                                          -.02

              TAMSAT                                                                                                           TAMSAT
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



                                                      Tanzania                                                                                                      Uganda
                                       p < 0.05             n.s.         p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                   .015                                                                                                         .04

                                                                                                   .01
                                                                                                                                                                                                                .02
                                                                                                   .005
                                                                                                                                                                                                                0
                                                                                                   0

                                                                                                   -.005                                                                                                        -.02
    *Weather Product*                                                                              -.01              *Weather Product*
              TAMSAT                                                                                                           TAMSAT                                                                           -.04
                 ERA5                                                                                                             ERA5
                 ARC2                                                                                                             ARC2
             MERRA-2                                                                                                          MERRA-2
                  CPC                                                                                                              CPC
               CHIRPS                                                                                                           CHIRPS


            *Dep. Var.*                                                                                                      *Dep. Var.*
                 Value                                                                                                            Value
               Quantity                                                                                                         Quantity


              * Model*                                                                                                         * Model*
  Weather + FE + Inputs                                                                                            Weather + FE + Inputs
          Weather + FE                                                                                                     Weather + FE
               Weather                                                                                                          Weather

                          0   4   8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                              56
                                  Figure A9: Specification Charts for Deviations in Days with No Rain

                                                         Ethiopia                                                                                                         Malawi
                                             p < 0.05          n.s.        p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                         .02                                                                                                      .04


                                                                                                         .01                                                                                                      .02

                                                                                                         0                                                                                                        0

                                                                                                         -.01                                                                                                     -.02
    *Weather Product*                                                                                                  *Weather Product*
              TAMSAT                                                                                     -.02                    TAMSAT                                                                           -.04
                 ERA5                                                                                                               ERA5
                 ARC2                                                                                                               ARC2
             MERRA-2                                                                                                            MERRA-2
                  CPC                                                                                                                CPC
               CHIRPS                                                                                                             CHIRPS


            *Dep. Var.*                                                                                                        *Dep. Var.*
                 Value                                                                                                              Value
               Quantity                                                                                                           Quantity


              * Model*                                                                                                           * Model*
  Weather + FE + Inputs                                                                                              Weather + FE + Inputs
          Weather + FE                                                                                                       Weather + FE
               Weather                                                                                                            Weather

                          0   4     8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



                                                             Niger                                                                                                    Nigeria
                                             p < 0.05          n.s.        p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                         .06                                                                                                      .02

                                                                                                         .04                                                                                                      .01
                                                                                                         .02                                                                                                      0
                                                                                                         0
                                                                                                                                                                                                                  -.01
                                                                                                         -.02
                                                                                                                                                                                                                  -.02
                                                                                                         -.04
    *Weather Product*                                                                                                  *Weather Product*                                                                          -.03
              TAMSAT                                                                                                             TAMSAT
                 ERA5                                                                                                               ERA5
                 ARC2                                                                                                               ARC2
             MERRA-2                                                                                                            MERRA-2
                  CPC                                                                                                                CPC
               CHIRPS                                                                                                             CHIRPS


            *Dep. Var.*                                                                                                        *Dep. Var.*
                 Value                                                                                                              Value
               Quantity                                                                                                           Quantity


              * Model*                                                                                                           * Model*
  Weather + FE + Inputs                                                                                              Weather + FE + Inputs
          Weather + FE                                                                                                       Weather + FE
               Weather                                                                                                            Weather

                          0   4     8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



                                                        Tanzania                                                                                                      Uganda
                                         p < 0.05             n.s.         p < 0.05                                                                       p < 0.05          n.s.        p < 0.05

                                                                                                                                                                                                                  .04
                                                                                                     .005

                                                                                                     0                                                                                                            .02

                                                                                                     -.005
                                                                                                                                                                                                                  0
                                                                                                     -.01
                                                                                                                                                                                                                  -.02
                                                                                                     -.015
    *Weather Product*                                                                                                  *Weather Product*
                                                                                                                                                                                                                  -.04
              TAMSAT                                                                                                             TAMSAT
                 ERA5                                                                                                               ERA5
                 ARC2                                                                                                               ARC2
             MERRA-2                                                                                                            MERRA-2
                  CPC                                                                                                                CPC
               CHIRPS                                                                                                             CHIRPS


            *Dep. Var.*                                                                                                        *Dep. Var.*
                 Value                                                                                                              Value
               Quantity                                                                                                           Quantity


              * Model*                                                                                                           * Model*
  Weather + FE + Inputs                                                                                              Weather + FE + Inputs
          Weather + FE                                                                                                       Weather + FE
               Weather                                                                                                            Weather

                          0   4     8   12              16            20        24    28   32   36                                           0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                                57
                                  Figure A10: Specification Charts for Percentage of Days with Rain

                                                    Ethiopia                                                                                                   Malawi
                                        p < 0.05          n.s.        p < 0.05                                                                 p < 0.05           n.s.        p < 0.05

                                                                                                3                                                                                                       4
                                                                                                2
                                                                                                                                                                                                        2
                                                                                                1
                                                                                                                                                                                                        0
                                                                                                0
                                                                                                                                                                                                        -2
                                                                                                -1
                                                                                                                                                                                                        -4
    *Weather Product*                                                                           -2          *Weather Product*
              TAMSAT                                                                                                  TAMSAT
                 ERA5                                                                                                    ERA5
                 ARC2                                                                                                    ARC2
             MERRA-2                                                                                                 MERRA-2
                  CPC                                                                                                     CPC
               CHIRPS                                                                                                  CHIRPS


            *Dep. Var.*                                                                                             *Dep. Var.*
                 Value                                                                                                   Value
               Quantity                                                                                                Quantity


              * Model*                                                                                                * Model*
  Weather + FE + Inputs                                                                                   Weather + FE + Inputs
          Weather + FE                                                                                            Weather + FE
               Weather                                                                                                 Weather

                          0   4    8   12          16            20        24    28   32   36                                     0   4   8   12          16             20        24    28   32   36



                                                        Niger                                                                                                  Nigeria
                                        p < 0.05          n.s.        p < 0.05                                                                 p < 0.05           n.s.        p < 0.05

                                                                                                                                                                                                        3
                                                                                                5
                                                                                                                                                                                                        2
                                                                                                                                                                                                        1
                                                                                                0                                                                                                       0
                                                                                                                                                                                                        -1
                                                                                                -5                                                                                                      -2
    *Weather Product*                                                                                       *Weather Product*
              TAMSAT                                                                                                  TAMSAT
                 ERA5                                                                                                    ERA5
                 ARC2                                                                                                    ARC2
             MERRA-2                                                                                                 MERRA-2
                  CPC                                                                                                     CPC
               CHIRPS                                                                                                  CHIRPS


            *Dep. Var.*                                                                                             *Dep. Var.*
                 Value                                                                                                   Value
               Quantity                                                                                                Quantity


              * Model*                                                                                                * Model*
  Weather + FE + Inputs                                                                                   Weather + FE + Inputs
          Weather + FE                                                                                            Weather + FE
               Weather                                                                                                 Weather

                          0   4    8   12          16            20        24    28   32   36                                     0   4   8   12          16             20        24    28   32   36



                                                   Tanzania                                                                                                Uganda
                                        p < 0.05          n.s.        p < 0.05                                                                 p < 0.05           n.s.        p < 0.05

                                                                                                2                                                                                                       4


                                                                                                1                                                                                                       2

                                                                                                0                                                                                                       0

                                                                                                -1
                                                                                                                                                                                                        -2
    *Weather Product*                                                                                       *Weather Product*
              TAMSAT                                                                                                  TAMSAT
                 ERA5                                                                                                    ERA5
                 ARC2                                                                                                    ARC2
             MERRA-2                                                                                                 MERRA-2
                  CPC                                                                                                     CPC
               CHIRPS                                                                                                  CHIRPS


            *Dep. Var.*                                                                                             *Dep. Var.*
                 Value                                                                                                   Value
               Quantity                                                                                                Quantity


              * Model*                                                                                                * Model*
  Weather + FE + Inputs                                                                                   Weather + FE + Inputs
          Weather + FE                                                                                            Weather + FE
               Weather                                                                                                 Weather

                          0   4    8   12          16            20        24    28   32   36                                     0   4   8   12          16             20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                     58
                          Figure A11: Specification Charts for Deviations in Percentage of Days with Rain

                                                       Ethiopia                                                                                                       Malawi
                                           p < 0.05          n.s.        p < 0.05                                                                     p < 0.05           n.s.        p < 0.05

                                                                                                       3
                                                                                                                                                                                                               5
                                                                                                       2

                                                                                                       1
                                                                                                                                                                                                               0
                                                                                                       0

                                                                                                       -1
                                                                                                                                                                                                               -5
    *Weather Product*                                                                                  -2          *Weather Product*
              TAMSAT                                                                                                         TAMSAT
                 ERA5                                                                                                           ERA5
                 ARC2                                                                                                           ARC2
             MERRA-2                                                                                                        MERRA-2
                  CPC                                                                                                            CPC
               CHIRPS                                                                                                         CHIRPS


            *Dep. Var.*                                                                                                    *Dep. Var.*
                 Value                                                                                                          Value
               Quantity                                                                                                       Quantity


              * Model*                                                                                                       * Model*
  Weather + FE + Inputs                                                                                          Weather + FE + Inputs
          Weather + FE                                                                                                   Weather + FE
               Weather                                                                                                        Weather

                          0   4   8   12              16            20        24    28   32   36                                         0   4   8   12          16             20        24    28   32   36



                                                           Niger                                                                                                      Nigeria
                                       p < 0.05              n.s.        p < 0.05                                                                     p < 0.05           n.s.        p < 0.05

                                                                                                   5                                                                                                           4

                                                                                                                                                                                                               2
                                                                                                   0

                                                                                                                                                                                                               0
                                                                                                   -5
                                                                                                                                                                                                               -2
    *Weather Product*                                                                                              *Weather Product*
                                                                                                   -10
              TAMSAT                                                                                                         TAMSAT                                                                            -4
                 ERA5                                                                                                           ERA5
                 ARC2                                                                                                           ARC2
             MERRA-2                                                                                                        MERRA-2
                  CPC                                                                                                            CPC
               CHIRPS                                                                                                         CHIRPS


            *Dep. Var.*                                                                                                    *Dep. Var.*
                 Value                                                                                                          Value
               Quantity                                                                                                       Quantity


              * Model*                                                                                                       * Model*
  Weather + FE + Inputs                                                                                          Weather + FE + Inputs
          Weather + FE                                                                                                   Weather + FE
               Weather                                                                                                        Weather

                          0   4   8   12              16            20        24    28   32   36                                         0   4   8   12          16             20        24    28   32   36



                                                       Tanzania                                                                                                   Uganda
                                           p < 0.05          n.s.        p < 0.05                                                                     p < 0.05           n.s.        p < 0.05

                                                                                                       3                                                                                                       4

                                                                                                       2                                                                                                       2

                                                                                                       1                                                                                                       0

                                                                                                       0                                                                                                       -2

                                                                                                       -1                                                                                                      -4
    *Weather Product*                                                                                              *Weather Product*
              TAMSAT                                                                                                         TAMSAT
                 ERA5                                                                                                           ERA5
                 ARC2                                                                                                           ARC2
             MERRA-2                                                                                                        MERRA-2
                  CPC                                                                                                            CPC
               CHIRPS                                                                                                         CHIRPS


            *Dep. Var.*                                                                                                    *Dep. Var.*
                 Value                                                                                                          Value
               Quantity                                                                                                       Quantity


              * Model*                                                                                                       * Model*
  Weather + FE + Inputs                                                                                          Weather + FE + Inputs
          Weather + FE                                                                                                   Weather + FE
               Weather                                                                                                        Weather

                          0   4   8   12              16            20        24    28   32   36                                         0   4   8   12          16             20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                            59
                                       Figure A12: Specification Charts for Longest Dry Spell

                                                   Ethiopia                                                                                                     Malawi
                                       p < 0.05          n.s.        p < 0.05                                                                   p < 0.05          n.s.        p < 0.05

                                                                                               .04                                                                                                      .06
                                                                                                                                                                                                        .04
                                                                                               .02
                                                                                                                                                                                                        .02
                                                                                                                                                                                                        0
                                                                                               0
                                                                                                                                                                                                        -.02

                                                                                               -.02                                                                                                     -.04
    *Weather Product*                                                                                        *Weather Product*
              TAMSAT                                                                           -.04                    TAMSAT
                 ERA5                                                                                                     ERA5
                 ARC2                                                                                                     ARC2
             MERRA-2                                                                                                  MERRA-2
                  CPC                                                                                                      CPC
               CHIRPS                                                                                                   CHIRPS


            *Dep. Var.*                                                                                              *Dep. Var.*
                 Value                                                                                                    Value
               Quantity                                                                                                 Quantity


              * Model*                                                                                                 * Model*
  Weather + FE + Inputs                                                                                    Weather + FE + Inputs
          Weather + FE                                                                                             Weather + FE
               Weather                                                                                                  Weather

                          0   4   8   12          16            20        24    28   32   36                                       0   4   8   12          16            20        24    28   32   36



                                                       Niger                                                                                                Nigeria
                                       p < 0.05          n.s.        p < 0.05                                                                   p < 0.05          n.s.        p < 0.05

                                                                                               .1                                                                                                       .1

                                                                                               .05                                                                                                      .05
                                                                                               0
                                                                                                                                                                                                        0
                                                                                               -.05

                                                                                               -.1                                                                                                      -.05
    *Weather Product*                                                                                        *Weather Product*
              TAMSAT                                                                                                   TAMSAT                                                                           -.1
                 ERA5                                                                                                     ERA5
                 ARC2                                                                                                     ARC2
             MERRA-2                                                                                                  MERRA-2
                  CPC                                                                                                      CPC
               CHIRPS                                                                                                   CHIRPS


            *Dep. Var.*                                                                                              *Dep. Var.*
                 Value                                                                                                    Value
               Quantity                                                                                                 Quantity


              * Model*                                                                                                 * Model*
  Weather + FE + Inputs                                                                                    Weather + FE + Inputs
          Weather + FE                                                                                             Weather + FE
               Weather                                                                                                  Weather

                          0   4   8   12          16            20        24    28   32   36                                       0   4   8   12          16            20        24    28   32   36



                                                  Tanzania                                                                                                  Uganda
                                       p < 0.05          n.s.        p < 0.05                                                                   p < 0.05          n.s.        p < 0.05

                                                                                               .03                                                                                                      .04

                                                                                               .02                                                                                                      .02

                                                                                               .01                                                                                                      0

                                                                                               0                                                                                                        -.02

                                                                                               -.01                                                                                                     -.04

    *Weather Product*                                                                          -.02          *Weather Product*                                                                          -.06
              TAMSAT                                                                                                   TAMSAT
                 ERA5                                                                                                     ERA5
                 ARC2                                                                                                     ARC2
             MERRA-2                                                                                                  MERRA-2
                  CPC                                                                                                      CPC
               CHIRPS                                                                                                   CHIRPS


            *Dep. Var.*                                                                                              *Dep. Var.*
                 Value                                                                                                    Value
               Quantity                                                                                                 Quantity


              * Model*                                                                                                 * Model*
  Weather + FE + Inputs                                                                                    Weather + FE + Inputs
          Weather + FE                                                                                             Weather + FE
               Weather                                                                                                  Weather

                          0   4   8   12          16            20        24    28   32   36                                       0   4   8   12          16            20        24    28   32   36



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 36 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                      60
                                  Figure A13: Specification Charts for Median Daily Temperature

                                                         Ethiopia                                                                                                   Malawi
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  .2
                                                                                                                                                                                                             .1
                                                                                                  .1
                                                                                                  0                                                                                                          0

                                                                                                  -.1                                                                                                        -.1
                                                                                                  -.2
    *Weather Product*                                                                                          *Weather Product*                                                                             -.2
                                                                                                  -.3
             MERRA-2                                                                                                    MERRA-2                                                                              -.3
                 ERA5                                                                                                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                          Niger                                                                                                     Nigeria
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  1                                                                                                          .6

                                                                                                                                                                                                             .4
                                                                                                  .5
                                                                                                                                                                                                             .2
                                                                                                  0
                                                                                                                                                                                                             0
    *Weather Product*                                                                                          *Weather Product*
                                                                                                  -.5                                                                                                        -.2
             MERRA-2                                                                                                    MERRA-2

                 ERA5                                                                             -1                        ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                     Tanzania                                                                                                       Uganda
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  .2                                                                                                         .2

                                                                                                  .1
                                                                                                                                                                                                             0
                                                                                                  0

                                                                                                  -.1                                                                                                        -.2
    *Weather Product*                                                                                          *Weather Product*
                                                                                                  -.2
             MERRA-2                                                                                                    MERRA-2                                                                              -.4

                 ERA5                                                                                                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                        61
                                  Figure A14: Specification Charts for Variance of Daily Temperature

                                                           Ethiopia                                                                                                   Malawi
                                            p < 0.05          n.s.        p < 0.05                                                                     p < 0.05         n.s.         p < 0.05

                                                                                                    .4                                                                                                         .5

                                                                                                    .2                                                                                                         0

                                                                                                    0
                                                                                                                                                                                                               -.5

    *Weather Product*                                                                               -.2          *Weather Product*
                                                                                                                                                                                                               -1
             MERRA-2                                                                                                      MERRA-2
                                                                                                    -.4
                 ERA5                                                                                                         ERA5                                                                             -1.5

                  CPC                                                                                                          CPC



            *Dep. Var.*                                                                                                  *Dep. Var.*

                 Value                                                                                                        Value

               Quantity                                                                                                     Quantity



              * Model*                                                                                                     * Model*

  Weather + FE + Inputs                                                                                        Weather + FE + Inputs

          Weather + FE                                                                                                 Weather + FE

               Weather                                                                                                      Weather

                          0   2     4   6              8             10        12    14   16   18                                      0   2   4   6              8             10       12     14   16   18



                                                            Niger                                                                                                     Nigeria
                                            p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                    3                                                                                                              1
                                                                                                    2
                                                                                                                                                                                                                   .5
                                                                                                    1
                                                                                                    0                                                                                                              0
                                                                                                    -1
    *Weather Product*                                                                                            *Weather Product*
                                                                                                    -2                                                                                                             -.5
             MERRA-2                                                                                                      MERRA-2

                 ERA5                                                                                                         ERA5

                  CPC                                                                                                          CPC



            *Dep. Var.*                                                                                                  *Dep. Var.*

                 Value                                                                                                        Value

               Quantity                                                                                                     Quantity



              * Model*                                                                                                     * Model*

  Weather + FE + Inputs                                                                                        Weather + FE + Inputs

          Weather + FE                                                                                                 Weather + FE

               Weather                                                                                                      Weather

                          0   2     4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                       Tanzania                                                                                                       Uganda
                                            p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                    .5                                                                                                             .4
                                                                                                                                                                                                                   .2

                                                                                                    0                                                                                                              0
                                                                                                                                                                                                                   -.2

    *Weather Product*                                                                               -.5          *Weather Product*                                                                                 -.4

             MERRA-2                                                                                                      MERRA-2                                                                                  -.6

                 ERA5                                                                                                         ERA5

                  CPC                                                                                                          CPC



            *Dep. Var.*                                                                                                  *Dep. Var.*

                 Value                                                                                                        Value

               Quantity                                                                                                     Quantity



              * Model*                                                                                                     * Model*

  Weather + FE + Inputs                                                                                        Weather + FE + Inputs

          Weather + FE                                                                                                 Weather + FE

               Weather                                                                                                      Weather

                          0   2     4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                          62
                                  Figure A15: Specification Charts for Skew of Daily Temperature

                                                         Ethiopia                                                                                                   Malawi
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  1.5
                                                                                                                                                                                                                    2
                                                                                                  1
                                                                                                                                                                                                                    1
                                                                                                  .5
                                                                                                                                                                                                                    0
                                                                                                  0
    *Weather Product*                                                                                          *Weather Product*                                                                                    -1
                                                                                                  -.5
             MERRA-2                                                                                                    MERRA-2                                                                                     -2
                 ERA5                                                                                                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14    16    18



                                                          Niger                                                                                                     Nigeria
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                   6                                                                                                                2

                                                                                                   4
                                                                                                                                                                                                                    1
                                                                                                   2

                                                                                                   0                                                                                                                0
    *Weather Product*                                                                              -2          *Weather Product*
                                                                                                                                                                                                                    -1
             MERRA-2                                                                               -4                   MERRA-2

                 ERA5                                                                                                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14    16    18



                                                     Tanzania                                                                                                       Uganda
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05         n.s.         p < 0.05

                                                                                                  .5
                                                                                                                                                                                                               1

                                                                                                  0                                                                                                            .5
                                                                                                                                                                                                               0
                                                                                                  -.5
                                                                                                                                                                                                               -.5
    *Weather Product*                                                                             -1           *Weather Product*                                                                               -1
             MERRA-2                                                                                                    MERRA-2                                                                                -1.5
                 ERA5                                                                                                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10       12     14   16    18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                        63
                                      Figure A16: Specification Charts for Deviations in GDD

                                                         Ethiopia                                                                                                       Malawi
                                          p < 0.05          n.s.        p < 0.05                                                                         p < 0.05          n.s.        p < 0.05

                                                                                                  .1                                                                                                                 .2

                                                                                                                                                                                                                     .1
                                                                                                  .05

                                                                                                                                                                                                                     0
                                                                                                  0
    *Weather Product*                                                                                              *Weather Product*                                                                                 -.1
             MERRA-2                                                                                                        MERRA-2
                                                                                                  -.05                                                                                                               -.2
                 ERA5                                                                                                           ERA5

                  CPC                                                                                                            CPC



            *Dep. Var.*                                                                                                    *Dep. Var.*

                 Value                                                                                                          Value

               Quantity                                                                                                       Quantity



              * Model*                                                                                                       * Model*

  Weather + FE + Inputs                                                                                          Weather + FE + Inputs

          Weather + FE                                                                                                   Weather + FE

               Weather                                                                                                        Weather

                          0   2   4   6              8             10       12     14   16   18                                          0   2   4   6              8             10        12    14   16   18



                                                          Niger                                                                                                         Nigeria
                                          p < 0.05          n.s.        p < 0.05                                                                         p < 0.05          n.s.        p < 0.05

                                                                                                                                                                                                                 .04
                                                                                                  .04
                                                                                                  .02                                                                                                            .02
                                                                                                  0
                                                                                                                                                                                                                 0
                                                                                                  -.02
                                                                                                                                                                                                                 -.02
    *Weather Product*                                                                             -.04             *Weather Product*
                                                                                                  -.06                                                                                                           -.04
             MERRA-2                                                                                                        MERRA-2

                 ERA5                                                                                                           ERA5

                  CPC                                                                                                            CPC



            *Dep. Var.*                                                                                                    *Dep. Var.*

                 Value                                                                                                          Value

               Quantity                                                                                                       Quantity



              * Model*                                                                                                       * Model*

  Weather + FE + Inputs                                                                                          Weather + FE + Inputs

          Weather + FE                                                                                                   Weather + FE

               Weather                                                                                                        Weather

                          0   2   4   6              8             10       12     14   16   18                                          0   2   4   6              8             10       12     14   16   18



                                                     Tanzania                                                                                                           Uganda
                                          p < 0.05          n.s.        p < 0.05                                                                         p < 0.05          n.s.        p < 0.05

                                                                                                      .2                                                                                                             .2

                                                                                                                                                                                                                     0
                                                                                                      .1
                                                                                                                                                                                                                     -.2
                                                                                                      0
                                                                                                                                                                                                                     -.4
    *Weather Product*                                                                                              *Weather Product*
                                                                                                      -.1
             MERRA-2                                                                                                        MERRA-2                                                                                  -.6

                 ERA5                                                                                                           ERA5

                  CPC                                                                                                            CPC



            *Dep. Var.*                                                                                                    *Dep. Var.*

                 Value                                                                                                          Value

               Quantity                                                                                                       Quantity



              * Model*                                                                                                       * Model*

  Weather + FE + Inputs                                                                                          Weather + FE + Inputs

          Weather + FE                                                                                                   Weather + FE

               Weather                                                                                                        Weather

                          0   2   4   6              8             10        12    14   16   18                                          0   2   4   6              8             10        12    14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                            64
                                            Figure A17: Specification Charts for z-Score of GDD

                                                         Ethiopia                                                                                                   Malawi
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  .6                                                                                                         1

                                                                                                  .4
                                                                                                                                                                                                             .5
                                                                                                  .2
                                                                                                                                                                                                             0
                                                                                                  0
    *Weather Product*                                                                                          *Weather Product*
                                                                                                  -.2                                                                                                        -.5
             MERRA-2                                                                                                    MERRA-2

                 ERA5                                                                                                       ERA5
                                                                                                                                                                                                             -1
                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                          Niger                                                                                                     Nigeria
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  .4                                                                                                         .2
                                                                                                  .2
                                                                                                                                                                                                             .1
                                                                                                  0
                                                                                                                                                                                                             0
                                                                                                  -.2
                                                                                                  -.4                                                                                                        -.1
    *Weather Product*                                                                                          *Weather Product*
                                                                                                  -.6
             MERRA-2                                                                                                    MERRA-2                                                                              -.2

                 ERA5                                                                                                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



                                                     Tanzania                                                                                                       Uganda
                                          p < 0.05          n.s.        p < 0.05                                                                     p < 0.05          n.s.        p < 0.05

                                                                                                  .4                                                                                                         .2

                                                                                                                                                                                                             0
                                                                                                  .2
                                                                                                                                                                                                             -.2
                                                                                                  0                                                                                                          -.4
    *Weather Product*                                                                                          *Weather Product*                                                                             -.6
                                                                                                  -.2
             MERRA-2                                                                                                    MERRA-2                                                                              -.8
                 ERA5                                                                             -.4                       ERA5

                  CPC                                                                                                        CPC



            *Dep. Var.*                                                                                                *Dep. Var.*

                 Value                                                                                                      Value

               Quantity                                                                                                   Quantity



              * Model*                                                                                                   * Model*

  Weather + FE + Inputs                                                                                      Weather + FE + Inputs

          Weather + FE                                                                                               Weather + FE

               Weather                                                                                                    Weather

                          0   2   4   6              8             10        12    14   16   18                                      0   2   4   6              8             10        12    14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                        65
                                  Figure A18: Specification Charts for Maximum Daily Temperature

                                                          Ethiopia                                                                                                       Malawi
                                           p < 0.05          n.s.        p < 0.05                                                                         p < 0.05          n.s.        p < 0.05

                                                                                                   .1                                                                                                                 .2

                                                                                                   .05                                                                                                                .1

                                                                                                                                                                                                                      0
                                                                                                   0
                                                                                                                                                                                                                      -.1
                                                                                                   -.05
    *Weather Product*                                                                                               *Weather Product*
                                                                                                                                                                                                                      -.2
             MERRA-2                                                                               -.1                       MERRA-2

                 ERA5                                                                                                            ERA5

                  CPC                                                                                                             CPC



            *Dep. Var.*                                                                                                     *Dep. Var.*

                 Value                                                                                                           Value

               Quantity                                                                                                        Quantity



              * Model*                                                                                                        * Model*

  Weather + FE + Inputs                                                                                           Weather + FE + Inputs

          Weather + FE                                                                                                    Weather + FE

               Weather                                                                                                         Weather

                          0   2    4   6              8             10       12     14   16   18                                          0   2   4   6              8             10        12    14   16   18



                                                           Niger                                                                                                         Nigeria
                                           p < 0.05          n.s.        p < 0.05                                                                         p < 0.05          n.s.        p < 0.05

                                                                                                       .6
                                                                                                                                                                                                                      .3
                                                                                                       .4
                                                                                                                                                                                                                      .2
                                                                                                       .2
                                                                                                                                                                                                                      .1
                                                                                                       0
                                                                                                                                                                                                                      0
    *Weather Product*                                                                                  -.2          *Weather Product*
                                                                                                                                                                                                                      -.1
             MERRA-2                                                                                   -.4                   MERRA-2

                 ERA5                                                                                                            ERA5

                  CPC                                                                                                             CPC



            *Dep. Var.*                                                                                                     *Dep. Var.*

                 Value                                                                                                           Value

               Quantity                                                                                                        Quantity



              * Model*                                                                                                        * Model*

  Weather + FE + Inputs                                                                                           Weather + FE + Inputs

          Weather + FE                                                                                                    Weather + FE

               Weather                                                                                                         Weather

                          0   2    4   6              8             10        12    14   16   18                                          0   2   4   6              8             10        12    14   16   18



                                                      Tanzania                                                                                                           Uganda
                                           p < 0.05          n.s.        p < 0.05                                                                         p < 0.05         n.s.         p < 0.05

                                                                                                   .1                                                                                                             .1

                                                                                                   .05                                                                                                            .05

                                                                                                   0                                                                                                              0

                                                                                                   -.05                                                                                                           -.05

    *Weather Product*                                                                                               *Weather Product*                                                                             -.1
                                                                                                   -.1
             MERRA-2                                                                                                         MERRA-2                                                                              -.15
                                                                                                   -.15
                 ERA5                                                                                                            ERA5

                  CPC                                                                                                             CPC



            *Dep. Var.*                                                                                                     *Dep. Var.*

                 Value                                                                                                           Value

               Quantity                                                                                                        Quantity



              * Model*                                                                                                        * Model*

  Weather + FE + Inputs                                                                                           Weather + FE + Inputs

          Weather + FE                                                                                                    Weather + FE

               Weather                                                                                                         Weather

                          0   2    4   6              8             10       12     14   16   18                                          0   2   4   6              8             10       12     14   16   18



Note : The figure presents specification curves, where each panel represents a different country, with three different
models presented within each panel. Each panel includes 18 regressions, where each column represents a single regres-
sion. Significant and non-significant coefficients are designated at the top of the figure. For each Earth observation
product, we also designate the significance and sign of the coefficient with color: red represents coefficients which are
negative and significant; white represents insignificant coefficients, regardless of sign; and blue represents coefficients
which are positive and significant.




                                                                                                             66
                                                       Figure A19: Bumpline: Days Without Rain
67




     Note : The figure presents a bumpline plot, where each panel represents a different country, where each column represents a different models and a different
     outcome variable for each model; each row represents the rank of each coefficient (one to six, with one representing the largest number and six representing
     the smallest number) within those regressions. Each panel includes 36 regressions. The bumpline package was developed by Naqvi (2024).
                                                      Figure A20: Bumpline: Growing Degree Days
68




     Note : The figure presents a bumpline plot, where each panel represents a different country, where each column represents a different models and a different
     outcome variable for each model; each row represents the rank of each coefficient (one to six, with one representing the largest number and six representing
     the smallest number) within those regressions. Each panel includes 18 regressions. The bumpline package was developed by Naqvi (2024).