Are CO 2 emissions a reason for drought in the United States?

causality analysis, CO₂ emissions, impulse-response analysis, panel data analysis, Standardized Precipitation Evapotranspiration Index (SPEI), US

In this study, the relationship between carbon dioxide (CO₂) emissions and drought (Standardized Precipitation Evapotranspiration Index) is investigated using data from 51 states in the US for the period 1997–2020. For this purpose, firstly, information about previous studies and the results of these studies are given in the study. Later, as empirical analysis, cross-section dependency tests, slope homogeneity tests, unit root analyses, cointegration tests, and causality analyses were performed. Then, short- and long-term parameter estimates are made for the entire panel and for each case. Lastly, impulse-response analyses are made. When the results are evaluated in general, it is found that CO₂ emissions and drought in the US affect each other in the short and long term. CO₂ emissions have been shown to have an increasing effect on drought, especially in most of the states located in the southeastern region of the US. On the other hand, CO₂ emissions have been found to cause an increase in the incidence of wet periods in a significant part of the states located in the northeastern region of the US.

HIGHLIGHTS

Listen

Studies show a close link between CO₂ emissions and drought.
The US is a leading CO₂ emitter globally.
This relationship was analyzed for 51 US states.
In southeastern states, higher CO₂ levels are linked to more dry periods.
In northeastern states, higher CO₂ levels are associated with more wet periods.

INTRODUCTION

Listen

Greenhouse gases, such as carbon dioxide (CO₂) and methane, are released into the atmosphere through the utilization of fossil-based energy sources, altering the chemical composition of the troposphere, the lower layer of the atmosphere. This phenomenon leads to the emergence of climate change and variability (Sen 2022). Over the past three decades, global warming and climate change have become among the most pressing environmental issues. Greenhouse gases play a pivotal role in the onset of these challenges (Ozturk & Acaravci 2010). Carbon dioxide, a prominent greenhouse gas, accounts for approximately 75% of the greenhouse effect globally and around 80% in the United States (US), according to the latest data from the WDI online database (World Development Indicators | DataBank 2022).

Climate change, associated with the escalating emissions of greenhouse gases, is anticipated to result in a rise in global average temperature and an intensified hydrological cycle, leading to more frequent and severe droughts and floods (Houghton et al. 2001). Drought conditions entail reduced precipitation, water scarcity, elevated average temperatures, sea level elevation, and an increase in both the frequency and intensity of extreme weather events (Troccoli 2010).

Figure 1, constructed using data from the Emissions Database for Global Atmospheric Research (EDGAR) fossil CO₂ booklet 2022, illustrates the total fossil CO₂ emissions of the countries emitting the highest amounts of CO₂. The figure indicates a significant global increase in CO₂ emissions, rising approximately 2.5 times over the past 50 years. Notably, the US stands out as the country with the highest CO₂ emissions during the initial 35 years of this 50-year period (EDGAR 2022).

Figure 1

Fossil CO2 totals of top emitting countries (1970–2021). Source: Emissions Database for Global Atmospheric Research (EDGAR), CO2 emissions of all world countries, 2022 Report.

Fossil CO₂ totals of top emitting countries (1970–2021). Source: Emissions Database for Global Atmospheric Research (EDGAR), CO₂ emissions of all world countries, 2022 Report.

In the Fifth National Climate Assessment (NCA5), it was stated that the 1.1 centigrade global warming observed in the industrial age is caused by greenhouse gas emissions resulting from human activities. It was stated that three-quarters of this global warming and emissions have occurred since the 1970s. It was reported that the US is most affected by this global warming and that disasters may occur due to warming caused by emissions in the future (NCA5 2023).

There are some studies in literature that examined the relationship between CO₂ emissions and climate change. Matthews et al. (2009) stated that CO₂ emissions will lead to climate change. Based on this, using carbon-climate response calculations, which they described as the ratio of temperature change to cumulative CO₂ emissions, they estimated that the carbon-climate response would be in the range of 1.0–2.1 °C per trillion tons of carbon emitted. Davis et al. (2010) calculated the CO₂ emissions that will result from fossil fuel use between 2010 and 2060, based on the current energy infrastructure. They stated that, according to their calculations, there will be 496 gigatons of CO₂ emissions, which will cause an average warming of 1.3 °C. Hao et al. (2016) used 1995–2011 data from 29 provinces of China to examine the potential effects of climate change on China's CO₂ emissions. According to their findings, they concluded that climate change has a significant but small impact on China's carbon emissions.

In addition to the studies examining the relationship between CO₂ emissions and climate change in the literature, there are also studies examining the relationship between climate change and drought. Some studies examining the effects of climate change on drought in different parts of the world follow.

The study conducted by Wilhite and Hayes examined the effects and severity of drought in the US and they concluded that the effect and severity of drought have increased in recent years (Wilhite & Hayes 1998). These drought events are estimated to be two to three times more damaging to the US economy than the 1989 San Fransisco earthquake (Riebsame 2019; Mishra & Singh 2010). At least 50% of the US has been affected by some level of drought event since mid-summer 2020. Especially since the fall of 2022, the drought has increased its severity, and more than 85% of the land area of the US has been affected by drought (Rodziewicz et al. 2023).

Droughts have become more frequent in many countries in Europe as well as in the US (Demuth & Stahl 2001). According to Spinoni et al. (2015), Northern and Eastern Europe showed the highest frequency and severity of droughts from the early 1950s to the mid-1970s. Considering the whole of Europe, the 1950s were considered to be the years when the effects of meteorological droughts were felt the most. Southern and Western Europe (especially the Mediterranean region) showed the highest frequency and severity of droughts from the early 1990s. Overall, a continuous upward trend in drought events was observed across Europe from the early 1980s to the early 2010s (Spinoni et al. 2015).

In recent years, studies for Asian countries have shown that crop production has decreased in many parts of Asia due to rising temperatures and decreasing rainfall days (Bates et al. 2008; Kienzler et al. 2012). The increasing trend in the frequency of drought due to the effect of global warming causes countries in East Asia such as China, Japan, Korea, the Peninsula, and Mongolia to be highly affected by drought every year (Zhai et al. 2010; Zhang & Zhou 2015). When historical records were examined, it was seen that two major drought events occurred in the Northern China region in 1972 and 1997, and this drought caused the Yellow River to dry out (Zhang & Zhou 2015). In addition, due to the drought in Southwest China between 2009 and 2010, 21 million people had difficulties in obtaining drinking water and an economic loss of 30 billion dollars occurred (Yang et al. 2012). One of the regions most affected by the effects of drought in recent years is the Australian continent. Floods and drought events are common climate features in Australia (King et al. 2020). Whetton et al.’s (1993) study analyzed the impact of climate change caused by the greenhouse gas effect on flood and drought events in Australia and evaluated that there is an increase in high precipitation frequency and a decrease in low precipitation frequencies due to greenhouse gas effect in the simulations, and therefore, an increase in flood events will occur (Whetton et al. 1993). It is observed that in the Australian region, especially after long-term droughts, precipitation values tend to occur in the form of heavy rains instead of returning to average conditions (King et al. 2020). The Murray Darling Basin, one of Australia's largest crop production basins, was greatly affected by drought between 2017 and 2019, and Australian average rainfall was at a record low level in 2019, resulting in severe drought and significant agricultural impacts in the second half of the year (Dunne & Kuleshov 2023). It has been stated that the winter months of the year before the drought, which has been ongoing since 2017, were the fourth wettest winter since the early 1900s. It is estimated that the reason for this is the abnormally high sea surface temperature of the Indian Ocean (King et al. 2020). It is also stated that the severity of drought has increased in Africa over the last few decades as a result of changes in the climate (Gebrechorkos et al. 2022). It is observed that the frequency of occurrence of drought events has increased in recent years (Meier et al. 2007; Ayana et al. 2016; Haile et al. 2019). According to the Centre for Research on the Epidemiology of Disasters, drought caused more than 800,000 deaths and $2 billion in economic losses in Sub-Saharan Africa in the 20th and 21st centuries (Masih et al. 2014; Ekolu et al. 2022). It has been observed that droughts in the region occasionally occur in the same year as other extreme climatic events (Hillbruner & Moloney 2012; Nicholson 2014; Haile et al. 2019). Droughts and unprecedented flash floods in 2006, 2007, 2009, and 2020 can be cited as examples of these events (Nicholson 2014; Haile et al. 2019).

The above-mentioned studies examine the effects of CO₂ emissions on climate change and climate change on drought. It is stated that the biggest cause of climate change is greenhouse gas emissions. A significant part of greenhouse gases comes from CO₂ emissions. Therefore, examining the relationship between CO₂ emissions, climate change, and drought is an important issue. However, there are a limited number of studies in the literature examining the relationship between CO₂ emissions, climate change, and drought. One of these studies was conducted by Strezepek et al. In their 2010 study, Strzepek et al. applied the Standardized Precipitation Indices and the Palmer Drought Severity Index to the Intergovernmental Panel on Climate Change General Circulation Models to assess the impact of climate change on the frequency and intensity of droughts in the US over the next century. Although drought risk was not assessed under CO₂ stabilization scenarios in the study, a comparison of SPI and Palmer Drought Severity Index results between emission scenarios provides evidence that lower CO₂ concentrations are related to lower drought risk across the US (Strzepek et al. 2010). Likewise, in a study conducted by Hardin et al. (2017), the relationship between drought and CO₂ emissions was discussed. In the study, the intense drought period between 2012 and 2014 in California and the period between 2009 and 2011, which is the pre-drought period, were compared and it was determined that there was a 33% increase in CO₂ emissions due to drought.

In the study of Gregory et al. (1997), it was evaluated that with the increase in CO₂ levels, precipitation decreased and evaporation increased. It was evaluated that these changes would lead to a decrease in soil moisture and drought events would become more frequent and severe. The research reveals that due to the high temperatures caused by increasing CO₂ levels, there is a decrease in snow cover in winter and a decrease in soil water level in spring. The decrease in soil moisture indicates that the large decrease in precipitation exacerbates drought, although evaporation remains limited in summer. Furthermore, this study emphasizes that the incidence of long dry spells increases significantly with increasing CO₂, which can have serious impacts on agriculture and water resources. These findings suggest that CO₂ emissions have lasting effects on the hydrological cycle in the long term and that droughts will become more widespread and intense with global climate change.

Räisänen used 20 model experiments participating in the second phase of the Coupled Model Intercomparison Project (CMIP2) in his 2005 study to examine the effect of CO₂ on high extreme and low extreme precipitation. In particular, he found that CO₂ is well correlated with changes in long-term average precipitation. It was observed that low extreme precipitation increased where mean precipitation was low, while high extreme precipitation increased where mean precipitation was high. The difference between this high and low extreme precipitation is determined to increase with CO₂.

In their 2024 study, Su et al. focus on understanding global drought changes under a carbon neutral scenario using CO₂ removal and increase models. They used the Standardized Precipitation Evapotranspiration Index (SPEI) to assess drought severity. In their study, they found that CO₂ increases in particular lead to an increase in potential evapotranspiration in excess of precipitation (PET), resulting in more intense droughts. In particular, low- and mid-latitude regions, including South Africa, Australia, and the Amazon, showed more significant drying due to CO₂ increase. In addition, one of the important points in the study is that even if CO₂ levels return to pre-industrial levels, the global water cycle continues to deteriorate, and many regions continue to experience persistent droughts. In addition, the study attributed about 65% of drought changes to increased PET, especially due to increasing temperatures, while the contribution of precipitation was assessed to be only around 35%. Therefore, although it seems likely that reducing CO₂ emissions will stabilize some aspects of the climate, it is still stated that temperature and evapotranspiration will cause drought to persist for longer.

Cook et al. (2020) analyzed future drought patterns using CMIP6 models to determine how increasing CO₂ emissions affect global drought risk. As a result of the studies, it is emphasized that as CO₂ levels increase, temperature increases lead to higher evapotranspiration rates, precipitation patterns change, precipitation decreases in some regions and dry periods are prolonged, which causes more frequent and severe droughts, especially in mid-latitude and subtropical regions. In the study, western North America, Central America, Europe, and the Mediterranean, the Amazon, southern Africa, China, Southeast Asia, and Australia were shown among the regions that will be most affected by drought.

Vahedifard et al. (2024) examined the feedback loop between drought, soil drying, and CO₂ emissions. The findings of the study show that as CO₂ levels rise, drought conditions become more severe, causing the soil to crack and dry out. This leads to the release of more soil surface area and increased CO₂ emissions from the soil surface, further intensifying global warming. This feedback loop reduces the soil's capability to retain moisture, accelerating evapotranspiration and weakening the soil's capability to sequester carbon, which in turn exacerbates drought conditions. As a result, CO₂-induced drought not only worsens climate change, but also directly contributes to higher atmospheric CO₂ levels.

Sen (2022) stated that there will be changes in flood frequency and severity with climate change, while rainfall intensity will decrease, and drought frequency and severity will be expected to increase in many regions. Another result obtained in his study is that it is possible for floods and droughts to increase sequentially in some water catchment areas.

When the literature is examined, it is seen that similar results are reached in the studies on drought and CO₂ emissions, and it shows that the increase in CO₂ emissions causes changes in climate parameters such as precipitation and temperature and may be related to drought events.

The hypothesis of this study is that ‘CO₂ emissions, by driving climate change, influence the frequency and intensity of droughts through their impact on regional precipitation and temperature patterns’. To investigate whether this hypothesis is valid for the US, this study analyzed statistically the relationship between CO₂ emissions and drought using data from 51 states in the US for the period 1997–2020. As seen in Figure 1, the US is one of the countries that emit the most CO₂ emissions in the world. Therefore, this study is especially focused on the US and will make significant contributions to the literature in terms of investigating whether CO₂ emissions cause drought for the 51 states of the US.

DATA AND METHODOLOGY

Listen

In the study, based on the SPEI drought index and CO₂ emissions data of the 51 states of the US for the period 1997–2020, the relationship between the variables is examined. The function of the model established in the study is as follows:

(1)

The model used in this study is given as follows:

(2)

Here, CO₂ refers to million metric tons of energy-related carbon dioxide, DR refers to SPEI drought indices, and m refers to months. For example, DR1 refers to SPEI January drought index, DR2 refers to SPEI February drought index, DR3 refers to SPEI March drought index, etc. Drought data are calculated for the 12-month period with the SPEI. CO₂ emissions data are obtained from the U.S. Energy Information Administration (EIA 2022), and the 12-month period droughts data are obtained from the SPEI Global Drought Monitor. CO₂ emissions data are taken in the logarithmic form.

Correlation matrix and descriptive statistics are given in Table 1. Examining the mean and median values of the variables, it is found that lnCO₂ takes the highest values, while DR12 takes the lowest values. While the CO₂ emission values range between 1 and 7, it takes an average value of 4. In this period, while the drought values vary between −3 and +3, the average drought value is 0.3.

Table 1

Descriptive statistics and correlation matrix for lnCO₂ and droughts

	lnCO₂	DR1	DR2	DR3	DR4	DR5	DR6	DR7	DR8	DR9	DR10	DR11	DR12
Mean	4.24	0.24	0.25	0.25	0.26	0.25	0.25	0.24	0.24	0.23	0.23	0.23	0.22
Median	4.38	0.25	0.31	0.30	0.31	0.27	0.31	0.29	0.29	0.30	0.27	0.29	0.22
Max.	6.53	3.06	2.97	2.88	2.69	2.82	3.05	3.06	2.88	2.95	2.76	2.93	2.99
Min.	0.88	− 2.54	− 2.52	− 2.68	− 2.30	− 2.36	− 2.24	− 2.25	− 2.37	− 2.26	− 2.54	− 2.58	− 2.60
Std. Dev.	1.04	1.04	1.04	1.05	1.03	1.01	1.03	1.04	1.04	1.01	1.00	1.02	1.02
Obs.	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224
lnCO₂	1
DR1	− 0.01	1
DR2	− 0.01	0.96	1
DR3	− 0.01	0.91	0.96	1
DR4	− 0.01	0.84	0.90	0.95	1
DR5	− 0.00	0.74	0.80	0.86	0.93	1
DR6	− 0,00	0.60	0.67	0.72	0.80	0.90	1
DR7	0.00	0.51	0.57	0.62	0.71	0.83	0.95	1
DR8	0.00	0.45	0.51	0.56	0.65	0.77	0.90	0.95	1
DR9	0.00	0.35	0.42	0.46	0.54	0.66	0.78	0.85	0.92	1
DR10	0.00	0.28	0.35	0.39	0.48	0.60	0.72	0.79	0.87	0.93	1
DR11	− 0.00	0.20	0.27	0.31	0.41	0.53	0.65	0.72	0.79	0.88	0.95	1
DR12	0.01	0.13	0.20	0.23	0.33	0.46	0.58	0.66	0.72	0.81	0.90	0.97	1

	lnCO₂	DR1	DR2	DR3	DR4	DR5	DR6	DR7	DR8	DR9	DR10	DR11	DR12
Mean	4.24	0.24	0.25	0.25	0.26	0.25	0.25	0.24	0.24	0.23	0.23	0.23	0.22
Median	4.38	0.25	0.31	0.30	0.31	0.27	0.31	0.29	0.29	0.30	0.27	0.29	0.22
Max.	6.53	3.06	2.97	2.88	2.69	2.82	3.05	3.06	2.88	2.95	2.76	2.93	2.99
Min.	0.88	− 2.54	− 2.52	− 2.68	− 2.30	− 2.36	− 2.24	− 2.25	− 2.37	− 2.26	− 2.54	− 2.58	− 2.60
Std. Dev.	1.04	1.04	1.04	1.05	1.03	1.01	1.03	1.04	1.04	1.01	1.00	1.02	1.02
Obs.	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224	1,224
lnCO₂	1
DR1	− 0.01	1
DR2	− 0.01	0.96	1
DR3	− 0.01	0.91	0.96	1
DR4	− 0.01	0.84	0.90	0.95	1
DR5	− 0.00	0.74	0.80	0.86	0.93	1
DR6	− 0,00	0.60	0.67	0.72	0.80	0.90	1
DR7	0.00	0.51	0.57	0.62	0.71	0.83	0.95	1
DR8	0.00	0.45	0.51	0.56	0.65	0.77	0.90	0.95	1
DR9	0.00	0.35	0.42	0.46	0.54	0.66	0.78	0.85	0.92	1
DR10	0.00	0.28	0.35	0.39	0.48	0.60	0.72	0.79	0.87	0.93	1
DR11	− 0.00	0.20	0.27	0.31	0.41	0.53	0.65	0.72	0.79	0.88	0.95	1
DR12	0.01	0.13	0.20	0.23	0.33	0.46	0.58	0.66	0.72	0.81	0.90	0.97	1

The correlation matrix shows that CO₂ emissions are positively correlated with droughts 7, 8, 9, 10, and 12, and negatively correlated with droughts 1, 2, 3, 4, 5, 6, and 11.

Standardized Precipitation Evapotranspiration Index

Listen

The SPEI was developed by Vicente-Serrano et al. (2010). The SPEI is a simple multiscale drought index combining precipitation and temperature data. The multiscale nature of SPEI allows the identification of different types of droughts and their effects on different systems. The SPEI was specifically proposed to investigate the impact of global warming on the severity of drought, based on precipitation and potential evapotranspiration (PET). When calculating SPEI, the monthly (or weekly) difference between precipitation and PET is used. The formulation of the difference is given as Equation ():

(3)

where

is the monthly deficit,

is the total precipitation value for month i, and

is the monthly potential evapotranspiration. Calculated

values are summed over different time scales as Equation ():

(4)

where n represents the month, and k represents the time scale. The probability density function form of the log-logistic distribution is expressed as Equation ():

(5)

where

represents the scale,

represents the shape, and

represents origin parameters for

⁠. According to the log-logistic distribution, the probability distribution function for the D series is expressed as Equation ():

(6)

Following the classical approach suggested by Abramowitz & Stegun (1965), with

the SPEI can be obtained as the standardized values of

as Equation ():

(7)

where

and

are the constants of the SPEI equation, and

2.515517,

0.802853,

0.010328,

1.432788,

0.189269, and

0.001308.

And W is obtained for from and for from ⁠, the sign of the resultant SPEI is reversed for (Vicente-Serrano et al. 2010; Tirivarombo et al. 2018; Abbasi et al. 2019; Danandeh Mehr & Vaheddoost 2020).

Obtained SPEI values can be interpreted as drought increases, it falls below −0.5 and as wetting increases, it rises above +0.5.

The SPEI is able to explain the possible effects of temperature variability and temperature extremes beyond the context of global warming. Hence, it is preferable to use SPEI for the detection, analysis and monitoring of droughts in any climatic region of the world, given the small additional data requirements of SPEI compared with SPI. According to its intensity and duration, the SPEI can measure the severity of drought and identify the beginning and end of drought periods. Similar to the SPI, the SPEI can be calculated over a wide range of climates, allowing for comparisons of drought severity over time and space. In addition, the SPEI drought index has a clear and understandable calculation procedure. However, an important advantage of the SPEI over the most widely used drought indices that take into account the impact of PET on drought severity is that its multiscale characteristics allow the description of different drought types and their effects in the global warming context (Vicente-Serrano et al. 2010).

Cross-section dependency tests

Listen

When applying panel data analysis, firstly, the cross-section dependency and the homogeneity of the slope should be analyzed. Cross-sectional dependence implies that anything that affects one state may also affect other states. The homogeneity of the slope, on the other hand, is whether the slope coefficients are homogeneous in the state sample since the states have different characteristics within themselves. Due to these two possibilities, it is crucial to investigate the presence of cross-sectional dependence and investigate whether the slope is homogeneous or heterogeneous.

In this study, Lagrange multiplier (LM), bias-adjusted LM (LM_adj), and CD_LM tests were applied to analyze cross-sectional dependence. The LM test, proposed by Breusch and Pagan in 1980, tests the impact on first-order conditions so that the probability of imposing the hypothesis is maximum. It is based on prediction, with the hypothesis introduced as parametric constraints. It provides researchers with a simple technique to evaluate the adequacy of their certain specifications (Breusch & Pagan 1980).

LM statistics are applicable when

and N is fixed and are defined as Equation ():

(8)

where T is time, N is the number of observations, i is cross-section units, and

is the sample estimate of the pairwise correlation of the residuals. The

value can be calculated as Equation ():

(9)

where

represents the ordinary least squares estimate of

⁠. The LM statistic is asymptotically chi-square distributed with

degrees of freedom (Emirmahmutoglu & Kose 2011). It has been widely discussed in the literature that the LM test is affected by severe size distortions for large panels with N > T (Li & Yao 2021).

In 2004, Pesaran proposed a scaled version of the LM test which is stated by the CD_LM test and has a distribution of

as

and

⁠. Different from the standard LM test, the CD_LM test is valid for a large number of cross-section units (N) observed over a time period T. The CD_LM test is non-zero centric for finite T and may exhibit large-scale degradations as n is increased (Baltagi et al. 2012). The CD_LM test is defined as Equation ():

(10)

This test indicates that under the null hypothesis of no cross-sectional dependence

for

and T is sufficiently large. A slightly modified version of the above equation is proposed for unbalanced panels as Equation ():

(11)

where

is the number of common time series observations between units i and j (De Hoyos & Sarafidis 2006).

The CD_LM test has a significant disadvantage. It may lack power in some cases where the population average pairwise correlations are zero, but the underlying individual population pairwise correlations are not zero. This may occur, for example, where the cross-dependence under the alternative hypothesis can be characterized as a factor model with a mean factor load of zero (Pesaran 2004).

Pesaran et al. (2008) proposed bias-adjusted LM (LM_adj) tests that use the exact mean and variance of the LM statistic if the panel data models have exogenous regressors and normal errors. The centering of the LM statistic is valid for constant Τ and N, as the adjustments were derived by using the results in Ullah (2004). The bias-adjusted test was shown to be stable even when Pesaran's CD test was not stable, and the cross-sectional average of factor loadings was close to zero. The LM_adj test is defined as Equation ():

(12)

where k is the number of regressors,

is the exact mean of

⁠, and

is the variance of

⁠. As

then

⁠, the

test has a distribution of

(Pesaran et al. 2008).

Homogeneity tests

Listen

To test the slope of the homogeneity, Pesaran & Yamagata (2008) proposed that dispersion-type tests based on the tests suggested by Swamy (1970) can be applied to panel data models where the cross-section size may be larger than the time series size. There are two versions of homogeneity tests suggested by Pesaran and Yamagata. The first version, indicated by

⁠, uses the Swamy statistic, while the other version, indicated by

⁠, is based on a revised version of the Swamy statistic in which the standard errors of regressions for individual cross-sectional units are calculated using the pooled fixed effects (FE) estimator instead of the ordinary least squares estimator. When the errors are normally distributed, the

tests are applicable as

irrespective of the relative expansion rates of N and T. The first version of the standardized dispersion statistics is formulated as Equation ():

(13)

where

is Swamy's statistic applied to slope coefficients that can be calculated as Equation ():

(14)

where

represents the weighted FE (WFE) pooled estimator of slope coefficients, and

represents an identity matrix of order T.

and

can be calculated as Equation ():

(15)

while N is constant and

⁠, Swamy statistic,

⁠, has a chi-square distribution with

degrees of freedom asymptotically.

The second version of the standardized dispersion statistics is formulated as Equation ():

(16)

where

is the modified version of Swamy's statistic applied to slope coefficients that can be calculated as Equation ():

(17)

where

represents the WFE estimator calculated by using

that can be shown as Equation ():

(18)

Under normal distribution errors, the small sample characteristics of the dispersion tests can be reformed according to the following mean and variance bias-adjusted versions of

and

as Equations () and ():

(19)

(20)

Unit root test

Listen

When the series are non-stationary, the findings obtained can sometimes be misleading. Therefore, it is necessary to test whether the series are stationary or not. To analyze the stationarity in models with cross-section dependence, second-generation unit root tests can be applied. One of these tests is the CIPS (cross-sectional increased panel unit root statistics) test, which is a cross-sectional augmented version of the IPS test, suggested by Pesaran (2007). The CIPS test statistic is the overall unit root test statistic for the entire panel and is calculated by averaging the CADF (cross-sectionally augmented Dickey–Fuller) unit root tests. The CIPS test can be calculated as Equation ():

(21)

where

represents the CADF statistic for the ith cross-section unit given by the t ratio of the

coefficient in the CADF regression. The CADF statistic can be calculated as Equation ():

(22)

The calculations of these symbols are given as Equations ()–():

(23)

(24)

(25)

where

represents the identity matrix.

Panel cointegration test

Listen

Cointegration analysis is necessary to test whether two variables move together in the long run. If there is no long-term relationship between two variables, short-term analyses would not give accurate results. Therefore, the cointegration test is a necessary analysis to verify the long-term relationship of our data. Westerlund (2007) proposed new cointegration tests based on error correction for panel data. Two of the four tests are developed to check the alternative hypothesis that the panel as a whole is cointegrated, whereas the other two tests are developed to check the alternative hypothesis that there is at least one cointegrated cross-sectional unit. The asymptotic results show that the tests have restrictive normal distributions and are coherent.

Westerlund developed Equation () to create the new statistics:

(26)

where

represents the deterministic components,

represents the parameters’ associated vector, and

represents the error correction parameter which is estimated by applying the least squares method.

The group mean test statistics, two of the four test statistics based on the least squares estimate of

in this equation and its t-ratio, are calculated as Equation ():

(27)

where

represents the semiparametric kernel estimator of

⁠, and

represents the conventional standard error of

⁠.

The panel test statistics can be calculated as Equation ():

(28)

AMG and PMG tests

Listen

The augmented mean group (AMG) test allows us to consider that the relationship between CO₂ emissions and drought in different states may be heterogeneous and that each region may have different dynamics. The pooled mean group (PMG) test, on the other hand, allows for short- and long-term assessments for the states as a whole.

The AMG estimator method, which is an estimation method used in cases of parameter heterogeneity and cross-section dependence, was suggested by Eberhardt & Bond (2009) and Eberhardt & Teal (2010). To estimate the unobserved co-dynamic effect, the AMG estimator uses a two-step method and allows for cross-section dependency by including the co-dynamic effect parameter. The empirical model which is given in Equation () is used to calculate the AMG estimator:

(29)

where

represents a vector of observable covariates,

and

represent unobserved common factors, and

represents country-specific factor loadings.

The AMG estimator is obtained by including a common dynamic effect in the country regression. The two steps of AMG estimation are given as Equations () and ():

(30)

(31)

The first step indicates a standard first difference OLS regression with T − 1 period dummy variables in first differences ⁠, from where the estimation coefficients are collected and relabeled as ⁠. This process is subtracted from the pooled regression in first differences, as non-stationary and unobservable variables can severely distort the estimates in pooled level regressions. In the second step, is included in each of the N group-specific regressions, including linear trend terms, to capture omitted idiosyncratic processes that evolve linearly over time (Eberhardt & Teal 2010).

Pesaran et al. proposed the PMG estimator in 1999, which includes both the pooling implicit in the homogeneity constraints on the long-run coefficients and averaging between groups to derive the means of the estimated error correction coefficients and other short-run parameters. They developed the PMG model using the Auto Regressive Distributed Lag (ARDL) model cointegration model and used this model to estimate short-term and long-term parameters through heterogeneous cross-sections. The PMG model can be formulated as Equation ():

(32)

where

represents coefficient vectors,

represents the coefficients of lagged variables,

represents the error correction term,

represents the long-term coefficients, and

represents the adjustment coefficients (Bilgili et al. 2017).

Dumitrescu-Hurlin panel causality test

Listen

After determining the correlation between CO₂ emissions and drought, the Dumitrescu-Hurlin panel causality test was applied to determine the direction of the relationship between these two variables. This test is a critical step to determine whether CO₂ emissions cause drought or vice versa by examining the directional relationships between the two variables.

Dumitrescu and Hurlin suggested a very simple Granger non-causality test for heterogeneous panel data models with the Dumitrescu-Hurlin panel causality test they developed in 2012. The Dumitrescu-Hurlin test statistic depends on the individual Wald statistics of Granger non-causality. They consider the linear model given as Equation ():

(33)

where K represents optimum lag interval and is assumed to be identical for all cross-section units of the panel and that the panel is balanced. The individual effects

are assumed to be fixed in the time dimension. Initial conditions of individual process

and

⁠,

⁠, are given and observable. The null hypothesis is defined as Equation ():

(34)

This statement means that there is no causality for all units in the panel. According to the Dumitrescu-Hurlin test, there may be causality for some units and not for others. Then, the alternative hypothesis is defined as follows:

(35)

where

is unknown but it is known to satisfy the

condition. If

⁠, there is no causality for any of the cross-section units in the panel, conversely when

⁠, there is causality for all cross-section units in the sample. And this means that we obtained a homogeneous result in terms of causality. But if

⁠, this means that we obtained a heterogeneous result in terms of causality. From this perspective, they suggest to use the mean of individual Wald statistics associated with the test of the non-causality hypothesis for

units. The mean statistic

related to the null hypothesis of homogeneous non-causality (⁠

⁠) is calculated as in Equation () (Lopez & Weber 2017; Aydin 2019):

(36)

where

represents the individual Wald statistics for the ith cross-section unit. The matrix

with (⁠

can be represented as

⁠, while e represents a

unit vector and

the parameter vector of the model. Here, the test for the Hnc hypothesis can now be implied as

⁠; where R is a

matrix with

The Wald statistic

corresponding to the individual test

is defined for each

as Equation ():

(37)

where

represents the estimation of the

parameter and is derived under the alternative hypothesis and

represents the estimation of the residuals' variance. The Wald statistics can also be written as Equation ():

(38)

Each individual Wald statistic converges to a

distribution with K degrees of freedom for

⁠, under the null hypothesis of non-causality given as Equation ():

(39)

Null hypothesis of no causality is not accepted when the Wald statistic is greater than the critical values (Bilgili et al. 2017).

Impulse response functions

Listen

A shock to one variable not only affects that variable but also spreads to all other endogenous variables. The impulse-response function analyzes the impact of a one-time shock to one of the variables on the present and future values of endogenous variables. If there is no simultaneous correlation between the variables, the interpretation of the impulse-response is simple. However, variables are often interrelated and can be viewed as having a common component that cannot be attributed to a particular variable. In this case, what needs to be done is to make the variables uncorrelated (Lütkepohl 1990; Killian 1998; Pesaran & Shin 1998).

RESULTS AND DISCUSSION

Listen

In this section of the study, the results obtained from the analyses applied in the study and the discussions about the results are given. In the study, first, cross-section dependency and homogeneity tests were performed. The results obtained from these tests are given in Table 2.

Table 2

Cross-sectional dependence and homogeneity tests

Test	Statistic	P value
Cross-sectional dependence tests
LM	2,931	0.00
LM_adj	43.77	0.00
CD_LM	22.63	0.00
Homogeneity tests
Δ	0.75	0.45
Δ_adj	1.11	0.27

Test	Statistic	P value
Cross-sectional dependence tests
LM	2,931	0.00
LM_adj	43.77	0.00
CD_LM	22.63	0.00
Homogeneity tests
Δ	0.75	0.45
Δ_adj	1.11	0.27

In Table 2, when the results of cross-sectional dependence tests are examined, all P values are less than 0.05. Then, it can be said that there is a cross-sectional dependency between the series. When the results of homogeneity tests are examined, all P values are greater than 0.05. So, it can be said that the null hypothesis that the slope is homogeneous is accepted and there is no state-specific heterogeneity.

Since it is determined that there is cross-section dependency according to the cross-sectional dependence test results, second-generation unit root tests should be used when investigating the stationarity of the series. In Table 3, the results obtained from the CIPS unit root test, which is the second-generation unit root test, are given.

Table 3

CIPS unit root test for panel of 51 US States (1997–2020)

Panel CIPS test	Intercept			Intercept and trend
lnCO₂	− 2,282^a			− 2,756^a
drought1	− 4,281^a			− 4,413^a
drought2	− 4,214^a			− 4,305^a
drought3	− 4,302^a			− 4,483^a
drought4	− 4,285^a			− 4,413^a
drought5	− 4,242^a			− 4,390^a
drought6	− 4,272^a			− 4,444^a
drought7	− 4,484^a			− 4,697^a
drought8	− 4,648^a			− 4,821^a
drought9	− 4,678^a			− 4,839^a
drought10	− 4,425^a			− 4,565^a
drought11	− 4,094^a			− 4,280^a
drought12	− 4,327^a			− 4,405^a
Critical values	10%	5%	1%	10%	5%	1%
Critical values	−2.02	−2.08	−2.19	−2.52	−2.58	−2.69

Panel CIPS test	Intercept			Intercept and trend
lnCO₂	− 2,282^a			− 2,756^a
drought1	− 4,281^a			− 4,413^a
drought2	− 4,214^a			− 4,305^a
drought3	− 4,302^a			− 4,483^a
drought4	− 4,285^a			− 4,413^a
drought5	− 4,242^a			− 4,390^a
drought6	− 4,272^a			− 4,444^a
drought7	− 4,484^a			− 4,697^a
drought8	− 4,648^a			− 4,821^a
drought9	− 4,678^a			− 4,839^a
drought10	− 4,425^a			− 4,565^a
drought11	− 4,094^a			− 4,280^a
drought12	− 4,327^a			− 4,405^a
Critical values	10%	5%	1%	10%	5%	1%
Critical values	−2.02	−2.08	−2.19	−2.52	−2.58	−2.69

^aIllustrates 1% statistical significance.

Examining the results derived from the CIPS test, it can be concluded that all variables are stationary at level. Since all series are stationary at level, it may be analyzed if there is a cointegration relationship between the variables. Since the Westerlund cointegration test provides more reliable and consistent results in the presence of cross-sectional dependence between the series, the Westerlund cointegration test is applied as a cointegration test.

When Table 4 is examined, it is seen that H₀ was rejected according to the test results for the cointegration between lnCO₂ and DR1, DR2, DR3, DR4, DR5, DR6, DR7, and DR8, but H₀ was accepted for the cointegration between lnCO₂ and DR9, DR10, DR11, and DR12. That is, there is a cointegration relationship between lnCO₂ and DR9, DR10, DR11, and DR12 variables in at least one cross-sectional unit. And according to the and test results, H₀ was rejected. That is, for the entire panel, there is a cointegration relationship between the lnCO₂ and drought variables.

Table 4

Westerlund Error Correction Model (ECM) panel cointegration tests

Relationship tested
lnCO₂ and drought1	−2.66^a	−11.87	−18.95^a	−11.53^a
lnCO₂ and drought2	−2.67^a	−11.99	−19.21^a	−11.71^a
lnCO₂ and drought3	−2.63^a	−11.67	−18.86^a	−11.44^a
lnCO₂ and drought4	−2.59^b	−11.39	−18.36^a	−11.07^a
lnCO₂ and drought5	−2.63^a	−11.43	−18.44^a	−11.04^a
lnCO₂ and drought6	−2.56^b	−11.03	−17.31^a	−10.53^b
lnCO₂ and drought7	−2.52^c	−10.86	−17.28^a	−10.31^c
lnCO₂ and drought8	−2.54^b	−10.72	−17.46^a	−10.19^c
lnCO₂ and drought9	−2.42	−10.44	−16.82^b	−9.92
lnCO₂ and drought10	−2.38	−10.40	−16.53^b	−9.91
lnCO₂ and drought11	−2.41	−10.73	−16.74^b	−10.17^c
lnCO₂ and drought12	−2.44	−10.78	−16.87^b	−10.19^c

Relationship tested
lnCO₂ and drought1	−2.66^a	−11.87	−18.95^a	−11.53^a
lnCO₂ and drought2	−2.67^a	−11.99	−19.21^a	−11.71^a
lnCO₂ and drought3	−2.63^a	−11.67	−18.86^a	−11.44^a
lnCO₂ and drought4	−2.59^b	−11.39	−18.36^a	−11.07^a
lnCO₂ and drought5	−2.63^a	−11.43	−18.44^a	−11.04^a
lnCO₂ and drought6	−2.56^b	−11.03	−17.31^a	−10.53^b
lnCO₂ and drought7	−2.52^c	−10.86	−17.28^a	−10.31^c
lnCO₂ and drought8	−2.54^b	−10.72	−17.46^a	−10.19^c
lnCO₂ and drought9	−2.42	−10.44	−16.82^b	−9.92
lnCO₂ and drought10	−2.38	−10.40	−16.53^b	−9.91
lnCO₂ and drought11	−2.41	−10.73	−16.74^b	−10.17^c
lnCO₂ and drought12	−2.44	−10.78	−16.87^b	−10.19^c

Notes: The optimal lag and lead lengths are both set to 1 and the optimal Barlett Kernel window width is set to 3.

^aIndicates 1% statistical significance.

^bIndicates 5% statistical significance.

^cIndicates 10% statistical significance.

Since it was understood that there was a cointegration relationship between the variables, the long-term parameters of the variables were estimated using the AMG estimation method.

In Table 5, the results obtained as a result of the analysis of the effects of CO₂ emissions on 12-month period SPEI drought index variable for each month are given. When Table 5 is examined, it is seen that CO₂ influences drought in 17 of 51 states in different months. The states with the longest impact of CO₂ emissions on drought are Pennsylvania and Tennessee, for 8 months. CO₂ emissions increase the January SPEI drought index in seven states: Arizona, Connecticut, District of Columbia, Hawaii, New York, Pennsylvania, and Virginia. CO₂ emissions increase the February SPEI drought index in eight states: Arizona, Connecticut, District of Columbia, Hawaii, New Jersey, New York, Pennsylvania, and Virginia. CO₂ emissions increase the March SPEI drought index in seven states: Arizona, Connecticut, District of Columbia, Florida, New York, Pennsylvania, and Virginia. CO₂ emissions increase the April SPEI drought index in six states: Connecticut, District of Columbia, Florida, New York, Pennsylvania, and Virginia. While CO₂ emissions reduce the May SPEI drought index in Tennessee, they increase it in five states: Connecticut, District of Columbia, New York, Pennsylvania, and Virginia. While CO₂ emissions reduce the June SPEI drought index in Tennessee, they increase it in two states: Ohio and Pennsylvania. While CO₂ emissions increase the July SPEI drought index in Pennsylvania, they reduce it in three states: Oregon, South Carolina, and Tennessee. While CO₂ emissions increase the August SPEI drought index in Pennsylvania, they reduce it in three states: Arkansas, South Carolina, and Tennessee. CO₂ emissions reduce the September SPEI drought index in five states: Alabama, Arkansas, Georgia, South Carolina, and Tennessee. CO₂ emissions reduce the October SPEI drought index in six states: Alabama, Georgia, Louisiana, North and South Carolina, and Tennessee. CO₂ emissions reduce the November SPEI drought index in five states: Alabama, Georgia, North and South Carolina, and Tennessee. CO₂ emissions reduce the December SPEI drought index in six states: Alabama, Georgia, Louisiana, North and South Carolina, and Tennessee.

Table 5

Cointegration coefficients from the AMG estimator (Independent variable: lnCO₂)

State	DR1	DR2	DR3	DR4	DR5	DR6	DR7	DR8	DR9	DR10	DR11	DR12
Alabama	1,666	1,807	1,211	0,711	− 1,469	− 3,368	− 4,642	− 4,971	− 6,125^b	− 7,364^a	− 7,574^a	− 7,971^a
Alaska	− 1,909	− 1,227	− 1,024	− 0,604	− 0,306	0,224	− 0,056	− 0,021	0,310	− 0,387	− 0,723	− 1,365
Arizona	5,323^a	4,887^b	3,579^c	2,539	2,555	1,724	1,181	0,615	1,059	1,198	2,362	2,366
Arkansas	− 2,974	− 2,453	− 3,073	− 2,982	− 3,925	− 3,190	− 4,229	− 5,536^c	− 6,282^b	− 5,264	− 3,840	− 5,204
California	1,332	4,452	3,063	0,126	0,173	− 0,585	− 1,883	− 1,965	− 1,551	− 1,687	− 0,403	2,170
Colorado	− 0,332	− 0,086	0,339	− 0,325	0,726	0,587	− 0,577	− 0,816	− 0,772	− 1,133	− 1,014	− 1,410
Connecticut	6,486^b	6,963^b	7,106^b	7,621^a	5,145^b	3,917	3,094	2,848	2,197	2,331	2,225	2,351
Delaware	2,601	3,380	3,277	2,503	3,582	2,551	1,954	− 0,305	− 1,723	− 2,576	− 2,911	− 2,772
Dist. of Col.	6,128^b	6,588^b	7,486^a	7,193^a	5,939^b	4,476	4,250	3,284	1,257	2,514	1,339	1,203
Florida	5,563	5,749	7,103^c	6,851^c	4,517	2,784	1,878	2,715	− 0,583	− 0,316	− 2,329	− 3,408
Georgia	2,652	1,816	1,615	0,584	− 1,037	− 2,002	− 2,822	− 3,305	− 4,110^c	− 4,735^b	− 5,066^b	− 5,268^b
Hawaii	3,616^b	4,047^b	1,798	0,765	0,042	− 0,307	− 1,051	− 1,192	− 0,425	− 0,878	− 0,544	− 0,572
Idaho	1,942	1,633	− 0,312	− 1,419	− 2,004	− 3,021	− 2,945	− 3,291	− 1,530	− 0,879	− 1,875	− 2,259
Illinois	− 1,789	− 0,992	− 2,256	− 3,032	− 2,970	− 2,196	− 3,566	− 3,651	− 1,861	− 2,406	− 2,340	− 2,583
Indiana	1,120	2,248	1,752	1,441	1,727	2,500	1,084	0,776	1,832	0,060	0,604	0,807
Iowa	1,091	1,378	0,336	− 0,083	0,111	1,714	1,100	1,980	2,651	2,674	2,788	2,557
Kansas	− 1,371	− 1,386	− 1,810	− 3,283	− 2,844	− 1,985	− 2,798	− 2,503	− 1,984	− 2,550	− 2,602	− 3,128
Kentucky	− 1,191	− 0,361	− 0,088	− 0,185	− 0,856	− 1,103	− 2,059	− 2,612	− 2,640	− 3,363	− 3,186	− 3,111
Louisiana	− 5,137	− 4,930	− 6,861	− 7,969	− 5,778	− 6,588	− 7,311	− 8,601	− 9,154	− 11,908^c	− 8,607	− 11,004^c
Maine	0,219	0,630	− 0,028	0,710	1,165	0,972	1,171	1,420	1,345	0,794	0,756	1,015
Maryland	1,835	2,870	2,523	1,829	1,865	1,395	0,566	− 0,642	− 0,994	− 0,971	− 1,820	− 1,565
Massachusetts	0,644	2,368	2,783	2,836	2,137	0,055	− 1,240	− 0,486	0,738	2,135	1,469	1,896
Michigan	0,448	1,371	− 3,020	− 2,980	− 3,303	− 0,503	− 1,987	− 3,254	1,810	− 1,227	− 1,568	− 0,047
Minnesota	− 0,724	− 0,218	− 0,633	− 0,049	− 1,597	− 1,792	− 3,414	− 4,110	− 1,545	− 0,945	0,402	0,392
Mississippi	0,512	0,763	0,649	− 0,495	− 0,912	− 0,517	− 1,046	0,349	0,317	0,121	0,213	− 0,756
Missouri	− 0,830	− 0,524	− 1,453	− 1,902	− 2,454	− 2,473	− 3,994	− 4,247	− 3,520	− 4,072	− 4,052	− 3,873
Montana	− 0,957	− 0,364	− 0,862	− 1,319	− 0,004	− 2,080	− 2,182	− 2,193	− 0,436	− 1,550	− 0,748	− 0,536
Nebraska	− 3,162	− 3,305	− 3,446	− 4,411	− 2,830	− 2,773	− 4,173	− 2,315	− 1,676	− 2,270	− 1,951	− 1,634
Nevada	− 1,626	− 0,514	− 0,342	− 1,182	1,010	− 0,527	− 1,700	− 2,240	− 1,235	− 1,242	− 1,449	− 0,750
New Hamp.	2,516	2,712	2,125	2,421	2,377	2,097	1,989	2,469	2,358	2,571	2,288	2,418
New Jersey	4,090	5,008^c	4,575	3,707	4,047	2,558	1,210	0,370	− 0,356	− 0,867	− 1,966	− 0,882
New Mexico	4,048	3,461	2,552	1,480	0,596	0,085	− 0,597	− 1,058	0,436	− 0,565	0,176	− 0,168
New York	5,949^b	7,610^a	8,551^b	9,020^a	7,486^a	4,587	3,270	1,918	0,684	1,614	0,504	1,500
N. Carolina	4,783	5,034	4,271	3,506	0,352	− 1,673	− 3,681	− 3,796	− 5,197	− 5,848^b	− 7,159^a	− 8,071^a
N. Dakota	4,931	4,104	3,525	1,544	1,763	2,303	1,670	0,823	2,424	2,846	5,110	4,107
Ohio	1,501	3,415	3,420	3,239	4,124	5,648^c	4,032	3,805	4,479	2,825	3,446	3,338
Oklahoma	− 0,090	− 0,142	− 1,102	− 2,065	− 2,001	− 2,502	− 4,238	− 3,672	− 3,247	− 4,137	− 3,822	− 4,440
Oregon	− 4,042	− 2,641	− 4,497	− 4,727	− 4,496	− 6,646	− 7,612^c	− 6,377	− 5,114	− 5,648	− 6,379	− 5,706
Pennsylvania	6,770^b	8,395^a	8,380^a	7,801^a	7,511^a	9,028^a	7,642^b	5,896^c	4,562	4,918	4,138	4,313
Rhode Island	1,738	1,504	2,910	2,501	1,428	0,485	− 0,517	0,681	0,843	1,945	1,529	0,677
S. Carolina	1,980	1,230	1,065	− 0,118	− 1,506	− 3,083	− 4,783^b	− 4,916^b	− 5,064^a	− 5,693^a	− 5,815^a	− 6,305^a
S. Dakota	3,266	2,201	2,458	2,603	1,534	2,262	6,069	5,653	6,814	6,759	7,206	7,300
Tennessee	− 2,577	− 2,768	− 3,128	− 3,491	− 4,601^b	− 4,952^b	− 6,307^a	− 7,057^a	− 8,131^a	− 8,470^a	− 7,926^a	− 8,435^a
Texas	4,760	3,231	3,040	3,372	3,604	4,932	4,514	5,309	2,183	2,738	3,692	2,508
Utah	2,912	3,374	2,965	2,312	4,461	3,238	2,024	1,478	1,461	1,568	2,104	1,554
Vermont	0,825	1,131	− 0,267	0,410	0,897	− 0,472	− 1,484	− 0,700	− 0,763	− 1,074	− 1,085	− 1,657
Virginia	4,738^c	5,678^b	5,271^b	5,182^c	4,883^c	3,633	2,299	2,121	0,759	0,191	− 0,966	− 1,745
Washington	0,770	1,047	− 0,722	− 0,887	− 1,831	− 2,873	− 2,557	− 2,043	− 0,259	− 1,072	− 2,859	− 4,011
West Virginia	− 0,489	2,107	1,669	1,403	1,013	1,412	0,421	− 0,629	− 1,137	− 1,644	− 0,732	− 0,584
Wisconsin	2,636	3,137	0,830	1,859	− 0,601	0,642	− 1,113	− 1,827	2,349	0,021	0,296	0,131
Wyoming	− 0,575	− 0,770	− 1,015	− 2,947	1,308	0,578	− 0,781	− 0,458	0,185	− 0,297	0,489	− 0,614

State	DR1	DR2	DR3	DR4	DR5	DR6	DR7	DR8	DR9	DR10	DR11	DR12
Alabama	1,666	1,807	1,211	0,711	− 1,469	− 3,368	− 4,642	− 4,971	− 6,125^b	− 7,364^a	− 7,574^a	− 7,971^a
Alaska	− 1,909	− 1,227	− 1,024	− 0,604	− 0,306	0,224	− 0,056	− 0,021	0,310	− 0,387	− 0,723	− 1,365
Arizona	5,323^a	4,887^b	3,579^c	2,539	2,555	1,724	1,181	0,615	1,059	1,198	2,362	2,366
Arkansas	− 2,974	− 2,453	− 3,073	− 2,982	− 3,925	− 3,190	− 4,229	− 5,536^c	− 6,282^b	− 5,264	− 3,840	− 5,204
California	1,332	4,452	3,063	0,126	0,173	− 0,585	− 1,883	− 1,965	− 1,551	− 1,687	− 0,403	2,170
Colorado	− 0,332	− 0,086	0,339	− 0,325	0,726	0,587	− 0,577	− 0,816	− 0,772	− 1,133	− 1,014	− 1,410
Connecticut	6,486^b	6,963^b	7,106^b	7,621^a	5,145^b	3,917	3,094	2,848	2,197	2,331	2,225	2,351
Delaware	2,601	3,380	3,277	2,503	3,582	2,551	1,954	− 0,305	− 1,723	− 2,576	− 2,911	− 2,772
Dist. of Col.	6,128^b	6,588^b	7,486^a	7,193^a	5,939^b	4,476	4,250	3,284	1,257	2,514	1,339	1,203
Florida	5,563	5,749	7,103^c	6,851^c	4,517	2,784	1,878	2,715	− 0,583	− 0,316	− 2,329	− 3,408
Georgia	2,652	1,816	1,615	0,584	− 1,037	− 2,002	− 2,822	− 3,305	− 4,110^c	− 4,735^b	− 5,066^b	− 5,268^b
Hawaii	3,616^b	4,047^b	1,798	0,765	0,042	− 0,307	− 1,051	− 1,192	− 0,425	− 0,878	− 0,544	− 0,572
Idaho	1,942	1,633	− 0,312	− 1,419	− 2,004	− 3,021	− 2,945	− 3,291	− 1,530	− 0,879	− 1,875	− 2,259
Illinois	− 1,789	− 0,992	− 2,256	− 3,032	− 2,970	− 2,196	− 3,566	− 3,651	− 1,861	− 2,406	− 2,340	− 2,583
Indiana	1,120	2,248	1,752	1,441	1,727	2,500	1,084	0,776	1,832	0,060	0,604	0,807
Iowa	1,091	1,378	0,336	− 0,083	0,111	1,714	1,100	1,980	2,651	2,674	2,788	2,557
Kansas	− 1,371	− 1,386	− 1,810	− 3,283	− 2,844	− 1,985	− 2,798	− 2,503	− 1,984	− 2,550	− 2,602	− 3,128
Kentucky	− 1,191	− 0,361	− 0,088	− 0,185	− 0,856	− 1,103	− 2,059	− 2,612	− 2,640	− 3,363	− 3,186	− 3,111
Louisiana	− 5,137	− 4,930	− 6,861	− 7,969	− 5,778	− 6,588	− 7,311	− 8,601	− 9,154	− 11,908^c	− 8,607	− 11,004^c
Maine	0,219	0,630	− 0,028	0,710	1,165	0,972	1,171	1,420	1,345	0,794	0,756	1,015
Maryland	1,835	2,870	2,523	1,829	1,865	1,395	0,566	− 0,642	− 0,994	− 0,971	− 1,820	− 1,565
Massachusetts	0,644	2,368	2,783	2,836	2,137	0,055	− 1,240	− 0,486	0,738	2,135	1,469	1,896
Michigan	0,448	1,371	− 3,020	− 2,980	− 3,303	− 0,503	− 1,987	− 3,254	1,810	− 1,227	− 1,568	− 0,047
Minnesota	− 0,724	− 0,218	− 0,633	− 0,049	− 1,597	− 1,792	− 3,414	− 4,110	− 1,545	− 0,945	0,402	0,392
Mississippi	0,512	0,763	0,649	− 0,495	− 0,912	− 0,517	− 1,046	0,349	0,317	0,121	0,213	− 0,756
Missouri	− 0,830	− 0,524	− 1,453	− 1,902	− 2,454	− 2,473	− 3,994	− 4,247	− 3,520	− 4,072	− 4,052	− 3,873
Montana	− 0,957	− 0,364	− 0,862	− 1,319	− 0,004	− 2,080	− 2,182	− 2,193	− 0,436	− 1,550	− 0,748	− 0,536
Nebraska	− 3,162	− 3,305	− 3,446	− 4,411	− 2,830	− 2,773	− 4,173	− 2,315	− 1,676	− 2,270	− 1,951	− 1,634
Nevada	− 1,626	− 0,514	− 0,342	− 1,182	1,010	− 0,527	− 1,700	− 2,240	− 1,235	− 1,242	− 1,449	− 0,750
New Hamp.	2,516	2,712	2,125	2,421	2,377	2,097	1,989	2,469	2,358	2,571	2,288	2,418
New Jersey	4,090	5,008^c	4,575	3,707	4,047	2,558	1,210	0,370	− 0,356	− 0,867	− 1,966	− 0,882
New Mexico	4,048	3,461	2,552	1,480	0,596	0,085	− 0,597	− 1,058	0,436	− 0,565	0,176	− 0,168
New York	5,949^b	7,610^a	8,551^b	9,020^a	7,486^a	4,587	3,270	1,918	0,684	1,614	0,504	1,500
N. Carolina	4,783	5,034	4,271	3,506	0,352	− 1,673	− 3,681	− 3,796	− 5,197	− 5,848^b	− 7,159^a	− 8,071^a
N. Dakota	4,931	4,104	3,525	1,544	1,763	2,303	1,670	0,823	2,424	2,846	5,110	4,107
Ohio	1,501	3,415	3,420	3,239	4,124	5,648^c	4,032	3,805	4,479	2,825	3,446	3,338
Oklahoma	− 0,090	− 0,142	− 1,102	− 2,065	− 2,001	− 2,502	− 4,238	− 3,672	− 3,247	− 4,137	− 3,822	− 4,440
Oregon	− 4,042	− 2,641	− 4,497	− 4,727	− 4,496	− 6,646	− 7,612^c	− 6,377	− 5,114	− 5,648	− 6,379	− 5,706
Pennsylvania	6,770^b	8,395^a	8,380^a	7,801^a	7,511^a	9,028^a	7,642^b	5,896^c	4,562	4,918	4,138	4,313
Rhode Island	1,738	1,504	2,910	2,501	1,428	0,485	− 0,517	0,681	0,843	1,945	1,529	0,677
S. Carolina	1,980	1,230	1,065	− 0,118	− 1,506	− 3,083	− 4,783^b	− 4,916^b	− 5,064^a	− 5,693^a	− 5,815^a	− 6,305^a
S. Dakota	3,266	2,201	2,458	2,603	1,534	2,262	6,069	5,653	6,814	6,759	7,206	7,300
Tennessee	− 2,577	− 2,768	− 3,128	− 3,491	− 4,601^b	− 4,952^b	− 6,307^a	− 7,057^a	− 8,131^a	− 8,470^a	− 7,926^a	− 8,435^a
Texas	4,760	3,231	3,040	3,372	3,604	4,932	4,514	5,309	2,183	2,738	3,692	2,508
Utah	2,912	3,374	2,965	2,312	4,461	3,238	2,024	1,478	1,461	1,568	2,104	1,554
Vermont	0,825	1,131	− 0,267	0,410	0,897	− 0,472	− 1,484	− 0,700	− 0,763	− 1,074	− 1,085	− 1,657
Virginia	4,738^c	5,678^b	5,271^b	5,182^c	4,883^c	3,633	2,299	2,121	0,759	0,191	− 0,966	− 1,745
Washington	0,770	1,047	− 0,722	− 0,887	− 1,831	− 2,873	− 2,557	− 2,043	− 0,259	− 1,072	− 2,859	− 4,011
West Virginia	− 0,489	2,107	1,669	1,403	1,013	1,412	0,421	− 0,629	− 1,137	− 1,644	− 0,732	− 0,584
Wisconsin	2,636	3,137	0,830	1,859	− 0,601	0,642	− 1,113	− 1,827	2,349	0,021	0,296	0,131
Wyoming	− 0,575	− 0,770	− 1,015	− 2,947	1,308	0,578	− 0,781	− 0,458	0,185	− 0,297	0,489	− 0,614

^aIndicates 1% statistical significance.^bIndicates 5% statistical significance.^cIndicates 10% statistical significance.

Accordingly, while CO₂ emissions generally increase the SPEI drought index in the first 6 months of the year, they reduce it in the last 6 months of the year. Negative values of the SPEI indicate dry periods, and positive values indicate wetting periods. Therefore, a decrease in SPEI values indicates a transition from a wetting period to a dry period, and an increase indicates a transition from a dry period to a more wetting period. Therefore, this situation can be explained as CO₂ emissions generally lead to drought in the last 6 months of the year. This study seeks to empirically test the hypothesis that CO₂ emissions, by driving climate change, can increase the frequency and intensity of droughts through their impact on regional precipitation and temperature patterns. The results of this study support the hypothesis that CO₂ emissions, through their impact on climate change, contribute to changes in drought patterns in the US.

Figure 2 shows the CO₂ emission map of the US. When the CO₂ emissions map is examined, it is seen that the CO₂ emission values of the US states, especially in the eastern region, are at very high levels. When the results obtained in this study are examined, it is seen that CO₂ emissions increase the number of dry periods, especially in the states located in the southeastern region of the US and increase the number of periods with excessive precipitation in the states located in the northeastern region, and the findings obtained are directly related to the areas shown on the CO₂ emission intensity map. It has been stated in previous studies in the literature that CO₂ emissions cause an increase in air temperature values in the region where they are emitted, and this causes a change in precipitation regimes, causing either a completely dry situation or climatic conditions with heavy rains. Therefore, it is considered that the findings obtained in this study closely match the situations stated in previous studies in the literature.

Figure 2

US CO2 emissions map. Source: Ecological and Environmental Informatics Research, Northern Arizona University.

US CO₂ emissions map. Source: Ecological and Environmental Informatics Research, Northern Arizona University.

After estimating the parameters on state bases with the AMG analysis, error correction-based PMG analyses are performed to see short- and long-term panel estimations.

According to Table 6, since the error correction term values are negative and statistically significant, it can be said that there are cointegration relationships between the variables. According to the long-term analysis results, CO₂ emissions have long-term effects on droughts in all months. According to the short-term analysis results, CO₂ emissions affect the SPEI drought index in February, April, May, June, July, September, November, and December, while they do not affect the SPEI drought index in other months. It can be seen that CO₂ emissions affect drought both in the short- and long-run, increase the SPEI drought index (in other words increase the wetting) in the short-run, and decrease the SPEI drought index (namely, increase the drought) in the long-run. Studies in the literature (such as Strzepek et al. 2010; Hardin et al. 2017) have also found that in periods when CO₂ emissions increase, drought also increases and there is a direct relationship between them. Therefore, the findings obtained from this analysis are in line with the previous studies in the literature.

Table 6

PMG analyses based on the error correction (short-run and long-run estimations)

Dep. var.	ΔDR1	ΔDR2	ΔDR3	ΔDR4	ΔDR5	ΔDR6
Long run	lnCO₂ −1.46^a (0.00)	lnCO₂ −1.67^a (0.00)	lnCO₂ −1.67^a (0.00)	lnCO₂ −1.89^a (0.00)	lnCO₂ −1.57^a (0.00)	lnCO₂ −1.29^a (0.00)
Short run	ΔlnCO₂ −0.22 (0.76)	ΔlnCO₂ 1.23^c (0.095)	ΔlnCO₂ 0.91 (0.22)	ΔlnCO₂ 1.32^c (0.09)	ΔlnCO₂ 1.43^b (0.03)	ΔlnCO₂ 1.35^a (0.0095)
Constant	6.23^a(0.00)	6.94^a (0.00)	6.98^a (0.00)	7.50^a (0.00)	6.45^a (0.00)	5.59^a (0.00)
Error correction	−0.96^a (0.00)	−0.93^a (0.00)	−0.95^a (0.00)	−0.90^a (0.00)	−0.93^a (0.00)	−0.97^a (0.00)
Dep. var.	ΔDR7	ΔDR8	ΔDR9	ΔDR10	ΔDR11	ΔDR12
Long run	lnCO₂ −1.23^a (0.00)	lnCO₂ −1.17^a (0.00)	lnCO₂ −1.09^a (0.00)	lnCO₂ −0.95^a (0.00)	lnCO₂ −1.20^a (0.00)	lnCO₂ −1.04^a (0.00)
Short run	ΔlnCO₂ 0.96^c (0.07)	ΔlnCO₂ 0.82 (0.13)	ΔlnCO₂ 0.99^c (0.08)	ΔlnCO₂ 0.33 (0.61)	ΔlnCO₂ 1.44^b (0.03)	ΔlnCO₂ 1.66^b (0.02)
Constant	5.48^a (0.00)	5.20^a (0.00)	4.83^a (0.00)	4.17^a (0.00)	5.07^a (0.00)	4.48^a (0.00)
Error correction	−1.01^a (0.00)	−1.01^a (0.00)	−1,01^a (0.00)	−0.98^a (0.00)	−0.95^a (0.00)	−0.97^a (0.00)

Dep. var.	ΔDR1	ΔDR2	ΔDR3	ΔDR4	ΔDR5	ΔDR6
Long run	lnCO₂ −1.46^a (0.00)	lnCO₂ −1.67^a (0.00)	lnCO₂ −1.67^a (0.00)	lnCO₂ −1.89^a (0.00)	lnCO₂ −1.57^a (0.00)	lnCO₂ −1.29^a (0.00)
Short run	ΔlnCO₂ −0.22 (0.76)	ΔlnCO₂ 1.23^c (0.095)	ΔlnCO₂ 0.91 (0.22)	ΔlnCO₂ 1.32^c (0.09)	ΔlnCO₂ 1.43^b (0.03)	ΔlnCO₂ 1.35^a (0.0095)
Constant	6.23^a(0.00)	6.94^a (0.00)	6.98^a (0.00)	7.50^a (0.00)	6.45^a (0.00)	5.59^a (0.00)
Error correction	−0.96^a (0.00)	−0.93^a (0.00)	−0.95^a (0.00)	−0.90^a (0.00)	−0.93^a (0.00)	−0.97^a (0.00)
Dep. var.	ΔDR7	ΔDR8	ΔDR9	ΔDR10	ΔDR11	ΔDR12
Long run	lnCO₂ −1.23^a (0.00)	lnCO₂ −1.17^a (0.00)	lnCO₂ −1.09^a (0.00)	lnCO₂ −0.95^a (0.00)	lnCO₂ −1.20^a (0.00)	lnCO₂ −1.04^a (0.00)
Short run	ΔlnCO₂ 0.96^c (0.07)	ΔlnCO₂ 0.82 (0.13)	ΔlnCO₂ 0.99^c (0.08)	ΔlnCO₂ 0.33 (0.61)	ΔlnCO₂ 1.44^b (0.03)	ΔlnCO₂ 1.66^b (0.02)
Constant	5.48^a (0.00)	5.20^a (0.00)	4.83^a (0.00)	4.17^a (0.00)	5.07^a (0.00)	4.48^a (0.00)
Error correction	−1.01^a (0.00)	−1.01^a (0.00)	−1,01^a (0.00)	−0.98^a (0.00)	−0.95^a (0.00)	−0.97^a (0.00)

Note: Probability values are shown in parentheses.

^aIndicates 1% statistical significance.

^bIndicates 5% statistical significance.

^cIndicates 10% statistical significance.

In the next step, the Dumitrescu-Hurlin panel causality test, which is used to test causality in heterogeneous panel data models, was conducted to investigate the mutual causality relationship between the variables.

In Table 7, when the causality relationships between the variables are analyzed, it is found that SPEI drought index values in all months are the causes of CO₂ emissions. When the causality relationships from CO₂ emissions to SPEI drought index values are examined, it is found that CO₂ emissions in all months except January and April are the causes of SPEI drought index values.

Table 7

Dumitrescu & Hurlin (2012) panel causality test

DR1 → lnCO₂

DR2 → lnCO₂

DR3 → lnCO₂

DR4 → lnCO₂

DR5 → lnCO₂

DR6 → lnCO₂

1.46^b (0.02)

1.75^a (0.00)

1.63^a (0.00)

1.79^a (0.00)

2.22^a (0.00)

2.08^a (0.00)

lnCO₂ → DR1

lnCO₂ → DR2

lnCO₂ → DR3

lnCO₂ → DR4

lnCO₂ → DR5

lnCO₂ → DR6

1.25 (0.20)

1.40^b (0.04)

1.38^c (0.06)

1.31 (0.12)

1.41^b (0.04)

1.50^b (0.01)

DR7 → lnCO₂

DR8 → lnCO₂

DR9 → lnCO₂

DR10 → lnCO₂

DR11 → lnCO₂

DR12 → lnCO₂

1.79^a (0.00)

1.60^a (0.00)

1.90^a (0.00)

1.49^b (0.01)

1.84^a (0.00)

1.82^a (0.00)

lnCO₂ → DR7

lnCO₂ → DR8

lnCO₂ → DR9

lnCO₂ → DR10

lnCO₂ → DR11

lnCO₂ → DR12

1.61^a (0.00)

1.64^a (0.00)

1.66^a (0.00)

1.51^a (0.0096)

1.53^a (0.008)

1.36^c (0.07)

Note: Wald Statistic: probability values are given in parentheses.

^aIndicates 1% statistical significance.

^bIndicates 5% statistical significance.

^cIndicates 10% statistical significance.

To examine the causality results by states, the individual results of the Dumitrescu-Hurlin panel Granger causality tests are given in Table 8.

Table 8

Dumitrescu & Hurlin panel causality test (state-specific causality)

	Null hypothesis: No causality
State	lnCO₂ → DR1	lnCO₂ → DR2	lnCO₂ → DR3	lnCO₂ → DR4	lnCO₂ → DR5	lnCO₂ → DR6
Alabama	3,993^c(0,059)	4,838^b(0,040)	4,090^c(0,057)	4,440^b(0,048)	4,527^b(0,046)	5,204^b(0,034)
Alaska	7,828^b(0,011)	7,607^b(0,012)	6,634^b(0,018)	5,943^b(0,024)	5,094^b(0,035)	6,345^b(0,020)
Arizona	0,031 (0,862)	0,005 (0,944)	0,102 (0,753)	0,078 (0,783)	0,027 (0,871)	0,024 (0,878)
Arkansas	0,004 (0,950)	0,043 (0,839)	0,214 (0,649)	0,352 (0,559)	1,644 (0,214)	2,182 (0,155)
California	0,288 (0,598)	0,142 (0,710)	0,031 (0,861)	0,001 (0,972)	0,001 (0,972)	0,011 (0,916)
Colorado	3,765^c(0,067)	4,042^c(0,058)	4,297^c(0,051)	2,975^c(0,100)	2,378 (0,139)	1,928 (0,180)
Connecticut	0,777 (0,389)	0,738 (0,400)	0,830 (0,373)	0,391 (0,539)	0,790 (0,385)	1,696 (0,208)
Delaware	1,068 (0,314)	1,241 (0,278)	1,270 (0,273)	1,446 (0,243)	0,965 (0,338)	0,427 (0,521)
Dist. of Col.	0,059 (0,810)	0,039 (0,846)	0,108 (0,745)	0,277 (0,604)	0,103 (0,752)	0,015 (0,904)
Florida	1,989 (0,174)	2,297 (0,145)	2,236 (0,150)	1,991 (0,174)	1,851 (0,189)	1,731 (0,203)
Georgia	1,726 (0,204)	2,658 (0,119)	2,873 (0,106)	3,436^c(0,079)	4,490^b(0,047)	4,997^b(0,037)
Hawaii	0,014 (0,908)	0,396 (0,536)	0,674 (0,421)	0,533 (0,474)	0,509 (0,484)	0,499 (0,488)
Idaho	0,380 (0,545)	0,074 (0,789)	0,235 (0,633)	0,223 (0,642)	0,078 (0,782)	0,809 (0,379)
Illinois	2,940 (0,102)	3,321^c(0,083)	2,481 (0,131)	3,402^c(0,080)	3,679^c(0,069)	3,606^c(0,072)
Indiana	2,103 (0,162)	2,215 (0,152)	1,771 (0,198)	1,156 (0,295)	1,212 (0,284)	0,855 (0,366)
Iowa	0,926 (0,347)	1,006 (0,328)	1,152 (0,296)	0,675 (0,421)	0,462 (0,505)	0,013 (0,911)
Kansas	0,187 (0,670)	0,172 (0,683)	0,195 (0,664)	0,113 (0,740)	0,019 (0,891)	0,019 (0,892)
Kentucky	5,922^b(0,024)	6,003^b(0,024)	6,074^b(0,023)	3,310^c(0,084)	4,022^c(0,059)	2,431 (0,135)
Louisiana	1,521 (0,232)	1,609 (0,219)	0,657 (0,427)	0,980 (0,334)	2,274 (0,147)	2,409 (0,136)
Maine	0,216 (0,647)	0,048 (0,829)	0,050 (0,825)	0,005 (0,946)	0,011 (0,918)	0,037 (0,850)
Maryland	0,204 (0,657)	0,344 (0,564)	0,496 (0,489)	0,695 (0,414)	0,609 (0,444)	0,476 (0,498)
Massachusetts	0,017 (0,897)	0,033 (0,858)	0,011 (0,918)	0,016 (0,900)	0,041 (0,842)	0,237 (0,632)
Michigan	2,631 (0,120)	2,833 (0,108)	1,919 (0,181)	2,090 (0,164)	2,682 (0,117)	2,924 (0,103)
Minnesota	0,029 (0,867)	0,022 (0,885)	0,022 (0,883)	0,134 (0,719)	0,266 (0,612)	1,310 (0,266)
Mississippi	0,746 (0,398)	0,745 (0,398)	0,717 (0,407)	0,587 (0,452)	0,049 (0,827)	0,058 (0,812)
Missouri	0,872 (0,361)	0,926 (0,347)	1,351 (0,259)	1,729 (0,203)	1,944 (0,178)	2,235 (0,150)
Montana	0,345 (0,563)	0,595 (0,450)	0,619 (0,441)	0,543 (0,470)	0,536 (0,473)	0,976 (0,335)
Nebraska	0,904 (0,353)	0,934 (0,345)	0,775 (0,389)	0,925 (0,348)	0,182 (0,674)	0,040 (0,843)
Nevada	0,089 (0,769)	0,000 (0,997)	0,008 (0,929)	0,093 (0,764)	0,003 (0,953)	0,017 (0,896)
New Hamp.	0,602 (0,447)	0,703 (0,412)	0,627 (0,438)	0,719 (0,407)	1,218 (0,283)	1,859 (0,188)
New Jersey	0,079 (0,782)	0,026 (0,874)	0,005 (0,947)	0,034 (0,856)	0,000 (0,994)	0,038 (0,848)
New Mexico	0,278 (0,604)	0,284 (0,600)	0,150 (0,702)	0,097 (0,759)	0,103 (0,752)	0,078 (0,784)
New York	0,012 (0,913)	0,001 (0,970)	0,008 (0,929)	0,139 (0,713)	0,077 (0,785)	0,008 (0,929)
N. Carolina	2,906 (0,104)	3,749^c(0,067)	3,907^c(0,062)	4,288^c(0,052)	5,163^b(0,034)	4,874^b(0,039)
N. Dakota	0,460 (0,505)	0,481 (0,496)	0,660 (0,426)	0,794 (0,384)	0,460 (0,506)	0,000 (0,983)
Ohio	0,692 (0,415)	0,556 (0,465)	0,707 (0,410)	0,661 (0,426)	0,841 (0,370)	0,741 (0,400)
Oklahoma	2,445 (0,134)	2,652 (0,119)	2,447 (0,133)	1,289 (0,270)	0,476 (0,498)	0,749 (0,397)
Oregon	0,416 (0,526)	0,870 (0,362)	0,324 (0,575)	0,335 (0,569)	1,050 (0,318)	0,295 (0,593)
Pennsylvania	0,029 (0,867)	0,013 (0,911)	0,010 (0,923)	0,013 (0,909)	0,002 (0,966)	0,096 (0,760)
Rhode Island	0,176 (0,679)	0,198 (0,661)	0,131 (0,722)	0,000 (0,986)	0,043 (0,838)	0,322 (0,577)
S. Carolina	3,466^c(0,077)	4,274^c(0,052)	4,235^c(0,053)	4,762^b(0,041)	6,042^b(0,023)	5,517^b(0,029)
S. Dakota	2,420 (0,135)	2,865 (0,106)	3,103^c(0,093)	2,372 (0,139)	2,308 (0,144)	2,158 (0,157)
Tennessee	4,092^c(0,057)	4,534^b(0,046)	5,699^b(0,027)	5,694^b(0,027)	7,065^b(0,015)	7,226^b(0,014)
Texas	1,170 (0,292)	1,327 (0,263)	2,078 (0,165)	2,552 (0,126)	3,333^c(0,083)	3,545^c(0,074)
Utah	0,489 (0,493)	0,177 (0,678)	0,081 (0,779)	0,325 (0,575)	0,296 (0,592)	0,415 (0,527)
Vermont	0,082 (0,777)	0,033 (0,857)	0,015 (0,902)	0,005 (0,947)	0,177 (0,678)	0,027 (0,872)
Virginia	0,112 (0,742)	0,322 (0,577)	0,271 (0,608)	0,385 (0,542)	0,351 (0,560)	0,592 (0,451)
Washington	0,188 (0,669)	0,354 (0,558)	0,244 (0,627)	0,112 (0,741)	0,044 (0,836)	0,001 (0,971)
West Virginia	0,292 (0,595)	0,407 (0,531)	0,777 (0,389)	1,147 (0,297)	0,808 (0,379)	0,585 (0,453)
Wisconsin	0,334 (0,570)	0,404 (0,532)	0,218 (0,645)	1,240 (0,279)	1,281 (0,271)	3,841^c(0,064)
Wyoming	1,531 (0,230)	2,330 (0,143)	2,552 (0,126)	1,086 (0,310)	0,228 (0,638)	0,214 (0,649)
	lnCO₂ → DR7	lnCO₂ → DR8	lnCO₂ → DR9	lnCO₂ → DR10	lnCO₂ → DR11	lnCO₂ → DR12
Alabama	4,879^b(0,039)	5,125^b(0,035)	4,301^c(0,051)	4,127^c(0,056)	3,842^c(0,064)	2,816 (0,109)
Alaska	8,972^a(0,007)	5,022^b(0,037)	6,078^b(0,023)	6,047^b(0,023)	7,085^b(0,015)	6,970^b(0,016)
Arizona	0,139 (0,714)	0,379 (0,545)	0,191 (0,666)	0,386 (0,541)	0,478 (0,497)	0,276 (0,605)
Arkansas	2,146 (0,158)	2,883 (0,105)	3,756^c(0,067)	4,518^b(0,046)	5,968^b(0,024)	5,010^b(0,037)
California	0,016 (0,901)	0,020 (0,888)	0,019 (0,892)	0,007 (0,936)	0,017 (0,897)	0,137 (0,715)
Colorado	1,207 (0,285)	0,985 (0,333)	1,335 (0,262)	0,906 (0,353)	0,892 (0,356)	0,764 (0,393)
Connecticut	1,267 (0,274)	0,747 (0,398)	0,256 (0,618)	0,437 (0,516)	0,172 (0,683)	0,050 (0,826)
Delaware	0,535 (0,473)	0,900 (0,354)	1,037 (0,321)	0,738 (0,401)	0,766 (0,392)	0,392 (0,538)
Dist. of Col.	0,056 (0,816)	0,510 (0,483)	0,561 (0,463)	0,489 (0,493)	0,730 (0,403)	0,736 (0,401)
Florida	1,395 (0,251)	1,352 (0,259)	2,300 (0,145)	1,232 (0,280)	0,821 (0,376)	0,012 (0,913)
Georgia	5,405^b(0,031)	6,591^b(0,018)	6,190^b(0,022)	6,181^b(0,022)	5,788^b(0,026)	5,096^b(0,035)
Hawaii	0,899 (0,354)	0,506 (0,485)	0,417 (0,526)	0,277 (0,605)	0,273 (0,607)	1,310 (0,266)
Idaho	1,031 (0,322)	1,044 (0,319)	0,259 (0,616)	0,323 (0,576)	0,521 (0,479)	0,202 (0,658)
Illinois	3,384^c(0,081)	2,702 (0,116)	1,716 (0,205)	0,957 (0,340)	0,840 (0,370)	0,570 (0,459)
Indiana	0,679 (0,420)	0,533 (0,474)	1,126 (0,301)	1,350 (0,259)	1,581 (0,223)	0,796 (0,383)
Iowa	0,001 (0,975)	0,021 (0,886)	0,082 (0,778)	0,067 (0,798)	0,064 (0,803)	0,043 (0,837)
Kansas	0,100 (0,756)	0,047 (0,831)	0,055 (0,818)	0,092 (0,765)	0,054 (0,819)	0,100 (0,755)
Kentucky	2,208 (0,153)	2,241 (0,150)	4,090^c(0,057)	5,100^b(0,035)	6,003^b(0,024)	5,748^b(0,026)
Louisiana	2,895 (0,104)	2,103 (0,162)	0,354 (0,559)	0,393 (0,538)	1,270 (0,273)	1,334 (0,262)
Maine	0,000 (0,997)	0,000 (0,996)	0,139 (0,713)	0,076 (0,785)	0,152 (0,701)	0,134 (0,718)
Maryland	0,799 (0,382)	1,652 (0,213)	2,268 (0,148)	2,415 (0,136)	2,693 (0,116)	2,595 (0,123)
Massachusetts	0,453 (0,508)	0,724 (0,405)	0,948 (0,342)	0,992 (0,331)	1,032 (0,322)	0,893 (0,356)
Michigan	3,049^c(0,096)	3,161^c(0,091)	2,472 (0,132)	2,171 (0,156)	1,830 (0,191)	1,639 (0,215)
Minnesota	1,718 (0,205)	1,581 (0,223)	1,989 (0,174)	1,763 (0,199)	1,473 (0,239)	1,162 (0,294)
Mississippi	0,003 (0,955)	0,002 (0,967)	0,217 (0,647)	0,380 (0,544)	0,503 (0,487)	0,815 (0,377)
Missouri	2,163 (0,157)	1,432 (0,246)	1,401 (0,250)	1,331 (0,262)	0,976 (0,335)	0,622 (0,440)
Montana	0,288 (0,598)	0,092 (0,765)	0,630 (0,437)	0,335 (0,569)	0,507 (0,485)	0,640 (0,433)
Nebraska	0,108 (0,745)	0,167 (0,687)	0,161 (0,692)	0,105 (0,749)	0,088 (0,769)	0,062 (0,806)
Nevada	0,038 (0,847)	0,006 (0,941)	0,015 (0,904)	0,106 (0,749)	0,138 (0,715)	0,351 (0,560)
New Hamp.	1,736 (0,202)	1,504 (0,234)	1,480 (0,238)	1,380 (0,254)	0,862 (0,364)	0,568 (0,460)
New Jersey	0,001 (0,982)	0,022 (0,883)	0,019 (0,891)	0,010 (0,921)	0,075 (0,787)	0,036 (0,851)
New Mexico	0,504 (0,486)	0,811 (0,379)	1,378 (0,254)	1,382 (0,253)	1,127 (0,301)	1,158 (0,295)
New York	0,000 (0,995)	0,003 (0,958)	0,048 (0,829)	0,013 (0,911)	0,068 (0,797)	0,067 (0,799)
N. Carolina	6,261^b(0,021)	7,565^b(0,012)	7,099^b(0,015)	6,552^b(0,019)	4,515^b(0,046)	3,899^c(0,062)
N. Dakota	0,089 (0,769)	0,002 (0,967)	0,000 (0,984)	0,000 (0,995)	0,007 (0,933)	0,011 (0,917)
Ohio	0,713 (0,409)	1,102 (0,306)	1,043 (0,319)	0,914 (0,350)	1,274 (0,272)	0,866 (0,363)
Oklahoma	0,490 (0,492)	0,516 (0,481)	0,491 (0,492)	0,268 (0,610)	0,077 (0,784)	0,151 (0,702)
Oregon	0,165 (0,689)	0,490 (0,492)	1,532 (0,230)	1,191 (0,288)	1,759 (0,200)	2,991^c(0,099)
Pennsylvania	0,086 (0,773)	0,026 (0,873)	0,001 (0,979)	0,001 (0,971)	0,063 (0,805)	0,086 (0,772)
Rhode Island	0,236 (0,632)	0,199 (0,661)	0,359 (0,556)	0,119 (0,733)	0,018 (0,896)	0,030 (0,863)
S. Carolina	6,962^b(0,016)	6,989^b(0,016)	6,984^b(0,016)	6,205^b(0,022)	5,734^b(0,027)	4,331^b(0,050)
S. Dakota	2,343 (0,142)	2,105 (0,162)	1,817 (0,193)	1,745 (0,201)	1,630 (0,216)	1,442 (0,244)
Tennessee	4,568^b(0,045)	4,739^b(0,042)	3,777^c(0,067)	3,424^c(0,079)	4,104^c(0,056)	3,711^c(0,068)
Texas	3,259^c(0,086)	3,656^c(0,070)	2,619 (0,121)	2,340 (0,142)	2,041 (0,169)	1,445 (0,243)
Utah	0,647 (0,431)	0,727 (0,404)	0,398 (0,535)	0,041 (0,841)	0,064 (0,804)	0,132 (0,721)
Vermont	0,037 (0,849)	0,111 (0,743)	0,110 (0,743)	0,000 (0,991)	0,004 (0,953)	0,056 (0,816)
Virginia	1,003 (0,328)	1,351 (0,259)	2,033 (0,169)	2,079 (0,165)	1,136 (0,299)	0,961 (0,339)
Washington	0,010 (0,922)	0,059 (0,811)	0,303 (0,588)	0,603 (0,447)	0,177 (0,679)	0,156 (0,697)
West Virginia	0,265 (0,612)	0,471 (0,500)	0,531 (0,475)	0,447 (0,512)	0,894 (0,356)	0,710 (0,409)
Wisconsin	6,547^b(0,019)	7,472^b(0,013)	6,835^b(0,017)	4,424^b(0,048)	4,827^b(0,040)	4,255^c(0,052)
Wyoming	0,133 (0,719)	1,116 (0,303)	1,193 (0,288)	0,732 (0,402)	0,841 (0,370)	0,852 (0,367)

	Null hypothesis: No causality
State	lnCO₂ → DR1	lnCO₂ → DR2	lnCO₂ → DR3	lnCO₂ → DR4	lnCO₂ → DR5	lnCO₂ → DR6
Alabama	3,993^c(0,059)	4,838^b(0,040)	4,090^c(0,057)	4,440^b(0,048)	4,527^b(0,046)	5,204^b(0,034)
Alaska	7,828^b(0,011)	7,607^b(0,012)	6,634^b(0,018)	5,943^b(0,024)	5,094^b(0,035)	6,345^b(0,020)
Arizona	0,031 (0,862)	0,005 (0,944)	0,102 (0,753)	0,078 (0,783)	0,027 (0,871)	0,024 (0,878)
Arkansas	0,004 (0,950)	0,043 (0,839)	0,214 (0,649)	0,352 (0,559)	1,644 (0,214)	2,182 (0,155)
California	0,288 (0,598)	0,142 (0,710)	0,031 (0,861)	0,001 (0,972)	0,001 (0,972)	0,011 (0,916)
Colorado	3,765^c(0,067)	4,042^c(0,058)	4,297^c(0,051)	2,975^c(0,100)	2,378 (0,139)	1,928 (0,180)
Connecticut	0,777 (0,389)	0,738 (0,400)	0,830 (0,373)	0,391 (0,539)	0,790 (0,385)	1,696 (0,208)
Delaware	1,068 (0,314)	1,241 (0,278)	1,270 (0,273)	1,446 (0,243)	0,965 (0,338)	0,427 (0,521)
Dist. of Col.	0,059 (0,810)	0,039 (0,846)	0,108 (0,745)	0,277 (0,604)	0,103 (0,752)	0,015 (0,904)
Florida	1,989 (0,174)	2,297 (0,145)	2,236 (0,150)	1,991 (0,174)	1,851 (0,189)	1,731 (0,203)
Georgia	1,726 (0,204)	2,658 (0,119)	2,873 (0,106)	3,436^c(0,079)	4,490^b(0,047)	4,997^b(0,037)
Hawaii	0,014 (0,908)	0,396 (0,536)	0,674 (0,421)	0,533 (0,474)	0,509 (0,484)	0,499 (0,488)
Idaho	0,380 (0,545)	0,074 (0,789)	0,235 (0,633)	0,223 (0,642)	0,078 (0,782)	0,809 (0,379)
Illinois	2,940 (0,102)	3,321^c(0,083)	2,481 (0,131)	3,402^c(0,080)	3,679^c(0,069)	3,606^c(0,072)
Indiana	2,103 (0,162)	2,215 (0,152)	1,771 (0,198)	1,156 (0,295)	1,212 (0,284)	0,855 (0,366)
Iowa	0,926 (0,347)	1,006 (0,328)	1,152 (0,296)	0,675 (0,421)	0,462 (0,505)	0,013 (0,911)
Kansas	0,187 (0,670)	0,172 (0,683)	0,195 (0,664)	0,113 (0,740)	0,019 (0,891)	0,019 (0,892)
Kentucky	5,922^b(0,024)	6,003^b(0,024)	6,074^b(0,023)	3,310^c(0,084)	4,022^c(0,059)	2,431 (0,135)
Louisiana	1,521 (0,232)	1,609 (0,219)	0,657 (0,427)	0,980 (0,334)	2,274 (0,147)	2,409 (0,136)
Maine	0,216 (0,647)	0,048 (0,829)	0,050 (0,825)	0,005 (0,946)	0,011 (0,918)	0,037 (0,850)
Maryland	0,204 (0,657)	0,344 (0,564)	0,496 (0,489)	0,695 (0,414)	0,609 (0,444)	0,476 (0,498)
Massachusetts	0,017 (0,897)	0,033 (0,858)	0,011 (0,918)	0,016 (0,900)	0,041 (0,842)	0,237 (0,632)
Michigan	2,631 (0,120)	2,833 (0,108)	1,919 (0,181)	2,090 (0,164)	2,682 (0,117)	2,924 (0,103)
Minnesota	0,029 (0,867)	0,022 (0,885)	0,022 (0,883)	0,134 (0,719)	0,266 (0,612)	1,310 (0,266)
Mississippi	0,746 (0,398)	0,745 (0,398)	0,717 (0,407)	0,587 (0,452)	0,049 (0,827)	0,058 (0,812)
Missouri	0,872 (0,361)	0,926 (0,347)	1,351 (0,259)	1,729 (0,203)	1,944 (0,178)	2,235 (0,150)
Montana	0,345 (0,563)	0,595 (0,450)	0,619 (0,441)	0,543 (0,470)	0,536 (0,473)	0,976 (0,335)
Nebraska	0,904 (0,353)	0,934 (0,345)	0,775 (0,389)	0,925 (0,348)	0,182 (0,674)	0,040 (0,843)
Nevada	0,089 (0,769)	0,000 (0,997)	0,008 (0,929)	0,093 (0,764)	0,003 (0,953)	0,017 (0,896)
New Hamp.	0,602 (0,447)	0,703 (0,412)	0,627 (0,438)	0,719 (0,407)	1,218 (0,283)	1,859 (0,188)
New Jersey	0,079 (0,782)	0,026 (0,874)	0,005 (0,947)	0,034 (0,856)	0,000 (0,994)	0,038 (0,848)
New Mexico	0,278 (0,604)	0,284 (0,600)	0,150 (0,702)	0,097 (0,759)	0,103 (0,752)	0,078 (0,784)
New York	0,012 (0,913)	0,001 (0,970)	0,008 (0,929)	0,139 (0,713)	0,077 (0,785)	0,008 (0,929)
N. Carolina	2,906 (0,104)	3,749^c(0,067)	3,907^c(0,062)	4,288^c(0,052)	5,163^b(0,034)	4,874^b(0,039)
N. Dakota	0,460 (0,505)	0,481 (0,496)	0,660 (0,426)	0,794 (0,384)	0,460 (0,506)	0,000 (0,983)
Ohio	0,692 (0,415)	0,556 (0,465)	0,707 (0,410)	0,661 (0,426)	0,841 (0,370)	0,741 (0,400)
Oklahoma	2,445 (0,134)	2,652 (0,119)	2,447 (0,133)	1,289 (0,270)	0,476 (0,498)	0,749 (0,397)
Oregon	0,416 (0,526)	0,870 (0,362)	0,324 (0,575)	0,335 (0,569)	1,050 (0,318)	0,295 (0,593)
Pennsylvania	0,029 (0,867)	0,013 (0,911)	0,010 (0,923)	0,013 (0,909)	0,002 (0,966)	0,096 (0,760)
Rhode Island	0,176 (0,679)	0,198 (0,661)	0,131 (0,722)	0,000 (0,986)	0,043 (0,838)	0,322 (0,577)
S. Carolina	3,466^c(0,077)	4,274^c(0,052)	4,235^c(0,053)	4,762^b(0,041)	6,042^b(0,023)	5,517^b(0,029)
S. Dakota	2,420 (0,135)	2,865 (0,106)	3,103^c(0,093)	2,372 (0,139)	2,308 (0,144)	2,158 (0,157)
Tennessee	4,092^c(0,057)	4,534^b(0,046)	5,699^b(0,027)	5,694^b(0,027)	7,065^b(0,015)	7,226^b(0,014)
Texas	1,170 (0,292)	1,327 (0,263)	2,078 (0,165)	2,552 (0,126)	3,333^c(0,083)	3,545^c(0,074)
Utah	0,489 (0,493)	0,177 (0,678)	0,081 (0,779)	0,325 (0,575)	0,296 (0,592)	0,415 (0,527)
Vermont	0,082 (0,777)	0,033 (0,857)	0,015 (0,902)	0,005 (0,947)	0,177 (0,678)	0,027 (0,872)
Virginia	0,112 (0,742)	0,322 (0,577)	0,271 (0,608)	0,385 (0,542)	0,351 (0,560)	0,592 (0,451)
Washington	0,188 (0,669)	0,354 (0,558)	0,244 (0,627)	0,112 (0,741)	0,044 (0,836)	0,001 (0,971)
West Virginia	0,292 (0,595)	0,407 (0,531)	0,777 (0,389)	1,147 (0,297)	0,808 (0,379)	0,585 (0,453)
Wisconsin	0,334 (0,570)	0,404 (0,532)	0,218 (0,645)	1,240 (0,279)	1,281 (0,271)	3,841^c(0,064)
Wyoming	1,531 (0,230)	2,330 (0,143)	2,552 (0,126)	1,086 (0,310)	0,228 (0,638)	0,214 (0,649)
	lnCO₂ → DR7	lnCO₂ → DR8	lnCO₂ → DR9	lnCO₂ → DR10	lnCO₂ → DR11	lnCO₂ → DR12
Alabama	4,879^b(0,039)	5,125^b(0,035)	4,301^c(0,051)	4,127^c(0,056)	3,842^c(0,064)	2,816 (0,109)
Alaska	8,972^a(0,007)	5,022^b(0,037)	6,078^b(0,023)	6,047^b(0,023)	7,085^b(0,015)	6,970^b(0,016)
Arizona	0,139 (0,714)	0,379 (0,545)	0,191 (0,666)	0,386 (0,541)	0,478 (0,497)	0,276 (0,605)
Arkansas	2,146 (0,158)	2,883 (0,105)	3,756^c(0,067)	4,518^b(0,046)	5,968^b(0,024)	5,010^b(0,037)
California	0,016 (0,901)	0,020 (0,888)	0,019 (0,892)	0,007 (0,936)	0,017 (0,897)	0,137 (0,715)
Colorado	1,207 (0,285)	0,985 (0,333)	1,335 (0,262)	0,906 (0,353)	0,892 (0,356)	0,764 (0,393)
Connecticut	1,267 (0,274)	0,747 (0,398)	0,256 (0,618)	0,437 (0,516)	0,172 (0,683)	0,050 (0,826)
Delaware	0,535 (0,473)	0,900 (0,354)	1,037 (0,321)	0,738 (0,401)	0,766 (0,392)	0,392 (0,538)
Dist. of Col.	0,056 (0,816)	0,510 (0,483)	0,561 (0,463)	0,489 (0,493)	0,730 (0,403)	0,736 (0,401)
Florida	1,395 (0,251)	1,352 (0,259)	2,300 (0,145)	1,232 (0,280)	0,821 (0,376)	0,012 (0,913)
Georgia	5,405^b(0,031)	6,591^b(0,018)	6,190^b(0,022)	6,181^b(0,022)	5,788^b(0,026)	5,096^b(0,035)
Hawaii	0,899 (0,354)	0,506 (0,485)	0,417 (0,526)	0,277 (0,605)	0,273 (0,607)	1,310 (0,266)
Idaho	1,031 (0,322)	1,044 (0,319)	0,259 (0,616)	0,323 (0,576)	0,521 (0,479)	0,202 (0,658)
Illinois	3,384^c(0,081)	2,702 (0,116)	1,716 (0,205)	0,957 (0,340)	0,840 (0,370)	0,570 (0,459)
Indiana	0,679 (0,420)	0,533 (0,474)	1,126 (0,301)	1,350 (0,259)	1,581 (0,223)	0,796 (0,383)
Iowa	0,001 (0,975)	0,021 (0,886)	0,082 (0,778)	0,067 (0,798)	0,064 (0,803)	0,043 (0,837)
Kansas	0,100 (0,756)	0,047 (0,831)	0,055 (0,818)	0,092 (0,765)	0,054 (0,819)	0,100 (0,755)
Kentucky	2,208 (0,153)	2,241 (0,150)	4,090^c(0,057)	5,100^b(0,035)	6,003^b(0,024)	5,748^b(0,026)
Louisiana	2,895 (0,104)	2,103 (0,162)	0,354 (0,559)	0,393 (0,538)	1,270 (0,273)	1,334 (0,262)
Maine	0,000 (0,997)	0,000 (0,996)	0,139 (0,713)	0,076 (0,785)	0,152 (0,701)	0,134 (0,718)
Maryland	0,799 (0,382)	1,652 (0,213)	2,268 (0,148)	2,415 (0,136)	2,693 (0,116)	2,595 (0,123)
Massachusetts	0,453 (0,508)	0,724 (0,405)	0,948 (0,342)	0,992 (0,331)	1,032 (0,322)	0,893 (0,356)
Michigan	3,049^c(0,096)	3,161^c(0,091)	2,472 (0,132)	2,171 (0,156)	1,830 (0,191)	1,639 (0,215)
Minnesota	1,718 (0,205)	1,581 (0,223)	1,989 (0,174)	1,763 (0,199)	1,473 (0,239)	1,162 (0,294)
Mississippi	0,003 (0,955)	0,002 (0,967)	0,217 (0,647)	0,380 (0,544)	0,503 (0,487)	0,815 (0,377)
Missouri	2,163 (0,157)	1,432 (0,246)	1,401 (0,250)	1,331 (0,262)	0,976 (0,335)	0,622 (0,440)
Montana	0,288 (0,598)	0,092 (0,765)	0,630 (0,437)	0,335 (0,569)	0,507 (0,485)	0,640 (0,433)
Nebraska	0,108 (0,745)	0,167 (0,687)	0,161 (0,692)	0,105 (0,749)	0,088 (0,769)	0,062 (0,806)
Nevada	0,038 (0,847)	0,006 (0,941)	0,015 (0,904)	0,106 (0,749)	0,138 (0,715)	0,351 (0,560)
New Hamp.	1,736 (0,202)	1,504 (0,234)	1,480 (0,238)	1,380 (0,254)	0,862 (0,364)	0,568 (0,460)
New Jersey	0,001 (0,982)	0,022 (0,883)	0,019 (0,891)	0,010 (0,921)	0,075 (0,787)	0,036 (0,851)
New Mexico	0,504 (0,486)	0,811 (0,379)	1,378 (0,254)	1,382 (0,253)	1,127 (0,301)	1,158 (0,295)
New York	0,000 (0,995)	0,003 (0,958)	0,048 (0,829)	0,013 (0,911)	0,068 (0,797)	0,067 (0,799)
N. Carolina	6,261^b(0,021)	7,565^b(0,012)	7,099^b(0,015)	6,552^b(0,019)	4,515^b(0,046)	3,899^c(0,062)
N. Dakota	0,089 (0,769)	0,002 (0,967)	0,000 (0,984)	0,000 (0,995)	0,007 (0,933)	0,011 (0,917)
Ohio	0,713 (0,409)	1,102 (0,306)	1,043 (0,319)	0,914 (0,350)	1,274 (0,272)	0,866 (0,363)
Oklahoma	0,490 (0,492)	0,516 (0,481)	0,491 (0,492)	0,268 (0,610)	0,077 (0,784)	0,151 (0,702)
Oregon	0,165 (0,689)	0,490 (0,492)	1,532 (0,230)	1,191 (0,288)	1,759 (0,200)	2,991^c(0,099)
Pennsylvania	0,086 (0,773)	0,026 (0,873)	0,001 (0,979)	0,001 (0,971)	0,063 (0,805)	0,086 (0,772)
Rhode Island	0,236 (0,632)	0,199 (0,661)	0,359 (0,556)	0,119 (0,733)	0,018 (0,896)	0,030 (0,863)
S. Carolina	6,962^b(0,016)	6,989^b(0,016)	6,984^b(0,016)	6,205^b(0,022)	5,734^b(0,027)	4,331^b(0,050)
S. Dakota	2,343 (0,142)	2,105 (0,162)	1,817 (0,193)	1,745 (0,201)	1,630 (0,216)	1,442 (0,244)
Tennessee	4,568^b(0,045)	4,739^b(0,042)	3,777^c(0,067)	3,424^c(0,079)	4,104^c(0,056)	3,711^c(0,068)
Texas	3,259^c(0,086)	3,656^c(0,070)	2,619 (0,121)	2,340 (0,142)	2,041 (0,169)	1,445 (0,243)
Utah	0,647 (0,431)	0,727 (0,404)	0,398 (0,535)	0,041 (0,841)	0,064 (0,804)	0,132 (0,721)
Vermont	0,037 (0,849)	0,111 (0,743)	0,110 (0,743)	0,000 (0,991)	0,004 (0,953)	0,056 (0,816)
Virginia	1,003 (0,328)	1,351 (0,259)	2,033 (0,169)	2,079 (0,165)	1,136 (0,299)	0,961 (0,339)
Washington	0,010 (0,922)	0,059 (0,811)	0,303 (0,588)	0,603 (0,447)	0,177 (0,679)	0,156 (0,697)
West Virginia	0,265 (0,612)	0,471 (0,500)	0,531 (0,475)	0,447 (0,512)	0,894 (0,356)	0,710 (0,409)
Wisconsin	6,547^b(0,019)	7,472^b(0,013)	6,835^b(0,017)	4,424^b(0,048)	4,827^b(0,040)	4,255^c(0,052)
Wyoming	0,133 (0,719)	1,116 (0,303)	1,193 (0,288)	0,732 (0,402)	0,841 (0,370)	0,852 (0,367)

^aIndicates 1% statistical significance.^bIndicates 5% statistical significance.^cIndicates 10% statistical significance.

If Dumitrescu-Hurlin panel causality results are examined by states, it is seen that there is causality from CO₂ emissions to the SPEI January drought index in 6 states, from CO₂ emissions to SPEI February and March drought indexes in 8 different states, from CO₂ emissions to SPEI April, May, June, August, September, October, November, and December drought indexes in 9 different states, and from CO₂ emissions to SPEI July drought index in 10 states. To be stated for all months, there are causalities from CO₂ emissions to SPEI drought indexes in Alabama, Alaska, Arkansas, Colorado, Georgia, Illinois, Kentucky, Michigan, Oregon, South and North Carolina, South Dakota, Tennessee, Texas, and Wisconsin.

It is seen that the results obtained from the AMG and PMG analyses and the causality analyses are generally consistent with each other. In the AMG and PMG analyses, it was determined that CO₂ emissions, as a driver of climate change, contribute to the occurrence and severity of droughts by altering regional climate patterns. As a result of causality analyses, it was found that there are mutual causality relationships between CO₂ emissions and drought.

Impulse-response analysis is used to analyze the effect of a random shock occurring in one variable on other variables. At this stage of the study, an impulse-response analysis was applied to analyze the effect of unexpected increases in CO₂ emissions on droughts and the results are presented graphically in Figure 3.

Figure 3

Results of impulse-response analysis between CO2 emissions and droughts.

Drought monitoring and prediction using SPEI index and gene expression programming model in the west of Urmia Lake

Results of impulse-response analysis between CO₂ emissions and droughts.

According to the results obtained from the impulse-response analysis, DR1, DR2, DR11, and DR12 respond negatively until the 5th period in the face of an unexpected increase in CO₂ emissions and follow a slightly fluctuating course after the 5th period. DR3, DR4, DR5, DR6, DR7, DR8, DR9, and DR10 react negatively to an unexpected increase in CO₂ emissions until the 3rd period and follow a fluctuating course after the 3rd period. The results obtained from the impulse-response analysis showed that CO₂ emissions decreased the SPEI drought indices, that is, increased drought, in line with our expectations.

CONCLUSIONS

Listen

The drought problem, which is increasingly on the agenda due to climate change, is being examined by many researchers as it is a vital issue. The drought problem is one of the most important problems of humanity. If its further progress cannot be prevented, it will cause people to face hunger, thirst, and various health problems after a while. Therefore, the causes and consequences of drought should be investigated, and various studies should be carried out to prevent it. In this study, firstly, a literature review was conducted and studies on the relationship between CO₂ emissions, climate change, and drought were analyzed. It was observed that there are a limited number of studies examining the relationship between drought and CO₂. Then the mutual relations between CO₂ emissions and droughts were discussed by using the 1997–2020 SPEI drought index and CO₂ emissions data of 51 states of the US.

In the study, cross-section dependence, slope homogeneity, Westerlund cointegration, AMG and PMG analyses, Dumitrescu-Hurlin panel and state-based causality analyses, and impulse-response analysis were conducted as empirical applications. According to the findings obtained from analyses, it was determined that there are mutual interactions between CO₂ emissions and the SPEI drought index.

The results of the AMG analyses show that CO₂ emissions were found to be effective on the SPEI drought index in 17 of 51 states, and the states where CO₂ emissions had the longest effect on the SPEI drought index were Pennsylvania and Tennessee. When the coefficients are examined, it is understood that from these two states, CO₂ emissions increase wetting in Pennsylvania and increase drought in Tennessee. This situation is an expected result for Tennessee, which experienced the worst drought in history from 2007 to early 2009 (Goodrich et al. 2011). It is determined that CO₂ emissions led to drought in Alabama, Arkansas, Georgia, Louisiana, North and South Carolina, Oregon, and Tennessee in various months, especially in the last 6 months of the year. According to the results obtained from PMG analyses, it was determined that for the whole panel, CO₂ emissions generally increase wetting in the short run, while CO₂ emissions increase drought in the long run. According to the results of causality analysis, it was determined that CO₂ emissions cause drought. And according to the results obtained from the impulse-response analysis, CO₂ emissions decrease the SPEI drought indices, that is, increase drought. When the results obtained from all these analyses are evaluated together, it is concluded that our hypothesis that CO₂ emissions, by driving climate change, can increase the frequency and intensity of droughts through their impact on regional precipitation and temperature patterns is valid for the states of the US. In order to prevent drought, which is an important problem for humanity, it is necessary to reduce CO₂ emissions. Therefore, it would be appropriate to take measures to reduce CO₂ emissions in the US.

Previous studies have extensively documented the impact of climate change, driven largely by greenhouse gas emissions, on precipitation variability and drought patterns. CO₂ emissions are recognized for exacerbating global warming, which leads to changes in regional hydrological cycles, impacting both drought and wet periods. In this study, considering that CO₂ emissions have an increasing effect on air temperatures in the US, it has been determined that dry periods may occur in some regions due to a decrease in precipitation regimes, and that periods with high amounts of precipitation may occur in some regions. It has been determined that especially in the northeastern states of the US, the rainfall expectation is high, and the number of wet periods is quite high, whereas in the states located in the southeastern region, the number of dry periods with low rainfall values is expected to be quite high. The findings obtained coincide with the findings of previous studies in the literature.

In order to prevent drought, which is an important problem for humanity, it is necessary to reduce CO₂ emissions. Therefore, it would be appropriate to take measures to reduce CO₂ emissions in the US. Natural disasters due to climate change occur in the US, which is the most polluting country in the world, along with China, due to industrial production. In addition, due to the damage caused by global warming to natural life, the natural order is disrupted, and new types of deadly diseases appear. Therefore, the effects of climate change will impact not only future generations but also people living today. The US did not participate in the Kyoto Protocol, which is one of the steps taken to limit global warming and greenhouse gas emissions, with the thought that the welfare level in the country would decrease. This study investigates how CO₂ emissions, as a major driver of global warming, contribute to altered precipitation regimes and increased drought risk in specific regions of the US, thus analyzing the indirect effects of emissions on hydrological cycles. As a consequence, it would be appropriate for the US to take the necessary measures against climate change and to take measures to limit CO₂ emissions so that the welfare level of the country does not decrease.

This study aims to explore the relationship between CO₂ emissions and drought occurrence in the US, focusing on how CO₂-driven climate change may contribute to drought conditions over time. For this purpose, various statistical analyses were applied both for the entire US (on a panel basis) and for the states of the US. As a result of the analyses, the relationships between CO₂ emissions and drought were determined on a state and panel basis. It is thought that this study will make significant contributions to the literature, as both panel-based and state-level relationships are presented and the relationships between SPEI drought indices and CO₂ emissions are analyzed statistically. Similar studies can be done for other countries. Different analyses can be made using different drought indices. Analyses can be differentiated by using different analysis methods or different variables.

FUNDING

Listen

The author declares that no funds, grants, or other support were received during the preparation of this manuscript.

AUTHOR CONTRIBUTIONS

Listen

The entire work was prepared by a single author.

DECLARATION OF GENERATIVE AI AND AI-ASSISTED TECHNOLOGIES IN THE WRITING PROCESS

Listen

During the preparation of this study, the author did not use any service tools.

CONSENT FOR PUBLICATION

Listen

The final version of the manuscript was reviewed and approved by the author.

DATA AVAILABILITY STATEMENT

Listen

All relevant data are available from an online repository or repositories: For the SPEI data: https://spei.csic.es/map/maps.html For the CO2 data: https://www.eia.gov/environment/emissions/state/.

CONFLICT OF INTEREST

Listen

The authors declare there is no conflict.

REFERENCES

Abbasi

A.

,

Khalili

K.

,

Behmanesh

J.

&

Shirzad

A.

(

2019

)

,

Theoretical and Applied Climatology

,

138

,

553

–

567

.

https://doi.org/10.1007/s00704-019-02825-9

.

Abramowitz

M.

&

Stegun

I. A.

(

1965

)

Handbook of Mathematical Functions, with Formulas, Graphs, and Mathematical Tables

.

New York

:

Dover Publications

.

Examining the relationship between environmental factors and conflict in pastoralist areas of East Africa

Ayana

E. K.

,

Ceccato

P.

,

Fisher

J. R.

&

DeFries

R.

(

2016

)

,

Science of the Total Environment

,

557

,

601

–

611

.

Renewable and non-renewable electricity consumption–economic growth nexus: Evidence from OECD countries

PubMed

Aydin

M.

(

2019

)

,

Renewable Energy

,

136

,

599

–

606

.

https://doi.org/10.1016/j.renene.2019.01.008

.

A Lagrange multiplier test for cross-sectional dependence in a fixed effects panel data model

Baltagi

B. H.

,

Feng

Q.

&

Kao

C.

(

2012

)

,

Journal of Econometrics

,

170

,

164

–

177

.

https://doi.org/10.1016/j.jeconom.2012.04.004

.

Bates

B.

,

Kundzewicz

Z.

&

Wu

S.

(

2008

)

Climate Change and Water

.

Geneva

:

Intergovernmental Panel on Climate Change Secretariat

.

The impact of urbanization on energy intensity: Panel data evidence considering cross-sectional dependence and heterogeneity

Bilgili

F.

,

Koçak

E.

,

Bulut

U.

&

Kuloğlu

A.

(

2017

)

,

Energy

,

133

,

242

–

256

.

The Lagrange multiplier test and its applications to model specifications in econometrics

Breusch

T. S.

&

Pagan

A. R.

(

1980

)

,

Review of Economic Studies

,

47

,

239

–

253

.

https://doi.org/10.2307/2297111

.

Twenty-first century drought projections in the CMIP6 forcing scenarios

Cook

B. I.

,

Mankin

J. S.

,

Marvel

K.

,

Williams

A. P.

,

Smerdon

J. E.

&

Anchukaitis

K. J.

(

2020

)

,

Earth's Future

,

8

(

6

),

e2019EF001461

.

1–20

.

Identification of the trends associated with the SPI and SPEI indices across Ankara, Turkey

Crippa

M.

,

Guizzardi

D.

,

Banja

M.

,

Solazzo

E.

,

Muntean

M.

,

Schaaf

E.

,

Pagani

F.

,

Monforti-Ferrario

F.

,

Olivier

J.

,

Quadrelli

R.

,

Risquez Martin

A.

,

Taghavi-Moharamli

P.

,

Grassi

G.

,

Rossi

S.

,

Jacome Felix Oom

D.

,

Branco

A.

,

San-Miguel-Ayanz

J.

&

Vignati

E.

(

2022

)

CO₂ emissions of all world countries – 2022 Report. EUR 31182 EN, Publications Office of the European Union, Luxembourg, 2022. JRC130363. Available at: https://edgar.jrc.ec.europa.eu/report_2022.

Danandeh Mehr

A.

&

Vaheddoost

B.

(

2020

)

,

Theoretical and Applied Climatology

,

139

,

1531

–

1542

.

https://doi.org/10.1007/s00704-019-03071-9

.

Future CO₂ emissions and climate change from existing energy infrastructure

Davis

S. J.

,

Caldeira

K.

&

Matthews

H. D.

(

2010

)

,

Science

,

329

,

1330

–

1333

.

https://doi.org/10.1126/science.1188566

.

Testing for cross-sectional dependence in panel-data models

PubMed

De Hoyos

R. E.

&

Sarafidis

V.

(

2006

)

,

The Stata Journal

,

6

,

482

–

496

.

https://doi.org/10.1177/1536867X0600600403

.

Testing for Granger non-causality in heterogeneous panels

Demuth

S.

&

Stahl

K.

(

2001

)

ARIDE: Assessment of the Regional Impact of Droughts

.

Final Report (No. C056.026)

.

Freiburg, Germany

:

Institute of Hydrology, University of Freiburg

.

Dumitrescu

E. I.

&

Hurlin

C.

(

2012

)

,

Economic Modelling

,

29

,

1450

–

1460

.

https://doi.org/10.1016/j.econmod.2012.02.014

.

Drought risk assessment and mapping for the Murray–Darling Basin, Australia

Dunne

A.

&

Kuleshov

Y.

(

2023

)

,

Natural Hazards

,

115

(

1

),

839

–

863

.

Long-term variability in hydrological droughts and floods in sub-Saharan Africa: New perspectives from a 65-year daily streamflow dataset

Eberhardt

M.

&

Bond

S.

(

2009

)

Cross-section dependence in nonstationary panel models: A novel estimator [MPRA Paper]. Available at: https://mpra.ub.uni-muenchen.de/17692/.

Eberhardt

M.

&

Teal

F.

(

2010

)

Productivity Analysis in Global Manufacturing Production

.

Discussion Paper 515

.

Oxford, U

K:

Department of Economics, University of Oxford

.

Ecological and Environmental Informatics Research, Northern Arizona University

.

Available at: https://ecoinfo.nau.edu/index.php/news/2237/ (Accessed 18 November 2023)

.

Ekolu

J.

,

Dieppois

B.

,

Sidibe

M.

,

Eden

J. M.

,

Tramblay

Y.

,

Villarini

G.

,

Pena-Angulo

D.

,

Mahe

G.

,

Paturel

J. E.

,

Onyutha

C.

&

Van De Wiel

M.

(

2022

)

,

Journal of Hydrology

,

613

,

128359

.

Testing for Granger causality in heterogeneous mixed panels

Emirmahmutoglu

F.

&

Kose

N.

(

2011

)

,

Economic Modelling

,

28

,

870

–

876

.

https://doi.org/10.1016/j.econmod.2010.10.018

.

Gebrechorkos

S. H.

,

Pan

M.

,

Lin

P.

,

Anghileri

D.

,

Forsythe

N.

,

Pritchard

D. M.

,

Fowler

H. J.

,

Obuobie

E.

,

Darko

D.

&

Sheffield

J.

(

2022

)

Variability and changes in hydrological drought in the Volta Basin, West Africa

,

Journal of Hydrology: Regional Studies

,

42

,

101143

.

Summer drought in northern midlatitudes in a time-dependent CO₂ climate experiment

Gregory

J. M.

,

Mitchell

J. F. B.

&

Brady

A. J.

(

1997

)

,

Journal of Climate

,

10

(

4

),

662

–

686

.

The 2007 mid-south summer drought and heat wave in historical perspective

Goodrich

G. B.

,

Thompson

J. K.

,

Wingard

S. D.

&

Batson

K. J.

(

2011

)

,

Southeastern Geographer

,

51

,

411

–

421

.

The influence of climate change on CO₂ (carbon dioxide) emissions: An empirical estimation based on Chinese provincial panel data

Hao

Y.

,

Chen

H.

,

Wei

Y. M.

&

Li

Y. M.

(

2016

)

,

Journal of Cleaner Production

,

131

,

667

–

677

.

https://doi.org/10.1016/j.jclepro.2016.04.117

.

Droughts in East Africa: Causes, impacts and resilience

Haile

G. G.

,

Tang

Q.

,

Sun

S.

,

Huang

Z.

,

Zhang

X.

&

Liu

X.

(

2019

)

,

Earth-Science Reviews

,

193

,

146

–

161

.

California drought increases CO₂ footprint of energy

Hardin

E.

,

AghaKouchak

A.

,

Qomi

M. J. A.

,

Madani

K.

,

Tarroja

B.

,

Zhou

Y.

,

Yang

T.

&

Samuelsen

S.

(

2017

)

,

Sustainable Cities and Society

,

28

,

450

–

452

.

https://doi.org/10.1016/j.scs.2016.09.004

.

When early warning is not enough – Lessons learned from the 2011 Somalia Famine

Hillbruner

C.

&

Moloney

G.

(

2012

)

,

Global Food Security

,

1

(

1

),

20

–

28

.

Houghton

J. T.

,

Ding

Y.

,

Griggs

D. J.

,

Noguer

M.

,

Linden

P. J.

,

van der Dai

X.

,

Maskell

K.

&

Johnson

C. A.

(

2001

)

Climate Change 2001: The Scientific Basis: Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change

.

Cambridge, UK

:

Cambridge University Press

.

Conservation agriculture in Central Asia – What do we know and where do we go from here?

Kienzler

K. M.

,

Lamers

J. P. A.

,

McDonald

A.

,

Mirzabaev

A.

,

Ibragimov

N.

,

Egamberdiev

O.

,

Ruzibayev

E.

&

Akramkhanov

A.

(

2012

)

Field Crops Research

,

132

,

95

–

105

.

Small sample confidence intervals for impulse response functions

Killian

L.

(

1998

)

,

Review of Economics and Statistics

,

80

,

218

–

230

.

https://doi.org/10.1162/003465398557465

.

The role of climate variability in Australian drought

King

A. D.

,

Pitman

A. J.

,

Henley

B. J.

,

Ukkola

A. M.

&

Brown

J. R.

(

2020

)

,

Nature Climate Change

,

10

(

3

),

177

–

179

.

Testing for Granger causality in panel data

Li

Z.

&

Yao

J.

(

2021

)

Extension of the Lagrange multiplier test for error cross-section independence to large panels with non normal errors. arXiv: 2103.06075. Available at: http://arxiv.org/abs/2103.06075.

Lopez

L.

&

Weber

S.

(

2017

)

,

The Stata Journal

,

17

,

972

–

984

.

https://doi.org/10.1177/1536867X1801700412

.

Asymptotic distributions of impulse response functions and forecast error variance decompositions of vector autoregressive models

Lütkepohl

H.

(

1990

)

,

The Review of Economics and Statistics

,

72

(

1

),

116

–

125

.

https:// doi.org/10.2307/2109746

.

A review of droughts on the African continent: A geospatial and long-term perspective

Masih

I.

,

Maskey

S.

,

Mussá

F. E. F.

&

Trambauer

P.

(

2014

)

,

Hydrology and Earth System Sciences

,

18

(

9

),

3635

–

3649

.

The proportionality of global warming to cumulative carbon emissions

Matthews

H. D.

,

Gillett

N. P.

,

Stott

P. A.

&

Zickfeld

K.

(

2009

)

,

Nature

,

459

(

7248

),

829

–

833

.

https://doi.org/10.1038/nature08047

.

Environmental influences on pastoral conflict in the Horn of Africa

PubMed

Meier

P.

,

Bond

D.

&

Bond

J.

(

2007

)

,

Political Geography

,

26

(

6

),

716

–

735

.

A review of drought concepts

Mishra

A. K.

&

Singh

V. P.

(

2010

)

,

Journal of Hydrology

,

391

(

1–2

),

202

–

216

.

A detailed look at the recent drought situation in the Greater Horn of Africa

Nicholson

S. E.

(

2014

)

,

Journal of Arid Environments

,

103

,

71

–

79

.

CO₂ emissions, energy consumption and economic growth in Turkey

Ozturk

I.

&

Acaravci

A.

(

2010

)

,

Renewable and Sustainable Energy Reviews

,

14

,

3220

–

3225

.

https://doi.org/10.1016/j.rser.2010.07.005

.

General diagnostic tests for cross section dependence in panels

Pesaran

M. H.

(

2004

)

,

SSRN Electronic Journal

,

2004

,

1

–

42

.

https://doi.org/10.2139/ssrn.572504

.

A simple panel unit root test in the presence of cross-section dependence

Pesaran

M. H.

(

2007

)

,

Journal of Applied Econometrics

,

22

,

265

–

312

.

Generalized impulse response analysis in linear multivariate models

Pesaran

H. H.

&

Shin

Y.

(

1998

)

,

Economics Letters

,

58

(

1

),

17

–

29

.

https://doi.org/10.1016/S0165-1765(97)00214-0

.

Testing slope homogeneity in large panels

Pesaran

M. H.

&

Yamagata

T.

(

2008

)

,

Journal of Econometrics

,

142

,

50

–

93

.

Pooled mean group estimation of dynamic heterogeneous panels

Pesaran

M. H.

,

Shin

Y.

&

Smith

R. P.

(

1999

)

,

Journal of the American Statistical Association

,

94

,

621

–

634

.

https://doi.org/10.2307/2670182

.

A bias-adjusted LM test of error cross-section independence

Pesaran

M. H.

,

Ullah

A.

&

Yamagata

T.

(

2008

)

,

The Econometrics Journal

,

11

,

105

–

127

.

Impact of increasing CO₂ on monthly-to-annual precipitation extremes: Analysis of the CMIP2 experiments

Räisänen

J.

(

2005

)

,

Climate Dynamics

,

24

,

309

–

323

.

Riebsame

W. E.

(

2019

)

Drought and Natural Resources Management in the United States: Impacts and Implications of the 1987–89 Drought

.

New York:

Routledge

.

Rodziewicz

D.

,

Dice

J.

&

Cowley

C. A.

(

2023

) '

Impacts of drought in the US Cattle Sector

',

Agricultural and Applied Economics Association Annual Meeting

,

Washington, DC

,

July 23–25, 2023

.

Sen

Z.

(

2022

)

İklim değişikliği ve Türkiye

,

Çevre Şehir ve İklim Dergisi

,

1

(

1

),

1

–

19

.

SPEI Global Drought Monitor

(

2023

)

Available at: https://spei.csic.es (Accessed 16 March 2023)

.

Spinoni

J.

,

Naumann

G.

,

Vogt

J. V.

&

Barbosa

P.

(

2015

)

The biggest drought events in Europe from 1950 to 2012

,

Journal of Hydrology: Regional Studies

,

3

,

509

–

524

.

Characterizing changes in drought risk for the United States from climate change

State Carbon Dioxide Emissions Data – U.S. Energy Information Administration (EIA)

(

2022

)

Available at: https://www.eia.gov/environment/emissions/state/ (Accessed 3 November 2022)

.

Strzepek

K.

,

Yohe

G.

,

Neumann

J.

&

Boehlert

B.

(

2010

)

,

Environmental Research Letters

,

5

,

1

–

9

.

https://doi.org/10.1088/1748-9326/5/4/044012

.

Su

X.

,

Huang

G.

,

Wang

L.

&

Wang

T.

(

2024

)

Global drought changes and attribution under carbon neutrality scenario

,

Climate Dynamics

,

62

,

7851

–

7868

.

Efficient inference in a random coefficient regression model

Swamy

P. A.

(

1970

)

,

Econometrica: Journal of the Econometric Society

,

38

(

2

),

311

–

323

.

Drought monitoring and analysis: Standardised Precipitation Evapotranspiration Index (SPEI) and Standardised Precipitation Index (SPI)

Tirivarombo

S.

,

Osupile

D.

&

Eliasson

P.

(

2018

)

,

Physics and Chemistry of the Earth, Parts A/B/C

,

106

,

1

–

10

.

https://doi.org/10.1016/j.pce.2018.07.001

.

Seasonal climate forecasting

Troccoli

A.

(

2010

)

,

Meteorological Applications

,

17

,

251

–

268

.

https://doi.org/10.1002/met.184

.

Ullah

A.

(

2004

)

Finite Sample Econometrics

.

New York

:

Oxford University Press

.

Amplifying feedback loop between drought, soil desiccation cracking, and greenhouse gas emissions

U.S. Global Change Research Program

(

2023

)

Fifth National Climate Assessment (NCA5)

.

Available at: https://nca2023.globalchange.gov/

Vahedifard

F.

,

Goodman

C. C.

,

Paul

V.

&

AghaKouchak

A.

(

2024

)

,

Environmental Research Letters

,

19

(

3

),

031005

.

1–7

.

A multiscalar drought index sensitive to global warming: the Standardized Precipitation Evapotranspiration Index

Vicente-Serrano

S. M.

,

Beguería

S.

&

López-Moreno

J. I.

(

2010

)

,

Journal of Climate

,

23

,

1696

–

1718

.

Testing for error correction in panel data

Westerlund

J.

(

2007

)

,

Oxford Bulletin of Economics and Statistics

,

69

,

709

–

748

.

Implications of climate change due to the enhanced greenhouse effect on floods and droughts in Australia

Whetton

P. H.

,

Fowler

A. M.

,

Haylock

M. R.

&

Pittock

A. B.

(

1993

)

,

Climatic Change

,

25

(

3–4

),

289

–

317

.

Wilhite

D. A.

&

Hayes

M. J.

(

1998

)

Drought planning in the United States: Status and future directions

. In: Bruins, H. J., Lithwick, H. (eds)

The Arid Frontier. The GeoJournal Library, vol 41

, pp.

33

–

54

.

Dordrecht: Springer. https://doi.org/10.1007/978-94-011-4888-7_2

.

Extreme drought event of 2009/2010 over southwestern China

World Development Indicators | DataBank

(

2022

)

Available at: https://databank.worldbank.org/source/world-development-indicators (Accessed 11 November 2022)

.

Yang

J.

,

Gong

D.

,

Wang

W.

,

Hu

M.

&

Mao

R.

(

2012

)

,

Meteorology and Atmospheric Physics

,

115

,

173

–

184

.

Spatial variation and trends in PDSI and SPI indices and their relation to streamflow in 10 large regions of China

Zhai

J.

,

Su

B.

,

Krysanova

V.

,

Vetter

T.

,

Gao

C.

&

Jiang

T.

(

2010

)

,

Journal of Climate

,

23

(

3

),

649

–

663

.

Drought over East Asia: A review

Zhang

L.

&

Zhou

T.

(

2015

)

,

Journal of Climate

,

28

(

8

),

3375

–

3399

.