Influence of Climate Conditions on Tax Revenues

In this paper, we investigate the effects of climate conditions on collected tax revenues, based on a panel-model approach. The dataset includes 123 countries and covers the period 1996-2010. The main results demonstrate that climate has a significant impact on tax revenues, the assumed function being nonlinear, with a cubic ⋂ and incomplete ⋃-shape. We also find that governments situated in temperate climate zones, with low to moderate temperatures (the ‘optimal temperature’ in our investigation), can ensure a good collection of tax revenues. The study suggests that a significant increase in collected tax revenues, without a major negative reaction by taxpayers, can be easily obtained by the public authority if situated in a temperate climate zone, that is, one with moderate temperatures.


Introduction
state that the climate also has a great impact on human behavior; even so, this type of approach was preferred during the later 19 th and much of the early 20 th century but has been rejected by geographers since the 1960s and 1970s.
The mentioned authors consider that the climate influence on humans is related to 'societal (e.g., civilization, culture and migration), psychological (e.g., aggression, cognition and mental illness), physiological (e.g., health, allergies, diet and nutrition), economic (e.g., energy production, tourism and agriculture) and ecological conditions (e.g., fauna and flora)' .
The connection 'tax revenues-climate' has two principal transmission channels: a direct one and an indirect one. The direct channel implies that the level of collected tax revenues depends on citizens' behavior, which differs from one climate zone to another. The climate influence on human behavior is sustained by many recent studies (e.g., Anderson, 2001;Berkowitz, 1993or Boyanowsky, 1999. The authors find that heat is associated with high degrees of aggressiveness and other complex behavioral changes. In an extended view, McDougall (2005) argues that persons situated in the cool climate zones are characterized by honesty, pragmatism, and also great independence of character, individual initiative, and tenacity of will. As tax compliance is a dimension of human behavior, the climate can play an important role in the level of collected tax revenues through the intensity of tax compliance. The indirect channel consists of the impact of climate conditions on tax revenues through a set of socio-economic determinants. In this case, GDP growth, and thus tax base growth, under climate conditions determines a positive impact on tax revenues. At the same time, the structure of the economy and economic openness can also play an important role for extension of the tax base under a considered climate environment.
The aim of this paper is to analyze the effects of climate conditions on collected tax revenues, based on a panelmodel approach. The data set includes 123 countries and covers the period 1996-2010. The main results show that the climate has a significant impact on tax revenues, the assumed function being nonlinear, with a cubic ⋂ and ⋃-shape. We also find that governments situated in temperate climate zones, with moderate temperatures (the 'optimal temperature' in our investigation), can ensure a good collection of tax revenues.
In the contemporary period, Sachs and Warner (1997) investigated the growth in 23 African countries, for the period 1965-1990, based on a cross-country regression model. The results demonstrate that poor economic policies determine slow economic growth, without any openness to international markets. Furthermore, the authors add new determinants of slow growth, i.e., lack of access to the sea and tropical climate conditions. One year later, Bloom and Sachs (1998) explored the influence of climatic conditions on some economic variables and emphasized that poorer zones are correlated with tropical climates, as a result of tropical ecology effects on human health and agricultural productivity. The temperate climate is associated with the rich zones. Dell, et al. (2012) change the type of climate with temperature in their investigations. The authors study the effect of temperature's fluctuations on economic activity, with a sample of 125 countries. The main conclusion shows that higher temperatures substantially reduce economic growth in poor countries. In parallel, they associate higher temperatures with reduced agricultural output, industrial output and political stability. More recently, Bansal and Ochoa (2012) investigated the connection between temperature, aggregate risk and expected returns, utilizing a sample of 38 countries in the period 1960-2008. The researchers note that temperature represents an aggregate risk factor of economic growth. In this respect, countries closer to the Equator reveal a positive temperature risk premium that decreases as one moves farther away from the Equator.

Influence of climate conditions on tax revenues
New evidence concerning the 'climate -economy' nexus was revealed by Gallup and Sachs (1999). They focus on food production and find a strong correlation between food production and climate zone type.
Because of certain inputs, such as capital, labor and fertilizers, harvests are much lower in tropical zones, compared with temperate zones, which experience greater harvests. Mellinger, et al. (1999) are more analytical, examining the connection between climate (eco zones), water navigability and economic development (GDP per capita) in each of 152 countries, with a population of 1 million or more, in 1995. They emphasize that GDP per capita and density of economic activity (in GDP per km 2 ) are high in temperate zones and in areas accessible to the sea (within 100 km of the ocean or a seanavigable waterway).
Hall and Jones (1999) choose a new topic: the impact of climate on revenue inequality. The authors focus on the issues of per capita income discrepancy across nations and illustrate a strong correlation between geography (measured as the distance from the equator) and per capita output by country. The location affects economic success because the position of human settlements can influence institutions. On the same topic, Gallup, et al. (1999) analyze the effects of location and climate on income levels and income growth. Novel is that a set of control variables is used, such as transport costs, disease burdens and agricultural productivity, among other channels. The main conclusion stresses that geography also affects economic policy choices.
Other authors link temperature, as an independent variable, with some national characteristics, such as institutional quality (e.g., Acemoglu, et al., 2002;Rodrik, et al., 2004). The results of Rodrik, et al. (2004) reveal that the institutions, geography and trade are the most important determinants of income levels around the world. Moreover, geography is considered to have weak direct effects on income levels. Sylwester (2004) juxtaposes history with geography as the important stimulus pair of institutions and income. Utilizing a cross-section of countries approach, he suggests that only being landlocked has a strong influence on revenue inequality.
In the present paper, we extend the literature in the field by focusing on the climate implications in the economy and find new evidence regarding the determinants of collected tax revenues. As there are cyclical climate changes, this contribution could help governments to adjust their tax policy taking into account this climate periodicity.
The rest of the paper is structured as follows: Section 2 highlights the methodology and data. Section 3 contains the results. Section 4 presents our conclusions.

Methodology and data
The influence of climate on tax revenues is explored based on a large sample, determined by 123 crosssections (123 countries), from 1996 to 2010 (Table   1, in the Appendix), using a panel-model approach. Two variables are considered for investigating this nexus: the tax revenues per capita, as a dependent variable, and the climate, as an independent interest variable. The dependent variable is represented by tax revenues per capita (τ) and reveals the level of tax revenues collected by general government per person in U.S. dollars.
The interest determinant is climate (δ). Climate zones are characterized by four coordinates: annual mean daily temperature, total annual precipitation, total annual potential evapo-transpiration (PET), and elevation. As the temperature is a very good proxy for climate, this coordinate is selected to capture weather conditions. Temperature measures annual mean daily temperature degrees, on the Fahrenheit scale, and quantitatively denotes the common notions of hot and cold.
The main hypothesis of this analysis is that climate conditions determine the level of government tax receipts. The function has this shape (the variables are treated as elastic): where τ equals the amount of tax revenues per capita in U.S. dollars, and δ equals the climate variable (the temperature). The scatter diagram of this function, presented in Figure 1, is performed using the nonparametric Nearest Neighbor Fit method (degree = 3; span = 0.9).
The scatter plot with Nearest Neighbor Fit method suggests that the association between the two variables is not linear, having a cubic ⋂ and incomplete ⋃-shape.
The Ramsey's Reset Test, assuming cubic and square functions, is performed to reinforce the nonlinear relationship between the ln tax revenues per capita and ln temperature. Additional tests are employed to verify the existence of polynomial degree greater than 3.
The robustness of these simple stylized facts is explored using several econometric techniques. The extended OLS naïve panel-model is as follows: where α is the intercept, β 0,1,2 are the slopes of the interest variables, i is the country, t is the time and remainder, and i t ε is the error term, which varies over both country, and time.
To isolate the effect of interest determinants, we enter three categories of control variables: one inspired by the appropriate tax literature, one derived from macroeconomic policy, and another one represented by robustness variables. Based on this adjustment, the extended linear model becomes: where β k is the coefficient of control independent variable k by n type, and X is the control independent variables.
The first group of control variables includes deter-    Table 3  The next section illustrates the variance inflation factor (VIF) test results for multicollinearity detection ( Table 4 in the Appendix) and the estimation outputs of considered function, following several econometric scenarios (models 1-7), as Table 5, in the Appendix, illustrates.

Results
The VIF tests for multicollinearity detection (Table 4, in the Appendix) demonstrate that, in all cases, the test values are less than 4. Hence, there is no multicollinearity issue between considered determinants (O'Brien, 2007).
The first outputs reveal that the suggested nonlinear relationship between ln tax revenues per capita and ln temperature is reinforced by Ramsey's Reset Tests, in the case of naive OLS model 1 (Table 5, in the Appendix). Ramsey's Reset Tests, assuming squares and cubes, confirm the scatter diagram: there is a nonlinear relationship between ln tax revenues per capita and ln temperature, leaving a cubic ⋂ and incomplete ⋃-shape, as model 3 (Table 5,

Influence of climate conditions on tax revenues
The tax cubic function with respect to δ has an oscillating trend, with two critical points: one maximum (τ max. ) and the other minimum (τ min. ). τ 1a and τ 1b are the roots of the first derivative cubic function, while τ 2 is the root of the second derivative of the same function.
Considering the definition interval of the function [27.86, 86.85], the tax revenues cubic function with respect to δ increases to τ 1a , decreases between two critical points (τ 1a , τ 1b ), and increases from τ 1b , but only in a very short interval. There is an inflection point τ inf. in which the accelerated decreasing trend slows.
In conclusion, the empirical findings in the case of 123 investigated countries for the period 1996-2010 reveal that the relationship between tax revenues per capita and temperature is cubic, with inverted-⋃ and incomplete ⋃ shapes. The results reveal that low tax revenues are associated with very low temperatures (cold climates) and that tax revenues tend to increase as temperatures increase, until reaching a maximum (i.e., 43.4 degrees Fahrenheit / 6.33 degrees Celcius, representing the root of the 1 st derivative of cubic function, given the coefficients of the interest variables in model 7). After that, even if the temperature increases, tax revenues decrease and reach a low level. In this point, the tax revenue is minimal under a quasi-high temperature (i.e., 72.3 degrees Fahrenheit / 22.38 degrees Celsius, denoting the 2 nd root of the first derivative of the considered cubic function). Regarding the next interval, taxes tend to exhibit an ascending incomplete trend under very high temperature (i.e., more than 72.3 degrees Fahrenheit /22.38 degrees Celsius). One may argue that this incomplete tendency is because a very high temperature represents the upper limit of human existence.

Conclusions
Tax revenues represent the main financial resources of governments. These revenues have several determinants (e.g., economic, socio-demographic and political types).
As the economic literature does not take into account geographical conditions concerning tax revenues, our exploration demonstrates this new evidence, using the temperature as the main climate determinant.
The main conclusion states that climate has a significant impact on tax revenues. We also find that governments situated in temperate climate zones, with low to moderate temperatures (the 'optimal temperature' in our investigation), can ensure a good collection of tax revenues. To be able to determine the climate type, we follow Peel, Finlayson and McMahon (2007)'s main contribution, which states that the temperate climate zone is characterized by a temperature with an average level of 0 to 10 degrees Celsius, during the coldest month of the year.
The study suggests that a significant increase of collected tax revenues, without a major negative reaction of taxpayers, can be easily obtained by a public authority situated in temperate climate zones, with low to moderate temperatures. Overall, these climate zones represent 'the best environment for tax revenues' , accelerating tax revenue collection. In