Anyway, the two equations, or the remaining part of Chapter I

My editorial

And so I continue my novel in short episodes, i.e. I am blogging the on-going progress in the writing of my book about renewable technologies and technological change. Today, I am updating my blog with the remaining part of the first Chapter, which I started yesterday. Just for those who try to keep up, a little reminder about notations that you are going to encounter in what follows below: N stands for population, E represents the non-edible energy that we consume, and F is the intake of food. For the moment, I do not have enough theoretical space in my model to represent other vital things, like dreams, pot, beauty, friendship etc.

Anyway, the two equations, namely ‘N = A*Eµ*F1-µ’ and ‘N = A*(E/N)µ*(F/N)1-µ’ can both be seen as mathematical expressions of two hypotheses, which seems perfectly congruent at the first sight, and yet they can be divergent. Firstly, each of these equations can be translated into the claim that the size of human population in a given place at a given time depends on the availability of food and non-edible energy in said place and time. In a next step, one is tempted to claim that incremental change in population depends on the incremental change in the availability of food and non-edible energies. Whilst the logical link between the two hypotheses seems rock-solid, the mathematical one is not as obvious, and this is what Charles Cobb and Paul Douglas discovered as they presented their original research in 1928 (Cobb, Douglas 1928[1]). Their method can be summarised as follows. We have three temporal series of three variables: the output utility on the left side of the equation, and the two input factors on the right side. In the original production function by Cobb and Douglas had aggregate output of the economy (Gross Domestic Product) on the output side, whilst input was made of investment in productive assets and the amount of labour supplied. We return, now, to the most general equation (1), namely U = A*F1µ*F21-µ, and we focus on the ‘F1µ*F21-µ’ part, so on the strictly spoken impact of input factors. The temporal series of output U can be expressed as a linear trend with a general slope, just as the modelled series of values obtained through ‘F1µ*F21-µ’. The empirical observation that any reader can make on their own is that the scale factor A can be narrowed down to that value slightly above 1 only if the slope of the ‘F1µ*F21-µ’ on the right side is significantly smaller than the slope of U. This is a peculiar property of that function: the modelled trend of the compound value ‘F1µ*F21-µ’ is always above the trend of U at the beginning of the period studied, and visibly below U by the end of the same period. The factor of scale ‘A’ is an averaged proportion between reality and the modelled value. It corresponds to a sequence of quotients, which starts with local A noticeably below 1, then closing by 1 at the central part of the period considered, to rise visibly above 1 by the end of this period. This is what made Charles Cobb and Paul Douglas claim that at the beginning of the historical period they studied the real output of the US economy was below its potential and by the end of their window of observation it became overshot. The same property of this function made it a tool for defining general equilibriums rather than local ones. As regards my research on renewable energies, that peculiar property of the compound input of food and energy calculated with ‘Eµ*F1-µ’ or with ‘(E/N)µ*(F/N)1-µ’ means that I can assess, over a definite window in time, whether available food and energy stay in general equilibrium with population. They do so, if my general factor of scale ‘A’, averaged over that window in time, stays very slightly over 1, with relatively low a variance. Relatively low, for a parameter equal more or less to one, means a variance, in A, staying around 0,1 or lower. If these mathematical conditions are fulfilled, I can claim that yes, over this definite window in time, population depends on the available food and energy. Still, as my parameter A has been averaged between trends of different slopes, I cannot directly infer that at any given incremental point in time, like from t0 to t1, my N(t1) – N(t0) = A*{[E(t1)µ*F(t1)1-µ] – [E(t0)µ*F(t0)1-µ]}. If we take that incremental point of view, the local A will be always different than the general one.

Bearing those theoretical limitations in mind, the author undertook testing the above equations on empirical data, in a compound dataset, made of Penn Tables 9.0 (Feenstra et al. 2015[2]), enriched with data published by the World Bank (regarding the consumption of energy and its structure regarding ‘renewable <> non–renewable’), as well as with data published by FAO with respect to the overall nutritive intake in particular countries. Data regarding energy, and that pertaining to the intake of food, is limited, in both cases, to the period 1990 – 2014, and the initial, temporal extension of Penn Tables 9.0 (from 1950 to 2014) has been truncated accordingly. For the same reasons, i.e. the availability of empirical data, the original, geographical scope of the sample has been reduced from 188 countries to just 116. Each country has been treated as a local equilibrium, as the initial intuition of the whole research was to find out the role of renewable energies for local populations, as well as local idiosyncrasies regarding that role. Preliminary tests aimed at finding workable combinations of empirical variables. This is another specificity of the Cobb – Douglas production function: in its original spirit, it is supposed to work with absolute quantities observable in real life. These real-life quantities are supposed to fit into the equation, without being transformed into logarithms, or into standardized values. Once again, this is a consequence of the mathematical path chosen, combined with the hypotheses possible to test with that mathematical tool: we are looking for a general equilibrium between aggregates. Of course, an equilibrium between logarithms can be searched for just as well, similarly to an equilibrium between standardized positions, but these are distinct equilibriums.

After preliminary tests, equation ‘N = A*Eµ*F1-µ’, thus operating with absolute amounts of food and energy, proved not being workable at all. The resulting scale factors were far below 1, i.e. the modelled compound inputs of food and energy produced modelled populations much overshot above the actual ones. On the other hand, the mutated equation ‘N = A*(E/N)µ*(F/N)1-µ’ proved operational. The empirical variables able to yield plausibly robust scale factors A were: final use of energy per capita, in tons of oil equivalent (factor E/N), and alimentary intake of energy per capita, measured annually in mega-calories (thousands of kcal), and averaged over the period studied. Thus, the empirical mutation of produced reasonably robust results was the one, where a relatively volatile (i.e. changing every year) consumption of energy is accompanied by a long-term, de facto constant over time, alimentary status of the given national population. Thus, robust results could be obtained with an implicit assumption that alimentary conditions in each population studied change much more slowly than the technological context, which, in turn, determines the consumption of energy per capita. On the left side of the equation, those two explanatory variables matched with population measured in millions. Wrapping up the results of those preliminary tests, the theoretical tool used for this research had been narrowed down to an empirical situation, where, over the period 1990 – 2014, each million of people in a given country in a given year was being tested for sustainability, regarding the currently available quantity of tons of oil equivalent per capita per year, in non-edible energies, as well as regarding the long-term, annual amount of mega calories per capita, in alimentary intake.

The author is well aware that all this theoretical path-clearing could have been truly boring for the reader, but it seemed necessary, as this is the point, when real surprises started emerging. I was ambitious and impatient in my research, and thus I immediately jumped to testing equation N = A*(E/N)µ*(F/N)1-µ’ with just the renewable energies in the game, after having eliminated all the non-renewable part of final consumption in energy. The initial expectation was to find some plausible local equilibriums, with the scale factor A close to 1 and displaying sufficiently low a variance, in just some local populations. Denmark, Britain, Germany – these were the places where I expected to find those equilibriums, Stable demographics, well-developed energy base, no official food deficit: this was the type of social environment, which I expected to produce that theoretical equilibrium, and yet, I expected to find a lot of variance in the local factors A of scale. Denmark seemed to behave according to expectations: it yielded an empirical equation N = (Renewable energy per capita)0,68*(Alimentary intake per capita)1 0,68 = 0,32. The scale factor A hit a surprising robustness: its average value over 1990 – 2014 was 1,008202138, with a variance var (A) = 0,059873591. I quickly tested its Scandinavian neighbours: Norway, Sweden, and Finland. Finland yielded higher a logarithm in renewable energy per capita, namely µ = 0,85, but the scale factor A was similarly robust, making 1,065855419 on average and displaying a variance equal to 0,021967408. With Norway, results started puzzling me: µ = 0,95, average A = 1,019025526 with a variance 0,002937442. Those results would roughly mean that whilst in Denmark the availability of renewable energies has a predominant role in producing a viable general equilibrium in population, in Norway it has a quasi-monopole in shaping the same equilibrium. Cultural clichés started working at this moment, in my mind. Norway? That cold country with low density of population, where people, over centuries, just had to eat a lot in order to survive winters, and the population of this country is almost exclusively in equilibrium with available renewable energies? Sweden marked some kind of a return to the expected state of nature: µ = 0,77, average A = 1,012941105 with a variance of 0,003898173. Once again, surprisingly robust, but fitting into some kind of predicted state.

What I could already see at this point was that my model produced robust results, but they were not quite what I expected. If one takes a look at the map of the world, Scandinavia is relatively small a region, with quite similar, natural conditions for human settlement across all the four countries. Similar climate, similar geology, similar access to wind power and water power, similar social structures as well. Still, my model yielded surprisingly marked, local idiosyncrasies across just this small region, and all those local idiosyncrasies were mathematically solid, regarding the variance observable in their scale factors A. This was just the beginning of my puzzlement. I moved South in my testing, to countries like Germany, France and Britain. Germany: µ = 0,31, average A = 1,008843147 with a variance of 0,0363637. One second, µ = 0,31? But just next door North, in Denmark, µ = 0,63, doesn’t it? How is it possible? France yielded a robust equilibrium, with average A = 1,021262046 and its variance at 0,002151713, with µ = 0,38. Britain: µ = 0,3, whilst average A = 1,028817158 and variance in A making 0,017810219.  In science, you are generally expected to discover things, but when you discover too much, it causes a sense of discomfort. I had that ‘No, no way, there must be some mistake’ approach to the results I have just presented. The degree of disparity in those nationally observed functions of general equilibrium between population, food, and energy, strongly suggested the presence of some purely arithmetical disturbance. Of course, there was that little voice in the back of my head, saying that absolute aggregates (i.e. not the ratios of intensity per capita) did not yield any acceptable equilibrium, and, consequently, there could be something real about the results I obtained, but I had a lot of doubts.

I thought, for a day or two, that the statistics supplied by the Word Bank, regarding the share of renewable energies in the overall final consumption of energy might be somehow inaccurate. It could be something about the mutual compatibility of data collected from national statistical offices. Fortunately, methods of quantitative analysis of economic phenomena supply a reliable method of checking the robustness of both the model, and the empirical data I am testing it with. You supplant one empirical variable with another one, possibly similar in its logical meaning, and you retest. This is what I did. I assumed that the gross, final consumption of energy, in tons of oil equivalent per capita, might be more reliable than the estimated shares of renewable sources in that total. Thus, I tested the same equations, for the same set of countries, this time with the total consumption of energy per capita. It is worth quoting the results of that second test regarding the same countries. Denmark: average scale factor A = 1,007673381 with an observable variance of 0,006893499, and all that in an equation where µ = 0,93. At this point, I felt, once again, as if I were discovering too much at once. Denmark yielded virtually the same scale factor A, and the same variance in A, with two different metrics of energy consumed per capita (total and just the renewable one), with two different values in the logarithm µ. Two different equilibriums with two different bases, each as robust as the other. Logically, it meant the existence of a clearly cut substitution between renewable energies and the non-renewable ones. Why? I will try to explain it with a metaphor. If I manage to stabilize a car, when changing its tyres, with two hydraulic lifters, and then I take away one of the lifters and the car remains stable, it means that the remaining lifter can do the work of the two. This one tool is the substitute of two tools, at a rate of 2 to 1. In this case, I had the population of Denmark stabilized both on the overall consumption of energy per capita (two lifters), and on just the consumption of renewable energies (one lifter). Total consumption of energy stabilizes population at µ = 0,93 and renewable energies do the same at µ = 0,68. Logically, renewable energies are substitutes to non-renewables with a rate of substitution equal to 0,93/0,68 = 1,367647059. Each ton of oil equivalent in renewable energies consumed per capita, in Denmark, can do the job of some 1,37 tons of non-renewable energies.

Finland was another source of puzzlement: A = 0,788769669, variance of A equal to 0,002606412, and µ = 0,99. Ascribing to the logarithm µ the highest possible value at the second decimal point, i.e. µ = 0,99, I could not get a model population lower than the real one. The model yielded some kind of demographic aggregate much higher than the real population, and the most interesting thing was that this model population seemed correlated with the real one. I could know it by the very low variance in the scale factor A. It meant that Finland, as an environment for human settlement, can perfectly sustain its present headcount with just renewable energies, and if the non-renewables are being dropped into the model, the same territory has a significant, unexploited potential for demographic growth. The rate of substitution between renewable energies and the non-renewable ones, this time, seemed to be 0,99/0,85 = 1,164705882. Norway yielded similar results, with the total consumption of energy per capita on the right side of the equation: A = 0,760631741, variance in A equal to 0,001570101, µ = 0,99, substitution rate 1,042105263. Sweden turned out to be similar to Denmark: A = 1,018026405 with a variance of 0,004626486, µ = 0,91, substitution rate 1,181818182. The four Scandinavian countries seem to form an environment, where energy plays a decisive role in stabilizing the local populations, and renewable energies seem to be able to do the job perfectly. The retesting of Germany, France, and Britain brought interesting results, too. Germany: A = 1,009335161 with a variance of 0,000335601, at µ = 0,48, with a substitution rate of renewables to non-renewables equal to 1,548387097. France: A = 1,019371541, variance of A at 0,001953865, µ = 0,53, substitution at 1,394736842. Finally, Britain: A = 1,028560563 with a variance of 0,006711585, µ = 0,52, substitution rate 1,733333333. Some kind of pattern seems to emerge: the greater the relative weight of energy in producing general equilibrium in population, the greater the substitution rate between renewable energies and the non-renewable ones.

At this point I was pretty certain that I am using a robust model. So many local equilibriums, produced with different empirical variables, was not the result of a mistake. Table 1, in the Appendix to Chapter I, gives the results of testing equation (3), with the above mentioned empirical variables, in 116 countries. The first numerical column of the table gives the arithmetical average of the scale factor ‘A’, calculated over the period studied, i.e. 1990 – 2014. The second column provides the variance of ‘A’ over the same period of time (thus the variance between the annual values of A), and the third specifies the value in the parameter ‘µ’ – or the logarithm ascribed to energy use per capita – at which the given values in A have been obtained. In other words, the mean A, and the variance of A specify how close to equilibrium assumed in equation (3) has it been possible to come in the case of a given country, and the value of µ is the one that produces that neighbourhood of equilibrium. The results from Table 1 seem to confirm that equation (3), with these precise empirical variables, is robust in the great majority of cases.

Most countries studied satisfying the conditions stated earlier: variances in the scale factor ‘A’ are really low, and the average value of ‘A’ is possible to bring just above 1. Still, exceptions abound regarding the theoretical assumption of energy use being the dominant factor that shapes the size of the population. In many cases, the value of the exponent µ that allows a neighbourhood of equilibrium is far below µ = 0,5. According to the underlying logic of the model, the magnitude of µ is informative about how strong an impact does the differentiation and substitution (between renewable energies, and the non-renewable ones), have on the size of the population in a given time and place. In countries with µ > 0.5, population is being built mostly through access to energy, and through substitution between various forms of energy. Conversely, in countries displaying µ < 0,5, access to food, and internal substitution between various forms of food becomes more important regarding demographic change. United States of America come as one of those big surprises. In this respect, empirical check brings a lot of idiosyncrasies to the initial lines of the theoretical model.

Countries accompanied with a (!) are exceptions with respect to the magnitude of the scale factor ‘A’. They are: China, India, Cyprus, Estonia, Gabon, Iceland, Luxembourg, New Zealand, Norway, Slovenia, as well as Trinidad and Tobago. They present a common trait of satisfactorily low a variance in scale factor ‘A’, in conformity with condition (6), but a mean ‘A’ either unusually high (China A = 1.32, India A = 1.40), or unusually low (e.g. Iceland A = 0.02), whatever the value of exponent ‘µ’. It could be just a technical limitation of the model: when operating on absolute, non-transformed values, the actual magnitudes of variance on both sides of the equation matter. Motor traffic is an example: if the number of engine-powered vehicles in a country grows spectacularly, in the presence of a demographic standstill, variance on the right side is much greater than on the left side, and this can affect the scale factor. Yet, variances observable in the scale factor ‘A’, with respect to those exceptional cases, are quite low, and a fundamental explanation is possible. Those countries could be the cases, where the available amounts of food and energy either cannot really produce as big a population as there really is (China, India), or, conversely, they could produce much bigger a population than the current one (Iceland is the most striking example). From this point of view, the model could be able to identify territories with no room left for further demographic growth, and those with comfortable pockets of food and energy to sustain much bigger populations. An interpretation in terms of economic geography is also plausible: these could be situations, where official, national borders cut through human habitats, such as determined by energy and food, rather than circling them.

Partially wrapping it up, results in Table 1 demonstrate that equation (3) of the model is both robust and apt to identify local idiosyncrasies. The blade having been sharpened, the next step of empirical check consisted in replacing the overall consumption of energy per capita with just the consumption of renewable energies, as calculated on the grounds of data published by the World Bank, and in retesting equation (3) on the same countries. Table 2, in the Appendix to Chapter I, shows the results of those 116 tests. The presentational convention is the same (just to keep in mind that values in A and in µ correspond to renewable energy in the equation), and the last column of the table supplies a quotient, which, fault of a better expression, is named ‘rate of substitution between renewable and non-renewable energies’. The meaning of that substitution quotient appears as one studies values observed in the scale factor ‘A’. In the great majority of countries, save for exceptions marked with (!), it was possible to define a neighbourhood of equilibrium regarding equation (3) and condition (6). Exceptions are treated as such, this time, mostly due to unusually (and unacceptably) high a variance in scale factor ‘A’. They are countries where deriving population from access to food and renewable energies is a bit dubious, regarding the robustness of prediction with equation (3).

The provisional bottom line is that for most countries, it is possible to derive, plausibly, the size of population in the given place and time from both the overall consumption of energy, and from the use of just the renewable energies, in the presence of relatively constant an alimentary intake. Similar, national idiosyncrasies appear as in Table 1, but this time, another idiosyncrasy pops up: the gap between µ exponents in the two empirical mutations of equation (3). The µ ascribed to renewable energy per capita is always lower than the µ corresponding to the total use of energy – for the sake of presentational convenience they are further being addressed as, respectively, µ(R/N), and µ(E/N) –  but the proportions between those two exponents vary greatly between countries. It is useful to go once again through the logic of µ. It is the exponent, which has to be ascribed to the consumption of energy per capita in order to produce a neighbourhood of equilibrium in population, in the presence of relatively constant an alimentary regime. For each individual country, both µ(R/N) and µ(E/N) correspond to virtually the same mean and variance in the scale factor ‘A’. If both the total use of energy, and just the consumption of renewable energies can produce such a neighbourhood of equilibrium, the quotient ‘µ(E/N)/µ(R/N)’ reflects the amount of total energy use, in tons of oil equivalent per capita, which can be replaced by one ton of oil equivalent per capita in renewable energies, whilst keeping that neighbourhood of equilibrium. Thus, the quotient µ(E/N)/µ(R/N) can be considered as a levelled, long-term rate of substitution between renewable energies and the non-renewable ones.

One possible objection is to be dealt with at this point. In practically all countries studied, populations use a mix of energies: renewable plus non-renewable. The amount of renewable energies used per capita is always lower than the total use of energy. Mathematically, the magnitude of µ(R/N) is always smaller than the one observable in µ(E/N). Hence, the quotient µ(E/N)/µ(R/N) is bound to be greater than one, and the resulting substitution ratio could be considered as just a mathematical trick. Still, the key issue here is that both ‘E/Nµ’ and ‘R/Nµ’ can produce a neighbourhood of equilibrium with a robust scale factor. Translating maths into the facts of life, the combined results of tables 1 and 2 (see Appendix) strongly suggest that renewable energies can reliably produce a general equilibrium in, and sustain, any population on the planet, with a given supply of food. If a given factor A is supplied in relatively smaller an amount than the factor B, and, other things held constant, the supply of A can produce the same general equilibrium than the supply of B, A is a natural substitute of B at a rate greater than one. Thus, µ(E/N)/µ(R/N) > 1 is far more than just a mathematical accident: it seems to be the structural property of our human civilisation.

Still, it is interesting how far does µ(E/N)/µ(R/N) reach beyond the 1:1 substitution. In this respect, probably the most interesting insight is offered by the exceptions, i.e. countries marked with (!), where the model fails to supply a 100%-robust scale factor in any of the two empirical mutations performed on equation (3). Interestingly, in those cases the rate of substitution is exactly µ(E/N)/µ(R/N) = 1. Populations either too big, or too small, regarding their endowment in energy, do not really have obvious gains in sustainability when switching to renewables.  Such a µ(E/N)/µ(R/N) > 1 substitution occurs only when the actual population is very close to what can be modelled with equation (3). Two countries – Saudi Arabia and Turkmenistan – offer an interesting insight into the underlying logic of the µ(E/N)/µ(R/N) quotient. They both present µ(E/N)/µ(R/N) > 2. Coherently with the explanation supplied above, it means that substituting renewable energies for the non-renewable ones, in those two countries, can fundamentally change their social structures and sustain much bigger populations. Intriguingly, they are both ‘resource-cursed’ economies, with oil and gas taking so big a chunk in economic activity that there is hardly room left for anything else.

Most countries on the planet, with just an exception in the cases of China and India, seem being able to sustain significantly bigger populations than their present ones, through shifting to 100% renewable energies. In two ‘resource-cursed’ cases, namely Saudi Arabia and Turkmenistan, this demographic shift, possible with renewable energies, seems not less than dramatic. As I was progressively wrapping my mind around it, a fundamental question formed: what exactly am I measuring with that logarithm µ? I returned to the source of my inspiration, namely to the model presented by Paul Krugman in 1991 (Krugman 1991 op. cit.). That of the two factors on the right side of the equation, which is endowed with the dominant power is, in the same time, the motor force behind the spatial structuring of human settlement. I have, as a matter of fact, three factors in my model: non-edible renewable energy, substitutable to non-edible and non-renewable energy, and the consumption of food per capita. As I contemplate these three factors, a realisation dawns: none of the three can be maximized or even optimized directly. When I use more electricity than I did five years earlier, it is not because I plug my fingers more frequently into the electric socket: I shape my consumption of energy through a bundle of technologies that I use. As for the availability of food, the same occurs: with the rare exception of top-level athletes, the caloric intake is the by-product of a life style (office clerk vs construction site worker) rather than a fully conscious, purposeful action. Each of the three factors is being absorbed through a set of technologies. Here, some readers may ask: if I grow vegetables in my own garden, isn’t it far-fetched to call it a technology? If we were living in a civilisation who feeds itself exclusively with home-grown vegetables, that could be an exaggeration, I agree. Yet, we are a civilisation, which has developed a huge range of technologies in industrial farming. Vegetables grown in my garden are substitutes to foodstuffs supplied from industrially run farms, as well as to industrially processed food. If something is functionally a substitute to a technology, it is a technology, too. The exponents obtained, according to my model, for particular factors, in individual countries, reflect the relative pace of technological change in three fundamental fields of technology, namely:

  1. a) Everything that makes us use non-edible energies, ranging from a refrigerator to a smartphone; here, we are mostly talking about two broad types of technologies, namely engines of all kind, and electronic devices.
  2. b) Technologies that create choice between the renewable, and the non-renewable sources of energy, thus first and foremost the technologies of generating electricity: windmills, watermills, photovoltaic installations, solar-thermal plants etc. They are, for the most part, one step earlier in the chain of energy than technologies mentioned in (a).
  3. c) Technologies connected to the production and consumption of food, composed into a long chain, with side-branches, starting from farming, through the processing of food, ending with packaging, distribution, vending and gastronomy.

As I tested the theoretical equation N = A*(E/N)µ*(F/N)1-µ’, most countries yielded a plausible, robust equilibrium between the local (national) headcount, and the specific, local mix of technologies grouped in those three categories. A question emerges, as a hypothesis to explore: is it possible that our collective intelligence expresses itself in creating such, local technological mixes of engines, electronics, power generation, and alimentary technologies, which, in turn would allow us to optimize our population? Can technological change be interpreted as an intelligent, energy-maximizing adaptation?

Appendix to Chapter I

Table 1 Parameters of the function:  Population = (Energy use per capita[3])µ*(Food intake per capita[4])(1-µ)

Country name Average daily intake of food, in kcal per capita Mean scale factor ‘A’ over 1990 – 2014 Variance in the scale factor ‘A’ over 1990 – 2014 The exponent ‘µ’ of the ‘energy per capita’ factor
Albania 2787,5 1,028719088 0,048263309 0,78
Algeria 2962,5 1,00792777 0,003115684 0,5
Angola 1747,5 1,042983003 0,034821077 0,52
Argentina 3085 1,05449632 0,001338937 0,53
Armenia 2087,5 1,027874602 0,083587662 0,8
Australia 3120 1,053845754 0,005038742 0,77
Austria 3685 1,021793945 0,002591508 0,87
Azerbaijan 2465 1,006243759 0,044217939 0,74
Bangladesh 2082,5 1,045244854 0,007102476 0,21
Belarus 3142,5 1,041609177 0,016347323 0,8
Belgium 3655 1,004454515 0,003480147 0,88
Benin 2372,5 1,030339133 0,034533869 0,61
Bolivia (Plurinational State of) 2097,5 1,019990919 0,003429637 0,62
Bosnia and Herzegovina (!) 2862,5 1,037385012 0,214843872 0,81
Botswana 2222,5 1,068786155 0,009163141 0,92
Brazil 2907,5 1,013624942 0,003643215 0,26
Bulgaria 2847,5 1,058220643 0,005405994 0,82
Cameroon 2110 1,021629875 0,051074111 0,5
Canada 3345 1,036202396 0,007687519 0,73
Chile 2785 1,027291576 0,003554446 0,65
China (!) 2832,5 1,328918607 0,002814054 0,01
Colombia 2582,5 1,074031013 0,013875766 0,44
Congo 2222,5 1,078933108 0,024472619 0,71
Costa Rica 2802,5 1,050377494 0,005668136 0,78
Côte d’Ivoire 2460 1,004959783 0,007587564 0,52
Croatia 2655 1,072976483 0,009344081 0,72
Cyprus (!) 3185 0,325015959 0,00212915 0,99
Czech Republic 3192,5 1,004089056 0,002061036 0,84
Denmark 3335 1,007673381 0,006893499 0,93
Dominican Republic 2217,5 1,062919767 0,006550924 0,65
Ecuador 2225 1,072013967 0,00294547 0,6
Egypt 3172,5 1,036345512 0,004306619 0,38
El Salvador 2510 1,013036366 0,004187964 0,7
Estonia (!) 2980 0,329425185 0,001662589 0,99
Ethiopia 1747,5 1,073625398 0,039032523 0,31
Finland (!) 3147,5 0,788769669 0,002606412 0,99
France 3557,5 1,019371541 0,001953865 0,53
Gabon (!) 2622,5 0,961643759 0,016248519 0,99
Georgia 2350 1,044229266 0,059636113 0,76
Germany 3440 1,009335161 0,000335601 0,48
Ghana 2532,5 1,000098029 0,047085907 0,48
Greece 3610 1,063074 0,003756555 0,77
Haiti 1815 1,038427773 0,004246483 0,56
Honduras 2457,5 1,030624938 0,005692923 0,67
Hungary 3440 1,024235523 0,001350114 0,78
Iceland (!) 3150 0,025191922 2,57214E-05 0,99
India (!) 2307,5 1,403800869 0,024395268 0,01
Indonesia 2497,5 1,001768442 0,004578895 0,2
Iran (Islamic Republic of) 3030 1,034945678 0,001105326 0,45
Ireland 3622,5 1,007003095 0,017135706 0,96
Israel 3490 1,008446182 0,013265865 0,87
Italy 3615 1,007727182 0,001245927 0,51
Jamaica 2712,5 1,056188543 0,01979275 0,9
Japan 2875 1,0094237 0,000359135 0,38
Jordan 2820 1,015861129 0,031905756 0,77
Kazakhstan 3135 1,01095925 0,021868381 0,74
Kenya 2010 1,018667155 0,02914075 0,42
Kyrgyzstan 2502,5 1,009443502 0,053751489 0,71
Latvia 3015 1,010440502 0,023191031 0,98
Lebanon 3045 1,036073511 0,054610186 0,85
Lithuania 3152,5 1,008092894 0,025234007 0,96
Luxembourg (!) 3632,5 0,052543325 6,62285E-05 0,99
Malaysia 2855 1,017853322 0,001002682 0,61
Mauritius 2847,5 1,070576731 0,019964794 0,96
Mexico 3165 1,01483014 0,009376118 0,36
Mongolia 2147,5 1,061731985 0,030246541 0,9
Morocco 3095 1,07892333 0,000418636 0,47
Mozambique 1922,5 1,023422366 0,041833717 0,48
Nepal 2250 1,059720031 0,006741455 0,46
Netherlands 2925 1,040887411 0,000689576 0,78
New Zealand (!) 2785 0,913678062 0,003946867 0,99
Nicaragua 2102,5 1,045412214 0,007065561 0,69
Nigeria 2527,5 1,069148598 0,032086946 0,28
Norway (!) 3340 0,760631741 0,001570101 0,99
Pakistan 2275 1,062522698 0,020995863 0,24
Panama 2347,5 1,007449033 0,00243433 0,81
Paraguay 2570 1,07179452 0,021405906 0,73
Peru 2280 1,050166142 0,00327043 0,47
Philippines 2387,5 1,0478458 0,022165841 0,32
Poland 3365 1,004848541 0,000688294 0,56
Portugal 3512,5 1,036215564 0,006604633 0,76
Republic of Korea 3027,5 1,01734341 0,011440406 0,56
Republic of Moldova 2762,5 1,002387234 0,038541243 0,8
Romania 3207,5 1,003204035 0,003181708 0,62
Russian Federation 3032,5 1,050934925 0,001953049 0,38
Saudi Arabia 2980 1,026310231 0,007502008 0,72
Senegal 2187,5 1,05981161 0,021382472 0,54
Serbia and Montenegro 2787,5 1,0392151 0,012416926 0,8
Slovakia 2875 1,011063497 0,002657276 0,92
Slovenia (!) 3042,5 0,583332004 0,003458657 0,99
South Africa 2882,5 1,053438343 0,009139913 0,53
Spain 3322,5 1,061083277 0,004844361 0,56
Sri Lanka 2287,5 1,029495671 0,001531167 0,5
Sudan 2122,5 1,028532781 0,044393335 0,4
Sweden 3072,5 1,018026405 0,004626486 0,91
Switzerland 3385 1,047790357 0,007713383 0,88
Syrian Arab Republic 2970 1,010909679 0,017849377 0,59
Tajikistan 2012,5 1,004745997 0,078394669 0,62
Thailand 2420 1,05305435 0,004200173 0,41
The former Yugoslav Republic of Macedonia 2755 1,064764097 0,003242024 0,95
Togo 2020 1,007094875 0,014424982 0,66
Trinidad and Tobago (!) 2645 0,152994618 0,003781236 0,99
Tunisia 3230 1,053626454 0,001201886 0,66
Turkey 3510 1,02188909 0,001740729 0,43
Turkmenistan 2620 1,003674668 0,024196536 0,96
Ukraine 3040 1,044110717 0,005180992 0,54
United Kingdom 3340 1,028560563 0,006711585 0,52
United Republic of Tanzania 1987,5 1,074441381 0,031503549 0,41
United States of America 3637,5 1,023273537 0,006401009 0,3
Uruguay 2760 1,014226024 0,019409309 0,82
Uzbekistan 2550 1,056807711 0,031469698 0,59
Venezuela (Bolivarian Republic of) 2480 1,048332115 0,012077362 0,6
Viet Nam 2425 1,050131152 0,000866138 0,31
Yemen 2005 1,076332698 0,029772287 0,47
Zambia 1937,5 1,0479534 0,044241343 0,59
Zimbabwe 2035 1,063047787 0,022242317 0,6

Source: author’s


Table 2 Parameters of the function:  Population = (Renewable energy use per capita[5])µ*(Food intake per capita[6])(1-µ)

Country name Mean scale factor ‘A’ over 1990 – 2014 Variance in the scale factor ‘A’ over 1990 – 2014 The exponent ‘µ’ of the ‘renewable energy per capita’ factor The rate of substitution between renewable and non-renewable energies[7]
Albania 1,063726823 0,015575246 0,7 1,114285714
Algeria 1,058584384 0,044309122 0,44 1,136363636
Angola 1,044147837 0,063942546 0,49 1,06122449
Argentina 1,039249286 0,005115111 0,39 1,358974359
Armenia 1,082452967 0,023421839 0,59 1,355932203
Australia 1,036777388 0,009700331 0,52 1,480769231
Austria 1,017958672 0,007854467 0,71 1,225352113
Azerbaijan 1,07623299 0,009740098 0,47 1,574468085
Bangladesh 1,088818696 0,017086232 0,2 1,05
Belarus (!) 1,017676486 0,142728478 0,51 1,568627451
Belgium 1,06314732 0,095474709 0,52 1,692307692
Benin (!) 1,045986178 0,101094528 0,58 1,051724138
Bolivia (Plurinational State of) 1,078219551 0,034143037 0,53 1,169811321
Bosnia and Herzegovina 1,077445974 0,084400986 0,66 1,227272727
Botswana 1,022264687 0,056890261 0,79 1,164556962
Brazil 1,066438509 0,005012883 0,24 1,083333333
Bulgaria (!) 1,022253185 0,190476288 0,55 1,490909091
Cameroon 1,040548202 0,059668736 0,5 1
Canada 1,02539319 0,005170473 0,56 1,303571429
Chile 1,006307911 0,001159941 0,55 1,181818182
China 1,347729029 0,003248871 0,01 1
Colombia 1,016164864 0,019413193 0,37 1,189189189
Congo 1,041474959 0,030195913 0,67 1,059701493
Costa Rica 1,008081248 0,01876342 0,68 1,147058824
Côte d’Ivoire 1,013057174 0,009833628 0,5 1,04
Croatia 1,072976483 0,009344081 0,72 1
Cyprus (!) 1,042370253 0,838872562 0,72 1,375
Czech Republic 1,036681212 0,044847525 0,56 1,5
Denmark 1,008202138 0,059873591 0,68 1,367647059
Dominican Republic 1,069124974 0,020305242 0,53 1,226415094
Ecuador 1,008104202 0,025383593 0,47 1,276595745
Egypt 1,03122058 0,016484947 0,28 1,357142857
El Salvador 1,078008598 0,028182822 0,64 1,09375
Estonia (!) 1,062618744 0,418196957 0,88 1,125
Ethiopia 1,01313572 0,036192629 0,3 1,033333333
Finland 1,065855419 0,021967408 0,85 1,164705882
France 1,021262046 0,002151713 0,38 1,394736842
Gabon 1,065944525 0,011751745 0,97 1,020618557
Georgia 1,011709194 0,012808503 0,66 1,151515152
Germany 1,008843147 0,03636378 0,31 1,548387097
Ghana (!) 1,065885579 0,106721005 0,46 1,043478261
Greece 1,033613511 0,009328533 0,55 1,4
Haiti 1,009030442 0,005061414 0,54 1,037037037
Honduras 1,028253048 0,022719417 0,62 1,080645161
Hungary 1,086698434 0,022955955 0,54 1,444444444
Iceland 0,041518305 0,000158837 0,99 1
India 1,414055357 0,025335408 0,01 1
Indonesia 1,003393135 0,008680379 0,18 1,111111111
Iran (Islamic Republic of) 1,06172763 0,011215001 0,26 1,730769231
Ireland 1,075982896 0,02796979 0,61 1,573770492
Israel 1,06421352 0,004086618 0,61 1,426229508
Italy 1,072302127 0,020049639 0,36 1,416666667
Jamaica 1,002749054 0,010620317 0,67 1,343283582
Japan 1,082461225 0,000372112 0,25 1,52
Jordan 1,025652757 0,024889809 0,5 1,54
Kazakhstan 1,078500526 0,007887364 0,44 1,681818182
Kenya 1,039952786 0,031445338 0,41 1,024390244
Kyrgyzstan 1,036451717 0,011487047 0,6 1,183333333
Latvia 1,02535782 0,044807273 0,83 1,180722892
Lebanon 1,050444418 0,053181784 0,6 1,416666667
Lithuania (!) 1,076146779 0,241465686 0,72 1,333333333
Luxembourg (!) 1,080780192 0,197582319 0,93 1,064516129
Malaysia 1,018207799 0,034303031 0,42 1,452380952
Mauritius 1,081652351 0,082673843 0,79 1,215189873
Mexico 1,01253558 0,019098478 0,27 1,333333333
Mongolia 1,073924505 0,017542414 0,6 1,5
Morocco 1,054779512 0,005553697 0,38 1,236842105
Mozambique 1,062086076 0,047101957 0,48 1
Nepal 1,02819587 0,008319264 0,45 1,022222222
Netherlands 1,079123029 0,043322084 0,46 1,695652174
New Zealand 1,046855187 0,004522505 0,83 1,192771084
Nicaragua 1,034941617 0,021798159 0,64 1,078125
Nigeria 1,03609124 0,030236501 0,27 1,037037037
Norway 1,019025526 0,002937442 0,95 1,042105263
Pakistan 1,068995505 0,026598749 0,22 1,090909091
Panama 1,001556162 0,038760767 0,69 1,173913043
Paraguay 1,049861415 0,030603983 0,69 1,057971014
Peru 1,06820116 0,008122931 0,41 1,146341463
Philippines 1,045289953 0,035957042 0,28 1,142857143
Poland 1,035431925 0,035915212 0,39 1,435897436
Portugal 1,044901969 0,003371242 0,62 1,225806452
Republic of Korea 1,06776762 0,017697832 0,31 1,806451613
Republic of Moldova 1,009542233 0,033772795 0,55 1,454545455
Romania 1,011030974 0,079875735 0,47 1,319148936
Russian Federation 1,083901796 0,000876184 0,24 1,583333333
Saudi Arabia 1,099133179 0,080054524 0,27 2,666666667
Senegal 1,019171218 0,032304226 0,49 1,102040816
Serbia and Montenegro 1,042141223 0,00377058 0,63 1,26984127
Slovakia 1,062546838 0,08862799 0,61 1,508196721
Slovenia 1,00512965 0,039266211 0,81 1,222222222
South Africa 1,056957556 0,012656394 0,41 1,292682927
Spain 1,017435095 0,002522983 0,4 1,4
Sri Lanka 1,003117252 0,000607856 0,47 1,063829787
Sudan 1,00209188 0,060026529 0,38 1,052631579
Sweden 1,012941105 0,003898173 0,77 1,181818182
Switzerland 1,07331184 0,000878485 0,69 1,275362319
Syrian Arab Republic 1,048889583 0,03494333 0,38 1,552631579
Tajikistan 1,03533923 0,055646586 0,58 1,068965517
Thailand 1,012034765 0,002131649 0,33 1,242424242
The former Yugoslav Republic of Macedonia (!) 1,021262823 0,379532891 0,72 1,319444444
Togo 1,030339186 0,024874996 0,64 1,03125
Trinidad and Tobago 1,086840331 0,014786844 0,69 1,434782609
Tunisia 1,042654904 0,000806403 0,52 1,269230769
Turkey 1,0821418 0,019688124 0,35 1,228571429
Turkmenistan (!) 1,037854925 0,614587094 0,38 2,526315789
Ukraine 1,022041527 0,026351574 0,31 1,741935484
United Kingdom 1,028817158 0,017810219 0,3 1,733333333
United Republic of Tanzania 1,0319973 0,033120507 0,4 1,025
United States of America 1,001298132 0,001300399 0,19 1,578947368
Uruguay 1,025162405 0,027221297 0,73 1,123287671
Uzbekistan 1,105591195 0,008303345 0,36 1,638888889
Venezuela (Bolivarian Republic of) 1,044353155 0,012830255 0,45 1,333333333
Viet Nam 1,005825608 0,003779368 0,28 1,107142857
Yemen 1,072879389 0,058580323 0,3 1,566666667
Zambia 1,045147143 0,038548336 0,58 1,017241379
Zimbabwe 1,030974989 0,008692551 0,57 1,052631579

Source: author’s

[1] Charles W. Cobb, Paul H. Douglas, 1928, A Theory of Production, The American Economic Review, Volume 18, Issue 1, Supplement, Papers and Proceedings of the Fortieth Annual Meeting of the American Economic Association (March 1928), pp. 139 – 165

[2] Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer (2015), “The Next Generation of the Penn World Table” American Economic Review, 105(10), 3150-3182, available for download at

[3] Current annual use per capita, in tons of oil equivalent

[4] Annual caloric intake in mega-calories (1000 kcal) per capita, averaged over 1990 – 2014.

[5] Current annual use per capita, in tons of oil equivalent

[6] Annual caloric intake in mega-calories (1000 kcal) per capita, averaged over 1990 – 2014.

[7] This is the ratio of two logarithms, namely: µ(renewable energy per capita) / µ(total energy use per capita)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s