1: \documentclass[12pt,a4paper]{article}
2:
3: \usepackage{pslatex}
4: \usepackage{graphicx}
5: \usepackage{setspace}
6: \usepackage{array}
7:
8: % Note: the hyperref package must be loaded last
9: %\usepackage[ps2pdf, bookmarks, pdfview={FitH}, pdfstartview={FitH}]{hyperref}
10: %\usepackage[ps2pdf, colorlinks, bookmarks, pdfview={FitH}, pdfstartview={FitH}]{hyperref}
11: %\usepackage[bookmarks]{hyperref}
12: \usepackage[ps2pdf, colorlinks, bookmarks, pdfview={FitH}, pdfstartview={FitH}]{hyperref}
13:
14: %\doublespacing
15: \singlespacing
16:
17:
18: \def\noi{\noindent}
19:
20:
21: \begin{document}
22:
23: \date{}
24:
25:
26: \title{{\bf Avoiding Chaos in Wonderland}\\
27: }
28:
29:
30: \author{G.A. Kohring\\
31: C\&C Research Laboratories, NEC Europe Ltd. \\
32: Rathausallee 10, D-53757 St. Augustin, Germany\\
33: kohring@ccrl-nece.de
34: }
35:
36: \maketitle
37:
38: \bigskip
39:
40: \begin{abstract}
41: Wonderland, a compact, integrated economic, demographic and environmental
42: model is
43: investigated using methods developed for studying critical phenomena.
44: Simulation results show the parameter space separates into two phases, one of
45: which contains the property of long term, sustainable development.
46: By employing information contain in the phase diagram, an optimal strategy
47: involving pollution taxes is developed as a means of moving a system initially
48: in a unsustainable region of the phase diagram into a region of sustainability
49: while ensuring minimal regret with respect to long term economic growth.
50: \end{abstract}
51:
52: \bigskip
53: \noindent{\bf PACS-2003:}
54: 89.65-s,89.65.Gh,89.75.-k
55:
56: \noindent{\bf Keywords:}
57: Wonderland Model, Critical Phenomena, Complex Systems, Sustainable Development
58:
59:
60: \bigskip
61:
62: \noindent Submitted to \textit{Physica A}.
63:
64: \newpage
65:
66:
67: \section{Introduction}
68:
69: Long-term planning to ensure a sustainable world is a difficult problem for
70: many reasons, not the least of which is the lack of certainty surrounding
71: mankind's interactions with his environment. Given a host
72: of models with varying levels of complexity \cite{CHA00}; it is difficult
73: to develop a unified methodology for understanding the intricate dynamics
74: contained within these models. A common
75: approach to problem of long-term planning is the \textit{Scenario} method
76: \cite{MRM04,HUG99,RBG02}; whereby, the goal is to create a small set of
77: reasonable scenarios extrapolating from the present into possible futures.
78: These scenarios are often supported by quite large computer simulations
79: of models
80: describing the different sectors of an integrated world. Generally, these
81: models
82: include economic, demographic and environmental sectors \cite{MRM04}, though
83: they may also include political \cite{HUG99} and security \cite{RBG02}
84: components as well. Basically, the idea behind the scenario approach is to
85: selects a reasonable set of
86: parameters for the model in question, then integrate the model forward over
87: the planning time frame in order to ascertain what type of world might evolve.
88:
89: While the scenario approach yields readily understandable projections based upon
90: the current state and justifiable assumptions; its basic weakness is the
91: complexity of the models themselves. The values of some of the parameters which
92: enter the models are not know with any accuracy and can only be inferred from
93: the current state of the system; at the same time
94: some of the equations describing the system are chaotic. This combination of
95: vaguely known
96: parameters and chaotic equations of motion cast doubt on the reliability of
97: the forecasts \cite{CAM05}, leading critics to charge the authors with
98: cherry picking, i.e., selecting sets of parameters to fit the authors
99: preconceived notions of how the world should evolve.
100:
101: Recently, Bankes, et al. \cite{BLP01,LPB03} suggested a more fundamental
102: approach
103: based upon large scale computer simulations, specifically aimed at avoiding
104: the charges of cherry picking. By using Monte Carlo techniques to create a
105: large ensemble of scenarios, they propose to avoid bias towards any one
106: scenario. Projections used for long-term planning are then based upon the
107: probability of any given scenario occurring in the ensemble.
108:
109: In this paper the ensemble approach is extended to a systematic exploration
110: of the parameter space, whereby the equations of motion are treated as
111: equations of state, leading to a multi-dimensional phase diagram. The phase
112: diagram allows a clean separation of the parameter space into regions
113: where sustainable development is possible and where it is not. This removes
114: the guess work from the construction of plausible scenarios and brings more
115: focus to the ensemble method.
116:
117: For discussing the current proposal in more depth, the Wonderland model
118: \cite{SAN94,HL98,LPB03}\nocite{LUT94} has been chosen as an example
119: because of its relative simplicity and tractability. Although it lacks the
120: details of many other models, it still captures enough general behavior
121: of more sophisticated models to make its study worthwhile.
122:
123: The next section describes the Wonderland model in more detail, after which
124: a detailed discussion of Wonderland's behavior and phase diagram is
125: undertaken. Based upon these findings, section \ref{sec:tax} presents a
126: strategy for long term planning aimed at ensuring sustainable development.
127:
128:
129:
130: \section{The Wonderland Model}
131:
132: Sanderson's Wonderland model \cite{SAN94,HL98}\nocite{LUT94}
133: describes in an integrative
134: framework the economic, demographic and environmental sectors of an idealized
135: world. The model is characterized by four state variables, $\{x(t),\ y(t),\
136: z(t),\ p(t)\}$, representing the population, per capita output, stock of
137: natural
138: capital and pollution flow per unit of output respectively. These four state
139: variables evolve according to following set of non-linear, difference
140: equations:
141:
142: \begin{eqnarray}
143: x(t+1) &=& x(t)\left[1+n\Big(y(t),z(t)\Big)\right],
144: \label{eq:x_state} \\
145: y(t+1) &=&
146: y(t)\left(1+\gamma-(\gamma+\eta)\Big[1-z(t)\Big]^{\lambda}\right),
147: \label{eq:y_state}\\
148: z(t+1)
149: &=&\frac{g\Big(x(t),y(t),z(t),p(t)\Big)}{1+g\Big(x(t),y(t),z(t),p(t)\Big)},
150: \label{eq:z_state}\\
151: p(t+1) &=&p(t)(1-\chi).
152: \label{eq:p_state}
153: \end{eqnarray}
154:
155: \noi The state variables for population ($x$) and per capita output ($y$) can
156: assume all non-negative real
157: values ($x,y\in[0,\infty)$), while the stock of natural capital ($z$) and the
158: pollution per unit of output ($p$) are confined to the unit
159: interval ($z,p\in[0,1]$). A value of $z=1$ represents a full stock of
160: unpolluted natural
161: capital and $z=0$ represents the fully polluted state. $p=1$
162: on the other hand represents maximal pollution per unit of output and $p=0$
163: implies no pollution per unit of output.
164:
165:
166: In eq.~\ref{eq:x_state},
167: the endogenous population growth rate, $n(y,z)$, can be written as
168: as the difference between the crude birth rate, $b(y,z)$ (number of births
169: per $1\,000$ population per time at a given time step) and the crude
170: death rate, $d(y,z)$ (number of deaths per $1\,000$ population per time at a
171: given time step):
172:
173: \begin{eqnarray}
174: n(y,z)&=&b(y,z) - d(y,z)
175: \label{eq:n} \\
176: b(y,z)&=&\beta_0\left[\beta_1 - \left(
177: \frac{e^{\beta y}}{1+e^{\beta y}}\right)\right],
178: \label{eq:b} \\
179: d(y,z)&=&\alpha_0\left[\alpha_1 - \left(
180: \frac{e^{\alpha y}}{1+e^{\alpha y}}\right)\right]
181: \left[1+\alpha_2(1-z)^{\theta}\right],
182: \label{eq:d}
183: \end{eqnarray}
184:
185: \noi The parameters $\beta$, $\beta_0$ and $\beta_1$ govern the birth rate,
186: while the parameters $\alpha$, $\alpha_0$, $\alpha_1$, $\alpha_2$ and
187: $\theta$ govern the death rate.
188: From eqs.~\ref{eq:b} and \ref{eq:d} it can be
189: seen how both the birth rate and death rate decrease with increases in the
190: per capita output, $y$. Furthermore, in eq.~\ref{eq:d}, the death rate
191: is seen to increase when the environment deteriorates, i.e., when $z$ decreases.
192: These effects are in line with recent studies relating population growth
193: with industrial output \cite{COH95,BLE05}.
194:
195: The function $g(x,y,z,p)$ which determines the time evolution of the
196: natural capital is given by:
197:
198: \begin{eqnarray}
199: g(x,y,z,p)&=& \frac{z}{1-z}\,e^{\,\delta z^{\rho}-\omega f(x,y,p)
200: },
201: \label{eq:g} \\
202: \end{eqnarray}
203:
204: \noi where, $f(x,y,p)$, is the pollution flow:
205:
206: \begin{equation}
207: f(x,y,p)=xyp.
208: \label{eq:f}
209: \end{equation}
210: The parameters $\delta$, $\rho$, and $\omega$ determine the pollution
211: flow at which the economic, environmental and demographic sectors are in
212: balance. This critical pollution flow,
213: $\delta z^{\rho}/\omega$, determines the rate at which the natural capital is
214: able to ameliorate the pollution flow, $f$. As can be seen from
215: eqs.~\ref{eq:z_state} and \ref{eq:g}, when $f=\delta z^{\rho}/\omega$
216: the level of natural capital remains constant and the economic sector is in
217: balance with the environmental sector.
218:
219: The form of $f$ originates from the I-PAT hypothesis \cite{EH71}. In its
220: original form, the I-PAT hypothesis states: a
221: population's impact on its environment is equal to its size multiplied by its
222: per capita
223: output and its level of technology. In wonderland, the impact is the
224: pollution flow and the level of technology is
225: represented by the pollution per unit of output; hence, $f=xyp$.
226:
227: Wonderland's economic sector is characterized by the parameters:
228: $\gamma$, $\eta$, $\lambda$ and $\chi$. $\gamma$ is the exogenous economic
229: growth
230: rate and determines how fast the economy could grow if its capital stock
231: were fully intact (see eq.~\ref{eq:y_state}). $\eta$ and $\lambda$ determine
232: how rapidly the economy deteriorates (recovers) when the capital stock declines
233: (increases). The parameter $\chi$ governs the economic decoupling rate,
234: i.e., the rate at
235: which technological innovations reduce the pollution flow per unit of output.
236:
237: Some variants of the Wonderland model include a term in eq.~\ref{eq:f}
238: governing pollution control expenditures \cite{GWM96,MPF96,HL98,LH02}.
239: Since pollution
240: control is essentially a policy decision and not an intrinsic part of the
241: model, in this paper, we
242: follow \cite{LPB03} and introduce pollution control later
243: in section~\ref{sec:tax} where long term planning is discussed in more
244: detail. Such an
245: approach allows us to cleanly separate the intrinsic dynamics of the
246: model as given above, and perturbations of the dynamics due to policy
247: decisions.
248:
249: As can be seen, equations~\ref{eq:x_state} to \ref{eq:f} depend upon
250: 15 positive parameters which govern the overall behavior of model; whereby
251: eight of them ($\alpha$, $\alpha_0$, $\alpha_1$, $\alpha_2$,
252: $\beta$, $\beta_0$, $\beta_1$ and $\theta$) determine the population growth,
253: four ($\delta$, $\rho$, $\omega$ and $\chi$) determine the state of the
254: environment and three ($\gamma$, $\eta$ and $\lambda$) determine the health
255: of the economy. In the next section we take some slices through this 15
256: dimensional parameter space to learn more about the models behavior.
257:
258:
259: \section{Phase Diagram \label{sec:phase}}
260:
261: As noted in the introduction, previous work on this model has mostly followed
262: a scenario based approach \cite{SAN94,GWM96,BLP01}, with the two most often
263: investigated scenarios being the so-called ``Dream'' scenario and the ``Horror"
264: scenario. The Dream scenario earns its name because it holds out the
265: possibility of continued economic growth combined with a stable population and
266: a healthy environment; whereas the Horror scenario depicts environmental
267: collapse followed by an economic collapse and a declining population.
268: Table~\ref{tab:dream} contains the parameters for the Dream scenario. In the
269: horror scenario, the only parameter whose value is changed, is the
270: decoupling rate, $\chi$, which takes on the new value $\chi=0.01$.
271:
272: Plots of the time evolution for the Dream and Horror scenarios are shown in
273: Figure~\ref{fig:traj}. Initially, both scenarios follow the same growth curves
274: in terms of per capita output and population; however, after
275: approximately 90
276: years, the world in the Horror scenario undergoes a spontaneous transition to
277: a phase marked by a depleted stock of natural capital,
278: an economic depression and population decline. The sudden collapse of the
279: environment after 90 years of
280: relative stability is indicative of a first order phase transition.
281:
282:
283: Until now, our analysis has concentrated upon the classical scenario approach.
284: As a first step in going beyond this approach, we examine the behavior of the
285: model when systematically moving through the parameter space along the line
286: $\chi=0.01$ to $\chi=0.04$ while holding all the other parameters fixed.
287: For this
288: purpose, define an order parameter, $t_c$, as the number of years before the
289: natural stock collapses. The justification of calling $t_c$
290: an order parameter \cite{SMA76} becomes apparent when examining
291: Figure~\ref{fig:scale}. As can be seen, $t_c$ follows the scaling
292: behavior expected in the presence of a second order phase transition:
293:
294: \begin{equation}
295: t_c\sim \Big(\chi -\chi_c\Big)^{\zeta},
296: \label{eq:tc}
297: \end{equation}
298:
299: \noi furthermore, $t_c$ is undefined for all $\chi \geq \chi_c$. From the data
300: in Figure~\ref{fig:scale}, the critical value of $\chi$ can be
301: estimated: $\chi_c\approx 0.0385$, along with the critical exponent,
302: $\zeta~\approx~0.945$. The symbols in Figure~\ref{fig:scale}, indicate different
303: starting states. Depending upon the exact initial state, the collapse may
304: take place sooner or latter, but the scaling behavior is the same. From this
305: data we can view $\chi_c$ as marking a phase boundary between the phase
306: of unsustainable development,
307: as epitomized by the Horror scenario, and the phase of sustainable development
308: as epitomized by the Dream scenario. (The reader should not take this diagram
309: to mean we are advocating a planning horizon extending to $100\, 000$ years!
310: Rather, the diagram indicates how rapidly $t_c$ changes given small changes in
311: $\chi$.)
312:
313: In the Wonderland model, the parameters most responsible for the
314: interactions of the economic and environmental sectors are $\gamma$ and
315: $\chi$. Therefore, if we repeat the above analysis for the remainder of
316: the $\gamma-\chi$
317: plane, we arrive at the phase diagram shown in Figure~\ref{fig:phase}.
318: The line of points delineates the parameter space into a phase of sustainable
319: development and one of unsustainable development. In the phase of sustainable
320: development the economic, demographic and environmental sectors of Wonderland
321: are in equilibrium, while in the phase of unsustainable development, these
322: sectors are out of equilibrium, eventually leading to a world wide collapse.
323: As one
324: approaches the phase boundary from below, the time before the collapse occurs
325: increases as a power law of the distance from the boundary.
326:
327: One can continue this exercise for the remaining parameters in the model;
328: however, most of the other parameters effect the quantitative results but not
329: the qualitative features described above. The parameters $\delta$, $\rho$ and
330: $\omega$, for example, determine the exact value of the critical
331: pollution flow above which
332: the environment begins deteriorating, but not the existence of a critical
333: pollution flow \cite{GWM96}.
334:
335: \section{Policy Planning \label{sec:tax}}
336:
337: In the previous section, the analysis focused on the intrinsic behavior of the
338: Wonderland model; however, this is only part of the problem, the other more
339: intricate question is whether or not it is possible to introduce control
340: mechanisms in order to avoid the phase transition in the Horror scenario and
341: the attendant catastrophic consequences.
342: Note, the model as described by
343: equations \ref{eq:x_state} through \ref{eq:f} has no steerable parameters. All
344: of the 15 parameters entering into the Model are in principal measurable
345: or can be estimated using available data \cite{SAN94,LPB03}; hence, once they
346: are known one can look on the phase diagram of Figure~\ref{fig:phase} to
347: determine whether the world of Wonderland is in a sustainable phase or
348: an unsustainable phase approaching a potentially catastrophic collapse.
349:
350: When faced with the question of how to handle sustainable development,
351: policy makers can choose to either control the emission of pollutants at their
352: source or to spend funds abating pollutants aftwards. Either of these
353: approaches may be effective or
354: they may induce undesirable side affects. As a first step, we look at the
355: the pollution abatement approach.
356:
357: \subsection{Pollution Abatement}
358:
359: To abate the effects of pollution, funds must be drawn from other sources
360: of income; furthermore, as the environmental degradation becomes more serious,
361: more funds are required. A simple, non-linear model governing the
362: expenditures for pollution abatement,
363: $c(y,z)$, has been proposed by previous authors \cite{SAN94,HL98,LH02}:
364:
365: \begin{equation}
366: c(y,z)=\phi(1-z)^{\mu}y.
367: \label{eq:c}
368: \end{equation}
369: In this model, as natural capital deteriorates, the expenditures to abate
370: pollution
371: increase, whereby the rate at which expenditures increase are governed by the
372: policy parameters $\phi$ and $\mu$. Withdrawing capital in this manner,
373: decreases the per capita output available for other uses; hence,
374: the per capita output, $y$, in equations
375: \ref{eq:b} and \ref{eq:d} must be replaced by $y^{\prime}=y-c$.
376: (Eq.~\ref{eq:y_state}
377: does not change, because goods and services needed for pollution abatement
378: are part of the overall output.) Furthermore,
379: since the aim is to improve the state of the environment, the pollution
380: flow is reduced by the effectiveness of the regulations, i.e.,
381: eq.~\ref{eq:f}, becomes\cite{SAN94,HL98,LH02}:
382:
383: \begin{equation}
384: f(x,y,p)=xyp - \kappa\frac{e^{\epsilon cx}}{1+e^{\epsilon cx}}.
385: \label{eq:freg}
386: \end{equation}
387: where, $\kappa$ determines the effectiveness of the expenditures, $c(y,z)$.
388:
389: Basically, policy makers have the parameters $\phi$, $\mu$ and $\kappa$
390: for control purposes, though none of these parameters can be varied without
391: limits. $\phi$ and $\mu$ determine how much output is diverted to pollution
392: abatement once the stock of natural capital starts deteriorating. Since we
393: require $c<y$, this limits $\phi < 1$. $\mu$ determines how quickly the
394: inhabitants of Wonderland respond to early signs of environmental degradation.
395: While these parameters cannot of themselves drive the system from the phase
396: of unsustainable development to the phase of sustainable development,
397: they do have an impact on the amount of chaos present in the
398: unsustainable phase.
399:
400: An intriguing new phenomena in this model is the environments recovery
401: from collapse. Indeed, one can simulate the Horror scenario over several
402: millennium,and observe how it undergoes repeating cycles of
403: collapse and recovery. To gain insight into the underlying dynamics it is
404: instructive to plot the orbits of the real growth rate as a function of
405: the stock of natural capital. The normalized real grow rate can be defined as:
406:
407:
408: \begin{equation}
409: r(t)=\frac{y(t)-y(t-1)}{\gamma\, y(t-1)},
410: \label{eq:r}
411: \end{equation}
412:
413: \noi where we have normalized the real growth rate by the exogenous growth
414: rate, $\gamma$. A plot of $r(t)$ versus $z(t)$ is shown in
415: Figure~\ref{fig:chaos}. When the environment is deteriorating, the economy
416: follows the upper curve and when the environment is improving, the economy
417: follows the lower curve. Hysteresis curves of this type are expected in the
418: presence of a first order phase transition.
419: From this figure it can be seen that the momentary
420: performance of the economy sheds little light on the overall health of
421: Wonderland, since the growth rate at first decreases only slowly with
422: the declining capital stock until the natural capital is nearly exhausted, at
423: which point the growth rate rapidly turns from positive to negative. The exact
424: shape of the orbits depend upon the parameters $\eta$ and $\lambda$ from
425: eq.~\ref{eq:y_state}. This
426: general picture is consistent with scenarios described by more detailed models
427: such as World3 \cite{MRM04}, thereby lending credence to the proposition of
428: using Wonderland as a tractable model for in depth studies.
429:
430: Notice how
431: the trajectories of the Horror scenario appear to contract to an aperiodic
432: recurrent attractor \cite{CAM05}. Using the techniques
433: described in \cite{RCL93}
434: we can estimated the Lyapunov exponent for this strange attractor:
435: $l\approx 0.026$. A positive value of the Lyapunov exponent is another
436: indication of the problems facing the Horror scenario as past experience does
437: not provide detailed guidance on future behavior. In the next cycle the
438: economy may
439: collapse more quickly or more slowly than it did in the previous cycle. This
440: type of behavior is typical of a dynamical system operating in the
441: chaotic regime \cite{CAM05}.
442:
443: While previous research
444: has shown it is possible to avoid the chaos and collapse in the
445: horror scenario by
446: setting $\kappa=100$ \cite{GWM96,MPF96,HL98}, such large values of
447: $\kappa$ seem unrealistic, because the first term in eq.~\ref{eq:freg}
448: is $O(1)$ initially.
449:
450: Unfortunately, pollution abatement alone, does not move the system from
451: the phase of unsustainble development to the phase of nsustainble development,
452: though it does
453: help to recover from an environmental collapse; in fact, as long
454: as the parameters remain within reasonable bounds they have no
455: impact on the phase diagram shown in Figures~\ref{fig:phase}.
456:
457: \subsection{Pollution Control}
458:
459: Pollution taxes have been introduced into the Wonderland model in
460: previous studies \cite{HL98,LH02,LPB03}. The goal of a pollution tax is to
461: increase the effective decoupling rate, $\chi$, by making pollution
462: unprofitable. In this paper, the pollution tax rate, $\tau$, enters the
463: model first via eq.~\ref{eq:p_state}, which changes to:
464:
465: \begin{equation}
466: p(t+1)=p(t)\left(1-\chi-\chi_0\frac{\tau}{1+\tau}\right).
467: \label{eq:p_w_tax}
468: \end{equation}
469: where, $\chi_0$ is the maximum \textit{additional} decoupling rate for the
470: assumed
471: level of technology, i.e., $\chi+\chi_0$ is the maximum achievable decoupling
472: rate for an assumed level of technology.
473:
474: The side affect of taxing pollution is a reduction in the real per capita
475: growth rate since resources are diverted into the pollution control; hence,
476: eq.~\ref{eq:y_state} becomes:
477:
478: \begin{equation}
479: y(t+1)=y(t)\left(1+\gamma-\left(\gamma+\eta\right)\Big[1-z(t)\Big]^{\lambda}
480: -\gamma_0\,\frac{\tau}{1-\tau}\right),
481: \label{eq:y_w_tax}
482: \end{equation}
483:
484: \noi where the parameter $\gamma_0$ determines the amount by which the
485: pollution tax retards growth. Theoretically, a modest pollution tax does not
486: diminish growth to the full extent of the tax, because the tax itself spurs
487: investment in pollution reduction technologies which in turn increase
488: growth; however, as the tax becomes larger, the numerator in
489: equation~\ref{eq:y_w_tax} tends toward zero and the tax can become a
490: considerable drag on the economy.
491:
492: By varying the decoupling rate, $\chi$ and the tax rate $\tau$, while keeping
493: the other parameters fixed to their values in Table~\ref{tab:dream} (with
494: $\gamma_0=0.5$ and $\chi_0=\chi/2$), the phase diagram
495: in Figure~\ref{fig:tax} can be constructed
496: using the techniques discussed in the last section. As can be seen, with the
497: help of a moderate pollution tax, the system can be moved from the phase of
498: unsustainable development to the phase of sustainable development.
499:
500: In todays political climate many of society's leaders do not dispute the
501: ability of pollution taxes or other remedies to improve the state of the
502: world's natural capital, rather they
503: claim the cost in terms of foregone economic development is too high. Indeed,
504: many leaders would prefer to maintain high per capita growth now and deal with
505: the aftermath of an environmental collapse later, in the same manner the
506: stock market bubble was allowed to expand at the end of 1990s until it burst
507: in 2000.
508:
509: Concentrating on short term growth rates alone, however, can be misleading. As
510: shown in Figure \ref{fig:volatility}, the average growth rate is the same
511: when the pollution tax rate is too low, or when the pollution tax is too high.
512: Hence, the average growth rate alone is not
513: definitive, policy makers must also be concerned about reducing economic
514: volatility \cite{RR95,EIS01}, i.e., reducing fluctuations in the growth rate.
515: Figure \ref{fig:volatility} also plots the variation in the normalized real
516: growth rate as a function of the pollution tax rate. As can be seen,
517: volatility drops to zero at the critical tax rate, which is also the point
518: where the real growth rate is a maximum. This is to be expected, since in
519: the phase of sustainable development, the Wonderland model contains no
520: mechanism to create endogenous variability in the economic growth rate.
521:
522: As an additional check on the viability of differing long term strategies, the
523: maximal regret\cite{SAV50} for alternative strategies should be calculated
524: and the strategy yielding the minimal of the maximal regret should be chosen.
525: Towards this end, we define the critical regret for per capita output for
526: different value of the tax rate, $\tau$ as:
527:
528: \begin{equation}
529: R_c(t,\tau)=\frac{y_c(t)-y(t,\tau)}{y_c(t)}
530: \label{eq:regret}
531: \end{equation}
532:
533: \noi where $y_c(t)$ is the per capita output at the critical value
534: of $\tau$, i.e., the value of $\tau$ on the phase boundary in
535: Figure~\ref{fig:tax} (all other parameters are assumed fixed).
536: Figure~\ref{fig:regret} depicts the critical regret as a function of time
537: for different
538: tax rates when all other parameters are held fix to the values they assume
539: in the Horror scenario. For tax rates lower than the critical tax rate, the
540: short term regret is negative, meaning faster economic growth than that
541: achievable at the critical tax rate; however, this faster growth is completely
542: dissipated once the environment collapses, leaving the long term regret at its
543: maximum possible value. For tax rates above the critical tax rate, economic
544: growth is slower, leading to increased regret. Hence, the critical tax rate is
545: the optimal tax rate in terms of Wonderland's long term prospects.
546:
547: The relative ineffectiveness of pollution abatement versus pollution control
548: was noted previously by Leeves and Herbert \cite{LH02} who studied a modified
549: version of the Wonderland model in which eq.~\ref{eq:y_state} is replaced by a
550: Cobb-Douglas production function. Their model showed a transient dynamics
551: consisting of large volatility in the per capita
552: output and the stock of natural capital. As in the present model, pollution
553: abatement expenditures were ineffective at eliminating the unwanted behavior,
554: whereas adequate levels of pollution control expenditures were effective in
555: eliminating the volatility.
556:
557: In summary, it seem reasonable to conclude that controlling pollution at the
558: source is more import as far as sustainability is concerned, than attempting
559: to abate its effects afterwards.
560:
561:
562: \section{Conclusion}
563:
564: The advantages of basically treating eqs.~\ref{eq:x_state}-\ref{eq:f} as
565: dynamical equations of state have been demonstrated. The phase diagram
566: depicted in Figure~\ref{fig:phase} contains an infinite number of scenarios,
567: some of which have the property of sustainable development, others of which
568: do not. In the lower right hand corner, for example, one can find
569: scenarios of environmental collapse due to over production; while in the
570: lower left hand corner are scenarios of collapse due to over population.
571:
572: The analysis of the preceding two sections has yielded the information
573: needed to formulate an optimal strategy for ensuring the long term health of
574: Wonderland. Such a strategy would consist of the following steps:
575:
576: \begin{enumerate}
577:
578: \item Determine the model parameters as accurately as possible.
579: \item Map out the phase diagram in that part of parameter space covered by
580: the parameters measured in step 1.
581: \item Determine the approximate critical pollution tax required for
582: sustainable development
583: \item Periodically repeat steps 1-3 adjusting the pollution tax as necessary.
584:
585: \end{enumerate}
586:
587: \noi As demonstrated by Figure~\ref{fig:regret}, this simple strategy is
588: optimal in the sense of yielding the minimal regret in the long term, at
589: the sacrifice of short term (and short lived) gains.
590:
591: One may question whether it is necessary to aim for the phase boundary, or
592: whether simply being ``not too far'' is sufficient, especially since the
593: time to collapse increases exponentially as one approaches the phase boundary.
594: The answer is given by the discussion in \ref{sec:phase} and in particular
595: Figures \ref{fig:traj} and \ref{fig:chaos}; namely, the system provides
596: little forewarning of an impending collapse. Relative stability may prevail for
597: decades, until the system suddenly undergoes a phase transition; hence, prudence
598: would dictate avoiding this parameter regime if at all possible.
599:
600: As noted above, the Wonderland model is a comparatively simple integrated
601: model. For example, the maximum growth rate, $\gamma$, is
602: treated in this model as an exogenous parameter; in reality the
603: maximum growth rate is a complex function of the state of the economy, the
604: the environment and the population. When all three
605: collapse as in Figure \ref{fig:traj}, $\gamma$ would also be expected to fall,
606: leading to a faster collapse and longer recovery times. Hence, the
607: phase diagram in Figure \ref{fig:phase}
608: should be taken as upper bound on the sustainability due to the use of a
609: constant $\gamma$. An extension of the Wonderland model to include more
610: realistic, endogenous growth through a Cobb-Douglas production function
611: has been studied by Leeves and Herbert \cite{LH02}.
612:
613:
614: In summary, this paper has demonstrated how concepts originally developed for
615: studying critical phenomena in physical systems can be successfully applied to
616: problems in long-term, socio-economic planning. Of course, Wonderland is a
617: toy model which captures the global features of the complete
618: human-environment interactions, but not its details. Indeed, detailed models
619: transferable to the real world, where the models can contain an
620: order of magnitude more equations \cite{CHA00}; however, we believe the
621: principals elucidated here can still apply. Furthermore, the
622: principled approach developed here can augment the ensemble
623: approach~\cite{LPB03} by applying large scale computer simulations more
624: systematically.
625:
626:
627: %\bigskip
628: %\bigskip
629: %\bigskip
630: %\noindent{\large\bf Acknowledgments}
631: %\smallskip
632: %
633: %\noindent The author would like to thank the reviewers for their useful and
634: %constructive commentary on this work.
635:
636:
637: \newpage
638:
639: \bibliographystyle{unsrt}
640: \bibliography{sd}
641:
642: \newpage
643: \appendix{\large\bf Tables}
644:
645: \bigskip
646: \bigskip
647:
648: \begin{table}[htbp]
649: \begin{center}
650: \begin{tabular}{||m{1cm}|m{1cm}||m{1cm}|m{1cm}||m{1cm}|m{1cm}||}
651: \hline
652: \multicolumn{2}{||c||}{\textbf{Economy}} &
653: \multicolumn{2}{c||}{\textbf{Environment}} &
654: \multicolumn{2}{c||}{\textbf{Population}}\\
655: \hline
656: name & value & name & value & name & value \\
657: \hline
658: $\gamma$ & 0.04 & $\chi$ & 0.04 & $\alpha$ & 0.09 \\
659: \hline
660: $\eta$ & 0.04 & $\delta$ & 1.0 & $\alpha_0$ & 10.0 \\
661: \hline
662: $\lambda$ & 2.0 & $\rho$ & 0.2 & $\alpha_1$ & 2.5 \\
663: \hline
664: & & $\omega$ & 0.1 & $\alpha_2$ & 2.0 \\
665: \hline
666: & & & & $\beta$ & 0.8 \\
667: \hline
668: & & & & $\beta_0$ & 40.0 \\
669: \hline
670: & & & & $\beta_1$ & 1.375 \\
671: \hline
672: & & & & $\theta$ & 15.0 \\
673: \hline
674: \end{tabular}
675: \caption{\label{tab:dream}Parameter values for the Dream scenario.}
676: \end{center}
677: \end{table}
678:
679: \newpage
680: \noi\appendix{\large\bf Figures}
681: \bigskip
682:
683: \bigskip
684: \noi Figure \ref{fig:traj} The basic scenarios Dream and Horror. (a) Semi-log
685: plot of the per capita output as a function of time. (b) Plot of relative
686: population as a function of time. (c) Plot of natural capital as a function of
687: time.
688:
689: \bigskip
690: \noi Figure \ref{fig:scale} Log-log plot of the time before the first
691: collapse of natural capital as a function of the distance $\chi-\chi_c$.
692: Different point types indicate different initial states of the system.
693:
694: \bigskip
695: \noi Figure \ref{fig:phase} Phase diagram in the $\gamma-\chi$ plane. The
696: line marks the boundary between the phases of sustainable, S,
697: and unsustainable, U, development. H marks the location of the Horror
698: scenario, while D marks the location of the Dream scenario.
699:
700: \bigskip
701: \noi Figure \ref{fig:chaos} Plot of the normalized, real growth rate as a
702: function of the stock of natural capital over a $40\, 000$ year time span.
703:
704: \bigskip
705: \noi Figure \ref{fig:tax} Phase diagram in the $\tau-\chi$ plane
706: when the other parameters are fixed to their values in Table~\ref{tab:dream}
707: and $\gamma_0=0.5$.
708:
709: \bigskip
710: \noi Figure \ref{fig:volatility} The average, $\mu_r$, and deviation,
711: $\sigma^2_r$ of the normalized real growth rate, $r(t)$ (see eq.
712: \ref{eq:r}) as a function of the tax rate, $\tau$, when the other parameters
713: are the same as those used in the Horror scenario.
714:
715: \bigskip
716: \noi Figure \ref{fig:regret} The critical regret as a function of time
717: for different values of the tax rate, $\tau$, when the other parameters are
718: the same as those used in the Horror scenario.
719:
720: \bigskip
721:
722:
723:
724: \newpage
725: \begin{figure}[htbp]
726: \begin{center}
727: \includegraphics[height=0.3\textheight,keepaspectratio]{fig_1a.eps}
728: (a)
729: \includegraphics[height=0.3\textheight,keepaspectratio]{fig_1b.eps}
730: (b)
731: \includegraphics[height=0.3\textheight,keepaspectratio]{fig_1c.eps}
732: (c)
733: \caption{\label{fig:traj}}
734: \end{center}
735: \end{figure}
736:
737: \newpage
738: \begin{figure}[htbp]
739: \begin{center}
740: \includegraphics[angle=90,width=\textwidth,keepaspectratio]{fig_3.eps}
741: \caption{\label{fig:scale}}
742: \end{center}
743: \end{figure}
744:
745: \newpage
746: \begin{figure}[htbp]
747: \begin{center}
748: \includegraphics[angle=90,width=\textwidth,keepaspectratio]{fig_4.eps}
749: \caption{\label{fig:phase}}
750: \end{center}
751: \end{figure}
752:
753: \newpage
754: \begin{figure}[htbp]
755: \begin{center}
756: \includegraphics[angle=90,width=\textwidth,keepaspectratio]{fig_2.eps}
757: \caption{\label{fig:chaos}}
758: \end{center}
759: \end{figure}
760:
761: \newpage
762: \begin{figure}[htbp]
763: \begin{center}
764: \includegraphics[angle=90,width=\textwidth,keepaspectratio]{fig_5.eps}
765: \caption{\label{fig:tax}}
766: \end{center}
767: \end{figure}
768:
769: \newpage
770: \begin{figure}[htbp]
771: \begin{center}
772: \includegraphics[angle=90,width=\textwidth,keepaspectratio]{fig_6.eps}
773: \caption{\label{fig:volatility}}
774: \end{center}
775: \end{figure}
776:
777: \newpage
778: \begin{figure}[htbp]
779: \begin{center}
780: \includegraphics[angle=90,width=\textwidth,keepaspectratio]{fig_7.eps}
781: \caption{\label{fig:regret}}
782: \end{center}
783: \end{figure}
784:
785:
786: \end{document}
787: