University of Amsterdam! Faculty of Economics and Business! January 2018!
!
Economic Predicting in a Complex and
Uncertain World!
!
Abstract!
The financial crisis of 2008 has not been the ignitor for a shift in economic thinking. Rather, the Neo-Classical and Neo-Keynesian schools have remained dominant, including the belief that the future course of history is explainable. Contrary to this belief, I argue that predicting the significant events in economic life is impossible. This conclusion is based on the (increased) complexity of the economic world, the fact that the aggregate cannot be represented by the sum of its parts, and irreducible uncertainty. The latter, the core of this paper, argues that it is impossible to predict the future growth of knowledge, which entails that predicting is impossible - we cannot know what we will only know tomorrow. The section on irreducible uncertainty is based on the works of Hume, Keynes, and Popper. The solution to this unpredictable world, I believe, is to become more diverse and pluralistic, not only in the economic world, but especially in the profession itself. A single economic model or truth does not exist, therefore diversity of thought and adaptability is the path to growth of knowledge. It entails to not be reliant on the success or failure of the single, rather to be a synergy of the many. !
!
!
!
!
!
!
!
Written by! Elias Z. Asselbergs!!
Supervised by! Dr. Dirk Damsma!!
Statement of Originality !
This document is written by Elias Asselbergs who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Table of Contents!
1. Introduction 4
2. Foretelling the Future 5
1. Complexity 5
2. The Problem of the Aggregated Sum 7
3. Irreducible Uncertainty 9
4. Conclusion: The Unpredictable World 13
3. Coping with Unpredictability 15
1. Pluralism 15
2. Qualitative Terms 16
3. Piecemeal Engineering 17
4. Conclusion: Openness, Diversity and Pluralism 17
4. Conclusion 19 5. Limitations 21 6. Conclusion 22
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
1. Introduction!
! The financial crisis of 2008 proved once more the inability of economist to predict the future. Yet, unlike Keynesian Economics, which was a byproduct of the Great Depression of the 1930s, and Monetarism, which had stagflation of the 1970s as its emulsifier, the Great Recession of 2008 did not set about a similar shift in economic thinking (Skidelsky, 2018). Rather, we have sticked with the belief of the New-Classical and New-Keynesian economists that everything can be reduced into the smallest probabilities, thereby believing that the course of history is explainable and can be understood. We have seem to forgone the simple notion of Keynes' irreducible uncertainty, something that has been understood throughout human life, however it seems to be no longer acknowledged in today's field of economics. !
! The belief in this fundamental uncertainty is as old as human life, uncertainty can be viewed as one of the bases for religion and for many schools of thought, e.g. Stoicism, for which it was the central starting point. Similarly, it was at the core of the economic thinking for two of the most influential economist of the last century, namely for Keynes and Hayek. Yet, with the shift from the dominant Keynesian Economics to the current New-Classical and New-Keynesian schools of thought, economics subscribed to the 'ergodic axiom', which entails that the outcome of any future event is the statistical shadow of the past and present (Davidson, 2012). It means that we can know the probabilities of future events, the opposite of the formerly held belief that there are things unknown to us, things that are fundamentally uncertain or as Donald Rumsfeld would call them 'unknown unknowns' - Keynes' irreducible uncertainty. !
! Since the financial crisis, it would have been likely to revise this notion that there are no 'unknown unknowns' in economics, and indeed the expertise of economist was seriously put into question, but ten years later a definitive change failed to happen. Furthermore, the influence of economist over society has remained intact and vast in scale. The legitimacy for this influence is partially rooted in their ability to predict the future. Therefore, this paper will examine this 'root of prophecy' and discuss the validity of the overthrown notion of irreducible uncertainty. It will focus on investigating what uncertainty and complexity entail for the economy at large and thus the way
economist should study it. ! !
! This research will be a literature based study, focusing mainly on uncertainty and complexity within economics. It is divided into two chapters. The first chapter will discuss our ability to foretell the future, in which the section on Keynes' irreducible uncertainty is most fundamental. The second chapter will discuss and introduce approaches - without giving too much away - on how to deal with a world that is not always as predictable as we would like it to be. It will conclude by making the case in favour of a more open, diverse and pluralistic economic world, both for the economy as a whole, as well as for the field of profession. !
!
!
2. Foretelling the Future!
! This chapter, the very core of this thesis, will discuss the difficulty that we pose on ourselves by trying to predict. In fact, it will be argued that it is impossible to predict the significant events in economic life. Three main issues, namely complexity, the problem of the aggregated sum, and irreducible uncertainty will be discussed, which, when combined, entail that predicting is impossible. This chapter will start out with the complexity of economic life and especially the increase of it over the last decades. Next, it will examine the problem of the aggregated sum, explaining why the aggregate cannot be represented by summing up all its parts. Finally, it will discuss irreducible uncertainty, which both stems from the two previously mentioned sections and can be the answer to them - a vicious cycle of unpredictability. !
!
Complexity!
! This section will discuss the complexity of social life and more specifically on the spectacular increase of complexity over the last century. A theme that has been discussed widely among many schools of philosophy, yet for the Greek and Roman school of Stoicism, it was the central starting point. Rather than imagining an ideal society, stoicism tries to deal with the world as it is: complex, uncertain and asymmetric. They believed that the universe operates according to a web of cause and effect, which although they believed to be rational, is impossible to comprehend. The current revival of stoicism (The Economist, 2016), might have to do with the rapid increase of complexity in recent times. Whilst, stoicism is sometimes merely viewed as a guide on how to approach difficulty, this section will rather use it as the fundamental stepping stone for understanding the current increase in asymmetry and thus nonlinearity within social life. This section works towards its central point, which is that the winner-take-all society leads to a world that is increasingly complex to predict. It will do so by firstly explaining the winner-take-all effect, secondly by discussing the implications it has on network theory, which ultimately explains why the increased periods of stability and reduced volatility will make predicting the significant events even harder. !
! In 1968, the sociologist Robert K. Merton described the Matthew effect, which explains how, for example, eminent academics will get more credit than an unknown researcher, even though they produced similar work, or why 'the rich get richer and the poor get poorer'. Both phenomena are due to their respective cumulative (dis)advantage. Thirteen years later, in 1981, economist Sherwin Rosen described a similar tendency in his 'The Economics of Superstars', in which he started his paper with the following: 'The phenomenon of Superstars, wherein relatively small numbers of people earn enormous amounts of money and dominate the activities in which they engage, seems to be increasingly important in the modern world.' Both explained, gave examples and discussed different domains of a tendency that share a common denominator, namely the winner-take-all effect. Robert Frank and Philip Cook argue that we are moving towards
a state in which a few winners take very much, whilst the rest is left with very little, due to today's structure of markets and technologies (2010 [1995]). However the trend towards a winner-take-all society is unsurprising. Mainstream economic theories, for example in international trade, such as the Heckscher-Ohlin model, developed in the 1930s, and the specific factor model, a few decades later, both based on the Ricardian theory of comparative advantage, describe a world in which specific countries specialise in a certain industry. Indeed, it even promotes the idea of a certain dominant country (a winner) in a certain sector. These models are the basis of international trade and equally within the leading textbook on international economics by Krugman, Obstfeld and Melitz. Take for example the United States of America, which, due to globalisation, has been able to specialise into the creative part of business, e.g. Apple, for whom it is more profitable to engineer software than to actually manufacture it. It is seen as the big positive of globalisation, since due this specialisation the USA can increase its overall welfare even though its manufacturing jobs are reduced. Yet, this section is not about the possible income inequality that globalisation creates, it is rather about the increased difficulty for predicting that is created by a winner-take-all society. Sticking with an example from 'international economics', China can be the winner in the manufacturing of buttons, even though Thailand has lower marginal cost than China, this can be due to the previously described Matthew effect: It can simply be mere luck that China started producing buttons first, causing a too high entry barrier for Thailand to enter the market. 1
Clearly, who gets a lucky break is unpredictable, however its consequences are far from insignificant (Taleb, 2001). !
! Yet, the unpredictability of whom gets lucky is not the main issue, the real problem rather lies in the fragility that is created by these winner-take-all dynamics, since it enhances 'low probability high impact' events, or as Nassim Taleb would call them 'Black Swan events' (2007). The difficulty that increasing complexity poses can be understood using Network Theory. In short, a network is an assemblage of elements called nodes that are somehow connected to one another by a link, e.g. financial institutions, airports, social connections etc. Albert-Laszlo Barabási and Duncan Watts, front-runners in the theory of networks and nonlinear-dynamics, uncovered the following feature within networks: there is concentration among a few nodes that serve as central connections. Networks have an instinctive tendency to gather themselves around a vastly clustered architecture: a few nodes are heavily connected; others barely so (Barabási, 2002). The tricky part is that, due to these 'winner-take-all nodes', the overall network seems rather solid, since the overall diversity is decreased and the volatility reduced, the perfect example being the banking industry pre-2008 (and still today). However, nothing could be further from the truth. In fact a blow
More specifically, Thailand will not start producing buttons, due to the fact that China is already producing buttons at its
1
lowest possible average cost, which is below the average cost for a small quantity of buttons produced by Thailand. Yet Thailand's average cost is lower than that of China for an equal quantity, meaning Thai buttons would have been cheaper on the world market than Chinese buttons. However, the barrier to entry was created due to the winner-take-all effect of China in a globalised world, which could have been the mere result of luck.
to poorly connected nodes will do nearly no harm, whilst if it were to hit a major node (think Lehman Brothers), a whole network could crumble. Battiston et al. call this 'too-central-to-fail' rather than the well known 'too-big-to-fail' (2012). Furthermore, the decreased diversity that comes with a winner-take-all-society will lead to a system which is less adaptive to shocks, meaning its crises are of vaster scale. Ultimately, it means a small disruption can trigger a domino effect that multiplies through the network and pushes it into a crisis state (Battiston et al. 2016). Not knowing which impact will hit what node and thus what overall effect it will have, makes predicting next to impossible. !
! Economic fluctuations are highly nonlinear (Buiter, 2009; Taleb, 2007). It does not matter how often predictions are right, what matters is how large the cumulative errors are (Taleb, 2007). For example, in the last fifty years, the ten most extreme days in the financial markets represent half the returns (Taleb, 2007). This means that these nonlinear results dominate the markets. The now increasingly popular field of complexity theory, of which the earlier network theory is a part as well, tries to predict these nonlinear dynamics. However, as much as this fields adds to our understanding and acknowledgement of asymmetries, by definition predicting nonlinearities will remain impossible, as argued by the French mathematician Henri Poincaré. He explained that dealing with nonlinear dynamics means that changing the mechanics ever so slightly can lead to explosive errors, or forecasting difficulty, e.g. the famous example of a butterfly clapping its wings in India, causing a hurricane in New York. His mathematics showed that nonlinearities limits forecasting to a certain extent. Therefore, Poincaré believed that we should only work with qualitative terms - features can be discussed, but not computed. !
! All in all, it seems that if any gains are made in our ability to model and predict the world, they may be offset by the increase in its complexity, due to a greater and greater role for the unpredictable (Taleb, 2007). The winner-take-all society implies that there will be more periods of calm and stability, with most problems concentrated into a small number of Black Swans. By doing so it will create even more devastating Black Swans. This has been perfectly shown by the example of the financial crisis of 2008: financial institutions merged into a smaller number of vast banks, all of them interrelated, using the same risk measures and business models, thus when one fails, they all fail. In short, the increasingly interrelated, less diverse and winner-take-all society has made crises less likely, but when they occur they are more global in scale and of greater magnitude. Moreover, the rarer the event, the less we know about its odds. It means that we know less and less about the possibility of a crisis. So, increased complexity has made the world increasingly nonlinear, making it even more difficult to know what cause will lead to what effect. !
!
The Problem of the Aggregated Sum!
! In recent decades, modern economists have subscribed to the fact that microeconomics ought to be the basis for macroeconomics. It is the very basis for the widely used Dynamic
Stochastic General Equilibrium (DSGE) model, the economic model used to explain and predict aggregate economic phenomena, such as economic growth, boom and bust cycles, and the impact of fiscal or monetary policy. More precisely, it assumes that the aggregate level can be thought of as a single representative economic agent (Solow, 2003). Furthermore, this weight placed on micro foundations stems from the attack on the, until then dominant, Keynesian paradigm. Best captured by the so-called Lucas-critique, who himself explained it as follows (1976): 'Given that the structure of an econometric model consists of optimal decision rules of economic agents, and that optimal decision rules vary systematically with changes in the structure of series relevant to the decision maker, it follows that any change in policy will systematically alter the structure of econometric models'. This famous critique was the starting point for using micro as the basis for macro in the 1970s, remaining so ever since. Theoretically, micro founded models are not prone to the Lucas-critique, because they reason from micro foundations based on the preferences of the decision makers (Lucas, 1976). However, as helpful as it would be, to be able to assume that the aggregate macroeconomy behaves like a single representative agent, it does not seem to be the case, not only empirically but also theoretically it can be questioned. !
! Buiter argued, shortly after the financial crisis of 2008, that DSGE models are unable to describe the highly nonlinear dynamics of economic fluctuations, calling the training in them and other 'state-of-the-art' macroeconomic modelling 'a privately and socially costly waste of time and resources' (2009). Robert Solow, founder of the neo-classical Solow growth model, goes one step further calling his critiquing paper on DSGE models 'Dumb and Dumber in Macroeconomics'. The heavy nonlinearity and asymmetry of economic life has been discussed in the previous section on complexity and the next section on uncertainty, yet the empirical failing of DSGE and thus the failing of macro predictions based on microeconomics seems evident after the financial crisis. However, it should not have taken a financial crisis of the magnitude of 2008 to come to this insight, since in the early 1970s Sonnenschein, Mantel and Debreu (later recipient of the Swedish National Bank's Prize in Economic Science in Memory of Alfred Nobel, however not established by Alfred Nobel's will, unlike the others) introduced their theorem. They showed that although the Law of Demand applies for a single rational individual, it does not hold for multiple rational individuals with multiple commodities. Thereby, they proved that every polynomial is an excess demand function for a specified commodity in any 'n' commodity economy (Sonnenschein, 1972). This means that any continuous function (in this case an aggregated demand curve) can move in any form of direction: up, down, wobble, etc, which is the complete opposite of what would be found if the entire economy would be treated as if it were a single utility maximising agent (Keen, 2013). In short, they found that aggregating individuals who obey the Law of Demand leads to markets that do not - let alone entire economies. Therefore, they showed that macroeconomics cannot be derived from microeconomics. The fact that the aggregate is more than the mere sum of its parts is nothing particular to economics, it has been realised by many throughout history and holds for
natural sciences as well, as described by Physics Nobel Laureate Philip Anderson in More is
Different (1972): 'The behaviour of large and complex aggregates of elementary particles, it turns
out, is not to be understood in terms of a simple extrapolation of the properties of a few particles. Instead, at each level of complexity entirely new properties appear, and the understanding of the new behaviours requires research which I think is as fundamental in its nature as any other.' Even elementary atoms and particles, which are as identical as can be found, still behave differently when aggregated - let alone unidentical irrational humans.!
! Theoretically, macroeconomics can be logically derived from microeconomics only if micro principles aggregate seamlessly to the macro level or if the sum of the parts added up perfectly to the actual aggregate; mathematically if f(x) + f(y) = f(x+y). However this does not only seem to be disproven by recent empirics or by the sound mathematical reasoning underlying the Sonnenschein-Mantel-Debreu theorem, but it has also been debated by philosophers since Aristotle. He was the first to describe the emergence of properties in his Metaphysics as follows: '… the totality is not, as it were, a mere heap, but the whole is something besides the parts …'. This problem of the aggregated sum, as it will be called in this paper, has been discussed under various different names and different versions, all describing a similar issue, such as the theory of emergence, but also the doctrine of organic unity as dubbed by Moore and subscribed to by John Maynard Keynes. Skidelsky summarises Keynes' thought on this issue and adds, the later discussed, irreducible uncertainty, which could explain, at least partly, the problem of the aggregated sum. Being a strong believer of the doctrine of organic unity, Keynes believed that the macroeconomy is not equal to the sum of individual choices. Further, he argues that individual decisions can have effects greater or lesser than intended because of the reactions to other. Thus, imparting irreducible uncertainty to many outcomes of the individual agent, due to the interaction between them. So, simultaneously, the aggregate level generates irreducible uncertainty, as well as that irreducible uncertainty can be viewed as the reason for why the sum of the parts is unequal to the aggregated level. !
! Ultimately, there is a natural synergy or anti-synergy between parts, whether this is explained by increased irreducible uncertainty or by reflexivity or by any other explanation, the fact of the matter is that a small sum of the parts or one representative part is unable to explain its aggregate. This 'co-creation' by the parts leads to unpredictable aggregates, which do not lend themselves to be studied using applied micro. Therefore the macroeconomy either has to be studied on an aggregate level, which would entail more than a healthy dose of uncertainty, or this unpredictability of aggregate levels has to be accepted. !
!
Irreducible Uncertainty!
! This section on, as Keynes called it, 'irreducible uncertainty' is the most fundamental one for understanding the impossibility to predict the future. This section will outline the works of three
great thinkers, who devoted part of their efforts to this subject, namely David Hume, John Maynard Keynes and Karl Popper. Their insights will be supplemented by other more recent works and empirical studies. This section will follow a chronological structure, starting with Hume and ending with Popper. In the final paragraph, their insights will be bundled and will be corroborated by recent studies, before concluding with the hard to accept notion that foretelling the future is above anyones pay grade. !
! First up, the man most associated with the problem of induction, David Hume. He argued that all our beliefs about the unobserved are based on a general principle or presumption of uniformity (similar to what Popper called generalisation), meaning that it is believed that the future will ultimately resemble, or at least be conformable to, the past (Hume, 2007 [1748]). However, as Hume rigorously argued there does not seem to be any solid rational basis for believing that 'the present testimony of our senses' or 'the records of our memory' will indeed be a blueprint for what we will witness in the future. This is Hume's famous critique on induction, or what is commonly known as the problem of induction. Hume's central argument may be best summed up in the following quote from 'An Enquiry concerning Human Understanding': "When it is asked, 'What is the nature of all our reasonings concerning matter of fact [note: a posteriori]?' The proper answer seems to be, that they are founded on the relation of cause and effect. When again it is asked, 'What is the foundation of all our reasonings and conclusions concerning that relation?' It may be replied in one word, 'experience.' But if we still carry on our sifting humour, and ask, 'What is the foundation of all conclusions from experience?' The best expedient to prevent this confusion, is to be modest in our pretensions; and even to discover the difficulty ourselves before it is objected to us. By this means, we may make a kind of merit of our very ignorance." From this quote we can deduce Hume's three main ideas: Firstly all reasoning concerning matter of fact (the observable) are based on the relation of cause and effect. However, according to Hume, it is impossible to ever discover anything more than just one event following another, without being able to comprehend any form of 'power' between them, 'they seem conjoined, but never connected.' Secondly experience, or empirical observations, are the only thing acknowledging the relation between two events, or cause and effect. Yet, all inferences from experience presume that the future will resemble the past, meaning that if there is any change in the course of nature, the past will no longer layout the future, then all inference or conclusion from experience becomes useless. Finally, Hume concludes that all inferences from experiences are due to custom, not to reasoning. Thus, asserting that the only option is to be modest and be aware that we are ignorant about the course of future events. Hume's arguments have a rather clear consequence: We cannot determine the future, by looking at the past and present. Hence predicting is impossible. !
! Second in, the man who introduced the term 'irreducible uncertainty', John Maynard Keynes. His use of 'uncertainty' does not merely distinguish between what is known for certain from what is only a possibility (2013 [1921]). Casino games, like roulette or blackjack, are excellent
examples to distinguish these two main types of uncertainty for Keynes. On the one hand, roulette deals with calculable risks, for which the odds and probabilities are known. Therefore, they are not subject to irreducible uncertainty. Rather, irreducible uncertainty for Keynes was that in which the prospect of a European war is uncertain, or the price of a barrel of oil and the interest rate in twenty years from now is uncertain, or maybe even more uncertain: the possibility of a new invention, or the position of wealth-owners in the social system in 2050. According to Keynes, there is no basis on which to establish any calculable probability for these instances, meaning we cannot reduce this uncertainty to any probability, therefore it is irreducibly uncertain. One of the main points of Keynes' 'Treatise on Probability' (1921) was his rejection of the frequency or statistical theory of probability, which he replaced by logical theory, in which probability is a function of propositions, not frequencies. To put this into context, the famous Efficient Market Hypothesis assumes that financial markets are similar to insurance markets, meaning EMH believes that the risks can be precisely calculated. Yet, this is to mistake the world in which irreducible uncertainty is the dominant factor (financial markets) and the world in which all risks are known and thus possible to reduce to calculable probabilities (Casinos and life insurance markets, note: not insurance markets, that start to give out insurance for financial markets, like AIG that needed a bail-out of 85 billion dollars by the US government, after it sold insurances against the crash of CDO's). Irreducible uncertainty is central to understanding Keynes' economics, it is, for example, the reason why Say's Law, in which there is a tight link between supply and demand, does not hold. Another example is its dominance in explaining business cycles: when confidence is high, the economy thrives, when it is low, it worsens (Keynes, 2016 [1936]). Note, uncertainty is only an issue in economics when our prosperity is conditional on a projection of the future. Crucially, under capitalism this will always be the case, meaning uncertainty is inescapable, in fact it is generated by the system itself, since it is based on the fact that rewards will not come now but in the future (Skidelsky, 2009). Furthermore, Keynes' irreducible uncertainty is not similar to irrationality, the basis of the increasingly popular behavioural economics, in fact Keynes treated individuals as rational (although in a different manner than new-classical economists would do today). Because, for Keynes evidence which refuted an initial belief did not mean it was irrational to hold this belief in the first place. Ultimately, it is closely related to Keynes' belief that luck plays a more significant role in success or failure than causes which we believed to be at play with hindsight. In more recent literature, such as 'Fooled by Randomness' by Nassim Taleb and 'The Tipping Point' by Malcolm Gladwell, similar emphasis is put on the role of luck in social life. To summarise in the words of the former United States Secretary of Defense, Donald Rumsfeld, it are the 'unknown unknowns' that make predicting impossible for Keynes. !
! Furthermore, Keynes subscribed to the difficulty that the earlier discussed problem of induction poses, which might be best exemplified by his stance on econometrics. He believed that no amount of data on past economic events can be a proxy for their true likelihood of occurring in
the future, simply, because no economic event ever runs twice (Keynes, 1939). Moreover, Keynes had three main points of critique on econometrics. Firstly, he argued that running a regression to find parameters, which are seen as constant, is fundamentally flawed, since he argued that there would be no reason why they could not be completely different each year, or month, or day - the relationship of economic events is non-homogenous throughout time. Secondly, he did not believe in 'via positiva', rather he believed in 'via negativa', since 'with a free hand to choose coefficients and time lags, one can, with enough industry, always cook a formula to fit moderately well a limited range of past facts. But what does this prove?'. Thirdly, he believed that there are certain important influences that cannot be captured in a quantitative form, similar to Einstein, who famously said 'not everything that counts can be counted, and not everything than can be counted counts'. Ultimately, he believed that econometrics could only be used for simple relationships, not for example to predict the dynamics of credit cycles. Keynes' view on the use of econometrics seems similar to the proposed piecemeal engineering of Popper, on which later more. However, both New Classical and New Keynesian economics have overthrown Keynes' work by accepting the 'ergodic' axiom, which holds that the outcome of any future event is the statistical shadow of the past and present (Skidelsky, 2009). This, in other words means that the spread of past events (or returns in the case of financial markets) gives us a range of possible scenarios for the future, known as the variance, a sort of maximum cap on uncertainty. Indeed, the mainstream economic thought of the past few decades holds the belief that all risks can be known and correctly calculated; irreducible uncertainty has been neglected all together. It is one thing to not be able to predict, but it is another thing to continuously - even when proven wrong on various occasions - hold the belief that foretelling the future is possible.!
! Thirdly, the man who made of fallibilism a constructive method, Karl Popper. His point is the clearest, yet most simple argument of why predicting significant events is impossible. His argument, which is set out in his 'The Poverty of Historicism' (1957) is best summarised by the following reasoning: 1. The course of human history is strongly influenced by the growth of human knowledge. 2. It is impossible to predict the future growth of our knowledge. Simply, assuming that there is growing human knowledge means that we cannot know today what we will only know tomorrow. Or explained in other words; No predictor (e.g. scientist, artificial intelligence etc.) can predict its own future outcomes. They can only do so after their prediction has turned into a retrodiction. 3. Therefore, it is impossible to predict the future course of history. 4. This means that a theoretical history (e.g. a historical social science that would correspond to theoretical physics) must be rejected. Ultimately, 'there can be no scientific theory of historical development serving as a basis for historical prediction.' His argument, in essence, is that to predict the future, we need to incorporate discoveries that will be made in the future. However, if we were to know about a future discovery, we would have already discovered them today. All in all, predicting future events, means knowing about discoveries that will be made in the future, which are fundamentally unpredictable -
we do not know today what we will know only tomorrow. So, it can be concluded that it is impossible to make predictions that are dependant on the growth of our knowledge. !
! All in all, Hume, Keynes and Popper, all subscribed to the idea that is known as the problem of induction. Whether we use Keynes' 'irreducible uncertainty', or Popper's 'we cannot know the knowledge of tomorrow', to explain our inability to predict, the fact of the matter is that there are elements in the future that are unknown to us, which make predicting impossible. In trying to illustrate Popper’s and Keynes line of reasoning we can look at the example of the oil market: Since we cannot know today what the range of innovations will be in the world of electricity, solar power, or anything that could substitute the use of oil or even disrupt the entire energy system, it is impossible to predict the price of oil for in 10 years from now. Not only are the outcomes of the various possibilities unknown, but also is the range of possibilities unknown, because, our imagination about future events stems from our present and past experiences, e.g. I cannot think of a colour which I have not experienced either intrinsically or externally. Ultimately, our past and present cannot be generalised to the future. Our inability to predict significant events has been shown as often as we have tried to make these grand predictions. Vast empirical studies done for instance by Tetlock and Makridakis, on the ability of political scientist and econometric models, respectively, to predict future events have corroborated this. Tetlock found a negative correlation between the predicted outcome and the actual outcome, meaning that the predicted outcomes are worse than random predictions, which would have given a correlation of zero (2005). Makridakis is the creator of the famous M-Competitions, in which various different forecasting methods are compared against each other. In his paper on the M3-Competition, the third and most recent competition in 1999, he and Hibbon concluded that 'statistically sophisticated or complex methods do not necessarily provide more accurate forecasts than simpler ones', a conclusion similar to the one reached by Christ a few decades earlier, when he compared 'naive models' to complex ones ex post (1951). In the spring of 2018 the results of the M4 will be published, in which 100000(!) different forecasting models have been tested. To compare in the year 2000 only/already 3000 models were included. However, I am not uncertain that their conclusion will be similar to the one in 2000. To conclude, in the words of Bertrand Russell: 'The demand for certainty is one which is natural to man, but is nevertheless an intellectual vice.' Ultimately, irreducible uncertainty will continue to make it impossible to predict. !
!
The Unpredictable World!
! This chapter has argued that making meaningful predictions about the future is rather difficult. The complexity of economic life and especially the increase of it in recent times makes predicting the significant events near impossible. On an individual or micro level, certain predictions are within our capabilities, yet when we try to predict on the aggregate level, things start to be come rather complex, especially due to the fact that the the aggregate is not the mere
sum of its parts. Both the increase of complexity and this problem of the aggregated sum result in a form of irreducible uncertainty, simply there are things we will never be able to predict or see coming. However the direction of the causality, does not have to be that complexity and the aggregated sum result in this irreducible uncertainty, rather irreducible uncertainty can also be seen as the explanation of both complexity and the problem of the aggregate sum. It is a vicious cycle. All three of the problems can set the whole cycle in motion and empower the others. E.g. irreducible uncertainty can be seen as the explanation of why the aggregate is not the sum of its parts, yet the aggregate level which is not the sum of its parts can also explain why there is irreducible uncertainty. Which causes what effect is pragmatically not all too important. The importance that this chapter tries to address is that economic life is unpredictable, whether in the recent increasingly complex times (which have made it even more difficult) or centuries ago, the fact of the matter is that irreducible uncertainty is one of the few certainties about life. We only know that there are things that we do not know. In this chapter only three explicit 'problems' have been described, yet far more accounts, with different starting points, could have been given, e.g. Kahneman's and Tversky's work on the psychology of predicting, which would all result in the same final conclusion: we cannot predict. !
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
3. Coping with Unpredictability!
! This chapter will introduce three possible approaches to deal with our inability to predict the future, as was argued in the previous chapter. It will give solutions that deal with this problem for the economy at large and thus for the way economist study it. Clearly, the impact that the introduction of these options would have on society are uncertain, the very point of this paper, yet all solutions are based on the fundamental premise that the world is unpredictable, in itself a significant shift away from the current belief that we can predict the future. The possible approaches that are given, are in no way the only options, there will probably exist an unlimited amount of possibilities. However, the few that have been chosen to be discussed, are here because they are approaches that were proposed by the earlier discussed thinkers, who accepted irreducible uncertainty, for example Popper's Piecemeal Engineering, and Keynes' and Poincaré's use of qualitative terms. Yet, the selection is of course limited and biased towards my own understanding of the world. Moreover, it is a sign of the times, in so far that the proposed solutions are chosen due to the current state of economics, because in the case of most proposed solutions the current state seems rather opposite from the herein discussed solutions, e.g. qualitative versus the current quantitative terms. !
! Three solutions will be shortly introduced, namely pluralism, qualitative terms, and piecemeal engineering. In doing so, this paper hopes to create a first foundation for further research and discussion, in which more of their possible opportunities and drawbacks could be researched. !
!
Pluralism!
! Accepting pluralism, that is accepting a diversity of views rather than one single approach, may be the most essential and fundamental stepping stone for acknowledging the fact that we cannot predict. As argued by Dani Rodrik in 'Economics Rules' a non pluralistic economic world would mean that one single model can be the ultimate model that could grasp everything (2015). However, this would defy the earlier presented logic of Popper as displayed in the section on irreducible uncertainty. It would namely entail that we know the ultimate path of history, meaning we would know today everything that we would know in the future. Yet, this is contradictory to any belief in progress of knowledge. Furthermore, if we would known everything today, then all innovations would happen today etc. Rather, a pluralistic view would entail that different assumptions, approaches and models could contain possible explanations. By accepting this notion, the foundation is laid for a world which opposes the notion that the world is full of certainties, e.g. the future is a statistical shadow of the past. Pluralism would give room to a world that searches for differences in beliefs, for differences in possible outcomes, for differences in thought. Pluralism should entail looking for discomfort, instead of comfort. Instead of looking for conformation, it should look for refutation. It would mean speaking with different minded people,
instead of similar minded ones. This should not only hold for the academic field of economics, but also for real life. Pluralism would foster diversity of thought. By increasing diversity, it would decrease fragility and increase adaptability to shocks and crises, a problem that has greatly increased by the winner-takes-all society, as explained in the section on complexity. Specifically for economics it would entail becoming less insular (Fourcade, Ollion & Algan, 2014), instead of being the field that quotes least from outside itself (Taleb, 2007). Ultimately, by accepting pluralism, we create an academic world that mirrors the actual world, namely full of diversity, full of subjective truths and full of downright uncertainty.!
!
Qualitative Terms!
! Ever since Paul Samuelson, economics has become more and more reliant on quantitative, instead of qualitative terms. Samuelson's 'Foundations of Economic Analysis', in which he argues that mathematics should be the language and thus foundation for economics, has set out the increasingly mathematical course of economics ever since its publication (1983 [1948]). He believed the use of quantitative terms would bring rigour and clarity to the field of economics. It was a significant shift away from the previously qualitative terms used by Keynes and other major economist before him. Although Samuelson was a vivid supporter of Keynes' economics, he seemingly had set aside Keynes' central argument of irreducible uncertainty, and thus for his disbelief in the use of overly precise models and mathematics in economics. The use of mathematics ought to have made economics clearer, however it has overemphasised its importance in the process, making it a sort of social mathematics game (Blaug, 1998; Friedman, 1999). As pointed out in the previous chapter, both Keynes and Poincaré, were proponents of the use of qualitative terms. Both believed that too complex mechanisms should only be described by the use of qualitative, instead of quantitive terms, since the latter would give a false sense of precision, as the actual outcome could deviate greatly. Furthermore, they believed that dynamics could be discussed, but not computed. For Keynes, irreducible uncertainty meant that modelling would be a useless exercise, because a model would give out a precise outcome, whereas the actual result would be highly uncertain. He believed it would be more useful to think about the possible direction that the effect could have, e.g. an increase in the money supply is likely to increase, rather than decrease the inflation rate, instead of giving a very precise prediction about the inflation rate after an 'x' increase in the money supply (Keynes, 2015). All in all, making precise predictions in a unpredictable world seems like a rather silly exercise. By using very precise quantitative terms and predictions, we construct the false impression that we understand complex things, creating an 'expert problem' and increasing fragility. To summarise with a quote often associated with Keynes, but coming from Carveth Read: 'It is better to be vaguely right than exactly wrong' (1898). !
Piecemeal Engineering!
! Accepting the notion that the path of history is unclear and uncertain, and that it is not predetermined, but rather has the ability to change, Popper concluded that we must reject any form of 'holistic, utopian engineering' (Popper, (2002) [1957]). Holistic, utopian engineering means to have the ability to change society in one single revolution, e.g. Mao's Great Leap Forward. Rather Popper believed in 'piecemeal engineering', a form in which changes are made piece by piece and the effects are being evaluated at every turn. Piecemeal engineering is fundamentally nothing more than remaining adaptable. It ask of us to not close ourselves off to truths we do not want to see, but rather accept that we will be wrong and adjust accordingly. It means building towards desired outcomes step by step instead of making giant leaps, of which the unforeseen consequences could be vast. It means evaluating at every step and readjusting if there are any undesired consequences. Ultimately, it means accepting irreducible uncertainty and thus to remain adaptable.!
! Piecemeal Engineering is not only an approach for society or the economy at large, but is also applicable for the way economist study it. For instance, it means to determine a domain, which is small enough to say meaningful things about, instead of studying the economy as a whole, which is too complex. Note, that there is a close resemblance between Popper's piecemeal engineering and Friedman's domains. Furthermore, piecemeal engineering calls for the ability of accepting to be wrong and adjusting accordingly. The whole notion of progress of scientific knowledge is based on being wrong and learning from mistakes, it is based on conjecture and refutation (Popper, 2014 [1963]). Piecemeal engineering means opening the door to irreducible uncertainty, instead of shutting it down by the belief that everything can be reduced to probabilities. It means to understand that the error rate is at least as important as the prediction itself. !
! All in all, piecemeal engineering is about remaining adaptable and learning from our mistakes. It is important to note that the aggregate is not the mere sum of its parts, so by adjusting piece for piece, or part for part, we do not necessarily get the desired outcome on the aggregate level, therefore the essence is to remain open to adjustment, open to adjusting our beliefs, that is the progress of knowledge. !
!
Openness, Diversity and Pluralism!
! This chapter has given an introduction into three approaches that help to deal with an unpredictable world, namely pluralism, that is accepting diversity of thought, secondly the use of qualitative terms, i.e. to discuss features, rather than compute them, and thirdly piecemeal engineering, meaning to go step by step and to learn from our mistakes. This chapter is intended to be a basis for further research and discussion on how to accept and deal with a unpredictable world. By no means are the given approaches the only ways to deal with unpredictability, neither have they been fully reviewed and have all their consequences been considered, as this is a mere
starting point. Yet, I believe that they are fundamental for a world that acknowledges irreducible uncertainty. Furthermore, they are sensible ideas that are reasonably close to the current state of affairs, nor do they entail passive nihilism, which should not be the conclusion drawn from unpredictability. With reasonably close it is meant that the ideas are nothing new and have been used in past and present to certain extends. For example, Rodrik argues that economics is (quite) pluralistic, however by being one of the most insular fields, that can hardly be the case, undoubtedly there is some pluralism, but there should really be an awful lot more of it, if it is to be seriously considered as the pluralistic field that it needs to be. Further, the overkill of mathematics and quantitative terms should be lessened, it needs to give room to 'the old' qualitative terms. Finally, piecemeal engineering is too a great extend applied in the sense that we tackle problems part by part, yet being open to being wrong, being adaptable, and learning from mistakes seems a lot more distant in the current world of economics. As mentioned previously, this paper and specifically this chapter is a sign of the times, in so far that it partially reacts to (and dismisses) the current state of economics. However, following Popper's line of reasoning, there is nothing created free from history, so in that sense it could be seen as an example of piecemeal engineering. Clearly, conducting economics is all a balancing act. !
! Ultimately, by accepting that the world is unpredictable, we would be almost there, meaning that by making this acknowledgement, we would have to discard the belief in an ultimate model (or truth) and thus accept openness, diversity and pluralism! !
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
4. Conclusion!
! Almost a decade has passed since the biggest economic crisis since the Great Depression of the 1930s. Many anticipated that the crash would be a shifting point for economic thinking, it looked to be the required ignition to revise certain assumptions that are underlying the dominant New-Classical and New-Keynesian schools of economics. Yet, a real shift has not occurred. The most important acknowledgement that I believe for economics to make is the acceptance of 'irreducible uncertainty'. Without this reestablished belief, economics will remain with the false hope that it can find a model that can explain 'everything', which, as has been argued, would violate the belief in any form of progress of knowledge and innovation. Therefore, I have tried to discuss what
uncertainty and complexity entail for the economy at large and thus the way economist should study it.!
! The fundamental thesis of this paper is that predicting the meaningful events of the future is a task too ambitious for mankind. It has done so on the following three grounds: Firstly, the complexity of economic life, and especially the increase of it over the last century, have created a world, which can be described as a winner-take-all society, causing an economic system that is less diversified, in which it is harder to predict which shock will have what impact. Secondly, the aggregate is not the mere sum of its parts, meaning that the bundling of individual agents will not add up to one representative agent. The aggregate will behave in a significantly different way than its individual parts, meaning micro can not be the basis for macro. Finally, these two problems are both explained by, as well as that they can be seen as cause for, irreducible uncertainty, or in different words the 'unknown unknowns'. Ultimately, the future has some fundamental uncertainties is store for us, which make the predicting of it impossible. !
! Clearly, there are many domains in which predicting is possible, such as various microeconomic related problems, e.g. the increasingly popular nudging that is being done. Yet, things like vast macroeconomic dynamics are hard to predict, just as the type of '10 year down the line' projections that are being done by financial economist. Moreover, the impossibility of predicting the significant events does not mean that all predictions are necessarily wrong, yet predicting is not about how often you can be right, but rather about your cumulative result, e.g. it is more hurtful to miss one significant event, like the financial crisis of 2008, than it is beneficial to have correctly predict the next interest rate hike for five times in a row.!
! Therefore, the second chapter of this paper has discussed how to deal with this irreducible uncertainty and thus the unpredictability of economic life. Rather than that the proposed approaches have been fully assessed, it serves a foundation for further research and discussion. Three specific approaches have been discussed, namely pluralism, the use of qualitative terms and piecemeal engineering. All of them share at least one fundamental belief, namely that of an uncertain world. Furthermore, it would be silly to summarise them individually, since they all point to the same direction: we need to create an open, diverse and pluralistic mindset, not only in the
'real' economy, but especially in the profession itself. Economics is unpredictable, it does not contain one ultimate truth, it can not be described by one single model. Rather, the very essence of economics should be its diversity, instead of closing itself of from other disciplines, it should open up, instead of trying to create a single dominant belief, it should be diverse in its thinking. This openness and diversity is the key to the progress of knowledge, it is how the natural conformation bias of men is being diminished. Instead of being a system that is reliant on one single stream of thought, which could singlehandedly lead us to the collapse of an entire system, it would be a system that relies on a multitude of thoughts, thereby becoming an adaptable, instead of fragile system. A system that is no longer reliant on the success or failure of the single, rather it will become a synergy of the many. !
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
5. Limitations!
! Clearly, our thoughts and beliefs are heavily biased, it makes any work limited to some extend, which especially holds true for this paper, the fact that there is no ultimate truth and any work is limited is the very reason why this paper has been advocating for diversity and pluralism so strongly.!
! Indisputable is the fact that I have not always done justice to the often (slightly) more nuanced position of the dominant economic schools. Furthermore, beliefs of the current dominant schools could have been discussed more in depth, to give a better overall picture of what is and what is not working. By doing this, the differences within the dominant schools would also have become more apparent than is currently portrayed. Similarly, I have been radical in saying that the world is unpredictable, again here should be room for nuances, clearly there are many situations which are rather predictable. Yet, the currently dominant belief that almost everything is predictable, instead of the belief that so much is not predictable, enhanced my 'radicalism'. Unfortunately, it has also meant that I did not focus on setting out more precise definitions to what is and what is not predictable. Although, I do not believe that very precise 'borders' exists, more could have been said about possible conditions and guidelines for what is predictable. !
! Another limitation can be found in the fact that I only discussed three themes which led me to the conclusion that we live in an unpredictable world. More accounts, with different initial starting points, could have been given, which would have all resulted in the similar conclusion, namely that we cannot predict. Equally, because the the answer to this unpredictable world is being a form of extended further research and discussion, it is far too brief to answer this enormously complex question. Yet, I do believe that the main point about openness, diversity and pluralism is central to its answer. !
! Hopefully, my master thesis will provide me with the opportunity to delve deeper into the approaches to an unpredictable world, and by doing so, hopefully some of the herein addressed limitations will thereby become less of an issue. !
!
!
!
!
!
!
!
!
!
!
!
References!
Anderson, P. W. (1972). More is different. Science, 177(4047), 393-396.! Aurelius, M. (2013 [c. 170 AD]). Meditations. Oxford: Oxford University Press.!
Barabesi, A. L. (2002). Linked: The New Science of Networks. New York: Perseus Books Group. ! Battiston, S., Puliga, M., Kaushik, R., Tasca, P., & Caldarelli, G. (2012). Debtrank: Too central to ! ! fail? financial networks, the fed and systemic risk. Scientific reports, 2, 541.!
Battiston, S., Farmer, J. D., Flache, A., Garlaschelli, D., Haldane, A. G., Heesterbeek, H., ... & ! ! Scheffer, M. (2016). Complexity theory and financial regulation. Science, 351(6275), !
! 818-819.!
Blaug, M. (1998). Disturbing currents in modern economics. Challenge, 41(3), 11-34.!
Buiter, W. H. (2009). The unfortunate uselessness of most state of the art academic monetary ! ! economics. Retrieved 17/12/17 from Financial Times: http://www.ft.com/maverecon/! ! 2009/03/the-unfortunate-uselessness-of-most-state-of-the-art-academic-monetary-!
! economics/?! !
Campbell, R. (1969 [c. 65 AD]). Seneca: letters from a Stoic. London: Penguin.!
Christ, C. (1951). A test of an econometric model for the United States, 1921-1947. Conference on
! business cycles, 35-135 !
Davidson, P. (2012). Is economics a science? Should economics be rigorous?. real-world !
! economics review, 59, 58-66.!! ! ! ! ! ! ! !
Frank, R. H., & Cook, P. J. (2010 [1995]). The winner-take-all society: Why the few at the top get
! so much more than the rest of us. New York: Random House.!
Friedman, M. (1953). Essays in positive economics. Chicago: University of Chicago Press.! Friedman, M. (1999). In Conversations With Leading Economist: Interpreting Modern ! !
! Macroeconomics, edited by B. Snowdon and H.R. Vane. Cheltenham: Edward Elgar Gladwell, M. (2006 [2000]). The tipping point: How little things can make a big difference. Boston: ! Little, Brown and Company. ! ! ! ! Hume, D. (2007 [1748]). An Enquiry concerning Human Understanding. Oxford: Oxford University
! Press.! ! ! ! ! ! ! ! ! ! ! !
Kahneman, D. (2011). Thinking, fast and slow. London: Macmillan.!! ! ! !
Keen, S. (2013). Discussing DSGE. Retrieved 12/11/17 from Youtube: https://www.youtube.com/
! watch?v=l1VS6u1V_Jo! ! ! ! ! ! ! ! !
Keynes, J. M. (1939). Official papers. The Economic Journal, 558-577.! ! ! ! Keynes, J. M. (2013 [1921]). A treatise on probability. New York: Courier Corporation.! !
Keynes, J. M. (2015). The Essential Keynes. London: Penguin. ! ! ! ! !
Keynes, J. M. (2016 [1936]). General theory of employment, interest and money. New Delhi: !
! Atlantic Publishers & Dist.! ! ! ! ! ! ! ! !
Lucas Jr, R. E. (1976). Econometric policy evaluation: A critique. Carnegie-Rochester conference
series on public policy, 19-46. North-Holland.! ! ! ! ! !
Makridakis, S., & Hibon, M. (2000). The M3-Competition: results, conclusions and implications. ! ! International journal of forecasting, 16(4), 451-476.! ! ! ! !
Manthorpe, R. (2016). Life Lessons From Ancient Greece: Stoicism's Revival. The Economist 1843 ! Magazine. Retrieved from: https://www.1843magazine.com/body-mind/life-lessons-from-!
! ancient-greece! ! ! ! ! ! ! ! ! !
Marion, F., Étienne, O., & Yann, A. (2014). The superiority of economists. MaxPo Discussion !
! Paper, 14(3).! ! ! ! ! ! ! ! ! ! !
Merton, R. K. (1968). The Matthew effect in science. Science, 159(3810), 56-63.! ! ! Poincaré, H. (2012 [1903] [1905] [1908]). Edited by Gould, J. The value of science: essential !
! writings of Henri Poincaré. New York: Modern library.! ! ! ! ! Popper, K. R. (2005 [1934]). The logic of scientific discovery. London: Routledge. ! ! Popper, K. R. (2002 [1957]). The poverty of historicism. London: Routledge.! ! ! Popper, K. R. (2014 [1963]). Conjectures and refutations: The growth of scientific knowledge. !
! London: Routledge. ! ! ! ! ! ! ! ! ! !
Rodrik, D. (2015). Economics rules: the rights and wrongs of the dismal science. New York: WW
! Norton & Company.! ! ! ! ! ! ! ! ! !
Rosen, S. (1981). The economics of superstars. The American economic review, 71(5), 845-858. Samuelson, P. A. (1983 [1948]). Foundations of economic analysis. Cambridge: Harvard University
! Press. !! ! ! ! ! ! ! ! ! ! !
Skidelsky, R. (2009). Keynes: the return of the master. London: Allen Lane.! ! ! Skidelsky, R. (2018). How Economics Survived the Economic Crisis. Retrieved 21/01/18 from ! ! Project Syndicate:
https://www.project-syndicate.org/commentary/why-no-intellectual-shift-! in-economics-by-robert-skidelsky-2018-01! ! ! ! ! ! !
Solow, R. M. (2003). Dumb and dumber in macroeconomics. Festschrift for Joe Stiglitz, 1-3.! Solow, R. M. (2010). Building a science of economics for the real world. House Committee on !
! Science and Technology Subcommittee on Investigations and Oversight, 20.!
Sonnenschein, H. (1972). Market excess demand functions. Econometrica: Journal of the !
! Econometric Society, 549-563. ! ! ! ! ! ! ! !
Taleb, N. N. (2001). Fooled by randomness: The hidden role of chance in life and in the markets.
! New York: Random House.! ! ! ! ! ! ! ! !
Taleb, N. N. (2007). The Black Swan: The Impact of the Highly Improbable. New York: Random !
! House.!! ! ! ! ! ! ! ! ! ! !
Tetlock, P. E. (2005). Expert political judgment: How good is it? How can we know?. Princeton: !
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, ! 185(4157), 1124-1131.!
