Expectations-based Processes – An Interventionist Account of Economic Practice
This paper is closed for comments.
Abstract
The paper starts distinguishing between two kinds of economic practice: theoretical practice (TP) (model and theory building) and direct economic practice (DEP) (the practical operation upon real economies). Most of the epistemological and philosophical considerations have been directed to the first type of practice, one of whose main goals is the discovery of particular sorts of economic laws, mechanisms and other regularities which throw light on relevant economic patterns. We do not deny that in some restricted domains these kinds of regularities may be found. Rather, we claim that (in many domains) the realm of economics is best understood as consisting of processes whose regular structure (if they have it at all) is not guaranteed beforehand but may be crucially influenced and successfully enforced by what we call DEP.
We claim that (a) some economic processes are a particular type of social processes that will be referred to as Expectations-Based Processes (EBP); (b) in those cases in which EBP exhibit a regular behavior, they depend on agents’ expectations and, crucially, we argue, on interventions upon them.
Characteristically, an EBP shows a connection between the information that individuals receive from the relevant economic context, the expectations they form, and the activities they perform. More importantly, authorities’ interventions may change agents’ expectations (and therefore, their decisions), contributing to shape EBP and helping to produce the patterns that lead to some targeted economic phenomena.
The features of an EBP show that they are not shielded from external influences and they do not run autonomously once triggered; on the contrary, they are processes that require continuous prodding on the part of policy makers to keep them running in the intended way. So they cannot be conceived neither as mechanisms nor as economic machines.
This interesting paper distinguishes between real economic processes (real markets and real economies) and their representations (theories and models). It also distinguishes between economists oriented towards “direct economic practice” (DEP) which consists in applying theoretical tools upon concrete socioeconomic systems while others are oriented towards theoretical practice (TP).
The paper focuses on DEP, the way in which given knowledge provided by theories and models, economists involved in key economic government positions contribute to influence or transform real economic processes. According to the authors, the two DEP and TP are closely related but “philosophical analysis” has focused on TP.
The authors claim that the realm of economics is best understood as consisting of processes whose regular structure (if they have it) is not guaranteed beforehand but may be crucially influenced and successfully enforced by DEP (direct economic practice). In other words, when these processes involve regularities, they depend on agents’ expectations and on interventions on them.
I have a few questions. The first question: economic theory assumes there is a particular dimension of social life that is amenable to the kind of ‘scientific’ analysis it carries out. Furthermore, economic theory assumes its task is to unravel the laws, regularities or internal dynamics that drive this field or dimension of social life. The authors, I think, will not have quarrel with this particular formulation (although perhaps we would need to discuss the semantics). This is what I call the primal assumption of economic theory: that economics has its own rationality, that of a system. Without this assumption, it is not possible to even attempt trying to identify and examine the mechanisms, laws, regularities and processes of economics.
So, as a point of clarification, what is the relation between mechanisms and processes, as defined in the paper, and this primitive assumption of rationality?
The authors mention there is widespread scepticism about the existence of universal laws. That may be so, but the primal assumption of a fundamental rationality in economics is a different thing. This is the way economic theory sees the world and thinks about itself. For example, if we examine Walras’s model, or an Arrow-Debreu model, or if we consider a Sraffian model of prices of production, we are dealing with attempts to unravel the laws that govern the economic world. Now, these models are used to describe the inner workings of the economic system. The architecture of each one of these models is different. For example, in Walras and Arrow-Debreu distribution and price-determination operate simultaneously and one can argue that the price formation (for example, the tâtonnement) process constitutes a mechanism in the sense of this paper. We return to this case below. Sraffa, on the other hand, reveals a different perspective in which distribution takes place independently of price determination and profits and wages are negatively related. But all of these models rely on the prior assumption that economics has a rationality of its own.
Now, if Sraffa builds a model in which prices of production are determined by technology as well as by the variables of distribution, is he describing a mechanism or a process? In fact there is no space for expectations here. The model is just an analytical tool. We do not need to talk about expectations here because this is not a process. And perhaps it does not make a lot of sense to talk about a mechanism because there is no sequence of events. The model just says that distribution is exogenous (a pretty big statement given the inclination of economic theory to insist on the autonomy of ‘economics’) and that it has a critical role in the determination of prices.
How do the authors deal with pure analytical models in economic theory?
Second question. The authors state that usually mainstream models assume all the ‘special conditions needed for the rise of convenient regularities’ and that the ex-ante produced regularities are attributed to targets of those representations. Here we find a first interesting problem. The authors should consider the case in which even when all the special conditions are assumed, the ‘convenient regularities’ fail to appear! This is not a weird exception. The model itself may be flawed.
Take for example general equilibrium theory (GET). Its main objective was to demonstrate that in a world of l consumers, m producers and n commodities, market forces lead the economy to a competitive equilibrium. But even after assuming all the special conditions needed general equilibrium models failed to show that in the general case the economy has the desired property of stability (that assures convergence to an equilibrium). In other words, even after assuming all the required assumptions the model failed to show the proper regularities. Thus, it appears that what the authors describe in this paper as theoretical practice is not as simple as it sounds. I think the authors need to integrate this type of difficulty in their analysis.
This is not the only example of failed models, but it is one of the most important. All policy recipes based on the efficiency of free markets are affected by this failure. There is no rational demonstration that market (competitive) forces lead in the general case to equilibrium and thus to some kind of efficiency. And this is regardless of what the expectations of agents may be. So it is important to understand the position of the authors with respect to cases such as this one. The model is flawed and no matter what you attempt to do with your agents’ expectations (through some kind of direct economic ‘intervention’) the desired result will never be attained.
I think the history of neoliberal market-friendly reforms in Latin America provides an interesting case. The model that provides the rational foundation for these reforms is flawed: tâtonnement does not converge to equilibrium except in two highly unsatisfactory cases (gross substitution or the weak axiom of revealed preferences at the market level). But nevertheless, governments argued in favour of these reforms and tried to influence expectations by promising the results the model itself could not guarantee. This is a case of distorted policy-making. Again, the situation in the real world with respect to the distinction between mechanisms and processes (and the role of expectations) is more complex than what the paper assumes.
The third question centres on what the authors describe as the Keynes effect. In introducing this example the authors state that in their view “contrary to mainstream philosophy of economics’ approach, most of the regularities that exist in concrete systems are the product of continuous interventions upon peoples’ expectations, social institutions, etc.”
The authors describe the KE a sequence of events, starting with a positive change in the money supply (as a result of policy by the monetary authorities) leading to a reduction in the rate of interest, an increase in investment, employment and output. The authors then examine what they call the deviations from the normal process: they use the qualifying conditions mentioned by Keynes himself in his analysis of the effects of an increase in the quantity of the money supply on the rate of interest (page 173).
I have a problem with this description of the Keynes effect. For the record let me say that the Keynes’ effect involves a longer sequence of events and in fact is of a different nature because it starts in the labour market and not with the expansion of the money supply. I’m aware that some authors do refer to the example in this paper as the Keynes effect, but that’s inexact. The Keynes effect is an important component in the evolution of macroeconomic theory because it sets the stage for the recovery of Keynes’ thought by orthodoxy. Having shown how a capitalist economy can maintain socially unacceptable levels of unemployment, Keynes goes on to describe a case in which the economy does tend to move towards an equilibrium position with full employment.
The Keynes’ effect is described throughout several pages of the General Theory. I could find the following: 232, 253, 261, 263-264, 266-267, 298, 308 (all pages correspond to the MacMillan Cambridge University Press edition for the Royal Economic Society, Collected Writings, Vol III). The story goes as follows. Keynes criticizes the Classical view that unemployment can and is reabsorbed whenever there is flexibility of wages. His answer, in the general case, is in the negative. However, he finds there is a case in which the flexibility in wages can lead to a decline in the interest rate. This is an important case because it generates what can be called Classical results within a Keynesian structure.
How does this come about? In a case where unemployment exists, the Classics think that workers will compete for jobs and this will bring down wages. As wages fall, production costs will drop leading to an expansion in employment. For the Classics flexibility of wages is enough to ensure equilibrium in the labour market. Keynes thinks this is not enough because the ensuing fall in the price level will leave real wages unchanged. There is no adjustment in the labour market and unemployment persists.
But Keynes also presents a case in which unemployment is reabsorbed because of the impact of flexibility of wages on the interest rate. The reasoning assumes that with unemployment wages are bid down, marginal costs drop and output expands. Keynes thinks the extra output cannot be all sold because the marginal propensity to consume is less than 1. Thus, there will be accumulation of inventories and this will lead to price reductions. The change in the price level will lower the demand for active balances, causing the demand for money function to shift and creating an excess supply of money at the prevailing rate of interest. This results in a corresponding excess demand for bonds, with the result that bond prices will increase causing the interest rate to fall (at least until the excess supply of money is channelled into speculative or idle balances). Because the interest rate is a key variable determining investment, the lower rate of interest will encourage higher levels of investment (and of aggregate demand). This leads to higher levels of output and to the elimination of involuntary unemployment.
Thus, when effects on the money market are taken into effect, Keynes’ analysis includes a case in which capitalist economies exhibit an inclination to move towards full employment. Keynes then posed the question of whether following a policy of wage cuts was the best policy to achieve full employment (GT, pages 266-7): instead, Keynes ventured his ironic comment, that we should have monetary management by the trade unions aimed at full employment instead of by the banking system.
The description of the Keynes’ effect in the paper distorts the nature of the process described by Keynes. The Keynes’ effect that we find in the General Theory starts with an endogenous process (wages are reduced as a result of what goes on in the labour market) and continues with the logical consequence in the realm of prices and then to the money supply, the demand for bonds and the drop in the interest rate. The Keynes effect described in the paper under review starts with an exogenous intervention.
What is the relevance of the ‘exceptions’ and ‘deviations’ to the analysis by Ivarola et al? This is perhaps the most important question. According to the authors “these deviations have their origin in the information obtained from the context, which significantly influences agents’ expectations”. Indeed, Keynes examines the different scenarios that unfold when agents have different expectations with respect to the drop in wages (page 263).
I think a clarification is required here. The paper appears to be saying that expectations are analogous to ‘exogenous disturbances’, not part of the core mechanism. But I think there is room for disagreement here because expectations are part and parcel of the components of the mechanisms or processes in many theoretical models (certainly this is the case of the family of models where rational expectations play a role). If expectations are core components of these models, what’s left of the analysis where policy interventions need to intervene on expectations in order to achieve the regularities promised by the model?
We appreciate very much Alejandro Nadal’s comments about our paper. They are sharp and point to very interesting questions. Not all of them however are strictly related to the main claims of our article. In what follows we hope to clarify our main arguments and dissipate some of Nadal’s worries.
First question. What is the relation between mechanisms and processes, as defined in the paper, and this primitive assumption of rationality?
We claim that most of the theoretical practice of economics (TP) takes as granted the following two assumptions: (1) economic regularities, mechanisms or laws exists by themselves out there, (2) they can be known by thought and empirical research. Nadal seems to have in mind both of them when he says that “economic theory assumes its task is to unravel the laws, regularities”, in order to “identify and examine the mechanisms, laws, regularities and processes of economics”. We think hat they are highly implausible in the case of economics (or social sciences), where individual decisions help to shape economic and social patterns. Anyway, we do not need to hold any definite position about this issue, even less to reject altogether the existence of laws or mechanisms. To develop our arguments we may endorse the weaker view that laws and mechanisms, even if they possibly exist, do not prevail in economic affairs. What we think permeates economics is open ended processes based on agent’s expectations. And we advocate for redirect our philosophical attention to this kind of processes.
So a direct answer to Nadal would be to say that we suspect that laws and mechanisms are wanted in the realm of economics (and cannot be discovered or known) but our claim is in favor of the existence of other kind of entity (processes). The existence of mechanisms could be in fact compatible with the existence of processes. Mainstream economics as well as mainstream philosophy of economics put emphasis on the first kind of entities (here we find “the primal assumption of economic theory”, we on the second.
Some additional comments may be of help to clarify the arguments of the paper.
Taking into account Nadal’s comments it may be of some help to distinguish between two types of PT: the one that he calls “analytical models” (which look for laws and automatic mechanisms in economics), and that which assumes strong uncertainty of the Keynesian sort. Our illustration belongs to this second kind of theoretical practice. The paper argues that the usual assumptions of conventional economics cannot be sustained when uncertainty prevails. As long as we think that uncertainty is a usual state of affairs in economics, a preference for the second kind of PT is entailed in our paper.
Anyway, it must be clear that we are contesting not economic theory itself (not even in its “analytical” form), but its philosophical and epistemological commitments. Our paper argues against mainstream philosophy of economics (rather than mainstream economics), that takes as granted these same assumptions and presumes that the only (or the main) task of economic theories and models is to deal with economic laws, mechanisms and strict regularities. We propose that philosophical attention should be given to the contribution of the second kind of economic theoretical practice (where the actual state of expectations really matters), and to the kind of “objects” it studies: processes.
More importantly, as Nadal notes, we also call to pay attention to another practice (DEP) and propose to subject it to philosophical consideration. These no strictly theoretical aspects of economics have been largely ignored by mainstream philosophy. To articulate our discourse we distinguished between Mechanisms (or, by the way, laws or strict regularities of any kind) and Processes. Mechanisms are regular sequences of entities related in stable manner which brings about regular sequences of events. Crucially, they keep running by themselves once started, so to speak. As an alternative to this picture, we propose to conceive economics as a set of processes (i.e., sequences of events whose precise direction neither are known nor guaranteed ex – ante). Processes are crucially shaped and made by agent’s decisions and external interventions. In our paper we focused on a particular type of external interventions (DEP), but a broader view, that admits the active participation of different kinds of groups, will be more realistic (and perhaps invested of even more political interest).
Regarding rationality, even if it was not a topic of our article, we think that standard economic rationality (expected utility) is conceived in an ex – ante way: before deciding one agent has to set his preferences and probabilities over lotteries and then calculate. The same idea is applicable to people or groups intending to (rationally) intervene in the economy: they have to know before hand which are the options and which of them is the best. Besides, if they are policy makers they have to be sure that the commended policies will attain the expected goals. However, if economics are mainly made of processes in our sense these sort of calculus and privileged knowledge are not possible, and rationality has to be characterized in other ways. This is why we introduce other kinds of skills and knowledge which are decisive in order to achieve actor’s (interventionist’s) goals. In short, we encouraged the interested people to develop a kind of practical rationality, which can not be based on the recognition of the best alternative before action is to be taken. Regarding uncertain processes (where decision makers matters) rational behavior cannot be known in advance. It has no a priori validation; decisions have to be validated on the field (there is no a-priori proof on his behalf). On this account rationality is related with the art of making things happens. In our paper we did not focused on rationality, but we agree it is an important point and deserves more elaboration. But maybe Nadal is predicating rationality not from individuals but from the system itself. A rational system would be that which is ruled by laws or mechanisms. Even it this is an acceptable way of speak we prefer to restrict rationality as a predicate of individual behavior.
A related question: How do the authors deal with pure analytical models in economic theory?
In fact theoretical economics has elaborated what Nadal calls “pure analytical models” designed to describe strong economic regularities (and where expectations play no role). Analytical tools like these ones are welcome. Theoretical economics needs a better use of mathematics, workable definitions, and analytical models. All this may be good and may be useful. Maybe they are of any help for some relevant purpose. But some justification of its epistemic value should be given by those who endorse and applaud this type of theoretical practice. Unfortunately, the following quotation shows that a sound philosophical justification is still wanted.
“Fact or fiction? Is economics a respectable and useful reality-oriented discipline or just an intellectual game that economists play in their sandbox filled with imaginary toy models? Opinions diverge radically on this issue, which is quite embarrassing from both the scientific and the political point of view”…….“Economics is a contested scientific discipline. Not only are its various theories and models and methods contested but, remarkably, what is contested is its status as a science. ….Suppose we take one of the characteristics of science to be the capability of delivering relevant and reliable information about the world. Suppose furthermore that this is not just a capability, but also a major goal and actual achievement of whatever deserves to be called by the name of “science.” How does economics do in this respect? This question is about as old as economics itself” (Mäki, U., (ed.), 2002, Fact and Fiction in Economics: Models, Realism, and Social Construction. Cambridge: Cambridge University Press, p. 3).
This failure of mainstream philosophical approaches makes us highly suspicious about the philosophical and epistemic foundations of this sort of theory. However, we are not rejecting a-priori this kind of theorizing and, more importantly, we do not need to reject it to advance our proposal of paying attention to EBP and DEP. As said earlier our view may coexist with analytical approaches to economics.
Second question. The case of “flawed models”.
Nadal correctly points out that carefully constructed models, which introduce convenient assumptions in order to reach a regular behavior on the part of the system may fail in this attempt (these are cases of unsuccessful or flawed models). In fact, TP may be “good” or “flawed”. In the first case intended regularities are obtained as the outcome of the model. In the latter, regularities do not occur even within the model! We do not see however why the existence of flawed models needs special accommodation within our position. Because one of the main claims in the paper is that self-sustained regularities (laws or mechanisms) are not found (or rarely found) in real economies, we think that the existence of flawed models do not compromise our general approach.
Maybe Nadal’s worries arise because he interprets that we suggest that given any theory any result can be obtained by means of carefully manipulation of expectations. But certainly this is not implied in our paper. This will be a case of rampant voluntarism, that we do not endorse. All we are claiming is that provided that “good” theory has identified different possible sequences of events (the ones that are feasible: that do not affront “physical” (material) limitations), the path that ultimately takes the economy is decided by external intervention (as was mentioned earlier the paper focused on government intervention, but other big economic actors also join the party).
Nadal claims, correctly we think, that, given that GET are flawed, “all policy recipes based on the efficiency of free markets are affected by this failure”, and he mentioned the case of “distorted policy-making” in Latin America. His claim seems to be that interventions based on bad theory cannot reach their desired goals. That is correct and we certainly agree with his remark. So said, it is important to see that we are not advocating for “neoliberal market-friendly reforms”. On the contrary, we are thinking in processes in which a heterodox policy based on fueling aggregate demand is being implemented in order to improve employment. This is why we illustrated our analysis with (a standard schema of) the so called Keynes Effect. In the case of Argentina the conventional economic view asserts that all the implemented policies designed to increase the income of the poorly remunerated people will be a failure. Conventional economists say so because they assume some kind of law or mechanism is well known in Economics according to which fueling the aggregate demand will result in inflation (not in more income and employment). Supposedly they know this ex – ante. Other heterodox economists, on the contrary, think that it is possible to reach an increase in employment starting with an increase in aggregate demand. Supposedly they have discovered an alternative mechanism opposed to the conventional one! We think that not automatic mechanism able to take us (necessarily) to a success or a failure is available. Political decisions will be crucial in the resulting events. Our paper is partially inspired in this controversy: we think that in the case of expectations based processes the final outcome is open (and there is room for a sound administration on the part of the government).
An additional remark is needed. As we explicitly say in the paper, we do not need to assume that KE is good economics in the sense that it portrays a “true” economic mechanism. We just claim that KE offers a good illustration of EBP, where its final outcome is not knowledgeable before hand: it has to be construed (administered). If the policies of the Argentinian Government were successful it will be because they managed to realize in concrete terms what was merely possible (an outcome that is actively sabotaged by many concentrated powers that are probably interested in coming back to “neoliberal market-friendly reforms”). In sum, we think that we have a substantial agreement with Nadal’s remarks on this point.
Third question: What the authors describe as the Keynes effect is inaccurate.
What is the relevance of the ‘exceptions’ and ‘deviations’ to the analysis by Ivarola et al?
Regarding our treatment of the Keyness Effect, we are basing our analysis in Snowdon, B., Vane, H., and Wynarczyk, P., eds, (1994), A Modern Guide to Macroeconomics, Edward Elgar, Great Britain. Certainly Nadal is right pointing out that in the General Theory a different or more complex story may be found. We did not put much emphasis on the true historical account because we recurred to KE as just an expository device. And this is the reason why in our text it was referred to as an illustration in order to develop our thesis. But the objection is welcome and some clarification has to be provided.
Nadal’s remark that “the paper appears to be saying that expectations are analogous to ‘exogenous disturbances’, not part of the core mechanism” deserves some comments. In fact, our view of KE assumes that there is no mechanism there, but an open ended process. He is right that speak of “deviations” of the main path is misleading. It is part of the mechanistic approach entailed in the usual approaches, but there is a dissonance with our view that claims that all the paths are in principle equally feasible. We agree that some clarifications should be provided on this point.
Expectations are certainly part of the process, but given uncertainty (in the radical sense used in some post-Keynesian approaches, like that of Davidson, Arestis or Lavoie), expectations are extremely volatile and this is why they can be exogenously influenced. Keynes agrees with this view. His reluctance about the success of interventionism is rather that it is difficult to see in advance how to intervene in a “scientific” form (that is, reaching a guaranteed result known before hand). But to believe that this kind of knowledge is attainable is a view shared by many (perhaps most) economists, a view that we contest in our paper. And this takes us back to one of the main tenets of the paper. In our view this kind of ex – ante (certain) knowledge is generally not attainable. But we do not need (nobody needs) to have such a privileged knowledge to intervene. Most of the time, Governments as well as different types of interest groups are intervening on expectations just because they do not know which the final result of the process will be (but in spite of this all of them try hard to push the economy in their preferred direction).
Ivarola, Marques and Weisman (henceforth IMW) present a methodological argument against the possibility of conceiving certain economic regularities as either “mechanisms” or “economic machines”. They claim that these regularities are better understood as “processes”, and that these processes must be distinguished from mechanisms and machines since they can be influenced by direct economic practice. In particular, those social processes that are of the expectation-based kind do not show the type of regularities one would expect from economic machines, since “the regularities that exist in concrete systems are the product of continuous interventions upon peoples’ expectations, social institutions, etc” (pag. 5).
IMW’s claim is mostly based on an example that is drawn from the history of economics, which they call the Keynes Effect. My comments concentrate on the example, which I do not find convincing.
As presented by IMW, the Keynes Effect is a generalised version of the principle of effective demand: an expansive monetary policy is supposed to induce an increase in income and employment through a transmission mechanism represented by movements in the rate of interest and reactions of consumers and investors to it. There is a typical chain of reaction, but since this depends on agents’ expectations there may be alternative courses of actions, and the actual result can be different from what one would expect: “deviations have their origins in the information obtained from the context, which significantly influences agent’s expectations” (pag. 8). So the Keynes Effect cannot be considered a machine since agents are involved: for instance, the link between a lowering of the interest rate and an increase in investment can be severed by agents who display a significant preference for liquidity. It is a process that, in order to work as expected, requires policy makers to exert control over agents’ expectations and their subsequent actions.
It is not clear to me why economic agents who perform activities, that is “reaction to the information they receive,” should not be considered part of a mechanism. Every representation of an economic system must account for the role of economic agents. Even if one uses a macroeconomic perspective which does not put emphasis on individual agents and markets, an assumption about expectations is needed in a theoretical explanation of an economic process.
So I find it difficult to understand the distinction IMW make (pag. 11-12) between “theoretical knowledge … needed in order to know which economic variables have to be manipulated” and “practical knowledge … needed in order to operate on expectations, so that agents’ activities are performed in the desired and expected way”. Couldn’t an economic theory of expectation formation be formulated at the theoretical level? Of course yes, I would argue. Apparently no, in IMW’s view (pag. 12): “in order to increase investment both the interest rate has to be lowered and entrepreneurs’ uncertainty about the future ought to be dissipated. Reducing the interest rate is a step that can be done in a rather direct way. However, dissipating the uncertainty is somehow more difficult to achieve, because it depends on a complex set of expectations. In particular, it presupposes a kind of knowledge that, properly speaking, is not scientific knowledge. On the contrary, it requires knowing how to manage peoples’ expectations.”
However, I do not know enough about the methodological literature on mechanism and processes to claim that IMW’s contribution is not methodologically significant. My point is that the example they use is not developed enough to make their argument convincing. Given that IMW concentrate on an example from Keynes, and attribute such an importance to the question of how to manage people’s expectations, one would expect that Keynes’s analysis of uncertainty and expectations were instrumental to make their point. But this is not so, and as a referee this is the major aspect I would like to see discussed in a revision of the paper. Either Keynes has no theory of uncertainty and expectation formation, something that is difficult to argue, or the way Keynes dealt with uncertainty and expectation formation must tell us something about whether the Keynes Effect is a machine or a process. My guess from part of the literature they list in the bibliography, but do not quote in the text, is that IMW thought about it. Most of the recent literature on behavioural aspects of the functioning of financial markets makes explicit reference to Keynesian insights. Behavioral explanations can be interpreted as an attempt to get the kind of (un-scientific?) knowledge that is needed to understand agents’ expectations and operate on them. This is an open, and controversial field of inquiry both theoretically and in historically oriented assessments (Skidelsky 2011 objects to Akerlof and Shiller’s 2009 claim that behavioural finance offers a formal representation of Keynesian animal spirits. On this point see Zappia 2012). But behavioural explanations can be of help to substantiate IMW’s claim that there is a practical knowledge which cannot be accounted for in the mechanics of the Keynes Effect. However, though they quote some relevant literature in the references, they do not elaborate on this, something that makes their “case-study” incomplete.
To sum up: IMW’s contention is that processes of the expectation-based kind are “less automatic” than is supposed in mechanistic accounts, since “they require that interventions take place not only upon their starting conditions (some economic variables), but also in context, providing an informational frame that prompts people to form those expectations which enable authorities to reach their goals”. This conclusion sounds to me of limited significance – every description of an economic system necessarily has human agency as a crucial component of the mechanism: do we really need a distinction between mechanisms and processes based on this aspect? But I leave this question to methodologically oriented commentators. My main point is that if IMW’s claim is to be made through the example they choose, this example must be developed in a much wider detail.
References
Akerlof, G., and Shiller, R. J. (2009). Animal Spirits. Princeton: Princeton University Press.
Skidelsky, R. (2011). The relevance of Keynes. Cambridge Journal of Economics, 35: 1-13.
Zappia, C. (2012). “Re-reading Keynes after the crisis: probability and decision.” Mimeo, University of Siena. Available at SSRN: http://ssrn.com/abstract=2103094
Thanks to Carlo Zappia for his interest in our article and his comments. Following are his main remarks and our answers.
Zappia: It is not clear to me why economic agents who perform activities, that is “reaction to the information they receive,” should not be considered part of a mechanism. Every representation of an economic system must account for the role of economic agents. Even if one uses a macroeconomic perspective which does not put emphasis on individual agents and markets, an assumption about expectations is needed in a theoretical explanation of an economic process.
Certainly, expectations and individual behavior may be represented within theoretical models or mechanisms. In fact, Hedström and Swedberg shows different mechanisms that include agent’s participation. We distinguished between mechanisms (fixed sequences of events) and processes (open ended sequences that may be influenced by the outside). Of course one may model individual reactions to information as part of a mechanism or as part of a process. In the case of mechanisms the reaction to a signal of the environment (that may include the behavior of another people) is usually well defined and some way fixed. Think for instance the “thresholds mechanism of collective behavior” of Granovetter or the “dying seminar” of Schiller (another threshold mechanism). There individuals are “programmed” to react like robots. Theoretical models usually fix expectations because they seek stable sequences and well known outcomes. This way of representing individual behavior exists as Zappia notes. But in real economic situations individual behavior is usually more volatile and subject to external influences. As our commentator points out this happens when uncertainty of the Keynesian sort prevails. But in Keynes’ work uncertainty is only related to investment decisions. We think that in real life open ended behavior permeates all kind of decisions. Essentially, public attitudes to economic policy are heavily influenced by many different groups intending to make good business.
Zappia: So I find it difficult to understand the distinction IMW make (pag. 11-12) between “theoretical knowledge … needed in order to know which economic variables have to be manipulated” and “practical knowledge … needed in order to operate on expectations, so that agents’ activities are performed in the desired and expected way”. Couldn’t an economic theory of expectation formation be formulated at the theoretical level? Of course yes, I would argue. Apparently no, in IMW’s view (pag. 12): “in order to increase investment both the interest rate has to be lowered and entrepreneurs’ uncertainty about the future ought to be dissipated. Reducing the interest rate is a step that can be done in a rather direct way. However, dissipating the uncertainty is somehow more difficult to achieve, because it depends on a complex set of expectations. In particular, it presupposes a kind of knowledge that, properly speaking, is not scientific knowledge. On the contrary, it requires knowing how to manage peoples’ expectations.”
We do not deny that “an economic theory of expectation formation (can) be formulated at the theoretical level”. Again, in Hedström and Swedberg some of these theories are described. And in some cases these representations may be useful for rendering account of some social situations (like why people decide to attend a seminar or enter to a restaurant). But we are not focusing on these mechanisms. We claim that there are other kinds of sequences, called processes, where individual expectations are nor fixed in advance and may be influenced by the outside. In reference to these processes the role of theory is to identify the different paths the process may plausibly take. What theory may usefully contribute in these cases is not a stable and recurrent mechanism but a “tree of possible (plausible) outcomes”.
Our distinction between theoretical and practical knowledge is introduced in connection to this idea. We try to point out to the difference between knowing the set of feasible sequences (this is provided by theory) and knowing how to realize one of the sequences of the set (this task is accomplished by practical knowledge and extra theoretical skills). Looking into more familiar scenarios, it is the difference between having a good plan (ex – ante knowledge) and having the skills to make your plan working in practice.
Zappia: the example they use is not developed enough to make their argument convincing. Given that IMW concentrate on an example from Keynes, and attribute such an importance to the question of how to manage people’s expectations, one would expect that Keynes’s analysis of uncertainty and expectations were instrumental to make their point. But this is not so, and as a referee this is the major aspect I would like to see discussed in a revision of the paper.
As we say from the start our treatment of KE is just an illustration in order to articulate around it our main arguments that are mainly epistemological. In fact we do think “that Keynes’s analysis of uncertainty and expectations was instrumental to make our point”. We took from Keynes analysis just what we needed. We did not develop further our simplified schema because we didn’t need to do it for our purposes. We hoped our arguments were convincing on philosophical grounds and not based in an historical case. Of course how much attention should be devoted to illustrations is debatable and we concede that some readers will be more pleased if more history and empirical material were used along the text.
Zappia: Either Keynes has no theory of uncertainty and expectation formation, something that is difficult to argue, or the way Keynes dealt with uncertainty and expectation formation must tell us something about whether the Keynes Effect is a machine or a process.
We have trouble with this dilemma. It is unclear to us whether “Keynes has or not a theory of uncertainty and expectation formation”. Keynes use the notion of uncertainty, related to investment decisions (but using a concept and have a theory about it are different things), and we are inclined to say that he does not have a theory of expectation formation. But our differences on this point may be to a large extent terminological and are rather beside the point. More importantly, whether KE is a mechanism or a process is something that in our view has to be decided on philosophical grounds.
Zappia: My guess from part of the literature they list in the bibliography, but do not quote in the text, is that IMW thought about it. Most of the recent literature on behavioural aspects of the functioning of financial markets makes explicit reference to Keynesian insights. Behavioral explanations can be interpreted as an attempt to get the kind of (un-scientific?) knowledge that is needed to understand agents’ expectations and operate on them. This is an open, and controversial field of inquiry both theoretically and in historically oriented assessments (Skidelsky 2011 objects to Akerlof and Shiller’s 2009 claim that behavioural finance offers a formal representation of Keynesian animal spirits. On this point see Zappia 2012).
But behavioural explanations can be of help to substantiate IMW’s claim that there is a practical knowledge which cannot be accounted for in the mechanics of the Keynes Effect. However, though they quote some relevant literature in the references, they do not elaborate on this, something that makes their “case-study” incomplete.
We agree with Zappia on this point, provided we understand his main claim correctly. Some of behavioral analysis may be used to influence people’s decisions (framing them). There is a growing literature showing how to help people decide among different saving or retirement plans (Benartzy and Thaler , 2007), and the authors of the paper under discussion have written a couple of papers on this subject (see below). So, it is true that this sort of theoretical knowledge may be used (and in fact it is used) to change people’s minds in a desired way. Then there is no sharp division between theoretical and practical knowledge, and was not our intention to sustain a kind of demarcation line among them. We have to do justice to this point in our paper.
In spite of this we still think that most of the interventionist actions of external powers are based not on scientific (theoretic) knowledge. And, more importantly, having theoretical knowledge about how people make decisions is generally not enough: it is also necessary to apply this knowledge successfully on real scenarios where other external actors are trying to push the process in other directions (this requires practical or administrative knowledge).
In short: once the distinction between theoretical knowledge and practical skills is done, in the case of broad economic processes it is still required to have a good amount of practical skill to use successfully any piece of theoretical knowledge.
Zappia: To sum up: IMW’s contention is that processes of the expectation-based kind are “less automatic” than is supposed in mechanistic accounts, since “they require that interventions take place not only upon their starting conditions (some economic variables), but also in context, providing an informational frame that prompts people to form those expectations which enable authorities to reach their goals”. This conclusion sounds to me of limited significance – every description of an economic system necessarily has human agency as a crucial component of the mechanism: do we really need a distinction between mechanisms and processes based on this aspect?
Our account of economic processes describes them as open ended sequences and takes into account two types of human agency: that of the agents (who receive signals from the environment) and the external actors that intervenes on agent’s expectations at any time. We do not think that these properties may be found in “every description of an economic system”. For this reason we think that “a distinction between mechanisms and processes” may be helpful.
References
Hedström, P. and Swedberg, R. eds., 1998. Social Mechanisms. An Analytical Approach to Social Theory. Cambridge: Cambridge University Press.
Benartzy, Sh., and Thaler, R., (2007), Heuristics and Biases in Retirement Saving Behavior, Journal of Economic Perspectives—Volume 21, Number 3—Summer 2007—Pages 81–104
Leonardo Ivarola y Gustavo Marqués. “Behavioral Procedural Models – A Multipurpose Mechanistic Account”, The Journal of of Philosophical Economics, Volume V Issue 2 (Spring 2012).
Gustavo Marqués y Diego Weisman. Framing Effects, preferencias, e intervencionismo paternalista, en María Llairó de Monserrat y Priscila Palacio (comp), Buenos Aires : Universidad de Buenos Aires. Facultad de Ciencias Económicas, 2011.