When an economic model fails, it is reality—and the people living in it—who pay the bills while the model lives on, unscathed.
This essay focuses on the global macroeconomic models produced by the IMF, the OECD, and the European Commission, and highlights reasons they are able to survive despite their failures. One key reason is the perpetuation of a belief system that authenticates intuitive, appealing policy narratives. The essay compares these theories to an alternative, demand-side model: the Global Policy Model used and developed by the United Nations. Views expressed do not necessarily represent those of the institution of affiliation. Comments by J. Capaldo, CP Chandrasekhar and P. Gehl Sampath are greatly appreciated. Usual disclaimer applies.
It is a truism of modern times that when an enterprise fails, it pays all it can and goes bankrupt. Large financial institutions are exceptions to this rule, but that is well known and widely critiqued. Less discussed is the fact that when an economic model fails, it is reality that pays the bills while the model remains unscathed—especially if it carries the stamp of an influential international organization (IO). There are a number of reasons why this problem has persisted for over half a century or more. Some are addressed in the first “Experts on Trial” collection from the Institute for New Economic Thinking (INET). This essay seeks to disentangle the theoretical justifications of the models themselves—and present an alternative.
Among those currently in use by IOs, the most influential are models designed for macroeconomic policy analysis at the global level. These global models include specifications of single-country models and further investigate outcomes or make policy recommendations after taking into account interactions between the world economy and its country-level parts. To be clear, this essay does not deal with forecasting models and trade models; it rather focuses on the global macro-models produced by the IMF, the OECD and the European Commission. When applicable, the propositions made about such models are contrasted with the Global Policy Model of the United Nations (UN GPM). The purpose of this essay is not to embark on a formalized, analytical review of the mentioned modelling frameworks and their roots in neoclassical economics; this is sufficiently covered elsewhere. Rather, the aim is to highlight aspects of the global models of the IOs that allow them to survive their own failures—in particular, the intuitive and appealing assumptions that veil their inadequacies, justify their simplifications and nullify objections.
A central proposition in this essay is that global models cannot be taken to represent objective and scientific tools for policy analysis. Clearly, all models have to make simplifications and in doing so they will fail to capture some dimensions of economic reality. Models are sets of mathematical equations consciously articulated and estimated econometrically; at times the critical coefficients of such equations are imposed (“parametrized” and “calibrated’). The decision about how best to represent economic relations is informed by assumptions and beliefs about policy choices and their limits. Unfortunately, the dominant models proposed by the mentioned IOs ignore essential features of the socio-economic system and therefore deliver a seriously distorted view of policy impacts. Most salient are their assumptions about economic growth, distribution, fiscal and monetary policy, and their failure to address problems of global aggregation—that is, of adding up variables for each world region to account for all relevant macroeconomic factors.
Economic growth in IOs’ models: A Mechanical Fantasy with Ideological Clout
A central preoccupation of policymakers is to ensure growth of GDP (domestic product) or GWP (world product), generally abbreviated as “output.” The term “output” has a clear connotation: It mechanically relates to “inputs,” the factors of production. Hence, most standard models, including certainly the IMF’s FSGM, the “Going for Growth” framework of the OECD, and the EC’s QUEST, are constructed around a so-called “aggregate production function.” This function defines total output mainly as a combination of two factors of production: capital and labour inputs, and their productivity.
From a modelling perspective, what matters is not the actual, measured amount of goods and services produced, but rather potential output, which denotes the ability of the economy (global or national) to produce enough to match expected demand. This is easily illustrated by referring to a situation upon which practically everyone seems to agree: the global financial crisis. When demand weakens, the capacity to produce will exceed actual demand. Households will hold money or pay out debts instead of spending, triggering a downward spiral. Facing a decline in sales, employers will seek to cut labour costs and postpone purchases of inputs and capital equipment. This in turn will lead to further weakening of demand in the aggregate, and the situation worsens. What is more, a slower pace or a contraction of investment in capital goods coupled with the gradual deskilling of workers who remain out of employment over time will contribute to an erosion of aggregate productivity growth. Back in 2008, there was broad agreement that such downward tendencies should be halted by policy stimuli.
The problem lies in deciding the degree, timing, and the nature of the stimuli. This is where model assumptions matter. Despite the seeming convergence of opinions at the peak of the crisis, the perspectives of economists on how to model production and demand are split into two camps. The models of the mentioned IOs assume that following a demand stimulus, if there is slack (spare capacity), production could initially rise, as the factors in place (labour and capital equipment) become more fully utilized. But after a while, as the stimuli are maintained or their lagging effects come through, capacity constraints will start to bite. Employers will increase demand for capital goods and labour, beyond what is actually available, which will lead to higher prices for them and eventually raise retail prices as well. The resulting inflation will erode purchasing power and demand will weaken in real terms, as incomes will buy less of the same goods. Unless that there is a meaningful increase in productivity—i.e., the total product that can be generated with the same amount of invested capital or labour—any excessive demand stimulus will not lead to a higher level of output, but rather about the same output at higher prices. This dynamic will in turn lead to a new wave of price instability, unemployment, and a whole range of undesired outcomes. The story is simple and almost convincing.
The problem is that while it is true that productivity cannot rise ad infinitum, we do not really know how productive an economy can potentially be; we only know the level of production actually observed. Moreover, taking capital and labour as inputs in a production function misses what is known as the “Solow residual” in the estimation of actual output. In the words of economist Moses Abramovitz (1956), the residual represents a “measure of our ignorance.” And this, “our ignorance,” is identified as “total factor productivity,” or the ability to generate more output with a given combination of inputs under changing technological and other conditions.
Since Abramovitz’s time, interpretations of that residual have evolved, with formulations of increasing degrees of sophistication that ultimately emphasise the role of “unmeasurable” factors on the supply side, thereby rescuing the validity of the above-mentioned production function.
But as ingenious as these models may seem, we cannot disregard their pernicious effects—particularly in their latter-day versions. With the help of mathematical algorithms, these models add dynamic features to their production function to trigger a turning point in labour markets where inflation starts accelerating. They also impose technological constraints that can only be lifted by assumptions, usually related to positive investment shocks under favourable business conditions (Andrle et al, 2015 p.12; Ratto et al, 2009, p. 223).
We should be clear about how this works. Granted, it takes a great deal of empirical and mathematical effort to build a logically consistent model, especially for the world economy. However, the crux of the matter rests on whether such mathematical formulations are viable representations of socio-economic realities. A big part of the success of IOs’ models consists in having reclaimed the vast territory lying between mathematical sophistications and socio-economic phenomena by deploying appealing narratives. Once the narratives are accepted, the role of the modellers is no more than changing a parameter, or imposing a “shock” on an equation that obtains a result that can be considered plausible in the sense of conforming to such narratives.
To illustrate how this works, it may be instructive to rehearse a typical syllogism that allows the modeller to impute technical progress in these kinds of models, based on aggregate production functions: Technology is knowledge incorporated in equipment. Knowledge is the result of investment in research. Investors will not increase research expenditure unless they have guaranties that allow them profits on the desired scale through various forms of investment protection. This calls for policies and regulation to ensure investor and patent protection, deregulation to remove “red tape,” and liberalization in the goods markets that promotes competition so that the best innovations are rewarded.
A complementary line of reasoning touches on budgetary decisions, such as tax exemptions and lower tax rates for investors, which are seen as sure ways to increase investment in new equipment, and therefore productivity. Likewise, fiscal discipline is treated as a necessary condition for improving the investment climate. With appropriate fiscal policies, productivity will increase in a business-friendly environment. These persuasive narratives offer the justification for changes to the parameters of the model that generate a scenario of faster growth of output in response to a “business-friendly” environment of profit protection, reduced bureaucracy and fiscal discipline. The model has the merit of providing mathematical precision to policy-makers who want to know how soon and to what extent they will see the benefits of their choices, but the model itself has been built based on an underlying narrative that is subject to debate.
Another critical limit to economic growth in the IOs’ models is wage inflation. To relax this constraint—that is, to allow model changes that would trigger a higher level of output and employment with less inflation—IOs put forward a persuasive change of policy regime: labour-market flexibility. It seems plausible—particularly from the perspective of an individual employer—that taking actions like removing rules governing wage setting or labour protection, lowering social security contributions, and shortening the duration of unemployment benefits, will increase the number of jobs on offer and induce more workers to accept them under the conditions set out by employers. Though this is not usually spelled out, there is a hidden expectation embedded in labour-supply functions that redirecting labour market policies towards skills training, education and promoting greater participation of women in the workforce will reduce wage pressures.
The appeal of these lines of reasoning seems indisputable. Variants of neoclassical economics taught in most universities worldwide offer the allure of apparent rigour by elaborating models that incorporate this intuitive content. Policy-makers can hardly resist embracing the simple intuition combined with quantitative sophistication the IOs’ models provide.
But for reasons detailed further below, outcomes are disappointing. Failure of these recipes for growth and high employment is the norm rather than the exception. How, then, do modellers insist on the same assumptions and prognostications in each successive period and maintain good reputations? Arguments advanced to explain failure are simple: “policy shifts were not strong enough,” “the sequencing was not right,” “more time is needed,” “other shocks disturbed the outcome,” “policy interferences did not allow markets to work properly,” and so on.
‘We demand aggregate demand’: The United Nations’ Global Policy Model (UN GPM) Alternative
In the other camp, which disregards entirely the aggregate production function, models are informed by theories of economic growth developed by economists like Roy Harrod, Evsey Domar, Petrus Johannes Verdoorn, Nicholas Kaldor and others belonging to the so-called post-Keynesian tradition. The United Nations uses an empirical global model along these lines to explore alternative policy scenarios. In essence, these alternative theories of growth hinge on two propositions. First, under the most general conditions, the main constraint faced by an economy (national, and especially global) is insufficient aggregate demand. This is observable in the resources—e.g., labour, capital goods, and knowledge—left unused or under-used across the socio-economic structure, and in the ability of modern financial systems to generate large amounts of credit to sustain investment. The low level of inflation in most economies during the last three decades is another indication. Second, as is evident from recent history, technical progress is usually a response to sustained increases in aggregate demand, especially if wage incomes are maintained over time and investment and innovation are supported by government programmes that promote research and the services and infrastructure required for widespread adoption of new products.
Here also, assumptions matter. The UN GPM model contemplates the risk that inflation could rise to levels that will eventually erode aggregate demand and cause instability. Demand could become stronger than the ability of an economy to produce at the desired pace. But this has not been the norm. Hence, productive potential is derived from a simple, path-dependent technical progress function (of the type outlined by P.J. Verdoorn, N. Kaldor and others) in which increases in current output adjust over time to aggregate demand pressures. Price pressures may arise from high labour demand and various categories of “cost-push.” The low inflation of recent decades justifies the expectation that growth of demand, productivity and output can proceed with little risk. Factors that could trigger unsustainable dynamics include excessive demand in commodity and energy markets, or balance-of-payments constraints that affect countries with limited external credit. Most critically, in the UN GPM, imbalances between countries in the absence of global policy coordination can add serious constraints to growth—but these are failures of political management, not constraints imposed by shortage of resources or inadequate technology per se.
The underlying narratives that allow a modeller using the UN GPM to generate sustained growth scenarios are counter to the advice of the IMF, the OECD and the EC. At their core, such narratives tend to stress that an effective role for government is the promotion of research, maintenance of social protection, income distribution, employment generation, and the coordination of demand stimuli at a global level. Far from acting as obstacles to rising aggregate productivity, these activities stimulate it. Related narratives include, for example, that research promoted by public universities and other public institutions has been, and should remain, at the core of major technological advances; research and development by private companies should be rewarded in ways that are conducive to its wider use and dissemination for future technological development, and regulated in accordance with social objectives; government expenditures on goods, services, infrastructure and social investment can effectively contribute to full employment; employment promotion policies and wage protection can help sustain stable growth of income and create incentives for adoption of new technologies; and patterns of global growth can be improved if imbalances are corrected by higher spending by surplus countries rather than reduced expenditure in deficit countries. The final section of this essay expands on some of these issues, after laying out how models from the IOs disentangle problems of distribution and macroeconomic policy.
Distribution: Heads Profits Win, Tails Wages Lose
Income distribution is an essential aspect of macro and global models. This is because unlike micro-economic models, where wage payments are a cost that erodes the potential for profit accumulation, a macro model explores the impact of wages and profits on the aggregate macroeconomic structure. Crucial among those are the effects of wages, which represent both costs for firms and incomes for households on the decisions of firms to increase productivity by investments in different sectors and the decisions of households to spend their wage income. Likewise, macro-economic models that contain a financial sector (surprisingly, few macro models do), help trace the impact of distributional changes—e.g., between profits and wages onto saving and portfolio decisions.
However, in empirical models, particularly for the world economy, distributional issues have been hard to crack because of conceptual and data issues. Early frameworks in the tradition of models from the IOs discussed here were based on a simple dichotomy in which profits and savings push up investment and wage growth constrains it. Recent versions mentioned above are more sophisticated. Regarding investment, while firms continue to operate as profit-maximizing units in which wage costs and interest payments impose a constraint, they can also take advantage of the assumed effects of technological change and of stock market valuations. While investment in these models is positively influenced by stock market valuations, with tight labour markets and high interest rates, stock markets may languish and investment will be less profitable.
Regarding consumption, these models specify two kinds of households. Liquidity-constrained households, which obtain income only from wages and public sector transfers, will likely increase consumption following wage increases and social policies, provided that inflation remains subdued. But more significant in the adjustment process are asset-rich households (also characterized as “overlapping generation” households), which will increase consumption in relation to the value of their accumulated financial wealth. Such wealth includes, most importantly, stock market assets.
There are therefore a multitude of forces at work: wage increases, price pressures, technology-rich investment decisions, savings, acquisition of financial assets, interest rates and asset prices. How these different forces interact is complex and generally involves considerable discretion and adaptability to specific circumstances. The general tenor, observed in detailed model results across many of the papers reviewed while preparing this essay, is to stress the advantages of promoting investments by keeping wage pressures low. Essential to these outcomes is a policy of maintaining low interest rates so long as (wage) inflation is tamed. Low interest rates benefit both investors and the bulk of consumers, who react positively to moderation of inflation and high asset prices. This narrative is fully consistent with the preceding discussion of the role of aggregate production functions.
In recent years, IOs have become increasingly aware of the limits to growth when the distribution of income worsens significantly. For example, the April 2017 issue of the World Economic Outlook (IMF, 2017a) devotes a chapter to analysing the dynamics of income distribution using a version of the IMF model presented above. The main findings strongly suggest that underlying a decline in labour’s share of income are the forces of technology and globalization. It points out that relative wage compression is significant in low- and middle-skill sectors, while in high-skill sectors wage incomes have grown faster than productivity. The central policy recommendation is instructive: Considering that technological advances and globalization are exogenous—i.e., beyond the control of policy makers—policy should be oriented to education, skills training and increasing labour force participation. That is, with the expectation of higher rewards for labour, it seems persuasive to aim at supporting labour-upgrading. Whether all this would lead to the increased availability of labour, labour market flexibility, the erosion of collective bargaining and reduced earnings is not spelled out.
Fiscal and monetary policy: the humbug of finance
The term “humbug of finance” (Patnaik, 2003, p. 226) captures how IOs see the role of fiscal policy, and by extension monetary policy, in their models. Despite the fact that here too the models of the IOs have increased in complexity and detail in considering fiscal issues, it is worth noting that the policy messages remain well in line with the well-known monetary approach to the balance of payments (MABP). The underlying MABP model, originally devised by Dutch economist J. Polak in the late 1950s, was the workhorse of the Bretton Woods institutions during the decades of “structural adjustment” imposed on developing countries, especially in the aftermath of the debt crisis that started with Mexico’s default in 1982. In this model, fiscal profligacy is the main explanation for balance-of-payments problems and uncontrolled inflation. As the fiscal deficit rises, the ensuing monetary expansion leads to inflation, making exports less competitive, raising imports, encouraging capital flight and worsening the external balance. There is no correction available to fix this “monetary” dynamic other than through drastic fiscal adjustment.
As explained by Patnaik (2003) and many other observers, the same view prevailed throughout the 20th century, and still permeates the latest incarnations of IMF models as well as those of the OECD and the EC considered here. A proactive fiscal policy and enhanced government expenditure are seen to offer benefits under some circumstances—as reflected in the emphasis in recent times on public investments in infrastructure (after all, good infrastructure increases the profitability of businesses at no direct cost for them). But fear of public sector debt is paramount. The models in question tend to establish an a priori ceiling for the public debt-to-GDP ratio, to which the government deficit is connected by stock-flow arithmetic.
The reader may be familiar with the traditional mainstream view that deficits crowd out private sector expenditure, at least to a certain extent. The argument assumes that debt financing leads to inflationary pressures and interest rate rises that adversely affect real consumption and investment. By this logic, each dollar of public expenditure sacrifices a defined amount of private expenditure. The net result of the public expenditure increase and the private sector contraction will be less than the dollar spent, while the increase in the debt will be one dollar.  Thus, while the deficit rises and accumulates to increase the debt stock, GDP will not increase as desired in the short run, and in the long run some adjustment will be required to contain a rising public debt and keep it below the stipulated target. Fiscal deficits must shrink. Often the recommendation is to maintain a balanced budget to “keep the house in order.”
Modern versions of these models invoke all sorts of devices to create this basic effect. Because of the implications for profitability, business promotion, and the like, it is generally recommended to keep up infrastructure spending, or alternatively, public sector financing of private sector ventures through “public-private partnerships.” Similarly, tax breaks and promotional transfers to business are assumed to have positive effects on the business climate and attract foreign investment. On the other hand, transfers to the poor in the form of subsidies and government current expenditure in various areas (often excluding the military) are suggested targets for downward adjustment. The government can reinforce fiscal consolidation by raising the rate of value added tax (also known as sales tax). These all are obviously regressive measures. But they are justified in the pursuit of financial stability and improving long-term growth prospects. They align with arguments centred on the aggregate production function, whose “total factor productivity” is presumably weakened by the mentioned redistributive fiscal policies.
In contrast to advice related to fiscal policy, it has become commonplace to espouse expansionary monetary policy. To stimulate demand, central banks can use a transparent, market-friendly policy rule that allows for lowering the interest rate in presence of sluggish demand. This will surely stimulate investment, it is assumed, and will not pander to vested interests of policy officials or overt corruption. Though the models of IOs allow for a rich interplay between policy choices, expansionary policy scenarios proposed by these models show a bias against fiscal relaxation in favour of monetary expansion.
In the aftermath of the global financial crisis of 2008-09, governments and central banks tried various combinations of fiscal and monetary policy responses. Monetary policy reactions came first, as they always do. Although policymakers remained fearful of the unambiguous position of IOs during the last many decades— condemning deficit spending—the gravity of the crisis was such that most countries resorted to fiscal stimuli. Despite the initial success of the relaxation in containing a freefall, the tables turned in less than two years. Public debt ratios were high, but the fact that these were for the most part induced by the crisis and emergency assistance programmes was disregarded. In most countries, with the exception of some of the larger emerging economies, the government brought on fiscal austerity, while major economies adopted more aggressive monetary policy in successively larger amounts in the form of quantitative easing.
In retrospect, it is possible to find policy communiqués that acknowledge that fiscal policy relaxation, when activated, works most effectively in the presence of monetary accommodation. But it is rare to find statements like “monetary policy accommodation works most effectively when accompanied by government deficit spending.” And yet, that should be obvious. William White, former chief economist of the Bank for International Settlements and keen observer of central bank policy, wondered: “It seems decidedly odd to be driving the economic car with one foot on the brake and one foot on the pedal” (White, 2012).
Therein lies the wisdom of policy-propositions in the models of dominant IOs: Policies that encourage private sector activity and risk-taking are naturally optimal, but policies that increase the size or the influence of the public sector amount to interference with the efficient market mechanism and should be avoided. Again, the narrative is powerfully intuitive: The private sector can take care of itself, since failures are punished with losses. This theory, a reassertion of the so-called Treasury View of the 1920s, was baptised the “Lawson doctrine” during the 1980s in the UK in honour of the then-Chancellor of the Exchequer. But experiences during the 1930s, the 1990s and most recently the global financial crisis of 2008-09, have demonstrated repeatedly that the most traumatic macro-financial crises occur in the wake of private sector excesses.
The Global Aggregation Problem: A Leap of Faith on a Set of Assumptions
A useful way to explore model properties is by assessing how a model resolves the many aggregation issues in the process of finding a global solution. In modelling terms, this is not as simple as it seems because the whole is not necessarily a sum of its parts. Modelling parts are not fixed bricks; rather, they represent organic components reacting to changes in other parts of the model. To complicate matters further, for similar reasons, aggregation over time is not always as simple as assuming that long-run outcomes are the cumulative result of short-run episodes.
Two examples could help highlight this issue. First, from the perspective of a single firm, it seems advantageous to increase activity when labour costs are reduced. Thus, if the model simply aggregates the expected results out of the profit-maximizing premises of all firms, one can assume that wage cuts will trigger greater activity and aggregate investment. But the additive properties of the model have to take into account whether other feedbacks—from wages to demand—matter, and by how much.
Second, if the public sector deficit rises, its immediate effect would be an increase in the stock of public debt. But if the effect of the deficit is to both accelerate GDP growth and increase government revenues, public debt could after one or two years shrink or rise less than the sum of deficits in the previous years. Coupled with a faster rise in GDP, the upshot could be, in fact, an effective fall of public debt-to-GDP ratios.
There are innumerable techniques that can help and are most often used. But as argued above, which aspects to include and which ones to exclude, as well as the way the relevant relations are specified, depend on the modeller’s assumptions. Furthermore, in testing a model, some degree of judgement is required. For example, the modeller’s assumptions and judgement will determine how strongly investment responds to unknown profit expectations, how consumers will react to lower wages, and how sensitive investors will be to regulatory changes. Time dynamics can also be complex.
Most models of the IOs reviewed here, as well as many other global and macroeconomic models, are estimated empirically using large amounts of data. Modellers aim at being “right,” at getting “sensible,” plausible results. But because many of the features of the models themselves defy measurement—e.g., such as factors like investor confidence, consumer expectations, desired wealth targets, and portfolio preferences—or because there is insufficient data for the required variables, key components of these models are “calibrated.” In other words, modellers derive key relationships from an assumed or prior economic effect that they consider crucial in representing reality, and adjust equation parameters to maintain consistency with those assumptions.
The most critical empirical relation in models reviewed in this essay is the aggregate production function. Essential for the generation of such a function is a measure of potential output that is consistent with the underlying assumptions: non-accelerating inflation ceilings, supply-driven technical progress, and the like. The output potential defined should, in turn, yield a measure of the “output gap.” That gap would indicate how much faster an economy can grow—say, with policy stimuli—without breaching technical limits, causing inflation, or triggering financial instability resulting from public deficits.
Considering the examples discussed above, it is clear that the estimated aggregate production function that corresponds with the underlying narrative should also be consistent with an investment function that responds positively to cuts in labour costs. On the other hand, the aggregate consumption of the two kinds of households presented above should respond in ways that do not prevent aggregate output from responding positively to wage moderation. (An alternative through export-led growth is briefly discussed below.) Thus, if wage cuts or wage decelerations adversely affect a portion of income of households and therefore their consumption, countervailing adjustments in other parts of the model will have to ensure that consumption is sustained, through for example increases in asset prices or rising wages in the high-skill sector. Bear in mind that with these premises, model parameters are calibrated—i.e., adjusted and tested to ensure that together they reflect the “right” properties.
Likewise, while a fiscal stance involving deficit spending may contribute in the short run to “closing the output gap,” there should be triggers in the model that prevent other bottlenecks from emerging. To increase product potential and avert an inflationary spiral, investment must rise and innovation must occur. But for that to happen, governments must adhere to fiscal discipline to foster a favourable investment climate. Thus, the model parameters are adjusted to ensure that fiscal multipliers are reduced as the model’s output gets close to the productive potential. In brief, at the core of these models are conditions relating to the supply side, while attention to demand-side features is restricted to the short term. The supply side is determined by the assumed characteristics of the production function while demand-side measures are seen as small corrections and stimuli which nevertheless carry huge potential risks, and therefore should only in exceptional circumstances be considered.
There are many other aspects of such global models—beyond the impact of distribution on investment, productivity and demand, or on fiscal multipliers—that need to be critically reviewed. An issue briefly mentioned above is whether wage remuneration and employment problems can be resolved by skills training and education. These are desirable measures per se, but they can only contribute to fair redistribution and move the system towards full employment when policies are oriented toward sustaining aggregate demand at the pace required to engage the full labour force.
Another aggregation problem that has significant weight in global models is the role assigned to structural reforms when global interactions are taken into account—for example, regarding the plausibility that improved competitiveness will promote growth through exports. As noted already, in an economic model driven by aggregate production functions, structural reforms that lower the cost of labour promote competitiveness in each country and may both attract investment and reduce the price of consumer goods. But it is unclear how the effects will add up for the world as a whole. In a different model, such as the UN GPM, structural reforms that depress wages and increase inequality in one country have negative repercussions in other countries that tend to reduce aggregate demand in the world as a whole (Capaldo and Izurieta; 2013, Kiefer and Rada, 2015; von Arnim et al, 2014). Under-consumption and excess financial saving can weaken demand and production growth in the long term. Budget cuts intended to reduce government debt present similar spillover issues, and are another potential source of low growth or stagnation in the global economy.
The Ultimate Role of a Model: Investigate Policy Options in the Real World
The same system of beliefs that authenticates policy narratives also influences a model’s design: its estimation of parameters and its very structure. Models, in turn, determine outcomes, their potential and constraints, and how these can be affected by policy action. Academics, policy makers, activists and citizens should be able to critically assess whether such narratives are convincing enough to justify affirmation of the policy choices made. Applying a critical lens is particularly important for models like those used by the IOs, the policy recommendations of which have been applied repeatedly—despite having failed so often to deliver the promised results. Other models have not enjoyed much support from policy makers, either because of the dominance of mainstream beliefs, or because policy makers with different viewpoints prefer to err with the herd rather than be right on their own.
Models are well-structured mathematical formulations that reflect a system of beliefs, and they allow the modeller to explore their implications. It is legitimate to dispute the results obtained if there are reasons to think that the beliefs are misplaced. At the end of the day, reality has to be the ultimate test, and models that continually fail to predict the most important phenomena or are unable to encompass alternative policy choices must be restructured, or fade away at last.
 Alex Izurieta is Senior Economist at the United Nations Conference on Trade and Development (UNCTAD). Views expressed do not necessarily represent those of the institution of affiliation. Comments by J. Capaldo, CP Chandrasekhar and P. Gehl Sampath are greatly appreciated. Usual disclaimer applies.
 For critical reviews, see for example Galbraith (2014), Kay (2014), Hudson (2015), Plender (2015) and Turner (2016), in the context of the global financial crisis; or Diaz-Alejandro (1985) for earlier crises.
 To briefly refer to these, the former are useful to project outcomes in the relatively immediate future. They do so by investigating statistical properties of economic variables within a global structure and building in advice from country-level specialists, but they take policies as given and assume “normal” patterns of other exogenous variables. Hence, their projected baseline scenarios reflect no unexpected changes ahead. Their forecasts can be way off the mark and have to be revised continuously. Similarly, trade models are useful to project changes in exports and imports at country and sectoral level, depending on hypothesized changes in tariffs and prices. They can be particularly useful to capture the effect of significant changes in tariffs and trade regimes, on a spectrum from highly protectionist to highly liberalized. But they are becoming increasingly irrelevant in modern times for two reasons. First, except in very exceptional cases trade tariffs in most countries are already very low. Hence, changes are so marginal that other effects not contemplated by such models tend to drive the outcomes, rather than tariff changes per se. Second, changes in trade regimes since the mid-1990s, especially after conclusion of the Uruguay Round and the agreement on Trade Related Intellectual Property Rights (TRIPs), have increasingly reached far beyond trade. They involve amendments affecting the financial sector, capital flows, investment protection, and even health and environment regulations. Trade models fail to capture these complex set of interactions because of simplistic assumptions about full employment and savings-investment and fiscal balance equilibrium conditions, and because they ignore the financial sector (see Capaldo and Izurieta, 2016).
 The three mentioned institutions use modelling frameworks that hook into the IMF’s Flexible System of Global Models (FSGM), which is derived from two earlier generation models of the IMF, the Global Economic Model (GEM) and the Global Integrated Monetary and Fiscal Model (GIMF); see Andrle et al (2015). The OECD assesses impacts of structural reforms and fiscal reforms on growth using the empirical framework called ‘Going for Growth’; see OECD (2015). The European commission uses the QUEST model; see Ratto et al (2009), or EC (2016) for a recent application. These models are typically policy oriented and their relevance can be explained by the fact that their analysis is nested into the economic policy evaluation of the G20. Other institutions, like the World Bank work with ‘forecasting’ models of global scope but conduct policy evaluations with country-specific models.
 Cripps and Izurieta (2014). This model is not to be confused with the IMF’s Global Projection Model, also often abbreviated GPM (see Carabenciov et al, 2013).
 To mention but a few: Dutt (1990), Felipe and McCombie (2013), Godley and Lavoie (2007), Storm and Naastepad (2012), Taylor (2004), Weeks (2012).
 Joan Robinson, a leading Cambridge (UK) economist, famously argued: ‘A map of the scale 1:1 is of no use to a traveller’ (Harcourt and Kerr, p.174)
 There remain a number of other fallacies in these models that are not discussed here for sake of brevity, such as the absence of financial sectors (or the unconvincing separation of finance from the real economy), the equilibrium properties of price changes, and even misconceptions about economic development.
 The debate between those two camps was coded as ‘the Cambridge controversies’ as it involved primarily the academic communities of Cambridge, UK versus Cambridge, MA. See, for example Cohen and Harcourt (2003), Harcourt (1969),Kregel (1971), and L. Pasinetti’s interview at INET (https://www.ineteconomics.org/perspectives/blog/economics-in-a-different-key)
 Despite their algebraic sophistications, though, it can be proved that ‘in reality they are just artifacts of algebraic identity accounting’ (Rada and Taylor, 2004, p.52)
 To be clear, there is a myriad of good reasons to promote education and gender equality, the point here is how to do that while also ensuring fair and decent pay and protecting labour rights; enlarging the reserve army of potentially unemployed workers is not one of them (see UNCTAD, 2017 forthcoming).
 Banner photographed at the ‘Occupy Wall Street Demonstration’ in October 2011 (Weeks, 2012).
 See http://debt-and-finance.unctad.org/Pages/GPM.aspx. The model is shared with partners like the International Labour Office (ILO) and the Global Development and Environment Institute (GDAE) at Tufts University, while an academic version, the Cambridge-Alphametrics Model, is maintained by researchers in various universities in Europe, South America and Asia.
 Verdoorn, J. P. “On the factors determining the growth of labor productivity.” Italian economic papers 2 (1949): 59-68.
 See McCombie et al (2002) for a comprehensive review of these mentioned authors and subsequent empirical essays.
 A richer representation of a technical progress function of this kind is proposed in Storm and Naastepad (2012, p. 57), estimated empirically for a number of OECD countries. It critically introduces a ‘wage-led’ factor which follows the intuition that in response to wage rises at par with productivity, employers devise new and more productive techniques, instead of dismissing workers (see also Galbraith, 2012, for a discussion of this feature in Scandinavian countries).
 See Izurieta and Singh (2010) for an exercise showing potentially unsustainable price dynamics in commodity and energy markets using an earlier version of the UN model.
 See Barbosa (2012), D’Arista (2007), McCombie and Thirlwall (2004), Moreno-Brid (1998).
 See UNCTAD (2013), Annex to chapter I for an empirical exercise.
 For a recent, alternative analysis of the relationship between skills, wages and growth, see Storm (2017).
 The author, in turn, refers to Joan Robinson’s exposition of JM Keynes’ critique of the dominant view in the UK during the 1920s, then called ‘Treasury View’. The central policy recommendation under this predicament was that the government should always balance expenditure with its income; else the public deficit will consume savings of the private sector which should otherwise be available for investment. The author explains this is a fallacy (the ‘humbug’ of finance) by appealing to the argument proposed by Richard Khan, pupil of JM Keynes, which stresses that the total pool of savings in an economy is not fixed but rather depends on income. As the fiscal deficit contributes to increasing income (unless the economy is already at full employment), then the volume of savings and hence investment need not to be affected.
 See, for example, Killick (1995). As a matter of fact, the ‘monetary approach to the balance of payments’ was usually named by its critics as ‘the fiscal approach to the balance of payments’.
 For a short while in the recent post-crisis, a number of empirical evaluations of fiscal multipliers at the IMF and other IOs recognized that government expenditure multipliers could be significantly greater than one (meaning that each dollar of extra fiscal spending leads to more than one dollar increase in GDP). See IMF (2012) and Coenen et al (2012). Following these IMF findings, Pettifor and Coe (2012) show that a fiscal reflation in the UK will pay for itself. However, the IMF position has departed from such assessments in subtle but unequivocal ways. In its WEO of April 2017 it accepts that policies to support demand should be implemented when feasible, but argues that fiscal measures should be supply-friendly (IMF, 2017). Likewise, IMF’s Fiscal Monitor of April 2017, after recognizing that the role of fiscal policy has been reassessed, argues that fiscal policy could achieve more efficient outcomes by adjusting expenditures and revenues in ways to avoid the threats of high public sector debts. In practice, and in line with the recommendations advanced in the WEO of October 2016, government deficits should be avoided and policy should focus on changing the composition of the budget (IMF, 2017b).
 See UNCTAD (2015, chapter II) for a brief historical account.
 See, for example Mohlmann and Suyker (2015). The authors replicate the modelling framework used by the IMF to demonstrate that during 2012-13 fiscal multipliers were greater than one and show that by 2015 multipliers should be smaller than one.
Abramovitz M (1956) Resource and output trends in the United States since 1870. American Economic Review, 46 (2): 5–23.
Andrle M et al (2015) The flexible system of global models - FSGM. IMF Working Paper WP/15/64, March.
Barbosa-Filho N (2012). Exchange rates, growth and inflation: What if the income elasticities of trade flows respond to relative prices? In: Deshpande A, ed. Capital Without Borders. Challenges to Development. Gurgaon (Haryana), Anthem Press.
Capaldo J and Izurieta A (2012) The imprudence of labour market flexibilization in a fiscally austere world, International Labour Review Vol 152, No 1, pp.1-26.
Capaldo J and Izurieta A (2016) Modeling the Trans-Pacific Partnership: Should policy-makers assume they operate in a perfect world? GDAE Globalization Commentaries, September (available at: http://www.ase.tufts.edu/gdae/Pubs/rp/GC103Sept16Capaldo.pdf)
Coenen G et al (2012) Effects of fiscal stimulus in structural models, American Economic Journal: Macroeconomics Vol 4 No 1, January, pp.22-68.
Cohen, A and Harcourt, GC (2003) What ever happened to the Cambridge capital theory controversies? Journal of Economic Perspectives, Vol 17, No 1, Winter, pp. 199-214.
Cripps F and Izurieta A (2014) The UN Global Policy Model: Technical description. Available at: http://unctad.org/en/PublicationsLibrary/tdr2014_GPM_TechnicalDescription.pdf.
D’Arista J (2007). US debt and global imbalances, PERI Working Paper Series No 136, May. Available at: http://scholarworks.umass.edu/cgi/viewcontent.cgi?article=1109&context=peri_workingpapers.
Diaz-Alejandro C (1985). Good-bye financial repression, hello financial crash. Journal of Development Economics, 19(1-2): 1–24.
Dutt A (1990) Growth, Distribution, and Uneven Development, Cambridge (UK) and New York: Cambridge University Press.
EC Directorate-General for Economic and Financial Affairs (2016) The economic impact of selected structural reform measures in Italy, France, Spain and Portugal. EU Institutional Paper 023, April.
Felipe J and McCombie JSL (2013) The Aggregate Production Function and the Measurement of Technical Change. ‘Not Even Wrong’, Cheltenham (UK) and Northampton (MA) Edward Elgar.
Galbraith JK (2012). Inequality and Instability: A Study of the World Economy Just Before the Great Crisis. Oxford and New York, Oxford University Press.
Galbraith JK (2014). The End of Normal: The Great Crisis and the Future of Growth. New York, NY, Simon and Schuster.
Godley W and Lavoie M (2007) Monetary Economics. An Integrated Approach to Credit, Money, Income, Production and Wealth, New York: Palgrave MacMillan.
Harcourt, GC (1969) Some Cambridge controversies in the theory of capital, Journal of Economic Literature, Vol 7, No 2, June, pp. 369-405.
Harcourt, GC and Kerr P (2009) Joan Robinson. New York and Basingstoke: MacMillan.
Hudson M (2015) Killing the Host: How Financial Parasites and Debt Destroy the Global Economy. Glashütte (Germany) and Baskerville (UK): ISLET-Verlag.
IMF (2012) World Economic Outlook, October 2012. Washington DC: International Monetary Fund.
IMF (2017a) World Economic Outlook, April 2017. Washington DC: International Monetary Fund.
IMF (2017b) Fiscal Monitor. Achieving More with Less. April. Washington DC: International Monetary Fund.
Izurieta A and Singh A (2010) Can fast growth in India and China help or harm US workers? Journal of Human Development and Capabilities, Vol 11, No 1, February, p 115-141.
Kay J (2015) Other People’s Money. Masters of the Universe or Servants of the People? London: Profile Books.
Kiefer D and Rada C (2015) Profit maximising goes global: the race to the bottom. Cambridge Journal of Economics Vol 39, No 5, pp. 1333-1350.
Killick T (1995) IMF Programmes in Developing Countries. Design and Impact. London: Routledge.
Kregel, J (1971) Rate of Profit, Distribution and Growth: Two Views. London and Basingstoke: MacMillan.
Lavoie M and Stockhammer E (2012) Wage-led growth: Concept, theory and Policies. ILO Conditions of Work and Employment Series No. 41.
McCombie JSL and Thirlwall (2004) Essays on Balance of Payments Constrained Growth. London: Routledge.
McCombie JSL, Pugno M and Soro B (2002) Introduction. In McCombie JSL, Pugno M and Soro B, eds Productivity Growth and Economic Performance. Essays on Verdoorn’s Law. New York: Palgrave MacMillan.
Moreno-Brid JC (1998). On capital flows and the balance-of-payments-constrained growth model. Journal of Post Keynesian Economics, 21(2): 283−298.
Mohlmann J and Suyker W (2015) An update of Blanchard’s and Leigh’s estimates in ‘Growth Forecast Errors and Fiscal Multipliers’, VOX CEPR’s Policy Portal, December 1.
OECD (2015) Economic Policy Reform: Going for Growth 2015. Available at http://www.oecd.org/eco/growth/going-for-growth-2015.htm.
Patnaik P (2003). The Retreat to Unfreedom. Essays on the Emerging World Order. New Delhi, Tulika Books.
Pettifor, A and Coe D (2012) The IMF and the end of austerity, PRIME Briefing, 24 October.
Plender J (2015). Capitalism: Money, Morals and Markets. London, Biteback Publishing.
Rada, C and Taylor L (2004) Empty sources of growth accounting, and empirical replacements à la Kaldor with some beef, EconomiA Selecta, Vol 5, No. 3, pp. 45-74, December.
Ratto M, Werner R and in ‘t Veld J (2009) Quest III: an estimated open-economy DSGE model of the Euro Area with fiscal and monetary policy. Economic Modelling, 26, pp. 222-233.
Storm S and Naastepad CWM (2012) Macroeconomics Beyond the NAIRU, Cambridge (MA) and London: Harvard University Press.
Taylor L (2004) Reconstructing Macroeconomics. Structuralist Proposals and Critiques of the Mainstream, Cambridge (MA) and London: Harvard University Press.
Turner A (2016). Between Debt and the Devil: Money, Credit, and Fixing Global Finance. Princeton, NJ, Princeton University Press.
UNCTAD (2013), Trade and Development Report 2013. Adjusting to the Changing Dynamics of the World Economy. New York and Geneva: United Nations Publications.
UNCTAD (2015), Trade and Development Report 2015. Making the International Financial Architecture Work for Development. New York and Geneva: United Nations Publications.
UNCTAD (2017-forthcoming), Trade and Development Report 2017. New York and Geneva: United Nations Publications.
von Arnim R, Tavani D and Carvalho L (2014) Redistribution in a neo-Kaleckian two-country model, Metroeconomica, Vol 65, No 3, January, pp. 430-459
Weeks J (2012) The Irreconcilable Inconsistencies of Neoclassical Macroeconomics. A False Paradigm, Oxon (UK) and New York: Routledge. White , W (2012) Overt Monetary Financing (OMF) and crisis management, Project Syndicate, June 12.