Akerlof 2020: Sins of Omission and the Practice of Economics

In a recent paper in the Journal of Economic Literature, George Akerlof describes how the reward structure of economics favors "hard" research over "soft" research. He argues that this leads to "sins of omission" as researchers ignore important topics that are difficult to approach in a "hard" way.

Akerlof pushes "for reexamination of current institutions for publication and promotion in economics" [and] greatly increased tolerance in norms for publication and promotion as one way of alleviating narrow methodological biases."

Overall I enjoyed the paper. I think it's important, especially in providing an understanding of how we got here. A couple notes are below. One important omission from the paper IMO is an examination of the role of network effects in maintaining the status quo. Without this, I find it even less likely that we'll overcome the coordination difficulties of moving to a more useful research agenda for the discipline. Also, the current mainstream benefits certain groups (such as people on the political right) who will resist a push for better questions and better methodologies, regardless of whether they provide a better understanding of our economy.

Three reasons for the bias toward hard research:

1. Place in the scientific hierarchy -- economists have physics envy
2. The evaluation process -- quantitative / hard papers are easier to judge than important ones
3. Selection into the profession -- mathematicians appreciate mathematicians

Three consequences of hardness bias:

1. Bias against new ideas -- old paradigms have tools that aid precision; also, this bias makes it harder to challenge existing paradigms
2. Overspecialization
3. The curse of the top five -- tenure committees give a lot of weight to the fop five journals

Examples of sins of omission:

1. Failure to predict the financial crisis
2. Motivations -- people act based on the stories they tell themselves rather than as rational deductive optimizers with perfect foresight
3. Four examples illustrating the unappreciated role of stories in economics:
1. The Soviet Union
2. Smoking and health
3. Global warming
4. Macroeconomics - Shiller's explanation for how the Great Depression unfolded

Additional notes

Akerlof mentions Kuhn's Scientific Revolutions (2012) which describes scientific progress as occurring when "normal science" uncovers "anomalies" within existing paradigms. Scientific revolutions that explain those contradictions lead the way to a better paradigm. But Akerlof argues that an issue with economics is that the paradigm is not just the subject matter but also the field's methodology. In other words, is a good economist someone who understands the economy or someone who understands linear algebra? The more powerful economists that think it's the latter (and I have a hunch it's a significant majority) the less chance economics has of progressing as Kuhn has described.

Lastly, Akerlof mentions that 'Colander and Klamer (1987, table 4 3, p. 100) thus found that only 3 percent of economists thought it "very important" for their success to "have a thorough knowledge of the economy"'. Three percent.

Gopinath on dollar dominance in international trade

In Chapter 2 of the Hoover Institution's Currencies, Capital, and Central Bank Balances, edited by John Cochrane, Kyle Palermo, and John Taylor, IMF Chief Economist Gita Gopinath "shows how private international financial intermediaries tend to focus on certain currencies, with the US dollar currently the dominant currency of choice. The dollar is often used for invoicing even when trade is between two non-US entities." A couple things that stuck out to me from her chapter, Dollar Dominance in Trade and Finance, are below.

Dollar dominance in trade, finance, and central bank reserves

1. Trade

The Mundell-Fleming economic paradigm states that the importance of a country's currency in international trade is tied closely to its share in world trade. This does not actually happen because the assumption that countries export goods in their own currencies is faulty. The dollar's share as an invoicing currency is 4.7 times its share in world imports and 3.1 times its share in world exports.

Even Japan and the UK invoice only 40 and 51 percent of exports in their own currencies. US invoices 93 percent of imports and 97 percent of exports in dollars.

2. Asset markets

Dollar liabilities of non-US banks are ~$10 trillion, roughly the same magnitude as dollar liabilities of US banks.

The vast majority of syndicated cross-border loans are made in dollars -- from a low of 61% in developed countries to 97% in Emerging Americas. Euro is second with 24% and 20% in Developed Countries and Emerging Europe.

Data source: BIS locational banking statistics

3. Central bank reserves

$10 trillion in official central bank reserves (2017Q4) -- dollar share is 63 percent, followed by euro at 20 percent.

[Update for 2019Q4: $6.7 trillion of dollar reserves out of $11.8 trillion total ($11.1 trillion allocated)]

Central banks keep dollars not just for trade but also so they can be the lender of last resort to their banking system.

Data source: IMF COFER (Currency composition of Official Foreign Exchange Reserves)

What makes a currency dominant?

In short, history and network effects.

1. History: How does a currency become dominant?

First, the historical evidence described by Eichengreen (2010). How to make a dominant currency

1. Encourage its use in invoicing and settling trade;
2. Encourage its use in private financial transactions;
3. Encourage its use by central banks and governments as a form in which to hold private reserves.

This process will lead to a high demand for the currency, which will decrease the interest rates (raise the price) on safe assets in that currency.

Once you get to preferred currency status, a feedback mechanism can keep it in place: "Why do exporters invoice in dollars? Because it is cheaper to finance in dollars. Why is it cheaper to finance in dollars? Because exporters invoice in dollars."

2. Network effects: How does a currency remain dominant?

Comments made in the discussion were insightful on this point:

Adrien Auclert: "In principle, going forward, we might see the equilibrium switch again, with the euro or the renminbi becoming the new dominant currency. But what this static model misses is that existing assets and liabilities have long maturities. So in a sense, the anchor of history is likely extremely strong--it would take a really long time for all assets and liabilities to be redenominated in any new currency, and the staggered nature of contracts makes such a coordination very large to imagine."

Robert Heller: "in Silicon Valley, we talk all day long about network effects. Isn't the dollar's dominance similar to network effects? The dollar almost took over the world, and it's very difficult for a second competitor to come up and to compete with the currency once it's dominant, just simply because of network effects."

There is interaction and reinforcement between low interest rates, dollar trade invoicing, dollar liabilities in global supply chains. So we have coordination difficulties, network effects, and lock-in by historical events. Dollar dominance seems established for the foreseeable future, but beware information cascades.

Legacy economists and penguins

Why have top economists not updated their views in light of evidence that their main macroeconomic models aren't working? In short: inflexibility and too much intellectual investment in the old system to change now.

The DSGE model is so attractive, as I said, to the neatness freak in many economists that it is very hard for anything else to get any traction, and I hope that we can change that.

- Robert Solow in 2010 Congressional testimony

Standardization

Legacy software denotes a system that has been superseded by better technology but is difficult to replace because of its wide use. Similarly, the relevance of mainstream / legacy macroeconomics to real world analyses has been supplanted by other schools of thought. However, as in the case with old IT systems, legacy macroeconomics’ prominence is the result of a type of market failure, in which inertia and path dependence have resulted in the persistence of an inefficient equilibrium.

The economic research agenda since the 1970’s has resulted in a relatively high degree of

standardization, as legacy economists have converged on “a largely shared vision of both fluctuations and of methodology.” (Blanchard, 2009) Standardization has many benefits. For example, it enables us to call or text anyone, anywhere, regardless of our service provider or phone brand. Paul Samuelson believed that mathematical foundations provide a “natural language” for economists; they contribute greatly to informed discussions about policies and trade-offs. But, as Joseph Farrell and Garth Saloner showed in a widely-cited paper, standardization benefits can trap an industry in an inferior standard when a better alternative becomes available. (Farrell & Saloner, 1985) For example, it doesn’t matter whether Google+ was a “better” social network than Facebook – Facebook’s millions of users gave it an unassailable position. If we each had the choice of only one social media platform, anyone who switched to Google+ would have had an account on the world’s best social media site, and an empty news feed. Farrell and Saloner dubbed the excess inertia that can occur when early adopters bear a disproportionate share of these switching costs ("transient incompatibility costs") the penguin effect. “Penguins who must enter the water to find food often delay doing so because they fear the presence of predators. Each would prefer some other penguin to test the waters first.” (Farrell & Saloner, 1986) Legacy economists, like penguins, are stuck in a case of excess inertia.

Network effects

When technologies exhibit network effects (increasing returns based on number of users), multiple outcomes and equilibria are often possible. In economics, network effects are visible in levels of research support for various models. The aggregate research of non-establishment economists “is a fraction of what the mainstream tradition mounts, especially in core, policy-relevant domains like macroeconomics and public finance. Relative to the mainstream, the dissenters do not get much attention, and do not carry much weight.” (Winter) For young academics seeking tenure, prestige, and funding, the rational course of action is to become an expert in the mainstream models (DSGE, RBC, OLG) and work to improve them, without questioning too deeply the core assumptions (rational expectations, perfect foresight, importance of frictions and technology shocks, etc.) on which the entire framework is based. “The successful scholar is always the one who adds some marginal improvement to the dominant theories everyone is already accustomed to. If, however, a new theory falls outside of customary channels, it is certain to face general opposition whatever its empirical justification.” (Allais) And the empirical justification for the dominant framework is shaky.

Paul Romer has written that, due to convoluted models that rely on assumptions that are opaque and not credible, “For more than three decades, macroeconomics has gone backwards.” This concurs with Kenneth Arrow’s assessment. The man who developed the modern conception of general equilibrium (along with Gerard Debreu and Lionel McKenzie) concluded that the limitations of assuming rational behavior are “in contradiction to the very large bodies of empirical and theoretical research, which draw powerful implications from utility maximization.” (Arrow, 1986) Yet the research has continued.

Ad hoc additions hinder the shift to a better economics

Economists have tried to address some of these problems by adding “frictions” to their workhorse models. The intent of these frictions is to represent aspects of reality such as unemployment, a financial sector, or a role for government. But as Nobel laureate Robert Solow explains, “To try to do it in a logically tight way is extremely difficult.” For example, even in models with frictions, unemployment does not happen because there is no market available for a worker’s output -- there are simply glitches in the labor market. These frictions function as converters, in the sense that they hinder the shift to a more efficient equilibrium.

Farrell and Saloner explored how converters between otherwise incompatible technologies can hinder the shift to a more efficient equilibrium, finding “that the existence of converters can actually lead to less compatibility than would occur in their absence”. (Farrell & Saloner, 1992) In a similar sense, the process of adding frictions to models is a mechanism for legacy economists to acknowledge that the macroeconomics being taught at top universities does not represent the real world, while preventing the entrance of contrasting opinions or alternative theories.

[These alternative theories and methods are the complexity economics and agent-based modeling that I'll focus on in this blog. For the moment, here's Leigh Tesfatsion's site. It provides a great summary of agent-based computational economics and many useful links.]

Herding and cascades

The change in macroeconomic research since the 1970’s has “come with the destruction of some knowledge, and suffers from extremism and herding.” (Blanchard, 2009) This behavior is consistent with the finding that, even when network effects are not present, herding ("uniform social behavior") occurs as a result of informational cascades – situations in which “it is optimal for an individual, having observed the actions of those ahead of him, to follow the behavior of the preceding individual without regard to his own information.” (Bikhchandani, Hirshleifer, & Welch, 1992). However, informational cascades are vulnerable to shocks and thus can be fragile: they explain not only uniform behavior but also instances of drastic change.

Cracks in legacy economics, exposed by the financial crisis, have continued to expand. Larry Summers has recently written that “We have come to agree with the point long stressed by Post Keynesian economists & recently emphasized by Palley that the role of specific frictions in economic fluctuations should be de-emphasized relative to a more fundamental lack of aggregate demand.” As it becomes increasingly clear that current mainstream economic modeling cannot effectively accommodate the world we live in – including low interest rates, corporate finance behavior, and not-perfectly-rational humans – alternative theories with stronger empirical justification should receive more support.

Next steps

There are some parallels between today and 1939. After a decade of economic hardship, the U.S. was facing slower population growth and uncertain productivity growth. A sense of unease pervaded the establishment that mainstream economic thought did not provide an effective framework for addressing new problems. “We are moving swiftly out of the order in which those of our generation were brought up, into no one knows what.” (Hansen, 1939)

The U.S. was morbidly “rescued” from its 1930's economic malaise by World War II. There are other ways to revitalize economic dynamism, whether we dig holes, build pyramids, or invest in infrastructure, research, and education. At the very least, I am convinced by the argument that “there needs to be more pluralism. Just like in the sixteenth century, after the Christian Reformation, there may no longer be a true church.” (Vines & Wills, 2018)

References

• Arrow, K. J. (1986). Rationality of Self and Others in an Economic System. The Journal of Business, pp. 385-399.
• Bikhchandani, S., Hirshleifer, D., &; Welch, I. (1992). A Theory of Fads, Fashion, Custom, and Cultural Change as Informational Cascades. Journal of Political Economy, Vol. 100, No. 5: pp. 992-1026.
• Blanchard, O. J. (2009). The State of Macro. Annual Review of Economics, 209-228.
• Cassidy, J. (2009). How Markets Fail. New York: Penguin Group.
• Farrell, J., & Saloner, G. (1985). Standardization, Compatibility, and Innovation. RAND Journal of Economics, Vol. 16, No. 1: pp. 70-83.
• Farrell, J., & Saloner, G. (1986). Installed Base and Compatibility: Innovation, Product Preannouncements, and Predation. The American Economic Review, 940-955.
• Farrell, J., & Saloner, G. (1992). Converters, Compatibility, and the Control of Interfaces. Journal of Industrial Economics, Vol. 40, No. 1: pp. 9-35.
• Hansen, A. H. (1939). Economic Progress and Declining Population Growth. The American Economic Review, Vol. 29, No. 1: pp. 1-15.
• Vines, D., & Wills, S. (2018). The rebuilding macroeconomic theory project: an analytical assessment. Oxford Review of Economic Policy, Vol. 34, Issue 1-2: pp. 1-42.