Machine as Metaphor
During [Sigmund] Freud’s university years (the late 1870s and early 1880s), young enthusiasts in the fuzzier disciplines, such as psychology, liked to borrow terminology from the more rigorous and established field of mechanical physics. The borrowed terms became, in fact, metaphor; and metaphor, like a shrewd servant, has a way of ruling its master. Thus, Freud wound up with the idea that libido or sexual “energy,” as he called it, is a pressure that builds up within a closed system to the point where it demands release, as in a steam engine:
Over the past twenty years . . . neurophysiologists have begun to study the actual workings of the brain and central nervous system. These investigators find no buildups of “pressure” or “energy,” sexual or otherwise, for the simple reason that the central nervous system is not analogous to an engine. They regard it as more like an electronic circuit, such as a computer or a telephone system.15
During [Sigmund] Freud’s university years (the late 1870s and early 1880s), young enthusiasts in the fuzzier disciplines, such as psychology, liked to borrow terminology from the more rigorous and established field of mechanical physics. The borrowed terms became, in fact, metaphor; and metaphor, like a shrewd servant, has a way of ruling its master. Thus, Freud wound up with the idea that libido or sexual “energy,” as he called it, is a pressure that builds up within a closed system to the point where it demands release, as in a steam engine:
Over the past twenty years . . . neurophysiologists have begun to study the actual workings of the brain and central nervous system. These investigators find no buildups of “pressure” or “energy,” sexual or otherwise, for the simple reason that the central nervous system is not analogous to an engine. They regard it as more like an electronic circuit, such as a computer or a telephone system.15
—Tom Wolfe, “The Boiler Room and the Computer”
Like psychologists, economists were taken with mechanical metaphors, particularly in the aftermath of World War II. Uppermost in their minds were mass-produced machines, such as the T-34 tanks that the Soviet Union used to turn the tide in its struggle against Nazi Germany. Postwar economists naturally thought of an economic problem con- sisting of allocating key resources, such as steel and oil, among alternative uses, primarily aircraft, warships, and battle tanks.
The graduate program in economics at MIT was at first heavily funded by the U.S. Department of Defense, as World War II seemed to show the importance of combining economics with engineering. That combination was particularly useful for addressing problems of constrained optimization. That is, given a fixed capacity to produce, say, steel and rubber tires, what is the optimal quantity of tanks and airplanes to manufacture?
Like psychologists, economists were taken with mechanical metaphors, particularly in the aftermath of World War II. Uppermost in their minds were mass-produced machines, such as the T-34 tanks that the Soviet Union used to turn the tide in its struggle against Nazi Germany. Postwar economists naturally thought of an economic problem con- sisting of allocating key resources, such as steel and oil, among alternative uses, primarily aircraft, warships, and battle tanks.
The graduate program in economics at MIT was at first heavily funded by the U.S. Department of Defense, as World War II seemed to show the importance of combining economics with engineering. That combination was particularly useful for addressing problems of constrained optimization. That is, given a fixed capacity to produce, say, steel and rubber tires, what is the optimal quantity of tanks and airplanes to manufacture?
Constrained optimization became a trademark of postwar economic theory. MIT economists modeled both the individ- ual consumer and the individual firm as solving constrained optimization problems. There was even a brief attempt to model all economic policy as solving a constrained optimiza- tion problem using a “social welfare function,” which would substitute society’s values for market prices. MIT and the economics profession as a whole developed confidence that with modeling, mathematics, and statistical information, they could fine-tune the economic machine.
The MIT revolution was led by Paul Samuelson, who in 1970 became the first American to win the Nobel Prize in Econom- ics, and whose textbook dominated the market in the 1950s and 1960s and serves as the template for popular current textbooks.
The MIT revolution was led by Paul Samuelson, who in 1970 became the first American to win the Nobel Prize in Econom- ics, and whose textbook dominated the market in the 1950s and 1960s and serves as the template for popular current textbooks.
Samuelson and his successors taught that the economic machine had a gas pedal that could be used to avoid economic slowdowns. That device was “aggregate demand,” which could be increased by the government’s printing money, running a budget deficit, or both. In this economic subfield, known as macroeconomics, the concept of specialization is forgotten entirely. Instead, economists employ an interpretive frame- work in which every worker performs the same job, toiling in one big factory that produces a homogeneous output. Macroeconomics replaces specialization with that GDP factory.
Fifty years ago, those researchers who were not seduced by Freudian metaphors were likely to be enamored of B.F. Skinner, who taught that human behavior could be inter- preted as if we were machines that respond predictably to past experiences of pleasure or pain. However, today, many researchers prefer the interpretive framework of evolutionary psychology. They see the human brain as being endowed with capabilities that evolved in the era of hunting and gathering but that can be adapted to very different environments. Both individual behavior and cultural norms respond to more than just the stimuli of reward and punishment.
Although psychology has moved on, the discipline of eco- nomics has remained stuck in its mechanistic metaphors. As a result, economists engage in an ultimately futile attempt to apply mathematical methods that are analogous to those used to measure a tank’s speed, firepower, and armor. Economists have yet to incorporate metaphors that pertain to the computer or communication networks. They have not come to terms with the reality that an economic system is much more complex than a T-34 tank.
Fifty years ago, those researchers who were not seduced by Freudian metaphors were likely to be enamored of B.F. Skinner, who taught that human behavior could be inter- preted as if we were machines that respond predictably to past experiences of pleasure or pain. However, today, many researchers prefer the interpretive framework of evolutionary psychology. They see the human brain as being endowed with capabilities that evolved in the era of hunting and gathering but that can be adapted to very different environments. Both individual behavior and cultural norms respond to more than just the stimuli of reward and punishment.
Although psychology has moved on, the discipline of eco- nomics has remained stuck in its mechanistic metaphors. As a result, economists engage in an ultimately futile attempt to apply mathematical methods that are analogous to those used to measure a tank’s speed, firepower, and armor. Economists have yet to incorporate metaphors that pertain to the computer or communication networks. They have not come to terms with the reality that an economic system is much more complex than a T-34 tank.
Computer networks, which today offer a powerful meta- phor for thinking about the brain or human society, were not around in that key period in the history of economic thought. Until the mid-1970s, the only computers were mainframes, referred to as “big iron.” Far larger than today’s computing devices, and yet less powerful, computers were classified as heavy equipment. Owning a single mainframe required a massive investment. In the private sector, only the largest businesses could afford them, and only a handful of firms, primarily IBM, could supply them.
Managers thought of mainframe computers as akin to mechanical calculators. Rather than working on general, multipurpose software, most people who wrote computer code were programming the machine to perform a specific calculation.
The advent of the personal computer in the late 1970s and early 1980s brought software to prominence, exemplified by Microsoft. Subsequently, the emergence of the public Internet in the 1990s demonstrated the significance of decentralized networks offering specialized sources of content and connection.
Managers thought of mainframe computers as akin to mechanical calculators. Rather than working on general, multipurpose software, most people who wrote computer code were programming the machine to perform a specific calculation.
The advent of the personal computer in the late 1970s and early 1980s brought software to prominence, exemplified by Microsoft. Subsequently, the emergence of the public Internet in the 1990s demonstrated the significance of decentralized networks offering specialized sources of content and connection.
I have come to see software and Internet resources as useful metaphors for the market economy. In my view, it is better to think of the economy in relation to the Internet than in relation to a T-34 tank. A tank performs only a few functions. It is deliberately designed by a small group of engineers. It can be understood and evaluated using a few simple measurements, such as speed, armor thickness, and gun capacity. In contrast, the ser- vices available on the Internet perform myriad functions. The resources on the Internet, and the patterns of specialization and trade in the market, emerge from the actions of countless individuals, not from the minds of a few designers. And the factors that affect the value of market production or Internet resources are many, complex, and not all quantifiable.
The “hardware” of the economy consists of its physical resources and physical outputs. However, as with comput- ers, economic “software” is at least as important. Most of the world’s wealth is intangible. It consists of our individual and collective knowledge. We know how to transform apparently useless gunk, called oil, into energy. We know how to transform an apparently inert element, called silicon, into computer chips that enable us to process information and to communicate more efficiently. As consumers, we know how to work with complex equipment, such as automobiles and cell phones. As workers, we have great stores of industry-specific and task-specific know-how.
The “hardware” of the economy consists of its physical resources and physical outputs. However, as with comput- ers, economic “software” is at least as important. Most of the world’s wealth is intangible. It consists of our individual and collective knowledge. We know how to transform apparently useless gunk, called oil, into energy. We know how to transform an apparently inert element, called silicon, into computer chips that enable us to process information and to communicate more efficiently. As consumers, we know how to work with complex equipment, such as automobiles and cell phones. As workers, we have great stores of industry-specific and task-specific know-how.
Most of the world’s economic backwardness also comes from intangible factors. Where widespread poverty exists, it can be traced to bad governance, violent conflict, counterpro- ductive social norms, and poor education.
Samuelson and his successors created a modern economic orthodoxy that is flawed in several important ways. As we have seen, the engineering approach is best suited to tangible, quantifiable elements, such as the number of hours worked in factories or the number of machines in use. However, eco- nomic reality is more subtle and complex.
The MIT-influenced approach that dominates the econom- ics profession treats individual markets and the economy as a whole as if they were simple machines. It embodies a view that economic behavior can be analyzed and predicted on the basis of mathematical equations. The economist plays a role analogous to that of a mechanical engineer, using models and equations to suggest ways for policymakers to make markets operate more efficiently.
Samuelson and his successors created a modern economic orthodoxy that is flawed in several important ways. As we have seen, the engineering approach is best suited to tangible, quantifiable elements, such as the number of hours worked in factories or the number of machines in use. However, eco- nomic reality is more subtle and complex.
The MIT-influenced approach that dominates the econom- ics profession treats individual markets and the economy as a whole as if they were simple machines. It embodies a view that economic behavior can be analyzed and predicted on the basis of mathematical equations. The economist plays a role analogous to that of a mechanical engineer, using models and equations to suggest ways for policymakers to make markets operate more efficiently.
However, economic models of markets are not as pow- erful as the engineer’s model of a machine. Economists’ mathematical modeling fails to come to terms with the complexity of economic phenomena. In the real world, too many factors have to be left out of the mathematical models.
Defenders of the modern orthodoxy will argue that the use of mathematics is a way to force economists to keep track of their assumptions. Formal models make assumptions explicit, rather than hiding them. Mathematical derivations demon- strate how the assumptions interact.
The use of mathematics helps verify the connection between assumptions and conclusions, but it does not guarantee that we are making good choices in our assump- tions. On the contrary, we often make very bad assump- tions, because better assumptions would be too difficult to handle mathematically. Thus, we use “two-by-two” models of international trade, when the reality consists of much more complex forms of specialization. We model “expectations” as a set of identical beliefs held by everyone in the economy, when in fact differences exist among peo- ple with regard to information and expectations. We take market imperfections as given, rather than consider how enterprises and institutions might evolve to address current problems.
Defenders of the modern orthodoxy will argue that the use of mathematics is a way to force economists to keep track of their assumptions. Formal models make assumptions explicit, rather than hiding them. Mathematical derivations demon- strate how the assumptions interact.
The use of mathematics helps verify the connection between assumptions and conclusions, but it does not guarantee that we are making good choices in our assump- tions. On the contrary, we often make very bad assump- tions, because better assumptions would be too difficult to handle mathematically. Thus, we use “two-by-two” models of international trade, when the reality consists of much more complex forms of specialization. We model “expectations” as a set of identical beliefs held by everyone in the economy, when in fact differences exist among peo- ple with regard to information and expectations. We take market imperfections as given, rather than consider how enterprises and institutions might evolve to address current problems.
One example of dubious analysis based on mathematics concerns the question of how retired people should allocate their assets between annuities and ordinary savings. For example, if you have $300,000 in savings, you could use it to obtain an annuity from an insurance company that will pay you, say, $30,000 a year for as long as you live. Instead, sup- pose that you gradually spend from the savings that you have. In that case, if you live exceptionally long, you face the risk of outliving your savings.
An extensive literature using mathematical modeling says that most or all of retiree savings should be converted to annuities. Economists have even suggested that because few people use annuities, it might be appropriate to force them to do so. For example, Davidoff, Brown, and Diamond write:
The near absence of voluntary annuitization is puzzling in the face of theoretical results that suggest large benefits to annuitization. While incomplete annuity markets may render annuitization of a large fraction of wealth suboptimal, our simulation results show that this is not the case even in a habit-based model that intentionally leads to a severe mismatch between desired consumption and the single pay- out trajectory provided by an incomplete annuity market. These results suggest that lack of annuity demand may arise from behavioral considerations, and that some mandatory annuitization may be welfare increasing.
An extensive literature using mathematical modeling says that most or all of retiree savings should be converted to annuities. Economists have even suggested that because few people use annuities, it might be appropriate to force them to do so. For example, Davidoff, Brown, and Diamond write:
The near absence of voluntary annuitization is puzzling in the face of theoretical results that suggest large benefits to annuitization. While incomplete annuity markets may render annuitization of a large fraction of wealth suboptimal, our simulation results show that this is not the case even in a habit-based model that intentionally leads to a severe mismatch between desired consumption and the single pay- out trajectory provided by an incomplete annuity market. These results suggest that lack of annuity demand may arise from behavioral considerations, and that some mandatory annuitization may be welfare increasing.
Knowing of that literature, I had always inferred that annuitization was a good idea, until I observed close relatives reach retirement age. Then, when they asked me for financial advice, I noticed the following considerations:
- As people reach their later years, their financial needs are dominated by health issues. That means that con- sumption needs are not smooth. On the contrary, the elderly can face sudden increases in the cost of deal- ing with disabilities (moving to assisted-living facil- ities, or requiring a home health aide) and a steady decline in non-health-related spending (less travel and entertainment).
- A lot of scope for insurance exists within a family. If one spouse lives an exceptionally long time, that spouse can be supported by savings left over from the spouse who died earlier. Alternatively, a long-lived parent can be supported by children.
3. If you wind up spending the last few years of your life in a nursing home, then you are not better off for having an annuity with nothing to spend it on.
In principle, mathematical models can be adapted to take into account real-world complications. In practice, however, economists tend to draw strong conclusions from simple models.
When it comes to the financial sector, mathematics has served economists mainly as a blindfold. Engineering models are poorly suited to articulating the role of financial intermediation in the economy:
In principle, mathematical models can be adapted to take into account real-world complications. In practice, however, economists tend to draw strong conclusions from simple models.
When it comes to the financial sector, mathematics has served economists mainly as a blindfold. Engineering models are poorly suited to articulating the role of financial intermediation in the economy:
- In the engineering model, the essence of economic activ- ity is turning resources into output. However, there is no tangible output from the financial sector to quantify.
- The engineering approach divides into “partial equilib- rium,” which looks at the behavior of a single market, and “general equilibrium,” which looks at the interaction among all markets. The financial sector is important for the way in which it interacts with other sectors, so that it is not well understood using a partial equilibrium approach. However, most general equilibrium models are posed as mathematical problems that can be solved without any financial sector at all.
In his recent book Foolproof, financial journalist Greg Ip says of economic policy analysts:
Philosophically, they fall into two schools of thought. One, which I call the engineers, seeks to use the maximum of our knowledge and ability to solve problems and make the world safer and more stable; the other, which I call the ecologists, regards such experts with suspicion, because given the complexity and adaptability of people and the environment, they will always have unintended consequences that may be worse than the problem we are trying to solve.
I fall on the ecologist side of this divide. Although an engi- neer thinks of machines as having stable, predictable proper- ties, an ecologist thinks of an evolving, adapting system.
The engineering approach requires a presumption that someone is standing outside the economy—an economic policy adviser—who, with the aid of simple models and equations, clearly sees what it would take to achieve outcomes that are superior to those that would emerge without government intervention. The engineers argue, quite reasonably, that we should not interpret market outcomes as perfect or ideal. However, they implicitly assume, much less reasonably, that the political process aided by economic models, will succeed in correcting the flaws in markets. Moreover, they assume that individuals and organizations acting in the context of markets will be unable to adapt to solve the problems that arise.
Philosophically, they fall into two schools of thought. One, which I call the engineers, seeks to use the maximum of our knowledge and ability to solve problems and make the world safer and more stable; the other, which I call the ecologists, regards such experts with suspicion, because given the complexity and adaptability of people and the environment, they will always have unintended consequences that may be worse than the problem we are trying to solve.
I fall on the ecologist side of this divide. Although an engi- neer thinks of machines as having stable, predictable proper- ties, an ecologist thinks of an evolving, adapting system.
The engineering approach requires a presumption that someone is standing outside the economy—an economic policy adviser—who, with the aid of simple models and equations, clearly sees what it would take to achieve outcomes that are superior to those that would emerge without government intervention. The engineers argue, quite reasonably, that we should not interpret market outcomes as perfect or ideal. However, they implicitly assume, much less reasonably, that the political process aided by economic models, will succeed in correcting the flaws in markets. Moreover, they assume that individuals and organizations acting in the context of markets will be unable to adapt to solve the problems that arise.
There are several fundamental concerns with the presump- tion of a wise, benevolent policy process. It treats the knowledge embedded in an economist’s simplified model as though it were complete knowledge. It ignores the ways that markets might adapt to solve problems. And it presumes that when the political process goes to work on problems, it arrives at solutions flawlessly.
Economists talk about “market imperfections” or “market failure.” However, economic models are themselves imperfect and capable of failure. If you ask different economic experts to predict the effect of a change in health insurance regulation or an increase in the corporate income tax rate, you will get different answers. The answer that appears to have the stron- gest support may turn out to be incorrect in practice.
Economists talk about “market imperfections” or “market failure.” However, economic models are themselves imperfect and capable of failure. If you ask different economic experts to predict the effect of a change in health insurance regulation or an increase in the corporate income tax rate, you will get different answers. The answer that appears to have the stron- gest support may turn out to be incorrect in practice.
As outsiders, economists see some of the conditions in a market, but they omit other factors. In that regard, econo- mists are no different from other outsiders. To the extent that there are outsiders who see a flaw in how the market serves consumers, those outsiders have the option of starting a busi- ness to address the problem. That is what entrepreneurs do all the time, and they are the main engine of economic progress.
However, entrepreneurs are often mistaken, and new busi- nesses often fail. By the same token, economists and policy- makers are also capable of making errors. What we should be comparing is not the existing market configuration with an ideal based on a simple model but the market process of error correction with the political process of error correction.
My skepticism of mechanistic, mathematical modeling leads me to reject the implicit assumption of a nearly omniscient economic adviser. I will return to this subject in the section on policy in practice.
However, entrepreneurs are often mistaken, and new busi- nesses often fail. By the same token, economists and policy- makers are also capable of making errors. What we should be comparing is not the existing market configuration with an ideal based on a simple model but the market process of error correction with the political process of error correction.
My skepticism of mechanistic, mathematical modeling leads me to reject the implicit assumption of a nearly omniscient economic adviser. I will return to this subject in the section on policy in practice.
No comments:
Post a Comment