Download PDF Spring Bridge on AI: Promises and Risks April 15, 2025 Volume 55 Issue 1 This issue of The Bridge features fresh perspectives on artificial intelligence’s promises and risks from thought leaders across industry and academia. Op-Ed Why Engineers Should Learn Economics Friday, April 11, 2025 Author: Debasis Mitra Engineers and economists have much to learn from each other, and there is much to be gained from them working collaboratively. Economics and economists wield great influence in policy-making at various levels of both the public and private sectors of our society. Examples in the public sector include policies for industry, trade, and regulations. In the private sector, top managers of engineering firms often find it to their advantage to talk up their economics know-how and economic reasoning in justifying their decisions, even when they lack depth in their knowledge. While economists hold sway in policy-making, engineers are in the unenviable position of being the policy-enablers without a major say in policy-making. Engineers are also typically directly affected by these policies, arguably more so than workers in other professions. Economists and engineers generally have an inadequate understanding of the other profession’s basic knowledge and goals, and they have much to gain from changing the status quo. Economics is understandably essential for policy-setting, and a reason that engineers are largely ignored is their ignorance of the field. It is true that “engineering economics” has a long history of support for inclusion in the engineering curriculum (Hayford 1917), but what has been proposed is largely accounting in substance. What is needed is a broad understanding of micro- and macroeconomics, including, at a bare minimum, models for and analysis of social welfare, consumption, investments, markets, growth models, business cycles, and fiscal policy. Decision-makers in engineering have much to gain from such knowledge. Do economists understand the workings of engineering systems? Take, for example, outsourcing, with globalization at the extreme. Globalization affected society broadly and engineers especially. Through the ’80s and ’90s, economists were the cheerleaders, making the model-based case for mutual benefits to the outsourcer and the contract manufacturer. Economists now have a better handle on all the implications (Autor et al. 2016; Krugman 2019). Had engineers been at the table when the outsourcing dogma was being built up, then, just possibly, they may have pointed out that their professional experience spoke strongly to the downside, which would have been a counterbalance to the arguments for globalization. Engineering systems’ success depends on tight feedback loops between different working groups, each with its particular specialty. This gives engineers an ingrained and intuitive understanding of the value of such couplings. The Bell System gives us a historical perspective on how these feedback loops worked. Consider Hendrik Bode’s career at Bell Telephone Laboratories, during which he worked on mathematics, circuit theory, control systems, military systems, and, after retirement, as a professor of systems engineering at Harvard (Brooks 1989). Bode’s 1971 monograph, Synergy: Technical Integration and Technological Innovation in the Bell System, champions “integration” of systems. Here is a quote from the book that so well describes the workings of integration: “In writing about the transistor discovery, Dr. M.J.Kelly, then President of BTL, said, ‘In accord with our policy of concentrating the efforts of our scientists on research, we immediately formed a closely associated fundamental development group to acquire that body of technological knowledge essential to the development and design of transistors for the many specific communications applications that would certainly follow. They have interested themselves in such problems as the factors controlling the bandwidth of amplification; the noise figure; the amount of amplification possible per stage; energy levels of output; basic materials, processing and structure studies essential for controlled development and design of transistors for specific functions; etc.’” Bode’s book makes the case that systems engineering helps in establishing feedback loops by linking specialized compartments and subsequently smoothing the coordination of their activities. Indeed, critical to engineering in the Bell System was the intimate interworking of systems and services with substantial help from a large contingent of systems engineers at the interfaces (Gilliam 2023). Economists and engineers generally have an inadequate understanding of the other profession’s basic knowledge and goals, and they have much to gain from changing the status quo. The breakup of AT&T and the spin-off of Lucent Technologies were based in part on the apparent value of separating “systems” and “services.” There were irreparable losses to both sides, which accumulated over the years. A more recent example of the losses that build up over time as a consequence of breaking the feedback loops is that of Boeing divesting various functions that had been integrated, including fuselage construction to Spirit Aerosystems. The costs so greatly outweighed the benefits that apparently a reunion is now in the works (Sindreu 2024). Does economic theory provide an understanding of the value of integration in engineering? Does it have a handle on the loss from breaking an existing integration? Surveying the economics literature, the answer would appear to be no, at least not to a degree that is adequate for the high goal of influencing national and corporate policy. If, on the other hand, engineers understood economics and spoke the language, then at least they would be able to make their case, which could, just possibly, be heard. It must be said that none of this is open and shut. That is, there are instances where engineering feedback loops have been broken with apparent success. For instance, from about 1970 to at least the turn of the century, Intel exemplified integrated chip design and manufacture. The integrated model started breaking up around 2010, to be replaced by the current model wherein the design is done in-house and manufacturing is outsourced to a foundry contractor, such as Taiwan Semiconductor Manufacturing Company (TSMC), the premier chip manufacturer. So, when do integrated operations out-perform outsourcing? In my opinion, this is a large area of engineering economics with many open questions. This is so even without taking national security into account. It should be noted that there are areas where engineering experience and economic theory converge. Take, for instance, learning-by-doing. The feedback loops connecting manufacturing and design run continuously to improve performance as measured by various metrics. The Nobel Prize-winning economist Kenneth Arrow modeled and analyzed this process in his classic 1962 paper, “The Economic Implications of Learning by Doing.” The following quote from the paper puts it succinctly: “…to produce the Nth airframe of a given type, counting from the inception of production, the amount of labor required is proportional to N -1/3. This relation has become basic in the production and cost planning of the United States Air Force.” No less consequential than cost reduction is the gain in knowledge. It should be noted that when work is outsourced the gain in knowledge capital from learning-by-doing accrues to the doer, not to the outsourcer. The position that TSMC occupies today in the semiconductor business must surely be in part due to the benefits of the knowledge capital that it has accumulated. This too should be part of the economics of outsourcing. Turning to the future, there would appear to be tremendous scope in engineers and economists working collaboratively to tackle the greatest challenge facing humankind today, climate change. This may not have happened in the past, yet the potential exists. Reasons to be hopeful emerge from looking in the rearview mirror at the occurrence of several related and parallel developments in the engineering and economic sciences. For instance, in recent years economists have been looking hard at the policy consequences of predictions of future global average surface temperatures due to global warming from the greenhouse effect. There is strong evidence that the probability distribution of the temperature is heavy-tailed, and the variance is infinite, which translates to probabilities of extreme future temperatures being substantially higher than would be the case with thin-tailed distributions. When it comes to policy implications, there is a schism, with one camp (see “dismal theorem” [Weitzman 2011]) championing extreme measures to avoid dire consequences, while another camp downplays the aura around fat tails when it comes down to policy implications, and goes on to promote far less drastic policy responses that put global warming on par with other possible calamities (Pyndyck 2011). This schism in the economics community mirrors one that enveloped the internet traffic engineering community beginning in the 1990s. Network engineers devoted considerable effort to collecting and analyzing data of internet packet traffic, which led to the discovery that the packet distributions have heavy tails. This discovery led a group of engineers to claim that the established processes of network control and design based on exponential and Markovian assumptions were invalid (Beran et al. 1995; Leland et al. 1994). Just as with the economists, counterarguments followed. Soon thereafter, the parameter domain in network design where such heavy-tail behavior mattered was substantially narrowed in the case of video traffic, the killer application on the internet at the time (Heyman and Lakshman 1966). A post-2000 study by statisticians examining network traffic (Cao et al. 2003) showed that in the scenario of higher-speed links and the multiplexing of many traffic streams, possibly having heavy-tail characteristics, the resulting traffic tends toward independence and exponentiality. The rationale for describing these debates and schisms in the economic and engineering communities is to take note that, in spite of the considerable differences in context, there is nonetheless commonality in the mindset, techniques, and approaches to problems. Yet another pair of parallel discussions involves the overhang of history in policy-making, which in economics is captured in the long-established concept of “path dependence.” It has taken on new importance with climate change, since it focuses attention on both the difficulties of weaning ourselves off of past dependencies on fossil fuel and the future consequences of the large investments currently being made in energy generation and storage (Aghion et al. 2014). In engineering systems there is a similar focus on the role of the past, which manifests when multiple equilibrium states exist and initial conditions determine trajectories to steady state at one of the equilibrium states. These concepts are long-established in engineering, especially in the analyses of nonlinear dynamical systems and control and electrical circuits. In certain cases, multiple equilibrium states exist and control strategies are devised for the system to equilibrate to a preferred state, and in other cases, systems are designed to ensure that a unique equilibrium state exists. This is yet another illustration of affinity and the potential gain in combining the knowledge stocks in the disciplines to tackle problems. There are other fundamental disciplinary skills that can be brought to bear in tackling climate change. For instance, in problem-solving, both in engineering and economics, an essential step is to settle on the operational time scale, and, unsurprisingly, time scale separation is in both disciplines’ toolboxes. These deep-rooted skills should provide common ground in jointly addressing the grand challenge. Engineers stand to benefit from being educated in economics. First, they will be better positioned to manage their own businesses. Globalization as we have known it may be diminishing, but the scale of business and, importantly, the need to address grand challenges, such as climate change, will undoubtedly become more global and complex, and knowledge of economic fundamentals will be ever-increasingly essential. Second, it will allow engineers to join economists in debating and setting policy at the higher levels of our society. Also, the skills and tools that are common to both knowledge bases should ease the path for engineers learning economics. Engineers stand to benefit from being educated in economics. (Note: A fair question is “What have you done to educate engineers in economics?” I have created two graduate courses, which I teach at Columbia, that intertwine engineering and economics, “Internet Economics, Engineering and the Implications for Society” and “Future Energy: Economics, Systems, Policies.”) References Aghion P, Hepburn C, Teytelboym A, Zenghelis D. 2014. Chapter 4: Path dependence, innovation and the economics of climate change. In: Handbook on Green Growth, 67–83. Fouquet R, ed. Edward Elgar. Arrow K. 1962. The economic implications of learning by doing. Review of Economic Studies 29:155–73. Autor DH, Dorn D, Hanson GH. 2016. The China shock: Learning from labor-market adjustment to large changes in trade. Annual Review of Economics 8:205–40. Beran J, Sherman R, Taqqu M, Willinger W. 1995. Long-range dependence in variable-bit-rate video traffic. IEEE Transactions on Communications 43(2/3/4):1566–79. Bode HW. 1971. Synergy: Technical Integration and Technological Innovation in the Bell System. Bell Laboratories. Brooks H. 1989. Hendrik Wade Bode. Memorial Tributes. National Academy of Engineering. Online at https://www.nae.edu/189189/HENDRIK-WADE-BODE-19051982. Cao J, Cleveland WS, Lin D, Sun DX. 2003. Internet traffic tends towards poisson and independent as the load increases. In: Nonlinear estimation and classification in Lecture Notes in Statistics (vol. 171), 83–109. Denison DD, Hansen MH, Holmes CC, B. Mallick B, Yu B, eds. Springer. Gilliam E. 2023. How did places like Bell Labs know how to ask the right questions? The Good Science Project, April 22. Online at goodscience.substack.com/p/how-did-places-like-bell-labs- know. Hayford J. 1917. The relation of engineering to economics. Journal of Political Economy 25(1):59–63. Heyman DP, Lakshman TV. 1966. What are the implications of long-range dependence for VBR-video traffic engineering. IEEE/ACM Transactions on Networking 4(3):301–317. Krugman P. 2019. Globalization: What did we miss? In: Meeting Globalization’s Challenges: Policies to Make Trade Work for All, 113–120. Catao L, Obstfeld M, eds. Princeton University Press. Leland W, Taqqu M, Willinger W, Wilson D. 1994. On the self-similar nature of ethernet traffic (extended version). IEEE/ACM Transactions on Networking 2:1–15. Pyndyck RS. 2010. Fat tails, thin tails, and climate change policy. 2010. National Bureau of Economic Research: working paper 16353. Online at: http://www.nber.org/papers/w16353. Sindreu J. 2024. Boeing calls time on the great American outsourcing. The Wall Street Journal, July 2. Weitzman ML. 2011. Fat-tailed uncertainty in the economics of catastrophic climate change. Review of Environmental Economics and Policy 5(2):275–92. About the Author:Debasis Mitra (NAE) is a senior research scientist in the Department of Electrical Engineering at Columbia University.