The Answer Is Transaction Costs

Little's Law: The Transaction Costs of (Re)Drawing Lines

• Michael Munger

Share episode

Send us a text

A conversation with Andrew Wagner, production and manufacturing engineer, now in aerospace, but with experience also in the auto industry.

We trace how transaction costs shape production, from Adam Smith’s pin factory to Toyota’s SMED, and why empowering workers and redesigning tools can raise quality while cutting cost. An aerospace manufacturing engineer joins us to unpack Little’s Law, line reconfiguration, and the culture that makes flexibility real.

• division of labor limited by the extent of the market
• sub shop and Chipotle as live line-balancing examples
• Smith’s three productivity drivers applied to modern factories
• Little’s Law guiding WIP, stations, and throughput
• costly line changes and capacity planning in auto plants
• meta-tools, CNC, and multi-operation automation
• stamping dies, SMED, and Toyota’s flexibility edge
• just-in-time, early error detection, and quality economics
• U.S. responses: robotics, platforms, and Deming at Ford
• NUMMI proof: same workforce, new system, better output
• CAD parametrics, modular design, and clay by robot
• structure by design: darts, curves, and manufacturability
• specialization, ergonomics, turnover, and the $5 day
• worker empowerment as applied Hayekian local knowledge
• letter on bureaucracy, spending, and the social order book pick

Some links:

Book o'da'Month:  Jacques Rueff, THE SOCIAL ORDER




If you have questions or comments, or want to suggest a future topic, email the show at taitc.email@gmail.com !


You can follow Mike Munger on Twitter at @mungowitz


SPEAKER_01:

This is Mike Munger, the knower of important things from Duke University. This episode is about division of labor and the extent of the market, particularly the transaction costs of changing a production line. My guest is Andrew Wagner, a production and manufacturing engineer, the sort of person Ronald Coase always said economists should talk to to find out how things really work. A new twedge, a letter, and a book at a month. Straight out of Creedmoor, this is Tidy C.

SPEAKER_04:

I don't want to talk about the system where there is no transaction cost. It's an imaginary system.

SPEAKER_00:

When it is costly to transact, institutions matter. And it is costly to transact.

SPEAKER_01:

This month on The Answer is Transaction Cost. We're talking to Andy Wagner. And Andy, as always, I ask that guests introduce themselves because I believe in comparative advantage and you know the most.

SPEAKER_03:

Fair enough. I am a manufacturing engineer in the aerospace industry. I've I've worked about 25 years in uh engineering, and most of that in aerospace. Uh I did my undergraduate at Vanderbilt University in uh materials science and engineering, and then I have a master's from Tufts University in engineering management. Where did you grow up? Oh, I grew up outside of Detroit. So uh part of my interest in in manufacturing comes out of the the uh the auto industry and being around that environment as a young boy.

SPEAKER_01:

I've talked to a number of people lately that were from the kind of Midwest who had gone to Vanderbilt. How did that happen?

SPEAKER_03:

Well, for me, uh I I lived uh maybe half an hour from Ann Arbor, and my entire high school was going to go to the University of Michigan, and I just wanted to be somewhere different. You were a rebel even then. That's right, absolutely.

SPEAKER_01:

Well, um, our our topic today, as always, is transaction costs, but you had contacted me with some very interesting thoughts about engineering and the problem of division of labor. I wanted to start by reading from Adam Smith's uh Wealth of Nations, the example of the Pin Factory, and say a little bit about what he says about division of labor. And then you and I will explore how that actually works. Because this the pin factory example is interesting, but it is something that takes place a very long time ago. It seems kind of abstract. And as Smith himself says in the first sentence, it's a very trifling manufacturer. So that was his point that even something so simple was a very trifling manufacturer. So here we go. So on page 14 of book one of the Wealth of Nations. To take an example, therefore, from a very trifling manufacturer, but one in which the division of labor has very often been taken notice of, the trade of the pinmaker, a workman not educated to this business, which the division of labor has rendered a distinct trade, nor acquired with the use of the machinery employed in it, to the invention of which the same division of labor has probably given occasion, could scarce, perhaps, with his utmost industry, make one pin in a day, and certainly could not make twenty. But in the way in which this business is now carried on, not only the whole work is a peculiar trade, but it is divided into a number of branches, of which the greater part are likewise peculiar trades. One man draws out the wire, another straights it, a third cuts it, a fourth points it, a fifth grinds it at the top for receiving a head. To make the head requires two or three distinct operations. To put it on is a peculiar business, to whiten the pins is another. It is even a trade by itself to put them into the paper. And the important business of making a pin is in this manner divided into about 18 distinct operations, which in some manufactories are all performed by distinct hands, though in others the same man will sometimes perform two or three of them. So he then goes on to say that, and then the the title of chapter three of book one is that the division of labor is limited by the extent of the market. So I had said when we were talking to Russ Roberts on Econ Talk, one of the questions I ask my students is how many people are there in a pin factory? And they think about that, and it seems like the answer should be technical. It's not. It depends on how many pins you're going to sell. And that's really interesting because you have to try to decide that before you decide how many people to hire, because that tells you how many distinct stations there are going to be. There's 18 distinct operations. There may be five people who perform those 18. There might be 18. There might be 50, and you have some of them concentrating on some of those, and some another. Those 18 operations might be further subdivided. So what Smith points out is that if you had 18 people each working by themselves, they could only make a few pins. If you have 18 people working in a production line where each of them is working individually on one of those 18 operations, you can make tens of thousands of pins in a single day.

SPEAKER_03:

And then a great example of this is if you go to your local sub shop, what you'll notice is if there's no line, there's one person who makes your sub, takes off their rubber gloves, goes to the cash register. There's a few people, you might have someone at the cash register and someone making subs. If there's a lot of people, you'll suddenly have two people making subs and they do something called line balancing, where one person will start making a sub and the second person will meet them in the middle and go all the way to the cash register, and then they'll come back and they'll meet in the middle. But that middle point where they meet the transaction can change depending on what the different people want on their subs. And that's how they handle the variation in different toppings on the sub.

SPEAKER_01:

And that's that's true at Chipotle also, where they will they they will hand they'll start the burrito, maybe put on whatever the meat is, and then the next person will just do the toppings. And that seems so inefficient. So Andy, I'm gonna ask you about why they they do that. But let me say uh on on page 17, Smith gives the explanation for why this might be useful. And the so what he's what he's just shown is that there's an enormous increase in the number of pins. You can have the same number of people working, but it depends how you organize it. Is each of them working as an artisan separately, or are the 18 of them like in the sub shop? Because in the subshop, you could have 18 different little tiny subshop stations, or you can have one little bit longer one with three people working on it, and the increase is going to be more than three times as much. It's not just three times, it's not linear, it's more than three times as much. And the explanation that Smith gives on page 17 is this great increase of the quantity of work, which in consequence of the division of labor, the same number of people are capable of performing is owing to three different circumstances. First, to the increase of dexterity of every particular workman. Second, to the saving of time, which is commonly lost in passing from one species to another, that is, taking off the gloves when you move from one part of the substation to another. And lastly, to the invention of a great number of machines which facilitate and abridge labor. Now that's pretty important for pins, maybe not so important for subs, but the the invention of machines and dexterity is something that you're going to be able to tell us something about. And so let me ask if you can give us the background of how you came to think about this and what examples you have seen.

SPEAKER_03:

Yeah, so I think one of the principles that came to mind when I listened to the econ talk episode was uh Little's law. And so John Little was an operations researcher at MIT, and his law is an observation about uh queuing theory, so so lines, but it also applies to a production line. And the the way we state Little's law is that the average work in progress or the average number of people in the line is equal to the process time times the delivery rate. And so delivery rate would be the extent of the market, you know, how many pins per second do you need to make? And so what we do when we think about constructing or designing a production line, that work in progress is a really important number because that's the the number of positions that will be in the production line. If you think of an automotive assembly line that snake through the building, the number of positions is going to be proportional to how many cars you want to build, and also proportional to the time it takes to manufacture the vehicle. So if it's um you know a long time to manufacture the vehicle and a very small market, like say um a Ferrari, you may have a fixed build position where there's one car and the workers kind of come and go, each doing their own their own task. But if you're talking about a uh you know a Toyota Corolla, you're gonna have hundreds of vehicles nose to tail moving through workstations.

SPEAKER_01:

Can you say something more about Little's Law? What's the definition of each of the parts of that?

SPEAKER_03:

So the the work in progress would be the number of workstations or really the number of the product, uh, number of piece uh work pieces that are that are being uh worked on or even in queue. So it could just be the number of people standing in line at the subway. Um, but it also could be the number of vehicles in the production line. And then the processing time is the time it takes from when you start production of that vehicle till it rolls out the door at the end of the factory.

SPEAKER_01:

And is this an empirical law or or a theoretical law? It is empirical.

SPEAKER_03:

It is empirical, yeah. And so it also refers to averages over a long time. So at any moment you might see some fluctuations in it, but over the long term, uh, we're all dead and Little's law holds.

SPEAKER_01:

And this it could be a fairly short medium term because it's just kind of a central tendency. There's going to be some differences, but it actually is something that you probably would observe working. Well, so the there are examples that you have seen where that kind of the problem of how many workers to hire is not really a theoretical one. Smith was doing this as kind of a one-off, and it's fine for me in my classroom to say how many workers will there be. But that's an empirical question, also. So, how do we adjust the number of workstations? Because it's easy to write on a blackboard and say, here's the number of workstations. But if you have a production line and you're making cars, adding another work line means you have to close down the line and actually change the entire production process. So, how do you plan for that?

SPEAKER_03:

Yeah, I was uh very early in my career. I worked for uh Jeep and they were going from a two-ship up shift, I'm sorry, a two-shift operation to one shift for this particular facility. And so uh, in order to do that, you know, the advantage is your production rates have gone down. You don't have to keep the plant operating for the whole of the second shift, but you need to make that production rate long or the line longer. And so they had to bring in a construction crew, tear down a wall of the building, and extend the wall by I don't remember how many positions, but actually physically altering the length of that production line. And so it was you know a year of uh probably a year of planning and a year of construction to do that.

SPEAKER_01:

And and did did it did it accomplish their object? Did it work out okay?

SPEAKER_03:

Yeah, yeah, they were able to do that. Now, of course, um, it also makes them more flexible for the future. So if they need to go up in production, if they were to bring back the the second shift, they would be able to, you know, each of those shifts would be that much more uh have that much more capacity.

SPEAKER_01:

So um are are there other examples that you have seen? One of the things that struck me when you you first wrote an email to Russ and to me saying that you had seen examples of this was I had not really thought about the problem of identifying. Smith looked at a pin factory and he said 18. And the the number of different steps of work is actually something that emerges in the process of having people work on them and the design of tools. Smith gives this great example that one of the most important tools for a steam engine, because there was a there was a problem with opening and closing valves in coordination with the movement of the piston in a steam engine. And it was the job, according to Smith, of a boy, because it was an easy job, and his job was to uh open the valve at the correct point. Well, the boy noticed that the movement of the piston coincided with the valve. It's it's a that makes perfect sense. So what he did was he tied a string and that that meant he could go play, but that cut the job out. And so it may be that the creation of tools, which tools and machinery, which are ways to make it easier for people to work, also get rid of jobs. And so the the thing about division of labor is that once you start to have tool use and machinery, you end up with fewer people, each of whom is dramatically more productive. But how does that actually work from an engineering perspective? It's easier for me to talk about it, but how does an engineer think about that?

SPEAKER_03:

Yeah, well, I think of uh uh machine tools. So if you go back um, you know, 40 or 50 years, there were a lot of highly specialized machine tools, and you would have to move that workpiece, the object you're working on, from one specialized machine to the next specialized machine. In a modern uh factory, we have automated tool changers. And so you can keep the the workpiece in one place, and the the tool itself reconfigures itself through automation to do multiple steps of the operation, multiple operations on the same machine.

SPEAKER_01:

Well, but that that seems like even an additional step then, because now you're thinking of meta tools. Uh and so the that that sounds great, but boy, is that a complicated process to try to have theories about.

SPEAKER_03:

Yeah, and I I think you know, we when we're we're doing that sort of work, you're understanding each of the transformations that need to happen to that piece of hardware. Uh, right. So you're going to drill a hole and then you're going to deeper the edges of that hole. And then you're going to um uh well, you'll drill the hole, then you'll ream the hole, then you'll deeper the edges of the hole, and then you'll uh do another feature on the part. And so you need to understand the design of the part and each of the features, and then have a technique for how you're going to create each of those features, and then a technique for how you're going to inspect and confirm that each one of those features are as intended. And so there's a process step associated with each one of each one of those, and it could be manual, it could be automation, um, it could be a combination.

SPEAKER_01:

Well, so one of the things that you had said in your email uh was that you gave an example of stamping car doors. And yeah, there there was a time when making sheet metal was really difficult to start, but really cheap once it was set up. The problem was that stamping machines were relatively fixed, and so you'd have to make a whole bunch of them, but there were enormous economies in in doing that. So it's also clear that if we can design, and this is the meta tool problem, if we can come up with tools that can change the those tools when we need them, we don't need to have such big inventories or such big production runs of identical cars. So uh if if you don't mind, start at the at the beginning about that and say something about the history of that.

SPEAKER_03:

Yeah. So if you think about you know, General Motors in the 1960s, the extent of the US market is massive. They're making hundreds of thousands of Chevrolets and hundreds of thousands of Pontiacs, and they would do clever things like the doors, the roof, and the floor panel on a Pontiac and a Chevrolet are exactly the same, but the fenders and hoods are different. And and with stamping, um, it can take maybe a whole day, at least a shift, you know, eight hours or so, to change the stamping dies. They're heavy steel plates. Uh, they have to be heated in order to do the work of forming the steel. And so to put the dies on the machine, to take the dies off of the machine, to get them aligned to the proper position, they'll do some tests, presses in order to make sure that that the two sides are aligned. It takes a the machine is not running, it's shut down for a whole day. So you're losing a day of production. But once you've got it uh set up and aligned and and completed, you can run dozens of parts. You can run for a week straight, and the parts only take a few seconds to make. The sheet goes in, the press comes down, the press opens, the sheet goes out, or the door panel goes out. So it's very quick. And so in the heyday of the US auto industry with that large market, we're able to just crank out panels. Toyota had a very different situation in Japan because their market was very small. And so it was very difficult for them to make a variety of different vehicles. So they were kind of stuck with just the basic vehicles because the cost, the transaction cost of going from, say, uh a pickup truck to a car to a small car was very high. And there was an industrial engineer who worked with them named uh Shigo Shingo, who is very famous today, who came up with this concept of what they call single-minute exchange of dyes. And the idea here is if we can take that time it takes to exchange your dyes from eight hours down to eight minutes, it changes the whole nature of the business. And so there's a set of methods that he follows, and it's not really an invention, it's it's a methodology for separating the things that have to be done on that stamping machine from the things that can be done external to the machine. So, for example, they warm up the dies not on the machine, but when they're sitting next to the machine. They get very efficient about how they move the new dies from a storage area before they shut down the machine. Um, they have uh rails that help with the alignment and movement of the dies onto the machine instead of using an overhead crane and more clunky improvised methods. They standardize the alignment of both the dies so that they don't have to futz with them once they're on. They have quick change um bolts and tools and clamps and things so that that that the um you know clamping down the die can be done in in a matter of seconds instead of hours of of turning bolts and and things like that. And what Toyota was able to do with this is they were able to switch from model to model to model much quicker. And so all of a sudden, when they started to come into the US market, they had this level of flexibility in their production that uh Detroit couldn't compete with. And so the the irony there is as we know, that the Japanese cars coming in had better quality than the American cars, uh, even though and and lower cost at the same time, even though they were coming across the the ocean. I I remember having uh economics debates as a young boy in Detroit about what do you mean your Japanese car is cheaper and better? It has to it has to be shipped over there. How's it possible? It's only logical, it has to be true. Yeah, yeah. And it was it was only later when I started to study these things and I understood exactly why.

SPEAKER_01:

Right. And the the the when you say it's a methodology, it's true that it pervades everything, including engine construction. But the US cars of that era, the 70s and 80s, were notorious for problems with fit and finish. And the Japanese cars are all perfect. That the and some of that is just the the their ability to make on-the-fly adjustments in sheet metal. And the the US was one size fits all. It was being put on by people that are probably relatively indifferent because there's there are other problems in Detroit, there are other things that are different. But the the the essence, what's cool about this story is that the Japanese actually faced higher cost. And the US had this advantage that land near Detroit is relatively cheap. You can have warehouses of sheet metal, it's cheap enough to do that. You don't have to do something better. The Japanese couldn't do that. Land was expensive and they their production runs were too small. The result was they made a change in the way that they not just used but thought about machine tools. And that gave them a big advantage in average cost.

SPEAKER_03:

And that that uh that cost of warehousing led to what they call the just-in-time system. They reduced their their quantities of inventory and the cost of that inventory, and then it would arrive to the the final assembly just in time for production. One of the advantages of that, advantages of that is is in the area of quality. If you have a problem with one of those parts in your warehouse, well, it was made a month ago. So, how do you root cause and understand what happened in your factory a month ago? Well, and maybe you have another 10,000 just like it. Exactly. Exactly. So you've dis your detection of a problem is delayed and it is impacted potentially thousands of more parts, and the cost of either repairing them or scrapping them is enormous.

SPEAKER_01:

So you just put them on the car and ship them out, and it's it's a it's a Chevrolet, people will like it.

SPEAKER_03:

The the joke is at the end of the line at Chevy, there were two doors, one that said repair and the other said heavy repair. The Cadillac, the the uh I got it backwards. I messed up the joke, but at Cadillac it was repair and heavy repair. At Chevy, it was repair and and out the door.

SPEAKER_01:

So well, right, Cadillac's the luxury brand, and so you're willing to repair that.

SPEAKER_03:

You would repair everything as opposed to just the really bad stuff.

SPEAKER_01:

Well, so the the the US automobile industry took kind of a beating, it lost a fair amount of market share that it has never really recovered. It has recovered some of its reputation for quality, particularly some lines and some cars, some more than others. Uh, one of the things that I remember about cars and trying to compete with the Japanese was by cutting costs. And Chrysler famously had the K-Car, which was basically the same car, no matter what it was. It was literally the same car. It didn't just use some of the same parts. Um, and that actually seemed a little bit cynical because it was hard to tell the story that these were actually different cars. Um, can you say anything about the US response? Because you've you've you've told an interesting story. Here's how the U.S. did it. It made sense. Japanese couldn't do that. They came up with a way that actually turns out to be cheaper and better, which seems paradoxical, but sometimes having to do that means that you're you're looking in ways that there was a complacency in the U.S. automobile industry. It's perfectly understandable. It was big and they were, they, they led the world until they didn't. So, what was what was the U.S. response to the Japanese?

SPEAKER_03:

Yeah, and uh favorite topic of mine. Um, so there were a lot of there were different responses, right? So the General Motors was very famous for uh investing in a lot of automation. They thought that robotics would be this the answer. And um, there's benefits to that, but there's also the transaction cost that to change from model year to model year and update the vehicles is is more expensive, particularly in those early days when the automation was less flexible. So you'd see the the General Motors cars would would be the same from year to year with very minor changes. Um Chrysler got scrappy and creative, and the K-car is a great example. They they did look very much the same, but the if you look at the details, the number of variations in that platform were actually very high. So there was, you know, the New Yorker was four inches longer than the uh you know the Plymouth equivalent. And the the first minivan was actually based on the K-car platform. So while the exterior is completely different, the the underpinnings and the mechanicals were very similar. Um, now I have to confess, my dad was at Ford for 37 years and I'm a bit of a partisan, but Ford really was the first to embrace the quality movement and it's clearly true.

SPEAKER_01:

You just look at sales, that's clearly true. If you just if you look at the F-150 alone, that's clearly true.

SPEAKER_03:

Yeah, and and it was in the in the early 80s, they started working with uh Dr. Deming, who was the famous quality guru of that time. And so they uh the CEO of Ford at the time, uh Donald Peterson was in monthly meetings with Deming. And so that culture around how do we execute first time past quality was deeply embedded in Ford in that era. And uh yeah, it it showed they they won a number of car of the year awards and so forth. And uh, you know, the F-150, the original Taurus were very successful programs for them.

SPEAKER_01:

Well, can you say something about how Deming's principles were implemented? So I I I knew that, but I don't know anything about it. Yeah, so Deming has it is his cycle is called Plan Do Check Act, and so it's a kind of method, and it was adopted sooner in Japan, and there's this this idea that Japan's going to take over the world. I remember I was a little bit skeptical of it, but in the in the mid-90s, uh in a lot of business schools, they said we're gonna have to do exactly like the Japanese do, and then the Japanese economy wandered off into the blackberry bushes. So it there was something about a period of time where Japan made a lot of progress very quickly. They adopted some of the deming principles, but uh then Japan became much less of an innovative player, at least.

SPEAKER_03:

Well, and I think that a lot of the deming principles have become uh ubiquitous, and and and most companies these days uh embrace a lot of those those techniques, and and so it's it's become uh uh table stakes instead of a differentiator. Um, I think that the fundamental insight that that deming shared was the idea that um, again, that that Cadillac example, it was it was perceived before Deming that quality was expensive, that quality required doing a bunch of extra steps in order to bring uh the product to a higher level. And what Deming's insight was that if you design the systems, the tools, the processes, train the people to execute the work correctly the first time, there's a paradox there. You can actually bring the cost down. And so and quality up, yeah, that that quality was free if you do the upfront work of creating a system that's repeatable and consistent. And so um, Toyota with their uh error-proofing philosophy, um, their deep uh root cause corrective action practices on the shop floor, and their uh approach to design, where they would um one of the stories is told is uh the American industry would have a spec for for a component and they would test and say it was supposed to last for 100,000 cycles. Well, they would test to 100,000 cycles, and if it passed, all right, this motor's perfect and we're gonna use it. Well, Toyota would test it until it broke, right? So they would find out if it only went to 101,000 cycles, uh, maybe we need to, maybe we need to redesign it. But if it went to 120, oh, maybe we can make it a little bit more fragile and save a little bit of money, right? So they by testing to failure, they understood exactly. Exactly where their designs were, and they could then use that knowledge on future products.

SPEAKER_01:

What has happened to the American automobile industry today? Um, one of the things that was interesting about the Deming principles was a connection to how I think about Friedrich Hayek's observations about local knowledge. So his claim was that one of the difficulties of top-down management, be it government or markets, is that uh centralized bureaucrats don't have an he called it the the knowledge of the man on the spot. And there was a mistrust in Detroit in the 60s and 70s about whether workers knew enough. Whereas at least the caricature, I don't know if it's true, and that's why I'm asking, was that in the Japanese system, a single worker could shut down the line and say, this is wrong, this is set up wrong. We're not going to continue, we're going to get it fixed. And in the long run, that ability sort of empowers workers more. You're going to have better quality and lower cost. Although at that moment, it means the line shuts down and the foreman loses his bonus. And so there's a conflict about the use of local knowledge there. But the the difference is the sense of empowerment of workers. Not only do we trust you, we depend on you to be the eyes about quality rather than just say, well, this the the line is moving. Yeah, that tender doesn't quite fit, but they'll fix it later.

SPEAKER_03:

Yeah. And those those stories uh they're they're told widely in in the engineering, you know, manufacturing engineering field. Um, there's folks who were brought in as manufacturing engineers when Toyota started opening plants in the US. And the first thing they did is they went to Toyota in Japan and they worked on the line for a few months. And so they had the personal experience of having made a mistake as a new production employee. And do I stop the line or not? And they were trained to stop the line, they'd stop the line, and then they'd they'd get applause. The group would come together and they'd say, Thank you, thank you, uh, you know, Mike's on for stopping the line. Um, and and the US to a great degree has embraced that approach. Um, but obviously labor relations ebb and flow. And so uh one of the things Deming worked with with Ford in the 80s was on improving labor relations. And so Ford had some of the best labor relations in that period of any of the big three, and it and it had the result in in quality.

SPEAKER_01:

Right, but it may have been that it's hard to know which it is. It partly the sense of empowerment and belonging, being partly in control rather than just being a cog in a machine, uh, might improve labor relations. But having better labor relations means that you can't just use stopping the line because you want a break or a work stoppage protest. And General Motors had much worse labor relations. And it, if they had done that, of the result might well have been people would just do it to sabotage because they're pissed off at management. And so there is a whole package of things you have to accomplish.

SPEAKER_03:

Yeah. And so again, the the the most famous story there is the story of the uh the Fremont, California uh General Motors plant. It was the worst plant in the General Motors system. It was shut down in the early 80s. And uh Toyota came to General Motors and said, We'd like to start a plant, a joint venture with you in the U.S. It's our first plant in the U.S. We want to work with you in order to learn what you know and you can learn from us. And uh they chose that Fremont, California location. We got a plan for you.

SPEAKER_01:

We know just the plant because this one, well, it sucks, but it'll be great.

SPEAKER_03:

Well, the thing was the facilities, General Motors built great facilities, the facility was excellent, but it'll be the same workforce or it'll be the same population they're drawing from. Well, so the the interesting part of the story, the original plan was hey, we're gonna hire all new people, but when they got into the discussions with the UAW, the UAW pulled the hard line. They said, No, you're gonna bring back the exact same people you laid off with some cause.

SPEAKER_01:

If it's a unionized place, then seniority and yeah, you would have to you might be able to negotiate out of that, but it would be expensive.

SPEAKER_03:

Yeah. But the product that came off of they called it New United Motor Manufacturing Incorporate NUMI, the product that came off of that line when Toyota was running it, it was the highest quality and most productive plant in the United States with the same workforce, but with the design of the Toyota vehicle. So it had the error proofing designed into it by the engineering staff, and it had the Toyota style of management on the shop floor, pulling, pulling the cord to stop the line, all that. Now they had to work to build trust with that workforce in order to get them comfortable with that idea, and that that took time, but uh the workers were willing to give them give them a chance, and and the results uh were remarkable. And of course, GM rotated managers through that plant and they learned those lessons and they learned those approaches and they took them back into the General Motors system, including, of course, Saturn, right? So Saturn was built very much on that philosophy.

SPEAKER_01:

What happened to Saturn? It seemed like it was such a success story for about a decade, and then it just disappeared.

SPEAKER_03:

Yeah, I think the the bottom line there is that the um it just got reabsorbed very slowly into the big General Motors system. But in the, and I don't remember the year of the strike, but I think it was around 2008 or so, there was a a strike at General Motors as they went through the financial crisis and all that. And the story was that that the the Saturn plant, they were only making engines, they weren't making cars anymore, but there was still a remnant of that unique culture. And so as the the management went through the the gate, they rolled down their windows and gave coffees to the guys on on the uh on the picket line because they still had that relationship that came from the Saturn when it was a unique uh culture.

SPEAKER_01:

Can you expand a bit? I mean we've we've talked a fair amount about automobiles, and that partly because that that is your experience and background. But somewhere in the last 10 or 20 years, one of the big advantages that China has in the production of plastics and textiles or clothing articles of well, I I just had the experience of uh I'm gonna be the president of the Philadelphia Society, which is uh academic and uh political organization in the US that meets and discusses ideas. And we had wanted to see how much it would cost to get a bespoke designed tie, so neckwear for men. And most of all the things we could find in the US were uh silkscreened onto the tie, which doesn't last very well, doesn't look very good. Uh, but having it knit or woven into the material is obviously much more expensive. There's a lot of setup costs. But I found a place in China that would do a run of only 40 of them for a cost of$15 per tie. It's a pretty complicated design, and they were able to turn it around in two weeks. So obviously, they're not setting up a machine where they're going to weave this, where they're making going to make 10, 20,000 of these. This is a tiny run, and yet it was perfectly woven. So, how is it that the automation, the sort of meta tool idea that we've been talking about, how has that what what knowledge experience do you have that in other industries? Is it taking over in a way that particularly China seems to have the lead on a number of things like that, where there it wasn't that the US prices were higher. There was that product wasn't available in the United States.

SPEAKER_03:

Yeah, it it's interesting. And obviously, there's there's layers of technology and culture and business and politics on top of it. Um, I assume what they're doing in that case is probably uh using automation, so some kind of computer control for the um for the the weaving machine. Uh, if you know your history of either computers or manufacturing, you know that Jacquard uh was the inventor of of some of the early looms in France. And the punch card actually came from the uh the code that contained the pattern of the weave and fed it to the the machine. So the way that the different um uh weft and and threads are are interwoven to make different patterns was done because controlled by by a punch card. So, of course, today your your design would have been digitized and fed to an automated uh loom of some sort that was able to do that. Now the question is why is there a no American company that buys that loom and does that work? Um, it seems like a business opportunity to me. But I'm just a manufacturing engineer and not a not a businessman.

SPEAKER_01:

Yes, but it seems like you could uh you don't have to build, you should be able to obtain the machines because absolutely, yeah. One of the reasons that paisley prints were so desired and so expensive was that it was very difficult to set up a loom to make that because it wasn't symmetric. You had you had to do it, you had to do it by hand. It's strange that that was just an ostentatious way of being able to demonstrate wealth because it it was completely the same fabric, but it it was much harder to create. So it was sort of a Veblinian uh conspicuous consumption. But with the ability to because what what they needed for these ties was to have an extremely high-resolution image, which they then just scanned, and then they could reproduce on basically any scale. And my impression is it took almost no time because the the machine would just do what you had told it to do, and so that raises if we combine that with AI, what's going to happen to engineers who used to make tools? Well, yeah, I mean, uh these are machines making themselves, right?

SPEAKER_03:

Yeah. So you think of a uh a machine tool that's computer controlled, and in the past we would have a a uh an NC programmer and numerically they even use the term numerically controlled because that was novel when we started doing computer control, but you have a programmer who has to identify the the machine path uh for a milling machine or something, and of course, a lot of that is going to be able to be automated. Um there's uh a lot of use of parametrics in computer-aided design. So if you want to go from using from building a small car to a large car, you can adjust one or two parameters and the whole thing will scale up or down based on a few clicks. Um, the latest generation of automobiles have not just a common platform like the K car, where it was generally the same size vehicle, but they've discovered that they can make the module around the driver and the controls and the instrument panel uh one one giant module, and then everything around it can be different. Um, so you only have to engineer that the the most difficult part of the vehicle once, and then make your whole product lineup around that that little core of systems and technology.

SPEAKER_01:

So I hadn't thought of that before. The and this may just be a uh stereotype, but my impression is that in design, what people did for cars was create a clay mock-up and actually used uh shaped the clay in a way that took forever. And it was possible to use it with uh if you have a CAD program, you have ways of visualizing even in 3D, but it's not the same. So, how are new cars designed now?

SPEAKER_03:

Yeah, well, they definitely still use clay mock-ups because to really understand things like how the light will reflect off it and to really get that feel, they they do go for the physical clay mock-up, but it's not carved by humans, it's carved by a computer arm with a cutting tool. Uh, and so um you can imagine most cartoons today aren't hand sketched, they're sketched on a tablet, and it's the same with cart.

SPEAKER_01:

So I can make cartoons. I I now make cartoons, which means that a lot of cartoonists are gonna, well, a lot of traditional cartoonists are lose their job. Another way to think of it is we're all cartoonists now, I suppose. But the the the so they they still make clay mock-ups.

SPEAKER_03:

Absolutely, yeah. And they will even, you know, when they get further along in the process, they'll even um paint them or coat them so you can see, you know, different the color and not just the the clay. But yeah, that's still part of the process. Um, but it it all comes from electric, you know, electronically done design work. Uh, and then you know, there's still an art to it, and there's still skill involved in in making some of those decisions. The other piece of it is you you have this interplay between the designer who's looking for something that's attractive, and the manufacturing engineer who's looking for something that's feasible to stamp repeatedly and affordably.

SPEAKER_01:

Well, and the parts, the parts will fit inside it. So it it it is nice to be able to have enough room to put the engine in.

SPEAKER_03:

Yeah, yeah. Um the uh the Tesla Cybertruck is kind of famous or infamous for these flat body panels. And of course, a flat panel is not as rigid as one that is is curved or or has some kind of a a jog to it. And so um, you know, I'm not quite sure how they did that, but that was a really challenging thing from a producibility standpoint because essentially those body panels are not very rigid. Um, now they're stainless steel, which makes them a little bit, you know, the material properties make up for that. But if you look at any other vehicle on the road, the the curvature and the angles and things like that, they uh serve an aesthetic purpose, but they also serve a structural purpose.

SPEAKER_01:

It is a very distinctive look, and in a way, it is sort of retro throwback because it was easier to make, it was weaker, but it's easier to make a flat panel than a curved panel. And so even it's a sign of wealth.

SPEAKER_03:

Yeah, yeah. It but even if you go back to the the 80s cars where they were quote unquote boxy, they always would have features in them to give them a few.

SPEAKER_01:

Well, a little a line of stuff indentation which would make it much more rigid.

SPEAKER_03:

That that's called a dart, and and that would be again, it's it's a structural component. And you know, you put a pinstripe on it or maybe a molding, but it's really about how do we keep it from wobbling.

SPEAKER_01:

Well, and it it's dramatically stronger, it's not just a little bit, but dramatically stronger. So you can see that. Well, um, we've covered a number of topics. Did you have other examples or things that you had thought about? The problem with the topic of transaction cost is that uh it is pretty encompassing. And that actually has been some of the point that I've tried to make with this podcast.

SPEAKER_03:

Yeah, it's funny. I had a conversation in preparing for this with my wife. And I was talking, we were, you know, we're talking about specialization, right? And I talked about the example of a production line being all these individuals with a very specialized task. And I mentioned, you know, Henry Ford and the Model T. And of course, our perception is that someone working on that production line doesn't need any skill. They can come off the street. They're not a specialist, but what's specialized is the tools they use, the way that the processes have been designed and chunked into these tiny little pieces. Um, so the fact that you have that 59 seconds of work to do, and it's the same 59 seconds over and over again, um, you might not need a lot of skill coming off the street to do that job. But other skilled people, other specialists have designed the work for you in order to make it so that you can you can work in that way. And again, going back to that um labor relations question, one of the things that that Toyota did extremely well was to listen to those workers who who knew that 60 seconds of work very well and take their feedback. And so that was one of the differentiators. And I think a lot of companies have have leaned into that, listening to their their workforce more and and learning from them.

SPEAKER_01:

Well, that that's that's a very Hayekian point. Perhaps I should have interviewed your wife instead, but it's too late. No, I I do thank her for her insight. Um, Russ Roberts recently did, obviously, Russ is my hero, so the the I I often bring him up, but he recently did a podcast about production lines in World War II and the making of B-24s, among other things. And one of the things that those factories had was a lot of rivets and a lot of the connections that they were making. And the airplane bodies are not flat, they're curved because they're much stronger. And if if you rivet it correctly, you've got a tube, which is a very strong structure, even if it's very light. But the rivets have to work and they're going into combat, so you can't fix them. Um, the problem was that the rivet guns and the setups were made for 180-pound men, not 110-pound women. And most of the men were gone. And so they had to redesign this. So, not just people who were not experienced riveters, but people whose upper body strength and the ability to do this repetitive motion with a 40-pound rivet gun was just different. And so they had to rethink the production process from the ground up. But the result was that it was actually far more efficient. They didn't change it back, even when the regular workers came back from the war. And so it does make you think that maybe sometimes uh having to rethink the production process might actually be an advantage, although nobody wants to have a war or some the depression or some big problem for that. But the the problem with production, the the kind of example that you used about uh the the Japanese, it is really interesting in those terms about designing the process itself so that it requires less skill. But it did it raises a question for me won't those workers get paid less? Because you can now hear someone off the street.

SPEAKER_03:

Well, and and so that's you know, this is kind of the story of the early early Model T and Henry Ford, right? Okay, you can you can pay them less because uh they're unskilled, but what you quickly find is there's other aspects of the job that are unpleasant because you're doing the same thing repetitively and you're not being treated very well. There's some ergonomic problems with doing the same thing every day. And so um Henry Ford had terrible turnover in his factory. He couldn't get people to come back, and so what did he have to do in order to keep people on the line? The$5 day. He had to actually pay them not a little bit more, but significantly more than than uh any any other company uh in order to get that labor force. But they were productive enough because they were turning out so many vehicles that he could afford to do that.

SPEAKER_01:

I see what you did there. He could afford to do that. Well played. Well, born in Dearborn. That that that is a terrific place to end because I I think people often tell the story that uh Henry Ford was this magnanimous forward-thinking progressive, and it wasn't. He was kind of politically a troglodyte. What he did see was that it was in his self-interest to pay the workers more because it would be more profitable. And a lot of changes in the workplace, work hours, pay, working conditions, have been the result of the fact that it's actually more profitable to do those things. And that I think is an interesting aspect of transaction cost. So, Andy Wagner, thank you so much for having been a part of the answer is transaction cost today. Thank you for the opportunity. It's been a fun talk. Whoa, that sound means it's time for the twedge. A wife asks her husband, an economist who taught at an engineering school, Darling, can you please go to the shop to buy one pint of milk? And if they have eggs, get a dozen. Off he goes. Half an hour later, the husband returns with twelve pints of milk. His wife stares at him and asks, Why in the world did you get twelve pints of milk? He said, Well, they had eggs. It's like writing code that syntax matters. She said, Go to the shop and buy one pint of milk, and if they have eggs, get a dozen. This month's letter. Mike, just finished the innovation is the exchange itself. Great podcast as usual. Two points. First, you keep doing your podcast your way. I love listening, and if we lived closer, I think I would enjoy talking to you and being your friend. I love I'd laugh through each episode and walk away with much to ponder from each one. Two, I'm a government bureaucrat. I'm a civilian in the U.S. Air Force conducting tests and evaluation of new weapons. I used also used to own my own business, so when I hear your Adam Smith and Marxist references to perpetually expanding bureaucracy, I both recognize it and try not to get upset that your observations are not universally true. I'm actively trying, I am actively trying to reduce the quantity and cost of the services to focus on what only we can do and what is most needed for the future warfighter. It's tough to persuade my subordinates to let go of the old and irrelevant, but I'm trying to run this show like a not-for-profit instead of a black hole. Keep up the good work and keep it fun. CM. End of letter. Well, thanks, CM. And uh you actually make both a valid protest and my anticipated response. Uh it's absolutely true that many people, perhaps most people in the bureaucracy, are actually trying to do a good job. They're they're called public servants, and they are public servants. I myself worked for the U.S. federal government. Many friends who have and still do work there. But as an empirical matter, it's just true that the size of bureaucracy almost always increases. Now, perhaps it doesn't increase as much as it would if there were not people like you out there doing their best to keep it from rising any faster. But in terms of the size of government spending by any standard, including the standard of as a proportion of GDP. Bureaucracy and the amount, the proportion of GDP that the government is grabbing is growing dramatically. It's gone from 30% or so in 1960 to 40% of GDP now. And so it's not just that it's growing in absolute terms, it certainly is. It's growing in relative terms. More and more of society is going inside the maw of government spending. Now, the problem with military spending is we want it to be wasteful. We literally want military spending to be wasteful. And so in peacetime, we're always trying to find ways to cut it. And then as soon as there's a war or a conflict, we blame people in the military for not having anticipated that conflict and spent even more a long time ago. So the problem with the military is that you're darned if you do and you're darned if you don't. I'm happy to have that perspective, so thanks for writing in. The book of the month this month is published by Liberty Fund. It's written by Jacques Rouef. That's R-U-E-F-F. The title is The Social Order, and it's about how the money and banking system constitute a social order or can cause disorder. It's frankly an odd book and difficult to read, but it does make you think about things differently. It's very serious about the role of money in organizing and enabling commerce. Next week, on Tuesday, October 21st, we'll have the sixth episode of the Adam Smith Wealth of Nations series and take up book three. Talk to you then.