Episode 2: Arvind Panagariya and Shruti Rajagopalan Talking Trade

Arvind Panagariya
Shruti Rajagopalan

Arvind Panagariya and Shruti Rajagopalan Talking Trade

Episode 2: Infant Industry Protectionism

RAJAGOPALAN: Welcome to the discussion series on free trade and liberalization as part of the 1991 Project at the Mercatus Center. I’m Shruti Rajagopalan. In this conversation, I will be talking trade with professor Arvind Panagariya, who is the director of the Deepak and Neera Raj Center on Indian Economic Policies and the Jagdish Bhagwati Professor of Indian Political Economy at Columbia University.

In the past, he has served as the first vice chairman of the NITI Aayog in the government of India and as the chief economist of Asian Development Bank. He is the author of a number of books. For today’s conversation, we will focus on his recent book, “Free Trade and Prosperity.” Welcome, Arvind.

PANAGARIYA: Thank you, Shruti. A pleasure to be with you.

Infant Industry Arguments

RAJAGOPALAN: I’m so thrilled to start part two of our conversation. When India embarked on protectionism and import substitutions and near autarky, in the ’60s and ’70s, it was the orthodoxy for developing countries. There were a lot of arguments for protectionism that were given, in those times. The most salient of those arguments was that infant industries need to be protected, especially in developing countries.

PANAGARIYA: There has been a checkered history here because the argument is a very old one, more than 200 years old, actually. Even in the 17th century, in some form the argument had been made. We trace it to Alexander Hamilton. I think that’s where we have a clear statement of the argument, at least, in its most rudimentary form. Then, of course, later on, John Stuart Mill had a paragraph blessing the argument.

Friedrich List wrote a whole book—John Stuart Mill was around in 1848—Friedrich List, who gave the argument a slightly different twist, applying it to the entire manufacturing sector. He was much more comfortable with protection to infant manufacturing sector, even for several decades and so forth. He wrote in 1856, a very influential book. Then Charles Francis Bastable in 1887 wrote on it, he set a different side to it. All this while, this was not taken very seriously, at least.

I mean, it was seen that that’s a possible interesting argument, but industrial countries were already way out of the range where you would say that any of their sectors were “infant” for protection. As a result, it didn’t play any major role in terms of countries invoking the argument to protect the sectors. After the Second World War ended and the developing countries began their process of development, the argument acquired new salience.

That is the time James Meade reexamined it very carefully in his book for which he actually eventually won the Nobel Prize, the first Nobel Prize in international trade. The last, I would say, very important contribution to it came from Robert Baldwin who in fact questioned the manner in which James Meade had reformulated the argument. James Meade was the first one who made the distinction between external economies of scale and internal economies of scale. Baldwin came along and said that, look, even when economies of scale are external, the argument really doesn’t hold. That’s a rough history and we can take it along.

RAJAGOPALAN: What exactly is the argument of infant industry protectionism, aspects like what is an infant industry? How long should the protection continue? What are the logical aspects, if any, of those who favor protectionism to further infant industries?

PANAGARIYA: First, I’ll give you what the basic idea is, and then I’m going to poke the holes in that idea from different angles. The basic idea is that, look, there are these learning economies of scale that when you carry out production activity, there is some learning process and that learning process leads to decline in the cost of production in the future. Production today basically lowers the cost of production tomorrow. That’s the working hypothesis initially.

What happens is that if today the production costs are higher than the world prices of the same product, then the industry is not able to stand competition from those outside producers who have had a longer history of production and therefore they have already done their learning. Nevertheless, tomorrow, if I actually allow production activity in my own country to take place, then tomorrow my costs will decline sufficiently, the resulting profits will more than outweigh the losses that I incur today in net present value terms.

Since the infant is unable to stand competition from the established foreign manufacturers today, it never comes to life, and so a socially desirable manufacturing activity fails to emerge. That is the premise. You say that, well, then if we give temporary protection today to the industry, then it’ll come into existence, it will do its learning and it will then become competitive tomorrow. That is the basic argument, basic set of circumstances laid out by the argument. Here I might point out that, originally, when we go back to Hamilton and Mill, they had not seen this nuance at the time that ultimately, even for the activity to be socially desirable, today’s losses have to be recovered in net present value terms by profits tomorrow.

The mere fact that the industry is profitable tomorrow is not sufficient for the industry to be protected justifiably. Those profits have to be large enough to more than offset the costs that I incurred today or the negative profits that I incurred today. That was the point that Bastable made afterward in 1887. Largely, by the time we come to the Second World War, the argument stands where Bastable has left, and that’s how I have just outlined it to you.

RAJAGOPALAN: I guess the crucial assumption is one about learning. I want to take you back one step behind before we get to trade. How do firms learn? Is firms’ learning internal? By doing, is it external? What is a good model to think about firms’ learning to then bring it in as an assumption which bolsters these infant industry protectionism arguments?

PANAGARIYA: Shruti, what I’m going to do is, I’m going to use a highly simplified example. I want to be very specific in the way we are specific in economics models. I’m going to lay down a very simple model which captures the essence of the argument and also allows me, therefore, to pick out where the problems are. We can pinpoint precisely where the problems are.

I’m going to in the next half hour simply lay out this and step by step make a distinction between internal economies of scale, external economies of scale, then also the sources of those economies of scale, how if it is an externality it travels from firms today to firms tomorrow. All that I think matters, and so let’s take a very clear simple example.

The example may appear overly simple, but that’s only done for keeping the exposition clear. None of these assumptions is actually necessary. I can generalize it. Generalization is not an issue here. All the simplification I’m doing is to make it clearer for the listeners to be able to see where the problems arise and how the argument is made in the first place.

What we’ll do is we’ll think in terms of a two-period model because it’s about learning. We have to have at least two periods. We can have multiple but we’ll simplify. We’ll just say two periods, that there is today and there is tomorrow. We will also say that this is a particular product produced domestically in the country where we are considering infant industry protection. Demand is fixed. We’ll just say the product is a widget and there is demand for 1 million widgets today, and there is demand for 1 million widgets tomorrow. Very simple demand, it’s a fixed demand.

If you want to think in terms of the demand curve, it’s a vertical demand curve with price on the vertical axis and quantity on the horizontal axis. Next, we say we also will rule out any exports by assumption of high transport costs. I want to think of a very simple situation in which the production required is only 1 million. As I said, none of these assumptions is necessary to make the argument but they’re useful for simplification.

That’s the basic domestic structure that we are going to work with. Total production activity to take place is 1 million widgets today, 1 million widgets tomorrow on the delivered import price. So we will also say that there’s a possibility of importing widgets, and the import price is fixed both today and tomorrow at $90 per widget. Finally, the per-unit cost of production: What is the per-unit cost of production at home? Today or period one, the per-unit cost of production is $100, which is more than the $90 delivered price by the foreign manufacturers, the import price is $90.

Today the production cost is $100, and therefore, it’s unprofitable to produce, and no firm will come to manufacture just based on that fact. In period two, tomorrow, the cost because of this learning, economies of scale, drops to let’s say $79. This is the learning effect of manufacturing activity in period one. At $79 tomorrow, there will be a profit, therefore, of $11 whereas the loss today is $10. We’ll also say finally one more technical assumption, that the social rate of discount is 0%.

Tomorrow’s income is as good as today’s income. Normally, social rate of discount is a positive number. It could be 5%, 3% or 2%. Again this is a simplification. These numbers are simplified by making the assumption of a zero discount rate. We can compare tomorrow’s $11 with today’s $10 and say that, well, loss of $10 today is more than offset by $11 income tomorrow. Therefore, the activity is a socially desirable activity. This is a project that the country should undertake because the benefits in terms of net profit in net present value terms are positive.

That’s my basic structure. If we do the cost-benefit analysis it’ll tell you tomorrow my profits are $11, today it’s a loss of $10 and because my social rate of discount by assumption is zero, on net, I’m making a profit or the benefit is $1 million. $1 on each widget, and on 1 million widget demand, total $1 million net benefits. Widget manufacturing at home is a socially desirable activity. That’s my starting point.

RAJAGOPALAN: Now, in this starting point the assumption that is being made is one that learning in time period one, is going to reduce the cost in time period two, such that overall there’s a benefit by keeping the industry protected, right?



RAJAGOPALAN: Now the second part of that question is, is that assumption a sensible assumption? Is that in fact how firms learn? Is all learning internal to the firm, or can firms also learn through competition?

PANAGARIYA: At this point, we are going to lay it out as the proponents of the argument. What you have said is extremely valid and important. When we come back to ultimately round off the discussion of the argument itself once I’ve gone through the conceptual basis on which the argument is made, then what you pointed out will be very important. In practice, I completely agree with you. Let’s come to that toward the end.

First what I’ve done actually is said, in fact, nothing about how the learning is taking place. It’s the production activity, it’s something which is leading to the learning. The next step you already anticipated. The next step is then to ask the question whether this learning is internal to the firm or it is external to the firm. We’re first going to say that learning is internal to the firm. What that precisely means is that when the cost declines tomorrow, the cost is going to decline tomorrow only for the firm which already engaged in production activity today. The new entrants tomorrow cannot come in and produce at the same cost of production.

That is what the internal nature of learning means, that whatever learning has happened, it may have happened because there was some innovation. It may have happened because the firm actually trained the workers, workers acquired skills. Whatever is the source, which we will actually identify more explicitly later on, right now all we are saying is that whatever learning takes place, takes place within the firm and it cannot be taken advantage of by any new entrants tomorrow. This is the assumption. Well, if that is the case, then what we know is that the per-unit cost of production will fall tomorrow only for those firms or that firm which produces the product today.

Then the firm knows that when it goes into production activity, it is going to incur a loss today of $10, but tomorrow it will make a profit of $11. It is only a matter of finding a way to cover my losses of $10 million today. There is a $10 per-widget loss, which on a million widgets is a $10 million loss. I just need some way to finance, basically, that loss. I can go to a bank, produce my blueprints of production to the bank, and borrow money from the bank to finance my losses today of $10, and then tomorrow, when I make the profit of $11 million, I would pay back the $10 million to the bank.

Here, again, if the capital markets are undistorted, in that case, I can borrow, given that the social rate of discount is zero, the interest rate will also be zero. This is again an assumption about the financial markets, about the capital markets, that with no distortions in the capital markets, if the social rate of discount is zero, then the interest rate is zero. Tomorrow I have to pay back $10 million to the bank, and I go back pay them back $10 million. There I have got the net profit of $1 million.

There is no need here for the government to step in because my profits tomorrow offset my losses today, and financial markets allow me to borrow and then repay the loan. In this case, with internal economies of scale, no intervention is actually required. That’s what James Meade pointed out. In 1955, James Meade pointed out that this was the next kind of big step from Bastable. Bastable had not made this distinction between internal and external economies of scale. Meade did. As long as the economies of scale are internal, no intervention is actually required.

RAJAGOPALAN: This obviously begs the question that if we took away that all learning is internal to the firm, if we tweak the assumption a little bit—and let’s not bring in free trade just yet, let’s just keep it to the domestic economy. If we assume, as is more realistic, that actually incumbent firms tend to share their learning with new entrants, how does that impact this argument for infant industry? This is still very much within the domestic market. We’re not yet introducing outside entrants to the market outside of that domestic economy.

PANAGARIYA: It may be clear here that, obviously, the possibility of outside imports is there because that’s how we are determining the $90 price. The domestic firms sell the product for $90 precisely because if they were to raise the price above $90, then the imports will beat it out. Although there are no imports because it is not even required, the domestic firm is able to actually compete with the imports at $90 as long as the economies of scale are internal.

Now, let’s turn to the possibility of external economies. What does it mean for the learning economies to be external to the firm? What externality here means is, conceptually speaking, that although the firm that produces today is the firm that entered the market today, but whatever learning it takes place—now, we are not saying what is the source, what’s the reason—but whatever learning takes place and whatever cost reduction happens tomorrow becomes available to other new entrants as well.

What is assumed is that for whatever reason, maybe the workers break away from this firm today, or something of that sort, and the productivity is gained by the workers and they go to the other firms tomorrow, or maybe the firm undertook some kind of innovation, but innovation becomes also available to the others tomorrow. For whatever reason, the cost reduction happens not just for the firm that is an entrant today, which we will call the incumbent firm, but it is also available to the new entrants tomorrow.

Now, what will happen in this case? What that means is that when tomorrow production activity begins, costs will fall to $79, not only for me, for the incumbent firm, for the firm that did the manufacturing activity in period one, but for all the new firms that enter also in period two. If that is the case, then all those firms have no losses to recover from period one because they were never producing anything in period one; they can produce at $79. When they enter and there is competition among them, the price itself will drop to $79.

This is the free ridership problem here that because the cost declines for anybody who wants to manufacture tomorrow because the innovation or the workers, more productive workers, become available to everybody. Cost declines for everybody to $79, and then competition among them drops the price to $79. The firm today, the incumbent firm which did the initial manufacturing activity whose production activity led to this decline in the cost of production tomorrow, simply cannot recover its losses in the period one.

It lost $10 million in period one, whereas with economies of scale staying internally, it could recover, in fact, make a profit of $11 million in the second period and more than recover the losses in the first period. It cannot do it now when it comes to period two. It will never enter the market in the first place. It will say, “Well, I know that tomorrow everybody’s cost will decline to $79. I will have to also sell at $79, and if I have to do that, then I cannot recover my losses in the first period, so there is no point. I will not enter the production activity.”

This is where then the legitimate argument is made that this socially desirable production activity will fail to emerge because markets failed in a sense. There is a market failure. If I introduce protection in period one, if I introduce a $10 per-widget tariff on imports, then imports today will also come at $90 plus the 10% customs duty, and therefore will be sold at $100. That will make my domestic firm today viable. The domestic firm, or number of firms, that choose to enter the market to undertake production activity today will see that although their cost of production is $100, imports will also be coming in at $100. Therefore, they can compete with them, so production activity will happen. As a result, of course, the protection will make it possible. Tomorrow you can remove the protection because learning would have then taken place. There is temporary protection which makes production viable, and it allows a sector which is socially desirable to actually emerge.

This is the essential logical argument that is made by the infant industry protection proponents. Papers have continued to come. We have an original paper by Pranab Bardhan which made this argument. Then, afterward, there is also around, 2003 or 2004, there is a paper by Marc Melitz, for instance, which also essentially exploits this particular argument.

Before we get to the prospects of exports and all, I still want to take an issue with this argument. This is really an issue that Robert Baldwin took with it way back, in fact, and pointed out that once you examine this argument more carefully—James Meade was the one who resurrected with this learning external economies—that once the learning is external and other firms can also take advantage of the reduced costs in period two, then indeed, a case of protection can be made.

Now, what I want to do is ask the more fundamental question that, what is it that allows the other firms, new entrants in period two, to actually be able to benefit from this learning that is happening from the production activity of the incumbent firm in period one? The whole question here is that, how is this externality traveling from the period one incumbent firm to the period two entrants? It almost looks like magic that somehow, although it was the incumbent firm that undertook the production activity, everybody is actually able to produce this product at this lower cost of production.

What is the source of this externality? Ultimately, where is learning taking place? How is this learning spilling over to these other firms that enter in period two? That is simply not made clear at all by this kind of mechanical argument. This is a mechanical argument that says that whatever is the production activity in period one, on account of that, the cost of production declines for everybody in period two, but that’s quite magical.

That really is the critical crux of the matter. I want to now go further and make this externality or the source of the externality a little more explicit. Let us say that workers employed in period one, by virtue of the production activity, acquired new skills. These are skills that they acquired on the job, and that raises their productivity in period two, and it is that increased productivity of the workers for period two that lowers the cost of production to $79 per widget.

Now we make it explicit that the source of externality or source of this learning is skills that are acquired on the job by the workers. We say now if period one firms employing these workers can prohibit them from offering their services to other firms in period two, learning will remain internal to the firm. If they could write a contract with the workers that you are going to also remain with me in period two to the extent that I need your services—only then I’m going to employ you—then, of course, you can internalize what could be potentially an externality.

In that case, of course, we get back to the internal economies, and there is no case for infant industry protection. Suppose that the firm is not in that position to actually write a contract with the workers and there is this possibility that these workers tomorrow might actually go in and work for the other firms, which would then drive the price tomorrow to $79 per widget. That would then create the problem of externality. That was the basis of the case for protection in the first place.

Now, the question is that even if there is that possibility, is protection necessarily required? Well, I’ll argue that—and this is what Baldwin’s argument was as well— even then, actually, the firm has the possibility of avoiding the losses that might accrue. How? Well, the firm knows that it is imparting a particular asset, the skills. That firm, when it employs these workers today, it knows that it is imparting them a skill or it’s imparting them an asset, which has a market value of about $11 because the cost will go down to $79 and the widget would still be worth $90 if is to compete with imports.

There is an $11 value per widget that’s associated with the skills that today’s firm is imparting to the workers. Recognizing that, what the firm will do is it’ll offer the workers a wage that is lower in period one such that the per-widget cost actually is also lowered by $11. It says that, “Well, look, I’m giving you some assets which will be valuable to you tomorrow. And therefore, I’m going to offer you a wage which is lower than the market wage today because tomorrow you’re going to get the wage which is higher than the market wage.”

The difference is exactly $11, which is the worth of $11 per-widget manufacturing, which is the benefit you’re getting in the form of this skill or this asset. Today, then I can lower my cost, in fact, by lowering that wage for $11 per widget, and that, of course, brings my cost to—$100 was the original cost—but because I’m able to pay low wage by $11, my cost comes down to $89. And that, of course, means that I can sell the widget at $89 per unit and outcompete the importer. I’m viable today, and then, of course, tomorrow, the workers go, price drops to $79, but no sweat. I incurred no losses in my first period in the first place. In fact, I can sell the product actually at $90 because I know that imports will come at $90. My cost is $89. I’ll make a profit of $1 million and I’m happy and all set. If the source of this external economy is skills that the workers acquire, then clearly logically the case for any protection collapses. That at least is one example where once we specify the basic microfoundation of the externality, we find that there is no case. That’s Baldwin’s first point.

Now we come to the second possibility. This is where we say that, well, suppose the cost reduction in period two results from an innovation by the incumbent firm. The firm actually invests in some sort of innovation, and that innovation in period one is the source of the cost reduction tomorrow. That’s what we are going to now look at. Rather than this source of externality being the skills of the workers, we’ll say that there is some sort of process innovation that the firm is going to undertake in period one, which will eventually lower its cost of production in period two.

Again, to consistently continue our example, let’s say that the innovation cost is about $5 million. The total cost that the firm is going to incur on the innovation, which will reduce my cost tomorrow, is $5 million. Then the regular production cost of a widget is another $95 per unit. I’m choosing the numbers so that the total cost of production per unit remains $100.

I’m just choosing those numbers. One could reformulate it with different numbers, but this is the easiest set of numbers that the part of the cost I’m incurring today on manufacturing the widget is my cost of innovation. $5 million I spend on innovation and then my remaining manufacturing cost is $95 per unit. Obviously, eventually it gives me $100 per-widget cost today. Now you can say that the externality will be of the kind where this innovation will spill over; that innovation becomes available to everybody tomorrow, so the cost will fall to $79 for everyone tomorrow.

Therefore, I’m still saying that the innovation actually, although done by this firm, is external. If it is internal, then there is no issue. Then we know that the firm can keep the innovation within itself, and, therefore, tomorrow the price need not drop to $79, but I’m ruling that out. That innovation is actually something that will become available tomorrow. Somehow the firms tomorrow will be able to observe something or the blueprints cannot be kept secret. Everybody tomorrow can produce a widget at $79. The externality problem remains.

The question I want to now ask is on temporary protection. You know that this will not work in terms of manufacturing or the sector emerging. It will not happen because if the innovation will become available tomorrow to everybody, then we know that product will have to be sold at $79, whereas the losses to the firm would be $10 million in period one, no profits to be recouped in period two. If that is the case, without some intervention, this manufacturing activity or even the beneficial innovation will actually not take place.

The question we now ask is—and this is a critical one—can temporary protection solve this problem? Well, suppose we impose a tariff of $10 per unit in period one, which leads to a rise in the import price to $100. That’s our contention that this might solve the problem. Now let’s follow this up. Now, suppose that this is done and the import price becomes $100, which is exactly what it costs me to produce widgets today, inclusive of the cost of innovation. That is where we are. Now the incumbent firm has two options.

It can invest in the innovation and just break even. Then it just breaks even. The cost is $100, but imports are also coming at $100 because of the $10 tariff. In period two, of course, the price drops to $79, profits are zero. You break even. It’s viable for sure. That’s one option for the incumbent firm. Now take option two. What is the second option for the incumbent firm? Instead of undertaking innovation, it says, “I’m not going to invest anything on innovation. Why should I? Instead, I will just produce the widget at the per-unit cost of $95 and now I’ll sell it for $100.”

There is a tariff, there’s a custom duty, so imports cannot come for less than $100. If I don’t do innovation, then my production cost is only $95 and I’m going to walk away with $5 million profit in period one. In period two, of course, tariff will return to zero and nobody will produce anything because no innovation happened. Therefore, there is nothing, the cost is not going to come to $79, but that is okay. Tariff will be removed, my cost will remain $95, imports will come in at $90. I will not do any production activity. I’ll just make my profit and speed it and go home.

Well, the firm will naturally choose the second option here because in the first option it barely breaks even, but in the second option, it walks away with a $5 million profit. No innovation will actually take place and the outcome will be a net social loss because if the innovation had taken place, then on net, the company would have been better off. In this case, it’s basically a social loss because the consumer got hit by the high price of which only $5 million was actually captured by the domestic firm but the other $5 million that the consumer paid in higher prices actually is a loss.

In effect, when innovation is costly, even if it spills over, it’s an externality. Therefore, protection will not solve that problem. You have to do something else. I mean, this is a situation that requires some intervention. The obvious intervention here is to grant a patent to the firm for the innovation, so that patent actually allows the firm to internalize the innovation itself. And then, of course, the production activity becomes feasible and the sector emerges. Alternatively, the government could directly subsidize the innovation. It gives $5 million to cover the cost of innovation. This is a kind of R&D subsidy. It can do that.

That will solve the problem, but clearly what we see here is that, logically speaking, protection cannot solve that problem. I think that really is the bottom line. That was the point of Baldwin. This is where he wrote actually—I’ll read out. He says it in very clear terms. It’s quite remarkable how the subsequent authors who wrote their papers on infant industry protection completely missed Baldwin’s point. They kept writing these models.

Magically, the externality travels from the incumbents to the new entrants tomorrow. That completely flies in the face of reality. What is the process through which the externality is traveling? Baldwin actually saw it all. He wrote it very clearly. I want to read out a quote from Baldwin in his 1969 paper. It’s more than 50 years ago.

He wrote that:  

He couldn’t have been clearer than that. Oddly enough, the future authors who made these infant industry argument cases, for protection, that is. Obviously, Baldwin very carefully says, “The infant industry argument for tariff protection is worthy of its reputation.” He’s very clear that he’s not questioning the intervention. There is some other intervention like the patent or subsidy on innovation, or it could be even subsidy on the skill formation and so forth. What he is questioning is that protection is not the solution to the problem. That is his point, and it is completely valid.

RAJAGOPALAN: No, this is an incredibly helpful and clear exposition. My intuition was going in a slightly different way, which was, does this assumption make sense? What you have told us is, leaving that aside, let’s take the assumption on merit. Even if this assumption actually holds, and we give it its due, protectionism is still not going to solve the problem that the assumption is hoping to solve, right?

If I can just look back to the older argument, which is, now let’s say that in your model, the importers are able to produce the same widget at $90. For whatever reasons, either because they were entrants to the market sooner, so they’ve already done the learning, they’re now at the point of reaping the benefits from learning. So they’re able to price their widget at $90. That’s the group that one needs to compete with if one enters the domestic market.

Now, if learning is only possible by doing but there’s an externality in the future, that is, once the firm has learned, other entrants to the market can actually learn from that firm, why doesn’t that translate to learning from importers? Or why is that not a reasonable assumption? Are these importers on Mars? Do they not speak the same language? Can workers not travel and train someone in the domestic industry? Just regular things that one would see all the time. Why has this assumption stuck so long and hard, that whatever they assume for the domestic industry, the same symmetrical assumption is not being made for the global industry?

PANAGARIYA: Absolutely. I think that’s a million-dollar question. If you look at really Koreans, for example, who did so well, and we talked about in our first conversation, they were often actually bringing in these imports and then doing reverse engineering. They were really learning through reverse engineering how to do it at a lower cost. Absolutely right.

Then, there is a whole literature of diffusion of technology through trade. And what the infant industry argument is doing is completely ignoring that. As you have quite correctly pointed out, there is an asymmetry in saying this. That if you can learn by observing your own manufacturers, why can you not learn by observing the manufacturers that are outside?

Actually, that gets to the broader point, which those of us who are on the side of free trade have often made that, look, in the end, how are you going to learn the most? By competing against the best in the world. If you shield yourself against the best in the world, then you’re not going to do that kind of learning. The example I frequently use from cricket that if you are trying to produce lots of Sachin Tendulkars and Virat Kohlis and all the fantastic cricket players, are your chances better when you play test cricket and all the international people, or if you play only domestically and the Ranji trophy or your county cricket matches? The answer is quite clear that you’ve got to play test cricket.

RAJAGOPALAN: I would actually take your argument one step further if I may—the cricket analogy—which is one you have to play test cricket with other test-playing countries. But even if we have an IPL model, you benefit more by having global immigration and mingling and trade in IPL because now every team—whether it’s Bangalore or Mumbai or Calcutta or Delhi—all the teams in the IP League can actually get foreign players. You actually learn from the best players in the world, except now they’re on your side, which is another model that a lot of firms have used for training their own people. They usually attract the best talent from abroad. Someone who’s already benefited from learning by doing, pay them a very high reward or remunerations for these kinds of training services, and that’s how the process takes place in any world where there is some trade and immigration.

PANAGARIYA: No, no, absolutely. I think you can further generalize it actually because cricket is only where the human skills are involved. But when manufacturing is involved, there is also other technologies at work. That also extends to foreign investment. If you allow foreign funds to come in, they will bring in a new know-how, technology of production, new management techniques, a lot of the management and organizational activities that are specific to the particular industry. They will bring those. The domestic firms will then also learn from the foreign-invested firms in bringing their costs down.

Concentrated Benefits, Dispersed Costs

RAJAGOPALAN: You’ve told us the benefits of keeping the free trade channel open because protectionism is not going to get you what you want, and if you treat importers as symmetrically as you treat domestic competition, then you can also learn from people who are already in the global incumbents but in the global industry, and you may be able to solve the problem many different ways the moment we start generalizing some of the assumptions of your very stylized example. I want to talk a little bit about the downsides of protectionism.

The assumptions made in the model is that in time period one, any protection that is brought in will disappear in time period two because by then the firms will recuperate profits. Now, this, of course, assumes that the firms always do the right thing and are interested in earning their profits the correct way. If we now think about the implications of that, it becomes quite clear that the moment you bring in protectionism, it is not going to be in the interest of any incumbent firm to allow the protectionism to go away, not just in time period two, but forever in the future.

PANAGARIYA: Absolutely.

RAJAGOPALAN: That has a tremendous downside. That downside is millions of consumers in the domestic economy will now lose because it is in the interest of some firms, which once upon a time were considered infant industries, to continue that protectionism for a very, very long time to come.

PANAGARIYA: Absolutely. I think there’s time-inconsistency problems. Not only this, what may happen is that—I gave you the innovation example—if you give protection, the firm happily comes in and produces, enters production activity, but it chooses not to innovate. Similarly, if for example, your expectation in the model was not innovation but one of training the workers, suppose instead of innovation, it could have spent the $5 million on training the workers, and in the hope that the firm will train the workers and create new skills, the government may give the protection.

In either case, whether it’s innovation or costly training of the workers, once you give protection, once protection has been acquired, it becomes in the interest of the firm not to innovate or not to train the workers. Why waste your cost? Because the protection is already there and you’ll make bigger profit if you simply did not do this learning, whether it’s through worker training or through innovation.

You’re not going to do that learning. That’s exactly what happens. Then, in fact the argument you made actually comes out quite endogenously there. Then, the firm has already entered the manufacturing activity in period one and then it says that, “Hey, if you now take away my protection, then I’m gone. All these workers I’m employing will become unemployed.”

In the political space, that argument has a lot of salience as if these workers cannot find any other job elsewhere in the economy. The policymakers are very sensitive to that and say, “Oh, yes, we’ll extend the protection.” And so the infant never grows into an adult, but protection becomes self-perpetuating. There are very rare examples. What this does is it begs the question that you’ve got to worry about the effectiveness of the government—are there effective governments?—which gets into whole host of questions.

That, first of all, how does the government know that this sector is my kind of infant sector which after protection will become very cost-competitive? You’re talking about picking the winners, but what ability does the government have to pick winners in the first place? Even if it manages to pick the right kind of sectors, what ability does it actually have to credibly make the protection temporary?

You find very rare examples. I mean, you do find examples, they are very rare. Singapore, in ’50s—not ’50s but maybe mid ’60s or something, for a very short period of time, five or six years, they would do it. But what they were doing was every six months, they were monitoring the firms. They would say, “Okay, next six months, we’ll keep this import restriction on, and if you don’t do any learning, then protection goes away.”

They were credible in all, and they did actually monitor continuously, but that also with very low level of protection, protection itself was not particularly high. You might find some capable governments, and it appears they successfully did it. Truth of the matter is that logically as we have seen, there was no case. The case was not for protection anyway, maybe case for R&D kind of subsidy or patents and so forth.

Even when they say, a lot of it is post hoc fallacy. Ha-Joon Chang has made a career out of it, saying that Toyota got protected in 1930s and look how successful they became in the 1980s. That really calls for your imagination to connect the two events. Just because you did this before—protection was provided 50 years ago—you somehow become successful in 1980s. Question also is what kind of cost did you impose on the consumers? This is one thing, the cost that the consumers pay is simply ignored. In the political process, the consumers simply don’t count.

This is my constant battle on the Indian auto industry. On the one hand, by allowing foreign investors to come in, we did, without doubt, actually improve the efficiency of the auto industry domestically. But at the same time, this kind of 100% protection that has been granted to the auto industry for the last 30 years—and before that, of course there was complete prohibition on imports on automobiles—the result has been that the consumer has been paying one and a half times the price that customers outside India pay for the same car.

RAJAGOPALAN: Absolutely.

PANAGARIYA: After 30 years, what happened? Obviously, not enough learning happened, right?

RAJAGOPALAN: Your background is of course in trade. Mine is in public choice economics. Now if you start bringing in some of those assumptions and some of the models of public choice, then it’s very clear the story that you’re talking about, which is, this is a classic case of concentrated benefits and dispersed costs. In time period one, you will give the infant industry protection.

That’s going to go to one or two firms, they are going to be protected, they’re going to get all the benefits. The costs are obviously going to be dispersed both in time and across all consumers or everyone who would have used the imports. Now, because it is difficult for consumers to collectivize, because the transaction costs of organizing themselves are very high, it’s going to be very difficult for them in democratic politics to start lobbying for taking the protectionism away even though they benefit from it. But the benefits of removing protectionism are small and dispersed, but in the case of protectionism, the benefits of keeping the protectionism alive, they are concentrated for a few manufacturers in a given infant industry or any other industry. They actually also have the transaction costs or the cost of collectivizing on their side to ensure that they can keep the protectionism going year on year. Now, once you bring in the models of democratic politics, the endogeneity only once again reinforces itself, this becomes self-perpetuating.

PANAGARIYA: Right, absolutely. I mean, this is really the story behind the Grossman and Helpman kind of models, that the producers are able to organize. In the political process, it’s the profits of the firms, profits of the enterprises that count. This is what we economists call the consumer surplus, the benefit to the consumer, that simply doesn’t count for how the political process works.

This also is what happens, once the government becomes willing to protect the industries, selected industries, then everybody comes in, “I’m infant.” Everybody wants to claim that they are infant. Once granted protection, as we discussed, it becomes permanent. It’s not temporary. The case is always made for temporary protection, except for rare examples. We don’t have these cases where protection was actually granted on infant industry grounds, and it was also temporary.

RAJAGOPALAN: Actually, that’s a great point, because one thing is how the economists are writing the academic models, and the model specifies that in time period T plus one or T plus two or T plus five, the protection should go away. But I have read a lot of these policies on the legal side. We rarely, I mean, even in the case of a pandemic or something like that might be the exception, we rarely see sunset clauses when we offer protectionism.

PANAGARIYA: Absolutely.

RAJAGOPALAN: If they’re not built into the political process to have a sunset clause, right?


RAJAGOPALAN: There’s a remarkable government, like, say, Singapore, which is constantly monitoring. But monitoring is different from “Hey, we looked at this particular model, this model made sense to us. We’re going to translate it exactly as policy. Now we’re going to tell you that in time period T plus five, this protection is going to go away.” We have rarely seen a credible commitment like that on the policy end.

PANAGARIYA: Very true. At least for temporary protection, I could think of this one Singapore example, but for the rest, I have not seen any, for the point you are making, which is, that there’s a prior sunset clause in protection, that I’ve never seen. At least I’m not aware of any such sunset clause. But anyway, that sunset clause is because of the other reasons, not on infant industries. Those are like safeguards and all.

RAJAGOPALAN: Yes. I’ve actually seen the reverse, which is, we are going to remove tariffs for a very brief period of time. This happened during COVID. There were many countries, including India, by notification, they say, “Just for six months, we’re going to suspend tariffs on like PPE, mask, protective gear, things like that.” And after six months, all the tariffs are going to come back in full form. If anything, I have always seen sunset clauses erring on the other side but not to further the infant industry argument, as you have so clearly laid out.

Protectionism and Building Export Capability

Now, I have a question about, infant industry protectionism with respect to building export capabilities. Now, you’ve shown that either the learning is not going to happen, or if the learning happens, it can also happen by reduced wages or by absorbing the cost of innovation, in which case, you don’t need the protectionism in the first place as you showed in your stylized example.

Now, a lot of the protectionist arguments in India are that we need to protect the infant industry so that it can grow strong enough to compete with the rest of the world and become exporters, not just serve the domestic market by competing with importers. To build export capability, is there any merit in infant industry argument? Or does that also face very similar problems?

PANAGARIYA: See, if the argument is made in this way, that we are protecting them until they become competitive and then they become exporters, then what we have already seen is that that argument doesn’t work because the infant never learns. If it’s innovation, then it’s not going to do the innovation. If it is worker training, it’s not going to invest its resources in worker training. Of course, if the learning was internal, then no intervention in the first place was required. Ultimately, to become an exporter, first, you have to be able to actually outcompete the import price in the first place, and that here will never happen.

On the export side, a different argument is sometimes given that, “I’m going to give export subsidies to exporters” because they are infant and they need to learn the market, the foreign market, they need to learn. And so, initially, when they go in there, they would be losing, simply because in the first few years, they have not learned about the market and all, and therefore, some learning will have to take place, and then they will become actually successful exporters and more than recover the losses in the first period. It’s the same problem, it comes back to the same issue, that if the learning is internal, then the firm will do it on its own. You don’t require any intervention.

And if it is external, then your interventions of the kind of subsidies, export subsidies, will not solve that problem, so you come back to the same set of issues. There is also a good paper by Gene Grossman and Henrik Horn, who make the point in a very different context where there is imperfect information about the firms and learning takes place in terms of revealing of the information and all. They again show there that, once you properly model the microfoundations, then there is no case for intervention.

There is one point I should make actually, that because from a practical standpoint, that logically, clearly there is not a case for infant industry protection. But governments kind of do come under pressure to intervene somehow or the other on behalf of the enterprises and do something good. All governments want to do that, and there, it seems to me, I’ve always argued that, if you feel compelled to intervene, then what you should do is you try to promote your infants who are likely to become exporters or who are already small exporters, at least, rather than those who are doing import substitution because there are many reasons.

First of all is whenever you say, “I’m going to do import substitution, and therefore, I’m going to promote the infant where I can do a lot of import substitution,” the industry you’re going to pick up is probably the most inefficient one. Because when you want to choose an infant for import substitution, you will ask yourself, where can I do the most import substitution?

RAJAGOPALAN: Absolutely.

PANAGARIYA: You will say that, “Ah, this is where I’m importing so much, and this is where the scope for import substitution is so large.” But there is a reason why you’re such a large importer of the product, because in that product, you are the least competitive. Your cost disadvantage is probably the largest, which is why you are such a large importer of the product in the first place, so you are going to start with the product where you are the least competent to actually become competitive.

Whereas, if you try to do this intervention on behalf of exporters or mere exporters, then your chances are a little better because ultimately exporters will have to compete in the world markets against the best in the world. That argument we already went over, and therefore, I think just exposing them through this intervention to the very best in the world and saying that, “Look, this is what you have to be.” That is more likely to give you some positive results.

It is also the case that for exports, you are going to have to use some kind of subsidies, export subsidies, and subsidies attract a lot more attention on the people that, “Ah, you’re giving my money to those fellows, and if they don’t deliver, then you can’t continuously in perpetuity give my money to them,” which is the opposite of the tariffs. In tariffs, the government thinks, “I’m collecting revenues. These bad guys, the foreigners, I’m really going to hit them hard,” although ultimately that is not falling on the foreigners. Largely it’s falling on your own domestic consumers, but it has the appearance of actually being charged from the foreigners, because they are sending you the goods and that’s where you’re collecting the duty and all. Politically, those become much harder to remove.

Then, you also get revenue departments or revenue ministries who say, “Oh, I can’t lose this revenue, customer revenue, and it generates this customer revenue.” They become addicted to that, the ministries, so they don’t want to lose that. That also makes import tariffs harder to remove. Whereas in export subsidies, the revenue guys come in, or at least the expenditure guys come in, and say, “Look, I have better use for this money. Why are we giving this to the profit-making corporations?”

RAJAGOPALAN: This is actually a very lovely point you’re making. Frédéric Bastiat, who fought against a lot of the protectionist arguments, talked about the seen and the unseen effects. Now, you’re telling us about seen and unseen redistribution instruments. Some redistribution instruments are more visible and some redistribution instruments are less visible.

PANAGARIYA: Absolutely.

RAJAGOPALAN: Once we take all the political economy implications, even though neither is ideal, one might be the lesser of the two evils than the other, because there might be more democratic pressure against it, which is really a lovely insight.

PANAGARIYA: Absolutely. Yes.

RAJAGOPALAN: Finally, on infant industries, if infant industries are really going to grow one day and become exporters and so on and so forth, then I feel like in India, export-led growth should have come from the PSUs, the public sector undertakings, because those are in one sense—all the learning is completely subsidized. You don’t even have to reduce the wages of the workers, they have tenure, they’re never going to leave the job. All the costs of innovation, you can protect completely. In fact, in many cases, public sector undertakings are also monopolies within the domestic sector.

They have protectionism on every side. If this is really a true model of how an infant industry grows, then every public sector undertaking should have been a massive— not just an efficient—enterprise domestically, should have had massive spillover effects and should have also been part of export-led growth. What you instead find is when you take this kind of protectionism to its extreme end, not what you came up with in your stylized model, you get nothing but incompetence and losses which are passed on to the taxpayers. If these models are really implemented to the full extent, you’re going to get Indian public sector enterprises.

PANAGARIYA: Absolutely. Absolutely. This tells you that the same problems that arise from lifetime employment—I have no incentive to do anything. Why should I even work?

I’ll give you an example of that as well, but in the context, the reason all this public sector stuff did not work, did not succeed and nobody became exporters is simply because there was no incentive for them to either reduce the cost, nor did the workers have an incentive to deliver on the work. And it is pretty much the same argument applied to the infant industries also, where once you protect them, the industry has every reason to remain lazy. “Why bother to learn? I’m protected, and I’ve got this market domestically, fully given to me, it’s reserved for me. If the consumer is going to suffer, so be it. It’s not my problem.”

And that is how we allowed the protection to really stay for 50 years. Remember, it’s a beautiful example, if you think about it, that when we liberalized in 1991, what did we liberalize first? The licensing went away from capital goods, licensing went away from raw materials, it went away from components, all the intermediate inputs. It did not go away from consumer goods imports. Consumers didn’t count, so the consumer was still denied. In 2001, finally the licensing on consumer goods imports went away, it didn’t go away because we did it on our own. It went away because we were challenged in the WTO by the United States, and we lost the case.

This whole thing about the political economy, which is so stacked against the consumer, even today, most people in India believe that auto industry is such a huge success. But if it is such a huge success, then why, as a consumer, I’m being penalized? But that has no salience at all, and that’s the kind of tragic side of this. This is why in some ways, the trade economists really see themselves as the champions of the consumers, of the people. Nobody else raises the concerns that the consumers have. They just have to take it upon themselves to represent the case of the consumers.

RAJAGOPALAN: The consumer interest is really important. I see this particularly in a developing country, one that has got as many poor people as, say, India did, especially in the pre-liberalization moment. If the cost of consumer goods dramatically reduces, the person who’s benefiting the most from it is actually the poorest people because their consumption inequality that they face on a daily basis just dramatically drops.


RAJAGOPALAN: One of the things that, people keep saying that you should ban all these cheap phones that are made in China or Taiwan, and India should make its own phones and things like that. One thing I always realize when I’m in India is I see people of all income stratas. I have a nice latest iPhone, whatever is available in the United States. Theirs is a much, much cheaper version of that, but that phone is able to do 90%, 95% of everything that my phone is able to do. In fact, some would go so far to argue that because it’s Android and not Apple, it might be able to do even more things than my phone is able to do because it has more applications and things like that. Now, the cost of my parents’ housekeeper taking a picture on her phone is so much lower that she does not have to spend the same amount as what I spend to take a picture from my phone. That I think is really important.

It matters of course for basics, it matters for food, it matters for T-shirts and shoes, and everything else. Because of the telecom revolution in India, we have really seen it take place in phones. And one part of it of course is the network, which is fairly low cost of 5G and data and cellular service in India, but a large part of it is also the hardware. The price of very cheap phones coming into India, if you put tariffs on that, the person who will experience the greatest pain because of that policy is actually not me and you or even my parents who live in India, it is actually the housekeeper, or their Uber driver or their rickshawala or their vegetable seller. These are the people who actually face the biggest costs of tariffs on consumer goods because they will actually be able to spend a much smaller proportion of their disposable income on these goods if cheap imports were available.

I think this is a battle worth fighting for because we normally separate trade economics and development economics or trade economics and inequality or something else. But trade economics is development economics, right?

PANAGARIYA: Absolutely.

RAJAGOPALAN: Trade economics is inequality reducing or growth enhancing. I think this is something really important. When people keep saying consumers, I don’t know who they have in mind, but they’re just regular people. There happen to be more consumers than producers in some of these economies. In that sense, you’re talking about imposing a massive tax on most people no matter how poor they are by protecting certain producer interests.

PANAGARIYA: The mobile example you gave, actually it’s a very interesting history on this. At that time, when we were in the liberalization phase, if you remember or read, that there was this information technology agreement in Singapore. We didn’t sign it right then, but two years later, we did, maybe 1998 or something. We signed that. This is now called the Information Technology Agreement 1 because there was a second one which we did not sign.

In this first one, we had signed it. What that Information Technology Agreement did was to say, “Look, all the technology products would go to zero tariffs for everybody who is a signatory on a nondiscriminatory basis. Even if you’re not a signatory, you will receive the benefit of that, but countries that sign it will eliminate their tariffs on all information technology products.” As a result, all the mobile phones—at that time, these are only feature phones, the smartphones had not come in yet—those tariffs went to zero in India also.

Prime Minister Vajpayee actually brought in the new telecom policy, and we implemented that. That, of course, revolutionized the whole industry. A very critical part of that revolution was the fact that we had signed in the Information Technology Agreement 1, so these mobiles, very cheap mobiles, could come in, and they spread like crazy. The spread was breathtaking. I used to keep track. Like there will be 20 million new phones sold in one month. Compare that to our entire stock of telephones in 1991, which was five million. In 100 years what we had done, five million phones, four times of that was being done in one month.

RAJAGOPALAN: Absolutely.

PANAGARIYA: Once this happened, some infant industry guys came forward later on and said, “Oh, we made a big mistake signing this Information Technology Agreement, because if we had not signed that agreement, then we would have been producing these billion mobile phones.” You forget that there will be no billion telephones if actually you had not allowed those imports to happen. That entire telecommunication revolution that happened with even the poorest people, the poorest households acquiring a phone in their hands was because these cheap imported mobile phones could come in.

RAJAGOPALAN: Absolutely.

PANAGARIYA: What you are saying about the tariffs and custom duties on these low-end, low-quality mobile phones today, actually applied with greater potential in the past actually. If we had not signed that agreement, there would have been no information technology revolution today that we have seen during those years.

RAJAGOPALAN: Yes. On the infant industry protection, what is stopping these phone manufacturers from producing the phones today? If this is about learning, so much of the design is available. I can pick a phone apart, one of these phones. I don’t have the capability of reverse engineering it, but I know some young bright people who do. In a sense, there are other roadblocks or bottlenecks in the Indian economy that are preventing manufacturers from producing very cheap phones. We need to think about that, but I don’t think it has anything to do with the phone being an infant industry.

PANAGARIYA: You see, that is the point we made very early on in the last conversation, that if you actually benchmark yourself to free trade and say that, “Look, I’m committed, and I want my guys to be competitive to those guys. They must compete against those guys.” Then you go back and ask if they are not able to do it. What is the problem? This is what you’re saying, that there’s something that’s preventing these guys.

RAJAGOPALAN: Then the bottlenecks get exposed, then you can actually pinpoint where they are actually.

PANAGARIYA: They get exposed. Exactly. Whereas if you say that, “Oh, they’ve got some disabilities in the domestic market, and I’ve got to neutralize that disability and then I’ll neutralize it by putting a barrier on the imports, which is 15%, 20% customs duty high.” Well, then you have not solved the fundamental problem.

RAJAGOPALAN: Absolutely.

PANAGARIYA: Then, effectively, you are trying to neutralize. You effectively kept your inefficiency in place and simply made it harder for the goods to enter from the cheaper sources, thereby hurting your own consumers and customers. After all, at the end of the day, you’re protecting it, you’re effectively—you call it protection.

Henry George had a beautiful way of putting it. He said that the buyer wants to buy it, the seller abroad wants to sell the product. Who are you protecting here? He even questioned the term protection. He said, “Well, in what sense is this protection? My people want to buy and their people want to sell it. By putting a custom duty, we are not protecting anybody.”

Are Network Goods a Special Case?

RAJAGOPALAN: Just to play devil’s advocate for a moment. I know this is old wine in new bottle every 20 years. That’s the infant industry argument. It just never goes away from the time of Alexander Hamilton till modern-day India, but there is a new version of the infant industry argument, and this is specifically for network goods, not for ordinary goods and services. The interesting thing about network goods are these are goods where firms are not competing in the market, they are competing for the market.

This is some of your platform services, like say Facebook or TikTok. Actually, telephones are a classic network good. The value of me getting a phone line increases depending on how many people I know are already using a phone line. Otherwise, what’s the point of me getting a phone line? I would have no one to speak with. In that sense, there are certain goods which are a special category, which have these network effects. And now, China has successfully kept the competition out for many of the big network goods.

Some of it was infant industry arguments, at least in popular imagination, some of it was about data and surveillance. We’ll keep the data surveillance arguments aside, but a large part was if you develop a domestic Alibaba or a domestic Google or domestic TikTok or Facebook and protect them for X number of years, then even when they’re exposed to foreign competition, all your users are not going to suddenly leave because the market has already been won by that particular network good or platform.

Now, many people are trying to make this argument for India, that India should engage in protectionism specifically in the network goods space, that India should develop its own Facebook and its own TikTok and its own Twitter. Do any of the infant industry style arguments hold more for network goods than ordinary goods? How does one think about this problem?

PANAGARIYA: One thing is clear that it’s a little different from infant industry, because in infant industry, we are saying there is a time element involved. That what I do today impacts tomorrow, but network industry has to do more with the size of the existing industry, it’s not the size of the past production. It is an externality, it’s there in any given period. If I could really start this whole thing on a very large scale, then I could do it. That’s the point.

Now, that being the case, first of all, one example that you gave, of course, goes the other way, where you said that the utility of a telephone to you depends on how many others are using it. There, you clearly want freer trade, because the cheaper my phones are, the more people will use them. One has to first of all make that particular distinction, meaning pinpointing what the source of network economy is. Just as we saw in the infant industry case, it is very important here in the network economy’s scale case as well.

That’s the first point that, in many of these cases, actually maybe free trade would reinforce your network economies, as clearly in your example of the mobile phone. Now, on the production side, when you say that, somehow, as a concrete example, take either Google or Facebook, the more users of the platform, the lower the cost of maintaining that platform. And the users may be concentrated here domestically and all, but the problem, first of all, in these cases, clearly, none of this is concentrated domestically.

Google, particularly if you think about it, the information has to be available from tons of sources. You have to be able to actually spread your platform around the world. We’re not China. China’s case is very different, because there is a government which wants to control the information. I think in India, people make this argument, “Oh, look, they threw out Google, and therefore they were able to build up their own platforms,” but this is nonsense actually.

They didn’t throw out Google. They told Google to follow what their domestic companies are following, which was to build firewalls and take the censorship. Google then decided that, well, this is not something they could do, and they therefore pulled out. That’s not how it happened. Now, in our case, in the Indian case, you want open platforms. You want information everywhere, flow in and flow out. India is not doing what China is doing. First question is therefore what has already started has started, if you tomorrow say that I’m going to shut off Google, who is going to suffer the most?


PANAGARIYA: Everybody. The information that you’re seeking cannot be because I’m going to develop my platform. Well, good luck, because this network externality is international, it’s not national. Either you come in with a huge amount of resources, get some VCs to come in and give you some $100 billion or something like that to build a platform which would be larger than Google. If you could do that immediately, then you could succeed, but if you think that by keeping Google out for 10 years, you’re going to build—I just don’t see how you could do that.

RAJAGOPALAN: Here, the infant industry argument takes a slightly different form. Here, they are infants because they don’t have the consumers yet, which will give them the entire market and therefore, until they capture the market, they remain infant.

PANAGARIYA: It’s a scale issue.

RAJAGOPALAN: It’s a scale issue. It’s not just a learning issue.

PANAGARIYA: It’s not learning, it’s a scale issue.

RAJAGOPALAN: But the time period comes in with scale. The time period one versus the second.

PANAGARIYA: No but if you could within your chosen time period, if you could get the scale, then—

RAJAGOPALAN: Then, fair.

PANAGARIYA: Yes. It’s not a learning issue. Nobody is learning here. It’s simply that the more people use it, the lower is my cost. That’s the point, which is a little different. That’s a case of external economy. In the case of external economy, sometimes you can write the case. Conceptually, you can certainly make the case. If it’s an external economy, then conceptually you can make a case for intervention. Typically, protection will not be the best remedy there. You need to pinpoint the source of externality itself.

RAJAGOPALAN: You’re saying that in the case of the network goods protections we’ve seen in other countries, they usually have to do with surveillance, data, secrecy, authoritarian regimes trying to have censorship and keep free speech out. They have more to do with those arguments than any kind of network externality argument because you don’t see them in nonauthoritarian regimes yet.

PANAGARIYA: I think so. Mainly, we are talking about China because China has this large market of its own consumers, but it is also the country which wants its consumers to use its own only. It doesn’t want them to be informed by the—

RAJAGOPALAN: By the others.

PANAGARIYA: Yes. It’s a very different objective.

RAJAGOPALAN: It is, that’s worth pondering. I would actually go one step further and say in India because of the border conflict with China, India did ban TikTok, and I was holding my breath for an Indian version of TikTok to emerge.

PANAGARIYA: Good point.

RAJAGOPALAN: Because there were already, I believe, 200 million users in India and people had already gotten a taste for TikTok, so you knew the consumer market existed. An entrepreneur could have just simply stepped in and said, “Hey, there’s this huge gap in the market. It’s completely protected. All I need is an Indian partner so that a thing like a ban doesn’t happen again, so let’s put in a lot of funds. We know the technology TikTok uses.”

The only people who did it in India were actually Instagram. Instagram brought in Reels, a TikTok-like feature, but I was holding my breath for an Indian firm to come out and produce an Indian version of TikTok. I’m still disappointed that it just never happened.

PANAGARIYA: It’s a good example. You see, this can be also very misleading, that if we stop the foreign companies from selling the product here, then automatically domestic supply will emerge. It’s a good example, at least, so far, it has not succeeded. It’s not impossible. After all, there are efforts underway. Well, even if I step back a little bit, take Google again, there were a lot of language areas, Tamil, Malayalam, etc. where Google was absent. We certainly had that market. How come no search engine actually came up which was catering to those languages? Why did everybody wait till Google actually got into that? It took a long time for Google to get into these specific languages.

RAJAGOPALAN: Absolutely.

PANAGARIYA: It is very easy to kind of just mislead people into thinking that the problem is a protection problem. It is not protection, it is something else.

RAJAGOPALAN: I completely agree with you. We forget each of the languages, we call them regional languages in India, but some of these languages you’re talking about, Tamil is spoken by tens of millions of people. West Bengal is 90 million people. Without including Bangladesh, you’re talking about 90 or 100 million Bengali-speaking people in the world. Now, these are the size of Germany and France in terms of the potential.

The fact that no one is doing it is probably because they don’t see a stream of advertising revenue. They don’t think these regions are rich enough or something to that extent or simply just lack of imagination on the part of Indian entrepreneurs that they didn’t spot the opportunity. But you’re right, in that both the numbers existed and the technology existed, you just didn’t see too many people come up with a simple solution for these kinds of problems.

Arvind, I’m so glad that we covered this infant industry argument in proper detail. It is of course not the only argument against free trade. There are a number of other arguments and they have to do with imperfect capital markets or they have to do with unemployment issues or diversification, coordination, externalities.

There are just so many arguments that keep popping up in favor of protectionism and against free trade. Hopefully in the next episode of this talk, in part three, you can walk us through many of the other arguments in favor of protectionism and against free trade, and tell us exactly where the problems lie or if in fact they have some merit.

PANAGARIYA: Right. Let’s do that. Let’s do that. We’ll do exactly that in the next episode, so wonderful.

RAJAGOPALAN: Thank you so much for this, Arvind.

PANAGARIYA: Thanks for holding these conversations, Shruti.

Arvind Panagariya

Professor of Economics

Arvind Panagariya is Professor of Economics and the Jagdish Bhagwati Professor of Indian Political Economy at Columbia University.

Shruti Rajagopalan

Senior Research Fellow

Shruti Rajagopalan is Senior Research Fellow at the Mercatus Center and a Fellow at the Classical Liberal Institute at New York University School of Law.