Economist Joseph Stiglitz recently published an interesting piece in Project Syndicate called “The US is at Risk of Losing a Trade War with China.” I am always surprised by claims that deficit countries like the United States stand to lose more from a trade war than surplus countries (that is certainly not what history suggests). I suspect it is because many commentators just do not understand why China is so susceptible to a trade war and why Beijing is so worried. Otherwise, I agree with much of what Stiglitz says about tariffs in the article (as I usually do). This includes the article’s main point—that tariffs are likely to have a limited or even adverse impact on U.S. and Chinese overall imbalances, even if they ostensibly improve bilateral imbalances. His article begins by saying this:

The “best” outcome of President Donald Trump’s narrow focus on the US trade deficit with China would be improvement in the bilateral balance, matched by an increase of an equal amount in the deficit with some other country (or countries). In fact, significantly reducing the bilateral trade deficit will prove difficult.

Regular readers of my blog know that I have made many similar arguments. In April 2017, for example, I explained why clumsy attempts to reduce the large U.S. bilateral trade deficit with Mexico are likely actually to increase the U.S. deficit with the rest of the world by more than they reduce the U.S. deficit with Mexico. This may seem counterintuitive at first, but there is nothing complicated about the logic that drives this process. It only seems counterintuitive because the trade models most people carry around in their heads involve implicit assumptions that used to be true but no longer are. Because these assumptions are almost never explicitly stated, it’s easy to fail to notice how changes in the dynamics of global trade and investment have made the old models that drive the debate obsolete.

The same dynamics apply to trade with China, as Stiglitz points out. While tariffs on Chinese imports are likely to reduce the bilateral U.S. trade deficit with China, they are unlikely to reduce the overall U.S. deficit, nor will they reduce the overall Chinese surplus. Tariffs will merely cause shifts in global trade patterns that might raise prices on some goods marginally but that do not affect the underlying causes of the imbalances.

Do Americans Save Too Little?

But for all my agreement with Stiglitz on how tariffs affect trade, I disagree with him on what drives U.S. trade imbalances and, more generally, on balance-of-payments dynamics. Stiglitz argues that the United States has been running trade deficits mainly because Americans save too small a share of their income.ⁱ According to this logic, if Washington wants to reduce its deficits, it must implement policies that force up U.S. savings:

The US has a problem, but it’s not with China. It’s at home: America has been saving too little. Trump, like so many of his compatriots, is immensely shortsighted. If he had a whit of understanding of economics and a long-term vision, he would have done what he could to increase national savings. That would have reduced the multilateral trade deficit.

The United States has indeed been saving “too little.” But it is easy to show that under certain conditions—ones that most of us, perhaps even Stiglitz, would agree characterize today’s global economy—low U.S. savings are an automatic consequence of balance-of-payments pressures originating abroad.

In fact, the same basic arithmetic shows that the United States cannot raise domestic savings relative to domestic investment (that is to say, it cannot reduce its trade deficit) without addressing problems that do indeed originate in China, and in all the other major surplus countries. The United States, in other words, doesn’t have a trade deficit because it saves too little: it saves too little because it has a trade deficit.

Again, I know that this may seem at first incredibly counterintuitive. (And here, inevitably, someone will foolishly intone that no one is putting a gun to a U.S. consumer’s head and forcing him or her to buy a flat-screen television.) But in fact this claim follows inexorably from the basic balance-of-payments arithmetic.

Before going on to explain why, I should point out that Stiglitz is not the only one who believes that low U.S. savings cause U.S. trade deficits. This has been an almost overwhelming consensus among economists and analysts for decades. In May 2017, for example, two other eminent economists, George P. Schultz and Martin Feldstein, proposed a seventy-word explanation of “everything you need to know about trade economics. . .” They wrote:

If a country consumes more than it produces, it must import more than it exports. That’s not a rip-off; that’s arithmetic. If we manage to negotiate a reduction in the Chinese trade surplus with the United States, we will have an increased trade deficit with some other country. Federal deficit spending, a massive and continuing act of dissaving, is the culprit. Control that spending and you will control trade deficits.

While Stiglitz argues that Washington must implement policies that raise national savings to reduce the U.S. trade deficit, Schultz and Feldstein are more specific and far more ideological: they state that Washington must reduce government dissaving by reducing the fiscal deficit.

Either way, these economists agree that only by taking steps to force Americans to save more—whether U.S. households, businesses, or the government—can Washington prompt the U.S. trade deficit to contract. They argue that this line of reasoning follows inevitably from the accounting identities that explain the relationship between savings, investment, and trade deficits.

A Hidden Assumption

I have often argued that economists too easily make categorical statements when they should be making conditional ones. Stiglitz, Schultz, Feldstein, and others base their argument on the accounting identity in which a country’s current account deficit is always and exactly equal to the excess of domestic investment over domestic savings. (For those who are interested, in a May 2017 blog response to the Schultz and Feldstein article, I list and explain the very simple equations behind the relevant accounting identities.)

The point is that U.S. investment has exceeded U.S. savings for decades, and the accounting identities let us see that, as long as this is true, the United States must run a current account deficit exactly equal to the gap between investment and savings. Narrow the gap between the two, Stiglitz argues, and you automatically reduce the U.S. deficit. This is true by definition.

Because very few economists would recommend reducing investment, Stiglitz then takes what seems like the logical next step. He argues that if the United States were to implement policies that caused U.S. savings to rise, the resulting higher savings would reduce the gap between U.S. investment and U.S. savings. Doing this, in turn, would reduce the U.S. current account deficit. In theory, this could perhaps be done by reducing the fiscal deficit, by making it harder for consumers to borrow, by increasing business profits at the expense of workers, or by increasing income inequality more generally.

But there is a hidden assumption at work here. It turns out that policies that increase domestic savings in the relevant sector of the economy would narrow the investment-savings gap only if U.S. investment and savings were wholly determined by domestic forces. If that were the case, it would also mean that Americans imported foreign capital specifically to bridge this investment-savings gap. To put it slightly differently, increasing domestic savings would narrow the investment-savings gap only if foreigners exported capital to the United States mainly in the form of trade finance, and only if this trade finance were designed specifically to fund the trade deficit or to fund the difference between domestic U.S. investment and domestic U.S. savings (which would amount to the same thing).

This is how the world used to work, but that is no longer true today. Economists too often fail to identify explicitly the assumptions that allow their models to work. This is probably why so many economists retain an obsolete model of balance-of-payments dynamics.

Why Does Capital Actually Flow to the United States?

There are, in fact, two very different explanations of why foreign savings flow into the United States, and each has completely different implications:

1. One explanation assumes that trade or capital imbalances originate in the United States, perhaps because Americans save too little and consume too much, in which case the rest of the world accommodates these imbalances. According to this explanation, the United States has domestic investment needs that cannot be satisfied by domestic savings, so Americans must bid up the cost of capital to attract foreign savings to fund the gap. This was almost certainly the case for much of the nineteenth century.
2. The other explanation assumes that trade or capital imbalances originate abroad. The thinking goes that the United States accommodates these imbalances, partly because it has very deep, liquid capital markets with highly credible governance, and partly because of its role as the capital shock absorber of the world. According to this explanation, surplus countries—usually, I might add, because of policies that suppress domestic consumption—have savings that exceed their domestic investment needs and must export these excess savings abroad to run trade surpluses and avoid unemployment. These surplus countries prefer to export a substantial portion of their excess savings to the United States and, as they do so, they push down the cost of capital.

The first explanation—which was valid for most of modern history—assumes that most capital flows consist essentially of trade finance. The second explanation—which was probably valid in the late nineteenth century and has now become valid again since the late twentieth century—assumes that most capital flows are driven by central banks, sovereign wealth funds, capital flight, and investors managing their capital. The presumption is that these capital flows represent independent investment decisions based on expectations of risk and returns.

Whichever explanation is correct, it is clear that the world must balance. And unless we believe that balance is achieved by an extraordinary coincidence at every point in time, causality must flow one way or the other. There is nothing in the accounting identity that tells us which way causality runs, but run it must.

It is wholly incorrect to assume, however—as most economists implicitly do—that it is the rest of the world that automatically accommodates U.S. imbalances. It could easily be the reverse. And I think it is very likely the reverse that holds true, given that interest rates do not typically rise as U.S. trade deficits rise. Interest rates suggest very strongly that capital isn’t sucked into the United States from abroad, but rather is pushed into the United States from abroad.

More importantly, capital flows into the United States do not consist only of trade finance. Instead, this influx does indeed consist mainly of independent investment decisions driven by central banks, sovereign wealth funds, capital flight, and investors managing their capital. Because of the depth and quality of its financial markets, the United States acts as an investor of last resort, absorbing excess foreign savings that need a safe home.

Whichever explanation any reader might prefer, my point is not to assert that one or the other is right. It is rather to insist on a fact that any trade model must recognize explicitly: a world in which U.S. capital imports are determined abroad, by countries and investors seeking to manage their excess savings, works very differently from one in which U.S. capital imports are determined domestically, as a reflection of structurally low U.S. savings rates that require the country to import foreign capital.

In the latter case, Stiglitz would be correct to argue that policies that force up U.S. savings must reduce the gap between savings and investment, and so must reduce the current account deficit. In the former case, however, the U.S. capital account surplus (that is, imports of foreign capital) is determined by conditions abroad, which in turn determine the gap between U.S. investment and U.S. savings. In this case, because policies aimed at increasing domestic savings have no predictable effect on the U.S. capital account surplus, the gap between U.S. savings and U.S. investment will remain unchanged, as will the current account deficit.

Will Raising U.S. Savings Cause Investment to Rise?

To put it in simpler terms, assume for a moment that foreigners have $100 in excess savings that they have decided to invest in the United States. Obviously, this creates a $100 U.S. capital account surplus and a corresponding $100 current account deficit; this also requires that U.S. investment exceed U.S. savings by exactly $100.

What would happen if, as Stiglitz proposes, Washington were to implement policies that are designed to raise U.S. savings by say $20? If foreigners continued to direct $100 of their excess savings to purchasing U.S. assets, the gap between U.S. investment and U.S. savings would not drop to $80, as Stiglitz assumes. It would still remain at $100 dollars because this gap is determined by the decision abroad to invest $100 in the United States.

How is it possible that savings rise without contracting the savings-investment gap? It turns out that if Washington were to implement policies designed to raise U.S. savings by $20, and if the gap between savings and investment were to remain unchanged at $100, there are basically two ways that the new policies could be accommodated (or some combination of the two):

1. In the first instance, U.S. investment would rise by $20 as Americans took advantage of the higher domestic savings to increase domestic investment. This would be true if the United States were a developing country in which desired investment exceeded actual investment, but the country was limited by insufficient domestic and foreign savings. In that case, both savings and investment would rise by $20, and the current account deficit would be unchanged.
2. But in the second case, if savings are already plentiful and interest rates are low, to the extent that all desired investment has been funded, U.S. investment wouldn’t rise. If the gap between U.S. investment and U.S. savings is unchanged (and investment doesn’t rise), then savings cannot rise. This means that policies designed to raise U.S. savings by $20 can only cause savings in one part of the economy to rise by $20 while simultaneously causing savings in another part to decline by exactly the same amount.

   This is the tricky part that is almost always missed. Policies designed to increase savings will only do so if the additional savings fund additional investment or are exported abroad. Otherwise, they simply cause savings to rise in one sector of the economy and to decline in another, as I explained in a May 2016 blog entry called “Why a Savings Glut Does Not Increase Savings.”

There are two important points here. First, if the U.S. capital account surplus is determined by conditions abroad—which is almost certainly the case—policies designed to raise U.S. savings will have no impact on the U.S. trade deficit. This is because (contrary to conventional opinion), in today’s world, the capital account drives the trade account, not the other way around. Again, this may seem counterintuitive, but I explain why this must be the case in a February 2017 blog entry called “Why Peter Navarro is Wrong on Trade.”

Second, this doesn’t mean that policies designed to raise U.S. savings will have no impact at all on the economy. What matters is whether or not these policies result in higher investment. If they do, the U.S. economy is probably better off. This is basic supply-side economics. But if these policies don’t result in higher investment, then the U.S. economy is almost certainly worse off.

Before I explain why these policies would leave the U.S. economy worse off, let me explain why investment is unlikely to rise. In a world characterized by excess savings, there is unlikely to be a significant amount of unfulfilled U.S. investment needs. The United States does need to invest in infrastructure, to be sure, but its failure to do so is political, not because of a lack of capital. In fact, capital is easily available to any credible U.S. borrower (and to quite a few noncredible ones) at the lowest rates in history, no less. And yet rather than invest massively in productive projects, U.S. companies (and those of most advanced economies) refuse to raise money to invest and instead sit on hoards of cash for which they seem unable to find productive use.

Increasing U.S. Savings Means Higher Unemployment or More Debt

This is why policies designed to raise U.S. savings are likely to leave the country’s economy worse off. If Washington were to cut the fiscal deficit, or to reduce taxes on the rich so as to increase income inequality, the result would not be higher domestic investment (as the supply-siders say) or a smaller current account deficit (as Stiglitz says). The result would be either higher unemployment or higher debt.

Why? Let’s return to the previous example. Assume Washington were to implement policies designed to raise U.S. savings by $20. If foreigners or conditions abroad determine the U.S. capital account surplus, there will be no reduction in the U.S. trade deficit. Living in a world of excess savings means that there is no pent-up demand for the additional productive U.S. investment that could theoretically be unleashed by a potential $20 increase in savings, so investment cannot rise.

But the gap between investment and savings must remain unchanged (because there was no change in the amount of money foreigners invested in the United States). This being the case, policies designed to raise U.S. savings by $20 inevitably can only cause savings in one part of the economy to rise by $20 while simultaneously causing savings in another part to decline by exactly the same amount. Total national savings cannot rise if the trade deficit doesn’t contract and if investment doesn’t rise.

How can policies that cause a $20 rise in U.S. savings in one sector of the economy also cause a $20 decline in savings in some other sector of the economy? I have discussed this issue before, perhaps most extensively in a May 2016 blog entry. To put it briefly, such policies can result in a rise in unemployment, which reduces household savings, or the policies can increase household debt by lowering interest rates, expanding credit, or setting off wealth effects.

This is the main point I hope to make here. Policies that Washington implements to try to raise U.S. savings rates can have very different effects on the U.S. economy, some benign but some very damaging. The outcome depends on underlying conditions that are implicit in the assumptions behind the balance-of-payments model that we use. We can broadly summarize the implicit assumptions and their consequences in this way:

1. If foreigners exported capital to the United States mainly to finance the U.S. trade deficit, policies designed to raise U.S. savings would cause the U.S. trade deficit to contract.
2. If foreigners export capital to the United States mainly to dispose of excess domestic savings, and if desired investment in the United States exceeds actual investment, policies designed to raise U.S. savings will cause U.S. investment to rise but will have no impact on the trade deficit.
3. If foreigners export capital to the United States mainly to dispose of excess domestic savings, and if there is no shortage of capital in the United States, meaning that desired investment in the country is broadly in line with actual investment, policies designed to raise U.S. savings will have no impact on the trade deficit but will cause an increase in either U.S. unemployment or U.S. debt.

What Can Washington Do?

So what are the policy implications if Washington is serious about reducing the current account deficit? Again, it depends on which underlying conditions apply.

1. If foreigners exported capital to the United States mainly to finance the U.S. trade deficit, Washington must implement policies that force up the domestic savings rate if it wants to reduce the trade deficit.
2. But if foreigners export capital to the United States mainly to dispose of excess domestic savings, Washington must implement policies that make it harder for foreigners to dump excess savings in the United States or policies that make it easier for Americans to send these flows abroad. Only by reducing net foreign capital inflows will Washington be able to drive down the trade deficit. (One way Washington might be able to reduce foreign capital inflows would be to require that central banks no longer accumulate U.S. dollars in their reserves but rather that they accumulate a synthetic currency that is backed by all major global currencies—perhaps even the International Monetary Fund’s Special Drawing Right (SDR).)

If I am right, then it is not the case that the United States runs a current account deficit because Americans save too little. It is the reverse: Americans save too little because the United States runs a current account deficit or because it runs a capital account surplus: foreign capital inflows automatically depress U.S. savings.

As counterintuitive as this conclusion may seem, this is the implication of very plausible assumptions about how the world works. The reason most economists are not aware of this is simply because they have not made explicit the assumptions that underlie the models they use. Consequently, they have not recognized how changes in global markets have made their models obsolete.

Aside from this blog I write a monthly newsletter that covers some of the same topics covered on this blog. Those who are interested in receiving the newsletter should write to me at chinfinpettis@yahoo.com, stating affiliation.

Notes

ⁱ In this sentence’s reference to trade deficits, the term current account deficit would be more correct, but for the purposes of this essay we can ignore the difference between the two.