Note: In a blog entry two months ago, I wrote that because of rather poor management of my blog, some entries seem to have disappeared. Fortunately for me, several other sites reproduce many of my blog posts, so when I have had to look for them and can remember the title, I have usually been able to find them. From time to time, if I think they might still be useful, I will repost them here on the Carnegie site with a little bit of editing to correct typos, clarify points, and add links. The first blog entry to receive this treatment was posted on February 28, 2017.
This is the second one, a reprint of a May 23, 2014, entry in which I tried to explain how savings gluts work their way through the global economy. My conclusion, and one that I repeat often, is that it is useless to approach the issue of savings without considering the underlying conditions that determine how changes in the savings rate are to be transmitted and counterbalanced. This is an especially important topic when we consider trade issues, income distribution, and a global economy suffering from structurally weak demand.
Debate about the global savings glut hypothesis is mired in various points of confusion, a fundamental one of which is the seemingly obvious but false claim that a global savings glut must lead to higher global savings. As one of my favorite economists, Barry Eichengreen, states in a recent piece, “There is only one problem: the data show little evidence of a savings glut. Since 1980, global savings have fluctuated between 22% and 24% of world GDP, with little tendency to trend up or down.”
But while the fact that global savings have not changed seems to present a pretty formidable refutation of the global savings glut thesis, in fact it refutes nothing. As surprising as it might sound, global savings gluts do not result in higher global savings except under specific, often unlikely, conditions. The main impact of a global savings glut is to create a mechanism that transfers savings from the glutted part of the global economy to another part of the global economy in which either investment consequently rises, resulting in an overall increase in global savings, or debt or unemployment rise, resulting in no change in overall global savings.
What Is a Savings Glut?
There is no formal definition of a savings glut, but whenever market conditions or policy distortions cause the savings rate in one part of the economy to rise excessively (itself an ambiguous word), we can speak of a savings glut. This phenomenon has at least two main causes.
- A rise in income inequality. We see this in China, Europe, the United States, and indeed in much of the world. As wealthy households increase their share of total income, and because they tend to save a larger share of their income than do ordinary households, rising income inequality forces up the savings rate.
- A decline in the household share of GDP. We’ve seen this mainly in China and Germany over the past fifteen years. When countries implement policies that intentionally or unintentionally force down the household share of GDP (usually to increase their international competitiveness), they also automatically force down the consumption share of GDP. Because savings is defined as GDP minus consumption, forcing down the consumption share forces up the savings share. There are many policies and conditions that do this, and I discuss these extensively in my book, The Great Rebalancing: the main ones are low wage growth relative to productivity, financial repression, and an undervalued currency.1
Notice that in both these cases—and completely contrary to the popular narrative that praises high savings as a virtuous consequence of household thrift—the rise in the savings rate does not occur because ordinary households have become thriftier. In the former case, household savings rise simply because the rich increase their share of total income. In the latter case, national savings rise without households in the aggregate increasing their savings at all. In fact, household savings in the latter case often declines as a share of GDP, as does household consumption. But total savings rise because lower household income is matched by higher income elsewhere (business profits or net government revenues) that is channeled into savings.
Because this seems so difficult for many to conceptualize, a simple numerical example might help. Consider the following economy in which households retain 70 percent of GDP, of which they spend half in the form of consumption and save the other half. For simplicity’s sake, we will assume that there is no government consumption, so that all consumption is household consumption:
|-Household income||$ 70|
|-Government income||$ 10|
|-Business income||$ 20|
In this economy, total consumption is $35 and total savings is $65.
Now assume that like Germany in 2003–2005, this economy implements labor reforms that increase business profits at the expense of household income, so that the share households retain of GDP drops to 60 percent. This is what the new economy would look like:
|Total GDP||$ 100|
|-Household income||$ 60|
|-Government income||$ 10|
|-Business income||$ 30|
Alternatively, assume that the government has engaged in financial repression or implemented other forms of hidden taxes, like those in China, that reduce the household share of GDP (also to 60 percent), except this time in favor of the government. In that case, there would be a small difference in the structure of the economy:
|Total GDP||$ 100|
|-Household income||$ 60|
|-Government income||$ 20|
|-Business income||$ 20|
Notice what happens to total savings in the economy. In either case, the household share of GDP has dropped to 60 percent, of which households continue to spend half in the form of consumption and to save half. In both cases, household consumption drops from $35 to $30, as does household savings. In both economies, however, total savings—which is simply equal to GDP minus consumption—rises from $65 to $70.
This is why we must stop thinking of savings purely in terms of what households do. It is very possible, as these two cases show, for total savings to go up even as household savings goes down and household thrift remains unchanged.
How Does the Economy Balance?
To continue my explanation of how savings gluts are transmitted, it is helpful to remember a couple of additional definitions. An economy’s total production of goods and services (GDP) can be defined either from the demand side (consumption plus investment) or from the supply side (consumption plus savings). By definition, in other words, in a closed system, savings is always exactly equal to investment.
An economy experiencing a savings glut must maintain this balance. It is consequently just a matter of logic that an increase in savings arising from a savings glut must be accompanied by a balancing adjustment—either by an increase in investment or by a reduction in savings in another part of the economy—and this adjustment must occur simultaneously. The necessary implication is that whatever causes a savings glut must also cause one or both of these balancing adjustments.
There are only two ways investment can rise and two ways savings elsewhere can drop. A savings glut, in other words, must result in one or more of the following adjustments, enough to fully offset the savings glut:
- Productive investment rises. If productive investment has been constrained by a lack of savings, productive investment will rise. There may be a rise in debt associated with an increase in productive investment, but because this debt can be fully serviced by the increase in productive capacity, higher debt will not result in a higher debt burden.
- The debt burden rises. Nonproductive investment can also rise. Excess savings can cause unsold inventory to rise because the economy produces more than it consumes. It can also cause large speculative flows into real estate or other assets, perhaps even setting off asset bubbles. When this happens, it can create additional investment outlets for excess savings in the form of projects, including most often real estate projects, whose economic value can only be justified by rising price expectations. The debt associated with a rise in unproductive investment represents a rise in the debt burden.
- The debt burden rises. Rising asset prices can unleash a consumption boom if this causes ordinary households to feel wealthier and so increase their consumption (the wealth effect). This increased consumption creates what I will call, perhaps clumsily, a consumption glut. A savings glut can also force changes through the financial system. As deposits pile up in the banking system or as money flows into funds, there is pressure within the banks and funds to deploy them into riskier assets. One way this happens is through a weakening of credit standards, so that households that once might have been denied access to consumer credit suddenly find it easy to borrow. They expand consumption. Needless to say, in either case, the debt associated with a rise in consumption represents a rise in the debt burden.
- Unemployment rises. If the reduced consumption caused by a savings glut is not matched by higher investment or by a consumption glut, total demand drops, resulting in higher unemployment. Unemployed workers stop producing goods and services but do not stop consuming. Because savings is simply the gap between production and consumption, unemployment causes the savings rate to drop.
Economists almost always miss this point. A global savings glut must be accompanied by one or more of the four adjustments listed above, two of which result in a rise in overall global savings and two of which don’t. The former occurs if the economy rebalances either in the form of higher productive investment or in the form of higher unproductive investment (although in the second case it is only temporary). The latter occurs if the economy rebalances either in the form of a consumption glut or in the form of a rise in unemployment.
Nothing else is possible. Notice that a savings glut must result in an increase in productive investment, an increase in the debt burden, or an increase in unemployment.
The best outcome, of course, is the first: when a savings glut is accompanied by higher productive investment. This is often referred to as trickle-down economics when both the rich and the poor benefit from productive investment, with the rich benefitting more. If productive investment has been constrained by an inability to access funding—that is, if the world is savings constrained—it will, but productive investment tends to be constrained by insufficient savings mostly in undeveloped countries. Most excess savings, by contrast, have originated or flow into rich countries.
In rich countries, there are often many productive projects that desperately await investment, but this failure to invest is driven by other factors, and usually not by a lack of savings, so a savings glut is unlikely to lead to higher productive investment. In his 1951 memoir, Beckoning Frontiers, former Federal Reserve chairman Marriner Eccles (1934–48) even argued that a savings glut could reduce productive investment. “By taking purchasing power out of the hands of mass consumers,” he wrote, “the savers denied to themselves the kind of effective demand for their products that would justify a reinvestment of their capital accumulations in new plants.” Eccles was only able to complete his high school education, going no further than that, and he did not have a banking background. But as a very successful, self-made businessman, he understood that whereas banking and finance can distribute wealth, wealth can only be created by producers, and their investment decisions reflect not just the cost of capital but also the expected returns on such investment.
The events Eccles warned about in fact seem to be what happened in Germany after the Hartz labor reforms of 2003–2005, to make a brief digression into current policy choices. The reforms increased desired savings by transferring wealth away from consuming households and toward saving businesses. Instead of increasing, however, after the reforms investment actually fell, perhaps because less domestic consumption reduced the incentive for German businesses to maintain existing investment levels.
This is why President Donald Trump’s proposals to cut taxes on the wealthy are likely to backfire. It is assumed that cutting taxes on the wealthy—which is effectively a wealth transfer from ordinary Americans to the rich—will increase American savings and, with that, American investment. But this is only likely to be the case if American businesses are capital starved and if American investment were currently constrained by inadequate savings. If not —and they almost certainly are not so constrained—investment will not rise and so savings cannot rise.
Often enough, when excess savings are high, they flow into real estate and stock markets, perhaps even setting off bubbles, with overinvestment in real estate an almost inevitable consequence of rapidly rising housing prices. (We saw this most obviously during the past decade in China, peripheral Europe, and the United States.) These speculative flows have another impact that allows the economy to balance savings and investment. A real estate bubble makes households feel wealthier, which encourages a consumption glut, so that between the real estate boom and the consumption glut, the savings glut is fully absorbed.
But this is temporary. When asset bubbles burst, the resulting surge in unemployment brings down the savings rate enough again to maintain the balance between savings and investment.
Savings Must Balance
The point here is that a savings glut need not result in an overall rise in savings. It can just as easily cause a consumption glut elsewhere whose positive impact on total demand fully mitigates the negative impact of the savings glut. The idea, however, that a savings glut can simultaneously create a consumption glut seems to be one of the most difficult things for many economists to understand, perhaps because it seems at first so counterintuitive.
The other way a savings glut need not result in an overall rise in savings is through higher unemployment. In fact, because neither an asset bubble nor a consumption glut is sustainable, unless productive investment has been constrained by a lack of savings, the only long-term consequence of a savings glut is a rise in unemployment and no rise in total savings.
In that case, there might be only two sustainable ways to address the resulting unemployment. Either the savings glut is reversed, or governments act to eliminate whatever previous constraints existed on productive investment (perhaps by liberalizing constraints to investment or even by initiating a kind of new deal in infrastructure investment). A third way, although not sustainable, is for another asset bubble to be inflated so as to encourage another consumption glut—this seems to currently be the preferred way of U.S. and European governments.
Which Way Does the Causality Point?
It is just a matter of logic that unless investment rises substantially, a savings glut must combine with a consumption glut or with a surge in unemployment, so that there is no net increase in savings. But logic only tells us that the two must occur simultaneously. It implies no obvious direction of causality. Does a savings glut cause a consumption glut, or does a consumption glut cause a savings glut? To put this question in contemporary terms:
- Did Chinese policies aimed at forcing up domestic savings (by forcing down the household income share of GDP) set off a consumption glut in the United States? Or did profligate U.S. consumption require that Chinese savings rise to accommodate it?
- Did German policies aimed at restraining workers’ wages force up the German savings rate, with excess savings pouring into peripheral Europe, setting off real estate bubbles, which then in turn set off consumption gluts? Or did overenthusiasm about the euro cause overly confident citizens of countries like Spain to embark on a consumption binge, which could only be balanced by a rise in the German savings rate?
One way of resolving these questions might be to examine the cost of capital. Pulling capital from parts of the economy that are low in savings to places that are high in savings might seem to require rising interest rates. Pushing capital from parts of the economy with high savings to those with low savings, meanwhile, might seem to require declining interest rates.
Underlying Conditions Matter
The point to recognize is that a savings glut does not mean a rise in savings. It means that within some sector of the economy conditions are such that they force desired savings levels to exceed the local need for savings. In the wider system—the global economy—whether a savings glut causes total savings to rise will depend on what happens next.
Under certain conditions, it is possible that this rise in desired savings in one part of the global economy is met with a flow of savings into capital-starved parts of the global economy, setting off a surge in productive investment. This is usually what occurs when capital flows from developed countries to developing countries, the latter being almost always capital constrained except in cases, like that of China, where significant distortions in income distribution have forced up the savings rate by constraining consumption.
But if the rise in desired savings in one part of the global economy is met with a flow of excess savings into developed economies or parts of the global economy that are not capital starved, the savings glut will not set off a surge in productive investment. In this case, the only possibility is some combination of the following three outcomes: either there is a temporary increase in nonproductive investment, in which case savings overall do rise, or there is a counterbalancing adjustment that lowers savings elsewhere so that savings overall do not rise—in the latter case, this counterbalancing must result either in higher consumer debt or higher unemployment.
There is so much misunderstanding about the savings glut hypothesis that much of the debate has verged on the nonsensical. Unless it unleashes a truly heroic surge in investment—productive or nonproductive, although the latter can only be temporary—a savings glut must always be accompanied either by a consumption glut elsewhere with rising debt or by a rise in unemployment. No other option is possible. This is why savings gluts rarely result in higher overall savings.
This is also why any serious discussion of the savings glut must eschew moralizing and must focus instead on the direction of causality. Did distortions that created a savings glut force the creation of a consumption glut or rising unemployment, or did distortions that created a consumption glut force the creation of a savings glut? More specifically, was it sudden surges in savings in China and Germany that caused the consumption surges in the United States and peripheral Europe, or was it sudden consumption surges in the United States and peripheral Europe that caused savings surges in China and Germany? Rather than simply assume one or the other based on possibly obsolete models, we should examine the conditions of each, including most obviously whether they were associated with rising or falling real interest rates. Any analysis that does not recognize that both must occur simultaneously, and so must be resolved simultaneously, cannot possibly be correct.
1 Although largely and maddeningly forgotten by mainstream economists, there exists an extremely sophisticated discussion of the economic impact of excess savings, most significantly in the work of the British John Hobson and the American Charles Arthur Conant, both of whom did much of their most important work during the last two decades of the nineteenth century. The main subsequent innovation to their work is the recognition of the second of these two great causes of savings gluts. In their work, only a rise in income inequality is recognized as a source of excess savings.