Thursday, June 05, 2014

Increasing and improving saving as a philanthropic cause

Summary: While labor's share of world GDP is over one half, capital's share is close to one third. When considering altruistic interventions to increase economic output, as in GiveWell Labs' exploration of U.S. policy, efforts to increase saving and investment should be considered alongside efforts to improve effective labor supply. Compulsory savings schemes and government savings schemes have been used in other developed countries to induce savings far above U.S. levels, and global adoption of such schemes could produce annual gains of many trillions of dollars, although the potential gains are substantially less than the potential gains of labor mobility. Regulatory changes to default pension/investment contributions might also capture important, albeit smaller, gains.

GiveWell recently published a list of potential U.S. policy focus areas, as it begins to experiment in policy-oriented philanthropy. One subset was "ambitious longshots: outstanding importance," areas where the maximum potential gains are exceptionally high, even if they are not obviously very tractable or uncrowded. Topics included migration liberalization, climate change, macroeconomic policy, improving foreign aid, improving political processes, and tax reform. One area that seemed conspicuously missing was non-tax measures to increase savings rates, in light of the properties of the listed candidates.

Tax reform can eliminate wasteful or harmful tax expenditures (such as housing subsidies for purchase of larger homes by the wealthy) or increase labor supply, but central to economists' estimates of large potential economic benefits is an increase in the stock of capital via shifting taxation from income (including saved income) to consumption. But other policies can increase savings and capital stock by significantly more than tax reform.

Tax reform and macroeconomic policy both seem less oriented towards immediate welfare benefits, which we might measure in QALYs, but also contribute to other measures of impact such as gross world product (GWP). In other words they seem to especially serve the second of these goals discussed by GiveWell:

In particular, we generally favor policy focused on benefiting low-income and otherwise disadvantaged people, even when it involves active government – an attitude often associated with the U.S. political “left” – but we place particularly high value on the developing world. Additionally, we place high emphasis on the value of economic growth and innovation (which we feel are likely to benefit future people).   
So policies that increased rich-country savings would seem similar qualitatively, and to have room to be quantitatively larger than some, although not all. Increasing migration to developed countries could increase the effective labor supply, and the labor share of GWP is over one half, but the capital share is in the vicinity of a third. Doubling effective labor supply would have larger effects than doubling the capital stock, but economic gains in each case would be many trillions of dollars per annum.

In this post I will focus on efforts to increase capital stocks through compulsory savings schemes and government savings, because their potential maximum impact is extremely large and they have been deployed in other developed countries to achieve much higher saving rates than that of the United States. There are also 'nudges' using regulatory policy that may increase savings, although the potential gains appear smaller.

General reasons for impartial altruists to favor increased investment
There are a number of basic theoretical reasons why increasing savings and investment should be a good thing from the perspective of patient impartial altruistic values, beyond the basic economic logic that increased capital investment will lead to higher wealth, incomes, and wages.

For one, market rates of interest reflect substantial rates of time preference: people substantially prefer a dollar of consumption now to one later, or enjoyed by their heirs. So supply and demand balance at rates of interest that allow investments to compound by several percent per annum in real terms. But from an impartial point of view, welfare enjoyed now or later is equally important, and so sacrifices of long-run welfare out of impatience are unfortunate.

Further, the individual impatience is often not reflectively endorsed, or endorsed as a forward-going policy. That is, while people may be reluctant to invest cash-in-hand, they are more willing to approve of automatic payroll deduction for investment, and still more willing to initiate automatic deduction that will be phased in later. Individual impatience is time-inconsistent: at a given time we wish we had been more patient in the past and hope we will be more patient in the future, even if we want to eat the marshmallow now.

Capital investment also has global externalities in a world of cross-border financial flows and open economies. Businesses and governments seeking capital can tap international investors, at a cost of capital set by global supply and demand. Increasing or substituting for such mobile capital will tend to increase total economic output, wages and land/resource rents around the world, while lowering investment returns. The United States makes up a little over 20% of global output at market exchange rates, China slightly over 10% and Japan slightly under, with the remainder highly fragmented. Cosmopolitan altruists should be concerned about global as well as national or individual impacts.

There is substantial policy room to increase savings rates in much of the world
The World Bank has data on gross savings as a percentage of GDP, with the world average fluctuating in the low 20s, high savings in East Asia, and unusually low savings in the United States:

Bringing U.S. savings rates up to the level of the Euro area, let alone East Asia, would substantially augment the world capital stock over time.

National savings rates can reflect many things, including the age distribution, foreign ownership in the economy, government fiscal policy, household human capital and income, and social welfare and security policies. Some of these are relatively resistant to policy control, but the basic levers of fiscal policy and compulsory savings on their own seem sufficient to produce drastic changes in savings rates.

Social security schemes around the world require citizens to pay a portion of their income into pension funds that can compound to finance payouts to retirees. Higher contribution rates and lower payout rates will directly increase savings, and there is substantial variation in these parameters among developed countries. For instance, the low-saving United States (17%) and high-saving Singapore (46%) differ dramatically in their contribution rates and the treatment of funds.

In the United States Social Security contributions amount to 6.2% of wages by the employee and 6.2% by the employer (although this is passed on to the employee through lower wages) under a cap which was $113,700 in 2013. Singaporean employees are required to contribute 20% of wages, with a matching contribution of 16% by employers, to the Central Provident Fund (CPF). The Singaporean wage cap is $60,000 Singaporean dollars per annum. Various subdivisions of the CPF accounts can be tapped to pay for education, medical expenses, purchase of a home, retirement, and other investment products.

Both the Social Security Trust Fund and the CPF are invested in the government bonds of their respective countries. However, in the United States revenues from those bonds are spent immediately in the general government budget, and the government savings rate is low. In Singapore the proceeds of bond sales to the CPF must be invested:
No Government borrowings are for spending. Under the Reserves protection framework in the Constitution and the Government Securities Act, the Singapore Government cannot spend the monies raised from Singapore Government Securities (SGS) and Special Singapore Government Securities (SSGS). SGS are issued to develop the domestic debt market and SSGS are bonds issued to the Central Provident Fund (CPF) Board with full Government guarantee.
The proceeds are largely invested in a highly successful sovereign wealth fund, which has taken advantage of the equity premium by investing in stocks and other risky assets to earn returns well above the rate of global economic growth. Government savings have consistently been high. However, since citizens do not hold equities, but only earn Singapore bond returns, the increase in assets has been disproportionately in government reserves. This prevents withdrawal of the return on equities by retiring individuals and thus further grows the capital stock, but amounts to a significant hidden tax relative. However, other developed countries allow compulsory savings to be invested in equities as well as government bonds, e.g. Australia.

Australia requires substantial compulsory savings (nominally paid by the employer, which may have made the scheme easier to sell politically, even though the cost ultimately falls on wages) in a superannuation scheme, introduced in 1992, with steadily increasing contribution rates:

1992-933 / 4*
1993-943 / 5*
1994-954 / 5*
1995-965 / 6*
2021-22 and subsequent years12
Additional private contributions can be made, with limited government matching. Despite the low early contribution rates, total superannuation assets had already increased to $1.84 trillion Australian dollars, or $1.71 trillion U.S. dollars by March 2014. This amounts to over $75,000 per Australian, and amounts would already be much higher if the system had been implemented earlier (no one has yet been in a position to contribute throughout their career) or with higher contribution rates.

The basic ingredients of high contribution rates and investment in private and foreign assets are relatively simple to duplicate as a matter of policy, if not politics. The actively-managed pension fund or sovereign wealth fund model could be replaced by passively-managed low-cost index funds (although some active investors need to exist), or individually-managed accounts (although poor investment decisions, overly expensive products, and individual losses would complicate matters).

Existing support for fiscal policies to increase saving rates
Efforts to increase savings rates through fiscal and social security policy enjoy significant support from across the political spectrum, and several recent U.S. Presidents have proposed major reforms that would have increased savings rates. However, there seems to be much disagreement about the means to increase savings, as different means have different ideological baggage and distributional consequences, e.g. tax increases or involvement of the financial industry.

The IGM Economic Experts Panel regularly surveys a group of eminent economists on different economic questions. On taxation of capital income, one way to affect savings rates:

In the respondents' comments, much of the uncertainty was about the responsiveness of investors to tax rates. The basic point that increased saving increases prosperity had broad buy-in.

Room for more philanthropic funding may be hard to come by at the U.S. national level without effort to find new approaches, as multi-billionaire and Giving Pledge member Peter Peterson's large foundation is focused on U.S. fiscal issues and has already received over $1 billion. Peterson has even funded a newspaper, the Fiscal Times, specifically focused on his issues of concern, and worked with the Obama administration when it was proposing a fiscal "grand bargain" with Republicans that would have increased saving rates.

However, it should be noted that this level of competition is far less than, e.g. the Gates Foundation's activities in global public health, although perhaps large compared to other high potential causes.

At the state level, the Laura and John Arnold Foundation (LJAF), another multibillion dollar philanthropy backed by Giving Pledge members, has worked to increase savings through public pension plans. Shifts from defined-benefit to defined-contribution structures, transparency, and other interventions aim to improve the balance of contributions, payouts, and bailouts to generate more future capital stock and welfare.

The IGM economist panel agrees with the LJAF that state accounting standards favor inappropriately low contribution rates and high payout rates that endanger the solvency of the pension plans:

The panel also, similarly, agreed that short time horizons for budgeting projections can distort public assessment of fiscal policy in favor of reduced savings:

Recently, over 1000 economists, including 15 Nobel Memorial Prize Laureates, endorsed bipartisan legislation to report long-run fiscal liabilities and assets, which had received support from the Peterson foundation.

Compensating behavior and elasticity
A number of complications reduce the most naive estimates of the gains of increased savings rates. Many of them fall under questions about elasticities and compensating behavior.

The point of increasing savings is to make capital available more cheaply and thus produce more productive investment. But a lower price of capital not only increases demand, but also reduces supply: some will invest less for a lower return (although others may invest more to meet stable retirement goals), and investors will have less in the way of returns to reinvest. So the elasticity of investment needs to be considered in assessing benefits.

Individuals and governments who are forced to save in one way may compensate to reach a preferred level of savings by saving less elsewhere. Individuals may take out loans based on future payouts from retirement accounts, mortgage their homes, and refrain from private savings. Such compensating behavior will normally be substantially less than needed to nullify mandated savings, particularly with a high contribution requirement, but does reduce impact. Credible commitments by governments not to raid pension assets, particularly in fiscal crisis, are also challenging.

Macroeconomically, while in the long-run investment increases wealth, under conditions of unemployment savings may not be as helpful as consumption in stimulating economic activity. Recently, there has been discussion of a global savings glut.

Comparison of savings policy with labor mobility
How would a near-maximal compulsory savings scheme, e.g. duplicating Australia's system with Singaporean contribution rates worldwide, compare to near-maximal migration liberalization, e.g. open borders or UAE-like guest worker programs throughout the OECD countries?  The labor share of GWP is over one half, while capital is in the vicinity of a third. This suggest an initial advantage for a 1% increase in labor supply over a 1% increase in capital stock in p, but only by a small factor. However, there are some additional burdens that make increased savings seem less promising economically than migration.

Doubling the savings rate will not generally double the capital stock. Increasing quantities of capital must take progressively less lucrative investments, and as the capital stock gets larger so will losses from depreciation which must be replaced with additional investment.

Increasing effective labor supply through migration would also increase the capital stock, as migrants saved some of their wages and created new profitable investment opportunities to attract additional savings from the existing rich country populations. In contrast, the size of the labor force is much less responsive to the supply of capital: factories do not spontaneously produce human children, and while increased wages may increase labor force participation to a degree, if anything fertility tends to decline with increased wages under post-demographic transition conditions.

Using the extremely simplified framework of a Cobb-Douglas production function (which questionably assumes factor income shares are constant, but can be used for a first pass) with a capital share of x, multiplying the equilibrium capital stock by n would multiply GDP by n^x. With a capital share of 1/3 and labor share of 2/3, doubling the capital stock would increase GDP by 26%. Doubling the labor force without capital adjustment would increase GDP by 59%, and in the long-run by 100% if the new migrants were just as likely to save.

To double the steady-state capital stock we need to double the amount spent annually replacing depreciated capital. The increased capital gives us 26% more income to be saved, but then to double saving we would need a saving rate 59% higher than current levels. The necessity for increased saving means that consumption will not increase as much as GDP, although it will achieve gains in the basic model until a Golden Rule saving rate is reached. However, one might want to increase capital intensity and output even beyond Golden Rule levels to expedite technological advancement and take advantage of experience curve effects.

While in theory one country like the United States might unilaterally absorb most of the developing world's population as migrants, changes in U.S.savings policies alone could only affect the U.S. portion of the world economy (although that is within a factor of 5-10 of the world economy). If major migration liberalization were to occur, then the gains of savings would go up: more economic activity would be affected by local savings policy, and the increase in effective labor supply would lower the ratio of effective labor to capital, increasing the benefit per dollar of investment. But such a shift cannot be relied upon, which favors a focus on labor.

Politically, both fiscal/savings reform and immigration are live issues that have commanded major legislative and media attention in recent years. In both cases proposals were far from potential limits (Singapore-like savings policy or UAE scale issuance of work visas or open borders), although the live fiscal proposals may have been closer to the limits of potential policy than the migration ones. I am not very confident about the degree of political opposition to large increases in savings rates vs tens or hundreds of millions of migrants, but my sense would be that overall labor mobility is more promising than capital deepening when combining the policy merits and political tractability.

Nonetheless, there may be exceptional opportunities, and the area deserves a place in any comprehensive list of candidates for major economic gains.

"Nudges" to increased savings
While compulsory household and government savings seem to have larger total potential, there are a number of less directly coercive measures to increase individual savings. Shifting taxation from income to consumption is one already mentioned by GiveWell.

One very widely discussed 'nudging' intervention is adjusting defaults for government and private investment plans, e.g. setting a default contribution rate, a default of low-cost index funds, investment in equities. Effect sizes are substantial, as many employees almost never attend to their pension schemes, and will even fail to set up an account when offered thousands of dollars of employer funds for doing so. The IGM panel on this:


Strong automatic defaults set by regulation might induce a significant portion of the saving from a compulsory scheme, perhaps hundreds of billions of dollars per annum.

A different approach is to reduce costs of investing, saving investors money which they are likely to save at an especially high rate (since it appears in their investment accounts, with the same inertia that interferes with starting to invest protecting the money, along with mental accounting).

Passively managed Index funds tend to have much lower costs for assets under management, and thus tend to deliver higher returns than actively managed funds net of costs, in broad accord with the efficient markets hypothesis (EMH). This is an especially serious problem for small investors (paying 1% of assets annually vs 0.1% makes a large difference over a life cycle), and from the point of view of economic efficiency there is a social deadweight loss in employing many bright and capable people in managing and promoting funds that largely do no better than chance. While active management helps to ensure that prices of securities reflect their value, and thus enable better allocation of capital, there is a reasonable case that the current allocation of resources to the sector is excessive.

A BCG report on the state of the global asset management industry in 2012 gives assets under management (AuM) of $62.4 trillion. While active management has been shrinking the "active core" and "active specialist" categories still held $46 trillion, and took revenues of $128 billion, 0.28% of AuM . The passive category (index funds and similar products) held $7.9 trillion but charged only $9 billion in fees, 0.11%. The reported fees and fee gaps are lower than for individual investors, which likely reflects the large role of pension funds, insurance companies, and other large-scale investors which enjoy economies of scale and lower fees relative to small investors. However, I would also like to investigate the precise measure of net revenues if seriously exploring interventions on active vs passive asset management.

Taking the report's figures at face value, bringing revenues for half of the active core and specialty funds down to passive levels would save almost $40 billion per annum, disproportionately reinvested as gains would be delivered into investment accounts.

Reduced trading would also delay capital gains taxes, allowing increased growth and capital investment before the tax cut is taken, with additional economic gains.

An Economist article suggests that regulatory changes governing the sale of asset management services have bolstered index funds and passive management:
One reason for the rise of ETFs is the changing behaviour of financial advisers. Historically, many earned commissions, paid by the fund-management company whose products they sold and incorporated in the annual management charge. This system created a conflict of interest: the products that were best for advisers to sell were not necessarily the best products for clients to own. Low-cost trackers did not have sufficient fees to reward the advisers, so tended not to be recommended. 
The best way to insure advisers’ independence is for them to be paid by the client, not the fund. But few clients wanted to pay an upfront fee when the cost of commission-based advice appeared to be free (because it was subsumed within the cost of the product). 
In Britain the introduction of the retail distribution review (RDR) in 2013 abolished commissions and required advisers to explain the true cost of advice to clients. Slowly but steadily this will expand the market for low-cost funds. Other countries are following suit; versions of the RDR have been created in India and Australia, and are being set up in Switzerland, Germany, Italy and South Africa. The PwC report reckons that by 2020 nearly all developed markets will have introduced rules that align the interests of the salesman (or broker) with those of the customer. In addition, some investors have the confidence to buy financial products directly, without the use of an adviser, just as they buy insurance online; the internet and the rise of product forums known as fund supermarkets make it easy for them to do so...  
Governments are also pushing pension providers to opt for low-cost funds. In Britain pension charges will be capped at 0.75% a year from April 2015. As part of a plan to nudge people into taking out private pensions, known as auto-enrolment, the British government set up a collective scheme called NEST, with annual fees that equate to just 0.5%. Such measures make it likely that more investments will flow into tracker funds.
Regulatory changes along these lines might produce annual gains in the billions, but shifts in investment costs would seem to be firmly dominated by shifts in the amount saved and invested.


Toby Ord said...

Carl, I would find it really helpful if you were to post an explicit estimate of how big a benefit would be produced for a shift in savings rate of a given amount. For example, you could measure shifts in savings rates as a certain number of percentage points increase in the world savings rate over a certain number of years, and could measure outputs as the GDP being and remaining a certain number of percentage points higher. What would happen if the world savings rate was increased by one percentage point for ten years, compared to the counterfactual (it is difficult to imagine any possible policy that would have an effect bigger than this)?

Then if one is interested in increasing long term GDP, one could think about how hard this would be to achieve versus other interventions such as freer migration, increased population, increased innovation etc.

Toby Ord said...

I'm also interested in more elaboration on why saving is better than spending. It seems intuitive to me that it is, but I've grown up with decades of hearing economists tell us to go out and spend and that this is what makes the economy do better. Would economists agree that "In the long run, the economy does better with savings rates being a percentage point higher at all times, versus the status quo?" You have a question a bit like that in your sample, but I'm not sure it is similar enough.

Rob Wiblin said...

I would be interested to know what you make of the idea that there is currently a savings 'glut' globally ( I presume the argument runs that appetite for risky investment in businesses is low, such that further 'savings' just drive up the price of existing assets or expand idle bank deposits. Perhaps in unusual situations like the present one, consumption spending will have a big effect on willingness to make productive investments in businesses/physical capital, and could even be more useful than bank deposits.

Re Toby, I think what you are looking for is this concept:

Most economists believe we are below this savings rate at most times, as you would expect given human impatience!

The unusual case economists worry about is when you are in a serious recession and monetary policy is not effective at driving people to consume or invest. In that case you would like people to spend money, though doing so in the form of investment in capital goods is just as good as hosting a party!

Ben Southwood said...

Toby, your confusion is understandable because (a) economists are not very good at explaining themselves; (b) economists do not try very hard to explain themselves; and (c) economics coverage in the news is often terrible.

If total nominal income (aka aggregate demand) is fixed, then extra spending means less saving (and less investment). There are a fixed amount of resources to spread between different things. We can use labour & capital to produce consumption goods or investment goods (that allow us to produce more & better consumption goods in the future).

But if total nominal income is not fixed, i.e. we are below potential, as we are in a recession, and there is unemployment because of wage stickiness (or debt stickiness, or informational stickiness) then consuming extra stuff will not reduce the amount available for producing investment goods.

Outside of recessions, when some economists think there is a "paradox of thrift" (more saving impossible in aggregate because lower consumption worsens the recession), economists would agree that more saving -> more investment -> more growth -> more consumption.

It's not necessarily desirable in developed countries, because we already do enough investment that our ancestors will likely be much wealthier than us. But in poorer countries that may not be true.

Carl said...

Toby, I added a paragraph mentioning the gains with a Cobb-Douglas aggregate production function.

Unknown said...

What about an "inverse wealth tax" where owners of capital are paid a certain percentage of their wealth each year? For example, if you wanted to enable investments that currently have a 0.5% lower rate of return than the market (and hence aren't competitive), you could give everyone 0.5% of their assets each year. This would have the advantage of being less distortionary (all types of investments are subsidized equally) and not decreasing aggregate demand, but two disadvantages would be that (1) it would be very expensive - doing this in the U.S. would cost 0.5% of the capital stock each year (~$225B), and (2) it would increase economic inequality.