One of the things I’ve learned over the years is that if you ask enough questions eventually you might catch the right person in the right mood and, with just a little luck, you can learn practically anything. Quite a few econbloggers including Brad DeLong, Kash Mansori, and Mark Thoma have posted on our low national savings rate and the potential implications for future growth. Recently, whenever one of these knowledgeable people has mentioned the national savings rate (and how low ours is), I’ve asked the same question (or something similar to it): “Doesn’t raising the national savings rate mean that people in the top income quintile need to save more?” A few days ago I asked this question in the comments section of this post on Dave Altig’s Macroblog and he was kind enough to respond. His response is sufficiently interesting that I’ll quote it in full:
“Dave — The arithmetic answer is “no” — the increase in saving can come from any part of the income distribution. But I think that the other way to think about your question is this: Has the decrease in saving rates been associated with the upper tail of the income distribution. The answer there seems to be yes — as documented in this study from the Federal Reserve Board.
These guys find that the dip in savings rates since the mid-80s has been concentrated in consumers who owned stock. There was no change in the behavior of non-owners. So, the answer to your question then becomes yes: Saving rates would recover if the upper income groups — they are the stock owners — reverted to their previous behavior. Of course, we might want to ask whether that would, indeed, be a good thing, as off-the-shelf economic theory suggests they are behaving more-or-less as they ‘should.’”
I want to look at his answer a little more closely.
First, I’m not convinced that this:
“The arithmetic answer is “no” — the increase in saving can come from any part of the income distribution.”
is completely correct in any but, as he puts it, the arithmetic sense, and not, in general, even for that. From the Census Bureau source linked above here are the household savings rates by income quintile:
| Quintile (lowest to highest) | Share of total income |
| First | 3.4 |
| Second | 8.7 |
| Third | 14.8 |
| Fourth | 23.4 |
| Fifth | 49.8 |
and here (borrowed from Kash) are some national household savings rates:
| Country | Household saving rate |
| China | 25.5% |
| Korea | 10.0% |
| Malaysia | N/A |
| Singapore | N/A |
| Thailand | 18.7% |
| India | 24.3% |
| Canada | 1.4% |
| Germany | 10.7% |
| Japan | 7.4% |
| United Kingdom | 5.5% |
| United States | 1.4% |
Let’s say, hypothetically, that we wanted to raise the U. S. savings rate to the still-modest 5.5% household savings rate of the U. K. If the lowest quintile saved 100% of their income we still wouldn’t get there. And, obviously, that’s unrealistic. Well, what if the lowest three quintiles were to save at the astonishing rate of the Chinese and Indians? My calculations suggest that that would achieve the target savings rate.
Frankly, I can’t imagine such a dramatic change in behavior by so many people in such a short period of time. And wouldn’t that change be catastrophic to a U. S. economy that’s 70% dependent on consumption for its GDP?
I’ve examined the paper that Dave Altig linked to and it looks pretty solid to me.
In the end I’m left with more questions than I started with:
Is my reasoning sound?
Are Juster, Lupton, Smith, and Stafford’s conclusions sound?
Is Dave Altig correct in saying “off-the-shelf economic theory suggests they are behaving more-or-less as they ‘should’”?
What rate of national savings should we have in order to sustain growth?
What policies would encourage an increase in the rate of national savings to the target rate (especially in light of the finds of J, L, S and S)?
The last thing I am is an economist. More like hopelessly naive. I’m hoping that this post may spark the interest of the many who are much more knowledgeable than I to address the questions I’ve raised.
Thank you for asking David this question, and thanks to him for the answer. I wasn’t aware of the distributional aspects in the change in saving. Now that I am, I want to think about what besides interest rate sensitivity (low rates, low discretionary saving) might explain this, e.g. changes in tax rates causing changes in differential rates of returns on, say, physical verusu financial assets. It sounds like David has a head start on this so maybe we’ll get lucky and he’ll do the hard workd for us. But low interest rates would cause saving to fall. I just can’t say, empirically, how much of the top of my head.
One measure of the optimal rate of saving is something called the Golden Rule rate which maximizes the well-being of societies members. Most estimates of this imply we are below this rate. Why? One idea is tha taxes drive a wedge between private and public optimality and cause a sub-optimal saving outcome. This requires active intervention to overcome.
The optimal saving rate is the one that is consistent with maximizing well-being, the optimal growth rate is a consequence of that exercise, i.e. sustaining growth is not a goal but a consequnce. I say that because many people believe economists maximize GDP growth with no other considerations, but we take preference as given (e.g. if people prefer a clean environment, that will be part of the maximzation) and maximize well-being.
If another country, say China, wants to lend us money dirt-cheap, why should we sacrifice and save as much now? We can finance investment by foregoing consumption ourselves, or by borrowing, and when borrowing is cheap we aren’t as willing to forego consumption ourselves (the reward, the i-rate, is too low).
I tried to be a skeptic on this for a bit and ask how we knew for sure saving was too low, but most of my questions were answered … and everyone seems to take that saving is too low as an a priori position.
I hope this quick response is helpful.
Dave –
When I made a similar plaint about needing to take an economics course before I dared to form an opinion (in this case, on the solution to working-class woes in an age of international economic competition), commenter Tony B suggested: ” If you don’t have the time to take an econ course, I would suggest reading Economics in One Lesson by Henry Hazlitt.”
For many, this is a confusing topic. What is — and what is not — included in National Savings is a bit counterintuitive for many.
The US Bureau of Economic Analysis computes National Income, which represents the total amount of money earned in a given year. This includes, mainly, payments of wages, rents, profits and interest to workers and owners of capital and property. It does not include other income items such as capital gains.
National Savings is National Income minus consumption costs and investment expenditures. Each year, personal and business savings are usually positive, but the Federal budget is often in a deficit and this Federal budget deficit reduces national savings.
Oddly enough, if we invested less as a nation, that would also increase our National Savings because capital invested reduces National Savings.
When you read laments about our ‘low’ savings rate, you should bear in mind that this is a statistical item rather than a commonsense use of the term savings. The BEA’s methodology of calculating savings does not include certain critical components that most of us think of as savings.
For example, the BEA’s calculations of income and savings do not include capital gains even though capital gains on stocks, real estate, businesses and other investments are a significant part of what we would consider our wealth. Also, the value of home equity and retirement plan holdings are also not fully-factored into this calculation of savings, so the published reports of National Savings significantly understate the true savings that Americans are making.
I was going to point out the problems with measuring savings, but Kurt beat me to it (with style, too). Savings rate should, intuitively, measure an increase in wealth. Instead, we try to back into it with some rather suspect calculations.
You can see a rather different picture, as derived from US Census data, here. Remember that this period includes both the last of the tech boom and the subsequent stock market crash, with a recession starting in 2000.