The Optimal Level of Governemnt Investment in the High-Tech World of 2010, not 1810

In response to Stephen Williamson's opus magnum last week, I left a series of comments that actually exceeded his word count. I encourage you to read them and the whole discussion. I think there are a lot of important points and insights. But I especially think the following part is important, and so I have reprinted it here, with modifications and expansions.

David Andolfatto asked:

    Richard,

    Would you mind elaborating on this:

    Due to externalities, especially positional/context/prestige and carbon, asymmetric information, the zero marginal cost of idea/information use, high transactions costs, great economies of scale and the problems of monopoly power, inability to price discriminate well leading to inefficient provision, and much more, the current level of government spending, insurance, and especially investment is far below the optimal level.

    Do you have any way of quantifying the importance of these "externalities" you highlight above? And what is the "optimal" level of government spending, investment, taxation, etc.?

Here is my reply (again, modified and expanded from the original comment):

David (and Stephen),

This is a very big question. A good complete answer can be book length (or books length). It's far from a few simple statistics and p-values. And I'm especially short of time right now, but I think I can give you some big important indicators that I hope will encourage you to really think about this – logically utilizing all of the evidence you've seen in your lifetime, not just the fraction that's considered "formal". As the great growth economist and perennial Nobel shortlister Paul Romer of Stanford said:

In evaluating different models of growth, I have found Lucas's (1988) observation, that people with human capital migrate from places where it is scarce to places where it is abundant, is as powerful a piece of evidence as all the cross-country growth regressions combined. But this kind of fact, like the fact about intra-industry trade or the fact that people make discoveries, does not come with an attached t-statistic. As a result, these kinds of facts tend to be neglected in discussions that focus too narrowly on testing and rejecting models.

Economists often complain that we do not have enough data to differentiate between the available theories, but what constitutes relevant data is itself endogenous. If we set our standards for what constitutes relevant evidence too high and pose our tests too narrowly, we will indeed end up with too little data...

My greatest regret is the shift I made while working on these external effects models...I suspect I made this shift toward capital and away from knowledge partly in an attempt to conform to the norms of what constituted convincing empirical work in macroeconomics. No international agency publishes data series on the local production of knowledge and inward flows of knowledge. If you want to run regressions, investment in physical capital is a variable that you can use, so use it I did. I wish I had stuck to my guns about the importance of evidence like that contained in facts 1 through 5.

– Journal of Economic Perspectives, Volume 8, Number 1, Winter 1994, Page 20.

That said, let's lay out some big chunks of evidence:

Stephen, regarding asymmetric information, you talk about contracts and credit markets, but asymmetric information (or just lack of information – or, as is often forgotten by economists, the lack of expertise to evaluate it well in a very advance complicated world, expertise that can take years or decades of full time study to obtain) is far more extensive than just this kind of thing.

Here is a a 2009 Eureka Alert. It notes a large scale survey which found 97% of climatologists who are active in research think that human activity is a significant contributing factor in global warming, but only 58% of the public thinks this.

And there's this AP poll from July which found that 59% of Americans would oppose any climate bill if it would cause their electricity bill to rise by even $10 a month.

Now that's what I call asymmetric information!

But this is asymmetric information that, David, will lead to massive government underinvesting in basic scientific research related to alternative energy, as well as related infrastructure, and as well as undertaxation of carbon (either direct or indirect).

But there's more than this; how many people understand the underlying economics, that the free market alone isn't the most efficient at all for providing many things (especially after the constant and massive disinformation efforts of the Republicans)? How many people understand the list of potential market problems referred to in the beginning of this comment? And how many people can you expect to have the time or willingness to learn this much economics, when the vast majority have careers that are in very different areas, and they have historically little free time?

Or foreign policy: How many people really understand the extent of the effect a declining price of oil has on positive change for some of the worst authoritarian and terrorist sponsoring regimes in the world (including the old Soviet Union in the 80s), where terrorism costs us trillions per decade in increased security costs, let alone the loss of life.

How many understand that that $10/month, or $100/month, on average, progressively applied, will essentially not make their monthly budgets any tighter after a period of adjustment (see here).

So this should give you an idea of the true extent of asymmetric information, and the magnitude of a large source of government underinvestment.

But now let's look at another huge one.

Cornell economist Robert Frank in 1999 wrote:

A cautious reading of the evidence suggests that we could spend roughly one-third less on consumption--roughly $2 trillion per year--and suffer no significant reduction in satisfaction. Savings of that magnitude could help pay for restoring our infrastructure, for cleaner air and water, and a variety of other things.

Now let's think about this.

I'm 47. I spied on David's vita and found he finished his undergrad degree in 1985, so he's probably around 47, and Stephen has admitted to 55. So all of us should remember 1978 very well.

Disco was sweeping the country and real per capita GDP was $25,503 (from the BEA, in 2005 dollars). In 2009 it's up to $41,890, for a difference of $16,387. Multiplied by the current population of 308 million, that's $5.05 trillion per year (and keep in mind here the yearly GDP figures are adjusted very little for increases in quality, like increased effectiveness of medical treatments).

Now, let's compare how people lived in 1978, their happiness, their utility, to how people live today. Have we added that $5.05 trillion per year efficiently from the standpoint of optimizing total societal utils? (and I care even less about the Pareto definition of optimality or efficiency than Stephen. What's Pareto optimal can have tragically comically small total societal utility).

In 1978, I lived like a pretty typical member of the middle class, in a three bedroom suburban house in Oak Park Michigan. Our floors were all carpet except for the kitchen and bathrooms which had nice linoleum. Our kitchen countertops were linoleum too. It was carpet and linoleum, not wood, stone tile, and granite. But for all that our home was much smaller and less expensive than a comparable one today, from my human experiences, our family (like those of our peers) found it just as beautiful and just as high quality as a comparable one with comparably middle class families today.

Now, there's no t-statistic here, but I can take assumptions just as mild as those on which t-statistics typically are based, or milder, and construct a logic chain showing that it's extremely unlikely that my experiences throughout 47 years in many cities and neighborhoods, with many people, were completely unrepresentative of the population as a whole – and that's what they would have to be to not conclude that a huge amount of that $5.05 trillion was positional/context/prestige utility, of zero sum game at the societal level. This is as opposed to intrinsic utility like increased incidence of air conditioning since 1978, or improved medical effectiveness – which is a quality increase little included in that GDP statistic.

As Paul Romer said, things that don't come with a t-statistic can be a lot more valuable pieces of evidence. And I'd add that their logic chains can be just as rock solid and anchored to assumptions as mild – or far milder – than those behind the t-stat, or other formal empirical evidence.

Now, today total government spending on basic scientific and medical research is approximately $34 billion per year. That includes all of the government spending on basic scientific and medical research on curing cancer, arthritis, backaches, headaches, obesity, robots building robots, solar power, everything. [Calculation: National Science Foundation data, table 2, 2007, the sum of columns E,H,N,O, and W, 2000 dollars adjusted to 2009 via GDP deflator]. And these are things well known to be usually provided more efficiently by the government, either directly, or largely indirectly, say through sponsorship, or purchase from the private sector – Paul Romer quote time:

As just one example, recall that the increasing returns to scale that is implied by nonrivalry leads to the failure of Adam Smith’s famous invisible hand result. The institutions of complete property rights and perfect competition that work so well in a world consisting solely of rival goods no longer deliver the optimal allocation of resources in a world containing ideas.

– Forthcoming American Economic Journal paper, page 8.

and

Think about the basic science that led to the discovery of the structure of DNA. There are some kinds of ideas where, once those ideas are uncovered, you'd like to make them as broadly available as possible, so everybody in the world can put them to good use. There we find it efficient to give those ideas away for free and encourage everybody to use them. If you're going to be giving things away for free, you're going to have to find some system to finance them, and that's where government support typically comes in...Because everybody can use the idea at the same time, there's no tragedy of the commons in the intellectual sphere. There's no problem of overuse or overgrazing or overfishing an idea. If you give an idea away for free, you don't get any of the problems when you try and give objects away for free. So the efficient thing for society is to offer really big rewards for some scientist who discovers an oral rehydration therapy. But then as soon as we discover it, we give the idea away for free to everybody throughout the world

– 2001 interview with Reason magazine.

Now, can you imagine how much faster, over the long run, science and medicine would advance if we increased this ten fold? One could easily see it advancing multiples as fast over the long run. At the end of this commentary, I'll leave references to some formal studies that do some quantifying, but for now I'll just leave this from the abstract of a 1998 Quarterly Journal of Economics paper by economists Charles I. Jones of Stanford and John C. Williams of the San Francisco Fed:

Is there too much or too little research and development (R&D) [This is all R&D combined together, not just basic scientific and medical research. It includes basic scientific and medical research, plus applied research, plus product development (even if it's for products of little or no intrinsic utility, but high positional/context/prestige externality, like "silky smooth transmission" or geo gravitational adjustment for a $100,000 mechanical watch for 0.1 seconds per month better accuracy, but still less accuracy than a $30 atomic clock radio-controlled watch) The estimated total for all of this is $308 billion per year from all sources combined, government, business, and non-profit, NSF data, table 1, column C, adjusted by the GDP deflator]? In this paper we bridge the gap between the recent growth literature and the empirical productivity literature. We derive in a growth model the relationship between the social rate of return to R&D and the coefficient estimates of the empirical literature and show that these estimates represent a lower bound. Furthermore, our analytic framework provides a direct mapping from the rate of return to the degree of underinvestment in research. Conservative estimates suggest that optimal R&D investment is at least two to four times actual investment [emphasis added].

– Vol. 113, No. 4 (Nov., 1998), pp. 1119-1135

A ten fold increase in government research spending is about an extra $304 billion per year, about 1/17th of our $5.05 trillion increase from 1978. Now you don't think if we channeled 1/17th of the increase since 1978 into this we'd produce much higher total societal utility over the long run? We went 1/17th of the way back, less giant wheels on cars, but they seem just as high quality and prestigious because everyone's wheels are smaller, do you really think an average person with a new 1978 Cadillac, with the crushed velvet and pile carpeting, got much less pleasure (or any less) out of it than their counterpart with a 2010 Mercedes of equal societal rarity and affordability? And technological advancement is very little in the GDP, so you can also compare such a Mercedes to the pleasure you'd get from one 2.1% less expensive ($304 billion divided by the current GDP of about $14.6 trillion), but it was just as rare and prestigious and high quality seeming because everyone else's car cost 2.1% less too? And remember, in return you get a ten fold increase in government basic scientific and medical research, and since 54% of basic scientific and medical research is government [same NSF data as before], that means an over five fold increase in total basic scientific and medical research from any source.

Think about it.

Or look at it this way: According to Berkeley economist Emmanuel Saez's data, the top 1/10th of 1% of earners in the United States in 2008 got 5.37% of all income including capital gains. U.S. GDP in 2008 was $14.6 trillion (2008 dollars), so depending on Saez's definitions and calculations, that 5.37% is in the neighborhood of $800 billion. And given that very few people are in the top 1/10th of 1%, his data estimates that that group makes a minimum – minimum – of $9.1 million per year. Now, suppose we take our $304 billion to increase government basic scientific and medical research ten fold and total basic scientific and medical research from all sources, government, business, and non-profit, five fold, from their $800 billion, so they all have about 3/8ths less, so they earn a minimum – minimum – of about $5.6 million per year.

But they all maintain their same relative position, same relative prestige, same relative feel of quality for what they have. Instead of making $9.1 million per year, it's $5.6 million per year. Instead of $91 million per year, it's $56 million per year. Instead of $910 million per year, $560 million per year. Do you really think there will be much of a loss of utils for these individuals, especially given that they will have the exact same level of prestige because all of their counterparts lose an equal proportion of income, the exact same feeling of quality for what they have?

And don't tell me the pie will shrink because they'll work less. They’ll only make $5,000 per hour after taxes instead of $8,000, so that's not enough incentive? Anyway, we all know the income and substitution effects. Stephen in his post just assumed a tax increase would get people to work more hours due to the income effect. In fact, tax rates in anything but a very extreme range have little effect over the long run on work hours, especially if they are constructed smartly to allay psychological effects (like a VAT with progressivity from using the proceeds progressively like for free universal pre-school and bachelors degree). From an expert well versed in this literature, MIT economist Jonathan Gruber:

Changes in tax rates appear to have relatively modest effects on total gross income; the total amount of income actually generated through work or savings does not respond in a sizable way to taxation.

–  "Public Finance and Public Policy", 2nd edition, 2007, page 734

I even got Scott Sumner to admit this! (see the comments of this post)

So again, I ask you, do you really think there will be much loss of utils for these super wealthy individuals with this 3/8ths diversion to basic scientific and medical research?

And you have to decide one way or another here. However happy you are with your data and evidence, not making a choice, or the status quo, is a choice. And one that must be justified with the data and evidence you have – not fancy non-existent data and evidence that you'd like to have – because the theory is not determinate due to all of the market imperfections of the invisible hand that are well acknowledged in economics, that are referred to at the start of this comment.

Do you really think that this little intrinsic utility loss per person – over this tiny an amount of people – would outweigh, over the long run, the utility gained from a ten fold increase in government basic science and medicine, which is a five fold increase in basic science and medicine from all sources, government, business, and non-profit combined?

Again, you have to choose a level of government investment too, based on the same data and evidence that I do. The data and evidence is no more formal and fancy for you (although I'll give some additional strong formal data and evidence at the end). You have to rely on the same data and evidence that I do to justify your decision if you say we should keep the level of government investment the same, if you say we should cut it, or if you say we should cut it by 99%. The theory is not specific. It's clear due to externalities etc. that government investment shouldn't be zero and it's clear that it shouldn't be 100%. To find out what it should be in between you have to use the data and evidence you have, and it's well established in statistics that it's inefficient to throw away data and evidence. Now, I hope to illustrate this with a little parable:

A snobby empirical economist is hiking with a friend who eats a berry off a bush and immediately keels over and dies. The snobby economist thinks to himself, well it's just a sample of one berry; you can't draw any conclusions from a sample of one. It's just anecdotal evidence. And being hungry himself, he eats a handful, keels over, and dies.

Moral of the story: It's very inefficient, and perhaps very dangerous, to ignore abundant a priori information at your disposal, even if that information is not formal, but still logical – the logic chains are completely solid – and based on relatively realistic assumptions. You just currently don't have a formal version of it.

It wasn't hard at all to put together a completely solid logic chain showing those berries were poisonous using common, but informal, a priori knowledge about biology, human physiology, evolution, similarity across humans, and extremely realistic assumptions.

Now, I think I could put together a similar case here with regard to the magnitude of positional/context/prestige externalities, the magnitude of the effect of the zero marginal cost of idea/information usage, etc., etc.

In any case, this is plenty for now. As I said at the start of this, a good complete answer can be book length (or books length), and I can't write that here, but I hope what I have written gets you to really think about this.

Before leaving you with a sample of strong formal evidence, I'll tie this post to it's title with this comment I recently left on the blog of Richard Green:

I think a big point is that as a country advances more technologically, high return government investment of the kind the free market will grossly or inefficiently underprovide (due to long established in economics market problems like externalities, etc.) becomes more and more important.

In 1810, there was little need for education. The vast majority just did low tech farming and there was not much to learn in school. Medical costs were tiny because there wasn't much for a doctor to learn and there wasn't much he could do.

As a country advances the need for these social investments increases and we've fallen far behind this increase over the last generation in California and in the country as a whole.

Now, I'll leave you with that sample of strong formal evidence:

Robert Barro and Jong-Wha Lee, "Educational attainment in the world, 1950–2010", Vox, http://www.voxeu.org/index.php?q=node/5058

Charles I. Jones, "Sources of U.S. Economic Growth in a World of Ideas", The American Economic Review, Vol. 92, No. 1 (Mar., 2002), pp. 220-239

Charles I. Jones and John C. Williams, "Measuring the Social Return to R & D", The Quarterly Journal of Economics, Vol. 113, No. 4 (Nov., 1998), pp. 1119-1135

Zvi Griliches, "Productivity, R and D, and Basic Research at the Firm Level in the 1970's", The American Economic Review, Vol. 76, No. 1 (Mar., 1986), pp. 141-154.