Sunday, September 18, 2016

AI and Krugman's Hot Dogs

Yes, I know, I missed a great clickbait headline opportunity… But as Wayne Campbell would say, I'm not worthy!

Anyway, I've been studying the issue of potential revolution in artificial intelligence and robotics extensively for the last few years. I've read thousands of pages, including most of the recent major economics papers, and general public books (this one just ordered). And I'm actually doing a good deal of learning about artificial intelligence, itself, as in studying textbooks.

Note to economists: You'll actually be very comfortable with much, or all, of AI. One of the five major types of machine learning (which basically is AI) is statistical modeling, especially Bayesian Monte Carlo. And all five, which can be powerfully combined, involve very advanced mathematics. And the goal of the ML is always maximizing a utility function! (or equivalently, minimizing a loss function)

So ML will be, in fundamental ways, very familiar and comfortable to an economist. But there are some important differences. It's not perfect optimization or bust (or no pub). You recognize that's usually not realistic. You just try to get the highest utility score you can, even if, as is usual, you don't know if that's the global optimum. And you don't just assume a local optimum is the global optimum.

I think there's a lot economists can learn from ML scientists, but there's also, I think, a lot ML scientists can learn from economists. For example, with evolutionary algorithms (one of the five ML methods), algorithms compete to be the highest scoring on the utility function, and they mutate to evolve, and can even reproduce sexually! where the features of two algorithms are spliced and combined. In the explanations of this I've seen, and it's benefits and costs, I have not seen something that occurred to me quickly from financial economics.

With options, you place much greater value on assets that have a high variance, all other things equal. As the upside benefits you far more than the downside costs you. Likewise, consider sexual reproduction. Suppose one species has good members, but not great ones; there's little variation, but they always reproduce for a good, but not great, average. Now, suppose another species is poor on average, and has some members which overall are not very good, but they do have some great feature, or features. That second species will evolve to take over the first with sexual reproduction, because eventually it will produce offspring which have the best of the greatest members all spliced together in some offspring. Those offspring, even if initially rare, will then eventually take over because they will be so super-fit, and become the bulk of the future populations. So, the lesson is that sexual reproduction of algorithms will be more likely to have benefits that outweigh the computational costs, the greater the variability (and it might be very beneficial to look for ways to juice up the variability.)

Note, in the natural world at least, there are complications, as genes proliferate in future populations, or not, in a tribe, so the success of a whole tribe is an important factor. And a tribe's success depends on some specialization and variety of members. Also, note that in the natural world, especially, sexual reproduction is extremely costly, yet it has nonetheless evolved to be in every organism beyond the most simple, so the logic for it is very powerful.

This whole thing does give you an idea of the interdisciplinary nature of artificial intelligence, and how the field could benefit from learning from, and collaboration with, other fields -- and vice versa.

The AI field is fascinating, and I have gained many insights already. I will eventually have a long blog post/article on all of this, but given the size of this issue, it's best not to wait, and get down some insights and ideas along the way. This is both to help my thinking and understanding, and hopefully to provide some.

So, without further ado, let's get to today's topics.

Krugman's Hot Dogs

The blog of The Bank of England, Britain's central bank, recently had a sanguine post on the whole robots/AI's issue. The author, B of E economist John Lewis, uses Nobel Prize winning economist Paul Krugman's hot dog mini-model as a justification for his optimism:
Technology can lead to workers being displaced in one particular industry, but this doesn’t hold for the economy as a whole.  In Krugman’s celebrated example, imagine there are two goods, sausages and bread rolls, which are then combined one for one to make hot dogs.  120 million workers are divided equally between the two industries:  60 million producing sausages, the other 60 million producing rolls, and both taking two days to produce one unit of output.  Now suppose technology doubles productivity in bakeries.  Fewer workers are required to make rolls, but this increased productivity will mean that consumers get 33% more hot dogs.  Eventually the economy has 40 million workers making rolls, and 80 million making sausages.
The problem, however, is this: Substitute roll makers and sausage makers with low-skilled workers and high-skilled workers. Suppose the economy starts at 4 billion low-skilled workers and 400 million high-skilled workers, and produces $80 trillion/year in goods and services.

Now suppose that advances in AI and robotics result in there being, for the purposes of producing pecuniary goods and services, 12 billion low-skilled workers, but only 500 million high-skilled workers. With the old production processes, you had a ratio of 1 high-skilled worker to 10 low-skilled workers. Keeping these processes, you would need just 5 billion low-skilled workers, not the 12 billion you now have, with the influx from Robotia and the AI Republic. So what happens to the other 7 billion low-skilled workers, machine and human?

Well, you could go to other production processes that use less high-skilled workers to low-skilled workers, again, where low-skilled workers now include the machine kind. But the problem may be, and I think is, in the real world, that production processes that have a low proportion of high-skilled workers produce a lot less per worker. And if they produce a lot less per worker then it does not make sense in the market to do them, unless they cost a lot less per worker. Thus, real cost per worker must drop.

Now, the high-skilled workers can be utilized for the old high-skilled production process, so the market must pay them at least that old wage. But now to get businesses in the free market to employ a very heavy low-skilled production process, they will only do it if the wage of the low-skilled workers drops, and maybe very dramatically, perhaps to poverty level – or below. And a human subsistence wage need not be a floor, as the subsistence wage for a machine may be well below that, with the cheap solar power of the future. And, with how smart and advanced these machines may become, they may have very low maintenance costs. They may mostly maintain themselves, and each other.

Essentially, the problem is, what if the roll makers don't have the skills to make the sausages. Then, you can't just shift to this new higher production economy with more hot dogs produced, but with a lower proportion of roll workers and a higher proportion of sausage workers. So what do you do? You train roll makers to now make sausages? What if sausage making requires far more education and skill? This may be very costly, and the benefit may be mostly hard to recoup for the payer of this education and training. Meanwhile, governments may be unwilling or unable to pay, especially with the horrifying power of the billionaire funded right, and its government always bad, always a waste, propaganda machine.

And aside from that, many workers may be too old to practically learn advanced new skills. And if even a fifth of the population do not have the cognitive and other abilities necessary to learn high-skilled sausage making, then do you have a fifth of the population permanently unemployed if the new robots can do any and all of the roll making at less than a human subsistence wage?  

So, what then? If you can't just shift enough workers into the high-skilled sausage making, what do you do?

You use a different production process? This may be a lot less productive. You end up with three rolls per sausage. It generates a lot less GDP per worker. It won't be done unless the roll workers' wages go down.

But the sausage workers' wages won't go down; if they did, they would be hired away into old style 1-1 facilities, but even moreso, with rolls very cheap now, sausages will command relatively more money (the price of their complement, some would say necessary complement, has gone down). And, if sausages command more money, then so also will the relatively rare sausage makers, all other things equal.

So, the sausage makers' wages go up, but the roll workers' wages will have to go down – and as far as it takes, to employ even the least productive of the roll workers, if they aren't to be unemployed. And this is just the strong trend we've seen over the last two generations for low-skilled workers.

Now, the B of E blog post author Lewis does concede, "In the interim, the transition might lead to unemployment, particularly if skills are very specific to the baking industry. But in the long run, a change in relative productivity reallocates rather than destroys employment"

But think about what that could mean here if robots become able to do almost anything that a low-skilled human could do, only at a cost below human subsistence. You would have to "in the long run" give almost every low-skilled human a college education, and not a Potemkin college education, but the academic and cognitive skills of the typical graduate of a well-respected major university, like my employer, the University of Arizona.

This may be a very, very long run, with massive, or catastrophic, poverty, unemployment, and homelessness in the interim. And the public will have to dramatically change their attitude about the size and value of government, because the private sector won't come close to funding this, given the severe free market problems with education. There's good reason government funds the vast majority of education in every first world country.

The key intuition here is if you just add a ton of low-skilled workers, including low-skilled workers from the countries of Robotia and the AI Republic, without high-skilled workers, or with proportionally way less high-skilled workers, then you're going to lower the marginal product and average product of low-skilled workers greatly, and thus their market wage.

You could say, like Clouseau, problem sol-ved, just have low-skilled workers become high-skilled workers, like the roll makers in Krugman's model just becoming sausage makers. But that is one incredible endeavor to suddenly massively increase the world's, or any country's, education level. I would love to do so, and would certainly vote to invest in it, but as I said in a recent post:
How are we supposed to get the vast majority of men, and women, up to this level of skill and education?
To do so would take a regime shift in our politics, and in public understanding of economics. By and large, one of our two major parties not only does not believe in global warming, or evolution for that matter, they don't believe in externalities, asymmetric information, natural monopoly, contracting limitations and costs, and basically anything that says the pure free market is imperfect (except in cases where it benefits the rich). But providing a massive increase in the education, skills, and general capabilities for most of the population is something that free market companies could only extract a small fraction of the benefits from in profits. And therefore they alone would grossly underprovide this. 
The externalities, contracting and enforcement problems and costs, adverse selection and other asymmetric information, and so on, are profound and enormous. This is why general education has historically been predominantly publicly funded. To say that now, so that most of the population won't go the way of horses, we have to enormously increase our investment in Heckman-style early human development, education, public nutrition, healthcare, and more, from prenatal until at least well into a person's 20's, is to say that we should have an unprecedented increase in governments' size and roles. 
Right now, this is impossible, as the Republican party is dogmatically against any government, except for a small number of areas; mainly military, police, courts, prisons, and perhaps minimal public infrastructure and education.
A recent OECD paper put it dryly, but seriously, "If the tasks that complement machines become increasingly complex and demanding, the employment prospects for workers lacking certain skills may deteriorate." (page 23)

Are robots and AI like shipping containerization?

Lewis gives the example of the revolution in shipping containerization, which plummeted the workers needed per ton to transport goods, yet workers were redeployed, and the amount of shipping increased dramatically to counter:
Take the humble shipping container.  Transporting goods in pre-packed locked containers, which can be lifted straight onto a lorry or train, yielded enormous savings relative to having cargo transported in crates which needed loading and packing individually at each port.   Their inventor estimated that the combined savings on labour costs, time at the dockside and insurance for breakage and theft reduced the price of a tonne of cargo 39-fold.  Bernhofen et al calculate this led to an eight fold increase in bilateral trade between countries with container ports. Whilst employment fell, productivity of labour increased nearly 20 fold. For the shipping industry this wasn’t a massively disruptive technology- though trade patterns changed, the industry became moreconcentrated and ironically less profitable. 
But by reducing the cost of trading, containerisation opened up the possibility of new supply chains and trading arrangements that were previously too expensive to undertake. And, inso doing, the resultant trade flows led to a substantial spatial reallocation of economic activity.
A point I'd like to make here is that with the kinds of robots, machines, and AI's we may see in the future, it won't be just a clever idea that eliminates many specific jobs in a specific business, so demand increases in other businesses and workers move there, or to different jobs in the same business. It will instead be whole classes of work eliminated. A whole class of work, or a whole class of skill or ability, say dexterous movement and sorting with good visual recognition, will be eliminated from pretty much every business, every industry, in the entire economy – permanently – as happened to horses.

It's whole skills that will be eliminated from employment by these new robots and AI's, across every business, not some specific jobs that will be eliminated in a specific business, so I'll take my skillset elsewhere. Your skillset may no longer be needed anywhere. It may be replaced everywhere, or replaced in 90% of businesses.

And again, there are problems with, well, then production will just increase 10-fold to sop up all those low-skilled workers. There are serious bottlenecks in high-skilled workers and raw materials, and there are inelasticities of demand, at least for certain types of products.

Is increasing GDP share of capital the only distributional concern, or even the biggest?

This is the issue you always hear, and the one Lewis discusses (and downplays). But you rarely hear a similar issue that's perhaps more important, and I think is probably far more important in the short and medium run: It’s not just that robots and other AI's may increase the share of the pie going to capital owners. It's that they could cause massive increases in inequality in how the labor share is distributed among laborers.

These technologies can send the superstar, or winner-take-all, phenomena to the moon. Moreover, an increasing share of workers, due to revolutionary advance in these technologies, may find that the wage the market will pay for their education, skills and other characteristics drops below minimum wage. Or well below, for those who think cutting the minimum wage is the answer.

What happens if these robots and AI's make it so that 10% of the population is basically unemployable at a wage above minimum? then 20%, 30%,… Suppose at a grocery store cameras watch everything shoppers put in their carts, and know what it is; the cart is mechanized and follows you, the cameras also recognize your face and have your payment info on file, so no need for any workers at checkout other than a security guard. Dexterous robots can do the vast majority of stocking and unloading,… We are not far from this now, and going the rest of the way looks not that hard in the next 30 years, from what I've learned of this technology.

Then, about 90% of the grocery store jobs are gone. Where are those 90% going to be redeployed? They're low-skilled laborers, where they can be redeployed are similar places where robots can also do those kinds of manual jobs; fast food, manufacturing, waiting tables,… And, as we've seen, no, people are not willing to pay much more, in money and time, for the human touch, to be able to chat with the checker, or teller, or waiter. They'd rather have more badly needed time and money to spend with their own families and friends.

You could say, ok, the economy would expand to have just 10 times the production then, but you hit the bottleneck of lack of high-skilled laborers to do this that I talked about earlier. In addition, you hit a lot of inelasticities of demand; people aren't going to just buy 10 times as many groceries, so we need 10 times as many grocery stores. Those who do still have jobs and money can only eat so much (the wealthy with money, I would think, would tend to buy high prestige items that require a high proportion of highly skilled labor to produce). And costly raw materials can also become bottlenecks.

So, this is not so easy to dismiss with, in the past,… We actually did have to increase educational levels and skills to prevent mass unemployment with past technological advance. But, (1) Government rose to the challenge. We didn't have such a dominant and billionaire powered, government always bad, pure free market always good, right wing. And, (2) The societal level of education and skill necessary with the kinds of robots and AI's we're probably going to see will be very high indeed. It will take an incredible increase in human capital investment, from prenatal to college and beyond.

A simple test of the veracity, or completeness, of normal-employment robot/AI arguments

Finally, I'd like to note a key problem with Lewis's post. His argument applies to any technology. There's no discussion of the specific technology here, AI and robotics, at least as far as what it's capable of. What the technology is, and how capable it is, or may become, is irrelevant to his arguments. These arguments are supposed to work for any technology, so no need to consider how good this technology might realistically get, what it might be capable of doing. That doesn't matter; his arguments will always apply. It wouldn't matter how good engine and motor technology got, we would always find alternative uses for horses, product prices would drop, production would expand,... Except these arguments didn't work for horses.

And with humans, and specifically low-skilled humans, you have the same technology irrelevance in Lewis's arguments. Even if robots and AI become capable, over the next generation, of doing any pecuniary work at all that a low-skilled human can do, at less than a human subsistence wage, this will still not be a problem for low-skilled humans "over the long run", where presumably "the long run" will not be long enough to ruin, or end, their lives.

At least explicitly and clearly, Lewis does not discuss how the long run could be generations, and it may be extremely hard to reach this long run. Nobel Prize winning economist Jan Tinbergen talked about "a race between education and technology" to prevent massive unemployment and/or income inequality. It is possible, depending on the technological advance, and the resources put into education, that technology can run far ahead for a very long time, even that education could never catch up for a potential super-technology like AI. I discuss this in a guest post at the blog of Haverford College economist and Roosevelt Institute fellow Carola Binder.

So, I'd like to offer a rule. Any time someone offers an argument that robots and AI cannot cause a massive unemployment/poverty problem, if those arguments do not consider how good the technology may, with significant probability, get, then those arguments have a fatal flaw, or are at least significantly incomplete.

Sunday, April 3, 2016

AI Veracity Programs, and My Dream for the 2020's and 2030's Trumps


The Trump of the 2020's in a CNN interview (adapted from interviews of the current Trump. All facts noted are for today):


Trump: NATO was set up at a different time. NATO was set up when we were a richer country. We’re not a rich country. We’re borrowing, we’re borrowing all of this money.

Interviewer: [Picks up cell phone/supercomputer] Siri, how does the US's wealth per capita rank? Siri: The United States ranks number two in the world in per capita net financial assets. Number one is Switzerland, but this is a country with less than 10 million in population. Source: The OECD. Interviewer: Mr. Trump, how can you say then that the US is no longer a rich country?

Trump: That’s [South Korea] a wealthy country. They make the ships, they make the televisions, they make the air conditioning. They make tremendous amounts of products.... I think that we are not in the position that we used to be. I think we were a very powerful, very wealthy country. And we’re a poor country now. We’re a debtor nation.

Interviewer: [turns to his cell phone] Siri, how does the US compare to South Korea in wealth and GDP per capita? Siri: The United States has 5.74 times the per capita net financial assets of South Korea. It has 2.03 times the GDP per capita. Sources: The OECD and The World Bank. Interviewer: How can you say, then, Mr. Trump, that the US is a poor country and South Korea is a wealthy one?

I've said before journalists should work in pairs in interviews, debates, etc., with an interviewer and a data person with a computer. The interviewer asks questions, and the data person, hopefully someone very expert, checks any facts and information in real time, and if there's a discrepancy he immediately brings this up in the interview. The reply can be, this is expensive and cumbersome, (although ridiculously worth it to society). But as Siri and Watson advance, this will be no excuse at all. How much do journalists, and the organizations they work for, really care about doing their jobs?


The 2030's Trump

In the 2030's, AI may advance to the point where people will commonly have very good, as I will call them, veracity apps, installed on their computing devices. So, when the 2030's Trump is interviewed, or in a debate, and he says something like, "South Korea is a rich country. We're a poor country now.", you will see red bars on your screen indicating the level of untruth, and scrolling across the bottom of the screen will be a comparison of the net worth's and GDP's per capita of both countries, and the sources of those figures.

Moreover, with a stream, as opposed to an old fashioned airwaves TV presentation, you can pause it anytime. So you can click to read, or hear, or view, the explanation for why the veracity AI just gave the politician four red bars. And then you can click, or say, continue, when you're done, and keep going with the interview. At the end, if you'd like, you can read, hear, view, a fact and source filled report from the veracity AI on all of its claimed untruths in that event.

And the journalist doing the questioning may be wearing glasses, or contact lenses, with a projected, or "heads up", view of information coming from the veracity app, so he can follow-up question on any untruths or misleading right away, in real time, and get a response.

If you follow my blog, you'll see that I have been studying AI extensively, reading a great number of books and articles, some very technical, from a wide variety of sources. I would say from this study that in the 2030's there is a very substantial chance that veracity AI's will be quite good, and quite common. Just look at how good Watson and Siri are already, and how fast they're improving.

When in real time people see the lies, or misleading, of politicians in questioning, interviews, debates, their TV commercials, anywhere, with exact contradicting facts and figures from standard respected sources, this will revolutionize politics. The kinds and amounts of propaganda we see today will be considered from a past dark age. It will be an amazing leap forward.

And there will be, surely, a variety of these veracity AI programs from a number of respected sources, like the google of the day, the Apple of the day, etc., with reputations worth many billions of dollars to protect, for accuracy, objectivity, competence, and trustworthiness. And there will also be open source versions. And meta versions, where the meta veracity AI checks a number of respected veracity AI's and gives a composite of them, and notes if any of those veracity AI's give a very different answer, as a check for one of the veracity AI's becoming biased or compromised.

And these veracity AI's will not be just for politics. Many, if not the great majority of people, won't choose a dentist, or doctor, or mechanic, or plumber unless his equipment is compatible with their veracity AI, sending all of the information from the equipment's readings; the dental x-rays, the blood test results, the video images from the plumber's endoscope, and so on, to your veracity AI service in the cloud.

If the dentist says you need major dental work that it looks like you clearly don't, the red bars will go up. If he says you need all of your silver fillings replaced because of a mercury risk, the red bars will go up, and your veracity AI will show you top scientific sources explaining how this is unscientific and well proven to be untrue, and how this is often used as a way for dentists to profit from extensive unnecessary work. And so on. And when the dentist gives the price estimate for his work, the veracity AI can tell you how this compares to the average price for such work in your area, and give a whole distribution, or histogram, if you'd like.

This could make scams in general far more difficult, perhaps only very possible with the most tribal, and otherwise non-analytical.

Veracity AIs can absolutely revolutionize markets and society, and make the worlds markets far more efficient, and its societies much smarter, richer, and better. But there's no reason to wait. Journalism easily can, and absolutely should, do so much more today.

Sunday, November 8, 2015

Robot/AI revolution decimating employment and wages, not just could it happen, has it largely happened already? Surprising data


"My big issue with the horse argument is that horses cannot innovate on their own behalf, nor are they capable of imitating new innovations. People can do both."

– Dietz Vollrath, University of Houston growth economist, January 27th, 2015


I've been studying this issue extensively for several years now, and have carefully read/watched many or most of the biggest books, papers, articles, posts, and videos. I'm interested in it for many reasons; concerned and hopeful citizen and father, businessman, entrepreneur, investor, economist, and science and technology enthusiast. But also because my main career is in personal finance. And there is a significant possibility that this technological advance will decimate personal financial security for most Americans, at least without strong and smart policy responses.

The impetus for this particular post is Berkeley economist Brad DeLong's "Peak Horse" post last September. I agree with what Brad says. But I'd like to add the following:

A big issue is not just will machines – robots, artificial intelligence or advanced computers – replace humans, substitute for them, but will they be able to inexpensively do so much of what low and medium skilled and educated humans can do, that they have a great effect on the supply and demand, and thus the market wage. And will that market wage drop below minimum for more and more of the population – those with less skill than average, or no college degree; or next, not even no college degree, but not one from a major nationally-recognized university; less health and endurance, less youth, good looks, attractive personality, clean record, etc. Will the market wage drop below minimum for more and more of these people? For 10% of the population, then for 20%, then 30%, 40%, 50%, 60%,… What then?

And, has it largely happened already?

Yes, has it happened already, at least to a large extent? Not just in some theoretical debated future. Please stay with me for the evidence, which may really surprise you.

You go into a grocery store. In 10 or 20 years will 90+% of those jobs be gone. Will there be very advanced cameras throughout the store, connected to a computer with amazing machine learning and other artificial intelligence (AI). And that computer is connected to the cloud where it learns from similar computers at grocery stores around the world, constantly, from their experiences, and their learning, with constant updates.

And what is it learning? How to correctly identify everything you put in your shopping cart with its machine eyes (as well as how to detect stealing).

You go to check out. No cashiers. The computer recognizes you, and everything in your cart, and says, “That will be $127.49 Mr. Delong. Would you like to pay with your Amazon Visa like last time?”

The (mechanized, computer driven) cart docks at a conveyer, and robots with amazing dexterity and speed bag up your groceries and call up your computer-driven car. Then, your mechanized cart (which you didn't have to push or steer. It stayed with you.) goes to your car, and robots load the bags in. The shelves are stocked by robots, and most janitorial and other functions are done by them too. And if you read The Second Machine Age, Rise of the Robots, or just out, The Master Algorithm, you’ll see that robots aren’t that far from a lot of this even now. And solar, at least in the sunbelt (reporting from Tucson), powers all these machines relatively inexpensively. The roof of the supermarket is covered with solar panels, and the parking lot is shaded completely with solar paneled canopies – This kind of thing is not that rare even today in Tucson, and Moore’s law in solar is only accelerating after more than a generation. The sun food for the machines is, and especially will be, a whole lot cheaper than the farmed food for the humans.

All those low skilled supermarket jobs reduced to just a human manager, and maybe a few humans, if that, to fill in the gaps. And the same for restaurants, factories, janitorial and maid service,… I find it very hard to think of what jobs those low-skilled people will get instead, in anywhere close to equal numbers to those lost, where those who still have jobs and wealth will want to pay at least minimum wage for their services.

How many people are going to want to pay a maid to sweep, vacuum, and mop their floors, when for a tiny fraction of the annual price they can buy a robotic machine that can plug itself in and manage its cord, so that it’s no anemic baby-battery Roomba, but a full powered plugged-in machine, exactly like the one a maid would push and guide.

How many people today want to pay the “minimum”, or subsistence, wage for a horse, rather than buy a car, or a tractor, or take the subway, train, or a plane? Not enough to employ even a tenth of one percent of the horses we had a hundred years ago.

Now, is this all just a theoretical debated future?

No, not at all, because if you look at the data, it's happened already, to a large extent. We're already well along. The statistics I'm about to give you are from MIT economist Michael Greenstone and Brookings senior fellow Adam Looney in a 2011 Milken Institute Review article:

1) "Between 1960 and 2009, the share of men [age 25 – 64] without any formal labor-market earnings for an entire calendar year rose from 6 percent to 18 percent." (page 11)

2) "The percentage of men working full time [age 25 – 64] has decreased from 83 percent to 66 percent over the same period." (page 12)

3) "Nonemployment for an entire calendar year among men without high school diplomas [age 25 – 64] increased by 23 percentage points (from 11 to 34 percent) and among those with only a high school degree by 18 percentage points (from 4 to 22 percent)". (page 12)

4) "One way to untangle the two phenomena is to examine the median earnings among all working-age men – this time including those who earn nothing at all. What appeared as stagnant earnings for workers is really an outright decline in wages for the median men of working age. The median wage of the American male has declined by almost $13,000 after accounting for inflation in the four decades since 1969. (Using a different measure of inflation suggests a smaller, but still substantial, drop in earnings.) Indeed, earnings haven’t been this low since Ike was president and Marshal Dillon was keeping the peace in Dodge City." (page 12)

5) "Consider just men between the ages of 30 and 50, a group for whom retirement is rare. The median earnings of all men in this group declined by 27 percent between 1969 and 2009, which is nearly identical to the 28 percent decline for those who are 25 to 64 years old." (page 12)

6) "Surely, the most astonishing statistic to be gleaned from the trend data is the deterioration in the market outcomes for men with less than a high school education. The median earnings of all men in this category have declined by 66 percent [not a misprint] [from 1969 to 2009]. At the same time, this group has experienced a 23 percentage point decline in the probability of having any labor-market earnings. Roughly 10 percentage points of the 23 percentage points is attributable to the fact that more men are reporting disabilities, even though work in physically demanding jobs has been declining for many decades. Men with just a high school diploma did only marginally better. Their wages declined by 47 percent and their participation in the labor force fell by 18 percentage points." (page 13)

Now, it's true that all of these statistics are just for men. The total number of jobs has increased, due to women entering the labor force en masse, and the population increasing. Still:

1) The total labor force participation rate, which considers all of this, has declined in the last 15 years from about 67%, where it was throughout the 1990's, to about 64% (from the Current Population Survey).

2) You bring up horses to some economists, and other smart people, and sometimes the reply is, humans are different; humans are just so much more flexible and adaptable and creative than horses, as in the Dietz Vollrath quote at the start of this post.

The machines eventually got horses. They shifted the demand curve inward so much that the supply had to decrease by over 99% to keep the market wage for those horses that remained above the subsistence level. And you could have made the same argument for horses that you hear all the time for humans – It never happens. Hundreds of years technology has advanced, and we've always found jobs for as big, or bigger, a population of horses.

Well, you know what, after hundreds of years of technological advance, it finally did happen. Machines reached the point where they were so good at almost everything you could employ a horse at that there was no way for 99+% of the horses to do anything else that would pay even a subsistence market wage.

So, relatively low-skilled, low-educated males are not the entire group of humans. But, they are a class of humans, and a big one. And what these data show is that it's not just a theoretical debatable thing about the future. It's already very largely happened. They've already very largely gone the way of the horse in the face of advancing machines (and I'll discuss alternative explanations).

Now, you could say, well, they could become more skilled humans, and not go the way of the horse. But, the bar keeps rising, and pretty fast, as these machines get smarter and smarter. Already that bar is pretty high, practically speaking, for a substantial percentage of our population to clear, and it keeps going up.

And, we have to then invest more and more in educating our population, and crucially, Heckman-style early human development (all of which will, very importantly, also greatly increase the total size of the pie, of GDP long run). But the more we vote Republican, the less we do this, as this money is instead shoveled into more yachts and mansions for the rich. And, no, raising taxes on the rich will not have a substantial effect on the hours they work; this is well shown empirically, and there is the long established in economics income and substitution effects and backward bending labor supply curve taught at every major university.

Cutting someone's after-tax wage from $10,000/hour to $5,000/hour still leaves plenty of incentive, and at these levels it's pretty much about attaining the prestige and feel of how your income and position and belongings and consumption rank, which is unchanged if your fellow rich pay the same higher tax rate. Your 10,000 square-foot house becomes just as prestigious and rare and awesome feeling and satisfying as a 20,000 square-foot one used to be before the tax increase.

But the main point I want to make here is that robot/AI revolution causing unemployment to soar and wages to plummet, for a large and growing percentage of the population, is not just some debate about the future; the evidence is it's already well on its way.

Now, of course, what are other explanations for the shocking collapse of employment and wages for lower-skilled males over the last half century. There are two big ones as far as I know, globalization, and the collapse of unions/bargaining power. Could these two be the predominant, or overwhelming, cause of this plunge in wages and employment? and, males without a college degree (from a well-respected university) could just shrug off the effects of technology over the last 50 years – and over the next 50 years – and adapt with their superior, highly flexible, human brains?

So all we’d have to do is legally well-protect unions, and put up strong barriers to globalization, and employment and wages and job security would be at least restored to the level of a generation or two ago for lower-skilled males? And for everyone else.

Advancing robotics, AI computers, and other machines wouldn’t have been, and won’t be, a major, or catastrophic, problem for males without a college degree.

Is that it?

Well, let’s examine these explanations.

Unions provide much greater bargaining power for lower-skilled workers, and so certainly raise wages greatly for those who have jobs at companies with strong unions. This has been extraordinarily true in the past, when many unionized workers without even a high-school diploma had better wages and benefits than many high-skilled workers. Wages and benefits that would stun the typical fast-food worker of today, or low-skilled member of the sharing economy (or is it the sharecropping economy?)

But that would be an explanation for the plunge in compensation. It wouldn’t explain the plunge in employment, in the labor force participation of lower-skilled males. If unions had remained just as strong, and had kept on negotiating the same dream-like compensation for low-skilled workers (dream-like to a Walmart worker today), if anything, this would have resulted in even less employment.

Now, you might think that with strong unions, employment for low-skilled males would be greater, as the unions would just force it. They would strike and negotiate against automation. But this admits that advancing machines are the cause. It just says that unions could stop, or slow, this cause.

Next, what about globalization. So the idea here is that trade barriers fell substantially, and the costs of transportation and communication absolutely plummeted, and this led to low-skilled workers in the U.S. having to compete far more with the vast numbers of low-skilled workers in the developing world. So essentially, the supply of available low-skilled workers increased dramatically. And this increase did not come with anything close to a proportionate increase in the supply of available high-skilled workers to need the additional low-skilled workers.

Certainly, this would depress wages and employment opportunities paying at least minimum wage for low-skilled American males. MIT Economist David Autor put it succinctly in a 2012 MIT News article on a paper of his on U.S. manufacturing job losses, “Trade may raise GDP, but it does make some people worse off.” Still, University of Oregon economist, and purveyor of the central economics blog, Economist’s View, Mark Thoma, recently wrote, “The evidence suggests that immigration and offshoring aren't the biggest source of the problems workers face. Technological change, particularly digital technology, appears to be a much bigger factor.”

I agree with Professor Thoma on this, based on the research and information I’ve studied. But even if globalization were the bulk of the reason for the collapse in wages and employment for low-skilled males over the last fifty years, this would still mean that advanced robots, computers, and other machines would likely devastate low-skilled male wages and employment in the not too far future anyway. It would still, nonetheless, happen, even if all of the globalization were ended tomorrow.

Why?

Here's the logic:

First, even though I don’t think globalization is the biggest factor, I have no doubt that the effect of it is very substantial on wages and employment of low-skilled males.

But what are the products and jobs in question?

For this phenomena, it’s predominantly manufactured and processed goods that can be transported in the ubiquitous uniform metal containers; on ships, trains, and trucks.

So, the foreign workers that are putting supply pressure on low-skilled American males are workers in production and processing facilities and environments, because that's where the bulk of goods that can be shipped in uniform containers are made, and that's where the bulk of the labor hours to make them comes from. 

And workers in such factories and facilities are one of the kinds of workers that advanced robots and machines are best suited to emulate. The work is relatively simple and repetitive, and in a relatively standardized predictable environment.  

So, the key point is that if foreign workers in these kinds of relatively standardized, simple, and predictable environments, with these kinds of also relatively standardized, simple, and predictable tasks, are able to so devastate employment and wages for lower-skilled American males over the last half-century, then machines would have done it anyway. Because advanced machines largely can do the same things already, and it looks like there’s a substantial probability they will be able to do the vast majority of this kind of work within the next generation or two.

So, in other words, if this is your explanation, then it would have happened anyway, or likely will happen anyway, from the advanced robots and machines doing the same thing as the low-skilled foreign workers.

If low-wage workers in China and India can so devastate wages and employment for low-skilled American males, then so can low-wage workers in Robotia and Machinia and The AI Republic.

So again we see that advanced robots and machines can devastate low-skilled male workers. But what is the evidence that machines will have this kind of capability to predominantly do what foreign low-skilled manual workers do?

Again, it’s largely in the books The Second Machine Age, Rise of the Robots, and The Master Algorithm; and, the large 2013 study by Oxford’s Benedict and Frey, which is based on the opinions and predictions of Oxford scientists and engineers. The authors conclude (page 45):
Our model predicts a truncation in the current trend towards labour market polarisation, with computerisation [instead] being principally confined to low-skill and low-wage occupations. Our findings thus imply that as technology races ahead, low-skill workers will [hopefully] reallocate to tasks that are non-susceptible to computerisation – i.e., tasks requiring creative and social intelligence. For workers to win the race, however, they will have to acquire creative and social skills. [emphasis mine]
And here is MIT economist Erik Brynjolfsson and MIT computer scientist Andrew McAfee in The Second Machine Age:
After visiting Rethink and seeing Baxter [a prototype, highly flexible, intelligent, and inexpensive robot] in action, we understood why Texas Instruments Vice President Remi El-Quazzane said in early 2012, “We have a firm belief that the robotics market is on the cusp of exploding.” There’s a lot of evidence to support his view. The volume and variety of robots in use at companies is expanding rapidly, and innovators and entrepreneurs have recently made deep inroads against Moravec’s paradox. (page 31)
And powerfully:
All these examples illustrate the first element of our three-part explanation of why we're now in the second machine age: steady exponential improvement has brought us into the second half of the chess board [1] – into a time when what's come before is no longer a particularly reliable guide to what will happen next. The accumulated doubling of Moore's Law [2], and the ample doubling still to come, gives us a world where supercomputer power becomes available to toys in just a few years, where ever cheaper sensors enable inexpensive solutions to previously intractable problems… 
Sometimes a difference in degree (in other words, more of the same) becomes a difference in kind (in other words different than anything else). The story of the second half of the chess board alerts us that we should be aware that enough exponential progress can take us astonishing places. Multiple recent examples convince us that we're already there. (pages 55-6, endnotes mine)
Other explanations for the collapse of wages and employment for lower-skilled American males, like an increased desire or ability to claim disability, or take on the role of homemaker, with a working wife, appear not to be large factors, as discussed in Greenstone and Looney’s article.

People today typically debate the future with regard to robot/AI revolution, will it be much harder to get and hold a job that can support a family decently, or even pay minimum wage. Will it take much more education to achieve this? Will this happen in some hypothetical advanced robot and AI computer future?

Well, for male humans, it’s not a will. It’s a has. It has, to a very large extent. The evidence I’ve presented I think is very strong on this. And it hasn’t happened to just mere horses this time. Lower-skilled male humans are humans, and they’re vastly beyond horses. And already the robots, AI computers, and machines have brought them a long way toward the fate of the horses. And these brilliant machines are just getting started.

You want to ask some will’s, then how about these:

Will it take a bachelor's degree to have at least a 50% chance of having a reasonably secure career that can pay wages high enough to support a family decently? And I emphasize secure, because I mean not just can you usually hold a decent job, but will you be marketable enough that you can avoid constant serious risk of long term unemployment before you find another decent job. Because this risk is incredibly costly. Even one or two episodes of 3 – 12+ months of unemployment can devastate a family. Or can lead to homelessness, which is so harmful and dangerous, there's a very good chance it's game over from there.

Will it take not just a bachelor's degree, but a bachelor's degree from a well-respected nationally-recognized research university, like my employer, the University of Arizona? and those kinds of skills? Or even moreso, the literacy, numeracy, and technical skills of only the top half of graduates from respected national universities today?

Because if you say: Oh, ok, it's just horses and men who don't make the effort to become skilled and educated enough, so no problem, they just get skilled and educated enough, and then the robots and AI's are no risk. Then, what if skilled and educated enough so all this is no problem goes from high school diploma to bachelor's at a nationally known respected research university, and with the commensurate skills? Or even the commensurate skills of just the top half of such graduates today, so we can't just grade hyper-inflate our way out of this, and throw up a bunch of Potemkin colleges.

How are we supposed to get the vast majority of men, and women, up to this level of skill and education?

To do so would take a regime shift in our politics, and in public understanding of economics. By and large, one of our two major parties not only does not believe in global warming, or evolution for that matter, they don't believe in externalities, asymmetric information, natural monopoly, contracting limitations and costs, and basically anything that says the pure free market is imperfect (except in cases where it benefits the rich). But providing a massive increase in the education, skills, and general capabilities for most of the population is something that free market companies could only extract a small fraction of the benefits from in profits. And therefore they alone would grossly underprovide this.

The externalities, contracting and enforcement problems and costs, adverse selection and other asymmetric information, and so on, are profound and enormous. This is why general education has historically been predominantly publicly funded. To say that now, so that most of the population won't go the way of horses, we have to enormously increase our investment in Heckman-style early human development, education, public nutrition, healthcare, and more, from prenatal until at least well into a person's 20's, is to say that we should have an unprecedented increase in governments' size and roles.

Right now, this is impossible, as the Republican party is dogmatically against any government, except for a small number of areas; mainly military, police, courts, prisons, and perhaps minimal public infrastructure and education.

Of course, to win elections they have to favor Social Security and Medicare for seniors. But, from my study of politics, I think that most of those who control the party would like, if they could get away with it without losing political capital, to end Social Security and Medicare. And, in fact, they fought Social Security and Medicare tooth and nail when they were first enacted. And I also think that many of those in control of the Republican party would like, if they could get away with it at no cost in political capital, to eliminate most, if not all, publicly financed education, infrastructure, and research.

This is my conjecture, but certainly it's obvious, and not controversial among experts, to say they would oppose any significant increase in investment in education and human development, let alone an increase to an unprecedented level. Just read their platform, and the positions of their major candidates; it's pretty obvious that the direction they'd like to go in is the opposite one.

So, if it's going to require a massive increase in human development, education, skills, and general capability for most men not to go the way of the horse, then that edification is not going to happen anytime soon. And things could get very bad. And for a large segment of the population, the statistics show it already has.

Republican control of the House (with gerrymandering, disproportionate weighting of rural votes, etc.), and the Supreme Court, could easily last another decade. And if the Republicans take the Presidency too, things could go viciously in the wrong direction. They would likely be able to enact extreme legislation and executive orders. Moreover, the Supreme Court, which could strike down any major program, in spite of what's in the Constitution, could be cemented in Republican control for another generation with a Republican President to appoint more Republican Justices.

So, I don't think we can take that much solace in the reply, all the low-skilled men have to do is become high-skilled to avoid going the way of the horse.

Nonetheless, longer run at least, I do think that this is a critical part of the solution (see, for example, my 2014 guest post at Carola Binder's.) I do think that if we had a massive increase in public investment in human development, especially Heckman-style early human development (Nobel Prize winning economist James Heckman's research has shown a high social return to public investment in high quality prenatal and postnatal care, training, and assistance, high quality developmental daycare and preschool, and other early assistance and intervention) that we would greatly decrease the percentage of our population without secure employment. And we would greatly increase the size of the economic pie long-run.

Moreover, an unprecedented increase in human development investment would, in of itself, create an enormous number of jobs.

But the benefits would extend way beyond jobs. A much smarter, healthier, stronger, less criminal, more knowledgeable and expert population would make for a much stronger and better country, and a much healthier democracy.

Eventually, there is a good chance that AI will reach the point where few if any humans will be smart and skilled enough to do anything pecuniary that a machine can't. At that point, substantial redistribution will be unavoidable. Most people have little wealth outside of their labor endowment. If that becomes worthless, they quickly starve without redistribution [3]. If we can maintain a democracy, in spite of the efforts of many plutocrats, then large-scale redistribution will probably be inevitable.

But, of course, there will be an aversion to just giving people money without it being earned. But there, a solution I see is allowing people to earn it by working to become better people, and citizens. The redistribution can be money paid for the job of being a student; getting a college degree, a graduate degree, passing expertise exams,... Or, money paid for hours spent on a physical fitness program, or studying chess and improving your intellect and game, or Tai Chi, Yoga, psychological study, or therapy..., at least until you have reached a certain number of hours worked, or age, and are allowed to retire with a pension if you wish [4].

So, in this way, there may be one job that we can never lose. In the end, our final job may be ourselves.



[1] This refers to an old story that the inventor of chess presented his game to the emperor of India, who was so impressed, he asked the man what he'd like as a gift in return. The man replied, simply some rice for my family, one grain on the first square of the chess board, then two on the second, four on the third, and so on, doubling until the end of the board. The emperor agreed, thinking it a modest request. But due to the power of exponential growth, the last square would require 263 grains, more than 18 quintillion; more rice than has been produced in the history of the world!

In the first half of the chess board, the number of grains grows relatively, or very, slowly. For a while the growth looks linear, and with a modest slope. By the end of the first half of the chessboard, the current square is about 4 billion grains, still just one large field. But now we're in the famed "second half of the chess board". Each doubling, or square, now causes not a small improvement, not a relatively modest, or ordinary, improvement, but a qualitative leap.

And the idea with computer advance, and Moore's law, is that we are at, or near, the second half of the chess board, where every routine doubling of Moore's Law, or similar doubling in computer-power-per-dollar-of-cost, due to software advance, 3-D circuits, parallel processing, etc., causes a startling increase in capability. An increase equal to every increase that came before it for the last 80 years. For supercomputers this means that a new square on the chess board, which happens every few years, now means an additional thirty-four thousand million operations performed every second! Very different than early doublings, which only added hundreds of calculations per second to the capability!

[2] Many experts expect Moore's Law to continue for decades or more, at least in spirit. Originally, Moore’s Law, or “Law”, as it’s an observed and predicted, approximate relationship, not an immutable physical law, was basically that the number of transistors per square inch would increase exponentially for a long period. But the term, “Moore’s Law”, is often used now to say, basically, that computing power, or capability, per dollar of cost will increase exponentially over a long period. And I think this is a much more useful definition. The number of transistors per square inch cannot double that many more times before a transistor becomes as small as an atom. So we're approaching the physical limit of matter's smallest units, and many experts say the practical limit will be hit in a decade or less. But there are many ways that the computing power, or ability, per dollar might keep doubling every few years for decades or more.

These include: (1) 3-D chips – micro-layers of chips on top of each other; (2) Directed self-assembly (DSA); fascinatingly, this is like the way DNA constructs our bodies. Certain molecules attract or repel others, and you use this to form the transistors, the connections between them, and other components and architecture in a chip. This can both plummet the cost of production, and create revolutionary increases in capability; (3) Improvements in architecture in general; (4) Massively Parallel Processing, like the human brain. This can lead to an astounding increase in capability, but it also can plummet power usage and heat. The human brain, which is incredibly massively parallel, has in many ways the capability of today's supercomputers, and more, but with about one one-billionth the energy usage; (5) Quantum Computers; (6) Advanced new materials like carbon nanotubes, potentially inexpensive room-temperature superconductors, and other new nanomaterials with amazing properties and applications; (7) Much faster and denser memory, such as RRAM (Restrictive Random Access Memory);(8) And there is more; this is not an exhaustive list.

And one thing I'd like to especially focus on is software advance. We always hear about hardware speed and capability doubling, but often as important, or even more, is the exponential advance of software. There are many examples you can give – Machine learning, which is revolutionary, is the reason for computers driving cars in dense city traffic, recognizing faces better than humans, and beating the best at Jeopardy. It is a new kind of software (at least in this form, and at this level). Certainly, it is made more powerful with hardware advance, but that hardware would never have been capable of coming close to what it does without this software advance.

MIT's Brynjolffsen and McAfee wrote in their first book on robot/AI revolution, Race Against the Machine (2011), "It also seems that software progresses at least as fast as hardware does, at least in some domains. Computer scientist Martin Grotschel analyzed the speed with which a standard optimization problem could be solved by computers over the period 1988-2003. He documented a 43 millionfold improvement...Processor speeds improved by a factor of 1,000, but these gains were dwarfed by the algorithms, which got 43,000 times better..." (page 18).

So, as a result of all of this, we see many different promising avenues by which "Moore's Law", at least a Moore's Law for doubling of capability per real dollar of cost, can continue for possibly decades, or more. And these doublings are now in the second half of the chessboard (see endnote 1 above), where each one is a gigantic, and possibly profound, advance.

[3] One might argue that with the technology explosion things would get so cheap that even a small amount of savings, maybe even $1,000 or less, would be enough to live ok for a lifetime. However, interestingly, I think this would be unlikely to work. The real price of at least many goods will plummet, but remember, the Fed, and all first-world central banks, hate even moderate deflation. For this to work, for your $1,000 to blow up in real value, they would have to allow hyperdeflation. Extremely unlikely. They are going to print as much as it takes to prevent more than at most modest deflation.

For you to have your wealth explode in value will require having it in real assets, like stocks, or ownership of raw materials that will be necessary, and relatively rare, in the robot-revolution future, not cash or fixed-interest bonds. This looks to me like the way to hedge against robot/AI unemployment, other than things like improving your and your children's education and skills.

I'll eventually do a post on this, but for now, if you'd like to find out more, please see my comments here, and this recent post.

[4] Of course, there would be a lot of specifics to work out here.

Monday, August 31, 2015

My Response to Shiller on Stock Prices and Historical P-E's

What is the better alternative to the "high" P-E's?

Bonds with near-zero real interest?

Real estate?

Let's consider real estate. A very serious bad future state [1], with a very substantial probability for most people, is artificial intelligence/robot revolution decimating jobs and wages [2]. What will happen to real estate in that state? The 80%+ have had their employment, and employment security, decimated, and wages have plummeted. How is that going to affect their ability to bid on homes and apartments?

And simultaneously, these robots will be able to build new homes and apartments at a fraction of the cost it takes today – Heard of massive 3-D printers on rails printing an entire track of homes' frames [3]? You will. When in the not too far future new homes, apartments, and offices can be built for half the cost, or a quarter of the cost, today, and be much higher quality, with incredible energy efficiency, solar built right into the structure, the walls even, and incredible computer control of the home or office built in seamlessly and beautifully, with amazing new materials for soundproofing, durability, and beauty, how is that going to affect the price of the real estate you're buying today?

So, real estate will plummet in this AI/robot revolution state, while in this same state the tech explosion may be very good for the owners of the robots – the owners of stocks. An asset that plummets in a majorly important bad state? Very risky. One which does well in that state, important insurance. The discount rate for the former is very high, all other things equal (and the former's expected return is also, I think, far lower than almost everybody thinks, because almost no one is considering this AI/robot revolution scenario with regard to real estate investing). The discount rate for the latter is very low, all other things equal.

Yes, there's also the possibility of the winner-take-all explosion getting the bulk of the tech revolution's bounty, but still, the 80% won't be winner-take-all's (WTA's), so at least they can own still needed robots, and raw materials, and patents, and have some negotiating power with the WTA's through the corporations they own, so they get some substantial percentage of the bounty; the WTA's don't get it all.

Because of the great future technological danger to the 80%+, stocks are becoming more and more insurance-like long run, and so meriting a lower and lower discount rate (even though very few people will think about this). And at the same time, bonds are currently paying nothing, and real estate is looking very risky long run.

So, what is better than stocks for long run savings? I ask seriously.

And, especially since Shiller wants to use historical comparisons for P-E ratio, we should also consider historical comparisons for other things too. Shiller writes, "… and within a year or two restore CAPE ratios to historical averages. This would put the S&P closer to 1,300 from around 1,900 on Wednesday."

Ok, so if we're invested in the S&P 500 now (a broader market index is better, like the Vanguard Total Stock Market Index Fund, or one of the many similar offerings available elsewhere) at 1,900, and it drops to 1,300 in 1.5 years, how long until it's back up to 1,900 (adjusting for inflation too, so in today dollars)?

Well, we're going with historical averages, like Shiller, right? So what's the historical average real return on the stock market? Wharton's Jeremey Siegel has possibly the best long run data base. In his seminal book Stocks for the Long Run, 5th edition, he finds that from 1802 until 2012, the geometric average return was 6.6% real, inflation adjusted. And it was about the same for three major sub-periods in this 210 year span (see chapter 5).

So, at 6.6%, how long until the 1,300 gets back to 1,900? Taking out my trusty Hewlett Packard 10bII, 5.93 years. So, in about 7.5 years we're back where we started, inflation adjusted. And that's better than government bonds, which have a somewhat negative real expected return, and about equal to high grade corporates which are at about the expected inflation rate.

But what about after 20 years pass? We keep up with that historical 6.6% average, and after 20 years, 12.5 years will pass since break-even. Those 12.5 years take us to 4,361. So, investing today, with these historical averages, 1,900 goes to 4,224 in 20 years. That gives us a 4.08% real return.

What's going to beat that today? And with a risk level that provides insurance against future inflation (stocks are ownership of real assets which go up with inflation) [4], and crucially, robot/AI revolution causing devastation to employment income and security for the 80%+.

Even if Shiller's right with his historical averages, what's going to beat stocks for long run investment? And yes, if you knew stocks would drop to 1,300 first, then you'd leave until it happened, and then go back. But that's risky to try. It's far from certain they will drop to 1,300. What if things start going really well in the economy? What if the robot/AI revolution starts to really take off within just the next few years? P-E's could start dropping fast not because prices drop fast, but because earnings rise fast.

I still can't see a better vehicle for long run savings than a well-diversified stock portfolio. If you know of one please let me know.

A final point: It is possible that investors are a lot more savvy than in the past, and so less unduly afraid of the risk of stocks. Therefore, they discount their future expected earnings less, resulting in a new era of higher P-E ratios.

So this would argue for stocks not being overpriced at their current CAPE's. INSEAD economist Antonio Fatas just made this point:
The other justification for high CAPE ratios is that the risk aversion of investors has gone down relative to previous decades. While talking about low risk perception this week might not sound right, the reality is that the years while the stock market had CAPE ratios of around 17 where also the years where academics wondered about why risk aversion was so high among investors (what we called the equity risk premium).
So, here we get to the famous equity premium puzzle in financial economics. I've talked about this a lot in previous posts because for years I've had an explanation that I haven't seen in the literature, and I've studied this literature very thoroughly as a finance PhD student, you can't help but. So, please bear with me, and let's talk about this; I really think there are important points to think about.

The equity premium puzzle is that going back a century or two in the US (and this has also been observed in almost every developed market) stocks appear to have had a very high real return relative to their risk. So the question, or "puzzle", is how can this have persisted for as long as two centuries. With people being oh so efficient, oh so smart, expert, and savvy; or, with the expert minority being willing and able to buy as much of assets as it takes to push them to efficient pricing, or so they say, why wasn't the return on stocks bidded down?

Or at least, why have most people put so relatively little of their savings into stocks when the expected return relative to the risk seems so good?

And the explanations that have gotten published in the top journals basically come down to maybe for some reason stocks are riskier than they appear to be, and the general public understands this better than finance professors with massive statistical, mathematical, theoretical, and empirical knowledge. Or, people are a lot more risk averse than they appear to be.

So there's this thought that if people discovered what a good deal stocks are, they would start investing in them a lot more, and that increased demand would push up the price of stocks, and thus down their expected return from that point on.

So, it's always a thought about demand. Once people wake up and see that stocks aren't that dangerous for getting so much higher a return on average, then they will start bidding down the expected returns permanently, until they are no longer such an abnormally good deal.

But, what I ask is if this were any other good, would we think that once consumers change their attitudes and greatly increase their demand, the price of the good must go up, not just over the short run, but over the long run too?

And the answer, from long established and well proven economics, is not necessarily at all.

It comes down to the production technology, the supply side. Is it constant returns to scale, the old CRTS, or increasing returns to scale, IRTS, or decreasing returns to scale DRTS, or is the supply fixed, perhaps like gold (although, while the supply of gold on the planet is fixed, the supply available to the market is much less fixed; it depends on surveying and mining technology).

And whether production is CRTS or IRTS or DRTS can depend on the quantity and time horizon. For a large range of quantities production may be IRTS, then it may go to CRTS after that, and finally if you keep producing more it will eventually go to DRTS.

Well, with stocks, what are you purchasing? Ultimately, the money put into stocks goes into corporate investment projects. And those investment projects ultimately produce goods. Now, what I argue in my explanation of the equity premium puzzle is that investment projects financed with equity, rather than debt, are intrinsically more productive. It's like a technology that produces more output. If you produce with technology X, for every $100 you invest in that kind of production you will permanently get $3/year of goods out. But if you invested that $100 instead with the more efficient technology Y, then you would produce more, $7/year.

And why might equity-financed investments be more efficient and higher productivity than debt-financed? Because equity financing gives managers the option much more to take much longer term projects whose exact fruition and cash flow is harder to predict with certainty, but have nonetheless a very high risk-adjusted expected return. If the financing is in debt, often these projects will be forgone, because the debt provides an often very substantial risk to managers from taking on these projects. If the cash flows don't come in in time, the company can be put under severe financial distress, really costing the manager who took on that project in his career, in just keeping his job. Equity is flexible, there are no fixed interest payments which must be made on precise schedule to avoid very bad consequences. So, basically we can expect, given this, that equity-financing will persistently provide a higher risk-adjusted return than debt. 

Now, what if suddenly people recognize this much more, and start investing much more in equity to take advantage of this, say twice as much. Will they bid down the risk-adjusted expected returns on equity? Well, not if there exist twice as many good equity projects. So, in other words, if the good equity type projects are CRTS over this range, then the increase in demand won't move the price. The long run supply curve is flat over this range, and so an increase in demand does not change the price. In fact, if the potential equity based projects are IRTS an increase in their demand would even push up the risk-adjusted expected return, allowing for excellent large scale projects that otherwise wouldn't have been possible.

So, this is my supply side explanation for the equity premium puzzle. For more, I have a brief write up here.  And so when Fatas writes, "But I wished that he [Shiller] would have considered as well the third possible scenario where current CAPE levels are fine and investors should get used to lower-than-historical returns but returns that are consistent with what is going on in other asset classes.", he's saying that maybe now investors are more savvy, and so are willing to invest a lot more in stocks and so will push down the risk-adjusted expected return, but be fine with that. What I'd reply is that even if what he hypothesizes about the demand side is true, it will not necessarily mean lower future risk-adjusted expected returns for stocks due to my supply side explanation, where higher demand will just mean more of these high risk-adjusted expected return equity-type projects get financed. And less will be financed through less efficient debt. Which may be a very good thing.

[1] The state of nature framework is standard in academic financial economics. You say in the future there are a number of possible states, where various things happen, or various levels of good or bad economic factors occur. And each state has a certain probability of occurring.
[2] To learn about this I recommend, The Second Machine Age (2014), by MIT Economist Erik Brynjolfsson and MIT information technology expert Andrew McAfee, and Rise of the Robots (2015), by software expert and entrepreneur Martin Ford. Ford's not an economist, and the economics is often off, or quite wrong, but the technological information is amazing, and often surprising.
[3] See Rise of the Robots above in endnote 2
[4] I'm not talking about Zimbabwe inflation, but over the next decade or two it is possible we'll start intelligently considering the dangers of lowflation, the ZLB, and lost decades, and raise the target to 3 or 4 percent, even 5, or at least make the current 2% a target, not a ceiling.