It’s often thought that what we have to fear from automation and AI is super-robots. Acemoglu and Restrepo make the useful point that what we actually should fear is mediocre robots, robots only slightly better than humans. Think about robots replacing labor in various tasks. A super-robot replaces labor but has an immense productivity advantage which generates wealth and increases the demand for labor elsewhere. A mediocre-robot replaces the same labor but doesn’t have a huge productivity advantage. As a result, the mediocre robot is the true jobs killer because it replaces labor without greatly increasing wealth. Think about automated phone systems or chat bots.
In an empirical breakdown, Acemoglu and Restrepo suggest that what has happened in the 1990s and especially since 2000 is mediocre-robots. As a result, there has been a net decline in labor demand with no big wealth increase. Thus, Acemoglu is more negative than many economists on automation, at least as it has occurred recently.
More generally, Acemoglu and Restrepo create a new type of production function and use that to reformulate how we think about production and how we measure what is happening in the economy with automation and AI. This is one of the most important new pieces on automation and the economy.
Men tip 12 percent more if their driver is a woman, but that’s entirely because they give more money to the youngest female drivers. The premium men pay to women behind the wheel shrinks as the women get older. By the time the drivers are age 65, it has virtually vanished. Women also tip other women more, but they don’t significantly change their tips based on the driver’s age.
Tips are highest between 3 a.m. and 5 a.m., and not surprisingly:
Tips are highest in small cities and the middle of the country. Riders in California and the Northeast weren’t great tippers.
The other day I asked whether our intuitions about minimum wages and also occupational licensing might be consistent. In particular, if we think occupational licensing is very bad for employment and prices and welfare, is that consistent with monopsony/low elasticity of demand for labor models?
That is a tough problem, here is one approach:
If you think minimum wage hikes are fine, typically you believe something like:
“A 20 percent hike in the required wage will not much damage employment, if at all.”
What then would you say to this?:
“A 20 percent training surcharge on all worker hires will not much damage employment, if at all.”
It seems you should believe the second proposition as well.
Now, consider occupational licensing. Typically it is not absolute (“only 300 goldsmiths in Florence!”), but rather it imposes a surcharge on entrants. They have to pass a test, or undergo training, or receive a degree of some kind. They must incur training costs to get the license, and you can think of those training costs as a tax on the employment relationship. But if those costs are incurred, a worker passes through the permeable membrane of the licensing restriction into the active labor force pool of that sector.
Of course we all know that a tax can be borne by either side of the market, depending on elasticities.
Now, if you believe minimum wage hikes don’t much harm employment, you believe the demand for labor is relatively inelastic. And if you believe the demand for labor is inelastic, the burden of the training costs for licensing fall on the employer, not the worker. Taxes fall on the inelastic side of the market.
Now, you’ve already assented to: “A 20 percent training surcharge on all worker hires will not much damage employment, if at all.”
So the occupational licensing should not much damage employment either. The employer simply picks up the tab, albeit grudgingly.
(The effect on consumer prices will depend on market structure, for instance you can have a local monopsonist shipping into in a largely competitive broader market — tricky stuff!).
The occupational licensing will not help workers as the minimum wage hike would, because there is (probably) greater rent exhaustion in the licensing case. The workers get higher wages, but they are paid the higher wages precisely to compensate them for and pull them through those arduous training programs.
So the licensing and the minimum wage hike are not equivalent, for that reason alone. But still, the licensing will not really harm the interests of the workers, again the burden being born by the employer, or possibly the consumers in the retail market to some extent.
So if you are finding occupational licensing results that damage overall worker welfare, you must not accept the premises of the low price elasticity demand for labor model!
Another way to put the point is that the occupational licensing papers are testing some of the common presumptions of minimum wage models, and flunking them.
First addendum: It is not an adequate reply to this post to reiterate, with multiple citations, that minimum wage hikes do not lower employment. Even assuming that is true, other simple models will generate that result, without clashing with the occupational licensing studies. For instance, the employer might respond to the minimum wage hike by lowering the quality of some features of the job. In essence you are then suggesting the demand for labor may be elastic, but the real wage hasn’t changed much in the first place, and then it is easy enough in the occupational licensing setting for the burden of licensing to fall on the class of workers as a whole.
Second addendum: There is a longer history of minimum wage assumptions not really being consistent with other economic views.
Have you ever heard someone argue for wage subsidies and minimum wage hikes? No go! The demand for labor is either elastic or it is not.
Have you ever heard someone argue for minimum wage hikes and inelastic labor demand, yet claim that immigrants do not lower wages? Well, the latter claim about immigration implies elastic labor demand.
Have you ever heard someone argue that “sticky wages” reduce employment in hard times but government-imposed sticky minimum wages do not? Uh-oh.
It would seem we can now add to that list. Maybe we will see a new view come along:
“Labor demand is elastic when licensing restrictions are imposed, but labor demand is inelastic when minimum wages are imposed.”
Third addendum: Of course there are numerous other ways this analysis could run. What is striking to me is that people don’t seem to undertake it at all.
One frequent theme is people objecting to a price increase. In Ecuador, a focal point of the protests has been a demand for restoration of fuel subsidies. Petroleum price subsidies also have been central to the Haitian protests. In Lebanon, citizens have been upset at a new tax levied on the use of WhatsApp, with a social media tax also having been an issue in Uganda. In Sudan cuts to food and fuel subsidies have been a major complaint. In Chile they are protesting subway fare hikes.
The trend is that price increases may continue to become less popular. And, crucially, the internet will help people organize against such changes.
Consider that an old-style labor-oriented protest can be organized through the workplace or plant itself, through on-the-ground techniques that long predate the internet. There is a common locale and set of social networks in place, including perhaps a union. Those who suffer from a price increase, in contrast, typically do not know each other or have common social ties. Just about everyone buys gasoline, either directly or indirectly. The internet, however, makes it possible to mobilize these people into protests with prices as the common theme.
In other words: Protests of workers seem to be becoming less important, and protests of consumers are becoming more important.
You may recall that one of the original demands of the “gilets jaunes” protests in France was for free parking in Disneyland Paris. If you think that sounds a little crazy, you haven’t yet internalized the nature of the new millennium.
In the future, efficiency-enhancing or austerity-induced changes in prices may be much harder to accomplish politically. The new trend is neither central planning nor market liberal reforms, but rather frozen prices, especially when those prices are set in the political realm.
Here is the rest of my latest Bloomberg column on that topic. Two further points: my global warming point I pulled from Noah Smith, though I could no longer find his tweet to cite. Furthermore, many of the recent protests, such as in Spain, fit a more political and ethnic model, I am not saying price increases are always the major factor.
Larry Summers is my favorite liberal economist because even while maintaining his liberal values he never stops thinking like an economist. That makes him suspect among the left but it means that he is always worth listening to. The video below with Saez, Summers and Mankiw (with Rampell moderating) is excellent throughout. I cribbed a number of points from Summers:
“I have studied last week’s twitter war very carefully and I have to say that I am 98.5% convinced by the critics that the Zucman-Saez data are substantially inaccurate and misleading.”
The arguments around political power are not persuasive. Most of what is wrong with politics is because that is what the people want (I’m filling in a bit here from comments throughout). A wealth tax does nothing about corporate lobbying and would increase the incentive to give to political organizations. If you cut wealth at the top by 30% that wouldn’t change relative political power in the slightest.
Wealth is up in large part because interest rates are down which means that permanent income hasn’t increased.
Forced savings programs like social security and unemployment insurance mean that people at the bottom need to save less and thus their wealth falls even as their welfare increases.
A wealth tax increases the incentive to consume instead of save and invest.
On employee stock ownership plans: “When you put workers in control of firms and you give them substantial control–see Israeli kibbutz’s, see Yugoslav cooperatives, see universities where faculties have a powerful voice–the one thing you do not get is expansion. You get more for the people who are already there. That does not seem to be an attractive position for progressives.”
In the Q&A Summers just goes to town on Saez when Saez claims 90% tax rates are a great American invention. “The people who were around in the Kennedy administration who were at least as progressive as you are were united in the belief that 90% tax rates were a bad idea….The number of people who paid those 90% tax rates was trivial and it wasn’t because there weren’t a lot of rich people.” Greg Mankiw, who gives a nice parable in his remarks, has to stifle a laugh as Summers lets rip.
The body language in the Q&A is very interesting.
An excellent short essay, with many points of note, here is one:
In Himalayan villages like mine, there is deep social uncertainty because of Airbnb and other online marketplaces. The opportunity cost of doing business with one’s nephews and cousins is now high. There is the real problem of nephews who run away on the flimsiest of pretexts. The stakes are higher, and there is much to gain by trading with outsiders. You can’t even run Airbnbs well without breaking free from closed relationships with your family and tribe, and forming spontaneous relationships with strangers. It’s hard for me to do justice to my Airbnb listings without being free to run them in a fairly entrepreneurial fashion.
And there is this:
Millions of people stay in Airbnb homes every night. It’s not trust which makes this possible. My pup is fearless when he sleeps with the door wide open, in a cottage in the woods. There are leopards around. Dogs here don’t live very long. He doesn’t trust leopards, but he knows they are afraid of humans. My pup sleeps on my bed, and so is well-protected from the vicissitudes of life. But I’m not the living proof that dogs can trust leopards. Dogs wouldn’t need humans to guard them if they could trust leopards. Similarly, Airbnb puts hosts and guests in a position where behaving badly would ruin their reputation. In one of my bad moods, I held my pup quite firmly. At midnight, he ran out of the cottage and barked for hours. I couldn’t bring him back to my bed. I did something he thought I wouldn’t consider. He felt I betrayed his trust in me. I’m, here, talking about a more meaningful form of trust. Intellectuals miss this obvious distinction, because they’re not the wonderful people they think they are. The distinction between trust and assurance is all too obvious. But if doing wrong doesn’t fill you with moral horror, you won’t get it. You can’t trust anybody who doesn’t feel that way, and there are not many such people. Unconditional trustworthiness is one of the rarest things in the world. Institutions can’t produce this kind of trust, because people aren’t conditionable beyond a point. In any case, how do you produce something you don’t even understand?
By Veridici, and I believe Shanu Athiparambath.
From Emergent Order the studio who brought you Keynes v. Hayek (and round two) comes Marx v. Mises. All hail to anyone who can rap, “Now here comes the bomb via Von Bohm-Bawerk.” I was also pleased to see Marginal Revolution makes an appearance as does the socialist calculation debate. Useful background notes here.
Would the two percent wealth tax apply to muni bonds? Because of their tax advantaged status, muni bonds are generally held by the wealthy, who get enough of a tax advantage to offset the lower yield. A wealth tax presumably causes more “reach for yield” among those affected, which would disproportionately affect munis.
On the other hand, a wealth tax that excepted wealth held in munis would create a massive tax advantage for them at the high end, much greater than their current income tax exemption.
That is from John Thacker in the comments. And, for the case where the wealth tax would apply to more people than just the very wealthy, Dallas ponders:
How would a wealth tax impact the fat civil service defined benefit pension plans? If you look at the actuarial value of my friend’s public pensions they have values in the 3 million+ range (up to 90% of a spiked salary at 55 years of age for life no-cut contract with a cost of living clause: if you claim disability, it becomes tax-free). A 2% wealth tax on that value would be $60,000+ per year.
Of course, since the people imposing the wealth tax would be bureaucrats with defined pension plans, they would be an asset (wealth) that is excluded and how can you charge a tax against an unfunded liability. Meanwhile, people like me who saved for his retirement would have their assets stolen (perhaps to fund that unfunded liability of the ruling bureaucrats).
The details of a wealth tax with the added variable of time would become even more complex than even the income tax system. With most long term assets value only becoming apparent upon sale having any real long-lived asset would become economically insane. You want some asset with near-zero value (as determined by the IRS bureaucrat) until the year you sell it. That will create a whole new class of privileged assets.
Ponder away on that one…
It covers privacy, Facebook, security vs. competition, Huawei, the proposed digital tax, and recent OECD corporate tax proposals. Here is the video.
From an email:
The eye-catching result here is they have consumption taxes being *sharply* regressive, e.g. 12% for the lowest income group. I’m not aware of any US state that has state + average local sales rates tax that high. And lots of goods are exempt from sales tax. So how do they get this? Well, suppose someone earns $1k in labor earnings and gets $9k in transfers, and consumes it all paying a 5% sales tax = $500 in tax. What sales tax rate have they paid (as a % of their income)? The method Treasury uses says 500/(1k+9k) = 5% (this is also what Auten-Splinter do). Saez-Zucman exclude transfers from the denominator, and thus say 500/1k = 50%. This is a matter of definition, so it’s hard to call it right or wrong, but it does seem misleading and yield some rather nonsensical implications. For example, it means that if welfare to the poor is increased, this will be measured as an increased tax rate.
Indeed, Saez-Zucman themselves seem to realise that this definition yields extreme numbers at the very bottom, where consumption tax rates can easily exceed >100%. In their appendix – https://eml.berkeley.edu/~saez/SZ2019Appendix.pdf – they note “People with very low pre-tax income (below half the federal minimum wage) earn transfer income (temporary assistance, SNAP, supplemental security income, veteran benefits, etc.), which is not part of pre-tax income. They pay sales taxes on that transfer income when it is consumed. As a result, they have high (sometimes very high) tax rates as a fraction of their pre-tax income. We avoid that problem by restricting the population to adults with more than half the minimum in pre-tax income.“
This is quite remarkable. If the sensible way of defining tax rates involves excluding transfers from the denominator (as they claim), the fact that it leads to very high rates by construction at the bottom should be because this is a sensible summary of reality. Yet, in their own words, it’s a problem. Rather than switching method, they drop the people at the very bottom which conveniently covers up the problem (but leaves a less severe version of the problem in their remaining lower income sample). Of course, they could have just used the standard definition which includes transfers in the denominator, but doing this destroys the entire headline result.
It also seems noteworthy that in choosing to excluding transfers, they nonetheless retain payroll taxes. It seems pretty egregious to call payroll taxes regressive when social security is implicitly an insurance scheme with a very large degree of aggregate progressivity, but this is a minor point by comparison.
That is the topic of my latest Bloomberg column. Mostly I am pro-prediction market, but my last two paragraphs contain the cautionary note:
Prediction markets have another potential flaw: They focus attention on clearly demarcated events that are easy to bet on, such as who will win an election or whether Rudy Giuliani will face federal charges. Sometimes these are important matters. Other times they are not.
There are more meaningful trends that are more difficult to measure, such whether Americans are feeling more lonely. These things certainly have an impact on politics, but they are not easy to bet on. Political prediction markets are undeniably useful and very often enlightening, but maybe they should come with a warning: Feel free to check the odds as often as you like, but do not let your obsession blind you to the larger issues at stake.
There is much more in the earlier parts of the piece.
…there is no logical or physical reason that the work year of a machine should not actually increase, say. But it would seem more likely that increased leisure over the next century should be accompanied by a smaller stock of capital (per worker), smaller gross investment (per worker), and thus a larger share of consumption in GDP. Of course, this tendency will almost certainly be offset by an ongoing increase in capital intensity, even in the service sector. Obviously there are other, totally moot, considerations. Will leisure time activities be especially capital intensive (grandiose hotels, enormous cruise ships) or the opposite (growing marigolds, reading poetry)? Show me an economist with a strong opinion about these things, and I will show you that oxymoron: a daredevil economist.
Of course if you really do think the capital-labor ratio will be falling, investment behavior is going to disappoint for a long time to come. The shift to intangible capital will strengthen that tendency all the more.
p.s. Leisure will become especially capital-intensive, at least in the United States.
Unusually for Africa, Ethiopia has sustained a high investment rate: currently 38 per cent of gross domestic product.
Here is more from Paul Collier at the FT.
Emmanuel Saez and Gabriel Zucman seem to think the correct answer is to assume that there is no substitution away from capital or from the corporate sector:
This paper proposes a new way to do distributional tax incidence better connected with tax theory. It is crucial to distinguish current distributional analysis from tax reform distributional analysis. Current distributional analysis shows the current tax burden by income groups and should assign taxes on each economic factor without including behavioral responses: taxes on labor should fall on labor earners, taxes on capital on the corresponding asset owners, and taxes on consumption on consumers. This allows to distribute both pre-tax and post-tax current incomes and measure the economically relevant tax wedges on each factor without having to specify behavioral responses. Tax reform distributional analysis shows the impact of a tax reform and should describe the effect on pre-tax incomes, post-tax incomes, and taxes paid by income group separately and factoring in potential behavioral responses. Various scenarios can be considered given the uncertainty in behavioral responses. We illustrate our methodology using a simple neo-classical model of labor and capital taxation.
No Western fiscal authority I have heard of thinks of tax incidence in these terms.
There is an argument that you first write down the “no-response” burden in order to arrive at the actual estimated burden, as the authors seem to note. That is not an argument for coming up with a “no adjustment” estimate and marketing it to The New York Times (and others?) as correct and based on normal assumptions, without first adjusting for incentives and capital responses and shifts in the ultimate tax burden. Would we have known about these underlying assumptions — which lie behind their subsequent calculation of wealth inequality — at all, if not for the tireless work of Phil Magness and Wojtek Kopczuk on Twitter?
Returning to the paper, it has some quite weak sentences, such as: “But it [no adjustment] also has the advantage of not being dependent on assumptions on behavioral responses.”
You might as well argue that assuming zero price elasticity of demand “has the advantage of not being dependent on assumptions on behavioral responses.” In reality, one is assuming about the least plausible behavioral response possible.
Here is some background material from Wojtek Kopczuk, which works through how the proffered inequality measures and corporate tax assumptions are related. And from Steven Hamilton. Here is also the recent David Splinter summary analysis on tax progressivity. Wojtek notes in his Twitter thread:
The bottom line: corporate tax should be felt by other forms of capital. That’s the standard assumption. CBO makes it, Auten-Splinter make it, Piketty-Saez-Zucman make it. Who does not? Saez-Zucman (2019) do not.
Here is the semantic innovation from the Saez-Zucman paper:
We think it is more useful to say that cutting corporate taxes could increase workers’ wages rather than say
that the tax burden on workers would fall.
Say both! Here are two well-known and also generally accepted AER papers suggesting that the corporate income tax places a burden on real wages.
Michael Smart agrees with me on the new Saez-Zucman piece:
Zucman has now kindly posted an early working paper to support the SZ assumption. I do not find this WP convincing. We’re simply told that the “natural description” of tax incidence is its legal incidence, i.e. 100% shareholder incidence of CIT.
I find this episode appalling, and I hope The New York Times is properly upset at having been “had.”
France among other nations has been calling for a three percent digital tax, for instance as might apply to Facebook revenue connected to France but booked say to Ireland, which has a lower corporate tax rate. (The exact meaning of “connected to France” is indeed murky here, if you are wondering, but proponents might have in mind a simple France-to-France transaction, such as selling an ad to a French buyer for a French product; there are more complicated grey areas.)
As is so often the case, the debate is focusing on how little tax some of the major tech companies pay directly to the French treasury, rather than on tax incidence. In reality, the major tech companies may already be bearing a quite significant tax burden.
Let’s say you believe that Facebook has significant market power over the advertising market in France. That is not exactly my view, but let’s run with it — a competitiveness assumption will hardly boost the case for taxing Facebook.
At this point your mind already may be thinking that the monopolist in the supply chain will bear some significant portion of a tax, just as land bears tax burdens in a Georgian land monopolist model.
Let’s now say that France boosts its VAT — how will that impact Facebook? Well, the short-run effect is that directly taxed good and services will tend to cost more. That in turn will create pressures for them to advertise less, because their potential market size and potential profits are smaller. If they advertise less, they are spending less money on Facebook ads. Facebook profits go down (remember, Facebook is selling those ads above marginal cost), and thus Facebook bears some of the burden of the tax.
Do the same analysis in terms of levels rather than changes, and you will see that Facebook bears some of the burden of the current French VAT.
So the French VAT brings money into the French treasury, and some of that money comes from Facebook in an indirect form, in addition to whatever direct tax liabilities Facebook may bear under the current French VAT structure. Furthermore, the net tax burden on Facebook is higher, the more monopolistic is Facebook in the ad market.
I should note that there are other ways you can play around with the assumptions.
A good rule of thumb is that you should place less weight on tax discussions that do not focus obsessively on tax incidence.