Assorted links


Re #6, Krugman almost sounds Sumnerian in his frustration.

Yes, there is an unequivocal and obvious reason why a **temporary** 'bridge' would have been useful for the US, because unlike Japan which is a relatively equal society with insanely high savings rates, most of our population excepting our plutocratic overlords is a massive underclass crawling out from under a mountain of debt that must constrain spending as anyone who is serious about this issue would immediately recognize.

Hyperbole much? Shall we compare median income between the U.S. and Japan? Let me give you a clue, the U.S. is 50% higher.

If equality is what Japan has, I would prefer to be unequal that is for sure. Somehow, giving your entire life over to one company, living in cramped house and taking even more cramped public transport is supposed to inspire feelings of envy? And that is not even getting into the sky high food and energy costs that the average worker faces. That is why Japanese workers do so badly on a purchasing power comparison.

Don't get me wrong. Japan has plenty of good points. But you ain't going to get this poor down trodden blue collar worker to envy the life of a blue collar worker over there.

That's pre-2000 Krugman, undoubtedly.

An issue I haven't heard much about: How do we avoid driverless vehicles turning into a Minority Report like scenario (you get in the car and the government can lock you in and redirect where it goes)? I like the idea of safer transport, less traffic, and less fuel cost, but how do you avoid handing over the management to the government? Is it legal or technological? Who is driving that bus (pun intended) right now?

How do we avoid driverless vehicles turning into a Minority Report like scenario (you get in the car and the government can lock you in and redirect where it goes)?

Through the same kind of mechanisms we already use to limit the government's power -- legal requirements such as warrants, court orders, probable cause, etc.

That doesn't work now, what makes you think it will work when they have more power.

I think it does work now. If you don't, you've got much bigger worries than self-driving cars.

You make it so the vehicle cannot be remotely operated. Or if it can be, you make sure the system is secure and open source.

" the government can lock you in and redirect where it goes"

Why do you need self driving technology for that? A cop with a gun suffices and already does it quite well.

if autodrive cars become popular, police => public interaction will go down by like 95%.

Cops will:

-lose huge chunks of their income from tickets,
-lose the chance to find probable cause to escalate a stop into a search
-lose most of their chances to check IDs for warrants
-lose the chance to profile (valid or not) suspicious people
-most cops out on traffic patrol will never be able to catch anyone - so why are we paying them?

How will they react? probably by trying to come up with lots of new things to do, new ways to talk to people, make lots of things illegal that are currently legal.

Oh holy! Good find on Krugman.

Please, somebody let Paul know. He will be relieved.

" Suppose that the factors depressing private spending are clearly temporary - for example, there is a clearly temporary financial crisis underway, or investment is on hold pending some sort of financial cleanup, etc.. That is, there are good reasons to think that EE will shift up in the not-too-distant future in any case; so propping it up artificially with fiscal stimulus is simply a holding action until the cavalry arrives. "

Krugman clearly thinks this is the case ("balance sheet recession") and I don't think that's an unreasonable claim in the least.
I sometimes prefer Slate Krugman to NYT Krugman, too, but it's important to remember why Krugman shifted: Because he became increasingly aware of the fact that one side wasn't debating in good faith.

Anecdotally, Richard Koo complained to me that it took him 10 frustrating years to finally get Krugman on board with his balance sheet recessions thesis.

Poor Krugman changed because his mean opponents were not debating in good faith? I used respect Krugman quite a lot when I was a teen (during the Asian crisis) and I seem to have clearer memory for the man's past then most of this friends. I can assure you, Krugman was just as condescending and arrogant towards his liberal opponents as he is to his conservative opponents today. Back when I was younger I appreciated it more than I do now.

I guess it speaks well of him that he is an equal opportunity jerk but I don't think it is fair to blame his current opponents for the change.

I'm not blaming the fact that he's arrogant and condescending on his opponents - that's probably a combination of being socialized as an economist and getting a ton of positive affirmation in his mid-career.
I'm attributing his much more partisan outlook to the fact that during GWB's first presidential campaign he found Republicans to be lying and deceitful about their planned economic policies. That experience and the continued lies of the Bush 43 administration - including the Iraq war, which Krugman was one of the few people in the mainstream to call correctly from the start - radicalized Krugman and made his writing more partisan. That's what he says, btw., not my hobby psychoanalysis. The fact that he was proven right about a lot of this - Iraq war, housing crisis as a serious problem (I know he wasn't among the first people to warn, but in comparison to most conservative pundits he was way ahead) etc. probably didn't "help". And so Krugman adapted the underlying assumptions of his analysis.

Re: #2, driverless cars, horizontal drilling, 3-d printing, etc. may be enough to put to bed the great stagnation (going forward) thesis. Karlgaard had a nice, concise piece on this awhile back:

Or prove the race against technology theory thesis. No more cab drivers, (non-union) truck drivers, fewer sales clerks, etc.

The rock star status conveyed by the Nobel changed Krugman from an economist seeking truth to one seeking to bask in the limelight. Pre-Nobel, he realized that his opinions would be judged based on their merit. Post-Nobel, he has counted on his opinions being judged based solely on the fact that he had won a Nobel prize. Today, he is more concerned with his status as a "star" in the largest circle that will fawn over his words rather than being concerned about how he is thought of by his peers. As such, he basis his opinions not on facts and sound economic theory, but on what the media wants to hear and because the media is chiefly composed of elements of the far left, he tailors his opinions to please his new masters. If it weren't so sad that a former Nobel prize winner had to grovel at the feet of the media, the contrast between his new and old views would be hilarious.

I will repeat myself. The myth of the nice pre-Noble Krugman and the mean Krugman post Nobel is a myth. He was always abrasive (which is one of the reasons he got shut out of the Clinton Administration while it was still in its liberal stage. The only thing that has changed is that he is doing more non-scholarly work where his abrasive side comes out more clearly.

I agree. He's always been an short asshole with an oversized ego. It's just that pre-Nobel he was an asshole with carefully crafted opinions, whereas post-Nobel he's an an asshole whose ego depends on media attention and who is easily offended when his authority is questioned. When you say his abrasive side comes out when he's doing non-scholarly work, I think we basically agree. He spouts off opinions that are based more on a leftist agenda than on serious scholarship and when he is challenged, having no scholarly defense, his response is even more caustic than usual. His Nobel "crown" has caused him to become lazy and as a result he defends his positions with bullying and abrasiveness instead of intellectual rigor.

The Sumner story requires multiple equilibria too, right?

If we all believe NGDP will be low then none of us will invest. If we all believe NGDP will be high we all invest because your investment makes mine likely to be profitable. Monetary NGDP targeting might keep us from falling into the bad equilibrium, but once in the bad equilibrium is it fortified FOMC communications policy that gets us out? Or do we need to spend our way out? Or is QE the same as spending our way out?

For those wondering the date on Krugman: Time on the cross: can fiscal policy save Japan? (9/21/99)

"The Sumner story requires multiple equilibria too, right?"

It's here that Sumner (to my ears) makes NGDP targeting sound like a coordinating mechanism to get out of a bad equilibrium and into a good one,

Along with (not Sumner)
followed by
in my post on negative demand externalities. The Fed's role can be viewed not as turning on the NGDP machine and creating more NGDP to distribute to private firms and households but as coordinating expectations across the private sector in order to arrive at a better equilibrium. If the Fed fails to play this coordinating role, the market may not be able to clear on its own. Firms are now building more and selling more, but we need them to build and sell a lot more if the stock of unemployed workers is to return to normal levels. But right now, firms are doing the equivalent of standing around saying to each other, "Well, I'll build a lot more if you buy a lot more", "Well, I'll buy a lot more, if you hire a lot more", and around and around. The Fed can and should put out word that now is the time for everyone to take the leap.

The philosophy post this morning left me wondering (among other things) what are "thought paparazzi." #6 seems like a good example of it ... snapping thoughts of big name thinkers out of context.

To add a little context back in, I thought the introduction page that set this essay up on Krugman's website was an interesting read: Also I do not really see the gotcha moment here. Just like I don't see it when Sumner (and others) compares Bernanke the academic with Bernanke the central banker. Thinking evolves, but more importantly the world changes. Seeing two economists to agree on anything, almost 100%, seems like a reason to celebrate to me.

Context is everything.

For example, had the BOJ expanded the money supply yet or aided the banks? No. First do that. Was there a shortage of aggregate demand. Japan was still exporting, and its banks were not lending, because they hadn't written off their bad loans.

Context is everything--except for a blogger who prescribes the same policies no matter what the context....a percent who agrees 100% with Krugmans comment----without telling his uninformed audience the time and context of the Japanese economy, what steps HAD NOT been taken, and which should be taken first.

Of course, if you want an uniformed audience, I suppose you could just post a quote and wait for the Krugman pinnata effect.

Now, let me see if I can find a quote from Friedman to take out of context--or rather not disclose context--so I can make a little trouble here.

"and its banks were not lending, because they hadn’t written off their bad loans."
Thats what bankruptcy is supposed to fix.

Your biting at Tyler's ankles is pathetic.

The suggestion that he's purposely hiding context, the better to fool the foolish commenters, is belied by the general trend of comments, which more or less was able to place the context of the article. A more plausible story is that he assumed his readers would be able to place the article in context, and he was vindicated in this assumption.

But your condescension is almost pitch-perfect mimicry of The Master. Bravo.

Why, this has gotten pretty modern now: search for some Krugman text from the past - go back to the 70s, if necessary - link to it, and imply something, though without being clear what that could be. (Usually an extremely subtle message for the more sensitives among us, like "Things are complicated, see!, and only this old Krugman piece can show this. Also, look, I hate snark and I especially hate that about Krugman, and I'd never be that snarky and offensive, but here is a link to Stephen Williamson, excellent, I tell you - and pure macro, nothing else, really *wink wink*"). TC didn't have a lot success with that last time around, but perhaps it works now.

Also, did you get the news that Krugman wrote an article on Friedman, thereby basically spitting on Friedman's corpse - and all the important implications that has for our daily Krugman talk, so that we should speak about it, now, please? And that Krugman said, some years ago, that he doesn't read conservative blogs, while Friedman clearly would have read Krugman's, or so, and that we should speak about that, now, too? Go check Sumner's blog.

And Delong deleted comments, in 2008, oh the news. Maybe someone should write a blog entry, now? Oh, wait...

I thought Cowen was being perfectly clear. Krugman expressed Cowen's ideas pretty clearly in the past. The only reason for digging up Krugman's work from the past is the amount of contempt that Krugman would heap on Cowen if Cowen expressed the same idea in his own words. It is all in good fun.

@ Ape Man

Yes he is clear about that. As TC doesn't agree with PK now, this implies that Krugman has a) shifted his opnion and that b) TC thinks that PK has done so wrongly.

However, he doesn't tell us why. There is a chance that PK thinks this analysis it pretty silly with regard to the actual situation and that those who don't have undergone his shift should be met with contempt (or perhaps not) Linking to an old PK piece doesn't say us a lot about how silly Krugman thinks (or does not think) it is in the current situation.

Also, note the comtempt TC is perfectly comfortable with if it is thrown towards PK - he'll happily provide you with a link to it and find it insightful, or excellent. He is a serious offender on this front (as Krugman is with his permanent vitriol), and you should stop treating TC as the angel he isn't (not in this respect).

I'm pretty sure DeLong deletes comments all the time. I know I've tried to comment a couple times without success. It's a hall of mirrors in there.

Will the real Paul Krugman please stand up.

'Cause I'm Paul Krugman, yes I'm the real Krugman
All you other Paul Krugmans are just imitating
So won't the real Paul Krugman please stand up,
please stand up, please stand up?

Let's assume that driverless cars come to market in 5 years and work perfectly. Let us assume it is a 5,000 dollar option. At the current rate of US auto turnover for cars with luxury options how long will it take for America to reach this driverless car nirvana? 30 years? 50 years? I surely won't be alive to see it.

We don't need all cars to be driverless to start getting enormous benefits. The most obvious benefit to individual driverless car owners is that they'll be able to use the time and energy they would have spent controlling and navigating their vehicle on more productive or enjoyable activities instead. They'll be able to use their car when they're drunk or sleepy. They won't have to worry about looking for parking. People who are currently dependent on buses and trains to get around will be able to use driverless taxis instead. Driverless delivery vehicles will reduce the cost of buying things online, and will likely make same-day delivery feasible for almost all purchases. External benefits like reduced congestion and accidents will start to appear as soon as there are driverless vehicles on the road. In fact, we don't even need full driverless operation to produce these benefits. Even partially autonomous operation, like Ford's Traffic Jam Assist feature, have great potential to reduce congestion and accidents.

So I take it you didn't read the article?

Extrapolating current turnover into the future, given affordable driverless car tech, would likely be an incorrect assumption.

If it means a 50% difference in your auto insurance, it won't take 30 years. Also, why would it be $5000? The marginal cost is just a few cheap cameras and servos.

Even if it is $5000 it would increase the turnover quite a bit. The NPV of the activities that you could do for an hour a day while commuting (sleep, read, work, not stress out, etc) spread over the life of a car would be much greater than $5000.

Not to mention the insurance savings. And the ability to drink and (not) drive. I think millions of people would be willing to pay $5000 just for that benefit alone, especially considering that a DUI costs thousands of dollars in fines and legal fees alone (never mind the prison time, loss of license, etc). I'd pay an extra $5000 just to avoid the hassles of finding parking (car drops me off at my destination, finds parking on its own, then comes and picks me up when I'm ready to leave). And the lower risk of being killed or injured in an accident. And the ability to send the car to give rides to friends and family when I'm not using it, or even to rent it out as a taxi. And so on. I'm kind of amazed at people who claim to see little or no benefit to self-driving vehicles. The benefits seem to me so large and obvious that as soon as this technology becomes available at a reasonable price I think it'll spread like wildfire. Even faster than the adoption of internet access or cellphones.

Cameras and servos aren't cheap when you have to build and test them to lives-are-at-stake levels of reliability. Even the wiring harnesses won't be cheap at that level. And it isn't just marginal cost that matters - the system won't be built unless it is certain to be sold at a price that also covers the non-recurring cost of developing it. The very cheapest aircraft autopilots - which don't even need the cameras - cost $5000 or so new. Autodrivers will (eventually) benefit from economies of scale, but they will also have to deliver a far higher standard of performance than a simple single-axis autopilot.

And why the assumption that insurance rates will go down? Yes, driver error is the leading cause of automobile accidents. Pilot error is the leading cause of aircraft accidents, for aircraft with pilots. We now have twenty years of experience that replacing human pilots with computers causes the accident rate to increase by roughly an order of magnitude, the accidents now caused by computer error (typically, doing absolutely nothing in response to what any ten-year-old would immediately recognize as a serious emergency).

Self-driving cars will exist, yes, and they will change the economy. Probably for the better, at least in the long run. But self-driving cars so safe that they don't need any insurance at all, that they can be trusted to also serve as untended robo-nannies, that's far enough in the future that it isn't worth thinking about except as a trivial subset of "how will the singularity change the economy?"

Assume self-driving cars exist, but for the first decade they will cause accidents at twice the rate of an average human-driven car while averaging half the speed in city traffic - or, same speed as human drivers but ten times the accident rate. Assume legislators nontheless allow them on the streets, presumably having bought into the hype before seeing the numbers. How does that change the economy?

Cameras and servos aren’t cheap when you have to build and test them to lives-are-at-stake levels of reliability.

When you're building millions of them, sure they are. Cars already have cheap cameras built to lives-are-at-stake levels of reliability. Rear-view cameras for backing up, forward-view cameras for collision avoidance systems, side-view cameras for lane-keeping systems.

And why the assumption that insurance rates will go down? Yes, driver error is the leading cause of automobile accidents. Pilot error is the leading cause of aircraft accidents, for aircraft with pilots. We now have twenty years of experience that replacing human pilots with computers causes the accident rate to increase by roughly an order of magnitude,

Huh? Show me this alleged data that autopilots increase accidents. Planes are both more automated and safer than ever. Insurance will go down for self-driving vehicles because they will be much, much safer than human-driven ones. Google's fleet of prototype self-driving vehicles have now completed 300,000 miles without a single accident. Automated collision avoidance systems are already reducing the rate of accidents in production vehicles.

I did not say that autopilots increase accidents, though Air France 447 might make a pretty good argunment. I said that replacing human pilots with computers, increases accidents. Autopilots do not replace human pilots but rather compliment them, and presuming the human pilots do not get complacent, the combination of autopilot plus human pilot is safer than either alone.

Actual pilotless drone aircraft, or even remotely-piloted drone aircraft, those have vastly increased accident rates compared to manned aircraft. Quoting from "Unmanned Aerial Vehicle Reliability Study", Office of the Secretary of Defense, 2003:

"These data show a cumulative mishap rate (i.e. class A accident per 100,000 hours of flight) of 32 for Predator, 334 for Pioneer, and 55 for Hunter (16 since the major reliability improvements of 1996). In comparison to mishap rates for manned aviation, general aviation aircraft suffer about 1 mishap per 100,000 hours, regional/commuter airliners about a tenth that rate, and larger airliners about a hundredth that rate. These statistics make it apparent that the reliability of UAVs needs to improve by one to two orders of magnitude to reach an equivalent level of safety with manned aircraft",

To be fair, the reliability of military UAVs should not be compared to that of commercial airliners. But, at least in peacetime use or in combat against an adversary without substantial air-defense assets, the military UAV mission is fairly similar to manned general-aviation operations. And in the ten years since that study, twenty years since the large-scale adoption of military UAVs, the best drones have improved only modestly over the post-1996 Hunter. There really is an order of magnitude difference in safety, and it isn't the steep learning curve of a technology in infancy, and it isn't Al Qaeda shooting down UAVs.

Now, the data supporting your claim that self-driving vehicles will be much, much safer than human-driven ones. Not the anecdotes, speculation, and handwaving, the actual data. Put up or shut up.

John Schilling,

Your claims about UAVs are completely irrelevant to the safety of driverless cars. UAVs are generally flown remotely by human pilots with a display screen and joystick. Nothing in your quote suggests that the high accident rate for UAVs has anything to do with errors or failures in automated flight systems. It doesn't provide any attribution of the "mishaps" at all. UAVs are likely to be more prone to accidents because of their unusual aerodynamic designs and because of problems arising from remote piloting, like interruptions or lags in communication.

Now, the data supporting your claim that self-driving vehicles will be much, much safer than human-driven ones.

I just gave you some examples: Google's accident-free testing of its driverless car prototypes. The reduction in accidents from automated collision avoidance systems. The Insurance Institute for Highway Safety found that automated collision avoidance systems have already reduced crashes by 14% in Mercedes and Acura vehicles, and by 27% in Volvos. And these systems are only going to get better and more sophisticated in future models.

I don't know why you're even questioning this. As far as I can tell, there is virtually unanimous agreement among experts in the field that self-driving vehicles will be much safer than human-driven ones. Computers attached to video, radar, and lidar sensors can measure speed and distance much faster and much more accurately than human drivers. They have much faster reaction times. They will have a 360-degree field of view around the vehicle. They will be able to operate effectively in poor weather or lighting conditions. They won't get tired or distracted or drunk. They won't engage in dangerous or aggressive driving because another driver cut them off, or to impress their girlfriend. And so on.

But why would your insurance go down 50%? There will still be plenty of human operated cars. Don't you think people in a hurry will override the autopilot so they can do 70 in a 55 mph zone when they are late to work?

Your insurance only pays out for accidents you (or your driverless car) cause. The insurance companies for those human drivers still on the road will be responsible for whatever accidents they cause.

Overriding the autopilot probably won't be possible without some record being made. You'll be incriminating yourself if you do it.

The adoption of driverless cars is likely to be comparatively rapid; the benefits are large. But such technology will be OEM equipment, not after-market. Therefore, it will be disseminated as a function of new car sales. By this route, it would take perhaps a decade to become "ubiquitous" (ie, something standard that most people have). So assuming a 2017 introduction, you're talking the second half of the 2020s until "everybody" has it.

Traffic will increase, not decrease, as the cost of travel (now excluding a captive driver) will decrease. Self-driving cars will increase mobility, as they should. The difference, of course, is that perhaps 30-40% of the cars on the road will be driverless, about the same ratio as passengerless taxis in New York City. (Expect some negative editorials on this in The New York Times, cc. 2022).

I believe the price point will fall between $2,500 and $5,000 for such a system (at steady state). At the bottom end, the price would be equal to a good DVD system today for your minivan. So that price point has already been established for a new feature and manufacturers are comfortable with that number. Much above $5,000, and manufacturers know they will be pushing their offering into another price bracket. Someone willing to pay $25,000 for a car is going to hesitate to buy a different version at $30,000. So I personally peg the cost of a SD System at $4,000. It is more valuable that a DVD system, but not enough to push the vehicle into the next price class.

Of course, such systems may first emerge in luxury vehicles like Mercedes and Cadillac, and in fact, we see early SD technology in both models already. However, I would anticipate relatively rapid diffusion as such systems are literally life-and-death for electric vehicle manufacturers--they will want them standard as soon as possible.

Street signs will not go away. SD technology is intended to operate in a heterogenous environment including both other SD vehicles as well as those with human drivers. It is unlikely that it will be illegal to drive your 1961 Thunderbird or 2016 Porsche on the roads for many, many years--certainly decades. Thus, roads will continue to have to assume human drivers are present and signage will be placed accordinly.

Also I am surprised that no one suggested that this will decrease the rate of car ownership. It would certainly help companies like zipcar. If your car spends most of its time parked then computers would help them be more productive wouldn't they?

The Zipcar model will be replaced with self-driving taxis. The effect of driverless technology on car ownership rates is unclear. When self-driving taxis become available, some people who would today reluctantly own a car may forgo ownership and use self-driving taxis instead. But other people who would not own a car today because they can't drive, don't like to drive, or aren't allowed to drive may choose to become owners. Car ownership costs will likely fall over the long term, due to lower insurance costs, lower parking costs and higher efficiency. Owning your own car would guarantee availability of rides on demand. Relying on taxis would mean you might have a long wait for a ride at times and in places of high demand. Manhattan is full of taxis, but New Yorkers are always complaining that you can't get a cab when you need one.

Another famous case of the Krugman vs. Krugman game. Jim Glass famously compiled these gems at Of course Jim, a lawyer of some renown, has a sharp mind and a biting wit. It was a pleasure to read him on sci.econ back in the day but blogs are harsh mistresses today so Jim is now lost to us and only his friends will enjoy.

To play Krugman vs. Krugman it is only necessary to find Paul's Slate columns (or MIT faculty archive) and compare them to his NYT's columns invariably you find libertarian/conservative Paul and then you find the Paul suffering from Bush Derangement Syndrome we imagine Sybil in these moments of aha.

Only that one twenty second recording of the whale?

Ummm did you guys actually read Krugman's article? There are two main ideas in it. First that fiscal stimulus would not work in the long term since Japan's problems were structural. Second that long term fiscal stimulus was not feasible. It seems people are taking the first point as Krugman vs Krugman where it clearly is not. Krugman currently argues that stimulus will work precisely because the problem is NOT structural. It is the second point where you actually see Krugman vs Krugman. And he has acknowledged he was wrong on that point and changed his viewpoint on the liquidity trap.

It's kind of sad that people just see he was against a fiscal stimulus at some point in the past and take that as meaning he is contradicting himself now. Can we maybe take a little time and examine the situation. It only takes like 5 minutes to actually read the piece folks. It was great troll baiting by Tyler however. He says he likes the piece but of course does not say why.

Who said Cowen was against fiscal stimulus? Who said Krugman was against all fiscal stimulus at anytime in his career? I think you are the one who is simplifying.

I see plenty of people here acting like the article shows a contradiction in Krugman's stance on stimulus. And TC intentionally did not mention why he liked that article so that we would all argue about it. But that's what makes reading these blogs fun.

Krugman is pointing out that productive fiscal stimulus is limited by the amount of new public investment that becomes viable as interest rates fall in response to monetary policy aiming at maintaining growth in NGDP and as shadow wages fall with a rise in unemployment.

Did anyone notice that Scott Sumner just made the case for drill, baby, drill.

Finally, an economist other than Jim Hamilton who is not asleep at the switch.

Comments for this post are closed