Adam Gurri writes, Kling clearly sees the need for the administrative state, and this makes him more clear-eyed in his criticism of it. Additionally, his insight that failures of governance can create the political conditions for further intrusions has considerable merit and generalizes well beyond the specific domains he discusses. Finally, his focus on the institutional
Thank you for taking the time to read and respond to my criticism. On the auditor first, I say, fair enough; most follow ups I could make here could be made against *any* proposed reform, which simply isn’t very helpful, so I won’t bother there. All I will do is reiterate your own point that this agency wouldn’t be very different in kind from ones we already have, so it’s unclear to me that the outcome would differ much from reconfiguring those. I also think you might find great interest in Congress Overwhelmed, a bipartisan collection of congressional scholars who are jointly concerned about, among other things, Congress’s diminished oversight capacity.
On the COO, I don’t want to make too much of the market analogy thing, or your statement that he’d have ideally been a top executive at a large corporation previously. Mostly I’m skeptical of such an enormous centralization of power, or that doing it under our clunky appointment system would result in greater accountability. I also again wonder about empirical precedent.
Empirical precedent was part of what drew me to the federalist approach to reform. I agree this can be taken too far. But many of our peers devolve a surprising amount of their governance to federal units. Looking at how Germany and Canada approach this was very eye-opening, especially when comparing to our fiscal federalism specifically. Agree however that ultimately a principle of subsidiary means thinking seriously about the proper level of each governance activity, and some really do only make sense at the very largest. Moving up medical licensure is definitely one such case (and of course, ideally drastically reforming it to be more open to entrants!)
As a foreigner, I find Arnold's posts and comments amazing. Everybody advances ideas to rebuild a system of government which many insiders have worked hard for decades to destroy. Arnold, the barbarians have taken over your government and you want to rearrange the chairs. Please tell me what you would advise to your senile President to get rid of Fauci&Co (if you are thinking about depriving them of their pensions, remember what happened to Andrew McCabe) or at least how you would ask him to go home now without empowering Queen Kamala or Nancy.
Re: "Gurri’s alternative proposal is to allocate more government responsibility to state governments. He claims that local governments are too parochial and not well run. The federal government is too clumsy and resistant to reform."
At stake are two issues: 1) What governance changes are feasible? 2) Which of the feasible changes which would most improve governance?
One may also ask: Which governance changes are more likely to occur?
Bryan Caplan highlights the emergence of "gubernatorial dictatorships" during the pandemic:
My intuition is that the overarching trend is towards political centralization, notwithstanding Brexit from the EU and gubernatorial dictatorships during the pandemic in the USA. In America, there is pressure for States to supersede local governance in zoning and in public education; and for the national government to supersede States in regulation. Witness growth in the DC region.
Therefore, I believe that Arnold Kling's proposal for reform--a COO in national government plus a check by a Chief Auditor--is more likely to occur than decentralization of governance to the States.
What is the probability that Arnold's proposal for reform (COO + CA) will be enacted? I hesitate to say. Would the reform, if enacted, substantially improve governance? Or would it mainly make government bigger by accretion, instead of better? I hesitate to say.
One must multiple the probability of enactment by the probability of improvement. Perhaps the resultant probability is very small.
Consider an alternative: Decentralization to local governance, plus reforms to create exit options. An example is radical vouchers for education, apprenticeships, internships. Perhaps the likelihood of "decentralization + exit options" is much less likely to occur than either Arnold's reform or Adam Gurri's proposal (governance by States). But if decentralization + exit options does occur, it is much more likely to improve governance, thanks to competition, innovation, experimentation, and accountability. The expected improvement--probability of the reform x probability that the reform, if enacted, will improve governance--seems greater in my 'nirvana' alternative, than in Arnold's or Adam's 'realist' reform proposals.
I believe a fruitful avenue forward would be to look at comparative civil service structures. Although we all had a good laugh at the television series "Yes Minister", the UK civil service has always struck me as far more professional that its US counterpart. Unlike the US, the UK actually has someone who is in charge of the entire civil service, the Cabinet Secretary. In many ways he is the COO that Arnold talks about. I'm not sure that there is an equivalent of the Chief Auditor role in the UK. In any event, we have seen massive institutional failure across all of the US federal government and it needs to be fixed. Arnold is one of the few people actually taking that challenge seriously.
Has Kling addressed the COO/Auditor proposal from a focus on the incentives therein? I agree that the idea should be judged against the current state and not an ideal, but even doing this I struggle to see how the incentives faced by the COO or Auditor wouldn't be exactly the same as any other political appointee. Either immediately or over time, this means that the positions would devolve into the same issues we have with Treasury, the OCC, CFPB, etc., no?
I don't think scale or degree of federalism or centralization has much to do with it at all. When I look down the list of countries ranked by population, my personal index of impression of governance quality is extremely volatile and pops up and down. Population size would be way down my list of important correlates.
Lots of people criticize China which famously has the largest population of any nation, but in terms of decrease in poverty and increase in economic production, military might, and geopolitical clout, has anybody else had a better 30 years? Since 2008 their industrial production is up over 150% while in the US it's ... 0%. Yeah, yeah 'comparative advantage', but still. Japan is huge at 125 million and like every place has plenty of its own specific problems, but traveling anywhere in the country you get the impression right away that it's run pretty well, and this even correcting for the huge advantage that government has of being tasked with running a country full of Japanese people. Mexico is the same size, and, well, not so much. Switzerland is the same size as Sierra Leone, and likewise for Israel and Honduras.
More important is the problem of unmanageably wide scope, which makes it hard to imagine that a COO could actually manage something like USG any better than any of the old conglomerates were managed, many of which broke up. GE is a recent example. To the extent management talent isn't purely generic (i.e., that one, true consultant book to rule them all doesn't exist), needs specialized expertise in a particular sector, and benefits from the ability to focus effort and learning on that sector, at some point in broadening of scope any human COO would be outclassed by the sheer immensity of the job. You can only keep so many plates spinning at any one time. One of the classic arguments for small, narrow government is that a government which tries to do everything ends up unable to do anything well.
One sees this happen with districts judges who are supposed to be these ultimate legal generalists as they are expected to be able to competently handle the whole variety of cases, but being merely human often demonstrate shall we say a somewhat uneven competence in the full spectrum of suits.
Or for another example, imagine that aside from the typical kinds of administration that go on in universities, that someone tried to manage the production of intellectual and scholarly output across the entire spectrum of academic fields. Even though this would be a lot more specialized than what any COO would be asked to do, it's still so broad that it's hard to imagine even a genuinely polymathic genius being good at knowing how to do this well for all the schools and disciplines. And I think what we see at many universities is that they have subordinate institutions and programs of uneven quality and when one has a deserved reputation for excellence it is more due to self-perpetuating good management on the inside than supervision from above.
And while academia is theoretically about everything in life, actual life is even more complex, and the trouble is that if life is complicated enough that we cannot escape the need for bureaucracy, then likewise it's too complicated for that bureaucracy to be 'managed', and at some level of scope and complexity a leader has to trade off capacity for genuinely management with something more like trust in the expertise and competence of more specialized subordinates, from whom one has no real choice but to take their word for it, most of the time.
Another classical argument is that this kind of reliance gives rise to principle-agent problems like being in the position of not really knowing whether what your doctor or auto mechanic is recommending is actually strictly necessary. We kind of hand-wave away these situations in a commercial context by saying that people can sometimes try to rely on reputation systems, but there isn't a good system for this for political appointees who are expected to be more loyal to what the boss wants than to some objective measure of 'performance', and who it would be unfair to hold accountable for instances in which they were indeed just following orders.
As for scale of population size, I don't really see any convincing mechanism for why that should be. My basic model is that if things scale, they can scale practically indefinitely. Militaries are a clear example of always being able to expand by cloning and putting another layer on top of the pyramid with an extra star. There doesn't seem to be much basis for the argument that says there is some optimum or maximum practical distance in hierarchy steps from grunt to top general (or SECDEF or President). Facebook and Amazon are 'governed' in a sense, and why not have every person and organization in the world as a user?
I could make a just-so story for democracies like this: Someone can't be a good representative of more than a limited number of constituents, there just isn't enough time, and on the geographic district model, without heroic assumptions about self-sorting, the bigger the number the more diverse in preferences and the less coherent in terms of a genuinely common interest in most matters. This kind of thing is brought up when people talk about the trend in consolidation and growth in population size of American school districts, such that the dilution of possible connection between a parent and a school board member or superintendent, and the number of other people one must compete with to try to petition or get attention, becomes totally distancing.
And likewise, like any group, a parliament designed for interactions between equally-stationed individual peers with perhaps only one extra level of hierarchy for party and committee leaders cannot actually function well or as intended when it goes super-Dunbar. So, if the limit of X is, say, 60K, and Y is, say, 500, then you get an upper bound of 30 million.
But in my view, that story doesn't line up very well to what goes wrong in Congress or in USG more broadly. One way to see this is that most states are like mini-countries run on the same model, only California is bigger than 30M, but even many very small states are still run terribly in ways that seem to closely mirror the terribleness of USG, the Little Sams apparently not falling very far from the tree of Big Uncle Sam.
As for federalism and subsidiarity, Singapore doesn't have any of that and is basically a fully centralized city-state that not just 'runs well', but, fairly or not, is a common go-to example of world class governance.
One reason to favor more federalism is not necessarily that it would produce better governing schemes, but simply because in the nature of things at least some states will do some important things at least a little differently, and by the power of example and also competition for people and businesses given legally easy internal mobility, that will help slow the race to insanity that is a failure mode of monopoly. The power of example, however, is severely limited by whether people are even aware of such examples, which just raises the question of how the prestige media (which is national-scale) wants people to think about those examples, if they want people to know about them at all. See: covid policies. Were Florida and Texas on the one hand, or New York and California, on the other, success stories, or death-traps, or anti-science tyrannies?
A national-scale media which tends to always converge on the same most fashionable and prestigious ideas and which adopts the pretense of presenting objectively correct answers is fundamentally incompatible with maintenance of attitudes of tolerance for political variation that federalism requires. Also, you can never convince progressives that "the race to the bottom" isn't a huge problem and doesn't justify the nationalization of everything. The EU is theoretically supposed to be all about respect for the principle of subsidiarity, but to the extent it wasn't just lip service, it seems a pipe-dream in the face of the forces behind the typical trend of bureaucracies to arrogate everything to the center.
But the bigger problem with federalism is that, pretending SCOTUS dictates weren't a binding constraint for the sake of explication, when one is trying to perform some kind of analysis for the best layer of government or size of government unit to handle some particular issue or question, most issues are a mix of practical matters which could be assessed empirically and objectively, and ideological matters which are deemed off the table for everybody no matter what the studies might say or what the costs may be.
And the trouble is that no one can agree on which of those ideological matters should be on or off the table. Even "14th Amendment Libertarians" (i.e., pretty much all libertarians these days) are quite insistent that there should be no federalism at all in the domains related to all the rights they like.
In practice we all have to defer to SCOTUS to simply dictate when various policy options are on or off the table, but the problem is that any entity with the power to determine the extent and limit of federalism ends up not just imposing constraints on a manager's ability to get things done well, but in effect *is* the real manager, but without any of the expertise, accountability, responsibility, or incentives required to manage well.
Economists are familiar with the possibility of regulations becoming so burdensome and onerous that even the best manager of some domestic company simply can't achieve the any performance level that produces a profit, and so the business just gets regulated out of existence. The problem in USG is that it is regulating *itself* into compulsory underperformance, not just driving itself into state fiscal bankruptcy, but also into state capacity bankruptcy.
"As an extreme thought-experiment, imagine if the U.S. had no central government. That this would not go well seems clear. States might all differ in how they regulate..."
This sounds very much like the EWG (Europäische Wirtschaftsgemeinschaft, 1957-1993), the organizational pre-predecessor of the European Union. Europe had one common market those days and countries competing economically, politically as well as regards monetary policies.
Some might look at the centralized EU today and say, subsidiarity is the way, competition the solution.
There are alternatives to a centralized political solution that seem quite attractive and more suitable to a complex and dynamic world: What about a confederation with limited central power that advances free flow of goods, services, people, but refrains from standardizing everything as a result of a bureaucratic initiative. Otherwise you'll have incentives to standardize even the size and form of cucumbers.
Lord Acton reminds us: Power tends to corrupt, absolute power ... It's equally true for standardizing life in the Rockys, South Phili and Manhattan as people living in rural Rumania, industrialized Milano and the port of Hamburg might tell you not to speak of small entrepreneurs.
Thank you for taking the time to read and respond to my criticism. On the auditor first, I say, fair enough; most follow ups I could make here could be made against *any* proposed reform, which simply isn’t very helpful, so I won’t bother there. All I will do is reiterate your own point that this agency wouldn’t be very different in kind from ones we already have, so it’s unclear to me that the outcome would differ much from reconfiguring those. I also think you might find great interest in Congress Overwhelmed, a bipartisan collection of congressional scholars who are jointly concerned about, among other things, Congress’s diminished oversight capacity.
On the COO, I don’t want to make too much of the market analogy thing, or your statement that he’d have ideally been a top executive at a large corporation previously. Mostly I’m skeptical of such an enormous centralization of power, or that doing it under our clunky appointment system would result in greater accountability. I also again wonder about empirical precedent.
Empirical precedent was part of what drew me to the federalist approach to reform. I agree this can be taken too far. But many of our peers devolve a surprising amount of their governance to federal units. Looking at how Germany and Canada approach this was very eye-opening, especially when comparing to our fiscal federalism specifically. Agree however that ultimately a principle of subsidiary means thinking seriously about the proper level of each governance activity, and some really do only make sense at the very largest. Moving up medical licensure is definitely one such case (and of course, ideally drastically reforming it to be more open to entrants!)
As a foreigner, I find Arnold's posts and comments amazing. Everybody advances ideas to rebuild a system of government which many insiders have worked hard for decades to destroy. Arnold, the barbarians have taken over your government and you want to rearrange the chairs. Please tell me what you would advise to your senile President to get rid of Fauci&Co (if you are thinking about depriving them of their pensions, remember what happened to Andrew McCabe) or at least how you would ask him to go home now without empowering Queen Kamala or Nancy.
Re: "Gurri’s alternative proposal is to allocate more government responsibility to state governments. He claims that local governments are too parochial and not well run. The federal government is too clumsy and resistant to reform."
At stake are two issues: 1) What governance changes are feasible? 2) Which of the feasible changes which would most improve governance?
One may also ask: Which governance changes are more likely to occur?
Bryan Caplan highlights the emergence of "gubernatorial dictatorships" during the pandemic:
https://www.econlib.org/the-american-experiment-in-federalist-dictatorship/
My intuition is that the overarching trend is towards political centralization, notwithstanding Brexit from the EU and gubernatorial dictatorships during the pandemic in the USA. In America, there is pressure for States to supersede local governance in zoning and in public education; and for the national government to supersede States in regulation. Witness growth in the DC region.
Therefore, I believe that Arnold Kling's proposal for reform--a COO in national government plus a check by a Chief Auditor--is more likely to occur than decentralization of governance to the States.
What is the probability that Arnold's proposal for reform (COO + CA) will be enacted? I hesitate to say. Would the reform, if enacted, substantially improve governance? Or would it mainly make government bigger by accretion, instead of better? I hesitate to say.
One must multiple the probability of enactment by the probability of improvement. Perhaps the resultant probability is very small.
Consider an alternative: Decentralization to local governance, plus reforms to create exit options. An example is radical vouchers for education, apprenticeships, internships. Perhaps the likelihood of "decentralization + exit options" is much less likely to occur than either Arnold's reform or Adam Gurri's proposal (governance by States). But if decentralization + exit options does occur, it is much more likely to improve governance, thanks to competition, innovation, experimentation, and accountability. The expected improvement--probability of the reform x probability that the reform, if enacted, will improve governance--seems greater in my 'nirvana' alternative, than in Arnold's or Adam's 'realist' reform proposals.
Exit beats voice + bureaucracy.
I believe a fruitful avenue forward would be to look at comparative civil service structures. Although we all had a good laugh at the television series "Yes Minister", the UK civil service has always struck me as far more professional that its US counterpart. Unlike the US, the UK actually has someone who is in charge of the entire civil service, the Cabinet Secretary. In many ways he is the COO that Arnold talks about. I'm not sure that there is an equivalent of the Chief Auditor role in the UK. In any event, we have seen massive institutional failure across all of the US federal government and it needs to be fixed. Arnold is one of the few people actually taking that challenge seriously.
Has Kling addressed the COO/Auditor proposal from a focus on the incentives therein? I agree that the idea should be judged against the current state and not an ideal, but even doing this I struggle to see how the incentives faced by the COO or Auditor wouldn't be exactly the same as any other political appointee. Either immediately or over time, this means that the positions would devolve into the same issues we have with Treasury, the OCC, CFPB, etc., no?
I don't think scale or degree of federalism or centralization has much to do with it at all. When I look down the list of countries ranked by population, my personal index of impression of governance quality is extremely volatile and pops up and down. Population size would be way down my list of important correlates.
Lots of people criticize China which famously has the largest population of any nation, but in terms of decrease in poverty and increase in economic production, military might, and geopolitical clout, has anybody else had a better 30 years? Since 2008 their industrial production is up over 150% while in the US it's ... 0%. Yeah, yeah 'comparative advantage', but still. Japan is huge at 125 million and like every place has plenty of its own specific problems, but traveling anywhere in the country you get the impression right away that it's run pretty well, and this even correcting for the huge advantage that government has of being tasked with running a country full of Japanese people. Mexico is the same size, and, well, not so much. Switzerland is the same size as Sierra Leone, and likewise for Israel and Honduras.
More important is the problem of unmanageably wide scope, which makes it hard to imagine that a COO could actually manage something like USG any better than any of the old conglomerates were managed, many of which broke up. GE is a recent example. To the extent management talent isn't purely generic (i.e., that one, true consultant book to rule them all doesn't exist), needs specialized expertise in a particular sector, and benefits from the ability to focus effort and learning on that sector, at some point in broadening of scope any human COO would be outclassed by the sheer immensity of the job. You can only keep so many plates spinning at any one time. One of the classic arguments for small, narrow government is that a government which tries to do everything ends up unable to do anything well.
One sees this happen with districts judges who are supposed to be these ultimate legal generalists as they are expected to be able to competently handle the whole variety of cases, but being merely human often demonstrate shall we say a somewhat uneven competence in the full spectrum of suits.
Or for another example, imagine that aside from the typical kinds of administration that go on in universities, that someone tried to manage the production of intellectual and scholarly output across the entire spectrum of academic fields. Even though this would be a lot more specialized than what any COO would be asked to do, it's still so broad that it's hard to imagine even a genuinely polymathic genius being good at knowing how to do this well for all the schools and disciplines. And I think what we see at many universities is that they have subordinate institutions and programs of uneven quality and when one has a deserved reputation for excellence it is more due to self-perpetuating good management on the inside than supervision from above.
And while academia is theoretically about everything in life, actual life is even more complex, and the trouble is that if life is complicated enough that we cannot escape the need for bureaucracy, then likewise it's too complicated for that bureaucracy to be 'managed', and at some level of scope and complexity a leader has to trade off capacity for genuinely management with something more like trust in the expertise and competence of more specialized subordinates, from whom one has no real choice but to take their word for it, most of the time.
Another classical argument is that this kind of reliance gives rise to principle-agent problems like being in the position of not really knowing whether what your doctor or auto mechanic is recommending is actually strictly necessary. We kind of hand-wave away these situations in a commercial context by saying that people can sometimes try to rely on reputation systems, but there isn't a good system for this for political appointees who are expected to be more loyal to what the boss wants than to some objective measure of 'performance', and who it would be unfair to hold accountable for instances in which they were indeed just following orders.
As for scale of population size, I don't really see any convincing mechanism for why that should be. My basic model is that if things scale, they can scale practically indefinitely. Militaries are a clear example of always being able to expand by cloning and putting another layer on top of the pyramid with an extra star. There doesn't seem to be much basis for the argument that says there is some optimum or maximum practical distance in hierarchy steps from grunt to top general (or SECDEF or President). Facebook and Amazon are 'governed' in a sense, and why not have every person and organization in the world as a user?
I could make a just-so story for democracies like this: Someone can't be a good representative of more than a limited number of constituents, there just isn't enough time, and on the geographic district model, without heroic assumptions about self-sorting, the bigger the number the more diverse in preferences and the less coherent in terms of a genuinely common interest in most matters. This kind of thing is brought up when people talk about the trend in consolidation and growth in population size of American school districts, such that the dilution of possible connection between a parent and a school board member or superintendent, and the number of other people one must compete with to try to petition or get attention, becomes totally distancing.
And likewise, like any group, a parliament designed for interactions between equally-stationed individual peers with perhaps only one extra level of hierarchy for party and committee leaders cannot actually function well or as intended when it goes super-Dunbar. So, if the limit of X is, say, 60K, and Y is, say, 500, then you get an upper bound of 30 million.
But in my view, that story doesn't line up very well to what goes wrong in Congress or in USG more broadly. One way to see this is that most states are like mini-countries run on the same model, only California is bigger than 30M, but even many very small states are still run terribly in ways that seem to closely mirror the terribleness of USG, the Little Sams apparently not falling very far from the tree of Big Uncle Sam.
As for federalism and subsidiarity, Singapore doesn't have any of that and is basically a fully centralized city-state that not just 'runs well', but, fairly or not, is a common go-to example of world class governance.
One reason to favor more federalism is not necessarily that it would produce better governing schemes, but simply because in the nature of things at least some states will do some important things at least a little differently, and by the power of example and also competition for people and businesses given legally easy internal mobility, that will help slow the race to insanity that is a failure mode of monopoly. The power of example, however, is severely limited by whether people are even aware of such examples, which just raises the question of how the prestige media (which is national-scale) wants people to think about those examples, if they want people to know about them at all. See: covid policies. Were Florida and Texas on the one hand, or New York and California, on the other, success stories, or death-traps, or anti-science tyrannies?
A national-scale media which tends to always converge on the same most fashionable and prestigious ideas and which adopts the pretense of presenting objectively correct answers is fundamentally incompatible with maintenance of attitudes of tolerance for political variation that federalism requires. Also, you can never convince progressives that "the race to the bottom" isn't a huge problem and doesn't justify the nationalization of everything. The EU is theoretically supposed to be all about respect for the principle of subsidiarity, but to the extent it wasn't just lip service, it seems a pipe-dream in the face of the forces behind the typical trend of bureaucracies to arrogate everything to the center.
But the bigger problem with federalism is that, pretending SCOTUS dictates weren't a binding constraint for the sake of explication, when one is trying to perform some kind of analysis for the best layer of government or size of government unit to handle some particular issue or question, most issues are a mix of practical matters which could be assessed empirically and objectively, and ideological matters which are deemed off the table for everybody no matter what the studies might say or what the costs may be.
And the trouble is that no one can agree on which of those ideological matters should be on or off the table. Even "14th Amendment Libertarians" (i.e., pretty much all libertarians these days) are quite insistent that there should be no federalism at all in the domains related to all the rights they like.
In practice we all have to defer to SCOTUS to simply dictate when various policy options are on or off the table, but the problem is that any entity with the power to determine the extent and limit of federalism ends up not just imposing constraints on a manager's ability to get things done well, but in effect *is* the real manager, but without any of the expertise, accountability, responsibility, or incentives required to manage well.
Economists are familiar with the possibility of regulations becoming so burdensome and onerous that even the best manager of some domestic company simply can't achieve the any performance level that produces a profit, and so the business just gets regulated out of existence. The problem in USG is that it is regulating *itself* into compulsory underperformance, not just driving itself into state fiscal bankruptcy, but also into state capacity bankruptcy.
"As an extreme thought-experiment, imagine if the U.S. had no central government. That this would not go well seems clear. States might all differ in how they regulate..."
This sounds very much like the EWG (Europäische Wirtschaftsgemeinschaft, 1957-1993), the organizational pre-predecessor of the European Union. Europe had one common market those days and countries competing economically, politically as well as regards monetary policies.
Some might look at the centralized EU today and say, subsidiarity is the way, competition the solution.
There are alternatives to a centralized political solution that seem quite attractive and more suitable to a complex and dynamic world: What about a confederation with limited central power that advances free flow of goods, services, people, but refrains from standardizing everything as a result of a bureaucratic initiative. Otherwise you'll have incentives to standardize even the size and form of cucumbers.
Lord Acton reminds us: Power tends to corrupt, absolute power ... It's equally true for standardizing life in the Rockys, South Phili and Manhattan as people living in rural Rumania, industrialized Milano and the port of Hamburg might tell you not to speak of small entrepreneurs.