4 ms·

Just include externalities in the oil taxes. It doesn't matter if it's 10 plane trips a year or driving 100 miles to work each day - both are bad and both shoul

by ajuc 6y ago

Just include externalities in the oil taxes. It doesn't matter if it's 10 plane trips a year or driving 100 miles to work each day - both are bad and both should be taxed into unaffordability.

hansvm 6y ago
One argument is that externalities aren't usually linearizable or localizable, so any attempt to tax them will result in loopholes that are patched with...you guessed it, more laws with their own loopholes growing into an insurmountable lovecraftian horror. Elaborating on the nonlinearizability, on a global scale it doesn't matter if a single person (or an entire city) drives 100 miles to work each day. In isolation their decision to drive has an externality of approximately $0 (wrt carbon). It's only a problem because of the global scale involved -- as an extremely simplistic (and wrong) model for the sake of fitting into a HN comment, there's a fuzzy cutoff point in total carbon production rates below which externalities are small and above which they're devastating and grow rapidly (and superlinearly). Adopting the obvious scheme of dividing total costs by total production, adding a healthy margin to further disincentivize carbon, and assigning that value as an externality tax (which gets updated periodically to never be too inaccurate) won't actually suffice to keep carbon production below the critical threshold because the marginal taxation of externalities for a unit of carbon is substantially less than the marginal externality itself. Underlying the above example is something that should be obvious to programmers -- not all problems can be efficiently solved by assigning a local value function to autonomous agents. In the case of people we value autonomy as a right in and of itself independently of its ability to benefit the whole, so many potential global optimizations are left on the table, but at the point where we need to enact taxes or legislation to prevent a global problem (e.g. the carbon issue) the illusion of autonomy no longer exists, and we might as well go straight for a solution that has a chance of working (e.g. carbon credits could work if implemented correctly) rather than handicapping ourselves with attempts to adjust our local reward function like trying to tax the externalities.
- ajuc 6y ago
  If your tax is linear whole classes of loopholes stop working (it doesn't matter how you split the usage - the total is still the same). > on a global scale it doesn't matter if a single person (or an entire city) drives 100 miles to work each day. In isolation their decision to drive has an externality of approximately $0 (wrt carbon) That's such a disingenuous argument. Their emissions add up. That's like saying at individual level gravity doesn't work because you don't feel a gravity from 1 atom. If you add oil tax the global consumption falls, it's only a matter of choosing a number big enough to make the total consumption fall below the threshold. I don't understand why you're trying to make it so complicated - we don't only have 1 try, we can introduce tax, look at effects, make corrections. Opening a valve in your tap isn't linear. But you can pretty easily choose the proper level after a few corrections.
  - hansvm 6y ago
    > That's such a disingenuous argument. Their emissions add up. That's like saying at individual level gravity doesn't work because you don't feel a gravity from 1 atom. I probably could have phrased it differently to make it more clear, but my argument is _not_ that individual contributions don't add up. Rather, they do add up, and after a certain "critical" threshold they add up to produce superlinear negative externalities and that attempting to price all carbon equally to add up to the sum of those externalities is potentially gameable and still allows for sufficiently profitable activities to effectively ignore them and push global climate change beyond acceptable levels. > If you add oil tax the global consumption falls, it's only a matter of choosing a number big enough to make the total consumption fall below the threshold. That's roughly true enough. It's subject to a form of "externality arbitrage" though. Let's say we're outputting enough carbon that externalities grow quickly as a function of carbon output rate. Legislators set a price on carbon based on how bad it currently is. Many activities are still profitable with respect to that tax and are still performed while others are axed. Assume that an agent scales up some profitable carbon-emitting activity, and to make the math simple they personally double global carbon output. By virtue of the fact that we're in a superlinear externality regime they now have to pay $N in taxes while causing damage much greater than $N. > we don't only have 1 try, we can introduce tax, look at effects, make corrections. Yes, eventually regulators step in and pick a bigger number, and presumably there are limits to any such profitable carbon-producing activity, so you're right that eventually we could level off total carbon output, but the large slope on that graph of externalities with respect to carbon production rate and the timescales involved in updating the tax code leave plenty of room for exploitation -- it's not an especially stable system that has a natural solution at our desired global output. > I don't understand why you're trying to make it so complicated It's...not thaat complicated? If your goal is to limit carbon production to a rate that leads to acceptable levels of climate change (which doesn't have to be your goal, but it's a central assumption in everything I've written, along with the notion that we can address carbon independently from other problems) then the most straightforward way to do that is to set a cap at the level you don't want to exceed and divvy out the credits as you see fit. You could tax those too if you'd like, or sell them, or whatever you want to do, but the critical piece is that to avoid exceeding your desired limit you just set the limit and ban exceeding it. Allowing a theoretically unbounded amount of carbon to be released and trying to manage a quickly-changing economic reality by imposing a flat tax to disincentivize the undesirable behavior is moderately more complicated and has likely, predictable failure modes. Flat externality taxes are not necessarily a bad idea in the abstract, but for the concrete case of carbon and global warming they don't seem like a good fit, at least not by themselves.

by ajuc 6y ago

Just include externalities in the oil taxes. It doesn't matter if it's 10 plane trips a year or driving 100 miles to work each day - both are bad and both should be taxed into unaffordability.

hansvm 6y ago

One argument is that externalities aren't usually linearizable or localizable, so any attempt to tax them will result in loopholes that are patched with...you guessed it, more laws with their own loopholes growing into an insurmountable lovecraftian horror. Elaborating on the nonlinearizability, on a global scale it doesn't matter if a single person (or an entire city) drives 100 miles to work each day. In isolation their decision to drive has an externality of approximately $0 (wrt carbon). It's only a problem because of the global scale involved -- as an extremely simplistic (and wrong) model for the sake of fitting into a HN comment, there's a fuzzy cutoff point in total carbon production rates below which externalities are small and above which they're devastating and grow rapidly (and superlinearly). Adopting the obvious scheme of dividing total costs by total production, adding a healthy margin to further disincentivize carbon, and assigning that value as an externality tax (which gets updated periodically to never be too inaccurate) won't actually suffice to keep carbon production below the critical threshold because the marginal taxation of externalities for a unit of carbon is substantially less than the marginal externality itself. Underlying the above example is something that should be obvious to programmers -- not all problems can be efficiently solved by assigning a local value function to autonomous agents. In the case of people we value autonomy as a right in and of itself independently of its ability to benefit the whole, so many potential global optimizations are left on the table, but at the point where we need to enact taxes or legislation to prevent a global problem (e.g. the carbon issue) the illusion of autonomy no longer exists, and we might as well go straight for a solution that has a chance of working (e.g. carbon credits could work if implemented correctly) rather than handicapping ourselves with attempts to adjust our local reward function like trying to tax the externalities.

ajuc 6y ago
If your tax is linear whole classes of loopholes stop working (it doesn't matter how you split the usage - the total is still the same). > on a global scale it doesn't matter if a single person (or an entire city) drives 100 miles to work each day. In isolation their decision to drive has an externality of approximately $0 (wrt carbon) That's such a disingenuous argument. Their emissions add up. That's like saying at individual level gravity doesn't work because you don't feel a gravity from 1 atom. If you add oil tax the global consumption falls, it's only a matter of choosing a number big enough to make the total consumption fall below the threshold. I don't understand why you're trying to make it so complicated - we don't only have 1 try, we can introduce tax, look at effects, make corrections. Opening a valve in your tap isn't linear. But you can pretty easily choose the proper level after a few corrections.
- hansvm 6y ago
  > That's such a disingenuous argument. Their emissions add up. That's like saying at individual level gravity doesn't work because you don't feel a gravity from 1 atom. I probably could have phrased it differently to make it more clear, but my argument is _not_ that individual contributions don't add up. Rather, they do add up, and after a certain "critical" threshold they add up to produce superlinear negative externalities and that attempting to price all carbon equally to add up to the sum of those externalities is potentially gameable and still allows for sufficiently profitable activities to effectively ignore them and push global climate change beyond acceptable levels. > If you add oil tax the global consumption falls, it's only a matter of choosing a number big enough to make the total consumption fall below the threshold. That's roughly true enough. It's subject to a form of "externality arbitrage" though. Let's say we're outputting enough carbon that externalities grow quickly as a function of carbon output rate. Legislators set a price on carbon based on how bad it currently is. Many activities are still profitable with respect to that tax and are still performed while others are axed. Assume that an agent scales up some profitable carbon-emitting activity, and to make the math simple they personally double global carbon output. By virtue of the fact that we're in a superlinear externality regime they now have to pay $N in taxes while causing damage much greater than $N. > we don't only have 1 try, we can introduce tax, look at effects, make corrections. Yes, eventually regulators step in and pick a bigger number, and presumably there are limits to any such profitable carbon-producing activity, so you're right that eventually we could level off total carbon output, but the large slope on that graph of externalities with respect to carbon production rate and the timescales involved in updating the tax code leave plenty of room for exploitation -- it's not an especially stable system that has a natural solution at our desired global output. > I don't understand why you're trying to make it so complicated It's...not thaat complicated? If your goal is to limit carbon production to a rate that leads to acceptable levels of climate change (which doesn't have to be your goal, but it's a central assumption in everything I've written, along with the notion that we can address carbon independently from other problems) then the most straightforward way to do that is to set a cap at the level you don't want to exceed and divvy out the credits as you see fit. You could tax those too if you'd like, or sell them, or whatever you want to do, but the critical piece is that to avoid exceeding your desired limit you just set the limit and ban exceeding it. Allowing a theoretically unbounded amount of carbon to be released and trying to manage a quickly-changing economic reality by imposing a flat tax to disincentivize the undesirable behavior is moderately more complicated and has likely, predictable failure modes. Flat externality taxes are not necessarily a bad idea in the abstract, but for the concrete case of carbon and global warming they don't seem like a good fit, at least not by themselves.