Blog

Monday, January 30, 2023

Thinking 4th Dimensionally About Costs

Part of the amazingly complex and voluminous analysis used to propose nearly quadrupling the social cost of carbon is based on converting future costs to a present-day dollar figure.

Simply put, the way the social cost of carbon is determined is that future costs of global warming are estimated (additional deaths, loss of agriculture, loss of property from climate damage, etc.), but to apply those future costs to determine today's policy they must be "discounted". Discounting is used often in the private sector to determine the value of future investments and decide whether spending money today is worth the returns in the future.

The social cost of carbon works similarly but in the opposite direction: should we spend money today to avoid the costs in the future? The reason this must be done is that we have a choice of spending the money today on reducing carbon emissions or also on building new factories or power plants, so the purpose of discounting is to decide which avenue will produce the most benefit/lowest cost.

In his review of the proposed policy, Maxim Lott goes into considerable detail about the discount rates used. I will not reiterate his arguments, but his main points are that the EPA is using a discount rate that is lower than was used before and relies on the most recent experience of low interest rates, which may not be representative of the future 200 years, and the discount rate they used is largely responible for the significant rise in the cost of carbon. These are essential points, but there are also other reasons the 2% rate that EPA relies on should not be used.

World Rate

Of the many differences between how Republicans determine a social cost of carbon and how the Democrats do, is the latter estimate costs incurred by the planet as a whole, while Republicans estimate only the costs incurred by Americans. Resources for the Future, a group deeply involved in this process and policy, estimated that the global costs are more than 80% of the total cost of carbon.

While whether the US should consider the global costs of its policies when considering regulations of the US economy is up for debate, it should not simultaneously be calculating the costs on a world-wide basis and discounting based on only on U.S. factors. The idea of discounting is asking yourself the question 'how much money would I need in the future to make me prefer investing it today instead of spending it?'

If you buy a 30-year Treasury Bond today, the interest rate is 3.63%, meaning you'd turn every $1 you invested into $2.91 in 30 years, but you couldn't spend it until then. So, in some sense, you'd be saying that you'd prefer the $2.91 in 30 years to the $1 today. That's the concept behind the discount rate. Every person has their own preference for long-term spending, and even that preference changes over time.

For the calculation, the EPA uses 2.0%, based on the recent experience in the U.S. For U.S.-exclusive applications, it is not a problem to attempt to take the average preferences of all Americans, but because this rule is discounting the world-wide costs of climate-change, that is no longer appropriate. This causes the question to be 'how much money would the world need to have to make me prefer investing it today instead of spending it?'.

This question is a non-sequiter. Perhaps there's an answer to the question, but the answer surely isn't the average interest rate on bonds in the United States.

Using the U.S.-based discount rate would also cause other countries that want to calculate their own social cost of carbon to settle on different values even if they estimated the exact same costs to climate change. Europe, for example, has even lower interest rates. This approach would mean that it would be acceptable for the United States to produce more CO2 than Europe, and perhaps even some developing countries. Japan's interest rate would lead to substantially lower ability to emit carbon than the United States.

This approach would mean that it would be acceptable for the United States to produce more CO2 than Europe

The only reasonable solution is to determine a discount rate based on global economic conditions, not just the United States

US Experience

One of Lott's arguments about the discount rate is that the recent experience with interest rates is not representative of its longer history, beyond the past 10 years. In addition to that, before 2021, US interest rates were at historic lows, in fact. From 2000 to 2020, the United States was regarded by many to be the safest place for other countries and investors to put their money, particularly during the financial crisis, the Euro crisis, and then the Covid pandemic. "[In 2008], Treasury bonds were in such hot demand that American interest rates plummeted." To base the next 200 years of policy off of interest rates during this time seems highly selective and ill-advised.

Given the purpose of the discount rate, it's not even clear that using 10-year treasury returns is the appropriate metric. Republican administrations tend to use a higher discount rate, such as 7%, based on stock returns. This, on the contrary may be too high. Essentially, this number should represent the tradeoff between investing and consuming. Most people who are saving their money for 30 years, aren't doing so expecting 2.0% interest. It would seem a more inclusive metric should be used than government bonds, which to a large extent, represent the security of the U.S. government that Americans and non-Americans expect.

Literature

The EPA's proposed rule cites three studies to support its choice of discount rate. The 2018 paper surveyed economists who have published a paper on discounting. The survey produced an average of 2.3% and a median of 2.0%. The 2020 paper also surveyed experts who had published, but the publication topic was climate change and not discounting. It also produced an average of 2.3% and a median of 2.0%.

Even climate scientists believe the discount rate should be much greater than that used by the EPA.

Lastly, a paper authored by Pindyck surveyed economists, but found a higher mean of 2.7%. This paper, also broke down the estimates by region and profession. Notably, climate scientists chose 3.13% on average, and developing country scientists chose 4.14% on average. European respondents produced the lowest average followed by Americans and then developing country scientists. This supports the notion that the discount rates used are highly location-specific, and Americans' recent experience in a low-rate environment might be biasing their considerations given the purpose of the discount rates. Even so, the average respondent in all three surveys opted for higher rates than what the EPA chose.

The average respondent in all three surveys opted for higher rates than what the EPA chose.

Thursday, December 15, 2022

Policy for the Experts or for the People?

Woe to anyone who tries to dig into this new EPA Social Cost of Carbon regulation (see previous post for background). While it's an effort to bring scientific rigor to an important policy, it is the very definition of byzantine. It is the culmination of hundreds of academic studies by thousands of researchers, marshaled by other researchers, bureaucrats, and activists to generate a single number that carries a titanic amount of underlying meaning.

The SCC calculation comprises many steps but, in summary, the objective is to use current projections of the harm that researchers expect will be done to the globe economically and divide that by the total amount of CO2, researchers expect will be emitted. This produces a cost per ton of CO2. Then economics is used to convert that dollar cost, which is incurred in the future, into today's dollars.

The National Academies created a characteristically difficult-to-parse diagram to outline the steps.

A more descriptive chart would look something like this.

Each of these steps, itself has numerous substeps, and is based on a large set of articles and research. It's not an overstatement to say that this process is inscrutable for a layman. It begs the question whether regulatory policy should be so complex that only highly, highly educated people can have any hope to understand it and critique it.

"It begs the question whether regulatory policy should be so complex that only highly, highly educated people can have any hope to understand it and critique it."

There's obviously a conflict here between having highly educated experts in their field produce the best possible answer and having policy-makers creating policy that is open to the public to be challenged and accepted democratically. It's unclear which is better, but we should acknowledge the drawbacks of each. One other drawback that's not immediately obvious to those who favor the expert approach is that experts are often ideologically motivated, their analysis isn't completely objective, nor is it subjected to rigorous skepticism and criticism from other experts who are subject to the same biases.

"Experts are often ideologically motivated, their analysis isn't completely objective, nor is it subjected to rigorous skepticism and criticism from other experts who possess the same biases."

Also, the bureaucrats designing the policy are themselves ideologically motivated, which causes them to favor research that matches their priors and pre-determined preferences. This leads to a multiplier effect that amplifies and enshrines incorrect conclusions.

Sunday, December 11, 2022

Billions of Taxpayer Funds Up in Smoke

Main Takeaways

• IRA spends $369 billion to reduce CO2 emissions.
• The EPA estimates 1 ton of CO2 does $190 worth of damage to world.
• IRA spends $369B to reduce world damage by $125B resulting in $244 billion or 2/3 of the cost being wasted.

The Biden EPA recently released a draft rule which will more than double what is referred to as the social cost of carbon. Essentially, this is an estimate of how much damage carbon dioxide does to the world. More specifically, it's the cumulative damage, in dollar terms that a ton of CO2, if emitted today will wreak. Its purpose is to inform environmental regulations. For example, if the EPA is considering requiring coal plants to install CO2 reducing methods, they are required to conduct a cost/benefit analysis and can only pass regulations that pass it. So, if a regulation is going to cost energy producers $5 billion, and it reduces CO2 by only 1 ton, then the regulation fails the cost/benefit test and can't be implemented. Conversely, if it only costs $1, but reduces CO2 by 100 million tons, then it passes and can be implemented.

There are a number of questionable assumptions made to generate the new estimate, but the process is extremely complicated and impenetrable so for now, I leave it to others to scrutinize it. But it is instructive to assume it's correct and apply it to recent environmental regulations to determine if the Democrats' energy policy as implemented through legislation would pass Biden's historically liberal cost/benefit analysis.

I have found energy policy to be extremely opaque. For instance, the best (only?) analysis of the IRA's effect on emissions that I can find is from the Rhodium Group, but most of the reporting is in percentage terms, and aggregates. The analysis I provide is surely not exact, and I push others who are more experienced to do better, but it does get the work started and hopefully provides some useful takeaways.

Just as a starting point, the IRA energy provisions' total cost is $369 billion from 2023 to 2031. The Rhodium analysis predicts that the IRA will reduce US emissions by between 439 and 660 million metric tons by 2030.1 This leads to an estimated cost of between $559 and $840/ton of CO2--i.e. Americans will pay between $559 and $840 to reduce CO2 emissions by one ton. Biden's EPA, on the other hand, says that it is worthwhile to spend up to, but not to exceed $190 per ton. Consequently, the IRA seems to be a huge waste of money, by Biden's EPA's own standards.

Several caveats should be noted here, and I again point them out in the interest that someone will take this and run with it to provide a better analysis. First, the IRA may have longer-run impacts than are estimated by Rhodium through 2035. If the IRA, for example, converts all energy production to solar, CO2 emissions will be reduced beyond 2035. Additionally, the $369 billion may not all be used to reduce carbon emissions. This will be addressed below, to some extent. Contrarily, in many ways, the IRA only accelerates processes that will happen no matter what, so these CO2 reductions are finite. In other words, if the economy were destined to transition to 100% solar power by 2075 without the IRA, but with the IRA, that year moves up to 2060, then the IRA would have no effect after 2075. Also, the time window applies to the money as well. Some of the programs might require additional funds after 2031 to maintain lower emissions which aren't accounted for.

To get around some of these complications, we could try to perform the same exercise on specific items in the bill. The Congressional Research Service provided the best breakdown of provisions I could find. According to it, the energy production provisions total $149B, transportation $49B, and industry $44B.2. By sector and scenario, not one of the areas will be cost effective.

The single estimate that comes closest to being cost effective are the energy provisions.

There is admittedly, much more work that can and should be done to determine whether the IRA is actually cost effective at reducing carbon reductions, but these estimates that are available publicly suggest not, and that's even using the proposed EPA regulations which are higher than any regulator in the world, even the Obama administration, thinks are appropriate.

Notes

1 Note that the highest reduction comes from the central emission scenario. It is unusual that the highest reduction isn't in the highest emission scenario and that the central emission scenario doesn't produce an estimate between low and high, but Rhodium does not provide their model or an explanation, only the results.
2 I excluded loans from the total and a list of my divisions and inclusions is available upon request.