Freight Railroad Regulation
“Economic Foundations for 21st Century Freight Rail Rate Regulation,” by John W. Mayo and Robert D. Willig. November 2018. SSRN #3286211.
When Regulation first appeared in 1977, the term “regulation” mainly referred to price and entry controls enacted with the intention of mitigating expected consumer harm in markets whose firms have large economies of scale and scope. The poster child for this was railroad regulation. Accordingly, when this type of regulation was shown to often harm consumer interests, railroads became one of the first candidates for deregulation.
The Staggers Act of 1980 largely, but not completely, deregulated the pricing of freight rail services. On the 30th anniversary of the Staggers Act, Regulation published two articles on how the industry flourished following deregulation and what issues remained. (See “The Staggers Act, 30 Years Later,” and “Railroad Performance Under the Staggers Act,” Winter 2010–2011.)
We are now nearing the 40th anniversary of the Staggers Act. This paper reminds us of the fundamental economics of industries with economies of scale and scope and how that economics informs the residual rate regulation that remains for freight railroads.
The authors of this paper state the fundamental economic problem clearly:
It is impossible to allocate, in any nonarbitrary way, a share of fixed and common costs to any one of a railroad’s many activities. There is simply no way to subdivide those costs in a mechanical fashion that is unique and has any foundation in economic logic. The significance of this problem is much magnified by the fact that a substantial share of total railroad costs is fixed and common. In addition, if the regulator attempts to force rates to equal marginal costs, overall revenues will endemically fall short of overall costs. For rail systems that are characterized by scale and scope economies, rates must generally lie above the costs economically attributable to individual services if the firm’s revenues are to cover its total costs.
…Compensatory rates cannot be determined by the regulator on the basis of cost data alone since the financial viability of any price depends also on the quantity of rail services customers are willing to buy at that price. Rational determination of prices must be based on both cost and demand conditions.
The authors go on to demonstrate the folly of two “mechanical” methods of allocating fixed costs among consumers: equal division (dividing total fixed costs by the number of shipments) or constant percentage markup above marginal costs. Both methods would drive away users that value the service at more than the marginal costs they impose but less than their share of fixed costs under the “mechanical” sharing rules and increase costs on the remaining customers. No one benefits from this type of pricing.
The Staggers Act retains federal rate-setting authority for so-called captive customers (those who have no shipping choices other than one railroad) who pay more than 80% above variable cost. The authors argue that the appropriate regulatory fallback for the captive shipper is stand-alone cost pricing, i.e., the long-run costs of the shared common portion of the railroad plus the long-run costs of the shipper-specific line.
Health Policy
“The Effect of Health Insurance on Mortality: Power Analysis and What We Can Learn from the Affordable Care Act Coverage Expansions,” by Bernard Black, Alex Hollingsworth, Leticia Nunes, and Kosali Simon. February 2019. NBER #25568.
How should we evaluate health insurance policy initiatives? One obvious metric is whether increases in the population covered by health insurance reduce mortality. Studies of important policy initiatives like the introduction of Medicare and the Oregon Medicaid expansion as well as the RAND Health Insurance experiment found no statistically significant effects on mortality. Those who oppose government-funded health insurance expansions cite such evidence in their criticism.
This paper evaluates the expansion of health insurance under the Affordable Care Act (ACA) and comes to similar no-effect conclusions. But the authors argue that the problem is not the real effect of the policy but the inability of any plausible research design to distinguish any mortality effect from no effect. To be able to detect an effect on mortality, the affected population must be reasonably large relative to the already insured population and the reduction in mortality also must be reasonably large. Neither is likely to be true.
Suppose first that out of 100,000 individuals aged 55–64 (those most likely to die and thus benefit from ACA coverage), half became newly insured because of ACA subsidies. The annual mortality rate in this group is around 600 per 100,000. If insurance were to reduce the probability of death by 25% among the newly insured, then insuring 50,000 individuals among 100,000 individuals would reduce the expected number of annual deaths by 75 (0.5 × 0.25 × 600) to 525. If mortality events are independent, the expected standard deviation of mortality per 100,000 persons would be around 24 and the expected t-statistic would be 3.07. Such a decrease in mortality would be easily discernible in the data with 95% confidence, which requires a t-statistic of only 1.96.
But the average increase in health insurance coverage attributable to Medicaid expansion over 2014–2016 was only around 1.1% for persons aged 50–64, and only around 4% for low-educated populations. Further, a 25% mortality reduction, as in the example above, would be extraordinary. The introduction of sulfa antibiotic drugs in the 1930s, by comparison, reduced maternal mortality by 24–36%.
If the increase in the insured population is only 5% (much greater than the actual 1.1%), and the mortality reduction for the newly insured is 10% instead of 25%, the expected population average treatment effect would be a reduction in the mortality rate of 3 (0.05 × 0.1 × 600) to 597. The standard deviation in the number of expected deaths would remain the same, so the expected t-statistic would be 0.11. According to the authors, to increase the t-statistic by a factor of 20 to 2.2, one would need a sample 400 times as large: 40 million people.
Such a study is not feasible, so the use of decreases in mortality to evaluate the effects of policies to increase the number of those with health insurance is also not feasible. Thus, the rationale for increasing the use of health insurance through public subsidies cannot be because such a policy saves lives. We never could tell whether the policy saved lives.
Bank Regulation
“The Impact of the Durbin Amendment on Banks, Merchants, and Consumers,” by Vladimir Mukharlyamov and Natasha Sarin. February 2019. SSRN #3328579.
“How Do Capital Requirements Affect Loan Rates? Evidence from High Volatility Commercial Real Estate,” by David Glancy and Robert Kurtzman. November 2018. SSRN #3289453.
The Durbin Amendment, a late addition to the Senate version of the 2010 Dodd–Frank financial reform legislation, passed without hearings or debate in May 2010 and became part of the final legislation. It directs the Federal Reserve to promulgate a rule to ensure that fees for debit transactions be “reasonable and proportional” to the actual cost incurred by the bank. The final Fed rule on this provision capped debit card fees (charged to merchants by banks with assets above $10 billion) at 22¢ plus 0.05% of the transaction amount. For banks above the asset threshold, debit card fees on an average transaction of $38 fell from 43¢ to 24¢ (exactly the maximum Durbin allows: 22¢ + 0.05% × $38). Fees for banks below the $10 billion threshold went unchanged—still 43¢. The result is a decrease of $6.5 billion in revenue annually for the affected banks (25% of their total debit card fee revenue).
In the first of these two working papers, the authors conclude that large banks affected by the debit-fee rule totally offset their $6.5 billion loss by charging higher checking account fees. Monthly maintenance fees on checking accounts doubled, decreasing the share of consumers with free checking accounts from 60% to 20%.
Did consumers at least benefit from merchants passing on the debit card fee reductions? The authors examined gasoline sales data from 65,000 stations in 10 states for the six months prior to and after the implementation of the debit-fee limit. Gas stations saved an average 0.76¢ per gallon. The paper compares retail gasoline profit margins in those ZIP codes with a large reduction (top decile) in debit card fees (34%) with identical ZIP codes (in terms of income, population density, and gas station density) with a low reduction (bottom decile) in fees (3.6%). The comparison found no difference in retail margins. The same comparison was performed on top-quarter and bottom-quarter and above- and below-median debit-fee reduction ZIP codes. Again, no difference in retail margins was found. So, in general, gasoline stations did not pass on their cost savings to consumers.
But gas stations did have lower margins in certain specific circumstances: in ZIP codes in which debit card usage relative to combined debit and credit card use was above the median. And in those high-debit-card-use ZIP codes, the pass-through to consumers was larger in ZIP codes with more competition (above the median number of gas stations per capita).
So, in general, merchants gained from the rule. Banks offset their losses with increased checking account fee revenue. Consumers, particularly low-income consumers, lost out through increased checking account fees and a decrease in free checking accounts after balance requirements increased dramatically, which led some lower-income consumers to become unbanked.
The second of these two working papers concerns shareholder equity in banks. The failure of financial firms during the Great Recession of 2008–2009 led many academics to recommend more equity and less debt in the capital structure of banks. In times of reduced loan repayment, equity owners of banks would lose their money, but as long as equity exceeds loan losses, the bank survives.
Previously in these pages, I described the work of Anat Admati (“Working Papers: Bank Capital Requirements,” Winter 2010–2011), who argues that regulatory mandates for more shareholder equity would allow banks to survive bad times. According to Admati, “Increasing equity requirements would reduce the cost to society of having a fragile and inefficient financial system where banks and other financial institutions borrow excessively, and thus it would be highly beneficial.”
The authors of this paper studied the increase in commercial real estate construction loan capital requirements from 8% to 12% in 2015. They found an increase in loan rates of 38 basis points (0.38%) relative to a baseline median interest rate of 3%, a 13% increase. The increase is not observed in construction loans not affected by the regulatory change. This increase is not the result of a change in the risk composition of borrowers, but rather an overall decline in construction loan risk-taking. This, in turn, keeps some construction projects from happening.
Increased capital requirements increase bank stability, but this benefit is not free.
