Editor’s Note: Conducting Benefit-Cost Analyses of federal programs traces its vintage to the passage of the Flood Control Act of 1936. It is possible that the current debate might be one of the few times that a prolonged attention is being given to addressing income distribution in the exercise of benefit-cost analyses by either the President or the Congress since such actions were championed by leading members of Congress during the Johnson Administration. The Administration has indicated that it is going to address income distribution in the rulemaking process. Stakeholders can either wait and comment on the Administration’s position or in the alternative take a proactive position by making a recommendation; the purpose of this note is to lay the foundation for the latter option by soliciting the views of experts on the alternatives presented herein.
Washington abounds with concerns that as a result of the Presidential Executive Order on the modernization of centralized regulatory review that the Biden Administration will take steps to reduce the rigor associated with conducting benefit-cost analyses of regulations as presently practiced pursuant to existing guidelines. In the absence of the immediate availability of a substitute measure and prior to the end of a forthcoming public comment period, as stated elsewhere, then under these conditions the Biden Administration is far too sophisticated to take on the entire community of professional economists.
We arrive at this position because the Administration would not want to incite the not so subtle regulatory think tanks and watchdogs as well as the myriad of regulated entities, some of whom have been supportive of many of the Administration initiatives. Of course there is also the need to keep Congress informed, as was the case when the Carter Administration was in the process of arming the Office of Regulatory and Information Policy, the precursor to OIRA, some forty years ago. Therefore it is likely that the Administration will continue to support the existing benefit-cost analyses of regulations, at least for the immediate future. In any event none of the proceeding statements should suggest that the Administration may not take actions independent of modifying the existing BCA regime.
For the discussion set forth herein the methodology currently used to generate benefit-cost analyses is only one of a number of such methodologies which some consider, probably inaccurately, to be under the broad umbrella of Social Welfare Functions. A welfare economist would probably be more precise and state the results of existing benefit-cost analysis, under a completely symmetric distribution of effects, is a proxy for a utilitarian social welfare function.
Nonetheless the Administration and affected stakeholders should begin immediately to develop other techniques for disclosing the complete range of impacts of a regulation as well as alternative mechanisms for rank ordering its merits relative to other alternatives. The magnitude and complexity of the issue suggests the need to organize work groups immediately. The available methodologies include, but are not limited to, those buried in the tomb of welfare economics in a crypt with the title of Social Welfare Functions (SWFs). A social welfare function might be viewed as a rule for aggregating individual preferences into a single metric that can be used to assess the merits of alternative statements of policy or more simply a policy assessment tool.
In a nutshell emerging SWFs might well address the documented shortcomings in benefit-cost analyses, namely no consideration of either income redistribution or the diminishing marginal utility of money. The emergence of a social welfare function, different from the methodology employed in conventional benefit-cost analysis, is addressed in this publication: OIRA: Past, Present and Future (pg. 17) published and reviewed by experts last year year in the Journal of Benefit Cost Analysis; in so doing the author benefitted from the views of Professor Matthew Adler of Duke University. Professor Adler has devoted a substantial part of his career to the analysis of Social Welfare Functions and it is highly recommended that those interested in the subject consult any one of a number of his publications on the subject.
The thrust of Option # 3 in the CRE White Paper of November 2020 is to preserve benefit-cost analysis as presently practiced and to accommodate the introduction of emerging SWFs as mechanisms to consider non-national efficiency gains, such as the distribution of income, in the assessment of the feasibility of a regulation but to do so within the confines of a multi-agency, multi-year regulatory program.
A First Step?
Option #3 above is not a proxy for a SWF but it allows, subject to refinement, income distribution concerns to be addressed immediately in the rulemaking process while the use of SWF’s are reviewed and refined.
On November 6, 2020 CRE distributed to affected and potentially affected parties a paper which outlined three options which could govern the daily operation of OIRA. Option #2, involving a pro-active role for OIRA, has received considerable attention for immediate implementation.
However for implementation over a longer period of time, and one that could address the incorporation of income distribution into to the existing modus operandi of OIRA, consideration should also be given to Option # 3. Option # 3 provides a window into the identification of the opportunity cost of foregoing gains in national efficiency as a result of adopting a regulation because of its distributive effects. The aforementioned information could be the basis for developing an interim algorithm for addressing the redistribution of income into the federal rulemaking process.
In summary, two days after the election CRE published a paper titled: “Questions to Nominees for the Administrator of OIRA” which was distributed widely to affected parties.
Option #2 received considerable attention in that it deals with OIRA assuming a proactive role in addressing existential threats.
Option #3 at least until now, received considerably less attention. The thrust of this option is the need for OIRA to develop a multi-year, multi-agency regulatory program which is reviewed by the public, the Congress and stakeholders.
Option #3 also provides a mechanism to address income inequality in rulemaking while Social Welfare Functions are being sharpened for widespread adoption throughout the federal government. Basically one could start with all available NPRM’s and rank order them in their descending level of net benefits. When a regulation does not pass the national efficiency test of positive net benefits it could be subjected to a qualitative comparison of its merits relative to the marginal regulation in the aforementioned list.
The analytical framework described above would become more rigorous if the multi-year, multi-agency regulatory program set forth in Option # 3 were subject to a financial ceiling. With further analysis additional decision rules could be formulated. Notwithstanding its shortcomings it might have advantages over waiting for the issuance of more explicit guidance on Social Welfare Functions.
Is the alternative set forth above perfect, obviously not(?). But it might well be preferred to waiting until the current advances in the development of SWF’s can be refined and issued in the form of Circular A-4 guidance to federal agencies. Besides advances are often made in the management of the administrative state when analysts begin to handle real world data.
The point of emphasis herein is that the work of welfare economists have progressed to the point where they have developed conceptual answers to macro policy issues but have a way to go in addressing regulatory specific issues. Professor Arrow’s Impossibility Theorem is such an example when he demonstrated that it is impossible to combine the preferences of individuals through the use of a social welfare function to produce a single metric of social well-being if there is in existence three or more established conditions. The point emphasized herein is that the aforementioned conceptual answers most certainly establish boundaries for allowable strategies but they do not lead directly to developing mechanisms for the simultaneous consideration of national efficiency gains and distributive impacts when deciding whether a proposed rule which is deficit in the national income account should be promulgated as a final rule.
In summary, if income inequality considerations are going to be introduced into federal rulemaking, at least in the immediate future, then such an action need not be based upon the adoption of a singular criterion similar to the one presently in use– that a proposed rule should have positive net benefits, however defined. Instead the ultimate decision could be based upon the adoption of a decision process which in this case is based upon the opportunity cost of a proposed rule, expressed in terms of national efficiency benefits forgone. The aforementioned opportunity cost can be determined by an identification of the rule(s) it displaces in a well defined multi-agency, multi-year regulatory program. This is not a new idea in that it was used some fifty years ago by the Corps of Engineers in the management of its construction program, see page 16 of this publication. What an ironical development if the price of incorporating income distribution into rulemaking is the implementation of a regulatory budget!
What’s Next on Social Welfare Functions?
As is explained in detail in the Appendix hereto, based upon our initial analysis to date, Social Welfare Functions have not progressed to the state where they can be implemented on a government-wide basis at this time. By implementation on a government-wide basis we mean with an attendant set of instructions similar to A-4 for conventional analyses.
That said, as also discussed in the Appendix, academic research is being accomplished at an accelerated pace and it is possible to introduce the use of SWF’s in select areas.
Those that are up to date on the most recent advances in the development of Social Welfare Functions are encouraged to present their views on how OIRA can utilize their work to address income distribution. That said, it might have to be a multistage process because ultimately the resultant procedures have to be capable of government-wide implementation across a large number of agencies and their attendant sub-parts.
If we are going to replace or augment, for all practicable purposes, the current decision rule that results in only those regulations whose national benefits exceed their costs are promulgated as final rules then it is imperative that we develop a reproducible decision criterion else there will be a race to the bottom.
The Biden Administration is going to address the issue of income redistribution with or without the assistance of the economic community. Stated somewhat more bluntly, it is time to exit Arrovian social choice theory and get our hands dirty with data.
To this end, consideration should be given to identifying a regulation which is in its formative stages, for example an ANPRM, and one which will undergo review by OIRA and subject it to analyses based on a range of alternative SWF’s and weighted BCA’s and share it with the relevant communities.
Further consideration should also be given to utilizing the concept of opportunity cost, as set forth above with respect to Option 3, as an interim approach to incorporating income distribution into rulemaking.
Accordingly relevant professional groups and associated thank tanks should prepare White Papers and conduct symposia on the same. Upon receipt of the said documents oversight and archival institutions should also take the necessary steps to ensure they are fully considered in the federal decision process.
The idea of using multiple indices to describe the impacts of a federal action, including income transfer effects, was described some fifty years ago; see this post and pages seven and ten of a DOD document titled Establishing Priorites for Public Investments (1969). Also see CRE Critique: Biden Regulatory Program. Relevant case studies are available here.
The emergence of a multi-agency, multi-year regulatory program (Option # 3 in the CRE White Paper of November 2020) could probably proceed simultaneously with the use of SWFs because the simple rule of demonstrating positive net national efficiency benefits as a condition for adopting a rule will probably vanish as the sole decision criterion. Therefore there needs to be a decision rule (forum) to sort out the acceptable rules from those that are not; a multi-year, multi-agency regulatory program could fulfill that role.
Subsequent to the passage of the Administrative Procedure Act in 1946 there was a considerable advancement in the development of the tools that could be used to address shortcomings in the administrative state. Considerably less attention was devoted to explaining the limitations of the aforementioned tools and the attendant need to meld them into a cohesive management regime.
More specifically most professions have a dead horse to beat. The legal profession alleges that regulatory guidance is “not-binding” when it states that although it is not binding on regulated parties it is binding on lower level regulators unless it is modified by their supervisors; many in the economic profession accept the Kaldor-Hicks test which argues that since in theory it is possible that those who benefit from a rule can compensate those who bear the cost of the rule that such a rule enhances consumer welfare even though the said payments seldom occur.
The aforementioned myths exist because academicians survive because they publish; practitioners survive because they do not publish. Consequently academicians are often forced to create the administrative record based upon an incomplete set of facts. For example, it took nearly fifty years to gain acceptance of the fact that centralized regulatory review did not begin with neither Executive Order 12291 nor Executive Order 12866.
However it should be emphasized that academicians are not disinterested bystanders. Academicians frequently view practitioners through dark lenses because practitioners frequently have “clients” who in the mind of many academicians are driven solely by the profit motive. Consequently any number of academicians believe they do not suffer from such a handicap when they are compensated by federal agencies, benefactors and private foundations.
This is not to suggest that both academicians and practitioners do not contribute to myopic panaceas in their refusal to disclose to the public during the promotion of their pet idea the potential resultant threat to the public’s well-being as a result of the cancer inherent in the monumental increase in federal and personal debt over a number of decades. Taking the high ground of promoting a particular concept is far safer and more enjoyable when it is done is a vacuum.
Given the above we believe that considerably more attention should be paid to the developing principles for the management of the administrative state even at the cost of reducing the time spent on the improvement of its protocols. It is time to give a serious look at SWF’s for the management of the administrative state.
A fundamental issue to be addressed is how can distributional concerns be addressed in the conduct of benefit-cost analyses? When we speak how it can be addressed we do not mean a theoretical explanation based upon a myriad of the concepts common to welfare economics. Instead, we mean the availability of procedures which can be used by federal regulators on a scale comparable to that set forth in OMB Circular A 4?
Whatever the shortcomings of BCA as presently practiced it does provide a screening process for separating the wheat from the chaff. Let us take a look back in history to examine how, if ever, such a problem was addressed in the past on a government-wide basis. Those who feel that history has little to contribute should stop reading at this point. This point is made because nowhere in the vast literature written by academicians from a wide range of backgrounds have we seen this particular topic addressed in the detail it deserves.
Benefit-cost analysis as now practiced had its beginnings with the passage of the federal Flood Control Act of 1936 when it stated that the civil works projects of the Corps of Engineers could be constructed only if the the “benefits to whomsoever they accrue exceed their costs”. With that statement the Corps amassed its considerable analytical talent not only to delineate the appropriate project evaluation procedures but to implement them across an entire federal agency consisting of dozens of District and Division offices under the management of the Corps headquarters in Washington.
Some twenty five years later the author entered the scene as an analyst in the Office of the Secretary of the Army which oversaw the Corps. There was one, if not the turning point, episode that brought distributional concerns to the forefront.
The Johnson Administration, along with the support of the Congress, wanted to utilize the civil works projects of the Corps, primarily dams and waterways, to assist in the redevelopment of impoverished areas in Appalachia. The Corps examined potential projects in terms of its conventional benefit-cost analyses procedures. These procedures examined national efficiency gains such as flood damages prevented and reduction in transportation costs. Unfortunately, it was very difficult to find projects which would meet the aforementioned criteria.
Consequently the proponents of the projects took their arguments to the Congress stating that the Corps failed to include “regional benefits” in its calculations. By regional benefits the proponents meant primarily income transfers to the local populace resulting from employment during the construction of the projects as well as “secondary benefits” resulting from salaries resulting from business induced to locate in the project area. The proponents of incorporating “regional benefits”, which include the aforementioned “secondary benefits”, into the traditional benefit-cost ratios of the Corps were leading members of Congress, some of which whose name continue to adorn Congressional office buildings.
The end result was that the Corps never agreed to combining regional and secondary benefits with those in the national account; however, the group in the Office of the Secretary of the Army who opposed such a combination and who eventually began applying benefit-cost analyses to regulations was abolished by an act of Congress.
Therefore, the bottom line is that any answer to the problem at hand must not only be functional and capable of being implemented across a number of large institutions but it must stand the test of time by accommodating the concerns of diverse stakeholders.
It should be noted that if you are unemployed and responsible for feeding a family the aforementioned “regional” and “secondary” benefits are very bit as important as those accruing to the national account. That said the economic profession was never overly supportive of addressing these non-national account issues.
Let’s explore some of the options for doing so.
One option is to argue that it is not the responsibility of economic evaluations to address distributional issues, instead it is argued that these concerns are best addressed by tax programs and direct income transfer programs.
Another option is to display national income effects simultaneously with the transfer payments.
The shortcoming of the first option is that there was no direct mechanism for energizing other federal programs in the same time frame as implementing a Corps project; with respect to the second option the transfer payments were not given serious attention because of the traditional emphasis accorded to national efficiency gains in the implementation of Flood Control Act of 1936.
It is in recognition of the aforementioned history that leads us to the conclusion that Option # 3 in the CRE White Paper of November 2020 be wedded into the OIRA review process. Option #3 allows a wide range of impacts associated with a proposed regulation to be considered and that the merits of a proposed regulation be reviewed simultaneously with other available regulations prior to including it into a multi-year, multi-agency government-wide regulatory program.
Option # 3 states:
3) Initiate a Holistic Program for the Management of the Administrative State: Assemble, publish, seek continued public input and manage a multi-year, interagency, government-wide regulatory program and codify a Magna Carta which defines the scope of the attendant administrative state. The adoption of this option would put an end to ad hoc proposals by agencies for OIRA to endorse a regulation void of any consideration as to how the recommended projects relate to complementary or substitute programs of other agencies. This option would also make it easier to implement a government-wide initiative of high priority to the incumbent Administration but subject to public announcement an review–not hidden in numerical analyses. Income distribution could be addressed within this option.
A Very Tentative Review of the Published Landscape
Let us begin with our ultimate objective: we wish to develop a viable set of instructions similar to OMB Circular A-4 which would provide detailed and understandable procedures for incorporating distributional impacts into the OIRA review process. The adoption of Option # 3 could, with sufficient work, be implemented in a manner which fulfills the aforementioned objective and could well be adopted as an interim measure until the option described below is developed in greater detail.
Basically we are confronting head-on a debate that has been underway in the economic community for decades and probably nearly a century. Benefit-cost analysis is based upon well established theories emanating from welfare economics which in turn are based on the neoclassical paradigm of a population composed of well informed and economically rational citizens. Neoclassical theory is claimed to be universal in nature and it is the foundation of a vast enterprise of model building used to describe and predict economic events under a wide range of market conditions.
Nonetheless there has been a growing minority of economists who question the existence of humans abiding to the strict laws of economic rationality. One of the leaders of this movement was Professor Herb Simon, a Nobel laureate in economics and then of the Carnegie Institute of Technology, who in the mid fifties took exception to the views of the neo-classical economists and attempted to educate engineering students, of which the author was one, of the existence of a populace whose behaviors did not reflect those of a completely economically rational person.
Dr. Simon was the advocate of “bounded rationality“. Bounded rationality “is the idea that, when individuals make decisions rationality is limited by: the tractability of the decision problem; the cognitive limitations of the mind; and, the time available to make the decision”. A more direct interpretation might be that Herb Simon believed the neoclassical economists were out to lunch in some of their theories of human behavior. Dr. Simon was also of the view that sound management is subject to the rules of scientific inquiry, thus his leadership in establishing the Graduate School of Industrial Administration at Carnegie and which set him on a public collision course with Dwight Waldo, the father of public administration.
In a nutshell, practitioners who are examining ways to incorporate distributional effects into benefit-cost analysis are in the middle of the Neoclassicists v. The Behavioral Economists brawl. Advocates of Social Welfare Functions took utility to be additive and measurable, even though the early Neoclassicists had cautioned against such assumptions. (International Encyclopedia of the Social and Behavioral Sciences, 2001).
Nonetheless let’s begin with the relevant work in welfare economics. Our review suggests that in the majority of instances that the quants that inhabit the world of welfare economics generally write for themselves and it will take a considerable effort to comb their writings and develop a usable algorithm comparable to Circular A 4 in the immediate future.
We do not pretend to be knowledgeable of all the relevant research but at this time we cannot find any documentation that comes close to meeting the aforementioned standard of the existence of operating procedures comparable to those set forth in Circular A- 4. Those who have access to such information should forward it to this address.
In the absence of a usable algorithm are there any case studies which address the issue of incorporating non-national efficiency impacts in BCA? In this instance we have identified some useful information.
One way to loose a considerable amount of your readers is to resort to detailed mathematical discussions which often lack the ease of qualitative analyses; consequently all the descriptions herein are qualitative. Nonetheless in that game theory is one foundation of welfare economics and for those interested in exploring its application to parties combating for limited military resources, they might be interested in this post.
This article describes three case studies dealing with the shortcomings in traditional benefit-cost analysis because the aforementioned studies do not quantify a range of benefits that traditionally fall in the “non-quantifiable” category. A number of the “non-quantifiable” categories of major import are discussed including “the costs borne by those continually suffering through chronic conditions beyond their out-of-pocket expenses”. The article also emphasizes the longevity of the continual psychic costs associated with chronic conditions that are also ignored in conventional benefit-cost analyses. The crux of the analysis is in this statement by the author in the aforementioned article:
The hedonic approach does not represent a substitute for traditional accounting of economic benefits, but it ensures that there is a least some accounting of how individuals will experience the effects of the rule in reality. Although the degree to which an accounting of those hedonic measurements might be balanced against economic costs remains debatable, simply put more information is superior to less when designing ideal regulatory policy.
We agree with this statement but with a caveat. Implementation of Option # 3 in the CRE White Paper allows to a degree a quantification of the aforementioned hedonic costs in that it identifies at the macro level their opportunity costs. That said, in some instances the enormity of the opportunity costs are self- evident.
Here is what two economists have written regarding the use of happiness data, which is integral to the hedonic approach, in a leading economic journal:
Happiness research is based on the idea that it is fruitful to study empirical measures of individual welfare. The most common is the answer to a simple well-being question such as “Are you Happy?” Hundreds of thousands of individuals have been asked this question, in many countries and over many years. Researchers have begun to use these data to tackle a variety of important questions in economics. Some require strong assumptions concerning interpersonal comparisons of utility, but others make only mild assumptions in this regard. They range from microeconomic questions, such as the way income and utility are connected, to macroeconomic questions such as the tradeoff between inflation and unemployment, including large areas in political economy. Public policy is another area where progress using happiness data is taking place. Given the central role of utility notions in economic theory, we argue that the use of happiness data in empirical research should be given serious consideration.
Benefit-Cost Analysis is not without a few warts; what about the use of Social Welfare Functions?
The objective of utilizing a Social Welfare Function (SWF) is to make improvements outside the realm of Benefit-Cost Analysis (BCA) through the use of a mechanism which allows one to assess the utility imparted to individuals who are subjected to a particular policy decision. This requires interpersonal utility comparisons which neoclassical economists have argued is impractical because they state that there is no objective unit of measurement to compare utility among participants. A response is that ” utility simply means the satisfaction that a consumer experiences from a product or service” and that although utility , which is measured in utils, cannot be measured directly economists have developed surrogate ways to do so.
Another shot across the bow of the use of Social Welfare functions came about as a result of the work of Nobel laureate Ken Arrow who concluded that “aggregating individual ranks is inherently flawed”. Others state it this way “Rarely are all the conditions present that would enable one to arrive at a true social order of available outcomes”. That said Arrow’s Theorem also highlights shortcomings in the foundations of Benefit-Cost Analyses with the resultant conclusion that both mechanisms must work within the constraints that make the utilization of such models of collective choice viable.
The bottom line is that both BCA and SWF’s have warts but nonetheless they both provide us methods for a rank ordering of alternative policies which address a particular objective.
The Use Social Welfare Functions to Augment or Replace Traditional Benefit-Cost Analyses
The literature is rampant with a wide range of alternatives to address deficiencies in BCA; they range from abolishing BCA, along with OIRA, to its replacement through the use of methodologies made available as a result of advances in the development of social welfare functions.
It is well beyond the scope of this paper to address or even try to address the current state of measuring social welfare through the prudent use of social welfare functions. We have, however, addressed the question as to whether the state of the art is sufficient to allow for the publication of an OMB circular similar to Circular A 4 to guide the implementation of a SWF based program to either augment or replace the existing BCA program administered by OIRA, a program used to manage the administrative state by analyzing the merits of a proposed regulation.
As was discussed in the previous section the use of SWF’s continue to guide research agendas in number of areas but they are at the macro level in the choice of analytical paradigms to address policy issues but they have not found their way into the daily operations of the administrative state. To that end we have found insufficient evidence to conclude that in its present state that the SWF methodology can be implemented on a government-wide basis and most certainly not at the micro level envisioned in the analytical framework set forth in OMB Circular A-4.
That said the question that remains is whether the SWF methodology can be incorporated into the BCA process on an incremental basis? For example there is literature on introducing weighting factors for certain components used in of the conventional BCA analysis. What is needed is a survey of the manner in which the SWF methodology is used by either federal regulatory agencies or practitioners to address issues of record before an administrative body. A cursory reading of the record reveals that the SWF methodology has been used in a number of instances by academicians to determine optimal strategies when studying policy issues, for example tax issues. However what is missing from the database reviewed by CRE are examples of how the SWF methodology can be used on a routine basis throughout large institutions to address issue specific regulatory issues embodied in a proposed regulation.
To this end, consideration should be given to identifying a regulation which is in its formative stages, for example an ANPRM, and which will undergo review by OIRA, and subject it to analyses based on a range of alternative SWF’s, Distributional Weights and explicit statements of their opportunity costs.
Proponents of incorporating income distribution concerns into the rulemaking process are encouraged to generate such a database.