Is cost-benefit analysis a good idea for financial regulation? Ostensibly an essay addressing that question, this piece expanded to a rather critical survey of financial regulation, as I thought about what the costs and benefits of financial regulation are. It's based on a presentation I gave at the Sloan Conference on Benefit-Cost Analysis at the University of Chicago Law School last fall, with many interesting papers, most of them more optimistic.
HTML here, to make it easy to read. Pdf and permanent link here which is where updates and a final (I hope) published version will reside.
Regulations should only be enacted if their benefits exceed their costs. Who can object to that?
That’s not the question. The question is whether legal requirements for cost-benefit analysis, a new legal and regulatory process erected around such calculations, a “judicially enforced quantification” (Coates 2014) on top of the current regulatory procedure, would produce better policies. Would laws forcing regulatory agencies to produce cost/benefit analysis, of certain specified types, with specified codified methods, and allowing proponents and opponents of regulation – who often have strong private reasons to favor one outcome or other – to challenge regulations on the basis of cost/benefit analysis – and especially, to challenge the cost-benefit process – overall produce better policy results?
As in environmental regulation, the specific proposal, I suppose, would be that Congress should pass a bill amending the Dodd-Frank act and similar thousand-page authorization bills with specific language directing the alphabet soup of agencies to conduct formal cost-benefit analyses, with the consequent right of affected parties to sue if the process is not followed correctly. Though our financial regulation is, in my opinion, of vast direct and indirect cost compared to its meager benefits, I am dubious of this proposition. An additional, formal, legal, adversarial, cost-benefit analysis process is likely not to produce regulations with more benefits and fewer costs, as economists understand the terms. The nature of the important costs and benefits resist the kind of objective quantification that is necessary for a formal and codified legal process. The nature and current state of financial regulation suggests that a formal cost-benefit process will become just as captured and derailed as the conventional regulatory process.
I salute the question. Surely after all these years – centuries, really – in which economists have howled into the wind their warnings of unintended consequences, of law and regulations with costs far exceeding ephemeral benefits, or delighted in their little optimal-planning and clever regulatory-construction schemes, only to be totally ignored, it surely is useful to pay more attention to the process question than just the result question. Go beyond writing optimal-policy advice to the benevolent regulator, he isn’t there. Instead, think harder about how to structure a policy process – a political process bringing together regulators, regulated industries and people, and, somewhere, representatives of consumer interest, and what quantifiable, scientific evidence that can be assembled – so that something resembling the economists’ “optimal policy” is likely to result. Ask how we should decide, not just what the answer should be. This question is good. I am simply skeptical that cost-benefit analysis, as implemented in the typical US regulatory process, and added to that process, is the central answer to that question in the case of financial regulation.
It is useful to have a few concrete examples in mind. The large-scale efforts to prevent another crisis such as happened in 2008 surely are on the top of the agenda. Should regulations address assets, liabilities, or prices? Assets: we can send an army of regulators out armed with an encyclopedia of rules, to try to regulate the investments of the too-big-to-fail banks so those banks never lose money again. Or perhaps regulators can micromanage the amount or form of executive compensation, so that executives do not choose risk profiles that taxpayers eventually regret. Liabilities: regulations can instead strongly increase capital requirements, either with quantity limits or (my favorite) Pigouvian penalties for debt and especially run-prone short-term debt. Markets: “Macro-prudential” regulation is the new hot idea. The Federal Reserve will intervene in a wide range of financial markets to “stabilize” prices, diagnose and pop “bubbles,” manipulate lending flows, and (inevitably) support collapsing prices, all so that the assets held by highly-leveraged banks never lose money again. Dodd-Frank: all of the above and more. Which of these approaches produces better benefits for its costs? Would a legal cost-benefit structure produce the right answer?
Pervasive small-scale financial regulations are also worth considering. Financial institutions were already highly regulated. Under the Dodd-Frank act, the various regulatory agencies are unleashing tens of thousands of pages of new rules governing every nook and cranny of the financial system. A quick look at, say, “interim final rule authorizing retention of interests in and sponsorship of collateralized debt obligations backed primarily by bank-issued trust preferred securities,” randomly chosen as the top item on the Fed’s website as I revised this paper, or “Prohibitions and Restrictions on Proprietary Trading and Certain Interests in, and Relationships With, Hedge Funds and Private Equity Funds,” likewise at the Commodity Futures Trading Commission , or “Adjustment to Asset-Size Exemption Threshold Home Mortgage Disclosure” at the new Consumer Financial Protection agency, makes any economist yearn for some sort of vague accounting of costs and benefits – but equally leery of additional layers of process.
When people envision cost-benefit analysis, I think they have something much more straightforward in mind, as in environmental regulation or automobile safety. Someone proposes that power plants should install more expensive scrubbers, to reduce mercury or sulfur dioxide emissions. But the scrubbers cost a lot of money. The Sierra club and the EPA staff are gung-ho; save the planet. The power industry and big power consumers oppose them, citing the costs of higher electricity prices. OK, let’s battle this one out with numbers: How much less mercury, how many more fish, how much lower concentrations in downstream humans, how many lives saved or improved, multiplied by a dollar figure; how many millions to install the scrubbers; how much higher prices. Similarly, forcing auto companies to install airbags has a defined cost, and one can estimate fairly reliably the number of lives saved.
Many proposed regulations have benefits thousands of times greater than costs, and vice versa for many others. Rabid proponents or opponents reveal the emptiness of their case when forced to come up with numbers. Such cases are, I think, the best case for cost-benefit analysis.
But even in simple environmental or auto-safety examples, cost-benefit analysis is often muddier. Economic elasticities are harder to calculate than equipment costs and medical benefits. How many businesses will close or move if they pay higher electric bills? How many people will move? What other kinds of power generation will expand, and what are their economic and environmental consequences? When we make planes safer but more expensive, how many people shift to driving much more dangerous cars?
One step further, once these decisions are in play, how much will power companies or environmental groups invest in lobbying, hiring ex-EPA staff, or other ways of capturing the regulatory process? How soon until businesses use the regulatory process to keep out new competitors and innovators? These costs are never mentioned.
Cost-benefit analysis attempts to enshrine the economist’s framework for evaluating policies: We worry about distortions, costs and benefits to society as a whole, not transfers. We express all costs and benefits in dollars – social, cultural, quality-of-life or other goals that proponents are not willing to reduce to a dollar value don’t count. We demand a documentable market failure and a clear cause-and-effect path by which regulation remedies it before acting.
Most of the interested parties, however, are deeply interested in transfers. The political process is all about transfers, and to heck with the distortions. All sides are happy to act on the flimsiest of cause-and-effect thinking.
The cost-benefit review process can turn in to one more element of regulatory capture, adversarial delay, or a way to induce transfers out of regulation. Finding a small endangered species on the lawn is a great way to stop the development next door that would block my view. See Keystone pipleline – a disaster no matter which side of the debate you agree with. The likelihood of this outcome is never considered in cost-benefit analysis.
Most deeply, legal and regulatory cost-benefit analysis is really a voluntary agreement by parties to adjudicate deep controversies using its methods. Parties must agree that cost-benefit analysis really should drive the outcome; for example that transfers or unquantifiable costs and benefits should be ignored, and thus to accept cost-benefit outcomes. They must agree on at least the basic framework for calculating costs and benefits and cause-and-effect mechanisms. Cost-benefit analysis falls apart – or becomes malappropriated as a tool for obstruction and rent-seeking – when parties are unwilling to agree on that framework. If “leave a clean planet” is a moral imperative above and beyond any documentable costs of a specific kind of pollution, or once social, esthetic, cultural, or other non-quantifiable policy goals are central to one or the other side’s position, cost-benefit analysis falls apart. If there is no agreed-on scientific framework, then it devolves to a battle of befuddlement by high-paid “experts.”
We don’t try to resolve social policy issues like, say, gay marriage or drug regulation by cost-benefit analysis, precisely because people of deeply held and widely varying views will not concede that the argument can be boiled down to dollars.
I think much of the wish among economists for cost-benefit analysis amounts to a wish that someone should put us in charge. “If I were in charge the first thing I’d do is…” echoed through comments made at the conference where this paper was first presented. Planning would have worked if only the planners had been as smart as we are.
But even if we were as smart as we think we are, non-economists also think everything would be fine if they were put in charge, to impose their views on the rest of us. Until parties agree on the rough framework of cost-benefit analysis, trying to institute a legal cost-benefit analysis framework in a democracy is not likely to lead to better policies.
And even the smartest of economists have trouble sticking to utilitarian analysis. It’s not obvious that an aristocracy of benevolent economists would do a whole lot better. Choose your favorite policy disaster enacted with lots of economists at the helm as an example. I like the ACA, but its sympathizers can surely find alternative examples just as disastrous in their minds.
In addition to a certain immodesty regarding Hayekian limitations of planning, the conference of academics at which this paper was presented evidenced no real agreement on my above list of principles. The session “Insurance versus Gambling,” for example, debated the proposition that the Federal Government should regulate in a paternalistic manner, anathema to standard utilitarian cost-benefit analysis. As in popular debate, the economists at this conference bubbled with ideas to regulate transfers. Offered a chance to testify that transfers represent “costs,” it will be easy to line up famous names.
With these thoughts in mind, financial regulation seems a poor candidate for top-down imposition of formalized cost-benefit analysis. The important costs and benefits are nearly impossible to quantify objectively. Policies and regulations do not come one at a time, but (ideally) in a coordinated mix, and (realistically) as stopgaps and a patchwork for unintended effects of existing policies. Financial regulation is, in a sense, all about regulating transfers, who gets how much money at whose expense. Yes, “illegitimate” transfers may impede markets, but the extent of market damage not desire to protect voluntary losers is supposed to end up in cost-benefit analysis. For every loser there is a winner, and finance is about taking risks to earn rewards. Far from clear documentable distortions, the cause-and-effect mechanisms adduced in many financial regulations are hazy cocktail-party stories relative to even the least scientific pollution studies. And that’s among academics. Wait until the agencies and lobbyists get to work. The amount of money at stake raises the potential for capture and other political-economy costs and dangers into the trillions. The parties in financial regulation seem nowhere near ready to agree that utilitarian cost-benefit analysis should govern policy outcomes. Certainly, utilitarian cost-benefit analysis makes essentially no sense of current regulation. Multiple overlapping agencies typically work to cross purposes.
Formal cost-benefit analysis requires a codification of procedure: what constitutes acceptable “science,” which cause and effect mechanisms matter, and so forth. The state of knowledge and professional agreement in financial economics does not approach this state. Experts will disagree on basic methods, the existence of causal channels, and even the definition of basic terms like “systemic,” “liquidity,” “bubble,” and so forth.
Financial regulation is a very complex part of economic regulation. Legal cost-benefit analysis is not a central procedure anywhere else in economic regulation – tariffs and quotas, farm price supports and subsidies, occupational licensing, zoning laws, labor and union legislation, price controls, anti-trust law, among many other examples. The Joint Committee on Taxation and Congressional Budget Office have only just started to incorporate behavioral responses to tax policy in their analyses – how much people will work more, invest more, etc. in response to lower marginal tax rates. Financial regulation is all about behavioral responses. The huge national argument over the minimum wage – a very simple economic regulation by the standards of the Dodd-Frank act – is not headed to a cost-benefit arbitration. If economic regulatory policy is to be mediated by legal cost-benefit analysis, it’s not obvious that we should start with finance, the hardest case of all!
One may say that it can’t hurt; at worst we add something like the hilarious paperwork reduction act notices and make agencies go through some extra hoops. Some real whoppers can get stopped out of the gate. Rose and Walker (2013) make a persuasive case along these lines.
But there is a real danger as well. A legal cost-benefit analysis will focus on the quantifiable, the kinds of costs that we can write a form and check the boxes to assess. For example, cost-benefit analysis can easily focus on compliance costs, the (large) number of employees who spend their days filling out forms, when the actual costs of a financial regulation in distorting market outcomes, causing or curing crises, or capturing the financial industry are orders of magnitude higher.
And, let us remember, not all regulations are stupid. A legal requirement for cost-benefit analysis drags out the regulatory process even more. More layers of process risk the danger that nothing gets done. Even critics of all regulation should not like this outcome, as it leaves the existing and clearly faulty regulatory structure in place.
For example, there was a run on money-market funds prompting a bailout in the financial crisis. Five years later, the child’s play regulation to fix them still has not been enacted by the Securities and Exchange Commission. (Restrict fixed-value funds to investments in treasuries, require floating values and a secondary market, or make them issue some equity. It’s not hard.) The industry has neatly been able to derail and capture the process. Imagine what they would do with another set of cost-benefit hoops to object to.
I conclude that the costs and benefits of financial regulation are of the type that a regulatory, legal and adversarial process focused on ritualized cost-benefit analysis, as it is currently implemented for environmental and transportation policy, is unlikely to produce better policies – though it may produce huge fees for financial economists and lawyers!
The alternative is to continue cost-benefit analysis in the less formal, less legalistic policy evaluation sphere. In the policy evaluation sphere, we can talk about important costs and benefits, even when they are not quantifiable or included in a ritualistic process. The kinds of costs and benefits, measurement techniques, and mechanisms of costs and benefits can remain fluid, with the court of public opinion and the existing regulatory process the final arbiter.
As an example, people in the policy debate seem to be converging on the idea that raised capital standards are better than asset regulation to address financial crises. That debate centers nicely on costs and benefits. But I doubt that a legalistic cost-benefit analysis of the Dodd-Frank act would have led to this reevaluation.
This is not a defense of the status quo! The regulatory process, with thousand-page monstrous bills authorizing tens of thousands of pages of complex regulation, years of sclerotic delay, and ultimately leading to whimsical discretion among increasingly politicized regulators, is clearly broken. It needs an overall reform, not just the addition of a formalistic set of cost-benefit hoops.
Preventing financial crises
The most important potential benefit of financial regulation, especially now, is the desire to prevent financial crises – a recurrence of the panic or run of Fall 2008. That was, of course, the avowed purpose of the Dodd-Frank act. What are the costs of a crisis, and thus potential benefits of a regulation that eliminated them? Real GDP fell 10% in 2008, or about one and a half trillion dollars, and five years later has really not regained much ground relative to the previous trendline or “potential.” So, by that back-of-the-envelope measure, the cost is $5-7 trillion dollars and counting. About 10 million people stopped working, and the employment-population ratio has not recovered since. And we should include the costs of government policy as well. Another trillion dollars per year or so of stimulus, automatic stabilizers, and other recession-induced spending will have to be paid by taxpayers or government creditors eventually.
But it is not really clear just how much of those costs one should attribute to the financial crisis itself. Even without a crisis, there would have been a boom and bust in housing, like the boom and bust in tech stocks of the late 1990s. Even without a crisis, we would have had a recession. Quite plausibly, we had a boom in housing, the appearance of a normal recession precipitated a bust in housing, and the bust in housing precipitated the shadow-banking run that caused the financial crisis. How deep would the recession have been without a crisis? Opinions can vary, and macroeconomic models are not reliable enough to produce anything like the kind of counterfactual one wants – though one can be sure a cost-benefit hearing would produce armies of high-priced economists bearing models and predictions accurate to three decimal places.
Many observers calculate the loss of housing and asset values as a “cost” of the financial crisis. But here again the definition of “cost” is not so clean. Much of the decline in asset values, including housing, turned out to be temporary. Famously, many (but not all) of the government’s bailout investments ended up turning a profit. The AIG portfolio of credit default swaps recovered, the AAA tranches of mortgage backed securities recovered, and stocks gained back their losses. Do we count the mark-to-market loss, which may have reflected the possibility that things could have gotten much worse, as they did in the great depression, or the long-term loss, if any?
Even if there are permanent losses in asset values, how much of those represent transfers and how much represent a loss to national wealth? A sharp decline in housing values is great news if you are 30 and have a job. You will spend a lot less of your lifetime income on housing, and therefore have the wherewithal to spend more on other things, or you will get a much nicer house for the same money. A sharp decline in stock price/earnings ratios has a similar, largely distributional effect. We would all welcome a technological discovery that cuts the price of cars in half, despite its effect on used-car prices. Houses that are still there, and factories that are still there, but at lower prices, are conceptually different from houses and factories that have washed in to the ocean. Macroeconomists have to assume asset values are important as collateral for borrowing, not as national wealth, to get them to have much effect in macroeconomic models, and even this argument requires “frictions” that might be more profitably be addressed directly.
Now, there is some loss to national wealth – we built too many houses, in the wrong places, as the current values do not cover the costs of construction. But how much? Once you recognize the benefit of lower prices, it suddenly becomes much harder to say.
More generally, how should cost-benefit analysis handle large transfers? Are huge bailouts, from taxpayers to bank creditors and stockholders, from equity investors to creditors, from old homeowners to young home buyers, really neutral in cost-benefit analysis? By standard utilitarian calculus, they are. But of course most of the fights over government policy are exactly about enacting or limiting zero-sum transfers, and much of the public outrage during the financial crisis was about transfers. Much of the explicit goal of the Dodd-Frank act is to limit future transfers – to end “too big to fail.” Since our goal is to think through a political structure that produces better regulation, yet is acceptable to the parties involved, it seems foolish to ignore transfers. Yet it is inconsistent with economic principles to enshrine them as “costs.”
Additionally, the costs of a financial crisis are not inevitable. Reinhart and Rogoff’s (2011) historical survey is more notable for the variety of historical experience rather than the average. Here’s a hard nut for cost-benefit analysis: If much of the cost of a financial crisis is due to the way financial crises spark inept government policies, or if much of the cause of a financial crisis is due to inept regulations and policies in advance of or during the crisis, does mitigating such crises, by regulations that themselves carry substantial costs, count as a “benefit?” If I persist in shooting myself in the foot, do we count the value of an iron cast in preventing broken feet as a benefit, when I could simply stop shooting myself in the foot to start with?
When we think of traditional cost-benefit analysis, we presume that we face some sort of free-market Eden gone wrong, in which unregulated markets suffer some dysfunction that is partially remediable by regulation. How do we address the situation that much of the cost is due to other poor regulations? Cost-benefit analysis treats each regulation in isolation. How do we treat regulations as a package?
The charge that the government mishandled the crisis, needlessly prolonged the recession, and that much of the crisis came from bungled previous regulations and policies, is shared by critics from every point of view, the only difference being which policies the critics dislike. To some, the recession was completely avoidable because the stimulus was too small. Had the government spent $2, $3, $4 trillion per year or more, they say, and even if the spending was completely wasted, output and employment would have recovered swiftly. Some of them claim that the multiplier is so huge, in fact, that extra spending would have been self-financed by the larger tax receipts coming from greater output, literally a free lunch. (For example, DeLong and Summers 2012.)
To others, the disincentives of vastly expanded social programs, continued meddling in housing markets, higher marginal tax rates, macro and micro policy uncertainty, the looming uber-regulation of health care and finance, and the constant meddling by aggressive actions of the NLRB, EPA, EEOC and others caused our stagnation. (For example, Mulligan 2013, Baker, Bloom and Davis 2013, Taylor 2012a.) Many of these actions were sparked by the deep recession, in turn (in some views) sparked by the crisis.
In either view, the true and necessary cost of a financial crisis is much lower than the pain we have suffered. Though they might not agree, Delong and Summers’ ultra-Keynesian view implies that the true cost of a crisis is zero, because a costless policy can eliminate the following recession. So, in measuring the cost of a crisis and benefits of crisis-preventing financial regulation, is it fair to treat inept policy responses as inevitabilities?
Critics from all points of view also identify ham-handed policies as major ingredient in causing the financial crisis in the first place. Some charge that the community reinvestment act was taken to heart by bank regulators who forced banks to make riskier loans, especially in return for approving mergers, and forced Fannie and Freddie to buy and guarantee those loans. Some charge that the Federal Reserve sparked the housing boom by holding interest rates too low for too long. The many subsidies for leveraged homeownership, not the least of which the tax deductibility of mortgage interest and the limitation on capital gains taxes for homes, encouraged housing speculation over less glamorous but financially more stable renting. (The first thing a “consumer financial protection bureau” should to is to heavily “nudge” vulnerable Americans not to invest dramatically in a highly leveraged, illiquid asset marked by huge idiosyncratic risk – the owner occupied home. Good luck with that.) Some believe that regulator’s failure to police “predatory lending” led people to take out mortgages they couldn’t afford. Fannie and Freddy, who went under in summer 2008, were hardly creations of the free market. AIG was a heavily regulated insurance company.
More deeply, the panic or run that was the defining event of the 2008 crisis revealed a morass of bad regulation, ineffective regulation, widely-recognized regulatory arbitrage that nobody did anything about, and failure to address obvious and building moral hazard. Auction-rate securities and off-balance sheet special-purpose vehicles, holding illiquid risky assets funded by rolling over run-prone short-term debt, were obvious end-runs to banking regulation, a way to create a synthetic “bank” without capital regulation or supervision. Collateralized debt obligations, trancheing pools of mortgage backed securities to the limits of ratings, and rating agency connivance in providing those ratings, happened only because regulators demanded that institutions hold securities blessed with particular letters by particular rating agencies.
The moral hazard by which creditors came to expect bailouts rather than bankruptcy had built up like underbrush, from the bailouts of Continental Illinois, the savings and loans, bank investments in Latin America, in the southeast Asian crisis, Long Term Capital Management, and finally Bear Stearns, which was intended to give Wall Street a little more breathing room and instead was interpreted to mean that investment banks like Lehman Brothers were now also guaranteed. (Cochrane 2010.) “Big banks are too complex to go through bankruptcy court,” the mantra repeated, but only because investors presumed them to be government-guaranteed, and so had not taken seriously to the task of fixing bankruptcy law or the bankruptcy provisions of their contracts so those banks could be unwound without chaos.
And direct, preventable, government actions contributed to the crisis. Treasury Secretary Paulson appeared before Congress, on national television, asking for $700 billion dollars, with no clear plan what he wanted to do with it other than an obviously hopeless quest to prop up the market prices of mortgage-backed securities, while over the previous weekend the government put in place a ban on short-selling bank stocks. It’s hard to think of a better way to start a panic. (See Taylor 2009, 2012b.)
So, how do we interpret any measure of the “costs of a financial crisis,” and the benefits of a new but costly regulation that might reduce the chance of a crisis, when so much of the crisis and so much of its severity was the effect of poor previous regulation, and poor policy and regulatory response, rather than a pathology of some mythical unregulated free market?
Financial regulation often works to cross purposes. One person’s or agencies’ “predatory lending,” by which financial companies are accused of forcing borrowing on unsuspecting customers who will not be able to repay the loans, is another’s (or that same person’s, a few years earlier) opening of credit markets to “underserved” income, geographical or racially-defined groups. One agency wants lower loan-to-value ratios, in the name of financial stability. Another wants higher loan-to-value ratios in the interest of community redevelopment. Taxes strongly distort decisions away from saving and investment and toward consumption, but then our government carves out a myriad of complex special deals for tax-advantaged savings. One arm of the government subsidizes short term debt, by the tax-deductibility of interest, the regular and too-big-to-fail guarantees, and by regulatory preference for such debt as an asset by other institutions, for example lower capital ratios for short term debt held as an asset. Another arm of the government wants to reduce short term debt, for its incendiary stability effects, with higher capital ratios, leverage ratios, clawbacks, and so on.
This sort of regulatory contradiction is pervasive. Our government subsidizes and require the use of corn ethanol to reduce emissions, yet bans the import of sugarcane ethanol which might actually have that effect. Our government heavily subsidizes solar cell production to lower prices, and then imposes tariffs against cheap Chinese solar cells to raise prices. A public choice economist might conclude that the purpose of regulation is simply to enhance regulator’s power to extract political and financial support from the regulated in return for subsidies and protection from competition. It would be hard to refute that view in the data.
So, do we measure the benefits of a new regulation on a backdrop of all the perverse and sometimes contradictory regulation that remains? Or do we measure them as a contribution to an ideal regulatory system? For example, do we address the costs and benefits of capital regulation assuming that debt remains subsidized and an effective too-big-to-fail guarantee remains in place? Do we address the costs and benefits of macroprudential bubble-pricking attempts, assuming that banks continue to run with ridiculously low capital ratios?
Financial cause and effect is nebulous. How do we measure the contributions of a specific policy measure to reducing the probability of crises? “Make the financial system more stable” is easy to say, but hard to prove and harder to measure even the sign of such an effect. Pretty much every section of the Dodd-Frank act is sold as a device to mitigate “systemic risk” and to reduce crises. But how many have any such effects? And by what scientifically documented mechanism?
“Resolution authority” is a good example. Its authors say it will end too big to fail and associated moral hazard. Without too big to fail, people will watch their own risks more carefully, charge appropriate premiums for risks, make sure exit plans, living wills and bankruptcy are in order, and otherwise endogenously create a more stable financial system. There, $5 trillion of benefits.
But will it work? I see a contradiction at its core (Cochrane 2010, 2013a). Given the presumption that large financial institutions are too complex to be unwound by a bankruptcy court, which has behind it centuries of law, centuries of case precedent, and the thousands of pages of what-happens-in-bankruptcy small type in every financial contract that critics decry as excessive complexity, will a few appointed officials be able to figure out who gets how many billions of dollars over a weekend? Or will that attempt lead merely to massive bailouts of politically well-connected creditors, who will surely scream of their own “systemic” nature, chaos while nobody knows which contract will be honored, the run of all time as less-well-connected creditors see this coming and try to get out of the way ahead of time, even bigger creditor bailouts to stem that run, and a huge investment by all parties in political influence over the discretionary power of the resolution authority?
So, seriously: would the sort of legalistic cost-benefit analysis followed by the EPA score the resolution authority of Dodd-Frank as eliminating too big to fail? Or as making it worse, the strong current if not rough consensus of current opinion?
We might say that all regulations are somewhat uncertain in their benefits. We don’t know entirely how many lives will be saved by reducing mercury emissions at a particular power plant. Dose-response relationships are debateable. But the uncertainty of even the sign of regulatory effects, to say nothing of the size, is many orders of magnitude larger for the systemic effects of financial regulation.
Policy-makers and financial economists bandy around words like “systemic,” “fire sales,” “illiquidity,” “liquidity spirals,” “bubbles,” or “imbalances” as if they had the same scientific standing as “morbidity,” “mercury concentration,” or “average pollutant transport distance.” But the scientific definition and measurability of any of these concepts would make evolution-deniers blush. The Dodd-Frank act does not even define “systemic.” “Systemic stability” in financial cost-benefit analyses is likely to be as solid a concept as as “preserving the American way of life” or “cultural” benefits are in transportation studies. Except that the numbers are in the trillions.
One might say, fine, let’s bring this argument out in hearings, public comments, and, inevitably, in court. But with huge sums at stake, a clearly important problem, and armies of easily hired “experts” who can befuddle regulators and judges with these stories, it is hard to place much faith in the outcome. Ideally, we would say that regulation should await documented scientific understanding of costs, benefits, causes, and effects. That won’t happen in our lifetimes, so the danger is that we instead give fairy tales the patina of scientific respectability and then enshrine them in law. Maybe it’s better to leave them as fairy tales and keep analyzing them.
Micro-financial regulation and the social contract
Most financial regulation is not aimed at preventing systemic crises, of course. And, to an economist, micro-financial regulation seem more amenable to cost-benefit analysis, since it stays away from nebulous general-equilibrium effects and behavioral responses -- i.e. predictions of how prices, moral hazard, contracts, industry structure, runs and “bubbles,” would react to regulations -- and concentrates on simple questions like what kinds of mortgages you should be allowed to buy.
Quantifying costs and benefits of micro-financial regulation is not that much easier, however. Consider a simple concrete example, regulations to limit payday loan interest rates, or similar regulations to limit mortgages offered to consumers, in the interest of reducing “predatory lending.” How would we measure the dollar value of social benefits of such a regulation? Some people will get loans at lower rates. Some won’t get loans at all. At best, we engineer a transfer from owners of existing companies, and excluded consumers, to the lucky recipients of lower-cost loans. Or consider a disclosure requirement, 10 more pages will be added to stack of forms a company must send to the SEC, or 10 `more pages of forms added to your mortgage boiler plate. How would we begin to define, let alone measure the benefits?
A big reason for this conundrum is that most financial regulation is an answer in search of a question. Policies are proposed, and then (maybe) subjected to analysis, perhaps now including cost-benefit analysis. In the economists’ framework, we are supposed to start with an analysis of an economic situation, find a distortion or externality, and then craft polices. Instead, we end up endlessly coming up with new theories to justify policies invented for other reasons. But those new theories rarely recommend the original policy as optimal anyway. Transactions taxes are a great example of a policy whose justification changes with the season.
This inversion of diagnosis and prescribed treatment reflects a deeper problem. Legal cost-benefit analysis is really an agreement of parties to have their disagreements adjudicated under its framework. Before that can happen, the parties have to agree on the rough aims of regulations and that utilitarian cost-benefit analysis, based on understood market failures, is a desirable arbiter. If imposed from on high, it is more likely that parties will conspire to undermine the whole procedure.
It would be reassuring to see that policy analysis and debate takes on a cost-benefit cast, and parties making cost-benefit arguments of the type recognizable to economists. As I look over financial regulation, and especially micro-financial regulation, however, I struggle to come up with any coherent and quantifiable aim describing current regulations that might be formalized in cost-benefit analysis.
Much Securities and Exchange Commission regulation, such as the regulations against trading on certain kinds of information, disclosure requirements, or rules on how exchanges must process orders (strict time order, time is continuous, price is discrete, orders must be routed to the highest price at the time, and so on) seem to be motivated by maintaining “orderly,” “fair” or “liquid” markets. (The pathologies of high-frequency trading seem to be a spectacular case of regulations having the opposite of the intended effects, see for example Budish, Cramton and Shim 2013.) At best these regulations are aimed straight at preventing or creating wealth transfers. Some regulations, such as perennial rules and occasional bans against short selling, are often motivated by a naked desire to prop up prices for powerful constituencies.
Most banking regulation, financial product regulation, the stack of forms you sign when you get a mortgage, or the huge amount of compliance and disclosure regulation, qualified investor rules, and the whole new so-named bureau, are defended on the basis of “consumer protection.”
But “consumer protection” is an economic controversy going back to Adam Smith and the guilds. The controversy remains with us pervasively in product, food and drug regulation, zoning, occupational licensing, and even taxicab regulation today, to say nothing of finance. For these centuries, economists have complained that “consumer financial protection” justifies regulation whose main actual point is to protect of incumbents from competition, protect their profits and subsidies, slow down disruptive innovation, and provide a steady source of political support for regulators and politicians. They point out that reputation, competition, and private sector ratings are far more effective protections.
The kindest view is that some of this regulation might decrease the political power of financial interests. For example, Zingales (2012) presents a novel view that branch banking restrictions did not just protect local banks from big-city competition, but also limited the national political power of big-city banks. Good luck enshrining that in a cost-benefit analysis.
In any case, despite the centuries that “consumer protection” arguments and counterarguments have played out, utilitarian, distortion-reducing, transfer-neutral cost-benefit analysis has never really taken over the framework for economic consumer protection policy analysis, let alone its regulatory process. The language of costs and benefits is often used, but not the basic idea of adding up dollar values and ignoring transfers.
As I mentioned above, regulation often works to cross purposes, simultaneously taxing or restricting and encouraging or subsidizing the same activity. That fact makes it even harder to find a coherent purpose or cost-benefit motivation of financial regulation.
Much financial regulation amounts to subsidizing credit for favored groups: small business loans via the small business administration, large business loans via the export-import bank, student loans, home loans, loans to “green” energy projects, and so on.
How might an economist approach these problems? First by stating the goal: The economic function of financial markets is to channel savings to investment, introducing as few distortions along the way as possible; and to foster risk sharing. For that goal, and for financial stability, consumer/investors should bear risks commensurate with greater returns directly and transparently rather than as taxpayers wherever possible. This goal makes no sense at all of current regulation.
So, the vast bulk of non-systemic financial regulation is motivated by all sorts of hazy, inconsistent and incoherent goals that have no quantifiable social benefit, no documentable mechanisms to produce its incoherent goals, no measurement of whether goals are reached, and the process is already deeply captured.
Perhaps that view should encourage me to advocate that forcing cost-benefit analysis on the whole process might lead us to a free-market financial nirvana and throwing out 90% of this regulatory structure. But I doubt it. With such strong traditions behind us, and with the true costs and benefits, and their causal mechanisms so hazy, surely the effect will be to enshrine in law and regulation “benefits” that are not benefits in any recognizable economic sense, cause-and-effect channels that defy rational analysis, and thus to further insure the regulator’s power and the industry’s desire to capture it. When the actual actors that do the “forcing” get to work, it is not the economist’s framework that will be enshrined.
We imagine that cost-benefit analysis will enshrine the utilitarian economists’ view of costs and benefits. But surely advocates will want to count as “benefits” the benefits they count now, such as numbers of customers who use a product, ignoring alternative products, numbers who receive lower interest rates when “protected” from “abusive” loans they would voluntarily have taken, ignoring those who get no credit at all, dollars transferred from banks to customers, ignoring alternative uses of the money, or the value of community redevelopment, ignoring communities who did not get development.
Allowing such arguments – which a political system will have to do – will make a mockery of cost-benefit analysis.
Cost-benefit analysis would have to be imposed from above by Congress, not the absent subject of the regulatory-passive voice. Surely, Congress’ intent and language would not be strict utilitarian cost benefit analysis, ignoring transfers. Congress wrote the Dodd-Frank act!
Environmental and safety cost-benefit regulation came in to force in an environment in which all sides of the debate could pretty much agree what costs and benefits mean, and that something like a tradeoff between costs and benefits measure desirable policy, and motivated previous rounds of regulation. Legalized cost-benefit analysis followed a cost-benefit tradition in less formal policy analysis. Legalized cost-benefit analysis falls apart when that isn’t the case – when the value of species diversity, of a general moral imperative to leave a clean planet, or protection against vaguely-understood tail events as in climate arguments are in fact central to one or the other side’s position. As in these cases, when existing regulation, the motivations of the various interested parties, and the language of the policy debate does not even vaguely conform to cost and benefit ideas that economists would recognize, I fear that introducing cost-benefit analysis will become just one more tool in an agonizing and increasingly sclerotic legal and regulatory process.
Costs, seen and unseen
Costs of financial regulation are just as nebulous and difficult to assess as are the benefits – and just as important.
The financial industry complains about compliance costs. They are real. But when compliance and paperwork costs become a large part of the argument, we know we are missing the point, because for finance especially the effects of regulation on the market outcome is certainly orders of magnitude larger than salaries of people filling out forms.
For example, Batkins and Brannon (2013) examine the costs of Dodd-Frank, savaging the agencies implementing the act for failing even to attempt cost-benefit analysis. They add up official estimates of $15.4 billion and 58 million paperwork hours, but point out these are vastly understated. Legal costs are more serious. One report (Jenkins 2013) totaled up $108 billion in legal fees since 2008 at the 6 big banks alone, an item usually overlooked in agency cost-benefit analysis.
But even if Dodd-Frank paperwork consumes $50 billion of annual paperwork hours, if it lived up to its promise of ending $700 billion bailouts, $2,000 billion per year stimulus, a $7,000 billion loss in GDP, and 10 million unemployed, it would be worth it.
Conversely, the true potential costs of financial regulation are orders of magnitude larger than its legal and compliance costs. The costs of financial regulation are its effects on the flow of credit, on innovation, competition, and entry into the financial system, on capture and misuse of regulation to protect incumbents, and the tendency of regulation to produce perverse outcomes such as making financial crises more likely rather than less. If the Dodd-Frank act results in 20 years of slow growth, and crony corruption of the whole financial sector of the economy, and a bigger crisis next time, perhaps triggered by sovereign debt, well, those costs also dwarf the paperwork. Good luck hoping that any of this will be measured in in a cost-benefit statement.
Take a very simple example. Regulations specify minimum amounts of capital that swap dealers should have. Well, that makes sense, no? We want safe, stable, well-capitalized swap dealers, right? Except that imposes a huge barrier to entry to anyone desiring to become a swap dealer. And swap dealing, especially in off-exchange over-the-counter markets, is enormously profitable, leading to a big argument about whether swaps should be cleared on exchanges.
We seem to be heading inexorably to a financial system based on 6 large, complex, too-big-to-fail banks, who are becoming regulated utilities. Lucchetti and Steinbnerg (2013) quote Morgan Stanley’s Chairman, “your No. 1 client is the government,” and that there are 50 full time government regulators working that firm alone, signing off on every deal. And the CFPB has not really started its work yet. Much as the SEC and other agencies love to make headlines attacking the big banks in court, and raking in billion dollar settlements, in the end if the big banks are too big to fail, the government must protect their profits. It’s a pretty safe bet that we will have the same 6 large banks in 20 years as we have today, just as the interstate commerce commission produced high prices, high wages, and the same few airlines on its demise that were there at its beginning. The Southwest airlines of banking will not intrude under the Dodd Frank act.
Where will cost-benefit analysis add up the costs of regulatory capture, the sickening and dramatically expanding crony-capitalist revolving door between Washington and Wall Street? When we think of regulation, we should put our public-choice, Stigler, Buchanan, Tullock hat on, not our advice to benevolent dictator hat on. Surely, and especially with financial regulation, this is one of the biggest costs of regulation. But it is just the kind of cost that traditional cost-benefit analysis and legal procedures are completely at a loss to consider.
The biggest costs are unseen. The businesses that didn’t get started, the people that didn’t get hired by those businesses, the great products those businesses didn’t produce for consumers, the innovative financial products that allowed consumers to live better lives, the savings and investment vehicles that improved people’s lives in their old age, the economic growth that didn’t happen. Looking across countries, the fact that so much innovation and such a large fraction of new innovative companies happen in the relatively unregulated US markets rather than Europe, more dependent on well-protected bank financing, can’t escape us. Once the US heads the same way, we will not have the comparison to tell us what might have been.
And then there are the unintended consequences. Special purpose vehicles, auction-rate securities, overnight repo, and even money market funds, which failed dramatically in the financial crisis, were unintended consequences of previous rounds of financial regulation. These structures were worse than pure free-market banks, because they exploited the weaknesses of the regulated system and government guarantees. For example, special-purpose vehicles and auction-rate securities bought mortgage backed securities, issued short-term money-like paper to finance them, but also had a credit guarantee from the sponsoring too big to fail bank, and no equity. In the relatively free-market era, banks operated with 40% equity or more – depositors would not trust anything else. The credit enhancement from the too-big-to-fail bank provided this reassurance in our system, neatly gaming the regulatory system but creating a banking system even less stable than the pure free-market system would have been.
Money market funds were an invention of regulatory arbitrage. Regulation Q of the now nostalgically remembered Glass-Stegall regulatory system limited the interest rates banks could pay, with the explicitly stated goal of maintaining the profitability of the banking system and reducing competition for deposits. (Well, at last there is a regulation where I can find the goal!) When inflation demanded higher interest rates, money market funds developed to make interest-paying banks where there were none before. And in 2008, there was a run on money market funds, which the government promptly bailed out.
Peer to peer lending and bitcoin are small shoots creating new alternatives to regulated banks which are no longer serving many classes of borrowers or providing efficient electronic transactions services. But as money market funds and special purpose vehicles turned out to have downsides, so may the next round of alternatives.
In addition to regulatory arbitrage, which can lead to results counter to the originally intended ones, financial regulations often directly and predictably cause the regulated system to become more fragile on its own. If we put a big new firehouse on every block, people tend to let their home fire extinguishers rot, they don’t install sprinklers, they store gasoline in the basement, they don’t trim back the trees, and they don’t watch their neighbors as closely.
Before the Fed became lender of last resort, banks had adopted a clearinghouse system which mitigated runs. Banks would declare a temporary suspension of direct convertibility, and exchange deposits for clearinghouse shares. Though arguably imperfect, this system at least provided a bulwark against runs. When the Fed came in, the clearinghouse was abandoned. When the Fed fell flat in its lender of last resort function in the 1930s, the banking crisis was worse than it would have been otherwise.
Before deposit insurance and too big to fail guarantees, banks voluntarily funded themselves with 20 to 40 percent equity capital. Depositors would not lend for less. Now we fight to get the banks to to issue 5 percent capital. And so on.
Policy actions can have the same de-stabilization results as regulations. In the financial crisis, the Fed stepped in aggressively to prop up the prices of various securities. But your fire sale is my buying opportunity, and each time the government props up a price that fell, it discourages the few prudent souls who did not lever to the hilt in the boom, and kept some cash handy to pick up bargains in the bust. They won’t be around next time. Looking forward, if the Fed starts “pricking bubbles” and limiting price rises, the incentive for investors to get in early and bear the risk that prices go down further before they recover is similarly reduced. Asset pricing requires deep-pocket fundamental investors to hang around and make profits from price dislocations. If the Federal Reserve outbids them at the bottom and cuts off their profits at the top, they won’t be around next time.
Where in a legalistic cost-benefit analysis can anyone even consider all these costs? Will there be a line, “anticipated regulatory capture, lobbying, political credit allocation, and corruption”, then a line “anticipated perverse effects of regulatory arbitrage, and gaming around the system,” and just after that a line “enhanced probability of financial crisis due to unintended effects of regulations?” Given the self-congratulatory nature of all legislation and regulation, this outcome is unimaginable. So we’ll fight about paperwork costs. Maybe Keynesians will start arguing that extra paperwork provides jobs.
Another aspect of financial regulation distinguishes it from the sorts of regulation that have benefited from cost-benefit analysis. In large part, financial regulation consists of giving regulators wide discretion rather than simple, clean, challengeable rules. Most rules are so vague, so overlapping, and so maddeningly complex, they mean whatever the regulator chooses the rules to mean. Procedurally, financial companies have to obtain regulatory approval for their actions, rather than follow an objective rule-book. This situation stands in stark contrast to much other regulation, where the rules are clear, and the regulated actor can successfully challenge a decision based on the facts.
The 50 regulators working full time at Morgan Stanley signing off on each deal have in fact wide discretion to decide which deals they like and which ones they don’t.
The Fed’s “stress tests” for big banks are another good example. At first glance, one would think that the Fed would announce the rules for the stress tests in advance, as Basel rules and capital ratios are in theory clearly spelled out objective rules. But the Fed’s regulators are smarter than that. They know that if they announce the rules of the stress tests, the clever MBAs and accountants at the banks will jigger the books to make sure the banks pass the tests – just as they all reported to be well capitalized on the eve of Fall 2008. So the clever regulators at the Fed dream up new and surprising stress tests each time to keep the banks on their toes. Until they quit and go work for the banks.
At a minimum, wide discretion makes cost-benefit analysis nearly useless. How would you analyze the costs and benefits of an energy policy if it consisted of a rule saying that regulators would visit each site and sign off on plants that were “necessary,” “proper,” and not “abusive?”
More deeply, wide discretion invites capture, and stifles dissent. Investment companies are loath to speak out against the Fed, SEC, or CFTC, no matter how silly they think the agencie’s actions might be. Just the announcement of an enforcement action can put firms out of business, even if the action goes nowhere. Companies who want Obamacare waivers know better than to talk about the law negatively in public.
You may think all of this is great. You may think this is all dastardly. Clearly, there are huge costs and benefits involved, and just as clearly they will not be considered in a formal cost-benefit process.
Cost-benefit applied to private parties
This discussion so far concerns only limitations on regulatory promulgation by Federal Agencies. Weyl and Posner (2013) ask a more provocative question – rather than a restraint on regulatory agencies, should cost-benefit analysis be applied preemptively to private actions? They write
“We propose that when firms invent new financial products, they be forbidden to sell them until they receive approval from a government agency designed along the lines of the FDA, which screens pharmaceutical innovations. The agency would approve financial products if they satisfy a test for social utility that focuses on whether the product will likely be used more often for insurance than for gambling.”
Every single one of my objections applies in this case. And more – financial products are by definition new, so cost-benefit analysis would have to rely on theory not measurement. Already in their abstract Weyl and Posner violate the utilitarian, non-paternalistic rules of cost-benefit analysis. If they can’t refrain from wishing to pronounce the line between “risk-taking” and “gambling” in financial markets, imagine how quickly the whole process would spiral out of control in the hands of lawyers and lobbyists.
To an economist, cost-benefit analysis is of course important. The issue is simply whether a legal and formalized cost-benefit process, imposed by Congressional action on regulators and regulated, will produce better policies. I have concluded that it will not. So what do we do instead?
The regular mechanism for policy analysis includes robust debate, academic research, think-tank and government agency analysis, and formal public comment periods for regulation. This informal process, we all must hope, can produce halfway reasonable policy without a specific formalistic procedure. That is, after all, the premise of democracy. If good law, regulation, and policy must depend on filling out a procedural rulebook that nobody believes in or understands, the chance of a good outcome is nil anyway.
The discussion surrounding capital requirements is a good example of this informal process. In the Dodd-Frank act, they pass as a whisper. In 2009, they were viewed with a bit of suspicion as one component of stabilization policy. But in the subsequent broad-based policy discussion, simple and very high capital requirements have come to the fore as probably the best idea a broad swath of involved parties can come up with that has a realistic chance of success.
Some touchstones for this process: The “Squam Lake Report” written in 2009-2010 by a team of academic financial economists (including myself) included a short chapter on “reforming capital requirements.” It includes a list of “costs” of capital requirements, for example,
“Capital requirements are not free. The disciplining effect of short-term debt, for example, makes management more productive… Similarly, some large banks may capture important economies of scale that reduce the cost of financial services.”
And it issues a clear call for at least voluntary cost-benefit analysis (though our inelegant language mistakes costs to banks with social costs appropriate in cost-benefit analysis)
“When designing capital requirements that address systemic concerns, regulators must weigh the costs such requirements impose on banks during good times against the benefit of having more capital in the financial system when a crisis strikes,”
along with a prescient forecast,
“Because they bear all the costs and receive only a small part of the societal benefits, we anticipate that banks will object to this proposal—even if regulators make the right tradeoff between the costs and benefits.”
But you can tell the authors’ collective heart isn’t in it. The chapters on “systemic regulator,” “new information infrastructure,” “regulation of executive compensation,” “improving resolution options,” two chapters on derivatives and prime brokers, and most of all a clever proposal for “regulatory hybrid securities” really draw their passions.
In the following years, my own thinking, and I think that of many economists, shifted away from this story that short-term debt and the threat of a run has a beneficial sword-of-Damocles disciplining device, to the opposite view advocated by Gorton and Ordoñez (2014), that short term debt is held precisely by people who do no monitoring whatsoever – it is an “informationally-insensitive” or “money-like” security, and it is quickly withdrawn when that monitoring might be necessary. And the larger consensus has shifted away from clever schemes for convertible debt, farsighted benevolent regulators, and any faith in resolution, to capital, just more capital.
Admati and Hellwig (2013) is a second touchstone in this process. (See also my review, Cochrane 2013b.) Admati and Hellwig argue s