I come not to praise behavioral economics and its relation to fundamental tax reform, but to bury it. To be more precise, I mean to bury the most commonly and particularly suggested application of behavioral economics to fundamental tax reform: using the insights of behavioralism to support proposals to promote individual savings through ad hoc, tax-favored vehicles. This may come as a bit of surprise because I have long advocated the relevance of behavioral economics or its intellectual parent, cognitive psychology, to the field of tax (McCaffery 1994a), most recently in a co-edited volume attempting to give birth to the field of “behavioral public finance” (McCaffery and Slemrod 2006; and see generally McCaffery and Baron 2006). And yet, while I do think that it is critically important to apply the insights of behavioral economics to fundamental tax reform, I also believe that it is critically important to get the terms of the marriage down right, preferably in advance of producing any (more) offspring.
The typical but, I shall argue, wrongheaded view of the possible marriage of the two subject areas is that reformers should consider substituting an irrational (“behavioral”) model of human behavior for a rational one such as the long-term dynamic optimization protocol at the core of the Life Cycle Hypothesis (LCH) for savings behavior first advanced by Modligiani and Brumberg (1954). The LCH holds that taxpayers use savings and other capital market transactions to spread out consumption over their lifetimes, maximizing utility by allocating the “right” amounts of their lifetime wealth to each time period. Of course, there are rational reasons to deviate from an LCH, such as having a bequest motive; a seminal paper by Kotlikoff and Summers (1981) found that as much as eighty percent of all savings gets transmitted intergenerationally. Others have questioned both the magnitude of this effect and its interpretation (Modigliani 1988; Gale and Scholz 1994) — after all, one reason for intergenerational transmissions can be imperfect implementation of an LCH strategy, perhaps because of the failure of private annuities markets (see generally Kotlikoff and Spivak 1981). But deeper and more persistent criticisms of the LCH have come from the realm of common sense, amplified and supported by behavioral economics. Challenges to the realism of the LCH and its demanding cognitive tasks are indeed compelling. Most notably, Shefrin and Thaler (1988) advocated a Behavioral Life Cycle Hypothesis (BLCH) to supplant the earlier, rational model (see also Thaler 1994). The BLCH holds that, for reasons relating to myopia, problems of self-control, and “mental accounting” (essentially, the cognitive tendency to ignore the fungibility of money/value, as in spending windfall gains (e.g., lottery winnings) on consumption binges not within one’s “normal” budget set or utility function), ordinary persons fail to adhere to the strict predictions of the LCH.
Since savings and the varying tax treatments of savings are at the core of the income-versus-consumption debate, and hence of all discussions of fundamental tax reform (McCaffery 2002 and 2005a), tax policy theorists are left to consider whether a BLCH model of individual savings behavior should lead to different tax reform options than the LCH model would—specifically, whether behavioral models suggest the use of ad hoc tax-favored savings vehicles rather than more fundamental reform (see for example Bernheim 2002, and Auerbach 2006, both discussing the possibilities of such an approach without specifically endorsing it).
So goes the traditional logic, which, I intend, like a rebellious child, to fight at every turn. Let me clarify my perspective from the outset. I am a tax lawyer, trained in economics but not an econometrician, who has also toiled in the terrain of behavioralism for some time now, with some impressive guides. Mine is not an empirical project, weighing in on the too-much-or-too-little-or-just-right savings debate (Pamela Moomau’s Comment surveys some of this debate, then backs off to note the “general consensus that there is not enough savings in the U.S. economy.”) Rather mine is a projected rooted in common sense, on the one hand, and a sound analytic understanding of the structure of our current tax laws, on the other. Simply put, if all people are indeed rational LCH savers, then there is no reason for ad hoc, tax-favored savings vehicles — comprehensive tax reform will get things just right. And if people — some or all or most people — are not rational LCH savers, then there is also no good reason to think that ad hoc tax- favored savings vehicles will get things right, and it is more likely than not that these vehicles will get things wrong, precisely for the “behaviorally challenged” persons for whom they are meant to help. (I shall address Jeff Strnad’s possible counter-example on this point in due course). And since these ad hoc tax-favored savings vehicles are in any event complex, confusing, and costly, we ought to get rid of them in the name of doing something structural and fundamental. While my comments here are limited to the particular case of tax-favored savings vehicles, I humbly suggest that the lesson is more general: complex tax law provisions, sometimes justified in the perceived light of “behavioral economics,” are rarely worth the candle.
I shall continue my rebel’s ride with a consideration of the state of play in fundamental tax reform today.
II. Tax Reform Today: Groundhog Day, All Over Again
In Harold Ramis’s 1993 film, Groundhog Day, Phil Connors, a big city weatherman played by Bill Murray, finds himself stuck inside of Punxsutawney, Pennsylvania (the “weather capital of the world”) where he had ventured in the midst of an horrific snow storm to report on whether “Punxsutawney Phil,” the famous groundhog, would see his shadow, in which case there would be six more weeks of winter. Each morning Connors awakes to discover that it is still February 2, Groundhog Day. Everyone around him is stuck in a perpetual present, endlessly repeating the same old actions and conversations, while he alone has knowledge of the movement of time.
Longtime followers of tax policy and the call for fundamental tax reform can easily empathize with Connors’ plight. The fundamental things continue to apply as time goes by. There is by now a no-longer surprising consensus in favor of certain basically well known (or easily knowable) principles: that the current tax system is too complicated and in need of some fundamental overhaul (President’s Advisory Panel on Federal Tax Reform, 2005); that any move should somehow be in the direction of consumption taxation, lowering if not eliminating the taxation of capital (Weisbach 2006; Auerbach 2006); that responsible tax policy means paying attention to incentive effects and maintaining low marginal tax rates (Diamond and Mirrlees 1971; Bankman and Griffith 1987) and minimizing transaction costs (Slemrod 1990; Weisbach 2000). These are discussions that have been going on for decades and by some measures centuries. The case for consumption taxation, after all, can be traced back to Thomas Hobbes (1660), Adam Smith (1776), and John Stuart Mill (1848), and has found currency among contemporary political philosophers such as John Rawls (1971) and Roberto Unger (2000). Proposals for broad-based cash-flow taxation, as in the failed Nunn- Domenici USA Tax of the mid 1990s (Seidman 1995; on current proposals, see McCaffery 2002, ch. 3) can be traced back at least to Nicolas Kaldor’s work on expenditure taxation (1955), which was in turn brought into the American tax policy mainstream by William Andrews (1972 and 1974). David Bradford and his colleagues in Treasury (1977) laid out the two great fault lines in comprehensive tax reform in the landmark Blueprints study—perfect or improve the income tax by broadening its base and lowering its rates (ideas that can be traced back at least as far as Stanley Surrey (Surrey et al. 1973; Surrey and McDaniel 1985 ), if not Henry Simons himself (1938)), or abandon the attempt to tax capital altogether and move to a systematic consumption tax along “postpaid” or cash-flow lines, as sketched out by Kaldor and Andrews. The basic analytic insight that prepaid (yield- exempt, wage-based) and postpaid (cash-flow, qualified account, sales-based) consumption taxes are equivalent under certain assumptions, mainly constant tax rates and rates of return (Andrews 1974; Warren 1975, 1980 and 1996; McCaffery
1992 and 2005a) led predictably enough to “flat-tax” proposals (Hall and Rabushka 1983), essentially prepaid consumption or wage taxes. Add on progressivity and get an X-Tax (Bradford 1986 and 1996), and most of the options now floating about—as for a national retail sales tax, a value-added tax (VAT), a “flat” tax without any capital taxation, a consistent cash-flow consumption tax— have lain before us for decades.
Proving perhaps that the less things change, the more they stay the same (with apologies to the wittier French subtitle of Auerbach, Gale and Orszag 2006), the recent Presidential Advisory Panel on Federal Tax Reform’s report (2005) basically tread in these same trenches, coming out, in the end, with an “expand the base/lower the rates” approach, along with a nod towards a consumption tax, that could easily have been written, but for a few specific details here and there, two or three decades ago. The papers in this volume, once again but for some admittedly important updating here and there—certainly the titles of the papers in this volume—could have featured in any of several tax policy conferences in the 1970s and 1980s, as reflected, for example, in several well known Brookings’ Institution publications (Pechman 1980 and 1984; Aaron and Galper 1985; Aaron, Galper and Pechman 1988).
All this consensus, and so little action; water, water everywhere, but not a drop to drink. The tax system is still a complex mess; the people are still unhappy; scholars are still debating and writing about the tax base (income versus consumption), transitions, implementation, and so forth, on and on. Yet little really changes: It’s still Groundhog Day. In hindsight, it seems clear that the then epochal Tax Reform Act of 1986 chose the wrong fork in the road (McCaffery 2005a) when it opted for non-fundamental, incremental reform under the existing “income” tax to anything more radical. In the subsequent two decades, the pace of tax law changes has dramatically accelerated (Doernberg and McChesney 1987; Steuerle 2005). The President’s Advisory Panel on Federal Tax Reform (2005) noted that there have been 15,000 changes to the tax code — two per day — since the 1986 act. The income tax is more detailed than ever, with features such as the capital gains preference, briefly eliminated by the 1986 act, back in full force. The people clamor for something—anything?—different, and yet practical politicians cannot see clear to recommend anything fundamental. Once again, the Presidential Advisory Panel — badly handicapped by its marching orders, (which included maintaining revenue and rough distributional neutrality, implementing a “fix” to the alternative minimum tax (AMT), and which also assumed the permanent extension of all recent tax cuts) — serves as Exhibit A. While repealing the AMT and the deduction for state and local taxes, (Internal Revenue Code (IRC) § 164), and substituting some low-rate credits for deductions may be good incremental ideas to clean up the income tax, they will hardly sate anyone’s appetite for fundamental reform.
What’s up? As Representative Rahm Emmanuel (D-Ill) (2004) said on the House floor, in regard to tort reform legislation, “We take up legislation that we have taken up before that is going nowhere and going nowhere fast. It is Groundhog Day here in this Congress” (quoted in McCaffery and Cohen, 2006, at 170).So too, it seems, with fundamental tax reform, for the tax policy community. Before this essay is through, I shall return to address this question, speculating that part of the problem is in our—the tax policy community’s—thinking about the problem. As Einstein once said (more or less), we cannot solve a problem by the same kind of thinking that led to the problem in the first place. Perfecting our economic and econometric models about labor and capital supply elasticities under alternative tax regimes, or sorting out the most nettlesome questions of transition from one hypothetical ideal tax system to another, are not what is needed to move on from the perpetual present of Groundhog Day. We need a sound, careful rethinking of the analytics of tax. The insights of behavioral economics, I shall argue, can be—and must be—enormously helpful in explaining and enacting fundamental tax reform. But first we have to escape from the clutches of “ad hocery,” where, I fear, behavioralism — as it exists in the service of tax policy and reform — is now situate.
III. The Usual Thought
Back to where we are, as a practical matter: We do not have a pure income or a pure consumption tax, but rather a hybrid income-consumption tax (Andrews 1974; Aaron, Pechman and Galper 1988; McCaffery 1992). This is largely because of the realization requirement occasioned by the 1920 Supreme Court case of Eisner v. Macomber. Although more particular hybrid features abound, the realization requirement alone, combined with the analytic fact that borrowing is not “income” under a pure income tax (because the positive consumption or savings represented by the use of the proceeds is offset by the liability representing the dissavings; in other words, there is no “net accession to wealth” when one borrows, (see McCaffery 2002 and 2005a), means that a significant amount of economic income, in the form of capital appreciation, escapes the tax base, even if the value is put to use as consumption. The “stepped-up basis” for assets passing on death, Internal Revenue Code (IRC) §1014, adds a third and decisive blow to the status of the income tax as a comprehensive ideal tax on all forms of material enhancement. The realization requirement, the nontaxation of debt, and the stepped-up basis rule lead to what I have called Tax Planning 101: the advice to buy/borrow/die (McCaffery 2002 and 2005a). By buying assets that appreciate without producing taxable cash flows, borrowing against the appreciation to finance present consumption (and/or wealth transfers to others), and dying with debt and appreciated assets in tow, the propertied classes can avoid all federal taxes.
Figure 1 U.S. personal savings rate, 2000-2005
Now whether self-consciously reflecting in this milieu, of the hybrid income-consumption tax, or considering the matter as if we had a pure, ideal income tax, a familiar thought comes to tax policy theorists. Savings isof course central to the intellectual task of fundamental tax reform, because the varying tax treatment of savings marks the distinctions between and among the major tax reform options, which all consist of income,prepaid or postpaid consumption taxes, or some combination thereof (McCaffery 2005b). Thus the topic of savings is featured in many of the papers in this volume. And savings or, rather, the lack of savings, is anenormous problem in the United States today; as Figure 1, based on data from the Bureau of Economic Advisors, shows, the personal savings rate (as a percent of disposable income), long drifting downwards, wentinto negative territory in 2005. Even with considerable questions about how to measure national savings, there seems little doubt that Americans save too little (Bernheim 2002, Moomau (this volume)). Since an income tax is designed, in theory, to double tax savings (Mill 1848), the perceived need for more savings is generally thought to lead to a systematic, fundamental conversion to a consumption tax (Bernheim 2002, Auerbach 2006). This first-best scenario typically assumes that individuals are rational life-cycle savers, who use their savings to solve a dynamic optimization problem, maximizing their utility across their lifespan by saving in order to smooth out consumption in each period. In the fully informed, “rational” setting, there seem to be few good reasons for ad hoc savings vehicles under the income tax— mechanisms of single taxation, superimposed on the income tax’s double-tax frame. It is a standard analytic result in the tax policy literature that, under the usual assumptions—mainly of constant tax rates and rates of return—“prepaid” consumption or wage taxes, where the single tax is levied upfront, and never again (such as the “Roth” Individual Retirement Account (IRA) treatment of current IRC § 408A), and “postpaid” or cash-flow consumption taxes, where the single tax is imposed later, on ultimate consumption or private preclusive use (such as for sales taxes, VATs, and “traditional” IRAs under current IRC § 408), lead to the same result (McCaffery 2005b).
Yet, analytically, there is no reason why a particular tax-favored vehicle using either mechanism should work to effect new savings under an ideal income or the existing income-with-realization tax, again assuming that the LCH holds. Prepaid accounts cannot prevent the movement of existing capital into the tax- favored accounts, simply giving a windfall bonus to existing accumulation while imposing transaction costs and theoretically reducing the need to save (on account of the windfall gain) (Bernheim 2002). Such accounts also afford little solace to mid-life wage earners, struggling to meet their household expenses from paycheck to paycheck, because they afford no cash-flow relief to help save (McCaffery 2005c). Further, now looking to the actual income-plus-realization tax, it is hard to see where prepaid savings accounts are much needed, given the ease with which capital investments can avoid further taxation under rather basic tax planning, such as buy/borrow/die (McCaffery 2002); the prepaid accounts simply seem to substitute one set of transactions costs (e.g., the rules for penalties on early withdrawal) for another set (e.g., the tendency to remain in non-interest or non- dividend-bearing securities, and the associated lack of diversification in a buy/borrow/die portfolio). Meanwhile, postpaid accounts cannot promise any net new savings at all, given the ease with which taxpayers can arbitrage out of them, simply by dissaving, or borrowing, at the same time as investing in the tax- favored accounts, thus generating a present deduction with no net savings (McCaffery 2005a and b). Note that this arbitrage operation is possible with or without an interest deduction for the debt (compare Auerbach 2006), which deduction was curtailed by the Tax Reform Act of 1986 in IRC § 163 (see especially IRC § 163(d)); the arbitrage comes about because the borrowing of the principal is not income, and yet its use in funding the tax-favored account triggers an immediate deduction. (Interest deductibility, where it obtains, only sweetens the deal for the taxpayer). More on this, anon.
Under these analytic facts, there is no reason why a rational life-cycle saver—who, by definition, would be optimizing her savings under the existing tax structure—should increase her savings in the presence of tax-favored vehicles. Such vehicles only at best increase the rate of return to savings, by reducing the effective taxation of the yield to capital, leading most likely to less savings under the dynamic optimization model (ignoring here possible effects on labor supply); or they do little at all, because the taxpayer is already (rationally, by stipulation) not saving; or, on yet another hand, for those taxpayers saving without bearing any effective tax on their savings (as by taking advantage of buy/borrow/die), the tax-favored vehicles do little, such as making it less transaction-costly to avoid the income tax’s double-tax sting, or nothing. If all taxpayers were fully rational life- cycle savers, in other words, the case for ad hoc tax-favored savings vehicles would be very weak, indeed—these provisions would impose complexity and transaction costs without evident gain—and policy theorists could return to the more global, welfare economic analysis of general consumption versus income taxation: fundamental, not ad hoc, tax reform (see for example Bernheim 2002).
Given these analytics, and a continued unwillingness of tax reform—in the Tax Reform Act of 1986 or otherwise—to address any of the basic planks in the buy/borrow/die tax advice (McCaffery 2005a), the case for ad hoc, targeted savings vehicles must therefore turn on a rejection of the LCH in favor of a BLCH or some other behavioral alternative. And so, in theory, advocates of these particular pro-savings provisions point to the possibility that myopic individuals, lacking self control, might focus on the tax-favored savings vehicles to increase their savings (Bernheim 2002, and Auerbach 2006, each noting the approach without specifically endorsing it). This then leads to the possibility—a possibility lacking if a strict version of the LCH were to hold—that the tax-favored provisions are “working,” that is, that they are increasing savings. Applied public finance theorists have rushed in to analyze the data to settled the perceived debate.
Unfortunately, perhaps, the econometric challenges involved in ascertaining whether or not ad hoc tax-favored savings vehicles “work” have been formidable, and the results, best understood, have been inconclusive. Studying the comparative statics of the introduction of IRAs in 1974, for example, Feenberg and Skinner (1989) and Venti and Wise (1986, 1992) found that IRAs did indeed lead to new savings, sometimes significantly so. But other research has looked at different numbers and reached different conclusions. Engen and Gale (1997) find for example that any increases in savings under 401(k) plans, largely a creature of the 1980s, were offset by higher mortgage borrowing and correspondingly lower home equity, a result that suggests a simple arbitrage operation was taking place, although Bernheim (1997) questions this result. What seems unquestionable is that the plethora of tax-favored provisions in the tax code is costly, under a static revenue analysis, and that our national savings is low and falling, notwithstanding decades of these creatures being out and about in the tax code. A recent Urban Institute study found —just to drive the point home — that the annual revenue loss from tax-favored savings provisions was greater in 2004 than the annual aggregate increase in private savings in that year (Bell, Carrasso, and Steurle 2004), suggesting, in a crude nutshell, that the Treasury is subsidizing taxpayers to consume more today by means of tax-favored savings vehicles.
Once again, I suspect that the econometrics will never yield decisive answers, although we could put the whole question of the efficacy of ad hoc tax- favored savings accounts in the “unknown but knowable” category. The real world is a messy laboratory, and the mechanisms for dissaving and moving around old savings are subtle and complex to track. Yet there are good analytic reasons to believe, as the next section presses, that the utility of ad hoc savings provisions is very limited—reasons that do not rest on accepting the LCH and rejecting a BLCH, in toto. Far from it: the lessons of behavioral economics, best understood, suggest that ad hoc tax favored savings vehicles should not work. But none of this has stopped actual tax policy from piling on one tax-favored savings provision after another. Tax-favored pension plans started in the 1940s; traditional IRAs were added in the 1970s; 401(k) plans were enacted in 1978 and began to take off in popularity in 1981; “Roth” IRAs were added in the 1990s; today we have Roth 401(k)s, medical savings accounts, Section 529 Qualified Tuition plans, and more. It is a little bit odd, and disconcerting, that while behavioralism still has to fight fierce battles to justify its very relevance inside the academy (compare Rabin 2002 with Epstein 2006a and b; Glaeser 2006), practical politicians and applied public finance theorists have been off and running with reforms that can only be justified if a particularly simplistic and naïve version of behavioralism were pervasively and specifically true. It is not.
IV. Critique of the Usual Approach.
A. The Known and the Knowable
From these basic principles, and drawing on what is known, we can construct a critique of the typical approach to engrafting behavioral economics onto fundamental tax reform. First, from the world of tax and economics, as traditionally practiced, abetted by common sense, here is what we know:
- Americans by and large do not save.
- An income tax is designed to tax savings, and thus favors present over deferred consumption.
- Since the 1940s, the United States tax system has engrafted pro- savings vehicles onto its income tax system, hoping to facilitate and even encourage savings, a trend that has accelerated dramatically in the last two to three decades, to where we now have a panoply of tax-favored savings vehicles in the Internal Revenue Code, for retirement, educational, lifetime savings and health accounts.
- These pro-savings vehicles within the Code are built on one of the two basic models for consumption taxation, namely postpaid (no tax now, tax later) or prepaid (tax now, no tax later).
- Analytically, there is no reason for postpaid accounts to lead to enduring new savings, because it is trivial to arbitrage against the savings by “saving” with one hand, and dissaving, or borrowing, with the other.
- Analytically, there is also no reason for the prepaid savings accounts to lead to new savings, either, because these vehicles offer no alleviation from the cash-flow stresses of the working classes, since they give no present tax (or other economic) benefit, and they also have no enforceable mechanisms to prevent old savings from simply being transferred into the new skins. Further, fairly basic tax planning under the current, income-with-realization tax can lead to this same prepaid consumption tax treatment (McCaffery 2002). And prepaid accounts have no mechanisms to prevent or deter dissaving, or spending down accumulated wealth.
- What does behavioral economics—also enlightened common sense—add to this knowledge?
- Most people, most of the time, are myopic, paying greater attention to the present, and lacking the self control to behave consistently with the solution to a dynamic, long run optimization problem (Thaler and Sheffrin 1981; Thaler 1994).
- People can focus on certain salient clues, are loss averse, and generally resistant to change (Tversky and Kahneman 1986; Kahneman, Knetsch and Thaler 1991; McCaffery and Baron 2006).
From these facts, the usual approach gets constructed. The two behavioral biases are set against each other. Let us make savings plans salient in the income tax, the thinking goes, to get people to act in their long-range self interest, by saving more, notwithstanding their myopia, and then hope that inertia (the “status quo bias”) will take hold, leading to enduring individual savings. This is a happy tale, to be sure. But a mere modicum of deeper, more careful thinking reveals it to be little more than a “just so” story, extremely unlikely to hold in the real world — and certain to lead to incremental, unhelpful, complexifying, and likely counter- productive tax reform.
B. Why Care about Savings?
We ought first to pause and reflect over just why, exactly, “we” want more savings. There are two broad sets of reasons. One we can call “micro,” looking to behavior at the individual level: People do not save enough, on their own lights, to maximize their own utilities, that is, they are not rational life-cycle savers. Two we can call “macro,” looking to the collective: Our aggregate capital stock is too small to maximize collective social welfare. It is common among at least some traditional welfare economists to assert that only the micro question has normative force: that, by definition, the optimal capital stock is what rational individuals on their own would choose, summed up. Under this view, the only way we can have “too little” savings—the only way for that concept to have meaning—is if some people are irrational, or “behavioral,” that is, wrong on their own lights. The macro is the micro summed, no more and no less.
This is puzzling. For one thing, the aggregate capital stock clearly has features of a classic public good: the total capital of society, which affects interest rates and the returns to labor and so forth, is non-rivalrous and non-exclusive. Under plausible conceptions of the social production function, labor — that is, workers, who, by and large, the non-savers are — would benefit from more social savings. It is a quite sensible social policy to help the poor and lower economic classes, not by tricking them into saving more, themselves, but rather by allowing the not-poor to save more, more easily (McCaffery 2002). Social theorists (e.g., Sen 1961; Rawls 1971) have long recognized, too, that the capital stock affects third parties not present at the micro-level decisions, such as the unborn: another reason why the problem of savings may not easily reduce to the sum of all micro problems. Certainly, the curiously non-parallel literature on the government’s budget deficit seems to presume that the macro level capital stock is the problem, if there is a problem. (Indeed, in a strict or “hyper” rational model, government deficits themselves would not matter, because rational self-interested agents would counteract any deficit in their own private savings decisions (Barro, 1974), and, since these individuals would be rational savers, too, there would be no problem at all. But that possibility has not stopped policymakers and applied public finance theorists from endlessly analyzing the government’s dissavings “problem,” in a story for another Groundhog’s Day.)
Even if there were indeed nothing to talk about except the micro foundations for the optimal social capital stock, in an ideal or first-best setting, the current status quo is far from ideal, with tax and other social economic policies distorting individual decisions. I shall give some more content to this thought below. For now, note that it is a heroic leap to conclude that there is no problem with savings at all, except for the myopia of some (many?), such that, if only we can get individuals to be better utility-maximizers on their own lights, we would have the “right” level of aggregate capital accumulation. And, finally, even if this were true — that the only “problem” lies in the irrationality of some (many? most?) — then, unless we had some confidence that we could cure the myopia of all, the social planner might still have to look to the aggregate capital stock: we would need the “over” savings (by stipulation) of some to compensate for the “under” savings of others (most?).
All of this suggests that there are, indeed, reasons for tax and fiscal policy to look to the macro level as well as the micro foundations of savings. The good news is that both views might lead to the same place — a place different from where we are heading, full speed forward, now.
C. Some Ground-rules
The rush towards ad hoc tax-favored savings plans has been furthered along by an ad hoc adoption of ad hoc findings from behavioral economics. Too much ad hocery, all in all. In general, I share Matthew Rabin’s (2002, at 659) thought that: “As a rule, it is bad to spend time on ‘methodological’ and broad- stroke issues rather than the nitty gritty of the phenomenon being studied. The goal of this research program [of psychology and economics] is that it become ‘normal science,’ and, as such, the nitty gritty is the point.” But rules have exceptions, and it is clear that the hoped-for marriage at the core of this essay — of behavioral economics and fundamental tax reform — is stumbling on the usual thing that marriages stumble over, a failure of communication and to fully understand each other. It rewards practical policy analysts to step back and reflect on some core analytic principles of behavioral economics. As at least a fellow- traveler with many fine behavioralists, I offer four ground rules of relevance to the task at hand, foundations for a possible happy and long term marriage.
One, theory matters. There is nothing in behavioralism to suggest any rejection of any facts of the matter. (Rabin 2002). The analytics of the tax system discussed a bove are facts of the matter. It is possible, given the fungibility of money/value, to move old savings into new prepaid tax-favored savings accounts. It is possible to arbitrage against postpaid tax-favored savings account by dissaving, that is, running up debts, or not saving as much as one otherwise would but for the tax-favored account, and so on. These facts of the matter must be taken into account in any sound analysis of tax reform, incremental or fundamental. They do not somehow disappear or lose their status as facts of the matter because individuals make systematic mistakes in their thinking about them or anything else.
Two, not all people are the same. It is a very large conceit of much social scientific policy writing — perhaps a vestige of the representative agent models common in graduate school training — that all people are identical (McCaffery 1994b, 2000). Much of the literature on the LCH and BLCH has an all-or-nothing quality, as if people must either by rational dynamic optimizers or hopelessly lacking in self-control; it reminds one of the futile reductionism of the pre- Socratics, debating whether there are one, two, three or four primal substances from which all else follows (such as earth, water, air, and fire). Clearly people are different. Savings have multiple and mixed motives. Some people save for life- cycle reasons, some for precautionary purposes; some of these are rational, some of these are irrational, as in being excessively risk averse and failing to take advantage of financial market vehicles such as annuities and insurance products. Some people save to make bequests; others make bequests only because they failed to annuitize their wealth. Some people save because they make more than they can think to spend; some people save because they think it is moral or right or just to do so. In thinking about fundamental tax reform — and whether we care about just the micro- or the macro-level capital questions — we have to think through the significance of this heterogeneity of people and savings behaviors. Similar conceits are that people act with single utility functions, and that theories must be all-right or all-wrong. There are no such meta-rules, certainly not coming out of “behavioralism,” per se.
Three, the widespread prevalence of cognitive heuristics and biases, central to behavioral economics, does not mean that people are stupid and helpless, or that behavioralists think that they are, or that “paternalism” in one or another of its dreaded garbs is warranted. It is always best, I believe, to treat people with respect. Ordinary citizens are busy, struggling to make do and juggle competing tasks and demands in a complex world. There is plenty of evidence that all of us mortals make “mistakes” in judgment and decision-making; there is very little evidence that these mistakes are devastating in their impacts, or keep us from living happy, meaningful, even flourishing lives. There are at least two reasons for this. One, the various rules of thumb that we employ are often, perhaps even usually, right. These heuristics can sometimes lead us astray, but our failure to counteract the biases does not mean that we are simply stupid and helpless. Further, not all actions that can be explained by “irrational” explanations are, in fact, irrational. Good social policy theory should always begin and end with good careful and respectful — thinking about the people to be affected by any policy.
Take savings behavior. The usual thought about the subject before us relies on a simple syllogism:
- People should save, for their own good,
- People do not save,
- Therefore people are irrational.
But this does not follow, logically. There are reasons that people might not save, having to do with the general structure of society and its institutions, and having nothing to do with irrationality — there is, or can be, “rational myopia,” as it were. For example, it may not be a rational strategy for a lower-income worker — and lower-income workers are especially unlikely to save, as Figure 2, from Kawachi, Smith and Toder (2006), showing the percent of qualified workers maximally investing in tax-favored
Figure 2: Percent of workers contributing maximum to 401(k) plans by earnings, constant 2004 dollars. Source: Kawachi, Smith and Toder 2006 401(k) plans by earnings level, illustrates — to save small sums of money over an extended period of time, in order to build up a modest nest egg. For example, $50 a month saved for 10 years at even a 6% interest(high for a real rate of return), compounded monthly, yields an amount considerably less than $10,000 by the end of the decade. Does it help to have $8,000 in the bank?
Such a relatively meager accumulation might only make one disqualified for financial or government aid programs, such as educational or medical benefits, as well as making one vulnerable to the wants or demands of family and other potential claimants. It is hard to afford good financial advice at such low wealth levels. In this social and institutional setting, I have argued (McCaffery 1994b), it can be fully rational to play thelottery — as the vast majority of U.S. citizens in fact do — notwithstanding the fact that lotteries are a wildly unfair actuarial bet. Whether we conceptualize this as “episodic risk preference” or whatever, the fact ofthe matter is that having a small nest egg may not mean much, and may even impose costs, such that the lower classes “invest” in a chance at a transformative wealth event: there is a kink in the social utility function,as it were. Yet another rational reason not to save, even under an LCH, is the thought or hope that things will get better tomorrow: a better job, a higher earning spouse or partner or child, more generous publicbenefits. These hopes may get dashed, more often than not, but they are not irrational — and they may even be, instrumentally, vitally important to have, for a host of reasons relating to health and well being.(Compare Loewenstein, Small and Strnad, 2006, discussing the instrumental value of the “identifiable victim bias” in public policy.)
This is not the same as the parlor game of converting all irrational behavior into a rational explanation by playing with semantics. To the contrary, it is first considering, and ruling out, rational explanations beforeconsidering irrational ones, looking to the people with respect. To me and most behavioralists I know, “irrationality” has a precise meaning, involving the violation of one of the basic axioms of rationality, such asconsistency, reflexivity, or transitivity (continuity is a bit more complex) (McCaffery and Baron 2006). If a person likes her glass half- full but rejects it half-empty, or if another person uses cash to avoid a penalty on credit cards but uses credit cards and forswears a bonus for using cash, something is up. But in the subject matter before us — savings — it is not obvious that “myopia,” or present-bias, which is easy to elicit inexperimental settings, is what is up. And even if it is so for some people, it need not be so for all, as Ground Rule Two had held. For a good many Americans, savings is simply not a viable option, and it would be bestfor social policy theorists to simply understand and accept this, as I shall discuss further in the next Part. And note, finally, that this “rational” set of explanations fits with the data — that we have and have had for decade pro-savings policies, but our savings rate is low and declining — in a way that the easy “behavioral” explanation do not.
Four, it is important to distinguish, in thinking about cognitive errors, between the cognitive and the experiential “spaces” of human endeavor. Danny Kahneman (2000) has of late been using a distinction between“decision utility,” or the weights people put on matters in forming a judgment or decision, and “experienced utility,” which is how people feel about the experience that their decision led to, during or after thedecision. Because there is a gap, Kahneman has been recommending paternalism of one sort or another. This is not exactly what I have in mind. The decisions that we care about lead to actions. But the time anddistance of the action from the decision, and the duration of the action’s effects, can vary. Sometimes, the decision alone is almost performative, as in voting, or forming an opinion to express to a public opinionpollster: the thought is the action. Other times, the thought leads to an action that will or can endure for a long time, out in the world. Of course, savings is a paradigm example of the latter: an action, or set of actionsthat, by definition, has consequences that endure through time. And so we ought to distinguish, as social policy theorists, between how we think about savings, and how we in fact, over time, save.
The reason for laying out and stressing this ground rule relates to the second reason why widespread cognitive errors have limited real-world harms, most of the time (the first was that these heuristics or rules ofthumb were often correct, or adequate to the tasks at hand). The real world of experiences is set up in the presence of people’s ordinary cognitive skills and shortcomings. Very often, in private financial markets,institutions arbitrage against and effectively “cure” the cognitive errors (McCaffery and Baron 2006). Thus competition and the market assure marginal cost pricing, even if few if any of us would be capable of discerning what anything should really cost. We can all venture into the supermarket and act as if we were fully rational agents. But this is not always so. The careful theorist should pay attention to society’s institutional structures to see when these counteract — and when they exacerbate — persistent, individual level cognitive bias. I shall have more to say on this point soon enough.
D. What’s Wrong with the Usual Approach
Putting the ground rules together with the facts of the matter leads to a deep critique of the trend in tax policy over the last several decades, towards ad hoc tax-favored savings vehicles. In essence, there is almost no reason why anyone should think these vehicles would “work,” which is why none ought to be surprised that the macro-level data typically suggests they are not in fact working.
Consider, first, the micro-level policy objective, to get individuals to save more, on their own lights, for their own good. Let us accept that some, perhaps many, even most, individuals are myopic, lacking in selfcontrol. Since we are only talking about savings, this means that they do not save enough for their own good. Now let’s look at the two broad policy options for ad hoc tax-favored savings vehicles.
Consider first prepaid accounts, which have been the dominant recent trend in tax policy (McCaffery 2005a). These accounts offer no solace to the myopic, cash-flow-constrained individual, because they offer noimmediate benefit for savings. Indeed, the reasoning for putting money into these accounts must be somewhat — perhaps even foolishly — far-sighted: One has to think ahead, to whether or not the tax advantagesof the Roth-style accounts outweigh their disadvantages. One ought even to reflect on whether or not Congress will renew or extend the tax-favored treatment; Section 529 qualified tuition plans, for example, are setto expire, with no automatic “grandfathering” of existing accounts on the books. Should one transfer wealth into a 529 plan, and should Congress decide not to renew or extend the “favorable” tax treatment, theinvestor will be (double) taxed, having given up an opportunity, afforded by present law, to avoid that second tax on one’s own. There is nothing very helpful to the myopic, here.
It is no surprise, then, that most commentators going down the path of considering tax-favored accounts as a cure for rampant myopia are thinking of postpaid accounts (Thaler 1994; Auerbach 2006; Bernheim 2002).But the tale is no better here. For one thing, for the fully “rational” myopic, postpaid accounts create a mechanism for more consumption today by greater dissaving, the simple arbitrage operation mentioned above.Imagine a taxpayer, Jill, in the 35% marginal tax bracket. Jill can borrow $2000 on a credit card on April 15, and open a tax-deductible, traditional IRA. There is no net savings here, the $2000 IRA account being offset bythe $2000 debt (which must be repaid, perhaps at a very high rate with non-deductible interest). But Jill has, immediately, gained $700 more to consume, today, by lowering her taxes, due by the stroke of midnight. A more prevalent problem may come from the “naively” myopic individual, Jack. Because of a persistent money illusion (Shafir, Diamond and Tversky 1997), or whatever, Jack simply forgets that taxes must be paid outof his traditional IRA or 401(k) plan. So the balance Jack sees in his monthly or quarterly statement leads him to a false sense of security, and he consumes even more, running up credit card or home equity debt, andso on.
Here is the deep, dark problem in all this: These “arbitrage” operations do not depend on exquisite calculations or the solving of complex, long term dynamic optimization problems. They depend, rather, on . . .myopia! That is, the very same cognitive or behavioral tendency on which the ad hoc “patch” is justified is the tendency that can or will undo the efficacy of the patch. The theory of postpaid tax- favored savingsaccounts within an income tax structure, in other words, is like the strategy of pouring water on a drowning man.
Now in the face of this critique, Professor Strand, drawing on the work of Bernheim and Rangel (2005a and b), suggests a counter-possibility. Perhaps some individuals are generally “rational” but suffer from “hotflashes,” so to speak, during which they temporarily lose control and would, if they could, spend everything they had. Such episodic myopics, as we might call them, would be benefited by tax-favored savingsaccounts with built-in liquidity constraints. Now I have little doubt, consistent with my general theme that not all people are the same (which Professor Strnad calls my “heterogeneity” point), that there are indeedpeople like this: some of my best friends are, as the saying goes. But so what? It cannot be the right approach to public policy to put in place a byzantine system and then, ex post, construct a personality profile forwhom the system maximizes individual welfare. In any event, the solution to the Strnad-Bernheim- Rangel “problem” is to create savings vehicles with liquidity constraints, with or without tax preferences: ChristmasClubs, for example, would do the trick perfectly well. The tax system need not be constructing Ulysses’ mast for him.
Further, because savings take place in the real world of experiences, and, by definition, must endure over and through time, the actual institutions of society will interface with the individual decisions (Ground RuleFour). This fact just makes things worse. Contrary to the general tendency of private markets to arbitrage against individual heuristics and biases and effect better, more efficient results, such as marginal cost pricing(McCaffery and Baron 2006), the institutions of society are perfectly content to encourage individual myopia. Financiers will not cease to send out invitations for new credit cards, and retailers will not cease to offer“no money down” financing, simply because Congress has decided to engraft ad hoc tax-favored savings vehicles onto an income tax structure. On the other hand, there is no reason to doubt that private markets would generate the kind of self-imposed liquidity constraints that would solve the Strnad-Bernheim- Rangel problem of episodic myopia, discussed above.
Consider next the macro-level concern, about the aggregate capital stock. Here again the ad hoc approach scores poorly. Looking first to prepaid savings accounts, these not only give no reason for any new savings,they also eliminate almost all barriers to dissaving. The wealthy, having initially benefited from the windfall gain to existing capitalists (and thereby perhaps reducing their savings, under the LCH), now face no barrierwhatsoever to spend away, tax-free. Note that they need not even wait until they turn 59.5 years of age, or whatever the statutory minimum age is for tax-free withdrawal, because they can borrow today, tax-free,and wait to pay off the debt until they hit the magic number. Since myopic individuals, who do not now save, are given no new reason or ability to save, and since the present savers, who are (almost by definition) lessmyopic, are given a windfall, a reason to save less, and an easier path to dissavings, there is little reason to expect greater social capital under prevalent ad hoc prepaid savings accounts.
Postpaid accounts fare no better. These offer no strong reason to save, as we have seen, and even encourage, both on account of rational myopia, and irrational money illusion, greater dissavings today. Once again,the myopic non- saver today is helped mainly in her present bias. Another, and potentially very large problem, with the postpaid savings account technique is that they compel the withdrawal of funds from theaccount, both by minimum distribution rules and by the potentially extreme taxation of the accounts on death (because they are subject to both “income tax in respect of decedent” rules, IRC § 691, and the estate tax,a combined level of 75% or more of their principal balance). But if most savings comes from rational or irrational long-term, intergenerational savers (Kotlkikoff and Summers 1981), these accounts can have a perverse effect on the aggregatecapital stock. By essentially forcing people into a “life-cycle” model, postpaid tax-favored savings accounts may be encouraging the dissipation of America’s best, most stable, and longest-term pools of private capital.
E. Summing Up: The Known and Knowable, Redux
A brief word from the author: My initial professional training is as a tax lawyer, and, while I added an economics degree after commencing my position as a legal academic, I lack the skills or resources to perform an in-depth, data- intensive, econometric analysis of the efficacy of tax-favored savings vehicles. But perhaps this is a blessing in disguise. Rather than “crunch the numbers,” myself, I have studied the work of others, and Ihave reflected on what I do know. I now know much about behavioral economics, and I have long been toiling in the terrain of fundamental tax policy, on a legal and analytic level. I have also tried, against alltemptations to the contrary, to maintain a modicum of common sense. Here, then, is what I consider to be known and/or knowable about the present state of affairs: One, Americans do not save. And we have notsaved any more since the 1940s, when a policy of hybridization to encourage savings was born in the income tax, with tax-favored pension plans. Nor has the situation improved since the 1970s, when we added IRAs tothe IRC; or the 1980s, when 401(k) plans took off; or the 1990s, when we added Roth IRAs; or the present decade, when we have seen the birth of medical and educational and lifetime savings accounts and Roth 401(k)sandonandon.
Two, these savings provisions in fact cost more in forgone tax revenue than we observe in new savings each year (Bell, Carraso and Steuerle 2005). Three, there is no reason, with or without the insights of behavioral economics, that this should be anything but as it appears.
V. A New Marriage: Escape from Groundhog Day
The cause of fundamental tax reform seems stuck in a perpetual present, endlessly debating the same old, same old questions of income versus consumption taxation, implementation, transitional rules, and so on.Savings behavior — and Americans’ lack of it, by and large — features prominently in this debate. Because, in a certain limiting case, it is hard to see any problem, by definition, if all people are fully rational life-cycle savers, tax policy theorists have looked to forge a marriage with behavioral economics, taking off-the-shelf findings that people are myopic and lack self-control, but can be tricked into changing their behavior — andthus, one hopes, saving — by salient vehicles. Hence tax policy has recommended engrafting ad hoc tax-favored savings vehicles onto the income tax. Like a football coach running the same play over and over until itno longer works, one thing has led to another, and we now have dozens of pro-savings vehicles in the tax code, with more on various drawing boards inside the Beltway. Only, unlike the football coach, there isprecious little evidence that the vehicles do work, and no reason, in theory, to think that they should. And so we see the past is what is keeping us in the present: we repeat the same techniques, and go down thesame intellectual paths, while the law gets more and more complex and the people seethe with anger and frustration for reform. It’s Groundhog Day, forever.
In this penultimate Part, I sketch out the outlines of a happier marriage, one that can get us out of the rut of the present, towards a better, more principled, future.
A. Fundamental Tax Reform
The analysis to this point leads to the surprisingly simple thought that fundamental tax reform should be fundamental. Reviewing the ground rules, recall that theory matters; ordinary people are not all the same butthey are not generally stupid, foolish, or helpless; and life plays itself out in the realm of experiences, where institutions interact with individual behaviors. The problem with ad hoc tax-favored savings vehicles, seenin this light, is that these can only work if all or most people are (1) the same, (2) myopic but easily distractable by salient cues to act in their self-interest, (3) persistently short-sighted enough not to learn thetechniques of arbitrage, so as to unravel the effects of the myopia-curing actions, and if (4) institutions in society do not in fact rise up to counteract the effects of the salient myopia cures, by teaching the people how to arbitrage (or just doing it for them). There is good reason to conclude that none of these four conditions obtain; certainly it is hard to conclude that all of them do. We can look to the real world and see, on the onehand, the plethora of tax-favored savings vehicles and, on the other hand, much evidence of low and declining personal savings rates.
In the face of this dire state of the world, there is an obvious strategy to try. We could tighten up the tax-favored savings accounts themselves, layering them with rules making them non-fungible, preventing easy arbitrage: keep them out of bankruptcy, do not allow them to be pledged as security for loans, have mandatory anti-withdrawal rules backed by penalties, and so on. But there are reasons to think that suchtechniques, like “spendthrift” or anti-alienation clauses in trusts, simply drive up the costs of any countervailing arbitrage mechanism. Certainly such rules, in themselves, do not change the fact of the matter that myopic individuals will want to arbitrage, or that social institutions, such as credit and retail markets, have every incentive to try to get individuals so to arbitrage. Patching up a Maginot Line is not far-sighted socialpolicy.
This then leads to the thought of a Copernican revolution, in which we invert our gaze, and look not to the tax-favored savings vehicles themselves, but to the residual, default tax system. If we were to convert thatsystem to a consistent, postpaid or cash-flow consumption tax, arbitrage would not be possible (McCaffery 2005c). Savings would not be taxable, but debt used to consume would trigger tax (Seidman, 1995; McCaffery 2002, 2005a and c). Hence there would be no reason to save on the one hand and borrow or dissave on the other: this would not generate any additional funds for immediate consumption. There would now be no tax-centric reason for particular, ad hoc tax-favored savings accounts, but of course that is true today, under the status quo, as discussed above. (Note, however, that it is a mistake to think that there are no reasons for special savings accounts given comprehensive cash-flow consumption taxation; I pick up this theme in the following section). Finally, a consistent, comprehensive conversion to postpaid consumption taxation — and Ibelieve this ought to be done without a separate, freestanding gift and estate tax (McCaffery 2002, ch. 4) — would do something more: it would make it easier for savers to save, for as much and as long as they want.None of this is to say that fundamental tax reform will be easy or seamless: transitions are hard, economically and politically; implementation details can be devilish; and opportunities for evasion arise under any tax system.
These recommendations follow from bearing in mind that theory matters, and that people are different. Tricking myopic non-savers into savings in a world in which there is no rational reason for the tricks to work isnot a sensible blueprint for reform. Changing the world to make the rules systematically pro- savings (or not anti-savings) does make sense. And it lays a better foundation for addressing the “problem” of savings, onboth micro- and macro-levels.
The insights of behavioral economics do not suggest or justify the use of ad hoc tax-favored savings within a basic income tax structure to address the problem of savings. Far from it: there is good reason, even andmaybe especially accepting that many and perhaps most of us are myopic, to suspect that such vehicles will not work. And while it may not yet be quite “known,” it seems pretty close to being “knowable” that thisstrategy has not worked. It is better to pay attention to the lessons of theory, and implement, probably sooner rather than later, a consistent, comprehensive consumption tax. What then to do about the problem ofsavings?
First, on a macro-level, the analysis above has suggested that most Americans do not and will not save. But some Americans clearly save a lot. There is much to be said for letting them do so. This is not to agree withJeff Strnad’s comment about a “barbell” shaped distribution, with some over-savers, some under-savers, and a vast middle. I have no reason to believe that the distribution will not be continuous, and some reason tohope that a sensible tax policy will inculcate – and encourage the private market to develop mechanisms to inculcate “better” approaches to savings. But I am certainly skeptical that tax policy can be any kind ofsimple, fair, or efficient mechanism to prop up undersaving, with its ill-thought through mechanism for the myopic, or to clamp down on oversaving, with its equally ad hoc mandatory distribution rules and so forth. A consistent, progressive, postpaid consumption-without-estate tax is an attractive vehicle for just such ends. This tax system allows and even encourages people to build up large stores of nominally “private” wealth.These tax-favored accounts can be monitored or regulated, however loosely, much as IRAs, 401(k)s, pension plans, and other large pools of “private” savings (e.g., the endowments of nonprofits) are today. Anywithdrawal from the accounts to finance private preclusive use will bear the brunt of progressive tax rates. Nor will this system let second and third generation heirs off the social hook, as the current, porous, income-with-realization-with-estate tax all too often does. To the contrary, the heirs will bear a tax when and as they withdraw “their” savings to consume. The system works like a global carryover basis regime, where thebasis of all savings and investments is zero. Professor Strnad is right to fret that some future government might view the tax-favored accounts as an object of confiscation, but this is no more true for generally tax-favored savings accounts than for the ad hoc tax-favored savings accounts we have to day. Whatever America’s failings, we seem to have found a way to remain credibly committed to not simply taking the private wealth of the most privileged few.
On the micro-level, such fundamental reform will let life-cycle savers be life-cycle savers and, as just suggested, will facilitate if not encourage bequest and precautionary savings. What of the myopic masses? Here,another mistake of the flawed early marriage of behavioral economics and fundamental tax reform is the conclusion that, if we adopt fundamental tax reform in lieu of ad hoc tax-favored savings vehicles, there is noremaining role for the lessons of behavioral economics. This is wrong. There is no reason why the ideas of Thaler and Bernatzi (2004), in their Save More Tomorrow™ plan, or of Choi, Laibson, Madrian and Metrick(2006), in their idea that the default should be set for workers to be “in” their employer-provided pension plans, should not take hold in a world of comprehensive postpaid consumption taxation. To the contrary: the argument above has led to the conclusion that these ideas can only work, in any enduring fashion, in such a world, where arbitrage against them bears a tax price. It is another characteristic of our biases as a tax policycommunity that we can only think of tax-based inducements, as if a man with a hammer could only use nails. Employers can still give matching funds, and the government can, too, to encourage private savings in aworld fundamentally set up not to discourage it.
Let a thousand flowers bloom, and let the market try to create “focal” and salient savings plans, in a setting where theory supports the potential practical efficacy of just such ideas.
The facts also suggest another approach to savings. I strongly suspect, at least given current cultural attitudes and mores, and the structures of our capitalist, consumerist marketplace, that, whatever the government does via fiscal policy, most Americans will not save, period. We can and ought to look at the full range of social policies — not just tax — that have led to this being the case. What is unknown and almost certainlyunknowable is what kind of people we would be, today, if we had not gone down the path we did in the first place. But there is in any event much to be said for accepting the facts as the facts. Letting those people soinclined to “over save” to do so seems one perfectly sensible response to this brute fact of the matter. Creating and providing non-fungible and secure government benefits, such as Medicare, old age and disability insurance, education, and so on, seems another. Of course, getting the government to save more, or dissave less, may be part of the answer, too, but let us leave that one for another Groundhog Day.
C. Behavioral Economics, Redux
I promised pages ago that I would come back and discuss how it is that fundamental tax reform and behavioral economics might have a stable, long term, productive marriage. Banishing ad-hocery from the union is animportant first step. I have argued to replace the ever-growing panoply of ad hoc tax-favored savings vehicles in an income tax structure with a consistent, fundamental conversion to a postpaid consumption tax,allowing the market or even the government to use salient savings vehicles within such a consistent theoretical structure. This advice follows from — it does not contradict — the lessons of behavioral economics andcommon sense. So, too, with the advice that a better, less fungible social safety net, wherein the society provides in-kind goods to those suffering from want of private funds, makes more sense than trying to trick people into savings in a world generally set up both to discourage (because of the tax system) and to require (because of the absence of more generous in-kind support) private savings.
There is another domain in which behavioral economics is relevant to fundamental tax reform. The topic to this point, of individual savings, has concerned a set of actions that by definition endure through time. Thusthe possibility — I have suggested the likelihood — of individuals coming to arbitrage against the tax-favored savings accounts, with an institutional, market-driven economy set up to help them do just that. But recallthe ground rule about some actions staying pretty close to the decisional bone. Sometimes, what is important is precisely what the people think about something. So it is, I believe, and have worked extensively totest and prove experimentally (see generally McCaffery and Baron 2006), with fundamental tax reform. It is a platitude that fundamental tax reform cannot occur without broad popular support and understanding,typically coalesced by presidential “leadership.” But how can the people, preoccupied by the demands of daily life, understand the lessons of public finance, such as optimal taxation, corporate tax incidence, and soon? And why should they try? The benefits to greater knowledge are miniscule, the costs significant. And yet it is far from clear that politicians have a properly aligned incentive here to welfare maximize: think of asingle politician suggesting abolition of the corporate tax, as an example. In private markets, arbitrage against behavioral anomalies is a private good; in public markets, it is a public good, predictably undersupplied (McCaffery and Baron 2006).
Behavioral economics can and I believe should teach those of us interested in fundamental tax reform how and why ordinary citizens think the way that they do about matters of public finance. Behavioral insights canalso help us think of mechanisms that might tend towards the adoption and widespread support for policies “we” know, from our expert vantage points, to be good for us all. These are important roles for a disciplineto play, more enduring than helping to craft ad hoc and counterproductive savings plans.
VI. Closing Thoughts.
We need fundamental tax reform. And fundamental tax reform must begin with fundamental—clear, good, rigorous—thinking and theory. It must also begin, I believe, with a respect for the people. Viewing themasses of citizenry as confused and helpless souls, on their own lights, who need to be tricked by sleight of hand into doing what is in their own self-interests to do, will not lead to success. Rather we theorists andpolicymakers should treat ordinary people as noble and thoughtful yet busy, overworked, and overtired workers, who have neither the time nor the reasons to discern the lessons of optimal public finance. We shouldnot use the insights of behavioral economics in an ad hoc way, for ad hoc reform that builds irrational mechanisms onto a system in the hope that two or more wrongs can make a right. Rather, we should identifyobjectives and move toward them with coherent strategies, predicated on our understanding of how people actually behave – which, after all, is the point of behavioral economics.
Let’s return to Punxsutawney, Pa., on Groundhog Day. Phil Connors explores every angle he can to manipulate the situation of his endlessly repeating the same day. But he cannot do what he really wants to do: getthe girl, of course, here named Rita (played by Andie MacDowell). Phil learns Rita’s secrets; he teaches himself great skills such as piano playing and French; he even amasses wealth and baubles and beads. None of this works to woo Rita. What he must do, to escape from the perpetual present, is to learn to look within, to change his internal self, and to come up with a more respectful view of others, especially the object of hisdesires. Then at last Phil can move on, to February 3.
It is high time for fundamental tax reformers to turn within, too. The ultimate lesson is about how we think. Like Phil Connors, tax policy theorists seem trapped in a perpetual present, unable to learn the lessons ofthe past to paint a better future. Our tax system, or at least the part of it (the income tax) that we endlessly discuss, is designed to tax — i.e. discourage – productive work, and to double tax savings. This made sense, once, when the tax was small in breadth and depth, and meant to exert a toll on East Coast financiers (McCaffery 2005a). But as the tax grew in scale and scope, its anti-savings bias became a problem. We tried to patchthings up, here and there, but we never went back to first principles. We got addicted to the quick fix, and started to pile on more and more ad hoc tax-favored savings vehicles into the code. This made little or nosense, in theory, and so we needed an irrational model of behavior to justify what we were doing. We turned to behavioral economics, which most theorists neither well understood nor generally accepted, andpicked findings, ad hoc, off the shelf — people were myopic, but they could be tricked by salient cues. The floodgates opened. As the tax system grew more and more complex, we debated the “data,” wondering against all reason if the policies we had advocated “worked,” comfortable in the knowledge, or hope, that we had a theory to make sense of it all. But we didn’t. If we had thought through the theory again, we could have come up with better answers. We could have fundamental tax reform, a more coherent tax policy, and a more sensible global approach to the problem of under-savings on the micro- and macro-levels. We do nothave any of that, today, and have not had any of it, for decades.
It is not, however, too late. It is still Groundhog Day, but that cannot last forever.
Aaron, Henry J., and Harvey Galper. 1985. Assessing Tax Reform. Washington, DC: Brookings Institution.
Aaron, Henry J., Harvey Galper, and Joseph Pechman, eds. 1988. Uneasy Compromise: Problems of a Hybrid Income-Consumption Tax. Washington, DC: Brookings Institution.
Andrews,WilliamD. 1972. PersonalDeductionsinanIdeal IncomeTax. Harvard Law Review 86: 309-85. ———. 1974. A Consumption-Type or Cash Flow Personal Income Tax. Harvard Law Review 87: 1113-88.
Auerbach, Alan C. 2006. Fundamental Tax Reform in the 21stCentury. Prepared for conference: “Is it time for Fundamental Tax Reform” (Austin, TX: Rice University James A. Baker III Institute, April 2006).
Auerbach, Alan J., William G. Gale, and Peter R. Orszag. 2006. The Budget: Plus Ca Change, Plus C’est La Meme Chose. Tax Notes 111:349-70.
Bankman, Joseph, and Thomas Griffith. 1987 Social Welfare and the Rate Structure: A New Look at Progressive Taxation. California Law Review 75:1905-67.
Barro, Robert J. 1974. Are Government Bonds Net Wealth? Journal of Political Economy 82(6):1095-117.
Bell, Elizabeth, Adam Carasso, and C. Eugene Steurle. 2004. Retirement Savings Incentives and Personal Savings. Tax Notes 105:1689.
Bernheim, B. Douglas. 1997. Rethinking Saving Incentives. In Fiscal Policy: Lessons from Economic Research, ed. Alan Auerbach (Cambridge, MA: MIT Press, 1997), 259-311.
———. 2002. Taxation and Saving. In Handbook of Public Economics, Volume 3, eds. Alan Auerbach and Martin Feldstein, (New York: Elsevier Science Pub. Co., 2002), 1174-1249.
Bernheim, B. Douglas, and Antonio Rangel. 2005. “Savings and Cue-Triggered Decision Processes.” Working paper. 2005b. Behavioral Public Economics: Welfare and Policy
Analysis with Non-Standard Decision Makers. NBER Working Paper 11518. Bradford, David F. 1986. Untangling the Income Tax. Cambridge, MA: Harvard Univ. Press. 1996. Fundamental Issues in Consumption Taxation. Washington, DC: AEI Press.
Bradford, David F. et al. 1977. Blueprints for Basic Tax Reform. Washington, DC: US Department of Treasury.
Diamond, Peter A., and James A. Mirrlees. 1971. Optimal Taxation and Public
Production I: Production Efficiency. American Economic Review 61(1):8-27. Doernberg, Richard L., and Fred S. McChesney. 1987. On The Accelerating Rate
and Decreasing Durability of Tax Reform. Minnesota Law Review 71:913. Engen, E.M., and William G. Gale. 1997. Debt, Taxes, and the Effects of 401(k)
Plans on Household Wealth Accumulation. Available at SSRN:
Epstein, Richard A. 2006a. Second Order Rationality. In Behavioral Public Finance, eds. Edward J. McCaffery and Joel C. Slemrod (New York: Russell Sage Publications, 2006), 355-89.
———. 2006b. Behavioral Economics: Human Errors and Market Corrections.
University of Chicago Law Review 73: 111-32.
Feenberg, Daniel R., and Jonathan Skinner. 1989. Sources of IRA Saving. In Tax
Policy and the Economy 3, ed. Lawrence Summers (pub, date), 25-46.
Gale, William G., and John K. Scholz. 1994. Intergenerational Transfers and the
Accumulation of Wealth. Journal of Economic Perspectives 8:145-60.
Glaeser, Edward L. 2006. Paternalism and Psychology. University of Chicago Law
Hall, Robert E., and Alvin Rabushka. 1983. Low Tax, Simple Tax, Flat Tax. New
Graetz, Michael J. 2002. 100 Million Unnecessary Returns: A Fresh Start for the
U.S. Tax System. Yale Law Journal 112(2):261-310.
Hubbard, R. Glenn. 1997. How Different are Income and Consumption Taxes?
American Economic Review 87(2):138-42.
Hobbes, Thomas. 1660. Leviathan. edited by J. C. A. Gaskin. New York: Oxford
University Press (1996).
Kahneman, Daniel. 2000. Experienced Utility and Objective Happiness: A Moment-Based Approach. In Choices, Values, and Frames, eds. Daniel Kahneman and Amos Tversky (Cambridge: Cambridge University Press,
Kahneman, Daniel, Jack Knetsch, and Richard Thaler. 1991. Anomalies: The Endowment Effect, Loss Aversion, and Status Quo Bias. Journal of Economic Perspectives 5:193-206.
Kaldor, Nicholas. 1955. An Expenditure Tax. London: George Allen & Unwin Ltd. Kawachi, Jannette, Karen E. Smith, and Eric Toder. 2006. Making Maximum Use of Tax-Deferred Retirement Accounts. Washington, DC: The Urban
Kotlikoff, Lawrence J., and Avia Spivak. 1981. The Family as an Incomplete
Annuities Market. Journal of Political Economy 89:372-91.
Kotlikoff, Lawrence J., and Lawrence Summers. 1981. The Role of Intergenerational Transfers in Aggregate Capital accumulation. Journal of Political Economy 89:706-32.
Loewenstein, George, Deborah A. Small, and Jeff Strnad. 2006. Statistical, Identifiable, and Iconic Victims. In Behavioral Public Finance, eds. Edward J. McCaffery and Joel C. Slemrod (New York: Russell Sage Publications,
McCaffery, Edward J. 1992. Tax Policy Under a Hybrid Income-Consumption
Tax. Texas Law Review 70:1145-218.
———. 1994a. Cognitive Theory and Tax. UCLA Law Review 41:1861-947.
———. 1994b. Why People Play Lotteries, and Why it Matters. Wisconsin Law
———. 2000. The Tyranny of Money. Michigan Law Review 98:2126-53.
———. 2002. Fair Not Flat: How to Make the Tax System Better and Simpler.
Chicago: University of Chicago Press.
———. 2005a. A New Understanding of Tax. Michigan Law Review 103:807-938.
———. 2005b. Three Views of Tax. Canadian Journal of Law & Jurisprudence 18:
———. 2005c. Good Hybrids/Bad Hybids. Tax Notes, June 27, 1699-705. McCaffery, Edward J., and Jon Baron. 2006. Thinking about Tax. Psychology,
Public Policy & Law 12(1):106-35.
McCaffery, Edward J., and Linda R. Cohen. 2006. Shakedown at Gucci Gulch: the
New Logic of Collective Action. University of North Carolina Law Review
McCaffery, Edward J., and Joel Slemrod, eds. 2006. Behavioral Public Finance.
New York: Russell Sage Publications.
Mill, John Stuart. 1848. Principles of Political Economy. edited by Jonathan Riley.
New York: Oxford University Press (1994).
Modigliani, Francisco. 1988. The Role of Intergenerational Transfers and Life Cycle Saving in the Accumulation of Wealth. Journal of Economic Perspectives 2:15-40.
Modigliani, Francisco, and R. Brumberg. 1954. Utility Analysis and the Consumption Function Function: An Interpretation of Cross-Section Data. In Post-Keynesian Economics, ed. K.K. Kurihara (New Brunswick, NJ: Rutgers University Press, 1954), 388-436.
Moomau, Pamela H. 2006. Comments on “Behavioral Economics and Fundamental Tax Reform?” by Edward J. McCaffery. Prepared for conference: “Is it time for Fundamental Tax Reform” (Austin, TX: Rice University James A. Baker III Institute, April 2006).
Pechman, Joseph A., ed. 1980. What Should Be Taxed, Income or Expenditure?
Washington, DC: Brookings Institute Press.
———. 1984. Options for Tax Reform. Washington, DC: Brookings Institute Press. President’s Advisory Panel on Federal Tax Reform. 2005. Simple, Fair, and Pro-
Growth. Washington, DC: U.S. Government Printing Office.
Rabin, Matthew. 2002. A Perspective on Psychology and Economics (Alfred
Marshall Lecture). European Economic Review 46:657-85.
Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press. Scholz, John Karl, Ananth Seshadri and Surachai Khitatrakun. 2006. Are Americans Saving ‘Optimally’ for Retirement? Journal of Political Economy
Seidman, Laurence. 1995. The USA Tax. Cambridge, MA: MIT Press.
Sen, Amartya K. 1961. On Optimising the Rate of Saving. Economic Journal
Shafir, Eldar, Peter Diamond, and Amos Tversky. 1997. Money Illusion. The
Quarterly Journal of Economics 112:2, 341-74.
Shefrin, Hersh M., and Richard H. Thaler. 1988. The Behavioral Life-Cycle
Hypothesis. Economic Inquiry 26(4):609-43.
Simons, Henry C. 1938. Personal Income Taxation: The Definition of Income as a
Problem of Fiscal Policy. Chicago: The University of Chicago Press.
Slemrod, Joel. 1990. Optimal Tax and Optimal Tax Systems. Journal of Economic
Smith, Adam. 1776. The Wealth of Nations. Intro. Alan B. Krueger. New York: Bantam CLA (2003).
Steuerle, Gene. 2005. Contemporary US Tax Policy. Washington, DC: The Urban
Strnad, Jeff. Comments on “Behavioral Economics and Fundamental Tax Reform.” Prepared for conference: “Is it time for Fundamental Tax Reform” (Austin, TX: Rice University James A. Baker III Institute, April 2006).
Surrey, Stanley S., et al. 1973. Pathways to Tax Reform: The Concept of Tax
Expenditures. Cambridge, MA: Harvard University Press.
Surrey, Stanley S., and Paul McDaniel. 1985. Tax Expenditures. Cambridge, MA: Harvard University Press.
Thaler, Richard H. 1994. Psychology and Savings Policies. American Economic
Thaler, Richard H., and Hersh M. Shefrin. 1981. An Economic Theory of Self- Control. Journal of Political Economy 89(2):392-406.
Tversky, Amos, and Daniel Kahneman. 1986. Rational Choice and the Framing of
Decisions. Journal of Business 59:4, 5251-78.
Unger, Roberto Mangabiera. 2000. Democracy Realized: The Progressive Alternative.
New York: Verso.
Venti, Stephen F., and David A. Wise. 1986. Tax-deferred Accounts, Constrained
Choice, and the Estimation of Individual Saving. Review of Economic Studies
———. 1992. Government Policy and Personal Retirement Saving. In Tax Policy and the Economy, ed. Lawrence Summers (Cambridge, MA: MIT/NBER Press, 1992), 6:1-41.
Warren, Alvin C., Jr. 1975. Fairness and a Consumption-Type or Cash-Flow
Personal Income Tax. Harvard Law Review 88:931-46.
———. 1980. Would a Consumption Tax Be Fairer Than an Income Tax? Yale
Law Journal 89:1081-124.
———. 1996. How Much Capital Income Taxed Under an Income Tax Would Be
Exempt Under a Cash-Flow Tax? Tax Law Review 52(1):1-16.
Weisbach, David A. 2000. Ironing Out the Flat Tax. Stanford Law Review 52:599-
664.———. 2006. The Case for a Consumption Tax. Tax Notes 110:1357-59.
[*]This paper is being prepared for a conference, “Is it Time for Fundamental Tax Reform?” to be held at the James A. Baker III Institute at Rice University, April 2006. I thank James Alm, Alan Auerbach , Colin Camerer, and George Zodrow for helpful conversations and editiorial assistance; Pamela Moomau and Jeff Strnad for the their thoughtful commentaries, and ted Dillman for excellent research assistance.
Once again, mine is an analytic claim about possibilities, not an empirical claim about the extent to which the “buy/borrow/die” strategy is implemented. But certainly the fully “rational” investor/saver would be aware of the possibilities, there is ample evidence that rich investors are, and the mere possibilities of the strategy constrain important elements of tax system design. I discuss all this at much greater length in McCaffery 2002 and 2005a.
Note that, even if many or most Americans are indeed saving enough (see Scholz, Seshadri and Khitatrakum, forthcoming, for an argument that they are or might be), this chapter’s basic conclusions obtain, as the “optimal” savers could continue to do so under a general consumption tax structure.
I understand that this graphic can be hard to read in black and white. The lines increase monotonically with earnings level. All four earnings categories under $75,000 a year — approximately double the median income for a family of four — are virtually at 0%, no higher than
3%, thoughout the time period, 1990 to 2003. The four more readily distinguishable lines start at the $75,000 to 100,000 level, where maximal contributions straddle the 10% level throughout the time period, and so on.
I do not share Professor Strnad’s skepticism on this point, though, if that is what it is. Strand, in his fourth section, lists a possible “arbitrage” mechanism under a consistent cash-flow consumption tax, in which, as I understand it, an individual borrows to consume today, as by taking a long-lived asset, and defers repayment until his death, when the government is left holding the bag. But it is hard to ascertain who such a lender would be, and I do not understand why the government would give him, her or it priority over its claims to the monies being paid over.
Previously published by the University of Southern California – Law School