All entries for Friday 04 May 2012
May 04, 2012
Writing about web page http://www.ft.com/cms/s/0/30177568-61fa-11e1-807f-00144feabdc0.html#axzz1tpB8SkZF
Always a hot potato, the top rate of tax is getting hotter. George Osborne is under fire for cutting taxes on incomes over £150,000 from 50p to 45p (this rises to 56p in the pound if you include payroll taxes). In the United States the marginal rate on incomes over $388,000 is 35 cents (rising to 42.5 percent with Medicare and payroll taxes) but many allowances mean it's not always paid. Barack Obama wants to put in place a "Buffet tax" to ensure those earning more than a million dollars pay at least 30 percent on average. François Hollande, likely winner of the French presidential election, wants to set a 75 percent tax rate on incomes above a million euros.
One way to understand the issues at stake is to recognize and reflect on two conflcting purposes of taxation in a democracy. In one view, taxes go along with citizenship. In the other, taxes help to re-engineer society. Where should we strike the balance?
- Taxation in a democracy
Many journalists and some economists write about taxes as if the optimum rate should be the one that maximizes the revenue from it. As a general principle, that can't be right. To understand why not, ask yourself what rate a coolly rational totalitarian dictator would set who treated the public finances as his personal housekeeping money and cared nothing for social welfare. This ruler would set taxes to maximize his revenue -- would he not?
What level would that be? Even for a dictator, the optimal tax rate is not 100 percent. The reason is that all taxes are in some degree "distorting," that is, they are a disincentive to engage in the activity that is taxed, so that the economy becomes less efficient and less is produced. Think of taxes as a proportion t of income, so if T is the revenue and Y is the income then T = tY. Because taxes are distorting, as you push up t, Y falls. When t = 100 percent, the disincentive to produce is so overwhelming that Y is zero and so is tY. Reduce t; Y will grow and so will tY up to a point. At some point, the product of t and Y is maximized.
Actually, a farsighted ruler might set today's taxes below even this level, and forego some of the revenue available today for the sake of more revenue tomorrow. Why's that? He would be aware that higher taxes may cause lower growth (Heady et al. 2008; Arnold et al. 2011). If so, the dictator who maximizes his revenue today will encounter lower revenue tomorrow. Given this, a dictator might well choose to hold back today and encourage the economy to grow so that he will be able to levy taxes from a larger base in the future.
If we take the dictator as a benchmark, how should a well-meaning democratic leader set taxes, having internalized the welfare of society? As Mançur Olson (1993) pointed out, the dictator values only his own welfare; the democratic leader values the welfare of society, which includes private income as well as public revenues. In the same spirit, the dictator seeks to avoid the distortions associated with taxation only to the extent that they undermine his own revenue, and will be indifferent to private losses; the democratic leader will set some value on the private losses too.
For these reasons the democratic leader should set income taxes at much less than the rate that a dictator would set. If even a dictator should go below the rate that maximizes revenue in the short term, the democratic leader should go some way below that.
Summary: maximizing revenue is not a legitimate purpose of taxation in a democracy.
- Taxing top incomes
Why then do some economists advocate setting the upper rate of income tax at the revenue-maximizing rate? In a recent paper Peter Diamond and Emmanuel Saez (2011) argue that the optimal rate that should be levied on the top 1 percent of U.S. incomes is more than 70 percent. They find this to be optimal because it would extract the maximum revenue from upper income recipients, taking into account the fact that they will probably reduce their taxable income in response. For given revenue needs, extracting the maximum out of the richest in society would then minimize the burden on taxpayers with lower incomes.
The underlying logic is diminishing marginal utility. The social benefit of a dollar consumed by the rich, Diamond and Saez argue, is much less than that of a dollar consumed by the poor. For example, if the benefit of consumption increases in proportion to the logarithm of income, then each doubling of income would increase well being by a given amount. They estimate that a dollar of consumption when income is at the top-1-percent margin in 2007 (around $1.4m) is worth less than 4 percent of a dollar when income is at the median ($53k) in the same year. One way of improving social welfare, they conclude, is take dollars from the rich family (which hardly misses them) and give them to the poor family (to whom it means a lot). On this argument, the result of putting up taxes on top incomes and reducing taxes on middle incomes is that well-being will rise on average.
Where do you stop? Because 4 percent is close to zero, Diamond and Saez summarize, the value of marginal income to the very rich is so low that we can ignore it. It follows that the tax rate on upper incomes that maximizes social welfare is pretty much the same as the tax rate that maximizes the revenue from the tax -- and this turns out to be around 73 percent.
Summary: Diamond and Saez make a case for maximizing revenues from top incomes -- even in a democracy. Since this is pretty much textbook stuff, what is there to disagree with? I'll set out two views of taxation. In one view, taxes are an instrument for redistribution. In the other, taxes go with citizenship. Both views can be taken to an extreme. I'll argue that both extremes are undesirable, but it's important to acknowledge the link from taxes to citizenship. There is no such link in Diamond and Saez.
- Taxation for redistribution
In this debate, Diamond and Saez are extremists for social engineering. That is, they subordinate all other considerations to distributing welfare as evenly as possible. Before I go on to the other view, I'm going to say why I think this is economically misconceived. There are several reasons.
First, the Diamond-Saez framework presumes that everyone maps income onto personal well being at the same rate at each income level. I've explained the average relationship. But people are not all average. An example can give the intuition. Suppose my income is $1.4m. (It's not, and never will be. I make this clear for those readers that don't easily get the difference between a working assumption and a tax declaration.) To repeat, suppose my income is $1.4m. At that point, I may well value a dollar of consumption at less than 4 percent of a dollar when my income falls to $53k. It does not follow that I will value a dollar of my consumption at less than 4 percent of the value the next guy puts on a dollar when his income is at $53k.
Think about it: If my income was 30 times his, was I luckier, or better connected, or more talented than him, or did I just want money more than he did? A new paper by Eugenio Proto and Aldo Rusticchini (2012) suggests that the slope of the relationshp between income and life satisfaction differs systematically for people with different personality traits, so this is not an abstract possibility. If so, taking money from me, given that I have more because I value it more highly, and giving it to him, when he has less because he values it less, is not an obvious way to improve average well being.
[Now for a short digression. Some readers might stop short here and ask: "Why ever not? That's exactly what I want to happen!" Maybe you'd get personal satisfaction from taking money away from me, precisely because I value it. I understand that; if so, however, your preference is likely to be rooted in a dislike of rich people and high consumption. Some people who take this line dislike bankers, neglecting the simple fact that many people with high incomes have nothing to do with finance. Others do not think anyone should receive an income 30 times the median on principle -- even though their own consumption may well be 30 times what was the median in their own country just 200 years ago, or 30 times the median income of some other country just 2,000 miles away. All these people are expressing a third view of taxation, namely taxing people because you don't like their values and want to punish them. That's enough about that.]
The same issue has a more general form. It is about giving the government the authority to treat everyone as an average citizen with average abilities and average preferences, to discount entirely the welfare of some of them at the margin, and radically to re-engineer the distribution of income, motivated by the wrong assumption that we are all the same (or if we're not that's too bad). It's egalitarian, certainly; but not, perhaps, in a good way.
If you followed the logic of taxes as social engineering to the limit, you'd want the government to set everyone an individual personal tax schedule, based on knowing deeply each person's individual capabilities, endowments, and preference map. That way lies a dystopian nightmare. We should neither pretend to have this information nor expect to be able to gather it. The collection of such knowledge in one place woujld be more consistent with a totalitarian state than with a decentralized, free market economy.
Finally, we all pay an economic price for social engineering. On top of the other evidence linking higher taxes to lower growth, Arnold et al. (2012) also show that a higher top rate of personal income tax is damaging specifically to productivity growth in industries that have high rates of firm entry. They suggest this arises because new firms have to take more risks and are also less likely to be incorporated. In other words, there is a channel through which raising taxes on top incomes is likely to make the market economy less dynamic. That matters too.
Summary: I've given several reasons not to follow the logic of taxation for redistribution to the limit.
- Taxation and citizenship
The alternative view can be summed up as "no representation without taxation." In a democracy, taxes are a membership due, or a moral obligation on the citizens to share society's burdens as well as benefits. Our democracy needs most families to pay their dues and so, when they vote, to have a stake, however small, in how (and how much) public money is raised. This view is also widely held, although it pops out most commonly when people react against tax loopholes for the rich and tax allowances for charitable giving. Shouldn't these people pay their taxes first?
The logic here is that you pay because you're an citizen. Equal citizens should contribute equally. Again, this is not something to take to extremes. Why not? The logic of taxes as membership dues is that everyone pays. To my British readers I'll say: Remember the poll tax. For others, the poll tax was Margaret Thatcher's attempt to fund local government through a level tax on all householders, and it arguably destroyed her premiership. In fact, not everyone can pay; their ability to pay is not equal. Many regular clubs and societies recognize ability to pay through concessionary rates for the young, the old, the sick, and those out of work. Just about all tax systems recognize ability to pay through income tax thresholds and marginal rates that rise with income. In this way, they modify the citizenship concept with a necessary nod to redistribution.
Diamond and Saez call their paper "The Case for a Progressive Tax." As far as I'm concerned, the case for income tax to be progressive was already made. I didn't need convincing. Given unequal ability to pay, the rich should pay more than the poor in proportion to their incomes. But the case Diamond and Saez make goes far beyond that: they advocate a tax system that completely confiscates the marginal social benefit of consumption from top incomes in order to minimize the burdens lower down. In their framework the middle and lower income families become clients of the state, not citizens. There is no acknowledgement that citizens have obligations, or that voters should have a private stake in how public revenue is raised. They push redistribution to an extreme, diminish citizenship, give the government too much power, and threaten long term damage to the market economy.
Summary: While remaining progressive, our fiscal system should leave room for an element of tax-paying as an attribute of citizenship. Everyone who can should pay something; only the neediest should pay nothing. The rich should pay at a rate higher than that on the middle and poor, but taxes on upper incomes should not aspire to confiscate all the gains from effort, enterprise, and talent. (And, if we leave untaxed a signficant share of the returns to effort, enterprise, and talent, we'll also have to accept that we leave undisturbed some part of the gains to connections and luck.)
In British terms, if the poorest pay nothing, and the middle pay 20 percent, then the rich can clearly pay 40 percent. That seems fine. The rest is politics.
- Arnold, Jens Matthias, Bert Brys, Christopher Heady, Åsa Johansson, Cyrille Schwellnus, and Laura Vartia. 2012. Tax Policy for Recovery and Growth. Economic Journal 121:550, pp. F59-F90. Available at http://ideas.repec.org/p/ukc/ukcedp/0925.html
- Johansson, Åsa, Christopher Heady, Jens Arnold, Bert Brys, and Laura Vartia. 2008. Tax and Economic Growth. OECD Economics Department Working Paper No. 620. Available at http://ideas.repec.org/p/oec/ecoaaa/620-en.html
- Olson, Mançur. 1993. Dictatorship, Democracy, and Development. American Political Science Review 87:3, pp. 567-576.
- Diamond, Peter, and Emmanuel Saez. 2011. The Case for a Progressive Tax: From Basic Research to Policy Recommendations. Journal of Economic Perspectives 25:4, pp. 165–190. Available at http://ideas.repec.org/a/aea/jecper/v25y2011i4p165-90.html.
- Proto, Eugenio, and Aldo Rustichini. 2012. Life Satisfaction, Household Income and Personality Traits. The Warwick Economic Research Papers no. 988. Available at http://ideas.repec.org/p/wrk/warwec/988.html