Why don't tax cuts boost growth?
Relative to current events, this probably seems less important at the moment, and a week or so too late, but the tax cut proposed by th Trump administration is still kicking around. There have been any number of responses to it, and responses in particular to the claim that it will “supercharge” economic growth, and ultimately pay for itself. In short, it won’t do either of those things, which is the conclusion you come to after reviewing all the evidence available. I did so here, in a prior post. Bruce Bartlett just posted an article where he also comes to that conclusion, and recants his earlier stand in the 1980’s pushing the whole “tax cuts increase growth” line.
The purpose of this post is not to rehash the arguments and evidence showing that tax cuts do not increase growth. Rather, I want to think a little about why they do not increase growth. Because the claim that they do sounds right, but fails in the real world.
Let’s start with some basics. If you want to raise economic growth - meaning an increase in real GDP - then you need to increase the amount of capital, labor, and/or productivity. Whether that is pulling new capital, labor, or innovations into the economy, or putting idle capital, labor, and innovations back to work is irrelevant. You want to increase the inputs to increase GDP.
The core of economics is that people respond to incentives. In the case of inputs, the incentives to provide them are the returns they earn, like wages, dividends or interest, or profits. If you tax those returns, that lowers the returns, and people will provide less labor, capital, or innovations. So if you lower that tax, the idea is that people will provide more labor, capital, or innovation. Nothing here is rocket science. Why does it not work the way we think?
We think that if you lower the tax rate, and hence raise the returns to inputs, we should get more of them. But to “supercharge” growth in GDP, or to have any appreciable effect on GDP at all, you need that the elasticity of that input supply with respect to those returns is really big.
For labor, there appears to be good evidence that this elasticity is in fact small. There is not some pent-up store of workers and human capital out there that is just a 35% tax bracket away from getting off their ass and going to work. This labor supply elasticity is found to be essentially zero in almost every case, with the exception of married women. You can see some citations on this in the review paper by Saez, Slemrod, and Giertz (SSG). With an elasticity close to zero, no matter how much you lower the tax rate, and raise the return to labor (i.e. the wage), you can’t induce a substantial increase in labor supply. And without a substantial increase in labor supply, you don’t get a big increase in GDP.
On the capital side, this is tougher to estimate because while it is easy to assert that reported wages are payments for real labor, it is much harder to say what part of reported business income is a payment to “capital”. Dividends are probably one of the cleanest things to look at, as they are direct payments to stockholders. But when Yagan looked at the 2003 dividend tax cut, he found there was no effect on corporate investment (i.e. real capital inputs). In other words, the elasticity of real capital inputs to the tax rate was zero.
For innovation, there is some better evidence of an effect of taxation on R&D activity (although we have to be careful to distinguish between R&D activity and innovation itself). Dechezlepretre, Einio, Martin, Nguyen, and Van Reenan have a working paper that uses a change in the threshold for qualifying for an R&D tax credit in the UK on both R&D spending and patenting, meaning they are trying to see an effect on actual innovation. They find significant effects of R&D tax incentives on R&D and patenting. Now, tax incentives (like an actual tax credit or allowing firms to write off R&D spending) are not exactly the same as tax rates. And the current tax proposal, so far as I can tell, does not include a specific change to R&D incentives. Changing the marginal tax rate on individual income or corporate income is not the same as providing R&D tax incentives. But, if you are looking for an avenue for tax policy to spur growth, then creating incentives for R&D and innovation appears more plausible than lowering tax rates on labor and capital.
We tax financial flows, not inputs
The second thing reducing the impact of tax cuts on GDP is that taxes are not levied on only the transactions that are part of GDP, they are levied on all kinds of transactions. Think of the fact that every year there are billions (and maybe trillions) of individual transactions that take place in the economy. Some of those transactions involve payment for a real input like labor or capital, which means they are part of GDP. But some of those transactions are purely financial, and so are not. But the tax code does not make this distinction; you are not taxed only on your transactions that contribute to GDP, you are taxed on all (okay, most) of your transactions where you receive some money.
Let’s break this down to four different kinds of transactions.
Taxed, and income is from providing a real input. Your paycheck is an example. You provide a real input (labor) to your employer, and hence this income is part of GDP. The income you earn is also subject to taxation. So if you lower the tax rate on this income, you presumably will supply more labor, which will raise GDP. As we saw above, this response would be small for both labor and capital, so GDP would not go up by much. Lower taxes on R&D or innovation might have a more tangible effect.
Untaxed, and income does not come from providing a real input. Think of a small gift of cash, under the IRS threshold, from your grandparents to you. This is income, in the sense that you have an inflow of money, but it is not in return for a real input. It is untaxed because it happens to be small in size.
Untaxed, and income is from providing a real input. You could think of small scale payments for work that fall below IRS reporting thresholds, like babysitting or some home services. Now, because they do not have to be reported to the IRS, they are untaxed, but its also likely they don’t get picked up and measured as part of GDP. But technically, because they involve someone providing a real input like labor, this income should be measured as part of GDP. Regardless, because they are untaxed, lowering tax rates would do nothing to change the real inputs people provide here; your babysitter is not going to offer to do a 2nd night a week if the marginal tax rate falls.
Taxed, and income does not come from providing a real input. Here we’ve got a whole world of financial transactions. If you buy shares of stock for cash, there is no real input provided, so this transaction does not add to GDP. But, this transaction is taxed, in the sense that the capital gain to the seller is subject to tax. If you lower tax rates, then you will get more of these financial transactions, but that will not add anything to GDP. [A side note is that the commissions on this trade do get counted as part of GDP, because they are made for providing a real service.]
One big reason that tax cuts do not boost growth in GDP is that tax cuts apply to transactions of type 4, as well as type 1. Lowering the tax rate should increase transactions of both types In practice, the elasticities involved in type 1 transactions are small, so there is little change in real inputs provided by tax cuts.
But tax cuts may have a big effect on those type 4, financial, transactions. Those type 4 transactions may boost taxable incomes by making the transactions more attractive, but have no impact on the real provision of inputs which would go into raising GDP. SSG use the example of the 1986 tax reform to demonstrate this. In response to this, there was a surge in reported S-corp income (i.e. “pass-through” income), which implies a legal shift from corporations to S-corp’s, but this had no practical effect on the supply of inputs to the economy. It is just reshuffling the paper ownership of companies.
In fact, you could make the argument that if you want to raise GDP it would be ideal to raise taxes on financial transactions (a la the Tobin tax) to reduce the incentives to do those type 4 transactions, but lower taxes on type 1 transactions dealing with the provision of real inputs, and definitely provide incentives to R&D spending and innovation.
If you wanted to convince me that these financial transactions of type 4 do have an important effect on real GDP growth, then you’d have to talk me into believing that facilitating the exchange of ownership leads to a more efficient allocation of management ability, or something like that. For that to be true, it would have to be that there are a lot of exchanges of ownership that are not currently taking place because they are not worth it due to high taxes. But the prima facie evidence works against you, because you’d want to show me that there has been some kind of decline, or stagnation in financial transactions like this over time, and that is just not the case. But that’s your in, if you are looking for one.
That aside, I think it’s key to reiterate that claiming tax cuts will not have an effect on growth is not a denial of the basic theory. Rather, the claim is based on the low elasticities, which mean that the responses are small, and a disconnect between the returns that are taxed and the returns that are paid to real inputs.
Back to blog