Who Pays for Corporate Tax?

Please don’t say “corporations”

Reminder: Progress Ireland’s next monthly drinks will be on Wednesday 25th March, upstairs in Doheny & Nesbitt, from 17:30 to 19:30. Looking forward to seeing you there!

---

It’s almost impossible to overstate how important corporation tax is to modern Ireland.

In 2024, 22 percent of all government revenue came from corporation tax. Only income tax is a bigger source of funds for the Irish state.

There is a smattering of other OECD countries that are even more dependent on corporate tax than us. But where Ireland is an extreme outlier internationally is in how much of that tax is paid by a relatively small number of American multinational companies. Only 11 percent of Irish corporation tax is paid by Irish companies. The distribution has become so skewed to the point of boggling the mind: in 2024, Apple and Microsoft combined paid 40 percent of all corporate tax. Ten companies pay €0.60 out of every euro of corporation tax.¹

In the bad old days, when Ireland had to be bailed out by the IMF, the European Central Bank, and the European Commission (The Troika), one of their key warnings was that Ireland would end up trading an overreliance on property-related taxes to an overreliance on corporate taxes. Reporting from the OECD and other international financial institutions routinely warn that Ireland has a narrow tax base.

All of these issues are well-known. The concentration of the tax base is a big problem, but remains difficult to solve for lots of reasons of incentives and political economy. The fiscal situation it has put the Irish government in can reasonably be described as a bonanza.

But this post is not about that. It’s about something more fundamental: How does Irish corporation tax actually work? Who pays for it? I’ll outline two influential models economists have developed relevant for thinking about corporation tax, then consider how these relate to the Irish case in the final section.

How Irish corporate tax works

Ireland’s corporate tax rate is 12.5 percent on trading profits, and 25 percent on any passive income.² Because of R&D tax credits, relief on income from certain forms of intellectual property, and a handful of other incentive schemes, the ‘effective rate’ on corporate profits is often significantly below 12.5 percent.

This, of course, is famously low. Most Irish adults probably know the 12.5 percent number. Is there another country on earth where the average educated person knows the rate of corporate tax off the top of their head? Most Irish people correctly understand low corporation tax to have been used very successfully to court foreign direct investment and American multinationals in a way that, for all its problems, greatly enhanced our standard of living.

I once spoke to a businessman who told me that he would be upset if the government announced that they were slightly reducing corporate tax, because that would imply that the rate was changeable. ‘Twelve and a half’ came to be as much a permanent fixture of Irish life as Guinness or Father Ted.

When you exclude Caribbean tax havens, there are very few countries with a lower corporation tax than Ireland. Across the OECD, the average is 24 percent.

Since the beginning of 2024, Ireland has been implementing the ‘Pillar Two’ agreement from the OECD, which is an attempt to harmonise global tax laws and enforce a minimum global corporation tax.³ Companies with an annual turnover of more than €750 million must now pay a minimum effective corporate tax rate of 15 percent. This effective rate is a weighted average of the tax rate across profits and passive income, net of deductions and tax incentives.

Part of what was required to get Ireland to agree to this framework was a change in wording so that a minimum tax of “at least” 15 percent was changed to be exactly 15 percent.⁴

Naturally enough, €750 million is an enormous amount of money, and the vast majority of companies operating in Ireland are still paying at most 12.5 percent tax on their profits.⁵ A working paper estimates that 1,000 companies will pay a top-up payment to the Irish government to reach the new 15 percent rate as part of Pillar Two. Since those payments will come from the largest companies, it’s reasonable to worry that Pillar Two will make Ireland’s tax concentration even worse. But we don’t yet know how big a deal the global minimum corporate tax will be for Ireland, because the filing deadline for the first payment of this tax is not until June this year.

Why does a tax that was paid since 2024 not have to be reported on until mid 2026? A story for another day…

The economic incidence of corporation tax

My favourite mantra in economics is “Legal incidence does not determine economic incidence.”

The economic incidence of a tax is the change in the distribution of welfare because of it, compared to the hypothetical world with no tax. This language is counterintuitive: economic incidence is about the change in how well off different individuals are. But the individuals affected (economic incidence) might be completely different from the ones who pay for the tax on paper (legal incidence).

This is a profound idea, worth deeply reflecting on. It took no less a genius than John Stuart Mill, in The Principles of Political Economy (1848), to first articulate it clearly. A textbook example would be a tax on cigarette factories. Demand for cigarettes is relatively inelastic, i.e., not very responsive to price changes. That means that the cigarette factories can raise their prices in response to the tax, and not lose many customers. The factory is able to pass the tax on to the consumer in the form of higher prices. Although all the cheques to the government are signed by the factory owner, the majority of the economic incidence is borne by smokers.

Analysing the economic incidence of tax is incredibly important for public policy. One of the first results you’ll look at in any microeconomics class is that the burden of a tax tends to fall on whichever factors are least elastic, i.e., least responsive to price changes. The part of the market which has fewer alternatives can’t easily escape the tax.

Generally speaking, elasticities are empirical quantities that need to be estimated from the data. In the case above, the split in how the tax is paid between smokers and factory-owners is determined by the price elasticity of demand: how much the demand for cigarettes declines as price increases.

This has important distributional consequences. Often, the legal incidence of a tax is on a group that is financially well-off (big companies, landlords, factories). But we need to look at whether the economic incidence is borne by much poorer or worse-off groups (workers, renters, consumers). When a policy sounds good on paper, but actually has the opposite of its intended effect, often someone involved has failed to internalise the deep wisdom of “legal incidence does not determine economic incidence.”

Let’s keep this in mind when thinking about corporation tax. Logically, there are three ways that the incidence of a corporate tax can be borne:

1. Lower wages for employees

2. Higher prices for consumers

3. Lower profits for shareholders

Most people think that corporation tax is primarily paid through channel #3.⁶ But why? We certainly can’t deduce it from first principles. It is logically possible that the entire effect of corporation tax on an economy is to reduce wages for employees.

The best empirical evidence I’m aware of on this front comes from Do Higher Corporate Taxes Reduce Wages? Micro Evidence from Germany, a 2018 paper by Clemens Fuest, Andreas Peichl, and Sebastian Siegloch. It exploits changes in corporate tax rates across German municipalities using a differences-in-differences methodology, which I previously wrote about in the context of measuring trends in academic entrepreneurship.

Their main finding is that, indeed, corporate tax reduces wages: half of the burden of corporate tax is paid by the workers, with the other half paid in the form of lower returns for shareholders. This is broadly in line with the educated guesses of economists, who on average believe that 40 percent of corporate tax incidence is on capital, with wide disagreement.

Interestingly, in Germany, corporate tax is disproportionately paid by low-skilled, young, and female employees.⁷

The belief that corporation tax is paid through channel #3 is presumably why it is widely believed to be one of the most progressive taxes around. The authors do a back-of-the-envelope calculation, in which they find that, if you take their results at face value, the overall progressivity of the German and US tax system is 25–40 percent lower than in the hypothetical world in which corporate tax was borne entirely by capital.⁸

Most of the papers in this literature aren’t able to evaluate channel #2 (corporation tax raising prices for consumers). In the case of Fuest et al., firms aren’t really going to charge different prices in different parts of the country. Even if corporate taxes are paid in the form of higher prices, this study wouldn’t be able to identify it.

Fuest et al. is an empirical paper about the short run. Although the dataset they managed to get their hands on is extremely impressive, they can only evaluate the effects of policy changes five years out. The authors manage to carefully be ‘theory neutral’ about what activities exactly corporation tax disincentivises. As will be explained in the next section, there are reasons to think that the share of corporate tax paid for by workers will be higher in the long run.

Open economies and capital flight

One of the most foundational papers in the field of public finance is The Incidence of the Corporation Income Tax from 1962. In it, Arnold Harberger introduced a general equilibrium model, which has surprising implications for efforts to tax businesses. The formal definition of general equilibrium is a bit technical, but loosely, he is considering the long-run effect across all goods, when all markets are allowed to clear. This is in contrast to partial equilibrium analysis.

On my personal blog, I recently cheekily wrote about how few Irish people know about Harberger’s general equilibrium, despite it being the intellectual cornerstone of our economy.

Harberger was considering the corporation income tax, which is a tax on the net return to capital employed in the corporate sector. This is not the same thing as what we now call ‘corporation tax’, but they are highly related, and we will return to understanding their differences in the final section.

The model is a bit technical to get into properly, but here are some of the intuitions. In the short run, capital is largely fixed (factories are already built, equipment is installed). But in the long run, capital is more mobile than labour is: human beings are sticky creatures, and will only move across borders in response to economic pressures to a certain degree.⁹ As such, more of the incidence of corporate income tax is borne by capital in the short than in the long run. This is a reason to think that the 50 percent number that Fuest et al. arrived at is a lower bound on the fraction of the incidence borne by workers.

In his original paper, Harberger was considering a closed economy. His headline result was that, in such a world, the corporate income tax burden falls entirely on capital. Without other markets that the capital can move to, when they get taxed, capital owners have to just “suck it up” and accept a lower rate of return.

Extensions of his model to small open economies came later. In the limit case, the central result flips: in the long run, all of the incidence will be borne by labour. Capital-owners won’t accept a lower rate of return, because they can just move their capital stock to another country (‘capital flight’).

Once again: The people you think you’re taxing aren’t necessarily the ones who really pay.

Usually, when you hear that an economic model is about a ‘small open economy’, it is based on very strong assumptions, like that capital is freely mobile, and that the economy in question has no ability to influence the global interest rate. Naturally enough, these assumptions are only ever true to a degree. Nevertheless, models of this kind have led to a general expectation that, the smaller and more open the economy, the more the burden of corporate income tax will be borne by workers.¹⁰

This mechanism is how the incidence of a corporate income tax connects to efficiency. When a small open economy starts taxing corporate income, you get capital flight. And a lower capital stock means workers are less efficient, reducing overall output for the whole country. This is also why the burden of a higher Irish corporate tax would be borne by Irish, not American, workers. The company paying the tax receipts might employ mostly Americans, but the path by which corporate tax hurts workers is in reducing the amount of capital in the country imposing the tax.

That is the logic leading to: If you’re a small open economy and capital is mobile, then, selfishly, you should tax corporate income at a low rate. And in practice, this is the pattern we see: small open economies tax corporations somewhat less than big closed ones.

Harberger’s ideas only get us so far. But this basic mechanism is probably about as relevant in Ireland as it is anywhere in the world. That there is a high share of foreign-owned capital and a high reliance on multinational investment suggest that Ireland’s tax base is highly mobile.

There is something that feels unjust about this. When your economy is closed, you can laser-target a tax on capital owners, who are presumably some of society’s wealthiest. But when it is open, capital flight can mean that, when you try to tax companies, you only end up hurting yourself. In the face of this cold hard logic of the market, the desire for a global minimum corporate tax is entirely understandable.

The inefficiency of taxing production

Corporation tax influences lots of decisions that a business makes: how much of its product to produce, how much to invest, and so on. There’s an influential framework for thinking about these distortions called the Diamond-Mirrlees production efficiency theorem. It’s another influential general equilibrium result, this one from 1971.

This theorem is sometimes summarised as saying that, in an optimal system, taxes should not distort production decisions.

It’s inevitable that taxes will create some distortion compared to the set of mutually beneficial trades that would otherwise occur. This is called deadweight loss, and is one of the most important concepts in all of economics.¹¹ We face a choice over whether to impose this distortion at the level of consumption decisions (“How many houses should I own?”, “How expensive a car should I buy?”) or production decisions (“How much steel should I sell?”, “How many workers should I employ?”). It’s certainly far from obvious which is best!

Again, the details are a bit technical, but Diamond-Mirrlees models a situation in which a hypothetical benevolent government is trying to maximise social welfare, and has a choice about whether to impose taxes on final goods, or on intermediate goods used in production. It then uses the revenue raised to enact redistribution. The key result, under some strong assumptions,¹² is that taxes on intermediate goods should be zero.

Suppose a government is raising revenue in an economy where bakeries are very important. They can choose from the following options:

• Tax consumers for buying bread.

• Tax the flour the bakery uses to bake the bread.

When taxing the bread, there is one distortion: in how much bread people buy. When taxing the flour, there are two distortions: in how much flour the bakery should use to bake its bread, and (indirectly) in how much bread people buy. Of course, a consumption tax on bread will influence a bakery’s decision about how many ovens to buy, but only indirectly, by changing the demand for its good.

It’s theoretically possible that taxing flour directly might be the best way to achieve redistribution, but the genius of Diamond and Mirrlees was to show that the redistributive effect of any production tax can also be achieved with a corresponding set of consumption taxes. Production taxes give us no benefits in terms of redistribution, and they multiply the opportunities for distortion.

As with everything in economics, it is debatable how realistic the assumptions are, and whether any of this is particularly relevant to real life. But Diamond-Mirrlees is a major reason why economists have long been wary of taxes that distort production efficiency, and favoured doing revenue-raising on the consumption side.

Thus, there has been scepticism of corporation tax insofar as it affects firms’ production decisions. The question of how much corporation tax really does that is itself complicated and messy. Let’s combine these insights with the earlier discussion of capital flight to understand why corporation tax is so difficult to evaluate.

Conclusion: A confusing mess

The Harberger and Diamond-Mirrlees models have both inspired large literatures in finance and international economics.¹³ They are probably the two most foundational results for how economists think about corporate tax.

However, neither of them are really about what we now call ‘corporation tax’. A corporation tax is a tax on the profits a company makes. But, as mentioned, Harberger-style models are about taxes on the net return to the capital employed in the corporate sector. While Diamond-Mirrlees is a tax on intermediate goods used in production.

The problem is that nobody seems to know what corporation tax actually is. I’m not being facetious. Corporation tax is genuinely one of the most conceptually ambiguous taxes.

The reason why is explained by the difference between economic profit and accounting profit. Economic profits are an instance of what economists call ‘rent’, and are also called supernormal profits. Economic profit accounts for all explicit and implicit costs, including opportunity cost. Accounting profit is just revenue minus cost, and in everyday speech is almost always what people mean when they say ‘profit’. It could be that zero or negative economic profit corresponds to an enormous accounting profit.

In theory, if corporation tax were a tax on economic profit, it would create zero distortion. Profitable activities would remain profitable. Such a tax would also have its entire burden fall on the owners of the capital; none would be passed on to workers. But corporation tax is a tax on accounting profit, which does not map cleanly on to any economic concept.

Corporation tax is partly a tax on actual economic profit. But it’s also partly a Harberger-style tax¹⁴ on corporate income, to the extent that the tax falls on the normal return to capital invested by companies. It’s also cut from the same cloth as Diamond-Mirrlees taxes, insofar as it directly influences decisions about production, not just decisions about what we should consume.

In the final chapter of The General Theory of Employment, Interest and Money, John Maynard Keynes famously wrote that “The ideas of economists [...] both when they are right and when they are wrong, are more powerful than is commonly understood.” After Harberger and Diamond-Mirrlees, the Atkinson-Stiglitz and Chamley-Judd results strengthened economists’ hostility toward taxing the normal return to capital. That might be part of the reason why, from the 1980s to the 2000s, the average corporate tax rate across the OECD fell from 46 to 31 percent.

Pace the great Keynes, my own feeling is that this resulted more from emergent dynamics of international tax competition, rather than any high-minded economic theory…

I was in Brussels two weeks ago, meeting and chatting with various EU policymakers. When I told a think tanker about a blog post I was working on, he teased me: “Why are the Irish asking so many questions about your magic money machine?”. I told him that it wasn’t clear to me whether anyone actually understood how the magic money machine works. One day, perhaps quite suddenly, the machine might turn off. We won’t understand who benefits, who loses, and why. That’s a scary thought.

Sam Enright is Innovation Policy Lead at Progress Ireland, and editor-in-chief of The Fitzwilliam. You can email him at sam@progressireland.org. He also runs a bounty system to find and catalogue mistakes he has made.

1. There is some useful discussion on the Irish Fiscal Advisory Council blog about how much ‘churn’ there is within this set of largest tax payers. It is not officially reported which companies are in this group, but the largest three are widely believed to be Apple, Microsoft, and Eli Lilly. This cannot be confirmed because of taxpayer confidentiality law. Also, because one ‘company’ may be structured into many subsidiaries, IFAC uses the term ‘group’; I am speaking loosely above.

2. The term ‘passive income’ is frustratingly unspecific. There is no great explainer online about what counts as passive income, but as far as I can tell, the 25% rate applies to any revenue stream that a business received from something other than the ‘excepted trade’ as defined by the Taxes Consolidation Act of 1997. There are many fewer exemptions and tax credits that apply to passive income than to trading income.

3. Notably, the US has not implemented Pillar Two. It’s striking how many times it’s happened that Americans have designed an international system for which they are envisaged to be a central component, and then don’t end up joining because of domestic political squabbling.

4. To take an example: if a business has equal trading profits and passive income, and it also benefited from an R&D tax credit reducing the rate on its trading income by one percentage point, then its effective rate would be (12.5 – 1) * 0.5 + (25) * 0.5 = 18.25 per cent.

However, tax credits sit outside the Pillar Two effective rate calculation if they are refundable. A refundable tax credit is one that is paid out in a grant, rather than as a change in the rate. It is now more common for businesses to pay the full 12.5% rate of tax on profits, and then have some of that money given right back to them in the form of a grant. The introduction of Pillar Two caused a major shift in Ireland, and some other countries, toward refundable and away from non-refundable credits.

Given that policy adjusted specifically to make it less likely that the newly agreed 15% rate is actually a binding constraint, it’s still a matter of debate whether the OECD agreement is toothless. Arguably, the “race to the bottom” in global tax policy has just shifted, away from offering lower tax rates and toward offering greater subsidies. Whether this represents an improvement is left as an exercise to the reader.

5. In fact, many Irish businesses pay no corporate tax at all. Very little was understood about these companies, until a working paper reviewing these ‘nil liable’ companies was circulated by the Statistical and Social Inquiry Society of Ireland. Hint: It’s a lot of aircraft leasing.

6. Fuest et al., page 393.

7. Fuest et al., page 393.

8. Fuest et al., page 396. This was on the basis of work by Thomas Piketty and Emmanuel Saez

9. Although, over the long run, Irish labour has arguably been the most mobile in the world!

10. See page 409 of Fuest et al.

11. For a useful explainer of the concept of deadweight loss, read this post from Matt Yglesias’s Substack.

12. One of the assumptions is that the government can’t use lump sum transfers as a way to raise revenue. From the standpoint of minimising distortion, the best tax code would be something like “every man, woman, and child pays €1,000 in tax”. This would not incentivise anyone to change their behaviour. The reason why nobody advocates lump sum taxes is that they are unacceptable from the standpoint of social equality and basic ethics. A small number of fringe libertarians have defended the idea of actually doing lump sum taxes in practice, known as a poll tax.

13. The Mirrlees review, identifying characteristics of a good tax system for the UK government, is particularly worth reading.

14. I had to stop myself many times from calling this a ‘Harberger tax’, which is a very different and equally fascinating concept.