Econ 551 Government Finance: Revenues Winter 2018

Transcription

1 Econ 551 Government Finance: Revenues Winter 2018 Given by Kevin Milligan Vancouver School of Economics University of British Columbia Lecture 9b: Dividend Taxation ECON551: Lecture 9b 1

2 Agenda 1. Overview 2. Different ways of taxing dividends 3. Theories of dividend taxation 4. Evidence: Auerbach and Hassett; Chetty and Saez ECON551: Lecture 9b 2

3 Overview Last lecture, we talked about different ways of measuring corporate profits. But, once firms generate profit, what should be done how get that value back to those who provided capital? Pay bond or bank loan interest if you borrowed money to fund the company. Pay dividends to shareholders if you raised funds by selling shares. Buy back shares with the profit. (Should induce capital gain to shareholders.) We will focus on the 2 nd of these methods paying dividends. There are two distinct questions that can be asked 1. Do existing shareholders care which method is used? 2. Do ongoing firm investment decisions depend on taxation / is cost of capital affected? ECON551: Lecture 9b 3

4 Accounting for dividends Before getting into the math and the theory, let s review quickly how the accounting works. Item Amount Revenue 100 Cost of Goods Sold 60 Operating Income 40 Interest payments 10 Profit 30 CIT 30% 9 After-tax profit 21 Dividends 15 Retained earnings 6 So, the basic point is that dividends will be coming out of after-tax profits. That means that any tax on dividend incomes will be on funds that have already paid corporate taxes. ECON551: Lecture 9b 4

5 Dividend Tax Credit Example Ontario 2016 Capital Wages/ rate gains interest Dividends Corporate profit before tax $ $ $ Corporate tax paid 26% $ $ Amount available for distribution $ $ $ dividend gross up 138% $ capital gain inclusion 50% $ taxable personal income $ $ $ Personal income tax (ON+FED) 53.53% $ $ $ Dividend tax credit (ON+FED) 25.02% $ personal taxes paid $ $ $ total taxes paid $ $ $ ECON551: Lecture 9b 5

6 80% 70% Total Taxation of $1 of Pre-Tax Corporate Earnings Distributed Three Different Ways Ontario, high-earner 60% 50% 40% 30% 20% 10% Prepared by: Kevin Milligan Vancouver School of Economics Dividends Wage/Interest Capital Gains Gap: Dividends less Interest Gap: Dividends less Capital Gains 0% -10% ECON551: Lecture 9b 6

7 Trends in dividend income: dollar value of dividend income Source: CRA Income Statistics ECON551: Lecture 9b 7

8 Trends in dividend income: proportion with any dividend income Source: CRA Income Statistics ECON551: Lecture 9b 8

9 Trends in dividend income: share of total income Source: CRA Income Statistics ECON551: Lecture 9b 9

10 Agenda 1. Overview 2. Different ways of taxing dividends 3. Theories of dividend taxation 4. Evidence: Auerbach and Hassett; Chetty and Saez ECON551: Lecture 9b 10

11 Taxation of dividend income First, we ll talk about how dividends are actually taxed in Canada and other countries. After that, we ll take a look at the different theories of dividend taxation. Finally, we ll look at some recent empirical evidence on dividends that might help us to distinguish between the theories. Notation τd τc τp τg dividend tax rate corporate tax rate personal tax rate accrual-equivalent capital gains tax rate Two main ways dividends are treated: Classical systems Integrated systems Let s dig in ECON551: Lecture 9b 11

12 Classical Dividend Taxation In classical systems, dividends are taxed as regular income in the hands of dividend recipients. No recognition is offered for the fact that taxes were already paid at the corporate level. So, for $1 of pre-tax earnings, the shareholder would receive: One argument against this method: Funds are subject to double taxation. Q: Should double taxation concern us? (1 τ c )(1 τ d ) Q: Should dividends face higher or lower taxation than personal income? ECON551: Lecture 9b 12

13 Integrated Dividend Taxation Many countries attempt to integrate the corporate and personal tax systems. Why? Schanz-Haig-Simons comprehensive measure of income: include all $ equally. If you re going to do this, may need to recognize taxes paid at the corporate level. Aim might be to find a τ d such that the following holds: (1 τ c )(1 τ d ) = (1 τ p ). If this can be done, interest income (exempt at corporate level; taxed at τ p at personal level) receives same tax treatment as dividend income. This all depends on the marginal dividend recipient being a taxable Canadian taxfiler. Pension funds and other tax-exempts? International capital market integration? (Boadway and Bruce 1992) ECON551: Lecture 9b 13

14 Three ways to integrate personal-corporate taxes: Complete integration: Personal taxation entirely accounts for the taxes paid (or not) at the corporate level. This can get complicated, as firms might have losses, loss carry-forwards, or could be based in different countries. Taiwan, Australia, and others do this. Corporate fiscal year lines up with personal fiscal year this makes it easier. Notional integration: Takes benchmark corporate tax rate and provides relief based on notional rather than actual corporate taxes. Canada does this. Trouble if the chosen benchmark starts to deviate from reality. ECON551: Lecture 9b 14

15 Three ways to integrate personal-corporate taxes: Split-rate integration: Tax applied at flat rate to dividend income. In Germany, rate is 25%. Common in Europe; Australia almost went this way recently too US has lower, but not flat, tax schedule for dividends. ECON551: Lecture 9b 15

16 ECON551: Lecture 9b 16

17 Combined Corporate+Personal rates in OECD: OECD Publication: Taxation of Dividend, Interest, and Capital Gain Income ECON551: Lecture 9b 17

18 How Dividends are Taxed in Canada: Canada uses a gross up and credit system. Korea uses similar method Australia and New Zealand also do some grossing up. Basic idea: undo corporate taxation and re-tax at personal level. Give a credit approximately equal to corporate tax paid. Gross up the income to a notional amount to reflect pre-corporate tax income. Apply personal tax rate to the grossed-up amount. ECON551: Lecture 9b 18

19 How Dividends are Taxed in Canada: Consider the following situation: Y in corporate income is available for distribution as a dividend. After paying corporate taxes, the following amount D is left: with Yτc paid in taxes at the corporate level. D = Y(1 τ c ) At the personal level, let s imagine that the gross up rate is g and the dividend tax credit is d times the grossed up amount. This leaves personal taxes to look like this: τ p Dg ddg Total taxes paid at the corporate and personal level is therefore: Yτ c + τ p Dg ddg ECON551: Lecture 9b 19

20 How Dividends are Taxed in Canada: Let s assume that the government chooses to set d = τ c and = 1 1 τ c. The total tax paid now looks like this: Yτ c + τ p Dg ddg = Yτ c + τ p D 1 1 τ c τ c D 1 1 τ c = Yτ c + (τ p τ c )D 1 1 τ c Substituting in the definition of D leads us to the final step: Yτ c + (τ p τ c )Y(1 τ c ) 1 1 τ c = Yτ p In other words, the corporate income is effectively taxed at the personal tax rate. ECON551: Lecture 9b 20

21 Comments on Canada s Gross-up and Credit: When this works, income is taxed the same whether as interest, personal income, or dividends, there is no bias in the form of corporate organization (incorporated or not) or corporate finance (through bonds or equity). What τ c to use? Two issues: o CCPCs have preferential tax rate. o Provinces have different corporate tax rates. Need to do it separately by province. Who gets to claim the d? o What about tax-exempts such as RRSPs or pension funds? o What about low income earners / non-taxable returns? o What about foreigners? What about dividends from foreign firms? What about firms that suffer losses? We end up with over-integration! Given opportunities for tax-exempt investing (RRSP, TFSA), why not just replace this with flat rate reflecting top tax bracket full integration level? ECON551: Lecture 9b 21

22 Agenda 1. Overview 2. Different ways of taxing dividends 3. Theories of dividend taxation 4. Evidence: Auerbach and Hassett; Chetty and Saez ECON551: Lecture 9b 22

23 Theories of dividend taxation There are several competing theoretical approaches to understanding the effect of dividend taxation on corporate decisions. Let s think about the following decisions that may be affected by dividend taxation: Financing: do you issue debt or equity to finance new operations? Corporate form: Do you incorporate or remain non-corporate? Income trust or not? Investment: Does dividend tax treatment affect investment? Distributions: Pay out distributions as capital gain or as dividends? Incidence: Who pays for the tax? Four models to consider: 1. Traditional view 2. New view. 3. Tax irrelevance view. 4. Agency view. ECON551: Lecture 9b 23

24 Traditional view The traditional view sees dividends as being paid even though they may be tax disadvantaged. Why would this occur? 1. Signaling (see Bernhein1991). To signal the quality of the firm, managers burn dollars in order to generate a separating equilibrium that differentiates them from firms with bad prospects. 2. Agency problems. Shareholders may prefer dividends because they lack the means to properly monitor what management is doing (see, for example Enron or Worldcom). ECON551: Lecture 9b 24

25 Traditional view If these are the reasons dividends are paid, then dividend taxation has the following effects: Imposition of dividend taxes makes discounted value of stream of income less valuable so the value of existing equity goes down stock tumbles in price. This means that new equity issued will have a higher pre-tax expected return; it now accounts for the fact that returns will be taxed more heavily. Taxes capitalized into prices. Incidence is on existing shareholders when tax imposed. New shareholders are compensated with higher pre-tax returns. Because new investment is financed in part by expensive new equity issuance, investment decisions now face a higher hurdle rate. That is, the dividend tax rate has a negative impact on the cost of capital and therefore on the investment decision. This is only going to matter if marginal investor is a domestic, taxable entity. ECON551: Lecture 9b 25

26 New View This dates from the Auerbach (1979), but is newer than the traditional view Investments are funded first through retained earnings. Why? Asymmetric information; lemons. Investors don t know enough about firm prospects, so cost of raising funds externally is higher than internally. Dividends are paid only when the firm has generated so much income that it has exhausted all investment opportunities within the firm. This means that firms don t have to issue new equity or bonds in order to finance new projects. Existing cash flow is more than sufficient for financing the new projects. This is sometimes also referred to as the trapped equity view. ECON551: Lecture 9b 26

27 New View If true, this has important implications: Dividend taxation still has impact on the valuation of outstanding shares. If taxes go up, value of the shares goes down so that after-tax returns are still acceptable. Incidence of a tax change will be on existing shareholders again. At the margin, new investment comes from cash flow, not from new shareholders. Therefore, dividend taxation is irrelevant to the investment decision. To see this, think about the following mental experiment: Take one dollar away from a shareholder s return (after all taxes) by reducing the dividend. Instead of paying the dividend, invest the retained earnings in the marginal project. Then, pay out the return to the new project as a dividend. What do you get back from that after tax? ECON551: Lecture 9b 27

28 New View Mental experiment, recast: Option 1: Receive one dollar after tax in dividends. Option 2: Firm retains the dividend and invests it in marginal project, with IRR of ρ before tax. Return from project is paid out as dividend. If we assume that the payout from Option 1 is $1, what is the payout from Option 2? 1 Retained amount of dividends is:. (This delivers $1 after tax to shareholder.) 1 τ d Because firm holds the funds, the value of the firm now rises so capital gains tax applies to this increase, leaving 1 τ g 1 τ d Marginal project returns ρ, so after corporate and dividend taxes return is ρ(1 τ c )(1 τ d ) Total return to option 2 is therefore the product of the two parts: Independent of τ d!! 1 τ g 1 τ d ρ(1 τ c )(1 τ d ) = ρ(1 τ c )(1 τ g ) ECON551: Lecture 9b 28

29 Tax Irrelevance View Could dividend tax by irrelevant not just to investment (like in new view) but also to value of firm? What if marginal investor not subject to Canadian dividend taxes? This is a tax clientele argument, based on theories and evidence that suggest that actors facing different tax situations specialize in holding assets beneficial to their situation. Who might this be? Foreigners not subject to taxation of dividend income. Big pension funds / RPPs-RRSPs / tax exempts. Tax arbitrage techniques that relieve the taxation of dividends. Boadway and Bruce (1992) make this theoretical case for a small open economy dividend taxation is irrelevant since the marginal investor is not a domestic taxpayer. ECON551: Lecture 9b 29

30 Tax Irrelevance View Empirical evidence from Mackenzie and Thompson (1995): Look at stock price reaction to dividend tax change in They find a substantial response, the magnitude of which indicates that the marginal investor is a clientele based on high-bracket domestic investors. That was 28 years ago, however, so maybe things are more global now Dividend taxation change In November, 2005 Finance Minister Ralph Goodale announced a change in the gross up and credit rates that lowered the effective tax rate on dividends. Stock market hit five-year high in the next few days. This is consistent with traditional or new views, but not tax irrelevance view. Still, the tax irrelevance view persists Many people who advocate full taxation of dividends (ie a classical system) use the tax irrelevance view as justification. ECON551: Lecture 9b 30

31 Agency View Chetty and Saez (2010) develop an agency model of dividend taxation. Dividend tax cuts of 2003 led to large changes in firms dividend payout policy. This seems consistent with the traditional view since taxes shouldn t change dividend payouts in the new view. However, they find that dividends change too fast in the traditional view the dividend increase should be a result of new induced investment, and so it should lag. They propose a different model that focuses on agency problems: the idea that management does not do a good job of maximizing shareholder returns. If management is keeping too much for itself, then cutting dividend taxes should increase the incentive to pay out larger dividends. This effect should be strongest where a) managers are also shareholders or b) when there are large dominant shareholders who can force management to pay out more. The evidence (which we discuss below) lines up with these predictions. ECON551: Lecture 9b 31

32 A Summary of the Views What is the source of marginal investment funds? Are dividend taxes capitalized into equity prices / do they affect the cost of capital? Are investment decisions affected by dividend taxation? Does monitoring of management matter? Traditional New Tax irrelevance Agency New share Retained Tax-exempts Retained issuance. earnings earnings YES YES NO YES YES NO NO YES NO NO NO YES ECON551: Lecture 9b 32

33 Agenda 1. Overview 2. Different ways of taxing dividends 3. Theories of dividend taxation 4. Evidence: Auerbach and Hassett; Chetty and Saez ECON551: Lecture 9b 33

34 Auerbach and Hassett (JPubE 2002): On the marginal source of investment funds Long and detailed literature review. They look for evidence that a) profit/cash flow and b) investment has some impact on dividends. o Under traditional view, more profit will lead to more dividends of course, but investment should not be coming out of dividends. o Under new view, dividends should go up as profits do, but dividends should go down when investment goes up, since investment being funded by foregone dividends. ECON551: Lecture 9b 34

35 ECON551: Lecture 9b 35

36 Comments: Investment one period before predicts a decrease in dividends. Some support here for the new view. cash flow lagged once predicts an increase in dividends. Effect is stronger for the more mature firms (eliminates dividend never-payers). IV specification uses contemporaneous cash flow and investment, instrumented by lags on those same variables. Coefficients are much larger. They proceed to look more deeply at which firms are driving the results. Is the finding consistent across all types of firms? ECON551: Lecture 9b 36

37 (3 rd column) Firms with analysts covering them and that have bond ratings do not display sensitivity of dividends to investment. This suggests that they may be using bonds to fund their investment, so investment decision / cost of capital is not affected by dividend taxation for these firms. ECON551: Lecture 9b 37

38 What about firms who issue new equity? 3 rd column: firms that have a high probability of issuing new equity do not show sensitivity of dividends to investment. This is the traditional view these firms are financing new investment with new equity issuance, so their cost of capital depends on the dividend taxes to be paid by these new shareholders. ECON551: Lecture 9b 38

39 Summary of Auerbach and Hassett: They don t look directly for a link from cost of capital to investment. Instead, they look for link between investment and dividends paid more indirect, but still informative. They find strong evidence that investment is financed out of foregone dividends. This supports the new view. However, there is substantial heterogeneity. Think of a spectrum of firms from small, new, and dodgy to big, bond-financed blue chips. Three categories: o The small and dodgy ones have problems accessing capital markets. They don t issue new shares because they don t like the price they get for them. They can t get reasonably priced finance from the bond market either. For them, the best deal is to finance out of cash flow. These are the guys who are new view firms. o In the middle are firms that do finance out of new share issuance. These are the ones most likely to have their cost of capital affected by dividend taxation. o At the other end are the big bond-borrowing blue chips. They don t care about dividend taxes so much because their cost of capital comes not from new share issuance but from bonds. Dividend taxation will have little effect on investment for these firms. ECON551: Lecture 9b 39

40 Chetty and Saez (QJE 2005): Dividend Taxes and Corporate Behavior: Evidence from the 2003 Dividend Tax Cut Tax cut in 2003 lowered rate of taxation on dividends to be close to that of Capgains (although because of deferral etc. that is always hard to nail down.) o Rate cut from 35% to 15%; bottom two tax brackets pay 5%. Study whether change in dividend taxation changes dividend pay-out policy. o Also study how this varies across different principal-agent scenarios to see if there is evidence that retained earnings may not be put to the best use of shareholders; ie that there are agency problems within firms that are increased by dividend taxation. Look at a sample of publicly traded firms in CRSP database and Compustat. Find about a 20% increase in total dividends paid, but since only about 23% of firms pay regular dividend it might be more informative to look at the extensive margin. o Initiations surged from 4.3/quarter to 29 per quarter; also big increase in firms increasing dividends. ECON551: Lecture 9b 40

41 Big jump in dividend initiations in 2003 ECON551: Lecture 9b 41

42 Compare to tax-exempts as control group Identify firm as control if largest institutional holder is a tax-exempt. Almost no change in initiations for these firms. This suggests that the measured effect is causal. ECON551: Lecture 9b 42

43 Could principal-agent issues be at play here? They look at two hypotheses for figuring out whether principal-agent issues contributed to this decision. Some firms have senior management with big stock option position taxed as CGs. Other firms have management with direct stock holdings. Were the firms with stock-heavy management more likely to increase dividends in face of tax change? Ie made decision based on their own interest? Some firms have shareholders who are tax exempts and some shareholders who are tax-paying. Were the firms with big powerful tax-paying stockholders more responsive to the change? ECON551: Lecture 9b 43

44 When management heave stock holder: initiated at rate 10 times that of heavy option execs. Large taxable shareholders showed a strong response, but not for large tax-exempt shareholders. Also: do share repurchases go down as dividends substitute for them? No evidence of this really. That suggests new dividends are coming out of retained earnings. But evidence here is weak. ECON551: Lecture 9b 44

45 Major findings of Chetty-Saez (2005) 1. Firms dividend policy reacted to dividend change mostly through extensive margin. This might support the traditional view, since dividends are driven by taxes in the old view. 2. Principle-agent concerns appear to be pivotal: Firms with stock-holding execs and tax-paying major shareholders were much more responsive to the change, suggesting that those having power over the firm looked after their own interests more than the interests of all the shareholders. In a way, this sets aside the whole old vs new debate because it says that either way the firm is not acting to maximize shareholder value so any model that presumes firms do so will lead to false conclusions about the economic impact. 3. No strong evidence of substitution of dividends for repurchases, suggesting that new dividends came out of funds that otherwise would have been retained. This is most consistent with new view. ECON551: Lecture 9b 45

46 More recent work: Yagan (2015): Uses 2003 dividend tax cut in the US. Finds no impact on investment or employee compensation. Suggests the cost of capital channel may not be there. Alstadsaeter et al (2017) Look at Swedish change in dividends taxation. Find no change in aggregate investment. But increase in cash-starved firms and decrease in cash-rich firms. Provides evidence in favour of cash-flow dependent models. ECON551: Lecture 9b 46