Taxes and the Fed: Theory and Evidence from Equities

Transcription

1 Taxes and the Fed: Theory and Evidence from Equities November 5, 217 The analysis and conclusions set forth are those of the author and do not indicate concurrence by other members of the research staff or the Board of Governors.

2 Introduction What are the effects of tax changes on equity markets?

3 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad.

4 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies:

5 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies: Sialm (25, 26, 29): Endowment Economies and Time Series

6 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies: Sialm (25, 26, 29): Endowment Economies and Time Series McGrattan and Prescott (25): RBC Model

7 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies: Sialm (25, 26, 29): Endowment Economies and Time Series McGrattan and Prescott (25): RBC Model In contrast, good news is not always good for stock market.

8 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies: Sialm (25, 26, 29): Endowment Economies and Time Series McGrattan and Prescott (25): RBC Model In contrast, good news is not always good for stock market. Boyd, Hu, Jagannathan (25): Good unemployment news can be bad

9 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies: Sialm (25, 26, 29): Endowment Economies and Time Series McGrattan and Prescott (25): RBC Model In contrast, good news is not always good for stock market. Boyd, Hu, Jagannathan (25): Good unemployment news can be bad McQueen and Roley (1993): Good economic news is bad

10 Introduction What are the effects of tax changes on equity markets? Conventional Wisdom: Tax cuts are good. Tax increases are bad. View shared by general media and multiple studies: Sialm (25, 26, 29): Endowment Economies and Time Series McGrattan and Prescott (25): RBC Model In contrast, good news is not always good for stock market. Boyd, Hu, Jagannathan (25): Good unemployment news can be bad McQueen and Roley (1993): Good economic news is bad Evans and Marshall (29): Exp. labor supply or pref. shock => higher profits => bigger response of rates

11 Intro: Theory I What we do: Take standard New Keynesian Model

12 Intro: Theory I What we do: Take standard New Keynesian Model Use MP rule based on Coibion + Gorodnichenko (212) Also use our own estimated rule for robustness

13 Intro: Theory I What we do: Take standard New Keynesian Model Use MP rule based on Coibion + Gorodnichenko (212) Also use our own estimated rule for robustness Conduct Campbell-Shiller News Decomposition

14 Intro: Theory I What we do: Take standard New Keynesian Model Use MP rule based on Coibion + Gorodnichenko (212) Also use our own estimated rule for robustness Conduct Campbell-Shiller News Decomposition Back out effect of distortionary income tax rate change (1) Cash Flow News (2) Discount Rate News a. Real Interest Rate News b. Future Excess Return News

15 Intro: Theory II Our Findings: For Pre-198 Rule: Response to tax cut Cash Flow News Discount Rate News

16 Intro: Theory II Our Findings: For Pre-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Positive Effect

17 Intro: Theory II Our Findings: For Pre-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Positive Effect For Post-198 Rule: Response to tax cut Cash Flow News Discount Rate News

18 Intro: Theory II Our Findings: For Pre-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Positive Effect For Post-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Negative Effect

19 Intro: Theory II Our Findings: For Pre-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Positive Effect For Post-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Negative Effect Why?

20 Intro: Theory II Our Findings: For Pre-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Positive Effect For Post-198 Rule: Response to tax cut Cash Flow News Discount Rate News Tax Cuts => Negative Effect Why? Fed s greater response to economic activity post-198.

21 Intro: Theory III Confirm similar findings in FRB/US Model. Demonstrate some asymmetric effects coming from downward nominal wage rigidities. Tax cuts have larger effects than tax increases.

22 Intro: Empirics Take Romer and Romer (21) shocks. Use surprise shocks (Mertens and Ravn, 212) Use SVAR surprise shocks (Mertens and Ravn, 213) Conduct CS News Decomposition => Split sample at 198

23 Intro: Empirics Take Romer and Romer (21) shocks. Use surprise shocks (Mertens and Ravn, 212) Use SVAR surprise shocks (Mertens and Ravn, 213) Conduct CS News Decomposition => Split sample at 198 Use Municipal Bond Yield Data from Leeper, Walker, and Yang (213) and Kueng (215) Use Sialm (29) Annual Data Find that tax increases => higher return for post-198.

24 Households Households choose state contingent c t, 1 h t to maximize with budget constraint E β t (cι t (1 h t) 1 ι ) 1 σ 1 t= 1 σ b t + c t + i t + τ t = R t 1 b t 1 + (1 τt D )(w t h t + u t k t + Φ t ) + δq t τt D k t π t and evolution of capital ( ) it k t+1 = (1 δ)k t + i t Ψ t i t 1 Calibration: IES =.2, ι =.25, β =.999, δ =.25

25 Labor and Wages Downward nominal wage rigidities: Kim & Ruge-Murcia (29) ( exp( φ(w Φ n t = Φ(Wt n /Wt 1 n n ) = φ t /Wt 1 n 1)) + φ(w n t /W t 1 n ) 1 ) Linex asymmetric adjustment cost function in nominal wages φ 2

26 Labor and Wages Downward nominal wage rigidities: Kim & Ruge-Murcia (29) ( exp( φ(w Φ n t = Φ(Wt n /Wt 1 n n ) = φ t /Wt 1 n 1)) + φ(w n t /W t 1 n ) 1 ) Linex asymmetric adjustment cost function in nominal wages Costs from wage decreases rise exponentially. Costs from wage increases rise linearly. φ 2

27 Labor and Wages Downward nominal wage rigidities: Kim & Ruge-Murcia (29) ( exp( φ(w Φ n t = Φ(Wt n /Wt 1 n n ) = φ t /Wt 1 n 1)) + φ(w n t /W t 1 n ) 1 ) Linex asymmetric adjustment cost function in nominal wages Costs from wage decreases rise exponentially. Costs from wage increases rise linearly. Households choose equate marginal costs and benefits of increasing W n where ω t = W n t /W n t 1. t ω t Φ = θ 1 ι ι φ 2 [ ] c t 1 leis t w t (1 τt d ) (θ 1)(1 Φ) + E h t m t+1 t,t+1 w h t+1 ω t+1 Φ t+1 t Calibration: Φ = 1 (Wage Adj. Costs), Ψ = 38 (Asymmetric)

28 Firms Firms maximize discounted value of profits subject to price adjustment costs max E t P t (j) t= s= where nominal profits are defined as Q t,t+s Ξ t+s Ξ t = P t (j)y t (j) mc t Y t (j)p t φ 2 which yields non-linear Phillips curve (1 η) + ηmc t φ π ( ) t πt 1 π ss π ss + φβe t ( ct+1 c t ) ι(1 1 ψ ) 1 ( 1 h t+1 1 h t ( ) 2 πt 1 Y t P t π ss ) (1 ι)(1 1 ψ ) ( ) y t+1 π t+1 πt+1 1 = y t π ss π ss Calibration: φ = 35 (Price Adj. Costs), η = 7 (Elasticity of Demand)

29 Government Fiscal budget constraint b t = R t 1 b t 1 + g t τ t π t τ t = τ D t y t + τ L t Tax shock analysis τ D t u tax t = τ D, + u tax t = ρ τ ut 1 tax + εtax t Calibration: ρ τ =.9999, σ tax =.22

30 Monetary Policy Monetary Policy Rule ln(r t /R ) = α r ln(r t 1 /R )+(1 α r ) [ α π ln(π t /π ) + α y ln( y t / y ) ] Pre-Volcker: α r =.91, α π = 1.32, α y =.94 Post-Volcker: α r =.9, α π = 1.58, α y = 2.21 Based on Coibion & Gorodnichenko and our updated estimates

31 Return I We define the return as follows: where and R DIV t+1 = P t+1 + D t+1 P t [ ] P t = E t M t+1 R DIV t+1 D t = Y t w t L t I t

32 Return II For the sake of our theoretical income tax experiments, we modify the cash flow in the return equation such that ( ) Dt = 1 τd,t D t where τd,t = γ τd t Follow Sialm (29) and use FED24 Flow of Funds => γ =.55 =>55% of equity is held by taxable investors.

33 Cash Flow vs Discount Rate Channels Income Tax Rate Increase Cash Flow Only Scenario: P 1,t = M SS E t [(P t+1 + D t+1 )] Discount Rate Only Scenario: P 2,t = E t [M t+1 (P t+1 + D SS )] Post Volcker: Excess Equity Return.4 Pre Volcker: Excess Equity Return.2.2 Cash Flow Only Cash Flow Only Dividends Output Why? Fed s greater response to economic activity post

34 Cash Flow vs Discount Rate Channels Income Tax Rate Increase Cash Flow Only Scenario: P 1,t = M SS E t [(P t+1 + D t+1 )] Discount Rate Only Scenario: P 2,t = E t [M t+1 (P t+1 + D SS )] Post Volcker: Excess Equity Return.4 Pre Volcker: Excess Equity Return.2 Discount Rate Only Cash Flow Only.2 Discount Rate Only Cash Flow Only Dividends Output Why? Fed s greater response to economic activity post

35 Cash Flow vs Discount Rate Channels Income Tax Rate Increase Cash Flow Only Scenario: P 1,t = M SS E t [(P t+1 + D t+1 )] Discount Rate Only Scenario: P 2,t = E t [M t+1 (P t+1 + D SS )].2.2 Why?.1 Post Volcker: Excess Equity Return Discount Rate Only Cash Flow Only Both Dividends Pre Volcker: Excess Equity Return Discount Rate Only Cash Flow Only Both Output.2.3.5

36 Cash Flow vs Discount Rate Channels Income Tax Rate Increase Cash Flow Only Scenario: P 1,t = M SS E t [(P t+1 + D t+1 )] Discount Rate Only Scenario: P 2,t = E t [M t+1 (P t+1 + D SS )].2.2 Post Volcker: Excess Equity Return Discount Rate Only Cash Flow Only Both Dividends Pre Volcker: Excess Equity Return Discount Rate Only Cash Flow Only Both Output Why? Fed s greater response to economic activity post

37 RBC.1 Excess Equity Return.5 Investment.5.5 RBC Model Dividends Output Real Interest Rate Consumption Income Tax Rate Increase: RBC Greater response No MP: Dividends more, real rates less

38 RBC vs Post-Volcker.1 Excess Equity Return.5 Investment.5.5 NK: Post Volcker, RBC Model Dividends Output Real Interest Rate Consumption Income Tax Rate Increase: Post-Volcker Greater response by MP: Dividends less, real rates more

39 RBC vs Post- vs Pre-Volcker.1 Excess Equity Return.5 Investment.5.5 NK: Post Volcker, NK: Pre Volcker RBC Model Dividends Output Real Interest Rate Consumption Income Tax Rate Increase: Pre-Volcker Smaller response by MP: Dividends more, real rates less

40 RBC vs Post- vs Pre-Volcker Channels Post-Volcker: Discount Rate news dominates RBC and Pre-Volcker: Cash Flow news dominates

41 RBC vs Post- vs Pre-Volcker Channels Post-Volcker: Discount Rate news dominates RBC and Pre-Volcker: Cash Flow news dominates

42 RBC vs Post- vs Pre-Volcker Channels Post-Volcker: Discount Rate news dominates RBC and Pre-Volcker: Cash Flow news dominates

43 RBC vs Post- vs Pre-Volcker Channels Post-Volcker: Discount Rate news dominates RBC and Pre-Volcker: Cash Flow news dominates

44 RBC vs Post- vs Pre-Volcker Channels Post-Volcker: Discount Rate news dominates RBC and Pre-Volcker: Cash Flow news dominates

45 Robustness I

46 Robustness II

47 Robustness III

48 Robustness IV

49 .1 With Downward Nominal Wage Rigidities Excess Equity Return Output Tax Inc W Labor (H) Investment Dividends Tax cut => Labor Supply rises but wages fall less (due to DWNR) Labor Demanded rises less => Output rises less => Dividends rise less => Less Positive Cash Flow News => Lower

50 .1 With Downward Nominal Wage Rigidities Excess Equity Return Output Tax Inc. Tax Cut W Labor (H) Investment Dividends Tax cut => Labor Supply rises but wages fall less (due to DWNR) Labor Demanded rises less => Output rises less => Dividends rise less => Less Positive Cash Flow News => Lower

51 .1 With Downward Nominal Wage Rigidities Excess Equity Return Output Tax Inc. Tax Cut Difference* W Labor (H) Investment Dividends Tax cut => Labor Supply rises but wages fall less (due to DWNR) Labor Demanded rises less => Output rises less => Dividends rise less => Less Positive Cash Flow News => Lower return

52 .1 With Downward Nominal Wage Rigidities Excess Equity Return Output Tax Inc. Tax Cut Difference* W Labor (H) Investment Dividends Tax cut => Labor Supply rises but wages fall less (due to DWNR) Labor Demanded rises less => Output rises less => Dividends rise less => Less Positive Cash Flow News => Lower return

53 .1 With Downward Nominal Wage Rigidities Excess Equity Return Output Tax Inc. Tax Cut Difference* W Labor (H) Investment Dividends Tax cut => Labor Supply rises but wages fall less (due to DWNR) Labor Demanded rises less => Output rises less => Dividends rise less => Less Positive Cash Flow News => Lower return

54 .1 No Downward Nominal Wage Rigidities Excess Equity Return Output Tax Inc. Tax Cut Difference* W Labor (H) Investment Dividends No downward nominal wage rigidity => No asymmetric effect Labor Demanded rises less => Output rises less => Dividends rise less => Less Positive Cash Flow News => Lower

55 FRB/US Model: Tax Increase Large-scale model of U.S. economy used by Federal Reserve We increase nominal personal tax receipts (TFPN) by.22% of taxable income (YPN GFTN GSTN) for ten years. Cash flow news declines by.7% Discount rate news declines by.98% Current excess return rises.28%

56 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks

57 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks Mertens and Ravn (212) => Split shocks up Anticipated vs Unanticipated

58 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks Mertens and Ravn (212) => Split shocks up Anticipated vs Unanticipated Mertens and Ravn (214) => Address measurement errors Due to necessary judgment calls in construction of narratives Censoring issues when changes deemed revenue neutral Discrepancies between actual changes and projected changes

59 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks Mertens and Ravn (212) => Split shocks up Anticipated vs Unanticipated Mertens and Ravn (214) => Address measurement errors Due to necessary judgment calls in construction of narratives Censoring issues when changes deemed revenue neutral Discrepancies between actual changes and projected changes To arrive at Proxy SVAR structural shocks:

60 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks Mertens and Ravn (212) => Split shocks up Anticipated vs Unanticipated Mertens and Ravn (214) => Address measurement errors Due to necessary judgment calls in construction of narratives Censoring issues when changes deemed revenue neutral Discrepancies between actual changes and projected changes To arrive at Proxy SVAR structural shocks: (1) Obtain reduced form residuals from VAR of T t, G t, Y t.

61 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks Mertens and Ravn (212) => Split shocks up Anticipated vs Unanticipated Mertens and Ravn (214) => Address measurement errors Due to necessary judgment calls in construction of narratives Censoring issues when changes deemed revenue neutral Discrepancies between actual changes and projected changes To arrive at Proxy SVAR structural shocks: (1) Obtain reduced form residuals from VAR of T t, G t, Y t. (2) Regress reduced form residuals on narrative shock series.

62 Empirics: Shocks Romer and Romer (21)=> narrative based shocks. 5 exogenous federal tax shocks Mertens and Ravn (212) => Split shocks up Anticipated vs Unanticipated Mertens and Ravn (214) => Address measurement errors Due to necessary judgment calls in construction of narratives Censoring issues when changes deemed revenue neutral Discrepancies between actual changes and projected changes To arrive at Proxy SVAR structural shocks: (1) Obtain reduced form residuals from VAR of T t, G t, Y t. (2) Regress reduced form residuals on narrative shock series. (3) Transform predicted values according to BP (22)

63 Shocks: Exogeneity Run ordered probit with state variables from CS News Decomp. BIC Optimal Lag = 1 Z t [r ex,t, r real,t, r t, SPREAD t, d t p t, REL t ] All Shocks SVAR Shocks Surprise Shocks Surprise SVAR Shocks Panel A: 198Q3-27Q4 Lag Lags Panel B: 1947Q1-198Q2 Lag Lags Panel C: 1947Q1-27Q4 Lag Lags Fail to reject null of no predictability for three out of the four shocks.

64 Shocks: Exogeneity Run ordered probit with state variables from CS News Decomp. BIC Optimal Lag = 1 Z t [r ex,t, r real,t, r t, SPREAD t, d t p t, REL t ] All Shocks SVAR Shocks Surprise Shocks Surprise SVAR Shocks Panel A: 198Q3-27Q4 Lag Lags Panel B: 1947Q1-198Q2 Lag Lags Panel C: 1947Q1-27Q4 Lag Lags Fail to reject null of no predictability for three out of the four shocks.

65 News Decomposition Post-198: Discount Rate News dominates Cash Flow News Tax increase associated with positive effect on current return

66 News Decomposition Post-198: Discount Rate News dominates Cash Flow News Tax increase associated with positive effect on current return

67 News Decomposition Post-198: Discount Rate News dominates Cash Flow News Tax increase associated with positive effect on current return

68 News Decomposition Pre-198: Cash Flow News dominates Discount Rate News Tax increase associated with negative effect on current return

69 Additional Results We also show the following: (1) personal income tax shocks (Mertens and Ravn, 213) yield similar results (2) tax cuts tend to have larger and more significant effects than tax increases.

70 Sialm (29) Results Sialm (29) constructs annual tax yield series: => Negative relationship between taxes & equity valuations Splits sample at halfway point, 196, shows robustness.

71 Sialm (29) Results Sialm (29) constructs an annual tax yield series: => Negative relationship between taxes & equity valuations Splits sample at halfway point, 196, shows robustness.

72 Sialm (29) Replication We replicate, but split in 198. Find same results for Pre-198, negative relationship

73 Sialm (29) Replication We replicate, but split in 198. Find same results for Pre-198, negative relationship

74 Sialm (29) Replication We replicate, but split in 198. Find different results for Post-198, positive relationship. Consistent with our other findings

75 Higher Frequency Results We use Go-2-Bond Municipal Bond index for yields on municipal bonds, which are exempt from federal taxes. We follow Leeper, Walker, and Yang (213) and Kueng (215) and construct the following spread: τ t = 1 Y t M Y t = T k=t ω k τ e k where Yt M is yield at time t of municipal bond maturing at time T Y t is yield at time t of taxable bond maturing at time T τk e is the expected future tax rate at time k

76 Higher Frequency Results Find positive effect between tax changes and equity returns. 1 % tax increase at 12 week horizon leads to.3438% increase in returns => Quantitatively similar to previously mentioned tax shocks => Surprise SVAR:.22% tax shock led to.7% => 1% shock.35%

77 Higher Frequency Results Find positive effect between tax changes and equity returns. 1 % tax increase at 12 week horizon leads to.3438% increase in returns => Quantitatively similar to previously mentioned tax shocks => Surprise SVAR:.22% tax shock led to.7% => 1% shock.35%

78 Conclusion Fed plays important role in effect of taxes on equity markets. Findings motivated with a standard New Keynesian model that exhibits a shift in the aggressiveness of monetary policy. For the Post-198 era, tax cuts lead to higher cash flow news and higher discount rates. The discount rate news dominates Tax cuts are associated with lower equity returns. Our results are confirmed across multiple measures of tax shocks at different frequencies.

79 Romer and Romer

80 Romer and Romer

81 Croce Tax Shock

82 Output Growth (1) Output growth is observable whereas the output gap is a function of two unobserved variables, the natural rate of unemployment and the potential level of output; (2) Real-time output gaps tend to significantly differ from ex-post output gaps (Belke and Klose, 211); (3) Smets and Wouters (27), Orphanides (24), and Campbell, Pflueger, and Viceira (217) find no significance on the output gap for Post-Volcker. =>In contrast, Smets and Wouters (27) do find significance on output growth for Post-Volcker.

83 Output Growth In addition, our own estimates based on Greenbook Data find significance for output growth but not the output gap for the Post-Volcker time period. Furthermore, Coibion and Gorodnichenko (215) find significant changes to the output growth variable across the Pre and Post-Volcker regimes whereas they find no significant difference in the output gap level.

84 MP Controls First, we incorporate the federal funds rate as an additional state variable and find quantitatively similar results. Second, we perform a Hall (1986) and Evans (1992) test by regressing our tax shocks on the monetary policy shocks used in Bernanke and Kuttner (25). =>Specifically, we regress our tax shocks on four lags of our tax shocks, contemporaneous monetary policy shocks and four lags of monetary policy shocks. =>We fail to reject the hypothesis of exogeneity (p-value =.4649) over the period where both of these shocks are available.

85 News Decomposition I

86 News Decomposition II

87 News Decomposition III