SK531 Economics of the rm, Diderik Lund, 8 September 2000 Taxation under uncertainty, Varian How do various tax regimes aect investment decisions when there is uncertainty? Two-period models only: Invest today, uncertain result next period Will show: Tax may lead to more investment. Paradox? First shown by Domar and Musgrave in 1944 article Warning: Economics literature has many dierent descriptions of { investment possibilities { tax system { preferences { equlibrium eects Thus many dierent results 1 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Portfolio choice without tax: Riskless vs. risky investment (Varian p. 227{228, originally Arrow (1965)) One risk averse individual maximizes expected utility A given wealth w divided between risky (x 0) and riskless (w , x 0) investment Risky rate of return r~ for x, only two possible outcomes, rg > rb Pr(rg ) = No rate of return for w , x. For consumption in second period: V~ = w , x + x(1 + r~) = w + xr~ Let Vg w + xrg ; Vb w + xrb, and maximize u(Vg ) + (1 , )u(Vb) Results (independent of closer specication of u function, but assume risk aversion): Choose corner solution x = 0 if and only if E (~r) < 0 When solution is interior (x), it is characterized by u0(w + xrg )rg + (1 , )u0(w + xrb)rb = 0 2 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Portfolio choice with tax: Riskless vs. risky investment (Varian p. 227{229, originally Domar and Musgrave (1944) and (with expected utility) Mossin (1968)) Same model as before, but tax on rate of return, tr~. Consumption in second period (w , x) + x(1 + r~) , txr~ = w + x(1 , t)~r: All of w remains untaxed. Choice consists in scaling the additional x(1 , t)~r. First order condition for interior solution, x^: u0(w+^x(1,t)rg )(1,t)rg +(1, )u0(w+^x(1,t)rb )(1,t)rb = 0 Simplied: u0(w + x^(1 , t)rg )rg + (1 , )u0(w + x^(1 , t)rb )rb = 0 Result: Can show that x x^ = 1,t When introduced into rst order condition, x x 0 0 u (w + (1 , t)rg )rg + (1 , )u (w + (1 , t)rb )rb = 0 1,t 1,t this becomes identical to the rst order condition without tax 3 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Portfolio choice with tax: Riskless vs. risky (contd.) x is solution without tax, independent of tax rate x^ = x=(1 , t) is solution with tax We then nd x x^ dx^ = = > 0: dt (1 , t)2 1 , t Paradox(?): Higher tax rate on risky rate of return leads to choice of higher x, i.e., more invested in risky alternative. Interpretation: By scaling up x by the factor 1,1 t , the same rate of return after tax is achieved as in the model without tax: xr~ = x^(1 , t)~r This is exactly optimal, since it is the optimal addition to w, chosen in both models. The choice is how to scale the risky addition to w, and the solution is the same in both models. More complicated if w , x has some riskless rate of return > 0, subject to the same tax rate. Still possible that dx^=dt > 0 holds. 4 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Underutilization of depreciation deductions Aarbu and MacKie-Mason Observe an unexpected phenomenon Review a number of (own and others') potential explanations Translate explanations into testable hypotheses Use very detailed data for testing What are depreciation deductions? Capital equipment with substantial value and durability (ex- ceeding NOK 10 000 and three years, respectively) may not be deducted as expenses at time of acquisition, but as depreciation In accounting regulations: \Book" depreciation, dB General rule: Let dB reect loss of economic value over time Dierent rules employed: dB exponentially decreasing, or constant General rule: Sum of dB over time equals acquisition value In tax regulations: Tax depreciation, dT Usually dB dT in the rst years, tax favorization of investment, but converse may occur 5 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Maximum present value of deductions? Firm wants to pay minimum tax, calculated as present value Delaying a deduction: Loss of present value Thus: Want tax depreciation as soon as possible But observe exceptions | why? (Precondition: The rm has a choice) Tax law prescribes maximum yearly percentage depreciation (xed percentage of remaining value, saldoavskrivning) When fully utilized: Remaining value falls exponentially When fully utilized: Deduction falls exponentially If less than full utilization one year: Maximum percentage next year is the same Thus: Underutilitation gives more than one year's delay of deduction 6 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Underutilization of depreciation deduction in Norway: Data Year 1988 1991 1992 1993 Sample size 1700 4400 4300 2400 Average percentage underutilization 10 21 8 9 Percentage of corporations underutilizing 39 44 25 20 Data from tax returns of corporations Relative underutilization: allowed , actual depr. deductions actual Similar phenomenon in Sweden and Finland (Why not elsewhere? Depreciation deductions compulsory?) Dierence: In Sweden and Finland mostly in large corporations, in Norway mostly in small 7 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Possible explanations for underutilization 1. Higher book prots allow higher dividend distribution 2. Bank debt is preferred to tax debt because banks monitor the corporation 3. Deducting losses carried forward is preferred to faster depreciation deductions 4. Window-dressing 5. Administrative costs of complicated accounting Regression analysis to test these ve hypotheses Observational unit: Norwegian corporation, sample Left-hand side (independent) variable: Underutilization 16 regressions: Separately for each of the four years, and for each of four size categories Linear regression, one or more explanatory variables for each hypothesis 8 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Notation: = operating surplus (= revenues , operating cost) rB = net interest expenses T = corporate income tax I = investment (expenses on capital equipment) F = increase in consolidation fund R = increase in reserve fund = increase in free equity D = dividend Three types of surplus after tax: Cash surplus: Book prots: , rB , T , I , rB , T , dB Taxable prots, minus taxes: , rB , T , dT = D + R + F + (Equation (5) in Aarbu and MacKie-Mason) Book prots can be increased by lowering dB , may imply lower dT (If = 0:) Lower dT allows higher D 9 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Increased book and/or taxable prots allows increased dividends Presumes the rm is willing to pay something (higher present value of taxes) to increase dividends, cf. Bernheim and Wantz Firm may have cash available for dividends without corresponding book prots Dividends limited by book or taxable prots Hypothesis: When equation (5) is a binding limitation: Higher tendency to underutilize depreciation Limitation disappears with 1992 tax reform Explanatory variable: Dummy = 1 when = 0 Expect positive coecient for 1988 and 1991 Result: No clear support for the hypothesis 10 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Bank debt preferred to interest free tax debt because banks monitor the rm Higher depreciation now implies deferred tax, a kind of interest free debt This may explain underutilization: The alternative, interest carrying debt, gives some advantages Advantage (for shareholders): Banks monitor the rm Observable variables: (i) Increase in long term bank debt (ii) Accumulated interest free debt (deferred taxes) Hypothesis: Lower tendency to underutilize if rm simultaneously increases its long term debt (| printing error in table 2, see instead section 5.3) Hypothesis: Higher tendency to underutilize if rm has previously accumulated much interest free debt Result: No clear support for any of the hypotheses 11 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Carry-forward of losses preferred to depreciation Equation (5): Neglects carry-forward of losses If rm had negative taxable prots previous year: Carry-forward of loss to this year, deductible in taxable prots May be accumulated: If also loss this year (after deduction for previous year's loss): Carry forward one more year Limitation: No more than ten years carry-forward When negative taxable prots this year, and carry-forward from previous years: Trade-o between using full depreciation and deduction of loss carried forward If only present value of tax payments counts: { Use maximum depreciation if no danger for hitting ten-year limitation { Use less than maximum depreciation if high probability for hitting ten-year limitation Observe dummy variable: Is rm in tax position (i.e., is taxable prots strictly positive after all claimed deductions)? Hypothesis: Higher tendency for underutilization when rm is in tax position Result: Clearly signicant in all 16 regressions 12 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Window-dressing Outsiders, both creditors and the stock market, have limited capacity to monitor rm May therefore emphasize accounting numbers, even when these do not reect economic result May emphasize positive prots, even if rm loses in the long run from underutilization of depreciation deductions Observable variables: (i) After-tax prots as if full depreciation were claimed, (ii) Financial strength (indebtedness, etc.) Hypothesis: Lower tendency for underutilization when high values for each of these variables Result: Conrmed, in particular for (i) After 1992 reform much of eect disappears for large rms 13 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Administrative costs of complicated book-keeping Simpler to have dT = dB , but rules allow dT > dB the rst years Observable variable: Accumulated dierence previous year, dummy = 1 when accumulated dierence is zero Hypothesis: Higher tendency for underutilization (i.e., setting dT = dB ) if dierence is zero Result: Conrmed, in particular for small rms 14 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Summing up the results These hypotheses are conrmed: Deduction of losses carried forward is preferred to depreciation Window-dressing Administrative costs of complicated book-keeping These hypotheses are not conrmed: Increased book and/or taxable prots allow increased dividends Bank debt is preferred to taxable debt because banks monitor the rm 15 SK531 Economics of the rm, Diderik Lund, 8 September 2000 Econometric method Estimating linear equations Hypothesis: Linear function of explanatory variables, plus nor- mally distributed error term, gives the desired underutilization Problem: Underutilitzation limited above by the maximum allowed depreciation, and below by zero, but right-hand side of regression will have values outside this range Solution: Assume right-hand side (linear function pluss error term) gives the rm's desired underutilization But: When desired underutilization is outside allowed range, the actual will be just inside, i.e., maximum or zero, respectively Need separate estimation method 16