Full Text: DRI/McGraw-Hill's 'Summary of Analysis' Available at JCT Symposium

MODELING THE MACROECONOMIC CONSEQUENCES OF TAX POLICY

JANUARY 1997

TABLE OF CONTENTS

Summary of Lessons Learned

Core Simulation Results

Overview of the DRI/McGraw-Hill Model of The U.S. Economy

Appendix 1: Expectations and "Regime Shift" Changes

Appendix 2: Inflation and Unemployment Impacts on Wage Demands

Appendix 3: Expectations, Wage Determination, Special Flat Tax

Appendix 4: Supplementary Tables and Charts

1. INTRODUCTION: SUMMARY OF MACROECONOMIC LESSONS LEARNED REGARDING

[1] The modeling and simulation exercises organized by the Joint Tax Committee have provided some significant insights into the real challenges and opportunities that will be faced if the current U.S. personal and corporate income tax structures are replaced by a consumption-oriented tax system. Our explorations with the DRI Model, and the contrasts or similarities with results from other models, suggest several conclusions to us.

[2] LONG-RUN BUSINESS CAPITAL SPENDING WOULD BE MATERIALLY ENHANCED BY ALMOST ALL OF THE TAX CHANGE VARIANTS SHIFTING THE U.S. TOWARD A CONSUMPTION-BASED TAX. The ability to immediately write off the cost of new plant and equipment against taxable business income, compared to current provisions allowing only phased depreciation, reduces the effective cost of long-lived investments quite significantly. IN THE LONG RUN, THIS ENCOURAGES MORE CAPITAL- INTENSIVE PRODUCTION AND THUS HIGHER OUTPUT PER EMPLOYEE. Greater capital intensity implies lower labor intensity, but this does not necessarily create unemployment because, other things equal, the Federal Reserve could choose to promote more rapid growth in total output to keep labor fully employed.

[3] STRONGER GROWTH AND AN IMPROVED LIVING STANDARD ARE THE EXPLICIT GOALS OF THE MAJOR TAX PROPOSALS, BUT THESE GOALS ARE MORE DIFFICULT TO ACHIEVE, OR SLOWER TO ARRIVE, THAN PROPONENTS MAY WISH TO ACKNOWLEDGE. Although the effective cost of business investment is promptly reduced by tax changes -- spurring investment -- prolonged weakness in real output due to inflation and transition problems is likely to reduce the need for business capacity -- curtailing investment spending. In the opening one-to-five years of most of the tax change scenarios, the disincentives of weaker capacity requirements tend to exceed the incentives of lower effective costs. Models that ignore or minimize the transition problems will thus tend to produce rosier medium-term scenarios because they do not recognize the short-run losses in capital formation that become longer-lasting losses in national supply potential.

[4] In the very long-run -- perhaps 25 years or more -- both macroeconomic and general equilibrium models will predict higher capital per worker and hence higher productivity; robust members of both classes of models are fundamentally guided by neoclassical models of investment (with the capital intensity determined by the relative costs of labor and capital), similar population and hence labor force projections, and neoclassical production functions combining capital and labor to define potential output.

[5] THE REPLACEMENT OF INCOME TAXATION WITH VALUE-ADDED TAXATION (IN A TRADITIONAL VALUE-ADDED TAX OR A FLAT TAX) HAS THE POTENTIAL TO CREATE A SEVERE INFLATION PROBLEM AS THE TAX IS INTRODUCED. In most variants, the taxes directly collected from the producer/employer rise while the taxes directly collected from the employee/household fall. To preserve or re-establish the required return on invested capital, the producer must pass these new taxes on to consumers in the form of higher prices or pass the taxes back to employees in the form of lower pre-tax wages and benefits. If employees do not immediately agree to pre-tax pay cuts, taking full recognition of the reduced income tax burden to be borne, then a nasty price-wage inflation cycle can be ignited. Many "general equilibrium" models ignore the transition problems, implicitly or explicitly expecting prompt nominal wage reductions, a pattern that does not seem consistent with historical responses to income tax changes.

[6] The productivity benefits of greater capital intensity are slow to arrive, and cannot offset the immediate inflation pressures of new value-added taxes if major wage concession are not forthcoming. Therefore, the third general observation is that the nation's central bank, THE FEDERAL RESERVE, WILL PROBABLY BE CHALLENGED TO EITHER ACCEPT A MUCH HIGHER PRICE LEVEL OR TO IMPOSE HIGHER UNEMPLOYMENT ON THE ECONOMY, or to compromise on an unsatisfactory combination of both of these negatives. "General equilibrium" models typically miss or understate this risk because they are not constructed to deal with short-run price and wage determination. All consistent models will eventually converge over five to ten years to a forecast with the price level tied to the liquidity provided by the central bank, thus higher inflation or even a higher price level is not a necessary long-run outcome of tax substitution. However, the best models will clearly identify and quantify the risk of labor market bargaining failing to produce the required nominal wage concessions unless pressured to do so by higher unemployment. Such economic stagnation will also grossly retard capital formation, delaying or eliminating the growth and productivity benefits potentially stimulated by more favorable tax treatment of long-lived assets.

[7] THE ECONOMY WILL LIKELY BE SUBJECT TO MAJOR WAVES OF SECTORIAL BUYING AND SELLING PRESSURE. A new value-added tax, in the absence of extraordinary wage concessions, will bring higher prices that are easily predictable in advance by the buying public. If the price of any durable good, from an expensive car to a simple box of frozen food, is expected to rise by 10 or 15% in the near future to cover a new tax, then a buy-in-advance mania will be followed by shopping-withdrawal after the tax becomes effective. The American public consistently behaves this way, waiting for bargains at Christmas, for department store clothing markdowns, and for special auto deals. The effect will be more pronounced with a new tax whose price effect and timing are easier to anticipate than current retail promotions. Few if any of the models quantify this shock because there are few historical parallels of the magnitudes under consideration here.

[8] THE PRICES OF KEY ASSETS WOULD CHANGE SUBSTANTIALLY, CREATING MAJOR WINDFALL GAINS AND LOSSES. As a key example, elimination of interest and property tax deductibility for owner- occupied homes would remove a valuable subsidy that is reflected in the price of land and, by extension, residences today. This subsidy raises the market price and quantity consumed, thus its sudden removal would generate huge windfall losses to current homeowners and would depress new construction activity. Market interest rates would not decline by a sufficient margin to buffer this shock. However, the use of scarce national savings to fund inflated land prices is not beneficial to the economy; allocation of these funds to productive investments would be preferable, but a long phase-out of the current benefit is almost certainly necessary. Prices of financial assets would also be subject to sharp shifts, depending on the exact formulation of the new taxes, the Federal Reserve responses, and the outcome of labor market negotiations. The uncertainty surrounding these outcomes, and the volatility of the economy during any transition, would tend to depress many prices until new trends emerge.

[9] IF THERE IS A GENERAL THEME OF CONCERN FLOWING FROM THE SIMULATIONS OF MANY, DIVERSE MODELS, IT IS THAT A THREE-TO-TEN-YEAR TRANSITION PERIOD COULD BE QUITE DIFFICULT TO PREDICT AND EQUALLY DIFFICULT TO COPE WITH AS CONSUMERS, HOME-OWNERS, EMPLOYEES AND EMPLOYERS. A KEY TO THESE PROBLEMS IS THE DECENTRALIZED ECONOMIC DECISION-MAKING IN OUR FREE ENTERPRISE ECONOMY, COUPLED WITH VERY POOR FUNDAMENTAL ECONOMICS EDUCATION.

[10] As an example of the former class of problems -- optimal microeconomic decisions producing problematic macroeconomic outcomes -- employers and employees do not view individual pay or pricing actions as material to the overall price level that is the concern of the Federal Reserve. Such micro decisions in response to a new tax regime will probably not be concerned with the conflict of individual actions tending to create inflation with the ambitions of the Federal Reserve to stabilize prices. Therefore, individuals or businesses facing directly higher taxes will tend promptly to push up prices, and be unable to obtain or wait for offsetting lower costs from those obtaining effective tax cuts in a new regime.

[11] As an example of the latter source of problems and conflict -- weak economies information -- the public and our politicians regularly assume that the burden of a tax is borne by the person or enterprise from whom the tax is directly collected; it is widely assumed that whoever writes the check to the Internal Revenue Service "pays" the tax, whereas the ultimate or true bearer of the tax burden is often quite different. There is little comprehension of how wages and prices are adjusted to shift taxes onto consumers, shareholders, or employees. Therefore the debate shaping any new tax will likely mistakenly focus on how direct tax bills will change for key income classes and voting groups; very little attention will be paid to the real income and asset changes that will occur as the economy settles into a new equilibrium. Very little effort will be made to inform all citizens so that the transition can be made with minimal friction, inflation, or unemployment.

[12] IT CAN BE FORCEFULLY ARGUED THAT THE MAJOR DIFFERENCES IN THE SIMULATIONS FOR ANY GIVEN MODEL OR CLASS OF MODELS REFLECT DIFFERENCES IN ASSUMPTIONS IN THESE KEY AREAS. THE DIFFERENCES DO NOT RELATE SOLELY TO WHETHER ONE MODEL OR CLASS OF MODELS HAS BETTER THEORETICAL PROPERTIES WITH ECONOMIC AGENTS ADHERING TO LOGICAL BEHAVIOR -- ALL MODELS ASSUME RATIONAL BEHAVIOR FOR THE MOST PART. THE DIFFERENCES ARE IN THEIR DESCRIPTIONS OF HUMAN BEHAVIORS BY INDIVIDUALS AND COLLECTIVE GROUPS. Two focal points are the assumed character of pay negotiations and the reaction of the Federal Reserve (or general credit costs) to inflation. As a class, "computable general equilibrium" models tend to assume away many transition problems, somehow expecting a decentralized economy to have perfect foresight and avoid conflict. In contrast, the "econometric" models based an prior historical evidence tend to assume that pay is negotiated on a largely pre-tax basis for the employee, and that price and wage increases flowing from a shock tend to reverberate until general economic stagnation -- perhaps intentionally created by the Federal Reserve to meet its mandate of pursuing price stability -- forces restraint.

2. SUMMARY OF SIMULATION RESULTS

[13] In general, replacing the current tax system with any of the proposed alternatives eventually leads to higher real output. Unfortunately, under every alternative, the path to this long-run improvement contains a period in which economic performance is worse than it would be under current law. This section briefly outlines the reasons for these results.

[14] PRICE IMPACTS: According to microeconomic theory, firms set prices so that the after-tax price (or the after-tax marginal revenue) of a unit of output equals the after-tax marginal cost of producing it. Assuming no first-order change in wage rates or fringe benefits, each alternative tax system raises the after-tax marginal cost of production, and thus boosts prices.

[15] Under current law, all costs of production are deductible from the tax base, so the after-tax cost of production is the pre-tax cost times one minus the tax rate. Under the unified flat income tax, non-wage labor costs (such as fringe benefits) are no longer deductible from the business tax base, so the after-tax cost of using labor rises. This produces a price inflation shock, unless management can convince employees to bear this burden by accepting lower pay or benefits. Under the consumption-based flat tax, the short-run price effect is similar to that of the unified flat income tax: fringe benefits are no longer deductible from the tax base, although both wages and the purchase price of machinery and equipment are deductible. Under a value-added tax, no labor costs can be subtracted from the tax base, so prices must rise substantially.

THE LARGER THE INFLATION SHOCK, THE

[Figure omitted]

[16] The inflation shocks resulting from the changes in tax regimes have substantial adverse effects. The Federal Reserve and private investors will push yields higher across the spectrum of yields. Inflation's demonstrably negative impact on credit costs and consumer attitudes promptly cuts consumer spending on durable goods. In addition, because housing purchases respond to nominal as well as real interest rates, a higher inflation rate reduces residential construction for any given level of real interest rates. These quickly generate negative accelerator consequences for business capital spending -- overwhelming temporarily the tax-related incentives.

THE GREATER THE BOOST IN CREDIT COSTS,

[Figure omitted]

[17] SUPPLY EFFECTS: In each of the alternative tax systems, the ratio of the after-tax cost of capital to the after-tax cost of labor significantly declines. The long-run impact will be a larger capital stock, and thus higher real output than in the baseline as the economy converges back toward full employment of its population. The long-run increase in the capital/labor ratio is largest in the alternative regimes that permit expensing of plant and equipment: the VAT and the consumption-based flat tax.

THE GREATER THE REDUCTION IN THE EFFECTIVE COST OF EQUIPMENT,

[Figure omitted]

HIGHER CAPITAL SPENDING GROWTH

[Figure omitted]

[18] The DRI Model carefully tracks and depreciates investment and capital stocks by type: autos, computers, other business equipment, public utility structures, mining and petroleum structures, government infrastructure, and other private buildings. In addition, business research and development spending is endogenous and treated as a capital stock that influences total factor productivity.

[19] Investment is the key link between tax policy and long-run performance. The full-employment labor force is little affected by the change in tax regimes. The labor force, within the DRI macro- econometric model, is sensitive to: the size of the U.S. adult population, the availability of jobs, and the after-tax real labor compensation. The real compensation term equals total compensation less employee-paid payroll taxes, multiplied by one minus the average marginal income tax rate and then divided by the chain-weighted personal consumption price index. The estimated elasticity of the labor force with respect to real pay is 0.2.

[20] HOUSING: Real home prices can be expected to drop by the capitalized value of lost mortgage interest and property tax deductions: a 15% decline. (Appendix 3 discusses this price shock in greater detail.) This generates a major short-run disruption to construction, as the drop in prices creates large windfall losses for current homeowners and lenders. Expectations of further declines also keep some buyers temporarily out of the market. The home price decline is ameliorated in the VAT tax substitution because all consumer goods prices have shot up, accomplishing part of the required real or relative decline in home prices.

[21] In the long run, the desired number of primary homes should be largely unchanged, although the loss of the huge federal subsidy for owner-occupied dwellings should reduce the outlays per home, the demand for second homes, the preference for owner-occupied versus rental housing, and the price of residential land.

[22] Mortgage rates will be buffeted by a host of influences, ranging from adverse short-run inflation shocks to beneficial long- run increases in national savings. The new tax benefits for business capital spending raises the demand for funds to absorb some of these new national savings. Adding up all these demographic, tax subsidy, and financial factors, residential construction recovers to, and actually exceeds baseline levels within 10 years, partially replacing construction lost during the early years of the tax change.

[23] A more long-lasting effect is the wealth effect from lower home prices. Owner-occupied homes make up a large share of personal wealth, so the 15% drop in prices reduces consumer wealth, and thus consumer purchases. Higher equity prices are not enough to offset this impact.

[24] FEDERAL RESERVE POLICY: The Federal Reserve is an independent agent of economic policy, subject to its own legislated goals and responsibilities. The primary objective is providing price stability, consistent with maximizing the long-run growth potential of the nation. As noted earlier, the new tax regimes would tend to boost short-run prices, antagonizing the Fed, as they attempt to prod the nation to higher investment and hence long-run productivity growth, supporting the Fed's mission.

[25] Because a radical shift of the tax structure has not been previously attempted, the Fed's reaction cannot be known or reliably driven off previous, econometrically-estimated "reaction-functions." Instead, as a reasonable set of initial reactions, DRI assumed that the Fed would increase nominal national banking system liquidity by an amount proportional to any price level shock. In addition, DRI assumed that real monetary reserve growth over the next decade would be raised by 1% (0.1% per year) to accommodate the enhanced supply- side potential of an economy with greater capital formation.

[26] In the first year of the simulations, this still creates higher interest rates because inflation and risk premiums are raised by the new taxes, with only partial offsets in some cases due to different tax treatment of interest income. Thereafter, short-term yields are higher or lower depending on the business cycle situation (i.e. the demand for funds given incomes, inflation, and transactions) and the emerging structural changes in private and public saving behavior. Long-term yields respond similarly, but with more sensitivity to the structural than the cyclical phenomena.

[27] The Joint Committee on Taxation requested all modelers to produce scenarios in which the central bank would adjust its policies such that unemployment rate would be promptly pushed back toward baseline levels after the year 2000. In the case of the VAT and FLAT tax substitutions, this required DRI to add substantial additional monetary reserves to offset the large initial recessionary shocks. This produces short-term interest rates that are exceptionally low, but perhaps not unreasonable in the particular circumstances of helping the nation absorb the repercussions of a major tax overhaul. The simulation results are separately reported.

INCLUDING FEEDBACK EFFECTS, THE VAT WOULD BOOST WHOLESALE PRICES BY

[Figure omitted]

NOMINAL WAGES SLOWLY ADJUST DOWNWARD TO THE SHIFT FROM INCOME TO

[Figure omitted]

INFLATION SHOCKS LEAD TO TEMPORARY BUT LARGE REAL GDP LOSSES.

[Figure omitted]

EMPLOYMENT LOSSES, AND EVENTUAL CONVERGENCE TOWARD BASELINE LEVELS,

[Figure omitted]

BY PERMITTING IMMEDIATE EXPENSING RATHER THAN SLOW DEPRECIATION, THE

[Figure omitted]

THE BENEFIT IS SIMILAR BUT NOT QUITE AS GREAT FOR PRODUCERS' DURABLE

[Figure omitted]

TOTAL REAL BUSINESS CAPITAL SPENDING RESPONDS TO THE CHANGED OUTPUT

[Figure omitted]

THE POTENTIAL OUTPUT (GDP) OF THE NATION SHIFTS IN LINE WITH THE NEW

[Figure omitted]

THREE-MONTH TREASURY BILL YIELDS SURGE WITH INFLATION, DECLINE WITH

[Figure omitted]

TEN-YEAR GOVERNMENT BOND YIELDS REFLECT SHORT-RUN AND LONG-RUN CREDIT

[Figure omitted]

THE LOSS OF THE CURRENT HUGE FEDERAL SUBSIDY TO OWNER-OCCUPIED

[Figure omitted]

HIGHER RATES, ECONOMIC WEAKNESS, AND THE LOSS OF MORTGAGE INTEREST

[Figure omitted]