2024-02-02-IMF-寻找工资增长_对_新机器时代_的政策回应_81页_3mb

Summary of "Searching for Wage Growth: Policy Responses to the 'New Machine Age'"

Core Content

This IMF Working Paper investigates how technological change, particularly the rise of "robot" capital (encompassing AI, computers, big data, digitalization, networks, sensors, and servos), affects the effectiveness of three key policies: corporate tax cuts (CTC), infrastructure investment (II), and education investment (IE). The study employs a neoclassical growth model calibrated to U.S. data, with a focus on how these policies influence wage growth and income distribution in the context of a rapidly evolving technological landscape.

Main Viewpoints

• Technological Change and Policy Impact: The introduction of "robot" capital significantly alters the effects of traditional policies on wage growth and income distribution.
• Substitution Elasticities: The model highlights the importance of substitution elasticities between different factors of production, especially between "robot" capital and low-skill labor.
• Policy Effectiveness: The effectiveness of each policy varies depending on the substitutability of "robot" capital with labor, with notable differences in their long-term impacts on GDP growth and wage distribution.
• Welfare Implications: The welfare rankings of the policies are influenced by the social discount factor and the weight given to distributional objectives.

Key Findings

1. Corporate Tax Cuts (CTC)

• Wage Effects: In a traditional economy, CTC leads to a rise in real wages for both low- and high-skill workers.
• With "Robot" Capital: CTC results in higher long-run GDP growth (1–2 percentage points), but the skill premium rises sharply. High-skill wages increase by the same amount as GDP growth, while low-skill wages may fall or even decline.
• Welfare: CTC is less effective in improving welfare for the poor compared to other policies, especially when the social discount factor is lower or when distributional concerns are emphasized.

2. Infrastructure Investment (II)

• Wage Effects: II leads to a greater increase in low-skill wages than CTC. It also results in a more substantial rise in skilled wages.
• GDP Growth: II increases GDP growth more than CTC, with strong private capital crowding-in effects. Even under pessimistic assumptions, II boosts GDP growth by 3–4 percentage points.
• Welfare: II consistently dominates CTC in terms of welfare gains, as it increases both aggregate capital and low-skill wages more effectively.

3. Education Investment (IE)

• Wage Effects: IE has a significant positive impact on low-skill wages and leads to the highest welfare gains in the long run.
• GDP Growth: IE stimulates the accumulation of "robot" capital more than CTC, and in some scenarios, even surpasses II in terms of welfare.
• Welfare: IE is most beneficial when the social discount factor is higher or when the welfare function places more weight on the poor. In a partial equilibrium setting, IE may be less effective due to lower direct returns compared to II and CTC.

Calibration and Model Structure

• Model Components: The model includes low-skill workers, high-skill workers, capitalists, and "robot" capital. It also incorporates international capital flows, public investment in education and infrastructure, and a representative agent who maximizes utility over time.
• Production Function: The production function is defined as $Q_t = G_{t-1}^{\eta} F[H(S_t, K_t), V(L_t, Z_t)]$, where $G_{t-1}$ is infrastructure capital, $H(S_t, K_t)$ and $V(L_t, Z_t)$ are CES aggregates for skilled and unskilled labor, and "robot" capital.
• Calibration Parameters:
  • Discount Factor: $\beta = 1 / (1 + \rho)$, where $\rho$ is the private time preference rate.
  • Tax Rates: The pre-2018 corporate tax rate is estimated at 40%, and the effective marginal tax rate on corporate profits is set at 27%.
  • Return on Capital: The pre-tax return on private capital is 10%, and the return on infrastructure is estimated at 10–15%.
  • Education Return: The return on education is 7%, with a depreciation rate of 3%.
  • Substitution Elasticities: The elasticity of substitution between "robot" capital and low-skill labor is found to be above 2, which is higher than other elasticities. This suggests that "robot" capital is highly substitutable with low-skill labor.

Implications

• Policy Prioritization: In the presence of "robot" capital, infrastructure investment is more effective than corporate tax cuts in boosting GDP and low-skill wages. Education investment, however, tends to produce the highest welfare gains, especially when the social discount factor is higher or when the welfare function emphasizes the poor.
• Empirical Support: The model is calibrated using empirical data and estimates from the literature, with a focus on ensuring that the production function and substitution elasticities are consistent with real-world observations.
• Robustness: The results are robust to different specifications of the production function and substitution elasticities, with the most plausible specification being the one that includes "robot" capital as a distinct factor of production.

Conclusion

The paper emphasizes that the "new machine age" fundamentally changes the way policies affect wage growth and income distribution. It provides a framework for understanding the differential impacts of CTC, II, and IE in a world where "robot" capital is a major factor of production. The findings suggest that policy design must take into account the substitutability of new technologies with labor and the broader welfare implications of different investment strategies.