Periodic Reporting for period 4 - APMPAL-HET (Asset Prices and Macro Policy when Agents Learn and are Heterogeneous)
Berichtszeitraum: 2022-04-01 bis 2024-02-29
The conventional approach in macroeconomics assumes that economic agents understand how market variables respond to different shocks, this is the full information Rational Expectations hypothesis (FIRE or RE). But in practice, financial investors and policy authorities have to make decisions with only partial knowledge about the economy. For example, standing in 2022, how could we know if the high inflation of that year was going to be a permanent or transitory phenomenon? How can a policy authority know if if it is influencing long or short run inflation?, This project is about economic agents acting under partial knowledge about the economy, be it in for policy makers influencing whole economies, investors in the huge US or Chinese stock markets, or farmers in a small village in Malawi.
We have been developing two methods to address these issues. First, a framework where consumers or investors learn about the economy, we call this Internally Rational (IR) learning. Second, Optimal Signal Extraction (OSE) allows to find the best policy when the government only has partial knowledge about the economy. In addition we have also done several applications more in line with the standard paradigm that serve as a reference.
IR considers investors who have a view about how stock prices are formed and they behave optimally given this knowledge. In this project we develop this concept in several ways: we extend the analysis to bond prices, option prices and business cycle models. We test the model using actual prices and survey data on expectations about bond yields, wage expectations, stock prices and inflation. We find that our approach to expectation formation helps to understand many aspects of the data. We study investors that have heterogeneous views about future market returns (disagreement), the secular evolution of the stock market, the dissemination of information through networks, the Chinese stock market, informal economy, heterogeneity in capital and labor income, preference heterogeneity, etc. both from an empirical and theoretical point of view.
The conventional approach justifies the so- called efficient market hypothesis, implying that policy interventions in asset markets are undesirable as they would only disrupt the allocation of investments. But if investors learn about prices monetary and macroprudential policy should take into account how expectations influence stock prices. Also, any policy reform needs to consider how expectations respond to policy, even more so when the population is heterogeneous.
We have been developing two methods to address these issues. First, a framework where consumers or investors learn about the economy, we call this Internally Rational (IR) learning. Second, Optimal Signal Extraction (OSE) allows to find the best policy when the government only has partial knowledge about the economy. In addition we have also done several applications more in line with the standard paradigm that serve as a reference.
IR considers investors who have a view about how stock prices are formed and they behave optimally given this knowledge. In this project we develop this concept in several ways: we extend the analysis to bond prices, option prices and business cycle models. We test the model using actual prices and survey data on expectations about bond yields, wage expectations, stock prices and inflation. We find that our approach to expectation formation helps to understand many aspects of the data. We study investors that have heterogeneous views about future market returns (disagreement), the secular evolution of the stock market, the dissemination of information through networks, the Chinese stock market, informal economy, heterogeneity in capital and labor income, preference heterogeneity, etc. both from an empirical and theoretical point of view.
The conventional approach justifies the so- called efficient market hypothesis, implying that policy interventions in asset markets are undesirable as they would only disrupt the allocation of investments. But if investors learn about prices monetary and macroprudential policy should take into account how expectations influence stock prices. Also, any policy reform needs to consider how expectations respond to policy, even more so when the population is heterogeneous.
“Optimal Policy with General Signal Extraction” derives the method OSE to solve for optimal policy under partial information. Technically this is a difficult problem because the choice of policy has direct consequences on the economy and it matters for the revelation of information about the state of the economy. The policy can explain why some governments react slowly to macroeconomic crisis and let deficits pile up before actually taking the necessary measures to avoid a debt default.
“Learning about Bond Prices” uses IR learning to explain the behavior of interest rates of US bonds at different maturities (yield curve). This has the potential to explain a number of observations relating to volatility of bond returns. It also improves dramatically the fit of the behavior of surveys.
“The Stock Market: a Factory of Uncertainty?” studies how belief heterogeneity (disagreement) under IR learning influences investors’ welfare. Disagreement about future returns fabricates volatility on individual consumption, hence restricting stock trades may improve welfare, just as it would in a standard model with incomplete markets: just because investors have different views, they make different bets and when the return is realized investors suffer a larger volatility of their wealth.
“Pareto-Improving Optimal Capital and Labor Taxes” studies a model where the government collects taxes both on labor and capital income and agents are heterogeneous. This changes considerably the nature of the policy as the distributive issues of different policies need to be considered: labor taxes should be low and capital taxes high for many years if all agents should benefit from a policy reform. More generically, the paper shows that policy analysis under heterogeneous agents should consider all possible efficient policies (the so-called Pareto-optimal frontier).
“On Leaving the Euro”. The main focus of the paper is the role of expectations in policy analysis with IR expectations. In the standard RE paradigm any change in policy should take into account that expectations will adapt to the policy change (“Lucas Critique”). But under learning agents’ expectations are a new parameter in the model, as most policy-makers understand, but the literature often ignores. The uncertainty about agents’ expectations needs to be taken into account in any policy reform.
In addition the project has dealt with a large number of additional issues including: empirical studies on the distribution of income; monetary and fiscal policy under partial information; the interaction between expectations about sovereign default and sovereign premium; how agents in a network exchange information about the contracts they take the interaction, both in theory and in the data on food transfers using data collected in a Malawi village; taste heterogeneity; how to limit stock price volatility due to wealth effect of stock fluctuations and disagreement; forecasting stock returns using survey expectations; interaction between capital gains taxes and stock price volatility under IR learning; option prices and expectations; learning about wages and employment volatility; the Chinese stock market; and inflation expectations adjust at the date of announcing or implementing Inflation Targeting across countries.
“Learning about Bond Prices” uses IR learning to explain the behavior of interest rates of US bonds at different maturities (yield curve). This has the potential to explain a number of observations relating to volatility of bond returns. It also improves dramatically the fit of the behavior of surveys.
“The Stock Market: a Factory of Uncertainty?” studies how belief heterogeneity (disagreement) under IR learning influences investors’ welfare. Disagreement about future returns fabricates volatility on individual consumption, hence restricting stock trades may improve welfare, just as it would in a standard model with incomplete markets: just because investors have different views, they make different bets and when the return is realized investors suffer a larger volatility of their wealth.
“Pareto-Improving Optimal Capital and Labor Taxes” studies a model where the government collects taxes both on labor and capital income and agents are heterogeneous. This changes considerably the nature of the policy as the distributive issues of different policies need to be considered: labor taxes should be low and capital taxes high for many years if all agents should benefit from a policy reform. More generically, the paper shows that policy analysis under heterogeneous agents should consider all possible efficient policies (the so-called Pareto-optimal frontier).
“On Leaving the Euro”. The main focus of the paper is the role of expectations in policy analysis with IR expectations. In the standard RE paradigm any change in policy should take into account that expectations will adapt to the policy change (“Lucas Critique”). But under learning agents’ expectations are a new parameter in the model, as most policy-makers understand, but the literature often ignores. The uncertainty about agents’ expectations needs to be taken into account in any policy reform.
In addition the project has dealt with a large number of additional issues including: empirical studies on the distribution of income; monetary and fiscal policy under partial information; the interaction between expectations about sovereign default and sovereign premium; how agents in a network exchange information about the contracts they take the interaction, both in theory and in the data on food transfers using data collected in a Malawi village; taste heterogeneity; how to limit stock price volatility due to wealth effect of stock fluctuations and disagreement; forecasting stock returns using survey expectations; interaction between capital gains taxes and stock price volatility under IR learning; option prices and expectations; learning about wages and employment volatility; the Chinese stock market; and inflation expectations adjust at the date of announcing or implementing Inflation Targeting across countries.
While APMPAL started to develop the foundations of IR and OSE the current project APMPAL-HET has produced a large number of applications, showing that these frameworks can become a standard tool for macroeconomists to understand data and to design policy. This would provide an alternative to the standard approach on how to perform macroeconomic and policy analysis which, we think, matters in practice. We have also shown that heterogeneity is a key issue in determining policy, and how to consider all possible (pareto optimal) efficient policies instead of picking an arbitrary policy as most papers currently do. This should be key in making policy reforms that are i) acceptable by democratic societies, where all parts of the population have a say in how policy is formed, and ii) take into account how expectations shape the effectiveness of a policy reform.
In the last two decades more policy institutions, including central banks, are using macroeconomic models to guide their decisions. Hopefully, the developments in the project can be incorporated in these models and this can help design better macro policies that affect the lives of billions of people.
In the last two decades more policy institutions, including central banks, are using macroeconomic models to guide their decisions. Hopefully, the developments in the project can be incorporated in these models and this can help design better macro policies that affect the lives of billions of people.