Final Report Summary - POLHIA (Monetary, fiscal and structural policies with heterogeneous agents)
The POLHIA's project general goal, as stated in the first annex of the grant agreement, consisted in 'exploring the role of macroeconomic policies and the nexus of macroeconomic and microeconomic/structural policies in a heterogeneous agents setting. Therefore, POLHIA aimed at providing new insights and useful suggestions for the implementation of both macroeconomic policies and structural policies.
The scientific basis of policy recommendations was developed during the first reporting period. In the second reporting period, research work was finalised and completed. The activities of most of the POLHIA units during this period aimed at exploiting the theoretical frameworks, analytical models, empirical and experimental databases and simulation software mainly developed in the first 18 months of the project to derive useful research based policy recommendations.
The research units exploited a wide range of tools. At the level of model building the development of macroeconomic frameworks in the new Keynesian tradition was paralleled with the extensive use of agent based (AB) models, which were particularly appropriate for the exploration of heterogeneous agents' environments. Empirical research was carried out by means of econometric models and experiments to study, for instance, the formation of expectations. The natural candidates to be beneficiaries of this type of analysis were therefore policy makers.
The project consisted of seven scientific work packages (WPs) which analysed the following issues:
1. heterogeneous expectations, macroeconomic performance and monetary policy
2. heterogeneous expectations and exchange rate dynamics
3. financing constraints and firms' growth
4. heterogeneous financial conditions and monetary policy
5. multi-agent models for policy analysis
6. fiscal policy, sustainability and coordination
7. technological change, skill mismatch and human resource policies.
During the first reporting period the research work of most of the POLHIA units aimed first and foremost at developing theoretical frameworks, analytical models, empirical and experimental databases and simulation software to explore the properties of economies in which agents were heterogeneous and to prepare the terrain for the derivation of useful research based policy recommendations.
During the second reporting period the research work of most of the POLHIA units aimed at exploiting the theoretical frameworks, models, databases and simulation software mainly developed in the first period to derive useful research based policy recommendations.
More specifically, in the first reporting period:
1. Units at Amsterdam and Leuven emphasised mainly the heterogeneity of mechanisms of expectations formation. This research showed that rational and fundamental expectations were indeed the exception rather than the rule when agents tried to gauge future developments of relevant macro-variables. Evidence collected and analysed during experiments with human subjects, moreover, showed that that only few rules were actually adopted by rational agents to form expectations and that agents might switch from one mechanism to another depending upon the forecasting performance of the rules. This evidence called for building macro-dynamic models in which expectations were not formed according to a unique rational scheme but might follow different competing rules.
2. Units at Catholic University, Milan, and Ancona developed multi agent models which started from the behaviour of heterogeneous agents and their local interactions and allowed for the possibility that interaction and individual rules changed in time. In these models, statistical regularities emerged in the aggregate that could not be inferred from the primitives of individuals. High-levels macroeconomic systems possessed new and different properties with respect to low-level microeconomic systems. Even in the absence of a centralised market clearing mechanism, the economy showed a tendency to self-organise towards a spontaneous order which was however characterised sometimes by persistent involuntary unemployment, unsold production or excess demands and, possibly, credit rationing.
3. The unit at Sciences Po, Paris, addressed essential issues pertaining to fiscal policy. This research showed that most of the arguments in favour of fiscal rules could justify the imposition of rules at the national level but not at the supranational level unless externalities or credibility considerations were invoked. Due to the inconclusiveness of the available evidence, research showed that the stability and growth pact (SGP) was a public social norm that countries obeyed in order to preserve reputation among the other members of the European Union. The role of the SGP as a coordinating device was thus debatable. It actually imposed coordination from the bottom, limiting the space for fiscal policies that could in principle be beneficial in the short as well as in the long run.
4. As to educational policies, research work in Rome showed that the interaction of educational policies and different labour market institutions affected the outcome of a skill biased technological transition. On the one hand, educational policies sped up skill formation and thereby allowed the system to escape low productivity equilibrium. On the other hand, they relaxed financial constraints at the household level associated to a large decrease in the unskilled wage.
The global financial crisis (GFC) and the great recession that unfolded after the project was drafted dramatically changed the terms of the policy debated in the academia, the political arena and on the media. The research agenda of POLHIA was therefore deeply affected by these developments. The research work on issues pertaining to the GFC nourished our thinking and challenged our view of the future and of the design of policies.
During the second reporting period:
1. Units at Amsterdam and Leuven expanded, extended and exploited for policy purposes models inspired by the idea that heterogeneous mechanisms of expectations formation were key to understand market outcomes, macroeconomic performance and the effectiveness of monetary policy. It was shown that empirical regularities that were considered puzzles if one adopted the efficient market expectations approach were quite straightforward market outcomes under heterogeneous expectations. In a macroeconomic perspective, new Keynesian dynamic stochastic general equilibrium (NK-DSGE) models with heterogeneous expectations behaved quite differently from the standard ones. Evidence collected and analysed during experiments with human subjects showed that only few rules were actually adopted to form expectations and that agents might switch from one mechanism to another depending upon the forecasting performance of the rules. This evidence had important implications for the effectiveness of monetary policy in an experimental environment which replicated the basic features of a macrodynamic NK model.
2. Units at Catholic University, Milan, and Ancona modified a multi agent model characterised by heterogeneous financial conditions of firms to assess the impact of monetary and fiscal policy. Even in the absence of a centralised market clearing mechanism the economy showed a tendency to self-organise towards a spontaneous order which was however characterised by booms and busts of economic activity. In the second reporting period simulations of the modified AB model with monetary policy showed the stabilising effects of a Taylor rule. Fiscal policy also had stabilising effects, especially when the public sector was not constrained by a balanced budget rule.
3. The unit at Sciences Po, Paris, pursued a line of research already inaugurated in the first reporting period essentially in the domain of fiscal policy. The different proposals under discussion at the European Union level were discussed and ranked based upon their macroeconomic outcomes. The new debt rule would certainly lead to lower debt levels, hence to larger fiscal margins for manoeuvring in the future but it would be very costly to implement. It was true, however, that the fiscal status quo in the European Union would be even worse.
4. The unit at La Sapienza, Rome, emphasised the two-way feedback between inequality, wage dispersion and labour market features on the one hand and educational attainment on the other hand.
The experimental and empirical evidence collected in WP1 showed that heterogeneity of expectations together with a mechanism of evolutionary switching between different prediction rules generally existed and had very important implication on macro-dynamics and monetary policy decisions. Theoretical results also suggested that the new Keynesian model with heterogeneous agent and prediction choice switching was capable of replicating many stylised facts on inflation persistence from macro and micro perspectives and received more and more interests, especially from central banks.
Firstly, the researchers found that the large intensity of selection between heterogeneous forecasting rules tended to cause high persistence in contrast to the more aggressive response of the nominal interest rate to the deviation of inflation from their target. Secondly, the researchers found that individuals tended to base their predictions on past observations following simple forecasting heuristics and that individual expectations were heterogeneous. Individual learning took the form of switching from one heuristic to another according to past relative performance. The stabilizing effect of different monetary policies depended on the ecology of forecasting rules, on agents' sensitivity to differences in forecasting performance and on how aggressively the monetary authority set the nominal interest rate in response to inflation. Thirdly, recent papers showed that multiple equilibria, periodic orbits and complex dynamics could arise in the new Keynesian models under dynamic predictor selection. Finally, in case financial markets were destabilised by financial innovation or learning, the effect could spill over to the macro economy and thus destabilise the economy as a whole.
The research of WP1 was presented in many conferences and workshops. Some researchers from central banks showed great interest in our results on heterogeneous agents and learning.
The main potential impact of research in WP2 was to add to the growing literature of heterogeneous agent models. This literature gained significant momentum over the years andoffered an alternative where mainstream models failed to provide a sensible explanation to several empirical regularities typically observed in financial markets. WP2 offered a clear contribution in this process. The type of models that the researchers proposed was richer than mainstream models in the sense that they allowed for the presence of multiple equilibria. In such a world, the emergence of asset price bubbles was natural. In mainstream models on the contrary assets price bubbles were not possible, and this was a remarkable failure. Understanding why asset price bubbles emerged was absolutely necessary if one wanted to create a more stable economic system. Researchers contributed towards gaining further understanding of economic complex systems in a world populated by heterogeneous agents.
The main dissemination activities that were carried out were the presentation of papers at several international conferences and workshops. In addition, the unit at Leuven organised an international conference, which was titled 'Quantifying and understanding dysfunctions of financial markets' and covered a range of topics, all related to the nature of the POLHIA project. Moreover, as part of the dissemination policy, a special issue of the Journal of Economic Dynamics and Control was organised. Several posters were printed and distributed in various economics departments, while an invitation by email was sent to a large list of economics researchers and practitioners. Moreover, the fifth workshop on monetary, fiscal and structural policies with heterogeneous agents was organised in Leuven.
The main potential impact of research in WP3 consisted in contributing to the literature on financial constraints and economic activity. This literature was potentially important for policy purposes.
Furthermore, the main potential impact of WP4 emerged from the development of a macroeconomic AB model whose policy implications were far reaching. In addition, WP4 contributed to the search for a syncretic framework that could reconcile the simplicity of the standard macroeconomic models with the realism of AB models.
The most important dissemination activity of WP5 was the organisation of the 16th annual workshop on heterogeneous interacting economic agents (WEHIA), which took place at the Faculty of Economics in Ancona from 23 to 25 June 2011.
WP6 contributed to the debate on the implementation of an optimal, or at least satisfactory, governance of fiscal policy in Europe. Most of the results of this research ran counter to the received view and conventional wisdom.
Finally, the main potential impact of WP7 was its contribution to the literature on technical progress, labour market institutions and educational policies. Some of the insights derived from this research ran counter to the received wisdom on these issues.
The scientific basis of policy recommendations was developed during the first reporting period. In the second reporting period, research work was finalised and completed. The activities of most of the POLHIA units during this period aimed at exploiting the theoretical frameworks, analytical models, empirical and experimental databases and simulation software mainly developed in the first 18 months of the project to derive useful research based policy recommendations.
The research units exploited a wide range of tools. At the level of model building the development of macroeconomic frameworks in the new Keynesian tradition was paralleled with the extensive use of agent based (AB) models, which were particularly appropriate for the exploration of heterogeneous agents' environments. Empirical research was carried out by means of econometric models and experiments to study, for instance, the formation of expectations. The natural candidates to be beneficiaries of this type of analysis were therefore policy makers.
The project consisted of seven scientific work packages (WPs) which analysed the following issues:
1. heterogeneous expectations, macroeconomic performance and monetary policy
2. heterogeneous expectations and exchange rate dynamics
3. financing constraints and firms' growth
4. heterogeneous financial conditions and monetary policy
5. multi-agent models for policy analysis
6. fiscal policy, sustainability and coordination
7. technological change, skill mismatch and human resource policies.
During the first reporting period the research work of most of the POLHIA units aimed first and foremost at developing theoretical frameworks, analytical models, empirical and experimental databases and simulation software to explore the properties of economies in which agents were heterogeneous and to prepare the terrain for the derivation of useful research based policy recommendations.
During the second reporting period the research work of most of the POLHIA units aimed at exploiting the theoretical frameworks, models, databases and simulation software mainly developed in the first period to derive useful research based policy recommendations.
More specifically, in the first reporting period:
1. Units at Amsterdam and Leuven emphasised mainly the heterogeneity of mechanisms of expectations formation. This research showed that rational and fundamental expectations were indeed the exception rather than the rule when agents tried to gauge future developments of relevant macro-variables. Evidence collected and analysed during experiments with human subjects, moreover, showed that that only few rules were actually adopted by rational agents to form expectations and that agents might switch from one mechanism to another depending upon the forecasting performance of the rules. This evidence called for building macro-dynamic models in which expectations were not formed according to a unique rational scheme but might follow different competing rules.
2. Units at Catholic University, Milan, and Ancona developed multi agent models which started from the behaviour of heterogeneous agents and their local interactions and allowed for the possibility that interaction and individual rules changed in time. In these models, statistical regularities emerged in the aggregate that could not be inferred from the primitives of individuals. High-levels macroeconomic systems possessed new and different properties with respect to low-level microeconomic systems. Even in the absence of a centralised market clearing mechanism, the economy showed a tendency to self-organise towards a spontaneous order which was however characterised sometimes by persistent involuntary unemployment, unsold production or excess demands and, possibly, credit rationing.
3. The unit at Sciences Po, Paris, addressed essential issues pertaining to fiscal policy. This research showed that most of the arguments in favour of fiscal rules could justify the imposition of rules at the national level but not at the supranational level unless externalities or credibility considerations were invoked. Due to the inconclusiveness of the available evidence, research showed that the stability and growth pact (SGP) was a public social norm that countries obeyed in order to preserve reputation among the other members of the European Union. The role of the SGP as a coordinating device was thus debatable. It actually imposed coordination from the bottom, limiting the space for fiscal policies that could in principle be beneficial in the short as well as in the long run.
4. As to educational policies, research work in Rome showed that the interaction of educational policies and different labour market institutions affected the outcome of a skill biased technological transition. On the one hand, educational policies sped up skill formation and thereby allowed the system to escape low productivity equilibrium. On the other hand, they relaxed financial constraints at the household level associated to a large decrease in the unskilled wage.
The global financial crisis (GFC) and the great recession that unfolded after the project was drafted dramatically changed the terms of the policy debated in the academia, the political arena and on the media. The research agenda of POLHIA was therefore deeply affected by these developments. The research work on issues pertaining to the GFC nourished our thinking and challenged our view of the future and of the design of policies.
During the second reporting period:
1. Units at Amsterdam and Leuven expanded, extended and exploited for policy purposes models inspired by the idea that heterogeneous mechanisms of expectations formation were key to understand market outcomes, macroeconomic performance and the effectiveness of monetary policy. It was shown that empirical regularities that were considered puzzles if one adopted the efficient market expectations approach were quite straightforward market outcomes under heterogeneous expectations. In a macroeconomic perspective, new Keynesian dynamic stochastic general equilibrium (NK-DSGE) models with heterogeneous expectations behaved quite differently from the standard ones. Evidence collected and analysed during experiments with human subjects showed that only few rules were actually adopted to form expectations and that agents might switch from one mechanism to another depending upon the forecasting performance of the rules. This evidence had important implications for the effectiveness of monetary policy in an experimental environment which replicated the basic features of a macrodynamic NK model.
2. Units at Catholic University, Milan, and Ancona modified a multi agent model characterised by heterogeneous financial conditions of firms to assess the impact of monetary and fiscal policy. Even in the absence of a centralised market clearing mechanism the economy showed a tendency to self-organise towards a spontaneous order which was however characterised by booms and busts of economic activity. In the second reporting period simulations of the modified AB model with monetary policy showed the stabilising effects of a Taylor rule. Fiscal policy also had stabilising effects, especially when the public sector was not constrained by a balanced budget rule.
3. The unit at Sciences Po, Paris, pursued a line of research already inaugurated in the first reporting period essentially in the domain of fiscal policy. The different proposals under discussion at the European Union level were discussed and ranked based upon their macroeconomic outcomes. The new debt rule would certainly lead to lower debt levels, hence to larger fiscal margins for manoeuvring in the future but it would be very costly to implement. It was true, however, that the fiscal status quo in the European Union would be even worse.
4. The unit at La Sapienza, Rome, emphasised the two-way feedback between inequality, wage dispersion and labour market features on the one hand and educational attainment on the other hand.
The experimental and empirical evidence collected in WP1 showed that heterogeneity of expectations together with a mechanism of evolutionary switching between different prediction rules generally existed and had very important implication on macro-dynamics and monetary policy decisions. Theoretical results also suggested that the new Keynesian model with heterogeneous agent and prediction choice switching was capable of replicating many stylised facts on inflation persistence from macro and micro perspectives and received more and more interests, especially from central banks.
Firstly, the researchers found that the large intensity of selection between heterogeneous forecasting rules tended to cause high persistence in contrast to the more aggressive response of the nominal interest rate to the deviation of inflation from their target. Secondly, the researchers found that individuals tended to base their predictions on past observations following simple forecasting heuristics and that individual expectations were heterogeneous. Individual learning took the form of switching from one heuristic to another according to past relative performance. The stabilizing effect of different monetary policies depended on the ecology of forecasting rules, on agents' sensitivity to differences in forecasting performance and on how aggressively the monetary authority set the nominal interest rate in response to inflation. Thirdly, recent papers showed that multiple equilibria, periodic orbits and complex dynamics could arise in the new Keynesian models under dynamic predictor selection. Finally, in case financial markets were destabilised by financial innovation or learning, the effect could spill over to the macro economy and thus destabilise the economy as a whole.
The research of WP1 was presented in many conferences and workshops. Some researchers from central banks showed great interest in our results on heterogeneous agents and learning.
The main potential impact of research in WP2 was to add to the growing literature of heterogeneous agent models. This literature gained significant momentum over the years andoffered an alternative where mainstream models failed to provide a sensible explanation to several empirical regularities typically observed in financial markets. WP2 offered a clear contribution in this process. The type of models that the researchers proposed was richer than mainstream models in the sense that they allowed for the presence of multiple equilibria. In such a world, the emergence of asset price bubbles was natural. In mainstream models on the contrary assets price bubbles were not possible, and this was a remarkable failure. Understanding why asset price bubbles emerged was absolutely necessary if one wanted to create a more stable economic system. Researchers contributed towards gaining further understanding of economic complex systems in a world populated by heterogeneous agents.
The main dissemination activities that were carried out were the presentation of papers at several international conferences and workshops. In addition, the unit at Leuven organised an international conference, which was titled 'Quantifying and understanding dysfunctions of financial markets' and covered a range of topics, all related to the nature of the POLHIA project. Moreover, as part of the dissemination policy, a special issue of the Journal of Economic Dynamics and Control was organised. Several posters were printed and distributed in various economics departments, while an invitation by email was sent to a large list of economics researchers and practitioners. Moreover, the fifth workshop on monetary, fiscal and structural policies with heterogeneous agents was organised in Leuven.
The main potential impact of research in WP3 consisted in contributing to the literature on financial constraints and economic activity. This literature was potentially important for policy purposes.
Furthermore, the main potential impact of WP4 emerged from the development of a macroeconomic AB model whose policy implications were far reaching. In addition, WP4 contributed to the search for a syncretic framework that could reconcile the simplicity of the standard macroeconomic models with the realism of AB models.
The most important dissemination activity of WP5 was the organisation of the 16th annual workshop on heterogeneous interacting economic agents (WEHIA), which took place at the Faculty of Economics in Ancona from 23 to 25 June 2011.
WP6 contributed to the debate on the implementation of an optimal, or at least satisfactory, governance of fiscal policy in Europe. Most of the results of this research ran counter to the received view and conventional wisdom.
Finally, the main potential impact of WP7 was its contribution to the literature on technical progress, labour market institutions and educational policies. Some of the insights derived from this research ran counter to the received wisdom on these issues.