New
Approaches to the Study of Economic Fluctuations
Training and Mobility of Researchers
Network
Research
During the last reporting period, the Network has been pursuing an intensive programme of research and training which involves not only the development of new statistical methodologies, but also the application of these methodologies to newly available sources of data, in order to better understand business cycle fluctuations, both at the aggregate level (ABC) and at the sectoral (SBC) one, in the EU. By fostering a close collaboration between econometricians and macroeconomists, the Network has made important progress in closing the gap between econometric and macroeconomic research in the EU, which for a long time seemed to be moving in independent directions.
Among the main analytical tasks undertaken by the teams participating in the Network, the following stand out:
Aggregate Business Cycles (ABC)
- Development of econometric techniques for the study of the ABC (Tilburg)
Francois (Tilburg) has opened a new avenue for analysing business cycles. He has shown that Schumpeterian growth models, which hitherto have been thought to create smooth growth processes, actually have coordination equilibria and thus business cycles, thus marrying these two previously separate literatures. This research is detailed in a joint paper with Lloyd-Ellis and CentER (Tilburg) DP 2001, No. 36. http://greywww.kub.nl:2080/greyfiles/center/2001/36.html
- Synchronisation and co-movements between (possibly) many time series (ECARES, Modena, CEPR, UCIII)
Here the research has proposed new measures of multiple correlation in a dynamic setting. These measures are based on frequency domain analysis. They have been used for the understanding of co-movements between regional ouput, labour market variables, exchange rates, etc. New techniques for forecasting nonstationary dynamic factor models have also been derived.
Modena has also been working on national and supranational business cycle analysis, and in particular to synchronization of cycles with the application of multivariate analysis.
- Leading Indicators (Modena, ECARES)
In 2001-2 the main contribution to this area of the project by the Modena team has been the proposal of a new forecasting method that exploits information from a large panel of time series. The method is based on the generalized dynamic factor model previously developed by network members. Further results have been derived also in the field of structural dynamic factor models, with the aim of proposing dynamic factor models instead of VAR models for the structural analysis of macroeconomic time series.
Team-members from Modena, ECARES and CEPR have contributed significantly to this area of research this year. Together with researchers from the Bank of Italy, they constructed EUROCOIN, the CEPR monthly coincident indicator of the business cycle of the euro area. The indicator is now published monthly in the CEPR web site (www.cepr.org) and is being frequently quoted in the leading financial press, like The Economist .
- Growth and ABC ( IMOP, UCIII)
This year work in this area has continued to develop new models on the propagation of cycles through the varying depreciation rate of capital goods in the different ABC phases. New theoretical work has been produced on the stability properties of the steady state in two-sector real business cycle models with sector specific externalities showing that the existence capital adjustment costs of any size precludes the existence of indeterminacy which is common in this type of models without adjustment costs.
Work in this area has concentrated on new identification methods of monetary shocks and its effects on the economy at different horizons. Research on the asymmetric effects of monetary policy and in its conduct by central banks has been followed up with new research on the presence of asymmetric features in the loss functions of central banks in terms of deviations of inflation from its target and of nonlinearities in the Phillips curve relating inflation and output gap. Further, theoretical analysis on the implications for inflation and employment variability of the choice of institutions for the labour market and the conduct of monetary policy has also been carried out.
- Co-integration in large models (UCIII)
New testing and estimation procedures for co-integrating relationships in large-dimensional systems have been derived. Likewise, several new methods of testing for long-memory in time series have been developed. Numerous empirical applications of those methods are provided. Further, new forecasting techniques in the presence of outliers have been developed.
Disaggregate Analysis of SBC and the Dynamics of large Cross- Sections
- Dynamic Factor Models (Modena)
Applications of a dynamic factor model have been investigated for the study of wage bargaining, and an analysis of the degree of specialization across manufacturing sectors. The synchronization of sectoral investment cycles has also been examined.
- Microfinance (ECARES)
A simple extension to existing credit contacts for the poor ("microfinance contracts") was examined, that would allow financial institutions to provide repayment insurance to their clients. The proposed contract uses the repeated nature of loans to build credit records that borrowers in good standing can use to insure themselves against default in case of adverse income shocks. After documenting borrowers' desire for insurance using data from a microfinance program in Guatemala, sufficient conditions for the proposed contract to reduce borrower vulnerability while improving repayment rates was derived. These conditions are quite similar to those that credit-card and automobile-insurance companies seem to apply to deter moral hazard and adverse selection among their subscribers. The question of why institutions lending to the poor may face particular implementation programs because of the history of past failures of credit programs for the poor was discussed.
With widely publicized high repayment rates, microfinance is gaining a great deal of attention. Using again data from Guatemala, ECARES (with the Federal Reserve Board) examined risk matching in credit groups. The literature often assumes that joint-liability will lead groups to form homogeneously in risk, and that risk heterogeneity emerges only as a second-best. It was found that they do not, even accounting for matching frictions. Data on mutual help within groups provides evidence consistent with the hypothesis that group lending provides insurance among borrowers. This intra-group insurance suggests that current credit contracts can be improved by incorporating insurance provisions
- Electoral Outcomes (ECARES and UCIII)
The literature on political campaigns has thus far mainly focused on the positioning of parties. Little work has been done on how parties communicate their platform to the electorate. Yet, with the proliferation and fragmentation of the media outlets, "getting the message out" has become increasingly important and expensive. In a model in which parties divide their budget between "creative costs" (the costs of designing the political message) and "media costs" (costs of distributing the message), ECARES (with INSEAD) examined how respective budget levels affect parties' advertising strategies during a political campaign, and how these affect their respective loci of electoral support. They used these results to investigate how the political landscape is affected by campaign spending limits.
From an empirical viewpoint, researchers at UCIII have developed a new method to forecast the results of elections relying on the modelling of long-memory processes. Those processes are the outcome of aggregating heterogenous autoregressive time series which differ in their degree of persistence. Insofar as voters can be classified among “committed” voters (with high persistence in their party preferences) and “floating” ones (with low persistence in their party preferences), long-memory provides a useful set-up to model aggregate poll opinions.
Estimation and Calibration Techniques for the Valuation of Equilibrium Macro Models
The Modena team has provided several contributions to macromodelling. The unifying theme in the work of the Modena team in this area has been the analysis of labor market structure and integration. The tools for the analysis have been: 1) dynamic political-economy models, 2) dynamic stochastic general equilibrium models with staggered price setting, and 3) game-theoretic models of wage bargaining.
A new method for analysing macroeconomic shocks, using sign restrictions on impulse responses in vector autoregressions which is becoming increasingly popular in the literature, has been developed by Harald Uhlig (Tilburg) and applied to an increasing set of questions to be addressed by this network. In the current reporting period, Harald Uhlig has analysed, whether the Alan Greenspan and the Federal Reserve Bank in the US surprised markets with their interest rate cuts in 2001, concluding that this was not the case (CentER DP 2001, nr. 88, http://greywww.kub.nl:2080/greyfiles/center/2001/88.html). More recent work has investigated the impact of fiscal policy shocks, see Mountford-Uhlig, ''What are the effects of fiscal policy shocks?'', CEPR DP 3338, April 2002. While these new methodologies have so far been used on US data first, applications to European business cycle issues are planned next.
Return to Introduction
Research