(Diese Seite ist nur auf Englisch verfügbar. Direkt zu
Sven Schreibers Hauptseite: hier. / This
page is available only in English. Directly to Sven Schreiber's
home page: here.
Most of the material on this page and then some more is also on my Repec page.)
(Latest update: October 2016)
Research interests: Quite broad (as you can see on this page), but mostly related to time series econometrics and macroeconomics. Other keywords are labor/unemployment issues, pensions, inflation.
Entries are (mostly) listed in reverse chronological order. Note that contact information given in older papers may be obsolete, but comments are still and always welcome!
Keywords: weather, business cycle, nowcasting
Download: version September 2016 based on the presentation at, and prepared for the conference proceedings of the Joint Statistical Meetings. This is a follow-up paper and fairly thorough revision of "Adjusting production indices for varying weather effects", IMK Working paper 171, June 2016 (with Erik Haustein).
Abstract: While recurring and regular variations of weather conditions are implicitly addressed by standard seasonal adjustment procedures of economic time series, extraordinary weather outcomes are not. We propose a way of measuring aggregate abnormal weather conditions based on available local measurements and a straightforward regression-based framework to analyze their impact on German monthly total industrial and construction-sector production data, and find noticeable effects. In the historical –and seasonally adjusted– construction sector growth data the extra explanatory power of the weather regressors over a benchmark univariate autoregressive model even exceeds 50% of the variation. The estimated effects of weather deviations can be subtracted from the already seasonally adjusted data to obtain (seasonally as well as) weather adjusted series, which might capture economic developments better. Given the timely availability of the weather data compared to the publication lag of economic measurements, we also point out how measuring the weather impact may help short-term forecasting or nowcasting of industrial production in real time.
Keywords: Granger causality, frequency domain, filter gain
Download: version June 2016 . The methods proposed in this paper can be applied with my following two function packages for gretl: "BreitungCandelonTest" and "delayspectral". These are available (from within the gretl program) on gretl's function package server.
Abstract: We extend the frequency-specific Granger causality test of Breitung and Candelon (2006) to a more general null hypothesis that allows causality testing at unknown frequencies within a pre-specified range of frequencies. This setup corresponds better to empirical situations encountered in applied research and it is easily implemented in vector autoregressive models. We also provide tools to determine the sampling uncertainty of the phase shift/delay at some pre-specified frequency or frequency band. In an empirical application dealing with the dynamics of US temperatures and CO 2 emissions we find that emissions cause temperature changes only at very low frequencies with more than 30 years of oscillation. Furthermore we analyze the indicator properties of new orders for German industrial production by assessing the delay at the frequencies of interest.
Keywords: consumption smoothing, retirement-consumption puzzle, SOEP
Download: version May 2016 . This paper has waited a long time for more data from recent SOEP waves, and for more explicit theory. Older versions had the title "Housing consumption after retirement: Is there a puzzle?", for example that's the title which appears in the programme of the Econometric Society World Congress 2010.
Abstract: We revisit the alleged retirement consumption puzzle. According to the life-cycle theory, foreseeable income reductions such as those around retirement should not affect consumption. However, we first recall that given higher leisure endowments after retirement, the theory does predict a fall of total market consumption expenditures. In order not to mistake this predicted drop for a puzzle we focus on housing consumption which can be plausibly regarded as complementary to leisure, and we control for the leisure change in our empirical specifications, using micro data for Germany (SOEP), where housing expenditures are observable as rents for the majority (60%), as well as dwelling relocations. We still find significant negative impacts of the retirement status on housing consumption, which is hard to reconcile with the life-cycle theory. For retirees we also find significant effects of the income reduction at retirement on housing. However, the effects are small in quantitative terms, given the lock-in nature of past housing decisions.
Keywords: Obama fiscal stimulus, fiscal multiplier, interest rate forecasts.
Download: version June 2015 .
Abstract: Cogan et al. (2009, 2010) claim that the stimulus package passed by the United States Congress in February 2009 had a multiplier far below one. However, the stimulus’ multiplier strongly depends on the assumed monetary policy response. Based on official statements from the Fed chairman, the general economic outlook, past behavior of the FOMC, optimal policy considerations, and from financial market expectations, we find that in February 2009, a reasonable prediction of the period of monetary accommodation would have exceeded 9 quarters. This implies that an plausible real time assessment of the stimulus’ effects would have been more optimistic than Cogan et al.’s.
Keywords: money growth, Granger causality, quantity theory, unemployment
Short answer: It helps a lot when other important variables are
excluded from the information set.
Longer answer: We revisit claims in the literature that money growth is Granger-causal for inflation at low frequencies. Applying frequency-specific tests to euro-area data in a system with various potentially important variables, money growth is not a significant low-frequency predictor of inflation. A general-to-specific testing strategy reveals a recursive structure where only the unemployment rate and long-term interest rates are directly Granger-causal for low-frequency inflation movements, and all variables affect money growth. We therefore interpret opposite results from bivariate inflation/money growth systems as spurious due to omitted-variable biases. We also analyze the resulting four-dimensional system in a cointegration framework and find structural changes in the long-run adjustment behavior, which do not affect the main conclusions, however.
(Please note that the contact information given in the finished papers below may be out of date.)
Keywords: transitory components, VECM, delta method, bootstrap
Download: only slightly outdated version March 2016 . Forthcoming in Econometric Reviews, see http://dx.doi.org/10.1080/07474938.2016.1235257. The methods proposed in this paper can be applied with my function package "PTconf" for gretl. This should be available shortly from gretl's function package server, or see my software page.
Abstract: The topic of this paper is the estimation uncertainty of the Stock-Watson and Gonzalo-Granger permanent-transitory decompositions in the framework of the cointegrated vector-autoregression. Specifically, we suggest an approach to construct the confidence interval of the transitory component in a given period (e.g. the latest observation) by conditioning on the observed data in that period. To calculate asymptotically valid confidence intervals we use the delta method and two bootstrap variants. As an illustration we analyze the uncertainty of (US) output gap estimates in a system of output, consumption, and investment.
Keywords: density forecasts, business-cycle turning points, real-time data, nowcasting, bootstrap
Abstract: For the timely detection of business-cycle turning points we suggest to use medium-sized linear systems (subset VARs with automated zero restrictions) to forecast monthly industrial production index publications one to several steps ahead, and to derive the probability of the turning point from the bootstrapped forecast density as the probability mass below (or above) a suitable threshold value. We show how this approach can be used in real time in the presence of data publication lags and how it can capture the part of the data revision process that is systematic. Out-of-sample evaluation exercises show that the method is competitive especially in the case of the US, while turning-point forecasts are in general more difficult in Germany.
Keywords: Demography, retirement age, cross-country comparison
Download: English version January 2014. The German version named "Europas künftige Rentenkluft" was published in Wirtschaftsdienst vol. 94 (5), pages 364-368, May 2014; also available as IMK Policy Brief 1/2014.
Abstract: There is a significant variation in demographic development between different European Union (EU) member states. Using the UN’s Population Prospects, we examine how different retirement ages in selected EU countries would lead to comparable relations between the working-age population and pensioners in the future. In the coming decades it seems that the French would be able to take retirement roughly four years earlier than Germans. There is, therefore, no apparent economic justification for the suggested alignment of retirement ages in accordance with the current German regulation, as is sometimes suggested. Even the EU Commission has prioritised life expectancy in its recommendations for greater sustainability in the pension system, despite the fact that it is an insufficient indicator.
Keywords: gretl, TestU01, RNG, random numbers
Download: free at the journal website. Published in Journal of Statistical Software, 2012, vol. 50.
Abstract: The increasing popularity and complexity of random number intensive methods such as simulation and bootstrapping in econometrics requires researchers to have a good grasp of random number generation in general, and the specific generators that they employ in particular. Here, we discuss the random number generation options, their specifications, and their implementations in gretl. We also assess the performance and the reliability of gretl in this department by conducting extensive empirical testing using the TestU01 library. Our results show that the available alternatives are soundly implemented and should be sufficient for most econometric applications.
Keywords: relative price variability, trend inflation, endogeneity bias
Download: Version February 2010 Published in Empirical Economics, 2012, vol. 43, no. 1, pp. 291-311.
Abstract: Distortionary effects of inflation on relative prices are the main argument for inflation stabilization in macro models with sticky prices. Under indexation of non- optimized prices those models imply a nonlinear and dynamic impact of inflation on the cross-sectional price dispersion (relative price or inflation variability, RPV). Using US sectoral price data we estimate such a relationship between inflation and RPV. We confirm the impact of inflation fluctuations but find hitherto neglected en- dogeneity biases, and our IV and GMM estimates indicate that average ("trend") inflation is significant for indexation. Lagged inflation is less important.
Keywords: euro-area unemployment, VECM, permanent-transitory decomposition
Download: Version June 2010. Published in Applied Economics 2012, vol. 44, no. 10, pp. 1315-1335 (DOI: 10.1080/ 00036846.2010.539548).
Abstract: Given that for France, Germany, Italy, and the Netherlands the unemployment rates are best classified as I(1), we apply permanent-transitory decompositions based on cointegrated VARs with relevant variables (labor productivity, wages, tax wedges, foreign relative prices) to estimate the time-varying natural unemployment rates. In general all variables seem to matter, and the results are quite different from published OECD Nairus. Our implied unemployment gaps are better than the OECD gaps in predicting unemployment changes and inflation gaps, but they are (except for Italy) as bad as the OECD gaps for forecasting inflation changes.
Keywords: productivity slowdown, growth, NAIRU level, common shifts
Download published version at www.bepress.com/bejm/vol9/iss1/art39. Published in The B.E. Journal of Macroeconomics, 2009, Vol. 9 : Iss. 1 (Topics), Article 39. DOI: 10.2202/1935-1690.1818.
Update: The Python/Numpy code for the co-breaking analysis is in the file cobreakprototype_web.py, which has a docstring at the top with explanations. To use it you also need the auxiliary file helpers.py and put it next to the other one, meaning in the same directory/folder.
Abstract: We investigate the controversial issue whether unemployment is related to productivity growth in the long run, using U.S. data in a framework of infrequent mean shifts. Univariate tests find (endogenously dated) shifts in 1974, 1986, and 1996. System co-breaking techniques indicate that the shifts are common features, and the implied long-run link between the two variables is negative. Therefore the secular decline of unemployment since the mid 1990's indeed seems related to higher average productivity growth. The initial and final regimes are essentially equal, which would be compatible with explanations of the productivity slowdown that point to historical learning costs of information technology adoption.
Keywords: Hausman test, negative chi^2 statistic, nuisance parameter
Download: last working paper version August 2008. Published in the Journal of Economics and Statistics (Jahrbücher für Nationalökonomie und Statistik), 2008, vol. 228 no. 4, pp. 394-405.
Abstract: We show that under H1 the Hausman chi-square test
statistic can be negative not only in small samples but even
asymptotically. Therefore in large samples a negative test
statistic is only compatible with H1 and should be interpreted
accordingly. Applying a known insight from finite samples, this
can only occur if the different estimation precisions (often the
residual variance estimates) under H0 and under H1 both enter the
test statistic. In finite samples, using the absolute value of
the test statistic is a remedy that does not alter the test under
the null hypothesis and is thus admissible.
[add the following paragraph for long summary:]
Even for positive test statistics the relevant covariance matrix difference should be routinely checked for positive semi-definiteness, because we also show that otherwise test results may be misleading. Of course the preferable solution still is to impose the same nuisance parameter (i.e., residual variance) estimate under the null and alternative hypotheses, if the model context permits that with relative ease. We complement the likelihood-based exposition by a formal proof in an omitted-variable context, we present simulation evidence for the test of panel random effects, and we illustrate the problems with a panel homogeneity test.
Keywords: worksharing, structural vector error correction model, employment
Download: Almost final version (August 2007), but still as eye-opening (or mind-boggling?) as the first one; this is a follow-up paper on Logeay&Schreiber (2005, "Testing the effectiveness...", see below), now in the European Journal of Political Economy, vol 24, no 2, pp. 478-490.
Abstract: Abstract French employment increased significantly after a labor-market reform in 2000. This paper analyzes whether that development was driven by worksharing (the mandated reduction of the workweek length) as claimed by the government. We use a structural VAR model in error correction form (SVECM) to assess the impact of shocks to the workweek length. It turns out that downward workweek shocks actually had adverse employment effects. We conclude that other reform components were responsible for the employment success in France, namely reduced non-wage labor costs and possibly higher firm-level flexibility of temporarily adjusting the workweek.
Keywords: NAIRU, Phillips curve, cointegration, VECM impulse response analysis
Download: This paper grew out of Free University Berlin discussion paper 2002/08 (with a different title) which was presented at the 2002 Econometric Society European Meeting (ESEM). A newer version is in the Journal of Macroeconomics, 2007, vol. 29, pp. 355-367. Slightly outdated version February 2005
Abstract: For the estimation of constant as well as time-varying NAIRUs it is customary to assume -sometimes implicitly- that the long-run Phillips curve is vertical. We point out that the observed data often do not possess the stochastic properties that are needed to impose this restriction, especially when unemployment is non-stationary. Using Germany as a prototypical example, we apply a VAR cointegration analysis and find a negative long-run Phillips curve relation between inflation and unemployment which is robust with respect to variations of the specification. The dynamic interactions indicate that real forces drive the system in the long run, such that the results are compatible with standard economic models.
Keywords: unemployment, work-sharing, France, VECM, forecasting
Download: More or less the version that was presented at EEA 2004 and which is now in Applied Economics, 2006, vol. 38, no. 17 (September), pp. 2053-2068: Final version . Es gibt auch eine deutsche Version des Papiers (Juni 2005) (830KB!). Earlier version presented at the European Association of Labour Economists (EALE) 2003 meeting, and an even earlier version in German appeared as DIW Berlin discussion paper no. 362.
Abstract: We analyze the macroeconomic impact of the French work-sharing reform of 2000 (a reduction of standard working hours in combination with wage subsidies). Using a vector error correction model (VECM) for several labor market variables as well as inflation and output we produce out-of-sample forecasts for 2000/2001. A comparison of these forecasts -which serve as a benchmark simulation without structural shifts- to the realized values (with shifts) suggests significant beneficial employment effects of the policy mix. Other shifts were absent and thus cannot explain the outcome. Output, productivity, hourly labor costs, and inflation are only transitorily affected or not at all.
Keywords: insider-outsider unemployment, learning by doing, pay-as-you-go pension system
Download: Drastically improved revision (with slightly changed title) of Free University Berlin economics discussion paper 2001/06 which was presented at the 2001 European Economic Association (EEA) meeting and at the 2001 Institute for the Study of Labor (IZA) conference "Pension Reform and Labor Markets". (An older version was posted as research report 2001/3 at the Centre for Pensions and Social Insurance, dead link now.) Published in JITE (Journal of Institutional and Theoretical Economics), 2005, vol. 161, no. 4 (December), pp. 708-728: Final version May 2005
Abstract: If workers gain an insider position through past activity, young workers will bear the resulting outsider unemployment burden. In a world where productivity of employed workers rises because of learning-by-doing, and where labor demand is sufficiently elastic, preventing this unemployment (by lowering wages) leads to a higher income tax base in the future. Thus the institution of certain intergenerational transfer schemes provides an incentive for insiders to lower wages. In a stylized overlapping generations model I show that this effect partially or fully abolishes unemployment in the steady state equilibria.
An earlier version was presented at the 2001 International Institute of Public Finance meeting. A still older version was presented in 1999 at the Fiscal Affairs Department of the International Monetary Fund (IMF), where it all started as my summer internship project.
Abstract: This paper estimates the cost of the minimum pension
guarantee in Chile’s individually funded pension system. It
uses a stochastic simulation to assess the fiscal impact of the
dynamics of the relevant variables, namely wages and asset
This is the first aggregate study with wage stochastics, and it also accounts for so-called “recognition bonds” reflecting workers’ contributions to the previous defined benefit system. It is found that the beneficiaries of the guarantee are mainly the currently non-contributing affiliates of the system, provided they will not permanently drop out of the labor force. Women as a group also tend to receive more transfers than men.
The ex-ante cost distribution for the baseline case lies in the range of 0 to 20% of Chile’s 1998 GDP with a median of about 4%. The impact of various institutional settings and parameter values is assessed.
Abstract: In the mid 1980s, the German Council of Economic Experts suggested a "productivity-oriented wage policy" rule. Some time series analytical considerations show that such a policy will fail to stabilize the labor share in the long run, except in the special case of a Cobb-Douglas production function. In general, the necessary cointegrating relationships cannot be established because the policy rule is formulated in terms of growth rates instead of levels. This finding holds for forward-looking rational as well as for backward-looking expectations. Some evidence is presented showing that the problem applies to the West German economy in the period 1980-1994.