Upload
tinbergen-institute
View
214
Download
1
Tags:
Embed Size (px)
DESCRIPTION
Tinbergen Institute Magazine highlights ongoing research at Tinbergen Institute for policymakers and scientists.
Citation preview
magazine
Fall 2001
tinbergen institute4
Tinbergen Magazine is published
by Tinbergen Institute, the
Institute for economic research of
Erasmus Universiteit Rotterdam,
Universteit van Amsterdam and
Vrije Universiteit Amsterdam.
Fat tails and the history of the guilder
Underinvestment in work-related training?
Recent research casts doubt
Transport economics
An interview with Prof. Dr. Piet Rietveld
Fat tails and the history of the guilder
Underinvestment in work-related training?
Recent research casts doubt
Transport economics
An interview with Prof. Dr. Piet Rietveld
In depth
References
In short
Up close
2
www.tinbergen.nl
In this issue
In depth
Fat tails and the history of the guilder
by Casper G. de Vries
Underinvestment in work-related training?
Recent research casts doubt
by Hessel Oosterbeek
Up close
Transport Economics
An interview with Prof. Dr. Piet Rietveld
by Dirk Brounen
In short
Discussion papers
Papers in journals
Theses
References
Discussion papers and theses that have appeared
in the last half year
3
10
13
14
16
tinbergen institute
magazine
Fall 2001
tinbergen institute4
Tinbergen Magazine is published
by Tinbergen Institute, the
Institute for economic research of
Erasmus Universiteit Rotterdam,
Universteit van Amsterdam and
Vrije Universiteit Amsterdam.
Fat tails and the history of the guilder
Underinvestment in work-related training?
Recent research casts doubt
Transport economics
An interview with Prof. Dr. Piet Rietveld
Fat tails and the history of the guilder
Underinvestment in work-related training?
Recent research casts doubt
Transport economics
An interview with Prof. Dr. Piet Rietveld
Highlighting ongoingresearch at TinbergenInstitute for policymakersand scientists.
7
15
Extremes and ExtremistsFinancial markets are still reverberating
with the aftershocks of the extremist attacksin New York and Washington. Seldom doesbad news travel alone, and subsequentabnormal movements in asset prices havewrought havoc with investors. Financial his-tory reveals that such episodes of financialturmoil exhibit a number of common fea-tures. As has been seen in the recent periodof turbulence, asset price gyrations appearextreme and tend to come in clusters ratherthan being one-off events, regardless ofwhether the data is analysed on a minute-by-minute basis or only once per year.
Consider the respectable 400-year histo-ry of the guilder-pound exchange rate by wayof illustration (see figure 1). The earliestrecorded contracts date back to 1609; thedata for the first 300 years are summarisedin Korthals Altes (1996) and were transcribedfrom the weekly “Beurscourant” byPosthumus. Gaps are due to lost editions andsuspensions of trading due to war.
Some historic episodes are clearly recog-nisable. The stability due to the gold stan-dard from 1875-1914 and the two short-lived
restoration periods after both world warsshow up as nearly horizontal lines. Over thelast century, the Bank of England almostdeterministically revealed her true colours asan agent of the Treasury, since the pound sliddown along a near straight line. Unlike theirfrugal Dutch counterparts, English govern-ments could not withstand the seductions offiat currency and came to regard monetarypolicy as the domain of the government. Afiat currency’s value is largely determined bythe expectations (trust) about the intentionsof the (fiscal) authorities. Only when the UKgovernment in 1997 disposed of their keys tothe printing presses (seignorage tax), did thecentury-long slide come to a halt.
The first figure also shows that theNapoleonic wars and the world wars, togetherwith the depression and the post Bretton-Woods free float era, are responsible for mostof the turmoil. This can be seen even moreclearly if we plot up the squared returns as anindicator of volatility. In figure 2 we see thatsome periods clearly stand out from the rest.This is the well-observed clustering of volatileepisodes and periods of quiescence that wasfirst observed by Mandelbrot (1963). Later,Engle (1982) proposed an attractive autore-
3
tinbergen magazine 4, fall 2001
I n d e p t h
Fat tails and the history of the guilder
Casper G. de Vries●
Casper G. de Vries is professor
of Economics at Erasmus
Universiteit Rotterdam,
where he teaches monetary
economics. He is a fellow at
the Tinbergen Institute and
EURANDOM.
This article reviews the history of the guilder as a way to describe the behaviour of
speculative prices. We pay special attention to the extreme movements caused by the
fat tail nature of the distribution.
4
tinbergen magazine 4, fall 2001
gressive scheme, dubbed ARCH, that couldboth exhibit the predictability in volatility,while retaining the zero mean for the returnsreflecting the absence of arbitrage opportuni-ties. The ARCH scheme is now widely used bypractitioners (such as option traders) to hedgetheir positions.
The Droste EffectInterestingly, the clustering of low and
high volatility is present in the data,no matter what level of detail we
choose for examining the data.Consider the plots for the month-ly returns from 1766-2000, andthe weekly returns from 1915-
2000 below. If we compare thesetwo figures with the previous plot for
the yearly volatility, we see a very similarpicture after we adjust for the scale. Thebouts of volatility come in clusters–regard-less of the frequency of observations.
Moreover, two or three outliers that deter-mine the scale of the vertical axis dominateeach graph.
The remarkable observation we takeaway from these three figures is that volatili-ty is self-scaling. In Dutch scientific slang,this is often referred to as the Droste effect,reminiscent of the nurse carrying a cup ofhot chocolate and a can of Droste displayinga label with a picture of the same nurse, adinfinitum. No matter how much we zoom inor out, the same pattern in the data appears.
Fat TailsBefore we can understand this self-scal-
ing, we have to discuss a property of theunconditional return data. In almost all sci-ences the normal distribution is widely usedbecause averages have a tendency to be nor-mally distributed. Furthermore, the normaldistribution is the only finite variance distri-
14
12
10
8
6
4
2
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.001650 1700 1750 1800 1850 1900 1950 2000 1650 1700 1750 1800 1850 1900 1950 2000
Figure 1: pound-guilder 1609-2000 Figure 2: volatility (yearly)
0.10
0.08
0.06
0.04
0.02
0.00
0.08
0.06
0.04
0.02
0.0080 00 20 40 60 80 00 20 40 60 80 00 1/08/15 3/09/34 5/08/53 7/07/72 9/06/91
Figure 3: volatility (monthly) Figure 4: volatility (weekly)
5
tinbergen magazine 4, fall 2001
bution that is self-scaling. Is it thus an obvi-ous candidate for modelling financial returns?Below are histograms of the daily returnssince 1971 (figure 5) and the yearly returnssince 1609 (figure 6), overlaid by the bell-shaped normal density curve estimated fromthe mean and variance in the sample.
As is clear to the naked eye, the normalmodel fails miserably. Relative to the normal,the empirical density is too peaked, and hastails that are too fat. Mandelbrot (1963) initial-ly proposed a model that would preserve boththe self-scaling nature and the fat-tail feature.But the infinite variance assumption thatcomes with this model imposes too much tailfatness and does not generate the observedclustering of volatility. The three data featurescan be reconciled, however, if we drop theself-scaling requirement for all the data. Itturns out that the clustering and the fat-tailproperty are commensurate with the self-scal-ing, if the latter property only applies to thetail area. This brings us to the work of...
ParetoMotivated by the social debates of his
time, Pareto (1896) discovered a remarkabletime invariance regarding the distribution ofthe highest incomes. His law survives todayas the well-known Pareto distribution instatistics. Suppose we adapt Pareto’s law todescribe the distribution of the loss returns X
Pr{X≤–x}=ax–a for a>0, and x > a1/a
Note that the x represents losses, andPr{X ≤–x} stands for “the probability that thereturn X will be less than -x”(i.e. a lossgreater than x). One can understand whythese distributions are fat tailed, since inte-gration shows that moments larger than aare unbounded due to the explosion of the
argument as x tends to infinity. Note that ifwe take logarithms on both sides of thisequation, we obtain a linear relationshipbetween the log-returns and their log-frequency with slope –a
lnPr{X ≤–x}= ln a–a ln x
What do the financial data tell us?We first plot all the monthly loss returns
logarithmically transformed against their log-rank order, when ranked in descending fash-ion (see figures 8 and 9 on page 6). The sec-ond plot zooms in on the 50 highest lossesonly. If Pareto’s model applies, one should seea straight line. It is clear this only applies forthe latter diagram, which gives a regressionslope coefficient of a
^= –2.26, implying that
moments up to the variance do exist, but theskewness may very well be unbounded.
Trinity: Fat Tails, Clustering andScalingThe three asset return data features can
now be brought into agreement with eachother if we assume that the tail of the distri-bution behaves like the Pareto distribution.This is similar to the case of the income dis-tribution, where the Pareto model is knownto fit well only for the highest income brack-ets. First, there are many fat-tailed distribu-tions, such as the Student-t distributions,which exhibit the Pareto law as the first termof a (Taylor) expansion of the distribution inthe tail area, but which are quite different inthe centre. Second, our theoretical researchhas shown that the distribution of theextremes from an ARCH process, which doesmodel the clustering of volatilities, indeedadheres to the Pareto-in-the-tail model (seeDe Haan et al. (1989)). Third, for some time ithas been known that if a distribution satis-fies the Pareto model in the tail area, then
3000
2000
1000
0
200
100
0
Std. Dev = .05 Mean = -.004 N = 313.00Std. Dev = .00 Mean = -.0001 N = 7372.00
-.350
-.0425-.0375-.0325-.0275-.0225-.0175-.0125-.0075-.0025.0025.0075.0125.0175.0225.0275.0325
-.300-.250
-.200-.150
-.100-.050
-.000.050
.100.150
.200
Figure 5: histogram (daily returns) Figure 6: histogram (yearly returns)
6
tinbergen magazine 4, fall 2001
for large x, the sum of two independentdraws from such a distribution satisfies
Pr{X1+X1 ≤ -x} ~~ 2ax–a
Compare this result with the first equa-tion: Except for the factor two, the right-handsides of both expressions are equal, implyingself-scaling (with factor 2
–1/a). Recall that log-returns are additive (i.e. the yearly return isjust the sum of the monthly returns), and therelevance of the scaling feature becomes evi-dent. Only recently has it been shown thatthis scaling also holds if the draws comefrom a process like ARCH, when the returnsare time dependent (see Basrak et al. (2000)).Nowadays, the Pareto-in-the-tail model isoften used in risk-management systemsdeployed by the larger commercial banks tocalculate their risk exposure to extremeevents (see Danielsson and De Vries (2000)).
The self-scaling feature is exploited to reducethe computational burden, since the exposurehas to be computed for different investmenthorizons. New estimation techniques go farbeyond the simple graphical proceduresemployed in the current essay and allow us,for example, to discern where the tail begins,based on a trade-off between bias and vari-ance (see Danielsson et al. (2000)).
Finally, why do economic data exhibitthe fat-tail property? Suppose, for example,that the asset price is uniformly distributedon the unit interval. It follows that the grossreturn, which is just the ratio of two consecu-tive prices, has a fat-tailed distribution (witha=1). While this explanation is clearly far toosimple, current research is actively address-ing this question with elaborate models. Theacademic programme is a never-ending tale,sometimes light and sometimes heavy.
●
I am grateful to Willem Korthals Altes, Martijn van
Harten and Nora Plaisier for collecting and sharing
data. The article is partly based on a joint project
commemorating the demise of the guilder. The tech-
nical statistical work owes much to the numerous
researchers at Erasmus Universiteit Rotterdam work-
ing on extreme value theory.
ReferencesBasrak, B., R.A. Davis, and T. Mikosch, 2000, A char-
acterization of multivariate regular variation,
EURANDOM report 2000-36.
Danielsson, J. and C.G. de Vries, 2000, Value-at-risk
and extreme returns, Annales D’Economie et de
Statistique 60, 239-270.
Danielsson, J, L. de Haan, L. Peng and C.G. de Vries,
2001, Using a bootstrap method to choose the sam-
ple fraction in tail index estimation, Journal of
Multivariate Analysis 76, 226-248.
Engle, R., 1982, Autoregressive conditional het-
eroscedasticity with estimates of the variance of UK
inflation, Econometrica 50, 987-1008.
Haan, L. de, S. Resnick, H. Rootzen and C.G. de
Vries, 1989, Extremal behavior of solutions to a
stochastic difference equation, with applications to
ARCH processes, Stochastic Processes and their
Applications 32, 213-224.
Korthals Altes, W.L., 1996, Van Pond Hollands tot
Nederlandse Gulden, NEHA, Amsterdam.
Mandelbrot, B., 1963, The variation of certain spec-
ulative prices, Journal of Business, 394-419.
Pareto, V., 1896, La Courbe de la Repartition de la
Richesse.
Figure 9: Pareto plot (monthly, loss returns)
log Pr
log x
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
0
-1
-2
-3
-4
-5
-6
-7
Figure 8: Pareto plot (monthly, loss returns)
log Pr
-12 -10 -4 -2 0
0
-1
-2
-3
-4
-5
-6
-7
-6-8
7
tinbergen magazine 4, fall 2001
Policy makers have become increasinglyinterested in topics like employability andlifelong learning. The Dutch governmenteven organised a “national programme ofaction” concerning these issues. This led topolicy measures intended to stimulate partic-ipation in training. These measures are basedon the premise that there is severe underin-vestment in training and that the returns totraining are substantial.1 Yet, results ofrecent research lend little support for thisassertion. Firms and workers seem to be ableto solve the underinvestment problem with-out any governmental intervention.Moreover, the returns to work-related train-ing are likely to be much smaller than earlierstudies suggested.
Why would there be under-investment in training?Becker (1962) laid the foundation for the
economic analysis of work-related training.He introduced the distinction between gener-al and specific human capital. General humancapital enhances a worker’s productivity,regardless of the particular firm for which heworks. Specific human capital, on the otherhand, only enhances productivity if theworker stays with the current firm.2 The eco-nomic implication of the distinction is thatsince workers are able to reap all the benefitsfrom general training, they should have to
bear the full costs of training. In contrast,firms and workers should share the costsand benefits of specific training.
Underinvestment in general trainingmight occur when workers are liquidity con-strained (so that they are unable to bear thecosts). When the worker’s current employerpays these costs, the worker has an incentiveto leave once the training is finished. This isthe case because other employers who didnot pay for the costs of training are readyand willing to pay a wage exceeding the wageoffered by the current employer. This mecha-nism is often referred to as poaching.Anticipating this, the current employer willnot pay the costs of training.
A first thing to note is that underinvest-ment in general training can only occur whenworkers are liquidity constrained. But evenin that case, there are some clear strategiesto prevent poaching. One is the use of pay-back clauses; if the worker should leave thefirm within a specified period after the train-ing has been completed, he agrees to remit (apart of) the training costs. Another solutionis the use of sectoral training funds; all firmswithin a certain sector pay a percentage oftheir payroll to the sectoral fund; when theyencounter training costs, they send the billto this fund. Both solutions are used in theNetherlands, and they seem to work well.
I n d e p t h
Underinvestment inwork-related training?Recent research casts doubt
By Hessel Oosterbeek
Hessel Oosterbeek is Professor of Economics of Education at the Universiteit van Amsterdam,
coordinator (jointly with Henriette Maassen van den Brink) of the priority program Scholar,
funded by the Netherlands Organization for Scientific Research (NWO), and a Research Fellow
of the Tinbergen Institute and IZA Bonn. He received his Ph.D. at the Universiteit van
Amsterdam in 1992. His research interests include the economics of education and training.
For further information, visit his homepage: http://www.fee.uva.nl/scholar/oosterbeek.htm
8
tinbergen magazine 4, fall 2001
Underinvestment in specific trainingmight occur due to the hold-up problem. Thisoccurs when, in situations in which theinvestment is not contractible, and in whichrenegotiations are possible, non-investors arein a position to appropriate a part of thereturns to the other parties’ investment. Theinvestor thus bears the full cost of the invest-ment, but receives less than the full benefit.Consequently, (s)he has no incentives tochoose the first-best investment level.
A growing theoretical literature analysesdifferent solutions to this underinvestmentproblem. Among these are the following:structuring the renegotiations stage suchthat the investor becomes a residualclaimant, removing opportunistic behaviourof the employer by introducing up-or-outcontracts, and stipulating in the contractpenalties for parties that breach the contract.
Evidence from experimental studiesTo examine how these (sometimes very
subtle) solutions work, we conducted a seriesof experimental economic studies (Oosterbeeket al. 1999, Sloof et al. 2000, Sloof et al. 2001,and Sonnemans et al. 2001). Experiments allowus to test the predictions of well-articulatedformal theories in a controlled environmentthat allows the observations to be unambigu-ously interpreted in relation to the theory. Thekey finding of the experimental studies is thatunderinvestment due to hold-up is often much
smaller than theory predicts. The theoreticalpredictions are based on a model in whichagents are utterly selfish. This implies thatnon-investors are prepared to take advantageof their bargaining position once the invest-ment costs are sunk. But this is not what wetypically observe in the laboratory; subjectsusually reveal reciprocal behaviour. Thismeans that they are prepared to forgo somefinancial gain in order to reward the fairbehaviour of others or to punish unfairbehaviour. In some of the experiments, suchresponses support outcomes that deviate fromthe theoretical predictions and enhance effi-ciency. It is important to note that subjects inthe laboratory already reveal such behaviour.In our experiments all subjects are completelyanonymous, and considerations regarding rep-utation are absent. In real employer-employeeinteractions, where such considerations docome into play, deviations from theoreticalpredictions are likely to be even greater.
New results on the (wage)returns to trainingAn important piece of empirical evidence
in favour of the presumed underinvestment intraining is found in the very high wagereturns to training that are reported in somestudies. An example is Frazis and Loewenstein(1999), who find an 8 percent wage increasefor 40 hours of training. Many of these esti-mates seem to be biased, however, becausethey do not take proper account of the non-random assignment to training. This meansthat firms are likely to offer training to thoseworkers who might benefit from it, and onlyworkers who think it is worthwhile will followthe training. This is of course a standard (self-)selection problem that hinders causal infer-ences. In the related field of measuring thereturns to (formal) schooling, much progresshas been made during the past 10 to 15 years.For decades it has been thought that the typi-cal OLS estimate of the return to schoolingoverestimates the true return to schooling.Recent studies using data from identical twinsand from natural experiments strongly sug-gest that this is not the case (see Card 1999for a review).
Experimental studies
suggest that underinvestment
due to hold-up problems
is often much smaller
than theory predicts.
9
tinbergen magazine 4, fall 2001
In a recent paper, we exploit a disconti-nuity of the Dutch tax system to identify theeffect of training on wages (Leuven andOosterbeek 2001a). In January 1998, theDutch government implemented a new taxlaw permitting extra tax deductions to firmsfor their expenditures on training. One partof this law specifies an additional deductionfor training expenditures pertaining to work-ers aged 40 or older. This additional deduc-tion makes it 14 percent cheaper to train aworker who is 40 years or older than to traina younger worker. Being 40 years or olderthen serves as an instrumental variable thatcan identify the causal effect of training onwages. When we take narrow samples aroundthe discontinuity of 40, workers older than40 are the treatment group and workersyounger than 40 represent the control group.The age-dependent tax law serves as the ran-dom device that sorts workers into the twogroups. It turns out that being 40 years orolder is a good predictor of training partici-pation. For samples close to the age of 40,the difference in training rates is 15 to 20%in favour of those older than 40. Using pre-dicted training participation as the instru-mental variable in the wage equation givesan estimated wage return to training not sig-nificantly different from zero.
In Leuven and Oosterbeek (2001b), weuse two specially designed survey questionsto identify the non-trained workers who want-ed to follow training but did not, due to somerandom event. It turns out that 22 percent ofthe workers who did not follow any trainingduring the 12 months prior to the interviewsaid that they actually did want to followtraining, but did not do so. The questionnairelists different reasons for not following thedesired training. Among these are “lack oftime”, “too expensive”, “lack of support ofemployer” and “other reason”. In addition, wealso explicitly listed “some random event”,and gave as examples family circumstances,illness of children or own illness. A total of77 respondents mentioned some randomevent as the reason for not following thedesired training course. These respondentsare arguably a much better control group for
the trained workers than the larger group ofall non-trained workers. They were motivatedto follow training (they wanted to), and onlydue to chance did they fail to do so. With allnon-trained workers as control group, theestimated wage return to training is at least9%. When the control group is restricted tothe 77 workers, the return is never signifi-cantly different from zero.
Thus, using two entirely differentapproaches, we find basically the same result.In the short run, the wage returns to work-related training are small. Given the actualamounts of training involved (the medianequal to 60 hours), this is hardly surprising.But the findings contrast strongly with otherresults that reported very high (short-run)returns to training-results that have been usedto motivate government intervention.
To summarise, recent research findingsprovide little support for the idea that there isunderinvestment in training. To precludeunderinvestment, firms and workers have attheir disposal many tools: training funds, pay-back clauses, wage contracts, up-or-out con-tracts, specially designed wage contracts andreciprocal behaviour. These instruments areused in practice, and seem to work very well.
Moreover, the high wage returns to trainingfound in earlier studies, which have beeninterpreted as evidence of underinvestment,seem to be seriously biased. Thus, the Dutchgovernment seems to have little reason to sub-sidise training activities in the private sector.
Notes1 In this contribution, the term “training” refers to
work-related training followed by workers who have
a job. It does not refer to training programs for
unemployed people.
2 Until recently, it has been thought that the difference
between the two types of training is determined by the
contents of the training course. Stevens (1994) and
Acemoglu and Pischke (1999) have shown that labor
market imperfections may render training that is gen-
eral in a technical sense, into training that is specific
in an economic sense.
In the short run, wage
returns to work-related
training are small.
ReferencesAcemoglu, D. and J.S. Pischke,
1999, The structure of wages
and investment in general
training, Journal of Political
Economy 107, 539-572.
Becker, G.S., 1962, Investment
in human capital: A theoretical
analysis, Journal of Political
Economy 70, 9-49.
Card, D., 1999, The causal
effect of education on earn-
ings, in: O.C. Ashenfelter and
D. Card, Handbook of Labor
Economics Vol. 3A, Amsterdam:
Elsevier.
Frazis, H. and M. Loewenstein,
1999, Reexamining the returns
to training: Functional form,
magnitude, and interpretation,
Bureau of Labor Statistics.
Leuven, E. and H. Oosterbeek,
2001a, Evaluating the effect of
tax deductions on training,
Scholar WP 24/01.
Leuven, E. and H. Oosterbeek,
2001b, A new method to esti-
mate the wage return to work-
related training, mimeo.
Oosterbeek, H., R. Sloof and
J. Sonnemans, 1999, Promotion
rules and skill acquisition: an
experimental study, Scholar WP
07/99.
Sloof, R., E. Leuven, H. Ooster-
beek and J. Sonnemans, 2001,
An experimental comparison
of reliance levels under alter-
native breach remedies,
Scholar WP 21/01
Sloof, R., J. Sonnemans and
H. Oosterbeek, 2000, Specific
investments, hold-up, and the
outside option principle: an
experimental study, Scholar WP
08/00.
Sonnemans, J., H. Oosterbeek
and R. Sloof, 2001, On the rela-
tion between asset ownership
and specific investments, Eco-
nomic Journal 111, 791-820.
Stevens, M., 1994, A theoretical
model of on-the-job training
with imperfect competition,
Oxford Economic Papers 46,
537-562.
For samples of employees
close to the age of 40,
the difference in training
rates is 15 to 20% in favour
of those older than 40.
10
Up
close
By Dirk Brounen An interview with Prof. Dr. Piet Rietveld
Transport Economics
Piet Rietveld is Professor of Transport Economics in the Department
of Spatial Economics at the Vrije Universiteit in Amsterdam.
After receiving his Ph.D. in economics at the Vrije Universiteit in
Amsterdam, Professor Rietveld worked at the International Institute
of Applied Systems Analysis in Vienna, and was research co-ordinator
at Universitas Kristen Satya Wacana in Salatiga in Indonesia.
Currently, he ranks among the most productive economists in
the Netherlands and is Fellow at the Tinbergen Institute.
Transport obviously comes in manyshapes and forms and has developed rapidlyin the past. Have economists studying this sec-tor been able to keep pace?
Transport is a field of economic researchwhere key concepts and methods were devel-oped and applied for the first time–like con-sumer surplus, and the logit model. Forinstance, the concept of ‘consumer surplus’stems from the 19th century and was devel-oped by a French transport economist whostudied the construction of channels. A morerecent example is Daniel McFadden, whoreceived the Nobel Prize of 2000 for his workin microeconometric applications in thechoice of transport modes. Transport eco-nomics has always offered unique opportuni-
11
tinbergen magazine 4, fall 2001
ties to apply and develop new concepts, andthe scientific challenges remain vast.Typically, transport economics is differentiat-ed into subfields relating to various means oftransport. I have worked in most of thesefields, on the price elasticity of aviationdemand, on the economic impact of high-speed rail systems, on congestion pricing onhighways, and even on future applications ofpipelines for underground logistical systems.Of all fields, road transportation has typicallyreceived the most attention, and I have writ-ten many papers on this subject as well.
Transport economics can result in usefulapplications for practical problems. Can youdescribe how science and practice interact inyour field of expertise?
Scientific insights can reach society indifferent ways. In the case of transport eco-nomics, one very fruitful link with societyexists through students who become policy-makers after graduation. Through thelessons these alumni have learned at the uni-versity, science can contribute to the solu-tion of transport-related problems. Directtransfers of scientific insights to decision-makers can sometimes be more difficult.Transport economists tend to base their solu-tions on optimal welfare outcomes, whichcould lead to distribution effects that are toodrastic for political success. Technology cansometimes hamper the implementation ofscientific breakthroughs as well. An exampleof this type of limitation was the develop-ment of congestion taxes by Pigou in 1920.Pigou created a tax system that charged roadtravellers for their contribution to traveldelays (congestion). Although the system wasoptimal in theory, the collection of thesetaxes in practice would easily create morecosts than the initial problem itself. Somecurrent technological developments, how-ever, such as ICT, are favourable.
Perhaps a good alternative example of theimportance of technological limitations couldbe found in the recent debate on the choicebetween the simple cordon charge (tolpoort) orthe more sophisticated kilometre charge (kilo-meterheffing). It appears that technology hasdominated this debate, don’t you think?
Indeed, technology has played an impor-tant role in these debates, but we should alsotake into account the aim of the proposed
policy instruments. Charge systems can beused to finance infrastructures (actually,infrastructural improvements), which used tobe the case for the Zeelandbrug in theNetherlands. But charge systems can also beused as regulatory tools. In that case, thegovernment tries to control road transport inorder to solve problems like environmentaldamage, road safety and congestion. JanTinbergen correctly claimed that fundamen-tally different problems cannot be solvedwith a single policy instrument. For example,the traditional cordon-charge system is sim-ple in nature and charges people crossing acertain point. Such a system can be used as afinancing tool and also as a means to addresscongestion, but it performs poorly when thegoal is to reduce environmental damage orthe number of traffic accidents. So far, tech-nological constraints have kept more sophis-ticated systems from competing with the cor-don charge, but recent ICT innovations haveexpanded the possibilities drastically.
Does the proposed kilometre charge fitthe profile of a scientifically optimal solution?
Well, consider the fact that commutersare associated with low price elasticities,whereas recreational travellers exhibit highelasticities. Charging systems that cannot dis-tinguish these variations might not be effec-tive in solving a problem like rush-hour con-gestion. The sophisticated kilometre charge isable to make the distinction along differentdimensions–both with respect to travellingtime and area, and regarding the pollution fac-tor of a car type. These complex systems,
The concept of ‘consumer surplus’ stems from
the 19th century and was developed by a
French transport economist who studied the
construction of channels.
12
tinbergen magazine 4, fall 2001
which involve roadside sensors, on-boardblackboxes and satellite communication,might therefore seem too futuristic to be true.Yet, Switzerland introduced a similar system–including all the high-tech trappings–at thebeginning of this year for lorry traffic, and sofar the system appears to be working satisfac-torily. Whereas economic theory discoveredthe optimal solution long ago, technology hasfinally made the necessary advances thatmake the solution feasible in practice.
According to the ‘Mobility and WelfareReport’ of CPB Netherlands Bureau forEconomic Policy Analysis, the Dutch economywould profit from the introduction of a flexi-ble kilometre-charge system. Total automobilemileage is predicted to decline some 14% by2020, while time losses due to road conges-tion will decrease by 32%. This would createan estimated annual gain in welfare of 4 to 5billion guilders. Do you share this positiveview?
Our team actually found similar results.Apart from examining these long-termeffects, we also studied the impact of a kilo-metre charge on welfare distribution. Thecharge itself will have a negative effect onthose who will have to pay the bill. However,if these tax revenues can be distributed toroad users through lower fixed vehicle taxes,then road users receive some compensation
and yet face an incentive to minimise theircar mileage. Those who frequently use theircars (like commuters) will face an increase intheir net expenses, whereas sporadic driverswill enjoy lower costs. As long as a tight reinis kept on the implementation costs of thecharging system, beneficial welfare effectsare likely to result.
The introduction of the system may facesome obstacles in the short term, but in thelong run the kilometre charge is likely to pre-vail. A very simple parallel to this discussioncan be found in the development of parkingpolicies. At first, people were accustomed to
free parking, which resulted in parking chaosin city centres. Although drivers protestedthe introduction of parking charges by localauthorities, they have gradually become usedto paying for parking spaces in certain areas.This system allows people a choice: freeparking in more remote areas or paid parkingin the more popular central areas.
Given all these recent innovations, whatdo you think the future holds for transportand transport economics?
When considering the future of trans-port, people tend to dream about futuristicsystems of gliding cars and magnetic levita-tion high-speed trains that shoot through theair. History, indeed, has proven that meansof transport are associated with life cycles inwhich innovations improve efficiency untilan alternative is developed that can replacethe existing type. Figure 1 (below) clearlyillustrates these transport lifecycles overtime. This graph suggests that roads appearto have matured by now, and that innova-tions are on the horizon. I expect that tech-nological innovations will remain important.But I also believe that future gains in trans-port efficiency will come from organisationalimprovements rather than technological dis-coveries alone. In a recent paper I showedthat much of the delays that occur in trans-port are due to poor internal connections.Currently, the different transport networkslike trains, metros and roads are not gearedto each other and operate as independententities. I believe that the future lies in net-work integration, leading to seamless trans-portation connections.
The introduction of the system may face some
obstacles in the short term, but in the long run
the kilometre charge is likely to prevail.
Whereas economic theory discovered the optimal solution
long ago, technology has finally made the necessary
advances that make the solution feasible in practice.
Source: Gruebler and Nakicenovi (1991)
Figure 1: Long transport lifecycles over time
Producing and
Manipulating
Information
What policy maker wouldn’tat least occasionally jump atthe chance of having a lookinto the future to see theoutcomes of certain policies?Since the consequences ofmany policies are complicat-ed and difficult to foresee,policy makers often try toward off failure by askingexperts to provide informa-tion on the effects of poli-cies. Sometimes, the expertalready possesses the infor-mation. Often, the informa-tion has to be produced.This paper is concerned withsituations in which a policymaker needs informationthat must be produced. Anexpert thus has to be moti-vated to collect information.However, a policy maker willfind it difficult to ascertainhow much effort an experthas put into acquiring infor-mation. Moreover, when anexpert has an interest in thepolicy outcome, he or shemay manipulate informationto distort policy decisions. This paper shows thatexperts who are unbiasedtoward the policy outcomeput in the highest effort incollecting information. Suchneutral experts thus produceinformation of high quality.Perfect communication, how-ever, requires that the prefer-ences of the policy makerand the expert coincide. For
instance, a policy maker whois strongly biased againstimplementing a particularpolicy may be reluctant tohire a neutral expert,because such an expert willtoo often recommend imple-mentation of the policy.Hence, when selecting anadvisor, a policy maker facesa trade-off between qualityof information and theextent of communication. Inaddition, this paper showsthat collecting informationgives rise to an externality:policy-driven experts wantother policy-driven expertsto give advice. The resultingcompetition can only driveup the price of hiring anexpert.
By Robert A.J. Dur,
Otto H. Swank (EUR)
“Producing and Manipulating
Information: Private
Information Providers vs.
Public Information Providers”
TI01-052/1
Time SeriesModelling of Daily TaxRevenuesThe production of daily fore-casts of tax revenues is animportant task of day-to-daycash management at theDutch Treasury Department.Statistical daily time-seriesmodels aim to process infor-mation regarding revenuesof previous days systemati-cally and efficiently. Dutchcentral government outflowsare usually known at leastone day ahead. Therefore,multiple-day-ahead time-series-model forecasts ofrevenues can also be used tomonitor the targets for thebudget.
Daily economic time-seriesdata often have propertiesthat make them harder to
model and to forecast thanmonthly or quarterly data(for which numerous stan-dard solutions exist). In addi-tion to the well-known fea-tures typical of monthlydata–trend, season, tradingday and calendar effects–there are two major prob-lems with daily data. First,the number of observationsvaries per month and peryear, which leads to a timeseries with irregular spacing.Second, account must betaken of daily heteroskedas-ticity, since the varianceoften depends on the day ofthe month. Many aggregateeconomic transactions havepatterns with a clear peakonce a month (e.g. salarypayments, money circula-tion, and tax revenues). It isoften difficult to stabilise thevariance by taking logs: the(persistently changing) sea-sonal pattern is not simplymultiplicative, and neither isthe irregular component.
The analysis starts with aperiodic regression modelwith time-varying parame-ters. The model is thenextended with a componentfor intra-month seasonality,which is specified as astochastic cubic spline. Thestochastic cubic splines weuse are smooth functions ofthe banking-day-of-the-month. These functions varyslowly but randomly overtime. State space techniquesare used for recursive esti-mation and evaluation, asthey allow for irregular spac-ing of the time series. Statespace techniques allow forautomatic online updating ofestimates of the trend com-ponents and seasonal pat-
terns in the tax revenues, asnew observations are addedover time. The correspond-ing statistical frameworkfacilitates a systematiconline evaluation of the dif-ferent components of themodel.
By Siem Jan Koopman
and Marius Ooms, (VU)
“Time Series Modelling of
Daily Tax Revenues”
TI01-032/4
MotivationsandPerformance Conditions for
Ethnic
Entrepreneurship
Ethnic entrepreneurship hasbecome a popular concept ina modern multi-cultural soci-ety for solving inter alia thestructural unemploymentproblems of ethnic groups incities. The concept hasbecome an important aspectof modern urban develop-ment policy. This case study of ethnicentrepreneurs in Amsterdamshows that there is no clearpanacea for successfulentrepreneurship. Rather, avariety of critical success (orfailure) factors determine thecommercial performance ofethnic firms (such as languageskills, commercial knowledge,market insight, network contacts, access to venturecapital, ICT skills, etc.).
13
tinbergen magazine 4, fall 2001
discussionpapers
All discussion papers
can be downloaded via
www.tinbergen.nl
14
tinbergen magazine 4, fall 2001
Successful entrepreneurs areable to develop effectivestrategies for enteringmature market segments,rather than focussing exclu-sively on reaching their ownethnic group. Once estab-lished, such entrepreneurswill eventually turn to diver-sification strategies. Thecase of Amsterdam confirmsthese informal economy fea-tures for the phenomenon ofethnic entrepreneurship.
By Enno Masurel, Peter
Nijkamp, Murat Tastan and
Gabriella Vindigni (VU)
“Motivations and Performance
Conditions for Ethnic
Entrepreneurship.” TI01-048/3.
Migration and
Immigrants
This paper surveys and doc-uments immigration andemigration flows for theNetherlands in the post-warperiod. Since 1961, annualimmigration has surpassedannual emigration. The posi-tion of immigrants in thelabour market has divergedmarkedly by their country oforigin. Indonesian immi-grants have strongly pro-gressed, yet those from theCaribbean occupy a lessfavourable position. Amongthe guest workers, thosefrom southern Europe havemarkedly improved theirpositions (or returned homeafter trade liberalisation inthe EEC), but those from
Turkey and Morocco stilloccupy an unfavourableposition and were hard hitby recession and de-indus-trialisation. Little is knownabout the effect of immigra-tion on the labour marketposition of natives. Onestudy quoted in the paperfinds that guest workers aresubstitutes for low-skillednatives and complements tohigh-skilled natives, thusdepressing the wages of theformer and increasing thewages of the latter. Thepaucity of information onrefugees is remarkable.
By Aslan Zorlu and
Joop Hartog (UvA). “Migration
and Immigrants: The Case
of the Netherlands” TI01-042/3.
Corporate Law
Enforcement
Perotti, E. and F. Modigliani.
Extending the recent litera-ture on the impact of thelegal system on the develop-ment of financial markets,this article argues that poor-ly enforced regulation mayexplain the relative impor-tance across countries ofbanking and security mar-kets in financing firms.Securities are most vulnera-ble to poor enforcement.When minority investors’rights are poorly protected,firms will find it difficult toraise equity capital, leadingto less financing for newrisky ventures. More general-ly, fewer firms will befinanced with outside equity,resulting in a low capitaliza-tion relative to GNP, and apredominance of internal(unlisted) equity and banklending over traded securities.A measure of poor protectionof minority investors can befound in the market valueattributed to control thepower to vote in generalshareholder meetings. A pricemeasure of this voting rightcan be extracted in countriesin which virtually identicalsecurities from the profile ofincome rights have differentialvoting rights. The differenceis termed the voting premium.In a subset of countrieswhere the votingpremium isvery large,
around 80%, corporatefinancing tends to be domi-nated by bank lending, andequity markets are muchsmaller. In the other subset,with voting premia below20/30%, external financingfor companies via listedsecurities is significantlymore abundant.
Perotti, E. and F. Modigliani,
2001. “Corporate Law
Enforcement and the
Development of Security
Markets: Theory and Evidence.”
International Review of Finance
(forthcoming).
Dealing withirrationalityMaking descriptive use of prospect theory toimprove the prescriptiveuse of expected utility
H. Bleichrodt (EUR), J.L. Pinto
and P. Wakker (UvA).
Many studies have foundthat actual human behaviordeviates from rationalitypostulates. Among thesefindings, positive construc-tive results are notoriouslyabsent. If the goal is toimplement optimal policies,but there is reason tobelieve that the measure-ments of utilities arebiased (e.g. overly riskaverse), then what arethe proper ways toproceed and to
papers in journals
15
tinbergen magazine 4, fall 2001
correct for those biases, andto obtain unbiased utilitiesafter all? This paper makesconcrete numerical propos-als for correcting biaseswhen measuring utilities forrisky decisions. The paper’sproposals are based onprospect theory, which pro-vides a logical model forirrational and biased choices.Some classical paradoxes ofutility measurement, wherelogically equivalent methodsfind different results–whilethey should find the same–are resolved by ourapproach. The differentmethods are reconciled andis found a consistent way tomeasure utilities for optimaldecisions and policies inrisky situations. The paperbriefly acknowledges thatthe ethical dilemma of eitheraccepting unreliable utilitymeasurements “as is” (simi-lar to consumer sovereign-ty), or changing these mea-surements as proposed,requires further discussion.
Bleichrodt, H., and
Wakker, P., 2001, Management
Science, forthcoming
The heart of the matter
Essays on EconomicGrowth and ImperfectCompetition
Economic growth is at theheart of economic policy.What must a country do toboost productivity and toraise the rate of growth?Much attention has beengiven to the German‘Wirtschaftswunder’ or themiraculous example of Japan.However, these countriesturned out to be ‘normal’after all–at least in terms ofeconomic performance. Thetruth is that a blueprint forhigh productivity and growthis not readily available. Evenif such blueprints should beformulated, though, ‘second-best economics’ provides thereasons for scepticism.Introducing or increasing a‘distortion’ in an already dis-torted economy may proveto be beneficial. For exam-ple, a national trade unionmay indeed be conducive tofast growth and solidmacroeconomic performance,or competition in goods mar-kets may actually lower theincentives to invest in inno-vative products, and progres-sive taxes may help to boostemployment. These exam-ples do not add up to ablueprint, let alone ablueprint in which marketsare given unrestricted lee-way. Instead, economic poli-cy benefits more from carefulanalysis of existing distor-tions and the interaction ofthese distortions with policymeasures.
Thesis: “Essays on Economic
Growth and Imperfect Markets”
by Paul J.G. Tang. Published in
the Tinbergen Institute
Research Series # 246
Customer relationships,defined as a series of repeat-ed exchanges between asupplier and the customerover time, are garneringincreasing attention in bothscience and practice. Thisattention is reflected in theenormous investments madein Customer RelationshipManagement. In this context,an interesting question ishow customers’ perceptionsof the relationship (such astheir satisfaction and thefeeling of identification withthe company (commitment))affect their behaviour. Arelated issue is how relation-ship-marketing instrumentssuch as loyalty programmesinfluence this behaviour.
The study examines cus-tomer retention, or the proba-bility that a customer keepsdoing business with the com-pany, buying new productsfrom the same firm, andspreading the informationabout the product (word-of-mouth). In addition, someaggregate indices of cus-tomer behaviour were consid-ered, like the share of a par-ticular customer in the totalpurchases of a given product,and the ability of the cus-tomer to generate profits. The results of a longitudinalstudy among customers of aDutch financial serviceprovider reveal some inter-esting patterns. The impact
of customer perceptions oncustomer behaviour appearsto be relatively small.Satisfaction positively affectscustomer retention and pur-chases of new products onlyfor those customers withlengthy relationships.Commitment positivelyaffects all types ofbehaviour. An interestingfinding is that those cus-tomers who perceive theprice of the financial prod-ucts to be fairer are less will-ing to buy additional prod-ucts from the same compa-ny; these customers seem tothus generate less profits. Apossible explanation is thatthese customers pay extraattention to prices, and“shop around” to find themost attractive offer in eachproduct category separately.As for the effects of loyaltyprogrammes, they seem tobe small, decreasing withthe customer’s commitmentto the company.
Thesis: “Analyzing Customer
Relationships. Linking
Relational Constructs and
Marketing Instruments to
Customer Behaviour.” by Peter
Verhoef. Published the in
Tinbergen Institute Research
Series # 255
theses
Analysing customer relationships:
Linking Relational Constructs and MarketingInstruments to Customer Behaviour.
16
Discussion papers
Institutions and Decision Processes
01-052/1
Robert A.J. Dur and Otto H. Swank, Erasmus
Universiteit Rotterdam, Producing and Manipulating
Information: Private Information Providers versus
Public Information Providers
01-053/1
Hsiang-Ke Chao, Universiteit van Amsterdam, Milton
Friedman and the Emergence of the Permanent
Income Hypothesis
01-054/1
Eduardo L. Giménez and Manuel González-Gómez,
Universidade de Vigo, Efficient Allocation of Land
between Productive Use and Recreational Use –
An Application to Galician Case
01-058/1
Ronald Bosman, Frans van Winden, Universiteit van
Amsterdam, Anticipated and Experienced Emotions
in an Investment Experiment
01-061/1
Peter Rodenburg, Universiteit van Amsterdam,
Tracing the Changing Measures of Unemployment in
Dutch Unemployment Statistics 1900-1940
01-062/1
David J. Dekker, Erasmus Universiteit Rotterdam,
Effects of Positions in Knowledge Networks on Trust
01-063/1
Ioulia V. Ossokina, Otto Swank, Erasmus Universiteit
Rotterdam, The Optimal Degree of Polarization
01-065/1
Maarten C.W. Janssen, Ewa Mendys, Erasmus
Universiteit Rotterdam, The Price of a Price: On the
Crowding out of Social Norms
01-068/1
Matthijs van Veelen, Vrije Universiteit Amsterdam,
Evolution in Games with a Continuous Action Space
01-072/1
Eelco Modderman, Cees Gorter, Jasper Dalhuisen,
Peter Nijkamp, Vrije Universiteit Amsterdam,
Labour Manoeuvrability and Economic Performance
in Township-Village Enterprises: The Case of China
01-075/1
J. Brinkhuis, Erasmus Universiteit Rotterdam,
V. Tikhomirov, Moscow State University, On the
Duality Theory of Convex Objects
01-083/1
Moez Bennouri, Université de Science Économique
et Gestion, Tunis, Sonia Falconieri, CREED,
Universiteit van Amsterdam, Price versus Quantity
Discrimination in Optimal IPOs
01-084/1
Sonia Falconieri, CREED, Universiteit van
Amsterdam, The Impact of Lobbying on the
Allocation of Political Authority
01-085/1
Cees Diks, Sebastiano Manzan, CeNDEF, Universiteit
van Amsterdam, Tests for Serial Independence and
Linearity based on Correlation Integrals
01-096/1
P. Jean-Jacques Herings, Universiteit Maastricht,
Gerard van der Laan, Vrije Universiteit Amsterdam,
Dolf Talman, Department of Econometrics &
Operations Research, and CentER, Tilburg University,
Measuring the Power of Nodes in Digraphs
01-099/1
Lola Esteban, José M. Hernández, Universidad de
Zaragoza, José Luis Moraga-González, Erasmus
Universiteit Rotterdam, Customer Directed
Advertising and Product Quality
Financial and InternationalMarkets
01-059/2
Bert Menkveld, Erasmus Universiteit Rotterdam,
Splitting Orders in Fragmented Markets
01-060/2
Koen G. Berden, Charles van Marrewijk, Erasmus
Universiteit Rotterdam, Maintenance Costs,
Obsolescence, and Endogenous Growth
01-069/2
Jón Daníelsson, London School of Economics, Bjørn
N. Jorgensen, Harvard Business School, Casper G. de
Vries, Erasmus Universiteit Rotterdam, Xiaogang
Yang, Chinese Academy of Sciences, Optimal Portfolio
Allocation under a Probabilistic Risk Constraint and
the Incentives for Financial Innovation
01-070/2
Namwon Hyung, Casper G. de Vries, Erasmus
Universiteit Rotterdam, and EURANDOM, Portfolio
Diversification Effects and Regular Variation in
Financial Data
01-071/2
P. Hartmann, European Central Bank, S. Straetmans,
Universiteit Maastricht, C.G. de Vries, Erasmus
Universiteit Rotterdam, Asset Market Linkages in
Crisis Periods
01-091/2
Menno Pradhan, World Bank and Vrije Universiteit
Amsterdam, David E. Sahn, Stephen D. Younger,
Cornell University, Decomposing World Health
Inequality
01-092/2
Menno Pradhan, World Bank and Vrije Universiteit
Amsterdam, Welfare Analysis with a Proxy
Consumption Measure – Evidence from a Repeated
Experiment in Indonesia
01-093/2
Menno Pradhan, World Bank and Vrije Universiteit
Amsterdam, Who wants Safer Streets? Explaining
Concern for Public Safety in Brazil
Wevalueyourinput
Please send
us an e-mail
• if you have addresschanges
• if you would like to order discussionpapers or to subscribe to e-mail notices of new discussion papers(please indicate inyour e-mail those areas in which youare interested): • Institutions and
Decision Analysis• Financial and
InternationalMarkets
• Labour, Region andthe Environment
• Econometrics andOperations Research
Thank you.
e-mail: [email protected]
http://www.tinbergen.nl
17
tinbergen magazine 4, fall 2001
Labour, Region and Environment01-049/3
Pim Klamer, Cees Gorter, Peter Nijkamp, Vrije
Universiteit Amsterdam, Retail Investments by Real
Estate Investment Trusts
01-056/3
Joelle Noailly, Jeroen van den Bergh, Cees Withagen,
Vrije Universiteit Amsterdam, Evolution of Harvesting
Strategies: Replicator and Resource Dynamics
01-057/3
Jasper M. Dalhuisen, Raymond J.G.M. Florax, Henri
L.F.M. de Groot, Peter Nijkamp, Vrije Universiteit
Amsterdam, Price and Income Elasticities of Residential
Water Demand: Why Empirical Estimates Differ
01-064/3
Galit Cohen, Vrije Universiteit Amsterdam, Marina
van Geenhuizen, Delft Universiteit van Technology,
Peter Nijkamp, Vrije Universiteit Amsterdam, Urban
Planning and Information and Communication
Technology: Ideas and Facts
01-066/3
Toshihiko Miyagi, Gifu University, Japan, Economic
Appraisal for Multiregional Impacts by a Large-scale
Expressway Project
01-074/3
D.B. Audretsch, Indiana University, CEPR, and EIM;
M.A. Carree, CASBEC, Erasmus Universiteit Rotterdam,
University Maastricht, and EIM; A.R. Thurik, CASBEC,
Erasmus Universiteit Rotterdam, and EIM, Does
Entrepreneurship reduce Unemployment?
01-080/3
James W. Albrecht, Georgetown University,
Washington DC, Pieter A. Gautier, Erasmus Universiteit
Rotterdam, Susan B. Vroman, Georgetown University,
Washington DC, Matching with Multiple Applications
01-086/3
Siv Gustafsson, Eiko Kenjoh, Universiteit van
Amsterdam, Cecile Wetzels, TNO/STB, Employment
Choices and Pay Differences between Non-Standard
and Standard Work in Britain, Germany, Netherlands
and Sweden
01-087/3
Eline C.M. van der Heijden, Tilburg University,
Jan H.M. Nelissen, Erasmus Universiteit Rotterdam,
Jan J.M. Potters and Harrie A.A. Verbon, Tilburg
University, Simple and Complex Gift Exchange in the
Laboratory
01-088/3
H.P. Huizinga, Tilburg University, J.H.M. Nelissen,
Erasmus Universiteit Rotterdam, R. Vander Vennet,
Ghent University, Efficiency Effects of Bank Mergers
and Acquisitions
01-089/3
Eline C.M. van der Heijden, Tilburg University,
Jan H.M. Nelissen, Erasmus Universiteit Rotterdam,
Harrie A.A. Verbon, Tilburg University, Should the
Same Side of the Market always move first in a
Transaction? An Experimental Study
01-090/3
Robert A.J. Dur, Coen N. Teulings, Erasmus
Universiteit Rotterdam, Education and Efficient
Redistribution
01-095/3
Giovanni Russo, Utrecht University, Aura Reggiani,
Bologna University, Peter Nijkamp, Vrije Universiteit
Amsterdam, Modelling and Estimating Modal Share
in European Transport
01-097/3
Shunli Wang, Peter Nijkamp, Onno Kuik, Vrije
Universiteit Amsterdam, Global Environment
Change Regimes
01-098/3
Shunli Wang, Peter Nijkamp, Erik Verhoef, Vrije
Universiteit Amsterdam, Modelling Externalities
between Ecological and Economic Systems
01-100/3
Aura Reggiani, Universiteit van Bologna, Thomas de
Graaff, Peter Nijkamp, Vrije Universiteit
Amsterdam, Resilience: An Evolutionary Approach
to Spatial Economic Systems
Econometrics andOperations Research
01-051/4
Michel Mandjes, Bell Laboratories/Lucent Techno-
logies, Nam Kyoo Boots, Vrije Universiteit Amsterdam,
The Shape of the Loss Curve and the Impact of Long-
Range Dependence on Network Performance
01-055/4
Paul A. Bekker, Universiteit van Groningen,
Frank Kleibergen, Universiteit van Amsterdam,
Finite-Sample Instrumental Variables Inference using
an Asymptotically Pivotal Statistic
01-067/4
Frank Kleibergen, Universiteit van Amsterdam,
Testing Parameters in GMM without assuming that
they are identified
01-073/4
Frank Kleibergen, Universiteit van Amsterdam, How
to overcome the Jeffreys-Lindleys Paradox for
Invariant Bayesian Inference in Regression Models
01-076/4
A. Galeotti, Erasmus Universiteit Rotterdam,
G. Salford, London School of Economics, Electoral
Cycles: Do they really fit the Data?
01-077/4
H. Peter Boswijk, Universiteit van Amsterdam, Testing
for a Unit Root with Near-Integrated Volatility
01-078/4
H. Peter Boswijk, Universiteit van Amsterdam, Block
Local to Unity and Continuous Record Asymptotics
01-079/4
Arjen H. Siegmann, Vrije Universiteit Amsterdam,
Optimal Saving Rules for Loss-Averse Agents under
Uncertainty
01-081/4
J.B.G. Frenk, Econometric Institute, Erasmus
Universiteit Rotterdam, G. Kassay, Babes Bolyai
University, Cluj, Introduction to Convex and
Quasiconvex Analysis
01-082/4
J.S. Cramer, Tinbergen Institute, Measures of Fit for
Multinomial Discrete Models
01-101/4
Emöke Bázsa, Peter den Iseger, Erasmus Universiteit
Rotterdam, Single Item Inventory Models
01-102/4
Emöke Bázsa, Peter den Iseger, Erasmus Universiteit
Rotterdam, Optimal Continuous Order Quantity (s,s)
Policies
18
tinbergen magazine 4, fall 2001
PhotographsHenk Thomas, AmsterdamLevien Willemse, Rotterdam
Editorial servicesJB Editing, Breda
DesignCrasborn Grafisch Ontwerpers bno,Valkenburg a.d. Geul
PrintingDrukkerij Tonnaer, Kelpen
ISSN 1566-3213
AddressesTinbergen Institute Amsterdam Keizersgracht 4821017 EG AmsterdamThe Netherlands
Telephone: +31 (0)20 551 3500Fax: +31 (0)20 551 3555
Tinbergen Institute RotterdamBurg. Oudlaan 503062 PA RotterdamThe Netherlands
Telephone: +31 (0)10 408 8900Fax: +31 (0)10 408 9031
e-mail: [email protected]
http://www.tinbergen.nl
Colophon
Tinbergen Magazine is published by
the Tinbergen Institute, an economic
research institute operated jointly by
the Economics and Econometrics
faculties of three Dutch universities:
Erasmus Universiteit Rotterdam,
Universiteit van Amsterdam and Vrije
Universiteit Amsterdam.
Tinbergen Magazine highlights
on-going research at the Tinbergen
Institute and is published twice a year.
Theses
249 GOVERT BIJWAARD (05-06-01), Structural
change in Central European agriculture. Studies
from the Czech and Slovak Republics.
251 BART VAN PRAAG (06-06-01), Earnings
Management: Empirical Evidence on Value Relevance
and Income Smoothing.
252 ERIK PEEK (19-06-01), Discretion in Financial
Reporting and Properties of Analysts’ Earnings
Forecasts.
253 NICOLE JONKER (13-06-01), Job performance
and career prospects of auditors.
254 MAURICE BUN (26-06-01), Accurate statistical
analysis in dynamic panel data models.
255 PETER VERHOEF (20-09-01), Analyzing customer
relationships: Linking relational constructs and mar-
keting instruments with customer behavior.
256 CHARLES BOS (13-09-01), Time varying parame-
ter models for inflation and exchange rates.
257 ARJAN O.J. HEYMA (06-09-01), Dynamic models
of labour force retirement; An empirical analysis of
early exit in the Netherlands.
258 SILVA DEZELAN (28-09-01), The impact of
institutional investors on equity markets and their
liquidity.
259 DAVID J. DEKKER (24-09-01), Network perspec-
tives on tasks in account management.
After serving three years om TI’s Board (with one year as chairman),Professor J.S. Cramer (left) will resign as a member of the Board as ofJanuary 1, 2002. We are pleased to introduce Dr. J.J.M. Kremers (right),who will succed Prof. Cramer.
Jeroen Kremers did his Msc. at Bristol and D.Phil. at Oxford.Together with J. Dolado, he wrote the much-cited paper, “The power ofcointegrations tests”, in the Oxford Bulletin of Economics and Statistics in1992. Since 1999, Dr. Kremers works at the Dutch Ministry of Finance.
19
tinbergen magazine 4, fall 2001
tinbergen institute
Tinbergen Research InstituteFour themes distinguish Tinbergen
Institute’s research programme:I. Institutions and Decision AnalysisII. Financial and International MarketsIII. Labour, Region and the Environment IV. Econometrics and Operations Research
Each theme covers the whole spectrum ofeconomic analysis, from theoretical to empiri-cal research. Stimulating discussions on theories, methodologies and empirical resultsarise from the interaction of the Institute’sfaculty – comprised of approximately 96research fellows. These fellows are facultymembers with excellent track records in eco-nomic research, active in organising researchactivities, teaching graduate courses andsupervising Ph.D. students.
Discussion PapersResearch is pre-published in the institute’s
own Discussion Paper Series. Download discussion papers at http://www.tinbergen.nl(section ‘Publications’). E-mail address for correspondence: [email protected]
Tinbergen Graduate SchoolThe Tinbergen Graduate School enrols
about 145 students in two programmes. Oneleads to a Master of Philosophy in economics,and the other to a Ph.D. in economics.
Master of Philosophy programmeTinbergen Institute’s intensive one-year
Master’s programme leads to a Master ofPhilosophy in economics. Both those studentsaiming for a Ph.D. in economics, as well asthose pursuing careers in top consulting- orpolicy advice organisations, stand to benefitfrom the excellent preparation offered by theprogramme. Core courses are offered in thefollowing: microeconomics, macroeconomics,mathematics for economists, econometrics,advanced econometrics, and organisation.Specialised courses are offered in the follow-ing: international trade and development,monetary economics, finance, labour eco-nomics, public economics, microeconomictheory and game theory.
Ph.D. programmeFour years of solid training in the princi-
ples of economics and econometrics (based onlectures, workshops, seminars and examina-tions), as well as the successful completion ofa supervised doctoral thesis, provide the basisfor Tinbergen Institute’s Ph.D. programme.The programme’s first year coincides with themaster’s programme. Ph.D. theses are pub-lished in the Institute’s Research Series.
For information on admission require-ments, application procedure, and scholar-ships, visit http://www.tinbergen.nl, or contact [email protected].
BoardA.G.Z. Kemna (Chair), J.S. Cramer,
S. Goyal, J. Hartog, P. Rietveld
General DirectorC.N. Teulings
Director of Graduate StudiesM. Lindeboom
Research Programme Co-ordinatorsInstitutions and Decision Analysis:
M.C.W. Janssen, F.A.A.M. van WindenFinancial Economics and
International Markets:C.G. de Vries, E.C. Perotti
Labour, Region and the Environment:J.C.J.M. van den Bergh, H. Oosterbeek
Econometrics:S.J. Koopman, R. Dekker
Scientific CouncilD.W. Jorgenson (Harvard University,
Chair), M. Dewatripont (CORE), P. de Grauwe(Leuven University), D.F. Hendry (OxfordUniversity), R.C. Merton (Harvard University),D. Mortensen (Northwestern University), S. Nickell (Oxford University), T. Persson(Stockholm University), L. Wolsey (CORE)
Social Advisory CouncilC.A.J. Herkströter (Chair), R.G.C. van
den Brink (ABN-AMRO), H.J. Brouwer (DNB),M.J. Cohen (Mayor of Amsterdam),F.J.H. Don (CPB), C. Maas (ING), F.A. Maljers, I.W. Opstelten (Mayor of Rotterdam), A.H.G. Rinnooy Kan (ING), H. Schreuder(DSM), J. Stekelenburg, R.J. in ’t Veld, P.J. Vinken, L.J. de Waal (FNV)
Editorial Board Tinbergen MagazineD. Brounen, J. Dalhuisen, B. Hof,
S. Manzan, E. Mendys, C.N. Teulings
How to subscribe?Address for correspondence/subscriptions:
Tinbergen Institute Rotterdam, Burg. Oudlaan 50, 3062 PA Rotterdam, The Netherlands. E-mail: [email protected] changes may be sent to the above e-mail address.
In this issue Fat tails and the history of the guilder
Underinvestment in work-related training?
Recent research casts doubt
Transport economics
An interview with Prof. Dr. Piet Rietveld
Discussion papers
Papers in journals
Theses