claude hillinger
homepage
Claude Hillinger, Ph.D. |
|
publications
My articles and working papers are listed and partly downloadable at two sites:
http://papers.ssrn.com/sol3/cf_dev/AbsByAuth.cfm?per_id=191500
http://econpapers.repec.org/RAS/phi14.htm
The research on business cycles done at my institute SEMECON at the University of Munich has led to the publication of two conference volumes and other books:
Barnett, William, A., Gandolfo, Giancarlo and Hillinger, Claude (eds.), (1996), Dynamic Disequilibrium Modeling, New York, Cambridge University Press.
Hillinger, Claude (ed.), (1992a), Cyclical Growth in Market and Planned Economies, Oxford, Clarendon Press.
Reiter, Michael (1995), The Dynamics of Business Cycles, Heidelberg, Physica-Verlag.
Süssmuth, Bernd (2003), Business Cycles in the Contemporary World, Heidelberg, Physica- Verlag.
Woitek, Ulrich (19978); Business Cycles, Heidelberg, Physica-Verlag.
Against the stream:
an unorthodox career in economics
My career in economics was shaped by a firm belief in the universal applicability of the scientific method, specifically the need to pay close attention to observations and to test theories against them. This sounds like an unobjectionable commonplace, but experience taught me that to an extent that I had not imagined at the beginning of my career, the succession of economic theories and research programs has been dominated by ideology, not by observation. By ‘ideology’ in this context I mean both the conventional political variety as well as various methodological commitments.
The fields in which I worked have a rough chronological order, with one set of interests leading logically to the next. They can be divided into two categories, those dealing with specific substantive issues and those that deal with a wider background in the philosophy of science, the history of social and economic thought and finally the role of divers ideologies. In the discussions below, I integrate these more general issues with the discussion of the applied fields.
A. Business Cycles
In the early 1960s. when I was a graduate student at the University of Chicago, there was a wide ranging consensus among economists and business analysts regarding the existence of an ‘inventory cycle’. By this was meant a characteristic alternation of aggregate inventory accumulation and decumulation with an average period of slightly more than three years. The phenomenon was regarded as so well confirmed and important that extensive congressional hearing on the subject were held. The phenomenon had also attracted the attention of Lloyd Metzler, distinguished theoretician on the Chicago faculty. He had formulated a series of models to explain how the output decisions of firms are related to their sales expectations and current inventory levels. He showed that plausible decision processes could generate the observed cycles.
I had come to economics with a strong interest in the philosophy of science, specifically in the methodology of the natural sciences. I was struck by the fact that while the existence of the inventory cycle was generally accepted, this was based on casual evidence. Little effort had been devoted to the quantitative analysis of relevant data using the best techniques available for the purpose. Moreover, no effort had been made to test Metzler’s models econometrically to see if they could quantitatively explain the observed features of inventory cycles. I decided to take on these tasks for my dissertation.
I confirmed the inventory cycle hypothesis and obtained reasonable results in fitting a modified Metzler model to the data. However, around this time the interest of the economics profession turned away from the analysis of business cycles, to the study of equilibrium growth. It was the heyday of Keynesianism and of faith in large-scale macro-econometric models. Economists believed that by means of Keynesian macroeconomic stabilization policies they had relegated the business cycle to the dustbins of history. I regarded these claims as being an ideology, unsupported by scientific evidence, but given the direction in which the profession had moved, I was tending to give up on the subject.
In 1972 I came to the University of Munich and in that year presented a paper on the inventory cycle at the IFO Institute in Munich. One of their researchers, Klaus Schüler, fitted my model to German data. It turned out that Germany had an exceptionally regular inventory cycle and my model fitted the German data substantially better than the US data. This, to me very surprising, result motivated me to pursue the research on business cycles further and to extend it in two directions. One extension was to consider the entire theory of business cycles that had evolved in the period of roughly l920-1960. This literature stressed, in addition to the short inventory cycle, the existence of a cycle of 80-10 years length in fixed investment. I determined to model both cycles and their interaction. The second extension involved the fitting of the models to most of the industrialized economies. I was able to obtain a generous grant to carry out this research.
Here I can only describe the most central finding. It is that the traditional theory of business cycles that viewed them as investment cycles is basically correct. Delayed adjustment of actual to desired capital stocks leads to overshooting and observable cycles in contemporary industrialized economies.
These findings have met with massive resistance on the part mainstream macroeconomists. The monetarist revolution against Keynesianism postulated an intrinsically stable economy disturbed only by exogenous shocks, particularly through foolish governmental policies. This was part of the tectonic shift from the political Left, committed to interventionist policies to the neoliberal Right that advocated hands-off policies. Subsequently, theoretical macroeconomics moved in the direction of imposing formal requirements that largely served to define what was regarded as acceptable: A publishable model had to feature an optimizing representative agent with rational expectations. These models have not been used by central banks and other institutions charged with economic forecasting and policy.
The legacy of the macroeconomic theorizing of the past decades is a belief in an essentially static economy that is buffeted by unpredictable shocks. The insight that emerged from the work on business cycles that I have directed is that on the contrary the economy has an inherent cyclical dynamics.
B. Economic Measurement
The history of economists’ attitudes towards measurement is an exceedingly strange one that cannot be recaptured in any detail here. The subject became involved in ideological battles, first between classical economics and marginalism, later between socialism and mainstream economics. The generation of social and economic statistics was part of the socialist agenda for social improvement via central planning. Mainstream economists argued that the common measures were largely meaningless since the only value free measure was the Pareto criterion. The upshot of a long story is that economic measurement has been largely abandoned as a research topic by academic economists and the theory as well as the concrete implementation of economic statistics has been relegated to national and international statistical agencies.
The theory of measurement is at the heart of empirical science. It is only when measurements are constructed in accordance with the theory of a subject that the resulting numbers can be meaningfully related to the theory. When I came to the subject, I found that none of the major measures employed in economics, such as the consumer price index, the real national product, or consumer surplus had an adequate justification in terms of economic theory. Moreover, I found some serious defects in the methods by means of which official statistics were produced.
Arguably, the most important economic statistics are those for the gross national product and its components in real, i.e. deflated terms. NIPA statisticians have traditionally deflated each sector with a sector specific deflator and the aggregate with an aggregate deflator. The consequence was that for real magnitudes the parts did not add up to the total. The statisticians then distributed the discrepancy over the sectors, to restore additivity. This procedure makes no sense and has since been abandoned, but without providing a solution to the problem.
My own work on economic measurement underwent a long evolution; here I will give a summary of the principal results. I have argued that all of the standard economic measures attempt to do the same thing, though possibly in different contexts. What they all attempt to do is to restore the measuring rod of the money metric that is lost when prices vary. Nominal expenditure is like a measuring rod made of rubber that is sometimes being compressed when there is inflation, or is being stretched in case of a deflation. The task then is to restore a uniform ‘length’. This can be done directly by means of an index or real expenditure, usually though inappropriately called a quantity index. Alternatively, the metric can be restored indirectly by deflating nominal expenditure with an inflation index. The two typed of index form a dual pair, given nominal expenditure and one index, the other can be inferred.
I have argued that there is only one satisfactory theory that can be constructed for the indexes. It involves the Divisia integral as the theoretical concept and the Törnqvist index as the observable approximation to it.
Regarding the national accounts I have argued that the aggregate and all sectors should be deflated using the same index. Ideally, this should be the Törnqvist inflation index based on the prices of consumer goods.
Knowledge of how to construct measurements is an important part of empirical sophistication. Another part is knowing how to use the data. Data are always discrete; this is true even when the process generating the data is continuous. The econometric mainstream assumes that since the data are discrete; the models estimated by means of these data should be discrete also. This contrasts with the practice in physics where explanatory models are almost always formulated in continuous time. There exists a substantial literature on ‘continuous-time econometrics’ where it was shown that the mainstream position is wrong. Given that economic data are generated nearly continuously and are by no means constant over the usual observation interval such as the quarter, it follows that the discrete models are seriously misspecified. The mainstream has never answered this criticism, instead ignoring it. The business cycle research described above did use continuous time econometrics to advantage.
C. Theories of Voting and Collective Choice
The most popular method used in general elections throughout the world is plurality voting. In the one round version, the candidate or issue receiving the most votes, usually less then a majority, wins. In the two rounds version, there is a runoff election between the two candidates receiving the highest vote in the first round. There exists a substantial literature, both theoretical and empirical, that demonstrates that this method has severe drawbacks. In particular, it may lead to the election of a candidate strongly opposed to by a majority of the electorate.
The theory of voting in its present, essentially mathematical form goes back to the work of Borda and Condorcet in the second half of the Eighteenth Century. This literature has grown rapidly since the 1950s, but without securing agreement as to the best, or at least an improved voting method. A decisive change occurred with the publication in 1951 of Kenneth Arrow’s Social Choice and Individual Values. This book started the more abstract theory of social choice which asks if it is possible to aggregate individual preferences so as to obtain a social preference, with the aggregation procedure satisfying certain reasonable appearing conditions. Arrow’s famous ‘impossibility theorem’ answers this question in the negative, a result that has been widely interpreted as proving the impossibility of democracy.
I was attracted to the subject by both the empirical and theoretical aspects. Voting is the central feature of democracy and if commonly used voting procedures are deeply flawed, then dealing with the problem appeared to be of paramount importance. The situation of collective choice theory also seemed quite remarkable. It is the common belief that democracies, however imperfect, do exist. The populations living in these countries hardly took notice of the proof that their form of government cannot exist!
I soon came to the conclusion that the problems of voting that theorists had vainly struggled with for 250 years, and that according to Arrow were actually insoluble, were not problems of the real world, but rather of the theorists own creation. This conclusion was not entirely original. Several theorists, most prominently John Harsanyi, had shown that perfectly reasonable axioms for collective choice could be formulated given a cardinal representation of individual preferences. This is precisely what was ruled out in voting theory and in collective choice theory in the tradition of Arrow. Harsanyi’s work was abstract and he did not apply his results to the solution of any concrete problem of collective choice such as voting. This was for me the obviously required next step. Concretely, I proposed evaluative voting in which the voter evaluates each alternative on a three point scale: +1 for a favored alternative, -1 for an alternative that is opposed, 0 or abstention in case of indifference. Under evaluative voting the voter gives a much more complete expression of her preferences than under plurality voting. The incentive to vote strategically is also much reduced.
Conclusion
About half a millennium ago Copernicus advocated the heliocentric hypothesis. For a long time it was vigorously opposed by the Church, but today it is a universally accepted scientific truth. Less well known is the fact that Copernicus also advanced the quantity theory of money as an equilibrium relationship between the supply of currency and the demand for goods. The theory was soon impressively validated when the influx of precious metals from the New World caused an expansion of currencies and concomitant inflation. Unlike the heliocentric hypothesis, economists have never agreed on the validity of the quantity theory and few seem to care. The Church is indifferent to the theory, but central bankers are not, because an implication of the theory is that discretionary monetary policy likely does more harm then good.
The historian and philosopher of science Ravetz defined an immature field as one that has no criteria for determining factual truth. That certainly applies to economics. However, nothing is static. I entered economics in the hope that it would evolve into a truly mature field of empirical science and wanted to contribute towards that end. Under the impact of various ideologies, economics moved in the opposite direction.