Bernhard Böhm; Gerhard Tintner · 1979
Bernhard Böhm and Gerhard Tintner’s essay, originally published in 1979 and supplied here in a republication of undocumented date, develops and estimates a disequilibrium model of Austria’s aggregate labour market. Its central claim is methodological: observed employment cannot consistently be treated as a market-clearing quantity within a model intended to explain disequilibrium. The essay proceeds from economic specification through statistical estimation and data description to comparisons of equilibrium and disequilibrium results. Its substantive outcome is a qualified argument that registered unemployment can conceal the cyclical extent of involuntary unemployment.
The conceptual starting point is that observed employment need not reveal both labour supply and labour demand:
Unterstellt man aber, wie beim Ungleichgewichtsansatz, daß die beobachtete Menge nicht dem Marktgleichgewicht entspricht, so muß diese Beobachtung entweder auf der Angebots- oder auf der Nachfragekurve liegen; man weiß allerdings von vornherein nicht auf welcher der beiden.
English translation: If, however, one assumes, as in the disequilibrium approach, that the observed quantity does not correspond to market equilibrium, then this observation must lie either on the supply curve or on the demand curve; one does not, however, know in advance on which of the two.
The authors therefore assume that employment equals the smaller of the quantities supplied and demanded at the prevailing wage. This “minimum condition” makes the unobserved side of the market an object of estimation and assigns observations to supply- or demand-constrained regimes. They acknowledge that the assumption is imperfect, adopting it partly to keep estimation manageable.
The departure from market clearing retains a largely neoclassical account of behaviour. Labour demand follows marginal-productivity theory and a CES production function incorporating technical progress. Labour supply follows utility maximization over consumption and leisure, with real wages and non-labour income as explanatory variables. Applying these individual-level relationships to aggregate behaviour is explicitly approximate. A wage-adjustment equation then connects real-wage growth to excess demand for labour; extensions introduce prices and union membership to investigate influences not captured by the basic adjustment mechanism.
The stochastic model adds independent normally distributed disturbances and derives the likelihood of observed employment and wages by integrating over unobserved supply or demand. Maximum-likelihood estimation consequently estimates both structural coefficients and the division of observations between regimes. Its implementation, however, imposes substantial limits. Numerical difficulties and computational expense prevent calculation of coefficient standard errors:
Auch findet das Programm nur lokale Maxima der Likelihoodfunktion, die den Startwerten am nächsten liegen.
English translation: The program also finds only local maxima of the likelihood function, those nearest to the starting values.
Sensitivity to starting values is thus central to interpreting the results, rather than a peripheral technical problem. The empirical investigation uses annual data for 1954–1975, including employment, real compensation, output, property and entrepreneurial income, registered unemployment, and labour-force potential. Equilibrium estimates obtained through ordinary, two-stage, and three-stage least squares provide the benchmark. Their high explanatory power does not resolve multicollinearity, unstable coefficients, or weak measurement of non-labour income. Union indicators fail to establish the expected influence on wages, while price changes add little explanatory power.
Disequilibrium estimation improves some theoretically important coefficient signs, but the authors do not equate a higher likelihood with economic credibility. The specification containing observed labour-force potential achieves the highest local maximum, yet effectively reproduces that variable in estimated supply. Its implication that labour demand exceeded supply in nearly every year would classify almost all unemployment as voluntary—an outcome they reject as unrealistic. Among the basic specifications, alternative starting values produce different regime assignments. For variant 2:
Die Stichprobenaufteilung ergibt, daß über den gesamten beobachteten Zeitraum Angebotsüberschuß herrscht.
English translation: The partition of the sample indicates that excess supply prevails throughout the entire observation period.
Other variants alternate between supply and demand constraints. The authors favour disequilibrium estimation because selected specifications yield more plausible wage responses and stronger adjustment to the estimated imbalance, but their comparison remains heuristic without coefficient standard errors. Adding prices does not produce a significant improvement in a likelihood-ratio test; union membership again fails to establish the hypothesized effect.
The final discussion distinguishes registered unemployment from the model’s estimated involuntary unemployment, defined through excess labour supply relative to estimated supply. The principal specifications identify excess supply in 1955–1959, 1967–1969, and 1975, periods associated with weaker activity. Their estimated unemployment paths have broadly similar shapes but different levels. Official figures may therefore understate involuntary unemployment during downturns, while including relatively more voluntary unemployment during prosperous years. The authors caution against taking the numerical estimates too literally and compare their 1975 finding with Rothschild’s independently revised unemployment rate.
Abschließend soll noch auf die interessante Tatsache hingewiesen werden, daß die durchschnittliche unfreiwillige Arbeitslosenrate des Modells der größten Plausibilität (Version 4), ziemlich genau mit der durchschnittlichen amtlichen (beobachteten) Arbeitslosenrate übereinstimmt.
English translation: In conclusion, attention should also be drawn to the interesting fact that the average involuntary unemployment rate of the model with the greatest likelihood (version 4) agrees quite closely with the average official (observed) unemployment rate.
This convergence does not erase the distinction between the measures: similar averages can conceal different cyclical movements and compositions. The essay’s enduring contribution is to make the interpretation of observed employment and unemployment depend explicitly on market-regime assumptions, while exposing how strongly empirical conclusions depend on specification, measurement, and numerical estimation.
This work was divided into 8 sections when it entered the library's research corpus—an apparatus for search and citation, not necessarily the author's own table of contents. Each title opens its summary.
Put a question to this work; the Librarian answers from its 8 sections and cites the passage.
Ask the Librarian