Johannes Leitner

Experiments on forecasting behavior with several sources of information - A review of the literature

Decision makers frequently have to forecast the future values of a time series (e.g. the price of a commodity, sales figures) given several sources of information (e.g. leading indicators, forecasts of advisors). As a subdomain of decision theory the explanation and the improvement of human forecasting behavior are interdisciplinary issues and have been subject to extensive empirical field and laboratory research. We here review the relevant experimental literature, demonstrate the significance of these results for decision science in general, and summarize the implications for practical forecasting applications.

Heuristic modeling of expectation formation in a complex experimental information environment

Abstract Academic subjects made judgmental forecasts of a graphically presented time series in a laboratory experiment. Besides the past realizations of the time series itself, the only available information for the forecasting task was provided by leading series, i.e. indicators with a constant lead period of one. The number and the quality of the leading series were varied systematically between seven versions of the experiment resulting in different levels of information complexity. We present a heuristic that explains the subjects’ average forecasting behavior better than the rational expectations hypothesis in all versions of the experiment. Furthermore, we find that the forecasting accuracy of the subjects increases with the number of reliable indicators but their efficiency declines with increasing complexity.

A systematic comparison of professional exchange rate forecasts with the judgemental forecasts of novices

Professional forecasters in foreign exchange markets are not able to beat naive forecasts. In order to find reasons for this phenomenon we compare the empirical forecasts of experts with the experimentally generated forecasts of novices for the EUR/USD exchange rate in three different forecast horizons. Although the subjects are only provided with the realizations of the exchange rate and are not supported by any statistical procedures they outperform experts in accuracy. Professionals consistently expect a reversal of forgoing exchange rate changes whereas novices extrapolate trends. The judgemental forecasts appear to be unbiased and professionals appear to be biased. We demonstrate that professionals are influenced by the fundamental value—an irrelevant anchor in speculative exchange markets. The poor performance of the experts is not a common failure of human decision-making in market environments but caused by misleading information.

Expected utility versus the changes in knowledge ahead

We present a decision theory appropriate for use in serious choices such as insurance. It extends standard decision theories like expected utility or cumulative prospect theory which are atemporal single stage theories. Instead it employs stages of knowledge ahead to track satisfactions and dissatisfactions. In the first stage of the risk, the uninsured face dissatisfactions of worries and planning difficulties (avoided by the insured), also perhaps positive satisfactions of thrills (missed out by the insured). In the second stage when the risk is past, the uninsured may face the dissatisfactions of ridicule and blame if they learn that they were unlucky. From experimental and questionnaire data, 80% of our subjects are influenced by such secondary satisfactions. Only five percent of our participants employ the usage of integrated quantitative aggregation rules for evaluating acts as assumed under expected utility theory.

Biases of Professional Exchange Rate Forecasts: Psychological Explanations and an Experimentally Based Comparison to Novices

The empirical performance of macroeconomic exchange rate models is more than disappointing. This dismal result is also reflected in the forecasting capabilities of professional analysts: all in all, analysts are not in a position to beat naive random walk forecasts. The root for this deficient outcome stems from the fact that professional forecasts are to a large extend influenced by actual changes in exchange rates. A reasonable explanation for this behaviour can be taken from the behavioural finance literature. To test whether this characteristic tends to be general human behaviour in an uncertain environment, we analyse the forecasting behaviour of students experimentally, using a simulated currency series. Our results indicate that topically-oriented trend adjustment behaviour (TOTA) is a general characteristic of human forecasting behaviour. Additionally, we apply a simple model to explain professional and student forecasts.

Modeling Expectation Formation Involving Several Sources of Information

Abstract Experimental studies of expectation formation of subjects are predominantly limited to the prediction of one single time series despite the practical relevance of expectations in situations with multiple sources of information. In this paper, we report on an experiment in which subjects are given time series (indicators) as additional information for the judgemental forecast of a stationary time series. The quality and the number of these indicators are varied in three versions of a forecasting experiment. We explore the effects on forecasting accuracy and we test the average forecasts of the subjects for consistency with the rational expectations hypothesis. A simple heuristic is presented that explains the average forecasting behavior better than the rational expectations if indicators are presented to the subjects. It is demonstrated by a simulation study that this result is representative for the considered stationary stochastic processes.

Expectation formation in an experimental foreign exchange market

Participants of an experimental foreign exchange market forecast an exchange rate with an unknown price reaction function. Aggregate demand is derived from their own forecasts and random shocks. Our experimental results indicate that the expectations of the subjects tend to be coordinated on a common prediction strategy. This strategy is best described as a trend-extrapolative, destabilizing expectation formation scheme. Deviations from common expectations are mainly caused by random shocks, which can be ascribed to the similarity of the subjects’ behavior within and between the different markets. The findings can be explained using insights of behavioral economics.

A bivariate Sarmanov regression model for count data with generalised Poisson marginals

We present a bivariate regression model for count data that allows for positive as well as negative correlation of the response variables. The covariance structure is based on the Sarmanov distribution and consists of a product of generalised Poisson marginals and a factor that depends on particular functions of the response variables. The closed form of the probability function is derived by means of the moment-generating function. The model is applied to a large real dataset on health care demand. Its performance is compared with alternative models presented in the literature. We find that our model is significantly better than or at least equivalent to the benchmark models. It gives insights into influences on the variance of the response variables.

Modelling Judgmental Forecasts under Tabular and Graphical Data Presentation Formats

The accuracy of statistical time series forecasts is a critical factor for the situationspecific application of a model. Makridakis and Hibon (1979) were the first to empirically explore the performance of various statistical models in forecasting competitions on a data set of thousands of real time series. It was found inter alia that simple procedures, such as exponential smoothing, perform equivalently to sophisticated models (Makridakis and Hibon, 2000) — a result supported by many other authors. Although statistical models were the initial interest of forecasting competitions probably the most common forecasting approach was incorporated soon: judgmental forecasting. Judgmental forecasts are based on subjective eyeballing of the past realizations of the time series without the support of statistical procedures — a technique which seems to be inferior to statistical procedures at first glance. Lawrence et al. (1985) applied 111 real-life time series of the Makridakis forecasting competition (Makridakis et al., 1982) in a forecasting experiment and compared the accuracy of judgmental forecasts to statistical models. Judgmental forecasts were at least as accurate as statistical models, and in some cases even superior to them. The authors also identified the influence of data presentation formats on the accuracy: Forecasts of time series presented in tables significantly outperformed graphs for annual time series (long run). They also found table forecasts to be more robust, i.e. smaller standard deviations of the forecasting errors. The authors attribute the differences to the inability of tabular forecasters to

Modifications in the bulk and the surface of unbleached lignocellulosic fibers induced by a heat treatment without water removal: effects on fibre relaxation of PFI-beaten kraft fibers

The effect of heating lab- and mill cooked refined and unrefined unbleached kraft pulp in a rotating lab reactor at different temperatures, pH and consistencies was evaluated based on sheet and fiber properties. The observed significant decrease in tensile strength with increasing temperature was attributed to the drop in fiber bendability and to the drastic increase in fiber curl. The increased fiber curl at increased temperature (labelled as “reversed latency”), was attributed to two potential mechanisms: (a) the removal of hemicelluloses from the fiber bulk and the re-aggregation of the fibrillar cell wall structure and (b) the local damage of the fiber surface and the presence of the fully softened hemicelluloses–lignin-matrix which caused the fiber to bend and kink.ZusammenfassungIn dieser Studie werden die Faser- und Papiereigenschaften von ungebleichtem ungemahlenem und PFI-gemahlenem Kraftzellstoff in Folge einer Behandlung bei unterschiedlichen Temperaturen, pH-Werten und Konsistenzen untersucht. Die Ergebnisse zeigten beim gemahlenen Zellstoff einen signifikanten Abfall der Zugfestigkeit mit steigender Temperatur. Die gesunkene Zugfestigkeit wurde mit einer geringeren Faserflexibilität und einem signifikanten Anstieg der Faserkräuselung mit steigender Temperatur begründet. Die steigende Faserkräuselung, gekennzeichnet als “reversed latency”, wurde zwei möglichen Mechanismen zugeschrieben: (a) dem Herauslösen von Hemicellulosen aus der Faserwand und der daraus resultierenden Reaggregation der fibrillären Zellwandstrukturen und (b) einer lokalen Schädigung der erweichten Faserwand in Folge der Temperaturbehandlung, wodurch die Fasern lokal knickten.