Management Science and Production Economics (MSAP)

The research group in Management Science and Production Economics (MSAP) focuses on research in and development of decision support systems, often utilizing big, complex, private, real-world data, as well as the analysis of market design, contracts, methods for cost sharing and models for the analysis of efficiency and productivity.

Management Science and Production Economics are both quite broad and well established research areas. Within these areas the Centre will pay particular attention to the following four main themes: 

Production and Efficiency Analysis: This topic is about understanding producer behaviour. Based on an aggregate description of the technology and cost structure a series of economic aspects such as, comparisons with relevant competitors, and firm specific improvement potentials, can be analyzed. We consider production planning and optimal production flows. We consider benchmarking and measurement of productive efficiency using various techniques from Production Economics such as Data Envelopment Analysis (DEA) and Stochastic Frontier Analysis (SFA).

Applied Game Theory and Market Design:This topic is about understanding agents strategic behaviour in various market situations and using this knowledge to design well-functioning markets. In terms of methods, we apply the equilibrium analysis of non-cooperative game theory as well as the analysis of various solution concepts in cooperative game theory. We further consider the approaches of mechanism design and auction theory.

Network Structure and Externalities: This topic is about understanding the structure of networks and the positive synergies that they create. The study of networks is interdisciplinary, but we mainly focus on issues of allocation, efficiency, network formation and how network synergies influence the market structure and business models for the firms involved. The term network should be understood quite broadly including both physical and social networks.

Cost Sharing and Pricing Methods: This topic is about understanding the effects of using different types of allocation rules when sharing costs or benefits connected with joint activities. Correct allocation of costs (benefits) is often a prerequisite for economic efficiency in the sense that it sustains cooperation among agents. For individual firms, allocating cost can becomes a vital part of ongoing efforts to improve various operations and processes and may also help to identify the relative profitability of products.

MSAP takes part in the following larger research projects:

• CFEM: Center for Foundation of Electronic Markets
• COBE: Confidential Benchmarking

Director Jens Leth Hougaard

Co-director Kurt Nielsen

By Kurt Nielsen

The technology is called Secure Multiparty Computation (SMC) and the exchange was also the first large application of this technology. The technology basically extend the way we usually think about trust by distributing the job as "trusted third party" among multiple parties. As a participant submitting private information to the central SMC coordinator, one does not have to trust any individual persons or institutions. In stead, one have to trust that the individuals that constitute the coordinator does not collaborate with the purpose of corrupting the system. This technological extension of traditional trust has interesting economic characteristics and applications, which is explored by members of the centre.

On the exchange used for trading sugar beet contracts, the purpose was to provide confidentiality about the bids and asks. A part from a general reluctance to reveal private information, the reason was also to avoid any leakages that could influence the sugar beet growers negotiation position in the frequent negotiations about the very same sugar beet contracts. The system may also be used more directly to address corruption. If the sealed information is competing bids for constructing a road or an airport, information in the competitors' bids is of large value. Bribing officials in procurement situations is not unusual; in fact, the World Bank estimated the cost of bribing in procurement auctions to be around 12 % of the turnover in 2000.

The figure below illustrates how a Secure Multiparty Computation system may look like. In this situation a number of participants submit encrypted information (e.g. bids on an auction). The information is kept encrypted forever and the 5 assigned TTPs (Trusted Third Parties) do the required computations collectively. Only the result is revealed (decrypted). As illustrated several of the TTPs may in fact be corrupt without compromising the system (illustrated by the two TTPs with horns): 

This cross-disciplinary research has been going on since 2004 and is continuing. The experiences have shown that the collaboration between computer scientists and economics is fruitful and necessary. As economists we contribute with:

• Analysis of the consequences and values of the control of information and central coordination in different economic situations. 
• Analysis of the properties of the system 
• Formulation or reformulation of the required computations in different economic situations

References:

The research project (2004-2006): Secure Computing Economy and Trust

The research project (2006-2009): Secure Information Management and Processing

The research project (2010-2013): Confidential Benchmarking

The research center (2010-2016): Center for Research in the Foundation of Electronic Markets

Internet auctions like eBay have made auctions available for everyone. Not only is this a relatively new market place for both regular (like bicycles) and rare (like arts) items, the strategic element stimulate competition and excitement. As a result bidding behavior can from an economists point of view look mysterious, and traditional game theory comes short. 

By Kurt Nielsen.

When a bank decides to lend out money it is critical for the bank to rate the likelihood that the customer will pay back the loan with interests - this is called credit rating.

If the borrower is a firm, then it is important for the bank not only to evaluate the associated assets that may be realized in case of failure, but also to evaluate how well the firm perform relative to its peer. The relative performance is a good indication of the firms chance of meeting its obligations with the bank.

The credit rating requires skilled judgement, proper analysis as well as representative data for the different types of customers. For the individual bank it is important to have access to critical performance data from a large number of representative firms and to monitor how exposed the bank is in different sectors relative to other banks.

In the research project COBE we are developing software and statistical analysis for confidential credit rating. The software is designed to merge encrypted data from different banks and other data sources without disclosing the data itself (data remains encrypted). The statistical analysis is done directly on the encrypted data and only the results intended for the individual bank, will be revealed. The case is agriculture and the demo software combine encrypted information from a large database with detailed performance data from up till 10.000 Danish farms with specific information from different banks. Hereby, the individual banks can run confidential analysis on their customers based against a very large encrypted data set and retain relative performance results for individual costumers as well as for the bank as a whole.

Confidential credit rating create more transparency not by opening up confidential data, but by sealing information such that it can be used by others without compromising the confidentiality. As such, the system is Proof-of-Concept for various collaborative systems that aim at better coordination of information and actions without sharing sensitive information. Another broad field of application may be various types collaborative supply chain coordination among independent firms.

The project is a collaboration with The Danish Bankers Association, Copenhagen as Financial IT Region (CFIR), The Knowledge Center for Agriculture (VFL) and Danish banks.

By Jens Leth Hougaard

Networks play an increasingly important role in the economy and the daily lives of people around the world both in the form of physical networks (electricity grids, pipelines, roads etc.) and in the form of virtual networks (wireless transmission systems, distribution systems, social networks; friends and social connections etc.). With the increasing role it also becomes increasingly important to understand the various forms of synergies that arise from networks as well as how network structure influences the involved agents behavior.

One important aspect concerns cost allocation issues in networks. Linking agents are typically associated with some kind of cost. It may be the very cost of establishing the connection itself, but it may also be maintenance cost or congestion costs that matters. In any case, a group of agents (persons, firms, organizations etc.) connected via the network face the total cost associated with the network, which has to be distributed among the agents involved. This is not a trivial matter because of the complex synergies (externalities) involved. If the allocation of costs is wrong it will lead to sub-optimality of the network. For example, history has shown us many examples of cases where some group of agents has been unfairly charged and consequently created their own network with a resulting inefficient structure of the net as a whole.

It is therefore important to find allocation rules, which sustain an efficient (cost minimal or utility maximizing) network structure. In the research project CFEM we analyze theoretically how various types of allocation rules perform in this respect. We further try to apply our analysis to problems like waste water disposal and cost sharing in various forms of computer networks and power grids.

Visit cs.au.dk/research/centers/cfem for more information.

By Mette Asmild

It is increasingly important to keep track of best practice within and among firms. Advanced benchmarking or efficiency analysis compare production units and provide valuable information about best practice, improvement potentials, benchmarks, and relative strengths and weaknesses.

One example of an application of such benchmarking analysis is for a large demolition company, where we compared different projects that had been completed recently. This way we manage to identify the more and the less successful projects, in a more comprehensive way than by simply looking at e.g. profit margins.

We were able to determine the characteristics of the more successful projects, which can then be used for the company when determining which types of projects to focus on in the future, for instance machine intensive as opposed to labor intensive projects.

We also investigated differences in performance between different subgroups of projects, for example those handled by different project managers. This is useful information in the evaluation of the performance of the manager, as represented by the portfolio of projects, all of which are assessed in a comprehensive way that takes into consideration the underlying characteristics of the projects.

A final outcome of the analysis was the comparison of projects undertaken by a newly acquired firm with those from the mother company. The underlying assumption was that the new company was more efficient, which was supported by the empirical analysis, but we found that any advantages were caused by superior management of projects in the new company rather than by having access to a superior technology on the contrary we could determine that it was the mother company which had the better production technology.

By Camilo Andres Franco de los Rios

Biogas plants allow producing heat and electricity in an efficient and sustainable way from the available natural resources and agricultural by-products. Hence, municipal planners in Denmark have to decide on the best locations for constructing new biogas plants. This is a multi-criteria problem where municipal planners need decision support in the form of recommended solutions, considering the multiple attributes and restrictions associated with location problems, along with the uncertainty existing over large amounts of data [1]. We propose an interactive methodology that handles imprecise data and ranks alternatives according to their relevance, noting that this methodology is also appropriate for handling complex decision problems under uncertainty such as:

Valuation of subjective perceptions such as meat tenderness, bitterness of beer, sweetness of coffee or body of the wine: A subjective concept such as meat tenderness depends on multiple one-dimensional components. For example, the meaning of tenderness can be represented by means of more basic attributes: peak-stress, water holding capacity, colour, etc., but also, from experts opinions on the perception of tenderness. Each of these components participates with a given weight in the aggregated multi-dimensional concept of tenderness. In consequence, using an automatic procedure for comparing different samples, they can be properly ranked according to the desired criteria, offering on-line and interactive support both for producers and consumers.

• Estimation of the resilience of a network with application to Disaster Risk Reduction (DRR) [2]: Resilience is the cornerstone of DRR. DRR consists in reducing disaster risks through a systems approach to understand and manage the different causal factors of natural disasters. Hence, resilience refers to the ability of a community to resist, absorb, react and recover from the effects of a catastrophic event. In this sense, resilience depends on distinct factors/criteria that can be directly measured, such as the proximity to water and power sources, population density, access to health services, etc.. As a result, different geographical areas can be analyzed and its respective degree of resilience can be estimated, offering support for identifying the zones with lesser degree of resilience, i.e., the zones where it is more urgent to act upon for DRR.

References
[1] C. Franco, M. Bojesen, J.L. Hougaard, K. Nielsen. The Fuzzy WOD Model with application to biogas plant location. Proceedings of the 8th International Conference on Soft Computing Models in Industrial and Environmental Applications. Salamanca, Spain, September 11-13, 2013.
[2] UN System Task Team on the POST-2015 UN Development Agenda. Disaster risk and resilience. UNISDR, WMO, May 2012. Link >>

By Camilo Andres Franco de los Rios

Under the common European Union strategy for developing new technologies for controlling the food production chain [1], it is necessary to develop reliable methodologies for estimating food quality parameters (such as sugar levels in sugar beans or in coffee, levels of vitamins in fruit or the tenderness of meat) in a non-destructive and automatic way.

Different imaging techniques can be used for extracting information from plantations and food samples, such as spectroscopic images which capture frequencies that are not visible to the human eye. As the technology for capturing better images increases, the mathematical techniques for exploiting the large amount of data (more resolution means more pixels) commonly refer to standard statistical procedures, which fail to extract all of the available knowledge that is hypothesized to be contained in the images.

We propose to use machine learning and computational intelligence methods (such as artificial neural networks and fuzzy logic) in order to improve the capacity of the mathematical software for extracting the desired knowledge from high-resolution images. In this way, the monitoring of the food production chain can be done at any stage of the process in an automatic and reliable manner.

Besides safety and control issues, the proper identification (by means of classification or prediction procedures) of quality parameters of food-commodities (how food is understood from a market-economics viewpoint) allows examining a basic and foundational problem, i.e., how the value of a commodity is formed. Such problem refers to how the prices reflect the products quality, what is the correlation between the quality value of the product and the market price, how the preferences of the consumer can be satisfied and how contracts can be followed and monitored. In this way, the development (under a proof-of-concept framework) of automatic methods for non-destructive estimation of food quality, permits controlling the complete food production chain, all the way from where it is raised until it reaches the consumers hands.

References
[1] Foodbest. Position paper for the EIT Strategic Innovation Agenda open consultation.