Data Mapping

As global investments become more decentralized, a growing number of assets are being auctioned off to the highest bidder, rather than relying upon traditional sales mechanisms. For governments and private corporations alike, auction mechanism design has become an increasingly important question. In particular, sellers of assets, ranging from intellectual property rights to physical land, are aiming to design auctions to maximize revenue – the type of auction chosen depends upon the nature of the good (for example, wireless spectrum rights are, often, non-exclusive) as well as the allocation of the rights (leased-temporary or permanent-fixed.) Analysts have actively worked to better understand auction data in order to determine the best strategy for both sales and bidding purposes. In order to better evaluate the underlying auction data, economists have formulated several classes of auction models:

Game Theory Models
Utilized to understand and model strategic auction behavior, each bidder has an unspecified demand function for the item in question. Certain models, which are based on private valuations, relate to instances where there is private information (such as private estimates on intellectual property), while other models are based upon public information (for example, when companies are bidding on rights to a revenue stream.) In particular, the marginal bidding decision is based upon the expected private bidder surplus, which is the difference between the valuation and the bid price. Often times, bidders end up overpaying for an item based upon the winner’s curse, a condition where competitive bidding leads to suboptimal outcomes in the context of competition.

Generalized Second-Price Auctions
A type of sealed-bid auctions, Vickrey auctions define a broad class of auctions where the winner pays the second-highest price for the good. Most commonly, this structure is used in Internet auctions where the winner bidder pays the second highest bid plus a marginal amount over that bid, as determined through proxy bids. Another form of generalized 2nd price auction were the FCC’s wireless spectrum auctions to allocate spectrum to telecommunications companies for cellular and data communications.

Ascending, English Auctions
The “traditional” auction format is based upon an opening suggested bid, followed by visible bids, which gradually increase the price until the winning bidder is solidified. These auctions are common in the art and collectibles world, although variants have emerged to allow for proxy bids by global bidders. In order to estimate and forecast English Auction outcomes, analysts must create an estimated value for the item, as many collectible auction houses do.

Just as innovations in science are crucial to improving health care outcomes, so are the resource allocation decisions that determine the most cost-effective course of treatment. At the intersection of these ethical and economic questions, analysts have been focusing on improving ways to measure health care outcomes to influence better public and private policy decisions.

The basic problem facing the field revolves around how to value various care options, as well as how to structure pro-health incentives outside of care; policy makers must structure an insurance system that provides for broad coverage without creating moral hazard, or conditions which might lead to sub-optimal consumer behavior. As a result, the problems facing health care economists are quite difficult, especially in the context of the various interest groups in the public policy sector. Modern health care outcome models are based upon consumers as both implicit producers, and, indirect consumers, of health. In this sense, health is a type of human capital, which can be augmented based upon care, education and proper decisions, parallel to the role that education plays.

Where health analysts commonly disagree is how to find the “optimal” level of health care – while early models suggested the optimal health investment occurs where marginal benefit equals marginal costs, measuring these variables (objectively) has proved to be difficult. Additionally, the public-private nature of the industry has made it difficult for policy makers to separate ethical and normative considerations from more objective measures. One of the most difficult questions is how to pool risk without denying the benefits of private coverage – the costs of treating uninsured patients has increasingly fallen upon government and private institutions, which are forced to pass on these costs to “healthy” patients. By pooling large groups of individuals together, policy makers are creating models that reduce overall risk and incentivize preventative care.

As legal concepts have evolved over recent decades, statistical techniques have come to play an increasingly important role in providing evidence for litigation. One of the greatest values that legal models provide is in helping to guide firms’ decisions with respect to legal complaints before a trial ever occurs; cost-benefit analysis can facilitate the process of settlements or contract conditions that can help companies to make better forward-looking decisions. As efficiency-based solutions have become a core part of corporate legal practice, risk modeling has emerged as a prominent cost-saving technique. Some of the key contributions to concepts of efficiency in law and economics include:

The Coase Theorem
Formulated by economist Ronald Coase, the Theorem poses an efficient solution of property rights in the context of externalities. At a basic level, the solution states that costs can be incorporated into decisions between private parties in the absence of regulation or government regulation. The original problem came about when Coase analyzed property rights in the context of radio frequency interference – while the FCC sought a solution by seeking to change the structure of the FM and AM allocations, Coase proposed in his paper “The Problem of Social Cost” that stations should be allowed to freely buy and sell frequency levels, allowing them to buy up blocks of frequencies close to theirs to eliminate complications. This solution has been, in turn, applied to a variety of legal problems related to externalities, especially in environmental law, and led to Coase being awarded the Nobel Prize for Economics. In practice, the Theorem has greatly influenced tort law, when Judge Hand began applying cost-benefit analysis to property dispute cases. Many scholars have argued that transactions costs and regulations are too costly to allow the private system to work, especially in cases of indirect pollution and unseen costs.

Kaldor Hicks Efficiency
A central concept in economic efficiency related to Pareto efficiency conditions, Kaldor-Hicks seeks to identify conditions in which utility of various parties might be improved through alternate allocation. In a standard sense, allocations are considered Pareto-efficient where the utility of at least one party is improved, while no others are harmed – on the other hand, Kaldor-Hicks allows for the “winning” parties to monetarily compensate the losers in order to clear a transaction. The theory has wide-ranging implications for property and environmental law, in addition to contribution broad solutions to management problems involving conflicting party interests.

The solution to many industrial problems can be understood through the lens of optimization: proper optimization techniques provide a solution in terms of inputs into a process to help businesses or researchers make resource allocation decisions. With applications across a broad set of problems, ranging from manufacturing to airline plane allocation decisions, optimization techniques have helped companies improve outcomes across nearly every sector:

Optimal Control Models
As the basis for variational calculus optimization, optimal control emerged from engineering and found a variety of applications in economic planning. In particular, the system defines a series of differential equations that model the path of costs according to changes in variables. By starting from an initial condition (state), the model seeks to understand how to maximize a function limited by a series of constraints – for example, engineers may want to evaluate the best possible fuel economy of a given design based upon road conditions and optimal control can help suggest improvements in controls to help improve these outcomes. In general, problems in the field are non-linear and often solved by way of numerical methods, which require computing solutions such as MATLAB to find a numerical solution.

Convex Optimization
As the basis for optimization on classical economic problems, convex optimization seeks to answer the question of finding the maximum or minimum values within a function. As a result, a variety of problems, from theoretical utility maximization to least-squares regression analysis, rely upon convex techniques to find solutions. A common problem within the field is to minimize a function (such as costs) subject to constraints (such as output requirements or labor costs) to determine optimal business allocation, especially when managers are allocating resources across various locations.

While traditional data modeling techniques are often applied to business problems, many of the innovative models actually originate from academic research. One of the fastest growing areas of scientific modeling is in the field of computational biology, which applies data models to cellular phenomena. Although analysts may not realize it, many of the techniques that are commonly applied today can be traced to this field. In order to better understand some of the forthcoming ideas in the modeling field, we review some of these research areas:

Bioinformatics
A new field within molecular biology, bioinformatics applies databases to help solve modeling problems in biology. Among the most innovative techniques within the field is the development of large scale databases to develop accurate models of protein structures. As a result, the field has led to innovations in data mining and machine learning techniques which have been ported over to business analysis.

Computational Genomics
In the racing to model the human genome, Craig Venter led a private research team which utilized computing power to out-paced government-funded efforts. Since that time, leading scientists have been working on techniques to improve the speed of genetic analysis, which has greatly accelerated many efforts in pharmaceutical research.

Molecular Modeling
A broad field that has helped to provide a better understanding of the behavior of molecular compounds for improvements in material science and chemical research, molecular modeling has been used to help scientists better understand protein folding and enzyme behavior. The field has, therefore, been central in helping to design new, improved materials and drugs, as well as leading to a number of software programs which are now also used in social sciences such as Gaussian.

As financial firms seek to better model and predict risk, especially across a variety of portfolio holdings, many economists are turning back to the Capital Asset Pricing Model (CAPM). During the recent market downturn, many investors sought increased returns, while paying little attention to the true risk of the underlying investments; CAPM, on the other hand, seeks to help structure portfolio allocation decisions based upon market risk factors which cannot be strictly controlled.

The model formed the basis for innovations in financial theory, which were recognized with a Nobel Prize awarded to economists Harry Markowitz and Merton Miller. In particular, the model allows for the expected returns on a class of assets to be the base interest rate (essentially, the rate on government bonds – the risk-free rate) plus the risk premium factors (scaled by beta, or the movement of the asset relative to the larger market.) Over the past few years, many forecasters created models in which the risk factors were under-stated based upon the belief that returns on investments would continue to grow (especially in real estate) while the overall health of the market would remain steady. In fact, these forecasts were created in order to justify securitization of assets, by passing them on to 3rd party investors. Modern portfolio theory, in practice, established an equilibrium of returns relative to risk; any anomalies in return will be adjusted based on improved information – where risk is greater, prices of assets will fall. One of the bases of portfolio theory is optimization – financial analysts seek to find the “efficient frontier”, which is the highest level of return for a given level of risk allowance – diversifying assets across several classes and markets, while adequately accounting for risk is the direction that financial models are beginning to return to.

Consumer theory is a field that has emerged from research in both psychology and economics to help marketers better understand (and serve) their customer base. In particular, models of consumer behavior seek to identify the variables which impact purchasing decisions, including price (endogenous) and external factors such as preferences, habits and product presentation. Business analysts increasingly turn consumer models to help forecast market trends in order to set prices, as well as determining future product lines.

Behavioral analysts go beyond traditional regression analysis to seek out variables which are difficult to measure, such as heuristics (the “rules of thumb” which form cognitive biases), as well as market inefficiencies (such as “bubbles” and other non-rational forms of behavior.) While traditional models of consumer decision making, attributed to pricing, competition and income constraints can explain many purchasing decisions, researchers have found that intangibles, including habits and recommendations, often have a large impact upon consumer decisions.

In order to incorporate these factors, economists have moved beyond basic marginal analysis to understand decision making through prospect theory: developed by Nobel Laureates Kahneman and Tversky, Prospect Theory analyzes consumer behavior through the lens of bounded rationality – consumer make decisions using their existing information (reference points) relative to their expected goals. Data analysts often critique traditional economic models for assuming perfect foresight and information on the part of consumers; the theory, on the other hand, allows for an explanation of biases such as risk aversion. As a result, when firms set prices or make decisions on their market offerings, it is common for product managers to review larger (qualitative) market trends, which are increasingly being incorporated into data models. The future of consumer theory is creating data sets from behavior that has traditionally been difficult to measure.