Data and Economics 101
As more organizations try to determine where best to deploy their limited budgets to support data and analytics initiatives, they realize a need to ascertain the financial value of their data and analytics – which means basic economic concepts are coming into play. While many of you probably took an economics class in college not too long ago, some more “seasoned” readers may be rusty.
The starting point for this topic began with a blog that I wrote several months ago titled “Determining the Economic Value of Data” and this key observation that started that conversation:
Data is an unusual currency. Most currencies exhibit a one-to-one transactional relationship. For example, the quantifiable value of a dollar is considered to be finite – it can only be used to buy one item or service at a time, or a person can only do one paid job at a time. But measuring the value of data is not constrained by those transactional limitations. In fact, data currency exhibits a network effect, where data can be used at the same time across multiple use cases thereby increasing its value to the organization. This makes data a powerful currency in which to invest.
So to better understand how economics can help determine the value of an organization’s data and analytics, I sought the help of an old friend who is passionate about applying economics in business. Vince Sumpter (Twitter: @vsumpter) helped deepen my understanding of some core concepts of economics, and consider where and how these economic concepts play in a business world that is looking for ways to determine the financial – or economic – value of their data and analytics.
The economic concepts that seem to have the most bearing on determining the economic value of data (and the resulting analytics) that this blog will cover include:
- Postponement Theory
- Multiplier Effect
It is our hope that this blog fuels some creative thinking and debate as we contemplate how organizations need to apply basic economic concepts to these unusual digital assets – data and analytics.
I found the below two definitions of “economics” the most useful:
- Economics is the science that deals with the production, distribution, and consumption of commodities. Economics is generally understood to concern behavior that, given the scarcity of means, arises to achieve certain ends.
- Economics is a broad term referring to the scientific study of human action, particularly as it relates to human choice and the utilization of scarce resources.
I pulled together what I felt were some of the key phrases to come up with the following definition of the “economics of data” for purposes of this blog:
Economics of Data: The science of human choice and behaviors as they relate to the production, distribution and consumption of scarce data and analytic resources.
Economics is governed by the law of supply and demand that dictates the interaction between the supply of a resource and the demand for that resource. The law of supply and demand defines the effect that product or service availability and the demand for that product or service has on price. Generally, a low supply and a high demand increases price, and in contrast, the greater the supply and the lower the demand, the lower the price tends to fall.
Now, we will explore the most relevant economic concepts in context to the Economics of Data.
Scarcity refers to limitations—insufficient resources, goods, or abilities to achieve the desired ends. Figuring out ways to make the best use of scarce resources or find alternatives is fundamental to economics.
Figure 1: Scarcity Ramifications
Scarcity is probably the heart of the economics discussion and ties directly to the laws of supply and demand. Organizations do not have unlimited financial, human or time resources, consequently as we discussed previously, organizations must seek to prioritize their data and analytic resources against those best opportunities. Scarcity at its heart forces organizations to do two things that they do not do well: prioritize and focus (see “Big Data Success: Prioritize ‘Important’ Over ‘Urgent’”).
Scarcity plays out in the inability, or the unwillingness, for the organization to share all of its data across all of its business units. For some business units, scarcity drives their value to the organization; that is, he who owns the data owns the power. This short-sighted mentality manifests itself across organizations in the way of data silos and IT “Shadow Spend.” For example, if you are a Financial Services organization trying to predict your customers’ lifetime value, having analytics that optimize individual business units (checking, savings, retirement, credit cards, mortgage, car loans, wealth management) without seeking to optimize the larger business objective (predict customer lifetime value) could easily lead to suboptimal or even wrong decisions about which customers to prioritize with what offers at what times through what channels.
Scarcity has the biggest impact on the prioritization and optimization of scarce data and analytic resources including:
- Are your IT resources focused on capturing or acquiring the most important data in support of the organization’s key business initiatives?
- Are your data science resources focused on the development of the top priority analytics?
- Does your technical and cultural environment support and even reward the capture, refinement, and re-use of the analytic results across multiple business units?
Consequently, the ability to prioritize (see “Prioritization Matrix: Aligning Business and IT On The Big Data Journey”) and carefully balance the laws of supply and demand are critical to ensure not only that your data and analytics resources are being prioritized against the “optimal” projects.
Postponement is a decision to postpone a decision (which is itself a decision). It can occur as one party seeks to either gain additional information about the decision and/or to delay for better terms from the other party.
Postponement has the following ramifications from an economics of data perspective:
- Case #1: Organizations may decide to postpone a decision in order to gather more data and/or build more accurate analytics in order to dramatically improve the probability of making a “better” decision
- Case #2: People and organizations may postpone a decision in order to get better terms especially given certain time constraints (e.g., car dealers get very aggressive with their terms near the end of the quarter)
While Case #2 may not have an impact on the economics of your organization’s data and analytics, Case #1 has direct impact. In order to make a postponement decision, organizations need to understand:
- What is the estimated effectiveness of the current decision given Type I/Type II decision risks (where a Type I error is a “False Positive” error and a Type II error is a “False Negative error)? See “Understanding Type I and Type II Errors” for more details on Type I/Type II errors.
- What data might be needed to improve the effectiveness of that decision?
- How much more accurate can the decision be made given these new data sources and additional data science time?
- What are the risks of Type I/Type II errors (the costs associated with making the wrong decision)?
Efficiency is a relationship between ends and means. When we call a situation inefficient, we are claiming that we could achieve the desired ends with less means, or that the means employed could produce more of the ends desired.
Data and analytics play a major role driving efficiency improvements by identifying operational deficiencies and proposing recommendations (prescriptive analytics) on how to improve operational efficiencies.
The aggregation of the operational insights gained from efficiency improvement might lead to new monetization opportunities in enabling the organization to aggregate usage patterns across all customers and business constituents. For example, organizations could create benchmarks, share, and index calculations that customers and partners could use to measure their efficiencies and create goals around efficiency optimization from the aggregated performance data.
The multiplier effect refers to the increase in final income arising from any new injection of spending. The size of the multiplier depends upon household’s marginal propensity to consume (MPC), or the marginal propensity to save (MPS).
The Multiplier Effect is one of the most important concepts developed by J.M. Keynes to explain the determination of income and employment in an economy. The theory of multiplier has been used to explain the cumulative upward and downward swings of the trade cycles that occur in a free-enterprise capitalist economy. When investment in an economy rises, it can have a multiple and cumulative effect on national income, output and employment.
The multiplier effect is, therefore, the ratio of increment in income to the increment in investment.
When applied to our thinking about the Economics of Data, the multiplier effect embodies the fact that our efforts to develop a new data source, or derived analytic measure, could have that same multiplier effect if the new data/analytics were to be leveraged beyond the initial project.
For example, when CPG manufacturers worked with retailers to implement the now ubiquitous UPC standard in the early 1980’s, their primary motivation was a desire to drive more consistent pricing at the cash register… Few imagined the knock-on benefits that would accrue by now having much deeper understanding of actual product movement through the supply chain…let alone the shift in balance-of-power that subsequently ensued from CPG Manufacturer to today’s Retailers!
Figure 2: Multiplier Effect
Price elasticity of demand is the quantitative measure of consumer behavior that indicates the quantity of demand of a product or service depending on its increase or decrease in price. Price elasticity of demand can be calculated by the percent change in the quantity demanded by the percent change in price.
In today’s big data environment, the price of data science resources (i.e. their salaries) seems almost price inelastic (inelastic describes the situation in which the quantity demanded or supplied of a good or service is unaffected when the price of that good or service changes). That means that the demand for data science resources is only slightly affected when the price of data science resources increases.
This price inelasticity of data science resources can only be addressed in a few ways: train (and really certify) more data scientists or dramatically improve the capabilities and ease-of-use of data science tools.
However, there is another option: train your business users to “think like a data scientist.” The key to this process is training your business users to embrace the power of “might” in collaborating with the data science team to identify those variables and metrics that might be better predictors of performance. We have now seen across a number of projects how coupling the creative thinking of the business users with the data scientists can yield dramatically better predictions (see forthcoming blog: “Data Science: Identifying Variables That Might Be Better Predictors”).
The “Thinking Like A Data Scientist” process will uncover a wealth of new data sources that might yield better predictors of performance. It is then up to the data science team to employ their different data transformation, data enrichment and analytic algorithms to determine which variables and metrics are better predictors of performance.
Capital is already-produced durable goods and assets, or any non-financial asset that is used in production of goods or services. Capital is one of three factors of production, the others being land and labor.
Adam Smith defined capital as “that part of a man’s stock which he expects to afford him revenue”. I like Adam Smith’s definition because the ultimate economic goal of data and analytics is to “afford organizations revenue.” And while it may be possible to generate that revenue through the sale of data and analytics, for most organizations data and analytics as capital get converted into revenue in four ways:
- Driving the on-going optimization of key business processes (e.g., reducing fraud by 3% annually, increasing customer retention 2.5% annually)
- Reducing exposure to risk through management of security, compliance, regulations, and governance, to avoid security breaches, litigation, fines, theft etc. to build customer trust and loyalty while ensuring business continuity and availability.
- Uncovering new revenue opportunities through superior customer, product and operational insights that can identify unmet customer, partner and market needs
- Delivering a more compelling, more prescriptive customer experience that both increases customer satisfaction and advocacy, but also increases the organization’s success in recommending new products and services to the highest qualified, highest potential customers and prospects
Probably the most important economic impact on data and analytics is the role of human capital. Economists regard expenditures on education, training, and medical care as investments in human capital. They are called human capital because people cannot be separated from their knowledge, skills, health, or values in the way they can be separated from their financial and physical assets. These human investments can raise earnings, improve health, or add to a person’s good habits over one’s lifetime. But maybe more importantly, an organization’s human capital can be transformed to “think differently” about the application of data and analytics to power the organization’s business models.
As my friend Jeff Abbott said after reviewing this blog: “What did I do wrong to have to review this blog?”
While the economic concepts discussed in this blog likely do not apply to your day-to-day jobs, more and more I expect that the big data (data and analytics) conversation will center on basic economic concepts as organizations seek to ascertain the economic value of their data and analytics. Data and analytics exhibit unusual behaviors from an asset and currency perspective, and applying economic concepts to these behaviors may help organizations as they seek to prioritize and optimize their data and analytic investments.
So, sorry for bringing back bad college memories about your economics classes, but hey, no one said that big data was going to be only fun!