It seems like the day people gain control of and sell their personal data is coming soon. Many blockchain projects are trying to keep up to this trend and are launching protocols for ‘my data’ sales.
This article introduces how the Harberger Tax that was originally proposed as a taxation system for the land market can be applied to my data sales, and analyze the benefits.
- What is the My Data Market.?
- What is the Appropriate Price for My Data?
- New Perspective on Appropriate Price: Land Market and Harberger Tax
- Harberger Tax Implementation
- Benefits of Harberger Tax
What is the My Data Market?
The data that we create on/off-line is a significant production element that bolsters the contemporary economy and industry. Companies use data to design strategy, products, and target marketing. As such, because data allows companies to earn profit it a more effective manner, data is perceived as an asset of companies and is also traded by brokers like commodity.
Then, here’s the question — is it fair that companies reap all the profits from data? The data that companies use have our social media activity records, purchase information, medical information, and financial information. Aren’t we entitled to have a share of that profit in return for usage of our data?
The my data market project started based on this particular question. Quite literally, my data market refers to the platform on which users can sell their data and earn profit. Some representative models of this in Korea are Airbloc Protocol, Carry Protocol that sells off-line purchase data, AMO and MVL that manages and sells automobile data, and Medibloc that manages and sells personal medical data.
In the my data market, individuals who sell their data are called data sellers and the users who purchase data are called data buyers. Data buyers purchase the identifier of a certain data seller (code that identifies a person’s device, like advertisement ID) and take part in the market for the purpose of utilizing data for target marketing. Though it may vary among projects, data sales in the my data market generally follows this procedure:
- Data sellers agree to the sale of their data and disclose a portion of their data to the market. At this time, the seller’s identifier is masked. For instance, a seller could disclose that he is male and recently bought an iPad, but nobody can know the seller’s identifier.
- Data buyers query the data that they are interested in buying. If their search shows the sellers that have the data that they are looking for, the buyers send a sale request.
- If the seller agrees to the request, the seller’s data or identifier is sent to the buyer. Based on the seller’s data, the buyer is able to use it for target marketing through the seller’s identifier.
Although this procedure may appear simple and smooth, it is missing a very important step — the process of request sale and transaction success between the buyer and seller. A transaction is made when one party makes an offer and the other accepts. In other words, the procedure elaborated above does not address how data is priced to a level of which both parties agree.
What is the Appropriate Price for My Data?
Finding the appropriate price for my data is a very tricky problem because the buyer and seller are placed in different circumstances.
As-Is of the Seller
The seller does not know what the appropriate price is for their data.
Typically, there are two ways that the price of goods is determined in the market: one by pricing the goods based on production cost and one based on the price buyers are willing to pay.
However, it is difficult for the sellers to use any of these two methods. No cost was incurred in creating their data and the sellers do not know how much buyers are willing to pay for certain types of personal data. If buyer A requests a purchase at 1,000 KRW, the seller has no idea whether that is the maximum amount buyer A is willing to pay, and has no accessible way to see if there are other buyers who are willing to pay more.
In this situation, it is likely that the seller will price their data based on the opportunity cost they feel for sharing their data. The more reluctant the seller is to share a certain type of data, the more money the seller is likely to ask for it. According to a study by Columbia Business School, people are more hesitant in sharing their purchase record, browsing history, social media activity, and other data that hints at a person’s preference and lifestyle. Such data is what is mainly sold in the my data market. Therefore, my data sellers will see that the opportunity cost of sharing data is high. If that price is higher than what the buyer believes is reasonable, then it is likely that the transaction will not happen.
The graph shows sellers’ Willingness to Share per type of data. You can see that people have a tendency to be reluctant in sharing data that directly show personal preference. (Source: “What is the Future of Data Sharing?”, Columbia Business School, Oct 2015)
As-Is of Data Buyers
Data buyers can predict how much the sellers’ data is worth because the expected profit from utilizing the data for marketing or re-selling equals the value of data. However, as we just addressed, the seller is unaware whether the buyer’s offer is the right price and will use their opportunity cost as the price. If the price is higher than what the buyer expected, then the buyer will not be able to purchase at the proper price.
To summarize, sellers price their data based on their subjective opportunity cost because they do not know how much their data is worth. This causes a situation similar to monopoly in the my data market. A monopoly is when there is only one supplier of a commodity. A monopoly occurs when a supplier with dominant comparative advantage blocks out the market entry of other competitors or because a certain commodity is unique. Monopoly in the my data market falls in the latter case. Why is that?
Data is not identical. My shopping preference and behavior pattern is unique to me alone. When I try to sell my data, there is very unlikely that another person pops up and sells the same data at a cheaper price. Because data is unique, monopoly by the seller could happen.
When there is monopoly of unique goods, it is difficult for the player who will use it in the most efficient way (create the most economic value) to obtain the goods. Such phenomenon is described as a degradation of allocative efficiency.
Then what pricing method is optimum in securing allocative efficiency in the data market? We were able to find some leads in the land market.
New Perspective on Appropriate Price: Land Market
and Harberger Tax
Let’s talk about land for a second.
Unlike industrial products and agricultural products, of which identical products can be reproduced, land is a unique asset that is fixed in terms of location and size. There are many other sites that have the same size as Yeouido, but there is only one Yeouido in the world. Land being unique means that there is only one supplier in the market that can supply a certain piece of land. When someone seeks to buy a plot of land, the owner of that land exercises monopoly. A representative example of such monopoly is land speculation.
Economics deems it ideal when the person who can most efficient use has possession of resources (which is the definition of allocative efficiency). However, land monopoly degrades allocative efficiency in society. Even if a person who make most use of a land appears, more value will not be created if the owner does not sell.
There is an intriguing idea set forth as a solution to achieve land allocative efficiency proposed by University of Chicago economist Arnold Harberger in 1962.
Later coined as the ‘Harberger Tax’, the two key points are as follows:
- The land owner prices his land based on his assessment. If there is a buyer, then ownership must be transferred.
- The land owner must periodically pay tax at a certain ratio of the self-decided price.
How can this idea contribute in achieving land allocative efficiency?
Let’s think about what the rational behavior would be for the land owner when the Harberger Tax is implemented since the effect of a policy is shown by the accumulation of rational behavior under that policy.
With the Harberger Tax, anybody who can match the determined land price can obtain land ownership regardless of the original owner’s opinion. If the owner wants to hold on to the land, he would set the price high.
But the owner has to pay tax in relation to the price he set. If the price is too high, the owner ends up paying more in tax. This would make the owner decide on a price that is not too much or too cheap.
Land value is a reflection of the future value that can be earned from the land. In other words, owners set their land price based on expected profit. Let’s consider the following example.
K built a shoe factory on his land. By calculating the profit and cost of the factory, K determines the land price as 1 billion KRW.
Meanwhile P believes it would be more profitable to build a semi-conductor factory on K’s land and thinks the land is worth 2 billion KRW. Eventually, P pays K 1 billion KRW and assumes ownership.
In the conventional land market, K would be able to drive up the price to 2 billion KRW and inflict considerable monetary impact on P. However, under the Harberger Tax, nobody has to suffer loss. P is able to purchase the land he wants without being ripped off and K is able to cash in the amount he believed he could profit from the land. Additionally, since the land is now owned by the person who can make the most out of the land, it is the optimum result for society as well.
In summary, Harberger Tax ensures that all land ends up in the property of the person who can most efficiently use land. Each land price is determined by the owner and that price is publicly disclosed. Reducing information divide and achieving allocative efficiency is the meat of the Harberger Tax.
Returning to the my data market, data and land share uniqueness and degradation of allocative efficiency due to monopoly. Wouldn’t this mean that we can apply Harberger Tax in the data market?
Applying Harberger Tax in My Data Market
Indeed, it is impossible to implement the Harberger Tax in the my data market in the same manner as the land market because data can be duplicated. Transferring ownership does not mean the original data owner loses the data.
However, data identifier is transferable because an identifier can always be re-set. If buyer B wants to purchase the identifier of seller C that buyer A is aware of, buyer B could pay buyer A and receive a reconfigured identifier from seller C. Then buyer A would no longer have knowledge of seller C’s identifier, and buyer B would be able to utilize seller C’s identifier.
Now that everything is on the table, let’s devise options on how to apply Harberger Tax.
- The data seller sets the number of slots (N) for his identifier. Slots refer to the number of buyers that can buy and utilize the seller’s identifier.
- If data buyers find the seller via query, they can buy a slot. For the transaction, the buyer would set a price for the slot based on his assessment of value and pay the seller a ratio of that price periodically. Buyers will earn profit by utilizing the seller’s data for the duration they have ownership.
- If a new buyer wants to buy an already sold slot, the new buyer can pay the original buyer can assume ownership. Through this transaction, the seller’s identifier is re-set and sent to the new slot owner. The new buyer also sets the price for the seller’s data and pays the seller periodically a proportion of the price.
The figure below illustrates this process.
Figure 1. The data seller opens three slots for his data identifier. Three data buyers each set the price for the slots as $30, $20, and $10, and maintains ownership. In this market, the Harberger Tax is 10%. The buyers each make a weekly pay of $3, $2, and $1 to the seller.
Figure 2. A new buyer (in red) sets the price for slot 3 as $15 and receives the slot from the original buyer who priced the slot at $10. With this change of ownership, the identifier is re-set.
Let’s consider a scenario of this method being actually used. I am a male in my 20s living in Gangnam. I enjoy eating u-dong and prefer to merge with my bed on holidays. My payment data is automatically uploaded to the marketplace so anybody who queries can request data purchase.
Mr. U, who wants to open a u-dong shop in Gangnam, queries ‘people near Gangnam who like u-dong’ and finds me. I only opened up one slot. Mr. U buys this currently empty slot and receives my identifier. Afterwards, Mr. U sends me coupons and ads on a daily basis through push alert. Mr. U priced the slot as 10,000 KRW. Because the marketplace rate is 5%, i received 500 KRW every week in return for selling my data.
But then comes Mr. Y who wants to start a u-dong delivery shop in Gangnam. Mr. Y finds me through querying ‘people near Gangnam who like u-dong and stay in during holidays’. Mr. Y sees that i am a great potential customer, pays Mr. U 10,000 KRW and receives a new identifier. Mr. Y sets the price for the slot at 20,000 KRW. Since then, i receive ads and points through push alert from Mr. Y’s delivery shop and i receive 5% of 20,000 KRW, which is 1,000 KRW. I can also transparently see that Mr. Y, not Mr. U, is utilizing my data.
Benefits of Harberger Tax
The proposed method above deals with transfer of ownership of a fixed asset, data, rather than person-to-person transaction. The transfer of ownership mimics the Harberger Tax principles. By doing so, the seller and buyer have certain utilities.
Utility for the Seller
Because the price set by the buyer is the appropriate price for the data, the seller has oversight of how much his data is worth. With mitigation of information divide, the seller can share his data at the right price instead of feeling an opportunity cost for sharing data. By adjusting the number of slots, the seller can balance control of one’s data and expected profit.
Utility for the Buyer
The buyer does not have to bear unnecessary loss from the seller setting the price higher than the proper price level. With the Harberger Tax system, asset owners always possess their asset at the appropriate price, allowing the buyers to use the seller’s data at the appropriate price. Even if a new buyer buys the slot, the original buyer is still compensated for the expected amount of profit for maintaining ownership, thus not bearing loss.
In the end, as data is distributed to buyers who can earn an anticipated amount of profit, and therefore, the market’s allocative efficiency improves.
We admit that the proposed method above is not the best pricing method for the my data market. There are many excellent transaction methods where buyers and sellers can directly make deals, and many data market projects are coming up with their own methods.
The important thing is that the value of data increases as it is processed and integrated, and a robust data circulation market will determine not just a company’s but also a country’s competitiveness in the near future. The Korean government is already striving to invigorate the data circulation ecosystem starting with public data. Now is the time to contemplate about allocative efficiency in the data market along with how to appropriately reward data creators.