Does privacy make us productive?

Here is the job market paper and abstract from Ethan Bernstein, who is on the job market from Harvard Business School:

Does Privacy Make Us Productive?

We have grown accustomed to calls for transparency. Transparency, or accurate observability, of an organization’s low-level activities, routines, behaviors, output, and performance provides the foundation for both organizational learning and operational control, and it has an untarnished reputation: rarely does one hear about any negative effects of transparency or problems stemming from too much transparency. Nonetheless, using data from embedded participant-observers and a field experiment at the second largest mobile phone factory in the world, located in China, I introduce the notion of a transparency paradox, whereby maintaining observability of workers may counterintuitively reduce their performance by inducing those being observed to conceal their activities through codes and other costly means; conversely, creating zones of privacy may, under certain conditions, increase performance. This research suggests that careful design and implementation of zones of visual privacy within an organization is an important performance lever but remains generally unrecognized and underutilized. Paradoxically, an organization that fails to design effective zones of privacy may inadvertently undermine its capacity for transparency.


But are the workers using codes to disguise the fact that they aren't doing things properly? Increasing productivity while decreasing work place safety etc.

This is ambiguous. Just like 'security' there real safety and then there is safety theater, the signaling waste required by safety bureaucrats. I've never had any safety incident in any of my jobs, prevented real safety problems and the safety theater drives me crazy, particularly when someone with stellar safety theater credentials repeatedly causes real safety problems. I also think that it is structurally corrupt. I used to wonder why there was a safety manager and a production manager. You could get fired by either. Then I started to think they wanted it exactly that way.

I'm reminded of David Brin's calls for transparency. It's not just that the lower levels of an organization receive better clarity, it's that this clarity must also extend into the upper levels, to be viewed by the lower levels. The idea is that no one should be able to hide.

In addition, I feel that there is something further that needs to be preached alongside transparency: tolerance. If someone is going to feel a desire to spend extra effort into hiding behaviors deemed undesirable by the observers, then efficiency can be increased by reducing the number of behaviors the observers consider undesirable. So, if the observers want to prevent sabotage, that's fine, but they shouldn't also not like, say, different (but still effective) working styles. If you, say, want to express your love of Dungeons & Dragons instead of football (assuming you don't work for a game designer or sports outfit), there shouldn't be a problem. It's important for the observers to only frown upon inefficient behaviors and not behaviors independent of work.

I wonder how this analysis translates to work-related information. Does an open exchange of ideas always produce better (more efficient) outcomes or is a certain amount of privacy better? More is better for transparency and clarity seems to be a current trend in say monetary policy, but what's the optimal stopping point? It seems like the potential audience and their attitude toward the organization, including tolerance, could matter to.

His actual paper looks more like something fluffy out of the McKinsey Quarterly or a case-study and less like a rigorous academic paper.

Yes, I've worked with Ethan before and he is a practical man who's held real jobs before.

Figure-2 from his paper was very interesting, rather self-recommending as Tyler might perhaps have said:

Table 3 is about as rigorous as it gets, I don't see the fluff. This is a straightforward DID. Although why you need this if you have a field experiment beats me.

You had a commenter on an earlier post saying "Privacy is a non-issue for most Indians", though in fairness I think there are two different types of privacy being discussed.

Please remember that in politics, law, and finance that the term Lack of Transparence or there needs to be more transparency is referring to hiding illegal activity!

In a production line the need to follow protocol is a top down management style that may produce inefficiency at the lower productive areas but it is efficient in keeping the jobs of managers’ in-between this production and the top.

When I was “older” it was discussed as “this is my rice bowl.”

The real job of the middle manger is to protect the lower managed personnel from the upper management and to protect upper management from a reality they can not accept. That is to say if the bottom of the organization knows best how to do the job why are we paid so much and why are we here.

It is not lack of transparency that you should be thinking about, it should be the lack of recognition and financial reward for the lower level workers who make the company successful in spite of the management.

I am procrastinating studying by sharing this with my girlfriend, who is sitting next to me. Thesis confirmed (n=1).

Thank you all for your comments (and thank you to Tyler for posting). I am always thrilled to discuss this paper with anyone and very much appreciate the thoughtful feedback.

Without going too deep into the research, allow me to be briefly responsive to several of the above concerns/questions.

First, as Bill points out, this research adopts the general meaning of the word "privacy," not the more narrow meaning (e.g., information privacy) it has come to represent in recent years. In the paper, privacy is defined as “the ability to control and limit physical, interactional, psychological, and informational access to the self or to one’s group," a distillation from the deep literatures on privacy in law and philosophy. Using that definition, privacy serves as an interesting antipode to transparency, or "accurate observability," in the management literature, with roots that are equally deep: as a human need, privacy originates in Babylonian and Sumerian historical texts; as an English word, it is first popularized by Shakespeare in 1598 in the Merry Wives of Windsor; and as a concept, its roots are squarely in the Victorian era where law officially punished acts of immorality but made it difficult to get the facts necessary to be successful in bringing such legal actions by providing protections against the revelation of private facts in the public sphere. (See the paper for citations on all of these points.) The basic point: a broader view of privacy is worthy of a look, particularly in the management literature where it really hasn't been given consideration (especially relative to the significant weight given to transparency).

Given that background, second, the goal of Figure 2 (referenced above) is to explain that privacy is functionally achieved in exactly two ways: either by closing the door (providing a boundary to visibility), or by encryption (hiding behavior that prevents the observation of activity from creating a full understanding of what is going on--essentially an illusion of transparency). When we talk about implementing transparency, we usually mean moving from the upper right hand corner of Figure 2 to the lower left. Indeed, many assume that simply making activities observable will make them transparent. But that ignores the behavioral response of individuals in an observable condition--which, at least in this field context (and others I have studied), is to hide through encryption. We all do things a little bit differently when no one is watching (and in many cases, more productively). So instead of choosing between the upper right and the lower left, management in organizations is actually choosing between the upper left and the lower right: in organizations where boundaries to visibility are eliminated, members use encryption for privacy (as Simmel said in the 1960s, "where privacy is prohibited, man can only imagine separateness as an act of stealth"); in organizations where privacy is legitimized through visibility boundaries, however, the need for the code of encryption dissipates.

Third, to demonstrate the performance implications of one form of privacy versus the other (which is the main purpose of the field experiment), I run what I believe is a tight experiment, although I am always open to suggestions for more rigorous methods. (whatsthat, a non-DID statistical model can also be run, but the DID helps to nicely capture the learning-focused context of this particular setting.) As shown in Table 3, I find significance and rather substantial effect size (i.e., 10% performance improvement for the lines with simple curtains around them).

There is much more in the paper, and I am happy to discuss it here or anywhere.

A few final thoughts:
Keeb and Andrew, in this case, safety and quality either improved or stayed the same with privacy. I was very careful to look for that. This was not about cutting corners that should not be cut.

IVV, I agree that tolerance is a related and important construct. I would caveat that, however, with my own observations that many managers in this setting were actually quite tolerant. Nonetheless, workers still acted this way. Think of Adler & Borys’ distinction between enabling and coercive bureaucracies. It may be the case that, no matter how tolerant (enabling) a bureaucracy is, too much transparency automatically pushes the context into one that seems coercive to those who ultimately choose whether to hide or not.

Davidt, on the question of financial rewards—interestingly, after this experiment was over, the organization tried to implement recognition and financial rewards to replicate the performance improvement of the curtains (in place of privacy). So far as I understand, the experiment failed to produce similar results. At least in this context, it may be that privacy was a stronger implicit motivation than could ever be overcome by any explicit motivational tool (like financial incentives). An interesting question to ask....

I hope this adds to the discussion, and I’d love to hear more. Thank you everyone for your thoughts.

There is known data that workers who perceive themselves to be trusted by managers are more productive; was that accounted for in the calculation?

If the result of reducing privacy is to reduce trust, then the loss of productivity is understandable and not counter-intuitive.

The challenge of management is how to monitor your workers' work without destroying the trust relationship that is needed for them to be productive.

I had a further though on the subject after I posted and another after reading your 2nd post.

First the inefficiency is key to pricing the job. Like repair job for a car the time to achieve a repair is inflated to arrive at the cost for the customer. The issue in this manufacturing job is the customer knows things about the efficiencies to be had. The instruction set for the worker begins the pricing and is actually given to the workers. Both customer and Manufacture knows there are efficacies to be had. The final price is a discount total cost of the activities in the instructions to the worker. The manufacture does not want the customer to know the efficiency as to keep its margin in negotiation. To confirm this or to refine this view it would be interesting to understand the negotiation activity between the parties or has it been set from much earlier work and price assumed and rewards and penalties inserted for on time delivery or late fees because of the huge cost of missing the market potential of being late or having too little volume to sell.

The second observation is about the one time monetary rewards of increased efficiency. Also to Richards comment about trust is in the same vein. At some jobs new workers are “asked” to slow down as they are showing a higher productivity level that the existing workers. It can not be know if the existing workers know that the pace can not be kept up over time, they do not want to work that hard or whether another issue is causing this to occur.

In this example of a “team” activity that can be sunk by a bottle neck at any point there is pressure on the individual to not take the safety time out of the production line for individual benefit of a one time income. This goes to Richards comment about trust. Do the workers trust management to not do this? Does management trust workers to include them in negotiating the delivery time? There are a lot of questions but I think these questions point to some of the interrelationships. All parties have a lack of transparency to one another and at present actually count on it to control their position in the process.

I think decreasing transparency would harm one or two of the parties to the benefit of one of the three in the beginning. Over time all three would fail as more is asked of the other two that they can give and the relationships fail and the customer leaves first because of price, lack of a timely delivery, or some other instability that could arises because of the huge cost of missing the market potential of being late or haveing too little volume to sell.

This data is basically good and Ill say will at all times be helpful if we strive it danger free. So when you can back it up. That will actually assist us all.

Comments for this post are closed