Tuesday, 4 October 2011

Is there a 'hidden cost of control' in naturally-occurring markets?

There is a new, interesting working paper from the NBER on incentives and their effects. The paper, Is there a 'Hidden Cost of Control' in Naturally-Occurring Markets? Evidence from a Natural Field Experiment (NBER Working Paper No. 17472, September 2011), is by Craig E. Landry, Andreas Lange, John A. List, Michael K. Price and Nicholas G. Rupp.

The authors note that,
Behavioral economics has matured to the point where theorists are leveraging psychological insights to improve their models and government officials are using behavioral results to fine tune policy. One particular result that has attracted increasing attention is the interaction of psychological and economic incentives [...]. For example, in a novel set of experiments, Gneezy and Rustichini [...] show that extrinsic incentives influence effort in an unexpected manner—small monetary incentives can crowd out intrinsic motivation, resulting in a perverse relationship between incentives and effort.

Complementing such insights is the line of work in the spirit of [papers by Fehr and Rockenbach, and Fehr and List], who find that the use of control and explicit incentives entail “hidden” costs: such control causes the principal’s actions to backfire, leading to lower profits. As this literature points out, such effects are first order and should be a concern for economists interested in studying labor markets. Yet, whether, and to what extent, such hidden costs manifest themselves outside the confines of the laboratory remains an important open empirical question [...]. Difficulties arise, however, in finding natural instances where agents are randomly allocated to appropriate treatment groups to permit a clean test of the relevant hypotheses. Because of these challenges, the literature has to date been unable to provide tests of the major hypotheses of ‘control’ in the field.
The point of the current paper is to make a first step in this direction.
We present an empirical approach that is composed of a set of field treatments that parallel the important economic features of the environments in Fehr and Rockenbach [...], Fehr and List [...], and the literature that has followed. To do so, we examine solicitor (worker) effort in a door-to-door capital campaign for the Center for Natural Hazards Research at East Carolina University. Importantly, we use natural incentives to exogenously change the action space of solicitors randomly assigned to one of three treatment groups. In the baseline treatment, workers are provided a pre-announced, fixed hourly wage of $10. In a second treatment, workers are provided an unconditional gift – a copy of Freakonomics – in addition to the pre-announced hourly wage. In our final treatment, the most opportunistic actions are ruled out by making the gift conditional – solicitors must raise at least $10 per hour to obtain the copy of Freakonomics. If trust is a characteristic rewarded by workers, then the control evoked in the final treatment might crowd out effort compared to the second treatment – particularly amongst those who would have raised more than $10 if the gift were provided unconditionally.
The authors argue that there are several important insights which emerge from their field experiment.
First, unconditional gifts have the ability to enhance worker productivity. Solicitors in our unconditional gift treatment were approximately 56 percent more likely to elicit a contribution than counterparts in the baseline. Similarly, solicitors in this treatment raised approximately 10.1 to 63.4 percent more per hour than those in our baseline treatment. These results are consonant with the existing literature [...] and suggest that reciprocal motives are an important determinant of worker behavior.

Second, conditionality proves a profit enhancing strategy. Participation rates and dollars raised per hour in our conditional gift treatment are higher than those observed in both the baseline and unconditional gift treatments. For example, solicitors in our conditional gift treatment elicit contributions from nearly 26 percent of all households approached – a rate of giving that is approximately 68 percent (8 percent) greater than that observed in our baseline (unconditional gift) treatment. Similarly, solicitors in our conditional gift treatment raise approximately 83 percent more per hour than counterparts receiving an unconditional gift and more than double that observed in our baseline group.

Finally, solicitors in the conditional gift treatment are approximately two and half to three times more likely to raise at least $10 per hour – the required threshold for receiving the gift – than are counterparts in our unconditional gift and baseline treatments respectively. As this threshold corresponds with the level of productivity necessary to cover labor costs, conditionality therefore has a positive effect on net revenues. Taken jointly, these data are at odds with models suggesting that agents respond adversely to control. Accordingly, our data suggest that hidden cost relationships identified in prior laboratory studies [...] do not arise in our field setting.
So, yes incentives do matter, just as economists normally argue.

No comments: