Voluntary Disclosure Not Covered in Textbook
You’re on a job interview and the interviewer knows what the distribution of GPAs are for MBA students at MSU: Expected/Average grade for everyone: .2*2.5+.3*3.0+.3*3.5+.2*4.0 = 3.25 Geoff Humphrys at the Lear Center advises anyone who has a 3.5 GPA or higher to volunteer their GPA. Is this a stable outcome?
Students remaining Original share • What does the potential employer believe about the people who stay quiet? • They know their GPA is below a 3.5, but how far below? • Guess the average grade of everyone who didn’t get at least a 3.5. • What is that? .3/.5 .2/.5 =.4 =.6 .4*2.5 + .6*3.0 = 2.8 People with 3.0s will reveal themselves because they don’t want employer to assume they have a 2.8
Voluntary disclosure • Full disclosure principle - if some individuals stand to benefits by revealing a favorable trait, others will be forced to disclose their less favorable values. • If disclosure is costless, only the lowest types will not reveal their quality
Voluntary Disclosure and Signaling • Voluntary Disclosure differs from Signaling because we are assuming that the cost of lying (i.e., saying you have a GPA of 4.0 when you have a GPA of 3.5) is so large than no one does it. Therefore, the decision is to either reveal your private information truthfully or don’t reveal.
Voluntary Disclosure • If it is true that only the lowest types don’t reveal and that consumers/employers (the uninformed party) can infer they are the lowest type, then government should not have to intervene in the market – for example, they should not require firms producing salad dressings to report the fat content and they should not require restaurants to report their hygiene score.
Voluntary Disclosure and the Fat content in Salad Dressing The Impact of Mandatory Disclosure Laws On Product Choice Alan Mathios http://www.jstor.org/view/00222186/ap020088/02a00130/0
Voluntary Disclosure and Hygiene Scores for Restaurants in Los Angeles The Effect of Information on Product Quality By Phil Leslie and Ginger Jin http://www.mitpressjournals.org/doi/pdfplus/10.1162/003355303321675428?cookieSet=1
Voluntary disclosure and SAT scores • Institutional Details • Voluntary disclosure question • Data • Results
Institutional Details • Increasing # of schools are adopting policies where submitting your SAT scores are optional • i.e., students can submit high school G.P.A., extracurricular activities etc, and intentionally exclude standardized test score on their application • School will judge students based on submitted material
Voluntary disclosure question • Theory: If it is fairly costless to reveal your scores, only the students with the very lowest scores should not reveal. Others want to avoid being considered the “average” of those who don’t reveal. • Is it only the students with very low SAT scores that don’t reveal?
Data • About the colleges • 2 Liberal arts college in the NE • 1800 students enrolled • Mean SAT score > 1200 (out of 1600) • 1020 is the mean SAT score of those who take it
Data - specifics • From recent years and when the policy was first implemented • 2 years for College X • 1 year for College Y (1st year of policy)
Data - specifics • Data contain • Whether student wanted SATs considered • SAT score if submitted • College Board provided SATs for those who didn’t submit • Race, gender, zip code, legacy status, early admissions, HS GPA, financial aid intent, type of high school • Survey results from SAT (self reported income, GPA etc) • Zip code information (race, income, etc.) from Census • College’s rating (total scale of 7 (X) 9 (Y))
Table 2 Voluntary Disclosure by SAT I Score and Demographics Share who chose not to reveal their SAT1 Score Does the voluntary disclosure principle hold? What else is related to NOT submitting your SAT score?
Predict SAT score based on other observables (like GPA, activities, etc). Compare to actual… Line → predicted = actual Right of line → predicted>actual • If predicted (based on other characteristics like GPA, activities, gender etc) is greater than actual (where actual is maximum if take SAT more than once)… More likely NOT to report! b. If predicted is < actual… More likely to report!
Regression Specification • OLS ordinary least squares Y = α0+ α1(SAT1 score/100)+ α2(HS GPA)+ α3(Female)+ α4 (attended private HS) … where, Y=1 if student does not submit and =0 if does submit -0.1284 A 100 point increase in SAT lowers probability of NOT submitting SAT1 by 0.12 percentage points (more likely to submit), ALL ELSE EQUAL
Linear Probability: Dependent Variable =1 if Applicant Chooses Not to Submit SAT I and=0 if Choose to Submit Economic significance: 100 point increase in SAT Lowers probability of NOT submitting SAT by 0.12 Percentage points (more likely to submit), ALL ELSE EQUAL Independent Variables Coefficient Standard Error * (stars) indicate statistical significance!
Dependent Variable (Chose Not to have SAT I Considered) All else equal (same SAT score, for example) More likely to NOT have SAT I considered if higher HS GPA Relative to a HS GPA of B! All else equal more likely to NOT have SAT I considered if a female student
Dependent Variable (Chose Not to have SAT I Considered) All else equal More likely to have SAT I considered if minority All else equal More likely to NOT have SAT I considered if attended private HS
Conclusion • Are the costs of revealing low? • Does the full disclosure seem to apply? • Why? • Who does withhold their SATs? • Students with other measures (HS GPA, SATIIs of high abilities • White students • Women • Private High School attendees
Why would Colleges go to Optional SAT Policy? • Attract a different type of student (those that don’t test well but do well in college) • Maybe more diverse? • Improve ratings • Average SAT score included in U.S. News and World Report • If don’t have SAT scores for lowest score students, reported average increases.