This section of the web site is devoted to my comments (drawn from presentations or correspondence) or speculations on what are interesting areas for future research.  It is grouped by exactly the same categories as my research page, as each of the topical areas in the research page that give my own research ends with "other sites" pointing to other research on the same topic, and a comments and puzzles section.

Consider these are free ideas, keeping in mind this is a page for economists who believe prices clear markets.

"Productivity shock"  It is well known in the USA that cost per student has risen enormously while measured learning achievement in the NAEP has not risen substantially.  Woessman, with various co-authors, has extended this result by linking NAEP scores (consistent over time) to various cross national tests (e.g. SIMSS) (consistent at a point in time) to create estimates of the evolution over time in many countries.  He also creates series on "real" expenditures per pupil (which corrects for "Baumol" effects by deflating with prices of non-tradable services).  He and his co-authors find that the decline in "test score answered correctly per dollar of expenditure" declined dramatically in nearly every OECD country. This is a big puzzle.  This "think piece" explores various alternative explanations for this generalized phenomena:     Educational Quality and Costs: A Big Puzzle and Five Possible Pieces. 

Interaction of education and growth.  The upshot of both "Where has all the education gone?" and "Does Learning to Add up Add up" is that the growth impact of education almost certainly varies across countries.  The puzzle is explaining why and how in ways that are consistent with both the micro-economic and aggregate data.  That is, education could have a low output impact because the micro-economic return is low--in which case this should be evident in the observed micro evidence on Mincer returns.  But it could also be the case that the gap between the micro-Mincer and aggregate returns differs across countries because in some countries there are positive externalities to education while in other countries there are negative externalities.  While positive externalities of schooling get a lot of attention, there are also cases with negative externalities.  In any model where "rent seeking" activities are a privately renumerative activity and rent seeking is a wealth reducing activity and is skill intensive (e.g. lawyers, a la Vishny et. al.) then in "distorted" environments there would be persistent high observed returns to education with no aggregate (or even negative) returns to increases in aggregate output.  The puzzle is:  is this true?  what is the appropriate interacting variable?  Are there ancillary hypothesis of this story that would add credence.

Late Enrollment.  The four figures below are from Deon Filmer's wonderful website showing enrollment and attainment from more than 70 countries around the world based on household data sets (which allows distinctions by household wealth, residence, and gender of child).  This figures constitute a puzzle for me.

 Why are these figures a puzzle?  Because the usual education story is that the marginal cost of schooling rises over time (as productivity and hence opportunity cost of time increases with age) while the marginal benefit of a year of schooling is roughly constant.  This would generate a single crossing:  enroll as soon as allowed (since MB>MC) and then quit school when MC>MB.  The Indonesia figure shows roughly this story--nearly everyone who will go to school is in by age 7.  But the others show other stories--in Ethiopia the enrollment ) particularly for the poor rises right through to age 14--there are more children coming into school at later ages than leaving it.  This is (a little less certainly) true in both urban and rural areas and true for boys and girls.  This must mean there is a "double crossing" of the MB and MC curves with respect to age.  One story (a la Glewwe and Jacoby) is that the MB rises sharply with age, especially among small or nutritionally deficit children, as they are not "school ready" so that the MB is low at young ages, but rises.  But it is not clear this can explain the variations across country and all the pervasiveness of the phenomena (e.g. even among the richest 20 percent in poor countries).  One story is that parents actually have a target years of attainment and then maximize the learning from those years (rather than being uncertain about termination and learning about MB and MC) so for instance, if a parent knows their child is only going for five years of schooling (primary) perhaps they would prefer ages 10 to 15 than 6 to 11.  The more concrete question are whether these decisions have negative consequences (e.g. ex post regret if a child turns out to have a high MB of schooling (high aptitude) but enrolled late so MC rises very fast) and if there any policy instruments for extending attainment by limiting late enrollment (rather than the usual focus on drop-out or lack of progression). 

Population