week 2 slides 2

Dynamic Labor Supply Model

• The dynamic labor supply model involves the intertemporal substitution of labor supply.
• Workers decide how to respond when earning extra and make working decisions daily, spending all income on consumption.
• Workers work more today and less tomorrow with the dominance of substitution effects.
• Workers will work less today and more tomorrow with the dominance of income effects.
• Workers could improve if they foresee a fall in tomorrow's earnings.
• Workers unable to anticipate tomorrow's wage change don't respond or change today's labor supply.
• Intertemporal substitution predicts workers buy leisure time when it’s cheap and work more today and less tomorrow when wage shocks are foreseen.
• If a permanent wage variation is foreseen, income effects re-enter the labor supply model, and behavioral responses mirror those in a static world, with workers working less if income effects are greater than substitution effects.
• Intertemporal substitution happens when wage changes are transitory and anticipated.

Evidence of Intertemporal Substitution

• Testing requires workers able to anticipate changes in hourly earnings and adjust their labor supply.
• Mixed evidence exists regarding the labor supply of taxi drivers that substitute intertemporarily.
• Taxi drivers should work longer when streets are full of passengers or take up the leisure on days when streets are empty.
• Farber's 2005 study finds that the taxi drivers work longer shifts when marginal wage rates rates are high.
• Camerer et al.’s 1999 study reveals taxi drivers to work longer hours when daily earnings were low, with earnings across working days being unrelated.
• The Camerer study introduces a behavioral model of targeting, where taxi drivers make daily labor supply decisions, set a daily income target, and quit working once reached.

RCT with Bicycle Messengers

• Two Zurich economists conducted a field experiment involving bicycle messengers for two months.
• The bicycle messengers were divided into two groups.
• One group received a premium for each message delivered in the first month, while the other group received the same premium in the second month.
• FG argue that finding proof in favor of intertemporal substitution is hard.
• Some of the reasons difficulty in finding proof are exogenous wage variation known in advance to be transitory, institutions that prevent workers' flexible adjustment of working hours, and standard labor supply models may be wrong (workers are guided by non-standard preferences).
• FG setup an experiment that made is possible for bicycle messengers in Zurich to flexibly choose their work hours and effort levels.
• The bicycle messengers each receive a commission rate per delivery and are assigned randomly to group A or B
• Messengers in group A receive a 25% higher commission rate in period 1 and messengers in group B receive a 25% higher commission rate in period 2.
• With standard labor supply models, there is predicted rise of total revenues and number of deliveries (total revenues) with higher commission rates.
• Behavioral models with reference-dependent preferences predicted that setting a daily income target would yield treated messengers to work more shifts with fewer deliveries or slower biking.

RCT with Bicycle Messengers: Loss in effort levels

• Average effort levels fall because messengers get tired with more shifts or people use reference dependent preferences with the same income target realized with less effort.
• It has been argued that loss averse bicycle messengers are the ones who exert effort up to the target.
• Loss aversion is measured via two lottery questions.
• Lottery A is where participants win CHF8 or lose CHF5 with 1/2 probability or reject it for CHF0.
• Lottery B is playing lottery A six times in a row, or reject it for CHF0
• Loss averse messengers reject A, B, or both.

Intertemporal Substitution and Added Worker Effects

• Macro-oriented examples of intertemporal substitution:
  • Labor supply in a life cycle model: With a high intertemporal substitution of labor supply, people will work less when they are young/old.
  • Labor supply in a business cycle model: Unemployment rates rise during recessions and decline during booms.
• Discouraged worker effects may dampen unemployment fluctuations if workers can’t find jobs in recessions and leave the labor force. Added worker effects may amplify unemployment fluctuations if nonworking partners try to compensate for earnings losses in recessions.
• These observations align, but not prove the intertemporal substitution.
• If husbands lose their job, do wives start to work to compensate for the loss in income?
• There are four reasons to expect at most a little labor supply response: generous programs, little opportunity for wives to work, wives aren't expected to respond or the selection.
• Wives in Austria whose husbands have involuntary job losses from plant closures and mass layoffs were examined.
• Findings show a small labor supply repsonse for these women.
• Wives' labor supply response varies by child age, their employment status, or local labor market conditions.
• The research indicates that few wives compensate for the loss in their displaced husbands' earnings.
• If women earn less than men is important to consider since mothers would work more.

Career Costs of Children

• What are the effects on the labor supply for women/men?
• Gender wage gap is important to consider because childcare programs are probably beneficial if children hinder women.

What Natural Experiments Tell Us

• The empirical strategy must identify the causal impact of children on parental labor supply by analyzing the ideal experiment.
• The natural event is an “Instrument” which provides a random assignment of children to some women, and thus affects the likelihood (relevance) of more children while affection career's women (exclusion).
• The instrumental variable is an identification strategy.
• Angrist and Evans consider the gender combination of the first two children as the instrument.
• Couples with same-sex first two children are more likely to have a third.
• In order to determine if the IV context is effective, 3 conditions must be met.
• Relevance: gender of the first two children must affect the likelihood of having a third child.
• Independence*: gender of the first two children must be as random as possible (can be checked).
• Exclusion: the labor must affect the maternal labor In 1998 economic review, the children and their parents labor supply exogenous variation in family size must affect the number of children.

Natural Experiments: Fertility and Labor

• Angrist and Evans (1998) focus on fertility measured at the intensive margin and examine how the arrival of a third child changes female labor supply.
• Lundborg, Plug and Rasmussen (2017) focus on fertility measured at the extensive margin.
• Another experiment studied childless women going through IVF (in vitro fertilization).
• If successful IVF treatment is random at first, motherhood effects can be identified.

Natural Experiments: Requirements to Fulfill

• Our IV-IVF strategy only works if IVF treatment success at the first IVF treatment satisfies the following conditions: relevance (affect the likelihood of having children), independence (treatment be random), and exclusion (labor supply exclusively affected through children).

Natural Experiments: The Modeling

• Variables are
  • Y represents labor supply,
  • F whether children are present (1 if true, else 0).
  • Z is a successful first IVF treatment.
  • X exogenous controls.
• A regression analysis is conducted to understand and calculate first stage equation, and second stage equation.
• The instrument that is being used must be satisfactory, that is, the instrument independence must be random and the instrument effects the relevant variables, as well as exclusively affecting F.

Instrumental Variable Regression

• There are two parts to the regression
  • Each part relates the regression used to calculate the Covariance to arrive at a reasonable conclusion. By taking the ratio we can arrive at the true treatment.
• The IV is run on two regressions, therefore, the estimation process is referred to as 2 stage least squares.
• There are 3 panels that relate fertility in years.
• All show a relationship between the variables.

Threats To Validity

• A depression or divorce interfere as other reasons to affect women's career path.

The Exclusion Restrictions

• The exclusion condition is generally the most substantive condition in IV Designs
• The effect if Z over and above. Therefore not knowing the data will lead to the negative affect.
• Within our IVF context, however, we only have compliers and always takers in binary fashion.
• Instruments are used to have treatment and potential outcomes. The fertility leads to the random assignment to treatment that is IVF or some other fashion. The 2 options that lead to under failed IVF are the treatment and control. The variable can work in any world, since that has a constant affect, by means of instruments.
• The successful attempt to test children through these means leads to causal attempt. The random instruments have a relationship, while potential outcome and treatments are made to be random or binary in type. In this random state, children are more easily accounted for. From the previous equation, this one is made for the complier relationship. All of the relationships that are made here are related to the first state instrument with a successful attempt at achieving certain stages within fertility.
• The next step in this model is to find children and have the same results affect the labor supply.

IVF Method (Further Exploration)

• By substituting the terms for Z we can arrive at the solution required. -The estimators required that all be valid and the right relationship has to be achieved or there will be inaccuracies in compliers and local treatment methods.
  • If and when completed, these estimators can point to the accurate treatment effect.