08 Feb 2014

India’s Child Labor Ban: Can Someone Draw Me a Picture?

Economics 45 Comments

James Schneider, the new guy at EconLog, has an interesting post on an unintended consequence:

Some take it as a matter of faith that increasing taxes will dull people’s desire to work. However, higher taxes can sometimes cause people to work more. When higher taxes reduce the after-tax wage, people are poorer for any given number of hours worked. When they become poorer, many people are more anxious to earn extra money. Economists call this the “income effect.” (Although in this context, I like to call it the “poverty effect.”)

Increasing tax rates is not the only way that the government can reduce wage rates. If the government punishes your employer for hiring you, your services will be less marketable, and your wage will fall. This phenomenon is important when countries enact bans on child labor. If a country is so poor that many parents send their children to work, then it is unlikely to have the wherewithal to perfectly enforce a ban on child labor. When the wages of children go down, the poor families that depended on child income will become even more desperate. This might cause parents to have their children work more.

A recent paper shows that this is exactly what happened when India enacted the Child Labor (Prohibition and Regulation) Act of 1986. The law banned children under 14 from working in many industries. After these rules took effect, the wages of children under 14 fell relative to those over 14.

Something about this doesn’t feel right to me. I’m worried that if free-market economists embrace this, we’re doing the mirror image of the economists who come up with ways to explain how the minimum wage might lead to more employment.

I mean think about it: We’re saying that initially, employers in India hired x million children. Then the government said, “We are going to start randomly fining you for doing that.” In the new equilibrium, employers hired more children.

(Yes yes, I understand the theoretical argument of how this can happen; wages fall so much that even when you factor in the risk of fines, employers end up hiring more child labor. But it just doesn’t feel right to me. For one thing, why does it “overshoot”? And notice that there is a lot of emphasis on the fact that the enforcement is lax. Well OK, if the enforcement is lax, then the demand shouldn’t fall as much, meaning wages shouldn’t fall as much, meaning the income effect on the supply of labor shouldn’t be as big a deal… See what I mean? The whole thing just seems really convenient, sort of like how the ACA just so happens to boost employment precisely when the economy needs it, but then reduces employment precisely when the economy needs it.)

The actual paper is here. They have a Figure 1 (page 11) that tries to illustrate their explanation in a Supply & Demand framework, but I can’t make sense of it. Also, if I’m understanding it, they’re showing a constant Demand curve for labor and expressing the y-axis in terms of wages net of fines.

That approach is totally non-intuitive (which isn’t to say counterintuitive) to me. I want to draw initial Supply and Demand curves, with wages being the actual wage paid (and received), and then when the government threatens fines, the demand curve shifts left.

Because of the Giffen good stuff going on, you have to have regions of the supply curve that are downward sloping. So my question: Can anybody draw me a Supply and Demand curve for child labor, where the leftward shift of demand leads to a lower wage and a higher quantity?

45 Responses to “India’s Child Labor Ban: Can Someone Draw Me a Picture?”

  1. Ken B says:

    I’m not going to read their paper. My intuition is that you are right.

    In the tax example the assumption is that there is a demand for the worker’s extra labor. Likely true if he can get a second lower wage job. There already exists a good spread of salaries, so existing demand at a lower one than the taxpayer paying the higher rate — excludes some — can get hours. Is that true with rug rats?

  2. Ken B says:

    Samantha Bee can do it. Her SD curves are magical.

  3. Tel says:

    Well OK, if the enforcement is lax, then the demand shouldn’t fall as much, meaning wages shouldn’t fall as much, meaning the income effect on the supply of labor shouldn’t be as big a deal… See what I mean?

    The employer demands additional risk premium, and probably needs to pay more bribes in order to keep doing business.

    Also, new employers may pop up now that more families are desperate and the new employers are already in a questionable legal position, but now they can take advantage of the cheap child labour.

  4. Jonathan Finegold says:

    Okay, at first I was going to agree with you. I almost finished my original comment before I realized that Schneider may have a point.

    Government increases regulatory costs of hiring child labor, causing firms to either reduce child employment at the going wage rate or push the wage rate down. Either way, families lose income and increase the number of hours they are willing to work. This increases the labor supply, pushing wages down again.

      • Bob Murphy says:

        Right but Jonathan, you’re moving the Supply curve of labor around in response to a change in wage rates. I’m wondering if you can draw just a single supply surve of labor, which takes into account both the substitution and income effects from the get-go. So that it actually plots a given wage rate with the quantity of labor supplied at that rate, which is what a supply curve is supposed to do.

        • Ken B says:

          Yes, I can J’s diagram does it, but you can do anything if you’re allowed to move both curves willy nilly. Why would it move? Bob’s comment makes a cogent objection out of my disquiet/confusion.

          • Ken B says:

            Bob help me out here.

            The only reason I can see why the S would shift would be as an independent reaction to the ban. It cannot be a reaction to the shift in D as you note, from the definition of what S shows. So what’s left? The direct efffect of the law on S is what.

            Is that right?

            If so, is that shift plausible?

        • Daniel Kuehn says:

          I agree with Bob – if your income effect isn’t already in your supply curve, something is wrong with how you’ve derived your supply curve.

          • Ken B says:

            Right. So the shift is only legit as an independent effect in response to the law. So You have to think that said shift would happen even if the D curve did not move.

            • Daniel Kuehn says:

              Or it’s shaped funny.

              I think lexicographic preferences should look a little like backward bending supply curves – see my comment below on that. Not 100% sure cause it’s been a while since I’ve worked with them (and never in the context of labor supply).

          • Ken B says:

            Watching you two agree this way, it’s like watching bonobos.

        • Jonathan Finegold says:

          Can you rephrase? Same goes for Daniel’s comment.

          I think I get what you and Daniel are saying (and I think I agree now) — I’m just trying to think through it.

          Is this pretty much what you guys are saying? When the demand curve shifts, we expect the substitution effect to outweigh the income effect (which is why the amount of labor supplied increases with increases in demand [despite a rise in incomes], and why the amount of labor supplied decreases with a fall in demand). I think that’s right, and I didn’t think about that yesterday.

          • Jonathan Finegold says:

            Of course, implicit in all of this is that the substitution and income effect are already in the supply curve — which is what you and Daniel are saying. I agree with you guys; I was wrong.

        • Major_Freedom says:

          “Right but Jonathan, you’re moving the Supply curve of labor around in response to a change in wage rates. I’m wondering if you can draw just a single supply surve of labor, which takes into account both the substitution and income effects from the get-go.”

          I don’t think you ever keep the supply curve fixed in the context of other factors changing. For this particular case of child labor, I think the supply curve SHOULD move in response to wage rates changing, for the following reason:

          If wage rates fall, then all else equal more children, or more children hours, should rise. We capture this by imagining a movement ALONG the demand curve. So at this point in the analysis, the supply curve can be assumed as fixed. But wait, that isn’t the end of the story. With more children, or more child labor hours, production would be higher than it otherwise would be. That increase in production of goods will reduce the need for child labor in the first place, regardless of child wage rates. So the supply of child labor should fall, i.e. shift, and not remain fixed.

          This happened in the US incidentally. During the 18th century, people went to work starting at around 8 years old. Then the age progressively increased as production increased, which allowed parents to raise their children on only parental income. The supply of child labor fell but not because child labor wage rates changed.

          • Ken B says:

            A real economist can correct me but I thought supply can shift in response to anything EXCEPT demand.

            • Major_Freedom says:

              I didn’t say DEMAND for labor shifts the supply curve. The demand for labor in nominal terms can be held fixed if you want, and a fall in wage rates will, ceteris paribus, increase the number of labor hours worked. My argument is that with a rise in the number of hours worked will increase production and supply of goods, and make it less necessary for children to work. So in this scenario, you can imagine the nominal demand for child labor remaining fixed, with fewer children in the workforce because their parent’s real income is higher.

              I am a “real” economist by the way. I am a person who studies economics and makes economics arguments. A person’s “label” isn’t determined by whether or not they get paid to do it.

              • Ken B says:

                It’s not a matter of credentials but coherence, sense versus gobbleeydee gook.

              • Major_Freedom says:

                Except I didn’t argue what you implied I argued. While I did make a typo, I think the problem here is that you are reading something in my post that I did not say.

            • Major_Freedom says:

              And interestingly enough, your statement that “supply can shift in response to anything EXCEPT demand” is actually something you might want to discuss with DK, because he wrote:

              “And what about adding costs to child labor shifts the supply curve in that way?”

          • Daniel Kuehn says:

            How are you moving along the demand curve if the supply curve is fixed? And what about adding costs to child labor shifts the supply curve in that way?

            This is completely garbled as far as I can tell, but maybe I’m missing something.

            • Major_Freedom says:

              Sorry, you’re right DK, it was a typo, I meant to say movement along the supply curve. Wage rates change, and ceteris paribus, the supply changes but not because the supply curve shifted, but because of a movement along the supply curve.

              With respect to your point about adding costs to child labor if the supply curve shifts, sure, that could very well take place as well. My point is only that because a fall in wage rates raises hours worked, and hours worked affects production and supply, and production and supply affects the number of children in the workforce and/or number of child hours worked, then the supply curve should not be held fixed in the face of changed child labor wage rates.

  5. GabbyD says:

    regarding being “lax” enforcement.

    the authors are comparing it to “perfect” enforcement, where NO children at all can be hired. in this case, the supply curve MUST stay the same.

    in an imperfect enforcement scenario, you may still increase the supply of labor (children), but it costs more as you may be found out (i.e. its a risk).

    you are right that as you decrease the “perfection” of the enforcement, the effect weakens. they would agree., but the comparison is NOT with stronger/weaker, but perfect vs imperfect.

  6. Tel says:

    That approach is totally non-intuitive (which isn’t to say counterintuitive) to me. I want to draw initial Supply and Demand curves, with wages being the actual wage paid (and received), and then when the government threatens fines, the demand curve shifts left.

    You can draw it, just picture the supply curve as having a constant dollar value. The child labour is compelled to bring in a certain number of total dollars per week… that’s your supply curve. Yes, it would imply the gradient of the curve is negative. The curve would be roughly hyperbolic like y = 1/x

    This isn’t completely dark side of the moon exploration, it’s a pretty well studied idea:

    http://en.wikipedia.org/wiki/Backward_bending_supply_curve_of_labour

    • Bob Murphy says:

      Tel, I’m not saying it’s dark side of the moon, just like Daniel Kuehn says that the pro-minimum wage literature is standard econometric stuff.

      Yeah I think I can draw two demand curves embedded on that picture you sent, to show how a drop in demand can lead to a higher equilibrium (and unique) quantity of labor.

      But note your constant dollar value suggestion doesn’t work. I didn’t spell it out in the post, but when I asked why does it “overshoot,” I mean that the authors say wages fall a certain amount, but labor supplied rises by much more. So that children end up earning a higher total amount of wages after the ban.

      • Tel says:

        Do they actually make the claim that total wages are increasing?

        I admit I read the summary and just skimmed through the main paper, but their only direct measurement is labour participation. They admit the don’t know how many hours are worked but they infer more child labour total from the increased participation.

        Finally, we show that other key components of household welfare, such as per capita expenditure and assets, seem to decrease slightly after ban. This suggests that overall the ban may have made households worse-off.

        This would suggest that total money coming home with those kids has reduced, despite them working harder. I don’t see any evidence of overshoot presented here. I agree with your theory BtW, there’s no reason I can think of to expect an overshoot.

    • Tel says:

      I’m looking at the main article and Fig. 1 does show a supply curve that looks pretty much hyperbolic, some relation to the y=1/x that I mentioned above. They call the curve LS(D) and you can easily see it is backwards bending.

      The demand curve is straight (labelled as LD), and this is drawn from the employer’s point of view with the stated presumption that the employer is indifferent to hiring adult or child workers, thus from an employer’s point of view the demand curve does not shift.

      It might be more intuitive drawn from the worker’s point of view… in which case the child worker would see it that their supply curve has not shifted, but employer demand for child labour has decrased (i.e. the demand curve moves downward). The gap between those two perspectives is made up of fines, bribes, etc that cost the employers more, but don’t actually pay the workers. What would an economics article be without pages of grinding prose and one somewhat ambiguous diagram?

  7. andrew' says:

    More black market?

    • andrew' says:

      I think I lived this, BTW, I wasn’t little Caesar’s material so I worked a lot more at lawns. I didn’t draw the SD curves at the time.

      Isn’t the goal that kids will work long hours for nothing in schools?

      • andrew' says:

        To your mirror image point, the problem with biased experts is they can stop the analysis wherever they want and emphasize whichever part they want.

  8. Garrett M. Petersen says:

    My understanding is that the ban reduces the number of children under 14 (those subject to the ban) working, but that more children over 14 (not subject to the ban) were drawn into the labour force, offsetting the reduction in under-14 labour.

    • Bob Murphy says:

      Garrett, if that’s what they were saying, it would make a lot of sense to me, but I don’t think that’s what they’re announcing to the world.

      However, you make a good point because in some of the empirical discussion, they made arguments that in my mind didn’t establish the age of the children working.

  9. Anonymous says:

    Always tough to draw conclusions about labor markets, especially those outside the USA…I enjoy and admire Murphy’s rare open-minded approach…

  10. Gamble says:

    Define wealth destruction:

    Make people work more so bureaucrats can waste more. Waste includes talking incessantly, feet on desk, web surfing, and other non productive activities day in, day out…

    “Some take it as a matter of faith that increasing taxes will dull people’s desire to work. However, higher taxes can sometimes cause people to work more. When higher taxes reduce the after-tax wage, people are poorer for any given number of hours worked. When they become poorer, many people are more anxious to earn extra money. Economists call this the “income effect.” (Although in this context, I like to call it the “poverty effect.”)”

  11. Daniel Kuehn says:

    Options seem to include backward bending supply curves, but that’s weird because it would imply the income effect is very strong for families sending their kids into child labor. That’s not plausible. These sorts of families should have normal supply curves. Rich families will have backward bending, but they don’t employ child labor.

    There is one story we MIGHT spin where backward bending supply curves make some more sense (I’m not 10% sure but it could work): lexicographic preferences. If the negative wage shock pushes you below subsistence you may work yourself to the bone until you’re up to subsistence again, at which point other goods that are somewhat lexicographic themselves (like time to sleep or rest) might re-enter the utility function. Once THEY’RE satisfied we have normal upward sloping demand again. So that would be a really wiggly supply curve, but I suppose it’s possible.

    I agree with Bob this seems implausible.

    The design is a DID, but the use of over and under 14 year olds for this seems a little dubious because of spillovers between labor markets, although (I think) the bias should be in the other direction (the bias from spillover should show a stronger negative impact than is actually the case because it would make the control group look better off), so the fact that it shows a positive effect is a little odd.

    • Gamble says:

      I think the unknown factor is money supply expansion. You send your spouse to work and children for more incomes, for a while this is extra but eventually the new money becomes the required money.

      Instead of CPI, an “essentials index” would shed light on the subject…

      • Daniel Kuehn says:

        I don’t think so. Those sorts of effects would be impacting the comparison group (> 14 year olds) too.

        • Anonymous says:

          Not arguing, could you expand your response?

          My statement was more of broad based phenom. Working more does not help if purchasing power drops equal amount. That is all I was saying.

          Money base is the governing factor of purchasing power.

          More work +same base= more purchasing power.
          More work+ more base= same or less purchasing power.

  12. Neal Provost says:

    If a government increases the difficulty of hiring a certain group of people, we know those people will be worse off. But since they can make choices, we do not know how they will be worse off. Maybe they will work fewer hours and be poorer. Or maybe they will work more hours for lower wages and still be poorer. Perhaps the latter happened in India, but I am not sure how we could predict ahead of time what people will choose in their fight against government coercion.

    I think this is different than the minimum wage debate. In that case, supporters are claiming that the variables will line up just perfectly to make the targeted group better off. In the India situation, I do not think anyone is claiming that increased child labor is an improvement.

    We all understand what the minimum wage supporters are claiming, and it is conceivable that low-wage workers (as some sort of aggregate) could benefit at the expense of others. We just believe it is highly improbable since those others will make choices to mitigate their losses. In other words, we are claiming that long-term consequences will be negative for the targeted group. However, in the India situation, is anyone better off? If not, why is this improbable, or “convenient”?

  13. Bill Woolsey says:

    The simple explanation is a negatively sloped supply curve.

    Make the supply curve more and more steep, hitting perfectly inelastic. Then keep going.

    The positive sloped supply curve occurs because the income effect is greater than the substitution effect.

    Lower labor demand creates a surplus of labor. It drives down wages. Quantity of labor demanded rises as does quantity of labor supplied. At the new equilibrium, they are equal, with a lower wage, as usual, and a higher quantity of labor (unusual.)

    Now, if the supply of labor is negatively sloped and more elastic than demand, so that a lower wage raises quantity of labor supplied more than quantity of labor demanded (just keep on testing it around so that it starts off below demand, crosses, and then is above,) then it is unstable. Push wages up, and it creates a shortage of labor. Which raises wages more, and so a greater shortage. And the opposite is true if wages go below equilibrium.

    Now, can the income effect be greater than the substitution effect for poor people? I think so.

    Economic development and higher demand for labor results in higher wages and a reduced quantity of labor. That is, more leisure as well as more consumption of material goods and services.

    • Major_Freedom says:

      “Economic development and higher demand for labor results in higher wages and a reduced quantity of labor. That is, more leisure as well as more consumption of material goods and services.”

      This is related to the point I made above. As “development” occurs, the supply of goods and services increases, and that makes parents and children less needful of the income derived from child labor, regardless of whether or not there are employers who are offering wages to any children who would accept them.

      Here, it is not that the supply of child labor is changing because of the price of child labor is changing, but rather, the supply of child labor is changing because of the supply of goods changing, which ends up shifting the child labor supply curve.

  14. Bill Woolsey says:

    The simple explanation is a negatively-sloped supply curve.

    A lower demand results in a lower wage and a higher quantity of labor.

    This occurs because the income effect Is greater than the substitution effect, which is highly plausible.

    Don’t try to make it bend. Just draw it negatively-sloped. It starts above the demand curve and ends up below it.

    Consider this. Economic development raises labor demand and wages. This raises the real incomes of workers, and they choose to enjoy more leisure as well as more material goods and services. The quantity of labor supplied decrease.

    Sadly, if the reverse occurs, the wage falls and the quantity of labor rises.

    If poor people really have the whole family working 18 to 20 hour days, 7 days a week, 52 week a year from 10 to death, then the supply of labor is perfectly inelastic. That isn’t true of most poor people.

    If the negatively sloped supply curve is flatter than the demand curve, (keep twisting it counter-clockwise) then it becomes unstable. If the wage rate falls below equilibrium, it creates a surplus, and lower wages make the surplus worse. And if the wage rises above equilibrium, it creates a shortage, and a higher wage makes the shortage worse.

    But as long as the supply curve, though negatively sloped, is steeper than the demand curve, a below equilibrium wage creates a shortage and a above equilibrium wage causes a surplus.

  15. James Schneider says:

    @ Daniel K

    If the negative wage shock pushes you below subsistence you may work yourself to the bone until you’re up to subsistence again

    I think this is more or less what is happening

Leave a Reply to Bob Murphy

Cancel Reply