Wednesday, December 16, 2009

Some Further Comments on Nominal Wage Flexibility

Tyler Cowan thinks that we should cut the minimum wage, and links to Bryan Caplan for an explanation. And Caplan thinks that it's all quite elementary:
Cutting wages increases the quantity of labor demanded.  If labor demand is elastic, total labor income rises as a result of wage cuts.
Even if labor demand is inelastic, moreover, wage cuts reduce labor income by raising employers' income.  So unless employers are unusually likely to put cash under their matresses, wage cuts still boost aggregate demand.
    Let's take this step by step. First, consider the claim that cutting wages increases the quantity of labor demanded. Through what mechanism does this occur? Consider a firm (McDonald's, say) that can now pay its workers less. It will certainly do so. But will it increase the size of its workforce? Not unless it can sell more burgers and fries. Otherwise its newly expanded workforce will produce a surplus of happy meals that will (unhappily) remain unsold. And this will not only waste the expense of hiring and training new workers, it will also waste significant quantities of meat, potatoes and cooking oil. So the firm will make do with its existing workforce until it sees an uptick in demand. And no cut in the minimum wage will automatically provide such an increase in demand. As a result, the immediate effect of a cut in the minimum wage will be a decline in total labor income.

    Employer income, of course, will rise. Some of this will be spent on consumption, but less than would have been spent if the same income had been received by low wage earners. The net effect here is lower aggregate demand. But wait, what will happen to the remainder of the increase in employer income? It will not be placed under mattresses, on this point I agree with Caplan. It will be used to accumulate assets. If these are bonds, then long rates will decline, and this might induce increases in private investment. Then again, it might not, unless firms believe that additions to productive capacity will be utilized. And right now they do not: private investment is not being held down by high rates of interest on long-term debt.

    Finally, what if employers use the unspent portion of their augmented income to buy shares? We would have a run up in stock prices not unlike that we have seen in recent months. Note that this would not be a speculative bubble: the higher prices would be warranted given that firms have lower labor costs. But would this asset price appreciation stimulate private investment in capital goods? Again, not unless the additional capacity is expected to be utilized.

    Mark Thoma has more on this, as does Paul Krugman. I discussed the opposing views of Becker and Tobin in an earlier post. What I cannot understand is why people of considerable intelligence persist in conducting a partial equilibrium Walrasian analysis of the labor market, as if we were dealing with the market for oranges. Please stop it.


    Update (12/16). Thanks to Mark for reposting this, and to Paul Krugman for following up. Tyler responds as follows on Mark's page:
    Graph a monopolist and shift the marginal cost curve down. Watch what happens. The first main paragraph of Sethi simply doesn't consider this mechanism but rather it assumes that changes at the margin don't matter.
    To which I have replied as follows:
    Tyler, a fall in wages won't just shift the marginal cost curve down, it will also shift down the demand curve and hence marginal revenue. This will happen for all firms who produce stuff that minimum wage workers consume, regardless of whether or not they themselves pay minimum wage. Now we may disagree about what the eventual outcome will be (I think there will be debt deflation, as described in Fisher 1933, and so admirably summarized recently on your blog). You may disagree. But we can't settle this with partial equilibrium models of monopolists and their cost curves.
    I'm sure that this debate will continue. My only hope is the we get beyond partial equilibrium models from introductory microeconomics in discussing these terribly important questions.  


    Update (12/17). The Economist points out (quite sensibly) that even demand constrained firms may increase hiring with lower wages in order to improve product quality: "If a firm can produce a particular number of burgers with either 5 workers earning $7 an hour or 6 workers earning $6 an hour it won't necessarily prefer the 5 worker set-up." True enough, but this is not the appropriate comparison: if the wage drops, employment will rise only if 6 workers earning $6 an hour are preferred to 5 workers earning $6 (not $7) an hour. In other words, the choice is between improving quality or pocketing the difference in wages.

    I do not doubt, by the way, that a lower minimum wage could raise employment in some firms. But unless the aggregate wage bill rises, some other firms will experience declining demand, and may respond by laying off workers across a much broader range of skills. My point is simply that in order to capture these effects, it is essential that one move beyond the partial equilibrium framework. More here.


    Update (12/18). Winterspeak adds:
    Any discussion of the minimum wage turns into a Rorschach test for the ideology of the commenter.
    This is unfortunately true, and ideology does seem to be clouding judgment and analytical clarity on the issue. My focus has remained on the methods one uses to reason about the effects of changes in the minimum wage, rather than on the magnitude or sign of these changes. I agree with Winterspeak, Tyler Cowan, and The Economist that the effects of such changes on aggregate demand, regardless of their sign, will likely be small. But it's important to get the reasoning right, because the same methods apply to discussions of flexible wages and prices more generally. To answer Andrew Gelman, this is why I think the debate is important. Hence the title of this post, and the content of the earlier post to which it refers.


    Update (12/19). Perhaps some signs of progress in this debate. Alex Tabarrok writes:
    A fall in wages increases the incentive to hire (call this the substitution effect) but it decreases the income of people who already have jobs and this in turn decreases their spending and other people’s income (call this the macro income effect).  In essence, Krugman and others are arguing that the macro income effect can dominate under certain situations.
    My response:
    Yes, but one needs to go a bit further than this. If the macro income effect dominates, then there will be layoffs in other sectors, most immediately in those producing outputs that low wage workers purchase. And these layoffs will not be confined to low wage workers. This means that even if the employment of low wage workers rises, there will be a decline in employment across other segments of the skill distribution. Concede this point and we're done. 
    It's worth thinking a bit more closely about what Alex calls the substitution effect. This is presumably what Tyler had in mind when he urged me to shift marginal cost curves down and watch what happens. He was arguing that the firm will lower price and use the additional hires to step up production to meet the increased demand. Now obviously this will not increase market share if one's competitors are doing the same thing, but it could increase the overall consumption of goods and services produced with low wage labor. Even disregarding the macro income effect, this implies a decline in demand for goods and services produced with skilled, high wage labor. In other words, there will be a "substitution effect" in unemployment patterns, more unskilled jobs and less skilled jobs. Add the macro income effect to the analysis and matters become worse. The only way that lower wages can increase aggregate demand is by stimulating private investment through changes in asset prices and interest rates. This is just not going to happen under current conditions (hence my support for this proposal). In any case, a partial equilibrium analysis is totally inadequate for thinking through these questions.


    1. I noted this on Thoma's blog, but not all workers are in production. What if McDonals's hired more janitors? More janitors would not lead to more burgers, but it could lead to a happier customer experience, potentially increasing revenue.

    2. Let me prefix this by saying that I agree with your conclusions on practical grounds.

      But I'm not sure if the argument is quite right. On the theoretical question, I believe Tyler is right. First, even at current production levels there is a marginal worker who is not being hired but might be hired at a lower wage. Second, a reduction in McDonalds' costs should lead to a (smaller) reduction in product prices, and an increase in demand and production. This in turn could increase employment.

      But this does not resolve the question of whether AD falls as a result of lower total wage income. This can only be determined empirically.

      My instinct is that on this point you probably have it right.

      How about if, instead of reducing minimum wage across the board, there is a reduction only for those who have been out of work for at least 3 months? This should avoid or mitigate the AD effect, while still increasing employment at the margin.

    3. Why is it that economists think everyone outside their campuses works in a factory or food service? This linear relationship between labor quantity and goods output quantity only exists in simplistic textbook models. The vast majority of labor today is hired for a variety of functions besides production. Reductions in any sort of cost (including labor), by improving margins, increase the (theoretical) hiring capacity of a business.

      Now, whether or not the business chooses to hire more minimum wage labor off the street with this savings as would be the stated policy intent, or instead hires a higher-wage worker, or just pockets the difference, is an open question. All a minimum wage reduction does is reduce a regulatory frictional cost of business that today cuts off one end of the labor demand curve.

      It may be AD-neutral, but at least it might redistribute income from the poor to the destitute.

      In any case, it should be trivial to prove that if a reduction in the minimum wage allowed McDonald's to hire 5 marginal economists, no marginal burgers would be made.

    4. He mispeled "mattresses" too. :)

    5. First, congratulations to Rajiv for his short post in the New York Times today having a go at Krugman (17th Dec 09). Rajiv just copied and pasted the last sentence of the above article to the New York Times. But good point all the same.

      Now for Rajiv’s article above. His first large para above (starting “Let’s take this step by step....”), essentially repeats Keynes’s point that wage cuts do not increase demand. True.

      However, what a cut in the minimum wage DOES do is to reduce NAIRU, which MAKES POSSIBLE an increased demand. Reason is that after the minimum wage cut, employers can employ labour that was previously so unproductive that its output did not cover the minimum wage. (This of course assumes that this low productivity labor is no dissuaded from taking minimum wage jobs by the attractions of living on social security.)

      The above para of mine says something very similar to Dan Nile above: “All a minimum wage reduction does is reduce a regulatory frictional cost of business that today cuts off one end of the labor demand curve.”

      At the end of this para, Rajiv claims “As a result, the immediate effect of a cut in the minimum wage will be a decline in total labor income.” My answer is “not necessarily”: it depends on to what extent employers are allowed to discriminate between employees with differing abilities, and pay them different wages to reflect this difference.

      To illustrate, with a simple example. Suppose all existing MacDonalds employees are on minimum wages and are actually worth this amount per week to employers. Suppose the minimum wage is cut. Employers continue to pay the latter employees the old minimum wage and pay new employees the new minimum wage. Net result will be an INCREASE in labour income.

      This is not to advocate minimum wage cut. Certainly such a cut is a daft solution to the recession. The level of the minimum wage is a LONG TERM question: it has nothing to do with whether the economy is currently in recession.

    6. "what a cut in the minimum wage DOES do is to reduce NAIRU"

      Yes, and it also decreases the weight of pink unicorns to the point where they can fly!

      I just think if you're going to talk about made-up things that don't exist in the real world, pink unicorns are a lot more fun than NAIRU...

    7. I think Tyler Cowen's argument has some force to it. Minimum wage workers, in their role as consumers, are responsible for only a small fraction of aggregate demand (aren't they? I don't have the numbers.) So a fall in their income resulting from a minimum wage cut is not likely to make a firm that employs such workers think that its demand curve will shift down by much (unless it sells its output mostly to such workers, probably a rare case). So it will cut its price and expand output when its labor cost falls. On the other hand, because minimum-wage workers don't account for a large share of costs in a modern economy, this won't have a large impact. What we have to remember is that most wages are downwardly rigid due to considerations of reciprocity and fairness, adverse selection, and moral hazard, not government intervention.

    8. Jimbo: can you be more precise in your criticisms?

      Are you suggesting that all abstract nouns are nonsense: love, hate, speed and NAIRU? Presumably not.

      Or are you saying that NAIRU is an invalid concept? If so, you presumably also claim that the Phillips curve and the “natural level of unemployment” (a very similar concept to NAIRU) is also nonsense. If so, are you saying that there is no relationship between employment and inflation? One possible interpretation of this is that we can let demand go thru the roof, which in turn brings full employment with no danger of inflation. Is that what you are saying?

      I’ve been in favour of more stimulus over the last year, but I think that ludicrously large amount of stimulus would be inflationary. Am I wrong?

    9. I know this is an old thread, but...

      Why always focus on McDonald's? McDonald's is not a typical restaurant. Restaurants open and close ALL THE TIME and very few of them are McDs. Do you really believe there are no marginal restaurants that would open or not close because of cost reductions?

    10. Ed, what I said applies to any business, not just McDonalds. Businesses close if they can't make money, and whether they can make money depends not only on costs but also on whether they have enough customers. If a single business could lower its costs while the costs of its competitors remain the same, it would thrive. But a general reduction in wages affecting all businesses will result in lower prices. This makes it harder to repay debts, which are fixed in nominal terms. The distributional effects of lower wages can also reduce demand for goods and services. Sufficiently reduced demand can cause business to close even if their costs have gone down.

    11. First, the wages of most employees are not prevented from falling by the minimum wage; they are prevented from falling by social conceptions.

      Second, gasoline probably pays a large part of the reason to avoid cutting wages, even if the annual cost of gasoline in 2009 was only $926 for the lowest quintile (5% of consumption) and $3,067 for the highest quintile (3% of consumption).

      That said: cutting wages is the way to reduce unemployment, when the wage cuts are designed to encourage people to work less and therefore buy fewer luxury goods from the subset of companies that captured 88% of growth of national income following the start of the 'recovery'.

      According to the latest NYT/CBS opinion poll 42% of those surveyed thought the gov't should spend more money to create jobs even if it meant borrowing money, while 52% thought the government should not spend money and should focus on lowering the country's debt. Therefore, any reduction of unemployment will likely have to take place without fiscal stimulus.

      The way to do so is described here: