Tag Archives: employment growth

The Impact of Increasing the Minimum Wage on Unemployment: No Evidence of Harm

Posted on by

Minimum Wage vs. Unemployment Rates, 1950-Jan 2013

Minimum Wage vs. Ratio of Unemployment Rates, 1950-Jan 2013

A.  Introduction

In his State of the Union speech last month, President Obama called for a rise in the federal minimum wage from the current $7.25 per hour to a new rate of $9.00 per hour.  This would be a 24% increase, but would still mean that someone working full time, 40 hours per week, 52 weeks a year (no vacation), would earn only $18,720 a year.  Such a full time worker would still be earning well less than the current federal poverty line for a family of four of $23,050 per year.  The proposed increase is modest.

Republican leaders nonetheless immediately denounced the proposal, asserting that raising the minimum wage would hurt, not help, the poor, as they would lose their jobs.  They assert that instead of seeing an increase in their wage, the minimum wage workers would be fired.  And unlike in many other areas (such as the impact of fiscal policies) the Republican leadership here is making an argument that one would find in an introductory economics course.  Elementary economics would indeed argue that in perfectly competitive markets, an increase in the minimum wage would lead to such workers losing their jobs rather than being paid more.

B.  Have Increases in the Minimum Wage Led to Higher Unemployment of Such Workers in the Past?

But what is the evidence?  If increases in the minimum wage lead to such workers being fired, one would see higher unemployment among such workers very quickly in the months following each increase in the minimum wage in the past.  The graphs above show what in fact has happened.

The first graph shows the federal mandated minimum wage since 1950, in real inflation adjusted terms (using the CPI), plus the unemployment rates for all workers and separately for workers aged 16 to 24.  The data comes from the Bureau of Labor Statistics, but was downloaded for convenience from FRED, the Federal Reserve Economic Data web site maintained by the Federal Reserve Bank of St. Louis.  The unemployment rate for workers aged 16 to 24 is shown separately as one would expect that increases in the minimum wage would increase unemployment especially sharply in that group, if the assertion is correct that increases in the minimum wage lead to such workers losing their jobs.  Approximately 51% of the hourly wage workers earning the minimum wage are in this 16 to 24 age group.

First of all, it is worth noting that the minimum wage, when adjusted to reflect general inflation, is a good deal lower now than a half century ago.  It reached a high (in prices of January 2013) of $10.82 in February 1968, and lows of $6.02 in March 1990 and $5.75 in June 2007.  It averaged close to $9.00 an hour over the twenty-five years of 1957-81 (inclusive), and is only $7.25 currently (almost 20% less).  The Obama proposal is modest, as he has only asked Congress to bring it back to that $9.00 an hour, the rate of a half century ago.

But does one see in the history that increases in the minimum wage lead to a jump in unemployment rates, particularly of the young (and decreases in the minimum wage leading to lower unemployment rates)?  Actually, no.  Unemployment rates do fluctuate a good deal, as they depend on macro conditions in the economy.  But it is hard to see any obvious jump in unemployment rates in the months following the sporadic increases in the real minimum wage we have had over the last more than 60 years.

It is also worth noting that due to the politics of the minimum wage, and the resistance of conservatives and businessmen to higher rates, increases in the minimum wage have been generally infrequent and then relatively large in percentage terms.  This thus provides good material for a test of whether increases in the minimum wage lead quickly to jumps in the unemployment rate (particularly of the young).  Yet one does not see it.

Any relationship might be hard to recognize in part because of the independent rises and falls in unemployment rates due to macro conditions.  The second graph therefore charts the real minimum wage along with a line that shows the ratio of the unemployment rate of those aged 16 to 24 to the unemployment rate for the entire labor force.  This ratio exploits the fact that a relatively high share (about 51%, as noted above) of minimum wage workers are young.  Thus, if it is in fact true that increases in the minimum wage will lead those making the minimum wage to become unemployed, the higher share of such workers in the ages 16 to 24 category will lead to an increase in that ratio.

But the graph does not show such a relationship.  While the real minimum wage has seen many sudden changes, the ratio of unemployment rates often does not then change, and in fact sometimes moves in a direction that is the opposite of what those opposed to increases in the minimum wage would predict.  The one possible exception appears to be a blip seen following the February 1968 increase in the minimum wage.  But this blip occurs in fact four months later (in June 1968; the time scale is compressed as the chart covers 63 years) and then drops back.  But one also sees that the steady decline in the real minimum wage during the 1980s in the Reagan years was accompanied by a rise in the ratio (the opposite of what they would predict), and that the ratio then declined (instead of rising) when the minimum wage was finally moved up in 1990 and 1991.  Similarly, increases in 2007-09 were accompanied by  a decline in the ratio.  One could pick out other examples, but basically what is shown is that there is no systematic pattern.

C.  More Rigorous Work on the Impact on Employment from Raising the Minimum Wage Also Shows No Harm

Graphs such as these are, however, simplistic.  While looking at the history is of interest, and should show at least some indication that increases in the minimum wage will lead to higher unemployment (especially of the young) if the minimum wage critics were correct, there is much more going on in the economy which should be taken into account.  Fortunately there have been rigorous studies that do this, and they too have found that the impact of increases in the minimum wage (of the magnitude historically seen) on unemployment rate is either non-existent or small.  Some studies have indeed found that increases in the minimum wage have reduced unemployment.

The seminal paper that launched the modern literature on the impact of the minimum wage was co-authored by economists David Card (of University of California, Berkeley) and Alan Krueger (then of Princeton, and now Chairman of the Council of Economic Advisors in the White House).  Card and Krueger controlled for extraneous effects that might be going on in the economy at the time a minimum wage is increased, by exploiting the fact that states may have separate minimum wage requirements from the federal requirements, and that states change their minimum wages at different times.  Specifically, they looked at employment in fast food establishments along the border between New Jersey and Pennsylvania on an occasion when New Jersey raised its minimum wage while there was no change in Pennsylvania.  They found no negative employment impact from New Jersey’s action.  Indeed, there might have been a small positive impact on employment following the increase in New Jersey’s minimum wage.

The Card and Krueger study, while rigorous, was limited as it looked at only one instance of a change in the minimum wage, and the impact on only one industry.  But more recent studies, following a similar approach, have extended the Card-Krueger work to many more cases.  A particularly comprehensive recent example is a paper by the economists Arindrajit Dube, William Lester, and Michael Reich.  Like Card-Krueger, they too found that increases in the minimum wage that have been periodically enacted in the US at the state level have not had a negative impact on employment of minimum wage workers.  And since changes at the state level would presumably have a bigger impact than changes at the national level (as jobs could in principle shift across state lines), there is no reason to believe the impact of a federally mandated change would be negative.

There is of course much more work on this issue, and a paper by John Schmitt issued last month provides a good summary of where the literature stands.  Less technical reviews of the issues and what economists have found are available here and here.  There are, of course, economists who would disagree, but the preponderance of the work done so far has found little or no evidence that increases in the minimum wage that we have seen in the past have led to decreases in employment of such workers.  Indeed, some studies have found that increases in the minimum wage increases employment.

D.  Why Does Standard Economic Theory Get This Wrong?

The real world evidence matters more than the theory.  But why would standard elementary economic theory appear to have gotten this wrong?  The truth is that there are many matters in the real world that do not behave as one might predict based purely on an economic abstraction.  Unemployment exists, for example, even though economic theory would predict that in a world of perfect competition, with full information and no transactions costs, and many other conditions, we should only see full employment.   But the assumptions of abstract economic models might well not apply in critically important ways once one faces the specifics of a particular issue.  That is why real world testing is important, as well as examination of the underlying assumptions and abstractions.

In the case of employment of workers at or close to the minimum wage, there are many reasons why the predictions of pure economic theory might well not apply.  Such labor markets are far from the perfect competition ideal.  There is a different balance of bargaining and other power between the employer and the potential minimum wage employee; the information available to each side of the transaction (on how productive the worker might be, and what his or her alternatives are) will differ; workers once they are in a job gain a good amount of job specific knowledge and abilities (not only on how to do the specific job, but also how best to work in a team with the specific colleagues there, where things are kept, and a million other details); that due to such job specific knowledge and imperfect information on who else might be available in the market, there will be substantial transactions costs incurred when an employee quits a job or is fired and a new one must be hired; and more.  Standard theory predicts that workers (in a perfect market) will be paid the value of their marginal product, but it can be difficult even to know what the marginal product of a specific worker might be when they work, as they typically do, in a team with others.  And one finds in the real world that wages are typically paid according to seniority and according to some hierarchy, rather than according to some strictly measured marginal productivity.

Good economists are therefore not surprised that markets in the real world can act quite differently from how the simple models might predict.  And they therefore accept as quite possible the finding of the real world empirical studies that increases in the minimum wage, such as those observed in the past, might well not have led to jumps in unemployment of workers earning the minimum wage.

E.  Conclusion

Full time workers in minimum wage jobs are poor, despite their evident willingness to work.  Even if the minimum wage is raised to $9.00 an hour from the current $7.25 an hour, as Obama has proposed, these working poor will still be earning well less than poverty line income.  And bringing the minimum wage to $9.00 an hour will only bring it back to where it was more than a half century ago.  Real GDP per capita has more than doubled over this period.  Yet minimum wage workers are currently earning 20% less.

Rigorous empirical studies do not show that increasing the minimum wage by an amount such as this will lead to an increase in unemployment of such workers.  Nor does one see such an increase in unemployment in a more casual examination of the evidence, such as in the graphs above.  While the poor need more assistance than just from this, increasing the minimum wage as Obama has proposed would certainly be an important help.

The Job Record in Obama’s First Term: Private Jobs Grew, and Government Jobs Were Cut

Posted on by

Cumul Private Job Growth from Inauguration, to Jan 2013

Cumul Govt Job Growth from Inauguration, to Jan 2013

With the recent release by the Bureau of Labor Statistics of the January job numbers, we can now look at the job record of Obama over his full first term, and compare it to that in the first term of Bush or others.  These new BLS numbers also reflect the impact of the re-benchmarking revisions (done each year at this time), which we noted in a post on this blog in October would likely show a substantial upward revision in the private job estimates in 2012, along with a substantial downward revision in the government job estimates.  The graphs above reflect these new numbers, and show cumulative job growth, private and government, over the full first terms of Obama and Bush.

Mitt Romney and his fellow Republicans repeatedly charged in the recent campaign that private job creation plummeted under Obama, while he boosted government spending and jobs for bureaucrats.  The exact opposite happened.  Private jobs were indeed plummeting when Obama took the oath of office in January 2009, as he inherited the economic crisis that had begun in the last year of Bush.  But through the stimulus package and other measures (including in particular aggressive action by the Fed), he was able to turn this around quickly.  The economy started to grow again six months after he took office, and private jobs began to grow a year after he took office (see the top graph above).  Private job growth has continued at a fairly steady rate since, and by the time Obama took the oath of office for his second term, there were over 1.9 million more workers employed in private sector jobs than when he took the oath of office for his first term.  More significantly, there were 6.1 million more private sector jobs when Obama ended his first term term than there were at the trough a year after he took office.  And as the graph above shows, the pace of new private job creation has not slowed since that trough three years ago.

In contrast to the Obama record, private jobs fell during the first term of George W. Bush.  There were 950,000 fewer workers employed in private jobs when Bush started his second term than when he started his first.  They were also not plummeting when he first took office, as they had been under Obama, but only started to fall a few months later.  They then continued to fall for the first two and a half years of his term before finally starting to rise.  And when they finally started to rise, they grew at a slower pace (102,000 per month) for the last year and a half of Bush’s first term, than they did (at a pace of 170,000 per month) over the final three years of Obama’s first term.

Yet Republicans continue to argue that the policies under Bush, of tax cuts and lax or no proper regulation, are necessary to support the “job creators” and lead them to create private sector jobs.  The record shows that the approach followed under Obama was far more successful.

Government jobs followed a very different pattern.  Government jobs (at all levels of government, including state and local) grew by 900,000 over the four years of Bush’s first term, but they fell by 720,000 over the four years of Obama’s first term.  This is a net difference of 1.62 million jobs.  (The sharp peak in quarter 16 was due to hiring to fill temporary jobs for the decennial census.  Government jobs soon returned to their previous declining path as these census jobs ended.)

With a current labor force in the US of 156 million, the simple direct impact, had one allowed government jobs to have grown during Obama’s term as they had during Bush’s first term (the net difference of 1.62 million jobs), would have been to reduce the unemployment rate by 1.0%.  That is, the direct impact would have been to reduce the unemployment rate to 6.9% from the current 7.9%.

But there would also have been indirect impacts, as the newly employed government workers would have purchased goods and services with their new income, which would have in turn employed workers to produce those goods and services.  With a conservative estimate of this multiplier at two, unemployment would now be at 5.9%, which is within the range of 5 to 6% unemployment which is generally considered to be full employment (unemployment will never be zero).

There would of course also be a budgetary cost to employing more government workers.  But it is not that much.  Using BLS data on the average total compensation costs (including benefits) for government workers, employing an additional 1.62 million public sector workers would cost $140 billion per year.  While significant, this is only 2.5% of the $5.7 trillion that government spends each year (at all government levels) in the US currently.  Furthermore, the net impact on the budget will be a good deal less as there will be increased tax revenues generated as more people are employed (both directly and indirectly).

The still high unemployment in the US can therefore be accounted for by the decline in government employment during Obama’s first term.  Had government jobs been allowed to grow as they had under Bush, we would now be at, or at least close to, full employment.  Furthermore, while the calculations here use the growth of government employment during Bush’s first term as the benchmark, that growth of 900,000 government workers under Bush was not out of the ordinary.  Government employment grew by a bit less during Clinton’s first term (by 690,000), but by more during the term of Bush’s father (by 1,240,000).  Government employment also grew by 850,000 during Bush’s second term.

One would expect government to grow in an economy that is growing with a population that is growing.  The growth in government employment during Bush’s first (and second) terms was not unusual nor was it inappropriate.  Rather, what was unprecedented was the sharp fall during Obama’s first term.  Never before in US history (at least as far back as 1939, when the BLS statistics start) has government employment fallen by so much during a presidential term.  The only instance that can rival it is the fall after World War II during the 1945-49 term, when government employment fell by half as much as it had under Obama (by 360,000 then, vs. by 720,000 under Obama).

The sharp cut-back in government jobs under Obama is therefore historic.  It can account for the still high rate of unemployment.  It would not cost that much to hire back the school teachers, health care workers, policemen and firemen that have lost their jobs or have not been able to get such jobs.  Yet despite such historic cuts, Obama is still seen by conservatives as a socialist presiding over a government exploding in size.

Employment Growth: Still Sluggish

Posted on by

Employment, Monthly Change, Dec 2005 - Jan 2013

The Bureau of Labor Statistics released its regular monthly report on employment this morning.  The report indicates that the trends seen since late 2010 have continued, with steady but modest growth in private employment, falling government employment, and a resulting growth in total employment that brings down the unemployment rate only slowly and with month to month fluctuations.  Total non-farm employment (from the Business Establishment Survey) rose by an estimated 157,000 in January, with private non-farm employment rising by 166,000 while government employment was cut by 9,000.

The figures in this report also reflect the impact of the regular annual re-benchmarking of the employment statistics.   As was discussed in an earlier posting on this blog, this re-benchmarking was expected to raise the estimates for employment in 2012, and it did.  But with revisions in the seasonally adjusted data going all the way back to 2008 (as is normal), the changes in 2012 itself were relatively modest as the base rose.  Under the earlier estimates, total non-farm employment growth in 2012 was estimated to have averaged 153,000 per month.  Under the revised figures, employment is estimated to have grown at about 181,000 per month, for an increase of 28,000 per month.  While an improvement, this still leaves monthly employment growth at less than the 200 to 250,000 per month that I had earlier noted in this blog would be needed to produce a steady and reasonably strong reduction in the unemployment rate.  But it is at least closer to what is needed.

And indeed, the unemployment rate from the separate Household Survey went up one notch to an estimated 7.9%, from 7.8% in December and November as well as September (it was an estimated 7.9% in October).  While such a small increase in the January rate is not statistically significant in itself (the numbers are all based on surveys), one can certainly say that the unemployment rate has not improved since September.

Like the Establishment Survey, the Household Survey that is used to calculate the unemployment rate (as well as labor participation and other figures) was revised to reflect new benchmarks and weights.  But unlike the Establishment Survey numbers (used for the employment estimates), the Bureau of Labor Statistics does not then go back and revise the previous estimates stemming from the Household Survey.  I do not know why.  One cannot then directly compare the January figures to those of December and earlier, to see what the changes were.

The BLS does, however, provide in a note the impact of the re-weighting on a few of the main estimates, including on the total number in the labor force, on those employed, and on those unemployed.  Taking this impact into account, the figures indicate that the number in the labor force was about unchanged in the month, while those employed (based on the Household Survey) fell by 110,000, and those unemployed rose by 117,000.  This is not good, and suggests that the unemployment rate rose (slightly) not because workers re-entered the labor force, but because workers lost jobs.  But too much should not be read into one month’s figures, particularly when the figures suggested by the Household Survey (that employment declined) contradict the figures indicated in the Establishment Survey (that employment rose).  Such contradictory moves in the month to month changes are not unusual between these two surveys, and the figures in the Establishment Survey are generally seen as more reliable for the total employment estimates.

Private employment growth is therefore continuing, and is indeed continuing at a rate that is somewhat higher than had been estimated before.  But total employment growth is being pulled down by continued cuts in government employment, and the overall rate of employment growth is not at what is needed to produce a steady and consistent decline in the rate of unemployment.  Perhaps the best that can be said is that the estimated fall in GDP in the fourth quarter of 2012 (discussed in a post on this blog yesterday) does not appear to have led to a sharp reduction in employment growth — at least not yet.