October 18, 2019 5:46PM

Costs of Higher Spending Would Fall on Middle Class

A Wall Street Journal editorial was right to zero in on Elizabeth’s Warren's “determined refusal to say if her plans would require taxes to increase on the middle class.” In the Democratic presidential debate the other night, the senator repeatedly ducked concerns that the costs of her huge spending plans would land on more than just the wealthy.

They surely would. The larger welfare states of Europe and some other high-income economies are not fueled by higher income taxes than ours, but by higher taxes on consumption and worker wages.

There are 36 high-income countries in the OECD. Thirty of them have higher overall taxes than we do as a percent of GDP. How do those higher-tax countries raise revenues? The chart below shows that they raise about the same amount as us in income taxes, but much more in Social Security (wage) taxes and goods and services (consumption) taxes. Those sorts of taxes hit lower- and middle-income households hard.

Warren and other Democrats say that they will pay for expanded programs with higher taxes on the rich. But that is unlikely. If their spending plans were enacted, the costs would ultimately be paid by higher taxes on all of us.


The five OECD countries with lower taxes than the United States are Chile, Ireland, Korea, Mexico, and Turkey. OECD tax data is here. The data includes all levels of government.

October 18, 2019 2:08PM

Do Illegal Immigrants Increase Drunk Driving Deaths?

The danger posed by illegal immigrant drunk drivers is a major point of contention in the debate over immigration policy. Significant media attention and the occasional tragedy contribute to the notion that illegal immigrants commit drunk driving (DUI) offenses at an alarming rate. In a recent interview, former Immigration and Customs Enforcement (ICE) director Tom Homan went so far as to label DUI offenses committed by illegal immigrants as a “public safety threat.” Indeed, law enforcement officials and immigration authorities alike continue to state that illegal immigrants are significant DUI offenders, despite research that finds no effect of illegal immigration on DUI related deaths or criminal activity in general.

In 2017, nearly 11,000 people were killed in alcohol-impaired-driving accidents, meaning that drunk driving is responsible for almost 29 percent of all traffic deaths. The National Highway Traffic Safety Administration (NHTSA) estimates that alcohol-related deaths in 2017 were highest in Texas, California, and Florida – three states with large illegal immigrant populations as well as large populations in general. If illegal immigrants are more likely to be drunk drivers, places with larger illegal immigrant shares of the population will have more alcohol-related traffic deaths. Below, we examine whether this is the case.


We collected raw data on traffic deaths from the NHTSA Fatality Analysis Reporting System (FARS) database for the 2010-2017 period. The FARS data provide comprehensive records on deaths in motor vehicle crashes, including details on drivers, victims, and details of the crash. To determine whether deaths were the result of alcohol impairment, we use data on drivers’ blood alcohol concentrations (BACs) and whether the police reported alcohol involvement through FARS. All 50 states and the District of Columbia passed laws setting the legal BAC limit at 0.08g/DL, so we adopt this same threshold for our filters. Since the NHTSA uses methods to impute missing BAC values, we were not able to replicate their annual estimates exactly. However, our figures using the two previous filters yield very similar results. For instance, NHTSA estimates a total of 10,874 alcohol-related deaths in 2017 and our filtering method yields 10,850 – a difference of just 0.22 percent. We aggregated these data up to the state-year level.

Our measure of the illegal immigration population and state-level demographic data come from the Pew Research Center and American Community Survey (ACS) 1-year estimates available through IPUMS, respectively. As a robustness check, we calculate state-level estimates for the share of Hispanic noncitizens as a proxy for the likely illegal immigrant population. At a minimum, their numbers are closely correlated with the number of illegal immigrants. That means that the results for regressions where the independent variable is the illegal immigration population as estimated by Pew should be similar to those where the Hispanic noncitizen population is the independent variable.

To test whether larger illegal immigrant population shares correlate with increased deaths in drunk driving accidents, we ran two-way fixed effects regressions that regress the log traffic death rate per 100,000 on the population share of illegal immigrants and Hispanic noncitizens by state. State fixed effects account for level differences across states, such as state drinking culture, while year effects control for national trends affecting all states such as secular changes in road injuries. Standard errors in each regression are clustered by state to account for arbitrary autocorrelation in the error term.


Our results show that illegal immigrant populations are not correlated with alcohol-related traffic deaths (Table 1). For both the overall level of traffic deaths and alcohol-related deaths, we find a negative partial correlation with the state population share of illegal immigrants. Interestingly, we find that places with larger illegal immigrant shares are significantly associated with fewer per capita traffic deaths. However, this result falls out of significance after accounting for state trends. We find no significant evidence for a relationship between illegal immigrant population share and the alcohol-related drunk driving death rate. Considering Hispanic noncitizens as a proxy for likely illegal immigrants yields similar results (Table 2). We continue to find no statistically significant association with an increased risk of death through alcohol-related crashes.

Table 1

Traffic Deaths and Illegal Immigrant Population Shares

Table 2

Traffic Deaths and Hispanic Noncitizen Population Shares

These regression above are also summarized by binned scatterplots in Figures 1-4. Figure 1 shows a statistically significant negative relationship between the illegal immigrant share of the population and the death rate from all traffic accidents by state from 2010-2017. Figure 2 shows no statistically significant relationship between alcohol-related traffic deaths and the illegal immigrant population share by state from 2010-2017. Figure 3 shows a statistically significant negative relationship between the noncitizen Hispanic share of the population and the death rate from all traffic accidents by state from 2010-2017. Figure 4 shows no relationship between the noncitizen Hispanic share of the population and the alcohol-related death rate from all traffic accidents by state from 2010-2017.

Figure 1

All Traffic Death Rate and Illegal Immigrant Population Share by State, 2010-2017


Sources: Authors’ calculations using the 2010-2017 FARS, Pew (2019), and 2010-2017 ACS 1-year estimates from IPUMS.

Figure 2

Alcohol-Related Traffic Death Rate and Illegal Immigrant Population Share by State, 2010-2017


Sources: Authors’ calculations using the 2010-2017 FARS, Pew (2019), and 2010-2017 ACS 1-year estimates from IPUMS.

Figure 3

All Traffic Death Rate and Noncitizen Hispanic Population Share by State, 2010-2017


Sources: Authors’ calculations using the 2010-2017 FARS and 2010-2017 ACS 1-year estimates from IPUMS.

Figure 4

Alcohol-Related Traffic Death Rate and Noncitizen Hispanic Population Share by State, 2010-2017


Sources: Authors’ calculations using the 2010-2017 FARS and 2010-2017 ACS 1-year estimates from IPUMS.


We find no statistical evidence to suggest that places with more illegal immigrants are more at risk for drunk driving deaths. Of course, there are individual instances to the contrary and those illegal immigrants who commit real crimes should be punished like everybody else, but their presence doesn’t seem to affect overall drunk driving deaths. Although our regressions results are correlative and not causal in nature, they suggest that illegal immigrants do not affect overall drunk driving deaths.

October 18, 2019 1:09PM

Senate Democrats’ RELIEF Act Would Double Legal Immigration Over 10 Years

Senators Dick Durbin (D-IL) and Patrick Leahy (D-VT) introduced the RELIEF Act this week. The legislation is their alternative to the Fairness for High Skilled Immigrants Act, which passed the House this year. Both bills would phase out the employment-based per-country limits that prevent any nationality from using more than 7 percent of the green cards granting permanent residence in any given year (unless they otherwise would go unused). The per-country limits create disproportionately long backlogs for Indian nationals. Both bills also double the family-sponsored per-country caps from 7 to 15 percent.

The key difference is that the RELIEF Act would also increase the total number of green cards, while the Fairness for High Skilled Immigrants Act contains no change in the amounts. It is only about “fairness,” treating Indians the same as all other immigrants with respect to their places in line, not dealing with the green card shortages that cause the backlog in the first place. As I’ve detailed before, it would dramatically decrease waits for Indians from decades to 7-8 years, while everyone would end up waiting about that long also rather than not at all. The RELIEF Act would solve both problems, ending both the inequity and the shortage.

The policies that the RELIEF Act changes

  1. Phases out the per-country limits over the course of 5 years for employment-based green cards and increases the per-country limits for family-based immigrants from 7 to 15 percent (pp. 2-7).
  2. Creates a pool of new green cards equal to almost the entire existing backlog of employment- and family-based applicants (pp. 8-9). In my April policy analysis, I estimated the entire backlog was about 4.7 million in 2018. It would award those over the course of 5 years without regard to the country limits.
  3. Uncaps the family-sponsored second preference F-2A category (pp. 9-10) for new spouses and minor children of existing legal permanent residents in the United States (e.g. new marriages that occur after the person receives a green card). This category has a cap of about 88,000 and backlog of about 150,000.
  4. Increases the number of green cards for the family-sponsored first preference category from 23,400 to 111,334, essentially reassigning the 88,000 from the F-2A category (pp. 11-12).
  5. Exempts spouses and minor children of primary applicants (e.g. the workers) from the green card limits (p. 9). In 2017, nearly half the green cards in the capped categories went to these “derivative” applicants, which means that the number of green cards in these categories would almost double as a result of this change.
  6. Ends “aging out” (pp. 12-13). This would maintain the eligibility of a minor child of a primary applicant (i.e. the worker) until the application was processed rather than removing the child from the line at their 21st birthday. It also maintains their eligibility for nonimmigrant status as the children of temporary workers. This prevents children who grew up in the United States as children of temporary workers from being forced to leave if they turn 21 before their parents receive green cards. Because the bill also exempts minor children from the overall caps, this change would further increase the number of immigrants receiving green cards

The effect of the RELIEF Act on legal immigration

The big picture is that over the next decade, the total number of legal immigrants receiving permanent residence would virtually double. From 2020 to 2024, the legal immigration rate (new immigrants as a percentage of the population) would increase to 0.75 percent, which would be the highest level since 1921, just before Congress slammed the gates on legal immigration. The effect is likely greater than the estimates here because an increase in the number of green cards will eventually result in more U.S. citizens who can sponsor spouses, minor children, and parents without a numerical limit, which will increase immigration even further.

Table 1 breaks down the changes by category. The total number of family-sponsored immigrants would increase by at least 375,000 and probably more once those family members receive citizenship and can sponsor more. The employment-based categories would also nearly double from 140,000 to 235,000. The diversity lottery program would also double. The primary driver of these changes is the exemption of spouses and minor children of new legal immigrants who are the primary applicants. They constitute half or more of the total immigrants in each category. In some cases, they are nearly two-thirds. For example, the EB-5 investors category would nearly triple with the removal of investors’ children from the cap.

The new numbers would likely stabilize the categories, meaning that wait times wouldn't continuously grow worse. On the employment-based side, wait times would likely be less than a year for everyone. The RELIEF Act lacks some important provisions contained in Senator Rand Paul’s BELIEVE Act (reviewed here). Among other things, the RELIEF does not provide green cards to children of H-1B workers who have already aged out, as Paul’s bill would, and Paul’s bill quadruples the employment-based figures, rather than merely doubling them. Nonetheless, by fixing both family- and employment-based immigration, the RELIEF Act probably contains the best legal immigration reforms overall since the comprehensive immigration reform bill (S. 744) that passed the Senate in June 2013 because it makes legal immigration viable across so many categories. 

Sen. Durbin appears to have introduced it because he has blocked the Fairness for High Skilled Immigrants Act in the Senate and needed to explain what he would do instead. Unfortunately, while his policy undoubtedly makes more sense, it politically has little chance of passage. Republicans already blocked Sen. Durbin’s attempt to have the bill receive floor consideration, and it is not likely that it will receive the bipartisan commitment needed to pass the Senate. President Trump would almost certainly veto it.

Meanwhile, the Fairness for High Skilled Immigrants Act, while it doesn't fix all the problems, could become law now. It's unclear whether the senators are trying to work out a compromise, but a deal seems unlikely given the distance between Sen. Durbin and Sen. Mike Lee (R-UT) who is the lead cosponsor of the Fairness for High Skilled Immigrants Act. Sen. Lee has already rebuffed his attempt as well as others' attempts to amend his bill to contain more green cards, making the prospect of a compromise here remote. 

October 18, 2019 11:53AM

Larry Summers Wisdom On Wealth Inequality

Chris Edwards argued earlier in the week that wealth inequality statistics alone tell us little to nothing interesting about the American economy. It seems Larry Summers agrees (see from 8h 27 mins).

In a speech at PIIE yesterday, the former Treasury Secretary under Bill Clinton outlined why wealth inequality wasn’t a particularly useful measure to consider the justness of a society.

He highlighted, for example, that the arguments about wealth inequality and political power appear to have almost no validity. Interest groups and corporate lobbying are much more important sources of political power than wealthy individuals. In the panel discussion afterward, he pressed economist Emmanuel Saez to give just one example or mechanism of how an extraordinarily rich individual losing a large fraction of their wealth would alter political outcomes. Little meaningful response was forthcoming.

I’ve looked in detail at the supposed relationship between wealth inequality and democracy for a forthcoming Cato paper with Chris Edwards, and the evidence for wild assertions we hear about inequality killing democracy is extraordinarily weak. But more on that next week. What was heartening was to hear Summers deliver some home truths to the left on wealth inequality and the redistributive welfare state:

Wealth inequality reflects many things that happen in a society. Suppose we successfully in the United States adopted a more generous and complete progressive social security system….I would assume that the lower half of the population would have much less need to accumulate or hold liquid assets because they were being properly insured. And so measuring the ratio of the wealth of the wealthy to the wealth of the less wealthy may reflect something about accumulation at the top or it may reflect something about the adequacy or inadequacy of social insurance arrangements.

This is exactly right, and supports what Chris and I have outlined about the disingenuousness of those who say wealth inequality shows the need for a more progressive welfare state.

Evidence from both here and abroad shows major social programs, not least Social Security, increase measured wealth inequality because they leave the non-rich with “proportionately less to save, less reason to save, and a larger share of their old-age resources in a nonbequeathable form than the lifetime rich.” Economists Baris Kaymak and Markus Poschke estimate that the expansion of Social Security and Medicare caused about one-quarter of the rise in the top one percent wealth share over recent decades.

As Chris writes today, the European experience shows the same results. It therefore makes no sense to say you believe reducing wealth inequality is an overwhelming imperative, and that expanding the welfare state is the way to do it.

Those who worry about measures of private wealth inequality and want a bigger welfare state therefore face two ways of squaring the circle. Either they can admit that reducing measured private wealth inequality is not an important and overwhelming public policy objective. Or they can seek to incorporate social programs into broader, “lived” measures of wealth inequality.

Both would be problematic for them, politically. If goals other than reducing wealth inequality drive their desire for more social spending, it’s much more difficult to argue that concerns about other priorities are illegitimate when discussing top taxes (for example, economic growth). And if you include assessments of the present value of promised government benefits as if they are real wealth, wealth inequality looks much lower and less problematic.

What’s dishonest though is to argue that private wealth inequality measures show we need an all-out war on narrowing the wealth distribution and that more social spending is the cure. Kudos to Larry Summers for pointing that out.

October 18, 2019 11:49AM

New York: Cold Showers, Black Eyes, and Public Employee Tenure

New York State has agreed to pay $6 million to settle claims that disabled residents of a Bronx group home for developmentally disabled adults were physically abused and neglected by staff, and the state has also spent a further $5.7 million thus far defending the staff members in court, as Benjamin Weiser reports in the New York Times. Among the allegations were that staff members hit and kicked residents, gave them cold showers, and "left some with black eyes and other bruises." However, don’t assume that any public employees lost their jobs:

A state investigation later substantiated allegations of misconduct by 13 workers.

But the state failed to fire any of the employees, The New York Times reported in June.

A state arbitration process shielded the workers who had been cited for abuse and neglect. They were typically sent to other jobs in the system.

As part of the settlement, lawyers representing families insisted that the group home be removed from the control of the New York state government.

Anti-privatization campaigns, often backed by public employee unions, inveigh against outside contracting for social services as a menace to accountability. But it is worth remembering that direct state provision of services can be among the least accountable alternatives of all. [adapted from Overlawyered]

October 18, 2019 10:28AM

Welfare State Causes Wealth Inequality — Euro Experience

Democrats running for president are condemning wealth inequality while calling for an increase in social spending. But expanding social spending would magnify wealth inequality, not reduce it, because it would displace private wealth accumulation by lower- and middle-income households.

Evidence comes from a study by Pirmin Fessler and Martin Schurz for the European Central Bank. The authors explore the relationship between government social spending and wealth distribution in 13 European countries using a survey database of 62,000 households. The database contains household balance sheet information.

Regression analyses by the authors confirm that “the degree of welfare state spending across countries is negatively correlated with household net wealth. These findings suggest that social services provided by the state are substitutes for private wealth accumulation and partly explain observed differences in levels of household net wealth across European countries.”

The authors found that the “measured inequality of wealth is higher in countries with a relatively more developed welfare state.” Why is this the case?

The substitution effect of welfare state expenditures with regard to private wealth holdings is significant along the full net wealth distribution, but is relatively lower at higher levels of net wealth. Given an increase in welfare state expenditure, the percentage decrease in net wealth of poorer households is relatively stronger than for households in the upper part of the wealth distribution. This finding implies that given an increase of welfare state expenditure, wealth inequality measured by standard relative inequality measures, such as the Gini coefficient, will increase.

Fessler and Schurz found, for example, that Austria, France, Germany, and the Netherlands have high social spending and low private wealth holdings by less well-off households. But other countries such as Luxembourg and Spain have lower social spending and higher private wealth holdings by less well-off households.

The relationship can be seen in this figure, which is their plot of social spending compared to the wealth of households at the 25th percentile (from the bottom) of each nation’s wealth distribution.


The authors note that their results are in line with the displacement, or crowding out, effects found in other statistical studies, such as economist Martin Feldstein’s work showing that Social Security substantially displaces private saving for retirement in the United States.

Data from Credit Suisse confirm that some nations with large welfare states have high wealth inequality. The bank points to Denmark, Norway, and Sweden:

strong social security programs—good public pensions, free higher education or generous student loans, unemployment and health insurance—can greatly reduce the need for personal financial assets, as Domeij and Klein (2002) found for public pensions in Sweden. Public housing programs can do the same for real assets. This is one explanation for the high level of wealth inequality we identify in Denmark, Norway and Sweden: the top groups continue to accumulate for business and investment purposes, while the middle and lower classes have a less pressing need for personal saving than in many other countries.

The bottom line for America is that expanding programs such as Social Security and Medicare will increase wealth inequality—the opposite effect Warren and Sanders may hope for. A better approach would be to cut the size of government and transition the nation to a leaner array of social programs based on personal savings accounts.

October 17, 2019 2:43PM

Mark Zuckerberg at Georgetown

Mark Zuckerberg gave an important speech today at Georgetown University about free expression and Facebook. I think the speech is important for four reasons.

Recently Zuckerberg wrote that free expression was a “paramount” value for Facebook. Paramount means “more important than anything else; supreme.” This speech articulates his understanding of “paramount” for Facebook. Free speech is immensely valuable and just not another value to be traded off against many other values. The limits to speech on Facebook are themselves limited.

Second, Facebook is about to launch an independent Oversight Board to consider appeals to its content moderation. The main job of the Board is to independently enforce Facebook’s values. Inevitably the Board will interpret and apply Facebook’s Community Standards. But difficult cases will arise. The Board will need some insight into Facebook’s values to properly interpret the standards. Here we have the Founder and CEO of Facebook articulating those values especially the status of free expression online. This speech will supplement the Community Standards in guiding the work of the Board. And the speech says: free expression is our paramount value.

Third, Zuckerberg goes beyond Facebook and pushes back against the current criticisms of free speech. He notes “we’re seeing people try to define more speech as dangerous because it may lead to political outcomes they see as unacceptable.” Zuckerberg counters that this view “is more dangerous to democracy over the long term than almost any speech.” And several times he says, “we must continue to stand for free expression.” Just words you say? Remember also that this part of the speech engaging the critics of free speech was optional for Zuckerberg. He could have just talked about his company and its obligations. He made the choice to reply to those critics and to endorse the fundamentals of freedom of expression.

Fourth, Zuckerberg is obviously a tech business leader. Content moderation takes place throughout social media and related platforms. We can hope this endorsement of free speech online will stiffen some spines during what has been and may be a difficult time.

The speech was not perfect. He provided no final answers on hate speech. That particular issue merits our attention and “more speech.” It is not an easy question for Facebook or any private organization.

By all rights, Zuckerberg’s speech should have been very different. Facebook and social media are under political attack, said to be a threat to democracy. And many of the critics do indeed assume that our traditional commitment to free speech should no longer hold in the new digital age. The easiest way forward would have been placating those critics. Zuckerberg has chosen a harder path here, and I can only conclude he has done so because he actually believes in free expression, not just as one value among many but as the most important one for his business and for the nation.

It’s good to have this statement from one of our leading businessmen. The next part is harder: Zuckerberg’s recognition of the paramount status of speech must be realized by his company and his new Oversight Board. But this speech marks a good start down that challenging path.