Macro Roundup

I work for the county out on ninety five, All day I hold a red flag and watch the traffic pass me by, "...One of the few things that Democrats and Republicans agree on is that spending billions of dollars on America’s roads would boost productivity and the U.S. economy’s growth prospects. Economists aren’t so sure. A wide body of research focused on the effects of highway spending suggests that major new investment in U.S. roads would generate little, if any, long-term economic gain....the U.S. already has an extensive system of roads, so building more wouldn’t add much to productivity…That’s not to say that billions of dollars in new government road spending wouldn’t boost growth in the short term. But the gains would come about as the result of the construction, and would dissipate once all the projects are completed. In a 2012 paper, San Francisco Fed economists Sylvain Leduc and Daniel Wilson found that new spending on roads can boost an area’s economy at two specific times: immediately after the new spending has been announced, and six to eight years later, when construction is under way. Beyond 10 years, there were no economic benefits to infrastructure spending, they found. Moreover, the immediate effect applies only during recessions, they wrote. It’s unclear whether the U.S. would see that short-term boost now that the economy is expanding rapidly...."

David Harrison, "Why More Highway Spending Won’t Rev Up the Economy,"Wall Street Journal, July 4, 2021, https://www.wsj.com/articles/why-more-highway-spending-wont-rev-up-the-economy-11625403600

Why More Highway Spending Won’t Rev Up the Economy

One of the few things that Democrats and Republicans agree on is that spending billions of dollars on America’s roads would boost productivity and the U.S. economy’s growth prospects.

Economists aren’t so sure. A wide body of research focused on the effects of highway spending suggests that major new investment in U.S. roads would generate little, if any, long-term economic gain.

While the projects would spur hiring and spending temporarily, both when they are announced and under way, they aren’t likely to raise the economy’s productivity and, in turn, its overall growth potential in a lasting way, many researchers find.

That is because the U.S. already has an extensive system of roads, so building more wouldn’t add much to productivity, economists say.

New spending for roads accounts for the largest single share—roughly 19%—of the $579 billion in new spending that the White House and a group of lawmakers have agreed to. Both Democrats and Republicans say that money would raise the economy’s productivity, defined as the level of output per hour worked.

President Biden last week touted the agreement as delivering “higher productivity and higher growth for our economy over the long run.”

Sen. Rob Portman, an Ohio Republican who helped craft the deal, said last month that the plan would “increase our productivity as a country.”

Development of the U.S. interstate highway system between the 1950s and 1970s—currently 47,000 miles of multilane highways stretching coast to coast—did make the economy much more productive, John Fernald, an economist at the Federal Reserve Bank of San Francisco, wrote in a 1999 paper.

The system meant a cross-country trip that used to take months could be accomplished in days. Businesses gained access to new suppliers and new customers. Cities were able to specialize in certain industries. International trade opened up. By one estimate, the U.S. economy would be 3.9% smaller today without the interstate highway system.

But those gains all came about when the highways were built. By now, the gains have been reaped.

“Building the interstate highway system was enormously productive,” Mr. Fernald said. “That does not imply that building a second one would be equally productive.”

Other research has reached similar conclusions.

A group of economists in Spain studying that country’s infrastructure spending between 1964 and 1991 concluded that the investment earlier in the period produced greater economic gains than investments later, when much of the infrastructure was already in place.

Researchers have also found that in developed countries, whatever local benefits come from highway improvements come at the expense of other locations. In other words, road spending reallocates the pie but doesn’t make it bigger.

Mr. Duranton and two co-authors, Geetika Nagpal and Matthew Turner, both of Brown University, suggested in a paper last year that new investments “lead to a displacement of economic activity while net growth effects are limited.”

That’s not to say that billions of dollars in new government road spending wouldn’t boost growth in the short term. But the gains would come about as the result of the construction, and would dissipate once all the projects are completed.

In a 2012 paper, San Francisco Fed economists Sylvain Leduc and Daniel Wilson found that new spending on roads can boost an area’s economy at two specific times: immediately after the new spending has been announced, and six to eight years later, when construction is under way. Beyond 10 years, there were no economic benefits to infrastructure spending, they found.

Moreover, the immediate effect applies only during recessions, they wrote. It’s unclear whether the U.S. would see that short-term boost now that the economy is expanding rapidly.

Some of the spending lawmakers are considering could ease congestion. But those improvements would also be temporary. Adding more highway lanes to ease congestion tends to encourage more people to use those lanes, making them congested once more, a phenomenon known as “induced demand.”

A 2011 paper by Mr. Duranton and Mr. Turner found that areas that added road miles saw a proportional increase in driving, resulting in the same overall traffic levels.

Even if long-term benefits are limited, there is still a case to be made for spending money on roads, economists say. Filling potholes could provide a more comfortable driving experience, for instance.

“The more comfortable ride is getting you the improved quality of life but it’s not necessarily adding tons of private sector productivity,” Mr. Fernald said.

New note from Michael Mauboussin highlight the impact of intangible assets, intangible intensive firms grow faster than their tangible counterparts but also face a greater risk of being overtaken by their competitors. Implying a higher standard deviation in the distribution of growth rates for intangible-based versus tangible-based industries.

Implication for base rates is the past distributions are not prologue given the impact of intangibles.

"...suggests two hypotheses that we can test. The first is that intangible-based businesses can grow faster than what the base rate data show...the right tail of the distribution of growth rates is extending outward from the average. Amazon’s results provide anecdotal evidence for this. The second is that we should observe greater variance in the distribution of growth rates for intangible-based businesses. That means that the left tail of the distribution of growth rates is also spreading further from the average. BlackBerry’s 26.7 percent average annual revenue decline in the past decade through February 2021 is a case in point....Companies grow by generating a return on investment. The nature of investment has changed markedly in recent decades, from one dominated by tangible assets to one mostly in the form of intangible assets. Intangible assets have some characteristics that distinguish them from tangible assets, including greater potential economies of scale and higher risk of obsolescence. The good news is that intangible-intensive companies can grow faster than their tangible counterparts. The bad news is they can also become irrelevant and shrink fast. As a consequence, we should see two effects in the data: higher growth and more dispersion in the outcomes. Our analysis of the results from companies in the Russell 3000 from 1984-2020 reveals both of these. The base rate of sales growth is getting stretched from the average in both the positive and negative direction. There are two main lessons for investors. First, it is important to be mindful of the potential shift in the base rate as the result of the rise of intangibles. Second, skillful investors may be able to identify the companies that will grow faster than expected, hence providing the potential for attractive returns...."

Evidence: “.. We calculated the median sales growth rate for companies in each of those categories using the constituents of the Russell 3000 from 1984-2020. We also examined the standard deviation, a measure of the dispersion, of the distributions. Exhibit 1 showsthe results for the full sample.

Michael Mauboussin and Dan Callahan, "The Impact of Intangibles on Base Rates," Morgan Stanley, June 23, 2021,
https://www.morganstanley.com/im/publication/insights/articles/article_theimpactofintangiblesonbaserates.pdf

The sales growth rates, measured either as the median or average, consistently go from highest for companies that are most intangible-asset intentive to lowest for those that are least intensive. While the short-term numbers are noisy, the relationship holds true over 1-, 3-, 5-, and 10-year periods. Consider the median compound annual sales growth rates over five-year periods. Growth was 10.4 percent for healthcare, 7.9 percent for technology, 6.0 percent for consumer, and 5.0 percent for manufacturing. The median across all companies was 6.5 percent. This supports the first hypothesis. The standard deviation of the growth rates follows the same pattern. Where intangible asset intensity is high, the standard deviation is also high. Assuming these distributions are normally distributed, an imperfect but illustrative assumption, about two-thirds of healthcare companies had 5-year sales growth rates between -18.0 and 43.2 percent. The comparable figures for manufacturing companies were -7.0 and 19.2 percent. This is consistent with the second thesis....One silver lining was the ability of digital companies, built largely on intangible assets, to thrive in the chaos. We examined the sales growth rates of the companies in the Russell 1000, the largest one thousand companies in the U.S., to see which companies fared well. Healthcare and technology, the industries with the highest intangible asset intensity, represented over 60 percent of the top 20, 50, and 100 growers despite being only 29 percent of the universe (see exhibit 2)...."

John Cogan and Daniel Heil report on their Research on the impact of Biden's planned expansion of the welfare state, "...the American Families Plan would add 21 million Americans to the list of federal entitlement beneficiaries. With these additional recipients,57% of all married-couple children would receive federal entitlement benefits, and more than 80% of single-parent households would be on the entitlement rolls. The share of households receiving assistance would be higher in some areas of the U.S. than in others. This is primarily because federal eligibility for many of the American Families Plan’s programs, particularly its refundable tax credits, don’t account for geographical differences in incomes and living costs. We estimate that most of the Biden plan’s entitlement benefits would go to middle- and upper-income households. Households in the upper half of the nonelderly income distribution would receive 40% of the new entitlement benefits...."

John F. Cogan and Daniel L. Heil,"Biden’s Plan for an Entitlement Society,"Wall Street Journal, June 28, 2021, https://www.wsj.com/articles/bidens-plan-for-an-entitlement-society-11624918494

Biden’s Plan for an Entitlement Society

The federal government’s system of entitlements is the largest money-shuffling machine in human history, and President Biden intends to make it a lot bigger. His American Families Plan—which he recently attempted to tie to a bipartisan infrastructure deal—proposes to extend the reach of federal entitlements to 21 million additional Americans, the largest expansion since Lyndon B. Johnson’s Great Society.

For the first time in U.S. history—except possibly for the pandemic years 2020 and 2021, for which we don’t yet have data—more than half of working-age households would be on the entitlement rolls if the plan were enacted in its current form. Contrary to Mr. Biden’s assertion that his plan “doesn’t add a single penny to our deficits,” his plan would add more than $1 trillion to the national debt over the next decade.

The American Families Plan proposes several new entitlement programs. One promises students the government will pick up the entire cost of community-college tuition; another promises families earning 1.5 times their state’s median income that Washington will cover all daycare expenses above 7% of family income for children under 5; still another promises workers up to 12 weeks of federally financed wage subsidies to take time off to care for newborns or sick family members.

The American Families Plan would follow longstanding government practice and make temporary emergency programs permanent. In March, Congress enacted the American Rescue Plan, which expanded Affordable Care Act subsidies and refundable tax credits for child care and low-wage workers. The expansions were sold as temporary measures to combat the effects of pandemic lockdowns. A month later, Mr. Biden asked Congress to make them permanent.

These programs extend eligibility for benefits high up the income ladder. Two-parent households with two preschool-age children and incomes up to $130,000 would qualify for federal cash assistance for daycare. Single parents with two preschoolers and incomes up to $113,000 would qualify. And some families with incomes over $200,000 would be eligible for health-insurance subsidies. Other parts of the plan, such as paid leave and free community college, have no income limits at all.

Our analysis shows that the American Families Plan would add 21 million Americans to the list of federal entitlement beneficiaries. With these additional recipients, 57% of all married-couple children would receive federal entitlement benefits, and more than 80% of single-parent households would be on the entitlement rolls.

The share of households receiving assistance would be higher in some areas of the U.S. than in others. This is primarily because federal eligibility for many of the American Families Plan’s programs, particularly its refundable tax credits, don’t account for geographical differences in incomes and living costs.

We estimate that most of the Biden plan’s entitlement benefits would go to middle- and upper-income households. Households in the upper half of the nonelderly income distribution would receive 40% of the new entitlement benefits.

Our estimates are for a full-employment economy, not one in recession. So the percentage of U.S. households receiving benefits from at least one federal entitlement program would only increase if the U.S. economy were to falter.

Where will the money come from to finance this largess? Mr. Biden claims that taxes on the rich will entirely finance his American Families Plan. But his proposed revenue heist falls woefully short of the plan’s true cost. Presidential budgets for years have been littered with gimmicks to hide their true expense. The American Families Plan is no exception.

The plan proposes that the $100 billion annual expansion of the child tax credit will suddenly expire at the end of 2025, reducing the tax credit from a high of $3,600 to $1,000. All other programs in the plan are assumed to be permanent. Why only phase out the child tax credit? The obvious answer: Its expiration reduces the 10-year estimated cost by $465 billion.

The gimmicks don’t stop there. The Biden administration proposes to use more than $200 billion in new business taxes to finance the American Families Plan. Amazingly, it also proposes to use that same money to finance future Medicare spending.

Properly accounting for these gimmicks, and the plan’s overly optimistic revenue assumptions from its Internal Revenue Service compliance initiatives, pushes the American Families Plan deficit to more than $1 trillion during the next 10 years. The president claims that his plan is part of a budget that is “putting the nation on a fiscally responsible path.” Hardly. If passed, it would accelerate the pace of entitlement expansions that began in the late 1960s. Improving the safety net is one thing, but spending more than $1 trillion on mainly middle-class entitlements and financing this expenditure with debt robs future generations while enriching today’s.

John Cogan analysis of the American Families Plan, "...Overall, the President’s plan would add 6.2 million households, containing 21.3 million persons, to the federal entitlement benefit rolls. The increase would lift the percentage of all non-elderly U.S. households that receive benefits from at least one federal entitlement to over 50 percent. This would be the first time in U.S. history that a majority of working age households are federal entitlement recipients, except perhaps for the pandemic years 2020 and 2021 for which we are yet to have data. Over 40 percent of the plan’s new entitlement benefits would go to households in the top half of the income distribution...."

Recently, VC Returns Have Consistently Outperformed Public Equity Markets: Outperforming S&P 500 Median Firm Returned 1.1x (top right), Outperforming S&P 500 Average Firm 1.4x (top left). Note There Is A Substantial Gap Btw Mean And Median Results3%

Absolute buyout performance, "...Let’s start with buyout funds. The charts below show average and median multiples of invested capital (MOIC) and internal rate of return (IRR) since the early 1990’s. MOIC is a simple measure of cash-in vs cash-out, while IRR is a time-weighted measure of return. Note how there’s only a small gap between average and median buyout manager results..."

Michael Cembalest, “Food Fight: An update on private equity performance vs public equity markets," JP Morgan Asset and Wealth Management, June 2021, https://privatebank.jpmorgan.com/content/dam/jpm-wm-aem/global/cwm/en/insights/eye-on-the-market/private-equity-food-fight-jpmwm.pdf

“…Figure 7 plots the average share of college-educated persons associated with job postings against the year since emergence on a technology-by-technology basis. The figure suggests that for the vast majority of technologies, there is a sharp decline in the skill level required for the positions associated with new technologies over time. Even in cases where demand for positions is sharply accelerating (such as AI and virtual reality), the share of skilled positions subsides over time. These results are consistent with the view that new technologies typically start with high-skill occupations and then involve larger parts of the workforce over time. The figure also shows a few notable exceptions to this general pattern: positions exposed to the Online Streaming, Cloud Computing, Search Engine, and Software Defined Radio technologies show no evidence of a declining average skill level over time (in fact, the trend for Online Streaming appears significantly positive)…. We summarize this information by presenting a binned scatterplot in Figure 8.This depiction shows the relationship across all 29 technologies between time elapsed after the emergence year and the mean share of the postings for college-educated people. It shows, on average, a strong negative linear trend, implying a declining requirement for a college-trained workforce as technologies mature…. Table 6 looks at this relationship formally. The sample consists of annual observations of each technology between 2007 and 2019. Here, we use the alternative measures of the skills required in the job postings associated with a given technology: the dependent variables include the share of the weighted SOC classes that are college educated (as of 2015), the share with graduate degrees, mean wages, and the mean years of schooling. Each regression uses as the key independent variable the years since the emergence date and controls for technology and calendar year fixed effects. The specification again follows Table 4 regarding the criteria for inclusion in the analysis and weighting. Using each measure, there is a strong negative relationship between the maturity of the technology and the reliance on a highly educated workforce. For instance, Columns 1 and 3 show that each additional year since the emergence of the technology is associated with a fall of about 0.96 percentage points in the share of job postings requiring a college education (an annual decline of -1.71%) and a decline of $1,023 in annual wages (measured in 2015 constant dollars) for the job postings associated with the technology. Similarly, the share of job postings in occupations requiring a post-graduate degree declines by a rate of 1.80% per year on average….”

New Nick Bloom looks at how new technologies diffuse though an economy using a series of factoids, "...Using the full text of millions of patents, job postings, and earnings conference calls over the past two decades, we introduce in this paper an approach to understand which new technologies affect businesses and to trace their diffusion across regions, industries, occupations, and firms. We can then map the spread of disruptive technologies in these dimensions, focusing on the hiring associated with each important innovation...."

Five takeaways, "...First,the locations where disruptive technologies are developed are geographically highly concentrated, both within and across technologies. Based on patenting activity ten years prior to each technology’s commercial breakthrough, we show that the typical disruptive technology in our dataemerged from only a handful of urban areas, which housed the majority of early patenting in the technology and the vast majority of its early employment up to the year in which it has its commercial breakthrough. We term these specific urban areas the technology’s “pioneer locations.”Although 23 of the 50 U.S. states host at least one pioneer location according to this definition, their distribution across technologies is also remarkably skewed: a few super-clusters are the birthplace of a surprising number of disruptive technologies in our data. Collectively, locations in California alone host a remarkable 40.2% of our technology-pioneer location pairs. Another supercluster along the Northeast Corridor from Washington to Boston accounts for additional 21.2%.More broadly, we find that the geographic distribution of patenting related to our 29 disruptive technologies is even more skewed than that of patenting in general....Second,despite this highly skewed initial distribution, as technologies mature and the number of new jobs related to them grows, they gradually spread geographically. For example, the coefficient of variation across the 917 core-based statistical areas (CBSAs) in the United States drops by 24% in the first decade after a technology emerges. We see this pattern of “region broadening” in virtually every technology that we examine....Third,while initial hiring is concentrated on high-skilled jobs, over time, the mean required skill levels of the jobs associated with the technologies declines, broadening the types of jobs that adopt a given technology. For example, the average earnings associated with job postings in a given new technology drop by about 15% within the first decade, falling from $70,468 per year to $60,608 per year on average, a drop of about one thousand dollars per year (all figures in 2015 dollars)....This pattern of an increasing share of low-skilled jobs that begin to use and adopt a given technology holds within most (though not all) of our disruptive technologies. Fourth, region and skill broadening interact: Low-skill jobs associated with a given technology spread out across space significantly faster than high-skill jobs. Our estimates suggest low-skilled jobs that use or produce new technologies are almost fully dispersed geographically within 20 years. For example, as technologies like the smart phone, cloud computing, and electric cars mature, the lower-skilled jobs associated with these - salespeople, technicians, repair specialists, etc. - spread across the United States at a fast clip. Fifth, despite region and skill broadening,disruptive technologies appear to yield long-lasting benefits for their pioneer locations, particularly when it comes to high-skill employment. Our estimates imply it takes almost 40 years for high-skilled job postings to fully disperse from their original pioneer locations.Perhaps not surprisingly, these pioneer locations tend to be located around universities and areas with more educated populations. Thus, regions with strong local education, research institutions, and universities appear to benefit from successful disruptive innovation for substantial periods of time..."

Mikhail Golosov, Michael Graber, Magne Mogstad and David Novgorodsky, "How Americans Respond to Idiosyncratic and Exogenous Changes in Household Wealth and Unearned Income," National Bureau Of Economic Research, July 2021, https://www.nber.org/papers/w29000

New NBER on the implied first order impact of a universal basic income on household earnings using evidence from lottery winners, “…we use variation in the timing of lottery wins as an instrument for yearly unearned income. On average, an extra dollar of unearned income in a given period reduces pre-tax labor earnings by about 50 cents, decreases total labor taxes by 10 cents, and increases consumption by 60 cents….UBI of $12,000 a year24 would reduce average household earnings by more than $6,000, and require an earnings surcharge of approximately 27 percent on all households, out of which 2.5 percentage points is due to the behavioral response. …Interestingly, the MPEs and MPCs vary systematically across the pre-win income distribution. For example, individuals in the bottom quartile of the pre-win income distribution use most of the increase in unearned income on consumption, while individuals in the top quartile prioritize reducing labor over increasing consumption…”

Core findings, ".... Our first contribution was to estimate the earnings responses to these windfall gains, finding significant and sizable wealth and income effects. On average, an extra dollar of unearned income in a given period reduces pre-tax labor earnings by about 50 cents, decreases total labor taxes by 10 cents, and increases consumption by 60 cents. These effects are heterogeneous across the income distribution, with households in higher quartiles of the income distribution reducing their earnings by a larger amount. Our second contribution was to develop and apply a rich life-cycle model in which heterogeneous households face non-linear taxes and make earnings choices along both intensive and extensive margins. By mapping this model to our estimated earnings responses, we obtained informative bounds on the impacts of two policy reforms: an introduction of UBI and an increase in top marginal tax rates. Our last contribution was to study how additional wealth and unearned income affect a wide range of behavior, including geographic mobility and neighborhood choice, retirement decisions and labor market exit, family formation and dissolution, entry into entrepreneurship, and job-to-job mobility. In our analyses, we described the key identification and measurement challenges that arise, motivated how lottery winnings as a source of variation in wealth or unearned income can address these challenges, and discussed how our findings relate and contribute to existing work... Our bounds demonstrate that the introduction of a UBI will have a large effect on earnings and tax rates. For example, even if one abstracts from any disincentive effects from higher taxes that are needed to finance this transfer program, each dollar of UBI will reduce total earnings by at least 52 cents. With the relatively modest disincentive effects of a Marshallian earnings elasticity of 0.3, a UBI crowds out earnings essentially one for one. By comparison, income effects have less pronounced implications for the earnings reduction from increasing the top marginal tax rates…”

The Evidence: “Figure 3.3 shows that various measures of labor market outcomes fall significantly for lottery winners (relative to later winners) after they win a lottery.Since our observations are at an annual frequency, it is difficult to interpret estimates in event time 0 since they are affected by the timing of the win within the year. For this reason we focus our discussion on years 1 through 5. The wage earnings of the winner fall on average by $3,572 (approximately 10 percent) in the first year following the lottery win, and continue to decline slightly in subsequent years. Per-adult wage earnings have a similar pattern to winner wage earnings, but they decline by slightly less: $3,234 (approximately 10 percent) in the first year after the win. Recall that per-adult wage earnings for married households is the average wage earnings of the winner and the spouse. The smaller decrease in per-adult wage earnings compared to winner wage earnings implies that the spouse of the winner decreases his or her wage earnings by a smaller amount. It is also evident that the winner is more likely to stop working, and this probability grows over time. This pattern also persists after accounting for spousal responses…We find that wealth effects for both earnings and employment decrease with prize size. For smaller winners, the average per-adult total earnings reduction per 100 dollars of additional wealth is in excess o 5.5 dollars, whereas for the largest winners, the reduction is less than 1 dollar per 100 dollars of additional wealth. This pattern of decreasing effects by prize size raises a natural question: How much of the crossprize difference in wage earnings responses is attributable to the extensive labor supply margin? To address this question, we decompose the difference in per-dollar earnings effects between smaller and larger prize winners into extensive- and intensive-margin contributions (see Appendix D). In particular, we focus our decomposition on the difference between the smaller ($30,000 to $300,000) prize bin and the largest ($1 million and up). We summarize the results of this decomposition of differences in Table 3.4. We find that the extensive margin explains a substantial fraction of the differences in per-dollar responses across prize size. For example, in aggregate, we find that for per-adult total labor earnings approximately half of the difference in wealth effects across prize size is attributable to differences along the extensive-margin response. The share of cross-prize-size differences attributable to the extensive margin declines as we look across the income distribution, falling from approximately 75% for lower-income (first quartile) winners to approximately 44% for higher-income (fourth quartile) winners…”

Christopher Esposito, Edward Leamer and Jerry Nickelsburg, "Who Paid Los Angeles' Minimum Wage? A Side-by-Side Minimum Wage Experiment in Los Angeles County," National Bureau Of Economic Research, June 2021, https://www.nber.org/papers/w28966

Restaurant Closures, “…The closure rate was notably higher in the southern portion of the City Wage Region. This is a part of the county with a larger proportion of lower income census tracts. The interior parts of the City Wage region and, Beverly Hills, West Hollywood, and Culver City in the State Wage region, more affluent parts of the county, had relatively low closure rates. In Table 10, we aggregate closures and compute closure rates at the strata level and, with raw closure rates reported for the reference group of restaurants (the State Non-Border) and closure rates relative to the reference group for the other three strata. Closure rates for restaurants in both high and low-income census tracts are shown to be higher in the City Border in Table 10. For robustness, we investigated the closure rate of restaurants using the alternative cross-border competitor construction described in Section 3. We report these results in Table 11. The first column of the table shows the closure rate for the reference group, State Wage restaurants not exposed to City Wage competitors. Columns 2-4 show closure rates relative to the reference group. State Wage Non-Border restaurants had a closure rate of 25.6% in low-income neighborhoods and 23.6% in high-income neighborhoods respectively. State Wage Border restaurants had lower closure rates, and City Wage Border restaurants had higher closure rates. The highest closure rates were for City Wage restaurants in low-income tracts that bordered on State Wage, low-income census tracts. In the City Wage region, the lowest closure rate was in the Non-Border high-income census tracts. is the same group of restaurants that also increased prices by the largest amount….”

Menu Changes, “…The above table provides evidence that restaurants changed their menus more frequently when subjected to the larger increase in the minimum wage. This result was particularly apparent among restaurants in high-income census tracts, where City Border and City Non-Border restaurants were respectively 4.5 and 13.3 percentage points more likely to change their menus than the reference group. If changes in menu items in response to the higher minimum wage are attempts by the restaurant to lower costs, then at least some of the incidence of the wage was borne by customers who purchased the inferior food items. Although we cannot determine if the menu changes served that purpose, the fact that they occurred more often at restaurants in low-income neighborhoods - where, as we found earlier, price increases were smaller- suggests that they may have been implemented to reduce costs…”

"...In Table 4 we display mean price changes taken between rounds 1 and 7 of our price survey broken out by these eight groups of restaurants. The first column shows mean price changes for the reference group, State wage restaurants not exposed to City wage competitors, while columns 2-4 show price increases for the other three groups of restaurants relative to the reference group. The results presented in Table 4 provide a robustness check on the results presented in Table 3. Both tables report exceptionally large price increases in the high-income State Border region which had the lower increase in minimum wages. Moreover, when price changes are computed between different rounds of the survey (Appendix A) the same result is obtained. This shifts attention to the pre-equilibrium hypothesis regarding exceptional closures of restaurants in the City Border region discussed in Section 8. The price changes reported in Table 4 for restaurants in low-income census tracts, however, are different from our Table 3 statistics. Table 3 reports no statistically significant difference in price changes in any of the three low income regions. Table 4 reports large (7.3) and statistically significant price increases in the City Border. Caution should be exercised when interpreting this result because few observations (33) were taken of items in low-income census tracts in the City Border that were exposed to State wage competitors...."

New NBER looks at how minimum wage increase are "paid for" via pass throughs, in high income neighborhoods the higher costs were passed on to customers, in low income neighborhoods they fell on landlords and restaurant owners"...We exploit a geographical discontinuity in Los Angeles County, where in 2015 the City of Los Angeles passed a minimum wage law and in 2016 the State of California passed a different minimum wage law. This created two minimum wage schedules in the county that remained unequal for over five years. Using a novel data set from a multi-year price survey, our analysis shows that the incidence of Los Angeles City’s higher minimum wage fell on customers in high-income neighborhoods, and on landlords and restaurant owners in low-income neighborhoods.We further show that the mix of responses at restaurants subject to the LA City minimum wage, including price increases, menu changes, and restaurant closures, was affected by proximity to restaurants subject to the lower California State minimum wage..."
Core Finding, "... Specifically, we found that restaurants in high-income neighborhoods and located far from lower-wage competition passed most of the City Wage differential through to their customers by raising prices. Restaurants in low-income neighborhoods including those located near lower-wage competition were found to not have larger price increases than those in the State Wage region. Instead, restaurants in low-income neighborhoods changed their menus more often and restaurants exposed to lower-wage competition were more likely to close. We infer that the spatial discontinuity in minimum wages caused demand for restaurant meals to spill across the wage border and to nearby restaurants subject to the State Wage. In support of this inference, we found that State Wage restaurants in high-income census tracts and located near City Wage competition increased prices and were less likely to close than their nearby City Wage competitors....

Evidence, "...In the summer of 2015, the City of Los Angeles enacted a schedule of minimum wage increases that would reach $15/hour in July 2020. Los Angeles County, which encompasses the City of Los Angeles, is comprised of 88 governmental jurisdictions not legally subject to the City’s minimum wage ordinance. In 2016 California enacted a statewide schedule of minimum wage increases which set a minimum for all 89 jurisdictions. The California state schedule of increases differed from the City’s in three important ways. First the California state minimum wage would not reach $15/hour until January 2022. Second, the two minimum wages are indexed to the CPI, but the indexation starts earlier for the City’s minimum wage. Third, in a recession, the Governor of California can suspend increases in the state minimum wage thereby extending the lower wage longer, while the City’s wage has no such provision. Absent the latter two differences, the minimum wage in Los Angeles City and in a handful of jurisdictions within Los Angeles County that elected to follow the City’s minimum wage schedule will be $1 to $1.50/hour higher than in the rest of the County between July 2017 and January 2022 (Figure 1)....In our study, we control for neighborhood income levels because customers prefer restaurants close to where they live, and because residents of high-income neighborhoods may have relatively inelastic demand for restaurant meals. Thus, restaurant owners in these neighborhoods should, on average, be able to pass along additional wage costs to their customers without a reduction in sales. Tellingly, in low-income neighborhoods we find that restaurants subject to the City Wage did not have statistically significant price increase for existing menu items, but they did change menu items more often. Finally, we note that small mom-and-pop restaurants are only subject to minimum wage requirements on statutory employees. Family members working at the restaurant are not considered employees. Mom and-pop restaurants are more concentrated in low-income neighborhoods, and this might also help to explain the lack of price increases in these regions...Although increases in the minimum wage occurred throughout the four years of the survey, we do not find restaurants responding to individual between-round increases in the minimum wage by raising prices in a way that is coordinated with the minimum wage increases…"
"...Table 1 has regressions that put the moving pieces in these figures together, explaining the log of restaurant employment and wages as a function of (1) a lagged dependent variable to account for the persistence of departures from normal, (2) the log of the California minimum wage, (3) the log of the overall employment or wages depending on which is the dependent variable, and (4) quarterly dummy variables that measure departures from QTR=1, which is the omitted effect absorbed by the constant C. An important problem with this regression is that much of the County is subject to the City minimum wage which generally exceeds the State minimum wage. This means there is an omitted variable equal to the City wage minus the State wage, which is pretty constant except for the fact that the City increment comes in July while the State increment comes in January. A second, and possibly more important problem, is that the minimum wage schedules are announced in advance, and the response of employment and wages need not be timed with the arrival of the pre-announced increases. The third, and most important problem, is that there is no control group without any minimum wage increases. With those caveats, the regressions in Table 1 are quite interesting. Attention in this table is focused with shading green or red on the t-statistics greater than 2 or less than minus 2. For the employment variables, the effect that is most statistically apparent is the lagged dependent variable (persistence). This persistence reinforces our view that the response of employment to a schedule of future minimum wage increases cannot be timed to the minimum wage increments but must take into account the future minimum wages. Employment in limited-service restaurants tracks overall employment in a statistically discernable way while employment in full service restaurants does not.For wages, the persistence effect is not detectable, but both the minimum wage and the overall LA County wage are statistically significant. It is worth mentioning that the minimum wage coefficients in both employment equations are negative, though not statistically significant. This could be summarized with the conclusion that effects on wages are apparent with this kind of regression model because the law requires actual wages to move with the minimum wages but the effects on employment are disguised by the fact that employment changes are not necessarily timed to the minimum increase: there can be adjustments in advance and adjustments long after. These results do conform with the minimum wage literature which finds apparent wage effects but often no apparent employment effects. This is natural segue into our study of restaurant prices: To what extent do LA county restaurants pass on the rising wages to customers via price increases, and to what extent to the City and State minimum wages produce different increments in prices?..."