The Unassuming Economist

From a new ILO working paper:

“The current wave of technological change based on advancements in artificial intelligence (AI) has created widespread fear of job losses and further rises in inequality. This paper discusses the rationale for these fears, highlighting the specific nature of AI and comparing previous waves of automation and robotization with the current advancements made possible by a wide-spread adoption of AI. It argues that large opportunities in terms of increases in productivity can ensue, including for developing countries, given the vastly reduced costs of capital that some applications have demonstrated and the potential for productivity increases, especially among the low-skilled. At the same time, risks in the form of further increases in inequality need to be addressed if the benefits from AI-based technological progress are to be broadly shared. For this, skills policy are necessary but not sufficient. In addition, new forms of regulating the digital economy are called for that prevent further rises in market concentration, ensure proper data protection and privacy and help share the benefits of productivity growth through a combination of profit sharing, (digital) capital taxation and a reduction in working time. The paper calls for a moderately optimistic outlook on the opportunities and risks from artificial intelligence, provided policy-makers and social partners take the particular characteristics of these new technologies into account.”

A new VOX post “reviews the evidence on the main causes of the secular decline in employment since the turn of the century. Labour demand factors – notably import competition from China and the rise of industrial robots – emerge as the key drivers. Some labour supply and institutional factors also have contributed to the decline, but to a lesser extent.”

“Among the effects for which we were able to construct an estimate, increased import competition from China is the single largest contributing factor to the decline in employment. A great deal of empirical evidence links import competition from China to the decline in manufacturing employment. These import pressures also had negative employment effects on ‘upstream’ intermediate goods industries, as well as other non-manufacturing industries (Acemoglu et al. 2016, Autor et al. 2015). Estimates from Acemoglu et al. (2016) are the basis for our estimate that the 302% increase in Chinese imports (measured in 2007 US dollars) from 1999 to 2016 led to displacement of approximately 2.65 million workers, a 1.04 percentage point shift in the employment-population ratio.”

“The employment effects from industrial robots are most clearly documented by Acemoglu and Restrepo (2017), who find that each additional robot per thousand workers between 1993 and 2007 reduced employment by 5.6 workers. This estimate implies that the rise in the stock of robots between 2007 and 2016 displaced 0.95 million workers, equivalent to a 0.37 percentage point decline in the employment-to-population ratio.Although improvements in computing technology have been if anything even more ubiquitous, Autor et al. (2015) find that competition from computing technology affects only routine task-intensive occupations, and that any employment losses in those occupations were offset by employment gains in abstract and manual task-intensive occupations.”

Labour supply factors as a group have been considerably less important than labour demand factors in driving the decline in employment.The claim that expanded safety net support through SNAP (food stamps) or Medicaid led to sizable decreases in employment is hard to square with either the institutional features of these programmes or the evidence on causal linkages. Careful work finds little to no labour supply effects of these programmes, and as a practical matter, they offer very little by way of income support to able-bodied childless adults (see, for example, Hoynes and Schanzenbach 2016 on SNAP, and Leung and Mas 2016 on Medicaid.)

That said, the availability of lifelong disability insurance benefits through the Social Security Disability Insurance (SSDI) programme and the Veterans Affairs Disability Compensation (VADC) programme has contributed modestly to falling employment rates. Applying age-group specific causal labour supply estimates from Maestes et al. (2013) to the growth in the SSDI caseload over and above that due to the population becoming older yields an estimated 0.14 percentage point decline in the e/pop ratio over our time period owing to growth in this programme. Applying the labour supply elasticity from a recent study of an exogenous policy change to VADC program eligibility (Autor et al. 2016) to an estimate of the excess VADC caseload implies that programme’s growth led to an e/pop decline of perhaps 0.06 percentage points.

One might expect the tremendous rise in incarceration in the US to have been a significant driver of declining employment, but because so many incarcerated individuals had low levels of labour force attachment even before their prison term, we estimate only a modest aggregate effect. Applying the causal estimates from Mueller-Smith (2015) to rough estimates of the number of former prisoners by length of time served and prior earnings history, our very rough estimate is that perhaps 0.13 percentage points of the decline in e/pop between 1999 and 2016 can be attributed to policy-induced increases in incarceration. Minimum wage increases probably also had a small but non-negligible impact, especially among younger, less-skilled workers. Taking account of the range of estimates produced by credible study designs, we estimate that increases in state and local minimum wages might have contributed 0.10 percentage points to the e/pop decline.

Other plausible factors driving the decline in employment are the sharp rise in occupational licensing (Kleiner and Krueger 2013), the decline in geographic mobility (Molloy et al. 2011), and increased difficulties securing affordable, high-quality child care. We do not attempt to assign a magnitude to these factors because we lack sufficient evidence to establish the causal impact of these factors or, in the case of child care, to assess how much the factor itself has changed. Another open question is to what extent anecdotes about worsening mismatch between the skills workers possess and those that employers need are borne out in the data.

Scholars have noted the connection of both increased leisure time (including time playing video games) (Aguiar et al. 2017) and increased opioid use (Krueger 2017, Currie et al. 2018) with non-employment, but whether one is causing the other or vice versa, or they are all manifestations of other societal changes, is not easy to disentangle.”

From a new article by Jessie Romero and Felipe F. Schwartzman at the Richmond Fed:

“Inequality in the United States has an important spatial component. More-skilled workers tend to live in larger cities where they earn higher wages. Less-skilled workers make lower wages and do not experience similar gains even when they live in those cities. This dynamic implies that larger cities are also more unequal. These relationships appear to have become more pronounced as inequality has increased. The evidence points to externalities among high-skilled workers as a significant contributor to those patterns.”

“A large body of research has identified several key facts about inequality across and within cities. First, larger cities have a greater concentration of high-skilled workers. In the Fifth District, for example, the share of the population over age twenty-five with a bachelor’s degree is 45 percent in the most urban areas, compared with 16 percent in the most rural areas. In the United States as a whole, the proportion ranges from 35 percent in the most urban areas to 17 percent in the most rural areas.”

“Second, nominal wages are higher in larger cities and in cities with a larger proportion of high-skilled workers. In the most urban areas of the Fifth District, average annual pay in 2016 was nearly $64,000; in the most rural areas, it was less than $35,000. Nationwide, workers in the most urban areas earned about $60,000 on average in 2016, while workers in the most rural areas earned about $36,000. (See Figure 2 above.) In recent research, Nathaniel Baum-Snow, Matthew Freedman, and Ronni Pavan find that nominal wages increase 0.065 percent for every percentage point increase in city size (based on data from 2005–07). They also find that the relationship between city size and wages has strengthened over time and that the wage gap between urban and rural areas has increased”

From a new IMF Staff Discussion Note:

“Opportunities and challenges. Women are underrepresented in science, technology, engineering, and mathematics (STEM) sectors anticipating jobs growth, where technological change can be complementary to human skills. There are some bright spots: job growth in traditionally female-dominated sectors, such as education and health services, will likely expand. The ongoing digital transformation is also likely to confer greater flexibility in work, benefitting women. But, breaking the “glass-ceiling” will be critical. Across sectors and occupations, underrepresentation of women in professional and managerial positions places them at high risk of displacement by technology.

Crucial role for policy. Fostering gender equality and gender empowerment in the changing landscape of work remains an imperative across countries.

• Endowing women with the requisite skills. Early investment in women in STEM fields, along with peer mentoring, can help break down gender stereotypes and increase retention. Fiscal instruments for those already in the workforce (e.g., tax deductions for training, portable individual learning accounts) can remove barriers to lifelong learning.
• Closing gender gaps in leadership positions. Family-friendly policies can play an important role in boosting women’s retention and career progression, but setting relevant recruitment and retention targets for organizations, promotion quotas, as well as mentorship and training programs to promote female talent into managerial positions should be considered.
• Bridging the digital divide. When it comes to the use of new technologies and access to them, countries must close gender gaps to improve women’s labor market prospects in the new world of work. Governments have a role to play through public investment in capital infrastructure and ensuring equal access to finance and connectivity.
• Easing transitions for workers. Ensuring gender equality in support for displaced workers through active labor market policies will be essential, given the high risk of automation faced by women. Ensuring that training and benefits are linked to individuals rather than jobs can help improve their reemployment prospects. Social protection systems will need to adapt to the new forms of work.”

From a new IMF Staff Discussion Note:

“While progress has been made in increasing female labor force participation (FLFP) in the past 20 years, the pace has been uneven, and large gaps remain. FLFP was 54 percent for the median Organisation for Economic Co-operation and Development (OECD) country in 2014, 14 percentage points below male labor force participation (MLFP); for the median middle-income country, FLFP was only 49 percent, 26 percentage points below MLFP; and for the median low-income country, FLFP was 64 percent, 13 points below MLFP.

Narrowing participation gaps between women and men is likely to engender large economic gains, with two mechanisms pointing to larger gains than previously thought:

• Gender diversity: Women bring new skills to the workplace. This may reflect social norms and their impact on upbringing, social interactions, as well as differences in risk preference and response to incentives, for example. As such, there is an economic benefit from diversity—that is, from bringing women into the labor force—over and above the benefit resulting from simply having more workers. This hypothesis finds support in the data—both cross-country macro data and firm-level data. This paper finds that male and female labor are complementary in production. The results also imply that standard models, which do not differentiate between genders in their analysis, understate the favorable impact of gender inclusion on growth, and misattribute to technology a part of growth that is actually caused by women’s participation. The results further suggest that narrowing gender gaps benefits both men and women, because of a boost to male wages from higher FLFP.
• Sectoral reallocation: As households get richer during the process of economic development, demand for services rises, and labor is reallocated to the growing sector. Because services are more gender equal in employment than other sectors, developing economies naturally become more inclusive. But barriers to FLFP (which include tax distortions, discrimination, and social/cultural factors) slow this process, reducing output and welfare. This paper estimates that these barriers can depress FLFP by as much as a tax of up to 50 percent on female labor, depending on the region. Barriers not only hold back gender parity, they have a direct cost: welfare gains from their removal would exceed 20 percent in India, Pakistan and other countries in the Middle East and North Africa, for example.

These mechanisms imply that reducing female underemployment should yield greater gains than an equivalent increase in male employment: gender diversity brings benefits all its own.”