Extra Empirical Project: Female labour supply and the macroeconomy Solutions

These are not model answers. They are provided to help students, including those doing the project outside a formal class, to check their progress while working through the questions using the Excel, R, or Google Sheets walk-throughs. There are also brief notes for the more interpretive questions. Students taking courses using Doing Economics should follow the guidance of their instructors.

Note

These solutions are based on data downloaded in January 2023. Your solutions may differ slightly if using more updated data. You can download the file used to create the solution figures here.

Part 1: Collecting and preparing the data

No solution is provided. Users can download an example dataset for the US to check that their final dataset has been formatted properly.

Solutions for the US are shown. In cases where the solution depends on the country chosen, some guidance is provided on how to interpret the statistical output.

  • Solution figure 1 shows the labour force participation rate by sex.
Labour force participation rates in the US, by sex (1960–2021).
Fullscreen

Solution figure 1 Labour force participation rates in the US, by sex (1960–2021).

  • Solution figure 2 shows the labour force participation rate, with shaded bars for recession periods.
Labour force participation rates in the US, by sex (1960–2021). Shaded bars indicate recession periods.
Fullscreen

Solution figure 2 Labour force participation rates in the US, by sex (1960–2021). Shaded bars indicate recession periods.

  • In the US, female labour force participation (LFP) increased steadily from around 40% in the 1960s to around 60% in the late 1990s. Since then, it stagnated and even declined after the global financial crisis.

    Researchers attribute this sharp rise from the 1960s to 1990s to (among other reasons) increased labour market opportunities for women due to the growing service sector, women becoming more educated, declining fertility rates, and women spending less time working in the household due to better household appliance technology.

    The male LFP rate has featured a continuous downward trend from around 80% in the 1960s to 68% in 2021. This leaves about a 12 percentage point gap between men’s and women’s LFP rates in 2021.

    Possible explanations put forward by researchers for the decline in male LFP include the decline of US manufacturing (a male-dominated industry), the decline in male educational attainment and the rise of substance abuse. Possible explanations put forward by researchers for the decline in male LFP include the decline of US manufacturing (a male-dominated industry), the decline in male educational attainment and the rise of substance abuse.

    The jumps in both series in the early 1960s are due to changes in the measurement methodology and have no economic meaning.

    Comparisons depend on the country chosen, but could include:

    • comparisons of the level (is the LFP higher/lower, and for which sexes?)
    • comparisons of trends over time.
  • No solution is provided.
  • No solution is provided.
  • Solution figures 3 and 4 show the charts for the US.
The cyclical component of log GDP for the US (1960–2021).
Fullscreen

Solution figure 3 The cyclical component of log GDP for the US (1960–2021).

Cyclical component of average hours worked in the US (1979–2021).
Fullscreen

Solution figure 4 Cyclical component of average hours worked in the US (1979–2021).

  • The great moderation is considered to have started in the mid-1980s. This can be seen by the dampened volatility of GDP after the early-1980s recessions, compared to the highly volatile period of the 1970s and early 1980s. There is no consensus on when the great moderation ended (if it has ended at all): some researchers argue that it ended with the global financial crisis in 2008, which featured a stark drop in GDP growth. Other researchers hold the view that the global financial crisis and the COVID-19 recession were extraordinary events, distinct from earlier recessions, and that (excluding these two abnormal shocks) the great moderation still continues today.

    We cannot see the great moderation in the employment series, as the data only starts in 1979, but we would have seen greater volatility in the 1960s and 1970s compared to the mid-1980s onwards.

    Comparisons depend on the country chosen, but could include:

    • comparisons of the duration of the great moderation (approximate start and end dates)
    • comparisons of the strength of the great moderation. (How much was the volatility of the cyclical components reduced compared to the pre-great moderation period?)
  • We obtain a cyclical (relative to GDP) correlation of average hours worked in the US of 0.72. This value is positive and relatively close to 1, which implies that employment (the intensive margin only in this case) is strongly procyclical in the US.
  • Employment is less volatile than GDP because its relative standard deviation (the standard deviation of employment divided by the standard deviation of GDP) is less than 1 (0.23 for total US employment). In the US, male employment is more volatile than female employment, indicated by the larger cyclical volatility (0.27 vs 0.19).

    Comparisons depend on the country chosen, but could include:

    • comparisons of the cyclical volatility of employment in the three groups (all persons, women, and men) between both countries
    • comparisons of the cyclical volatility between the three types of employment in your other country. (Do the same patterns as in the US apply there?)
  • Solution figure 5 shows the required chart for the US.
Cyclical volatility of hours worked by sex in the US (1979–2021).
Fullscreen

Solution figure 5 Cyclical volatility of hours worked by sex in the US (1979–2021).

  • In both periods, male employment is more volatile than female employment, in line with the full-sample result in Question 2d. This gender gap is more pronounced during the pre-global financial crisis period. At the same time, female labour force participation (LFP) was increasing steadily up to the 1990s. This implies that women constituted a larger share of the labour force (and therefore of aggregate hours worked), which caused aggregate employment to become less volatile during this period. As employment is highly correlated with GDP (that is, strongly procyclical; see Question 2e), this also stabilized GDP fluctuations during the same period, leading to the great moderation.

    The post-2008 results indicate that female and male employment have been almost equally volatile since the global financial crisis, while employment overall has become less volatile compared to pre-2008. But as this period contains only a decade of data (and therefore not enough variation over time), these facts do not provide sufficient evidence on whether or not the great moderation has ended with the global financial crisis.

  • The answer depends on the chosen country.
  • Solution figure 6 shows the trend component for the US.
Trend component in labour productivity for the US (1971–2021).
Fullscreen

Solution figure 6 Trend component in labour productivity for the US (1971–2021).

Productivity growth in the US declined rapidly during the 1970s. After an acceleration of growth during the early 1980s, there was a second period of subdued productivity growth that lasted until the mid-1990s.

Comparisons depend on the country chosen, but could include:

  • comparisons of the approximate start and end dates of the productivity slowdown (if there was one in the other country)
  • comparisons of how pronounced that slowdown was (in terms of growth rate of trend productivity).
  • Solution figure 7 shows the trend components for male and female LFP, along with that of labour productivity in the US.
Trend components of ALP, female LFP, and male LFP for the US (1963–2021).
Fullscreen

Solution figure 7 Trend components of ALP, female LFP, and male LFP for the US (1963–2021).

The trend component of the female LFP rate grew until 2000 (positive vertical axis values indicate growth) and has been declining since then (indicated by negative vertical axis values). It grew fastest during the 1970s and early 1980s, coinciding with the productivity slowdown. The male LFP rate features a slight but continuous downward trend throughout the sample, which does not appear to be related to changes in trend productivity at all. Since 2000, female and male trend LFP are practically comoving, without any discernable differences.

  • Trend productivity growth in the US slowed down at the same time as female LFP was strongly increasing (during the 1970s and 1980s) and accelerated the most when female LFP growth started to stagnate around the year 2000. This indicates that these two variables are inversely related. One possible explanation for this is that the women who entered the labour force during the 1970s and 1980s had relatively low levels of education and work experience compared to employed men. A larger share of women in the labour force therefore reduced productivity (as measured by output per hour worked). The productivity boom in the 1990s can be explained by women becoming more educated and experienced, while their hours worked started to stagnate (therefore increasing output per hour). In the long run, this trend is expected to persist, with women continuing to contribute positively to aggregate labour productivity.
  • Comparisons depend on the country chosen, but could include:
    • comparison of the relationship between LFP trends by sex and the ALP trend component between the two countries
    • whether female and male LFP growth have been comoving recently in the other country as well
    • whether the explanation for the US productivity slowdown applies to the other country’s potential productivity slowdown as well.
  • Solution figure 8 shows the (log) labour force participation for the US.
Log labour force participation rate for the US (1963–2021).
Fullscreen

Solution figure 8 Log labour force participation rate for the US (1963–2021).

  • The plot shows that after the early (pre-2000) recessions, the labour force participation in the US recovered quickly back to (and even exceeding) the pre-recession level. This indicates that the jobs that were lost during these recessions were ‘recreated’ relatively quickly. After more recent recessions (early-2000s, global financial crisis) however, the LFP rate has not recovered at all, making these recoveries ‘jobless’. In this data, the recovery after the early-2000s recession can be considered the first jobless recovery.
  • Solution figures 9 and 10 show the labour force participation rates in the US, by sex.
Female labour force participation rate in the US (1963–2021). Shaded bars indicate recession periods.
Fullscreen

Solution figure 9 Female labour force participation rate in the US (1963–2021). Shaded bars indicate recession periods.

Male labour force participation rate in the US (1963–2021). Shaded bars indicate recession periods.
Fullscreen

Solution figure 10 Male labour force participation rate in the US (1963–2021). Shaded bars indicate recession periods.

  • The female LFP graph shows that female employment did not decline during the earlier recessions, as female LFP was growing steadily up to the 1990s. Male LFP, however, has always declined during recessions, and has not recovered after recessions. That is, male recoveries have always been jobless, as recessions have had a scarring (long-term negative) impact on male labour force participation.

    Taken together, these sex-specific empirical facts indicate that the quick recovery of employment following the pre-1990s recessions was driven by the continuous growth of female employment. Then in the 1990s, as female LFP started to flatten out, overall employment recoveries became jobless, as female LFP growth could no longer make up for men’s jobless recoveries (because there was basically no more female employment growth). In fact, recent female recoveries are looking very similar to men’s today (that is, jobless), which is why recent (and probably future) recessions have had such devastating and scarring effects on employment.

  • The answer depends on the chosen country.
  1. Our results for the United States are in line with Albanesi’s results on the great moderation, productivity slowdown and jobless recoveries, despite using different data sources (Albanesi uses micro data; we used aggregate OECD and World Bank data).

This exercise has illustrated how changes at the household (micro) level, such as a growing number of women starting to work in the labour market, can explain macroeconomic phenomena. That is why most of macroeconomic research today follows a ‘micro-to-macro’ approach, where researchers look at the micro level (by using household or firm-level data) when trying to make sense of macroeconomic inequalities and trends.