Code for the paper Unemployment Volatility: When Workers Pay Costs upon Accepting Jobs. The paper is included in the top of the directory in unemployment_volatility_when_workers_pay_costs.pdf.

Here’s the 100-word abstract:

Hiring workers is costly. Firms’ costs reduce resources that can go to recruitment and amplify how unemployment responds to changes in productivity. Workers also incur up-front costs. Examples include moving expenses and regulatory fees. Workers’ costs lessen unemployment volatility and leave resources available for recruitment unchanged. Their influence is bounded by properties of a matching function. Using adjusted data on job finding, I estimate a bound that ascribes limited influence. The results demonstrate that workers’ costs affect outcomes (firms threaten workers with paying the fixed costs again if negotiations fail), but their influence on volatility is less than firms’ costs.

Here is code that collects relevant data and plots the adjusted and unadjusted rates of job finding.

Here’s code that

library(tidyverse) # for useful R functionality
library(here)      # for accessing the data

dat <- read_csv(here("out", "dat_02-adjust-transition-rates.csv"))

dat_useful <- dat |> 
  select(date, ends_with("_crude"), find, sep)

# Save the useful data
fout <- here("out", "dat_funDecomp.csv")
write_csv(dat_useful, file = fout)

head(dat_useful)
#> # A tibble: 6 x 5
#>   date       find_crude sep_crude  find    sep
#>   <date>          <dbl>     <dbl> <dbl>  <dbl>
#> 1 2000-12-01      0.963    0.0391 0.618 0.0402
#> 2 2001-01-01      0.950    0.0417 0.616 0.0416
#> 3 2001-02-01      0.871    0.0382 0.580 0.0394
#> 4 2001-03-01      0.900    0.0404 0.583 0.0430
#> 5 2001-04-01      0.830    0.0382 0.562 0.0389
#> 6 2001-05-01      0.869    0.0390 0.577 0.0409

The paper contains a detailed account of the variables. Briefly, the variables are:

• date : monthly date
• find_crude : number of hires divided by the unemployment level
• find : rate of job finding adjusted to account for workers’ transitions and hires recording all hires, even hires fired before the end of the month
• sep_crude : unadjusted separation rate
• sep : adjusted separation rate

Here is a figure that uses the data:

datfig <- read_csv(here("out", "dat_funDecomp.csv")) 

# Put the data into long form for plotting
datfig <- datfig |> 
  select(date, starts_with("find")) |> 
  pivot_longer(starts_with("find")) 
  
csub_blue <- rgb(0, 26, 112, maxColorValue = 255)

ggplot(data = datfig) +
  geom_hline(yintercept = 1.0, color = "maroon", linetype = "dotted") +
  geom_line(mapping = aes(x = date, y = value, color = name, linetype = name), size = 0.9) +
  scale_color_manual(values = c("find_crude" = "black", "find" = csub_blue)) +
  scale_linetype_manual(values = c("find" = "solid", "find_crude" = "dashed")) +  
  labs(x = "", y = "", title = "Monthly probability of finding a job", 
       color = "Series", linetype = "Series") +
  theme_light()  

Code that produces better-looking figures is contained in cde/.

Code that replicates that project is contained in cde/. The scripts are prefixed with the order in which they need to be run.

• 01-get-FRED-data.R: Retrieves data from FRED.
• 02-adjust-transition-rates.R: Uses the FRED data to adjust rates of job finding and job separation to account for
  • workers’ transitions between employment and unemployment and
  • how the Job Openings and Labor Turnover Survey (JOLTS) records hires—all hires for a month are recorded, even those who separate before the month ends.
• 03-est-matching.R: Estimates the nonlinear mathching technology, $M \left( u, v \right) = \mu \frac{uv}{\left( u^{\gamma} + v^{\gamma} \right)^{1/\gamma}}$. Various plots are produced showing how the Beveridge curve shifted since JOLTS data on vacancies became available and the fit produced by the estimated model.
• 04-plot-elasticity.R: Plots the elasticity of matching with respect to unemployment, $\eta_{M,u}$, and the upper bound for $\Upsilon$, $\max \left( \frac{1}{\eta_{M,u}}, \frac{1}{1-\eta_{M,u}} \right)$. The factor $\Upsilon$ is one of the factors in Ljungqvist and Sargent’s (2017) fundamental decomposition of the elasticity of market tightness with respect to productivity, $\eta_{\theta,y}$. The code also generates the tables in the text.
• 05-plot-replacement-rates.R: Plots two measures of how unemployment-insurance benefits replace claimants’ labor income.

After running an script, it is possible to view a function’s documentation using ?my_function. For example, docstring::docstring(compute_c) returns the documentation of the function compute_c.

This feature is accomplished with R’s docstring package.

Ryan, Rich. (2024) Unemployment volatility: When Workers Pay Costs upon Accepting Jobs. International Journal of Economic Theory, 20, 303–333. https://doi.org/10.1111/ijet.12405

Replication Material for “Unemployment Volatility: When Workers Pay Costs upon Accepting Jobs” by Rich Ryan is licensed under CC BY-SA 4.0