Published on Development Impact

Wage work, what is it good for? Labor market dynamics in urban Ghana: Guest post by Peter Deffebach

This page in:
Wage work, what is it good for? Labor market dynamics in urban Ghana: Guest post by Peter Deffebach

This is the twelfth in our series of posts by job market candidates.

In developing countries, only a small share of workers are engaged in wage employment. Why? Recently, Kevin Donovan, Will Lu, and Todd Schoellman made progress on this classic question by showing high exit rates out of the wage sector, as opposed to low entry rates into the wage sector, drive low overall levels of wage work.

However, limited data means we don’t know what causes such elevated rates of exit. Volatility faced by firms, for instance, could be driving lots of layoffs. On the other hand, forces on the worker’s side could be leading to frequent quits. To shed light on the causes of high exit rates out of the wage sector, I conducted a new survey of job-seekers and firms in Accra, Ghana.

In my job market paper, I argue that exits are driven by quits, which are in turn caused by volatility workers face outside the wage sector. Wage jobs are taken up as temporary options of last resort, but workers quit as opportunities in the non-wage sector improve. I build a general equilibrium model of job search capturing this intuition. I conclude a quarter of the difference in exit rates between the USA and Ghana can be explained by differences in the volatility of non-wage income faced by workers in the two countries.

New surveys in Accra, Ghana

My first new data source is a two-round panel survey of job-seekers. I recruited 465 individuals looking for work, with no other conditions attached, and asked these job-seekers about their current employment and job search strategies. I followed up with these job-seekers eight months later to ask about their current employment and experiences in the intervening eight months.

Overall, job-seekers were heavily male (likely due to my online recruitment method) and more educated than the average urban job-seeker: 50% have some college education. Job-seekers are 29 years old on average and all had previous work experience, with the average respondent having worked for 5 years. In this way, my sample differs from those studied to recent work on labor markets in Sub-Saharan Africa which focus on younger populations often entering the labor market for the first time.

My sample features high rates of on-the-job search. At baseline 45% of my sample was engaged in wage work, and even though almost everyone had worked for someone else in the past year, the median respondent had spent almost two years searching for a job. In sum, my job-seekers appear to be moving in and out of work, but continuously searching for better options in the wage sector.

I use the USA as a benchmark, employing Current Population Survey and the Survey of Income and Program Participation, and use a re-weighting method to make apples-to-apples comparisons.

My second new data source is a separate survey of 110 firms. I looked for formal firms employing positions desired by my job-seekers, mostly secretarial and administrative roles. The goal of the survey was to measure hiring, vacancy, quit, and layoff rates in a way that could be compared to the USA’s Job Openings and Labor Turnover Survey.

Exits are driven by worker-side, not firm-side, forces

As expected, jobs in Ghana didn’t last. Job-seekers in Ghana and a matched sample in the USA found work at equal rates, but those in Ghana exited at such high rates that eight months later, wage employment was half that of the USA.

Differences in exit rates between the two countries can’t be attributed to the prevalence of casual day labor. In Ghana, jobs that were created, then destroyed lasted about 2.6 months on average, and though “manual labor” jobs, such as construction and factory work, had the highest rate of destruction, at 66%, administrative jobs in offices also didn’t last, with a destruction rate of 40%.

I asked workers who entered, then exited, the wage sector, why they left that last job: A quit, a layoff, or because the job was only ever temporary work. Figure 1 shows 68% of workers left their last job in a quit, and less than 10% left their last job due to a layoff. The exact opposite is true in the USA. In both countries only 20% of workers who exited their job did so because their temporary contract expired, so it’s unlikely the temporary nature of jobs drives differences in exit rates between the two countries.

My firm survey tells the same story: Quits are far larger share of total separations in Ghanaian firms relative to firms in the USA. I interpret this to mean forces on the worker’s side, rather than the firm’s side, drive exit.

Figure 1: Exits in Ghana are much more dominated by quits than in the U.S.

Image

Quits are related to new income opportunities outside the wage sector

Why are workers quitting so much? To answer this, I examine what happened to workers before and after a quit. Figure 2 shows workers in Ghana who quit their work and exited the wage sector reported 20% higher incomes outside the wage sector, through a mix of self-employment income and transfers from family, than they did at the job before they quit. In the USA, there was no such income gain, and quitters saw large income drops of 70% after leaving the wage sector.

When development economists hear “income in the non-wage sector”, they usually think of self-employment income, such as running a small shop. In my paper, I show this is only half the story: Income gains from quitters are driven by self-employment income and transfers from friends and family in roughly equal measure. Transfers from friends and family might play a larger role in driving income dynamics in developing countries than previously assumed.

Figure 2: Quitters in Ghana see income gains, those in the U.S. see losses

Image 

Quits are concentrated among those without non-wage income sources at baseline

Given workers can see income gains after quitting, why take these jobs at all?

Consider two groups of job-seekers at baseline: Those with some source of non-wage income, whether from self-employment or transfers from friends and family, and those with no income source at all, forced to draw down savings to finance daily needs.

Somewhat surprisingly, I show this last group of workers is 26pp more likely to exit. In other words, when job-seekers with a source of non-wage income at baseline found work, they stayed employed. When job-seekers without a source of income found work, they quickly quit.

What’s more, when these workers quit their jobs, the appear to have encountered some new source of income, whether from self-employment or transfers from friends and family, in sharp contrast with workers who were laid off. Of those without income flows at baseline, only 20% of quitters were again without income flows at endline, compared to 67% of those who were laid off.

Coining a new term: Subsistence wage-employment

“Subsistence self-employment” refers to a host of income-generating activities that workers take up as an option of last resort in the absence of functioning labor markets. Subsistence self-employment is, of course, well documented and forms the basis of canonical models of the labor market in developing countries.

My findings bring a new twist on this concept by arguing subsistence wage employment is also a way workers manage risk in developing countries. For my job-seekers, wage work is easy to come by, but undesirable. Just as subsistence self-employment is taken up and dropped to cope with risk in the wage sector, subsistence wage-employment is taken up and dropped to cope with risk in the non-wage sector. Chris Blattman and Stefon Dercon document a similar dynamic studying factory workers in Ethiopia.

Volatility in non-wage earnings drives differences in exit between the USA and Ghana

Finally, I quantify the role volatility of non-wage income plays in driving difference in exit rates between the USA and Ghana. To this end I build a general equilibrium model of job search in which workers face idiosyncratically uncertain income flows outside the wage sector. I calibrate my model separately to the USA and Ghana, incorporating all the findings described so far.

I compare what underlying structural parameters in the USA and Ghana drive differences in exit rates. When I give Ghana the USA’s volatility of non-wage income, the gap in exit rates between the two countries shrinks by 25%.

The relative desirability of the wage compared to non-wage sector also drives differences in exit rates. In the USA, the wage sector is very productive compared to the non-wage sector. In Ghana, this premium is smaller, and when I give Ghana the USA’s wage-sector productivity, exit rates fall dramatically. In other words, when jobs pay high wages, workers don’t quit as much, even when volatility outside the wage sector remains high.

This final result indicates the declining exit rates along the path of development documented by Kevin Donovan, Will Lu, and Todd Schoellman may arise not only from declines in the volatility of non-wage earnings, but also secular improvements in the relative productivity of the wage sector, such as biased structural change.

Limitations and next steps

It’s clear policymakers want to reduce high exit rates out of wage work, but my model doesn’t capture the interesting dynamic effects that make longer job spells desirable. For instance, workers get more productive the longer they stay at a job, either from working or learning in general. A richer model, and better data, will help me measure the costs of exit more clearly and better inform optimal policy.

Peter Deffebach is a PhD student at Boston University. 


Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000