The job market speaks. Are we listening?

Imagine losing your job in the spring of 2020, then spending the better part of a year watching more people crowd into the same shrinking pool of openings. We can now see it almost week by week, not because a government survey told us, but because a job site did.

In our research, supported by a grant from the International Growth Centre (IGC), we mined a decade of data from Bangladesh’s largest online job portal, BDJobs.com, covering mid-2015 to the end of 2025. We used it to build a weekly readout of how the job market felt to people living through it. When Bangladesh’s Covid-19 lockdown began in March 2020, job postings on the platform collapsed by 70 percent and advertised vacancies fell by 75 percent. Applications fell by only 59 percent. People kept searching as opportunities vanished. The result was a more competitive labour market: 287 people were chasing every open position, up from a pre-pandemic norm of 173.

Tap here to add The Daily Star as a trusted source

Image

Bushra Chowdhury

That squeeze did not ease quickly. Our modelling shows market “tightness”, the number of applicants per vacancy, was still nearly 30 percent above normal three months after the lockdown began. Bangladesh’s quarterly Labour Force Survey, with its publication lags, was not built to capture that in real time. For context, comparable data from the US and the UK showed similarly sharp initial falls in vacancies, but faster rebounds, likely because furlough schemes and unemployment insurance cushioned workers in ways that are far less available in Bangladesh’s largely informal economy. Our data, in other words, reveals not just the labour market shock but the gap in the country’s social safety net.

Compare that with the mass uprising of July 2024. Our data shows an initial dip, followed by sharp spikes. Postings briefly jumped more than 200 percent above baseline, then returned towards normal. Vacancies and postings recovered within about six months, while market tightness normalised in roughly four. It was volatile but short-lived, rather than the long squeeze of the pandemic. They left different fingerprints, and only high-frequency data lets us see the shape of each one. A quarterly survey would have blurred both events into a single noisy data point. Just as interesting is what our dataset failed to detect. The Bangladesh Bank raised and cut its policy interest rate 17 times over the decade, including an aggressive tightening cycle from 2022. Basic economic theory says higher borrowing costs should cool hiring. In our data, however, that signal all but vanishes. At most, a rate rise shaves off a few hundred postings for about a week before fading into statistical noise. Job postings actually grew fastest during the period of the most aggressive rate rises, overwhelmed by the stronger trend of firms moving hiring online.

Image

Atonu Rabbani

Taken together, our findings offer a lesson for policymakers. Real-time digital data is good at sounding the alarm on sudden shocks, such as a pandemic, uprising or natural disaster, but nearly useless for tracking the slow burn of monetary policy. It is a smoke detector, not a thermostat. Bangladesh, and countries like it, should build lightweight public early-warning dashboards using platforms like this one, refreshed weekly and sitting alongside, never replacing, slower and more representative official surveys. Such a dashboard would not have prevented the Covid shock, but it could have told policymakers within a week how bad it was and how unevenly it was affecting jobseekers.

Of course, this window is not the whole house. Online postings capture the formal, urban and internet-connected slice of the economy, leaving out roughly 85 percent of Bangladeshi workers who work informally. What happens on this job site is a signal, not a census. But it is a signal we have never had before, updated in real time, at essentially no cost beyond the will to look at it. The next crisis is a strange time to be flying blind when the data has been sitting there all along.

The writers are researchers at Bangladesh Institute of Development Studies (BIDS)