This is the 6th in this year’s series of posts by PhD students on the job market
Online freelancing platforms have expanded access to global digital job opportunities for hundreds of millions of workers from low- and middle-income countries (LMICs). Yet many new freelancers struggle to secure jobs without an established reputation. This challenge is particularly severe for workers from LMICs whose qualifications, such as education, are unfamiliar or difficult for foreign employers to verify.
A natural strategy for market entry would be to offer lower initial wages to attract employers and build a reputation. But according to job application data, few novice freelancers do this. Why don’t more workers lower their wages to get that first job?
In my job market paper, I test two hypotheses: first, workers believe that employers interpret low wages as signals of low quality; and second, workers are uncertain about their own ability to perform well. Using two field experiments on a leading global freelancing platform, I examine how these beliefs shape worker outcomes and whether correcting them can help qualified workers break into global digital markets.
Do Employers Really Penalize Low Prices?
In our baseline survey of 481 novice freelancers from 37 LMICs, 44% believed that offering a wage below the posted job budget signals low quality to employers. This concern seems reasonable: extensive evidence from product markets shows that consumers infer lower quality from lower prices when evaluating unfamiliar goods. But is this perception accurate in online labor markets?
To test this, we conducted a demand-side experiment, where we submitted applications to 703 data entry jobs on the platform from both novice and veteran freelancer profiles located in LMICs, and randomly assigned the wage offers.
We find novices misperceive employer behavior: low wage offers double the chance that an employer read the application and increase callback rates by 54%, relative to very low base rates of 4.5 and 3.4 percentage points, respectively. Employers respond much more positively to low wages from novices than from veterans, suggesting that they interpret low wage offers from novices as a willingness to invest in reputation rather than as signals of poor quality.
To keep the experimental profiles comparable in the number of completed jobs, we declined any callback messages, so we cannot observe whether these applications would have resulted in final job offers. Nonetheless, the higher callback rates indicate that lowering prices substantially improves novices’ chances of being seriously considered by employers.
Back-of-the-envelope calculations show that novices who consistently offer low wages earn twice as much within their first year as those who do not, assuming a one-to-one conversion from callbacks to job offers. Offering low prices is, in fact, an effective and profitable entry strategy. So why don’t more workers do it?
Novice workers lack information not just about demand, but also about their own abilities.
We have shown that one potential reason novice freelancers do not offer low wages is their (mis)beliefs about employer responses. Another potential reason is workers’ uncertainty about their own abilities. We hired the same 481 novice freelancers to complete a standardized data entry task and measured their performance. We found substantial variation across workers: some performed well, others struggled.
When asked to estimate their performance relative to other novices, very few workers guessed correctly. If workers are uncertain about whether they can deliver good work and earn positive reviews, they may be hesitant to make the upfront investment of lowering wages to build a reputation.
Can correcting workers' beliefs change their behavior?
To answer this, we ran a second experiment focusing on the supply side with the same group of 481 novice freelancers hired for the standardized data entry task. After workers completed the initial task, we cross-randomized two treatments at the individual level.
1. The feedback treatment: we provided half of the workers with information about their task performance relative to other novices in our sample.
A few days later, we posted a higher-value data entry job from another employer account and referred it to all experimental participants by sharing the job link.
2. The employer evaluation info treatment (hereafter, employer info treatment): we added one sentence for randomly selected workers, stating that the employer would not judge worker quality based on wage bids to alleviate their concerns. We chose to give this neutral message to avoid pushing freelancers to lower prices due to experimenter demand effects.
We then tracked workers’ application decisions and wage offers for the referred job.
Figure 1. Treatment Effects on the Share of Workers Offering Low Wages
The results were striking. As shown in Figure 1, while only 5.5% of control-group freelancers bid below the job budget, those receiving either information were 9–15 percentage points more likely to do so, a 92–373% increase relative to the baseline. This demonstrates that correcting worker beliefs about demand responses and their own abilities increases their willingness to adopt the low-wage strategy for market entry.
Moreover, high-performers and low-performers responded differently. As shown in Figure 2, when freelancers didn't know their performance (control and employer info groups), treatment effects were similar across ability levels. But when freelancers received feedback on their performance, the response diverged sharply: treatment effects were driven by high-performers, who were more likely to believe at baseline that low wage offers improve hiring prospects for high-ability novices and were less concerned that employers penalize low wages.
Figure 2. Heterogeneous Treatment Effects by Worker Performance Type (High/Low)
Therefore, our information interventions address belief frictions through different channels. Correcting workers’ misperception about employer responses to low wage offers encourages those with pessimistic beliefs about employer responses to offer low wages to secure jobs. Resolving workers’ uncertainty about own abilities strengthens high-ability workers’ incentives to invest in reputation through low wage offers, thereby accelerating talent discovery.
Why Does External Support Matter?
We have shown that inaccurate beliefs prevent novice workers from offering low wages to overcome entry barriers. A natural question is whether workers could eventually correct these beliefs on their own through experience in the market.
Using estimates from both experiments, we simulate how long it would take a perfectly rational worker to learn the true demand curve through trial and error. Even under systematic experimentation and perfect updating, learning is slow and costly: a median worker with moderately pessimistic beliefs about demand would need around 200 job applications and incur about $290 in application fees, equivalent to one-quarter of annual income in low-income countries.
For novice workers from our settings, these costs make it difficult to adapt market entry strategies purely through experience. Low-cost external interventions, such as guidance on employer hiring behavior, can correct these misperceptions and meaningfully expand access to digital work opportunities.
Policy Implications
Studies have shown that inexperienced workers from LMICs often have limited understanding of the market fundamentals and their abilities, which hinders their labor market access. Existing programs like Kenya’s Ajira Digital Program and Jordan and Lebanon’s Mastering the World of Online Freelancing already incorporate mentorship and business strategy guidance to help new workers navigate global platforms.
Integrating similar information-based interventions into digital skills training could help workers form realistic expectations about hiring and returns to reputation. Correcting these misbeliefs can unlock access to global opportunities for skilled workers in LMICs and improve job matching in online labor markets.
Ruoxuan (Rebecca) Wu is a PhD candidate in Economics at the University of Chicago (bluesky: @rebeccawurx.bsky.social). This paper is coauthored with Hamna Ahmed and Zunia Saif Tirmazee from Lahore School of Economics.
Join the Conversation