Feature Articles
Measuring Firm-Level Artificial Intelligence Use and its Effects on Firm Performance and Employment in Singapore
11 August 2026
This study develops a firm-level measure of Artificial Intelligence (AI) use in Singapore based on data from online job postings. Linking the constructed measure to firm-level administrative data, the study then analyses the characteristics of AI-using firms and empirically estimates the impact of firms’ AI use on their performance and employment.
The main findings of the study are as follows. First, using network analysis, the study finds that AI use is not a stand-alone leap for firms. Instead, it tends to build on the prior efforts of firms to adopt foundational and complementary digital technologies.
Second, using a liner probability model, the study finds that firms in digital- and data-intensive sectors/clusters (i.e., information & communications, electronics, professional services and finance & insurance) are more likely to use AI. Larger firms are also more likely to use AI, even after accounting for their sector and age.
Third, using a combination of matched staggered difference-in-differences and fixed effects regressions, the study finds that:
Initial AI use is associated with an increase in firms’ revenue and total employment, and these gains continue to rise as AI capabilities deepen. This result could reflect the potential scale and/or innovation effects of AI use (e.g., AI reduces costs and/or enables the creation of new products, thus allowing firms to expand output). At the same time, the increase in firm-level employment suggests that any worker separations that may occur at the firm are outweighed by new hiring. It may also be the case that AI tends to automate specific tasks rather than entire jobs, thus allowing workers to focus on the remaining and/ or new complementary tasks.
Firms do not see a statistically significant increase in their productivity and profit within four years of first AI adoption. This could be because the full benefits of AI may only materialise after complementary investments are made, similar to the experience of earlier transformative technologies such as the internet.
Employment effects of firms’ AI use differ across worker types. Initial gains are found to be concentrated among local higher-earners and mid-career workers, as well as skilled foreign professionals. However, as firms deepen their AI capabilities, the local employment gains broaden across wage and age groups.
Fourth, heterogeneity analyses show that the impact of firms’ AI use varies across sector and firm size. For example, firms in the finance & insurance, wholesale trade and information & communications sectors, as well as micro and medium firms, see larger gains in revenue on average.
As the study analyses firms’ AI use up to 2024, the estimates should be seen as reflecting the early impact of AI diffusion, and would not capture the effects of more recent advances in AI capabilities (e.g., agentic AI) as well as broader enterprise adoption. Future studies could analyse the effects of the different AI use cases that firms are implementing using more recent data that better capture the advances in AI technology and new patterns of adoption.
The views expressed in this paper are solely those of the author and do not necessarily reflect those of the Ministry of Trade and Industry or the Government of Singapore.