When governments cannot track incomes and poverty in real time, social support can arrive late, miss entire communities or be shaped by what is loudest rather than what is most urgent.

Lebanon’s recent years have shown how quickly living standards can change. Yet measuring those changes is hard when surveys are infrequent, incomplete or interrupted. This directly translates into ineffective policies, which are guided by anecdotes rather than evidence. 

In a recently published paper, Dr. Walid Marrouch, professor at the Adnan Kassar School of Business, in collaboration with University of Ottawa researchers Dr. Paul Makdissi and Dr. Myra Yazbeck, investigated ways to use available data to inform policymaking.  

The study, published in Review of Income and Wealth, explores how to monitor poverty in a “data-deprived” setting, where respondents often report income in broad ranges while others do not answer income questions at all. Rather than discarding these surveys, the authors reconstructed the income distribution by filling in plausible values within each reported bracket, then quantifying uncertainty using ranges.  

Going further, the researchers tested what happens when the missing answers are not random, producing bounds that reflect a more cautious reading of the data. To check if this approach held up, they validated it using surveys that included exact income values: When they modified those exact values into ranges and reran the method, the reconstructed distribution closely matched the original one. 

The results attribute numbers to the economic crisis that has plagued Lebanon for the ‎past six years.

At the poverty line commonly used for upper-middle-income countries and the most usual assumption that missing answers are random, the ‎estimated poverty rate rose to 71.0 percent in 2021–2022, compared with 5.8 percent in ‎‎2016 and 10.0 percent in 2018.

At the poverty line used for lower-middle-income countries, it reached 38.1 percent in 2021–‎‎2022. Even with stricter ‎assumptions about missing income answers, the 2021–2022 poverty estimate at the first ‎line remained between 59.0 percent and 75.9 percent. 

The study also distinguishes between what is statistically significant and what is not. The pre-crisis comparison between 2016 and 2018 depends more heavily on assumptions, and the authors show that the two years can overlap once a missing answer is treated more cautiously, meaning the data does not support a confident ranking there.  

But the post-crisis deterioration is robust: Poverty in 2021–2022 is reported to be 5.2 to 7.7 times higher than in 2018, regardless of how one models missing answers.  

Conducting such a study comes with its own challenges. “The main challenge is the lack of real household income and expenditures survey data, as well as the lack of access to such data even when it exists,” explained Dr. Marrouch. For instance, access to data from the survey conducted by the World Bank–Lebanon’s Central Administration of Statistics is restricted. When it comes to the MICS survey conducted by the UNICEF, Dr. Marrouch noted how Lebanon is the only Arab country for which this data is not made available to researchers.

The methodology presented in the study, tackling a longstanding problem in both data collection and policymaking, enables policymakers to rely on accurate estimates rather than anecdotal evidence. 

Nevertheless, there is no substitute for proper data, said Dr. Marrouch. “The approach we propose in the paper is intended to be complementary, serving to bridge the gap between official survey waves,” he clarified. “However, to design effective, evidence-based policies, we as academic researchers ultimately need consistent access to real household income and expenditure data.”  

To browse more scholarly output by the LAU community, visit our open-access digital archive, the Lebanese American University Repository (LAUR).