Strengthening Nepal's Management Information System for effective Covid-19 responses

Key Insights

  • While Nepal’s Crisis Management Information System (CMIS) was set up to help support the allocation of Covid-19 response funds to local governments, CMIS data entry and storage methods are currently inadequate for robust analysis of local government needs. 
  • Data poverty and technical staff shortages may be contributing to poor Covid-19 financing response in vulnerable municipalities, especially in the southern region of Terai. 
  • To address these issues, the database should be improved and technical support should be provided to local governments that need it.  

Background

The Crisis Management Information System (CMIS) is a web-based platform set up by Nepal’s Ministry of Federal Affairs and General Administration (MoFAGA) in April 2020 to facilitate reporting by local governments on activities and expenditures related to the Covid-19 pandemic. 

The CMIS is a potentially valuable source of data for tracking local government responses to the pandemic, such as funding levels, health activities (including quarantining), and numbers of beneficiaries from government relief programs. 

To analyze the effectiveness of the CMIS system as a Covid-19 policy tool, we extracted CMIS data at four points in 2020 during Nepal’s nationwide lockdown: June 8, 11, and 22, and July 6. For cross-validation and analysis we also used data from our Local and Provincial Government Survey (LGPS), the Federal Capacity Needs Assessment Survey (FCNA), and data from the United Nations Resident Coordinator Office (UNRCO). 

The LGPS survey was completed in June 2020 by Yale University, Nepal Administrative Staff College, GovLab Nepal, and the London School of Economics. The FCNA was completed in 2019 by the Nepal Administrative Staff College (NASC), Georgia State University, and the World Bank. 
 

Key Findings

Finding 1: Data reporting is infrequent and imprecise

Our analysis found that local governments are updating responses in CMIS infrequently. Missing values are common in the database, and often coded as zero, making the cases of no data and measurements of zero indistinguishable. 

Finding 2: Data responsiveness (and quality) matters for federal funding

Figure 1 shows that local governments that report zero quarantine numbers in the system receive less funds for quarantine centers from the federal government. The difference in average quarantine funding available for municipalities that report people in quarantine at the time the data were extracted from CMIS ranges from $2,760 (3.2 lakh Nepalese rupees) to $3,709 (4.3 lakh Nepalese rupees). In this case, UNRCO’s survey of quarantine facilities shows that these municipalities had people in quarantine. This suggests the issue is missing data/lack of data entry. 

Bar Graph showing CMIS funding
Figure 1: Average CMIS funding for quarantine 

Finding 3: Staff shortages affect reporting 

Municipalities in the Terai region, especially Province 2, have the lowest reporting levels. Municipalities with technical staff shortages (as recorded by FCNA survey) in digital data-entry positions report roughly 7.5 percentage points fewer variables.

Finding 4: The database lacks system checks to ensure data consistency across categories

Discrepancies in data fields suggest poor validation efforts. Among local governments who report any individuals in quarantine, most of the time the reported sum of all individuals by quarantine type (government, home, private, etc.) and gender did not match the reported “total individuals in quarantine.”  Similarly, for 61% of local governments that responded to questions on funding, the reported total of all funding sources did not match the reported “total funding.” 

Recommendations

MoFAGA can take multiple steps to enhance quality data entry by local governments. First and foremost, they could reach out to local government to determine what reporting barriers they are facing, and accordingly resolve any issues in the system. Such efforts could focus on municipalities with lower reporting rates. Second, they could work with CMIS technical staff to allow for missing values in data (to distinguish between “not reported” and “0”). It is impossible to draw conclusions about testing, tracing, funding, etc. when it is unclear whether zeros reflect on-the-ground reality or are evidence of missing data. Finally, they could build in system and logic checks to the CMIS system to ensure consistency across variables within a category. In addition, MoFAGA might consider a region-specific reallocation of personnel. Technical staff support would be valuable in Terai region municipalities.


Appendix

A full report of this analysis, titled “The Crisis Management Information System (CMIS): Summary, Data Quality, and Significance”, is available upon request. Below, we briefly summarize some additional findings on CMIS data.

A. Most variables are rarely reported, and some are reported with significant lag. 

Figure 2 shows that 73% and 63% of questions were coded zero (or missing) in June and July, respectively. Question detail matters: questions referencing broad categories, such as total number of quarantine shelters (92%) or beds available (87%), or total funding (72%) were more likely to be reported, while very specific questions, such as number of women in quarantine shelter created by a private organization (1.3%) or the total fruits or vegetables distributed (9.4%) were less likely to be reported.

line graph showing percentage of CMIS entries over time
Figure 2:  Percentage of non-zero CMIS entries over time

B. Many variables that are reported are not updated regularly.

Many variables that were reported were not updated over the time referenced in this brief, including those that would be expected to change on a regular (if not daily) basis. For example, among municipalities that reported at least one male in quarantine on June 8, 70% reported the same figure on July 6th. Figure 3 shows that, among selected variables that we would expect to change regularly over time, fewer than 30% of municipalities made alterations between successive report updates, and fewer than 45% made any changes to these values between June and July.

Bar Graph with Municipalities that altered data for selected questions
Figure 3: Municipalities that altered data for selected questions