"Labor intensive" has a different meaning in places like Sri Lanka than in (e.g.) the US (though I doubt 98% of Sri Lanka still slaves away in the mud). "Costs more", similarly, is probably different.

By the report, the sudden switchover made a 20% reduction in absolute yield, for whatever was the reduction in inputs. That is not surprising for any un-pilot-tested step function change in process. Whether it is really a disaster depends on the cost of the inputs abandoned, which was not reported.

There are very strong incentives for this to be treated and reported as a disaster. Omitting key details from the reporting, as here, helps establish that narrative.