In my undergraduate study, I studied physics alongside economics. By the time I got to my third year (and when the financial recession of 2008 started), I observed and learnt that economic theories were not very good at predicting the future. In contrast, physics could dictate, with ridiculous accuracy, the motion or state of a particle or body (though quantum mechanics is a little different than complete accuracy).
In trying to figure out what exactly is wrong with economics theories (particularly macroeconomics) I was looking for an example to illustrate what I felt was wrong with economic theory.
The best example I could come up with has to be the story of how the atom’s structure was settled on.
The initial idea was proposed by J. J. Thompson around the beginning of the 20th century. The atom was imagined to a big region of positive charge with negatively charged electrons embedded in it. A positive pudding with negative plums embedded in it, which is where the name “plum pudding model”1 comes from. A decade or so later Geiger-Marsden had carried out an experiment where they fired helium nuclei at a sheet of gold at the directions of Ernest Rutherford. Rutherford had calculated that nearly all alpha particles would end up passing through the foil without deflection. However, the experiment results showed that nearly 99% passed through and a small fraction had bounced back, which should have been impossible. He figured out that it could be the shape of the atom itself that might be what was incorrectly specified and ran the calculations again assuming a new shape of the atom. Finding his assumption matching reality, he then presented his idea to the world. Note that his theory overturned the previous model after that was found to not be able to sufficiently explain the observations obtained by carrying out an experiment.
The old theory was not simply tweaked to fit the new results, the old theory was completely discarded and replaced with a model that matched the experimental results.
Does this ever happen with economics? No. Most of the theories are temporal and valid in certain conditions, which is why many studies are actually not replicable[2, 3].
What is the major problem with theories not being replicable or being invalidated when conditions change? Not much, just that the theory is invalid and moves into the realm of theology. For it is only in theology that one is required to hold that the theory (or ideology) is true even in the face of evidence against it.
References:
1 Plum pudding model in Wikipedia, https://en.wikipedia.org/wiki/Plum_pudding_model
2 Regarding the reproducibility problems in economics, http://retractionwatch.com/2016/03/03/more-than-half-of-top-tier-economics-papers-are-replicable-study-finds/
3 Economics research not replicable, https://www.timeshighereducation.com/news/papers-in-economics-not-reproducible