Market failures or inefficiencies, the impact of taxes and subsidies, positive/negative externalities are things. Job training may have enough positive externalities such that it would be efficient to subsidize it, but not because of a concept of fairness to someone. Economic actors should be rational enough to seek it out themselves if it maximizes their net benefit.
Indeed, however economics does allow you to input your own definition of what 'fair' is, and it makes suggestions on how to fulfill that definition in the most efficient way.
In terms of % of GDP spent on public spending on the labor market, US ranks near the bottom. When you look at spending on job training directly, US spends 0.03% of GDP, while Germany spends seven times as much as a fraction of GDP [1].
A better comparison would dollars per unemployed person.
The unemployment rate in the US and Germany are very similar (4-5%), so Germany spends 7 times as much training per unemployed person.
The most galling thing about this is that there is a reasonable compromise here, which is that you let the market decide things properly then you fairly pay off the people who were negatively affected, by intensive job training, relocation, early retirements + pensions, etc. For example compared to Germany, the job training in the US is a complete joke.
But American (and especially Republican) ideology of the rugged frontier individualist would never allow the blow to the psyche that would come from officially being on the 'government dole' or admitting needing help from anyone else.
Hence we have all these disguised welfare schemes (defense contractor jobs are another example) that are economically inefficient and environmentally damaging, mostly for reasons of personal and regional pride.