Economists are taught strange things.
Their problem here is that for example they see going from $400$ to $500$ as $+25\%$ but going from $500$ to $400$ as $-20\%$, and they do not like this as it disrupts their elasticity calculations.
A solution which could work for them would be to use logarithms, and here $\log_e(500)-\log_e(400) =\log_e(1.25)\approx 0.22314$ and in the opposite direction $\log_e(400)-\log_e(500) =\log_e(0.8) \approx -0.22314$.
But that is a bit complicated to explain and is not really a percentage change. So instead they redefine percentage change (why not - they already plot price and quantity a strange way round on the axes, mix superscripts with exponents in equations, and see no need in general to be consistent with mathematicians or statisticians), and here get $\frac{500-400}{450}\approx 0.22222$ and $\frac{400-500}{450}\approx -0.22222$ which they think is close enough and they are willing to call a percentage change.