You base your estimates of the ERP on US history alone
Your ERP estimate is too high because your calculation suffers from survivorship bias and errs by including the largest economy for which the most data is currently available rather than a random sample.
The US has been a spectacularly successful economy over the last hundred years. At the turn of the previous century, Imperial Russia and the US were at about the same level of economic development. Investing in Imperial Russia would have given you a -100% return. If we estimate the ERP from Imperial Russia, we get a very different answer than if we calculate it based on US data.
The US is the largest economy today. It is not a good place to estimate an average, prospective ERP.