In a sense ETF's are valued in exactly the same way as the companies are., The value of the ETF should closely track the Sum of the values of the shares it owns. If it doesn't the ETF Sponsor will arbitrage the discrepancy. Here is a simplistic example. ETF owns 100 Shares of company A worth $10,000 and 100 Shares of company B worth $20,000. The ETF trades at $25,000, not the $30,000 the assets are worth. So, the ETF Sponsor buys all of the ETF's own shares in the market at $25,000, and, sells 100 shares of A&B short for $30,000, borrowing the shares from the ETF. The selling and buying pressure cause the prices to converge.. If it doesn't want to wait for the convergence, it can unwind the structure, leaving it with 100 Shares of A&B, worth $30,000, that it bought for $25,000, and it covers the short.

The better question is why a Closed-End Fund, which has a fixed amount of capital, wouldn't trade at a pice equal to NAV. They usually trade at discount to NAV, as much as 15% to 20%