Authoritarian governments are kept in power by a small, elite part of their society. When they need additional support (perhaps to deter rebellion) they build domestic institutions - effectively sharing a part of their power in order to buy stability. Dictatorships can also use develop their own legitimacy (without sharing power), preventing people from wanting to resist in the first place.
The Structure of an Authoritarian Regime
Authoritarian regimes are built on the support of small, strategic parts of their society. This is in contrast to a democracy, which requires broad electoral support for a variety of pieces to be effective. For these small interest groups, the standard of living will be relatively high and members may enjoy more freedoms than the average person.
This article by Boix and Svolik contains a discussion of this structure.
This is the first deterrent to overthrowing an authoritarian regime. It's not that the government holds all the valuable resources (including military ones); it's that the government's allies hold many of them.
Of course, this is also a big opportunity: those same allies may be open to defecting.
The government may need additional support, beyond what it can get from domestic allies by dolling out extra economic incentives. When this is the case, the government will invest in developing political institutions (political parties, courts, legislatures, laws, formal departments, etc.).
This is risky, because it means limiting the government's power in exchange for (hoped) stability. For example, a government facing violent conflict from various groups might instate a national legislature with open elections. That will give the rebellious groups the opportunity to participate in a process, providing an outlet for their demands.
The government may or may not obligate itself to follow through with the legislature's proposals, but it's still a marginal victory for the rebels.
The linked paper by Gandhi and Przeworski focuses on providing empirical support for this.
Legitimacy is one of the cornerstones of political stability. Generally, people don't revolt against governments they perceive to be legitimate, even if those governments perform relatively poorly.
Nathan's article discusses this as part of a case study of Chinese politics. There are countless articles about identity politics (examples: Irish nationalism, trans-Africanism, Islamic solidarity) which illustrate how founding mythologies and other stories provide legitimacy for regimes. These actions are especially valuable for dictatorships, which sit atop a fragile network of political alliances.
From the citizen's perspective: economics
It's also worth noting that from the ordinary person's perspective, an authoritarian government might be significantly better than the regime that came before it. Or there might be real fear that the next regime will be worse. Both of these are reasons not to defect.
Of course, there is the possibility that a rebellion will improve things. People in poverty are notoriously risk-averse: any loss of income or resources could put them below sustenance-level (which means they will eventually die of disease, starvation, etc.). Even if the rebellion is successful, it likely won't meaningfully improve things for many years - can the poor afford to wait that long?