My dissertation investigates how economic growth is determined in the long run. To do this, I mainly focus on culture and institutions as fundamental growth factors and develop a general theoretical framework in which culture, institutions and growth are endogenously determined. Using the framework, I highlight the crucial role of the interaction between culture and institutions in the long-term growth. Chapter 1. Endogenous Financial Friction and Growth 2 investments, resulting in lower

more »

... We call this a direct growth effect. 3 Notably, this negative growth effect disappears as the economy grows in our model. This is because the credit market becomes asymptotically efficient although an entrepreneur borrowing for R&D investment is assumed to be credit constrained. 4 This is because the incentive constraint is endogenously relaxed as the borrower's profit (entrepreneurial profit) increases along with growth of the economy. 5 This implies that the negative growth effect emerges during the transition phase only, and thus, the growth effect is temporary. We call this asymptotic irrelevance to growth. This finding contradicts the permanent growth effect of financial frictions. If financial frictions adversely affect growth, then this negative impact is permanent in the previous literature adopting the exogenous borrowing constraint as long as it is assumed to be binding. This is simply because the tightness of the financial friction is arbitrarily given and fixed at a certain level. In contrast, the tightness is endogenously determined in our framework, and thereby, the negative growth effect disappears gradually as the economy grows. 6 Hence, the growth effect of the financial friction is heterogeneous across countries 3 Similar growth effects of financial frictions, led by the inefficient resource allocation, can be found in, for example, de Gregorio (1996) , Berentsen, Rojas Breu and Shi (2012), Bencivenga and Smith (1991) Aghion, Howitt and Mayer-Foulkes (2005), Trew (2014) and Laeven, Levine and Michalopoulos (2015) among others. In line with this theoretical result, there is extensive empirical evidence that consistently shows negative growth effects of financial frictions. Some prominent examples include King and Levine (1993) and Levine, Loayza and Beck (2000) . 4 Recently, Azariadis, Kaas and Wen (2016) develop a parsimonious model that features endogenous credit friction led by limited commitment. In the model, the credit market operates efficiently provided that the reputation value is large enough for a borrower to honor his debt voluntarily. 5 This intuition seems plausible from the observation that the average firm size correlates positively with income per capita (the average income of a country), e.g., Poschke (2014). Since a firm's profit is increasing, on average, in its size, the average firm will have higher profits in a richer country. Similarly, Donovan (2014) empirically shows that there are more entrepreneurs who are "poor" (subsistence entrepreneurs) in a poorer country. 6 This endogenous relaxation is equivalent as the model-implied tightness decreases, converging asymptotically to the steady state value that guarantees the first-best optimum. Looked at another way, we theoretically show that the interest rate spread, which proxies the severity of the financial friction, decreases along with growth due to the endogenous relaxation of the financial friction. Galor and Zeira (1993) also derive the spread of borrowing and lending rates led by the borrower's moral hazard problem and connect the interest rate spread to the degree of the financial friction severity. However, the interest rate spread in their model is essentially exogenous since it is independent of the state of the economy, and therefore, the negative effect of the financial friction persists forever in their framework. 10 We also look at the case of a closed economy briefly in Appendix A.1. 11 This implies that F (x) is strictly increasing in z. 12 These contrasting features are similar to those of AABM. 7 whether or not invest in R&D in order to enhance his own productivity level A i after drawing his individual ability z i . Investing in R&D requires x (z, A) amount of consumption goods. We assume that 0 denotes the total factor productivity (TFP) in period t. The last assumption reflects that it is more costly to conduct R&D innovations as the economy develops further. 13 We denote x t (z) ≡ x (z, A t ) for convenience. R&D investment is risky. Once investment has begun, he succeeds in the R&D project with probability p ∈ (0, 1) and obtains γA t as individual productivity where γ > 1. 14 For simplicity, we assume that the entrepreneur obtains zero productivity if he failed in his R&D project, i.e., A i = 0, so that produces and consumes nothing in the following period t + 1 (he goes underground). 15 Meanwhile, an entrepreneur who does not invest has the same individual productivity with the TFP given by A t . Finally, entrepreneurs are risk-neutral, and they consume at the end of second period. Then, an entrepreneur with z solves the following problem: 13 Jones (2008) shows with empirical evidence that successive generations of innovators have a heavier educational burden as technology develops further. Similarly, the last condition reflects that it is harder to imitate or adopt more advanced technologies from other countries as the economy develops since the TFP, A t , increases over time through successful R&D projects. Technically, this assumption is necessary for stationary growth and, in fact, widely adopted in many other studies. For instance, the R&D cost function is assumed to be linear in A t for stationarity in AHM and Acemoglu, Aghion and Zilibotti (2006) . As shown later, the linearity is a sufficient condition for stationary growth in our model; see Proposition 1.3. 14 One can introduce project varieties such that G = {γ 1 , γ 1 , · · · , γ N } where γ i > γ j , i > j, and p i ≡ Pr(ith project is done successfully) < 1. Here, for simplicity, we consider the simplest case, N = 1, since there is no significant change in the qualitative results. 15 This assumption can be easily relaxed such that the entrepreneur can still produce some amount of consumption goods. However, there is no change in the results qualitatively. 8 We denote the gross borrowing rate by 1 + i. The subscript 1 indicates that the producer is a (successful) R&D investor while the subscript 0 is for the one who did not invest. The gross borrowing rate 1 + i t+1 is required to be equal or larger than the gross return to capital 1 + r t+1 − δ > 0; if not, all entrepreneurs can make profits arbitrarily by borrowing consumption goods at 1 + i t+1 from a lender and lending them at 1 + r t+1 after depreciation by δ. That is, the net borrowing rate i has the lower bound at r − δ in any equilibrium to preclude the arbitrage opportunity. Then, entrepreneur i's borrowing is fixed at x (z i , A), and therefore, only inputs are choice variables.