1Long-run economic growth in Turkey
Sources, pitfalls, and prospects
M. Aykut Attar
Introduction
Living standards in Turkey exhibit sustained growth in the long run. The Maddison Project’s data indicate that, from 1870 to 2016, real Gross Domestic Product (GDP) per capita in purchasing power parity corrected terms increased by about 13 times. For the post-1923 republican era, the average annual growth rate of real GDP per capita is roughly equal to 3% (Bolt, Inklaar, de Jong, & van Zanden, 2018).
Figure 1.1 pictures the long-run evolution of real GDP per capita in Turkey and some selected economies. The performance with an average growth rate of around 3% is not sufficiently strong for Turkey to converge to frontier countries such as the United Kingdom and the United States. But it is also strong enough to let Turkey maintain the relative level difference with these countries. In short, Turkey’s long-run growth story is neither a miracle nor a disaster.
Figure 1.1 Long-run economic growth in selected economies, 1923–2023.
Source: Bolt et al. (2018).
The main objective of this chapter is threefold. It deciphers the sources of intensive economic growth in Turkey from the 1950s to the present. Growth in real GDP per capita originates from productivity growth, human capital accumulation, and fertility decline. The analysis also identifies some pitfalls of the Turkish economy in achieving a long-run growth rate higher than its stable 3% rate. Finally, the analysis focuses on counterfactuals for a quantitative exploration of growth prospects of Turkey for the rest of the 21st century.
The chapter develops an overlapping generations general equilibrium (OLG-GE) model with endogenous technological progress, endogenous human capital accumulation, and endogenous fertility choice. The model is eclectic, merging different theoretical mechanisms that explain economic growth within a single framework. It largely builds upon the simple multi-sectoral Schumpeterian model of Aghion and Howitt (2009, Ch. 4) and a version of the model of child quality-quantity (Q-Q) tradeoff constructed by de la Croix and Gosseries (2012). More specifically, it features horizontal and vertical dimensions of innovation as in the second-generation Schumpeterian (2GS) models. Incentivized via monopoly rents, firms purposefully invest in technology and endogenously choose the optimal level of research intensity. Innovation then leads to productivity increases, and, hence, to growing real GDP per capita. On the side of preferences, adult individuals face the child Q-Q tradeoff as in, e.g. Becker (1960), Becker and Lewis (1973), Becker, Murphy, and Tamura (1990), and Galor and Weil (2000): an adult individual’s lifetime utility is increasing in both the quality of each child, i.e. human capital, and the quantity of children, i.e. fertility. Over the course of economic growth and development, fertility decreases, and human capital per child increases.
The decentralized equilibrium of this model economy is unique and converges to a growth steady-state with a fixed level population that represents the carrying capacity of the economy. This steady-state satisfies two long-run feasibility restrictions emphasized in the recent literature: first, populating the economy with an increasing number of people has an upper limit as fertility responds endogenously to the cost of reproduction that partially depends on land endowment per worker. Thus, as in Peretto and Valente’s (2015) model of growth in a finite planet, economic growth is sustained in the long run with population growth rate being exactly equal to zero. Second, product proliferation has also an upper limit as in Peretto and Connolly’s (2007) model of the Manhattan Metaphor, and the number of firms per capita converges to a fixed level. Since the love-of-variety has no effect on aggregate productivity by construction, economic growth in productivity is entirely due to innovation.
Taking the length of a model period as 20 years, the chapter uses several sources of aggregate data for the Turkish economy for the years 1955, 1975, 1995, and 2015 to calibrate the model’s structural parameters and initial values. As most of these model inputs are analytically or numerically identified, they are calibrated in a rigorous manner using a multi-stage quantitative algorithm in the fashion of Simulated Method of Moments (SMM) to match certain data values. The chapter then uses the calibrated model for the four years (i.e. generations) to decompose growth into the contributions of productivity growth, human capital accumulation, and demographic change.
The analysis simulates the model for the years 2035, 2055, 2075, and 2095. This second part first analyses the evolution of the model economy under the benchmark scenario without any change in fundamentals. Experiments then focus on parameters that more directly affect the Q-Q tradeoff. Next presented are the simulations of the model economy with different constellations of certain parameters that more directly affect innovation processes.
Results indicate that real GDP per capita growth is driven mainly by innovation from 1955 to 1975 but human capital accumulation becomes increasingly more important from 1975 to 2015 and beyond. Among counterfactuals that increase Turkey’s 2015 human capital to the South Korean level, changes in preference parameters imply largest gains. On the other hand, the counterfactual that boosts productivity of R&D yields larger growth and welfare effects than the one that weakens competition among innovative firms.
The most directly related studies in the existing literature analyse economic growth issues by developing models with microeconomic foundations, e.g. Adamopoulos and Akyol (2009), Çiçek and Elgin (2011), İmrohoroğlu, İmrohoroğlu, and Üngör (2014), Voyvoda and Yeldan (2015), and Yılmaz and Saracoğlu (2016). These papers deliver concrete messages about (i) the sources of relative underperformance of the Turkish economy, (ii) how Turkey can avoid being in the so-called middle-income trap (MIT), and (iii) which alternative policies may foster productivity growth in the future. These studies, however, do not develop equilibrium models with endogenous growth where (i) both education and fertility decisions are interactively endogenous and (ii) technological progress has both vertical and horizontal dimensions over which in-house R&D and firm entry take place, respectively. The first point here is essential given that Turkey has experienced a fast demographic transition after the 1950s. Models that assume away fertility and education decisions would irremediably be misleading. The second point should also be a defining characteristic of a growth model of the Turkish economy since Attar’s (2017) model-based evidence implies and the TurkStat’s innovation surveys show that certain fractions of Turkish manufacturing firms are active in vertical innovation and horizontal innovation.
The next section introduces the model economy, the third section calibrates the initial values and structural parameters, the fourth section presents the results, the fifth section discusses these results and associates them with the main messages of related studies, and, finally, the last section concludes with some remarks.
The model economy
This section defines (i) the model environment, i.e. demographic structure, endowments, preferences, and technologies, (ii) market and ownership structures, (iii) decision problems, (iv) market clearing conditions and the static general equilibrium (SGE), and, finally, (v) the dynamic general equilibrium (DGE).
Environment
Time in the model, denoted by t, is discrete and diverges to +∞; t ϵ {0, 1, …}. Demographic structure features two overlapping generations, those who become adults at the beginning of period t and those who are born at the beginning of t as the children of these adults. Fertility is endogenous, and population therefore evolves endogenously. Some assumptions that are not uncommon in the literature simplify the analysis: fertility and population variables are real numbers, reproduction is asexual, and fertility is common across adults. Adult population and the number of children each adult has in period t are denoted by Nt and nt, respectively. Clearly, we have
(1.1)
$${N_{t + 1}} = {n_t}{N_{t}}.$$
Adults have two types of endowments. First, each has one unit of perfectly divisible time endowment to be allocated to leisure activities, child rearing, and labour supply. Besides, each adult in period t has a skill set or human capital stock denoted by ht, originating from the parent’s educational investment decision in period t – 1.
Preferences are represented by a Cobb–Douglas utility function of four arguments. The first is the amount ct ≥ 0 of consumption of the unique all-purpose good produced in this economy. The second is leisure time denot...