WEBVTT NOTE Success from Satisficing and Imitation: Entrepreneurs’ Location Choice and Implication of Heuristics for Local Economic Development 1 00:00:00.000 --> 00:00:07.000 This paper is motivated by a broader question about the potential value to Austrian economics, 2 00:00:07.000 --> 00:00:12.000 to asking questions about the decision process that entrepreneurs use. 3 00:00:12.000 --> 00:00:22.000 On the one hand, von Mises and Rothbard set out to carefully delineate what it is they do 4 00:00:22.000 --> 00:00:28.000 from whatever it is that other stuff that psychologists and decision theorists do. 5 00:00:28.000 --> 00:00:36.000 There's certainly some clear statements against the idea of opening the black box of what goes on in the mind. 6 00:00:36.000 --> 00:00:46.000 But on the other hand, when we encounter Hayek's The Sensory Order and see his keen interest in psychology, in models of mind, 7 00:00:46.000 --> 00:00:54.640 and then one can't help but be interested in how Austrian market process relates to Hayekian 8 00:00:54.640 --> 00:01:01.600 and other Austrian ideas about the way the mind works. 9 00:01:01.600 --> 00:01:07.120 Kirzner and Schumpeter in places they seem to write dismissively about psychology and 10 00:01:07.120 --> 00:01:15.980 decision process and certainly the overriding focus in the entrepreneurial studies in the 11 00:01:15.980 --> 00:01:23.580 The Austrian literature focuses on action, judgment and the nice taxonomy that you presented 12 00:01:23.580 --> 00:01:25.740 for us. 13 00:01:25.740 --> 00:01:29.500 What I'm hoping to do is propose a research program and show you what I've been doing 14 00:01:29.500 --> 00:01:36.460 of talking to entrepreneurs and thinking about what this might portend in terms of potential 15 00:01:36.460 --> 00:01:38.620 for future Austrian research. 16 00:01:38.620 --> 00:01:44.380 I would submit that I'm certainly undertaking to analyze the action of entrepreneurs but 17 00:01:44.380 --> 00:01:49.780 but also to complement the study of action with a more detailed description of underlying 18 00:01:49.780 --> 00:01:53.720 decision process. 19 00:01:53.720 --> 00:01:55.100 So what are the questions? 20 00:01:55.100 --> 00:01:59.500 I wanted to know how entrepreneurs choose the locations that they do. 21 00:01:59.500 --> 00:02:03.340 And one thing that's different than in the discussion we just heard was that the tacit 22 00:02:03.340 --> 00:02:05.420 knowledge is quite public. 23 00:02:05.420 --> 00:02:11.420 When I choose a location, then it's quickly observed by competitors in the market. 24 00:02:11.420 --> 00:02:19.140 I want to know how long do they search? How many locations do entrepreneurs, good ones, 25 00:02:19.140 --> 00:02:24.580 consider before finalizing a decision? What constitutes a stopping rule? What are the 26 00:02:24.580 --> 00:02:30.180 metrics for evaluating what a good location is? Certainly we wouldn't from an Austrian 27 00:02:30.180 --> 00:02:35.580 perspective expect it to be one thing, although in the neoclassical literature we tend to 28 00:02:35.580 --> 00:02:41.580 to reduce this to moments of known probability distributions, expected payoffs. 29 00:02:41.580 --> 00:02:46.580 And then we want to know what are the criteria that entrepreneurs use to stop search 30 00:02:46.580 --> 00:02:52.580 and then finally make a decision how to bet their own personal capital in choosing a location. 31 00:02:57.580 --> 00:03:01.580 If you go to the Economics 101 textbook, 32 00:03:01.580 --> 00:03:09.020 The benefit-cost calculus presumes that the decision maker has a known menu of alternatives. 33 00:03:09.020 --> 00:03:20.740 So if I were buying a house for a budget of $200,000, then the idea here is that I would have an exhaustive list of all the affordable homes everywhere, all continents, 34 00:03:20.740 --> 00:03:30.100 and that presumably there could be some hedonic attributes of living in a faraway place that would outweigh or offset the costs of going there. 35 00:03:30.100 --> 00:03:38.220 So, the standard textbook model is that exhaustive search has to be undertaken, that each element 36 00:03:38.220 --> 00:03:44.340 in this choice set has to be scored numerically, usually by expected payoff, without saying 37 00:03:44.340 --> 00:03:46.860 exactly what payoff means. 38 00:03:46.860 --> 00:03:51.540 And then, by definition, we hear this also from Austrian economists, interestingly, that 39 00:03:51.540 --> 00:03:58.660 choice is definitionally the selection of the best feasible alternative, and I'm hoping 40 00:03:58.660 --> 00:04:04.660 I wanted to enlarge that discussion a little bit based on my collaborations with psychologists and biologists. 41 00:04:04.660 --> 00:04:13.660 Well, neoclassical economists realized that smart people sometimes stop searching before the entire menu of alternatives is completed. 42 00:04:13.660 --> 00:04:18.660 That gave rise to optimal search theory. Many of you probably know the secretary problem, 43 00:04:18.660 --> 00:04:23.660 which was an early example of statistics that led to the 1 over E rule. 44 00:04:23.660 --> 00:04:53.660 So, basically, if you've got a list of unknown job candidates in a known sequence, and the only goal is to maximize the probability that you'll choose the best one, then it can be proved, and mathematical economists like to prove this, that the optimal stopping rule is you look up about a third of them and then pick the next dominating alternative that comes down the pike, 45 00:04:53.660 --> 00:05:01.660 Most choice problems that entrepreneurs face, though, don't simply have to do with durations of search, they also have to do with paths of search. 46 00:05:01.660 --> 00:05:09.660 In the secretary's problems, it's assumed that there's an exogenously given list of alternatives and they come in a fixed order. 47 00:05:09.660 --> 00:05:21.660 So, the proponents of neoclassical search theory, in trying to get a more realistic model of the choice problem 48 00:05:21.660 --> 00:05:26.660 This is a problem that people under time and decision cost pressures make. 49 00:05:26.660 --> 00:05:35.660 They make even more heroic assumptions about the givens in terms of the data that are available to the presumed users of this kind of optimization theory. 50 00:05:35.660 --> 00:05:41.660 And you can even think about it, if I just have three alternatives, then there would be three possible durations of search. 51 00:05:41.660 --> 00:05:48.660 I could look up all three, I could look up two, I could look up one, but then there are multiple paths I can go. 52 00:05:48.660 --> 00:05:55.300 and when you get into the combinatorics of the search path space combined with the duration length 53 00:05:55.300 --> 00:05:58.900 then the choice problem just becomes mathematically more formidable 54 00:05:58.900 --> 00:06:03.140 and computer scientists refer to this kind of problem as NP-hard, it's not computable 55 00:06:03.140 --> 00:06:09.060 for as the length of and as the sizes of choice sets get bigger 56 00:06:09.060 --> 00:06:12.660 and nobody, no supercomputer could compute it 57 00:06:12.660 --> 00:06:17.700 and we don't expect people to actually do it so then neoclassicals 58 00:06:17.700 --> 00:06:22.940 put this forward as a good as-if story. 59 00:06:22.940 --> 00:06:27.060 Okay, here's the world I'm interested in, the one that 60 00:06:27.060 --> 00:06:30.740 all of our heroes in the Austrian canon talk about 61 00:06:30.740 --> 00:06:34.500 under various labels, profound uncertainty 62 00:06:34.500 --> 00:06:38.420 I'm talking about. We don't know all the possible outcomes and we certainly don't 63 00:06:38.420 --> 00:06:39.180 have a list 64 00:06:39.180 --> 00:06:43.100 of probabilities. One can draw on 65 00:06:43.100 --> 00:06:46.700 numerous sources, I just listed a few. 66 00:06:46.700 --> 00:06:57.260 My point of departure here is to get into the behavioral economist's idea about how to deal with this problem. 67 00:06:57.260 --> 00:07:03.620 So I was trained as a mathematical modeler and I began working on models of financial market equilibrium 68 00:07:03.620 --> 00:07:06.100 where people sometimes believe the wrong thing. 69 00:07:06.100 --> 00:07:12.500 And I was intrigued with interesting results that sometimes equilibrium requires everybody to believe something different 70 00:07:12.500 --> 00:07:19.780 and sometimes you can get Pareto dominating equilibria when everybody believes the wrong thing. 71 00:07:19.780 --> 00:07:29.140 And this calls into question some of the basic normative tenets about the whole rational explanations conversation in the neoclassical theory. 72 00:07:29.140 --> 00:07:34.060 Now, when I first went to UT Dallas 13 years ago, I ran into Peter Lewin, 73 00:07:34.060 --> 00:07:46.620 an Austrian economist who began mentoring me and finding a richer literature to probe my interest in uncertainty and subjective beliefs. 74 00:07:46.620 --> 00:07:54.940 And I was always intrigued with this kind of this tension between a professed lack of interest in the introspective mind, 75 00:07:54.940 --> 00:08:00.620 but also a recognition in this radical heterogeneity of goals, of information sets 76 00:08:00.620 --> 00:08:06.700 that seems to have so much potential for the descriptive richness of models of mind. 77 00:08:07.900 --> 00:08:18.300 So, I could talk about a gazillion behavioral economists that have, at various stages in the arc of behavioral economics, 78 00:08:18.300 --> 00:08:23.260 tried to flirt with various authors in the Austrian tradition 79 00:08:23.260 --> 00:08:31.260 and then quite a few well-known Austrian authors, Lachmann and Buchanan and Van Berg. 80 00:08:31.260 --> 00:08:38.960 You'll see talking about taking psychology seriously and being interested in the decision process of entrepreneurs. 81 00:08:38.960 --> 00:08:45.160 I'll leave it there and move on to what the behavioral economists have done. 82 00:08:45.160 --> 00:08:54.840 I think there's a huge arrogance and a methodological misguidedness in the so-called behavioral economics revolution. 83 00:08:54.840 --> 00:09:03.800 So this was the methodological fight. When I was a PhD student, I was interested in working with otherwise standard neoclassical models, but where the beliefs were wrong. 84 00:09:03.800 --> 00:09:09.000 And already that was considered exotic, weird, behavioral. 85 00:09:09.000 --> 00:09:15.480 Then in the last 15 years, we've had this self-described methodological opening up. 86 00:09:15.480 --> 00:09:19.640 Now we can dump everything into the utility function, exotic preference models, 87 00:09:19.640 --> 00:09:27.960 and the self-description of the people who are editing the top neoclassical, 88 00:09:27.960 --> 00:09:33.000 or formerly neoclassical journals, now profess a new kind of methodological openness. 89 00:09:33.000 --> 00:09:41.000 But my point would be that this is more or less a retrenchment of a very rigid, singular methodological norm 90 00:09:41.000 --> 00:09:43.500 based on the mathematics of constrained optimization. 91 00:09:43.500 --> 00:09:49.000 What's new is we have a few new inputs in the utility function, we have a few new decision parameters, 92 00:09:49.000 --> 00:09:54.500 asymmetric weighting of gains and losses, you well know the work of Danny Kahneman 93 00:09:54.500 --> 00:09:59.500 and this claim of widespread irrationality. 94 00:09:59.500 --> 00:10:09.260 My point of view is that by deviating from the axioms, the rationality axioms of the neoclassical model, 95 00:10:09.260 --> 00:10:21.860 this is sometimes required to be an adaptive, smart, intelligent agent acting purposefully the way that Austrian heroes have described. 96 00:10:21.860 --> 00:10:29.540 and I have tried to write about this expansion of what counts as a legitimate question in economics 97 00:10:29.540 --> 00:10:36.260 and a paradoxical hardening of its normative interpretation that really everybody should be 98 00:10:36.260 --> 00:10:41.460 maximizing something in a small world that's so simple that we can list all the alternatives, 99 00:10:41.460 --> 00:10:47.380 write down probability distributions, describe payoff functions. I don't believe those models 100 00:10:47.380 --> 00:10:54.660 Work, and many of the people contributing to this have a diverse range of perspectives 101 00:10:54.660 --> 00:10:58.540 about it, but what they typically come back to is after putting a bunch of exotic stuff 102 00:10:58.540 --> 00:11:05.620 in the utility function, they hold the neoclassical axioms as the right normative model for what 103 00:11:05.620 --> 00:11:10.380 we all should be aspiring to, that business schools should be teaching everybody to maximize 104 00:11:10.380 --> 00:11:15.220 their expected utility theory, and if you look at the thousands of papers that are written 105 00:11:15.220 --> 00:11:21.700 and non-Basian behavior and loss aversion, standard behavioral topics, you might get 106 00:11:21.700 --> 00:11:25.760 the feeling that there's well established that there are billions and trillions of dollars 107 00:11:25.760 --> 00:11:30.140 of losses out there to the economy simply because people aren't maximizing the right 108 00:11:30.140 --> 00:11:31.400 thing. 109 00:11:31.400 --> 00:11:34.060 You can look wide and far for that empirical evidence. 110 00:11:34.060 --> 00:11:41.940 There's actually almost no data at all linking pathological decision process to any kinds 111 00:11:41.940 --> 00:11:47.940 and the kinds of norms that an Austrian would care about, accumulated wealth, life span, happiness. 112 00:11:47.940 --> 00:11:54.940 I will leave it at that and hope we can continue discussion about that. 113 00:11:54.940 --> 00:11:59.940 So what I did is I started talking to entrepreneurs in Dallas. 114 00:11:59.940 --> 00:12:09.940 A few of them, 17 of them had very large projects, over a million of their personal capital invested 115 00:12:09.940 --> 00:12:16.560 and the rest, smaller projects, and what I was most interested in was eliciting a 116 00:12:16.560 --> 00:12:22.580 description about how the choice of location was made, the length of 117 00:12:22.580 --> 00:12:27.020 search, the number of items considered, and what kinds of criteria were used to 118 00:12:27.020 --> 00:12:32.900 stop the search and finally make the decision. Here is a typical account. The 119 00:12:32.900 --> 00:12:37.860 idea struck me as I was driving by that area that it could be developed into a 120 00:12:37.860 --> 00:12:52.060 Property of Note. I told my husband to drive by to get a feel for the area. We liked it. It felt right. Then I ran the numbers and it looked like we could get at least 20% any return on capital within two or three years. That was enough to make it worthwhile to go ahead. 121 00:12:54.060 --> 00:13:07.740 You'll notice there's nothing about equating marginal benefit with marginal cost or considering all the possible search lengths and finding the optimal duration or the optimal path of search. There's 122 00:13:07.860 --> 00:13:18.860 and, instead, a discussion of threshold conditions that have to be fulfilled in order for this decision to be okay. 123 00:13:18.860 --> 00:13:25.860 So this, to me, looks a lot like the satisfying behavior that Herb Simon described. 124 00:13:25.860 --> 00:13:34.860 And I think this raises some methodological problems in some parts of the Austrian literature. 125 00:13:34.860 --> 00:13:37.380 And I'm eager to have conversations about this. 126 00:13:37.380 --> 00:13:41.380 The satisfying implication of this tells me 127 00:13:41.380 --> 00:13:43.900 that if you live in an opportunity-rich world, 128 00:13:43.900 --> 00:13:46.140 there are many, many ways to be successful. 129 00:13:46.140 --> 00:13:47.980 There are many good enough paths. 130 00:13:47.980 --> 00:13:50.260 Therefore, the imperative should not 131 00:13:50.260 --> 00:13:54.620 be to maximize something written down in a small world, 132 00:13:54.620 --> 00:13:59.820 but to take this massively high-dimensional action space 133 00:13:59.820 --> 00:14:02.540 and simply find a good enough path. 134 00:14:04.860 --> 00:14:14.060 Here is the empirical distribution of the number of elements in the consideration set. 135 00:14:14.060 --> 00:14:28.260 The entrepreneur is considered very few locations and this is kind of what you'd expect if you are reading Schumpeter and talking about a man of action and intuition. 136 00:14:28.260 --> 00:14:36.060 Nine of the 49 looked at only one, did not even think about opportunity cost or next best. 137 00:14:36.060 --> 00:14:41.160 The modal number of items considered was three. 138 00:14:41.160 --> 00:14:43.160 That may be enough. 139 00:14:43.160 --> 00:14:49.360 I probed, so I was using this scripted interview with numerous follow-up questions, 140 00:14:49.360 --> 00:14:55.260 trying to get as rich an informational set as I could about the self-described decision process. 141 00:14:55.260 --> 00:14:57.960 And the kinds of things that you heard from these entrepreneurs, 142 00:14:57.960 --> 00:15:06.960 They were sophisticated quantitatively. This is not a function of being incapable to collect information or not being good at numbers. 143 00:15:06.960 --> 00:15:13.560 Quite to the contrary, we even talked to a couple consulting firms that do do a kind of exhaustive search using census data, 144 00:15:13.560 --> 00:15:19.160 going down to the block group level and linking those with real estate openings. 145 00:15:19.160 --> 00:15:27.460 So there are services out there. If you wanted to do an exhaustive search of a state or of a city, it certainly can be done. 146 00:15:27.460 --> 00:15:31.900 I didn't find anybody who was remotely interested in that, and then when I asked point blank, 147 00:15:31.900 --> 00:15:36.700 well, why didn't you search more, or did you think about searching more, then I would often 148 00:15:36.700 --> 00:15:43.060 get the response, oh, I know what you're thinking, you're thinking about the optimization 149 00:15:43.060 --> 00:15:47.900 model, you're thinking, many of them had studied economics, and they said, yeah, you're probably 150 00:15:47.900 --> 00:15:52.020 wondering why more wasn't better. 151 00:15:52.020 --> 00:15:56.580 And you have to understand in the kind of environment I live in, things are changing 152 00:15:56.580 --> 00:15:58.580 The Theory of Money and Credit 153 00:16:26.580 --> 00:16:40.580 I'm interpreting this as a purpose of openness to serendipity and evidence of the kind of alertness and intuition that one finds discussed in Austrian writings on entrepreneurial behavior. 154 00:16:40.580 --> 00:16:51.580 The purpose of openness to serendipity and evidence of the kind of alertness and intuition that one finds discussed in Austrian writings on entrepreneurial behavior. 155 00:16:51.580 --> 00:16:59.580 Although I'm quite aware there would be alternative interpretations and would welcome some conversation about that. 156 00:16:59.580 --> 00:17:06.580 Okay, the threshold conditions. Nobody said they were setting any two variables equal to each other. 157 00:17:06.580 --> 00:17:18.580 I need 20% return on investment, I need an occupancy rate over 80%, if I think I can get X return after Y years, where Y is one year, two years, maybe five years. 158 00:17:18.580 --> 00:17:27.580 This is the kind of conditional threshold that defines nearly everybody I talk to as stopping rule. 159 00:17:27.580 --> 00:17:39.580 Imitation. This is something I know that bothers a lot. Some writers in the Austrian traditions considered too automatic, too much of an automaton kind of algorithmic rule. 160 00:17:39.580 --> 00:17:47.580 It doesn't seem to draw on the intuition, great man, great woman leader that we write about. 161 00:17:47.580 --> 00:17:55.580 What I found though is that the smaller projects, once I asked about their self-described return, 162 00:17:55.580 --> 00:18:02.580 whether their returns exceeded expectation, met expectation, fell below expectation, 163 00:18:02.580 --> 00:18:06.580 many of the smaller projects did quite well imitating. 164 00:18:06.580 --> 00:18:13.580 And there was lots and lots of talk about the presence of Starbucks or a laundromat or a drugstore nearby. 165 00:18:13.580 --> 00:18:20.580 People choosing locations are very conscious of nearby business activity. 166 00:18:20.580 --> 00:18:27.580 So many people mentioned wanting Starbucks nearby so that their employees would have good coffee to drink 167 00:18:27.580 --> 00:18:30.580 that it makes you wonder why there's even positive rent. 168 00:18:30.580 --> 00:18:33.580 These guys should be paid to locate. 169 00:18:33.580 --> 00:18:37.580 If you're selling real estate or renting, 170 00:18:37.580 --> 00:18:45.580 It's questionable whether Starbucks is actually earning a fair return on its positive spillover effects. 171 00:18:45.580 --> 00:18:50.580 Okay, then I told you a little bit about this, and I will cut to the chase. 172 00:18:50.580 --> 00:19:01.580 There was one of the things that, as an Austrian, I have a PhD student that became head of local economic development initiatives in Dallas. 173 00:19:01.580 --> 00:19:13.580 The cities often want to do things like target tax incentives to motivate and incentivize firms to invest in particular sectors of the city. 174 00:19:13.580 --> 00:19:26.580 In Dallas, there are big islands that are devoid of any commerce or any visible commerce and of frustratingly small responsiveness to tax incentives. 175 00:19:26.580 --> 00:19:29.980 And then we also ask questions about this. 176 00:19:29.980 --> 00:19:33.580 Most of the people we talked to said they would never put it in their consideration set. 177 00:19:33.580 --> 00:19:37.180 You could give me rent for free for a year, I still wouldn't consider it. 178 00:19:37.180 --> 00:19:49.580 Which really raises questions about whether these small marginal incentives have any chance of getting a location that's not in the consideration set into the consideration set. 179 00:19:49.580 --> 00:19:54.980 Of course their thinking is that if you see a neighborhood that has no profit, that has no business activity, 180 00:19:54.980 --> 00:19:58.020 Then if everybody's profit maximizing and everybody does exhaustive search, 181 00:19:58.020 --> 00:20:02.540 then that means that those places have no opportunity. 182 00:20:02.540 --> 00:20:05.780 But then if you talk to the way entrepreneurs actually make their decisions, 183 00:20:05.780 --> 00:20:10.380 many of them wisely economizing on search costs and information 184 00:20:10.380 --> 00:20:12.820 and accumulating in well-known neighborhoods, 185 00:20:12.820 --> 00:20:16.980 then this does portend the possibility that there could be profitable opportunities 186 00:20:16.980 --> 00:20:23.100 in unexploited neighborhoods that go unexploited for a long time 187 00:20:23.100 --> 00:20:25.100 simply because of the decision process 188 00:20:25.100 --> 00:20:27.300 that the entrepreneurs are using. 189 00:20:27.300 --> 00:20:28.900 So, I will leave it there 190 00:20:28.900 --> 00:20:32.420 and hope we have a continued conversation 191 00:20:32.420 --> 00:20:35.260 about decision process in Austrian economics. 192 00:20:35.260 --> 00:20:36.340 Thanks. 193 00:20:36.340 --> 00:20:38.340 Thank you very much.